The role of mechanical forces in
osteogenic differentiation, BMP signaling
and early tissue formation processes in the
context of bone healing
vorgelegt von
Dipl.-Ing. Sophie Görlitz
geborene Schreivogel
ORCID: 0000-0002-2302-8351
an der Fakultät III – Prozesswissenschaften der
Technischen Universität Berlin
zur Erlangung des akademischen Grades
Doktor der Ingenieurwissenschaften
-Dr.-Ing.-
genehmigte Dissertation
Promotionsausschuss:
Vorsitzender: Prof. Dr. Lorenz Adrian
Gutachter: Prof. Dr. Jens Kurreck
Gutachter: Prof. Dr. Roland Lauster
Gutachter: Prof. Dr. Georg N. Duda
Tag der wissenschaftlichen Aussprache: 02. Juni 2020
Berlin 2021
Für meine Familie
Danksagung
Mein ganz besonderer Dank geht an Dr. Ansgar Petersen, der mir nicht nur ermöglicht hat
den Weg der Promotion einzuschlagen, sondern mir auch während dieser Zeit ein
hervorragender Mentor war. Durch seine Vorschläge und konstruktive Kritik haben sich diese
Arbeit, aber auch vorherige Präsentationen und Veröffentlichungen entscheidend verbessert.
Weiterhin hat er durch die Zusammenstellung einer internationalen und interdisziplinären
Arbeitsgruppe, deren Mitglieder sich sehr gut ergänzen, ein konstruktives und stimulierendes
Arbeitsumfeld geschaffen. Ich bin mir sicher, dass mir das Wissen und die Fähigkeiten, die ich
unter seiner Betreuung erworben habe, auf meinem weiteren Weg sehr helfen werden.
Besten Dank für deine wertvolle Unterstützung.
Ein großer Dank geht auch an Prof. Georg Duda. Neben seinen wertvollen wissenschaftlichen
Beiträgen, habe ich durch ihn einen guten Einblick in die Mechanismen der akademischen
Welt erhalten. Weiterhin möchte ich mich bei Prof. Petra Knaus für ihre Unterstützung
bedanken. Die Zusammenarbeit mit ihr und ihrer Arbeitsgruppe hat fruchtbare Ideen in mein
Projekt gebracht. Auch Prof. Jens Kurreck danke ich, dass er die Betreuung dieser Arbeit aus
universitärer Seite übernommen hat.
Natürlich möchte ich mich auch ganz herzlich beim gesamten jetzigem und ehemaligem CBM
Team bedanken. Ihr habt meinen Arbeitsalltag enorm bereichert und ich habe die zahlreichen
Frühstücke, Ausflüge und Abendveranstaltungen mit euch sehr genossen. Insbesondere
möchte ich mich bei Aaron Herrera und Erik Brauer für die Zusammenarbeit bedanken. Ich
konnte immer auf eure Unterstützung zählen und weiß eure Meinungen und Vorschläge sehr
zu schätzen. Auch die verscheiden wissenschaftlichen Konferenzen, die wir gemeinsam
besucht haben, werde ich in sehr guter Erinnerung behalten. Weiterhin bin ich sehr froh, dass
Isabel Orellano, Ehsan Soodmand, Martina Tortorici, Janina Stadter und Parisa Khalaghi zum
Team dazu gestoßen sind. Mit euch macht es einfach Spaß zusammen zu arbeiten. Auf keinen
Fall darf ich Gabriela Korus und Iwona Cichocka vergessen. Sie haben mich nicht nur in die
Welt der Histologie und Zellkultur eingeführt, sie hatten auch immer eine helfende Hand und
ein offenes Ohr. Ich freue mich auch, dass unser Team nun durch Simone Cho unterstütz wird.
Die tägliche Labor- und Büroarbeit lebt von der gegenseitigen Unterstützung, daher möchte
ich mich auch bei Maria Reichenbach, Aline Lückgen, Christian Bucher, Dag Wulsten, Thomas
Sips, Julia Berkmann, Taimoor Hasan Qazi, Janosch Schoon und so vielen anderen bedanken.
Meinen allergrößten Dank verdient meine Familie, insbesondere mein Mann Richard, meine
Eltern, meine Schwester und meine Großeltern. Ich kann mir keine bessere wünschen. Bei
euch kann ich Ballast abwerfen und neue Energie tanken. Ihr seid immer für mich da und
stärkt mir den Rücken. Auch meinen Schwiegereltern, meiner Schwägerin und meinem
Schwager danke ich von Herzen für ihre Unterstützung. Ohne euch alle wäre ich nicht da wo
ich jetzt bin.
Abstract 1
Abstract
During bone fracture healing, cells are simultaneously subjected to extrinsic mechanical
forces and to a variety of biochemical signals including the indispensable and clinically
applied growth factor Bone Morphogenetic Protein 2 (BMP-2). In vivo experiments provide
evidence that mechanical forces promote BMP-2-induced bone defect healing and in vitro
studies report about a potentiation of BMP signaling by mechanical stimuli. Clinically, supra-
physiological BMP-2 concentrations are used for fracture treatment, which can cause various
side effects. Fine-tuned mechanical stimuli, either resulting from extrinsic loading or featured
by advanced biomaterials, could in future improve the growth factor application by increasing
its efficiency. However, to employ the power of the mechano-biochemical interaction, a
deeper understanding how both stimuli control cell behavior independently and in
combination is needed.
In this dissertation, the influence of extrinsic mechanical forces and BMP-2 on osteogenic cell
differentiation and early tissue formation processes was investigated in vitro and the
molecular mechanism underlying the mechano-regulated BMP signaling were explored. To
realize in vivo loading scenarios in the well-controlled environment of an in vitro screening
system, mechano-bioreactors in combination with 3D biomaterial matrices were utilized
throughout this study.
In contrast to data from literature, osteogenic differentiation of primary human mesenchymal
stromal cells was found to be down-regulated under cyclic compression. This could be
explained by the specific experimental conditions that excluded autocrine stimulation. When
the enrichment of secreted factors including BMP-2 in the cell culture medium was permitted,
cyclic compression promoted osteogenic differentiation as it was observed under direct
supplementation of BMP-2. Based on these observations, it was concluded that mechanical
stimulation induces osteogenesis indirectly through a mechanically controlled secretion of
BMP-2 and the resulting biochemical self-stimulation. This interpretation was underpinned
by the absence of load-induced osteogenic differentiation when a specific BMP inhibitor was
supplemented.
Besides a mechano-regulated increase in BMP-2 expression and secretion, mechanical stimuli
trigger mechanotransduction pathways that directly crosstalk to BMP signaling enhancing
Smad phosphorylation and target gene expression. However, the mechanical requirements
and the molecular mechanism causing the crosstalk are poorly understood. By a systematic
variation of the mechanical loading schemes, it was shown for the first time that cells feature
a mechanical memory that leads to an increased signaling response to BMP-2 even when the
mechanical signal has vanished. The mechanical memory is active upon long-term stimulation
and is based inter alia on an enhanced and sustained expression of the BMP receptor type 1B.
While transcriptional regulations are suggested to be an integral part of the mechanical
memory, the immediate early induction of Smad phosphorylation upon concurrent
mechanical and biochemical (BMP-2) stimulation is independent of any transcriptional
regulation. Instead, specific integrin knockdown and F-actin stabilization experiments
2 Abstract
revealed that integrin αv as well as load-induced integrin and actin cytoskeleton remodeling
are required for the immediate mechano-regulation of BMP signaling.
The relevance of the crosstalk for early tissue formation was investigated in the last part of
the project. While, cyclic compression alone specifically altered mechanical, structural and
compositional matrix cues, BMP-2 treatment had only minor effects. In a combination of both
stimuli, the effects of cyclic compression were therefore dominating and no synergistic effects
could be observed. Even though a role of the crosstalk for early tissue formation could not be
verified, new insides into how mechanical stimulation influences ECM formation have been
gained.
Taken together, this dissertation contributes to a profound understanding of how mechanical
forces regulate osteogenic differentiation, BMP signaling and early tissue formation,
processes, which are relevant in the context of bone regeneration. In a long-term perspective,
these findings could help to optimize mechanical boundary conditions with respect to BMP
signaling to increase the efficacy and safety of therapeutically used BMP-2. This study
highlights the role of mechano-biochemical interactions in controlling cell behavior and
motivate further research on growth factor signaling in a mechanical context.
Zusammenfassung 3
Zusammenfassung
Während der Knochenheilung sind Zellen gleichzeitig mechanischen Kräften und einer
Vielzahl von biochemischen Faktoren, einschließlich des klinisch angewandten Bone
Morphogenetic Protein 2 (BMP-2), ausgesetzt. In vivo Experimente deuten darauf hin, dass
mechanische Kräfte die BMP-2-induzierte Knochendefektheilung fördern und in vitro Studien
konnten eine Verstärkung des BMP Signalweges durch mechanische Stimulation zeigen. Zur
klinischen Behandlung werden noch immer supra-physiologische BMP-2 Konzentrationen
verwendet, die verschiedene Nebenwirkungen verursachen können. Optimierte mechanische
Stimuli, ausgehend von extrinsischer Belastung oder von einem Biomaterial, könnten
zukünftig dazu genutzt werden, die Effizienz des Wachstumsfaktors zu überhöhen. Um sich
diese Interaktion zunutze zu machen, muss jedoch zunächst verstanden werden wie beide
Stimuli unabhängig und abhängig voneinander das Zellverhalten beeinflussen.
In dieser Dissertation wurde sowohl der Einfluss von extrinsischen mechanischen
Kräften und BMP-2 auf die osteogene Zelldifferenzierung und Gewebebildung untersucht, als
auch der molekulare Mechanismus der der mechanischen Regulierung des BMP-Signalweges
zugrunde liegt, erforscht. Um in vivo Belastungsbedingung in einer 3D Umgebung
nachzubilden, wurde ein Bioreaktorsystem in Kombination mit makroporösen
Biomaterialien in dieser Studie verwendet.
Im Gegensatz zu Literaturdaten, wurde die osteogene Differenzierung primärer humaner
mesenchymaler Stromazellen durch zyklische Kompression herunterreguliert. Dies konnte
durch die speziellen experimentellen Bedingungen erklärt werden, die eine autokrine
Stimulation ausschlossen. Wenn eine Anreicherung von sezernierten Faktoren, einschließlich
BMP-2, im Zellkulturmedium zugelassen wurde, förderte die zyklische Kompression jedoch
die osteogene Differenzierung, was auch unter Supplementierung von BMP-2 beobachtet
wurde. Basierend auf diesen Ergebnissen wurde der Schluss gezogen, dass die mechanische
Stimulation die Osteogenese indirekt durch eine mechanisch kontrollierte Sekretion von
BMP-2 und die daraus resultierende biochemische Selbststimulation induziert wird. Diese
Interpretation wurde dadurch untermauert, dass eine Zugabe eines spezifischen BMP-
Inhibitors zum Ausbleiben einer belastungsinduzierten osteogenen Differenzierung führte.
Neben einer mechano-regulierten Erhöhung der BMP-2-Expression lösen mechanische
Stimuli Mechanotransduktionswege aus, die direkt mit dem BMP-Signalweg interagieren und
die Smad-Phosphorylierung und die Expression von Zielgenen verstärken. Allerdings sind die
mechanischen Anforderungen für eine Interaktion und der zugrundeliegende molekulare
Mechanismus nur unzureichend verstanden.
Durch Variation der Belastungsparameter wurde erstmals festgestellt, dass Zellen bei
langfristiger Vorstimulation ein mechanisches Gedächtnis entwickeln, welches sich auf den
BMP-Signalweg auswirkt. Dieses Gedächtnis wird unter anderem durch eine
belastungsinduzierte Erhöhung der BMP-Rezeptor-Typ-1B-Expression verursacht. Während
transkriptionelle Regulationen für die Ausbildung eines mechanischen Gedächtnisses von
großer Wichtigkeit sind, ist die sofortige und frühe Induktion der Smad-Phosphorylierung
4 Zusammenfassung
durch gleichzeitiger mechanischer und BMP Stimulation unabhängig von einer
Transkriptionsregulierung. Stattdessen konnte durch einen spezifischen Integrin-
Knockdown und eine F-Aktin-Stabilisierung gezeigt werden, dass αv Integrine und der
belastungsinduzierte Integrin- und Zytoskelettumbau für die sofortige Mechano-Regulation
des BMP-Signalweges erforderlich sind.
Die Bedeutung der Interaktion für die frühe Gewebebildung wurde im letzten Teil des
Projekts untersucht. Während die zyklische Kompression spezifisch mechanische,
strukturelle und kompositorische Matrixeigenschaften veränderte, hatte die BMP-2-
Behandlung nur geringe Auswirkungen. Bei einer Kombination beider Stimuli dominierten
daher die Effekte der zyklischen Kompression und es konnten keine synergistischen Effekte
beobachtet werden. Obwohl eine Rolle der Mechano-Regulation des BMP-Signalweges für die
frühe Gewebebildung nicht verifiziert werden konnte, wurden neue Erkenntnisse darüber
gewonnen, wie mechanische Stimulation die Bildung der extrazellulären Matrix beeinflusst.
Zusammengenommen tragen die Ergebnisse diese Dissertation zu einem tiefgreifenden
Verständnis darüber bei, wie mechanische Kräfte die osteogenen Differenzierung, den BMP-
Signalweges und frühe Gewebebildungsprozesse im Kontext der Knochenheilung regulieren.
In Zukunft könnten diese Erkenntnisse dazu beitragen, die mechanischen Randbedingungen
in Bezug auf den BMP-Signalweg zu optimieren, um die Wirksamkeit und Sicherheit von
therapeutisch eingesetztem BMP-2 zu erhöhen. Die Ergebnisse heben die Rolle mechano-
biochemischer Wechselwirkungen bei der Steuerung des Zellverhaltens hervor und
motivieren zu weiteren Forschungen zur Wachstumsfaktorsignalwegen in einem
mechanischen Kontext.
Table of Contents 5
Table of Contents
Abstract ....................................................................................................................................................... 1
Zusammenfassung .................................................................................................................................. 3
Table of Contents ..................................................................................................................................... 5
List of figures ............................................................................................................................................. 7
List of tables .............................................................................................................................................. 9
1 Introduction ................................................................................................................................... 11
1.1 Repair versus regeneration – bone as a model system for tissue regeneration ....... 11
1.2 Bone fracture healing and regeneration ................................................................................... 11
1.2.1 Bone Morphogenetic Proteins - growth factors essential for bone healing ...... 13
1.2.2 Mechanical forces influence bone healing ....................................................................... 13
1.2.3 Mechanical forces enhance BMP-2-induced bone healing ....................................... 15
1.3 Sensing, transmitting and responding to mechanical cues ............................................... 16
1.3.1 Integrin-mediated adhesions ................................................................................................ 16
1.3.2 Integrin-mediated mechanotransduction ....................................................................... 18
1.3.3 Mechanical forces influence cell fate decisions ............................................................. 19
1.4 BMP signaling pathway .................................................................................................................... 20
1.5 Mechanical signals integrate into the BMP pathway ........................................................... 23
1.5.1 Integrin-BMP receptor crosstalk ......................................................................................... 24
1.6 Mechanical forces and BMP-2 influence ECM formation .................................................... 26
1.7 Motivation and Aims .......................................................................................................................... 28
2 Materials .......................................................................................................................................... 30
2.1 Optimaix collagen scaffold .............................................................................................................. 30
2.2 Bioreactor used for mechanical stimulation ............................................................................ 30
2.3 Bioreactor equipment and consumables .................................................................................. 31
2.4 Flow chamber setup........................................................................................................................... 32
2.5 Devices..................................................................................................................................................... 33
2.6 Chemicals, reagents and kits .......................................................................................................... 33
2.7 Buffer ingredients ............................................................................................................................... 34
2.8 Cell culture ............................................................................................................................................. 35
2.8.1 Cells ................................................................................................................................................. 35
2.8.2 Cell culture material ................................................................................................................. 36
2.8.3 siRNAs, transfection reagents and transduction material ........................................ 36
2.8.4 Growth factor and small molecular inhibitor ................................................................ 37
2.9 Materials for histology ...................................................................................................................... 37
2.9.1 Primary and secondary antibodies .................................................................................... 38
2.9.2 Small molecular dyes for immunohistochemistry ....................................................... 39
2.10 RNA isolation, reverse transcription and qPCR ..................................................................... 39
2.10.1 Primer ............................................................................................................................................. 39
3 Methods ............................................................................................................................................ 42
6 Table of Contents
3.1 Cell biological methods...................................................................................................................... 42
3.1.1 Cell thawing and cultivation .................................................................................................. 42
3.1.2 Passaging and cryo-preservation......................................................................................... 43
3.1.3 Seeding of collagen scaffolds ................................................................................................. 44
3.1.4 Lentiviral transduction of an actin marker ...................................................................... 44
3.1.5 Transfection of small interfering RNA ............................................................................... 45
3.1.6 Bioreactor cultivation, mechanical loading and BMP stimulation ......................... 46
3.2 Molecular biological methods ......................................................................................................... 49
3.2.1 Ribonucleic acid isolation from collagen scaffolds ....................................................... 49
3.2.2 Reverse transcription, quantitative polymerase chain reaction and data
evaluation ...................................................................................................................................... 49
3.2.3 Cell lysis for protein analysis and extraction of ECM proteins ................................ 50
3.2.4 Sodium dodecylsulfate polyacrylamide gel electrophoresis .................................... 51
3.2.5 Western blotting and protein detection ............................................................................ 51
3.3 Immunocytochemistry ...................................................................................................................... 52
3.3.1 Sample preparation including fixation and cryo-trimming ...................................... 52
3.3.2 Immunofluorescence staining ............................................................................................... 52
3.4 Confocal multiphoton microscopy ................................................................................................ 53
3.4.1 Image acquisition and analysis ............................................................................................. 53
3.4.2 Live-cell-imaging and analysis .............................................................................................. 54
3.4.3 Mechano-imaging ....................................................................................................................... 55
3.5 Scaffold contraction analysis .......................................................................................................... 56
3.6 Mechanical compression tests ........................................................................................................ 56
3.7 Decellularization after in vitro tissue formation ..................................................................... 57
3.8 Mass spectrometry .............................................................................................................................. 58
3.9 Statistical analysis and data presentation ................................................................................. 58
4 Results .............................................................................................................................................. 59
4.1 Load-induced osteogenic differentiation via BMP-2 ............................................................. 59
4.1.1 Cell morphology, proliferation and oxygen concentration inside scaffolds
cultured in the bioreactor ....................................................................................................... 59
4.1.2 Cyclic mechanical compression downregulates the expression of key osteogenic
marker genes but upregulates BMP-2 expression ........................................................ 61
4.1.3 Limited biochemical conditioning of the culture medium during bioreactor
culture ............................................................................................................................................. 62
4.1.4 Cyclic mechanical compression enhances RUNX2 mRNA expression only in a
BMP-enriched environment ................................................................................................... 63
4.2 Mechanistic investigations on the crosstalk between mechano-transduction and
BMP signaling ........................................................................................................................................ 68
4.2.1 The crosstalk is relevant in primary human cells of the mesenchymal lineage ....
........................................................................................................................................................... 68
4.2.2 Correlation between loading frequency and crosstalk duration ............................ 69
List of figures 7
4.2.3 Focal adhesion number and size is increased by both, BMP-2 and mechanical
loading in a frequency-dependent manner ..................................................................... 72
4.2.4 Cells develop a mechano-memory impinging on BMP signaling ........................... 74
4.2.5 Mechanical signals regulate the BMP-pathway via integrins .................................. 76
4.2.6 Actin cytoskeleton remodeling is crucial for the crosstalk ...................................... 78
4.3 The influence of the crosstalk on ECM formation ................................................................. 86
4.3.1 Load- induced tissue contraction and stiffening is reduced by BMP-2 ............... 86
4.3.2 Mechanical loading increases collagen synthesis but reduces fibrillar collagen
density and fiber alignment .................................................................................................. 88
4.3.3 BMP-2 and cyclic compression induce distinct gene expression changes ......... 92
4.3.4 ECM protein composition is specifically altered by mechanical loading ............ 94
5 Discussion ....................................................................................................................................... 97
5.1 Mimicking mechanical loading conditions during the early phase of bone healing 97
5.2 Towards a deeper understanding how cyclic compression influences osteogenic
differentiation....................................................................................................................................... 98
5.3 Cyclic compression possess an osteoinductive potential only in a BMP-enriched
environment .......................................................................................................................................... 99
5.4 Cyclic compression integrates into the BMP signaling pathway only in a ligand
dependent manner ........................................................................................................................... 100
5.5 The mechano-sensitivity of BMP signaling is dependent on the loading frequency
and timing ............................................................................................................................................ 101
5.6 Integrin αv and load-induced integrin and F-actin reorganization processes are
required for the crosstalk .............................................................................................................. 105
5.7 Mechanical forces specifically alter mechanical, structural and compositional matrix
cues ......................................................................................................................................................... 110
6 Summary and Conclusion ....................................................................................................... 117
7 Outlook .......................................................................................................................................... 120
8 Bibliography ................................................................................................................................ 121
Supplement ........................................................................................................................................... 135
Abbreviations ...................................................................................................................................... 139
List of figures
Figure 1-1: Important phases and factors in bone regeneration. ...................................................... 12
Figure 1-2: Mechanical conditions at the fracture site define the route of tissue differentiation
and the mode of healing. ..................................................................................................................................... 14
Figure 1-3: Mechanosensation by integrin adhesions............................................................................ 17
Figure 1-4: The complexity of integrin signaling...................................................................................... 18
Figure 1-5: BMP signaling pathway and selected regulatory mechanisms. .................................. 22
Figure 1-6: Regulation of BMP signaling by mechanotransduction. ................................................ 24
Figure 2-1: Bioreactor setup. ............................................................................................................................ 30
8 List of figures
Figure 2-2: Schematic representation of the ibidi Pump System. ...................................................... 32
Figure 3-1: Analysis of protrusion remodeling. ......................................................................................... 54
Figure 3-2: Bioreactor-Microscope-Setup. ................................................................................................... 55
Figure 3-3: Schematic representation of scaffold contraction and calculation of total volume
contraction (V(V0-Vt)). .............................................................................................................................................. 56
Figure 4-1: Bioreactor setup validation. ....................................................................................................... 60
Figure 4-2: Cyclic compression downregulates the expression of key osteogenic marker genes
but upregulates BMP2 expression. .................................................................................................................. 62
Figure 4-3: Low BMP-2 concentrations in the conditioned medium. ............................................... 63
Figure 4-4: Cyclic compression only increases RUNX2 expression, if rhBMP2 is added or an
enrichment of cell-secreted BMP2 in the cell culture medium was permitted. ........................... 65
Figure 4-5: Cyclic compression does not increase RUNX2 expression, if BMP signaling is
inhibited by rhNoggin. .......................................................................................................................................... 67
Figure 4-6: Cyclic mechanical compression significantly increases the BMP-2-induced
Smad1/5/8 phosphorylation in human primary MSCs and dermal fibroblasts (hdF). ............. 69
Figure 4-7: Loading frequency influences strength and duration of Smad1/5/8
phosphorylation and ID gene expression. .................................................................................................... 70
Figure 4-8: Heat map summarizing the gene expression changes in response to 24h BMP-2
stimulation and/or mechanical loading of 1 Hz or 10 Hz. ..................................................................... 71
Figure 4-9: Mechanical stimuli regulate gene expression in a frequency dependent manner.
........................................................................................................................................................................................ 72
Figure 4-10: Focal adhesion number and size is increased by BMP-2 treatment and by
mechanical loading in a frequency-dependent manner. ........................................................................ 73
Figure 4-11: Mechanical pre-stimulation induced a crosstalk on p-Smad and on gene
expression level. ...................................................................................................................................................... 75
Figure 4-12: Only prolonged mechanical pre-stimulation induced a crosstalk on p-Smad level.
........................................................................................................................................................................................ 76
Figure 4-13 Validation of integrin αv knockdown in hFOBs. ............................................................... 77
Figure 4-14: Integrin αv knockdown reduced the crosstalk on Smad phosphorylation level.
........................................................................................................................................................................................ 78
Figure 4-15: Cyclic compression induced focal adhesion kinase and myosin light chain
activation. .................................................................................................................................................................. 79
Figure 4-16: Effects of Jasplakinolide on actin cytoskeleton integrity and dynamics. .............. 81
Figure 4-17: Dynamic actin remodeling induced by fluid shear stress is inhibited by
Jasplakinolide. .......................................................................................................................................................... 83
Figure 4-18: F-actin stabilization by Jasplakinolide inhibits load-induced Smad
phosphorylation and ID1 expression. ............................................................................................................ 85
Figure 4-19: Load- induced scaffold contraction and stiffening is reduced by BMP-2. ............. 87
Figure 4-20: Cyclic mechanical compression increases collagen synthesis but reduces fibrillar
collagen density. ...................................................................................................................................................... 89
Figure 4-21: Cyclic compression changes the dependency of collagen density and tissue
contraction. ............................................................................................................................................................... 90
Figure 4-22: Cyclic mechanical compression reduces fiber and cell alignment. .......................... 92
Figure 4-23: Gene expression analysis of selected ECM proteins and ECM modulators. ......... 92
List of tables 9
Figure 4-24: Cyclic compression induced distinct changes in the protein composition of the
ECM. ............................................................................................................................................................................. 95
Figure 5-1: Cyclic compression promotes osteogenic differentiation via BMP. ........................ 100
Figure 5-2: Schematic representation of how mechanical forces integrate into BMP signaling.
..................................................................................................................................................................................... 109
Figure 5-3: The regulation of collagen cross-linking by periostin. ................................................. 114
Figure 5-4: Cyclic compression disturbs collagen cross-linking. ..................................................... 114
Figure 0-1 Pubmed search term statistics. ................................................................................................ 135
Figure 0-2: Gene expression changes of TGFβ1, TGFβ3, FGF2, PDGF-A and VEGF-A, growth
factors that influence osteogenic differentiation. .................................................................................. 135
Figure 0-3: BMP2 stability during bioreactor culture. ......................................................................... 136
Figure 0-4: Dynamic actin remodeling induced by cyclic compression and scaffold wall
deformation under compression visualized using the Bioreactor-Microscope-Setup. .......... 137
Figure 0-5: Fibrillar collagen density is reduced by cyclic compression after 2 weeks of
cultivation. .............................................................................................................................................................. 137
Figure 0-6: Cyclic compression did not induce ERK1/2 or Src phosphorylation. .................... 138
List of tables
Table 2-1: List of bioreactor equipment ...................................................................................................... 31
Table 2-2: List of bioreactor consumables .................................................................................................. 32
Table 2-3: Consumables for the ibidi Pump System ............................................................................... 32
Table 2-4: List of devices .................................................................................................................................... 33
Table 2-5: List of chemicals, reagents and kits .......................................................................................... 33
Table 2-6: Buffer compositions ........................................................................................................................ 34
Table 2-7 Material for siRNA transfection and lentiviral transduction .......................................... 36
Table 2-8: List of growth factors and small molecular inhibitors ..................................................... 37
Table 2-9: Consumables for histology ........................................................................................................... 37
Table 2-10: List of primary and secondary antibodies .......................................................................... 38
Table 2-11: List of small molecular dyes used in immunohistochemistry .................................... 39
Table 2-12: Consumables and kits for RNA isolation, reverse transcription and pPCR........... 39
Table 2-13: List of primers for qPCR ............................................................................................................. 39
Table 3-1 Summary of expansion media composition ........................................................................... 43
Table 3-2: Volumes used for culture and trypsinization depending on the culture format ... 43
Table 3-3 Cell concentrations used for different cell types .................................................................. 44
Table 3-4 Plate layout for the transduction efficiency test. ................................................................. 45
Table 3-5 Calculations for seeding of one scaffold (volume = 80 µl) per condition................... 46
Table 3-6: Summary of experimental conditions ..................................................................................... 47
Table 3-7: FBS concentration and medium amount inside bioreactors depending on the type
of experiment .......................................................................................................................................................... 48
Table 3-8 qPCR reaction steps ......................................................................................................................... 50
Table 3-9 General IF staining protocol ......................................................................................................... 52
Table 3-10: Perfusion protocol ........................................................................................................................ 57
10 List of tables
Table 0-1: Fold change (F.I.) gene expression in response to 24h BMP-2 stimulation and/or
mechanical loading of 1 Hz or 10 Hz. .......................................................................................................... 136
Table 0-2: Gene expression analysis of selected ECM proteins and ECM modulators. .......... 138
Introduction 11
1 Introduction
1.1 Repair versus regeneration – bone as a model system for tissue
regeneration
Regeneration, a process by which the original structure and function of a tissue, organ or even
whole body parts are fully restored, is fascinating and motivates the whole field of
regenerative medicine. The question why some organisms regenerate while others not is still
unanswered but there is a great endeavor to identify common themes of regeneration that if
fully understood might revolutionize medical treatment [1]. In order to unravel regenerative
processes, model organs and organisms are extensively studied, one of which is bone.
In the adult human body, bone is one of the few tissues that have the ability to fully regain
their initial functionality after injury [2]. In other tissues, however, a repair process is
initiated, which mainly involves the deposition of fibrous matrix and wound contraction by
fibroblasts. The resulting scar tissue closes the wound but possesses, in comparison to the
original tissue, different compositional, structural and mechanical properties, impairing the
tissues functionality [3], [4]. Future regenerative therapies aim at scar-less healing by
resembling endogenous regeneration cascades but therefore, a full understanding of
regenerative processes like in bones is needed.
1.2 Bone fracture healing and regeneration
Depending on the size of the fracture gap and the mechanical stability at the fracture site,
bone healing can follow two mechanisms, which are termed primary (also referred to direct
healing) or secondary healing (also referred to endochondral ossification) [5]. Primary
healing, a process in which bone is formed directly without the intermediate step of cartilage
formation, requires absolute mechanical stability, a very small fracture gap (< 500 µm) and
aerobic conditions [6]. In other instances, bone healing follows a complex multiphase process
that is commonly divided into four overlapping phases: inflammatory, soft callus, hard callus
and remolding phase [7] (see Figure 1-1). Directly after injury, a hematoma is formed and
chemotactic signaling molecules are released attracting immune cells. They in turn secrete
pro-inflammatory cytokines such as tumor necrosis factor-α (TNF-α), interleukin-1β (IL-1β)
or IL-6 to further attract inflammatory cells [8]. The initial pro-inflammatory phase is
gradually transferred into an anti-inflammatory phase characterized by the presence of
Fibroblast Growth Factor -2 (FGF-2), Transforming Growth Factor-β (TGF-β), Platelet-
Derived Growth Factor (PDGF) and Bone Morphogenetic Proteins (BMPs) [9]. These growth
factors are essential for the recruitment, proliferation and differentiation of progenitor cells
12 Introduction
[10]. Due to the secretion of chemoattractants and growth factors as well as their ability to
remove necrotic tissue, immune cells play an important role in initiating the healing cascade
[9]. During the inflammatory phase, the primary hematoma is remodeled into a fibrin-,
fibronectin-, but also collagen-rich granulation tissue that serves as a scaffold for the
establishment of the cartilaginous soft callus. Importantly, the structural organization of the
fibrillar collagen network within the granulation tissue was recently shown to guide the
following process of endochondral bone formation [11]. Recruited and proliferated
mesenchymal stromal cells (MSCs) differentiate into chondroblasts and synthesize a
cartilaginous matrix consisting of collagen II and glycoproteins. All fibrous tissue in the
fracture gap will eventually be replaced by cartilage that bridges and stabilizes the fracture
[12]. This process is followed by cartilage calcification. Hypertrophic chondrocytes and later
osteoblast release membrane-derived-vesicles containing alkaline phosphatases (ALP),
calcium phosphate complexes and proteoglycanases [13]. Glycoaminoglycans become
degraded, calcium phosphate complexes are integrated into the collagenous matrix and the
region will be revascularized. The established hard callus is still consisting of unorganized
woven bone that is eventually remodeled into lamellar bone to fully restore the mechanical
strength [14].
Figure 1-1: Important phases and factors in bone regeneration. (A) Consecutive and overlapping phases of bone
regeneration after fracture. Figure adapted from [15] with permission from the publisher. (B) “The card house of bone
healing”. The authors’ interpretation of the diamond concept for bone healing [16]–[18]. Factors for successful
healing are assembled in a card house, relating to the fragility of the process.
All factors that influence and contribute to fracture repair, which are summarized in the
“diamond concept” [16]–[18] (Figure 1-1B), must be tightly controlled in order to allow
successful healing. The misbalance of one of the factors due to for example aging or obesity,
can lead to delayed healing or non-union formation. Moreover, the natural self-healing
capacity of bone is limited by the fracture size. Large bone defects resulting from e.g. tumor
resection where large tissue quantities need to be restored, resemble an especially
challenging situation [19]. The gold standard for the treatment of large bone defects is still
the harvest of autologous material from the iliac crests of the pelvis, or the intramedullary
Introduction 13
canal of long bones [20]. However, due to the known disadvantageous like donor site
morbidity and limited amount of material [21], alternative treatment strategies are needed,
which could include biomaterials, cells or growth factors. In fact, the clinical use of the growth
factor BMP-2 has gained increasing importance.
1.2.1 Bone Morphogenetic Proteins - growth factors essential for bone healing
BMPs are one of the major signaling molecules orchestrating bone healing by regulating
cellular processes like proliferation [22], migration [23] and differentiation [24]. The family
of BMPs consists of 30 different members but not all of them are associated to bone [25].
During the different healing phases, the expression of individual BMP types is tightly
regulated. BMP-2, -4 and -7 are highly upregulated in the periosteal region at the early phase
of healing [26], whereas BMP-3, -4, -7 and -8 are expressed at later stages of endochondral
ossification when the cartilaginous matrix is remodeled and calcified. BMP- 5 and -6 instead
are expressed almost throughout the healing cascade [27]. Therefore, some BMP types are
potent inducers of differentiation, while others regulate cell maturation. From in vivo
knockout experiments it is known that specifically BMP-2 is indispensable for the initiation
of fracture healing [28].
With the FDA (Food and Drug Administration) approval of recombinant human BMP-2
(rhBMP-2, InFUSE™, Medtronic, USA) for the treatment of tibial non-unions and spinal
fusions, the growth factor has gained significant clinical relevance [25], [29]. Recombinant
hBMP-2 is used either as an alternative or complementary treatment option for autologous
bone grafting. Its administration has reduced the length of hospital stay [30], the rate of
secondary interventions and the treatment failure rate [31]. However, still excessive and non-
physiological amounts of rhBMP-2 are required (1.5 mg/ml are FDA-approved) to promote
bone formation increasing the treatment costs and the risk of side effects like ectopic bone
formation and osteoclast-mediated osteolysis [32]. In order to optimize the growth factor
treatment, it is necessary to better understand signaling cascades and regulatory
mechanisms.
1.2.2 Mechanical forces influence bone healing
Mechanical boundary conditions at the fracture site, mainly determined by the mechanical
properties of the fixation device and musculoskeletal loadings, critically influence the course
and outcome of bone healing [33]–[35]. Complete stability or excessive movements delay
healing or can even cause non-union formation [36], [37]. In case of excessively rigid fixation,
the healing bone is protected from normal stresses (stress shielding) resulting in bone-end
resorption due to the lack of mechanical communication [38]. However, if the mechanical
stability is too low, blood vessels become repeatedly disrupted and the healing cascade
14 Introduction
cannot continue [37], [39]. These findings demonstrate, that mechanical conditions have to
be adjusted carefully in order to promote healing.
Movements at the fracture site (interfragmentary movements (IFMs)) are differentiated
in axial compression of the fracture fragments or relative movements causing shear stress.
While the latter is considered to be detrimental, it is widely accepted that a moderate amount
of axial compression promotes healing by stimulating callus formation [40]–[43]. In vivo
animal studies and numerical simulation suggest that IFMs smaller than 15% of the fracture
height allow undisturbed healing via endochondral ossification [37], [40], [44], [45]. Using FE
modeling and histological data, mechanobiological models have been developed, describing
how mechanical conditions define tissue differentiation and consequently the mode of
fracture healing (Figure 1-2). According to the model by Claes and Heigele (1999),
endochondral ossification occurs for strains less than ±15% and compressive pressure larger
than −0.15MPa. However, larger mechanical stimuli result in the formation of connective and
fibrous tissues [40].
Figure 1-2: Mechanical conditions at the fracture site define the route of tissue differentiation and the mode of healing.
Model is based on the correlation between mechanical condition and types of tissues in a fracture callus.
Intermembranous ossification takes place in regions which are defined by a surface strain <±5% and a hydrostatic
pressure <±0.15 MPa (region A). Endochondral ossification takes place in regions which are defined by surface strains
<±15% and negative hydrostatic pressure values greater than −0.15 MPa. Figure taken from [40] with permission of
reuse from the publisher.
Besides the magnitude of IFMs, the timing is critically important. Moderate movements
during the early healing phase contributed to increased bone mineral density and stiffness
[43], [46], while increased movements at later stages had contrary effects [46]. From this, the
concept of reverse dynamization evolved, in which the fracture fixation stiffness changes from
Introduction 15
low to high during the course of healing [47]. When this strategy, the change from flexible to
rigid fixation, was applied at 1 week after surgery in a rat osteotomy model, healing was
accelerated [48]. In humans, this beneficial effect could be reproduced in a pilot study [49]
but further clinical studies are missing until now.
In summary, specifically the early healing phase is highly mechano-sensitive and
optimization of biomechanical conditions during this stage has great potential to promote the
subsequent healing cascade.
1.2.3 Mechanical forces enhance BMP-2-induced bone healing
Even though, rhBMP-2 is a very potent inducer of bone formation, it is increasingly recognized
that the efficiency of rhBMP-2 is controlled by mechanical cues [47], [50]–[54]. The interplay
between BMP-2 and mechanical forces was mainly investigated in critical-sized femoral
defects in rats but clinical studies are missing so far. In animal experiments, the mechanical
environment was tuned by using different fixator stiffnesses or by the active application of
axial compression during healing.
Low- stiffness fixation (stiffness of 114 N/mm) accelerated rhBMP-2 induced bone
healing in comparison to medium and high stiffness fixation (185 N/mm, 254 N/mm,
respectively) [47]. Using the reverse dynamization approach it was found that the effect of
BMP-2 can be even further promoted if mechanical loads are high during the early phase but
rather low at later stages. Strikingly, given the strong osteoinductive capacity of rhBMP-2,
fracture fixation using too low fixator stiffnesses (25.4 N/mm) can be detrimental to healing
[50].
Additionally, external mechanical loading (10% axial compression, f = 0.01Hz, applied
once a week) under rhBMP-2 treatment significantly enhanced mineralized tissue volume
and mineral content at 2 weeks post-operation in comparison to the non-loaded control. In
agreement with the study mentioned before, it was found that early mechanical stimulation
seems to be beneficial but once a bony bridging is established loading loses its stimulatory
capacity. Interestingly, in a critical-sized bone defect, mechanical stimulation alone does not
induce bone formation but rather negatively affects callus formation [51].
These findings demonstrate that even if rhBMP-2 is a powerful therapeutic tool, its
effectiveness is largely dependent on mechanical boundary conditions. It becomes clear, that
certain amounts of mechanical stimulation have the ability to enhance the efficacy of rhBMP-
2. This indicates a cooperative interaction of BMP-2 and mechanical forces that can be
expected to be true also for endogenously expressed BMP-2. Still further investigations are
needed to define optimal timing and magnitude of mechanical loading in combination with
rhBMP-2 treatment. Although in vitro studies cannot resemble the physiological complexity,
16 Introduction
they help to understand concepts of cellular behavior, which can be translated back into in
vivo situations. Therefore, in this study, the sensitivity of the BMP signaling pathway for
loading frequency, duration and timing were investigated on cellular level. These findings
might in future help to improve mechanical boundary conditions during healing.
1.3 Sensing, transmitting and responding to mechanical cues
Due to intensive research in the field of cellular biomechanics during the past 20 years (see
supplementary Figure 0-1), it is nowadays accepted that physical signals are as important as
biochemical cues to control cellular behavior. Cells are not only influenced by external
mechanical forces like tension, compression, shear and hydrostatic pressure [55] but also by
the mechanical properties of the surrounding extracellular matrix (ECM) like rigidity [56],
stress relaxation behavior [57] and topology [58]. Specialized organelles and structures
including the primary cilium, ion-channels, G-protein-coupled receptors (GPCRs), cell
adhesions sites, the cytoskeleton and also the nucleus perceive and transduce mechanical
stimuli [59]. Given that a variety of different physical stimuli act simultaneously on the cell, it
is likely that many of these sensors contribute to define cell behavior.
1.3.1 Integrin-mediated adhesions
Integrins are one of the most important mechanotransducers. They not only provide a
molecular link between the ECM and the actin cytoskeleton but also mediate the conversion
of physical into biochemical signals via cytoplasmic adaptor proteins [60]. Integrins form
non-covalently linked heterodimeric complexes of α and β subunits, each of which is a single-
spanning type I transmembrane protein with an extracellular ligand binding site, a
transmembrane domain and a short cytoplasmic tail. From a combination of 18 α-subunits
and 8 β-subunits, 24 different integrin subtypes are formed that specifically recognize ECM
ligands such as fibronectin, collagen or laminin [61] (Figure 1-3A). Due to their ligand
specificity, integrin expression patterns dependent on the composition of the ECM and are
tissue-specific. Importantly, each integrin type triggers different combinations of
downstream signaling pathways, which differentially affect cellular behavior [62]. Integrins
are usually in their low affinity (bent-V shape) conformation. Unfolding into the high affinity
(active) conformation is triggered either upon binding to the specific ECM ligand (outside-in
activation) or upon binding of Talin to the cytoplasmic tail of the β-subunit (inside-out
activation).
Introduction 17
Figure 1-3: Mechanosensation by integrin adhesions. (A) Integrin subunits α and β form heterodimers, which
specifically recognize ECM proteins or motifs (Arg-Gly-Asp (RGD)). (B) Illustration of an adherent, migrating cell
containing diverse integrin-mediated adhesions. Focal complexes (FC) form within the lamellipodium and eventually
mature to focal adhesions (FA), which are connected to thick actin stress fibers. Some FA transition into fibrillar
adhesions, which are important for the remodeling of fibronectin. (C) Schematic drawing showing the principle
structure of integrin-mediated adhesions. Integrin heterodimers bound to the ECM are connected via adapter proteins
to the actin cytoskeleton allowing a bidirectional force transmission. Figure information taken from [63]–[65]
Integrin activation is followed by the recruitment of numerous adaptor proteins, by the
assembly of actin filaments and the clustering to other integrins. Early adhesions formed
within the lamellipodia or filopodia of motile cells are termed focal complexes (FC) or nascent
adhesions (NA) (Figure 1-3B). As the leading edge is pushed forward by actin polymerization,
the lamellipodium moves over the stationary NA. Within the lamella, most of the NA
disassemble, whereas some eventually undergo a force dependent growth and maturation
into focal adhesions (FA) [66]. FA are much larger, elongated and coupled to thick actin
filaments cross-linked by α-actinin and myosin II (actin stress fibers). Myosin II, a motor
protein, generates the tension that is necessary for FA assembly and stability [67], [68]. The
contractile forces exerted by the actomyosin-network (cell traction force) are transmitted
towards the ECM via FAs (Figure 1-3C). Mature FAs can transition into fibrillar adhesions, a
specialized type of integrin-mediated adhesion, essential for the remodeling of fibronectin
that contain tensin instead of talin [69].
18 Introduction
1.3.2 Integrin-mediated mechanotransduction
Upon force-dependent FA maturation various adaptor proteins are recruited, which either
reinforce the connection to actin filaments or mediate downstream signaling. The integrin
adhesome within FAs can consist of 180 different proteins with approx. 700 interactions
demonstrating the complexity of mechanotransduction [70], [71]. The conversion of
mechanical into biochemical signals is mediated inter alia via the recruitment and activation
of kinases such as focal adhesion kinase (FAK) and proto-oncogene tyrosine-protein kinase
(Src) (see Figure 1-4). FAK is upstream of many pathways ultimately controlling cell motility,
proliferation and survival. In the context of this work, it is important to note that FAK was
shown to control adhesion stability and the activity of Rho-familiy GTPase, thereby taking
part in the regulation of actin cytoskeletal dynamics. Autophosphorylation of FAK at Tyr397
leads to binding of Src, which in turn phosphorylates other tyrosine residues in FAK
promoting its activity. The FAK-Src complex mediates the phosphorylation of p130Cas that in
turn activates Rac1 leading to enhanced lamellipodia formation [72], [73]. Additionally, this
complex phosphorylates paxillin resulting in increased focal adhesion turnover [74].
Figure 1-4: The complexity of integrin signaling. Pathway map showing integrin-mediated downstream signaling
with consequences for cell motility, proliferation and survival. Map taken from KEGG (Kyoto Encyclopedia of Genes
and Genomes) database with permission from Kanehisa Laboratories [75]–[77].
Integrin signaling-mediated remolding of FAs and the actin cytoskeleton is crucial for the
adaptation to mechanical stimuli. Upon mechanical tension, cell traction forces increase
through a cascade that involves RhoA/ROCK and the subsequent activation of myosin II [78].
Cyclic stretch and fluid flow trigger the reorientation of cells along with actin stress fiber
realignment, which is further more regulated by FAK and Rho-kinases. Interestingly, stress
Introduction 19
fibers oriented in the direction of stretch disassemble and reassemble in perpendicular
direction, which is accompanied by FA reorientation [79], [80]. Additionally, FAK-mediated
Rac1 activation induces the formation of lamellipodia upon tensile strain [81]. Mechanical
forces not only induce cytoskeletal remodeling through kinase activation but also triggers the
reorganization of protein interactions within the adhesom of FAs [82].
Whereas forces like fluid shear stress or compression are actively transmitted through
FAs, rigidity is a passive mechanical parameter whose sensation requires active probing of
the substrate. To deform the ECM, cells apply traction forces generated by the actomyosin
cytoskeleton and transmitted through integrins. The generated traction force is thereby
adapted to the ECM rigidity through a feedback mechanism. On stiff versus soft matrices,
stress fiber assembly and cytoskeletal contractility is increased through the RhoA pathway
[83].
In summary, integrin-based adhesions are important for the sensation, transduction and
response to changes in the mechanical environment. As an immediate response, cells adapt
by remodeling their cytoskeletal organization and adhesion sites to establish a new force
equilibrium. On the long-term, this will have consequences for gene expression affecting
proliferation, matrix production and differentiation.
1.3.3 Mechanical forces influence cell fate decisions
Since Engler et al. (2006) reported that MSCs can be directed into the neurogenic, myogenic
and osteogenic lineage by plating them on matrices mimicking the tissue-specific stiffness of
brain, muscle and collagenous bone, respectively, mechanical cues have gained increasing
attention [56]. To date it is accepted that the physical environment, which is defined by the
mechanical properties of the substrate as well as the type, frequency and magnitude of
mechanical loading, affects stem cell differentiation. In the following, the response of stem
cells, specifically MSCs, to external mechanical forces is summarized.
MSCs, adult progenitor cells residing in different mesodermal tissues such as bone
marrow and fat, are often used to study the influence of physical cues in the context of
regeneration, as they are actively recruited to the site of injury where they are subjected to
increased tissue deformation [84], [85]. A wide range of experimental setups have been used
to stimulate MSCs with mechanical forces in vitro. They can be generally categorized by the
type of mechanical loading including tension, compression, fluid shear stress, ultrasound and
vibration. These different modes of stimulations can be applied to cells cultured in tissue
culture plates or in bioreactors in 2D or 3D. Despite this, many different experimental
parameters influence the cell response, for example, scaffold properties (type of matrix,
stiffness), biochemical medium supplements, donor age, cell density, and of course different
20 Introduction
loading parameters (duration, magnitude, and frequency). As cell responses can be distinctly
different depending on the culture dimension [86], a direct comparison of 2D and 3D results
is often difficult. Therefore, here only studies investigating the influence of compressive force
applied to MSCs seeded in a 3D biomaterial are summarized.
Depending on the scaffold material, loading parameters and -very importantly- medium
supplements used, cyclic biomaterial deformation was reported to induce/enhance
osteogenic or chondrogenic differentiation of MSCs. Interestingly, studies in which MSCs were
cultured under chondrogenic or osteogenic medium found an enhanced chondrogenic [87]–
[89] or osteogenic [90]–[92] differentiation under cyclic compression, respectively. However,
if adipogenic medium was added, mechanical stimulation suppressed adipogenesis of MSCs
[93]. Therefore, cyclic compression promoted the differentiation towards the osteo-chondral
lineage predefined by the medium supplements. These investigations yet don’t show whether
mechanical stimulation directly induces osteo-chondral differentiation or just promotes the
biochemical trigger.
Only a few studies used basal medium to investigate the direct impact of loading on MSC
commitment. Michalopoulos et al. (2012) reported that cyclic compression of collagen-
alginate sponges (f = 1 Hz, 4h/day, 21 days) induced osteogenic or chondrogenic
differentiation of MSCs in a magnitude dependent manner. While 10% compression induces
the expression of Runt-related transcription factor 2 (RUNX2), an early osteogenic
transcription factor, 15% compression enhanced the expression of chondrogenic markers
Sox 9 (SRY (sex determining region Y)-box 9) and aggrecan [94]. Furthermore, low magnitude
compression of MSC seeded PCL/PLGA/TCP scaffolds increased Runx2 protein levels and the
expression of other important osteogenic markers as osterix (OSX), ALP and osteopontin
(OPN) [95]. Moreover and in addition to the classical osteogenic markers, mechanical loading
was reported to induce the expression of the growth factor BMP-2 in MSCs [96], [97]. The fact
that BMP signaling regulates the transcription of RUNX2 through the Smad pathway [98],
points towards an involvement of BMP signaling in load-induced osteogenic differentiation.
However, if the observed pro-osteogenic effects of cyclic compression on MSCs are a
direct consequence of mechano-regulated gene expression, or an indirect consequence of
load-induced autocrine or paracrine signaling (e.g. via secretion and signaling of BMP2)
remains an open question. This study aims to address this question by dissecting the direct
mechanical influence from a mix influence of mechanics and BMP-2.
1.4 BMP signaling pathway
The BMP family belongs to the Transforming Growth Factor (TGFβ) superfamily of cytokines
that fulfill functions in tissue development, homeostasis and healing but also disease [99].
Introduction 21
Even though originally described as bone growth factors [100], BMPs regulate processes in
many different tissues including cartilage [101], muscle [102], tendon [94], heart [104],
vessles [105] and the neuronal system [106]. On the cellular level they control proliferation,
migration, differentiation and apoptosis [107], [108].
After posttranslational processing, secretion and dimerization (homo- and heterodimers
exist), BMPs can bind to hetero-tetrameric receptor complexes consisting of two type I and
two type II transmembrane receptors. Both BMP receptor types feature an extracellular
ligand-binding motif and an intracellular serine/threonine kinase domain. While type II
receptors are constitutively active, type I receptors carry an additional glycine/serine-rich
region (GS-box) that controls the kinase activity. Upon ligand binding, this region becomes
phosphorylated by the type II receptor, leading to the activation of the type I kinase, which in
turn activates Smad or non-Smad signaling pathways [109], [110]. The mode of ligand-
receptor oligomerization defines the way of signal propagation. The canonical Smad pathway
is activated when BMPs bind to preformed complexes (PFCs) of type I and II receptors. BMPs
can furthermore bind to single type I receptors, forming a so-called BMP-induced signaling
complex (BISC) to which type II receptors are recruited activating the non-Smad pathway
[111].
The canonical Smad pathway is activated by C-terminal phosphorylation of recruited
receptor-regulated Smads (R-Smads) by BMP receptor type I and subsequent complex
formation with the common mediator Smad4. The trimeric transcription factor complex
composed of two phosphorylated R-Smads and one Smad4 molecule, translocates into the
nucleus and binds to elements in the promotor regions of BMP target genes to control their
expression [112], [113]. Downstream target genes controlled by the Smad pathway are inter
alia the family of Inhibitor of DNA binding (ID)-genes as one of the earliest [114], but also the
osteogenic transcription factor RUNX2 [98].
Non-Smad pathways include a diversity of other downstream effectors like mitogen
activated protein kinases (MAPK), such as p38 and ERK, which induce transcriptional
responses by the activation of ATF2, c-Jun or c-Fos and further control the expression of
osteopontin, ALP or collagen type 1 [115]. In addition, BMPs can induce immediate non-
transcriptional responses like actin rearrangement and migration via phosphatidylinositol 3-
kinase, small RhoGTPases and LIM kinases [107], [116].
22 Introduction
Figure 1-5: BMP signaling pathway and selected regulatory mechanisms. The dimeric ligand binds to its BMP receptor
complex, which becomes activated. Subsequently, Smad1/5/8 transcription factors are phosphorylated, form a
trimeric complex with Smad 4 and translocate into the nucleus. Together with transcriptional cofactors, they control
the transcription of multiple target genes. Non-Smad pathways include a diversity of downstream effectors like
MAPKs such as p38, ERK, JNK and others. Their activation leads to transcriptional and non-transcriptional responses.
BMP signaling is controlled via multiple factors including the ECM, BMP antagonists, co-receptors and inhibitory
Smads (I-Smads). Figure inspired by [111], [117].
The pleiotropic signaling responses upon pathway activation are based on the diversity
of BMP ligands, of BMP receptors (4 type I and 4 type II receptors are known) [110] with
different ligand-binding affinities, the tissue specific expression of receptors [118] and the
modes of ligand-receptor oligomerization. To ensure context-specific and precise signal
propagation, pathway activation and inactivation must be under tight control. This is realized
by a large number of regulation systems on different levels. At the ligand level, secreted
antagonists (e.g. Noggin, Chordin, Gremlin) inhibit signaling by binding BMPs and masking
the receptor-binding epitope [119]. Furthermore, ECM proteins such as fibrillin bind and
sequester BMPs in their inactive from [120]. At the receptor level, multiple co-receptors
attenuate or enhance signaling activity. Intracellularly, inhibitory Smad 6 and 7 (I-Smads)
compete for receptor binding [121] and MAPK and GSK3β-mediated R-Smad linker
phosphorylation targeting R-Smads for proteasome-dependent degradation [122].
These regulatory mechanisms enhance or attenuate signaling and are crucial for
physiological tissue function. However, in recent years it became more and more clear, that
also the mechanical environment adds to these regulatory mechanisms.
Introduction 23
1.5 Mechanical signals integrate into the BMP pathway
Several in vitro studies described a direct regulation of the BMP signaling pathway by
mechanotransduction [96], [123]–[126]. Mechanical loading was described to activate Smad
signaling in both a ligand dependent and independent manner. In osteoblasts this relation
was discussed somewhat controversial. Some studies reported that loading alone was
sufficient to activate R-Smads [123], [127], while others described a ligand dependent
activation [96], [125]. These contradictory results might be explained by the experimental
design that included a pre-cultivation on the biomaterial up to one week prior to loading
versus a direct load application. In the case of pre-cultivation an autocrine ligand secretion
could have led to the activation of the BMP pathway. Indeed, this assumption was supported
by Wang et al. (2010) who reported that Noggin treatment abolished the load-induced BMP
pathway activation [96].
A study performed by Kopf et al. (2012) sets the basis for the work presented here. They
showed a ligand-dependent force-specific activation of R-Smads in human fetal osteoblast.
Under BMP-2 stimulation and concurrent mechanical loading (f=1Hz, 10%), the
phosphorylation of Smad1/5/8 was increased in intensity and duration in comparison to the
BMP-2-only treated control. As the positive regulation was visible already after 15 min of
stimulation, it was suggested that mechanotransduction events integrate into the BMP
pathway already at the receptor level.
Moreover, the alterations on Smad-level were transmitted into transcriptional responses, as
the expression of direct BMP target genes (ID1, ID2) as well as BMP ligands (BMP-2, -6) and
the antagonist Noggin were regulated by mechanical stimulation. Mechanical loading alone
was not sufficient to activate R-Smads, but components of the non-Smad pathway were
induced ligand-independently. A strong induction of Akt, p38 and Erk1/2 phosphorylation
after 15 min could be detected in response to loading, but no further enhancement under
concurrent BMP-2 stimulation was observed [125].
24 Introduction
Figure 1-6: Regulation of BMP signaling by mechanotransduction. Mechanical forces like substrate deformation and
fluid flow trigger mechanotransduction events, which enhance Smad1/5/8 phosphorylation and consequently BMP
target gene expression. The exact mechanotransduction pathway feeding into the BMP pathway is unknown.
Even though the crosstalk between mechanotransduction and BMP signaling was
described in several studies, the underlying mechanism is still not understood (Figure 1-6).
Wang et al. (2010) attributed the regulation of BMP signaling by mechanical forces to the
load-induces downregulation of Smurf1 expression, which mediates the protein degradation
of Smads [96]. Others suggest an involvement of integrins and/or integrin mediated signaling,
which is described in more detail in the following section [128], [129].
1.5.1 Integrin-BMP receptor crosstalk
Influence of integrins on basal BMP signaling
Different integrin subtypes have been found to co-localize with type I and II BMP receptors
but contradicting statements were made concerning the influence of their association on BMP
signaling.
Positive regulation of basal BMP signaling by integrins was described in human
osteoblasts and osteosarcoma cells. In those cells, both BMP-receptor type I and II were found
to co-localize with αv and β1 containing integrins [130]. Treatment with function-blocking
αvβ, α1 and α2 antibodies reduced BMP-2-induced Smad transcriptional activity causing
reduced ALP, osteocalcin, osteopontin and bone sialo protein mRNA levels, thereby leading
to a reduction in osteogenic differentiation [130], [131]. Since blocking antibodies did not
affect the BMP receptor- integrin co-localization, it is suggested that integrin signaling rather
Introduction 25
than the physical interaction is responsible for the positive regulation [130]. In a recent study,
genomic deletion of β1 integrins in osteoblasts also caused a reduction in BMP-2-induced
expression of osteogenic marker genes like RUNX2, ALP, OCN and OSX [132].
However, other studies reported inhibiting effects of BMP receptor- integrin interaction
on BMP signaling. In MC3T3, CHO cells, primary osteoblasts and bone marrow derived MSCs,
BMP receptor type IA (BMPRIA) associated with integrin α1β1 shown by immunostainings
and immunoprecipitation. The site of integrin binding was identified the same as for BMP-2-
receptor binding [133] and knockdown of α1 integrin in CHO cells increased the level of Smad
phosphorylation. Together this led to the suggestion that α1 integrin and BMP-2
competitively associate with the BMPRIA receptor [134]. In neuronal stem cells, β1 integrins
interact with BMPRIA and IB and negatively influence the BMP- mediated astrocytic
differentiation, while a loss of β1 integrins result in an enhanced differentiation [135].
The role of integrins for the mechanoregulation of BMP signaling
Integrin activation, clustering and signaling is influenced by extracellular substrate
composition and stiffness as well as external mechanical stimuli like fluid flow or cell
straining. The effect of substrate stiffness or fluid flow on BMP signaling events have been
previously implicated with the interaction of BMP receptors and mechano-responsive
integrins.
A study comparing the influence of soft versus stiff polyacrylamide gels (elastic moduli:
Esoft ∼ 0.1–1k Pa and Estiff ∼ 50–100 kPa) on bone marrow MSCs differentiation provided
evidence for an integrin regulated trafficking of the BMP receptor. Cells on soft substrates did
not spread and displayed reduced surface levels of β1 integrins due to increased caveolae-
mediated internalization. Furthermore, on soft gels, BMP signaling was repressed indicted by
a strong reduction in Smad phosphorylation in comparison to stiff gels. Due to the observed
co-localization of BMPRIA and β1 integrins in intracellular vesicles, it was suggested that β1
integrins promote caveolae-mediated internalization of BMPRIA causing the decrease in
Smad phosphorylation and promoted neuronal differentiation of MSCs on soft gels [136].
Interestingly, covalent incorporation of BMP-2 into soft substrates could override the
influence of substrate elasticity on cell spreading. Via β3 integrins, C2C12 cells were able to
spread and organizes their cytoskeleton as they would on stiff substrates. In addition, BMP-
2-induced Smad signaling was found to be dependent on the inhibitory effect of β3 integrin
signaling on glycogen synthase kinase 3 (GSK3) activity. This mechanism requires the
activation of the downstream integrin signaling pathway Cdc42-Src-FAK-ILK [129], with ILK
previously shown to negatively regulate GSK3 via phosphorylation [137].
One study also related the regulation of the BMP signaling cascade by fluid flow to the
interaction of BMP receptors and integrins. In vascular endothelial cells, disturbed flow with
26 Introduction
oscillatory shear stress (OSS) induces the ligand independent phosphorylation of Smad1/5
and activation of the BMP signaling cascade, which was proposed to be pro-atherogenic. The
activation of Smad1/5 by OSS was attributed to the activation of the Shc/FAK/ERK pathway
following the interaction of αvβ3 integrins and BMPRIB. This association was interestingly
found to be mediated by the cytoplasmic kinase domain of BMPRII. Phosphorylated Smad1/5
in turn activated Runx2, mammalian target of rapamycin (mTOR) and p70S6 kinase signaling
leading to increased proliferation of endothelial cells [128].
Even though the mechanism of fluid flow-induced Smad phosphorylation was proposed
for vascular endothelial cells in an atherosclerosis model, it still remains to be elucidated in
the context of bone healing, where different cell types and mechanical stimuli are relevant.
This thesis aims to contribute to an enhanced understanding of how mechanical signals
integrate into the BMP pathway in the context of bone by investigating the role of intergins
and load-induced focal adhesion and actin cytoskeleton remodeling processes in human fetal
osteoblasts.
1.6 Mechanical forces and BMP-2 influence ECM formation
Mechanical properties, structure and composition of the ECM influences how external
mechanical forces are transmitted to the cell, how BMPs are recognized and in the end, how
cells respond in terms of proliferation, migration and differentiation. Vice versa, it will be
described in this section that external mechanical forces and BMP change the mechanical
properties and composition of the ECM by regulating the expression of ECM proteins and
remodeling enzymes. To understand these regulations, at first the composition and formation
of the extracellular matrix will be explained.
A small excursion into the extracellular space: Although the ECM composition is highly
tissue- specific, fundamentally it is composed of two main classes of molecules: fibrous
proteins and proteoglycans, which are further divided into subgroups. The main fibrous
proteins are collagens, elastins, fibronectins and laminins. Proteoglycan such as decorin,
aggrecan or perlecan, are composite molecules consisting of a core protein, which is
covalently linked to glycosaminoglycan (GAG) polysaccharide chains. Due to their hydrophilic
character, proteoglycans hydrate the ECM and serve as ion storages [138].
The ECMs` mechanical properties are greatly defined by the amount and structural
organization of collagens. Collagens are summarized in a family of 28 members that all feature
triple helical motifs within their structure. Based on their supramolecular assembly, collagens
can be further subdivided into fibril-forming, fibril-associated and network-forming
collagens. Fibrillar collagens (type I, II, III, V and XI) assemble into higher ordered, long, cable-
Introduction 27
like fibers, which mostly consists of not only one collagen type (heterotypic fibrils) [139]. The
formation of fibrils is a complex processes starting with the assembly of the triple helix after
numerous posttranslational modifications of the synthesized single proα-chain. The triple
helix is secreted into the extracellular space where the propeptides at each end of the triple
helix are cleaved enzymatically. The resulting tropocollagens self-assemble into staggered
collagen fibrils, a process which is regulated by cell-adhesions like integrins and ECM proteins
such as fibronectin and other collagens. Lysyl oxidases (LOX) covalently crosslink the
tropocollagens, stabilizing the fibril and strengthening its mechanical properties [140]. Non-
fibrillar collagens associate to and interconnect fibers and also serve as binding partners for
proteoglycans [139].
Especially during regeneration processes, but also for tissue maintenance, ECM
remodeling is essential. Dysregulation of ECM remodeling, however, can be the cause of many
different diseases such as cancer, fibrosis or arthritis. Matrix degradation is mostly mediated
by matrix metalloproteinase (MMPs), zinc-dependent endopeptidases that cleave both matrix
and non-matrix proteins with different specificities and efficacies. Their proteolytic activity
is controlled by tissue inhibitors of metalloproteinases (TIMPs) [141].
The interplay between matrix protein synthesis and degradation must be tightly
orchestrated and mechanical forces and BMP take part in this regulation.
Tension applied to fibroblasts on 2D substrates or embedded in gels induced the
expression of collagen type I and fibronectin [142]. Compression or relaxation, however,
reduced the ECM production but increased the secretion of collagenases [143]–[145]. On the
other hand, cyclic compression of open porous collagen scaffolds was shown to enhance
procollagen-I and fibronectin secretion, but also collagen degrading MMP1, pointing towards
and increased remodeling of the established ECM [146]. In tissue engineering approaches,
mechanical stimulation have been employed to enhance the mechanical properties of the
construct. Especially in the context of cartilage regeneration, cyclic mechanical compression
was shown to induce collagen II and aggrecan deposition of chondrocytes [147], [148].
Interestingly, also BMP-2 has been shown to stimulate the secretion of cartilaginous
matrix like collagen II, aggrecan and other proteoglycans but also matrix degrading enzymes
in chondrocytes [149], [150]. As a potent osteoinducer it is not surprising that MSCs and
osteoblasts increase their expression of bone ECM including osteocalcin, osteopontin, bone
sialo protein and collagen I upon BMP-2 treatment in a dose-dependent manner [151]–[153].
The response of fibroblasts to BMP-2 was less studied, and if, rather in the context of scar
formation. Scar tissue derived fibroblasts stimulated with BMP-2 showed an enhanced
28 Introduction
deposition of collagen I, suggesting a role of BMP-2 in the formation of hypertrophic scars
[154].
During bone healing, extracellular matrix formation is initiated directly with the end of
the pro-inflammatory phase [15] and the early structural organization of collagen fibers
within the fracture gap was shown to critically influence healing [11]. Given the importance
of early ECM formation processes and the fact that both BMP-2 [149], [155] and mechanical
forces [142]–[146] were independently described to influence such processes, it is even more
important to study how ECM formation is influenced by their mutual interaction. As those
mutual interactions might change individual effects, in this study the individual and mutual
influences of cyclic mechanical loading and BMP-2 stimulation on ECM formation were
compared. Since mechanical forces have already been shown to promote BMP signaling, it
was hypothesized here that these effects are transduced to the ECM-level, meaning that cyclic
loading is expected to increase the effect of BMP-2 stimulation on ECM formation.
1.7 Motivation and Aims
Even though bone is one of the few tissues in the human body that possess a great
regeneration potential, its natural self-healing capacity faces limitations resulting in delayed
healing or non-unions [2]. In such cases, BMP-2 is an established clinical treatment, which is
applied instead of, or in combination with autologous bone grafting [32]. However, the high
treatment costs and potential severe side effects due to supra-physiological concentrations
used, motivate further research on how to optimize its application. Interestingly, in vivo
experiments provide evidence that mechanical forces promote BMP-2-induced bone defect
healing [51] and in vitro studies report about a potentiation of BMP signaling by mechanical
stimuli [123], [125], [127], [128], [156]. Fine-tuned mechanical stimuli, either resulting from
extrinsic loading or featured by advanced biomaterials, could in future improve the growth
factor application by increasing its efficiency. However, to employ the power of the mechano-
biochemical interaction, a deeper understanding how both stimuli control cell behavior
independently and in combination is needed.
Therefore, this dissertation aims for an enhanced understanding of the molecular
mechanisms mediating the described mechano-regulation of the BMP signaling and the role
of the mutual interaction of mechanical signals and BMP-2 in controlling cell fate decision and
ECM formation.
To accomplish this, the first objective was to dissect the direct effect of mechanical forces
on osteo-differentiation of hMSCs from a mutual influence of mechanics and BMP. The
underlying hypothesis was that mechanical stimulation would directly induce osteogenic
Introduction 29
differentiation of hMSCs independent of BMP-2. To analyze the pure loading effect,
experiments were conducted under diminished autocrine signaling, under BMP-2
supplementation as well as under the specific exclusion of BMP from the system.
After investigating the individual and mutual influences of BMP-2 and mechanical forces
for osteogenic differentiation, the second step was to gain a deeper molecular understanding
of how mechanical signals integrate into the BMP signaling pathway. The basis for further
molecular investigations was set by a precise characterization of how mechanical parameters
influence the signaling dynamics. Thereafter, the hypothesis was tested if mechanical forces
integrate into the BMP pathway via integrin-mediated mechanosensation and the resulting
actin cytoskeletal adaptation. For this, first focal adhesion and actin reorganization in
response to mechanical loading and BMP stimulation were investigated and second, integrin
expression and actin remodeling dynamics were manipulated.
Besides the known osteoinductive properties of BMP-2, evidences point towards an
additional role in regulating tissue formation [149], [155], a process induced early during
bone healing [15]. Since the early ECM is believed to influence subsequent healing processes,
in a third step, the influence of mechanical loading and BMP-2 stimulation on early ECM
formation processes was studied. It was hypothesized that mechanical stimulation would
foster the growth factors` effects on ECM formation by enhancing BMP signaling. As collagens
greatly define ECM structure and its mechanical properties [140], which vice versa influence
cell behavior [56], [157], a particular focus was laid on collagen formation, as well as
microtissue structuring and stiffening.
In summary, the findings reported in this dissertation aim to contribute to a deeper
understanding how mechanical forces regulate osteogenic differentiation, BMP signaling and
early tissue formation processes, thereby influencing bone regeneration. In a long-term
perspective, the knowledge gained here about mechano-regulated cellular processes in the
context of BMP-2 signaling might help to better employ the power of mechanical forces in
critical bone healing scenarios.
30 Materials
2 Materials
2.1 Optimaix collagen scaffold
Marcoporous collagen scaffolds fabricated from purified porcine collagen suspensions using
a directional freeze and freeze drying method [158] were provided by Matricel GmbH
(Kaiserstraße 100, 52134 Herzogenrath, Germany). The collagen sponges are characterized
by an aligned channel-like open-porous architecture providing optimal oxygen and nutrient
supply. In its wetted state, the biomaterial exhibits a purely elastic behavior under repeated
compression up to 20% of the scaffold height, therefore suitable for cyclic mechanical
stimulations. In this study, scaffolds with collagen contents of 1.1 and 1.5 wt-% were used,
which differ in their elastic moduli, while biomaterial architecture is not affected. Collagen
scaffold were delivered dry, sterile and in a bulk material size of 30x40x3mm. Opened
packages were stored in sterile containers at 4°C for further use.
2.2 Bioreactor used for mechanical stimulation
A custom-made mechano-bioreactor system, previously described by Petersen et al. (2012)
was used to apply cyclic monoaxial compression to cell-seeded collagen scaffolds. The system
was designed to mimic the mechanical environment in the hematoma during the early phase
of fracture healing. The bioreactor can be separated in two compartments, the cell culture
unit and the mechanical unit. The cell culture unit can be assembled under sterile conditions
and consists of a reactor chamber, a medium reservoir allowing gas exchange and a micro
pump (pump rate approx. 2.5 ml/min). The bioreactor chamber can be equipped with
different scaffold-holders made from silicone, which offer space for up to 5 scaffolds with a
diameter of 5 mm.
Figure 2-1: Bioreactor setup. Schematic view of the mechanical unit in cross-section (A), showing piezo actuator (1),
cantilever for displacement amplification (2), linear actuator (3), wedge for translation to vertical movement (4),
lower arm (5), and force sensor (6). Arrows indicate movement direction. Picture into the opened mechanical unit
(B). Schematic view of the bioreactor chamber in cross-section (C), with glass housing (1), upper (2a) and lower (2b)
Materials 31
silicone sealings, threaded rings (3), upper (4a) and lower (4b) plunger, polyether ether ketone meshes (5), specimen
(6), and centering pins (7). Picture of the bioreactor chamber with collagen scaffold inserted (D). View inside the
chamber showing the position of the scaffold (E). Fully assembled bioreactor consisting of bioreactor chamber (1),
medium reservoir (2), micropump (3), 5µm filter (4), pressure equalization tube (5) and mechanical unit (6).
Schematic representations (A and C) were adapted from Petersen et al. 2012 [159].
The mechanical unit allows the application of defined loading patterns and the online-
measurement of the force acting on the sample. It contains two electro-mechanical devices, a
linear actuator and a piezo actuator for sample positioning and dynamic mechanical
stimulation, respectively. The horizontal motion of the linear actuator is translated into
vertical movement by a wedge, which slides underneath a bolt. The bolt deflects the lower
bioreactor arm leading to a movement of the lower plunger in the bioreactor chamber. Inside
the lower arm, a force sensor is mounted, which detects applied loads up to 15 N with a
resolution of 1.5 mN. The piezo actuator is attached to a pivoted cantilever, which increases
the displacement by three-fold. Both actuators are connected to their respective motor
controllers. In addition to the eight individual bioreactor units, the system contains a
micropump controller, a measurement data acquisition unit and a gassing system, listed in
Table 2-1. The system can be controlled and automated via a LabView interface. Mechanical
loading protocols, not only including the loading parameters (frequency, amplitude and
duration) but also mechanical compression tests during culture time can be run up to weeks,
while the data (force sensor, actuator positions) are recorded.
2.3 Bioreactor equipment and consumables
Table 2-1: List of bioreactor equipment
Equipment Model Manufacturer
Data Acquisition System Spider 8 Hottinger Baldwin
Messtechnik
Frequency Generator 180LF Wavetek San Diego
Gassing system MX 4/4 DASGIP Technologies
Incubator for bioreactors - Memmert
Interface for Pump Control NI USB-6501 National Instruments
Linear Actuator Servocontroller C-863 Mercury™ Physik Instrumente
Micropump mp6 Bartels Mikrotechnik
Mircopump Controller 8CH MTL Charite- Med. Tech.
Labore
Micropump Desktop Controller EDP0604 thinXXS
Piezo Actuator E-621.SR in E-500.621 Physik Instrumente
32 Materials
Table 2-2: List of bioreactor consumables
Item Ordering # Manufacturer
30 ml syringes 629502 CODAN Medical
Blue Filters Minisart 0.2 µm 16534K Sartorius Stedim
Brown Filters Minisart 5 µm 17594Q Sartorius Stedim
Perfusor 8722935 B.Braun
2.4 Flow chamber setup
The Ibidi Pump system (10902, Ibidi GmbH) used in this dissertation was kindly provided by
the Lab of Petra Knaus at the Freie Univerität Berlin. The setup, shown in Figure 2-2, consists
of a fluidic unit, a pump which is controlled via a PumpControl software and disposable parts,
such as perfusion sets and flow chamber slides (Table 2-3). The fluidic unit is equipped with
a sterile perfusion set (fluidic reservoirs and tubing) connected to the flow chamber slide and
performs the switching operation to create the unidirectional flow. The fluidic unit is further
coupled to the ibidi pump via two connectors, one electrical connection to control the valve
and a tubing for the pressurized air. To protect the pump from the moisture inside the cell
culture incubator, in which the fluidic unit is placed, the warm air passes through a drying
bottle.
Figure 2-2: Schematic representation of the ibidi Pump System. The fluidic unit is placed into the cell culture
incubator, while the pump and drying bottle are outside. Image taken from the instruction manual of the ibidi Pump
System.
Table 2-3: Consumables for the ibidi Pump System
Item Ordering # Manufacturer
μ-Slide I 0.8 Luer 80176 Ibidi
Perfusion set 10962 Ibidi
Materials 33
2.5 Devices
Table 2-4: List of devices
Device Model Manufacturer
Autoklave 5L steam pot Fissler
Cell counter Casy Modell TT Casy
Centrifuge (1.5-2 ml Eppendorf
tubes) Heraeus Fresco 17 Thermo Fischer
Centrifuge (15, 50 ml Falcon tubes) Rotofix 32 Hettich
Clean bench Safe2020 Thermo Fischer
Herasafe Heraeus
Cryotome CM1950 Leica
CO2 incubator cell culture APT.line™ Binder
Confocal multiphoton microscope TCS SP5 Leica Microsystems
Gel electrophoreses chamber SureLock™ Mini-Cell Thermo Fischer
Heating plate/stirrer - VWR
Ibidi Pump System 10902 Ibidi
Microplate reader Infinite pro 2000 Tecan
Microscope DM IL LED Leica
Multipipette Picus 5-120L BIOHIT/Satorius
Spectrophotometer (NanoDrop) ND-1000 Thermo Fischer
PCR cycler Mastercycler gradient Eppendorf
pH Electrode SenTix® WTW Weilheim
Power supply Power Pac HC Bio-Rad
Scale Scout pro 400g Ohaus
Thermomixer Thermomixer comfort Eppendorf
Ultrasound bath SONOREX SUPER RK 100H Schalltec
Vacuum pump AC1 PH-MTR Serie 71 Pfeiffer Vacuum
Water bath WNB 7-45 Memmert
Western blot XCell II™ Thermo Fischer
Western blot imaging system Odyssey Li-Cor
2.6 Chemicals, reagents and kits
Table 2-5: List of chemicals, reagents and kits
Item Ordering # Manufacturer
10% SDS solution 15553027 Thermo Fischer
10% Triton X-100 solution 93443 Sigma Aldrich
34 Materials
Item Ordering # Manufacturer
4x protein loading buffer 928-40004 Li-COR
Acetonitril, Ultra LC-MS (ROTISOLV > 99,98%) HN40.2 ROTH
Acetonitrile with 0.1% trifluoracetic acid (LC-MS) 34976 FLUKA
Alamar Blue® DAL1025 Thermo Fischer
Ammonium bicarbonate (for LC-MS) A6141-500G Sigma Aldrich
Bovine Serum Albumin 8076.2 Carl Roth
Cell tracker green (CMFDA) ab145459 abcam
cOmplete™, Mini Protease Inhibitor 4693124001 Roche
DMSO (anhydrous) ≥99% 276855 Sigma Aldrich
DNase I (700U) 10104159001 Roche
Luciferase assay system E4030 Promega
HEPES (hydroxyethyl-piperazineethane-sulfonic
acid buffer)
L1613 Merck
Nitrocellulose membrane 10600002 GE healthcare
NuPAGE™ 4-12% Bis-Tris Protein Gels NP0336BOX Thermo Fischer
NuPAGE™ MES SDS Running Buffer NP0002 Thermo Fischer
NuPAGE™ MOPS SDS Running Buffer NP0001 Thermo Fischer
Odyssey® Blocking Buffer (TBS) P/N 927-50000 Li-Cor
PageRuler™ Plus Prestained Protein Ladder 26619 Thermo Fischer
Paraformaldehyde 4 wt.% 1.04005.1000 Merck Millipore
Phosphate-Buffered Saline with Ca2+/Mg2+ 14040 - 091 Thermo Fischer
PhosSTOP™ phosphatase inhibitor 4906845001 Roche
Pierce™ BCA Protein Assay Kit 23225 Thermo Fischer
RIPA lysis buffer 806 Cell Signaling
Trifluoroacetic acid (Uvasol for spectroscopy) 1.08262.0025 Merck-Millipore
Trypsin (Sequencing Grade Modified, LC-MS) V5111 Promega
Water with 0.1% (v/v) formic acid (for LC-MS) 84867.320 VWR Chemicals
Water with 0.1% Trifuoroacetic acid (for LC-MS) 84871.320 VWR Chemicals
2.7 Buffer ingredients
Table 2-6: Buffer compositions
Chemicals concentration Ordering # Manufacturer
Ammonium chloride solution
NH4Cl 25 mM 1.01145.0500 Merck Millipore
PBS 1x 14190-094 Thermo Fischer
Materials 35
Chemicals concentration Ordering # Manufacturer
Western blot transfer buffer
Glycin 192 mM 0079.2 Carl Roth
Tris base 25 mM T6066 Sigma Aldrich
Methanol 20 vol.% 0082.3 Carl Roth
TBS-T (1x) Western blotting, pH 7.6
NaCl 136 mM 567440-1KG Merck Millipore
Tris 20 mM 108.382 Merck Millipore
Tween-20 0.1% P1379-100ml Sigma Aldrich
TBS-T (1x) Immunofluorescent staining, pH 8.2
NaCl 150 mM 567440-1KG Merck Millipore
Tris 50 mM 108.382 Merck Millipore
Tris-HCl 50 mM T3253 Sigma Aldrich
2.8 Cell culture
2.8.1 Cells
Human fetal osteoblasts (hFOBs 1.19)
hFOBs, immortalized by SV40 pUCSVtsA58 vector transfection, were purchased from ATCC
(Manassas, Virginia, USA) and used for mechanistic investigation on the crosstalk between
BMP signaling and mechanotransduction.
Primary human mesenchymal stromal cells (hMSCs)
Human MSCs were isolated from bone marrow aspirates of patients who underwent total hip
endoprosthesis. The isolation was performed by the Core Unit “cell harvesting” at the Berlin-
Brandenburg Center for Regenerative Therapies (BCRT). Cells were validated for their
osteogenic differentiation potential by stimulation with osteogenic medium and the
subsequent analysis of ALP activity and calcium amount in the ECM.
Primary human dermal fibroblasts (hdF)
Human dermal fibroblasts were isolated form skin samples. Experiments were conducted
using cells obtained from one male donor at the age of 26.
The isolation of primary human mesenchymal stromal cells and fibroblasts from patient-
derived material was approved by the Institutional Review Board of the Charité Berlin. All
patients gave their written consent.
36 Materials
2.8.2 Cell culture material
Item Ordering # Manufacturer
Biopsy Punch 5 mm 48501 Pfm medical AG
Chamber slides (8 well) 80826 Ibidi
Culture Flasks (75 cm², 175 cm², 300 cm²) 734.2315 TPP
DMEM Dulbecco’s Modified Eagle Medium
(low glucose (LG)) D5546 Biochrom GmbH
DMEM Dulbecco’s Modified Eagle Medium
(high glucose (HG)) 41965-039 Gibco® Life
Technologies
DMEM Dulbecco's Modified Eagle Medium
and Ham’s F-12 11320-082 Thermo Fischer
FBS Superior (fetal bovine serum) S0615 Biochrom GmbH
Geneticin disulphate (G418)-solution CP11.3 Carl Roth
GlutaMAX™ 35050-038 Thermo Fischer
MEM Non-Essential Amino Acids Solution
(NEA) K0293 Biochrom GmbH
Mr. Frosty™ freezing container 5100-0001 Thermo Fisher
Multiwell-Plates (6, 12, 24, 96 wells) 353046 BD Biosciences
Nutridoma-SP 11011375001 Roche
Penicillin/Streptomycin (P/S) A2213 Biochrom AG
PBS- Phosphate-Buffered Saline (w/o
Ca2+/Mg2+) 14190-094 Thermo Fischer
Trypsin/EDTA (10x) 59418C-10ml Biochrom GmbH
2.8.3 siRNAs, transfection reagents and transduction material
Table 2-7 Material for siRNA transfection and lentiviral transduction
Item Ordering # Manufacturer Storage
siRNA transfection
Lipofectamine™ RNAiMAX 13778500 Thermo Fisher 4°C
Opti-MEM™ serum-free 51985034 Thermo Fischer 4°C
Materials 37
siRNA (Silencer® select) ITGαv s7568 Thermo Fisher -20°C
siRNA Lincode Non-targeting #1 D-001320-01-05 Dharmacon -20°C
Lentiviral transduction
Hexadimethrine bromide
(Polybrene) TR-1003-G Sigma-Aldrich -20°C
Puromycin A1113802 Thermo Fisher -20°C
rLV-Ubi-LifeAct-TagGFP2
Lentiviral Vector 60141 Ibidi -80°C
2.8.4 Growth factor and small molecular inhibitor
Table 2-8: List of growth factors and small molecular inhibitors
Item Dissolved
in
Stock
concentration
Ordering
# Manufacturer Storage
rhBMP-2 1 mM HCl 1mg/ml - AG Prof. Dr.
Thomas Müller* -80/4°C
Jasplakinolide DMSO 1 mM 420107 Calbiochem -20°C
2.9 Materials for histology
Table 2-9: Consumables for histology
Item Ordering # Manufacturer
Antibody diluent S3022 Dako
Bovine Serum Albumin A7906-100g Sigma Aldrich
Cover slips 01-2446 Langenbrinck
Fluoromount-G® 0100-01 Southern Biotech
Microscope slides 08.100 00 Marienfeld
Normal donkey serum 017-000-1212 Jackson Immunoresearch
Normal goat serum S-1000 Vector Laboratories
Normal horse serum S-2000 Vector Laboratories
Scalpels 5518067 Aesculap AG
Tissue-Tek® Cryomold 4566 Sakura Finetek
Tissue-Tek® O.C.T.™ Compound 4583 Sakura Finetek
*Molekulare Pflanzengenetik, Lehrstuhl für Molekulare Pflanzenphysiologie und Biophysik, Julius-Maximilians-Universität
Würzburg
38 Materials
2.9.1 Primary and secondary antibodies
Table 2-10: List of primary and secondary antibodies
Target source clonality Ordering # Manufacturer
Primary antibodies
Collagen 1 Rabbit Monoclonal ab138492 Abcam
FAK Rabbit Monoclonal 13009 Cell Signaling
Technology
GAPDH Rabbit 14C10 monoclonal 2118 Cell Signaling
Technology
pERK1/2
(Thr202/Tyr204)
Rabbit Monoclonal 4370 Cell Signaling
Technology
pFAK(Y397) Rabbit Monoclonal 8556 Cell Signaling
Technology
pMLC (S19) Mouse Monoclonal 3675 Cell Signaling
Technology
pPaxilin (Y118) Rabbit Polyclonal 2541 Cell Signaling
Technology
pSmad 1/5
(S463/465)
Rabbit Monoclonal 9516 Cell Signaling
Technology
p-Src(Y416) Rabbit Monoclonal 6943 Cell Signaling
Technology
Secondary antibodies
Target source conjugation Ordering # Manufacturer
Anti-mouse
IgG
Goat Cy3-conjugated 405309 Biolegend
Anti-mouse
IgG
Goat Alexa Fluor 488-
conjugated
A11001 Thermo Fischer
Anti-rabbit
IgG
Donkey Cy3-conjugated 711-165-152 Jackson
Immunoresearch
Anti-rabbit
IgG
Donkey Alexa Fluor 488-
conjugated
A21206 Thermo Fischer
Anti-rabbit IgG Goat IRDye® 800CW 925-32211 Li-Cor
Anti-mouse
IgG
Goat IRDye® 680RD 925-68070 Li-Cor
Materials 39
2.9.2 Small molecular dyes for immunohistochemistry
Table 2-11: List of small molecular dyes used in immunohistochemistry
Item Concentration/ storage Ordering # Manufacturer
Alexa Fluor™ 488
Phalloidin
6.6 μM in MeOH at -20°C A12379 Thermo Fischer
Alexa Fluor™ 633
Phalloidin
6.6 μM in MeOH at -20°C A22284 Thermo Fischer
DAPI (4',6-Diamidino-2-
Phenylindole,
Dihydrochloride)
5 mg/ml in dH2O at -20°C
D1306 Thermo Fischer
DRAQ5 5 mM at 4°C 424101 Biolegend
2.10 RNA isolation, reverse transcription and qPCR
Table 2-12: Consumables and kits for RNA isolation, reverse transcription and pPCR
Item Ordering # Manufacturer
Ethanol EMPROVE® 1009861000 Merck Milipore
iQ™ SYBR® Green Supermix 170-8882 Bio-Rad
iScript™ cDNA Synthesis Kit 170-8891 Bio-Rad
Nuclease-free water AM9937 Thermo Fischer
Optically clear flat 8 cap strips TCS1080 Thermo Fisher
PureLink® DNase 12185010 Thermo Fischer
PureLink® RNA Mini Kit 12183018A Thermo Fischer
RNaseZAP® AM9780 Thermo Fischer
Semi- skirted 96 well PCR plate AB0900 Thermo Fisher
2.10.1 Primer
Table 2-13: List of primers for qPCR
Gene Protein Function Primer sequence 5’
3’
forward (fwd) and reverse (rev)
BMP-1
Bone morphogenetic
protein 1
Matrix metallo-
proteinase
fwd CTCCATCAAAGCTGCAGTTCC
rev CGGGATCTACCTCTCCATCTC
BMP-2 Bone morphogenetic
protein 2
BMP ligand fwd CATGCCATTGTTCAGACGTT
rev CAACTGGGGTGGGGTTTT
BMP-4 Bone morphogenetic
protein 4
BMP ligand fwd CCACGAAGAACATCTGGAGAAC
rev ATACGGTGGAAGCCCCTTT
40 Materials
BMP-6 Bone morphogenetic
protein 6
BMP ligand fwd GCAGACCTTGGTTCACCTTATG
rev
AGAATGTGTGTCCCCAGCA
BR1A BMP type I receptor
1a
BMP signaling fwd TTCGATGGCTGGTTTTGCTC
rev ACGACGTCTGCTTGAGATGC
BR1B BMP type I receptor
1b
BMP signaling fwd CCTGGAGAATCCCTGAGAGAC
rev AGTCCTTTGGACCAGCAGAG
BR2 BMP type I receptor
2
BMP signaling fwd GTTGGAGCTGATTGGCCGAG
rev TTTACAGCAACTGGACGCTC
c-fos
FBJ murine
osteosarcoma viral
oncogene homolog
Mechano-sensitive
TF
fwd CAAGCGGAGACAGACCAACT
rev AGATCAAGGGAAGCCACAGA
COL1A2 Collagen alpha-2(I)
chain
ECM proteins fwd AGCCGGAGATAGAGGACCAC
rev GGCCAAGTCCAACTCCTTTT
COL6A1 Collagen alpha-1(VI)
chain
ECM proteins fwd ACTGCGTATCAAGAAGGGG
rev TCGTTCACAGCATCCTCCAG
DLX2 Distal-less
homeobox 2 BMP target fwd GGCGTTTCCAAAAGACTCAA
rev CGAAGCACAAGGTGGAGAAG
EEF1A1
eukaryotic
translation
elongation factor 1
alpha 1
House-keeping
gene
fwd AACACAGGTGTCGTGAAAAC
rev AAGACCCAGGCATACTTGAA
ELN Elastin ECM proteins fwd TTTTATCCAGGGGCTGGTCTC
rev AGAGCCCCCGGAAAGGTAAC
FGF2 Fibroblast growth
factor 2
growth factors
fwd
AGCGGCTGTACTGCAAAAAC
rev AGCCAGGTAACGGTTAGCAC
FN1 Fibronectin ECM proteins fwd CAGCCAGTAGCTTTGTGGTC
rev GCATCAGGCGCTGTTGTTT
FBLN1 Fibulin 1 ECM proteins fwd CGGATGGCCACTCATCAGAAG
rev GCACCATCCTGCATTCTTTGG
HPRT1
hypoxanthine
phosphoribosyl-
transferase 1
House-keeping
gene
fwd TATGGACAGGACTGAACGTC
rev TGATGTAATCCAGCAGGTCA
ID1 Inhibitor of DNA
binding 1
BMP target fwd GCTGCTCTACGACATGAACG
rev CCAACTGAAGGTCCCTGATG
ID2 Inhibitor of DNA
binding 2
BMP target fwd GTGGCTGAATAAGCGGTGTT
rev TGTCCTCCTTGTGAAATGGTT
ITGα1 Integrin subunit
alpha 1
Cell adhesion fwd ACGCTGCTGCGTATCATTCA
rev CACCTCTCCCAACTGGACAC
ITGα5 Integrin subunit
alpha 5
Cell adhesion fwd TGGCCTTCGGTTTACAGTCC
rev GGTGCAGTTGAGTCCCGTAA
ITGαv Integrin subunit
alpha v
Cell adhesion fwd TCAGCAAGGCAATGCTCCAT
rev GAGGGCAAGATCCCGCTTAG
ITGβ1 Integrin subunit beta
1
Cell adhesion fwd CTGCGAGTGTGGTGTCTGTA
rev CACAGGATCAGGTTGGACCG
ITGβ3
Cell adhesion fwd ACCAGTAACCTGCGGATTGG
Materials 41
Integrin subunit beta
3
rev TCCGTGACACACTCTGCTTC
ITGβ5 Integrin subunit beta
5
Cell adhesion fwd ATACCTGGAACAACGGTGGAG
rev AGATCCTCAGGCTGATCCCA
LOX Lysyl oxidase ECM regulation fwd TGGCCGACCCCTACTACATC
rev TGGGGAAATCTGAGCAGCAC
LOXL1 Lysyl oxidase
homolog 1
ECM regulation fwd TGTACCGGCCCAACCAGAAC
rev GATGCTTGCACATAGTTGGGG
MMP1 Interstitial
collagenase
ECM regulation fwd ACATGAGTCTTTGCCGGAGG
rev ATCCCTTGCCTATCCAGGGT
MMP13 Collagenase 3 ECM regulation fwd TTGAGCTGGACTCATTGTCG
rev TCTCGGAGCCTCTCAGTCAT
Noggin Noggin BMP antagonist fwd GCCAGCACTATCTCCACATCC
rev GGGTGTTCGATGAGGTCCAC
TGFB1 Transforming
growth factor 1
growth factors fwd GGCCTTTCCTGCTTCTCAT
rev GTCCTTGCGGAAGTCAATGT
TGFB2 Transforming
growth factor 2
growth factors fwd ACTGTCCCTGCTGCACTTTT
rev GGGGTCTTCCCACTGTTTTT
TGFB3 Transforming
growth factor 3
growth factors fwd ATGAGCACATTGCCAAACAG
rev ATTGGGCTGAAAGGTGTGAC
TGFBI TGF-beta induced
protein
ECM proteins fwd TACGAGTGCTGTCCTGGATATG
rev GTTTGAGAGTGGTAGGGCTGC
TNC Tenascin ECM proteins fwd GTGAAAAACAATACCCGGGGC
rev CCGTAGGTCAGCTCAATGCC
THBS1 Thrombospondin ECM proteins fwd TCTGCAAAAAGGTGTCCTGC
rev AGAACAGGAGGTCCACTCGG
POSTN Periostin ECM proteins fwd CAGCAGTTTTGCCCATTGACC
rev CAGCAGTTTTGCCCATTGACC
RUNX2
Runt-related
transcription factor
2
Osteogenic marker fwd CTCCTACCTGAGCCAGATGA
rev CGGGGTGTAAGTAAAGGTGG
Smad7 Smad familiy
member 7
BMP signaling fwd TGCAACCCCTACCACTTCAGC
rev GAGACATGCTGGCGTCTGAG
Smurf1
SMAD specific E3
ubiquitin protein
ligase 1
BMP signaling fwd AATGAAGATGCGACCGAAAG
rev AGCCCGTAATAAGGATTCAGC
Smurf1
SMAD specific E3
ubiquitin protein
ligase 2
BMP signaling fwd TCCTCGGCTGTCTGCTAACT
rev GGGACTGTCAGGCATTCTGT
SPP1 Osteopontin Osteogenic marker fwd TCACCTGTGCCATACCAGTTA
rev TCATGGCTTTCGTTGGACTT
VEGFA Vascular endothelial
growth factor A
growth factors
fwd
CAGAAGGAGGAGGGCAGAAT
rev
CTGCATGGTGATGTTGGACT
42 Methods
3 Methods
3.1 Cell biological methods
3.1.1 Cell thawing and cultivation
Cryopreserved cells (1x106 per cryo-vial) were thawed at 37°C in the water bath under
constant movement of the cryo-vial until only a small piece of ice remained. The vial was
transferred into the clean bench and as soon as the ice disappeared completely, cells were
transferred into a cell culture flask containing the respective pre-warmed expansion medium
(Table 3-1). The day after, the medium was exchanged in order to remove the DMSO
containing freezing medium.
Human fetal osteoblasts (hFOBs) were cultured in a 1:1 mixture of Dulbecco's Modified
Eagle Medium (DMEM) and Ham’s F-12 (11320-033; Thermo Fischer Scientific)
supplemented with 1 vol.-% penicillin/streptomycin (P/S: A 2212; Biochrom), 0.3 mg/ml
Geneticin (CP11.3; Carl Roth) and 10 vol.-% fetal bovine serum (FBS: S0615; Biochrom). hFOB
were grown at 34°C and 5% CO2 in a humid incubator until ~80% confluence was reached
(after 3-4 days). For further expansion, cells were split in a ratio of 1:4, which corresponds
to approx. 2800 cells/cm². Cells were used from passage six to 15.
Primary human mesenchymal stromal cells (hMSCs) isolated from bone marrow were
expanded in DMEM low glucose (D5546; Sigma Aldrich) supplemented with 1 vol.-% P/S, 1
vol.-% GlutaMAX™ (35050-038, Life Technologies) and 10 vol.-% FBS. hMSCs were expanded
at 37°C and 5% CO2 in a humid incubator until ~80 to 90% confluence was reached and split
for further cultivation at a density of approx. 3300 cell/cm² (1x106/T300 flask). Cells were
used between passages three to five.
Primary human dermal fibroblasts (hdF) isolated from skin biopsies were cultured in
DMEM high glucose (# 41965; Gibco, Invitrogen) supplemented with 10% FBS, 1% P/S and
1% Nonessential Amino Acids (NEA: K0293; Biochrom). hdF were expanded at 37°C and 5%
CO2 in a humid incubator until 100% confluence was reached. For further culture cells were
split at a density of approx. 3300 cell/cm² (1x106/T300 flask). Cells were used between
passages four to nine.
C2C12-BREluc were expanded in DMEM low glucose supplemented with 10 vol.% FBS, 1
vol.% GlutaMAX™ and 0.5 mg/ml Geneticin at 37°C with 5% CO2 in a humidified incubator
until ~70% confluence was reached (every 2-3 days). For further expansion cells were split
at a concentration of 1100 cells/cm² (2x105/T175 flask) into a new culture flask. Cells were
used from passage eight to 15.
Methods 43
Table 3-1 Summary of expansion media composition
Component hFOB hdF hMSC C2C12-BREluc
DMEM (high glucose) x
DMEM (low glucose) x x
DMEM F-12 x
FBS 10 vol.-% 10 vol.-% 10 vol.-% 10 vol.-%
P/S 1 vol.-% 1 vol.-% 1 vol.-% 1 vol.-%
G418 0.6 vol.-% 1 vol.-%
GlutaMAXTM 1 vol.-%
NEA 1 vol.-%
3.1.2 Passaging and cryo-preservation
Cell passaging was performed by trypsinization using 1x Trypsin/EDTA solution (59418C,
Biochrom) at 37°C after two washing steps with 1x phosphate buffered saline (PBS: 14190-
094, Thermo Fisher). Cells were incubated for approx. 2-3 min with trypsin until they
detached and the reaction was stopped with the according expansion medium. Cells were
resuspended and given through a cell strainer to separate and remove aggregates. Cell
concentration was determined using the CASYTM Cell Counter (Model TT, Roche) by diluting
70 µl of the cell suspension in 7 ml CASY® ton. At the same time, cell suspension was
centrifuged at 375 x g for 6 min‡ or 325 x g for 8 min†. Thereafter, supernatant was sucked off
and cell pellet was resuspended in expansion medium to a concentration required for the type
of experiment.
Table 3-2: Volumes used for culture and trypsinization depending on the culture format
Flask Medium vol. PBS wash Trypsin/EDTA Medium to stop
T75 15 ml 5 ml 1 ml 4 ml
T175 35 ml 10 ml 2† / 2.5‡ ml 8†/7.5‡ ml
T300 60 ml 25 ml 4† /5‡ ml 16†/15‡ ml
For cell freezing, a concentration of 2x106 cells/ml was adjusted and suspension was
incubated on ice for 5 min. The freezing medium containing the respective expansion
medium, 20% FBS and 20% DMSO (anhydrous, Sigma Aldrich) was prepared and chilled on
ice to 4°C. Cryo-vials were filled with 500 µl of the cell suspension and 2x250 µl of the freezing
† hMSC, hFOB, C2C12-BREluc
‡ hdF
44 Methods
medium in two consecutive steps (final ratio 1:1). Vials were transferred into the -80°C
freezer overnight and finally stored in the gas phase of the liquid nitrogen container.
3.1.3 Seeding of collagen scaffolds
Cylindrical samples of 5 mm diameter were cut from the collagen scaffold sheet using a sterile
biopsy punch. Scaffold cylinders were dipped into the prepared cell suspension (see Table
3-3) that was immediately soaked up until the scaffold was completely filled. Seeded scaffolds
were placed into a 12-well-plate without additional medium. During a 60 min incubation, the
cells were allowed to adhere to the scaffold walls. Subsequently, the scaffolds were washed
in fresh expansion medium to remove unattached cells and placed in a 12 well plate on PEEK
(polyether ether ketone) meshes. The meshes enable improved supply from the bottom.
Depending on the type of experiment performed, scaffolds were either pre-incubated for one§
or two** days in the cell culture incubator prior to bioreactor experiments.
Table 3-3 Cell concentrations used for different cell types
Cell type Cell
concentration
Cell number
/ scaffold
Experiment
hFOBs, hdF, hMSC 5000 cells/µl ~3.25x105 mechanistic investigations
hdF 7500 cells/µl ~4.875x105 ECM formation
hMSC 2500 cells/µl ~1.625x105 load-induced osteogenesis
3.1.4 Lentiviral transduction of an actin marker
To visualize and follow the remodeling of filamentous actin (F-actin) in living cells, the GFP
tagged F-actin binding peptide LifeAct (17-amino acid long) was introduced into hFOBs by
viral transduction. In contrast to GFP-actin or other actin labeling methods, LifeAct was not
found to interfere with the actin dynamics[160]. A lentiviral vector was selected since it
mediates efficient transduction and stable integration into the genome of many different cell
types. The rLV-Ubi-LifeAct-TagGFP2 Lentiviral Vector (60141) containing 100µl of 1x107
TU/ml was purchased from Ibidi and the transduction was performed according to the
manufacture´s instruction under Safety Level 2 conditions.
The transduction efficiency was evaluated before by testing MOIs (multiplicity of
infection) of 0.5, 1 and 2. hFOBs were seeded in a 48 well plate at different concentrations
(30, 50 and 70% confluence). The next day, medium was removed and 200 µL transduction
medium, containing no antibiotics, 10% heat inactivated FCS and 8 µg/ml hexadimethrine
§ ECM formation, load-induced osteogenic differentiation
** Mechanistic investigations
Methods 45
bromide (Polybrene), was added. A 1:10 dilution of the lentiviral vector was prepared in PBS
and added to the cells according to Table 3-4. After a 20 hours incubation the medium
containing the lentiviral particles was removed and 200 µl fresh expansion medium was
added.
Table 3-4 Plate layout for the transduction efficiency test. Volumes taken from the 1:10 dilution of the vector
Lentiviral Titer 10
6
TU/mL
3200 cells/well
4800 cells/well
6500 cells/well
MOI
µl
µl
µl
0.5
1.6
2.4
3.25
1
3.2
4.8
6.5
2
6.4
9.6
13
total
50.75
At the fifth day after transduction, the transduction efficiency was evaluated under the
fluorescence microscope according to that a MOI 2 and a confluence of 50% was selected for
the final transduction.
Finally, hFOBs in P4 were seeded at 50% confluence (4.8x104 cells/well) in a 6 well plate
and transduced as described. At the fifth day, cells were transferred into a 25cm² cell culture
flask for expansion. The positive selection started on the next day by adding 1µg/ml
puromycin to the culture medium. The day after, dead cells were removed by medium
exchange that was repeated every 3 days until only transduced cell remained. The efficiency
of the selection process was thereafter evaluated by FACS analysis. Stable expressing hFOBs
are called in the following hFOB-LA.
3.1.5 Transfection of small interfering RNA
Knockdown of integrin (ITG) αv was performed by lipid-based transfection of small
interfering RNAs (siRNAs) into hFOBs using Lipofectamine™ RNAiMAX (13778500, Thermo
Fisher) according to the manufacture´s instruction. siRNA (A) and RNAiMax (B) dilutions
were prepared separately in serum-free Opti-MEM™ (51985034, Thermo Fischer) as shown
in Table 3-5. To from siRNA-lipid complexes, solution A and B were mixed in a 1:1 ratio and
incubated for 5 min. Cell suspension at a concentration of 1x105 cells/µl in antibiotic-free
expansion medium was added to the siRNA-lipid complexes in a 1:1 ratio and incubated for
further 5 min. Thereafter, cell suspension was transferred into a 48 well plate and scaffolds
(Ø = 5 mm) were seeded as described in section 3.1.3 and incubated for two days prior to
bioreactor experiments. The reverse transfection method, meaning simultaneous cell seeding
and transfection, was selected since it reached higher efficiencies then transfection of already
seeded scaffolds. A nonspecific siRNA (scrambled, scr) was used as a negative control to
determine the effect of lipid-based transfection in all RNAi-experiments. In general, nuclease-
46 Methods
free consumables (filter tips, tubes and water) and antibiotic-free cell culture medium for
transfection and the subsequent experiment was used.
Table 3-5 Calculations for seeding of one scaffold (volume = 80 µl) per condition
A1
A2
B
siITG αv
scr
RNAiMAX dilution
siRNA stock
concentration
20 µM
100 µM
-
siRNA working
concentration
25 nM
25 nM
-
siRNA (µl)
1 µl of a 1:10
dilution in water
2 µl of a 1:100
dilution in water
-
Opti-MEM
20 µl
20 µl
2 x 20 µl
RNAiMAX
-
-
2 x 0.25 µl
3.1.6 Bioreactor cultivation, mechanical loading and BMP stimulation
The assembly of the bioreactor cell culture units was performed under sterile conditions. Pre-
warmed medium was filled into syringes and transferred via perfusor tubes inside the
reservoirs. Depending on the type of experiment and its duration, the medium amount and
the FBS concentration varied (see Table 3-7). Additional medium supplements were added
according to the used cell type as summarized in Table 3-1. The medium was pumped into the
bioreactor chamber until the lower sample holder was covered, thereby preventing sample
dehydration during scaffold positioning. Carefully the cell-seeded scaffolds were placed into
the custom-made silicon holders and covered with another PEEK mesh. The upper plunger
was inserted, the chamber was sealed and the complete unit was mounted onto the
mechanical subunit in the incubator. Thereafter, gas mixing and pump control units were
connected and activated. Rough positioning of the plungers was carried out manually using
the LabView interface, while fine tuning was performed using a force-controlled automated
sample positioning protocol. Further experimental settings were dependent on the respective
research aim and are described in the following:
3.1.6.1 Load-induced osteogenic differentiation
To study the influence of load-induced autocrine signaling, in particular the enrichment of
BMP-2, on hMSC differentiation, three experimental conditions were selected, which are
listed in Enrichment of autocrine factors in the cell culture medium during bioreactor
cultivation was promoted by placing five scaffolds (Ø = 5 mm) into each chamber and by
reducing the volume of cell culture medium to 12 ml. On the other hand, autocrine factors
Methods 47
were strongly diluted in reactors containing only one scaffold and 27 ml medium.
Additionally, to compare the impact of medium conditioning to direct BMP-2 stimulation, 135
ng/ml recombinant human BMP-2 was added at day 4 of cultivation into bioreactors
containing one scaffold and 27 ml medium.
Table 3-6.
Enrichment of autocrine factors in the cell culture medium during bioreactor cultivation
was promoted by placing five scaffolds (Ø = 5 mm) into each chamber and by reducing the
volume of cell culture medium to 12 ml. On the other hand, autocrine factors were strongly
diluted in reactors containing only one scaffold and 27 ml medium. Additionally, to compare
the impact of medium conditioning to direct BMP-2 stimulation, 135 ng/ml recombinant
human BMP-2 was added at day 4 of cultivation into bioreactors containing one scaffold and
27 ml medium.
Table 3-6: Summary of experimental conditions
Experimental condition Scaffolds per
reactor
Medium
volume
Cell to medium
ratio (cells/ml)
1 Enabling autocrine stimulation 5 12 ml 6.25x104
2 Disabling autocrine stimulation 1 27 ml 0.56x104
3 BMP-2 addition 1 27 ml 0.56x104
hMSC were subjected to cyclic compression with a frequency of 1 Hz and an amplitude of
10%. Mechanical loading was applied periodically with 3h stimulation and 5h break. Loading
resulted in a compression of the scaffold in the direction of the scaffold pores. At the end of
the experiment, cells were either fixed in 4% PFA (IF) or lysed in the RNA isolation lysis buffer
(qPCR).
3.1.6.2 Mechanistic investigations
Two days after scaffold seeding and culture in growth medium containing 10 % FBS, scaffolds
were transferred into the bioreactor. For short- term bioreactor experiments up to 120 min,
FBS-free cell culture medium was used. After positioning, samples were left untreated for
three hours in order to decrease unspecific background signaling due to the presence of
growth factors in FBS and as a consequence of mechanical deformations that the sample
experiences when mounted in to the reactor chamber. Next, BMP-2 was diluted to 4185 ng/ml
in starvation medium and 0.5 ml was injected into the bioreactor reservoir to reach a final
concentration of 135 ng/ml. The respective loading protocol was started immediately
afterwards. To study the crosstalk dynamics, cells were stimulated with cyclic uniaxial
mechanical loading with different frequencies (0.03, 1 and 10 Hz), amplitudes (5% and 10%
48 Methods
of the scaffold height) and durations (15, 30, 90, 120 min, 24h). At the end of the experiment,
cells were either fixed in 4% PFA (IF) or lysed in the respective assay buffer for further
analysis (WB, qPCR).
To investigate whether load-induced actin cytoskeleton rearrangement processes are
necessary for the crosstalk, different small molecular inhibitors (Table 2-8) were used during
bioreactor experiments. Jasplakinolide (Jas) was used to stabilize actin filaments. Jas at a
concentration of 0.1 µM was supplemented to the medium at the beginning of the experiment
so that cells were treated for three hours prior to BMP-2 stimulation and mechanical loading.
As a control to Jas treated samples, DMSO was supplemented to the medium. After the
starvation phase, scaffolds were subjected to cyclic mechanical compression of 10% with a
frequency of 1 Hz for 90 min in this particular experiment. Thereafter, cells were either fixed
in 4% PFA (IF) or lysed in the respective assay buffer for further analysis (WB, qPCR).
3.1.6.3 Consequences of the crosstalk for ECM formation
As described in section 3.1.6.2, samples were left untreated for three hours in order to
decrease unspecific background signaling. Next, BMP-2 was diluted and injected into the
bioreactor reservoir (final concentration of 135 ng/ml in 15 ml) and loading protocol was
started. hdF in collagen scaffolds were allowed to form ECM during one and two weeks
bioreactor culture under BMP-2 stimulation and cyclic mechanical loading. To enable fibrillar
collagen formation, cell culture medium contained ascorbic acid at a concentration of 1.36
mM. Bioreactor culture of two weeks required half a medium exchange on day seven.
Scaffolds were subjected to cyclic compression with a frequency of 1 Hz and an amplitude of
10%. Cyclic compression was applied periodically with 3h stimulation and 5h break in the
direction of the scaffold pores. BMP-2 (135 ng/ml) was added at day one, five and 10, while
medium without BMP-2 was added to unstimulated controls. BMP-2 was injected at the start
of each mechanical stimulation phase. At the end of the experiment, cells were either fixed in
4% PFA (IF) or lysed in the respective assay buffer for further analysis (WB, qPCR).
Table 3-7: FBS concentration and medium amount inside bioreactors depending on the type of experiment
Type of
experiment Cell type Experiment
duration
Medium
volume
FBS
concentration
Mechanistic
studies
hFOBs, hdFs, hMSCs ≤ 2h 15 ml 0%
hFOBs 24h 15 ml 1%
ECM formation hdFs one and two
weeks
15 ml 2%
Osteogenic
differentiation
hMSCs one week 27 ml or 12 ml 10%
Methods 49
3.2 Molecular biological methods
3.2.1 Ribonucleic acid isolation from collagen scaffolds
For RNA isolation, exclusively nuclease-free materials (filter tips, tubes, water) were used and
all surfaces and the equipment were wiped with RNaseZap™ (AM9780, Thermo Fisher).
Total RNA was isolated using the PureLink® RNA Mini Kit (12183018A, Thermo Fischer)
and DNA digestion was performed using ON-column PureLink® DNase (12185-010, Thermo
Fischer). Isolation from cells grown in 2D were performed according to the manufacture’s
instruction. For 3D collagen scaffold cultures, however modifications were implemented,
which are described in the following.
At the desired endpoint of the experiment, scaffolds were placed onto a sterile filter paper
to removethe culture medium and subsequently transferred into a tube containing 500 µl
lysis buffer supplemented with 1 vol.-% β-mercaptoethanol. Tubes were vortexed and frozen
at -80°C at least overnight. For the isolation, samples were thawed and centrifuged at 3000 x
g through a 10 µl pipette tip, which was loaded with a glass bead hindering the scaffold from
passing, while collecting the total lysate at the tube bottom. The empty scaffold was discarded
and the lysate was processed according to the manufacture’s instruction.
Finally, the elution of the RNA was performed twice with nuclease-free water pre-warmed
to 60°C to increase the yield. RNA concentrations were measured using a NanoDrop
Spectrophotometer.
3.2.2 Reverse transcription, quantitative polymerase chain reaction and data
evaluation
For each sample, 500 ng RNA was transcribed to complementary DNA using the iScript™
cDNA Synthesis Kit (#170-8891, BIO-RAD) according to the manufacture’s instruction. The
transcription was performed inside the Mastercycler® gradient (Eppendorf). The resulting
cDNA was stored for further qPCR analysis at -20°C.
Quantitative determination of messenger RNA transcription was performed using the
SYBR green- based PCR. The reaction mixture contained 5 ng cDNA, 500 nM Primer (100 µM
stock concentration) and 50% of the iQ™ SYBR® Green Supermix (170-8882, Bio-Rad). The
reaction was performed in an iQ™5 Real-Time PCR Detection System (Bio-Rad) following the
steps listed in Table 3-8. The assessed CT values were processed according to the efficiency
corrected ΔΔCT–method [161]. The primer efficiencies listed in Table 2-13 were determined
by measuring a cDNA standard curve as described elsewhere [162]. Melting curves were
recorded to check for non- specific amplifications, like primer dimers. Three different
housekeeping genes (Hypoxanthine Phosphoribosyltransferase 1 (HPRT), Beta-2-
50 Methods
Microglobulin (B2MG) and Eukaryotic Translation Elongation Factor 1 Alpha 1 (EEF1A))
were tested for their stable expression under mechanical loading and BMP stimulation and
HPRT was selected for normalization.
Table 3-8 qPCR reaction steps
Cycle 1 Cycle 2 (40x) Cycle 3 Cycle 4 (80x)
step initiation denaturation annealing elongation denaturation melting
°C 95 95 60 72 95 55-95(0.5 incr.)
min 3 0:30 0:30†† 0:30†† 1 0:10††
3.2.3 Cell lysis for protein analysis and extraction of ECM proteins
For the analysis of intracellular proteins, cells were lysed using 1x RIPA lysis buffer (9806,
Cell Signaling Technology) supplemented with protease (cOmplete™, 4693124001, Roche)
and phosphatase inhibitors (PhosSTOP™, 4906845001, Roche), while extracellular proteins
were extracted by an adapted protocol described previously[163].
Intracellular protein extraction: Cells harvested from 2D surfaces were washed once in 500
µl ice-cold PBS and subsequently lysed in 100 µl RIPA buffer. After 5 min incubation on ice,
cells were scraped off using a pipette tip and lysate was collected and frozen at -20°C. Cells
from scaffold cultures were washed in 200 µl ice-cold PBS by placing it on a filter paper, which
first removed the culture medium and subsequently the PBS. The sample was transferred into
100 µl RIPA buffer vortexed thoroughly and incubated on ice for 4 min. Thereafter, lysates
were sonicated for 30 seconds to increase the extraction efficiency and again vortexed
thoroughly. A 10 µl pipette tip with a glass bead was placed into the tube and loaded with the
scaffold. By centrifugation for 2 min at 3000 x g at 4°C, the scaffold was dried and the lysate
was collected and subsequently stored at -20°C.
Extracellular matrix extraction: Scaffolds were transferred to -80°C and frozen samples
were pulverized under liquid nitrogen conditions using custom-made silicone vessels and
steel pestles. Scaffold powder was dissolved in 200 µl of ECM extraction buffer I, vortexed
thoroughly and sonicated at 4°C for 2 min inside an ultrasound bath. Thereafter, 100 µl of
ECM extraction buffer II was added, vortexed and samples were centrifuged at 5000 x g for 2
min to remove insoluble fragments. The supernatant was collected and stored at -20°C.
†† Measurement of fluorescent intensity
Methods 51
3.2.4 Sodium dodecylsulfate polyacrylamide gel electrophoresis
The gel electrophoresis was conducted using the NuPAGE® electrophoresis system (Thermo
Fischer) and NuPAGE™ 4-12% Bis-Tris protein gels. Lysates were mixed with 4x LDS loading
buffer (Li-Cor) and heated for 4 min at 95°C to denature the proteins. Samples were cooled
down to 4°C on ice, centrifuged briefly and lysates were loaded onto gradient gels clamped
into XCell SureLock™ Mini-Cell container (Thermo Fisher). In addition, a pre-stained protein
marker (26619, Thermo Fisher) was loaded to monitor the gel electrophoresis and to indicate
the location of the proteins of interest. SDS-PAGE was run with 1x MES buffer at 150V for 80-
90 min until the desired protein separation was achieved. Finally, polyacrylamide gels were
removed from the plastic case and processed as described in the following section.
3.2.5 Western blotting and protein detection
Gel and nitrocellulose membrane were equilibrated for 5 min in transfer buffer, which was
prepared according to Table 2-6. Thereafter, the blotting sandwich was assembled in the
following order from bottom to top: filter paper, gel, membrane, filter paper. The sandwich
was placed between blotting sponges soaked with transfer buffer and fitted inside the XCell
Blot module (Thermo Fisher). The module was mounted inside the XCell SureLock™ Mini-Cell
container, filled with transfer buffer and the transfer was run at 30V for one hour.
Next, the membrane was rinsed in TBS, incubated for another hour in Odyssey® Blocking
Buffer (TBS) (P/N 927-50000, Li-Cor) under constant shaking, before it was cut as desired.
Membrane sheets were further incubated overnight at 4°C in the respective primary antibody
dilution, which was prepared according to the manufacture´s instruction. The next day,
membranes were washed three times with TBS-T and incubated with the secondary antibody
(Li-Cor) diluted 1:15000-1:20000 in 3% BSA/TBS-T for two hours at room temperature. The
secondary antibody is coupled to an InfraRed-Dye, therefore the subsequent steps were
performed protected from light.
Finally, membranes were washed three times for 10 min with TBS-T and one time with
TBS to remove excess secondary antibody, before proteins of interest could be detected using
the Li-Cor Odyssey® infrared imaging system (Li-Cor Biosciences). The signal intensity of the
detected protein bands were quantified in the Li-Cor software by contouring the respective
band with a rectangular ROI. The signal intensity of the protein of interest was normalized to
GAPDH or β-Actin.
52 Methods
3.3 Immunocytochemistry
3.3.1 Sample preparation including fixation and cryo-trimming
Cells cultured on 2D surfaces or in collagen scaffolds were fixed in 4% paraformaldehyde
(PFA) at room temperature for 15 min or for at least 5 hours, respectively. To quench the
reaction, samples were incubated in 25 mM ammonium chloride solution (in PBS) for one
hour at room temperature. After two times consecutive washing in PBS, scaffold samples
were infiltrated at 37°C by a 5% gelatin/sucrose solution overnight. Gelatin was solidified at
4°C for 30-60 min and scaffolds were cut along the symmetry axis using a scalpel. Scaffold
halves were transferred into PBS and gelatin was washed out at 37°C during repetitive
exchange (3-4x) of PBS every one hour. To generate a plane imaging surface, the cut sides of
the scaffold halves were trimmed in a cryotome. Therefore, the samples were embedded into
Tissue-Tek® (Sakura Finetek) and snap-frozen on a metal bar placed into liquid nitrogen. In
the cryotome, approx. 200 µm was trimmed off the surface in 10 µm steps and the Tissue-
Tek® was subsequently washed out with PBS at 37°C.
3.3.2 Immunofluorescence staining
Proteins of interest were stained indirectly by coupling a fluorophore-conjugated antibody to
a primary antibody that specifically recognized and bound to an epitope of the protein. The
immunofluorescence (IF) staining protocol was adapted to each antibody combination and
combined with nuclei or actin labeling using small molecule probes. An overview of the
procedure including pre-treatments, blocking and antibody incubation is listed in Table 3-9.
Wash buffer composition and materials used for IF stainings can be found in Table 2-6 and in
section 2.9, respectively.
Table 3-9 General IF staining protocol
Step Condition Time
Wash (optional) 0.025% TBS-T 10 min
Permeabilization (optional) 0.1-0.5% TBS-T pH 8.2 10 min
Wash 0.025% TBS-T 3 x 10 min
Blocking I 1% BSA/TBS 10 min
Blocking II 5% normal serum (NS)/1%BSA/TBS 30 min
Primary antibody different concentrations in diluent (Dako) overnight, 4°C
Wash 0.025% TBS-T 3 x 10 min
Methods 53
Secondary antibody different concentrations diluted in
5%NS/1%BSA/TBS
2h
Wash 0.025% TBS-T 3 x 10 min
DNA staining DAPI (in Ampuwa) or Draq5 (in TBS) 15 or 60 min
Wash Ampuwa or TBS 3 x 10 min
Storage PBS 4°C
Actin labeling via Alexa Fluor-coupled phallotoxins (Phalloidin) was combined with the
secondary antibody incubation step. All steps including the fluorophore were conducted in
the dark to avoid photobleaching.
3.4 Confocal multiphoton microscopy
3.4.1 Image acquisition and analysis
Images were acquired using a Leica TCS SP5 confocal microscope equipped with an Argon-,
two Helium-Neon- and a Mai Tai HP multiphoton laser. Overview scans and images of small
structures such as focal adhesions were obtained by using the 25x (image size 620 x 620 µm)
or 63x (98.5 x 98.5 µm) water immersion objective, respectively. Second harmonic generation
(SHG), a phenomenon in which two photons of the same wavelength within specific molecular
structures generate one photon with halve the wavelength but twice the frequency, was
utilized to visualize fibrillar collagen in a label-free manner [164]. Both, porcine collagen of
the scaffold as well as newly in vitro-deposited collagen mediates the photon conversion by
its ordered fibrillar architecture. In this case, the wavelength was set to 910 nm and signal
was detected in the range of 450-460 nm. In general, photons were detected either using an
internal photomultiplier or an external non-descanned detector (NDD). To compare different
samples, all settings like laser power, z-spacing, detection range etc., were kept constant.
Recorded images were analyzed in Fiji, a packaged distribution of ImageJ, using different
custom-made macros.
Analysis of focal adhesion number and size per cell was performed from Phospho-Paxillin
(Tyr118) (#2541, cell signaling) stainings. Recorded z-stacks were transferred into a
maximum projection, cell outlines were contoured in the actin-channel and the p-Pax-channel
was binarized. Number and size of adhesions per cell were determined by particle tracking.
Particle size was categorized and normalized to the total number of particles per cell.
Analysis of signal density and orientation was performed for fibrillar collagen recorded by
second harmonic imaging (SHI). For the quantification of collagen density, z-stacks were
54 Methods
summed up and scaffold pores were contoured manually so that only the in vitro-deposited
collagen was captured. The density was calculated by summing up a background subtracted
histogram and dividing it by the ROI (region of interest) area. Orientation distribution was
analyzed from maximum projections, which were previously aligned to the pore orientation,
using the ImageJ plugin OrientationJ [165].
3.4.2 Live-cell-imaging and analysis
Time-lapse live-cell imaging was performed to investigate migration or actin reorganization
processes. Therefore, the Leica TCS SP5 confocal microscope described above, was used in
combination with a custom-made incubation chamber and a gassing unit to maintain cell
culture conditions.
For migration experiments, cells seeded in collagen scaffolds were stained by cell tracker
green (ab145459, abcam) at a dilution of 1:1000 in 0% FBS containing medium for 1h at 37°C.
Cells were washed once in cultivation medium and scaffolds were transferred into a custom-
made silicone holder mounted in a stainless steel chamber with an optical view field. At
minimum four different scaffolds positions were marked and images were acquired every 30
minutes using the 25x objective at a resolution of 512 x 512 pixel with 4 μm z-spacing. Cell
migration velocity was analyzed from 3D stacks using the TrackMate [166] ImageJ pugin.
For imaging of actin remodeling processes, hFOB-LA were seeded in 8-well chamber slides
(80826, Ibidi) and 5 µm image stacks with a z-spacing of 1 µm were recorded every 6 seconds
over 3 min using the 63x objective at 1024 x 1024 pixel resolution. Protrusion dynamics of
whole cell protrusions were analyzed using a custom-made Image marco. In brief, z-stacks
were projected, cell outlines were contoured for each time point and the area in between two
consecutive ROIs was determined. Mean area change over time was calculated for the image
sequence. Representative images showing the cell outline change over time were prepared
using the QuimP [167] ImageJ plugin.
Figure 3-1: Analysis of protrusion remodeling. Exemplary image sequence showing a LifeAct®-transduced hFOB
during the course of 90 min (images acquired every 5 min) (A). The recording was processed using the QuimP [167]
ImageJ plugin to illustrate cell morphology changes over time by the color-coded outline. Assessment of the cell area
Methods 55
change as a measure for actin remodeling dynamics (B). The shaded area between cell outlines indicates the area
change from time point 1 to 2 which was measured between all consecutive images using ImageJ.
3.4.3 Mechano-imaging
Bioreactor-Microscope-Setup: To image cell seeded scaffolds in the bioreactor, both the
mechanical and the cell culture unit needed to be modified. The lower silicone sealing of the
bioreactor chamber, containing the lower plunger and the silicone sample holder, was
replaced by a silicone sealing with circular glass window (8 mm diameter, 0.2 mm thickness).
The scaffold sample is now positioned directly on the glass window bottom. Due to the
changed sample position, the upper star-shaped plunger needed to be replaced by a small
round stamp (6 mm diameter), glued to a PEEK mesh of the same size. In the original setup,
the bioreactor chamber is resting on the lower arm of the mechanical unit that would block
the newly introduced optical window. Therefore, the lower arm was modified by introducing
an opening of the size of the glass window. The window can be inserted and clamped tightly
into the opening, enabling both sample positioning and imaging.
Figure 3-2: Bioreactor-Microscope-Setup. Schematic representation of the modified bioreactor chamber (A) with
scaffold (1), modified upper piston (2), modified silicon sealing (3) sample cup (4) and glass window for optical access
with inverted microscope (5). Pictures showing the modified bioreactor setup on the Leica TCS SP5 confocal
microscope (B). The yellow arrow indicates the microscope objective below the optical window of the chamber.
The modified bioreactor setup was combined with the Leica TCS SP5 confocal microscope
by placing the whole mechanical unit on top of the microscope table after removing the
condenser head. The gas mixing unit, pump and motor controller were connected and the
sample was brought into contact with the upper stamp by using a force-controlled automated
sample positioning protocol. After precise positioning, the optical window of the bioreactor
chamber was aligned with the 25x or 63x objective for imaging. For straining experiments,
hFOB-LA or hFOBs stained with cell tracker green were used. Images were acquired either
directly after mechanical loading or during stepwise scaffolds compression.
Flow chamber setup: The Ibidi Pump System, which is owned by the Lab of Petra Knaus at
the Freie Univerität Berlin, was used for the application of fluid shear stress to hFOB-LA. The
experiments were conducted according to the manufacture´s instruction in collaboration
with Dr. Maria Reichenbach. In brief, hFOB-LA were seeded inside µ-slides (80176, Ibidi) at a
concentration of 1.2x105 cells/ml one day prior to the flow experiment. Growth medium was
56 Methods
exchanged to FBS-free starvation medium buffered with 20 mM HEPES (L 1613, Merck) and
incubated for two hours. µ-Slides were transferred to the Leica TCS SP5 confocal microscope
and connected to the pump system also containing HEPES-buffered FBS-free starvation
medium. Thereafter, cells were allowed to rest for an additional hour before actin remodeling
under static conditions was recorded according to section 3.4.2. Laminar shear stress of 5
dyn/cm² was applied and time-lapse images under flow were recorded at time point 10, 30
and 90 min. Subsequently, the inhibitor Jasplakinolide (0.05 µM) was added into the medium
reservoir and time-lapse videos were recorded after 10 and 20 min. Actin remodeling
dynamics was quantified as described in section 3.4.2.
3.5 Scaffold contraction analysis
Cell-mediated scaffold contraction was assessed by scanning the sample at day one after
seeding (t0) and at the end of the experiment (t1) using a commercially available digital
scanner (Epson Perfection V200). Therefore, samples were placed inside a 48-well-plate filled
with expansion medium specific for the cell type used (Table 3-1) and scanned both in top
and side view. The cross-sectional area was determined from top views by manual contouring
of the scaffold outline. The side view was used for the measurement of the scaffold height by
calculation the mean distance between bottom and top. From these values, the total volume
contraction (in %) was calculated as described in Figure 2-2.
Figure 3-3: Schematic representation of scaffold contraction and calculation of total volume contraction (V(V0-Vt)).
3.6 Mechanical compression tests
The scaffolds mechanical properties were assessed by performing mono-axial compression
tests using the BOSE ElectroForce Mechanical TestBench equipped with a 50 g load cell
(Model 31 Low load cell, Honeywell Corp.). To calculate the elastic modulus from the data
obtained by compression testing, the scaffold-dimensions were assessed prior to the
measurement. Empty scaffolds or native cell seeded scaffolds were placed into a custom-
made chamber filled with PBS. The chamber consists of the same upper and lower star-
shaped plunger that were also used in the bioreactor, to which PEEK meshes are glued
distributing the applied force. Three consecutive compression cycles at a speed of 0.05 mm/s
and a displacement of 10 or 20% of the scaffold height (adjusted for each sample individually)
were performed with a resting time of 30 seconds at 0, 10 and 20% displacement.
Methods 57
Recorded load/displacement curves were converted into stress (σ)/strain (ε) curves
according to:
𝜎𝜎[𝑃𝑃𝑃𝑃] =
𝑚𝑚 ∗ 𝑔𝑔
𝐴𝐴
m = mass [g]
g = 9.81 m/s2 (gravitational acceleration)
A = cross sectional area [m²]
3.1
𝜀𝜀=
𝑙𝑙
∆𝑙𝑙
l = length [m]
∆l= change in length [m] 3.2
The Young's modulus, as a measure for the material stiffness, corresponds to the slope of
the stress/strain curve in the linear region (equation 3.3).
𝐸𝐸 [𝑃𝑃𝑃𝑃] =
𝜎𝜎
𝜀𝜀
3.3
3.7 Decellularization after in vitro tissue formation
Scaffolds seeded with hdF were cultured for two weeks in the bioreactor as described in
section 3.1.6.3. In preparation for mass spectrometry analysis of the ECM, cellular
components were removed from in vitro-grown micro-tissues by detergent-based
decellularization and DNA digestion using DNase I. Decellularization was performed in an in-
house developed perfusion system consisting of individual sample chambers connected via
silicone tubes to a peristaltic pump and detergent reservoirs. Samples were actively perfused
with a fluid velocity of 2.5 ml/min. The decellularization protocol used here (Table 3-10) was
previously established and optimized to preserve many soft ECM components while
removing most of the cellular components.
Table 3-10: Perfusion protocol
Step
Condition
Time
ddH
2
O
sterile deionized water with 1x cOmplete™
protease inhibitor and 5 mM Tris (pH 8)
60 min
detergent treatment
0.05% SDS in deionized water
20 min
PBS wash
PBS with Ca
2+
/Mg
2+
2 x 20 min
DNA digestion
DNasesI (350U/ml) dissolved in PBS (with
Ca2+/Mg2+)
5h
detergent treatment
0.025% SDS in deionized water
20 min
wash
sterile deionized water
20 min
Thereafter, samples were frozen at -80°C and subsequently freeze dried.
58 Methods
3.8 Mass spectrometry
Mass spectrometry of decelluarized samples was performed by the Core Unit “Tissue Typing”
and conduced as described previously [168]. In brief, decelluarized and freeze dried samples
were subjected to tryptic digestion at 37 °C for 3h and overnight. Peptides were extracted
with trifluoridic acid (0.1% (w/v)), desalted with ZipTip and analyzed by LC/ESI–MS.
Peptides were separated using an analytical UHPLC System (Dionex Ultimate 3000 RSLC,
Thermo-Fisher) and analyzed by a ESI-QTOF-mass spectrometer (Impact II, Bruker). Mass
spectra were evaluated using PEAKSX+ software (PEAKS Studio 10.5 (Bioinformatics
Solutions Inc., Waterloo, Canada) [169] automatically searching the SwissProt database.
MS/MS ion search was performed with the following set of parameters: a) taxonomy: homo
sapiens (human) (20366 sequences); b) proteolytic enzyme: trypsin; c) maximum of accepted
missed cleavages: 2; d) mass value: monoisotopic; e) peptide mass tolerance 10 ppm; f)
fragment mass tolerance: 0.05 Da; and g) variable modifications: oxidation (M), deamidation
(N,Q) and acetylation (N-therm). Only proteins with scores corresponding to p < 0.01 and
with at least two identified peptides were considered.
3.9 Statistical analysis and data presentation
Data analysis was performed using Microsoft Excel 2016. The OriginPro 2015G (OriginLab
Corporation) software was used for the graphical presentation and statistical analysis of the
obtained data. Box and line plots show mean values with standard deviation. Box and whisker
plots display the maximum and the minimum, the upper and lower quartile and the median
marked as a horizontal line of all data points. For statistical analysis, the non-parametric, two-
sided Mann- Whitney-U test was performed. For the comparison of multiple groups the p-
value was corrected according to the Bonferroni method using the following equation:
p* = p·n , with n=number of statistical tests. P values < 0.05 were considered as statistical
significant. Different significance levels are indicated as: # p<0.1; * p<0.05; ** p<0.01; and ***
p<0.001.
Results 59
4 Results
4.1 Load-induced osteogenic differentiation via BMP-2
In the first part of this thesis, the direct influence of cyclic mechanical loading on the
osteogenic differentiation of primary human bone marrow MSCs (hBMSCs) is being
investigated and dissected from the effect of load-induced autocrine signaling, in particular
of BMP-2. Multiple studies examined the influence of mechanical loading on stem cell
differentiation, including osteogenic commitment (see section 1.3.3 and reviews [55], [170]).
Motivated by a tissue engineering approach, the majority of these studies used osteoinductive
medium supplements, bone derived scaffolds or hydrogels with limited supply masking
effects of loading on cell fate decision. Even in studies working without additional osteogenic
triggers, the influence of load-induced autocrine signaling was not investigated. Therefore, it
still remains unclear whether the observed mechano-sensitivity is a direct consequence of
cyclic compression, an indirect effect of altered supply or a specific modulation of autocrine
BMP signaling.
To elucidate this, special emphasis was put on the selection of the experimental setup and
its physiological relevance. Therefore, the in vitro setup used here was specifically chosen to
resemble the mechanical environment during the early phase of bone healing as it was
observed in vivo. The bioreactor was used to simulate interfragmentary compression that
occur as a consequence of weight bearing in the rage of reported data for external fixation in
bone healing in sheep [37], [43], [159]. The utilized scaffolds are characterized by low elastic
moduli mimicking the soft tissue matrix in the fracture gap and have been shown to
successfully induced endochondral ossification in a rat bone defect model [11]. Due to its
elastic deformation behavior, the material withstands repetitive compression, as it was
shown previously [171]. As the stiffness of the substrate that cells adhere to is known to be
an important regulator influencing cellular behavior [56], scaffolds with bulk stiffnesses of
3.4 kPa (scaffold A) and 12.3 kPa (scaffold B) were used in this study (Figure 4-1A). To
additionally strengthen the in vivo relevance, primary hBMSCs obtained from at least five
donors were used.
4.1.1 Cell morphology, proliferation and oxygen concentration inside scaffolds
cultured in the bioreactor
Bioreactor cultivation and cyclic compression might have altered the cell morphology and
proliferation that could cause differences in the later on investigated migration and
differentiation behavior. Therefore, it was verified in the beginning that neither bioreactor
culture, nor cyclic compression have major influences on cell morphology and number by
60 Results
comparing hBMSC-seeded scaffolds cultured for seven days in the bioreactor (with and
without cyclic compression) to hBMSC-seeded scaffolds cultured for one and seven days in
the cell culture incubator (static). Images acquired using confocal multiphoton imaging
(Figure 4-1B), showed no differences in cell distribution and morphology between different
culture time points (one or seven days) and conditions (static, bioreactor without cyclic
compression, bioreactor with cyclic compression). Cells were homogeneously distributed
throughout the scaffold and showed an elongated morphology in the direction of the scaffold
pores. Analysis of the cell density seven days after seeding neither showed significant
differences between static and bioreactor culture nor an alteration in response to cyclic
compression (Figure 4-1C). The comparison with the cell density one day after seeding
revealed that the cells remained viable but did not proliferate significantly in the
microenvironment provided by the scaffold. The oxygen concentration, measured by opto-
chemical microsensors introduced into the sample, showed only a slight decrease from the
surface (20.7/20.5%) to the center of the sample (18.9/18.5%) at day 3/7 of culture (Figure
4-1E).
Figure 4-1: Bioreactor setup validation. (A) Electron microscopy image of the two scaffold prototypes with collagen
solid contents of 1.1 and 1.5 wt-%. (B) Bioreactor consisting of reactor chamber (1), medium reservoir (2), micro
pump (3), filter (4), pressure equalization tube (5) and the mechanical unit (6). (C Close-up of the bioreactor chamber
with collagen scaffold inserted. (D) Human BMSCs obtained from three donors were seeded in collagen scaffolds and
Results 61
cultured for one or seven days in well plates under static conditions or in the bioreactor with and without cyclic
compression (f=1Hz, 10% axial compression). Representative confocal images showing hBMSCs in the collagen
scaffold (white), stained with phalloidin (green) and DAPI (blue) to visualize the F-actin fibers and the cell nuclei,
respectively. (E) Cell density (cells/mm³) inside the scaffold as analyzed from confocal image stacks (mean ± SD, 3
donors, ns = not significant). (F) Concentration of oxygen measured inside the scaffold depending on the distance from
the scaffold surface (mean ± SD, n=2 scaffolds per time point). Figure modified from [172] with permission form the
publisher.
This verified that the cells were well-supplied even in the center of the scaffold throughout
the duration of the experiment. Consequently, a potentially improved supply resulting from
enhanced fluid flow under cyclic compression could be excluded. Thus, the effects of cyclic
compression on gene expression and protein secretion reported in this study could be linked
to direct mechanical consequences of cyclic compression. In contrast, supply in other, less
open-porous biomaterials might be significantly enhanced by cyclic compression (reduced
hypoxia, increased viability) as reported before for cell-seeded fibrin hydrogels [173].
4.1.2 Cyclic mechanical compression downregulates the expression of key
osteogenic marker genes but upregulates BMP-2 expression
Next, the impact of cyclic mechanical compression on the mRNA expression of osteogenic
marker genes was quantified by qPCR (Figure 4-2). Therefore, hBMSCs were seeded in
collagen scaffolds and subjected to cyclic compression of 5% and 10% magnitude.
Surprisingly, the median fold-change mRNA expression levels of RUNX2 (early transcription
factor for osteogenesis) were reduced in both scaffold types in response to 10% compression,
with statistical significance for scaffold A [FC
�(RUNX2)scaffA,10%= 0.8, p = 0.0002;
FC
�(RUNX2)scaffB,10%= 0.81]. Cyclic compression of 5% significantly decreased the RUNX2
expression in scaffold A [ FC
�(RUNX2)scaffA,5%= 0.7, p = 0.01] while no change was visible in
scaffold B [FC
�(RUNX2)scaffB,5%= 0.98]. The median mRNA expression of osteocalcin (BGLAP)
and collagen type 1 α 2 (COL1A2) were also decreased for both scaffold stiffnesses and
loading magnitudes in comparison to the uncompressed controls. Statistical significant
downregulation was reached for BGLAP in response to 10% compression in scaffold B
[FC
�(OC)scaffB,10%= 0.56, p= 0.0007].
The median expression of osteopontin (SPP1) under 5% cyclic compression remained
unchanged, whereas 10% compression significantly increased the SPP1 expression for the
softer scaffolds [FC
�(OP)scaffA,10%= 2.24, p = 0.01] but not for the stiffer. Interestingly,
mechanical stimulation induced an upregulation of BMP-2 mRNA expression in scaffold B at
5 and 10% and in scaffold A at 10% compression. Statistical significance was reached at 10%
compression in scaffold B [FC
�(BMP2)scaffB,10%= 1.5, p = 0.02].
62 Results
Figure 4-2: Cyclic compression downregulates the expression of key osteogenic marker genes but upregulates BMP2
expression. Human BMSCs obtained from various donors were seeded in collagen scaffolds of 3.4 kPa (n=8 donors)/
12.3 kPa (n=6 donors) stiffness, respectively. Scaffolds were cultured in the bioreactor and stimulated with 5% or 10%
intermittent cyclic compression or left unstimulated (0%). mRNA expression was analyzed by qPCR. Fold change
expressions to the 0% control group of the respective scaffold type are depicted in box and whisker plots. Figure
reproduced from [172].
In summary, we found a clear downregulation of important osteogenic marker genes,
especially of RUNX2, in response to cyclic mechanical compression. However, BMP-2, a potent
inducer of osteogenic differentation, was clearly upregulated under 10% cyclic compression.
4.1.3 Limited biochemical conditioning of the culture medium during
bioreactor culture
To study whether the observed increase in BMP-2 gene expression resulted in an increased
protein secretion, BMP-2 protein concentrations were analyzed using an enzyme-linked
immunosorbent assay (ELISA) specific for human BMP-2 (Figure 4-3). Since BMP-2 gene
expression was significant upregulation in hBMSCs seeded in scaffold B (12.3 kPa) and
stimulated with 10% cyclic compression, this condition was analyzed in comparison to the
unstimulated control (0%).
Results 63
Figure 4-3: Low BMP-2 concentrations in the
conditioned medium. BMP-2 concentration in
the conditioned bioreactor media was
analyzed using a human BMP2 ELISA. Only the
medium of hBMSCs seeded in scaffold B (12.3
kPa) and stimulated with 10% intermittent
cyclic
compression was analyzed in
comparison to the unstimulated control (0%),
as this condition induced a significant
upregulation of BMP-2 gene expression. Figure
modified from [172].
In agreement with the expression data, a slight increase of BMP-2 secretion was detected
in culture media of samples stimulated with 10% compression [𝛽𝛽
�(𝐵𝐵𝐵𝐵𝑃𝑃2)0%= 252 pg/ml,
𝛽𝛽
�(𝐵𝐵𝐵𝐵𝑃𝑃2)10%= 280 pg/ml]. However, the detected BMP-2 concentrations were in general
very low and only slightly above the BMP-2 concentration detected in the culture medium
without cells. According to Katagiri [174], these concentration are not capable to stimulate an
osteogenic response. This result can be explained by the comparably low cell number in
respect to the large medium volume (1.5x105 cells/ 27 ml medium = 5.6x103 cells/ml) in the
bioreactors. The cell-to-medium ratio in the bioreactor was approximately nine times lower
compared to typical 2D culture conditions (5x104 cells/ml). Therefore, the gene expression
data shown in Figure 4-2 was obtained from bioreactor experiments under a strong dilution
of secreted proteins. The observed gene regulations thus represented consequences of cyclic
compression without relevant contributions of autocrine biochemical stimulation.
4.1.4 Cyclic mechanical compression enhances RUNX2 mRNA expression only
in a BMP-enriched environment
With the goal to promote medium conditioning and autocrine BMP-2 signaling, the ratio
between cell number and medium volume was increased in subsequent experiments. The
number of scaffolds per bioreactor was increased from one to five and the medium volume
was reduced from 27 ml to 12 ml (minimal filling volume of the bioreactor) resulting in an
increase in cell-to-medium ratio from Rlow=0.56x104 to Rhigh=6.25x104 cells/ml. Additionally,
we conducted separate control experiments where 5 nM (135 ng/ml) recombinant human
BMP-2 (rhBMP-2) was added at day 4 of bioreactor cultivation, to compare the impact of
medium conditioning to direct BMP-2 stimulation. These experiments were conducted using
the 12.3 kPa scaffold and the five donors of the six that showed a consistent downregulation
in RUNX2 expression upon cyclic compression.
64 Results
As expected, the increase in cell-to-medium ratio enabled a significant 5.5-fold increase
of BMP-2 concentration (p = 0.004) in the cell culture medium from β
�(BMP2)Rlow,−cyclic comp.
= 252 pg/ml to β
�(BMP2)Rhigh,−cyclic comp.=1395 pg/ml (Figure 4-4B, light grey vs. light blue
box). In response to cyclic compression the BMP-2 concentration increased slightly but
significantly (Fig. 4B, dark blue vs. light blue box, [β
�(BMP2)Rhigh,+cyclic comp. = 1623 pg/ml]).
This is in agreement with the observed increase in BMP-2 gene expression under load
strengthening the assumption that cyclic compression triggers a positive feed-forward loop
for BMP-2.
Next, we analyzed the concentration of BMP-2 in the conditioned media collected from
bioreactors that were supplemented with rhBMP-2. Also here mechanical loading increased
BMP-2 concentration slightly but non-significantly (Figure 4-4B, dark vs. light orange box). It
has to be mentioned that the detected BMP-2 concentrations were overall very low compared
to the initially added amount of rhBMP-2 (135 ng/ml). To understand this discrepancy, we
analyzed the BMP-2 stability in the bioreactor. Therefore, 135 ng/ml rhBMP-2 was added to
the bioreactor experiment (conducted without cells) and medium samples were collected
after 30min and at day 1, 3, 5 and 7 (supplementary Figure 0-3). Already after one day, the
BMP2 concentration decreased to about one third and declined further during culture.
Together this indicated that BMP-2 stability is transient under cell culture conditions.
Results 65
Figure 4-4: Cyclic compression only increases RUNX2 expression, if rhBMP2 is added or an enrichment of cell-secreted
BMP2 in the cell culture medium was permitted. Data was obtained from three different experimental conditions: 1.
Human BMSCs were seeded in collagen scaffolds (12.3 kPa) and cultured with or without cyclic compression (f=1Hz,
ε=10%) for six days. 2. BMSCs were additionally stimulated with recombinant human BMP2 (rhBMP2, 5nM) added at
the fourth day of cultivation (+ rhBMP2). 3. The cell number was increased five times and the medium volume was
reduced from 27ml to 12ml (“high cell-to-medium ratio”). (A) Illustration of low vs high cell-to-medium ratio. (B)
Human BMP2 ELISA of collected bioreactor media from all loading experiments was conducted. The relative gene
expressions of (C) RUNX2, (D) BMP2, (E) BMP4, -6 and Noggin normalized to the untreated control (low cell-to-
medium-ratio, without cyclic compression) (n=5 hBMSC donors B-E). Figure reproduced from [172].
Signaling initiated by BMP-2 directly stimulates the expression of RUNX2 and balances its
transcriptional activity [175], [176]. Therefore, BMP-2 is an important trigger for early
osteogenic responses. Consequently, the observed upregulation of BMP-2 expression in
response to mechanical loading led us to the assumption that hBMSCs would trigger
themselves towards osteogenic differentiation (enhance RUNX2 expression) in response to
cyclic loading under increased cell-to-medium ratio Rhigh.
Strikingly, under Rhigh conditions, RUNX2 expression was significantly upregulated in
response to cyclic compression in comparison to the uncompressed control (Figure 4-4C,
dark blue box, [FC
�(RUNX2)Rhigh,+cyclic comp. = 1.3, p = 0.036]). This finding stands in strong
contrast to the regulation of RUNX2 in response to cyclic loading under Rlow conditions in the
original setup (Figure 4-4C, gray box, [FC
�(RUNX2)Rlow,+cyclic comp.= 0.8, p = 0.021]).
Additionally, RUNX2 expression under rhBMP-2 supplementation was also further enhanced
by cyclic compression (Figure 4-4C, dark orange box [FC
�(RUNX2)+rhBMP2,+cyclic comp.= 1.3]).
This indicated that BMP-2 is capable to alter the cell’s gene expression response to mechanical
loading. Moreover, in response to (i) the increase of the cell-to-medium ratio from Rlow to Rhigh
66 Results
and (ii) the supplementation of rhBMP-2, concurrent cyclic compression further enhanced
the expression of BMP-2, suggesting a positive feed-forward regulation by biochemical self-
stimulation (Figure 4-4D, dark blue and dark orange box, [ FC
�(BMP2)Rhigh,+cyclic comp.= 2.5,
FC
�(BMP2)+BMP2,+cyclic comp.= 3.5]).
To investigate the involvement of possible further feed-forward components we also
analyzed the expression of BMP-4, -6 and -7, as well as the expression of the BMP antagonist
Noggin (Figure 4-4E). Expression of BMP-4 was not affected by neither condition and BMP-7
transcription was not detected. BMP-6 expression, instead, was found to be sensitive to cyclic
compression. Under rhBMP-2 supplementation, loading induced a 1.4-fold increase in BMP-6
transcription, while in the Rhigh condition, only a slight 1.1-fold increase was detected. Taking
into account the observed mechano-sensitivity of BMP-6 expression, a potential contribution
to the RUNX2 regulation cannot be excluded. However, in comparison to the according
changes in BMP-2 expression, the contribution of regulations in BMP-6-expression are
regarded to be low. Noggin expression strongly increased by 4.3-fold in response to rhBMP-
2 treatment and increased further to 5.6-fold by cyclic compression. Under high cell-to-
medium ratio, cyclic compression increased Noggin transcription by 1.9-fold. The regulation
of Noggin expression under both conditions is line with the increased BMP2 expression and
medium concentration. Noggin inhibits BMP2, -4, -7 and -14 from binding to the BMP
receptor. BMP6 however is more resistant to noggin inhibition [177]. Additionally, the
expressions of Transforming growth factor beta-1 and 3 (TGFβ1 and β3), Fibroblast Growth
Factor-2 (FGF2), Platelet-derived Growth Factor-A (PDGF) and Vascular Endothelial Growth
Factor-A (VEGF-A) that are of relevance in osteogenic differentiation were analyzed
(supplementary Figure 0-2). FGF-2 and PDGF-A expressions were not regulated by any of the
treatments. TGFβ1, TGFβ3 and VEGF-A expressions were increased in response to cyclic
compression under rhBMP-2 stimulation. However, in the high cell-to-medium ratio group,
only the expression of TGFβ3 was increased by 1.35-fold upon cyclic compression, while the
others were not consistently regulated. No statistical significant differences could be found.
Therefore, in comparison to the 2.5-fold change in BMP-2 expression change under cyclic
compression ( FC
�(BMP2)Rhigh,+cyclic comp.= 2.5) only modest changes were detected.
Results 67
Figure 4-5: Cyclic compression does not increase RUNX2 expression, if BMP signaling is inhibited by rhNoggin. (A)
Validation of rhNoggin efficiency by western blot analysis of p-Smad1/5 level after rhBMP2 stimulation.
Phosphorylation was normalized to GAPDH (n=3, one donor). (B) Scaffolds were cultured under high cell-to-medium
ratio with or without 10% cyclic compression and with or without rhNoggin stimulation (100ng/ml, added at day 1,
3, 5) and RUNX2 and ID1 expression were analyzed. Gene expressions were analyzed by qPCR. HPRT1 was used as the
reference gene and expressions were normalized to the untreated control (low cell-to-medium-ratio, without cyclic
compression) (n≥4 for one donor). Figure reproduced from [172].
To finally verify that the load-induced increase in RUNX2 expression is mediated by BMP-
2, in the next experiment BMP-2 was depleted from the system by recombinant human
Noggin. At first, the effect of the rhNoggin on BMP signaling was validated in a separate
experiment by western blot analysis investigating the phosphorylation of the transcription
factor Smad1/5 (Figure 4-5A). While Smad1/5 phosphorylation was increased by 3-fold in
response to 5 nM rhBMP-2, no increase was detected if cells were treated with rhBMP-2 and
rhNoggin (12 nM). Next, for the proof of concept experiment, one hBMSC donor representing
the group was selected. Cells were cultured under Rhigh conditions with or without cyclic
compression and rhNoggin (100 ng/ml equals 2.2 nM) which was added at day 1, 3 and 5.
Again RUNX2 expression was significantly upregulated in response to cyclic compression,
however treatment with rhNoggin abolished the effect of compression completely. The
RUNX2 expression was significantly reduced when rhNoggin was supplemented.
Furthermore, the expression of ID1 (inhibitor of DNA binding 1), a common only used BMP
target gene, was investigated. Under Rhigh conditions, ID1 expression was significantly
induced by cyclic compression and significantly reduced by rhNoggin treatment (Figure
4-5B).
Taken together, the results show that cyclic compression strongly downregulated the
expression of RUNX2 when the cell-to-medium ratio was low and BMP self-conditioning was
impeded by dilution. However, after increasing the ratio, cyclic compression significantly
upregulated RUNX2 expression. Both observation, (i) that rhBMP-2 stimulation led to a very
similar result as the increase in cell-to-medium ratio and (ii) that the load-induced effect on
RUNX2 and ID1 expressions were abolished by rhNoggin treatment, verify the role of BMP-2
as the relevant mechano-regulated signaling factor controlling RUNX2 and consequently
osteogenesis.
68 Results
4.2 Mechanistic investigations on the crosstalk between mechano-
transduction and BMP signaling
So far it was found, that the increased BMP expression in response to cyclic compression
contributes to a positive feed-back loop enhancing RUNX2 expression. In addition, cyclic
compression not only increases the expression of but also the sensitivity for BMP-2. In a
ligand dependent manner, mechanical stimulation was shown to enhanced BMP-2-signaling.
Even though the mechano-regulation of BMP signaling was described previously, still many
questions are unanswered: Is the observed mechano-regulation a general phenomenon, or
exclusive for some cell types? What mechanical requirements need to be met to regulate BMP
signaling? How is the dynamics of BMP signaling altered? Which mechanotransduction
pathway is involved? The following part aims to address those questions.
4.2.1 The crosstalk is relevant in primary human cells of the mesenchymal
lineage
The regulation of the BMP signaling pathway by external mechanical stimuli has been
described in several cell types, including cell lines like C2C12 myoblasts [178], MC3T3-E1
[179] and human fetal osteoblasts (hFOB) [125] as well as primary cells like murine and rat
osteoblasts [123], [127] or human vascular endothelial cells [128]. However, primary human
cells from the mesenchymal lineage have not been tested so far.
For the relevance of the study and to investigate the influence of BMP-2 and mechanical
stimulation on extracellular matrix formation (section 4.3) and osteogenic differentiation
(section 4.1), it was required to verify the existence of the crosstalk in primary human
fibroblasts and MSCs. Therefore, Smad1/5/8 phosphorylation, an immediate early event
downstream of the BMP receptor, was investigated in hMSCs and hdFs upon treatment with
BMP-2 and mechanical loading. The cell line hFOBs, which was used in a previous study
investigating the mechano-regulation of BMP signaling [125], was selected as a reference. The
conditions selected for the experiments (duration and loading parameters) have been
validated in hFOBs (see section 4.2.2) before and were found to result in maximum crosstalk
strength.
Cell-seeded collagen scaffolds were subjected for 90 min to BMP-2 stimulation, cyclic
uniaxial compression (e=10% of the scaffold height, f= 1Hz) or a combination of both and
Smad1/5/8 phosphorylation was analyzed by western blotting. BMP-2 treatment resulted in
an increase in Smad phosphorylation, which was notably the highest for hdF. Importantly,
concurrent mechanical loading led to the significant increase of the BMP-2-induced Smad
phosphorylation consistently in all cells tested. Even though the fold change increase of B/L
Results 69
to the control is the same for hFOBs and hdF, the effect of loading reflected by the difference
between B/L and B is the highest for hFOBs. In none of the cells, mechanical stimulation alone
was not able to induce a phosphorylation of Smads. Taken together the results obtained here
with the studies mentioned above, it can be suggested, that the effect of external mechanical
stimuli on BMP signaling represents a fundamental regulatory mechanism to control the
effectiveness of BMPs.
Figure 4-6: Cyclic mechanical compression significantly increases the BMP-2-induced Smad1/5/8 phosphorylation in
human primary MSCs and dermal fibroblasts (hdF). hFOBs, hMSCs and hdFs seeded in collagen scaffolds were
subjected for 90 min to BMP-2 stimulation, cyclic uniaxial compression (e=10% of the scaffold height, f= 1Hz) or a
combination of both. Thereafter, cells were lysed and Smad1/5/8 phosphorylation levels were determined via western
blotting. Signal intensities were related to GAPDH and the fold changes to the untreated controls were calculated, n=3
from one donor.
4.2.2 Correlation between loading frequency and crosstalk duration
The crosstalk between mechanotransduction and BMP signaling was shown to be induced by
different mechanical forces, including laminar and oscillatory shear stress [124], [127], [128],
mechanical stretch [180] and compression [123], [125]. However, a systematic investigation
of how different loading parameters of the same force magnitude influence the duration and
strength of BMP signaling is missing. Such investigations would, however, be important to
defined parameters optimally supporting BMP signaling and furthermore, to gain insides into
the dynamics of this regulation. Therefore here, the impact of the loading frequency on early
BMP signaling events was investigated in a time-dependent manner.
Collagen scaffolds seeded with hFOBs were subjected for 30, 90, or 120 min to BMP
stimulation and cyclic mechanical compression of selected frequencies (0.03 Hz, 1 Hz and 10
Hz) and Smad1/5/8 phosphorylation was examined via western blotting (Figure 4-7 A-D).
The expression of ID1 and ID2, early BMP target genes, was analyzed after 90 min by qPCR
(Figure 4-7 E).
Already after 30 min, mechanical stimulation induced a significant increase in Smad
phosphorylation in comparison to the BMP-only treated control for all frequencies applied.
However, the crosstalk induced by 0.03 Hz loading was significantly reduced in comparison
to 1 Hz, while no significant difference was detected between 1 Hz and 10 Hz. After 90 min of
70 Results
stimulation, the maximum increase in Smad phosphorylation was reached for 1 Hz and 10 Hz
loading, while 0.03 Hz could not maintain the crosstalk. Whereas the Smad phosphorylation
induced by 1 Hz loading decreased after 120 min to the level of the BMP-only treated control,
high frequency loading with 10 Hz maintained the maximum phosphorylation level, therefore
inducing a prolonged crosstalk. The frequency dependent phosphorylation of Smads is
furthermore reflected in the expression level of ID1 and ID2. Especially the ID1 transcription
increases with increasing frequency. Interestingly, even 0.03 Hz loading, which only mildly
and transiently increased the Smad phosphorylation, increased the ID1 expression by two-
fold.
Taken together, a positive correlation between the frequency of mechanical loading and
the duration of the crosstalk was observed, which is indicated by a prolonged increase of
Smad phosphorylation.
Figure 4-7: Loading frequency influences strength and duration of Smad1/5/8 phosphorylation and ID gene
expression. Human FOBs seeded on collagen scaffolds were subjected for 30, 90 or 120 min to BMP-2 stimulation,
mechanical loading (10% compression) or a combination of both. Loading frequencies of 0.03 Hz, 1 Hz and 10 Hz
were applied to analyze the impact on (A-D) SMAD1/5/8 phosphorylation, determined using western blot analysis
and on (E) ID1 and ID2 expression, determined via qPCR (relative to HPRT expression) (n = 3, *p < 0.05, **p<0.01).
To further examine whether the initial frequency-dependent effects on Smad
phosphorylation and early gene expression persist or equilibrate at a later time point, hFOBs
were continuously stimulated for 24h with BMP-2 and mechanical loading. Since 0.03 Hz was
not expected to cause any changes in comparison to the BMP-treated control, this condition
was excluded from further examinations. Direct BMP-targets, osteogenic markers and genes
Results 71
related to the perception of mechanical forces were included in the evaluation. The heatmap
summarizes the regulation of the tested genes with induction or reduction of expression
labeled in red or blue, respectively (Figure 4-8). The predominantly red-colored heatmap
clearly illustrates the overall anabolic effect of the treatments. The expression analysis indeed
revealed a frequency-dependent increase of transcription even after 24h. For all genes
regulated, the response to 10 Hz loading was stronger in comparison to 1 Hz.
Figure 4-8: Heat map summarizing the gene expression changes in response to 24h BMP-2 stimulation and/or
mechanical loading of 1 Hz or 10 Hz. Induction and reduction of expression is labeled in red and blue respectively,
while white indicates no change in comparison to the untreated control (c), n=4. The values of the fold changes and
log2(F.I.), on which the heat map is based, are depicted in Table 0-1 in the supplement.
Mechanical stimulation with 10 Hz further enhanced significantly the BMP-induced
expression of both, positive (ID1, ID2) and negative regulators (Noggin, Smad7) of the BMP
pathway, while 1 Hz had only minor effects (Figure 4-9). Interestingly, the expressions of BMP
receptor type 1B but not type 1A or type 2 were found to be significantly increased by
mechanical loading in a frequency-dependent manner, whereas BMP-treatment alone had no
effect. The expressions of the osteogenic marker genes RUNX2 and COL1A2 were unaffected
by the treatment, but osteopontin (SPP1) was significantly up-regulated in response to 10 Hz
loading. C-fos, a transcription factor known to be a target of mechanotransduction [181], was
used as a positive control for mechanical loading. As expected, the expression of c-fos
increased with increasing frequency.
72 Results
Moreover, the expression of specific integrin subtypes was analyzed, which were
selected according to the previously determined integrin expression profile of hFOBs
(obtained in personal communication with Dr. Maria Reichenbach, Knaus lab, FU Berlin).
Integrin αv, β1 and β3 expressions were increased by mechanical loading, while α1, α5 and
β5 were not affected. Mechanical loading especially induced the integrin β3 expression, which
was further promoted under concurrent BMP-2 treatment, even though BMP-treatment alone
had no impact. Interestingly, the heterodimer of integrin αvβ3 binds to RGD-containing ECM
proteins, like osteopontin, which were all mechano-sensitive
Figure 4-9: Mechanical stimuli regulate gene expression in a frequency dependent manner. hFOBs were seeded in
collagen scaffolds, transferred into the bioreactor and stimulated for 24h with BMP-2 and/or mechanical loading (1
Hz or 10 Hz). Thereafter, cells were lysed and gene expression was analyzed via qPCR (n≥3).
In summary, frequency-dependent effects on early Smad phosphorylation persisted and
transduced to the level of BMP target gene expression. The results revealed that mechanical
loading with 10 Hz significantly increases the crosstalk duration in comparison to 1 Hz.
4.2.3 Focal adhesion number and size is increased by both, BMP-2 and
mechanical loading in a frequency-dependent manner
To investigate whether the increased integrin expression was transduced into an increased
assembly of focal adhesions (FAs), hFOBs were stained for the focal adhesion marker
phospho-Paxillin (pPax) after a 24 hours treatment with BMP-2 and/or mechanical loading
(1 Hz or 10 Hz). The confocal microscopy images and the corresponding quantifications show
a strong influence of the treatments on cellular attachment to collagen walls (Figure 4-10 A-
C). While untreated cells assembled little and small FA complexes, treatment with BMP-2,
mechanical loading and a combination of both increased the amount of FAs significantly. The
total number of FAs in cells treated with BMP-2 and 1 Hz loading was comparable and not
Results 73
increased by a combination of both treatments. However, mechanical stimulation with 10 Hz
under concurrent BMP-2 treatment, further increased the BMP-2- and load-only effect
significantly (Figure 4-10 B). Furthermore, the percentage of cells with FAs larger than 0.7
µm² increased about 1.5 fold under a combined treatment of BMP-2 and 1 Hz loading and
about 2 fold under BMP-2 and 10 Hz loading.
Interestingly, BMP-2 and 1 Hz mechanical stimulation induced equal FA number and size
distributions, even though BMP-2 in contrast to 1 Hz loading did not enhance integrin
expressions. Therefore, it is suggested that BMP-2 treatment mainly promoted integrin
clustering. Both, the strong increase in integrin expression under 10 Hz loading and the
increased integrin clustering under BMP-2 consequently led to the synergistic increase of FA
size and amount under concurrent stimulation.
Figure 4-10: Focal adhesion number and size is increased by BMP-2 treatment and by mechanical loading in a
frequency-dependent manner. hFOBs were seeded in collagen scaffolds, transferred into the bioreactor and
stimulated for 24h with BMP-2 and/or mechanical loading (1 Hz or 10 Hz). Cells were fixated and stained for phospho-
Paxillin (green), F-actin using phalloidin (pink) and nuclei using DAPI (blue). (A) Representative confocal images of
stained hFOBs. Scale bar represents 50 µm. (B) Amount of phospho-Paxillin positive FA per cell and (C) percentage of
cells with FAs of different size classes was assessed in ImageJ (see paragraph 3.4.1) (in total >110 cells pre condition,
n=3). (D) Schematic drawing illustrates the increase in FA size and amount due to BMP-2 and mechanical stimulation.
The observed increase in integrin expression and FA assembly together with the
increased expression of BMP receptor type IB under mechanical stimulation without the
74 Results
application of BMP-2, was especially interesting since both, an increased amount of BMP
receptors and the described BMP receptor-integrin interaction (see paragraph 1.5.1) would
promote BMP signaling. A 24 hours stimulation with cyclic compression, could therefore
induced gene expression changes, which could enhance BMP signaling once cells are exposed
to BMP-2.
Consequently, two further hypothesis arose:
(1) Long-term mechanical loading sensitize the cells for BMP-2 so that the crosstalk between
BMP signaling and mechanotransduction induced by subsequent concurrent mechanical
and BMP-2 stimulation would be even further increased.
(2) Long-term mechanical loading leads to the establishment of a “mechano-memory”, which
would be sufficient to mediate the crosstalk, even though mechanical and BMP-2
stimulation are not concurrently applied.
4.2.4 Cells develop a mechano-memory impinging on BMP signaling
To examine whether mechanical pre-stimulation even further promotes the crosstalk
between BMP signaling and mechanotransduction in comparison to no pre-stimulation
(hypothesis (1)), cells were continuously stimulated for 24h before and during BMP-2
treatment. To examine whether cells establish a “mechano-memory”, which would mediate
the crosstalk (hypothesis (2)), cells were mechanically stimulated prior to but not during
BMP-2 treatment. The pre-load conditions were compared to 90 min concurrent BMP-2 and
mechanical stimulation with a frequency of 1 Hz (= crosstalk control), as this induced the
maximum crosstalk strength (see Figure 4-7).
Results 75
Figure 4-11: Mechanical pre-stimulation induced a crosstalk on p-Smad and on gene expression level. hFOBs were
seeded in collagen scaffolds, transferred into the bioreactor and stimulated with mechanical loading for 24h.
Thereafter, cells were stimulated with BMP-2 (5nM) for 90 min with or without concurrent mechanical loading. Cells
were lysed and (A) p-Smad1/5 levels were determined by western blot and (B) gene expression was analyzed via qPCR
(n≥3).
Gene expression analysis confirmed the previously observed significant increase in BMP
receptor type 1B, integrin αv and β3 expression by 24h mechanical loading (Figure 4-11 B).
However, in contrast to hypothesis (1), this adaptation to mechanical stimulation did not
further promote Smad phosphorylation under BMP-2 treatment in comparison to the
crosstalk control (Figure 4-11 A). It could be assumed, that differences could not be observed
because Smad phosphorylation was already in saturation. Therefore, it would be interesting
to investigate earlier time points in potential follow-up studies.
Most strikingly, when cells were mechanically stimulated prior to but not during BMP-2
treatment, Smad phosphorylation was equally increased as in the crosstalk-control. This
demonstrated that mechanical pre-stimulation was able to induce a mechano- memory,
which was sufficient to promote Smad phosphorylation even if the direct mechanical trigger
was missing. Consequently, this experiment suggests that the mechano-regulation of BMP
signaling at the Smad-level does not exclusively depend on concurrent BMP and mechanical
stimulation.
Even though Smad phosphorylation levels were equal for all three crosstalk-conditions,
the expression of BMP target genes ID1 and ID2 were reduced by mechanical pre-stimulation
in comparison to the crosstalk-control (Figure 4-11B). This implies, that intracellular negative
regulators of the BMP pathway, which might interfere with the translocation of Smads into
the nucleus or suppress the binding of Smads at the promoter region, have been activated by
mechanical stimulation decreasing BMP target gene expression.
In summary, cellular adaptation processes in response to 24 hours mechanical pre-
stimulation persisted and were able to promote BMP signaling with different efficiencies on
different signaling levels.
It is assumed that during 24 hours cyclic compression, cells have fully adapted to the
changed mechanical environment, meaning non- transcriptional such as adhesion and
cytoskeletal adaptations but also transcriptional responses including some negative control
mechanisms have been initiated until a new mechanical equilibrium was established
(assumption based on [182]). However, what happens if the time of pre-stimulation is
reduced so that cells not jet fully adapted on all levels to the change in mechanics? By
investigating this, the contribution of load-induced gene expression regulation (slow
processes) in comparison to adhesion and cytoskeletal remodeling (fast processes) for the
establishment of the mechano-memory can be estimated.
76 Results
Therefore, the pre-stimulation time was reduced to 90 and 30 min prior to 90 min BMP
stimulation and phosphorylation levels of Smad1/5 were again analyzed (Figure 4-12A).
Figure 4-12: Only prolonged mechanical pre-stimulation induced a crosstalk on p-Smad level. (A) hFOBs were seeded
in collagen scaffolds, transferred into the bioreactor and stimulated with mechanical loading for 90 min or 30 min.
Thereafter, cells were stimulated with BMP-2 (5nM) for 90 min with or without concurrent mechanical loading. Cells
were lysed and p-Smad1/5 levels were determined by western blot (n=3). (B) Summary of pre-loading conditions and
their efficiencies (in %) to potentiate Smad1/5 phosphorylation (=crosstalk). The maximum crosstalk strength (90
min concurrent mechanical loading (1Hz) and BMP-2 stimulation) was set to 100%, while BMP-2-only treatment was
set to 0%.
Interestingly, 90 min pre-loading could already induce an increased Smad1/5
phosphorylation in comparison to BMP-only control. Compared to the crosstalk control,
however, phosphorylation levels were reduced. In fact, this condition reached around 50% of
the crosstalk strength induced by the crosstalk-control (f = 1Hz, t = 90 min), if the BMP-only
control is set to 0% (Figure 4-12B). Short pre-loading of 30 min had no effect on the BMP-
induced Smad phosphorylation and was therefore insufficient to trigger a crosstalk.
Figure 4-12B summarizes the pre-loading conditions and their efficiency to induce an
enhanced Smad phosphorylation. The calculated percentages of crosstalk strength show that
the shorter the time of pre-loading, the weaker the crosstalk.
4.2.5 Mechanical signals regulate the BMP-pathway via integrins
Stimulation with BMP-2 and mechanical loading increased the expression of BMP receptor
type B1, integrin αv and β3 (Figure 4-11) as well as the size and amount focal adhesions
(Figure 4-10). Taking into account the described interaction of different integrin and BMP
receptor subtypes (see paragraph 1.5.1), it was hypothesized that mechanical stimulation
would increase BMP-2-induced Smad phosphorylation through interactions between BMPRs
and integrins. Due to the specific regulation of integrin αv and β3 expression, the role of αvβ3
Results 77
integrins for the crosstalk was investigated via siRNA mediated integrin αv knockdown. Since
αv integrin is the only relevant interaction partner for integrin β3 in hFOBs (the fibrinogen
receptor αIIbβ3 is mainly relevant in platelets [183]) a knockdown of integrin αv in turn
reduces the amount of active integrin β3. Therefore, both integrin αv and β3 are no longer
available as interaction partners for the BMP receptors.
Scaffold seeding and lipofectamine-mediated siRNA transfection was performed at the
same time, since transfection of already seeded scaffolds was less efficient. A nonspecific
siRNA (scrambled, scr) was used as a negative control in all RNAi-experiments. Knockdown
efficiencies were validated via qPCR and western blot analysis two days after seeding and
transfection. Integrin αv mRNA expression and protein amount were reduced respectively by
about 90% and 60% in comparison to the scr control (Figure 4-13 A and B).
Immunofluorescence staining of integrin αv after transfection (cells seeded on 2D chamber
slides) show the absence of integrin αv-positive focal adhesion complexes in most of the cells.
Interestingly, cells were less spread and established a more spindle-like morphology in
comparison to the scr control (Figure 4-13 C).
Figure 4-13 Validation of integrin αv knockdown in hFOBs. Human FOBs were transfected with siRNAs targeting
integrin αv or a non-targeting control (scr) (30nM) using lipofectamin and simultaneously seeded either into
collagen scaffolds or on 2D chamber slides. Two days after seeding, knockdown efficiencies were validated via qPCR
(A) and western blot analysis (B) from cells grown in the scaffold. Representative images show hFOBs on chamber
slides stained for integrin αv (green), F-actin (pink) and nuclei (blue). Scale bar represents 50 µm.
After the successful validation of the integrin αv knockdown, its impact on the crosstalk
between BMP signaling and mechanotransduction was investigated. Therefore, bioreactor
experiments were performed with transfected cells under crosstalk-control conditions (1 Hz,
10%, 90 min, 5nM rhBMP-2) and Smad phosphorylation was analyzed.
In the scr control, mechanical loading further increased the BMP-2-induced Smad
phosphorylation significantly by 2-fold. Since the induction strength is comparable to results
of previous experiments using non-transfected hFOBs (see Figure 4-7), the transfection
procedure itself was not affecting the crosstalk. Knockdown of integrin αv, however, reduced
the total integrin αv protein levels consistently and significantly by about 50-60% in
comparison to the scr control. Strikingly, this influenced the sensitivity of BMP signaling to
mechanical stimulation, while basal signaling was unaffected. The Smad phosphorylation
78 Results
upon concurrent BMP-2 and mechanical stimulation was still significantly increased, however
only by 1.36-fold. Therefore, the sensitivity to mechanical stimulation was reduced by about
65% in comparison to the scr control with a p- value of 0.06. It is assumed that an increase in
knockdown efficiency would also increase the effect on the crosstalk. In summary, αvβ
integrins play an important role for the crosstalk between BMP signaling and
mechanotransduction.
Figure 4-14: Integrin αv knockdown reduced the crosstalk on Smad phosphorylation level. Human FOBs were
transfected with siRNAs targeting integrin αv or a non-targeting control (scr) (30nM) using lipofectamin and
simultaneously seeded into collagen scaffolds. Two days after, scaffolds were transferred into the bioreactor, starved
for 3h and subsequently stimulated with 5nM BMP-2 and/or cyclic mechanical compression (1 Hz, 10%) for 90 min.
Protein levels of integrin αv and phosphorylated Smad1/5 were quantified via western blotting. Bar charts depict
relative fold changes in comparison to the unstimulated scr control (n=4).
If a direct interaction of intergins and BMP receptors is assumed to mediate the
integration of mechanical signals into the BMP pathway, is this interaction already existing
under static conditions, or is it only established in response to mechanical stimulation? The
latter scenario would imply that load-induced remodeling/reorganization of intergins and
BMP receptors must have preceded an interaction. In the following, it was analyzed whether
a remodeling process is a prerequisite for the crosstalk.
4.2.6 Actin cytoskeleton remodeling is crucial for the crosstalk
It was found that mechanical stimulation induces the expression of specific integrin subtypes
and the clustering into FAs (Figure 4-10), already indicating a load-induced remodeling
process at the plasma membrane. The rearrangement of FAs upon mechanical stimulation is
mediated through integrin signaling that also induces the remodeling of structurally and
Results 79
mechanically connected actin fibers. A primary integrin effector controlling FA turnover and
lamellipodia formation is focal adhesion kinase (FAK). Upon integrin engagement, FAK is
recruited and activated via autophosphorylation at tyrosine 397. Its phosphorylation level in
relation to the total protein amount was examined in response to cyclic compression in a time
dependent manner (Figure 4-15A). In comparison to the static control, a significantly
increased Y397 phosphorylation was detected after 30 min of mechanical loading that
remained high until the 90 min time point. Interestingly, this time dependent behavior
correlated with the load-induced increase in p-Smad level at 30 and 90 min. The total FAK
levels remained constant during the loading period and were equal to the static control. The
increased FAK activation points towards an enhanced integrin clustering and signaling in
response to cyclic compression. A prominent downstream pathway of FAK activation is the
RhoA/ROCK pathway, which controls the activity of myosin II by phosphorylation of the
myosin light chain (MLC) subunit. Cyclic compression increased the MLC phosphorylation in
a time dependent manner reaching significance after 90 min of stimulation (Figure 4-15B).
Figure 4-15: Cyclic compression induced focal adhesion kinase and myosin light chain activation. Human FOBs
seeded in collagen scaffolds were subjected for 15, 30 or 90 min to cyclic compression (1Hz, 10%). The total protein
level of FAK, its phosphorylation at tyrosine 397 and the phosphorylation of MLC was analyzed by western blot
(n=4).
This indicated a load-induced reinforcement of the actin cytoskeleton that goes along
with an actin remodeling and adaptation process. The myosin-mediated increased tensile
force of the cytoskeleton acts on integrin adhesion sites and fosters the clustering of FA. Since
integrin signaling induces actin cytoskeleton remodeling that in turn cause the remodeling of
integrin-mediated adhesions, there is a mutual interaction of integrins/ FAs and the actin
cytoskeleton. Due to this connection, it was hypothesized that load-induced actin cytoskeletal
remodeling is an important mechanotransduction process to mechanically enhance BMP
signaling, as it in turn regulates integrin remodeling. To investigate the relevance of load-
80 Results
induced actin cytoskeletal remodeling for the crosstalk, cells were treated with the actin
cytoskeleton stabilizer Jasplakinolide (Jas). Given the great diversity of actin modulatory
agents, Jas was chosen specifically for its described inhibitory function on actin filament
depolymerization, stabilizing the network and interfering with its remodeling. Jas, an actin
binding macrocyclic peptide, was originally isolated from marine sponge and its
concentration dependent effect on the actin cytoskeleton were described before [184].
Therefore, hFOBs were treated over three hours (the duration of starvation in bioreactor
experiments) with different Jas concentrations and cell morphology and actin cytoskeleton
organization was analyzed. Representative fluorescent images of phalloidin-stained hFOBs
(in 2D) treated with 0.05 µM or 0.1 µM Jas or equal amounts of DMSO (serving as negative
control) are shown in Figure 4-16A. DMSO treatment did not affect the organization of the F-
actin stress fibers into a dense network. Jas at concentration of 0.05 µM increased the
thickness but reduced the amount of actin stress fibers in most of the cells. Furthermore, actin
aggregates formed in the perinuclear region. Due to the competitive binding of Jas and
phalloidin, phalloidin signal intensity was reduced. The cell morphology was barely affected
by the treatment with 0.05 µM in comparison to the DMSO treated control. However, at a
concentration of 0.1 µM, the actin cytoskeleton completely collapsed into an unorganized
actin mass around the nucleus. In contrast, cells seeded into collagen scaffold were less
sensitive to the inhibitor. Cell morphology was maintained but some actin aggregates were
present at the concentration of 0.1 µM Jas. Therefore, for following 2D experiments 0.05 µM
Jas and for 3D experiments 0.1 µM Jas were selected. These concentrations already show
effects on actin organization without inducing major changes to the cell morphology.
Results 81
Figure 4-16: Effects of Jasplakinolide on actin cytoskeleton integrity and dynamics. (A) Concentration and culture
system dependent effects of Jas. Human FOBs seeded onto collagen coated glass slides or into the collagen scaffold
were treated for 4.5h with 0.05µM Jas, 0.1µM Jas or with the solvent DMSO in starvation medium. Thereafter, cells
were fixed and stained. Representative confocal images show F-actin (green) and nuclei (blue) and collagen scaffold
(white) (scale bar = 50µm). (B) Representative images showing the dynamic protrusion remodeling of GFP-LifeAct
expressing hFOBs during a time frame of three minutes before and after inhibitor treatment. The remodeling
dynamics is visualized by an overlay of all cell outlines colored according to frame number from blue to pink. The
kymographs, taken along the red lines, illustrate the different protrusion dynamics. A zoom into the lamellipodia
region illustrates the fast remodeling. Blue lines indicate the cell border changes between (C) Quantification of the
protrusion/retraction area change before and after Jas treatment (20-45 min after). The protrusion/retraction area
change was normalized to the untreated control (=before treatment), n = 12.
To validate its effects on actin cytoskeleton remodeling processes, time-lapse microscopy
was performed using GFP-LifeAct expressing hFOBs. LifeAct is an F-Actin binding peptide
coupled to GFP, which was reported to be less interfering than a direct coupling of GFP to
actin monomers[160]. Cells seeded on collagen-coated glass slides were recorded every 10
seconds during a time frame of 3 min. Since protrusion dynamics differ from cell to cell,
depending on the current phase of cell cycle and whether a cell is migrating or static,
protrusion dynamics of the same cell before and after Jas treatment were compared. The cell
outlines were marked and the area in between two consecutive outlines was measured to
assess the area change (extension and retraction) per time frame. Area changes during one
minute were summed up and averaged over the three recorded minutes. To better illustrate
82 Results
the protrusion dynamics, all cell outlines colored according to frame number from blue to
pink were overlaid and depicted in Figure 4-16B for a representative cell. Additionally,
kymographs show the time dependent movement of a selected position within the
lamellipodia. The resulting image sequences showed a highly dynamic remodeling of
protrusion sites before Jas supplementation (Figure 4-16B). The kymograph illustrates the
fast extension and retraction of the lamellipodium covering approx. ±5 µm and also visualizes
the time dependent accumulation of actin in this region. Jas treatment drastically decreased
the fast protrusion remodeling observed in control cells. The protrusions almost remained
“frozen”, only slowly sliding forward. Additionally, the change in fluorescence signal intensity
over time visible in the kymograph shows, that in contrast to the control, actin was retracted
from the outer cortex. Quantification of the protrusion dynamics, underlines the impression
of the image sequences: Jas significantly reduced the area change per min. Taken together, Jas
inhibited fast actin remolding processes especially visible in protruding lamellipodia.
The here collected data concerning FA growth and increased FAK and MLC
phosphorylation in response to mechanical loading suggest, that cytoskeletal remodeling
processes are triggered. However, a direct proof of the assumption that the dynamics of actin
remodeling is increased in hFOBs upon exposure to mechanical stimulation was missing.
Furthermore, it was necessary to validate the efficiency of Jas to stabilize the cytoskeleton
even under loading conditions. In a first attempt to visualize and quantify actin remodeling
dynamics in response to mechanics, the bioreactor was modified to allow in situ time-lapse
confocal microscopy. Therefore, a glass window was implemented at the bottom of the
bioreactor chamber and the upper plunger was elongated. The collagen scaffold seeded with
GFP-LifeAct expressing hFOBs was positioned onto the glass, the plunger was adjusted on top
and the bioreactor was placed onto the microscope table. Time-lapse imaging was performed
before and after cyclic compression, as imaging during compression was impossible since the
rate of image acquisition was lower than the stimulation frequency. Due to the increased
imaging volume in 3D the frequency of image acquisition decreased drastically so that, to
investigate the very fast protrusion dynamics, not the whole cell body could be recorded.
Furthermore, position selected before cyclic compression could not be re-imaged afterwards
because of a slight, uncontrollable drift of the scaffold within the bioreactor system. Moreover,
it became obvious that scaffold walls deform inhomogenously under compression, aggravating
the comparison of actin remodeling at different positions within the scaffolds. Due to these
limitations, recorded image sequences could only exemplary show the change in protrusion
dynamics, but no quantifications were possible (see supplementary Figure 0-4).
Although substrate deformation caused by cyclic compression of the scaffold, is regarded
to be the most prominent mechanical trigger, cells also experience fluid flow induced by the
Results 83
cyclic compression of the scaffold in the bioreactor. By using a 2D flow chamber setup (Ibidi)
it was interestingly found that fluid shear stress alone enhances early and late BMP signaling
events (data obtained by Dr. Maria Reichenbach, FU Berlin, Knaus Lab). To simplify imaging
of actin remolding in response to mechanical stimuli, this flow chamber setup was combined
with the confocal microscope (in cooperation with Dr. Maria Reichenbach, FU Berlin, Knaus
Lab). GFP-LifeAct expressing hFOBs were seeded into the flow chambers and image stacks
were recorded every 10 seconds over 3 min before and during fluid flow stimulation at 90
min (=flow) as well as after 30 min Jas treatment.
Figure 4-17: Dynamic actin remodeling induced by fluid shear stress is inhibited by Jasplakinolide. GFP-LifeAct
expressing hFOBs seeded in Ibidi flow chambers were stimulated with fluid flow (5dyn/cm²) and Jas (0.05µM) was
supplemented after 90 min. Fast actin remolding processes were recorded during 3 min before (=static) and during
fluid flow stimulation 90 min (=flow) as well as after 30 min Jas treatment (=flow + Jas). (A) Representative images
show the cell outline change over 3 min. The cell outlines are colored according to frame number from blue to pink.
The kymographs, taken along the red lines, illustrate the different protrusion dynamics. The motility map below shows
the speed of nodes along the cell outline normalized to the maximum speed. Red shades represent expanding regions,
blue shades contracting regions. (B) Quantification of the cell area change per time frame divided by the total cell
area and normalized to the static control (n=15 cells in 3 experiments).
As described before, changes of protrusion dynamics were compared within the same
cell. Under static conditions, cells dynamically extended and retracted the protrusion sites
representatively illustrated in Figure 4-17A by the overlaid cell outlines and the kymograph.
Additionally, the speed of nodes positioned in regular distances along the cell outline was
tracked using Quimp an ImageJ plugin and displayed in motility maps, with red shades
represent expanding regions and blue shades contracting regions (middle vertical panel). The
84 Results
node speed was normalized to the maximum node speed measured in all images sequences
acquired for the cell. The dynamic changes of the cell outline, the increased extension and
retraction visible in the kymograph and the enhanced speed of extension and retraction
depicted in the motility map, clearly showed the increased protrusion dynamic of cells
stimulated with fluid flow. Especially obvious is the increased accumulation of actin in the
lamellipodia region in a fluctuating manner. Furthermore, fluid flow induced the polarization
of the cell in this example. This was however not observed for all the recorded cells. For the
analysis of this particular experiment, the area change per minute was normalized to the cell
area as the cell morphology was changing significantly during the time course of the
experiment. Without normalization, area changes of a large cell in comparison to a small cell
would always be greater. A statistically significant increase in cell area change was quantified
for cells stimulated with fluid flow at both time points. Strikingly, also under continuous fluid
flow, Jas strongly reduced the protrusion dynamics of the cells. In some cases, the lamellipodia
were even completely retracted. As in the previous 2D experiments under static condition
(Figure 4-16), cells appeared “frozen” and a fluctuation of actin accumulation was completely
inhibited. The area change was significantly reduced by the treatment with Jas. Even though
a direct experimental proof is missing, it can be speculated that this is also happening in 3D
during the loading experiments in the bioreactor using 0.1µM Jas.
In summary, it was found that Jas inhibited the load-induced dynamic remodeling of the
actin cytoskeleton, especially visible in the reorganization of protrusion sites. Since the actin
network is linked to integrin–mediated adhesions, a disturbed actin reorganization will in
turn impair remodeling of adhesion sites. However, the load-induced remodeling of adhesion
sites and the actin cytoskeleton, are fundamental first events leading to the establishment of
a new force equilibrium. Therefore, it was proposed that an interference with the load-
induced actin remodeling would disturb mechanotransduction and would moreover hinder a
load-induced interaction between integrins and BMP receptors.
Since Jas was proven to efficiently inhibit actin remodeling dynamics induced by
mechanical stimulation, the agent was regarded a good candidate to test the hypothesis
whether load-induced dynamic remodeling of the actin cytoskeleton mediates the mechano-
regulation of BMP signaling. Therefore, bioreactor experiments with and without Jas were
performed under crosstalk-control condition (t=90 min, f=1Hz, A=10%, 5nM BMP-2), which
is known to increase Smad phosphorylation significantly. Jas or equal amounts of DMSO were
supplemented to the starvation medium, bioreactors were assembled, hFOB-seeded scaffolds
were positioned and starved for 3h. Thereafter, cells were subjected for 90 min to BMP-2
stimulation, mechanical loading (1 Hz, 10%) or a combination of both. Subsequently, cells
Results 85
were harvested for western blotting or qPCR to analyze p-Smad levels or ID1 expression,
respectively (Figure 4-18A B).
Figure 4-18: F-actin stabilization by Jasplakinolide inhibits load-induced Smad phosphorylation and ID1 expression.
Human FOBs seeded in collagen scaffolds were incubated for 3 h in starvation medium supplemented with 0.1 µM
Jasplakinolide (Jaspl). Subsequently, scaffolds were subjected for 90 min to BMP-2 stimulation, mechanical loading (1
Hz, 10%) or a combination of both. (A) The phosphorylation of Smad1/5/8 was analyzed by western blot and (B) ID1
expression was determined via RT-qPCR (n = 3).
In the DMSO treated control samples, cyclic compression significantly increased
Smad1/5/8 phosphorylation under concurrent BMP-2 stimulation in comparison to BMP-2-
only stimulation. Under Jas treatment, however, cyclic compression did not increase p-Smad
levels and in comparison to the crosstalk-control, phosphorylation was significantly
downregulated. This is mirrored in the expression of ID1. Whereas cyclic compression under
concurrent BMP-2 stimulation significantly increased ID1 expression under DMSO treatment,
there was no difference under Jas treatment (Figure 4-18B). Consequently, Jas treatment
abolished the positive effect of cyclic compression almost completely, while basal BMP signaling
remained unaffected. Together with the observed inhibition of actin remodeling, this led to the
conclusion, that actin cytoskeletal adaptation in response to cyclic compression is a
prerequisite for the mechano-regulation of BMP signaling.
86 Results
4.3 The influence of the crosstalk on ECM formation
BMP-2 is known for its strong osteoinductive potential and its influence on cell differentiation
in the context of bone healing is well studied. However, the growth factor should not be
reduced to this feature alone, as it was described to influence cell migration [23], proliferation
[22], angiogenesis [219] and evidences point towards an additional role in steering early
extracellular matrix formation processes [149], [155]. Therefore, the question arose whether
mechanical stimulation would not only increase the growth factors` potential to induce
osteogenic differentiation but also foster its effects on tissue formation.
The influence of cyclic compression and BMP-2 stimulation on extracellular matrix
formation was investigated using human primary fibroblasts known as tissue forming cells.
For these experiments BMP-2 concentration (5nM) and mechanical loading protocol (f=1Hz
and ε=10%, 3h cyclic compression, 5h break) used were identical to previously performed
differentiation experiments described in section 4.1. The selected experimental parameters
were proven to induce a crosstalk on Smad phosphorylation level at 90 min in fibroblasts as
shown in section 4.2.1. Therefore, it was hypothesized that cyclic compression increases the
BMP-2 effect on early tissue formation.
4.3.1 Load- induced tissue contraction and stiffening is reduced by BMP-2
Tissue contraction is regarded as an essential process during tissue healing to re-establish
the lost tissue pretension [185]. Therefore, macroscopic deformation of the scaffold was
investigated after bioreactor culture. Scaffold dimensions were assed in radial and axial
direction, before (0d) and after bioreactor culture (7d), (example images in Figure 4-19 A) to
calculate the scaffold volume contraction (%). A significant increase in volume contraction of
mechanically loaded compared to control samples was observed. The separation into axial
and radial contraction demonstrates, that this is not only due to increased tissue compaction
resulting from the axially applied load (Figure 4-19 C), but also due to active cell-mediated
contraction perpendicular to the loading direction (Figure 4-19 D). BMP-2 stimulation slightly
increased the volume contraction in comparison to the control. However interestingly, in
combination with cyclic compression, BMP-2 slightly reduced the load-induced tissue
contraction.
Mechanical properties of the cell substrate are known to influence cell behavior [56]. To
characterize the effect of cyclic compression and BMP-2 on the mechanical properties of the
early tissue, mono-axial compression tests of native samples were performed to assess the
compressive stiffness. Since the compression tests were conducted in the direction of the
scaffold pores that provide resistance to the newly deposited tensed collagen fibers, it is
expected, that collagen contributes manly by occupying the space in the pores and not by its
Results 87
tensile load-bearing capacity. The measured stiffness, therefore, mostly depends on the cross-
sectional area of the bulk material (a reduction in cross-sectional area due to contraction
results in densification) but also on the amount of deposited ECM.
Indeed, an increased radial contraction correlated with an increased Young’s modulus (E)
here expressed as compressive stiffness (Figure 4-19 E). Both cyclic loading and BMP-2
enhanced tissue stiffness in comparison to the control, reaching statistically significant
difference for the loaded group. Unexpectedly however, under concurrent BMP-2 and
mechanical stimulation, BMP-2 reduced the load-induced scaffold stiffening.
In summary, both tissue contraction and stiffening due to cyclic compression were
reduced by BMP-2 stimulation.
Figure 4-19: Load- induced scaffold contraction and stiffening is reduced by BMP-2. Human fibroblasts seeded in 1.5-
wt% collagen scaffolds were cultured for seven days in the bioreactor under intermitted cyclic compression (f=1Hz,
ε=10%, 3h load, 5h break), rhBMP2 (5nM) or a combination of both. (A) Representative images of fibroblast-seeded
collagen scaffold cultured under control conditions showing radial and axial contraction at day 0 and 7. (B-D)
Quantification of volume, axial and radial contraction [%] of scaffolds after seven days dependent on the culture
condition (n=8). (E) Compressive stiffness E [kPa] of native samples after seven days dependent on the culture
condition (n=5).
Tissue contraction is not only dependent on cell traction forces but also on the amount
of fibrillar collagen. Indeed, a linear dependency between biomaterial contraction and
fibrillar collagen density was shown previously, indicating a mechanical contribution of
tensioned collagen fibrils in the contraction process [171]. In the following, it was
investigated whether the observed differences in scaffold contraction can be correlated to
differences in collagen content. Based on previous findings that BMP-2 and cyclic
compression individually increased collagen I expression [154] and the secretion of
procollagen I C-peptide [159] in fibroblasts, respectively, it was hypothesized that both
stimuli independently but specifically in combination increase contraction and tissue
stiffening due to increased fibrillar collagen formation.
88 Results
4.3.2 Mechanical loading increases collagen synthesis but reduces fibrillar
collagen density and fiber alignment
To verify that fibroblasts increase collagen secretion upon cyclic loading, the concentration of
procollagen I C-peptide (PIP) was measured in the harvested conditioned culture medium
using ELISA (Figure 4-20 C). Collagen is synthesized and secreted in its pro-form that is
cleaved by collagen peptidases to remove the end-termini before it is eventually assembled
into collagen fibrills. The concentration of the soluble C-terminal peptide directly correlates
with the amount of collagen synthesized. Cyclic compression significantly increased the PIP
concentration in the medium. BMP-2 stimulation, however, had no effect on collagen I
synthesis. In agreement with this the PIP concentrations under the BMP-2+load condition
were similar to the load-only group. Consequently, the question arose whether the increased
collagen secretion under cyclic compression would also result in an increased deposition of
collagen fibrils that are known to serve as a structural network for tissue repair processes.
Collagen type I was visualized in the ECM using an antibody staining and confocal
microscopy (Figure 4-20 A). The monoclonal collagen I antibody is directed to the amino acid
sequence at position 1200-1300 of human collagen I, binding to fibrillar and non-fibrillarized
collagen, therefore visualizing the whole proportion of deposited collagen. Interestingly and
in contrast to the PIP measurement, the signal intensity per mm³ quantified from confocal
images was similar in control, loaded and BMP-2 stimulated samples, with a slight reduction
under combined treatment (Figure 4-20 D). The increased secretion was therefore not
leading to an increased deposition of collagen I into the ECM. However, since scaffold
contraction was increased under cyclic compression (Figure 4-19 B-D) and contraction was
previously described to correlate linearly with the amount of fibrillar collagen [171], it was
hypothesized that the proportion of fibrillar vs non-fibrillar collagen would increase under
load.
Results 89
Figure 4-20: Cyclic mechanical compression increases collagen synthesis but reduces fibrillar collagen density. Human
fibroblasts seeded in 1.5-wt% collagen scaffolds were cultured for seven days in the bioreactor under intermitted
cyclic compression (f=1Hz, ε=10%, 3h load, 5h break), rhBMP2 (5nM) or a combination of both. (A) Representative
confocal multiphoton images showing stainings for collagen 1 (green), F-actin (red), nuclei (blue) and (B) fibrillar
collagen visualized by SHG (white) after seven days. Yellow arrows indicate newly deposited collagen fibers within
the collagen walls of the scaffold. Scale bar = 100µm. (C) Quantification of Procollagen type I C-peptide (PIP)
concentration inside culture medium using ELISA. Quantification of (D) collagen 1 density (antibody staining), (E)
fibrillary collagen density and (F) cell density from confocal images.
To investigate this, second harmonic imaging (SHI), a label-free method to visualize
fibrillar collagen was used. Under control conditions and BMP-2 treatment, thick bundles of
newly deposited fibrillar collagen were visible in scaffold pores (Figure 4-20 B). Cyclic
compression, however, reduced the second harmonic signal intensity drastically. Even in
combination with BMP-2, the inhibitory effect of cyclic compression was dominating.
Quantification of the fibrillar collagen density inside collagen pores confirmed the visual
impression; cyclic compression significantly reduced the density of SHI-visualizable collagen
bundles even in combination with BMP-2 (Figure 4-20 E). This is not related to the cell
number, as the cell density in loaded samples even increased due to scaffold contraction
(Figure 4-20 F). Frist investigations on further tissue maturation at the 2 week time point
indicate, that fibrillar collagen density remains decreased under cyclic compression, even
though BMP-2 stimulation seems have a small rescuing effect (supplementary Figure 0-5).
90 Results
The observation that tissue contraction increased while at the same time fibrillar
collagen density was reduced stands in contrast to the previously observed linear correlation
between macroscopic contraction and fibrillar collagen depositions by fibroblasts from
different donors, which strongly differed in their ability to deposit fibrillar collagen [171]. To
better compare the current finding with previous results obtained by Brauer et al. [171], the
tissue contraction was plotted in relation to the fibrillar collagen density combining the data
obtained here with the data obtained by Brauer et al. (2019) (Figure 4-21A). In comparison
to the linear dependency between tissue contraction and fibrillar collagen density observed
by Brauer et al. (2019), it becomes obvious, that cyclic compression drastically changed the
described interdependency as these conditions strongly deviate from the linear fit. Control
and BMP-2 treated samples, however, nicely fit into this linear correlation, with the BMP-2
treated samples slightly shifted upwards along the line.
Against the assumption, that cyclic compression would increase the proportion of
fibrillar vs non-fibrillar collagen, the opposite was the case. This becomes even clearer when
calculating the ratio between Col1 density (determined by antibody staining, showing both
fibrillar and non-fibrillar collagen) and fibrillar collagen density (Figure 4-21B). This ratio
shows the proportion of non-fibrillar collagen in the samples and demonstrates that the
proportion of non-fibrillar collagen strongly increased under cyclic compression in
comparison to control and BMP-2 treated samples.
The ratio between PIP concentration and fibrillar collagen density is 3-fold higher in
samples treated with cyclic compression in comparison to the control or BMP-2 treated
samples. This demonstrates that under cyclic compression much more collagen I has been
secreted than assembled into collagen bundles (Figure 4-21C).
Together this indicates that under cyclic compression the amount of collagen is reduced
with progression of collagen maturation; secretion (PIP), embedding (Col1
immunofluorescent staining), fiber assembly (SHI of collagen fibrils).
Figure 4-21: Cyclic compression changes the dependency of collagen density and tissue contraction. (A) Correlation
of fibrillar collagen density and scaffold contraction. Data for linear correlation (black dots and dotted line) based on
Results 91
7 fibroblast donors was obtained and kindly provide by Erik Brauer [171]. (B) Ratio of collagen 1 (staining) and
fibrillar collagen density. (C) Ratio of PIP and fibrillar collagen density.
Even though almost equal amounts of Col1 were deposited into the ECM under all conditions
(Fig. 4-20D), the assembly into collagen fibers, which can be detected using SHI, was different.
A closer look onto the structure of collagen fibers within the pore reveals an altered
organization under cyclic compression (Figure 4-22A). In control and BMP-2 treated samples,
fibers show a uniform alignment along the scaffold pores, while under cyclic compression a
more unorganized meshwork of fibers was established. When quantifying the collagen fiber
orientation distribution from both SHI and Col1 staining under the different culture
conditions (Figure 4-22B), it becomes clear that the anisotropic alignment adopted under
control and BMP-2 conditions is strongly reduced towards a more isotropic orientation upon
cyclic compression. As the ECM alignment follows the orientation of the cell, it was not
surprising that F-actin and collagen fibers orientation are similar. Also on the level of ECM
organization, the effect of cyclic compression dominated over the BMP-2 effect.
92 Results
Figure 4-22: Cyclic mechanical compression reduces fiber and cell alignment. (A) Representative confocal
multiphoton images showing collagen 1 (green) visualized by an antibody staining, fibrillar collagen (white)
visualized by SHG and F-actin (red) visualized by phalloidin staining. Scale bar = 50µm (B) Comparison of fiber
orientation distribution (percent of total) of fibrillary collagen, collagen 1 and F-actin signal relative to local pore
orientation. Polar diagrams show mean value as solid line and standard deviation as color/gray band.
4.3.3 BMP-2 and cyclic compression induce distinct gene expression changes
To further investigate the regulation of collagens and collagen modulating proteins by BMP-
2 and cyclic compression, gene expression analysis were performed. Gene expression changes
in fibroblasts cultured for seven days in the bioreactor under intermitted cyclic compression
were assessed using qPCR. Specifically, the expression of important fibrillar collagens, soft
ECM proteins and enzymes involves in collagen fibrillogenesis and remodeling were
investigated and are summarized in the heat map (Figure 4-23). These candidates have been
selected due to their abundance in MS analysis (described in the paragraph below) and
because of their regulatory capacity on collagen metabolism. At the first glance it can be seen,
that cyclic compression increased the expression of most of the genes investigated, while a
more diverse regulation was observed under BMP-2 stimulation. In a combination, the impact
of mechanical stimulation on gene regulation seemed dominant, although BMP-2 treatment
dampened the strong load-effect. This could lead to the suggestion, that the ECM established
under cyclic compression is subjected to a higher turnover than under control conditions.
Figure 4-23: Gene expression analysis of selected ECM proteins and ECM modulators. Human fibroblasts seeded in 1.5-
wt% collagen scaffolds were cultured for seven days in the bioreactor under intermitted cyclic compression (f=1Hz,
ε=10%, 3h load, 5h break), rhBMP2 (5nM) or a combination of both. mRNA expression was determined via qPCR
(expressions relative to HPRT). Heat map shows the log2 of the fold change towards the untreated control (n=3). The
values of the fold changes and log2(F.I.), on which the heat map is based, are depicted in Table 0-2 in the supplement.
In detail, cyclic compression induced a strong 2-fold up-regulation of fibulin (±0.3), elastin
(±0.4), TGFβ-induced protein (±0.3, TGFBI) and 1.5-fold on bone morphogenetic protein type-
1 (±0.4, BMP-1) and matrix metalloproteinase-13 (±0.2, MMP13) expression. In contrary to
what the name suggests, BMP-1 is a matrix metalloproteinase that cleaves the C-terminal
Results 93
propeptide of secreted tropocollagen. The cleaved C-peptide was quantified in the medium
(Figure 4-20C) and found to be elevated under cyclic compression, which would be in
agreement to the increased BMP-1 expression. However, even though the increased PIP
concentration in the medium would also suggest an enhanced collagen I synthesis, the
expression of COL1A2 after one week is only slightly increased by cyclic compression. It might
be suggested that an increased collagen I expression at an early time point caused an
increased secretion and an enrichment of soluble tropocollagen I. In contrast to the reduction
in fibrillar collagen density in response to cyclic compression, lysyl oxidase (LOX) and lysyl
oxidase like protein (LOXL1) were slightly increased. Since these enzymes mediate the
covalent cross-linking of staggered collagen fibrils, an increase in their expression should in
turn increase fibrillar collagen density, which stands in contrast to the histological
observation (Figure 4-20). This discrepancy might be explained by an increased expression
of the collagenase MMP13, suggesting increased collagen degradation. The increase in
expression of TGFβ-induced protein, could point towards elevated levels of TGFβ secreted by
fibroblasts under compression, since its expression is, as the name suggests, induced by the
growth factor. The proteins specific function, however is still a matter of debate [186].
BMP-2 stimulation alone induced a strong 2.25-fold up-regulation of MMP1 (±0.4), a 1.5-
fold increase in fibulin-1 expression and a down regulation of elastin and COL1A2 expression
with a fold change of 0.6(±0.2) and 0.8(±0.3), respectively. Especially the regulation of elastin
is distinctly different under cyclic compression or BMP-2 stimulation. Elastin is a fibrillar
protein, which provides elasticity to tissues [187]. In this regard, it would be especially
interesting to investigate if the changed elastin levels would change the tissues´ stress
relaxation behavior, a material property which was previously shown to influence cell
function [188]. Although, MMP1 expressions was elevated and COL1A2 expression was
decreased under BMP-2 treatment in comparison to the control, previous histological
quantifications of collagen 1 and fibrillar collagen density (Figure 4-20) do not indicate an
increased collagen degradation. In the follow-up investigations, the activity of MMPs should
be assessed, as MMP gene expression and enzyme activity might differ.
Under concurrent BMP-2 and mechanical stimulation, gene regulations are not as
pronounced as under their individual influence but similarities can be observed. As in
response to mechanical loading alone, a combination of both treatments increased the
expression of TGFβ-induced protein in comparison to the control, however with reduced
strength as in comparison to mechanical treatment alone (L=2-fold(±0.3) vs B/L=1.5-
fold(±0.2)). As under BMP-2 treatment alone, a combination of both treatments increased the
MMP1 expression by 1.5-fold (±0.2) and decreased the COL1A2 expression by 0.8-fold (±0.03)
in comparison to the control. The regulation of MMP1 might contribute to the decreased
94 Results
fibrillar collagen density observed in the ECM. The decreased COL1A2 expression is, however,
in disagreement with the increased PIP concentration detected in the culture medium. In
general, as qPCR analysis only represent the transcriptional activity of the cell at the time of
lysis, it might be necessary to investigate the gene expression in a time dependent manner.
In summary, the gene expression analysis of selected ECM proteins and ECM modulators
show that each condition; cyclic compression, BMP-2 or a combination of both, trigger
distinctly different gene expression patterns, which suggest the formation of distinct,
biochemically different ECMs. The strength of gene regulation under concurrent BMP-2 and
mechanical stimulation was overall reduced in comparison to the individual treatments. This
is pointing towards a mutual balancing of effects that might be of importance for regenerative
processes.
To investigate whether gene expression regulations reflect the ECM protein composition,
in the following mass spectrometry (MS) analysis were conducted. Since here the effect of
cyclic loading on ECM formation particularly collagen formation was most intriguing, first MS
analysis were focused on a comparison between static and dynamic culture.
4.3.4 ECM protein composition is specifically altered by mechanical loading
To investigate the composition of the ECM established under static conditions and under
cyclic compression (both cultured in the bioreactor) via mass spectrometry, samples had to
be decellularization, as the amount of cellular proteins would greatly overlay ECM proteins.
The detergent-based decellularization approach using a perfusion system was previously
established [171] and the protocol was adapted with the aim to preserve many ECM proteins,
while removing most of the cellular components.
Figure 4-24A, represents a summary of all proteins detected in the matrices of both control
and mechanically stimulated samples after decellularization. For a better overview, proteins
were grouped into extracellular (blue) and cellular (gray) compartments. To increase the
accuracy and reliability of the analysis, here only those proteins are depicted, which were
identified by two peptides.
Results 95
Figure 4-24: Cyclic compression induced distinct changes in the protein composition of the ECM. Human dF seeded in
collagen scaffolds were cultured for 2 weeks in the bioreactor with (L) or without (c) 10% cyclic compression (f=1Hz,
cycles of 3h stimulation and 5h break). Samples were decellularized, freeze dried and mass spectrometry analysis was
performed. (A) Protein networks of all proteins detected in the two conditions grouped into extracellular (blue) and
cytosolic (grey) proteins. (B) Network of proteins with significantly changed abundance between the conditions
grouped into extracellular (blue) and cytosolic (grey) proteins. (C) Heat map of the log2 ratios of the abundance of
each sample relative to the average abundance (only ECM proteins) (n=4).
The MS analysis securely identified 15 ECM proteins and 53 cellular proteins in all
samples. Among those ECM proteins detected, five were significantly regulated with a fold
change of ≥2 (Figure 4-24B). The abundance of elastin was increased, while fibulin-1,
periostin, TGFβ-induced protein and tenascin were very consistently reduced under cyclic
compression (Figure 4-24C). Interestingly, the regulation of elastin by cyclic compression is
in agreement with the gene expression analysis (Figure 4-23). However, protein regulations
of fibulin-1, periostin, TGFβ-induced protein and tenascin are in disagreement with the gene
expression data. Although the expressions of fibulin-1, periostin, TGFβ-induced protein and
tenascin were increased, their amount in the ECM was significantly reduced. This again
highlights the importance to investigate all levels of the protein synthesis to gain a
96 Results
comprehensive understanding. In agreement with the quantification of collagen 1 density
(antibody staining against COL1A1), also MS analysis did not detect a change in the abundance
of COL1A1. This observation again points towards a decelerated collagen fibrillogenesis
under cyclic compression. In this context, it is interesting to mention that periostin is known
to be involved in collagen-crosslinking (for details see section 5.7 of the discussion).
Therefore, the MS analysis helps to understand why collagen fibrillogenesis might be
disturbed under mechanical loading
In summary, the results obtained by MS analysis revealed distinct differences in the
biochemical composition of extracellular matrices grown under static conditions or under
cyclic compression. Since integrins specifically recognize ECM proteins, it is assumed that
alterations in the biochemical ECM signature alter integrin adhesion and expression patterns
and thereby potentially also BMP signaling.
Discussion 97
5 Discussion
5.1 Mimicking mechanical loading conditions during the early
phase of bone healing
To increase the physiological relevance of this work, special emphasis was laid onto the
experimental design. The early healing phase is especially sensitive to mechanical signals and
is believed to lay the ground for the entire repair process. Indeed, allowing limited
interfragmentary movement in the early phase was shown to enhance fracture healing in a
sheep model [43]. Axial interfragmentary movement was reported to be the main loading
regime in animal osteotomy models with an external fixator [189]. Interfragmentary
compression that occurs as a consequence of weight bearing in experimental bone healing
studies was reported to range between 10 - 33% [37], [43] and 2 - 20% [40], [190] of the
fracture gap for sheep and rat osteotomies, respectively. According to the reported in vivo
data and in agreement with previous studies investigating the impact of mechanical loading
on cell differentiation [91], [92], [94], [191], strain regimes of 5% and 10% were selected. To
mimic the load pattern during human locomotion [192], a sinusoidal compression with a
frequency of f= 1 Hz was applied in this study using the mechano-bioreactor. In summary, the
loading parameters used in this study were chosen to represent regimes occurring in vivo,
more specifically, to mimic the mechanical environment in the early fracture gap.
The biomaterial used in this study was made from fibrillar collagen, a component of the
extracellular matrix that is a relevant cell substrate throughout the bone healing process [11].
The macroporous architecture assures a three-dimensional cell morphology and provides
enhanced nutrient and oxygen supply for the cells even in the center for the scaffold. Collagen
crosslinking during production protects the scaffolds from fast enzymatic degradation. Both,
collagen crosslinking and its elastic deformation behavior, allowed repetitive compression
without major shape-changes even over long periods of time [11], [171]. By changing the solid
content, the wall stiffness was tuned without affecting the pore architecture (Figure 4-1). As
the stiffness of the substrate that cells adhere to is known to be an important regulator
influencing cellular behavior [56], scaffolds with bulk stiffnesses of 3.4 kPa and 12.3 kPa were
to investigate load-induced osteogenic differentiation. The utilized scaffolds are
characterized by low elastic moduli mimicking the physical environment in the fracture gap
early after bone injury where a soft tissue matrix is present within the fracture gap.
98 Discussion
5.2 Towards a deeper understanding how cyclic compression
influences osteogenic differentiation
Multiple studies examined the influence of mechanical loading on stem cell differentiation,
including osteogenic commitment (see section 1.3.3 and reviews [55], [170]). Motivated by a
tissue engineering approach, the majority of these studies used osteoinductive medium
supplements, bone derived scaffolds or hydrogels with limited supply masking effects of
loading on cell fate decision. Even in studies working without additional osteogenic triggers,
the influence of load-induced autocrine signaling was not investigated. Therefore, it remained
unclear whether the observed mechano-sensitivity is a direct consequence of cyclic
compression, an indirect effect of altered supply or a specific modulation of autocrine BMP
signaling. In this dissertation, the direct influence of cyclic mechanical loading on the
osteogenic differentiation of primary human bone marrow MSCs (hBMSCs) was investigated
and dissected from the effect of load-induced supply changes and autocrine signaling, in
particular of BMP-2. To investigate osteogenic differentiation, the expressions of common
osteogenic markers were analysed.
Runx2 (Cbfa 1), a member of the RUNT domain gene family, is an indispensable
transcription factor for osteoblast differentiation [193]. In this study the expression of RUNX2
in hBMSCs was found to be downregulated upon cyclic compressive loading (Figure 4-2). This
observation stands in contrast to previous reports in which comparable experimental setups
were used [91], [92], [94], [95]. It is known that Runx2 regulates the expression of osteoblast
specific genes such as collagen type 1, bone sialoprotein, osteocalcin and RUNX2 itself [194].
Thus, the downregulation of COL1A2, which encodes the pro-alpha2 chain of type I collagen
and osteocalcin observed in this study is most likely a consequence of the downregulation of
RUNX2. Only one study was identified that reported an inhibitory effect of mechanical
stimulation on the RUNX2 expression and, consequently, on other osteogenic markers. In this
study, continuous application of mechanical loading over up to 10 days, might be responsible
for the negative impact [195]. However, this can be excluded as an explanation in our study,
since here only intermittent loading (repeated cycles of 3h cyclic compression and 5h break)
was applied.
In contrast to RUNX2, COL1A2 and osteocalcin, a significant upregulation was found for
osteopontin mRNA expression in response to 10% compression in the softer scaffold A (E=3.4
kPa). A possible explanation for this discrepancy is that osteopontin expression is regulated
by an alternative mechanism independent of RUNX2. The expression of osteopontin was
previously found to be sensitive to mechanical stimulations [91], [196]. Osteopontin is an
abundant non- collagenous protein in the extracellular matrix of bones and serves as a cell
Discussion 99
attachment point mediated through integrin binding [197]. It is conceivable that hBMSCs
establish stronger attachments to their substrate in response to cyclic deformation of the
walls by increased osteopontin secretion.
In summary, aside from osteopontin, all investigated osteogenic marker genes were
found to be downregulated in response to cyclic compression. This surprising finding
contradicts the majority of literature on this topic. A reason for the discrepancy may be found
in the specific experimental conditions. In comparison to others, we can exclude previously
reported indirect effects of mechanical loading on oxygen concentration and nutrient supply
inside the biomaterial due to the chosen macroporous architecture [173]. A further major
difference is the low total cell number in relation to the volume of cell culture medium in the
bioreactor (1.5x105 cells/ 27ml medium) compared to the cultivation of 3D cell seeded
constructs in well plates with low medium volume [91], [94], [95]. This in combination with
the small but decisive fluid flow in the bioreactor, lead to a strong dilution of signaling
proteins secreted by the cells (here demonstrated for BMP-2, Figure 4-4) and hinder
autocrine biochemical self-stimulation.
5.3 Cyclic compression possess an osteoinductive potential only in
a BMP-enriched environment
As an indicator for an osteogenic response to cyclic compressive loading, and in contrast to
the osteogenic genes mentioned above, BMP-2 expression and secretion were found to be
enhanced. This is in line with previous studies, reporting about a mechanosensitive BMP-2
expression [97], [198]. The growth factor BMP-2 is known to be an indispensable player
during bone repair [28] and it’s in vivo administration leads to bone defect healing [199]. In
fact, BMP signaling regulates the transcription of RUNX2 through the activation of the Smad
transcription factors [176]. Based on this connection, we hypothesized that cyclic
compression does not induce osteogenic differentiation per se but only in presence of BMPs.
The addition of rhBMP-2 in combination with cyclic compression caused a strong
enhancement of RUNX2 and BMP-2 gene expression. This observation is remarkable, since in
the absence of rhBMP-2, cyclic compression suppressed the transcription of RUNX2 mRNA.
Hence, it indicates that BMP-2 is capable to alter the cell’s gene expression response to cyclic
compression. Strikingly, after increasing the cell-to-medium ratio in bioreactor experiments
the fold change expression of RUNX2 was significantly increased in response to 10% cyclic
compression even in the absence of an additional rhBMP-2 stimulus. The changed culture
conditions enabled a significant enrichment of BMP-2 in the culture medium, confirmed by
ELISA quantification (Figure 4-4B). Cyclic compression further increased BMP-2 expression
and secretion pointing to an autocrine BMP-mediated increase of RUNX2 expression.
100 Discussion
Treatment with rhNoggin verified the importance of cell-secreted BMP-2 in inducing RUNX2
expression under cyclic compression (Figure 4-5). That rhNoggin treatment did not fully
recapitulate Rlow conditions under which RUNX2 expression was downregulated, could have
two reasons, either the concentration of rhNoggin was not sufficient to fully inhibit BMP-2
signaling, or a BMP-independent mechanism is additionally acting under Rhigh conditions. A
BMP-2-mediated induction of osteogenic differentiation under cyclic compressive load was
previously postulated [96], [97]. Rui et al. (2011) correlated an upregulation of BMP2
expression under loading conditions to an increased expression of either RUNX2 or ALP in rat
tendon derived stem cells. However, the hypothesis was not proven by the exclusion of BMP-
2 from their systems as it was done here by the dilution effect (in Rlow condition) and the
addition of rhNoggin (in Rhigh condition). Wang et al. (2010) instead showed in a 2D setting
using the MC3T3-E1 cell line, that Noggin treatment abolished ALP expression induced by
mechanical loading (four-point bending device), thereby demonstrating the role of BMP in
this context. Our findings highlight the physiological relevance of load-induced effects for cell
differentiation processes in a 3D bone healing context, since primary human bone marrow-
derived MSCs from multiple donors were utilized.
Figure 5-1: Cyclic compression promotes
osteogenic differentiation via BMP. In this
study, mechanical stimulation alone did not
promote osteogenic differentiation but
enhanced the expression of BMP-2. It is
proposed that mechanical loading is able to
promote osteogenic differentiation via
autocrine signaling through BMP-2.
It was found that the increase of BMP expression in response to cyclic compression
contributes to a positive feed-back loop enhancing osteogenesis (see Figure 5-1). In addition,
cyclic compression not only increases the expression of but also the sensitivity for BMP-2. It
was shown in human fetal osteoblasts [125] and in hBMSCs (Figure 4-6) that cyclic
compression enhances BMP-2-signaling events under concurrent BMP-2 stimulation. These
two mechano-regulated processes – an increased expression of BMP-2 and an increased
sensitivity for BMP-2 – seem to contribute jointly to the osteogenic commitment of hBMSCs
under cyclic mechanical compression.
5.4 Cyclic compression integrates into the BMP signaling pathway
only in a ligand dependent manner
The transcriptional regulation of RUNX2 is controlled by several transcription factors and
mechanisms and is not jet fully understood. The described transcriptional regulators,
Discussion 101
Smad1/5 [176] and DLX5 [200] (Distal-Less Homeobox 5) are important enhancers binding
to the promoter region of RUNX2. Smad proteins need to be activated by phosphorylation and
form a trimeric complex to function as transcription factors. Phosphorylation is mediated by
the BMP receptor complex upon BMP-ligand binding [113]. DLX5 is a direct target gene of
BMP-Smad signaling specifically induced by BMP-2 or BMP-4 [200]. The Smad pathway is
therefore crucially important to enhance RUNX2 expression. To induce RUNX2 expression via
mechanical stimulation, mechanotransduction events would need to directly (ligand
independently) activate the BMP-Smad pathway. This was tested in short-term bioreactor
experiments analyzing the BMP pathway under cyclic compression. However, in all cell types
tested here (hFOB, hMSC, hdF, Figure 4-6), cyclic compression alone did not induce
Smad1/5/8 phosphorylation or ID1 expression. In the literature it is still discussed
controversial whether mechanical forces activate Smad signaling in a ligand dependent or
independent manner. Some studies reported that loading alone was sufficient to activate R-
Smads [123], [127] while others described a ligand dependent activation in osteoblastic cells
[96], [125]. This discrepancy may be explained by the use of different experimental setups
including the type of force (fluid shear stress vs cyclic compression) and pre-incubation of
cells up to 7 days prior to loading without medium exchange. Although, fluid shear stress and
cyclic compression might trigger distinct mechanotransduction pathways differently
affecting Smad signaling, shear stress alone did not stimulate R-Smad phosphorylation in
hFOBs (data obtained by Dr. Maria Reichenbach, FU Berlin, Knaus Lab), which is in contrast
to Kido et al. [127]. In case of a pre-incubation of cells up to 7 days prior to loading without
medium exchange, it is likely that mechanical forces induced Smad phosphorylation due to
autocrine BMP secretion. This assumption is supported by a study showing, that Smad
phosphorylation in response to mechanical stimulation was abolished after the addition of
noggin during the loading experiments [96]. For these experiments, stimulation by autocrine
BMP secretion was avoided by exchanging the culture medium to starvation medium only 3h
prior to cyclic compression as well as by using large medium volumes. Therefore, it is
concluded here that mechanotransduction pathways only activate BMP-Smad signaling in a
ligand dependent manner. This also serves as an explanation for the observation that cyclic
compression enhances RUNX2 expression only in a BMP-enriched environment as it was
described in section 5.3.
5.5 The mechano-sensitivity of BMP signaling is dependent on the
loading frequency and timing
As discussed in the previous section, mechanical forces have no influence on BMP signaling
in the absence of the BMP ligand. However, in the presences of BMP, mechanical forces are
102 Discussion
strikingly able to enhance the growth factor´s signaling. This mechano-regulation of BMP
signaling, here also referred to as the crosstalk between mechanotransduction and BMP
signaling, was described before using different cell types and experimental conditions. While
different types of forces like oscillatory fluid shear stress [124], [127], [128], cyclic tension
[180] and compression [123], [125] were found to promote BMP signaling, systematic
investigations of how different loading parameters of the same force influence the duration
and strength of BMP signaling were missing. Such investigations would, however, be
important to defined parameters optimally supporting BMP signaling and furthermore, to
gain insides into the dynamics of this regulation.
Therefore here, the impact of the loading frequency on early BMP signaling events was
investigated in a time dependent manner. Cyclic biomaterial compression with a frequency of
1 Hz was chosen as it represents the frequency of human locomotion and was previously
shown to enhance Smad1/5/8 phosphorylation [125]. Additional frequencies of 10 Hz,
representing muscle contraction [201] and a comparably low frequency of 0.03 Hz were
selected for comparison. As expected, the loading frequency strongly influenced the strength
and the duration of early and late BMP signaling events. Whereas the crosstalk strength was
saturated at 1Hz, the crosstalk duration at Smad and gene expression levels increased with
increasing frequency. The increase in the crosstalk duration was especially obvious when
analyzing the gene expression after 24 hours showing an increase in BMP target genes (ID
genes, noggin, Smad7) only for 10 Hz. Therefore, frequency-dependent effects on early Smad
phosphorylation persisted and transduced to the level of BMP target gene expression,
confirming the importance of the loading frequency for a long-term cell response.
It is proposed that the magnitude and duration of Smad-phosphorylation following cyclic
loading are a measure of how strong mechanotransduction pathways are activated and how
fast cells adapt to the changed mechanical environment. Evidence for this is provided by the
frequency dependent regulation of c-fos expression (see fig. 4-8). C-Fos is a highly mechano-
responsive gene [181] that was shown to be induced via integrin-FAK signaling [202].
Furthermore, this conclusion is based on literature reports analyzing frequency dependent
mechanotransduction responses. An increasing alignment of endothelial cells perpendicular
to the stretch direction, going along with stress fiber reorientation, was observed with
increasing stretch frequency (0.01 Hz , 0.1 Hz, 1 Hz) [203]. This was attributed to a frequency-
dependent increased in p38 phosphorylation, as the reorientation was hindered by the
inhibition of the p38 pathway [204]. Furthermore, cyclic pull on a fibronectin coated
ferromagnetic beads attached to vascular cells increased ERK1/2 phosphorylation in a
frequency dependent manner (from 0.5 to 2 Hz) by two-fold. Since fibronectin binding is
mediated via integrin receptors, activation of the ERK1/2 pathway was attributed to
Discussion 103
frequency-dependent integrin signaling [205]. The frequency-dependent regulation of
mechanotransduction pathways is also relevant on the tissue level, as the bone formation rate
in the rat tibiae [206] or ulna [207] increased with increasing frequency of cyclic bending
(0.05, 0.1, 0.2, 0.5, 1 and 2 Hz) or compression (1, 5, 10 Hz), respectively. Interestingly, it was
found that lower compression load (N) was needed at 10 Hz in comparison to 1 Hz
compression to yield the same bone formation rates [207].
Based on the similar frequency dependency observed in in vivo studies, it is therefore
suggested, that the crosstalk between mechanotransduction and BMP signaling is involved in
the adaptation of bones to mechanical loading.
Strikingly, cyclic compression alone was found to increase the expression of BMP receptor
1B but not BR1A or BR2, whereas BMP-2 treatment had no effect on either of the receptors.
A thorough literature research did not reveal other studies reporting about a mechano-
regulation of BMP receptor expression. The present work thus adds another important
information to understand how mechanical signals regulate BMP signaling.
Cyclic compression was also found to increase the expression of integrin αv and β3 in a
frequency-dependent manner, whereas the expression of integrin α1, α5 and β5 were not
regulated. Even though BMP-only treatment had no effects on integrin expression, BMP-2
stimulation under concurrent cyclic compression further enhanced integrin β3 expression
highlighting the mutual interaction between mechanotransduction and BMP signaling.
Despite the unchanged integrin expression levels upon BMP-2 stimulation, both BMP-2
treatment and cyclic compression increased the size and amount of integrin-mediated
adhesions visualized by p-paxillin staining. This led to the suggestion that BMP-2 treatment
mainly promoted integrin clustering, but comparable literature data is missing to confirm this
assumption. It can be concluded that the strong increase in integrin expression under 10 Hz
loading and the increased integrin clustering under BMP-2 led to the synergistic increase of
FA size and amount under concurrent stimulation. Consistent with the results presented here,
mechanical forces were previously shown to induce the expression of specific integrin
subtypes [208] and growth of adhesion sites [209].
The observed increase in integrin expression and FA assembly, together with the increased
expression of BMP receptor type 1B under mechanical stimulation was especially interesting
since both, an increased amount of BMP receptors and the described BMP receptor-integrin
interaction [128], [130] could promote BMP signaling.
Consequently, two hypotheses arose:
(1) Long-term mechanical loading sensitize the cells for BMP-2 so that subsequent
concurrent mechanical and BMP-2 stimulation would even further increase the
crosstalk.
104 Discussion
(2) Long-term mechanical loading leads to the establishment of a “mechano-memory”,
which would be sufficient to mediate the crosstalk, even though mechanical and
BMP-2 stimulation are not concurrently applied.
Against the hypothesis (1), mechanical pre-stimulation did not further increase BMP
signaling, neither on p-Smad level nor on gene expression level, under subsequent concurrent
mechanical and BMP-2 stimulation. On p-Smad-level, this might be explained by a potential
saturation of phosphorylation already under crosstalk-control conditions (90 min, 1Hz, 10%,
5nM BMP-2). As a consequence a further increase of Smad phosphorylation due to pre-
stimulation could not be detected. Evidence for this is provided by the analysis of frequency-
dependent Smad phosphorylation, where the p-Smad level reached saturation at 90 min for
1Hz and 10 Hz. Therefore, the result could be further supported by the investigation of an
earlier time point (30 min), where a saturation is not jet reached.
Hypothesis (2) was tested by an experiment where mechanical stimulation was applied
for 30 min, 90 min and 24 hours prior to but not during BMP-2 treatment. Intriguingly, it was
observed that long-term mechanical pre-stimulation over 24h induces a persistent
mechanical activation sufficient to increase Smad phosphorylation under BMP stimulation
even in the absence of a direct mechanical trigger. This observation is the first prove for the
existence of a so-called mechano-memory concerning the crosstalk between BMP-2 and cyclic
mechanical compression. Interestingly, the shorter the time of mechanical pre-stimulation
was chosen, the weaker was the effect. While 24 hours were as efficient as the crosstalk-
control, 90 min only reached 50% of the controls’ crosstalk strength and 30 min was too short
to induce a mechano-memory (Figure 4-12). This observation provides valuable information
about the kinetics of the mechano-memory which is relevant to identify the underlying
mechanisms. Cellular adaptation processes involving cytoskeletal reorganization or
conformational changes of proteins are taking place within minutes after the onset of
mechanical stimulation [210], [211]. Adaptation processes due to transcriptional regulations,
however, take several hours [182]. Therefore, during 30 min cyclic compression cells
potentially have remodeled their cytoskeleton and might already have reinforced their
adhesion sites, but transcriptional regulations haven’t had any effects yet. After 90 min of
mechanical stimulation, cytoskeleton and adhesion site have adapted and the first effects of
transcriptional regulations might contribute to meet the new mechanical requirements. After
24h non-transcriptional and transcriptional responses have potentially established a new
mechanical equilibrium. Since only 24 hours pre-loading was as efficient as the crosstalk-
control, it is assumed that mechanical information need to be translated and stored in
transcriptional regulations in order to persist beyond the phase of mechanical stimulation. It
is suggested, that the observed mechano-memory is inter alia mediated by the load-induced
Discussion 105
increase in BMP receptor type 1B expression that would lead to an increase in the amount of
receptors available for ligand binding and Smad phosphorylation. Additionally, taking into
account the here described role of integrin αvβx, a load-induced increase in expression of
integrin αv and β3 as well as FA clustering could lead to increase in BMP receptor-integrin
interactions, thereby promoting BMP signaling.
Physiologically, the existence of a mechanical memory is of high importance for tissue
homeostasis, growth and adaptation like exercise-driven muscle or bone strengthening. In
this context, the here described mechano-memory concerning the crosstalk between BMP-2
signaling and mechanical loading might play an important role in bone metabolism. It is
suggested that cyclic loading during exercise increases the expression of endogenous BMP-2
(paragraph 4.1). The growth factor accumulates over time in the extracellular environment
and triggers BMP signaling, which is enhances by mechanical stimulation even if the person
is already resting. However, it still needs to be elucidated how long-lasting this effect is, in
other words on what time-sale the information in the memory is vanishing. Therefore, in
potential follow-up experiments, BMP-2 should be added at different time points (e.g. 15, 30
and 90 min) after pre-loading. Concerning the existing literature, the phenomenon of
mechanical memory has been exclusively investigated with respect to passive biophysical
cues like substrate rigidity [182] but not with respect to active and alternating mechanical
forces. For example, prolonged culture of hMSCs on stiff matrices that led to nuclear
translocation of YAP (yes-associated protein 1), a mechano-sensitive transcription factor,
prevented YAP re-localization into the cytosol even when cells were subsequently cultured
on soft matrices. These extended pre-cultures on stiff substrates biased hMSCs towards
osteogenic differentiation on soft substrates [212]. The application of prolonged external
forces might have a similar effect on YAP and osteogenic differentiation but a YAP mediated
mechano-memory on BMP signaling is unlikely, since the effects on Smad phosphorylation
are at the level of the receptor and not at nuclear level.
5.6 Integrin αv and load-induced integrin and F-actin
reorganization processes are required for the crosstalk
While transcriptional regulations are suggested to be an integral part of the mechano-
memory, the early induction of Smad phosphorylation upon concurrent mechanical and BMP-
2 stimulation must be, due to its immediateness, independent of any transcriptional
regulation. However, the question remains which mechano-responsive structure facilitates
the fast integration of mechanical signals into the BMP pathway? The Smad phosphorylation
investigated in detail in this thesis is an immediate early BMP signaling event. Since
mechanical forces enhance BMP receptor-mediated Smad phosphorylation already after 15
106 Discussion
min, mechanoresponsive structures at the level of the plasma membrane are potential
candidates that facilitate the crosstalk.
Two prominent mechanosensitive structures located at the plasma membrane are ion
channels and integrins. But also the BMP receptors themselves could possibly function as
mechano-receptors. Reports from literature motivate to investigate a possible integrin- BMP
signaling crosstalk. Integrins were reported to interact with BMP receptors leading to positive
or negative regulations of basal BMP signaling depending on the context (cell type and
experimental conditions) [130], [213]. But intergins could also be involved in the mechano-
regulation of BMP signaling. Interestingly, a study by Zhou et al. (2013) investigating the fluid
flow- mediated phosphorylation of Smad1/5 in endothelial cells (EC), proposed that:
”oscillatory shear stress induces synergistic interactions between specific BMPRs and integrin to
activate Smad1/5 through the Shc/FAK/ERK pathway, which leads to the activation of the
Runx2/mTOR/p70S6K pathway to promote EC proliferation” [128]. However, this study
distinguishes itself strongly from the present work. Firstly, ECs in contrast to hFOBs respond
in a BMP-ligand independent manner to mechanics and secondly, the application of fluid flow
in 2D instead of cyclic compression in 3D. Therefore, it still remains to be elucidated in the
context of bone healing, where different cell types and mechanical stimuli are relevant. Here
it was hypothesized, that integrin αv and β3, which were specifically increased in expression
upon mechanical stimulation, are important for the integration of mechanical signals into the
BMP pathway. Since the knockdown of integrin αv reduced Smad1/5 phosphorylation under
mechanical stimulation in comparison to the crosstalk-control (Figure 4-14), it is suggested
that αv- βx integrins are involved in the mechano-regulation of BMP signaling. However, since
αv not only interacts with β3 but also β1, β5, β6 and β8 [64] it is still open which integrin αv-
heterodimer is responsible. In a next step a β3 integrin knockdown is suggested to be
performed to verify or falsify the hypothesis. It is however important to mention that all αv-
heterodimers are RGD-binding receptors (fibronectin, osteopontin, tenascin and other soft
ECM components). Changing the ECM composition or the substrate stiffness, both influencing
the expression/activation of integrins, will therefore indirectly control the integration of
active mechanical signals into the BMP pathway. It is hypothesized that cells on soft vs. stiff
substrates will differently respond to mechanical stimulation in regards to BMP signaling
amplification. However until today, only stiffness dependent basal BMP signaling was
investigated [136], [214].
Here it was shown that αv integrins are involved in the mechano-regulation of BMP
signaling, but further investigations need to clarify whether this is due to a direct physical
integrin-BMPR association, or indirectly via integrin signaling events. Zhou et al. (2013)
proposed an activation of Smad1/5 through the Shc/FAK/ERK pathway in endothelial cells.
Discussion 107
However, while FAK phosphorylation increased under cyclic compression (Figure 4-15), ERK
and Shc phosphorylation were not found to be regulated in the present study (supplementary
Figure 0-6). Therefore, different mechanisms of integration might be existing in osteoblasts
versus endothelial cells.
If a direct interaction of intergins and BMP receptors is assumed to mediate the
integration of mechanical signals into the BMP pathway, the question remains whether this
interaction already existed under static conditions, or whether it needs to be established in
response to mechanical stimulation? The latter scenario would imply that load-induced
remodeling/reorganization of intergins and BMP receptors must have preceded an
interaction. A load-induced reorganization of focal adhesions was observed here (Figure
4-10) and is described in the literature [215]. To prevent reorganization and thereby an
integrin-BMPR interaction in response to mechanical loading, the actin cytoskeleton
stabilizer Jasplakinolide was used. Since activated integrins are connected to the actin
cytoskeleton and a structural reorganization of the actin cytoskeleton under fluid flow
orchestrates the reorganization of adhesions [80], an inhibition of actin remolding would
consequently inhibit integrin-adhesion remodeling.
At first it was verified by time-lapse imaging under fluid flow that mechanical stimulation
indeed triggers a dynamic actin remolding process, as it was previously described [79], [80],
[216]. In cells stimulated with fluid flow, area and speed of protrusion extension and
protrusion retraction was increased, which is a measure for the actin remodeling dynamics.
Additionally, the phosphorylation of myosin light chain, the regulatory subunit of the myosin
motor protein, was increased significantly under cyclic compression, indicating a load-
induced reinforcement of the actin cytoskeleton that goes along with a remodeling and
adaptation process.
Secondly, the previously described stabilizing-effect of Jasplakinolide on actin remodeling
processes [217] was proven under static and flow conditions (Figure 4-17). In agreement
with Cramer et al. (1999) Jasplakinolide inhibited the dynamic remodeling of protrusion. It
should be noted that the effect of Jasplakinolide was extremely concentration dependent.
High concentrations (≥0.1µM in 2D (Figure 4-16) and >0.5 µM in 3D (data not shown)) led to
a gross disruption of actin organization and the formation of actin aggregates, which is in
accordance with previous investigations [184], [218]. Since a disintegration of the actin
cytoskeleton structure would not only alter immensely the cell morphology but also cellular
mechanosensation, the concentration and timing of Jasplakinolide stimulation was adjusted
carefully. By doing so, actin remodeling could be blocked without major alterations of the
actin organization and cell morphology. In this case, the actin filaments are still taking part in
mechanotransduction processes by transmitting mechanical tension to other proteins.
108 Discussion
However, certainly by changing the actin dynamics, also other load-induced processes were
altered. Even though in turn the remodeling of adhesions sites is reduced, it is assumed that
integrin signaling is not altered.
After proving the ability of Jasplakinolide to efficiently inhibit actin remodeling dynamics
in response to mechanical stimulation, the agent was used to test the hypothesis whether
load-induced dynamic remodeling of the actin cytoskeleton and associated integrin adhesion
remodeling is essential for the mechano-regulation of BMP signaling. Jas treatment during the
bioreactor experiment abolished the positive effect of cyclic compression almost completely,
while basal BMP signaling remained unaffected (Figure 4-18). It is thus proposed here that
actin cytoskeletal adaptations in response to cyclic compression is a prerequisite for the
mechano-regulation of BMP signaling. It is suggested that the stabilization of the actin
cytoskeleton hinders the remodeling of integrins in the plasma membrane and in turn their
interaction with BMP receptors. Certainly, this needs to be elucidated further, for example by
using proximity ligations assays to assess a reduction of integrin-BMPR interaction upon
Jasplakinolide treatment.
Discussion 109
Figure 5-2: Schematic representation of how mechanical forces integrate into BMP signaling. Mechanical stimulation
induces mechanotransduction via integrin signaling (via FAK) which leads to the remodeling of the actin cytoskeleton
associated with increased protrusion dynamics, which in turn stimulates integrin-adhesion remodeling. The dynamic
reorganization and maturation of adhesion sites causes increased interactions of αv-βx integrins and BMP receptors,
which leads to the amplification of Smad1/5 phosphorylation and early target gene expression (ID1). Knockdown of
integrin αv reduces the amount of interaction partners for the BMP receptor, thereby reducing the mechanosensitivity
of BMP signaling. Inhibition of load-induced actin remodeling hinders the remodeling of adhesions and in turn the
interaction of αv-βx intergins and BMP receptors. Long-term mechanical stimulation will trigger the expression of
BMP receptor Ib (BMPRIB) and integrin αv (ITGav) and β3 (ITGb3) possibly responsible for the mechano-memory of
BMP signaling.
In summary, (i) the load-induced growth of FA adhesions going along with a
reorganization of integrins, (ii) the decrease in crosstalk efficiency after integrin αv
knockdown (iii) the decreased crosstalk after F-actin stabilization and (iv) the known BMP
receptor-integrin interaction [128], [130] led to the interpretation that the induction of
intergin reorganization in response to mechanical loading and BMP-2 stimulation causes
increased interactions of αv-βx integrins and BMP receptors, which leads to the amplification
of Smad1/5 phosphorylation. A knockdown of integrin αv reduces the amount of interaction
partners for the BMP receptor, thereby reducing the mechano-sensitivity of BMP signaling.
Inhibition of load-induced actin remodeling hinders the reorganization of intergins and in
turn the interaction of αv-βx integrins and BMP receptors. Long-term mechanical stimulation
110 Discussion
triggers the expression of BMP receptor Ib and integrin αv and β3 proposed to be responsible
for the mechano-memory of BMP signaling.
5.7 Mechanical forces specifically alter mechanical, structural and
compositional matrix cues
BMP-2 is known for its strong osteoinductive potential and its influence on cell differentiation
in the context of bone healing is well studied. However, the growth factor should not be
reduced to this feature alone, as it was described to influence process which happen much
earlier during the bone healing cascade. These processes include, cell migration [23],
proliferation [22], angiogenesis [219] and evidences point towards an additional role in
steering early extracellular matrix formation processes [149], [155]. During bone healing,
extracellular matrix formation is initiated directly with the end of the pro-inflammatory
phase [15] and the early structural organization of collagen fibers within the fracture gap was
shown to critically influence healing [11]. Given the importance of early ECM formation
processes and the fact that both BMP-2 [149], [155] and mechanical forces [142]–[146] were
independently described to influence such processes, it is even more important to study how
ECM formation is influenced by their mutual interaction. To better understand a potential
crosstalk, in this study the individual and mutual influences of cyclic mechanical loading and
BMP-2 stimulation on ECM formation were compared. Since mechanical forces have already
been shown to promote BMP signaling, it was hypothesized here that this positive effect is
transduced to the ECM-level. This means that cyclic loading was expected to further enhance
BMP-2-specific ECM alterations under concurrent treatment.
For the investigation of ECM formation processes in 3D, the macroporous collagen scaffold,
which was previously used as an in vitro tissue formation model system [146], [171], [220],
was selected. Due to its high porosity, it provides space for the fibroblasts to de-novo deposit
ECM and furthermore allows ECM imaging and reliable quantification of the cell-derived
fibrillar collagen, as it is spatially and structurally distinguishable from the scaffold walls. In
addition, the macroscopic stiffness of the scaffold selected here (5.9 ± 0.6kPa) allows for a
slow cell-mediated biomaterial contraction during culture (Figure 4-19). This feature is very
interesting, since tissue contraction represents an important process during tissue repair to
re-establish the tissue pretension, which was destroyed upon injury [185], [221]. However,
cells alone are not capable to contract the scaffold. Instead, contraction requires the gradual
conversion and storage of cell forces into cell-deposited, pre-tensioned collagen fibers [171].
Additionally, fibrillar collagens are an integral structural component of the ECM, greatly
defining its mechanical properties [140]. Material properties like structure and stiffness in
Discussion 111
turn greatly influence cell behavior such as cell signaling, specifically BMP signaling [136],
[222] and differentiation [56], [58]. Therefore in this thesis, ECM formation processes were
analyzed with a special focus on how mechanical and BMP-2 stimulation impact tissue
contraction, stiffening and structuring, processes which are influences by collagen formation
and which vice versa affect cell behavior.
Effects of BMP-2 treatment: Here is was found that, BMP-2 treatment led to a slightly
enhanced tissue contraction and stiffening in comparison to the control (see Figure 4-19).
Due to the increased contraction and the accompanied bending of the collagen walls, the
alignment of collagen fibers and cells were in turn slightly reduced (see Figure 4-22). No
changes could be observed on the level of collagens. Neither the expression of collagen I or VI,
nor the concentration of secreted pro-collagen type I, nor the amount of collagen type I
deposited into the matrix, nor the fibrillar collagen density was changed in comparison to the
control (see Figure 4-20 and Figure 4-23). However, since the amount of fibrillar collagen
correlated with the amount of tissue contraction [171], it might be assumed that BMP
treatment potentially stimulates cellular contractility, which is supported by other studies
[54], [223]. Gene expression analysis revealed a strong upregulation of MMP1 transcription
by BMP stimulation (see Figure 4-23), which is in agreement with a previous study [224].
MMP1 is a collagenases cleaving collagen type I, II and III [225], while only type I and III are
secreted by fibroblasts. This would point towards an increased degradation of the
collagenous ECM, which was however not reflected in the quantification of collagen. To
further validate the regulation of MMP1 and other MMPs, it would be necessary to investigate
the MMP protein levels and their activity using for example zymographic analyses.
Additionally, the expression and protein levels of tissue inhibitor of metalloproteinases
(TIMPs) should be examined since they control the activity of MMPs [141]. Even if no changes
on the level of collagens were observed, gene expression analysis showed an upregulation of
fibulin-1 and a strong downregulation of elastin expression in response to BMP stimulation.
Fibulin-1 is a family member of eight glycoproteins and its function has been associated with
cell adhesion and matrix remodeling due to integrin and metalloproteinase interactions,
respectively [226]. Furthermore, fibulin-1 deficient mice show a clear bone phenotype with
reduced bone volume and mineralization. It can be found in the ECM surrounding osteoblasts
and was suggested to function as a positive regulator of BMP signaling in mice [227]. Elastin
is secreted as tropoelastin that is cross-linked via lysyl oxidases to form elastin fibers onto
preformed bundles of fibrillin microfibrils. As the name suggests, elastin fibers provide
elasticity to tissues. While a regulation of elastin expression by BMPs has not been described
so far, TGFβ1 was found to have pro-elastogenic activities [187].
112 Discussion
Taken together, even though some minor changes on tissue contraction, stiffening and
gene expression have been observed, a clear and dominant BMP-effect on ECM formation was
not detected for the analyzed parameters. Prolonged culture times and a profound mass
spectrometry analysis might be necessary to identify the expected BMP-matrix phenotype.
Even though the responsiveness of fibroblasts to 5nM BMP-2 stimulation was verified before
the experiment (Figure 4-6), it could be possible that higher BMP-2 concentrations are
required to observe an effect.
Effects of cyclic compression: Most striking was the effect of cyclic compression on matrix
formation. At first, a significantly increased tissue contraction was observed in axial and
radial direction (see Figure 4-19). Although the collagen scaffold is fully elastic, upon
deposition of ECM a composite material is formed consisting of collagens, elastic fibers and
water-binding glycosaminoglycans that is characterized by viscoelastic mechanical
properties and a certain stickiness. Repeated axial compression consequently led to a tissue
compaction in the loading direction. However, since most of the materials respond with radial
expansion upon axial compression, the radial contraction perpendicular to the loading
direction observed here, must have been driven by active cell forces triggered by mechanical
stimulation. Going along with the increase in contraction and the accompanied deformation
of the collagen walls, the cells and the ECM adopted a more isotropic orientation (see Figure
4-22). Furthermore, a tissue stiffening was observed likely as a result of tissue densification.
Interestingly, mass spectrometry revealed an increased abundance of elastin in the matrices
of loaded samples. As described above, elastin fibers provide elasticity to tissues, therefore, it
is found in great amounts in tissues, which are subjected to dynamic loads, like blood vessels,
ligaments and skin. Static stretching has been shown before to induce elastin expression in
smooth muscle cells [228]. However, cyclic stretching (5 and 10% for 24h) of periodontal
ligament fibroblasts decreased elastin expression [229]. It might be speculated that
fibroblasts try to counteract the increasing compaction of their environment by introducing
an elastic ECM component.
Intriguingly, the increase in tissue contraction and stiffening was not accompanied by an
increase, but a significant reduction in the amount of fibrillar collagen (see Figure 4-21). This
again might suggests that cellular contraction forces might be accelerated in response to
loading. Indeed, increased matrix stiffness [230] and mechanical tension [231] triggered the
transdifferentiating of fibroblasts into myofibroblast, which feature a strong contraction
ability. An immunohistological staining for α-smooth muscle actin, a myofibroblast marker,
might be a way to clarify whether mechanical loading indeed increases cell contractility
leading to the here-observed contraction of scaffold-based in vitro tissues.
Discussion 113
The significantly reduced fibrillar collagen density observed under mechanical stimulation
stands in striking contrast to (i) the strong and significantly upregulated secretion of pro-
collagen type I C-peptide, (ii) the unchanged collagen I density assessed via specific antibody
staining and (iii) unchanged collagen type I and VI abundance assessed via MS analysis. Also
after 2 weeks of culture, fibrillar collagen density remained decreased in comparison to the
control indicated by preliminary experiments (see supplementary Figure 0-5). Therefore in
conclusion, cyclic compression disturbed the assembly of thick fibrillar collagen bundles,
raising the question how and at which stage fibrillogenesis is disturbed?
Collagen fibril formation is a complex multistage process starting with the intracellular
assembly of triple helices from synthesized and modified single proα-chains. After secretion
of the soluble procollagen molecules, the propeptides at each end of the triple helix are
cleaved enzymatically. The C-propeptide, which was measured in the culture medium (see
Figure 4-20), is cleaved by the metalloproteinase BMP-1 and by other members of tolloid-like
metalloproteases [140]. In agreement with the increased C-propeptide concentration, the
expression of BMP-1 was increased under mechanical stimulation (Figure 4-23). The
resulting tropocollagens self-assemble into staggered collagen fibrils, a process which is
regulated by cell-adhesions like integrins and other ECM proteins like fibronectin. To stabilize
the fibril and strengthen its mechanical properties, lysyl oxidases (LOX) need to covalently
crosslink the tropocollagens [140]. Even though the expression of lysyl oxidases was slightly
increased in mechanically stimulated samples, the density of fibrillar collagen was
significantly reduced (see Figure 4-20). This discrepancy might be explained by an
insufficient activation of lysyl oxidases in the extracellular space. This assumption is based on
the finding that periostin, a protein participating in lysyl oxidase activation, was significantly
reduced the matrices of mechanically stimulated samples (see Figure 4-24). Due to its multi-
domain structure, periostin is an important scaffolding protein binding to ECM proteins
(collagen I and V, fibronectin, tenascin C, and laminin), enzymes (BMP-1, LOX) and integrins
[232]. Importantly, BMP-1 not only cleaves the propeptide of collagens but also the
propeptide of LOX to activate its enzymatic activity [233]. The activated LOX in turn catalyzes
the covalent cross-linking of staggered collagen fibrils. Periostin supports BMP-1-mediated
proteolytic activation of LOX, by bringing the interacting proteins in close proximity, thereby
facilitating and accelerating collagen cross-linking (Figure 5-3) [232], [234].
114 Discussion
Figure 5-3: The regulation of collagen cross-linking by periostin. Periostin binds to fibronectin via its EMI domain and
to BMP-1 through its FAS-1 domain, thereby promoting the deposition of BMP-1 into the ECM. BMP-1 activates LOX
by proteolytic cleavage of the precursor LOX (Pro-LOX). Active LOX synthesizes pyridinium cross-links to interconnect
collagen fibers. Figure modified from [235] with permission from the publisher.
Its physiological relevance is highlighted even more as periostin deficient mice exhibit
reduced collagen cross-linking [236]. Here, the reduced periostin abundance in the ECM of
samples subjected to cyclic compression could be one reason for the load-induced reduction
of fibrillar collagen density. Collagen fibrils might self-assembly into staggered fibrils but their
insufficient cross-linking leads to fibril disruption under cyclic compression. The increased
secretion of collagen I and expression of BMP1, periostin and lysyl oxidases might be a
compensation mechanism, which was for some reason insufficient.
The finding that cyclic compression reduced the density of fibrillar collagen is
contradicting the expectation that cells in a highly mechanically unstable environment aim to
stabilize the tissue by an increased deposition of load-bearing fibrillar collagen. This together
with the finding that cyclic compression strongly increase the secretion of pro-collagen type
I, leads to the suggestion, that cell indeed aim at stabilizing their environment with the help
of collagen, but mechanical loading disturbs collagen fibrillogenesis (see Figure 5-4).
Figure 5-4: Cyclic compression disturbs collagen
cross-linking. After transcription, translation and
translocation into the rough endoplasmic
reticulum and posttranslational modifications
single proα-chains associate via the C-terminus.
The triple helix is formed from C-to-N terminus and
thereafter secreted into the extracellular space.
After enzymatic cleavage of the propeptides the
helices assemble into ordered fibrils. These fibrils
are finally stabilized by inter and intramolecular
cross-links. Mechanical stimulation is proposed to
disturb the step of collagen cross-linking. Figure
information taken from [140].
Discussion 115
Next to the reasonable explanation, that collagen cross-linking is insufficient due to the
significantly reduced deposition of periostin into the ECM, it might also be a pure mechanical
interference with the process of fibril formation. It might be that 5 hours rest during the
loading intervals are too short to sufficiently stabilize collagen fibrils leading to mechanically
disruption during the subsequent loading phase.
Although mechanical stress has been previously reported to alter collagens on different
levels of its biosynthesis, comprehensive studies investigating the influence of mechanics in
a 3D environment on all levels of collagen synthesis, as it was performed here, are missing.
Therefore, it is difficult to integrate the current findings into the existing literature and only
individual aspects will be compared here. The expression of collagen type I was described to
be differently regulated depending on the loading regime. While cyclic tension applied to
fibroblasts seeded onto collagen coated silicon membranes induced an increase in collagen
expression [142], relaxation of fibroblast- seeded pre-stretched membranes induced a
reduction of collagen type I expression [144]. In line with the here presented data, the
secretion of the procollagen C-peptide was found to be upregulated in response to cyclic
biomaterial compression [159]. In contrast to the present finding, collagen fibril formation,
visualized by SHG imaging, was reported to increase upon the application of cyclic stretch to
osteoblasts seeded onto flexible PDMS membranes coated with fibronectin [237]. The
contradiction might be due to the differences in culture dimensions. However, studies
visualizing collagen fibril formation under cyclic compression in a 3D environment by SHG
imaging are missing. The present work is the first to performed comprehensive investigations
from collagen gene expression to final collagen fibers and therefore delivers new insights on
how collagen biosynthesis is regulated by cyclic compression.
An interpretation of the current findings in the context of wound healing, suggests that
different loading regimes might be favorable for different processes. While BMP signaling and
osteogenic differentiation are promoted by cyclic compression with a frequency of 1Hz and
an amplitude of 10%, fibrillar collagen formation and consequent mechanical stabilization is
disturbed. In bone regeneration, mechanical instability (in the range of 5-15% strain [40])
would lead to healing via endochondral ossification, a process in which a cartilaginous phase
is proceeding mineralization [6]. If mechanical forces hinder the early formation of a fibrillar
collagen-rich ECM, which characterizes bone tissue and serves as a template for mineral
crystal deposition [238], an intermediate step aiming at tissue stabilization is required to
proceed. Cartilage most probably fulfills this function: its proteoglycan-rich ECM could act as
a “shock absorber” and establishes the mechanical stability that is needed for successful
collagen fibrillogenesis. This assumption is supported by the fact that bones heal via direct
ossification (intermembranous healing) in case of mechanical stability (and small fracture
116 Discussion
gaps) [6]. Following the interpretation line, these conditions would allow for a formation of a
collagen fiber network, in which mineral crystal can be immediately deposited – an
intermediate step of cartilage is not required.
Effects of concurrent BMP-2 treatment and cyclic compression: Due to the previously
described mechanosensitivity of the BMP pathway, it was hypothesized, that mechanical
stimulation would also enhance the BMP-effects on early tissue formation. However, due to
the weak influence of BMP-2 stimulation and the strong effects of cyclic compression on ECM
formation, the load-effect was dominating in samples concurrently stimulated. Also, no
synergistic effects have been observed, however conversely, BMP-2 dampened the impact of
mechanical stimulation for some parameters investigated. This was the case for tissue
contraction and stiffening (see Figure 4-19). Samples concurrently stimulated with BMP-2
and mechanical loading were, even though not significant, less contracted and softer than
samples only treated mechanically. This might be due to increased tissue remodeling but
further investigations are needed to prove this assumption. Also, the strong increase in gene
expression induced by mechanical stimulation of for example fibulin-1, elastin or TGFβ-
induced protein, was reduced by a stimulation with BMP-2 (see Figure 4-23). Additionally,
although after one week of culture BMP-2 did not influence the load-induced downregulation
of fibrillar collagen density, preliminary experiments over two weeks indicate a rescuing
ability of BMP-2 (see supplementary Figure 0-5).
An in vivo study investigating the interaction of BMP-2 stimulation and mechanical
boundary conditions in a rat critical-sized femoral defect model (5 mm) using three distinctly
different external fixator stiffness, indeed showed that their mutual interaction does have
implications beyond the induction of osteogenic differentiation [54]. Using gene expression
profiling performed at day 3 and 7, distinct differences in the expression patterns of genes
involved in extracellular matrix formation and cellular contractility were observed. While a
rigid fixation led to an increased expression of genes related to ECM remodeling, flexible
fixation triggered the expression of genes related to inflammatory response and cellular
contractility. Since the semi-rigid fixation showed the best healing outcome, it becomes clear
that mechanical stimuli need to be tightly balanced in order to positively cooperate with BMP-
2. Overall, literature concerning the influence of BMP-2 and mechanical stimulation on ECM
formation is very limited and we are just starting to understand their mutual interactions.
Further research will be needed to interpret the present findings in the context of bone
regeneration and wound healing.
Summary and Conclusion 117
6 Summary and Conclusion
Mechanical forces are, as one factor of the diamond concept [16], critically influencing
bone healing with detrimental effects if interfragmentary forces are too low or too high, but
beneficial effects if optimized. However, in order to employ the great potential of mechanical
forces, its effects on molecular, cellular and tissue level need to be understood and combined
to a universal model. To contribute to a profound understanding, in the first part of this thesis,
the direct influence of cyclic mechanical loading on osteogenic differentiation of primary
hBMSCs was investigated excluding effects of altered supply, autocrine stimulation and
further osteogenic triggers (e.g. medium supplements). The outcomes revealed that cyclic
compressive loading per se does not trigger osteogenic differentiation but instead causes a
downregulation of RUNX2 and osteocalcin expression. Osteogenic differentiation, indicated
by increased RUNX2 expression, was only promoted by cyclic compression, if an enrichment
of secreted factors including BMP-2 in the cell culture medium and resulting autocrine
signaling was permitted. This is striking, as it implies that the presence of BMP-2 changes the
cells response to mechanical stimulation. The proposed BMP-mediated osteogenesis under
cyclic compression was underpinned by the absence of load-induced osteogenic
differentiation when a specific BMP inhibitor was supplemented. This observation provides
evidence that mechanical stimulation induces osteogenic differentiation via a mechano-
regulated autocrine feedback mechanism involving BMP-2 [172].
Mechanical forces promote BMP signaling not only indirectly via the regulation of ligand
expression, but also directly by enhancing BMP signaling events at the receptor level that
further translates into the level of target gene expression. The aims of the second part of this
project, were to investigate the influence of mechanical loading parameters on the mechano-
sensitivity of BMP signaling to define optimal mechanical parameters and to gain a deeper
molecular understanding of how mechanical signals regulate the BMP signaling pathway. It
was found that the intensity and duration of Smad phosphorylation and target gene
expression was strongly affected by the loading frequency. While the intensity of Smad
phosphorylation reached saturation at 1Hz, the duration of the crosstalk increased with
increasing frequency, which was indicated by a prolonged increase of Smad phosphorylation.
Moreover, these frequency-dependent effects on early Smad phosphorylation persisted and
transduced to the level of BMP target gene expression. The results revealed that high
frequency loading is most effectively supporting BMP signaling. Strikingly, these
investigations also revealed a so far unknown mechano-regulation of BMP receptor type 1B
expression. Together with the observed mechanically-induced increase in integrin expression
118 Summary and Conclusion
and clustering this lead to the hypothesis that cells would establish a “mechano-memory”
upon long-term mechanical pre-loading. Indeed, here it could be shown for the first time that
long-term mechanical pre-stimulation induces a persistent mechano-memory sufficient to
increase Smad phosphorylation under BMP stimulation even in the absence of a direct
mechanical trigger. Since this effect decreased with decreasing pre-loading durations, it is
concluded that the crosstalk induced by a mechano-memory requires additional
transcriptional adaptations like BMP receptor 1B expression. While transcriptional
regulations are suggested to be an integral part of the mechanical memory, the immediate
early induction of Smad phosphorylation upon concurrent mechanical and biochemical
(BMP-2) stimulation is independent of any transcriptional regulation. However, the way how
mechanical signals integrate into the BMP pathway in such an immediate manner is still
unknown. To gain a deeper understanding of the molecular mechanism responsible for the
mechano-regulation of BMP signaling, the hypothesis was tested that mechanical forces
integrate into the BMP pathway via mechanotransduction through the integrin-F-actin-axis.
In this thesis, it was found that mechanical stimulation induces an increased integrin
clustering, integrin activation (indicated by increased p-FAK(Y397) levels) and F-actin
cytoskeleton remodeling. Furthermore, integrin αv knockdown and F-actin stabilization
decreased the efficiency of mechanical forces to amplify BMP signaling. Together with the
known BMP receptor-integrin interaction, it is concluded that the increased adhesion
remodeling in response to mechanical loading and BMP-2 stimulation causes increased
interactions of αv-βx integrins and BMP receptors, which leads to the amplification of
Smad1/5 phosphorylation. The knockdown of integrin αv reduced the amount of interaction
partners for the BMP receptor, thereby reducing the mechano-sensitivity of BMP signaling.
Inhibition of load-induced actin remodeling blocks the remodeling of adhesions and in turn
the interaction of αv-βx integrins and BMP receptors.
In the third part for the thesis, the influence of mechanical stimulation and BMP
treatment on extracellular matrix formation was investigated. The hypothesis was tested that
mechanical stimulation would not only increase the growth factors` potential to induce
osteogenic differentiation but also foster its effects on tissue formation. Therefore, human
fibroblasts were subjected to cyclic compression and BMP-2 stimulation and resulting ECM
properties were investigated with the focus on collagen. However, only minor changes on
tissue contraction, stiffening and gene expression have been observed under BMP-only
treatment and a clear and dominant BMP-effect on ECM formation could not be detected for
the analyzed parameters. Mechanical stimulation on the other hand, significantly increased
tissue contraction, stiffness and collagen type 1 secretion but surprisingly reduced the density
Summary and Conclusion 119
and structural alignment of fibrillar collagen fibers. This led to the conclusion that cyclic
compression disturbed the process of collagen fibrillogenesis. Analysis of the ECM
composition by MS revealed a consistent and significant increase in elastin, but a reduction of
fibulin1, periostin, tenascin and TGFβ-induced protein. Therefore, mechanical forces
specifically altered mechanical, structural and compositional matrix cues, which might in turn
alter cellular behavior. Under concurrent cyclic compression and BMP-2 stimulation, the
effect of cyclic compression was dominating. Synergistic effects were not observed, however
conversely, BMP-2 slightly dampened the impact of mechanical stimulation on some
parameters investigated such as tissue contraction, stiffening and gene expression. This might
point towards an increased tissue remodeling that could be beneficial in the context of
regeneration but further investigations are needed. Even though the hypothesis that
mechanical stimulation would increase the BMP-effects on early tissue formation could not
be proven here, important new insides into how mechanical stimulation influences ECM
formation have been gained.
It became clear, that different processes in the healing cascade favor different mechanical
boundary conditions loading regimes. While, BMP signaling and osteogenic differentiation
were promoted by mechanical signals, fibrillar collagen formation requires mechanical
stability.
Taken together, the main conclusions of this dissertation are:
(1) Mechanical stimulation induces osteogenic differentiation indirectly through a
mechanically controlled secretion of BMP-2 and the resulting biochemical self-
stimulation.
(2) Cells feature a mechanical memory, established via transcriptional regulation that leads
to an increased signaling response to BMP-2 even when the mechanical signal has
vanished.
(3) The immediate mechano-regulation of BMP signaling requires the presence of integrin
αv as well as load-induced integrin and actin cytoskeleton remodeling.
(4) Mechanical stimulation specifically modulates mechanical, structural and compositional
extracellular matrix cues, which are suggested to in turn influence cell behavior
This thesis therefore contributes to a profound understanding of how mechanical forces
regulate osteogenic differentiation, BMP signaling and early tissue formation - important
processes during bone regeneration.
120 Outlook
7 Outlook
The mechanical boundary conditions at the fracture site critically influence bone healing
and a mechano-biological optimization of interfragmentary movements, especially during the
early phase of healing, has great potential to promote the subsequent healing cascade. The
power of mechanical forces is in part based on the amplification of the BMP signaling pathway
enhancing the effectiveness of endogenous and therapeutic BMP-2. With future personalized
medicine, the fracture fixation system should account for the individual mechano-biological
requirements that depend on the location and type of fracture. Personalized computational
models of the patients’ fracture could help to simulate in vivo loads resulting from different
fixation systems. Intelligent fixation systems, which are equipping with strain gauges and load
sensors could realize postoperative validations and further control. However, even though
optimal mechanical parameters could be derived from this work and studies in animal
models, there needs to be further research in humans using for example the mentioned
intelligent fixations systems (which would need to be developed). Knowing from this study
that high frequency loading most effectively amplifies BMP signaling, it could be possible to
implement postoperative physiotherapy using whole body vibration training with optimized
parameters that indeed has been shown to be beneficial for bone healing in rodents [239],
[240]. However, knowing that collagen fibrillogenesis favors mechanical stability, also the
timing of load application post-operation would need to be optimized. Furthermore, local
external mechanical stimulation could be conceivable. Currently, in-house developed air
pressure controlled compression cuffs are under establishment in the Julius Wolff Institute
that apply 1Hz cyclic compression externally to a femoral fracture in mice. The external
massage shows promising first results on bone healing, although the underlying mechanism
needs further elucidation.
A profound understanding of the molecular mechanism underlying the regulation of BMP
signaling by mechanical forces might in the future lead to the identification of therapeutic
candidates, which if targeted could amplify BMP signaling even without the application of
external forces. This could be for example relevant for elderly immobile patients.
With regard to the specific alteration of ECM properties by mechanical stimulation, future
work should investigate progenitor cell responses like proliferation, migration,
differentiation and signaling on those matrices. Since material properties like stiffness
influence cellular fate decision [56] and BMP signaling (unpublished data), it is likely that the
ECM established under mechanical stimulation specifically modulates cell behavior. The
resulting findings could inspire specific biomaterial designs for bone regeneration.
Together, this work contribute to a better understanding of the signaling cascades and
matrix formation processes involved in bone regeneration process. A mechanistic
understanding of those processes is very valuable as is allows the knowledge-transfer to
other tissues.
Bibliography 121
8 Bibliography
[1] M. Maden, “The evolution of regeneration - where does that leave mammals?,” Int. J.
Dev. Biol., vol. 62, no. 6-7–8, pp. 369–372.
[2] K. Schmidt-Bleek, A. Petersen, A. Dienelt, C. Schwarz, and G. N. Duda, “Initiation and
early control of tissue regeneration – bone healing as a model system for tissue
regeneration,” Expert Opin. Biol. Ther., vol. 14, no. 2, pp. 247–259, Feb. 2014.
[3] D. T. Corr and D. A. Hart, “Biomechanics of Scar Tissue and Uninjured Skin.,” Adv. wound
care, vol. 2, no. 2, pp. 37–43, Mar. 2013.
[4] A. M. Segura, O. H. Frazier, and L. M. Buja, “Fibrosis and heart failure,” Heart Fail. Rev.,
vol. 19, no. 2, pp. 173–185, Mar. 2014.
[5] C. Sfeir and L. Ho, “Fracture repair,” Bone Regen. repair, no. 11, pp. 21–44, 2005.
[6] T. Tosounidis, G. Kontakis, V. Nikolaou, a. Papathanassopoulos, and P. V Giannoudis,
“Fracture healing and bone repair: an update,” Trauma, vol. 11, no. 3, pp. 145–156,
2009.
[7] M. Greenbaum and I. Kanat, “Current concepts in bone healing. Review of the
literature,” J. Am. Podiatr. Med. Assoc., vol. 83, no. 3, pp. 123–129, Mar. 1993.
[8] P. Kolar et al., “The Early Fracture Hematoma and Its Potential Role in Fracture
Healing,” Tissue Eng. Part B Rev., vol. 16, no. 4, pp. 427–434, Aug. 2010.
[9] K. Schmidt-Bleek et al., “Inflammatory phase of bone healing initiates the regenerative
healing cascade,” Cell Tissue Res., vol. 347, no. 3, pp. 567–573, Mar. 2012.
[10] R. Dimitriou, E. Tsiridis, and P. V. Giannoudis, “Current concepts of molecular aspects
of bone healing,” Injury, vol. 36, no. 12, pp. 1392–1404, 2005.
[11] A. Petersen et al., “A biomaterial with a channel-like pore architecture induces
endochondral healing of bone defects,” Nat. Commun., vol. 9, no. 1, p. 4430, Dec. 2018.
[12] M. E. Bolander, “Regulation of Fracture Repair by Growth Factors,” Exp. Biol. Med., vol.
200, no. 2, pp. 165–170, Jun. 1992.
[13] T. A. Einhorn, A. Hirschman, C. Kaplan, R. Nashed, V. J. Devlin, and J. Warman, “Neutral
protein-degrading enzymes in experimental fracture callus: A preliminary report,” J.
Orthop. Res., vol. 7, no. 6, pp. 792–805, Nov. 1989.
[14] H. M. Frost, “A determinant of bone architecture. The minimum effective strain.,” Clin.
Orthop. Relat. Res., no. 175, pp. 286–92, May 1983.
[15] K. Schmidt-Bleek, B. J. Kwee, D. J. Mooney, and G. N. Duda, “Boon and Bane of
Inflammation in Bone Tissue Regeneration and Its Link with Angiogenesis,” Tissue
Engineering - Part B: Reviews, vol. 21, no. 4. Mary Ann Liebert Inc., pp. 354–364, 01-
Aug-2015.
[16] P. V Giannoudis, T. a Einhorn, and D. Marsh, “Fracture healing: The diamond concept,”
4th Eur. Clin. Symp. Bone Tissue Regen., vol. 38, Supple, no. 0, pp. S3–S6, 2007.
[17] A. Moghaddam, B. Thaler, T. Bruckner, M. Tanner, and G. Schmidmaier, “Treatment of
atrophic femoral non-unions according to the diamond concept: Results of one- and
two-step surgical procedure,” J. Orthop., vol. 14, no. 1, p. 123, 2017.
[18] B. M. Willie et al., “Designing biomimetic scaffolds for bone regeneration: why aim for
a copy of mature tissue properties if nature uses a different approach?,” Soft Matter,
vol. 6, no. 20, p. 4976, 2010.
[19] P. H. Locker, J. Arthur, T. Edmiston, R. Puri, and B. R. Levine, “Management of Bone
Defects in Orthopedic Trauma.,” Bull. Hosp. Jt. Dis., vol. 76, no. 1, pp. 278–284, Mar.
2018.
122 Bibliography
[20] R. Dimitriou, E. Jones, D. McGonagle, and P. V Giannoudis, “Bone regeneration: current
concepts and future directions,” BMC Med., vol. 9, no. 1, p. 66, 2011.
[21] T. A. St John et al., “Physical and monetary costs associated with autogenous bone graft
harvesting.,” Am. J. Orthop. (Belle Mead. NJ)., vol. 32, no. 1, pp. 18–23, Jan. 2003.
[22] O. Fromigué, P. J. Marie, and A. Lomri, “Bone morphogenetic protein-2 and
transforming growth factor-beta2 interact to modulate human bone marrow stromal
cell proliferation and differentiation.,” J. Cell. Biochem., vol. 68, no. 4, pp. 411–26, Mar.
1998.
[23] S. Liu et al., “Recombinant human BMP-2 accelerates the migration of bone marrow
mesenchymal stem cells: Via the CDC42/PAK1/LIMK1 pathway in vitro and in vivo,”
Biomater. Sci., vol. 7, no. 1, pp. 362–372, Jan. 2019.
[24] G. Sánchez-Duffhues, C. Hiepen, P. Knaus, and P. ten Dijke, “Bone morphogenetic
protein signaling in bone homeostasis,” Bone, vol. 80, pp. 43–59, Nov. 2015.
[25] I. Dumic-Cule, M. Peric, L. Kucko, L. Grgurevic, M. Pecina, and S. Vukicevic, “Bone
morphogenetic proteins in fracture repair,” Int. Orthop., vol. 42, no. 11, pp. 2619–2626,
Nov. 2018.
[26] A. H. Reddi and N. S. Cunningham, “Initiation and promotion of bone differentiation by
bone morphogenetic proteins,” J. Bone Miner. Res., vol. 8, no. S2, pp. S499–S502, Dec.
2009.
[27] H. CHENG et al., “OSTEOGENIC ACTIVITY OF THE FOURTEEN TYPES OF HUMAN BONE
MORPHOGENETIC PROTEINS (BMPS),” J. Bone Jt. Surgery-American Vol., vol. 85, no. 8,
pp. 1544–1552, Aug. 2003.
[28] K. Tsuji et al., “BMP2 activity, although dispensable for bone formation, is required for
the initiation of fracture healing.,” Nat. Genet., vol. 38, no. 12, pp. 1424–9, Dec. 2006.
[29] O. P. Gautschi, S. P. Frey, and R. Zellweger, “Bone morphogenetic proteins in clinical
applications.,” ANZ J. Surg., vol. 77, no. 8, pp. 626–31, Aug. 2007.
[30] K. R. Garrison et al., “Clinical effectiveness and cost-effectiveness of bone
morphogenetic proteins in the non-healing of fractures and spinal fusion: a systematic
review.,” Health Technol. Assess., vol. 11, no. 30, pp. 1–150, iii–iv, Aug. 2007.
[31] S. Wei, X. Cai, J. Huang, F. Xu, X. Liu, and Q. Wang, “Recombinant human BMP-2 for the
treatment of open tibial fractures.,” Orthopedics, vol. 35, no. 6. pp. e847–e854, 01-Jun-
2012.
[32] A. W. James et al., “A Review of the Clinical Side Effects of Bone Morphogenetic Protein-
2.,” Tissue Eng. Part B. Rev., vol. 22, no. 4, pp. 284–97, 2016.
[33] D. C. Betts and R. Müller, “Mechanical regulation of bone regeneration: theories,
models, and experiments.,” Front. Endocrinol. (Lausanne)., vol. 5, p. 211, Jan. 2014.
[34] G. N. Duda et al., “Mechanical boundary conditions of fracture healing: borderline
indications in the treatment of unreamed tibial nailing.,” J. Biomech., vol. 34, no. 5, pp.
639–50, May 2001.
[35] V. Glatt, C. H. Evans, and K. Tetsworth, “A Concert between Biology and Biomechanics:
The Influence of the Mechanical Environment on Bone Healing.,” Front. Physiol., vol. 7,
p. 678, 2016.
[36] M. Steiner, L. Claes, A. Ignatius, U. Simon, and T. Wehner, “Numerical Simulation of
Callus Healing for Optimization of Fracture Fixation Stiffness,” PLoS One, vol. 9, no. 7,
p. e101370, Jul. 2014.
[37] H. Schell et al., “Mechanical induction of critically delayed bone healing in sheep:
Radiological and biomechanical results,” J. Biomech., vol. 41, no. 14, pp. 3066–3072,
Oct. 2008.
Bibliography 123
[38] K. Aalto, T. Holmström, E. Karaharju, J. Joukainen, P. Paavolainen, and P. Slätis,
“Fracture repair during external fixation: Torsion tests of rabbit osteotomies,” Acta
Orthop. Scand., vol. 58, no. 1, pp. 66–70, Jan. 1987.
[39] D. R. Carter, P. R. Blenman, and G. S. Beaupré, “Correlations between mechanical stress
history and tissue differentiation in initial fracture healing,” J. Orthop. Res., vol. 6, no. 5,
pp. 736–748, Sep. 1988.
[40] L. . Claes and C. . Heigele, “Magnitudes of local stress and strain along bony surfaces
predict the course and type of fracture healing,” J. Biomech., vol. 32, no. 3, pp. 255–266,
Mar. 1999.
[41] D. R. Epari, W. R. Taylor, M. O. Heller, and G. N. Duda, “Mechanical conditions in the
initial phase of bone healing.,” Clin. Biomech. (Bristol, Avon), vol. 21, no. 6, pp. 646–55,
Jul. 2006.
[42] M. Mehta, G. N. Duda, C. Perka, and P. Strube, “Influence of gender and fixation stability
on bone defect healing in middle-aged rats: a pilot study.,” Clin. Orthop. Relat. Res., vol.
469, no. 11, pp. 3102–10, Nov. 2011.
[43] P. Klein et al., “The initial phase of fracture healing is specifically sensitive to
mechanical conditions.,” J. Orthop. Res., vol. 21, no. 4, pp. 662–9, 2003.
[44] P. J. Prendergast, R. Huiskes, and K. Søballe, “ESB Research Award 1996. Biophysical
stimuli on cells during tissue differentiation at implant interfaces.,” J. Biomech., vol. 30,
no. 6, pp. 539–48, Jun. 1997.
[45] M. S. Ghiasi, J. Chen, A. Vaziri, E. K. Rodriguez, and A. Nazarian, “Bone fracture healing
in mechanobiological modeling: A review of principles and methods.,” Bone reports,
vol. 6, pp. 87–100, Jun. 2017.
[46] A. E. Goodship, J. L. Cunningham, and J. Kenwright, “Strain rate and timing of
stimulation in mechanical modulation of fracture healing.,” Clin. Orthop. Relat. Res., no.
355 Suppl, pp. S105-15, Oct. 1998.
[47] V. Glatt et al., “Improved Healing of Large Segmental Defects in the Rat Femur by
Reverse Dynamization in the Presence of Bone Morphogenetic Protein-2,” J. Bone Jt.
Surgery-American Vol., vol. 94, no. 22, pp. 2063–2073, Nov. 2012.
[48] N. Bartnikowski et al., “Modulation of fixation stiffness from flexible to stiff in a rat
model of bone healing,” Acta Orthop., vol. 88, no. 2, pp. 217–222, Mar. 2017.
[49] C. B. Howard, O. Elishoov, Y. Matan, R. Mosheiff, and M. Liebergall, “Can changing the
mechanical environment increase the speed of fracture healing? A pilot study in tibial
fractures,” Inj. Extra, vol. 40, no. 10, p. 230, Oct. 2009.
[50] V. Glatt, N. Bartnikowski, N. Quirk, M. Schuetz, and C. Evans, “Reverse Dynamization:
Influence of Fixator Stiffness on the Mode and Efficiency of Large-Bone-Defect Healing
at Different Doses of rhBMP-2.,” J. Bone Joint Surg. Am., vol. 98, no. 8, pp. 677–87, Apr.
2016.
[51] C. Schwarz, D. Wulsten, A. Ellinghaus, J. Lienau, B. M. Willie, and G. N. Duda, “Mechanical
Load Modulates the Stimulatory Effect of BMP2 in a Rat Nonunion Model,” Tissue Eng.
Part A, vol. 19, no. 1–2, pp. 247–254, 2013.
[52] J. D. Boerckel, Y. M. Kolambkar, H. Y. Stevens, A. S. P. Lin, K. M. Dupont, and R. E.
Guldberg, “Effects of in vivo mechanical loading on large bone defect regeneration,” J.
Orthop. Res., vol. 30, no. 7, p. 1067, Jul. 2012.
[53] S. Herberg et al., “Combinatorial morphogenetic and mechanical cues to mimic bone
development for defect repair.,” Sci. Adv., vol. 5, no. 8, p. eaax2476, Aug. 2019.
[54] C. Schwarz et al., “The Interaction of BMP2-Induced Defect Healing in Rat and Fixator
Stiffness Modulates Matrix Alignment and Contraction.,” JBMR plus, vol. 2, no. 3, pp.
174–186, May 2018.
124 Bibliography
[55] R. M. Delaine-Smith and G. C. Reilly, The Effects of Mechanical Loading on Mesenchymal
Stem Cell Differentiation and Matrix Production, 1st ed., vol. 87. Elsevier Inc., 2011.
[56] A. J. Engler, S. Sen, H. L. Sweeney, and D. E. Discher, “Matrix Elasticity Directs Stem Cell
Lineage Specification,” Cell, vol. 126, no. 4, pp. 677–689, 2006.
[57] O. Chaudhuri et al., “Hydrogels with tunable stress relaxation regulate stem cell fate
and activity,” Nat. Mater., vol. 15, no. 3, pp. 326–334, Mar. 2016.
[58] M. Werner et al., “Surface Curvature Differentially Regulates Stem Cell Migration and
Differentiation via Altered Attachment Morphology and Nuclear Deformation.,” Adv.
Sci. (Weinheim, Baden-Wurttemberg, Ger., vol. 4, no. 2, p. 1600347, 2017.
[59] D. E. Ingber, “Cellular mechanotransduction: putting all the pieces together again.,”
FASEB J., vol. 20, no. 7, p. 27, 2006.
[60] M. Barczyk, S. Carracedo, and D. Gullberg, “Integrins.,” Cell Tissue Res., vol. 339, no. 1,
pp. 269–80, Jan. 2010.
[61] R. O. Hynes, “Integrins: bidirectional, allosteric signaling machines.,” Cell, vol. 110, no.
6, pp. 673–87, Sep. 2002.
[62] S. Seetharaman and S. Etienne-Manneville, “Integrin diversity brings specificity in
mechanotransduction,” Biol. Cell, vol. 110, no. 3, pp. 49–64, Mar. 2018.
[63] G. Marsico, L. Russo, F. Quondamatteo, and A. Pandit, “Glycosylation and Integrin
Regulation in Cancer,” Trends in Cancer, vol. 4, no. 8. Cell Press, pp. 537–552, 01-Aug-
2018.
[64] S. Seetharaman and S. Etienne-Manneville, “Integrin diversity brings specificity in
mechanotransduction,” Biol. ot Cell, vol. 110, pp. 49–64, Jun. 2018.
[65] Z. Sun, S. S. Guo, and R. Fässler, “Integrin-mediated mechanotransduction.,” J. Cell Biol.,
vol. 215, no. 4, pp. 445–456, Nov. 2016.
[66] C. K. Choi, M. Vicente-Manzanares, J. Zareno, L. A. Whitmore, A. Mogilner, and A. R.
Horwitz, “Actin and α-actinin orchestrate the assembly and maturation of nascent
adhesions in a myosin II motor-independent manner,” Nat. Cell Biol., vol. 10, no. 9, pp.
1039–1050, 2008.
[67] M. Chrzanowska-Wodnicka and K. Burridge, “Rho-stimulated contractility drives the
formation of stress fibers and focal adhesions.,” J. Cell Biol., vol. 133, no. 6, pp. 1403–
15, Jun. 1996.
[68] A. M. Pasapera, I. C. Schneider, E. Rericha, D. D. Schlaepfer, and C. M. Waterman,
“Myosin II activity regulates vinculin recruitment to focal adhesions through FAK-
mediated paxillin phosphorylation,” J. Cell Biol., vol. 188, no. 6, pp. 877–890, Mar. 2010.
[69] C. J. McCleverty, D. C. Lin, and R. C. Liddington, “Structure of the PTB domain of tensin1
and a model for its recruitment to fibrillar adhesions,” Protein Sci., vol. 16, no. 6, pp.
1223–1229, Jun. 2007.
[70] R. Zaidel-Bar, S. Itzkovitz, A. Ma'ayan, R. Iyengar, and B. Geiger, “Functional atlas
of the integrin adhesome,” Nat. Cell Biol., vol. 9, no. 8, pp. 858–867, Aug. 2007.
[71] E. R. Horton, P. Astudillo, M. J. Humphries, and J. D. Humphries, “Mechanosensitivity of
integrin adhesion complexes: role of the consensus adhesome.,” Exp. Cell Res., vol. 343,
no. 1, pp. 7–13, 2016.
[72] S. K. Mitra, D. A. Hanson, and D. D. Schlaepfer, “Focal adhesion kinase: In command and
control of cell motility,” Nature Reviews Molecular Cell Biology, vol. 6, no. 1. pp. 56–68,
Jan-2005.
[73] D. Chodniewicz and R. L. Klemke, “Regulation of integrin-mediated cellular responses
through assembly of a CAS/Crk scaffold.,” Biochim. Biophys. Acta, vol. 1692, no. 2–3, pp.
63–76, Jul. 2004.
Bibliography 125
[74] R. Zaidel-Bar, R. Milo, Z. Kam, and B. Geiger, “A paxillin tyrosine phosphorylation switch
regulates the assembly and form of cell-matrix adhesions,” J. Cell Sci., vol. 120, no. 1,
pp. 137–148, Dec. 2006.
[75] M. Kanehisa and S. Goto, “KEGG: kyoto encyclopedia of genes and genomes - Release
72.1, December 1, 2014 ,” Nucleic Acids Res, vol. 28, no. 1, pp. 27–30, 2000.
[76] M. Kanehisa, “Toward understanding the origin and evolution of cellular organisms.,”
Protein Sci., Aug. 2019.
[77] M. Kanehisa, Y. Sato, M. Furumichi, K. Morishima, and M. Tanabe, “New approach for
understanding genome variations in KEGG,” Nucleic Acids Res., vol. 47, no. D1, pp.
D590–D595, Jan. 2019.
[78] E. C. Lessey, C. Guilluy, and K. Burridge, “From mechanical force to RhoA activation.,”
Biochemistry, vol. 51, no. 38, pp. 7420–32, Sep. 2012.
[79] C. F. Lee, C. Haase, S. Deguchi, and R. Kaunas, “Cyclic stretch-induced stress fiber
dynamics - Dependence on strain rate, Rho-kinase and MLCK,” Biochem. Biophys. Res.
Commun., vol. 401, no. 3, pp. 344–349, Oct. 2010.
[80] D. Verma, F. Meng, F. Sachs, and S. Z. Hua, “Flow-induced focal adhesion remodeling
mediated by local cytoskeletal stresses and reorganization.,” Cell Adh. Migr., vol. 9, no.
6, pp. 432–40, 2015.
[81] J. T. Parsons, “Focal adhesion kinase: the first ten years.,” J. Cell Sci., vol. 116, no. Pt 8,
pp. 1409–16, Apr. 2003.
[82] H. B. Schiller and R. Fässler, “Mechanosensitivity and compositional dynamics of cell-
matrix adhesions,” EMBO Reports, vol. 14, no. 6. pp. 509–519, Jun-2013.
[83] A. Elosegui-Artola, X. Trepat, and P. Roca-Cusachs, “Control of Mechanotransduction
by Molecular Clutch Dynamics,” Trends in Cell Biology, vol. 28, no. 5. Elsevier Ltd, pp.
356–367, 01-May-2018.
[84] Z.-J. Liu, Y. Zhuge, and O. C. Velazquez, “Trafficking and differentiation of mesenchymal
stem cells,” J. Cell. Biochem., vol. 106, no. 6, pp. 984–991, 2009.
[85] R. Bielby, E. Jones, and D. McGonagle, “The role of mesenchymal stem cells in
maintenance and repair of bone,” Injury, vol. 38, no. 1, pp. S26–S32, 2007.
[86] A. Riedl et al., “Comparison of cancer cells in 2D vs 3D culture reveals differences in
AKT-mTOR-S6K signaling and drug responses,” J. Cell Sci., vol. 130, no. 1, pp. 203–218,
Jan. 2017.
[87] P. Angele et al., “Cyclic, mechanical compression enhances chondrogenesis of
mesenchymal progenitor cells in tissue engineering scaffolds,” Biorheology, vol. 41, no.
3, pp. 335–346, Jan. 2004.
[88] Z. Li, S. J. Yao, M. Alini, and M. J. Stoddart, “Chondrogenesis of human bone marrow
mesenchymal stem cells in fibrin-polyurethane composites is modulated by frequency
and amplitude of dynamic compression and shear stress,” in Tissue Engineering - Part
A, 2010, vol. 16, no. 2, pp. 575–584.
[89] A. H. Huang, M. J. Farrell, M. Kim, and R. L. Mauck, “Long-term dynamic loading
improves the mechanical properties of chondrogenic mesenchymal stem cell-laden
hydrogel.,” Eur. Cell. Mater., vol. 19, pp. 72–85, Feb. 2010.
[90] A. Sittichockechaiwut, A. M. Scutt, A. J. Ryan, L. F. Bonewald, and G. C. Reilly, “Use of
rapidly mineralising osteoblasts and short periods of mechanical loading to accelerate
matrix maturation in 3D scaffolds,” Bone, vol. 44, no. 5, pp. 822–829, 2009.
[91] A. Sittichokechaiwut, J. H. Edwards, a M. Scutt, and G. C. Reilly, “Short bouts of
mechanical loading are as effective as dexamethasone at inducing matrix production
by human bone marrow mesenchymal stem cell.,” Eur. Cell. Mater., vol. 20, pp. 45–57,
2010.
126 Bibliography
[92] M. Jagodzinski et al., “Influence of perfusion and cyclic compression on proliferation
and differentiation of bone marrow stromal cells in 3-dimensional culture,” J. Biomech.,
vol. 41, no. 9, pp. 1885–1891, 2008.
[93] B. Sen, Z. Xie, N. Case, M. Ma, C. Rubin, and J. Rubin, “Mechanical strain inhibits
adipogenesis in mesenchymal stem cells by stimulating a durable β-catenin signal,”
Endocrinology, vol. 149, no. 12, pp. 6065–6075, Dec. 2008.
[94] E. Michalopoulos, R. L. Knight, S. Korossis, J. N. Kearney, J. Fisher, and E. Ingham,
“Development of Methods for Studying the Differentiation of Human Mesenchymal
Stem Cells Under Cyclic Compressive Strain,” Tissue Eng. Part C Methods, vol. 18, no. 4,
pp. 252–262, 2012.
[95] K. Shin Kang, Y. Hun Jeong, J. Min Hong, W.-J. Yong, J.-W. Rhie, and D.-W. Cho, “Flexure-
based device for cyclic strain-mediated osteogenic differentiation.,” J. Biomech. Eng.,
vol. 135, no. 11, p. 114501, 2013.
[96] L. Wang et al., “Involvement of BMPs/smad signaling pathway in mechanical response
in osteoblasts,” Cell. Physiol. Biochem., vol. 26, no. 6, pp. 1093–1102, 2010.
[97] Y. F. Rui, P. P. Y. Lui, M. Ni, L. S. Chan, Y. W. Lee, and K. M. Chan, “Mechanical loading
increased BMP-2 expression which promoted osteogenic differentiation of tendon-
derived stem cells.,” J. Orthop. Res., vol. 29, no. 3, pp. 390–6, Mar. 2011.
[98] K. S. Lee et al., “Runx2 is a common target of transforming growth factor beta1 and
bone morphogenetic protein 2, and cooperation between Runx2 and Smad5 induces
osteoblast-specific gene expression in the pluripotent mesenchymal precursor cell line
C2C12.,” Mol. Cell. Biol., vol. 20, no. 23, pp. 8783–92, Dec. 2000.
[99] C. Sieber, J. Kopf, C. Hiepen, and P. Knaus, “Recent advances in BMP receptor signaling.,”
Cytokine Growth Factor Rev., vol. 20, no. 5–6, pp. 343–55, Jan. 2009.
[100] M. R. Urist and B. S. Strates, “Bone morphogenetic protein.,” J. Dent. Res., vol. 50, no. 6,
pp. 1392–406.
[101] N. Tsumaki et al., “Bone Morphogenetic Protein Signals Are Required for Cartilage
Formation and Differently Regulate Joint Development During Skeletogenesis,” J. Bone
Miner. Res., vol. 17, no. 5, pp. 898–906, May 2002.
[102] A. Stantzou et al., “BMP signaling regulates satellite cell-dependent postnatal muscle
growth,” Development, vol. 144, no. 15, pp. 2737–2747, Aug. 2017.
[103] F. Klatte-Schulz et al., “An investigation of BMP-7 mediated alterations to BMP
signalling components in human tenocyte-like cells.,” Sci. Rep., vol. 6, no. 1, p. 29703,
Sep. 2016.
[104] H. Chen et al., “BMP10 is essential for maintaining cardiac growth during murine
cardiogenesis.,” Development, vol. 131, no. 9, pp. 2219–31, May 2004.
[105] A. García de Vinuesa, S. Abdelilah-Seyfried, P. Knaus, A. Zwijsen, and S. Bailly, “BMP
signaling in vascular biology and dysfunction,” Cytokine Growth Factor Rev., vol. 27, pp.
65–79, Feb. 2016.
[106] S. V. Hegarty, G. W. O’Keeffe, and A. M. Sullivan, “BMP-Smad 1/5/8 signalling in the
development of the nervous system,” Prog. Neurobiol., vol. 109, pp. 28–41, Oct. 2013.
[107] C. Hiepen et al., “BMP2-induced chemotaxis requires PI3K p55γ/p110α-dependent
phosphatidylinositol (3,4,5)-triphosphate production and LL5β recruitment at the
cytocortex,” BMC Biol., vol. 12, no. 1, p. 43, May 2014.
[108] D. O. Wagner, C. Sieber, R. Bhushan, J. H. Börgermann, D. Graf, and P. Knaus, “BMPs:
from bone to body morphogenetic proteins.,” Sci. Signal., vol. 3, no. 107, p. mr1, Feb.
2010.
[109] K. Miyazono, S. Maeda, and T. Imamura, “BMP receptor signaling: Transcriptional
targets, regulation of signals, and signaling cross-talk,” Cytokine Growth Factor Rev.,
Bibliography 127
vol. 16, no. 3, pp. 251–263, Jun. 2005.
[110] V. S. Salazar, L. W. Gamer, and V. Rosen, “BMP signalling in skeletal development,
disease and repair,” Nat. Rev. Endocrinol., vol. 12, no. 4, pp. 203–221, Apr. 2016.
[111] A. Nohe et al., “The Mode of Bone Morphogenetic Protein (BMP) Receptor
Oligomerization Determines Different BMP-2 Signaling Pathways,” J. Biol. Chem., vol.
277, no. 7, pp. 5330–5338, Feb. 2002.
[112] B. M. Chacko et al., “Structural Basis of Heteromeric Smad Protein Assembly in TGF-β
Signaling,” Mol. Cell, vol. 15, no. 5, pp. 813–823, Sep. 2004.
[113] J. Massague, J. Seoane, and D. Wotton, “Smad transcription factors,” Genes Dev., vol. 19,
no. 23, pp. 2783–2810, Dec. 2005.
[114] A. Hollnagel, V. Oehlmann, J. Heymer, U. Rüther, and A. Nordheim, “Id genes are direct
targets of bone morphogenetic protein induction in embryonic stem cells.,” J. Biol.
Chem., vol. 274, no. 28, pp. 19838–45, Jul. 1999.
[115] C.-F. Lai and S.-L. Cheng, “Signal Transductions Induced by Bone Morphogenetic
Protein-2 and Transforming Growth Factor-β in Normal Human Osteoblastic Cells,” J.
Biol. Chem., vol. 277, no. 18, pp. 15514–15522, May 2002.
[116] C. Gamell et al., “BMP2 induction of actin cytoskeleton reorganization and cell
migration requires PI3-kinase and Cdc42 activity,” J. Cell Sci., vol. 121, no. 23, pp. 3960–
3970, Dec. 2008.
[117] J. Kopf, P. Paarmann, C. Hiepen, D. Horbelt, and P. Knaus, “BMP growth factor signaling
in a biomechanical context,” BioFactors, vol. 40, no. 2, pp. 171–187, 2014.
[118] T. D. Mueller and J. Nickel, “Promiscuity and specificity in BMP receptor activation,”
FEBS Letters, vol. 586, no. 14. No longer published by Elsevier, pp. 1846–1859, 04-Jul-
2012.
[119] O. Avsian-Kretchmer and A. J. W. Hsueh, “Comparative Genomic Analysis of the Eight-
Membered Ring Cystine Knot-Containing Bone Morphogenetic Protein Antagonists,”
Mol. Endocrinol., vol. 18, no. 1, pp. 1–12, Jan. 2004.
[120] A. P. Wohl, H. Troilo, R. F. Collins, C. Baldock, and G. Sengle, “Extracellular Regulation
of Bone Morphogenetic Protein Activity by the Microfibril Component Fibrillin-1,” J.
Biol. Chem., vol. 291, no. 24, pp. 12732–12746, Jun. 2016.
[121] G. Murakami, T. Watabe, K. Takaoka, K. Miyazono, and T. Imamura, “Cooperative
Inhibition of Bone Morphogenetic Protein Signaling by Smurf1 and Inhibitory Smads,”
Mol. Biol. Cell, vol. 14, no. 7, pp. 2809–2817, Jul. 2003.
[122] L. C. Fuentealba et al., “Integrating patterning signals: Wnt/GSK3 regulates the
duration of the BMP/Smad1 signal.,” Cell, vol. 131, no. 5, pp. 980–93, Nov. 2007.
[123] B. Rath et al., “Biomechanical forces exert anabolic effects on osteoblasts by activation
of SMAD 1/5/8 through type 1 BMP receptor,” Biorheology, vol. 48, pp. 37–48, 2011.
[124] J. Zhou et al., “Force-specific activation of Smad1 / 5 regulates vascular endothelial cell
cycle progression in response to disturbed flow,” Proc., vol. 109, pp. 1–6, 2012.
[125] J. Kopf, A. Petersen, G. N. Duda, and P. Knaus, “BMP2 and mechanical loading
cooperatively regulate immediate early signalling events in the BMP pathway,” BMC
Biol., vol. 10, no. 1, p. 37, 2012.
[126] Z. Zeng, X. Yin, X. Zhang, D. Jing, and X. Feng, “Cyclic stretch enhances bone
morphogenetic protein-2-induced osteoblastic differentiation through the inhibition
of Hey1.,” Int. J. Mol. Med., vol. 36, no. 5, pp. 1273–81, Nov. 2015.
[127] S. Kido et al., “Mechanical Stress Activates Smad Pathway through PKCδ to Enhance
Interleukin-11 Gene Transcription in Osteoblasts,” PLoS One, vol. 5, no. 9, p. e13090,
2010.
128 Bibliography
[128] J. Zhou et al., “BMP receptor-integrin interaction mediates responses of vascular
endothelial Smad1/5 and proliferation to disturbed flow,” J. Thromb. Haemost., vol. 11,
no. 4, pp. 741–755, 2013.
[129] L. Fourel et al., “β3 integrin-mediated spreading induced by matrix-bound BMP-2
controls Smad signaling in a stiffness-independent manner.,” J. Cell Biol., vol. 212, no.
6, pp. 693–706, Mar. 2016.
[130] C.-F. Lai and S.-L. Cheng, “αvβ Integrins Play an Essential Role in BMP-2 Induction of
Osteoblast Differentiation,” J. Bone Miner. Res., vol. 20, no. 2, pp. 330–340, 2004.
[131] A. Jikko, S. E. Harris, D. Chen, D. L. Mendrick, and C. H. Damsky, “Collagen Integrin
Receptors Regulate Early Osteoblast Differentiation Induced by BMP-2,” J. Bone Miner.
Res., vol. 14, no. 7, pp. 1075–1083, Jul. 1999.
[132] M. Brunner et al., “β1 integrins mediate the BMP2 dependent transcriptional control of
osteoblast differentiation and osteogenesis.,” PLoS One, vol. 13, no. 4, p. e0196021,
2018.
[133] A. Kotzsch et al., “Structure Analysis of Bone Morphogenetic Protein-2 Type I Receptor
Complexes Reveals a Mechanism of Receptor Inactivation in Juvenile Polyposis
Syndrome,” J. Biol. Chem., vol. 283, no. 9, pp. 5876–5887, Feb. 2008.
[134] Y. Zu, X. Liang, J. Du, S. Zhou, and C. Yang, “Binding of integrin α1 to bone morphogenetic
protein receptor IA suggests a novel role of integrin α1β1 in bone morphogenetic
protein 2 signalling,” J. Biomech., vol. 48, no. 14, pp. 3950–3954, Nov. 2015.
[135] H. A. North, L. Pan, T. L. McGuire, S. Brooker, and J. A. Kessler, “β1-Integrin Alters
Ependymal Stem Cell BMP Receptor Localization and Attenuates Astrogliosis after
Spinal Cord Injury,” J. Neurosci., vol. 35, no. 9, pp. 3725–3733, Mar. 2015.
[136] J. Du et al., “Integrin activation and internalization on soft ECM as a mechanism of
induction of stem cell differentiation by ECM elasticity.,” Proc. Natl. Acad. Sci. U. S. A.,
vol. 108, no. 23, pp. 9466–71, Jun. 2011.
[137] M. Delcommenne, C. Tan, V. Gray, L. Rue, J. Woodgett, and S. Dedhar,
“Phosphoinositide-3-OH kinase-dependent regulation of glycogen synthase kinase 3
and protein kinase B/AKT by the integrin-linked kinase.,” Proc. Natl. Acad. Sci. U. S. A.,
vol. 95, no. 19, pp. 11211–6, Sep. 1998.
[138] C. Frantz, K. M. Stewart, and V. M. Weaver, “The extracellular matrix at a glance,”
Journal of Cell Science, vol. 123, no. 24. pp. 4195–4200, 15-Dec-2010.
[139] S. Ricard-Blum, “The Collagen Family,” Cold Spring Harb. Perspect. Biol., vol. 3, no. 1, pp.
1–19, Jan. 2011.
[140] J. Bella and D. J. S. Hulmes, “Fibrillar collagens,” Subcell. Biochem., vol. 82, pp. 457–490,
Jan. 2017.
[141] H. NAGASE, R. VISSE, and G. MURPHY, “Structure and function of matrix
metalloproteinases and TIMPs.,” Cardiovasc. Res., vol. 69, no. 3, pp. 562–573, 2006.
[142] D. Kessler et al., “Fibroblasts in Mechanically Stressed Collagen Lattices Assume a
‘Synthetic’ Phenotype,” J. Biol. Chem., vol. 276, no. 39, pp. 36575–36585, Sep. 2001.
[143] V. Dumas et al., “The effect of dual frequency cyclic compression on matrix deposition
by osteoblast-like cells grown in 3D scaffolds and on modulation of VEGF variant
expression,” Biomaterials, vol. 30, no. 19, pp. 3279–3288, Jul. 2009.
[144] Y. He, E. J. Macarak, J. M. Korostoff, and P. S. Howard, “Compression and tension:
Differential effects on matrix accumulation by periodontal ligaments fibroblasts in
vitro,” Connect. Tissue Res., vol. 45, no. 1, pp. 28–39, 2004.
[145] J. D. Glaeser et al., “Modulation of matrix metalloprotease-2 levels by mechanical
loading of three-dimensional mesenchymal stem cell constructs: Impact on in vitro
tube formation,” Tissue Eng. - Part A, vol. 16, no. 10, pp. 3139–3148, Oct. 2010.
Bibliography 129
[146] A. Petersen, P. Joly, C. Bergmann, G. Korus, and G. N. Duda, “The Impact of Substrate
Stiffness and Mechanical Loading on Fibroblast-Induced Scaffold Remodeling,” Tissue
Eng. Part A, vol. 18, no. 17–18, pp. 1804–1817, Sep. 2012.
[147] J. N. A. De Croos, S. S. Dhaliwal, M. D. Grynpas, R. M. Pilliar, and R. A. Kandel, “Cyclic
compressive mechanical stimulation induces sequential catabolic and anabolic gene
changes in chondrocytes resulting in increased extracellular matrix accumulation.,”
Matrix Biol., vol. 25, no. 6, pp. 323–31, Aug. 2006.
[148] S. D. Waldman, D. C. Couto, M. D. Grynpas, R. M. Pilliar, and R. A. Kandel, “A single
application of cyclic loading can accelerate matrix deposition and enhance the
properties of tissue-engineered cartilage.,” Osteoarthr. Cartil., vol. 14, no. 4, pp. 323–
30, Apr. 2006.
[149] E. N. Blaney Davidson, E. L. Vitters, P. L. E. M. van Lent, F. A. J. van de Loo, W. B. van den
Berg, and P. M. van der Kraan, “Elevated extracellular matrix production and
degradation upon bone morphogenetic protein-2 (BMP-2) stimulation point toward a
role for BMP-2 in cartilage repair and remodeling.,” Arthritis Res. Ther., vol. 9, no. 5, p.
R102, Jan. 2007.
[150] J. Li, K. S. Kim, J. S. Park, W. A. Elmer, W. C. Hutton, and S. T. Yoon, “BMP-2 and CDMP-
2: Stimulation of chondrocyte production of proteoglycan,” J. Orthop. Sci., vol. 8, no. 6,
pp. 829–835, 2003.
[151] D. Chen et al., “Bone morphogenetic protein 2 (BMP-2) enhances BMP-3, BMP-4, and
bone cell differentiation marker gene expression during the induction of mineralized
bone matrix formation in cultures of fetal rat calvarial osteoblasts.,” Calcif. Tissue Int.,
vol. 60, no. 3, pp. 283–90, Mar. 1997.
[152] A. Marupanthorn, C. Tantrawatpan, P. Kheolamai, D. Tantikanlayaporn, and S.
Manochantr, “Bone morphogenetic protein-2 enhances the osteogenic differentiation
capacity of mesenchymal stromal cells derived from human bone marrow and
umbilical cord,” Int. J. Mol. Med., vol. 39, no. 3, pp. 654–662, Mar. 2017.
[153] F. Lecanda, L. V Avioli, and S. L. Cheng, “Regulation of bone matrix protein expression
and induction of differentiation of human osteoblasts and human bone marrow
stromal cells by bone morphogenetic protein-2.,” J. Cell. Biochem., vol. 67, no. 3, pp.
386–96, Dec. 1997.
[154] P. Owens et al., “Bone Morphogenetic Proteins Stimulate Mammary Fibroblasts to
Promote Mammary Carcinoma Cell Invasion,” PLoS One, vol. 8, no. 6, Jun. 2013.
[155] J. Li, S. T. Yoon, and W. C. Hutton, “Effect of bone morphogenetic protein-2 (BMP-2) on
matrix production, other BMPs, and BMP receptors in rat intervertebral disc cells.,” J.
Spinal Disord. Tech., vol. 17, no. 5, pp. 423–8, Oct. 2004.
[156] Y.-K. Wang et al., “Bone morphogenetic protein-2-induced signaling and osteogenesis
is regulated by cell shape, RhoA/ROCK, and cytoskeletal tension.,” Stem Cells Dev., vol.
21, no. 7, pp. 1176–86, May 2012.
[157] M. Werner, N. A. Kurniawan, G. Korus, C. V. C. Bouten, and A. Petersen, “Mesoscale
substrate curvature overrules nanoscale contact guidance to direct bone marrow
stromal cell migration.,” J. R. Soc. Interface, vol. 15, no. 145, 2018.
[158] H. Schoof, J. Apel, I. Heschel, and G. Rau, “Control of pore structure and size in freeze-
dried collagen sponges.,” J. Biomed. Mater. Res., vol. 58, no. 4, pp. 352–7, 2001.
[159] Petersen, Joly, Bergmann, Korus, and Duda, “The Impact of Substrate Stiffness and
Mechanical Loading on Fibroblast-Induced Scaffold Remodeling,” Tissue Eng. Part A,
vol. 18, no. 17–18, pp. 1804–1817, 2012.
[160] J. Riedl et al., “Lifeact: a versatile marker to visualize F-actin,” Nat. Methods, vol. 5, no.
7, pp. 605–607, Jul. 2008.
130 Bibliography
[161] M. W. Pfaffl, “A new mathematical model for relative quantification in real-time RT-
PCR.,” Nucleic Acids Res., vol. 29, no. 9, p. e45, 2001.
[162] C. Ramakers, J. M. Ruijter, R. H. L. Deprez, and A. F. M. Moorman, “Assumption-free
analysis of quantitative real-time polymerase chain reaction (PCR) data.,” Neurosci.
Lett., vol. 339, no. 1, pp. 62–6, Mar. 2003.
[163] B. Kappler et al., “The cytoprotective capacity of processed human cardiac extracellular
matrix.,” J. Mater. Sci. Mater. Med., vol. 27, no. 7, p. 120, Jul. 2016.
[164] X. Chen, O. Nadiarynkh, S. Plotnikov, and P. J. Campagnola, “Second harmonic
generation microscopy for quantitative analysis of collagen fibrillar structure,” Nat.
Protoc., vol. 7, no. 4, pp. 654–669, Apr. 2012.
[165] R. Rezakhaniha et al., “Experimental investigation of collagen waviness and orientation
in the arterial adventitia using confocal laser scanning microscopy,” Biomech. Model.
Mechanobiol., vol. 11, no. 3–4, pp. 461–473, Mar. 2012.
[166] J.-Y. Tinevez et al., “TrackMate: An open and extensible platform for single-particle
tracking.,” Methods, vol. 115, pp. 80–90, Feb. 2017.
[167] P. Baniukiewicz, S. Collier, and T. Bretschneider, “QuimP: analyzing transmembrane
signalling in highly deformable cells.,” Bioinformatics, vol. 34, no. 15, pp. 2695–2697,
Aug. 2018.
[168] M. Becker et al., “Towards a Novel Patch Material for Cardiac Applications: Tissue-
Specific Extracellular Matrix Introduces Essential Key Features to Decellularized
Amniotic Membrane.,” Int. J. Mol. Sci., vol. 19, no. 4, p. 1032, Mar. 2018.
[169] B. Ma et al., “PEAKS: powerful software for peptide de novo sequencing by tandem
mass spectrometry.,” Rapid Commun. Mass Spectrom., vol. 17, no. 20, pp. 2337–42, Oct.
2003.
[170] K. H. Vining and D. J. Mooney, “Mechanical forces direct stem cell behaviour in
development and regeneration,” Nature Reviews Molecular Cell Biology, vol. 18, no. 12.
Nature Publishing Group, pp. 728–742, 01-Dec-2017.
[171] E. Brauer et al., “Collagen Fibrils Mechanically Contribute to Tissue Contraction in an
In Vitro Wound Healing Scenario,” Adv. Sci., vol. 6, no. 9, p. 1801780, May 2019.
[172] S. Schreivogel, V. Kuchibhotla, P. Knaus, G. N. Duda, and A. Petersen, “Load-induced
osteogenic differentiation of mesenchymal stromal cells is caused by mechano-
regulated autocrine signaling,” J. Tissue Eng. Regen. Med., Aug. 2019.
[173] F. Witt, G. N. Duda, C. Bergmann, and A. Petersen, “Cyclic Mechanical Loading Enables
Solute Transport and Oxygen Supply in Bone Healing: An In Vitro Investigation,” Tissue
Eng. Part A, p. 140120073517006, Jan. 2014.
[174] T. Katagiri et al., “Bone morphogenetic protein-2 converts the differentiation pathway
of C2C12 myoblasts into the osteoblast lineage.,” J. Cell Biol., vol. 127, no. 6 Pt 1, pp.
1755–66, Dec. 1994.
[175] K. Yang et al., “BMPR2-pSMAD1/5 signaling pathway regulates RUNX2 expression and
impacts the progression of dedifferentiated chondrosarcoma.,” Am. J. Cancer Res., vol.
6, no. 6, pp. 1302–16, Jan. 2016.
[176] R. Nishimura, K. Hata, T. Matsubara, M. Wakabayashi, and T. Yoneda, “Regulation of
bone and cartilage development by network between BMP signalling and transcription
factors.,” J. Biochem., vol. 151, no. 3, pp. 247–54, Mar. 2012.
[177] C. Krause, A. Guzman, and P. Knaus, “Noggin,” Int. J. Biochem. Cell Biol., vol. 43, no. 4, pp.
478–481, Apr. 2011.
[178] I. S. Kim, Y. M. Song, T. H. Cho, J. Y. Kim, F. E. Weber, and S. J. Hwang, “Synergistic action
of static stretching and BMP-2 stimulation in the osteoblast differentiation of C2C12
myoblasts.,” J. Biomech., vol. 42, no. 16, pp. 2721–7, 2009.
Bibliography 131
[179] L. Wang et al., “Involvement of p38MAPK/NF-κB Signaling Pathways in Osteoblasts
Differentiation in Response to Mechanical Stretch,” Ann. Biomed. Eng., vol. 40, no. 9, pp.
1884–1894, Sep. 2012.
[180] L. Wang et al., “Involvement of BMPs/Smad signaling pathway in mechanical response
in osteoblasts.,” Cell. Physiol. Biochem., vol. 26, no. 6, pp. 1093–102, Jan. 2010.
[181] M. A. Peake and A. J. El Haj, “Preliminary characterisation of mechanoresponsive
regions of the c-fos promoter in bone cells,” FEBS Lett., vol. 537, no. 1–3, pp. 117–120,
Feb. 2003.
[182] V. Kanoldt, L. Fischer, and C. Grashoff, “Unforgettable force – crosstalk and memory of
mechanosensitive structures,” Biol. Chem., vol. 400, no. 6, pp. 687–698, Jun. 2019.
[183] J. S. Bennett, “Structure and function of the platelet integrin αIIbβ3,” J. Clin. Invest., vol.
115, no. 12, p. 3363, Dec. 2005.
[184] M. R. Bubb, I. Spector, B. B. Beyer, and K. M. Fosen, “Effects of jasplakinolide on the
kinetics of actin polymerization. An explanation for certain in vivo observations.,” J.
Biol. Chem., vol. 275, no. 7, pp. 5163–70, Feb. 2000.
[185] D. P. Berry, K. G. Harding, M. R. Stanton, B. Jasani, and H. P. Ehrlich, “Human wound
contraction: Collagen organization, fibroblasts, and myofibroblasts,” Plast. Reconstr.
Surg., vol. 102, no. 1, pp. 124–134, Jul. 1998.
[186] K. Runager, J. J. Enghild, and G. K. Klintworth, “Focus on molecules: Transforming
growth factor beta induced protein (TGFBIp),” Experimental Eye Research, vol. 87, no.
4. pp. 298–299, Oct-2008.
[187] E. P. Sproul and W. S. Argraves, “A cytokine axis regulates elastin formation and
degradation,” Matrix Biology, vol. 32, no. 2. pp. 86–94, 11-Mar-2013.
[188] O. C. and D. J. Mooney, “Chaudhuri, Mooney - 2012 - Stem-cell differentiation Anchoring
cell-fate cues.pdf,” Nat. Mater., vol. 11, pp. 568–569, 2012.
[189] M. Bottlang and F. Feist, “Biomechanics of Far Cortical Locking,” J. Orthop. Trauma, vol.
25, pp. S21–S28, Feb. 2011.
[190] B. S. Klosterhoff et al., “Wireless Implantable Sensor for Noninvasive, Longitudinal
Quantification of Axial Strain Across Rodent Long Bone Defects,” J. Biomech. Eng., vol.
139, no. 11, p. 111004, Oct. 2017.
[191] C. Liu et al., “Influence of perfusion and compression on the proliferation and
differentiation of bone mesenchymal stromal cells seeded on polyurethane scaffolds,”
Biomaterials, vol. 33, no. 4, pp. 1052–1064, 2012.
[192] M. Morlock et al., “Duration and frequency of every day activities in total hip patients.,”
J. Biomech., vol. 34, no. 7, pp. 873–81, Jul. 2001.
[193] P. Ducy, R. Zhang, V. Geoffroy, A. L. Ridall, and G. Karsenty, “Osf2/Cbfa1: a
transcriptional activator of osteoblast differentiation.,” Cell, vol. 89, no. 5, pp. 747–54,
May 1997.
[194] R. T. Franceschi and G. Xiao, “Regulation of the osteoblast-specific transcription factor,
Runx2: Responsiveness to multiple signal transduction pathways,” J. Cell. Biochem., vol.
88, no. 3, pp. 446–454, 2003.
[195] Y. Shi et al., “Continuous cyclic mechanical tension inhibited Runx2 expression in
mesenchymal stem cells through RhoA-ERK1/2 pathway,” J. Cell. Physiol., vol. 226, no.
8, pp. 2159–2169, Aug. 2011.
[196] J. Klein-Nulend, J. Roelofsen, C. M. Semeins, A. L. Bronckers, and E. H. Burger,
“Mechanical stimulation of osteopontin mRNA expression and synthesis in bone cell
cultures.,” J. Cell. Physiol., vol. 170, no. 2, pp. 174–81, Feb. 1997.
[197] C. D. Toma, S. Ashkar, M. L. Gray, J. L. Schaffer, and L. C. Gerstenfeld, “Signal transduction
132 Bibliography
of mechanical stimuli is dependent on microfilament integrity: identification of
osteopontin as a mechanically induced gene in osteoblasts.,” J. Bone Miner. Res., vol. 12,
no. 10, pp. 1626–36, Oct. 1997.
[198] R. D. Sumanasinghe, S. H. Bernacki, and E. G. Loboa, “Osteogenic differentiation of
human mesenchymal stem cells in collagen matrices: effect of uniaxial cyclic tensile
strain on bone morphogenetic protein (BMP-2) mRNA expression.,” Tissue Eng., vol.
12, no. 12, pp. 3459–65, Dec. 2006.
[199] D. Wulsten et al., “Time kinetics of bone defect healing in response to BMP-2 and GDF-
5 characterised by in vivo biomechanics.,” Eur. Cell. Mater., vol. 21, pp. 177–92, Feb.
2011.
[200] M.-H. Lee et al., “BMP-2-induced Runx2 expression is mediated by Dlx5, and TGF-beta
1 opposes the BMP-2-induced osteoblast differentiation by suppression of Dlx5
expression.,” J. Biol. Chem., vol. 278, no. 36, pp. 34387–94, Sep. 2003.
[201] R. P. Huang, C. T. Rubin, and K. J. McLeod, “Changes in postural muscle dynamics as a
function of age.,” J. Gerontol. A. Biol. Sci. Med. Sci., vol. 54, no. 8, pp. B352-7, Aug. 1999.
[202] S. R. L. Young, R. Gerard-O’Riley, J. B. Kim, and F. M. Pavalko, “Focal adhesion kinase is
important for fluid shear stress-induced mechanotransduction in osteoblasts,” J. Bone
Miner. Res., vol. 24, no. 3, pp. 411–424, Mar. 2009.
[203] H.-J. Hsu, C.-F. Lee, and R. Kaunas, “A Dynamic Stochastic Model of Frequency-
Dependent Stress Fiber Alignment Induced by Cyclic Stretch,” PLoS One, vol. 4, no. 3, p.
e4853, Mar. 2009.
[204] M. J. Qu, B. Liu, H. Q. Wang, Z. Q. Yan, B. R. Shen, and Z. L. Jiang, “Frequency-dependent
phenotype modulation of vascular smooth muscle cells under cyclic mechanical
strain,” J. Vasc. Res., vol. 44, no. 5, pp. 345–353, Aug. 2007.
[205] M. E. Goldschmidt, K. J. McLeod, and W. R. Taylor, “Integrin-mediated
mechanotransduction in vascular smooth muscle cells: frequency and force response
characteristics.,” Circ. Res., vol. 88, no. 7, pp. 674–80, Apr. 2001.
[206] C. H. Turner, M. R. Forwood, and M. W. Otter, “Mechanotransduction in bone: Do bone
cells act as sensors of fluid flow?,” FASEB J., vol. 8, no. 11, pp. 875–878, 1994.
[207] Y. F. Hsieh and C. H. Turner, “Effects of loading frequency on mechanically induced
bone formation,” J. Bone Miner. Res., vol. 16, no. 5, pp. 918–924, 2001.
[208] K. Marycz, D. Lewandowski, K. A. Tomaszewski, B. M. Henry, E. B. Golec, and M.
Marȩdziak, “Low-frequency, low-magnitude vibrations (LFLM) enhances
chondrogenic differentiation potential of human adipose derived mesenchymal
stromal stem cells (hASCs),” PeerJ, vol. 2016, no. 2, 2016.
[209] R. Qin et al., “Phosphorylation and turnover of paxillin in focal contacts is controlled by
force and defines the dynamic state of the adhesion site.,” Cytoskeleton (Hoboken)., vol.
72, no. 2, pp. 101–12, Feb. 2015.
[210] D. Choquet, D. P. Felsenfeld, and M. P. Sheetz, “Extracellular Matrix Rigidity Causes
Strengthening of Integrin–Cytoskeleton Linkages,” Cell, vol. 88, no. 1, pp. 39–48, Jan.
1997.
[211] D. Riveline et al., “Focal Contacts as Mechanosensors,” J. Cell Biol., vol. 153, no. 6, pp.
1175–1186, Jun. 2001.
[212] C. Yang, M. W. Tibbitt, L. Basta, and K. S. Anseth, “Mechanical memory and dosing
influence stem cell fate,” Nat. Mater., vol. 13, no. 6, pp. 645–652, Jun. 2014.
[213] Y. Zu, X. Liang, J. Du, S. Zhou, and C. Yang, “Binding of integrin α1 to bone morphogenetic
protein receptor IA suggests a novel role of integrin α1β1 in bone morphogenetic
protein 2 signalling,” J. Biomech., vol. 48, no. 14, pp. 3950–3954, Nov. 2015.
[214] H. L. Ashe, “Modulation of BMP signalling by integrins.,” Biochem. Soc. Trans., vol. 44,
Bibliography 133
no. 5, pp. 1465–1473, Oct. 2016.
[215] K. Haase, Z. Al-Rekabi, and A. E. Pelling, “Mechanical cues direct focal adhesion
dynamics,” in Progress in Molecular Biology and Translational Science, vol. 126, Elsevier
B.V., 2014, pp. 103–134.
[216] E. A. Osborn, A. Rabodzey, C. F. Dewey, and J. H. Hartwig, “Endothelial actin
cytoskeleton remodeling during mechanostimulation with fluid shear stress,” Am. J.
Physiol. Physiol., vol. 290, no. 2, pp. C444–C452, Feb. 2006.
[217] L. P. Cramer, “Role of actin-filament disassembly in lamellipodium protrusion in motile
cells revealed using the drug jasplakinolide,” Curr. Biol., vol. 9, no. 19, pp. 1095–1105,
Oct. 1999.
[218] X. Zhang et al., “Actin stabilization by jasplakinolide affects the function of bone
marrow-derived late endothelial progenitor cells.,” PLoS One, vol. 7, no. 11, p. e50899,
Jan. 2012.
[219] W. C. Chen et al., “BMP-2 induces angiogenesis by provoking integrin α6 expression in
human endothelial progenitor cells,” Biochem. Pharmacol., vol. 150, pp. 256–266, Apr.
2018.
[220] D. Könnig, A. Herrera, G. N. Duda, and A. Petersen, “Mechanosensation across borders:
fibroblasts inside a macroporous scaffold sense and respond to the mechanical
environment beyond the scaffold walls,” J. Tissue Eng. Regen. Med., vol. 12, no. 1, pp.
265–275, Jan. 2018.
[221] R. Rudolph, “Contraction and the control of contraction.,” World J. Surg., vol. 4, no. 3,
pp. 279–87, May 1980.
[222] G. Sedlmeier and J. P. Sleeman, “Extracellular regulation of BMP signaling: welcome to
the matrix.,” Biochem. Soc. Trans., vol. 45, no. 1, pp. 173–181, Feb. 2017.
[223] N. Ghosh-Choudhury, S. L. Abboud, B. Chandrasekar, and G. Ghosh Choudhury, “BMP-2
regulates cardiomyocyte contractility in a phosphatidylinositol 3 kinase-dependent
manner,” FEBS Lett., vol. 544, no. 1–3, pp. 181–184, Jun. 2003.
[224] T. Rothhammer, S. Braig, and A. K. Bosserhoff, “Bone morphogenetic proteins induce
expression of metalloproteinases in melanoma cells and fibroblasts,” Eur. J. Cancer, vol.
44, no. 16, pp. 2526–2534, Nov. 2008.
[225] R. Visse and H. Nagase, “Matrix metalloproteinases and tissue inhibitors of
metalloproteinases: Structure, function, and biochemistry,” Circulation Research, vol.
92, no. 8. pp. 827–839, 02-May-2003.
[226] C. Cangemi, M. L. Hansen, W. S. Argraves, and L. M. Rasmussen, “Fibulins and their role
in cardiovascular biology and disease,” in Advances in Clinical Chemistry, vol. 67,
Academic Press Inc., 2014, pp. 245–265.
[227] M. A. Cooley et al., “Fibulin-1 is required for bone formation and Bmp-2-mediated
induction of Osterix,” Bone, vol. 69, pp. 30–38, Sep. 2014.
[228] M. C. Sutcliffe and J. M. Davidson, “Effect of Static Stretching on Elastin Production by
Porcine Aortic Smooth Muscle Cells,” Matrix, vol. 10, no. 3, pp. 148–153, 1990.
[229] P. S. Howard, U. Kucich, R. Taliwal, and J. M. Korostoff, “Mechanical forces alter
extracellular matrix synthesis by human periodontal ligament fibroblasts,” J.
Periodontal Res., vol. 33, no. 8, pp. 500–508, Jun. 1998.
[230] B. Hinz, “Formation and function of the myofibroblast during tissue repair,” Journal of
Investigative Dermatology, vol. 127, no. 3. Nature Publishing Group, pp. 526–537, 2007.
[231] J. P. E. Junker, C. Kratz, A. Tollbäck, and G. Kratz, “Mechanical tension stimulates the
transdifferentiation of fibroblasts into myofibroblasts in human burn scars,” Burns, vol.
34, no. 7, pp. 942–946, Nov. 2008.
134 Bibliography
[232] I. Kii and H. Ito, “Periostin and its interacting proteins in the construction of
extracellular architectures,” Cellular and Molecular Life Sciences, vol. 74, no. 23.
Birkhauser Verlag AG, pp. 4269–4277, 01-Dec-2017.
[233] M. I. Uzel et al., “Multiple Bone Morphogenetic Protein 1-related Mammalian
Metalloproteinases Process Pro-lysyl Oxidase at the Correct Physiological Site and
Control Lysyl Oxidase Activation in Mouse Embryo Fibroblast Cultures,” J. Biol. Chem.,
vol. 276, no. 25, pp. 22537–22543, Jun. 2001.
[234] T. Maruhashi, I. Kii, M. Saito, and A. Kudo, “Interaction between periostin and BMP-1
promotes proteolytic activation of lysyl oxidase,” J. Biol. Chem., vol. 285, no. 17, pp.
13294–13303, Apr. 2010.
[235] B. Merle and P. Garnero, “The multiple facets of periostin in bone metabolism,”
Osteoporosis International, vol. 23, no. 4. Springer-Verlag, pp. 1199–1212, Apr-2012.
[236] I. Kii et al., “Incorporation of tenascin-C into the extracellular matrix by periostin
underlies an extracellular meshwork architecture,” J. Biol. Chem., vol. 285, no. 3, pp.
2028–2039, Jan. 2010.
[237] K. Sato, O. Matsubara, E. Hase, T. Minamikawa, and T. Yasui, “Quantitative in situ time-
series evaluation of osteoblastic collagen synthesis under cyclic strain using second-
harmonic-generation microscopy,” J. Biomed. Opt., vol. 24, no. 03, p. 1, Jan. 2019.
[238] C. Aparicio and M. P. Ginebra, Biomineralization and Biomaterials: Fundamentals and
Applications. 2015.
[239] J. Chen et al., “Therapeutic effects of whole-body vibration on fracture healing in
ovariectomized rats: A systematic review and meta-analysis,” Menopause, vol. 26, no.
6. Lippincott Williams and Wilkins, pp. 677–686, 01-Jun-2019.
[240] R. Uchida et al., “Vibration acceleration promotes bone formation in rodent models,”
PLoS One, vol. 12, no. 3, p. e0172614, Mar. 2017.
Supplement 135
Supplement
Supplement to section 1.3 of the introduction:
Figure 0-1 Pubmed search term statistics. Number of publications related to growth factor signaling (gray), cellular
biomechanics (blue) or mechanotransduction (orange) listed per year. Since a long time, biochemical cues are
recognized as important regulators of cellular behavior while the field of biomechanics only emerged around the year
2000. Biomechanics has gained increasing attention until now but is still underrepresented.
Supplement to section 4.1.4 of the results:
Figure 0-2: Gene expression changes of TGFβ1, TGFβ3, FGF2, PDGF-A and VEGF-A, growth factors that influence
osteogenic differentiation. Under Rhigh conditions, only TGFβ3 expression is slightly increased in response to cyclic
compression. Human BMSCs were seeded in collagen scaffolds (12.4 kPa) and stimulated with and without cyclic
compression (f=1Hz, ε=10%) under increased cell-to-medium ratio or under concurrent rhBMP2 (5nM) stimulation.
Fold change gene expression was analyzed by qPCR in comparison to the untreated control (low cell-to-medium ratio,
without cyclic compression). HPRT1 was used as the reference gene. Box and whisker plots showing the following
values from bottom to top: minimum value, low quartile, median, upper quartile, maximum value, □ mean; × outlier
(n=5 hBMSC donors). Statistical significance was tested via Mann–Whitney test (two-sided) with Bonferroni
correction.
0
2500
5000
7500
10000
12500
15000
17500
20000
22500
number of publications
year
Pubmed statistic
growth factor signaling cellular biomechanics mechanotransduction
136 Supplement
Figure 0-3: BMP2 stability during bioreactor culture. Bioreactor experiment without cultivation of cells. rhBMP2
(135ng/ml) was added at day 0 and medium samples were taken after 30min (0d), on day 1, 3, 5 and 7 to measure
the BMP2 concentration by ELISA. Expansion medium without additional supplements served as a control (n=2).
Supplement to section 4.2.2 of the results:
Table 0-1: Fold change (F.I.) gene expression in response to 24h BMP-2 stimulation and/or mechanical loading of 1
Hz or 10 Hz. The heat map in Figure 4-8 is based on the log2(F.I.), n=4.
F.I.
c
B
L
1Hz
B/L
1Hz
L
10Hz
B/L
10Hz
log2(F.I.)
c
B
L
1Hz
B/L
1Hz
L
10Hz
B/L
10Hz
ID1
1
1.36
0.91
1.76
1.19
3.92
ID1
0
0.44
-0.13
0.81
0.25
1.97
ID2
1
1.16
0.98
1.43
1.19
2.06
ID2
0
0.21
-0.02
0.51
0.25
1.05
DLX2
1
1.29
1.19
1.24
1.47
1.86
DLX2
0
0.37
0.25
0.31
0.55
0.89
SMAD7
1
1.21
1.16
1.50
1.72
1.97
SMAD7
0
0.27
0.21
0.58
0.78
0.98
Noggin
1
1.18
1.13
1.42
1.31
2.87
Noggin
0
0.24
0.18
0.51
0.39
1.52
Smurf1
1
1.03
1.13
1.10
1.36
1.22
Smurf1
0
0.04
0.18
0.13
0.45
0.28
Smurf2
1
1.14
1.31
1.30
1.36
1.33
Smurf2
0
0.19
0.39
0.38
0.44
0.41
BR1A
1
1.08
1.14
1.15
1.21
1.06
BR1A
0
0.11
0.19
0.20
0.27
0.09
BR1B
1
0.96
1.20
1.41
1.74
1.62
BR1B
0
-0.06
0.27
0.50
0.80
0.70
BR2
1
1.05
1.23
1.22
1.22
1.10
BR2
0
0.07
0.30
0.28
0.29
0.13
c-fos
1
0.87
1.28
1.28
1.85
1.90
c-fos
0
-0.20
0.36
0.35
0.89
0.93
RUNX2
1
1.03
1.13
1.05
1.02
0.93
RUNX2
0
0.04
0.17
0.08
0.03
-0.11
SPP1
1
0.81
0.87
0.87
1.99
1.77
SPP1
0
-0.30
-0.20
-0.20
0.99
0.83
COL1A2
1
1.34
1.02
1.06
0.95
1.15
COL1A2
0
0.42
0.03
0.09
-0.08
0.20
RhoA
1
0.99
1.05
1.07
1.30
1.17
RhoA
0
-0.01
0.08
0.09
0.37
0.23
ROCK2
1
1.06
1.13
1.26
1.30
1.22
ROCK2
0
0.08
0.17
0.34
0.38
0.29
ITG a1
1
1.13
1.11
1.10
1.28
1.17
ITG a1
0
0.17
0.16
0.13
0.36
0.22
ITG a5
1
1.00
1.09
0.98
1.16
1.11
ITG a5
0
-0.01
0.13
-0.03
0.22
0.15
ITG av
1
0.98
1.06
1.14
1.46
1.31
ITG av
0
-0.03
0.09
0.19
0.55
0.39
ITG b1
1
1.16
1.32
1.31
1.42
1.32
ITG b1
0
0.21
0.40
0.39
0.50
0.40
ITG b3
1
1.03
1.26
1.85
1.57
1.92
ITG b3
0
0.04
0.34
0.89
0.65
0.94
ITG b5
1
1.08
1.10
1.10
1.03
1.07
ITG b5
0
0.12
0.14
0.14
0.04
0.10
Supplement 137
Supplement to section 4.2.6 of the results:
Figure 0-4: Dynamic actin remodeling induced by cyclic compression and scaffold wall deformation under
compression visualized using the Bioreactor-Microscope-Setup. (A) GFP-LifeAct expressing hFOBs seeded in collagen
scaffolds were stimulated with cyclic compression (1Hz, 10%) for 30 min. Fast actin remolding processes were
recorded during 3 min before and after cyclic compression. Representative images show the cell outline change over
3 min. The cell outlines are colored according to frame number from blue to pink. (B) Scaffold wall crimping due to
10% compression (=160µm). Red lines indicate position of collagen scaffold walls before compression (scale bar =
100µm).
Supplement to section 4.3.2 of the results:
Figure 0-5: Fibrillar collagen density is reduced by cyclic compression after 2 weeks of cultivation. Human fibroblasts
seeded in 1.5-wt% collagen scaffolds were cultured for 2 weeks in the bioreactor under intermitted cyclic compression
(f=1Hz, ε=10%, 3h load, 5h break), rhBMP2 (5nM) or a combination of both. Representative confocal multiphoton
images showing fibrillar collagen visualized by SHG (white). Yellow arrows indicate newly deposited collagen fibers
138 Supplement
within the collagen walls of the scaffold. Scale bar = 100µm. Quantification of fibrillary collagen density inside scaffold
pores (n=2).
Supplement to section 4.3.3 of the results:
Table 0-2: Gene expression analysis of selected ECM proteins and ECM modulators. Fold change (F.I.) gene expression
analyzed after seven days in the bioreactor under intermitted cyclic compression (f=1Hz, ε=10%, 3h load, 5h break),
rhBMP2 (5nM) or a combination of both. The heat map in Figure 4-23 is based on the log2(F.I.), n=3.
F.I.
c
L
B
B/L
log2(F.I.)
c
L
B
B/L
COL1A2
1
1.11
0.82
0.80
COL1A2
0
0.15
-0.29
-0.32
COL6A1
1
1.09
1.14
0.96
COL6A1
0
0.13
0.19
-0.06
FN
1
1.16
1.00
0.92
FN
0
0.21
-0.01
-0.13
FBLN1
1
1.84
1.42
1.26
FBLN1
0
0.88
0.50
0.33
ELN
1
1.91
0.64
1.22
ELN
0
0.94
-0.65
0.29
TNC
1
1.36
0.97
1.11
TNC
0
0.44
-0.04
0.15
THBS1
1
1.22
1.02
0.97
THBS1
0
0.29
0.03
-0.04
TGFBI
1
2.25
0.92
1.52
TGFBI
0
1.17
-0.12
0.60
POSTN
1
1.38
0.81
1.21
POSTN
0
0.47
-0.30
0.28
BMP1
1
1.48
1.00
1.18
LOX
0
0.39
-0.14
0.36
LOX
1
1.31
0.91
1.28
LOXL1
0
0.33
0.15
0.00
LOXL1
1
1.26
1.11
1.00
BMP1
0
0.57
0.00
0.24
MMP1
1
1.30
2.33
1.60
MMP1
0
0.38
1.22
0.68
MMP13
1
1.47
1.16
1.13
MMP13
0
0.55
0.21
0.18
Supplement to section 5.6 of the discussion:
Figure 0-6: Cyclic compression did not induce ERK1/2 or Src phosphorylation. Human FOBs seeded in collagen
scaffolds were subjected for 15, 30 or 90 min to cyclic compression (1Hz, 10%). The phosphorylation of ERK1/2 and
Src were analyzed by western blotting. Phosphorylation intensities have been normalized to the uncompressed control
(n=4).
Abbreviations 139
Abbreviations
A
ALP
alkaline phosphatases
B
BISC
BMP
BMPR
BSA
BMP-induced signaling complex
Bone Morphogenetic Protein
BMP receptor
Bovine Serum Albumin
C
c
Col
control
collagen
D
DMEM
DMSO
Dulbecco's modified Eagle
medium
Dimethyl sulfoxide
E
e.g.
ECM
exempli gratia
Extracellular matrix
F
FA
FACS
FAK
FBS
FC
FDA
FGF-2
focal adhesion
fluorescence-activated cell sorting
focal adhesion kinase
fetal bovine serum
focal complex
Food and Drug Administration
Fibroblast Growth Factor -2
G
GAG
GAPDH
GPCRs
GSK3
glycosaminoglycan
glyceraldehyde 3-phosphate
dehydrogenase
G-protein-coupled receptors
glycogen synthase kinase 3
H
hdF
hFOBs
hMSCs
human dermal fibroblasts
human fetal osteoblasts
human mesenchymal stromal cells
I
I-Smads
ID
IF
IFM
IL
ITG
inhibitory Smad
inhibitor of DNA binding
immunofluorescence
interfragmentary movement
interleukin
integrin
J
Jas
Jasplakinolide
K
L
L
LA
LOX
loading
LifeAct
lysyl oxidases
M
MAPK
MLC
MMPs
MNE
mRNA
MSC
mitogen activated protein kinases myosin
light chain
matrix metalloproteinases
mean normalized expression
messenger RNA
mesenchymal stromal cell
N
NA
NEA
nascent adhesion
non-essential amino acids
140 Abbreviations
O
OCN
OSX
OPN
OSS
osteocalcin
osterix
osteopontin
oscillatory shear stress
P
P/S
PAA
PBS
PCR
PDGF
PDMS
PEEK
PFA
PFCs
PGs
PI3K
POM
PTHrP
Penicillin/Streptomycin
polyacrylamide
Phosphate buffered saline
polymerase chain reaction
Platelet-Derived Growth Factor
polydimethylsiloxane
polyether ether ketone
paraformaldehyde
preformed complexes
proteoglycans
phosphatidylinositol 3-kinase
polyoxymethylene
parathyroid hormone-related
peptide
Q
qPCR
quantitative polymerase chain reaction
R
R-Smad
rh
RNA
ROI
RT
RUNX2
receptor-regulated Smad
recombinant human
ribonucleic acid
Region of interest
Reverse transcription
Runt-related transcription factor 2
S
SHG
SHI
siRNAs
Sox 9
Src
second harmonic generation
second harmonic imaging
small interfering RNA
SRY (sex determining region Y)-box
proto-oncogene tyrosine-protein kinase 9
T
TGF-β
TIMP
Transforming Growth Factor-β
tissue inhibitors of
metalloproteinases
U
V
W
WB
wt-%
western blot
weight percent
X
Y
YAP
yes-associated protein 1
Z
