Development and operation of a
perfusion bioreactor for the cultivation of
mammalian cells inside a sponge-like ceramic
matrix
vorgelegt von
Dipl.-Ing.
Vicky Goralczyk
aus Berlin
Von der Fakultät III - Prozesswissenschaften
der Technischen Universität Berlin
zur Erlangung des akademischen Grades
Doktor der Ingenieurwissenschaften
Dr.-Ing.
genehmigte Dissertation
Promotionsausschuss:
Vorsitzender: Prof.Dr.rer.nat.Lothar W.Kroh
Gutachter: Prof.Dr.-Ing.habil.Rudibert King
Gutachter: Prof.Dr.-Ing.Udo Reichl
Gutachter: Prof.Dr.rer.nat.Helmut Schubert
Tag der wissenschaftlichen Aussprache: 13.Juli 2010
Berlin 2010
D83
i Contents
Contents
1 Abstract 1
2 Introduction 5
2.1 Stateoftheart ................................ 6
2.1.1 Ceramics for cell cultivation . . . . . . . . . . . . . . . . . . . . . 6
2.1.2 Cell cultivation modes . . . . . . . . . . . . . . . . . . . . . . . . 10
2.1.3 Inoculationmodes .......................... 13
2.1.4 Perfusion dynamic . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2 Goalsofthisthesis .............................. 16
3 Materials and methods 19
3.1 Preparation of alumina ceramics for cell culture . . . . . . . . . . . . . . 19
3.2 Reactor device for cell cultivation on alumina foams . . . . . . . . . . . . 20
3.2.1 Tubular design for series connection of ceramics . . . . . . . . . . 21
3.2.2 Revolver design for parallel connection of ceramics . . . . . . . . . 22
3.2.3 Small block design for single foam analysis . . . . . . . . . . . . . 24
3.3 Inoculation procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.3.1 Staticinoculation........................... 24
3.3.2 Dynamic inoculation by stirring or agitation . . . . . . . . . . . . 26
3.3.3 Dynamic inoculation by convectional forces . . . . . . . . . . . . . 26
3.4 Cultivation of cells on ceramics . . . . . . . . . . . . . . . . . . . . . . . 27
3.4.1 Perfusion cultivation inside the reactor device . . . . . . . . . . . 27
3.4.2 Static cultivation of cells outside the reactor device . . . . . . . . 29
3.4.3 Cultivation of cells by medium convection outside the reactor device 29
3.5 Celllinesandorigin.............................. 29
3.6 Assays..................................... 31
3.6.1 Microscopic evaluation of cell vitality by dyeing according to FDA/EB
protocol ................................ 31
Contents ii
3.6.2 Microscopic evaluation of cell distribution by dyeing according to
hematoxylin/eosin protocol . . . . . . . . . . . . . . . . . . . . . 32
3.6.3 Scanning electron microscopy of ceramic surface and cells on ce-
ramics................................. 32
3.6.4 Metabolic evaluation of glucose consumption and lactate formation 33
3.6.5 Reduction of resazurin . . . . . . . . . . . . . . . . . . . . . . . . 34
3.6.6 Carrier hot gas extraction . . . . . . . . . . . . . . . . . . . . . . 35
4 Reactor characterization 37
4.1 Ceramics.................................... 37
4.1.1 Porosity ................................ 37
4.1.2 Flowresistance ............................ 37
4.2 Characterization of flow inside the reactor . . . . . . . . . . . . . . . . . 39
5 Analysis of CHO-K1 by resazurin assay and carbon content determination. 43
5.1 Reduction of resazurin by CHO-K1 ..................... 43
5.1.1 Determination of rate of reduction . . . . . . . . . . . . . . . . . 43
5.1.2 Modeling resazurin reduction . . . . . . . . . . . . . . . . . . . . 46
5.1.3 Adaptation of the resazurin reduction model to describe bioreactor
performance.............................. 48
5.2 Carbon content of cells cultivated on ceramics . . . . . . . . . . . . . . . 54
5.2.1 Carbon content of pure cells . . . . . . . . . . . . . . . . . . . . . 54
5.2.2 Carbon content of cells on foams . . . . . . . . . . . . . . . . . . 55
6 Influence of mode of inoculation on cellular growth and distribution 59
6.1 Staticinoculation............................... 59
6.1.1 Static inoculation on foams in culture plates . . . . . . . . . . . . 59
6.1.2 Static inoculation into the tubular reactor . . . . . . . . . . . . . 60
6.1.3 Static inoculation into revolver reactors . . . . . . . . . . . . . . . 61
6.1.4 Static inoculation in ceramics with flow channels . . . . . . . . . . 65
6.1.5 Reproducible cultivation following static inoculation into revolver
reactors ................................ 70
6.2 Dynamic inoculation by agitation . . . . . . . . . . . . . . . . . . . . . . 72
6.3 Oscillatory perfusion inoculation . . . . . . . . . . . . . . . . . . . . . . . 75
6.3.1 Influence of module orientation, flow velocity, initial cell count on
celldistribution............................ 75
iii Contents
6.3.2 Introduction of more porous ceramics and augmentation of flow
velocity during cultivation . . . . . . . . . . . . . . . . . . . . . . 76
6.3.3 Reduction of foam volume by reducing cylinder height . . . . . . 81
7 Reproducibility of the chosen operation methods 85
7.1 Standardfoams................................ 85
7.2 Foams with larger pores . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
8 Long-term cultivation 97
9 Applicability of the reactor system for other cell types 103
9.1 Human lung carcinoma cells A549 ......................103
9.2 Human primary fibroblasts . . . . . . . . . . . . . . . . . . . . . . . . . . 105
9.3 Madin-Darby canine kidney cells (MDCK).................106
10 Conclusion 111
Appendix 117
Literature 130
v Abbreviatons and Symbols
Abbreviations
Al2O3aluminium oxide
CO2carbon dioxide
CHGE carrier hot gas extraction
DNA desoxyribonucleic acid
EB ethidium bromide
ECM extracellular matrix
FBS fetal bovine serum
FDA fluorescein diacetate
HA hydroxy-apatite
IFS Interdisciplinary Research Priority Program
PBS++ phosphate buffered saline containing magnesium and calcium
PFR plug flow reactor
ppi pores per inch
SD standard deviation
SEM scanning electron microscopy
STR stirred tank reactor
TCP tri-calcium phosphate
TU Berlin Technische Universität Berlin
vii Abbreviatons and Symbols
Symbols
∆ppressure drop [Pa]
˙
Vvolumetric flow [ml/min]
ηcoefficient of viscosity [mPas]
φporosity
ρdensity [g/ml]
σstandard deviation
θnormalized time
Aarea [m2]
Bo Bodenstein number
cconcentration [g/l]
Dax axial coefficient of dispersion [m2/s]
Eextinction
Fnormalized extinction
Hheight [m]
KdDarcy’s constant [m2]
Llength [m]
mmass [g]
nquantity
Abbreviatons and Symbols viii
ppercentage
R2coefficient of determination
ttime [s]
Vvolume [ml]
xnormalized concentration
ix List of figures
List of Figures
2.1 Cellular embedment in connective tissue. . . . . . . . . . . . . . . . . . . 7
2.2 Al2O3structureandsurface.......................... 9
2.3 One foam replaces 6 T-flasks. . . . . . . . . . . . . . . . . . . . . . . . . 9
3.1 Pore size distribution of foamed alumina. . . . . . . . . . . . . . . . . . . 20
3.2 Module connection configurations. . . . . . . . . . . . . . . . . . . . . . . 22
3.3 Tubular design reactor module for series connection of foams. . . . . . . . 23
3.4 Tubular configuration connected to the Biostat B plus. . . . . . . . . . . 23
3.5 Ceramic foam holding magazines. . . . . . . . . . . . . . . . . . . . . . . 24
3.6 Reactor modules for parallel perfusion. . . . . . . . . . . . . . . . . . . . 25
3.7 Single foam reaction module. . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.8 Reactor module with glass olives. . . . . . . . . . . . . . . . . . . . . . . 28
4.1 Experimental setup for evaluation of Kd................... 38
4.2 Color step experiment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.1 Dynamic of resazurin reduction for CHO-K1................. 44
5.2 Relation between rate of reduction and cell number. . . . . . . . . . . . . 46
5.3 Model fit for reduction of resazurin for identification experiments. . . . . 49
5.4 Model validation for reduction of resazurin for non-identification experi-
ments. ..................................... 50
5.5 Scheme for division of reactor volume into compartments for flow modeling. 51
5.6 Model validation for reduction of resazurin inside the bioreactor. . . . . . 54
5.7 Carbon and sulfur content of CHO-K1.................... 55
5.8 Carbon content of CHO-K1 onceramics................... 56
6.1 Resazurin reduction in static approach. . . . . . . . . . . . . . . . . . . . 60
6.2 Cell leakage for cells statically inoculated and cultivated with agitation. . 61
6.3 Resazurin reduction after static inoculation in reactor module. . . . . . . 62
List of figures x
6.4 Static inoculation of cells into revolver reactor. . . . . . . . . . . . . . . . 63
6.5 Medium metabolites for static inoculation into reactor moules. . . . . . . 64
6.6 FDA/EB staining for static inoculation in reactor modules. . . . . . . . . 65
6.7 Medium metabolites for foams with blind holes. . . . . . . . . . . . . . . 67
6.8 Medium single foam analysis for blind hole foams. . . . . . . . . . . . . . 67
6.9 Resazurin single foam assay for blind hole foams. . . . . . . . . . . . . . 68
6.10 Vitality assay for blind hole foams. . . . . . . . . . . . . . . . . . . . . . 69
6.11 Dynamic resazurin reduction after static inoculation in reactor module. . 70
6.12 Medium metabolites for static inoculation into conical magazines. . . . . 71
6.13 Scanning electron microscopy of CHO-K1 on ceramic foam. . . . . . . . 72
6.14 Cell count for agitation inoculation. . . . . . . . . . . . . . . . . . . . . . 73
6.15 Cell density and distribution following agitation seeding. . . . . . . . . . 74
6.16 Cell distribution following dynamic seeding by help of perfusion. . . . . . 77
6.17 Scanning electron microscopy following dynamic seeding by help of per-
fusion. ..................................... 78
6.18 Comparison of resazurin reduction for small and wider pore sizes. . . . . 79
6.19 Glucose consumption and lactate formation for different pore sizes. . . . 80
6.20 Reduction of resazurin for flat foams with larger pores. . . . . . . . . . . 82
6.21 Vitality assay for CHO-K1 on flat ceramic foams with larger pores. . . . 83
7.1 Cumulative glucose consumption and lactate formation for reproducible
inoculation of CHO-K1 on standard foams. . . . . . . . . . . . . . . . . . 86
7.2 Reduction of resazurin for CHO-K1 reproducibly inoculated on standard
foams. ..................................... 87
7.3 Cell distribution for CHO-K1 reproducibly inoculated. . . . . . . . . . . 89
7.4 Cumulative glucose consumption and lactate formation for reproducible
inoculation of CHO-K1 on flat foams with larger pores. . . . . . . . . . . 90
7.5 Reduction of resazurin for CHO-K1 reproducibly inoculated on flat foams
withlargerpores................................ 91
7.6 Cell distribution for CHO-K1 reproducibly inoculated on flat foams with
largerpores................................... 93
7.7 Scanning electron microscopy for CHO-K1 on flat foams with larger pores. 94
8.1 Cumulative glucose consumption and lactate formation for long-term cul-
tivation of CHO-K1. ............................. 98
xi List of figures
8.2 Metabolic reduction of resazurin by cells cultivated on standard foams for
7 weeks under medium perfusion. . . . . . . . . . . . . . . . . . . . . . . 99
8.3 Cell proliferation for long-term cultivation. . . . . . . . . . . . . . . . . . 100
8.4 Vitality staining for CHO-K1 cultivated for several weeks in standard
foams and foams with larger pores. . . . . . . . . . . . . . . . . . . . . . 100
8.5 Scanning electron microscopy for long-term cultivation of CHO-K1. . . . 102
9.1 Vitality staining of A549 following 2 weeks of cultivation. . . . . . . . . . 104
9.2 Scanning electron microscopy of A549 following 2 weeks of cultivation. . 105
9.3 Vitality staining of primary fibroblasts following 7 days of cultivation. . . 106
9.4 Scanning electron microscopy of primary fibroblasts following 7 days of
cultivation. ..................................107
9.5 Staining of MDCK following 2 weeks of cultivation. . . . . . . . . . . . . 108
9.6 Scanning electron microscopy of MDCK following 2 weeks of cultivation. 109
10.1 Illustration of self-made shaker. . . . . . . . . . . . . . . . . . . . . . . . 118
List of Tables xiii
List of Tables
2.1 Requirements of scaffolds used for cell cultivation in packed bed bioreac-
tors [Meuwly et al., 2007]. . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.1 Demands for bioreactor design. . . . . . . . . . . . . . . . . . . . . . . . 21
3.2 Cultivation requirements for CHO-K1, A549, MDCK and human fibrob-
lasts....................................... 30
4.1 Flow distribution inside ceramics. . . . . . . . . . . . . . . . . . . . . . . 40
4.2 Characteristic numbers for revolver reactor design. . . . . . . . . . . . . . 42
5.1 Linear regression for reduction of resazurin. . . . . . . . . . . . . . . . . 45
5.2 Conditions for modeling resazurin reduction inside bioreactors. . . . . . . 52
5.3 Linear regression for carbon analysis. . . . . . . . . . . . . . . . . . . . . 57
6.1 Agitation inoculation for different cell numbers. . . . . . . . . . . . . . . 73
6.2 Optical analysis of foams two days after perfusion inoculation. . . . . . . 75
6.3 Optical analysis of flat foams two days after perfusion inoculation. . . . . 82
7.1 Reproducibility of cell cultivation in reactor modules. . . . . . . . . . . . 88
9.1 Optical analysis of A549 on ceramics three days after perfusion inoculation.104
9.2 Optical analysis of primary fibroblasts on ceramics 7 days after perfusion
inoculation. ..................................105
9.3 Optical analysis of MDCK on ceramics three days after perfusion inocu-
lation. .....................................107
10.1 Adaptation of CHO-K1 to growth in medium containing 1% FBS. . . . . 121
1. Abstract 1
1 Abstract
Cultivation of adherently growing cells inside three dimensional scaffolds gains more
importance as specific demands from medicine and industry arise. A huge challenge
herein lies in nutrient supply all over the scaffold’s volume. This work presents a perfu-
sion bioreactor device for cell cultivation inside porous aluminium oxide ceramics. The
bioreactor supports vital growth of CHO-K1,A549,MDCK and human primary fibrob-
lasts on and in ceramic cylinders of 5mmheight and 10mmdiameter with perfusion
velocities from 0.33-0.66mlmin−1cm−2. Proliferation was assured by a metabolic assay
based on resazurin reduction and carbon content analysis. Cell distribution and vitality
was assessed by staining cross sections of the ceramic.
Fibrous structures were observed inside ceramic scaffolds for perfusion cultivation of
CHO-K1 and human primary fibroblasts for several weeks by evaluation of scanning
electron microscopy pictures. This clearly supports the hypothesis of micro milieu for-
mation inside porous ceramics and consequently abundance of ECM-proteins.
Perfusion therefore enables three-dimensional cell cultivation inside porous alumina ce-
ramics.
Kurzfassung 3
Kurzfassung
Mit steigenden Anforderungen aus Medizin und industrieller Wertstoffherstellung gewinnt
die Kultivierung tierischer Zellen in dreidimensionalen Verbünden zunehmend an Be-
deutung. Die Versorgung von im Inneren der Strukturen wachsenden Zellen stellt
hierbei besondere Ansprüche an die Kultivierungstechniken. Die vorliegende Arbeit
beschreibt Entwicklung und Betrieb eines Perfusionsbioreaktors, in dem adhärent wach-
sende Zellen in einer porösen Aluminiumoxid-Keramik kultiviert werden. In zylinder-
förmigen Gerüsten mit 10mmDurchmesser und 5mmHöhe konnte das Wachstum von
CHO-K1,A549,MDCK und primären humanen Fibroblasten gezeigt werden, wobei
Wachstum über die metabolische Reduktion von Resazurin und Masse elementaren
Kohlenstoffs, Zellverteilung und -vitalität über Schnittfärbungen nachgewiesen wurden.
Überdies wurden in den Poren der Keramiken mittels Rasterlektronenmikroskopie pro-
teinartige Strukturen gefunden, die den Schluss nahelegen, dass CHO-K1 und Fibrob-
lasten Proteine der extrazellulären Matrix sezernieren. Anzunehmen ist, dass sich wegen
der geringen Strömungsgeschwindigkeiten von 0.33-0.66mlmin−1cm−2und damit stark
diffusionsgeprägter Nährstoffversorgung eine für das Zellwachstum besonders günstige
Mikroumgebung ausbildet.
Durchströmte Aluminumoxid-Keramiken sind daher bestens geeignet für die Kultivierung
tierischer Zellen unter naturähnlichen Bedingungen.
2. Introduction 5
2 Introduction
In the 1980s, for the first time ceramics were discovered as scaffolds for cell cultivation
[Lydersen et al., 1985]. Lydersen in his frequently cited paper proved the feasibility of
the scaffold to support growth of different cell types as mammalian cells, insect cells,
chicken and primate cells. In the following years, other groups also used scaffolds with
a plain ceramic surface for cell cultivation [Lydersen et al., 1985, Mitsuda et al., 1991,
Suck et al., 2008]. In contrast, a ceramic with a highly interconnected porous structure
not only supports cellular growth but moreover was found to promote cellular differ-
entiation [Schubert et al., 2004]. It is assumed, that the porous structure allows the
creation of a micro environment favorable for cell differentiation, as cells presumably
are exposed to various growth conditions depending on surrounding nutrient status.
Since then, porous ceramics have been used as cultivation surface for proliferation and
differentiation of production cell lines [Park and Stephanopoulos, 1993], many kinds of
primary cells [Wang et al., 2003, Schubert et al., 2004, Janssen et al., 2006] and stem
cells [Wang et al., 2003, Janssen et al., 2006].
The great potential of interconnected structured ceramics is in its ability to provide a
number of variable micro milieus in different pores at once. Therefore, the formation
of a well defined extracellular matrix is stimulated which is an inevitable step in order
to cultivate cells under natural conditions [Minuth et al., 2003]. Moreover, the huge
surface to volume ratio of a porous structure spares the need for permanent cell pas-
saging. Extensive cell passaging is a standard procedure in cell cultivation techniques
but it leads to cellular dedifferentiation [Bonassar and Vacanti, 1998], which should be
avoided in many applications. Promotion of a natural extracellular matrix together with
the option to grow a huge number of cells without phenotypical changes makes porous
ceramics useful for tissue engineering applications or for production of specific drugs
or therapeutics. As demands for very specific therapeutics and tissue engineered con-
structs are rising, simple ceramics turn out to be a great alternative to other systems.
The ceramics used in this work are unique in such a manner that the manufacturing
process differs strongly from that of porous ceramics used to date. For the commonly
6 2.1 State of the art
used ”Schwarzwalder process“, forming of ceramics is realized by coating a porous base
body in ceramic slurry and removing the base body by heat, which leaves a double-
layered structure of pores and bridges. Instead, the unique technique developed in the
group of H.Schubert, Technische Universität (TU) Berlin, see [Garrn et al., 2004], uses
a ceramic-protein-slurry for foam forming, which leaves a more solid porous ceramic due
to the one-layered structure obtained by the process.
The purpose of the work in hand is now to answer the question of maintaining a useful
quantity of cells in a vital, proliferating status deep inside a ceramic scaffold obtained by
the novel process described above. Of particular interest would be a three-dimensional
(3D) system which provides high ingrowth into the scaffold due to nutrient perfusion. At
the same time diffusion-controlled mass transport perpendicular to the main flow direc-
tion into the caverns should ensure the setup of local micro milieus. For cells statically
cultivated on ceramics (which is done mainly to study cells under close to physiolog-
ical conditions) [Orlandi et al., 1997, Bagley et al., 1999, Schubert et al., 2004], cellu-
lar ingrowth is limited to a few hundred microns. Therefore, to lure cells into scaf-
folds, these can be overflown with nutrient solution hereby increasing nutrient exchange
rate [Wang et al., 2003] or perfused with nutrients which enhances cellular ingrowth
strongly. Often, perfusion approaches have to deal with severe inhomogeneities in cel-
lular distribution and therefore inconsistent cultivation outcomes [Mitsuda et al., 1991,
Park and Stephanopoulos, 1993, Janssen et al., 2006], which is ascribed to poorly ad-
justed operational strategies in many cases. Moreover, upscaling of those perfused sys-
tems cannot be achieved by simply enlarging the ceramic volume as due to the scaffold’s
structure perpetuation of specific micro milieus will be impeded. Therefore, for many
systems the quantity of cells growing into a specific, environmentally defined compart-
ment is strongly limited as will be discussed below.
2.1 State of the art
2.1.1 Ceramics for cell cultivation
The human body consists of more than 200 cell types, which differ in form and localiza-
tion depending on their function in the organism [Alberts et al., 2007]. Among those,
less than 20 cell types are found to be free-moving through the organism, i. e. blood cells
and cells of the immune system. The vast majority of cells is organized in a three-
dimensional environment - the extracellular matrix - where the cells communicate with
2. Introduction 7
Figure 2.1: Cellular embedment in connective tissue. Adapted from [Alberts et al., 2007].
each other and the surrounding environment by chemical signals (see figure 2.1).
This embedded cell organization, which is mainly imprinted by the structure of extra-
cellular proteins, proteoglycans and polysaccharides, in turn models the cell’s shape and
hereby strongly affects the intracellular spatial organization of cytoplasmic components.
Therefore, intracellular structural proteins, functional units and cytoplasmic sugars build
up a highly organized compartment itself whose geometric constitution impacts cellular
processes.
Sustaining this cytoplasmic order during in vitro cultivations of cells demands sophisti-
cated cultivation procedures as well as adequate scaffolds for cell attachment. Meuwly
and colleagues in [Meuwly et al., 2007] recently proposed a list of material proper-
ties relevant for cell cultivation, which are summarized in table 2.1. Recent advances
for cell cultivation scaffolds allowing three-dimensional cellular growth mainly are fo-
cused on polymers as poly-lactic acid (PLA) or poly-glycolic acid (PGA). These are
rather soft but pressure-resistent materials possessing good manufacturing properties
and therefore facilitate the manufacturing of scaffolds with various shapes and inter-
nal structures [Vunjak-Novakovic et al., 1996, Marler et al., 1998, Kitagawa et al., 2006,
Moutos et al., 2007]. Other growth substrates as further polymers, hydrogels or hy-
brid materials as mineralized collagen were also reported [Vunjak-Novakovic et al., 1996,
Leach and Schmidt, 2004, Liao and Cui, 2004, Moutos et al., 2007].
8 2.1 State of the art
Table 2.1: Requirements of scaffolds used for cell cultivation in packed bed bioreactors
[Meuwly et al., 2007].
simple physical configuration and made of non-toxic materials
high surface to volume ratio
optimal diffusion from the bulk phase to the center of the carrier
chemical and mechanical stability
autoclavable
suitable for adherent and non-adherent cells
chemically and biologically inert, no reaction with the product
low cost and reusable if possible
of nonanimal origin
Opposite to those organic materials, fewer approaches concentrate on anorganic solid
bodies which often display a more homogeneous structure and allow an easier and more
controllable fabrication process. Herein, the group of ceramic substrates, in particu-
lar hydroxy-apatite ceramics (HA) or tri-calcium phosphate (TCP), is most commonly
known, as it is used in clinical applications as dental prosthesis or implantation devices
mostly for bone repair. Both HA and TCP are biodegradable, meaning the stiff ce-
ramic material is little-by-little replaced by tissue and organism cells of the acceptor
body [Krajewski et al., 1996, LeGeros, 2002]. Meanwhile, the anorganic material sup-
ports cell proliferation and differentiation, therefore encouraging physical regeneration.
Unfortunately, the decomposition properties of HA and TCP coincidently expose their
greatest challenge: depending on porosity and web thickness, ceramics can be very fragile
which affects their usability. Moreover, calcium particles disengaging from the ceramic
are known to evoke inflammation in vivo which of course has to be avoided not only in
immuno deficient patients [Harada et al., 1996].
As an alternative to calcium ceramics, aluminium oxide (Al2O3) ceramics are reported as
cell cultivation matrices since the nineteen-eighties, when Lydersen proved the feasibility
to cultivate various cell types on and in the non-modified porous scaffold [Lydersen, 1987].
Due to the fabrication process, foamed alumina represents a sponge-like three-dimensional
matrix with huge porosity of around 90%. Moreover, number and diameter of pore
openings can be adjusted in a way, that cells in the foam’s caverns can be supplied with
nutrients by perfusion of medium through the scaffold, see figure 2.2 for an overview of
the foam’s structure (for ceramics used in this work).
With a surface area to volume ratio of around 0.3m2/ml foamed alumina provides a
growth area for many cells in just a small reaction volume. One ceramic foam cylin-
der of 5mmheight and 10mmdiameter could replace 6 in laboratories commonly used
2. Introduction 9
Figure 2.2: Scanning electron micrograph of foamed alumina’s structure and surface.
Take note of the plaster stone-like surface structure in the last image which is owed to the
manufacturing process where small Al2O3particles are pressed against each other.
Figure 2.3: Comparison of cultivation surface: One alumina foam cylinder of 5 mm height
and 10 mm diameter as used in this work (foam situated left to the e0.10 coin) can replace
6 175 cm2cell culture flasks.
175cm2-T-flasks regarding cultivation surface, see figure 2.3.
The alumina foam’s characteristic purity with respect to organic material is ensured
by the last step of the fabrication process. During the sintering process temperatures
of around 1650◦Care set which incinerate any impurities that may have arisen during
former fabrication [Garrn et al., 2004].
Recent publications clearly indicate that the sum of the alumina foam’s beneficial prop-
erties promotes cell differentiation in the foam’s caverns - without further modification
of the ceramic surface or medium manipulation [Schubert et al., 2004].
The presented qualification of aluminium oxide ceramic foams for the cultivation of
10 2.1 State of the art
mammalian cells as well as its appropriateness for cultivations in perfusion mode estab-
lishes a basis for the experiments in this thesis. I propose the hypothesis that cellular
growth is not only easily achieved on the surface of porous ceramics but also can be
maintained inside a ceramic foam by help of perfusion of nutrients through the scaffold
for long-term cultivation.
2.1.2 Cell cultivation modes
Animal cells in vivo are embedded in a pattern composed of polysaccharides and fibrous
proteins, the extracellular matrix (ECM). Here, cells are supplied with nutrients and
signal substances via a circulating blood system, which is separated from the ECM by
a layer of epithelial cells (endothel) and basal lamina, see figure 2.1.
Thus, it appears that nutrients primarily approach convectively with the blood flow,
however, secondarily nutrients diffuse through the ECM to the consumpting cells. Phys-
iological studies in vertebrates revealed a maximal diffusion distance between ECM-
embedded cells and capillaries of 50-100µm and with capillary diameters of around
6µm[Alberts et al., 2007].
For in vitro cultivations, the challenge of nutrient supply of cells in monolithic columns
of three-dimensional substrates is addressed by a set of different operation strategies as
follows.
Static nutrient supply
The most simple operation strategy is to supply cells growing in a three-dimensional
scaffold by simply holding the scaffold in a non-moved nutrient solution (static ap-
proach). Hereby, nutrients reach the cells solely by diffusion which is driven by the
nutrient gradient between near-cell-environment and the solution and which is main-
tained as long as the gradient upholds. Due to oxygen diffusion limitations, cell growth
inside the scaffold is limited to a few hundred microns [Fassnacht and Poertner, 1999,
Martin and Vermette, 2005].
In a static environment, equalization of concentration profiles is a rather slow event as
it is restricted by diffusion rates.
2. Introduction 11
Nutrient supply by convection
While static operation is acceptable for cell cultures in monolayer, it clearly cannot
support cellular growth in three-dimensional structures. To overcome the lack of cells
growing in the scaffold’s interior, dynamic operation strategies were proposed, in which
scaffolds are flooded with medium in a way that increases the probability of medium
accessing the scaffold’s inward.
Hereby, two different approaches can be distinguished: Either medium is conducted
over a firmly anchored scaffold [Suzuki et al., 1994, Vunjak-Novakovic et al., 1996] con-
tinously, or the scaffold is moved through the more or less unruffled nutrient solution
[Vunjak-Novakovic et al., 1998, Neves et al., 2005, Timmins et al., 2007]. In a mixed
approach, scaffolds are exposed to stationary free fall in a rotating, so called micro-
gravitational reactor, favoring the penetration of medium at the scaffold’s skin face
[Freed and Vunjak-Novakovic, 1997]. To increase shear forces that augment the process
of convection, the scaffold can be moved through a phase boundary as with Zellwerk’s
Z®RP system (Zellwerk GmbH, Eichstädt, Germany). Here cell populated ceramics
are moved consecutively through a liquid medium phase and a gaseous air phase, hereby
increasing oxygen nourishment compared to submerged cultivations.
Penetration of medium into the flooded scaffold will apparently depend on scaffold poros-
ity (i.e.pore size) and enforcement. The inner volume of a scaffold with small pores
therefore requires intense fluid forces, which can be realized in highly oscillating sys-
tems.
Compared to static cultivation methods, spatial nutrient concentration will be rather
homogeneous, due to well mixing. Local gradients in nutrient composition due to cell
metabolism therefore do not establish near cells but will be equalized over the total
medium rapidly.
Nutrient supply by perfusion
In order to reach cells which grow deep inside the scaffold’s volume, the scaffold has to
be perfused with medium. Such an enforced convection, however, requires the sterile
incorporation of the scaffold into an operation periphery of tubings and pumps. Then,
two principle operational strategies can be distinguished: i) medium is pumped through
the scaffold into a medium reservoir and thereby recirculated over the volume (recircu-
lation) or ii) after one pass over the scaffold volume medium is discarded (single pass),
allowing a more or less homogenous nutrient distribution over the cultivation period.
12 2.1 State of the art
The first strategy is more common for economical reasons, and concurrently supports
the establishment of a cell-defined signal substance environment as essential cell prod-
ucts are retained in the system.
Applegate and Stephanopoulos in [Applegate and Stephanopoulos, 1992] proved for a
10cmperfused ceramic monolith that medium recirculation led to enhanced cell death
inside the system. Cell mortality mainly was deduced to oxygen limitations which are
associated to bed height. Nevertheless, total cell number still was increased compared
to single-pass mode [Grampp et al., 1996].
In perfused packed-bed bioreactors, the growing demand for nutrients, which is related
to cellular growth, is considered by augmenting medium flow velocity. Therefore, mass
throughput is augmented, but likewise laminar or even turbulent shear stresses increase,
establishing very new environments for the cells. To isolate cells from the nutrient flow
therefore might be a good solution and is addressed in hollow-fiber bioreactors which are
known since first experiments by Knazek et al.in the 1970s [Knazek et al., 1972]. Hereby
nutrients flow through artificial capillaries and reach cells which grow on the outside of
the capillaries by diffusion. This approach resembles more the nutrient supply in organ-
isms but has to deal with high metabolite gradients and accumulation of by-products in
the cell-space [Piret et al., 1991] due to large diffusion distances or inhomogenous flow
through the hollow fibers which leads to inhomogenous cell growth and often necrosis
[Knazek et al., 1972, Ku et al., 1981, Tharakan and Chau, 1986].
Both perfused packed-bed reactors and hollow fiber reactors have to overcome up-scaling
difficulties. The height of the bed and the length of the system respectively are strongly
limited by axial concentration gradients for nutrients and oxygen. Moreover, the bed
diameter for packed-bed bioreactors is limited due to dispersion effects, which influence
flow homogeneity.
Nevertheless, direct perfusion bioreactors for cell cultivation in general promote growth
and proliferation inside cell containing scaffolds [Goldstein et al., 2001, Navarro et al., 2001,
Bancroft et al., 2002]. Moreover, they have shown very promising cultivation results for
many kinds of cells including differentiated primary cells which expressed tissue-specific
markers in vitro [Pazzano et al., 2000, Goldstein et al., 2001, Bancroft et al., 2002, Carrier et al., 2002,
Davisson et al., 2002]. However, perfusion velocity has to be carefully adjusted depend-
ing on cultivated cell type and maturation stage [Davisson et al., 2002].
2. Introduction 13
2.1.3 Inoculation modes
Recent work revealed a strong relationship between a scaffold’s initial cellularity (i. e. cell
number and distribution directly after seeding) and distribution of tissue within an engi-
neered scaffold following further cultivation [Kim et al., 1997, Holy et al., 2000]. There-
fore, cell inoculation into three dimensional scaffolds mainly pursues two goals to enhance
the rate of tissue development: spatially homogeneous distribution of cells all over the
scaffold’s volume and an initial cell density high enough to support excellent growth
during further cultivation. Hereby inoculation efficieny defined as
inoculation efficiency =cell number in scaffold following inoculation
inoculated cell number (2.1)
or
inoculation efficiency = 1 −ratio of non attached to attached cells (2.2)
is a major quality factor as cell availability is often limited (i.e.when expanding primary
human cells) [Radisic et al., 2003]. The choice of a suitable cell inoculation method is a
crucial step for cell cultivation whereby aggressive inoculation methods should be avoided
in particular for three dimensional scaffolds as occurring shear stresses can reduce cell
viability critically.
Therefore, one can distinguish three main strategies, which will be faced in more detail:
i) static inoculation, where cells are inoculated on the scaffold’s surface and allowed to
settle before cultivation starts, ii) dynamic inoculation facilitated by stirring or shaking
the scaffold in a cell suspension, iii) dynamic inoculation promoted by convective flow
of the cell suspension through the scaffold.
Static inoculation
For static inoculation, scaffolds are either deposited in a cell suspension [Wang et al., 2003]
or the cell suspension is dropped on top of the scaffold [Dar et al., 2002]. Cell migration
into the porous structure is slightly amplified by capillary action, particularly when scaf-
folds were thoroughly dried before inoculation [Dar et al., 2002]. Cell leakage is rather
unlikely as cells are entrapped by the scaffold’s tortuous structure.
Static inoculation methods often are linked to low seeding efficiency [Kim et al., 1997,
Xiao et al., 1999, Holy et al., 2000, Li et al., 2001, Kitagawa et al., 2006]. Moreover,
static inoculation leads to a very non-uniform cellular distribution over the scaffold’s vol-
ume [Vunjak-Novakovic et al., 1996, Kim et al., 1997, Holy et al., 2000, Li et al., 2001,
14 2.1 State of the art
Wendt et al., 2003]. Dense cell layers are reported on top and bottom sides of the scaf-
fold [Du et al., 2008] with maximal ingrowth depth at about 1 mm [Wendt et al., 2006].
Also, cell aggregation is observed [Dar et al., 2002], which is an important issue as cell
spheroids often contain necrotic centers [Sutherland et al., 1986]. Cell constructs ma-
turing from statically inoculated scaffolds are often reported to have an inhomogeneous
structure which consists of dense cell layers and concentrated ECM along the periphery
together with necrotic interior regions [Wendt et al., 2006]. Static inoculation there-
fore should be avoided unless for cell types that are very sensitive to mechanical forces
[Xiao et al., 1999].
Dynamic inoculation by stirring or agitation
Dynamic inoculation without perfusion of the scaffold often is performed in stirred flask
bioreactors (spinner flasks) [Kim et al., 1997, Carrier et al., 1999, Wendt et al., 2003],
whereby either both cells and scaffolds are combined in a stirred solution or the con-
structs are mounted inside the reactor and the cell suspension is stirred around it.
Another option is to combine cells and scaffolds in a cultivation tube and to shake it on
an agitator [Kim et al., 1997]. In addition to the physical effects described above, cell
migration into the scaffold is supported by convective flow occuring at the cell suspen-
sion/scaffold surface boundary.
Mixing inoculation at the most is related to low seeding efficiency [Kim et al., 1997,
Carrier et al., 1999] and non-uniform cell distribution [Wendt et al., 2003]. However,
high seeding efficiencies with first order kinetics recently have been observed for spin-
ner flasks by [Ouyang and Yang, 2007]. For both mixing and agitation seeding, cells
are found inside the scaffold with much higher densities of cells lining the scaffold’s
surface [Vunjak-Novakovic et al., 1998]. Moreover, small cell aggregates and cell clus-
ters have to be remarked [Kim et al., 1997, Vunjak-Novakovic et al., 1998]. As stated
by [Kim et al., 1997], agitation seeding results in higher seeding efficiencies than mixing
seeding.
Directly compared to static inoculation methods, mixing and agitation seeding re-
sult in a higher adherent cell number inside the scaffold and better cell uniformity
[Kim et al., 1997] and are proven to promote a more tissue-like behavior, e. g., for engi-
neered cartilage constructs [Vunjak-Novakovic et al., 1996].
2. Introduction 15
Dynamic inoculation by convectional forces
For both static and mixing/agitation assisted seeding, following inoculation, the cell con-
taining scaffold has to be assembled in a cultivation device when perfusion cultivation is
pursued. This critical and potentially non-sterile step is avoided in most dynamic seed-
ing approaches utilizing perfusion, as inoculation is carried out in a combined bioreactor
setup which allows cultivation directly after inoculation without any change of the setup
[Sodian et al., 2002, Radisic et al., 2003, Wendt et al., 2003, Kitagawa et al., 2006]. For
dynamic inoculation by convectional forces the scaffold is perfused with the cell suspen-
sion, whereby flow direction can be alternated [Radisic et al., 2003, Wendt et al., 2003],
oxygen supply might be regulated [Wendt et al., 2006] and even horizontal rotation of
the reactor device is possible [Janssen et al., 2006]. Cell migration into the scaffold’s
interior, as it is coupled to flow through the scaffold therefore is strongly promoted with
cell scaffold contacts being more likely than in static approaches [Xiao et al., 1999]. Ap-
parently, this dynamic approach also causes higher mechanical forces than non-perfusing
methods which may be intolerably high for some cell types [Xiao et al., 1999] but can
be beneficial for others [Gooch et al., 2001].
If the scaffold’s pore size is very small and initial seeding density rather high, one often
speaks of filtration seeding, as cells are entrapped in the scaffold’s structure as in a fil-
tration unit [Li et al., 2001].
Compared to the seeding methods above, following dynamic inoculation cell distribu-
tion is more uniformly [Li et al., 2001, Wendt et al., 2006] with higher seeding efficien-
cies both for alternating directions [Wendt et al., 2003] and unidirectional inoculation
[Zhao and Ma, 2005]. Additional horizontal rotation is reported to yield even more
promising results, yet depending on the inner scaffold geometry [Janssen et al., 2006].
Filtration seeding leads to very high cell numbers which seems to hinder cells from fur-
ther proliferation (a phenomenon known as community effect) [Li et al., 2001]. All in
all perfused inoculation results in a good distribution of cells all over the scaffold with
high cell vitality [Du et al., 2008], but can also yield a gradient in cell number from top
to bottom [Janssen et al., 2006].
When applying dynamic perfusion inoculation, flow velocity carefully has to be adjusted:
high flow rates on the one hand wash dead and unattached cells out of the scaffold
[Radisic et al., 2003]. If the flow is too high, on the other hand, already entrapped cells
might also be washed out [Kitagawa et al., 2006]. For alternating flow directions flow
velocity must not be too low to cope with gravitational forces as otherwise cells will
settle at the bottom of the inoculation unit [Wendt et al., 2003, Timmins et al., 2007].
16 2.2 Goals of this thesis
Besides initial cell number which has to be chosen wisely as seeding efficiency drops
with higher cell numbers [Li et al., 2001], initial cell density (i.e.cell number divided by
volume utilized for perfusion) is an important parameter, hereby it is suggested to use
as less volume as possible [Kim et al., 1997, Du et al., 2008].
2.1.4 Perfusion dynamic
For both dynamic perfusion assisted cell inoculation and perfusion cultivation adjusting
the medium flow profile is crucial. Cell supply with nutrients and oxygen as well as
removal of waste substances has to be maintained at a specific level in order to support
cellular growth best. In a porous perfused system, exchanging processes are ascribed
to a superposition of convection and diffusion. The rate of exchange can be easily
manipulated by applying laminar or turbulent flow and naturally by tuning flow velocity.
[Kitagawa et al., 2006] suggest to adjust flow velocity for inoculation and cultivation
independently, using a slower profile during inoculation to decrease cell wash out and
a higher flow for cultivation, where sufficient nutrient supply is the most crucial issue.
Clearly, there is an optimum range for flow velocity beyond which cells will either die
because of lack of nutrients or due to intolerable shear forces.
As nutrient demand will rise during cellular growth, it should be considered to control
flow velocity during cultivation. This is particularly important when cultivating in single-
pass mode as economical considerations dictate flow velocity to be as low as possible.
2.2 Goals of this thesis
Regarding findings as discussed above, among others, the following questions are ad-
dressed: How can we unite cells, ceramic scaffolds and a reaction device in a way that
cells are grown in a large scaffold volume under steady conditions? Which are the main
influencing factors for cellular growth and distribution in a porous ceramic and how
do they have to be adjusted to provide cells with a micro milieu they can grow in as
close to natural conditions as possible? To answer these questions, scientists of multi-
ple disciplines have joined together in IFS3/2, a TU Berlin funded project. Members
were responsible for designing a reactor device (Department of Design Methodology),
composing the scaffold’s structure (Department for Food Chemistry) and fabricating a
porous ceramic (Department of Ceramics) that satisfies all demands for cell cultivation
(performed at Chair of Measurement and Control).
2. Introduction 17
The current work presents part of the progress of the ambitious project followed up by
the last group1. The herein pursued main goal of cultivating cells inside the porous
ceramic scaffolds breaks down into three smaller packages:
•Find and establish suitable strategies for cell inoculation and cultivation on and
in porous ceramics,
•establish qualified assay systems to verify vital growth of cells inside the ceramic
matrix,
•chose appropriate scaffold geometries and design and operate a reactor device
holding ceramic scaffolds (in strong cooperation with the departments mentioned
above).
Progress of the work is outlined as follows: At first, cultivation surface, developed biore-
actor device and operation techniques, used cell lines and employed analytical methods
are introduced (chapter 3). After a short characterization of the utilized ceramics and
flow inside the reactor system in chapter 4, two selected assay methods are discussed in
more detail (chapter 5). A large part of the work is devoted to the development of an
inoculation strategy which leads to good cell distribution inside ceramics, see chapter 6.
This is followed by a demonstration of the reactor’s feasibility as cell cultivation device
in chapter 7 and 8 and proof of its suitablity for different cell types (chapter 9). The
work closes with concluding remarks on accomplished goals and an outlook concerning
open questions (chapter 10).
1A few comments are in order with respect to some more goals set up initially for this project: It was
planned to test many different ceramic morphologies in the reactor device with and without chemical
modification of the surface. Moreover, first steps towards a mathematical description of the growth
process were envisaged. Along this line, experimental protocols were set up to measure amino acids
compositions, a reduction of fetal calf serum content in medium was obtained, growth experiments
with different surface coatings were performed, to mention just a few. However, after some time,
the solution of many more or less small problems concerning the design of the reactor modules, the
preparation of ceramics for cell cultivation, sealings, etc. and the initial wish to work with as little
serum as possible led to the decision to concentrate forces towards a thorough study concerning set-
up and operation of the reactor based on just two ceramic species using higher amounts of serum.
For this, reports about all other activities, although done, are not found here. Some information
with respect to growth on ceramic surfaces which were coated with various materials can be found
in [Driemel, 2010].
3. Materials and methods 19
3 Materials and methods
3.1 Preparation of alumina ceramics for cell culture
Ceramics were supplied by H.Schubert, TU Berlin, whose working group consolidated
the ceramic bodies as described in [Garrn et al., 2004]. As raw material highly purified
aluminium oxide (AKP50, Sumitomo Chemical CO., LTD) was used, which was foamed
by the help of protein (Albumin Bovine Fraction V, MP Biomedicals LLC) and disper-
sant Dispex A40 (Ciba Specialty Chemicals LTD). The resulting ceramics consisted of
pore sizes mainly ranging from 150-200µmwith openings of 50-150µm, porosity around
80-90% (see chapter 4.1.1) and specific surface of about 1m2/g, see figure 3.1.
All substances were milled for 15minand consolidated in a microwave oven (µWaveVa0150,
Püschner Mikrowellen Energietechnik) for 25minat 400W. Following drying in a dry-
ing closet (Memmert) for 15minat 110◦C, the ceramic bodies were cut to cylinders of
10mmdiameter using an auger. The protein scaffold was then removed at 600◦Cfor
30minin a muffle kiln (N20/HS, Naber Industrieofenbau) and ceramics were sintered
(HT04/17, Nabertherm) at 1650◦C for 60 min. Afterwards, the long foam cylinders were
cut to cylinders of 5mmor 10mmheight using a diamond saw. During this step, (mainly
carbon-associated) tool residues that could interfere with subsequent analyzation meth-
ods can be deposited on the ceramic and, therefore, are removed by an additional torch-
ing step in the muffle kiln at 600◦Cfor 30min.
After this final purifying step, ceramics should be strictly prevented from any contact
with a carbon source, being it proteins, fats, etc. Therefore, usage of gloves is strongly
suggested.
To prepare ceramic cylinders for cell culture, foams were rinsed for 2 h in aqua dest., a
detailed description of other not further specified substances and utensils is given in the
Appendix, see page 117. Then, the lateral surface was covered in teflon tape (PTFE) and
the foams were rinsed again to remove residual ceramic fragments. The teflon-sealing
step ensures medium flow through the ceramic volume as the ceramic’s flow resistance
was found to be very high and medium otherwise tends to flow past the ceramic foam
20 3.2 Reactor device for cell cultivation on alumina foams
Figure 3.1: Pore size distribution for foamed alumina. Values were obtained by graphical
analysis of scanning electron microscopy pictures (A. Berthold, TU Berlin). “Class of pore
size” describes an interval of 10 µm below the listed index.
(see chapter 4.1.2). Foams were found to be robust against disruption by flow for at
least seven weeks under continuous operation.
Sterilization of the teflon-sealed ceramics was performed at 2barand 121◦Cfor at least
20min.
3.2 Reactor device for cell cultivation on alumina
foams
Allowing for initial demands discussed in chapter 2.1.1 and 2.1.2 and considering further
requests as listed in table 3.1 a cultivation device was designed and constructed by
Bischof [Bischof and Blessing, 2006]. Considering all mandatory demands, the deduced
prototype is a modular system consisting of standard subunits (further referred to as
reactor modules) which can be easily converted to meet the needs of different operation
strategies. Three main designs were manufactured which will be presented in more detail
below. By providing nutrients out of one common medium reservoir, several reactor
modules can be connected by one another either in line to simulate sequential reaction
steps or side by side to adjust similar medium conditions, see figure 3.2a. By this,
upscaling of the system is achieved rather by multiplication than volume enlargement,
therefore, micro environments inside the porous scaffold can be maintanined. Moreover,
3. Materials and methods 21
Table 3.1: Demands for bioreactor design (as suggested by the author) at the starting
point of reactor evolution.
mandatory preferred desired
autoclavable easy to assemble sterile disassembly
bubblefree operation shakeable addition of test ports
biologically inactive materials easy to scale up
parallel or series connection of
single reactor units
continuous or pulsed
medium perfusion
laminar flow
constant inlet process values
(T, p, pO2...)
holds at least 6ceramics at
once
sterile sampling
sterile addition of supplements
close to the foams
incorporation of foams of dif-
ferent geometry
during initial testing of foams, e.g., for new cell lines, foams with different characteristics
such as porosity, thickness of the internal structures, pore volumes, etc.can easily be
tested in one run using different modules or even within one module, see below. For all
experiments described hereafter, reactor modules were connected individually to separate
medium reservoirs if not stated otherwise, see figure 3.2b. Inside one reactor module, up
to seven individual ceramic cylinders can be used in the actual design. They are either
supplied with medium in series or in parallel as explained next.
3.2.1 Tubular design for series connection of ceramics
For series connection of ceramic foams, one to seven non-sterile foams (height 10mm)
otherwise prepared as in chapter 3.1 were assembled in a series reactor module made of
poly-ether-ether-ketone (PEEK) and teflon (PTFE), which is connected to the medium
reservoir by tubings. Medium perfusion is imprinted by a peristaltic pump. To separate
foams from one another, seal rings (see Appendix) are used as spacers. A steel spring
is used to hold foams in place if less than seven foams were used. The reactor unit is
further equipped with inoculation ports made of aluminium and sight glasses (glass) to
optically ensure bubble-free operation, see figure 3.3.
For cultivation experiments, the reactor module was joined to a controlled bioreactor
22 3.2 Reactor device for cell cultivation on alumina foams
a)
b)
Figure 3.2: Module connection configurations. a) Connection of four reactor modules
to one medium reservoir by an in line configuration (left) and side by side configuration
(right). b) Testing configuration for individual reactor modules.
(Biostat B plus, Sartorius), which was prepared for cell cultivation and equipped with a
gas permeable tube. The system was then autoclaved as one unit, the bioreactor and the
reactor module were filled with medium and the reactor was put in a climatic chamber
for temperature controlled cultivation, see figure 3.4.
3.2.2 Revolver design for parallel connection of ceramics
For parallel connection of ceramic foams, six or seven non-sterile ceramic foams otherwise
prepared as in chapter 3.1 were assembled in a revolver reactor module via a foam holding
magazine made of PEEK which was selected depending on the foams outer geometry
and the inoculation strategy pursued, see figure 3.5. The magazine is embedded between
two PEEK shells and sealed by a seal ring and a silicone sealing disk. The shell’s and
magazine’s geometry was continuously adapted during the course of the project. Two
versions will be used in this work, see figure 3.6. The module is bolted by an aluminium
throw and connected by tubings either to a controlled bioreactor (Biostat B plus) or
to a medium reservoir that can be put into a CO2-incubator (see figure 3.2). Medium
perfusion then is imprinted by a peristaltic pump operated outside the incubator. The
reactor module is further equipped with inoculation and sampling ports (and if desired
with sight glasses to ensure bubble-free operation). For a complete sketch of the two main
3. Materials and methods 23
Figure 3.3: Tubular design reactor module for series connection of ceramic foams
[Bischof and Blessing, 2006]. One to seven foams can be cultivated by perfusion at once.
Figure 3.4: Connection of three reactor modules to a common medium reservoir. “Tubular
reactor” refers to the design described in chapter 3.2.1, “revolver reactor” refers to the design
described in chapter 3.2.2.
24 3.3 Inoculation procedures
Figure 3.5: Ceramic foam holding magazines for static inoculation (plain mag-
azine, upper row) and dynamic perfusion inoculation (cone magazine, lower row)
[Bischof and Blessing, 2006]. Two depth of the six bowls for the lower design are available:
5 mm and 10 mm. For cultivation, perfusion is imprinted from above (second column).
reactor designs see figure 3.6. The system was autoclaved prior to usage and residual
water was removed before the module was filled completely with culture medium.
3.2.3 Small block design for single foam analysis
For some experiments, it was beneficial to analyze a single foam at once. Therefore,
a small block capable of holding one ceramic foam was designed, see figure 3.7. The
reactor module consists of two shells which are bolt together and hold a sealing ring with
a teflon sealed ceramic foam plugged in. The module is connected to a medium reservoir
by tubings and operated in a CO2incubator. Due to the simple design and therefore lack
of ports, the foam had to be inoculated with cells outside the reactor device (chapter
3.3.1) and then aseptically assembled into the separately sterilized module. Filling of the
module and perfusion with culture medium is performed by help of a peristaltic pump
as above.
3.3 Inoculation procedures
3.3.1 Static inoculation
For some experiments static inoculation of cells was inevitable and performed as follows:
Following preparation of ceramics as in chapter 3.1, foams were soaked in culture medium
for at least 2hat 37◦C. I hereby expect the ceramic’s surface to absorb serum proteins
of the medium and therefore cause a beneficial ground for cell attachment. Cell inocu-
3. Materials and methods 25
a)
b)
Figure 3.6: Reactor modules for parallel perfusion of several ceramic foams
[Bischof and Blessing, 2006]. a) Early design with conically, flow-calming zones before and
after the foam holding magazine for potentially turbulent flow velocities. b) Final design
with less void volume and double-cone magazine to direct medium flow uniformly over all
foams (compare to figure 3.5).
26 3.3 Inoculation procedures
Figure 3.7: Sketch of a reactor module for cultivation and/or analysis of a single foam
[Bischof and Blessing, 2006].
lation is then performed from above, whereby cells (suspended in pre-warmed medium)
are dropped directly onto the foam’s surface in 100-200µlvolume at the maximum. Fol-
lowing a 30minadherence time, whose adequacy in terms of strong attachment of cells
on substrate was confirmed experimentally (data not shown), experimental procedures
continued.
If static inoculation in the bioreactor module was to be performed, sterile reactor mod-
ules filled with medium and pre-warmed at 37◦Cinside a climatic chamber or a CO2-
incubator were inoculated with cells via a needle and syringe from above, the module
was slightly shaken for better cellular distribution, and cells were allowed to settle for
2-4hbefore cultivation proceeded.
3.3.2 Dynamic inoculation by stirring or agitation
For dynamic inoculation, unsealed ceramic foams with 5 mm height and a concentrical
borehole of 2mmdiameter otherwise prepared as in chapter 3.1 were hitched into a
50mlreaction tube with a flexible steel wire. Tubes were filled with 15 ml cell suspension
in culture medium and immediately moved back and forth at 37◦Con an agitator (self-
build, suitable for use in a CO2incubator) for 72hat 1/s.
3.3.3 Dynamic inoculation by convectional forces
For perfusion assisted inoculation, revolver reactor units with conical foam holding mag-
azines as described in chapter 3.2.2 were further equipped with additional components
3. Materials and methods 27
prior to sterilization, see figure 3.8. Modules were placed in the CO2incubator and filled
with medium. Perfusion was performed overnight to ensure adjustment of temperature
(37◦C) and pH-value of medium. An upper glass olive was filled with an extra medium
volume of 10-20mland cells were inoculated in 500µl medium via the upper inoculation
port. In order to degas foams thoroughly, internal pressure was reduced for 30 s by re-
moving 1mlmedium out of the closed system via the lower sampling port and passing
it back into it. The extra medium was then pumped from the upper glass olive to the
lower, passing over the ceramic foams and flushing cells throughout the foam volume.
Pumping medium back and forth was repeated several times with one pass of cells over
the foam corresponding to 0.5cycles. During dynamic seeding, modules normally were
aligned vertically. However, for some experiments, modules were aligned horizontally
and turned for 90◦around the horizontal axis at every 0.5cycles.
For another experimental setup, modules were aligned vertically and held horizontally
every 0.5cycles for 90swith stopped pumps, rotating the horizontal alignment by 90◦for
every half cycle.
After dynamic inoculation, cells were allowed to adhere for 30minbefore perfusion
started from above.
3.4 Cultivation of cells on ceramics
The development and optimization of a perfused bioreactor system based on ceramic
foams was the main focus of the TU Berlin funded project IFS3/2, see chapter 2.2.
Therefore, perfusion cultivation of cells on ceramic foams was performed in the reac-
tor devices described above. However, for some experiments, cultivation of cells was
performed outside reactor devices, leading to different operation strategies as follows.
3.4.1 Perfusion cultivation inside the reactor device
Following attachment, cells on ceramics in reactor devices were cultivated under per-
fusion conditions, i.e.medium was pumped through ceramics. Depending on visual
evaluation of the medium’s color in the medium reservoir, which is orange for pH=7.2
and proceeds to bright yellow for pH=6 (due to pH-indicator phenole red), medium was
partially replaced at pH .6.5to prevent accumulation of lactate. In doing so, the
medium containing reservoir was replaced with another vessel containing fresh medium,
which had been equilibrated in the CO2-incubator for at least 4h. This procedure took
28 3.4 Cultivation of cells on ceramics
Figure 3.8: Setup of a reactor module and medium reservoir additionally equipped with
glass olives for dynamic inoculation in a CO2incubator. The module (left) comprises a
total volume of about 20 ml, reservoir (right) contains 100-200 ml culture medium. Medium
is pumped through the modules via silicone tubes by peristaltic pumps, which are operated
outside the incubator. The filled arrow points to an upper glass olive, the blank one to a
lower olive.
3. Materials and methods 29
about 10min, therefore perfusion cultivation was interrupted for 10minat the longest.
Sampling of the reactor module medium was performed via the lower inoculation port,
meaning medium samples of 1mlwere taken from behind the ceramics in flow direction.
3.4.2 Static cultivation of cells outside the reactor device
For static cultivation, cell populated foams were transferred to empty culture vessels
and covered with pre-warmed medium. Depending on optical evaluation of the medium’s
color, medium was replaced at pH .6.5to prevent accumulation of lactate. In doing so,
all medium was discarded and replaced by fresh medium, which had been equilibrated
in the CO2-incubator for 4h.
3.4.3 Cultivation of cells by medium convection outside the
reactor device
For convective cultivation of cells, cell-populated foams with a concentrical borehole
were hitched into 50mlreaction tubes containing 15mlpre-warmed culture medium by
a flexible steel wire. Tubes were placed on an agitator (self-build, suitable for use in a
CO2incubator) and moved back and forth at 1/sand 37◦C.
3.5 Cell lines and origin
For all experiments identifying operation strategies and parameters as well as scaf-
fold geometries, which support cellular growth best, I used the mammalian cell line
CHO-K1 (chinese hamster ovary, provided by U. Reichl, MPI Magdeburg) due to its
decade-long prominence in scientific literature and its rather simple handling techniques.
Moreover, data regarding CHO cell metabolism is widely accessible [Wu et al., 1992,
Hansen and Emborg, 1994, Altamirano et al., 2001, Deshpande and Heinzle, 2004] and
therefore interpretation of cultivation experiments is facilitated. A significant body of
work was done to adapt the CHO-K1 cell line to growth in medium with very low serum
concentrations and to experiments on ceramic foams with these adapted cells. However,
as growth on ceramics was poor, e. g., for a serum content of 1%, all but one set of
experiments in this report are based on medium with 10% fetal bovine serum. More
details relating to the serum-deprived cells can be found in the Appendix.
To prove the generality of the established bioreactor system along with the elaborated
30 3.5 Cell lines and origin
Table 3.2: Cultivation requirements for CHO-K1, A549, MDCK and human fibroblasts.
* refers to cells adapted to growth in medium containing 1% fetal bovine serum, see Ap-
pendix.
cell medium average population
doubling time (as
observed in experi-
ments)
CHO-K1 90% Ham-F12 (Sigma N4888),
10% FBS (Gibco, 10270-106, Lot
41G6972K), 2mmol/lglutamine
(Merck, 1.00289)
17 hours
CHO-K1,
1% FBS*
99% Ham-F12 (Sigma N4888), 1% FBS
(Gibco, 10270-098, Lot 41G1360K),
2mmol/lglutamine (Merck, 1.00289)
17 hours
A549 90% DMEM-Low Glucose (Sigma
D5523), 10% FBS (Gibco, 10270-
106, Lot 41G6972K), 3,7g/lsodium
bicarbonate (Merck, 1.06329)
70 hours
MDCK 90% Ham-F12 (Sigma N4888),
10% FBS (Gibco, 10270-106, Lot
41G6972K), 2mmol/lglutamine
(Merck, 1.00289)
17 hours
Fibroblasts 90% DMEM+GlutaMAX (Gibco
61965), 10% FBS (Gibco, 10270-106,
Lot 41G6972K)
3 weeks
operation modes, I furthermore introduced three more cells: A549 is a lung carci-
noma cell line [Lieber et al., 1976] widely used in toxicity-testing [Narula et al., 1998,
Davoren et al., 2006] and was gently provided by A.Hartwig, TU Berlin. MDCK (Madin-
Darby canine kidney) is a mammalian cell line known for efficient virus production
[Genzel et al., 2004, Möhler et al., 2008] and was gently provided by U.Reichl. Cells
were adapted to growth in Ham-F12 medium prior to usage in experiments. Human
primary fibroblasts from small skin biopsies were gently provided by R. Lauster, TU
Berlin. Table 3.2 summarizes the requirements for cell cultivation for all cell types.
Cultivation experiments and metabolic assays as in chapter 3.6.4 and 3.6.5 were per-
formed at 37◦Cand a CO2atmosphere of 5% for all cells. Cells were used for experiments
in their exponentially growth phase, unless otherwise stated. Cultivations preceeding de-
scribed experiments were performed according to the Appendix.
3. Materials and methods 31
3.6 Assays
To evaluate the qualification of a cultivation device, it is necessary to analyze cell number
and cell distribution on the underlying scaffold. Common attempts to detach cells and
count them by standard techniques are not easily transferred to porous structures and
failed. As many scaffolds consist of a cell entrapping structure similar to that of a porous
ceramic, recently other approaches were published, among them counting detached nuclei
[Lee et al., 1991] and measuring dissolved DNA (Cyquant®Cell Proliferation Assay Kit,
Invitrogen). Attempts to adapt these methods to the system at hand, however, failed
(more detailed results can be found in [Goralczyk et al., 2009]). I therefore established
an entirely new approach of estimating biomass on and in the scaffold by carrier hot
gas extraction, see below. Moreover, I indirectly assessed cell number by evaluating
the metabolic status of cells grown on ceramics and compare it to those of cells grown
in culture flasks, hereby measuring culture medium glucose and lactate concentration
and performing a metabolic assay based on the reduction of resazurin, see below. I am
aware, of course, that such a metabolic evaluation gives only a very rough estimate, as
factors such as inhomogeneous nutrient supply, time-varying metabolism, etc.strongly
affect the results obtained.
To visually validate cellular distribution on and inside the ceramic scaffold, I adapted
two dyeing protocols as stated below.
3.6.1 Microscopic evaluation of cell vitality by dyeing according
to FDA/EB protocol
Fluorescein diacetate (FDA) and ethidium bromide (EB) are often used dyes to differ-
entiate live and dead cells in cell cultivation. Hereby, colorless FDA on the one hand
is incorporated into cells and intracellularly hydrolyzed to fluorescein, which fluoresces
green when excited with blue light. This reaction being catalyzed by active esterases
only takes place in alive, metabolic competent cells.
EB on the other hand can only invade dead cells as it only passes disintegrated cell
membranes. Inside the cell, it binds to DNA and fluoresces red. The combination of
both dyes consequently allows a simple classification of a cell’s vital status.
Statically cultured foams were rinsed with prewarmed PBS++ (37◦C, see appendix,
chapter 10) and gently wiped with paper towels to remove liquids. If ceramics were
cultured in a reactor device, modules were rinsed with 80 ml prewarmed PBS++ for
30minand foams were removed from the reactor magazine and gently wiped with pa-
32 3.6 Assays
per towels. Foams were then soaked in a 5 µg ethidium bromide (Sigma-Aldrich) and
25µgfluorescein diacetate (Sigma) physiological salt solution (FDA/EB staining). Cells
were differentiated by fluorescent microscopy (Axioscope, Zeiss), whereby live cells flu-
oresce green and dead cells red. Digital images were assembled and annotated using
PaintShop Pro and Flash.
To include cells growing inside the scaffold into analysis, foams were cut with a scalpel
blade and cross sections were also analyzed as described.
3.6.2 Microscopic evaluation of cell distribution by dyeing
according to hematoxylin/eosin protocol
Dyeing according to a hematoxylin/eosin protocol differentiates almost all cells and tis-
sues into cellular background and nuclei. Following dyeing procedure, cells appear in
shades of pink and nuclei appear blue.
Statically cultured foams were rinsed three times with prewarmed PBS++ and gently
wiped with paper towels. If ceramics were cultured in a reactor device, modules were
rinsed with 80mlprewarmed PBS++ for 30minand foams were removed from the re-
actor magazine and gently wiped with paper towels. Foams were then fixed in ice-cold
methanol for 10minand thoroughly dried under the hood before storing at -80◦C.
For dyeing, foams were washed twice in aqua dest for 2mineach, then stained with Harris
Hematoxylin (Chroma, 2C 165) for 1 min. Cells were blued under tap water for 1minand
cross-stained in 1% Eosin (Chroma, 2C 284) for 1 min. Following a washing step in aqua
dest for 1minand dehydration in 70%, 96% and 100% ethanol for 2min each, foams
were thoroughly dried under the hood and cells were differentiated by light microscopy
(Axioscope, Zeiss). Digital images were assembled and annotated using PaintShop Pro
and Flash.
To include cells growing inside the scaffold into analysis, foams were cut with a scalpel
blade and cross sections were also analyzed as described.
3.6.3 Scanning electron microscopy of ceramic surface and cells
on ceramics
For scanning electron microscopy (SEM), following cultivation, ceramic foams were
washed twice in prewarmed PBS++, then cells were fixed in Karnovsky buffer overnight
at 4◦C. Then, samples were washed twice in cacodylate buffer and dehydrated in so-
lutions containing increasing percentages of acetone (10%, 30%, 50%, 70%, 90%) for
3. Materials and methods 33
15mineach and four times 15mineach in 100% acetone [Anton et al., 2008]. Dry foams
were imaged with an XL20 (Philips) and images further assembled and annotated using
PaintShop Pro and Flash.
3.6.4 Metabolic evaluation of glucose consumption and lactate
formation
Medium was analyzed for the concentration of glucose (Glucose GOD FS, DiaSys) and
lactate (Enzytec fluid L-lactate, DiaSys). Whereas manufacturers specify limits of quan-
tification to be 0.01g/lfor glucose and 0.003g/l for lactate, at laboratory performance
(data not shown) 95% confidence interval CI95% for measurement of glucose concentra-
tion cglucose encompassed ±0.12 ∗cglucose for concentrations of 0.25 g/l and higher. Re-
garding measurements of lactate concentration clactate,CI95% encompassed ±0.2∗clactate
for concentrations of 0.15g/land higher. For lower concentrations confidence intervals
were at ±0.18 ∗cglucose for glucose and ±1∗clactate for lactate.
For cultivation in reactor modules, accumulated glucose consumption and lactate for-
mation were calculated as follows.
Glucose consumption:
mglucose(tn) = mglucose(tn−1)+∆mglucose(tn, tn−1)(3.1)
∆mglucose(tn, tn−1) = Vsystem ∗(cglucose(tn−1)−cglucose(tn)) (3.2)
mglucose(t0) = 0 (3.3)
Lactate formation:
mlactate(tn) = mlactate(tn−1)+∆mlactate(tn, tn−1)(3.4)
∆mlactate(tn, tn−1) = Vsystem ∗(clactate(tn)−clactate(tn−1)) (3.5)
mlactate(t0) = 0,(3.6)
where mxis the accumulated mass of species x,cxis the concentration of species xin
culture medium, tnis a discrete time point of sampling and Vsystem is the total medium
volume recirculating over the module. To simplify the calculation, at each time point
medium concentration in the reservoir was assumed to equal medium concentration in
the reactor module behind the ceramic foams. Being clearly an incorrect assumption, it
still enables a simple comparison of different test runs.
34 3.6 Assays
3.6.5 Reduction of resazurin
Simple metabolic assays as MTT, XTT [Scudiero et al., 1988] or Alamar Blue assay
[Ahmed et al., 1994] are long time known and well characterized. A quite simple method
to evaluate the metabolic status of a cell culture is to monitor the reduction of resazurin
(which by the way is the main component in Alamar Blue [O‘Brien et al., 2000]). Re-
sazurin is a blue, “nonradioactive, nontoxic, water-soluble (eliminating the need for
extraction) and readily detectable by either absorbance or fluorescence spectroscopy”
[Voytik-Harbin et al., 1998] dye whose color turns into pink when reduced to resorufin
and which becomes colorless when further reduced to hydroresorufin. For resazurin re-
duction up to 50%, [Anoopkumar-Dukie et al., 2005] found a linear correlation to cell
number, whereby reduction occurs by mitochondrial, cytosolic and microsomal enzymes
[Gonzalez and Tarloff, 2001].
Resazurin assays were performed either inside the reactor module itself or for single
foams in 12-well culture plates. For assays in the reactor device, medium was discarded
and the module was rinsed with 80mlprewarmed PBS++ for 30min. PBS++ was dis-
carded and the module was filled with 40mlpre-equilibrated culture medium containing
22µmol/lresazurin (Sigma). Perfusion proceeded and medium samples were taken at
discrete times from the lower inoculation port and photometrically analyzed (Tecan
Sunrise, Tecan Trading AG) at 570 and 600 nm. Percentage of reduced resazurin was
calculated as in equation 3.7 (provided in Alamar Blue datasheet, Serotec).
p= 100 ∗117.216 ∗AS570nm −80.586 ∗AS600nm
155.677 ∗AB600nm −14.652 ∗AB570nm
,(3.7)
where pis the percentage of reduced resazurin (here resorufin), ASwavelength represents
the absorbance of the sample at a given wavelength and ABwavelength that of a negative
control.
For resazurin assays in small block reactors as described in chapter 3.2.3, following sterile
assembly of populated ceramics into the small block reactor, the reactor was filled with
6ml(Vsystem) pre-equilibrated culture medium containing 22 µmol/l resazurin. Perfusion
proceeded for 165minand medium was collected and photometrically analyzed as above.
Percentage of resorufin was calculated as in equation 3.7 and accumulation of resorufin
3. Materials and methods 35
was calculated as in equation 3.8.
mresorufin =Vsystem ∗csum ∗p−p0
100 (3.8)
csum = 5µg/ml, (3.9)
where mresorufin is the mass of reduced resazurin, Vsystem is total medium volume in
which the reduction takes place, pand p0are percentage of resorufin at the end of the
assay and its beginning, respectively. As percentage of resorufin is not zero at t=t0(as
calculated from equation 3.7), resazurin in batch medium is a mixture of both resazurin
and resorufin (corresponding to csum).
For resazurin assays for cells on ceramics performed in 12- or 24-well culture plates,
foams were covered with pre-equilibrated culture medium containing 22µmol/l resazurin.
Following static cultivation, medium was vigorously pipetted up and down and samples
were analyzed for resazurin reduction as above (equation 3.8).
For dynamic resazurin assays for cells statically cultivated in culture flasks or culture
wells, cells were covered with culture medium containing 22 µmol/l resazurin. Medium
samples were taken in intervals and cumulative reduction of resazurin was calculated as
in equation 3.10.
mresorufin(tn) = mresorufin(tn−1)+∆mresorufin(tn−1, tn)(3.10)
∆mresorufin(tn−1, tn) = Vsystem(tn)∗csum ∗1
100(p(tn)−p(tn−1)) (3.11)
Vsystem(tn) = Vsystem(tn−1)−Vsample (3.12)
where mresorufin(tn)is the accumulated mass of reduced resazurin (resorufin) at a given
sampling time point n,Vsystem(tn)is the total medium volume in which the reaction
takes place for each time interval [tn−1, tn](and which is reduced during the assay due
to sampling) and p(tn)is the percentage of resorufin at a given time point as calculated
in equation 3.7.
3.6.6 Carrier hot gas extraction
For a first approximation of the existing biomass on and in a ceramic scaffold, I devel-
oped a very new method, further referred to as carrier hot gas extraction (CHGE), in
collaboration with O.Goerke from TU Berlin. The principle relies on the coupling of
biomass to mass of elementary carbon, which can be determined by incinerating ceram-
36 3.6 Assays
ics completely and analyzing combustion gases, in particular CO2.
Following cultivation, ceramic foams were washed in prewarmed PBS++ and dried in
a laboratory oven at 50◦Covernight before storage at -80◦C. For carrier hot gas ex-
traction, foams were dried at 50◦C , then crushed to pieces between two layers of thin
tin foil (Carl Roth) in a Zwick Z00z (Zwick GmbH & Co.KG, Ulm) and thoroughly
dried at 50◦Covernight. Ceramic pieces were put in ceramic cups (IVA-Analysetechnik
e.K.) which were cauterized prior to use in order to ensure total removement of carbon.
For even heat distribution during the incinerating process, 0.3 g tin (IVA-Analysetechnik
e.K.) and 3gtungsten (IVA-Analysetechnik e. K.) were added and cups were placed in a
Carbon/Sulfur Elemental Metal Infrared Analyzer (EMIA-320V, Horbia). Incineration
occurred at 2500-3000◦C. Combustion gas was analyzed simultaneously for carbon diox-
ide and sulfur dioxide, which was computed as percentage of the ceramic pieces’ total
weight. Calculation of elementary carbon mass was performed following equation 3.13.
mcarbon =pcarbon
100 ∗mceramic,(3.13)
where mcarbon equals the total mass of elementary carbon, pcarbon is the percentage of
carbon as determined in CHGE and mceramic is the total mass of the ceramic pieces.
Calculation of elementary sulfur mass was performed analogously.
4. Reactor characterization 37
4 Reactor characterization
4.1 Ceramics
4.1.1 Porosity
Porosity of ceramics was calculated to be φ= 0.87 with a standard deviation of σ=
0.005 following equation 4.1. Hereby, three dry ceramic cylinders of 20mmheight and
25mmdiameter each (corresponding to an outer ceramic volume Vtotal of 9.8ml) were
considered for determination.
φ=Vvoid
Vtotal
=Vtotal −VAl2O3
Vtotal
(4.1)
VAl2O3=mAl2O3
ρAl2O3
(4.2)
Variables correspond to: Vtotal outer ceramic volume, VAl2O3volume of ceramic material,
mAl2O3mass of ceramic material as determined by weighing (HM-200-EC, A&D Instru-
ments Ltd) and ρAl2O3density of pure alumina which is 3.98g/ml.
In order to determine pore distribution, scanning electron microscopic images from the
top of one selected ceramic were analyzed for nominal pore count (ppi, pores per inch).
Six images were evaluated and revealed a ppi of 135, corresponding to a mean pore
diameter of 190µm. For this pore size, porosity for porous ceramics is around φ= 0.9
[Innocentini et al., 1998] which is in good agreement with experimental results.
4.1.2 Flow resistance
As a scale for flow resistance, I evaluated Darcy’s constant Kdwhereby a lower value of
Kdindicates a higher flow resistance. For a sketch of the experimental setup see figure
4.1. Ceramics of 15mmheight and 10mmdiameter were fitted in a heat shrinking tube
(SDH 19 SW, Reichelt Elektronik) of approximately 30mmlength and assembled hori-
zontally between two silicone pipes. Water was pumped through the construct whereby
volumetric flow was adjusted with a branching drain valve. Overpressure upstream of
38 4.1 Ceramics
Figure 4.1: Experimental setup for evaluation of Kd. Pressure drop is evaluated from
water column as read in vertical pipe branch. Direction of water flow is indicated by blue
arrows.
the ceramic was evaluated from height of water column in a vertical branch of the pipe
ahead of the ceramic and pressure downstream of the ceramic was assumed to equal at-
mospheric pressure due to the open setup. Water was collected at the end of the pipe and
weighed for time intervals of 30 s and volumetric flow velocity was calculated according to
equation 4.4. Following equation 4.3 Darcy’s constant (i.e.darcian permeability) was es-
timated to be about Kd= 1.7·10-10 m2with a standard deviation of σ= 7.1·10-12 m2for
10 measurement points. Permeability therewith is rather low compared to ceramics with
a higher porosity or lower ppi as summarized in [Innocentini et al., 1998]. I therefore
decided to seal ceramics laterally for perfusion experiments, as a high flow resistance
otherwise could lead to medium flowing by the ceramics.
Kd=˙
V∗η∗H
A∗∆p(4.3)
˙
V=mflow
ρH2O∗30 s(4.4)
Variables correspond to: ˙
Vvolumetric flow velocity, ηmedium coefficient of viscosity,
Hscaffold height in direction of perfusion. Aperfused scaffold area, ∆ppressure drop
across perfused scaffold, mflow water throughput for 30 s and ρH2O= 998 kg/m3.
To demonstrate efficiency of lateral foam sealing, in another experimental setup, ceram-
ics of 20mmheight and 25mmdiameter were assembled into vertically aligned silicone
pipes with variations in sealing, see table 4.1. Water was pumped through the pipes with
a constant water column above the ceramics and a constant drain of around 50ml/min.
4. Reactor characterization 39
For direct optical evaluation of flow through the ceramic, red aqueous ink was applied
directly on top and in the middle of the ceramic via a long needle and syringe and the
arising flow filament was observed and documented, see table 4.1, right column.
For evaluation of flow distribution inside the ceramics, ceramics were coated with 0.25%
bovine serum albumine (Merck, 1.12018) in aqua dest.at 37◦Covernight directly before
experimental procedures. Here, 10% coomassie blue (Fluka, 27813) in aqua dest.was
used for perfusion of ceramics. Following 5 to 10 min of perfusion, ceramics were dis-
mantled and cut vertically for determination of color distribution, see table 4.1 middle
column, blue color corresponds to flow having past through the analyzed volume element
as proteins bind the coomassie dye.
Both ink and coomassie blue distribution showed a homogenous liquid throughput
through the total ceramic volume only for thorough lateral sealing, last row in table
4.1. Therefore lateral sealing is strongly advised if homogeneous flow distribution is
sought.
4.2 Characterization of flow inside the reactor
A simple way to characterize a perfusion reactor is to compare its flow profile to that of an
idealized stirred tank reactor (STR) and that of an ideal plug flow reactor (PFR). Here,
Hagen [Hagen, 2005] considers the dimensionless Bodenstein number Bo as a measure
for the ratio of enforced convection to dispersion which is zero for an ideal STR and
converges to infinity for an ideal PFR. From that, the axial coefficient of dispersion Dax
in m2/sis calculated following equation 4.5.
Dax =u∗L
Bo (4.5)
u=L∗˙
V
V,(4.6)
where Lis the length of the reactor, ˙
Vis the volumetric flow velocity, Vis the total
reactor volume and Bo is the Bodenstein number as deduced from step experiments.
The Bodenstein number was determined for the revolver design reactor of chapter 3.2.2
for both a plain foam holding magazine and a double conical magazine as in figure 3.5.
In order to obtain comparable results, for a third experiment the bowl in the middle of
the plain magazine was clogged.
Step experiments were performed as follows: Modules were assembled as for cultivation
40 4.2 Characterization of flow inside the reactor
Table 4.1: Flow distribution inside ceramics for different sealing strategies, see text for
further explanation. Blue arrows indicate direction of flow; a deeper color for a volume
element indicates higher throughput for that area.
sealing setup coomassie blue staining evalution of ink flow
filament
ceramic hooked inside
pipe, no lateral sealing
dye is found solely on top of
the ceramic with poor inva-
sion in the middle
ink mainly passes by
the ceramic
ceramic hooked inside
pipe, lateral sealing at
top of ceramic
homogeneous distribution
of dye at top of ceramic
ink flows laterally into
the ceramic and leaves
it directly behind the
sealing
ceramic hooked inside
pipe, lateral sealing at
bottom of ceramic
inhomogeneous distribution
of dye
ink flows directly into
the ceramic and es-
capes all over the bot-
tom of the ceramic
ceramic grafted into
pipe, lateral sealing
with teflon tape
homogenous distribution of
dye all over the ceramic’s
volume
ink flows directly into
the ceramic and es-
capes all over the bot-
tom of the ceramic
4. Reactor characterization 41
0.8
1
1.2
plain foam magazine, 7 ceramics plain foam magazine, 6 ceramics
cone magazine ideal PFR
0
0.2
0.4
0.6
0.0 0.5 1.0 1.5 2.0 2.5
F(θ)
θ
Figure 4.2: Color step experiments for different foam holding magazine geometries (data
from one experiment per reactor design). Shown is normalized extinction F(θ)over nor-
malized time θ.
with 6 or 7 ceramic foams of 10mmheight as in chapter 3.2.2 and filled with aqua dest.
Perfusion was performed from above at 1.4mlmin−1. As a tracer, 1mg/mlresazurin in
water was joined at time t0and samples of the flow off were analyzed photometrically
at 600nm. For evaluation, time was normalized to dimensionless time θvia equation
4.7 and extinction was normalized to dimensionless F(θ), which is the function of sum
of retention time via equation 4.9.
θ=t
teff
(4.7)
teff =
m
X
n=1
(1 −F(tn)) ∗∆tn−1,n (4.8)
F(θ) = E(θ)
E∞
,(4.9)
where tnis a time point of sampling, tmis the time point of sampling when E(tn)equals
E∞,E(θ)is the fluid extinction at each normalized time point and E∞is the maximal
extinction reached in the experiment. F(tn)is calculated analogously to equation 4.9.
Figure 4.2 shows the experimental results for reactor characterization.
Bodenstein number and coefficient of axial dispersion were calculated as in equation 4.10
and equation 4.5 and are listed in table 4.2.
Bo =π∗(2 ∗a)2,(4.10)
42 4.2 Characterization of flow inside the reactor
Table 4.2: Reactor characteristics for plain and cone magazine holding 6 or 7 ceramic
foams.
magazine geometry slope Bo Dax [m2/s]
plain, 7 foams 0.9 10 0.00057
plain, 6 foams 0.8 9 0.00065
conical 1.1 16 0.0003
where ais the slope of the graph calculated from two points around θ= 1.
Comparing flow for 6 and 7 ceramics in a plain holding magazine shows that a set-up
of seven ceramics resembles more an ideal PFR, see figure 4.2. Therefore, it is assumed
that the open structure with a flow path in the middle of the magazine maintains a
more laminar flow profile than for a magazine with a blocked flow path in the middle.
However, the difference between 6 and 7 integrated ceramics is not very high, see table
4.2.
Clearly, the shape of the foam holding magazines as well as the shape of the magazine
shells influences the perfusion flow profile. A conical magazine with narrow flow entry
resembles more a PFR as the Bodenstein number is higher and the color profile in figure
4.2 is closer to the ideal profile than for a plain magazin in a reactor with a widening
shell. Nevertheless, the difference between both is rather small, and therefore I rejected
the widening shell design for the narrow one in favor of the material savings.
Regarding the magazine’s shape, I performed a similar experiment as described above
with a tracer dye sticking to the ceramics. I hereby proved that flow mainly passes
through the foam in the middle of the plain magazine for the 7-foam assembly - as
optical evaluation revealed a strong staining of the middle foam towards the ceramics at
the rim (data not show). Therefore, during the course of the presented work, I focused
on employing the conical magazine to get a homogenous cell distribution and medium
supply among all foams.
5. Analysis of CHO-K1 by resazurin assay and carbon content determination. 43
5 Coupling of cell number to
metabolic reduction of resazurin
and to elementary carbon
analyzed in CHGE
5.1 Reduction of resazurin by CHO-K1
5.1.1 Determination of rate of reduction
In order to evaluate the applicability of the resazurin assay for determination of the
CHO-K1 cell number, two dimensional experiments were performed in cell culture
flasks and plates. Different cell numbers (passage 3) were inoculated in culture medium
onto flasks or plates with varying surface area to modulate cell densities around 4 ·104
cells/cm2, which are assumed to be exponentially growing, and cell densities around
8·104cells/cm2or higher, which are assumed to be in a stationary phase (boundaries
were obtained in earlier experiments, data not shown). Cells were allowed to adhere
for 2h, then medium was discarded, cells were washed with PBS++ and covered with
medium containing 22µmol/l resazurin. Medium volumes were set to be 20mlfor flasks
with a growth surface of 175cm2, 10mlfor 75 cm2flasks, 6 ml for 25 cm2flasks and 5 ml for
6-well plates with a growth surface of 9.6cm2for each well. Samples of 0.5 to 1.5mlwere
taken in intervals from 30 to 45min and analyzed for reduction of resazurin as described
in chapter 3.6.5. The top figure 5.1 shows the dynamic of resazurin reduction as cal-
culated from equation 3.7 and the bottom figure 5.1 the dynamic as calculated from
equation 3.10. To distinguish the individual experiments the total numbers of cells used
are given. See table 5.1 for a reference concerning the cell number with respect to growth
area. Filled circles in figure 5.1 correspond to experiments with just 5mlstarting vol-
ume for resazurin assay. Due to mixing with residual PBS++ from the washing step it
44 5.1 Reduction of resazurin by CHO-K1
30
35
40
45
30
35
40
45
1.50E+07 4.64E+06 2.50E+06 2.50E+06 1.00E+06 3.00E+05 1.50E+05
inoculated total cell number:
0
5
10
15
20
25
30
0
5
10
15
20
25
30
resorufin [%]
resorufin [%]
0.2
0.3
0.4
0.5
0.6
15
20
25
accumulated resorufin [µg]
accumulated resorufin [µg]
-0.3
-0.2
-0.1
0
0.1
0.2
-5
0
5
10
0 50 100 150 200 250
accumulated resorufin [µg]
accumulated resorufin [µg]
assay time [min]
Figure 5.1: Dynamic of resazurin reduction for CHO-K1. Circles represent experiments
where cell density was around 4 ·104cells/cm2, squares represent experiments with inocu-
lated cell density around 8 ·104cells/cm2or higher. Experiments were repeated with n as
the numbers of repetitions varying from 1 to 6 as indicated in table 5.1. Top: Percentage
of resorufin. Bottom: Accumulated resorufin, filled circles belong to vertical axis on the
right side.
5. Analysis of CHO-K1 by resazurin assay and carbon content determination. 45
Table 5.1: Values for experimental evaluation of rate of reduction for CHO-K1.
inoculated
cell number
cell density
[cells/cm2]
(culture flask
area)
n time points for
linear regression
slope
[µg/min]
coefficient
of deter-
mination
(R2)
1.5·1078.6·104(175) 2 30; 60; 90 0.137 0.988
4.64·1066.2·104(75) 1 30; 60; 90; 123 0.04 0.999
2.5·1062.6·105(9.6) 3 30; 60; 105; 135;
165
0.025 0.995
2.5·1063.3·104(75) 3 60; 90; 123; 150;
185; 215
0.026 0.997
1.0·1064.0·104(25) 3 60; 90; 123; 185 0.014 0.999
3.0·1053.1·104(9.6) 6 60; 135; 165; 197 0.003 0.994
1.5·1051.6·104(9.6) 6 135; 165; 197;
227
0.002 0.999
is assumed that absorbances for resazurin species at t= 0 are significantly lower than
those from batch medium, leading to a different value pfor initial resorufin percentage
(see equation 3.7). As initial absorbances from culture vessels were not measured but
set to equal those from batch medium, calculation of accumulated resorufin therefore
may result in a decline in mass of resorufin at first as it is the case for some experiments.
From figure 5.1, time points were selected for linear regression, whereby time points were
excluded that belong to samples with difficulties during sampling or following analysis,
as well as time points when reduction apparently converges saturation. Table 5.1 sum-
marizes the results of linear regression.
For all experiments, cell free medium containing resazurin was included as a negative
control. Herein, spontaneous reduction of resazurin was not observed.
There is a strong coherence of the rate of reduction of resazurin as expressed in the slope
of the reduction dynamic and the inoculated cell number. Figure 5.2 shows a linear re-
lationship for a wide range of cell numbers. With a coefficient of determinination R2of
0.997, following linear regression the cell number can be evaluated according to equation
5.1.
Obviously, the growth phase does not influence the metabolic status regarding resazurin
reduction, as for 2.5·106cells inoculated with a high cell density (stationary growth
phase) the same slope was observed as for 2.5·106cells inoculated with a low density
(exponential growth phase), see table 5.1. Moreover, the slope for 1.5 ·107cells (station-
46 5.1 Reduction of resazurin by CHO-K1
0.08
0.12
0.16
rate of resazurin reduction [µg/min]
0
0.04
0.08
0.0E+00 5.0E+06 1.0E+07 1.5E+07
rate of resazurin reduction [µg/min]
inoculated cell number
Figure 5.2: Slope determined from figure 5.1 plotted against inoculated cell number.
ary growing) also is in alignment, see figure 5.2.
ncell =slope (in µg/min) −0.00182 µg/min
0.009 µg/min ∗106(5.1)
As with increasing accumulating mass of resorufin the percentage of reduced resazurin
approaches saturation, for later experiments it is strongly recommended to evaluate the
reduction dynamic rather than an endpoint value. Out of the first measurements, a time
range should be estimated in which reduction is nearly linear and the slope should be
determined for that range as shown above. Then, equation 5.1 gives a good hint of the
cell number ncell of underlying experiments.
For endpoint assays, however, the slope can be estimated by dividing the difference in
resorufin content for the start and end of assay by the assay time. The determined cell
number as evaluated by equation 5.1 is then a lower bound for cell number taking part
in the assay, at least.
5.1.2 Modeling resazurin reduction
To gain a more elaborate insight into resazurin reduction, data from chapter 5.1.1 was
evaluated to design a mathematical model of resazurin reduction in collaboration with
G.Gelbert, TU Berlin. The model describes the percentage of resorufin over time de-
pending on the underlying cell number for resazurin assays. Model assumptions were
formulated as follows:
5. Analysis of CHO-K1 by resazurin assay and carbon content determination. 47
•The dynamic of resazurin reduction to resorufin was set to follow a kinetic with a
power law influence of resazurin concentration and a linear influence of cell number.
Linear reduction within the tested frame was assured experimentally, see figure 5.2.
However, for adherently growing cells, metabolic rates are not only determined by
cell number but also by cell density on a growth surface. This influence will not
be modeled.
•Further reduction of resorufin to hydroresorufin was assumed to be very slow for
low concentrations of resorufin, therefore, reduction was set equal to zero. Fur-
thermore, hydroresorufin, which most likely occurs at very low concentrations in
batch medium at t=t0, but is not determined experimentally, is assumed to equal
zero at assay start.
•Reduction of resazurin is assigned to enzymatical catalysis by cells solely. This
assumption was validated in experiments where cell and ceramic free resazurin
medium was analyzed in a time frame of several hours without significant increase
in resorufin (data not shown).
•Cell number was assumed to be constant during assay time, as the population time
of CHO-K1 is around 20h.
The deduced model equations read as follows:
dmR
dt =−µmax ∗xK
R∗ncell (5.2)
dmF
dt =µmax ∗xK
R∗ncell (5.3)
where mRand mFare the total mass (in µg) of resazurin and resorufin, respectively,
xR=cR
c∗
Ris a normalized concentration of resazurin with cRbeing the concentration of
resazurin and c∗
R= 1 µg/ml. cRis calculated by cR=mR
Vwith Vbeing the experimental
volume (in ml). ncell is the number of cells and µmax and Kserve as model parameters.
Due to sampling, Vis reduced during the experiment, which was taken into account
for the model. Percentage of resorufin was calculated by pF= 100 ∗cF
csum , whereby cF
is the concentration of resorufin and csum the total concentration of all reaction species
resazurin and resorufin which is 5µg/ml. Model parameters were identified from 4 out
of 7 experiments described in chapter 5.1.1 using Matlab’s “fminsearch” optimization
with a least squares error. As some experiments showed an initial drop of percentage of
resorufin (see figure 5.1) probably due to mixing with PBS++ as discussed in chapter
48 5.1 Reduction of resazurin by CHO-K1
5.1.1, simulations for parameter identification and later also validation were performed
from t= 30 min which was the time point of first measurement of medium out of each
culture vessel. Parameters obtained read as follows:
µmax = 2.5·10-10 µgmin−1cell-1 (5.4)
K= 2.7(5.5)
Figure 5.3 shows the experiments used for parameter identification. The quality of the
model was validated by simulating three experiments from chapter 5.1.1 not used for
parameter identification, see figure 5.4.
Acceptable results were obtained for modeling resazurin reduction, see figure 5.4 and 5.3.
However, measurement data from the experiment with a very high cell number of 1.5 ·107
showed a slowdown in resazurin reduction during the assay which might be attributed
to a decrease in metabolic rates. As reduction of medium volume during sampling here
was performed faster than for all the other experiments, lactate accumulation and/or
lack of glucose in the culture vessel could have occurred, resulting in a decrease in cell
metabolism during the course of the experiment.
5.1.3 Adaptation of the resazurin reduction model to describe
bioreactor performance
For resazurin assays performed inside the reactor modules, a simple model was designed
that returns the percentage of reduced resazurin depending on the underlying growth
performance of cells in the reactor. Therefore, the total volume of the reactor was divided
into three main compartments being A the medium container, B the foam containing
magazine where the cell catalyzed reduction takes place, and C a small volume behind
the magazine from where sampling is performed (see figure 5.5). Model assumptions
read as follows:
•Flow inside modules and periphery is assumed to be laminar without backmixing
or diffusion of reaction species. Modules are specified as ideal plug flow reactors.
•Dotted areas in figure 5.5 do not take part in reaction or flow although filled with
medium; diffusion from those areas is assumed to equal zero.
•Cells are assumed to be equally distributed inside compartment B. A medium
volume entering compartment B is immediately mixed ideally and supposed to
5. Analysis of CHO-K1 by resazurin assay and carbon content determination. 49
0 50 100 150 200 250
10
15
20
25
30
35
40
45
resorufin [%]
1.5E+05 / 1.6E+04
0 50 100 150 200 250
10
15
20
25
30
35
40
45
1.0E+06 / 4.0E+04
0 50 100 150 200 250
10
15
20
25
30
35
40
45
resorufin [%]
assay time [min]
2.5E+06 / 2.6E+05
0 50 100 150 200 250
10
15
20
25
30
35
40
45
assay time [min]
measurement data
simulation
4.6E+06 / 6.2E+04
Figure 5.3: Model fit for resazurin reduction by CHO-K1. Figures show experimental
data and simulation results for experiments used for parameter identification, cell number
and cell density in cells /cm2is indicated on top of each panel.
50 5.1 Reduction of resazurin by CHO-K1
0 50 100 150 200 250
10
15
20
25
30
35
40
45
resorufin [%]
2.5E+06 / 3.3E+04
0 50 100 150 200 250
10
15
20
25
30
35
40
45
3.0E+05 / 3.1E+04
0 50 100 150 200 250
10
15
20
25
30
35
40
45
resorufin [%]
assay time [min]
measurement data
simulation
1.5E+07 / 8.6E+04
0 50 100 150 200 250
0.5
1
1.5
2
2.5
3
3.5
4
4.5
assay time [min]
concentration of resazurin species [µg/ml]
cresazurin
cresorufin
Figure 5.4: Model validation for resazurin reduction by CHO-K1. Figures show experi-
ments not used for parameter identification. Cell number and cell density in cells /cm2is
indicated on top of each panel. Bottom right figure shows the concentration development
for resazurin and resorufin for the panel above.
5. Analysis of CHO-K1 by resazurin assay and carbon content determination. 51
Figure 5.5: Scheme for division of reactor volume into compartments for flow modeling.
Three different volumes for A, B and C, respectively, were calculated depending on maga-
zine geometry and ceramics’ height. a) Geometry for 7 ceramics with 10 mm height in plain
magazine, b) geometry for 6 ceramics with 10 mm height in conical magazine, c) geometry
for 6 ceramics with 5 mm height in conical magazine.
leave B directly.
•Ideal mixing is assumed for compartment A.
•Sampling is performed from compartment C, reducing the medium volume in the
tube going to compartment A. Hereby, backmixing in C is assumed to be zero and
the tube to A fills with air from the medium container.
•Reduction of resazurin is only attributed to enzymatic activity of cells. This as-
sumption is valid, as no reduction of resazurin was observed during a resazurin
assay inside a reactor module with 6 cell free ceramics over a period of 3h.
Moreover, model assumptions as stated in chapter 5.1.2 apply.
Volumes were determined experimentally and by help of mechanical drawings of mod-
ules (see Appendix), confer to table 5.2. Total medium volume inside the reactor and
periphery at t=t0was 23mlfor module type a and 40mlfor module types b and c in
figure 5.5, flow velocity was 1.4 ml min−1for all simulated experiments. For simplifica-
tion, the reaction was assumed not to start before the reactor volume was totally filled
with medium (see chapter 3.6.5). During dynamic resazurin assay, the total volume
is reduced due to sampling, which had to be taken into consideration for the model.
Therefore, the model is separated into two parts. One describing operation between two
52 5.1 Reduction of resazurin by CHO-K1
Table 5.2: Conditions for modeling resazurin reduction inside bioreactors. Lag times TA,0,
TB,0and TC,0were calculated for a flow of 1.4 ml/min and starting medium volume of 40 ml.
module type as
in figure 5.5
volume A
(t=t0)
[ml]
volume B
[ml]
TA,0[min] TB,0[min] TC,0[min]
a) 5.8 4.9 8.7 5.6 4.6
b) 16.5 4.2 8.7 5 4.1
c) 17.1 2.1 9.8 5 4.1
samples (I) and one describing the process after sample removal (II). In part I medium
circulates all over three compartments A, B, and C and all tubes are thoroughly filled
with medium. Resorufin concentration profiles are described by model equations 5.6 to
5.10. Following sampling, the tube going to compartment A is partly filled with air,
which was sucked in during the sampling process. Therefore, no medium passes into A
for a time period of tsample =sample volume
flow velocity . Hence, in part II equations 5.9 to 5.10 and
equations 5.11 to 5.13 are solved for a time period tsample.
Following part II, all compartments and tubes are filled with medium again but as a
liquid volume has been removed during sampling, delay times of part I do not apply for
equations 5.6 an 5.7. Therefore, at first equations of part I are solved with a revised
delay time T0
B,0for a duration t0until the back drawn medium entirely leaves the tube
to compartment A (part I’). Hereby T0
B,0is calculated as TB,0+tsample, whereby TB,0is
the lag time needed for a volume element of B to approach to compartment A, see table
5.2. t0is calculated as TB,0−TC,0−tsample with TC,0being the lag time needed for a
volume element of B to approach to compartment C.
Then, delay times of part I apply again and model solving continues with equations from
part I as above.
I:
dmA,R
dt =˙
V∗(cB,R,0−cA,R)(5.6)
dmA,F
dt =˙
V∗(cB,F,0−cA,F )(5.7)
dVA
dt = 0 (5.8)
dmB,R
dt =˙
V∗(cA,R,0−cB,R)−µmax ∗xK
B,R ∗ncell (5.9)
dmB,F
dt =˙
V∗(cA,F,0−cB,F ) + µmax ∗xK
B,R ∗ncell (5.10)
5. Analysis of CHO-K1 by resazurin assay and carbon content determination. 53
II:
dmA,R
dt =−˙
V∗cA,R (5.11)
dmA,F
dt =−˙
V∗cA,F (5.12)
dVA
dt =−˙
V(5.13)
Variables are calculated as
cy,z =my,z
Vy
(5.14)
cy,z,0=my,z,0
Vy,0
,(5.15)
where my,z and cy,z are mass and concentration of resazurin species z (resazurin and
resorufin) in compartment y (A, B), respectively, xB,R =cB,R
c∗
B,R is a normalized concen-
tration with c∗
B,R = 1 µg/ml. ˙
Vis the volumetric flow during the assay. cB,z,0is the
concentration of resazurin species z in compartment B at time t−TB,0. Likewise, cA,z,0
is the concentration of resazurin in compartment A at time t−TA,0as well as VA,0is
the volume in compartment A at time t−TA,0, where TA,0is the lag time needed for a
volume element of A to approach to compartment B. Vyis the total volume of medium
in compartments A and B, respectively. ncell is the number of cells taking part in the
reaction and µmax and Kare the model parameters as identified in chapter 5.1.2.
Concentrations of reaction species in compartment C were calculated to be those of
compartment B at time t−TC,0.
The model is solved with “dde23” in Matlab. For validation, 3.15 ·106cells/foam were
statically inoculated on dry standard foams as in chapter 3.3.1 and allowed to adhere for
2h. Six foams were then assembled aseptically into reactor module type b) as in figure
5.5, washed with PBS++ and subjected to a resazurin assay as in chapter 3.6.5. The
experimental outcome and simulation results are depicted in figure 5.6. Simulation and
experimental data were in adequately good agreement, so for the experiments described
later in this work this model can be considered a good first evaluation of the dynamics
of resazurin reduction and therefore allows an estimation of the cell number.
54 5.2 Carbon content of cells cultivated on ceramics
0 20 40 60 80 100 120 140 160
14
16
18
20
22
24
26
28
30
32
34
assay time [min]
resorufin [%]
measurement data
simulation for compartment C
simulation for compartment B
simulation for compartment A
1.9E+07 cells in compartment B
0 20 40 60 80 100 120 140 160
3.4
3.5
3.6
3.7
3.8
3.9
4
4.1
4.2
4.3
4.4
assay time [min]
resazurin [µg/ml]
compartment B
compartment A
Figure 5.6: Model validation for reduction of resazurin inside the bioreactor. Solid lines
represent simulation data, + is for experimental measurement data. Left: percentage of
reduced resazurin (resorufin) in reactor compartments. Right: Simulation of concentration
development for resazurin in compartments A und B.
5.2 Carbon content of cells cultivated on ceramics
5.2.1 Carbon content of pure cells
In order to determine a correlation between cell count and measurable carbon content,
cells were centrifuged, counted by a hemocytometer and aliquots of cells in PBS were
subjected to carrier hot gas extraction (CHGE) as in chapter 3.6.6. Total carbon and
sulfur content are depicted in figure 5.7, whereby cell free samples were below detection
limits and therefore excluded for illustration. Whereas carbon content correlated very
good with cell number (regression coefficient 0.999), analyzed sulfur content did not
correlate with cell number at all, see sulfur measurements and standard deviations in
figure 5.7. From carbon measurements in figure 5.7, the following linear regression for
the cell number ncell can be estimated:
ncell =mc
0.000116 µg,(5.16)
where mcis analyzed carbon content inµg.
5. Analysis of CHO-K1 by resazurin assay and carbon content determination. 55
400
600
800
1000
8000
10000
12000
14000
sulfur content in µg
carbon content in µg
carbon sulfur
-400
-200
0
200
0
2000
4000
6000
0.0E+00 5.0E+07 1.0E+08 1.5E+08
sulfur content in µg
carbon content in µg
inoculated cell number
Figure 5.7: Carbon and sulfur content of CHO-K1 in PBS. The number of repetitions n
ranges from 3 to 6.
5.2.2 Carbon content of cells on foams
To determine the influence of ceramics and medium composites on carbon analysis, ce-
ramic cylinders of 10mmheight, 10mmdiameter and with one blind hole from above
with 6mmdiameter and 7mmdepth were prepared as in chapter 3.1 with additional
teflon sealing for the ceramics’ bottoms. Ceramics were dried overnight at 50◦Cin a
laboratory oven prior to static inoculation of cells in PBS or culture medium.
For cells inoculated in PBS, foams with removed teflon tape were dried over night at
50◦Cin a laboratory oven directly after inoculation.
Cells inoculated in medium were allowed to settle for 135min, then the teflon tape was
removed and ceramics were washed in PBS and dried overnight at 50◦C. To simulate cul-
tivation of cells on ceramics under perfusion, some ceramics were analyzed for resazurin
reduction 120minafter inoculation in medium. Single ceramics hereby were assembled
sterile into the small block reactors as in chapter 3.2.3 and perfused with 6ml culture
medium containing 22µmol/lresazurin for 165min. For this, the teflon tape sealing at
the ceramics’ bottom was removed. Total medium was collected and analyzed for re-
sazurin reduction as in chapter 3.6.5 and foams inside the reactor modules were rinsed
with 30mlPBS for 20min. Ceramics were disassembled, the teflon tape was removed
and foams were dried overnight at 50◦C.
Ceramics were stored at -80◦Cprior to CHGE as in chapter 3.6.6 and carbon content
was calculated according to equation 3.13.
Whereas carbon content of cell free ceramics incubated in PBS was below detection limit,
cell free ceramics incubated in medium showed a carbon content of about 500 µg per foam,
56 5.2 Carbon content of cells cultivated on ceramics
2500
3000
3500
carbon content in µg
PBS inoculation
medium inoculation
medium inoculation, perfusion washed in PBS
0
500
1000
1500
2000
0.0E+00 5.0E+06 1.0E+07 1.5E+07 2.0E+07 2.5E+07
carbon content in µg
inoculated cell number
Figure 5.8: Carbon content of CHO-K1 on ceramics. Cells were inoculated in PBS or
medium as stated in legend and dried prior to CHGE. For medium inoculated cells, ceramics
were washed in PBS before drying, see text for further explanation. n ranges from 1 to 3.
although ceramics were thoroughly washed in PBS prior to CHGE, see filled circles in
figure 5.8. This number was reduced to only 70 µg per foam if foams were perfused with
PBS in small block reactors (filled squares in figure 5.8). Therefore, the elimination of
residual medium inside ceramics’ pores is a crucial step for CHGE and was performed
for all subsequent experiments utilizing reactor modules by perfusion washing in PBS
for 30min.
Ceramics with cells inoculated in PBS without further cultivation showed a linear corre-
lation between carbon content and inoculated cell number. Similar results were obtained
for ceramics with cells inoculated in medium and cultivated for another 285 min (partly
during resazurin assay) prior to perfusion washing with PBS and subsequent CHGE.
Following resazurin assays, ceramics showed a spectrum of colors ranging from blue to
pink, depending on reduction of resazurin and inoculated cell number. Nevertheless,
this variation in color did not influence the outcome of carbon analysis, see filled squares
in figure 5.8. However, ceramics with cells inoculated in medium and washed in PBS
without perfusion after 135minof cultivation did not show such a good correlation con-
cerning carbon content and cell number (filled circles in figure 5.8), probably due to poor
medium elimination as ceramics solely were flushed in PBS for washing. Table 5.3 gives
an overview on calibration curves deduced from figure 5.8.
Nevertheless, for those foams with experimental performance closest to the following ac-
tual experiments (filled squares in figure 5.8) foams with the highest inoculated cell num-
ber of 1.88·106cells/foam seemed to match more the trend of foams not so thoroughly
5. Analysis of CHO-K1 by resazurin assay and carbon content determination. 57
Table 5.3: Linear regression for carbon analysis as in figure 5.8. ncell, cell number; *
marks linear regression as for the experiment in the above row but with excluded value for
1.88 ·106cells/foam, see text for explanation.
method carbon content following linear
regression [µg]
coefficient of
determination
(R2)
inoculation in PBS, no fur-
ther cultivation; n=1-2
0.000088*ncell 0.993
inoculation in medium,
washing in PBS, cultivation
for 135min; n=1
0.000119*ncell+647 0.972
inoculation in medium, cul-
tivation for 135 min, re-
sazurin assay for 165min,
perfusion washing in PBS;
n=2-3
0.00015*ncell+21 0.991
inoculation in medium, cul-
tivation for 135 min, re-
sazurin assay for 165min,
perfusion washing in PBS*;
n=2-3
0.000122*ncell+96 0.989
washed (filled circles). This most probably is attributed to the inoculation method,
where a highly concentrated cell suspension is dropped onto the foam, leading to cell
blockage with therefore poor washing in PBS and hereby unwanted measurement of
medium species in CHGE. Therefore, this value was excluded for linear regression (see
last row in table 5.3) leading to equation 5.17 for the calculation of cell number ncell for
the following experiments.
ncell =mc−96 µg
0.000122 µg,(5.17)
where mcis analyzed carbon content inµg.
Equation 5.17 clearly only applies if cell distribution on and in ceramics is homogeneous
without cell blockage, allowing efficient PBS-washing before carbon analysis in CHGE.
Moreover, abundance of matrix proteins as described in chapter 7.1 and chapter 8 will
influence cell number estimation derived from equation 5.17 significantly due to their
high carbon content, which will be assigned to a high cell number mistakenly.
6. Influence of mode of inoculation on cellular growth and distribution 59
6 Influence of mode of inoculation
on cellular growth and
distribution
As discussed in chapter 2.1.3 initial cellularity of a porous scaffold strongly influences
further cell growth inside the scaffold. To obtain a high number of cells homogeneously
distributed all over the ceramic volume different approaches of inoculation were evalu-
ated. Besides experiments described below, I also employed two gel inoculation tech-
niques hoping for strong cell penetration of the ceramic due to impeded cell flush out.
Whereas [Li et al., 2008] and [Blan and Birla, 2008] found promising results, in the sys-
tem at hand neither gelation of a 1% sodium alginate (Fluka, 71238)/cell suspension
in ceramics with 0.2mMcalcium chloride (Merck, 1.02381) nor application of different
concentrations of Matrigel (BD Biosciences, 356234) in cell suspensions on ceramics re-
sulted in even cell distribution. Moreover, cells were found to die inside the viscous
matrix, even though flow channels were drilled into the constructs to improve nutrient
access. Cell number and distribution inside the ceramics was found to be the poorest
of all realized experiments, therefore a more detailed experimental description of gel
inoculation was neglected for the current chapter.
6.1 Static inoculation
6.1.1 Static inoculation on foams in culture plates
To prove the ability of cells to proliferate on alumina, CHO-K1 were statically inoculated
as in chapter 3.3.1 with 2.7·105and 1.6·105cells/foam, respectively, and statically
cultivated (chapter 3.4.2) for five to eight days. Ceramic foams were transferred to
new 12-well culture plates before resazurin assay to avoid analysis of cells attached to
the bottom of the well. Resazurin assays (see chapter 3.6.5) were performed for 4 h at
60 6.1 Static inoculation
6
8
10
reducecd resazurin [µg] in 4h
-assay
1.6E+5 cells/foam inoculum 2.7E+5 cells/foam inoculum
0
2
4
0 40 80 120 160 200
reducecd resazurin [µg] in 4h
cultivation time [h]
Figure 6.1: Mass of reduced resazurin (n=3) after 4 h resazurin assays for two inoculation
densities following static inoculation and static cultivation of cells on ceramic foams in
culture plates.
selected days with results depicted in figure 6.1.
Metabolic reduction of resazurin clearly increased during cultivation time for 2.7 ·105
and 1.6·105cells/foam inoculation densities, respectively. Moreover, with equation 5.1
on page 46, mean cell number per foam was estimated to be 3.9·106for inoculation
with 2.7·105cells and cultivation for 8 days, and to be 1.3 ·106for inoculation with
1.6·105cells and cultivation for 5 days. Therefore, cell number apparently multiplied
by factor 14 and 8, respectively, and proliferation can be assumed. The increase in
cell number, however, is significantly below the theoretical limit assuming a doubling
time of 20hand absence of contact inhibition. Likewise, dyeing of cells on the ceramic
surface and microscopic evaluation as in chapter 3.6.2 displayed poor cellular growth
on the lower side of the foams and cellular ingrowth only to approximately 200µm. In
another experiment, foams were subjected to agitation cultivation as in chapter 3.4.3
210minafter static inoculation with 5.5·105and 2.2·106cells/foam. After one, three
and five days of cultivation, no cellular growth deeper than 200 µm was observed for all
foams. Moreover, cells attached to the foam appeared less vital and huge cell numbers
were observed in the surrounding medium (as counted by help of a hemocytometer), see
figure 6.2.
6.1.2 Static inoculation into the tubular reactor
For static inoculation into tubular reactors (see figure 3.3), 4 ceramic foams were as-
sembled into the reactor in such a way that the first foam’s upside was located next to
6. Influence of mode of inoculation on cellular growth and distribution 61
60
80
100
cell count in % of inoculum
inoculation
live cells in medium after 72h agitation
dead cells in medium after 72h agitation
0
20
40
60
5.5E+5 cells/foam 2.2E+6 cells/foam
cell count in % of inoculum
cell count for inoculation
foam 1 foam 2
foam 1 foam 2
Figure 6.2: Cell leakage for cells statically inoculated on ceramic foams in culture plates
and cultivated with agitation for 72 h. Shown is cell count in medium for two foams for
each inoculation density.
the upper inoculation port. Following preparation as in 3.2.1 and vertical alignment of
the module in the climatic chamber, 1·107CHO-K1 cells were inoculated directly onto
the first foam and allowed to settle for 4hbefore perfsuion was imprinted from above
at 0.08mlmin−1cm−2for 20d. The module was then disassembled and single foams
were subjected to microscopic evaluation as in chapter 3.6.2. While disassembling, thick
clumps of cells and cell debris were found between the sealing rings and the module
shell, probably remaining cells from inoculation. Nevertheless, cells were found to be
very dense on top of the first foam, and much lesser dense on all other top and bottom
sides of the foams with very inhomogeneous cell distribution. No cells were found in the
foam’s insides.
6.1.3 Static inoculation into revolver reactors
For static inoculation into revolver reactors with plain magazines (see figure 3.6a), seven
ceramic foams were inoculated at once with 7·106CHO-K1 cells as in chapter 3.3.1
and cultivated for 31 d under perfusion conditions with 0.14 ml min−1cm−2as described
in chapter 3.4.1. The reactor module was then disassembled and three foams from the
magazine’s rim (positions 1-3) as well as the foam in the middle of the magazine (po-
sition 7) were subjected to a resazurin assay for 2.5 and 4 h, see figure 6.3. Obviously,
inhomogeneous cell cultivation inside the reactor module led to very different cell num-
bers on single foams, as metabolic reduction of resazurin fluctuated strongly between
62 6.1 Static inoculation
6
8
10
reduced resazurin [µg]
2.5h-assay 4h-assay
0
2
4
1 2 3 7 mean value
reduced resazurin [µg]
foam position in the magazin
Figure 6.3: Mass of reduced resazurin in a 2.5 and 4 h resazurin assay for static inoculation
with 1 ·106cells/foam in a revolver reactor after 31 days of perfusion cultivation.
foams.
Whereas all foams’ upsides and even the foam holding magazine’s upside were covered
with a thick biofilm (which turned out to contain dead cells in a mucous matrix), total
cell number seemed to be rather low - by equation 5.1 on page 46 a mean cell number
of 2.0·106cells/foam was calculated from overall resazurin reduction in the 2.5h assay.
Compared to ceramics statically inoculated with only 2.7·105cells/foam and cultivated
under static conditions, cell growth therefore seemed to be very slow (see figure 6.1).
Moreover, FDA/EB staining as described in chapter 3.6.1 revealed a dense layer of vital
cells on the foams’ upside, very few and less vital cells at the bottom, see figure 6.4, and
no cells inside the ceramic.
Nevertheless, to evaluate the influence of flow velocity on cellular growth, in a further ex-
periment 1·107cells/module were inoculated into five reactor modules carrying 7 foams
each and allowed to settle for 3hbefore perfusion took place. Perfusion was then carried
out at 0.14, 0.29, 0.33, 0.44 and 0.61 ml min−1cm−2, respectively, for 13 days. Foams
were analyzed in a vitality assay (chapter 3.6.1) and glucose consumption and lactate
formation were monitored (chapter 3.6.4). A flow velocity of 0.29ml min−1cm−2hereby
supported cellular growth best as there was the highest glucose comsumption and lactate
formation over a culture period of 13d(see figure 6.5). The same results were obtained
for 0.33mlmin−1cm−2(data not shown). At 0.44 ml min−1cm−2an equal consumption
was found for glucose and production of lactate as for 0.29 or 0.33ml min−1cm−2(data
not shown) and foams also yielded good growth, but FDA/EB staining clearly revealed
that cell vitality (i.e.percentage of vital cells over total cell number) did not approach
6. Influence of mode of inoculation on cellular growth and distribution 63
a) b)
c)
Figure 6.4: Optical analysis of ceramic foams statically inoculated into a revolver reactor
and followed by 31 d of perfusion cultivation. a) Biofilm on foam holding magazin. b)
Biofilm on the upside of all 7 single foams after disassembling ceramics from the magazine,
ceramics in the culture plate’s last row are a cell free control. c) FDA/EB staining for
cells on one representative foam’s upside (left) and on the bottom side of that foam (right).
Vital cells fluoresce green, dead cells red, white arrows point to single vital cells, red arrows
point to single dead cells.
64 6.1 Static inoculation
1.0
1.2
1.4
1.6
glucose [g/L]
0.14 ml/(min*cm²) 0.29 ml/(min*cm²) 0.61 ml/(min*cm²)
0.4
0.6
0.8
1.0
glucose [g/L]
0.4
0.6
0.8
1.0
1.2
lactate [g/L]
0.0
0.2
0.4
0 60 120 180 240 300
lactate [g/L]
cultivation time [h]
Figure 6.5: Glucose consumption and lactate formation for CHO-K1 statically inoculated
into reactor modules and cultivated with different perfusion velocities for 13 d (1 ·107cell-
s/reactor module, 7 foams/module). Dashed horizontal lines indicate medium replacement.
6. Influence of mode of inoculation on cellular growth and distribution 65
Figure 6.6: FDA/EB staining of CHO-K1 cells statically inoculated into reactor modules
and cultivated with different perfusion velocities for 13 days (1 ·107cells/module inoculum,
7 foams/module). Green fluorescence corresponds to vital cells, red fluorescence to dead
cells, bars represent 100 µm, arrows point to single cells. Each column represents foams
of a specific perfusion velocity during cultivation as indicated underneath. For each flow
velocity representative pictures are shown for the foam’s face (top row), a cross section in
about 4 mm depth (middle row) and the end area (bottom row).
that of the lower flow velocity. At a flow rate of 0.61 ml min−1cm−2cell metabolism was
slightly reduced compared to 0.29mlmin−1cm−2, and total cell density was lower. More-
over, dyed cross sections revealed a relation between flow velocity and cellular ingrowth
into the foam. For 0.14, 0.44 and 0.61ml min−1cm−2almost no vital cells were found
in 5mmdepth, whereas cell colonies were observed for 0.29 and 0.33ml min−1cm−2(see
figure 6.6).
6.1.4 Static inoculation in ceramics with flow channels
Obviously, the filter-like ceramic shape hinders cells to penetrate the foam’s inside dur-
ing static inoculation. Therefore, in another experiment, I introduced a ceramic with
blind holes from both face areas to provide a larger, accessible surface. As these experi-
ments were performed very early in the project’s progress, CHO-K1 adapted to growth
66 6.1 Static inoculation
in medium containing 1% FBS (see chapter 3.5 and Appendix) were used, which are very
sensitive to flow velocities higher than 0.12mlmin−1cm−2. Thus, this section should be
regarded as one individual unit and not be compared to experimental outcomes of all
other chapters.
Revolver modules with plain magazines were prepared with 7 foams per magazine each,
whereby ceramics were modified as follows: Prior to the last torching step in chapter
3.1, four blind holes with 7 mm depth and 0.2 mm diameter were drilled into the ceramic
cylinders from the upside of the foam and four blind holes were drilled from the bottom.
Blind holes were aligned in a square with distance from the foam’s middle 1.5mmfor
the upside and 3mmfrom the bottom. By that, holes did not touch each other and flow
was forced to pass the foams’ 3D structure. Module assembly otherwise was performed
as described in chapter 3.2.2.
For static inoculation, 1·106CHO-K1 cells adapted to growth in medium containing
1% FBS were inoculated per foam from above and cells were allowed to settle before
perfusion took place with flow velocities 0.03, 0.08 and 0.12ml min−1cm−2, respectively.
Modules were subjected to perfusion cultivation for 1 and 7 days, then, foams were sep-
arately subjected to resazurin assays in culture wells as in chapter 3.6.5 and medium
was analyzed for glucose consumption and lactate formation. Single foams were also
subjected to vitality assays (FDA/EB staining). As a control, 1·106cells/foam were
statically inoculated (chapter 3.3.1) on foams modified as above and statically cultivated
(chapter 3.4.2) for 1 and 7 days, respectively.
Regarding overall glucose consumption and lactate formation, there were only small dif-
ferences for the three flow velocities, see figure 6.7. However, for 0.08mlmin−1cm−2,
lactate formation occurred slightly faster than in the other modules. As for single foam
analysis for 3hfollowing cultivation, glucose consumption and lactate formation differed
strongly for foams cultivated for 1 or 7 days, see figure 6.8. However, there was almost
no difference between glucose consumption between the flow velocities, lactate forma-
tion was slightly higher for 0.03mlmin−1cm−2and 7 days of cultivation. Reduction of
resazurin in a 4hassay again revealed an increase in metabolic activity after 7 days of
cultivation, see figure 6.9. Here, 0.08mlmin−1cm−2showed the highest increase for the
three flow velocities. Remarkably, both medium and resazurin assays showed the highest
increase in metabolic activity for the control group.
However, vitality assays of cross sections revealed cellular ingrowth into the blind holes
both from top and bottom for all foams after 1 day of cultivation. Following 7 days of in-
oculation, growth was observed in pores next to the blind holes for 0.08ml min−1cm−2flow
6. Influence of mode of inoculation on cellular growth and distribution 67
150
200
250
300
cumulative glucose comsumption
[mg]
0.03ml/(min*cm²) 0.08ml/(min*cm²)
0
50
100
150
0 30 60 90 120 150 180
cumulative glucose comsumption
[mg]
cultivation time [h]
150
200
250
300
cumulative lactate formation [mg]
0.12ml/(min*cm²)
0
50
100
150
0 30 60 90 120 150 180
cumulative lactate formation [mg]
cultivation time [h]
Figure 6.7: Cumulative glucose consumption (left) and lactate formation (right) for CHO-
K1 adapted to growth in medium with reduced serum content (1% FBS) statically inoc-
ulated into revolver reactors (7 ·106cells/module, 7 foams/module) equipped with foams
with blind holes (n=1). Cultivation was performed for 7 days at three flow velocities as
indicated. Medium replacement was performed twice for 0.03 and 0.08 ml min−1cm−2on
day 3 and day 5 and three times for 0.12 ml min−1cm−2on day 2, 4 and 6.
0.6
0.8
1
glucose consumption [mg]
0.03ml/(min*cm²) 0.08ml/(min*cm²)
0
0.2
0.4
0.6
1 day after inoculation 7 days after inoculation
glucose consumption [mg]
0.6
0.8
1
lactate formation [mg]
0.12ml/(min*cm²) control w/o perfusion
0
0.2
0.4
0.6
1 day after inoculation 7 days after inoculation
lactate formation [mg]
Figure 6.8: Single foam analysis for ceramics with blind holes statically inoculated with
CHO-K1 adapted to growth in medium with reduced serum content 1% FBS (1 ·106cell-
s/foam) and followed by cultivation under perfusion conditions for 1 or 7 d (shown is data
from a 3 h static cultivation of ceramics in wells of culture plates with fresh medium). De-
picted are mean values for glucose consumption (left) and lactate formation (right) of all
seven foams of one reactor module (n=7) and of three foams of the control group (n=3)
which was cultivated under static conditions in culture plates.
68 6.1 Static inoculation
6
8
10
reduced resazurin [µg] in 4h
-assay
0.03ml/(min*cm²) 0.08ml/(min*cm²)
0.12ml/(min*cm²) control w/o perfusion
0
2
4
6
1 day after inoculation 7 days after inoculation
reduced resazurin [µg] in 4h
Figure 6.9: Reduction of resazurin (4 h assay) for single foams with blind holes after static
inoculation with CHO-K1 adapted to growth in medium with reduced serum content 1%
FBS (1 ·106cells/foam) and perfusion cultivation. Depicted are mean values of all seven
foams of one reactor module (n=7) and of three foams of the control group (n=3) which
was cultivated under static conditions in culture plates.
velocity, but not for any other foams, see figure 6.10. Moreover, cells were found in the
blind holes from top in all foams after 7d, but only for 0.08mlmin−1cm−2and the control
group, vital cells were found in bottom blind holes. Consistent with analyses depicted
in figures 6.7, 6.8 and 6.9, 0.08 ml min−1cm−2yielded best results regarding 1 week long
cultivation. Again, foams of the control group showed comparable good results for vi-
tality staining, as growth was observed inside bottom blind holes after 7 days of static
cultivation.
Still, it has to be mentioned that medium and resazurin assays as performed in sections
6.1.1 to 6.1.4 do not properly evaluate cells growing inside the porous structure. Here,
assays were performed statically, i. e. cells growing inside the matrix do not take part
in metabolic processes, or, if they do, medium is not analyzed as it clearly cannot
be washed out thoroughly. Therefore, for the experiments above, data obtained during
medium perfusion as in figure 6.7 should be regarded as more trustworthy than data from
static assays as in figure 6.8 and 6.9 if the total foam volume is considered. However, in
later experiments, I implemented the resazurin assay into the reactor module in order
to allow for cells growing inside the ceramic matrix.
6. Influence of mode of inoculation on cellular growth and distribution 69
Figure 6.10: Vitality assay for ceramics with blind holes statically inoculated with CHO-
K1 adapted to growth in medium with reduced serum content 1% FBS (1 ·106cells/foam)
and cultivated for 1 and 7 days at 0.08 ml min−1cm−2. White arrows point to single vital
cells, red arrows to dead cells, bars represent 500 µm.
70 6.1 Static inoculation
0 20 40 60 80 100 120 140 160 180 200
10
15
20
25
30
35
40
45
50
55
assay time [min]
resorufin [%]
simulation for 5.0E+07
cells in compartment B
samples RM8
samples RM11
Figure 6.11: Percentage of reduced resazurin in a dynamic resazurin assay for CHO-K1
statically inoculated into a revolver reactor (1 ·107cells/module, 6 foams/module) after 8
days of perfusion cultivation. Revolver module RM10 was eliminated from analysis due to
experimental obstacles during resazurin assay. Solid line presents data of a simulation as
indicated in the legend.
6.1.5 Reproducible cultivation following static inoculation into
revolver reactors
To prove repeatable operation procedures, another experiment was designed for static
inoculation. Three revolver reactors (RM8, RM10 and RM11) with conical magazines
(see figure 3.6b) were equipped with six ceramic foams and inoculated each with 1·107
CHO-K1 cells as in chapter 3.3.1. Cultivation was performed for 8 days under perfusion
conditions with 0.33mlmin−1cm−2as described in chapter 3.4.1. The reactor modules
then were subjected to a resazurin assay as in chapter 3.6.5. Data of dynamic resazurin
reduction was close to that of a simulation with 5.0·107cells in reactor modules (chapter
5.1.3), see figure 6.11. Therefore, I assume total cell number has increased about fivefold
in ceramics during cultivation of 8 days. Medium samples were further analyzed for
glucose consumption and lactate formation as in chapter 3.6.4. Figure 6.12 gives an
overview of metabolism, which was very homogeneous for all three reactor modules.
Again, microscopic evaluation of single foams by FDA/EB staining revealed a dense
layer of cells on the foam’s upsides with very few cells at the bottom and no cells inside
6. Influence of mode of inoculation on cellular growth and distribution 71
300
400
500
cumulative glucose consumption [mg]
module RM8 module RM10
0
100
200
0 50 100 150 200 250
cumulative glucose consumption [mg]
cultivation time [h]
300
400
500
cumulative lactate formation [mg]
module RM
11
0
100
200
0 50 100 150 200 250
cumulative lactate formation [mg]
cultivation time [h]
Figure 6.12: Cumulative glucose consumption (left) and lactate formation (right) for
CHO-K1 statically inoculated into cone-magazine reactor modules (1 ·107cells/module).
Perfusion was performed for 8 days at 0.33 ml min−1cm−2. Dashed horizontal lines indicate
medium replacement.
the ceramic. Moreover, by scanning electron microscopy as in chapter 3.6.3, I did not
find many cells inside the ceramic, see figure 6.13.
Static inoculation - synopsis
Static inoculation of CHO-K1 in all experiments described above led to a dense layer
of cells on the foams’ surface. Cells proliferate on the ceramic surface (figure 6.1 and
6.9), but growth to higher cell numbers is prevented, probably due to effects such as the
community effect [Li et al., 2001]. Cells were found on the foams top and bottom with a
mucous matrix covering foams and magazines (figure 6.4), but cellular ingrowth into the
three dimensional structure did not occur - which is in good agreement with previous
findings outlined in chapter 2.1.3. For static inoculation followed by perfusion cultivation
in tubular or revolver reactors, dense layers of cells might even prevent medium from
flowing through the scaffold. This is supported by the experiments described in chapter
6.1.2, where cells seemed to pass the inoculated top foam laterally and therefore were
able to populate top and bottoms of the following foams. Clearly, if medium does not
pass through the foam, cellular ingrowth is inhibited. As shown in chapter 6.1.3 and 6.1.4
there is an optimal flow velocity for the cultivation of CHO-K1 following static inocu-
lation, which not only determines cellular growth but also affects cellular ingrowth into
the matrix. Low flow rates correspond to poor nutrient and oxygen supply and therefore
promote cell death. High flow rates promote cellular ingrowth into the ceramic, but also
correspond to a decrease in total cell number inside the foam and decreased cell viabil-
72 6.2 Dynamic inoculation by agitation
Figure 6.13: Scanning electron microscopy of CHO-K1 on ceramic foams following static
inoculation. Images are acquired in 1 mm depth of the ceramic from above after 8 days of
perfusion cultivation. Bars represent 20 µm.
ity, probably due to higher shear stresses. For reactor modules with plain foam holding
magazines, I found flow velocities of 0.29 or 0.33 ml min−1cm−2to support growth best
following static inoculation and velocities not higher than 0.1ml min−1cm−2for static
inoculation of serum reduced CHO-K1 into ceramics with blind holes. Nevertheless, pro-
liferation in foams in reactor modules was not as high as in statically cultivated foams,
where 8 day cultivation led to an increase of cell number by factor 8 to 14 (chapter 6.1.1)
compared to only fivefold by perfusion cultivation (chapter 6.1.5).
Reproducible cultivation of statically inoculated cells in a perfusion reactor is possible,
but cell distribution is weak and thick layers of cells on the foam’s surface are made
up of cell aggregates which will eventually die due to bad nutrient supply as shown
by [Sutherland et al., 1986]. Moreover, perfusion cultivation does not enhance cellular
growth and therefore is unnecessary, as proven by static cultivation experiments (see
figure 6.8).
6.2 Dynamic inoculation by agitation
For agitation inoculation as in chapter 3.3.2 CHO-K1 with 5.5·105and 2.2·106cells
per foam, respectively, were inoculated in 15 ml medium and agitated for 72 h. Cells in
medium were collected and centrifuged to evaluate number of live and dead cells using
a hemocytometer (see Appendix). Figure 6.14 and table 6.1 summarize the results of
dynamic inoculation using agitation. Percentage of cells in medium after 72 h of agitation
6. Influence of mode of inoculation on cellular growth and distribution 73
80
100
cell count in % of inoculum
inoculation
live cells in medium after 72h agitation
dead cells in medium after 72 hour agitation
0
20
40
60
5.5E+5 cells/foam 2.2E+6 cells/foam
cell count in % of inoculum
cell count for inoculation
foam 1 foam 2 foam 1 foam 2
Figure 6.14: Cell absorption by agitation assisted dynamic inoculation of foams in a cell
solution. Shown is the cell count in medium solution following 72 h of agitation inoculation
for two foams for each inoculation density.
Table 6.1: Agitation inoculation of different cell numbers of CHO-K1 on ceramic foams.
Whereas seeding efficiency (calculated as in equation 2.2) strongly differs, optical evaluation
did not show remarkable differences in cell number and distribution on the ceramics, see
figure 6.15.
cell number for inocu-
lation
total dead cell count in
medium after 72hof agita-
tion in % of inoculum
calculated inoculation effi-
ciency
5.5·105(foam 1) 20 0.6
5.5·105(foam 2) 7 0.9
2.2·106(foam 1) 5 0.9
2.2·106(foam 2) 13 0.8
was much lower for inoculation with 2.2·106cells/foam than for 5.5·105cells/foam,
therefore, more cells are assumed to have adhered to the ceramic surface for the higher
inoculum. Nevertheless, foam dyeing as in chapter 3.6.2 did not reveal differences in
overall cell density on the foams. Here, cell distribution was inhomogeneous for the
foams’ surface area with no cells inside the ceramics, see figure 6.15.
74 6.2 Dynamic inoculation by agitation
Figure 6.15: Cell density and distribution of CHO-K1 following agitation seeding. Left:
Inoculated cell number 5.5 ·105, right: 2.2 ·106cells/foam inoculum. Upper pictures show
the top, lower pictures the bottom of the ceramic, foams were aligned for agitation inocu-
lation as in picture on top. Bars represent 500 µm, arrows point to single cells.
6. Influence of mode of inoculation on cellular growth and distribution 75
module alignment initial cell
count/-
foam
cycle
count
flow velocity
[ml min−1cm−2]
cell vitality / cell
distribution
module horizontically
aligned with rotation* 3.5 ·1063.5 0.33 - -/-
0.24 ·/·
module vertically aligned
with horizontal holding*
2·106
3.5 0.24 ++/+
2.5 ·106++/++
module vertically
aligned
3.5 ·1063.5
0.33 -/ ·
0.24 ++/+
0.19 ·/·
5·1067 0.24 ++/+
2.5 ·1064.5 0.24 ++/++
2.5 ·1063.5 0.33 ++/++
Table 6.2: Optical analysis of CHO-K1 on foams two days after dynamic perfusion inoc-
ulation in reactor modules. Cell vitality is expressed as ++ for more than 90% and + for
80-90% vital cells, as ·for 70-80%, - for 50-70% and - - for less than 50% vital cells. Cell
distribution is expressed as - for no cells in the foam’s interior, ·for much less cells inside
the foam than on the upper and lower face area, + for cells all over the foam’s volume with
colony forming, and ++ for a homogeneous cell distribution all over the foam’s volume (*:
see chapter 3.3.3 for further explanation).
6.3 Oscillatory perfusion inoculation
6.3.1 Influence of module orientation, flow velocity, initial cell
count on cell distribution
For perfusion assisted inoculation as in 3.3.3, experiments were performed whereby mod-
ule orientation, initial cell count, number of cycles and flow velocity were varied, see table
6.2. The upper glass olive volume was set to 10mlprior to inoculation. Data as evalu-
ated in table 6.2 is selected from a wider range of experiments with experiments omitted
when results did not lead to additional information. The results, see table 6.2, reveal a
strong influence of initial seeding density and initial flow velocity on cell vitality. Cell
vitality was very low for the 0.33mlmin−1cm−2/3.5·106cells/foam experiments, but
very high for the combination 0.33mlmin−1cm−2/2.5·106cells/foam when the module
is vertically aligned. A flow velocity of 0.24mlmin−1cm−2with 2.5·106cells/foam initial
seeding density also yielded good results for both vertical alignment and vertical align-
76 6.3 Oscillatory perfusion inoculation
ment with horizontal holding, whereas higher cell densities often were found to result in
low vitality.
Moreover, initial seeding density played an important role regarding even distribution
of cells. For the system at hand it should be around 2.5 ·106cells/foam to get cells
distributed all over the foam’s volume. Both cycle count and module orientation did not
influence cell vitality or distribution within the tested frame.
I found best results for a seeding density of about 2.5·106cells/foam and an inocu-
lation velocity of 0.24 or 0.33mlmin−1cm−2,with 3.5cycles and vertical alignment of
the module as most convenient parameters. Hereby cells were distributed uniformly all
over the scaffold’s volume, cell vitality was around 100% and cell contact was found to
be close without cell aggregates after two days of cultivation. For the higher flow of
0.33mlmin−1cm−2cultivation for another five days did not affect cellular distribution
but I observed enhanced cell death in the foams caverns, see figure 6.16 (second column),
presumably due to bad nutrient supply. Therefore, initial flow velocity was chosen to be
at least 0.33mlmin−1cm−2for the following experiments.
On both top and bottom, scanning electron microscopy as in chapter 3.6.3 revealed cel-
lular growth and additional fibrous structures, presumably matrix proteins, see figure
6.17.
6.3.2 Introduction of more porous ceramics and augmentation of
flow velocity during cultivation
It was assumed that poor nutrient supply in 7-day-experiments was attributed to small
pore sizes and a flow velocity too low to support the growing cells,therefore in another
experiment I introduced a more porous ceramic with mean pore radius approximately
50µmhigher. Ceramics hereby were fabricated as in chapter 3.1 except all substances
were milled for 45mininstead of 15minin the first step. Moreover, flow velocity was
augmented during bioreactor operation to ensure sufficient nutrient supply. Three ex-
periments were performed, one with small pore ceramics (standard foams) and constant
perfusion flow, one with standard foams and flow augmentation and another with larger
pores and also flow augmentation. Cells were inoculated as before in vertically aligned
modules with conical magazines with an upper glass olive volume of 10mland 2.5·106
cells/foam, 3.5 cycles and seeding flow velocity 0.33ml min−1cm−2followed by an adher-
ing time of 30minbefore perfusion started. Flow was augmented to 0.4 ml min−1cm−2on
day 5 and ceramics were analyzed following 2 and 7 days, respectively, by a resazurin
6. Influence of mode of inoculation on cellular growth and distribution 77
Figure 6.16: Fluorescent staining of CHO-K1 cells on ceramic foam slices subjected to
dynamic seeding as in chapter 3.3.3 with initial seeding density of 2.5 ·106cells/foam,
3.5 cycles and seeding flow velocity 0.33 ml min−1cm−2. Left: Ceramic foam two days after
dynamic inoculation. Middle: Ceramic foam seven days after dynamic inoculation. Right:
Ceramic foam with larger pores seven days after dynamic inoculation with augmentation
of flow velocity (see chapter 6.3.2).Green fluorescence corresponds to vital cells, red flu-
orescence to dead cells, bars represent 500 µm, arrows point to single cells (white arrow,
live cell; red arrow, dead cell); images in the middle row are aquired in approximately
5 mm depth.
78 6.3 Oscillatory perfusion inoculation
Figure 6.17: Scanning electron microscopy of CHO-K1 cells on ceramic foam subjected
to dynamic seeding as in chapter 3.3.3 with initial seeding density of 2.5 ·106cells/foam,
3.5 cycles and seeding flow velocity 0.33 ml min−1cm−2. Bars represent 40 µm, black arrows
point to single cells , white arrows to fibrous structures observed all over the scaffold. Left:
View on top of ceramic foam seven days after dynamic inoculation. Right: View on bottom
of the same foam.
6. Influence of mode of inoculation on cellular growth and distribution 79
0 20 40 60 80 100 120 140 160 180
0
10
20
30
40
50
60
assay time [min]
resorufin [%]
simulation for 1.0E+08 cells
in compartment B
small pores, 2 day cultivation
small pores, 7 day cultivation
small pores, 2 day cultivation*
small pores, 7 day cultivation*
larger pores, 2 day cultivation*
larger pores, 7 day cultivation*
Figure 6.18: Comparison of reduction of resazurin by CHO-K1 on ceramics with dif-
ferent pore sizes dynamically inoculated and cultivated in reactor modules with augmen-
tation of flow velocity. * labels experiments for which flow was augmented from 0.33 to
0.4 ml min−1cm−2during cultivation on day 5. Solid line presents data of a simulation as
indicated in the legend.
assay performed inside the reactor module as in chapter 3.6.5 and vitality dyeing as
in chapter 3.6.1. I thereby found 100% vital cells following seven days of cultivation
only in the foams caverns with higher mean pore radius (figure 6.16, last column), and
an increase in metabolic activity compared to constant flow velocities and small pores
as evaluated by resazurin assay, see figure 6.18, but not for glucose consumption and
lactate formation (chapter 3.6.4), see figure 6.19. From a simulation of resazurin reduc-
tion as in chapter 5.1.3, cell number was estimated to be around 1.0·108(corresponding
to about 2·107cells per mlfoam volume), thereby having increased by factor ≈7for
ceramics with larger pores. Main proliferation hereby occurs during the first two days,
as resazurin reduction is not much higher 5 days later, see figure 6.18. For small pores,
reduction following 7 days of cultivation is even lower than for 2 day cultivations, which
is in good agreement with cell depletion observed in vitality assay, see also first two
columns in figure 6.16.
80 6.3 Oscillatory perfusion inoculation
0.4
0.5
250
300
350
400
flow [ml/(min*cm²)]
cumulative glucose consumption [mg]
wider pores
small pores
flow velocity
0.2
0.3
0
50
100
150
200
250
flow [ml/(min*cm²)]
cumulative glucose consumption [mg]
0.4
0.5
200
250
300
350
400
flow [ml/(min*cm²)]
cumulative lactate formation [mg]
0.2
0.3
0
50
100
150
200
0 20 40 60 80 100 120 140 160
flow [ml/(min*cm²)]
cumulative lactate formation [mg]
cultivation time [h]
Figure 6.19: Cumulative glucose consumption and lactate formation for CHO-K1 dynam-
ically inoculated in reactor modules equipped with ceramics with small and larger pores
(2.5 ·106cells/foam, 6 foams/module) and cultivated with augmentation of flow velocity
during cultivation. Dashed horizontal lines indicate medium replacement.
6. Influence of mode of inoculation on cellular growth and distribution 81
6.3.3 Reduction of foam volume by reducing cylinder height
In all perfusion inoculation experiments above, cellular ingrowth in depths of about
2mmwas observed to be very homogeneous from both sides of the foam. Depending
on operation strategies and foam geometry, ingrowth could be extended to 5 mm from
both sides but cell density was found to decrease with depth. Therefore, I introduced
a ceramic, which is only 5 mm in height, hoping for homogeneous growth all over the
scaffold.
6 ceramics with larger pores as described in chapter 6.3.2 were assembled into a cone
magazine bioreactor and prepared for cultivation. Inoculation was performed with three
different cell numbers and three starting volumes in the upper glass olive as in table 6.3 at
0.33mlmin−1cm−2and 3.5 cycles. Following 30 min of adhering time, perfusion started
from above with 0.33mlmin−1cm−2and was performed for two days. Then, foams were
subjected to resazurin assays inside the reactor modules and further analyzed in a vitality
assay or prepared for scanning electron microscopy as in chapter 3.6.3.
Regarding metabolism, all modules showed a similar behavior except RM8 which was
inoculated with the smallest number of cells in the smallest volume, see figure 6.20.
For 2.28·107cells/module (RM9 and RM10), there was no difference in metabolism for
different inoculation volumes. Vitality assay moreover revealed the most homogeneous
cellular distribution for the combination 0.76·107cells/module and 20mlinitial glass
olive volume (RM12), see table 6.3 and figure 6.21. Simulation of the resazurin reduction
model in chapter 5.1.3 revealed a number of 4.0·107cells per module at maximum (see
data for RM12 in figure 6.20), which means an increase in cell number by factor ≈5
for just two days of cultivation for RM12 corresponding to exponential growth with a
doubling time of 20h. All in all, flat foams with larger pores yielded very good results
regarding cellular distribution and metabolic activity. Here, initial seeding density plays
an important role and has to be adjusted carefully.
Oscillatory perfusion inoculation - synopsis
Clearly, regarding cell distribution, dynamic inoculation yields best results. Whereas
for inoculation procedure module alignment and cycle count do not influence cell dis-
tribution, there is a strong influence for initial seeding density and initial flow velocity
(see table 6.2). Remarkably, initial cell distribution following dynamic inoculation is
not compellingly maintained, but fluctuates with cultivation time. Whereas cell vitality
inside the ceramic can be improved by augmenting flow velocities, a much better exper-
82 6.3 Oscillatory perfusion inoculation
0 50 100 150 200 250 300 350
5
10
15
20
25
30
35
40
45
50
55
assay time [min]
resorufin [%]
simulation for 4.0E+07 cells
in compartment B
RM8, few cells, few volume
RM12, few cells, much volume
RM9, many cells, few volume
RM10, many cells, much volume
RM11, middle cells, middle volume
Figure 6.20: Reduction of resazurin for ceramics reduced in height and with larger pores.
CHO-K1 were dynamically inoculated into reactor modules as in table 6.3 and cultivated
for two days. Resazurin assays were performed inside the reactor devices. Solid line presents
data of a simulation as indicated in the legend.
denomination initial cell
count/foam
initial volume
in upper glass
olive [ml]
top cross sec-
tion
bottom
RM8 1.3 ·10610 ++ o o
RM9 3.8 ·10610 + + o
RM11 2.5 ·10615 o o o
RM12 1.3 ·10620 + + +
RM10 3.8 ·10620 ++ + +
Table 6.3: Optical analysis of ceramic foams with reduced height two days after dynamic
perfusion inoculation. Cell vitality in all experiments was found to be 100%. Cell distri-
bution is expressed as o for many inhomogeneously distributed cells, + for a homogeneous
cell distribution, and ++ for a dense layer of cells.
6. Influence of mode of inoculation on cellular growth and distribution 83
Figure 6.21: Vitality assay for CHO-K1 on ceramic foams reduced in height and with
larger pores two days after dynamic inoculation. Shown are representative pictures of
modules inoculated with 3.8 ·106cells per foam (RM 10) and 1.3 ·106cells per foam (RM
12) as indicated in table 6.3. Images in the middle row are aquired from cross sections in
2.5 mm depth. Bars represent 500 µm, arrows point to single cells.
84 6.3 Oscillatory perfusion inoculation
imental outcome is achieved when ceramics with larger pores are used (chapter 6.3.2).
In summary, operation procedures for standard foams should be inoculation performed
with 1.5·107cells/module holding a conical magazine with 6 foams, upper glass olive
volume should be 10ml, flow velocity 0.33 ml min−1cm−2with 3.5cycles at vertical align-
ment. For ceramics with larger pores and reduced in height, inoculation should be
performed with 7.6·106cells/module (also holding a conical magazine with 6 foams) in
20mlupper glass olive volume, 3.5 cycles and 0.33 ml min−1cm−2flow velocity at vertical
alignment. During further cultivation, flow velocity should be augmented approximately
every 5 days.
7. Reproducibility of the chosen operation methods 85
7 Reproducibility of the chosen
operation methods
As discussed on page 84, a set of operation procedures was found that yielded good
cellular distribution all over the foam and maintained cells in a vital status. Until
then, to minimize costs, experiments were performed with only one to two modules
per variation, therefore reproducibility of the established system still had to be proven.
Ceramics with larger pores showed better cellular growth and distribution, but due to
the not yet completely elaborated fabrication process, there are still issues of material
decomposition for this foam geometry. Therefore, reproducibility tests were performed
for standard foams as described in chapter 3.1 as well and for ceramics with larger pores.
7.1 Standard foams
Vertically aligned reactor modules with conical magazines and equipped with glass olives
were inoculated as described in chapter 3.3.3 with an upper glass olive volume of 10ml,
1.5·107cells per module, 3.5 cycles and initial flow velocity 0.33 ml min−1cm−2. Fol-
lowing 30minadhering time, perfusion started from above at 0.33 ml min−1cm−2and
cultivation was performed for 14 days. All in all, 10 modules were started at 5 dif-
ferent days. Flow was augmented to 0.38 ml min−1cm−2on day five and further to
0.42mlmin−1cm−2on day ten. On day 14 a resazurin assay was performed inside the
reactor modules and single foams were analyzed for cell vitality (FDA/EB staining).
Moreover, 2 foams per module were analyzed in CHGE for carbon content (see chapter
3.6.6).
Figure 7.1 gives an overview for glucose consumption and lactate formation during the
course of cultivation. Metabolic status as examined by resazurin assay is delineated in
figure 7.2. Disassembling the modules on day 14, I found a thick mucous layer on top of
the magazines and all over the inner reactor material. Vitality staining of that matrix
was found to contain many cells, not attached to each other but arranged inside the
86 7.1 Standard foams
1000
1200
1400
cumulative glucose consumption [mg]
RM9-2 RM10-1 RM11-1 RM12-2 RM15-3
0
200
400
600
800
cumulative glucose consumption [mg]
800
1000
1200
1400
cumulative lactate formation [mg]
0
200
400
600
0 100 200 300 400
cumulative lactate formation [mg]
cultivation time [h]
Figure 7.1: Cumulative glucose consumption and lactate formation for CHO-K1 dynam-
ically inoculated on standard foams with 1.5 ·107cells per module and cultivated for 14
days with augmentation of flow velocity (see main text, five runs were selected for analysis).
Medium replacement was performed every second day.
7. Reproducibility of the chosen operation methods 87
0 50 100 150
0
10
20
30
40
50
60
70
assay time [min]
resorufin [%]
simulation for 1.6E+08 cells
in compartment B
RM10−1
RM12−2
RM15−3
RM6−4
RM16−4
RM14−5
Figure 7.2: Reduction of resazurin for CHO-K1 dynamically inoculated on standard foams
with 1.5 ·107cells per module and cultivated for 14 days with augmentation of flow velocity
(see main text). Shown is data from six modules out of five different runs as indicated in
the legend above. Resazurin assays were performed inside the reactor devices. Solid line
presents data of a simulation as indicated in the legend.
88 7.2 Foams with larger pores
Table 7.1: Reproducibility of cell cultivation in reactor modules for CHO-K1 as exam-
ined by dynamic resazurin assay and CHGE. (*) Data from resazurin assay is a rough
estimation from simulations with different cell numbers as in chapter 5.1.3 and compared
to measurement data in figure 7.2 and 7.5.
experiment cell number per module
/ SD from resazurin as-
say*
cell number per module
/ SD from CHGE
small pores, cylinders 10 mm height,
inoculum 1.5 ·107cells per module,
14 d cultivation (n=6)
1.4 ·108/ 4 ·1072.5 ·108/ 5.1 ·107
wider pores, cylinders 5 mm height,
inoculum 7.6 ·106cells per module,
4 d cultivation (n=6)
2.0 ·108/ 3 ·1079·107/ 3.0 ·107
structure. However, cell vitality inside this matrix was lower than 50%, see figure 7.3.
Vitality assay of cells on ceramics revealed high vitality all over the foam’s volume and
good to few cell numbers inside the ceramic scaffolds, see figure 7.3. Cell number as
evaluated by equation 5.17 following CHGE was found to be around 4.1 ·107cells/foam,
resulting in 2.5·108cells per module, see table 7.1. As foams were covered with a mu-
cous layer as described above, cell number as evaluated from carbon content analysis
clearly is estimated too high, explaining the divergence to cell number as estimated from
resazurin assays, see table 7.1.
Whereas overall glucose and lactate metabolism gives a very homogeneous picture for
all experiments, resazurin assays showed remarkable divergence, see figure 7.2. How-
ever, simulation of the resazurin reduction model in chapter 5.1.3 with 1.6 ·108cells per
module was very close to data of some modules (all modules except RM6-4 and RM
16-4 in figure 7.2). Therefore, an increase in cell number by factor of about 11 for 14
day cultivation can be achieved, compared with 7fold increase following two days of
cultivation in figure 6.18.
Nevertheless, out of ten started modules just six could be analyzed properly. Other
modules leaked or were contaminated after a medium replacement.
7.2 Foams with larger pores
Vertically aligned reactor modules with conical magazines and equipped with glass olives
were inoculated as described in chapter 3.3.3 with an upper glass olive volume of 20ml,
7.6·106cells per module, 3.5 cycles and initial flow velocity 0.33mlmin−1cm−2(conditions
7. Reproducibility of the chosen operation methods 89
Figure 7.3: Cell distribution for CHO-K1 dynamically inoculated on standard foams with
2.5 ·106cells/foam and cultivated for 14 days with augmentation of flow velocity. Left: Cell
containing mucous matrix found on top of the foams and inside the reactor modules. Right:
Representative images of cell distribution in and on foams after two weeks of perfusion
cultivation. The cross section image is aquired in 5 mm depth. Bars represent 500 µm,
white arrows point to live cells, red arrows point to dead cells.
90 7.2 Foams with larger pores
200
250
300
cumulative glucose consumption [mg]
RM12-1 RM15-1 RM16-1
RM12-2 RM15-2 RM16-2
0
50
100
150
cumulative glucose consumption [mg]
150
200
250
300
cumulative lactate formation [mg]
0
50
100
150
0 20 40 60 80 100
cumulative lactate formation [mg]
cultivation time [h]
Figure 7.4: Cumulative glucose consumption and lactate formation for CHO-K1 dynam-
ically inoculated on ceramics with reduced height and larger pores with 7.6 ·106cells per
module. Perfusion cultivation was performed for four days.
as for RM12 in table 6.3). Following 30minadhering time, perfusion started from above
at 0.33mlmin−1cm−2and cultivation was performed for 4 days. All in all, 6 modules
were started at 2 different days with 6 ceramics with larger pores and 5mmheight each.
On day 4 a resazurin assay was performed inside the reactor modules and single foams
were analyzed for cell vitality (FDA/EB staining). Moreover, foams were prepared for
scanning electron microscopy as in chapter 3.6.3 and two foams per module were ana-
lyzed for carbon content in CHGE (chapter 3.6.6).
Figure 7.4 gives an overview for glucose consumption and lactate formation during the
course of cultivation. Metabolic status as examined by resazurin assay is delineated
in figure 7.5. Vitality staining revealed high vitality all over the foams’ volumes and
good to few cell numbers inside the ceramic scaffolds, see figure 7.6. Cell growth was
7. Reproducibility of the chosen operation methods 91
0 10 20 30 40 50 60 70 80 90 100
10
15
20
25
30
35
40
45
50
55
60
assay time [min]
resorufin [%]
simulation for 2.2E+08 cells
in compartment B
RM12−1
RM15−1
RM16−1
RM12−2
RM15−2
RM16−2
Figure 7.5: Reduction of resazurin for CHO-K1 dynamically inoculated on ceramics with
reduced height and larger pores with 7.6 ·106cells per module. Perfusion cultivation was
performed for four days. Two runs were performed with three reactor modules each as
indicated in the legend above. Resazurin assays were performed inside the reactor devices.
Solid line presents data of a simulation as indicated in the legend.
92 7.2 Foams with larger pores
dense on the foams’s tops and very homogeneous on the foams’ bottoms and inside the
foams. Cell density was equal for both bottoms and cross sections but lower than on top
of the ceramic. Nevertheless, cells were found homogeneously distributed all over the
cross sections with a sufficient cell density for all foams of each reactor module likewise.
Scanning electron microscopy revealed dense layers of cells on the foam’s surface and
fibrous structures in which more cells were observed, see figure 7.7. Cell number as
evaluated by equation 5.17 following CHGE was found to be around 1.5 ·107cells/foam,
resulting in 9·107cells per module, see table 7.1.
For cells inoculated on foams of 5mm height and with larger pores, cell distribution and
metabolic activity as evaluated by resazurin assays was rather homogeneous for all six
experiments. However, metabolic data from medium over four days of cultivation showed
a broad distribution (figure 7.4). Compared to cells on standard foams as in chapter 7.1,
glucose consumption was around the same depending on the considered reactor module
although inoculum was only one half of that for standard foams. Furthermore, reduc-
tion of resazurin was even higher for foams with larger pores even though cultivation
was performed only for four days compared to two weeks for standard foams and twice
as much cells for inoculation. Hereby, cell number was found to have increased by a
factor of 29 as evaluated by a simulation as in chapter 5.1.3, see figure 7.5. However, cell
number as evaluated by CHGE was found to be much less with a cell incrementation
factor of around 12, see table 7.1. Clearly, metabolic assays will take into account cells
growing all over the reactor’s periphery, whereas CHGE only determines cell number on
and inside ceramics. Still, proliferation in ceramics with larger pores is higher, as the
incrementation factor for small pores is around 17 (as calculated from CHGE analysis,
see table 7.1), but cultivation was performed for 14 days compared to only 4 days for
larger pores.
Cell distribution all over the scaffold was more homogeneous for foams with wider pores,
therefore local cell density was lower, which might have led to an increase in metabolism
and proliferation compared to higher foams with smaller pores.
Regarding operation techniques, opposed to modules in chapter 7.1 all six modules were
operated sterilely for four days - most likely due to the shorter cultivation time and no
needs for medium replacements.
Reproducibility - synopsis
Cells can be grown more or less reproducible on standard ceramics with operation tech-
niques as found in chapter 6.3.1. Whereas overall metabolism regarding glucose con-
7. Reproducibility of the chosen operation methods 93
Figure 7.6: Cell distribution for CHO-K1 dynamically inoculated on ceramics with re-
duced height and larger pores with 7.6 ·106cells per module. Perfusion cultivation was
performed for four days. Representative images of cell distribution in and on foams culti-
vated for four days after dynamic inoculation with 7.6 ·106cells/module are shown. The
cross section image is aquired in 2.5 mm depth. Bars represent 200 µm, white arrows point
to live cells, blue arrow points to a pore’s rim totally overgrown by cells.
94 7.2 Foams with larger pores
Figure 7.7: Scanning electron microscopy for CHO-K1 dynamically inoculated on ce-
ramics with reduced height and larger pores with 7.6 ·106cells per module. Perfusion
cultivation was performed for four days. Cells grow on the foam’s surface and on fibrous
structures, presumably secreted by themselves. Images are taken from the top of the foam
and in 2.5 mm depth for image on bottom right. Bars represent 50 µm, black arrows point
to single cells, white arrows to fibrous structures.
7. Reproducibility of the chosen operation methods 95
sumption and lactate formation does not vary so much (figure 7.1), cell distribution
over the scaffold volume varies strongly between foams of one reactor module, i. e. with
a factor 2 for single foam analysis in CHGE following 14 days of cultivation. More-
over, cellular ingrowth into ceramics is limited to 400µmat maximum and thick mucous
layers containing dead cells were found on top of the magazines, see figure 7.3. A bet-
ter experimental outcome is achieved when cells are inoculated into foams with only
5mmheight and larger pores as in chapter 6.3.3. Here, cell distribution all over the
scaffold is very homogeneous and metabolic rates as found in a resazurin assay following
4 days of cultivation do not differ much (figures 7.6 and 7.5). Although total inocu-
lum is smaller, metabolism of cells on foams with larger pores is higher and scanning
electron microscopy revealed protein resembling structures with strong cellular ingrowth
on the pores’ surface (see figure 7.7). From that, as a cell culture scaffold foams with
5mmheight and a coarser pore structure should be favored for cell cultivation in the
presented devices. Yet, metabolic data derived from glucose analyzation showed a dis-
persion of factor 2.5 (see figure 7.4) and single foam analysis in CHGE one of 2, therefore,
experimental performance was not quite satisfactory all in all.
8. Long-term cultivation 97
8 Long-term cultivation
For many (especially tissue engineering) demands, it is absolutely necessary to main-
tain cells consistently over a longer cultivation period of a few weeks. Therefore, with
operation procedures found in chapter 6.3 four to seven week long experiments were
performed to demonstrate the qualification of the established bioreactor system. Cul-
tivations were performed for standard foams as described in chapter 3.1 and for foams
with larger pores, respectively.
Two vertically aligned reactor modules with conical magazines and equipped with glass
olives were inoculated as described in 3.3.3 with 1.5·107cells and an upper glass olive
volume of 10mlfor RM8 and 7.6·106cells and an upper glass olive volume of 20mlfor
RM14, 3.5 cycles and initial flow velocity 0.33ml min−1cm−2. Thereby, reactor module
RM8 consisted of 6 ceramix foams of 10mmheight and with standard pore size, the
magazine for RM14 contained 6 foams of 5mmheight and with larger pores. Following
30minadhering time, perfusion started from above at 0.33 ml min−1cm−2and cultiva-
tion was performed for 4 or 7 weeks. Flow was augmented to 0.38 ml min−1cm−2on day
four and further to 0.42mlmin−1cm−2on day nine (thirteen for RM14). On day 49 a
resazurin assay was performed inside the reactor module for RM 8, and single foams
were analyzed for cell vitality (FDA/EB staining) on day 28 for RM14 and on day 49
for RM8, respectively. Moreover, two foams each were prepared for scanning electron
microscopy as in chapter 3.6.3 and two foams were analyzed for carbon content in CHGE
as in 3.6.6 for RM14.
Following 2 weeks of operation, cellular growth was observed in the tubings of the reac-
tors’ peripheries with cell clumps of approximately 1-2mm diameter and cells grew all
their way into the medium container during the following cultivation time. Following 4
weeks of cultivation, reactor module RM14 had to be dismantled, as cell growth inside
the tubings led to blockage and inhibition of medium perfusion. Therefore, the resazurin
assay for RM14 was omitted.
Figure 8.1 gives an overview for glucose consumption and lactate formation during the
course of cultivation. Metabolic status for reactor module RM8 as examined by resazurin
98
3000
4000
5000
cumulative glucose consumption [mg]
small pores, 10mm cylinders wider pores, 5mm cylinders
0
1000
2000
3000
cumulative glucose consumption [mg]
3000
4000
5000
cumulative lactate formation [mg]
0
1000
2000
0 200 400 600 800 1000 1200
cumulative lactate formation [mg]
cultivation time [h]
Figure 8.1: Cumulative glucose consumption and lactate formation for CHO-K1 dy-
namically inoculated on standard foams with 1.5 ·107cells per module (circles) and on
5 mm foams with larger pores with 7.6 ·106cells per module (squares). Cultivation was
performed for 49 and 28 days, respectively. Medium replacement is indicated by filled
symbols.
8. Long-term cultivation 99
0 10 20 30 40 50 60 70
15
20
25
30
35
40
45
50
55
60
65
assay time [min]
resorufin [%]
simulation for 3.6E+08 cells in compartment B
small pores, 10mm cylinders, 49 day cultivation
Figure 8.2: Metabolic reduction of resazurin by cells cultivated on standard foams for 7
weeks under medium perfusion (1.5 ·107cells/module inoculum). Solid line presents data
of a simulation as indicated in the legend.
assay after 7 weeks of cultivation revealed an increase in cell number by a factor of about
24 regarding inoculation density, see figure 8.2. Therefore, overall cell proliferation in
ceramics can be assumed to have increased as summarized in figure 8.3.
Vitality staining revealed high vitality all over the foams’ volume and good to few cell
numbers inside the ceramic scaffolds for RM8. For RM14, cell growth was dense on top
of the foams, but no cells were found inside the foams and cell vitality on bottom was
only around 90%, see figure 8.4, right. As cell growth was very good for cells cultivated
for 4 days (figure 7.6, page 93), the bad experimental outcome here is associated with a
poor nutrient supply due to blockage of tubings.
Disassembling reactor module RM8 on day 49 revealed a thick mucous layer on top of the
magazine and all over the inner reactor material. Again, this matrix was found to contain
many cells with about 50% of them dead as in figure 7.3. Cell growth, however, was
very good on both top and bottom of the ceramics’ with few cells inside the scaffold, see
figure 8.4, left. Scanning electron microscopy moreover disclosed abundance of protein
like structures all over the foams’ volume. For standard foams, cells seemed to grow
through that protein matrix, see figure 8.5, left. For RM14, cells were found to grow
densely on top of ceramics, see figure 8.5, right, but very sparsely inside the ceramics.
100
3.E+08
4.E+08
cell number/reactor module
inoculum, evaluated with hemocytometer
comparison of data from resazurin assay with model simulation
see figure 8.2
1.E+07
1.E+08
2.E+08
0 10 20 30 40 50 60
cell number/reactor module
cultivation time [d]
see figure 6.18;
small pores, 2 day cultivation
see figure 7.2; RM10-1
Figure 8.3: Cell proliferation for CHO-K1 as examined by dynamic resazurin assay for
various experiments on standard foams.
Figure 8.4: Vitality staining for CHO-K1 cultivated for several weeks in standard foams
and foams with larger pores. Left: Analysis of cells inoculated with 1.5 ·107cells per
module on standard foams and cultivated for 7 weeks. Right: Analysis of cells inoculated
with 7.6 ·106cells per module on foams of 5 mm height and with larger pores, cultivation
for 4 weeks. Bars represent 500 µm, arrows point to single cells.
8. Long-term cultivation 101
As above, due to tubing blocking, nutrient supply is assumed to have been too low
to support cellular maintenance inside ceramics for RM14. However, cell number as
calculated from equation 5.17 following CHGE was around 1.2·108cells per module,
i.e.increased by factor 16 during 4 weeks of cultivation. Obviously, these numbers have
to be interpreted very critically, as CHGE determines carbon content from all biological
material, which would also be highly abundant mucous matrix. Therefore, total cell
number in all ceramics will be substantially lower than estimated by equation 5.17.
Sterile operation of reactor module RM8 with medium replacements twice a week could
be performed over seven weeks.
102
Figure 8.5: Scanning electron microscopy for CHO-K1 cultivated for several weeks in
standard foams and foams with larger pores. Perfusion cultivation was performed for 49
days and 28 days, respectively. Cells grow densely on the foam’s surface and on fibrous
structures, presumably secreted by themselves. Images are taken from the top of the foam
and in 2.5 mm depth as indicated. Bars represent 50 µm unless otherwise stated, white solid
arrows point to cells, white dashed arrow to fibrous structures, circle surrounds an area of
carpet-like cell growth on the ceramic’s surface.
9. Applicability of the reactor system for other cell types 103
9 Applicability of the reactor
system for other cell types
So far, all experiments were performed with CHO-K1. Chapter 7 and 8 finally verified
reproducible cultivation of CHO-K1 with good cellular distribution all over the ceramic
scaffold and the bioreactor’s qualification for long-term cultivation, respectively. Nev-
ertheless, the system was thought to be more or less universal from the beginning.
Therefore, three more cell types described in chapter 3.5 were tested for their ability to
grow inside the ceramic scaffold in the perfused bioreactor system.
All experiments were performed in reactors with conical magazines as in figure 3.6b
with ceramic foam cylinders of 5mmheight and larger pores as in chapter 6.3.3. To find
best operational strategies, tests with variations in inoculated cell number, cycle count,
initial inoculation volume in glass olives or perfusion velocity were performed.
9.1 Human lung carcinoma cells A549
Cells were used in their passage 35 for inoculation. Following injection of cells, the
volume of the lower glass olive was pumped through the module into the upper volume
at first, then perfusion inoculation continued for another 3.5 cycles. Cells were cultivated
with perfusion flow for 3 days before vitality assays (chapter 3.6.1) of populated foams
were done. First test runs revealed an influence of initial cell count and flow velocity on
cells’ distribution and vitality, see table 9.1. Cell growth for the last experiment (IV)
was very dense on the foam’s top and bottom and less denser for the cross section with
still a high number of cells.
For another experiment, cells were cultivated for 14 days. Here, the experimental setup
IV from table 9.1 was chosen, as it yielded the best results regarding cell vitality and
distribution within the tested range. Cells were optically analyzed by vitality assay
and also scanning electron microscopy (chapter 3.6.3) and revealed very good results
regarding cell distribution, see figure 9.1. Cell growth on both top and buttom was
104 9.1 Human lung carcinoma cells A549
Table 9.1: Optical analysis of A549 on ceramic foams with reduced height and larger
pores three days after dynamic perfusion inoculation. * stands for the optical analysis of
the FDA/EB staining following three days of cultivation for the top and bottom side of the
foams and a cross section, respectively. Indicated are cell vitality in % and cell distribution
compared to experiment IV, which was set to excellent (exc). o cell distribution is equal
to that in experiment IV, - cell distribution is poor compared to experiment IV, and + cell
distribution is better.
label initial cell
count/
foam
flow for
inoculation
[ml/min ·cm2]
flow for
cultivation
[ml/min ·cm2]
view
from
top*
view on
cross
section*
view
from
bottom*
I 1.3 ·1060.33 0.33 100%/o 100%/- 90%/-
II 2.5 ·1060.33 0.33 100%/o 100%/o 100%/-
III 2.5 ·1060.66 0.33 100%/+ 100%/o 100%/-
IV 2.5 ·1060.66 0.66 100%/exc 100%/exc 100%/exc
Figure 9.1: Vitality staining of A549 following 2 weeks of cultivation with setup IV from
table 9.1. Top left shows a picture of A549 grown in a culture flask. The image from the
middle of the foam is taken in approximately 3 mm depth. Bars represent 500 µm, arrows
point to single cells.
9. Applicability of the reactor system for other cell types 105
Figure 9.2: Scanning electron microscopy of A549 following 2 weeks of cultivation, images
were taken from the top of the ceramic. Bar represents 10 µm, arrows point to single cells.
Table 9.2: Optical analysis of primary fibroblasts on ceramic foams with reduced height
and larger pores 7 days after dynamic perfusion inoculation. * stands for the optical analysis
of the FDA/EB staining following cultivation for the top and bottom side of the foams and
a cross section, respectively. Indicated are cell vitality in % and cell distribution, which is
+ for good distribution with dense growth, o for acceptable distribution with single cells
and cell clumps occuring, and - for inhomogeneous cell distribution.
label initial cell
count/foam
cultivation
duration [d]
view from
top*
view on
cross
section*
view from
bottom*
I 3.3 ·1057 100%/o 100%/- 100%/o
II 1.3 ·1067 100%/+ 100%/o 100%/o
dense, but single cells still to be recognized. For the cross section, cell distribution was
excellent and cell density was good as well. From scanning electron microscopy, single
cells were found on the ceramic surface, see figure 9.2.
9.2 Human primary fibroblasts
Cells were used in their passage 10 for inoculation. Following injection of cells, the vol-
ume of the lower glass olive was pumped through the module into the upper volume at
first, then perfusion inoculation continued for another 3.5 cycles at 0.33mlmin−1cm−2.
Cells were cultivated at 0.33ml min−1cm−2for 7 days before vitality assay (chapter 3.6.1)
of populated foams. As cultivation of cells before inoculation takes very long (see Ap-
pendix), only two runs were performed, see table 9.2. For cultivation II of table 9.2,
cell growth on both top and buttom was very homogeneous and well distributed and
106 9.3 Madin-Darby canine kidney cells (MDCK )
Figure 9.3: Vitality staining of primary fibroblasts following 7 days of cultivation with
setup II from table 9.2. Top left shows a picture of primary fibroblasts grown in a culture
flask. The image from the middle of the foam is taken in approximately 3 mm depth. Bars
represent 500 µm, arrows point to single cells.
cellular ingrowth was observed for the total foam volume, see figure 9.3. Scanning elec-
tron microscopy (chapter 3.6.3) moreover revealed a variety of different structures on
the ceramic’s surface and between pore connections, which probably are matrix proteins
secreted by the fibroblasts, see figure 9.4.
9.3 Madin-Darby canine kidney cells (MDCK )
Cells were used in their passage 5 after thawing (see appendix) for inoculation. Following
injection of cells, the volume of the lower glass olive was pumped through the module into
the upper volume at first, then perfusion inoculation continued for a cycle count as indi-
cated in table 9.3 at 0.33mlmin−1cm−2. Cells were cultivated at 0.33ml min−1cm−2for
3 days before populated foams were dyed as in chapter 3.6.2 and analyzed for cell distri-
bution. The experimental outcome is depicted in table 9.3. For another experiment, cells
9. Applicability of the reactor system for other cell types 107
Figure 9.4: Scanning electron microscopy of primary fibroblasts following 7 days of culti-
vation with setup II from table 9.2, images were taken from the top of the ceramic. Bars
represent 20 µm, black arrows point to single cells, white arrows to fibrous structures.
Table 9.3: Optical analysis of MDCK on foams with reduced height and larger pores
three days after dynamic perfusion inoculation. * stands for the optical analysis of hema-
toxylin/eosin stained ceramics following three days of cultivation for the top and bottom
side of the foams and a cross section, respectively. Indicated are cell vitality in % (as
evaluated from FDA/EB staining whenever performed) and cell distribution, which is +
for good distribution with dense growth, o for acceptable distribution with single cells and
cell clumps occuring and - for inhomogeneous cell distribution.
label initial cell
count/foam
cycle count view from
top*
view on
cross
section*
view from
bottom*
I 8.3 ·1050.5 + o -
II 8.3 ·1054.5 + o -
III 2.5 ·1060.5 +/100% o/100% -
IV 2.5 ·1064.5 +/100% +/100% -
V 3.6 ·1062.5 + o -
VI 5 ·1060.5 o/100% -/100% -
VII 5 ·1064.5 o/100% -/100% -
108 9.3 Madin-Darby canine kidney cells (MDCK )
Figure 9.5: Hematoxylin/eosin staining of MDCK following 2 weeks of cultivation with
setup IV from table 9.3. Top left shows a picture of MDCK grown in a culture flask
(FDA/EB staining). The image from the middle of the foam is taken in approximately
3 mm depth. Bars represent 500 µm, arrows point to single cells.
were cultivated for 14 days. Here, experimental setup IV from table 9.3 was chosen, as it
yielded the best results regarding cell vitality and distribution within the tested range.
Cells were optically analyzed by hematoxylin/eosin dyeing and also by scanning electron
microscopy (chapter 3.6.3), revealing very good results regarding cell distribution, see
figure 9.5. Cell growth on both top and buttom was dense, but single cells still to be
recognized. For the cross section, cell distribution was excellent and cell density was
good as well. However, both cell density and cell distribution decreased with depth, re-
sulting in much less cellular growth on the foams’ bottom. Scanning electron microscopy
showed a dense layer of cells covering the ceramic surface in the cross section, see figure
9.6.
Introduction of other cell types - synopsis
Growth and high viability of A549, primary fibroblasts and MDCK was confirmed on
the ceramic’s top, bottom and inside the ceramic structure for cylinders of 5mmheight
9. Applicability of the reactor system for other cell types 109
Figure 9.6: Scanning electron microscopy of MDCK following 2 weeks of cultivation with
setup IV from table 9.3, images were taken from a cross section of the ceramic. Bar
represents 20 µm, black arrows point to single cells, white arrows point to the plaster
stone-like ceramic surface.
and with larger pores compared to standard foams. Here, operation techniques only
had to be adjusted regarding inoculation density and flow velocity, whereby only a few
experiments had to be performed. Growth and therefore sufficient nutrient supply inside
the ceramic was maintained for 14 days. For fibroblasts, the ceramic was partly covered
by a layer of fibrous structures (see figure 9.4), presumably structural proteins of the
ECM as collagen or fibronectin, confirming the hypothesis that cells find a close-to-
natural environment inside the porous ceramic structure. For MDCK a dense layer of
cells was observed even inside the ceramic structure (see figure 9.6) with cells growing
on the ceramic’s surface and the bridges between pores.
Summarizing, next to CHO-K1 three further cell types were proven to grow on and in
the ceramic structure in the developed bioreactor system.
10. Conclusion 111
10 Conclusion
I demonstrated the feasibility of cell cultivation inside porous ceramics by a combined
approach of oscillatory inoculation and medium perfusion for cultivation. For this pur-
pose I designed and operated a modular bioreactor system that allows incorporation
of ceramics with different geometries, adjustment of multiple operational strategies and
allows addition of further components (e.g.glass olives). Four different cell types were
cultivated inside the reactor device growing all over the ceramics’ surface and volume.
Operation techniques
I found a strong influence of initial inoculation strategy on cell vitality and distribution
over the scaffold’s volume. Hereby, best results were achieved for seeding cells dynami-
cally in an oscillatory perfusion approach directly inside the reactor device (chapter 6.3)
and cultivating them further by means of perfusion, which is in good agreement with
findings from the literature [Burg et al., 2000].
To evaluate cell number and metabolism, I established an in-line metabolic assay based
on the kinetics of reduction of resazurin (chapter 5.1.3) which allowed the comparison
of different experimental outcomes and a first evaluation of cell number inside reactor
modules. For an approximation of biomass on and in ceramics I moreover conceived a
protocol for the determination of carbon content in ceramics based on carrier hot gas
extraction. Carbon content then was used to estimate cell number inside ceramics and
hence also for comparison of experiments.
A perfusion flow velocity of 0.33mlmin−1cm−2was found to support growth of CHO-K1,
MDCK and human primary fibroblasts, whereas for A549 a flow velocity of 0.66 ml min−1cm−2
should be favored. Clearly, due to ceramics’ porous structure and therefore tortuosity
effects [Habisreuther et al., 2008], actual flow velocity will strongly depend on location
inside the scaffold. However, for several flow paths convection will not much be hindered
and flow through the porous matrix can be assumed to be 0.33-0.66mlmin−1cm−2(cor-
responding to 0.055-0.11mms−1), which then is in the range of sinusoidal blood flow
(0.01-0.3mms−1) in bone marrow [Wickramasinghe, 1975]. As porous ceramics display
112
strong resemblance with trabecular bone structure hosting bone marrow, diffusion from
flow paths with velocities as above is anticipated to be sufficient for nourishment of cells
deep inside sparsely accessible pores.
Ceramic scaffold
Regarding foam geometry, a strong influence of length of the ceramics was found, as
cylinders of 10mmheight supported cell growth inside the scaffold directly after seed-
ing. However, cell density strongly decreased with scaffold depth and cultivation time
(figure 6.16, page 77). Therefore, dynamic oscillatory inoculation was not performed for
foams longer than 10mm or for foams series connected in a tubular reactor as described
in chapter 3.2.1. I assume filtration effects very common for porous growth substrates
which lead to pore blockage on top of the ceramics and therefore prevent further cell
penetration as well as medium from flowing through the total foam volume during fur-
ther course of cultivation.
For ceramics of 5mmheight I found very good results regarding cell vitality and dis-
tribution. Here, total foam volume is occupied by cells (see figure 7.6, page 93). For
these more flat foams, pore size was adjusted additionally to be slightly higher than for
ceramics of 10mmheight, leading to a coarser foam structure. Good cellular growth in
these scaffolds is ascribed to smoother flow through ceramics depending on its shape
and internal structure. Therefore, influence of pore size, pore openings and pore edge
structure on cellular growth behavior should be studied in more detail to gain deeper
insight into cell-ceramic interaction. With hereby gained results, internal pore structure
could be adjusted in micro-metre range and prediction and control of cell growth and
distribution inside ceramics would be possible. Clearly, feasibility of modulating µm-
scaled pore sizes, openings and even flow channels in a scaffold strongly depends on the
used material. Therefore, the list of material characteristics suitable for cell cultivation
as suggested by [Meuwly et al., 2007] (table 2.1, page 8) should be expanded by the item
of good material ductility.
Cell growth
For long-term cultivation, ECM resembling layers were found on top of the ceramics (see
figure 7.3, page 89) and fibrous structures were observed inside the ceramics’ pores (figure
6.17, page 78). I therefore hypothesize that cells find an excellent micro environment
inside ceramics and secrete extra cellular matrix proteins. Nevertheless, following several
weeks of cultivation, cell clumps were observed mainly on the ceramics’ surface and in
10. Conclusion 113
the bioreactor’s periphery which led to blocking of tubings and a decrease in overall cell
vitality as cells inside the aggregates undergo necrosis or apoptosis (chapter 8).
Therefore, especially for week-long cultivations usage of tubings should be avoided and
the reactor’s interior should be easily accessible to remove cell debris. Timmins and
colleagues in a recent publication present a tube-less perfusion bioreactor, the T-CUP
[Timmins et al., 2007]. Here, ceramic scaffolds are moved through the medium solution
therefore enforcing medium to pass through. As scaffolds are actuated back and forth,
not only seeding is performed by help of oscillatory perfusion, but also nutrient supply of
cells inside ceramics is ensured by oscillatory fluid flow. Compared to unidirectional flow
during cultivation, oscillatory perfusion is known to strongly increase cell vitality inside
the scaffold [Du et al., 2008]. In the performed experiments, I found cells which were
brought into scaffolds’ pores by oscillatory perfusion to die when further cultivated under
unidirectional flow (see figure 6.16, page 77). I therefore hypothesize an occurrence of
different flow paths through the scaffold for each flow direction. Then, cells in flow paths
that can be accessed only from the bottom clearly have to be supplied by medium flow
from the bottom of the foam.
Critical remarks
The present essay is a synopsis of quite different fields of science. Data had to be
collected for the reactor device (materials and physics, ceramic characteristics, flow pat-
terns), employed assays (cell number determination, mathematical modeling of resazurin
reduction, etc.), and bioreactor operation (inoculation, nutrient supply). Herein, besides
gained knowledge as how to seal ceramics laterally in order to direct flow through them,
several setbacks had to be accepted, e.g., problems in direct assessment of cell number
inside the ceramic foams. I however established two methods for cell determination in-
stead, but these are clearly more qualified for comparison of test runs than for specifying
explicit cell counts. The resazurin assay, e.g., assumes laminar plug flow behavior inside
the reactor device, which clearly is invalid as implied by figure 4.2 (page 41).
I found perfusion inoculation to yield good results regarding cell distribution for ceram-
ics with larger pores, but reproducibility of the experiments was poor (i.e.a dispersion
factor of 2.5 for glucose consumption, see figure 7.4 on page 90). Moreover, ceramics
with larger pores were found to be less robust, therefore, the manufacturing process
should be revised.
Medium flow and hereby nutrient distribution inside the very inhomogeneous ceramic
scaffold is a rather random event and probably leads to uneven cell growth. I found
114
mucous layers containing dead cells on top of the scaffolds and inside the reactor pe-
riphery in conjunction with unpopulated sites inside the ceramic (see figure 7.3 on page
89), therefore nutrient supply seems to be very inhomogeneous and should be adjusted
considering the ceramic’s structure.
However, homogeneous cell distributions and growth inside the ceramic scaffolds confirm
qualification of the reactor system at hand. Biomasses corresponding to about 4 ·107
cells/ml scaffold volume (as calculated from table 7.1, page 88) were achieved raising
confident hope for even more if scaffold structure and nutrient supply are adjusted more
elaborately.
Outlook
A huge challenge still needs to be overcome is the abundance of cell aggregates inside
the reactor device. Naturally, avoiding tubings - as for the T-CUP mentioned above -
only minimizes the risk of bad nutrient supply due to cell blockage but does not hinder
cell clump occurrence. Clearly, it would be more useful to prevent the occurrence of cell
aggregates in the first place. As cell clumps mainly occur when cell proliferation is high
due to nutrient abundance, a well considered strategy of nutrient provision should be
incorporated into reactor operation. Hereby, cells would be supplied with just enough
nutrients to guarantee cell survival, but further proliferation would be prohibited. More-
over, if deeper insight was gained into how internal structure affects cellular growth (as
suggested above), a combined approach of imprinting flow channels and control of nu-
trient supply could modulate cellular growth spatially.
Recently, other bioreactor systems employing ceramics as cell cultivation scaffolds came
up. Here, good results regarding cell distribution over the scaffold, cell viability and
even cellular differentiation were shown for cells cultivated under perfusion conditions
in scaffolds of a similar shape as those presented in this work [Holtorf et al., 2005,
Timmins et al., 2007, Du et al., 2008]. However, trial-and-error-based concepts to im-
prove bioreactor performance are time consuming and expensive. A more elaborate
strategy would be to outsource experiments to computations in silico, which evidently
requires an excellent underlying mathematical model of cellular growth inside porous
perfused scaffolds. [Mehta and Linderman, 2005] proved the successful design and op-
eration of a bioreactor depending almost exclusively on theoretical considerations and
avoiding laboratory experiments.
All in all, mathematical models describing cellular growth spatially resolved in a porous
scaffold would help us to understand cellular growth inside ceramics in more detail, could
10. Conclusion 115
enable adjustment of favorable micro milieus by nutrient supply and even promote re-
actor and scaffold improvement.
Then, reactors as presented in this work would represent an excellent basis for three
dimensional cell cultivation, comforting cells to behave as in organs and consequently
allow fulfilling tissue engineering demands or requirements for production of highly spe-
cific therapeutics.
Appendix 117
Appendix
118 Appendix
List of laboratory equipment
designation description origin
shaker for agitation of ceramics in reaction tubes,
works inside a humidified CO2incubator
home made design, see
figure 10.1
clean bench for aseptic cell cultivation procedures LaminAir HB2448, Her-
aeus Instruments
centrifuge cell harvesting Centrifuge 5804R, Ep-
pendorf
autoclave sterilization of cell culture material/reac-
tor modules
Certoclav, certoclav
sterilizer GmbH (Aus-
tria)
laboratory oven drying of ceramics prior to inoculation T6030, Heraeus Instru-
ments
CO2incubator cell cultivation under defined conditions,
5% CO2, 90% water saturated air
Heracell 150, Heraeus
Instruments
Biostat B plus serving as a recirculation container for re-
actor modules
Biostat B plus, Sarto-
rius
climatic chamber conditioning of modules connected to Bio-
stat B plus
home made design, see
figure 3.4
light microscope evaluation of cell growth for experiments
in culture flasks or plates
Diavert, Leitz Wetzlar
upright fluores-
cence microscope
live/dead differentiation in vitality assay
for ceramics
Axio Scope.A1, Zeiss
hemocytometer determination of cell concentration by mi-
croscopic evaluation
Thoma, LaborOptik
photometer color absorption for resazurin assays and
reactor characterization
Tecan Sunrise, Tecan
Trading AG
Figure 10.1: Illustration of self-made shaker.
Appendix 119
Cell culture materials
designation description origin
culture flask flask for 2-dimensional cell
cultivation (growth surface 25,
75 or 175 cm2
Greiner bio-one, 690175/ 658175/
660175
cell culture plate 6-, 12- or 24-well plate for 2-
dimensional cell cultivation or
static cultivation of cells on
ceramics
Greiner bio-one, 657160/ 665180/
662160
reaction tube 50 ml PP tube for agitation
cultivation of cells on ceram-
ics
Greiner bio-one, 227261
aqua dest. double-distilled water double distilled in Köttermann 1080
PBS washing solution for cells prior
to cell detachment
137mM NaCl (Merck, 1.06404);
2.7mM KCl (Merck, 1.04935);
5.7mM Na2HPO4(Merck, 1.06580);
1.5mM KH2PO4(Merck, 1.04873);
pH=7.2, in aqua dest.
PBS++ washing solution for cells prior
to experiments
137mM NaCl (Merck, 1.06404);
2.7mM KCl (Merck, 1.04935);
4.3mM Na2HPO4(Merck, 1.06580);
1.8mM KH2PO4(Merck, 1.04873);
1mM CaCl2(Merck, 1.02381);
1.5mM MgCl2(Fluka, 63064);
pH=7.2, in aqua dest.
Karnovsky buffer fixing solution for SEM 0.16M sodium cacodylate trihy-
drate (Sigma, C0250); 3.6mM CaCl2
(Merck, 1.02381); 0.04% glutaralde-
hyde (Serva, 23114), in aqua dest.
cacodylate buffer washing solution for SEM 0.2M sodium cacodylate trihydrate
(Sigma, C0250); pH=7.3, in aqua
dest.
acetone dehydration solution for SEM Riedel-de Haen, 65469
TrypLE Express cell detaching solution, can be
used 3x diluted in PBS
Gibco, 12605
120 Appendix
Additional equipment for bioreactor system
designation description origin
peristaltic pump imprinting perfusion to reactor mod-
ules by tubings
1B.1003-R/65, Petro Gas
Ausrüstungen Berlin
tubings conducting medium to and from re-
actor modules, attachment of gas fil-
ters to bottles
silicone
gas filter gas exchange for medium recircula-
tion vessels during cultivation
Acro®50, Pall Corporation
teflon tape sealing of ceramics EN 751-3 GRp R10D10, MB
Dichtungen GmbH Jetzendorf
glass olives for dynamic oscillatory cell inocula-
tion
self-made
seal rings sealing ceramic-PEEK joints, also
used as spacers
O-rings coated with 70 EPDM
281
steel spring separation of ceramics in tube reac-
tor
steel 1.4301 and 1.4571
silicone sealing
disks
sealing PEEK-PEEK joints; used as
membranes
silicone, shore hardness 60/80
syringe inoculation of cells through silicone
membrane
5 ml, Omnifix B Braun
needle inoculation of cells through silicone
membrane
0.8x80 ml, Erosa
Appendix 121
day passage by
split
passage into
medium with
serum content
cultivation
duration
3 1:6 10% 3 d
6 1:6 8% 3 d
9 1:6 8% 2 d
11 1:6 6% 2 d
13 1:6 6% 2 d
15 1:6 4% 2 d
17 1:6 4% 2 d
19 1:6 4% 2 d
21 1:2 3% 3 d
24 1:4 3% 2 d
26 1:6 2% 2 d
28 1:4 2% 3 d
31 1:6 2% 2 d
33 1:6 2% 2 d
35 1:6 2% 3 d
38 1:6 2% 2 d
40 1:6 1% 2 d
42 1:6 1% 3 d
45 1:6 1% 2 d
Table 10.1: Adaptation of CHO-K1 to growth in medium containing 1% FBS.
Adaptation of CHO-K1 to growth in medium with
reduced serum concentration
•thaw cells and transfer them directly in prewarmed culture medium on day 0
•for the next weeks, passage cells according to table 10.1
•on day 47, freeze cells at 5 ·106cells/ml in culture medium Ham-F12 containing
10% dimethyl sulfoxide (Sigma, D5879)
122 Appendix
Pre-cultivation for experiments in reactor modules
CHO-K1 and CHO-K1 adapted to growth in medium containing 1% FBS
•thaw cells and transfer them directly in prewarmed culture medium
•the next day, passage cells by a 1:5 split
•three days later, passage cells by a 1:4 split
•cells are used for inoculation the next day (exponential growth phase)
A549
•thaw cells and transfer them directly in prewarmed culture medium
•the next day, passage cells by a 1:5 split
•three days later, passage cells by a 1:4 split
•cells are used for inoculation the next day (exponential growth phase)
MDCK
•thaw cells and transfer them directly in prewarmed culture medium
•the next day, passage cells by a 1:5 split
•three days later, passage cells by a 1:4 split
•two days later, passage cells by a 1:3 split
•cells are used for inoculation two days later (exponential growth phase)
human primary fibroblasts
•thaw cells and transfer them directly in prewarmed culture medium
•three days later, passage cells by a 1:5 split
•about two weeks later, passage cells by a 1:4 split
•perform a medium replacement after one and four weeks of cultivation
•cells are used for inoculation six weeks after last passaging
Appendix 123
Calculation of flow velocity
Superficial velocity vfor flow through ceramics in reactor modules was calculated by
taking into account porosity (φ) and perfused face area (A). Volumetric flow velocity ˙
V
was calculated from pump characteristic and number of pump head revolutions.
v=˙
V
A(10.1)
A=π∗r2
foam ∗nfoam ∗φ(10.2)
nfoam and rfoam are number and radius of ceramic cylinders inside revolver magazine,
respectively. For ceramics with drillings as in chapter 6.1.4 perfused face area was
calculated as follows:
A= (Atotal −A0
drillings +Adrillings)∗φ(10.3)
Atotal =π∗r2
foam ∗nfoam (10.4)
A0
drillings =π∗r2
drilling ∗ndrilling ∗nfoam (10.5)
Adrillings = 2π∗rdrilling ∗hdrilling ∗ndrilling ∗nfoam +A0
drillings (10.6)
ndrilling,rdrilling, and hdrilling correspond to number of drillings per foam, radius, and
height of each drilling, respectively.
Engineering drawings
The following pages contain engineering drawings of bioreactor devices used in this work,
gently provided by A.Bischof, TU Berlin. These are in order of appearance:
•left shell of revolver module type A
•right shell of revolver module type A
•magazine type A
•revolver module type B
•left shell of revolver module type B
•right shell of revolver module type B
!" !!" #$%
!
&
!
'(
)
*
+!
&
"
% ,-
*
'*
"
%,
!!
+!-
!
!
./#01("*
23345$$
!""
"
#$%#&%##& !'()
*
#+,
*-
.*
*
#
/#
-
$
.$
&0##
$
+$
) 10*
&
)1
#0##
/
**
#
&) 10*
23'"45,&$!
6"778((!
!""##
"
$%&%& !'(
)
*
+,
+$
)
-
## ./'"# 01,&$!
23"445((!
&6
&6
!"!" #$
%&
'
%$$
()*(
((
+
!+)
%$$(,
%$$(("
-./)
0
,
+
"
.1#23 "4
!" !!" #$%
!
&
!
'(
)
*
"
% +,
*
'*
"
%+
!!
-!,
!
!
./#01("*
23345$$
!
&
!""
"
#$%#&%##& !'()
*
#+,
*-
.*
*
#
$
.$
&
$
+$
) /0*
&
)/
#
1
**
#
&) /0*
23'"45,&$!
6"778((!
#
*
*-
Literature 131
Literature
[Ahmed et al., 1994] Ahmed, S. A., Gogal, R. M., and Walsh, J. E. (1994). A new
rapid and simple non-radioactive assay to monitor and determine the proliferation
of lymphocytes: an alternative to [3H]thymidine incorporation assay. J Immunol
Methods, 170(2):211–224.
[Alberts et al., 2007] Alberts, B., Johnson, A., Lewis, J., Raff, M., Roberts, K., and
Walter, P. (2007). Molecular Biology of the Cell. Garland Science, Taylor & Francis
Group, Fifth Edition.
[Altamirano et al., 2001] Altamirano, C., Illanes, A., Casablancas, A., Gamez, X.,
Cairo, J. J., and Godia, C. (2001). Analysis of CHO cells metabolic redistribu-
tion in a glutamate-based defined medium in continuous culture. Biotechnol Prog,
17:1032–1041.
[Anoopkumar-Dukie et al., 2005] Anoopkumar-Dukie, S., Carey, J. B., Conere, T., Sul-
livan, E. O., van Pelt, F. N., and Allshire, A. (2005). Resazurin assay of radiation
response in cultured cells. Br J Radiol, 78:945–947.
[Anton et al., 2008] Anton, F., Suck, K., Diederichs, S., Behr, L., Hitzmann, B., van
Griensven, M., Scheper, T., and Kasper, C. (2008). Design and characterization of a
rotating bed system bioreactor for tissue engineering applications. Biotechnol Prog,
24:140–147.
[Applegate and Stephanopoulos, 1992] Applegate, M. and Stephanopoulos, G. (1992).
Development of a single-pass ceramic matrix bioreactor for large-scale mammalian
cell culture. Biotechnol Bioeng, 40:1056–1068.
[Bagley et al., 1999] Bagley, J., Rosenzweig, M., Marks, D. F., and Pykett, M. J. (1999).
Extended culture of multipotent hematopoietic progenitors without cytokine augmen-
tation in a novel three-dimensional device. Exp Hematol, 27 (3):496–504.
132 Literature
[Bancroft et al., 2002] Bancroft, G. N., Sikavitsast, V. I., van den Dolder, J., Sheffield,
T. L., Ambrose, C. G., Jansen, J. A., and Mikos, A. G. (2002). Fluid flow increases
mineralized matrix deposition in 3D perfusion culture of marrow stromal osteloblasts
in a dose-dependent manner. Proc Natl Acad Sci U S A, 99(20):12600–12605.
[Bischof and Blessing, 2006] Bischof, A. and Blessing, L., editors (2006). Geschäumte
Keramikwerkstoffe: Neue Herausforderungen für den Produktentwicklungsprozess, 17.
Symposium “DESIGN FOR X”, Neukirchen, 12.-13. Oktober 2006.
[Blan and Birla, 2008] Blan, N. R. and Birla, R. K. (2008). Design and fabrication
of heart muscle using scaffold-based tissue engineering. J Biomed Mater Res A,
86(1):195–208.
[Bonassar and Vacanti, 1998] Bonassar, L. J. and Vacanti, C. A. (1998). Tissue engi-
neering: The first decade and beyond. J Cell Biochem Suppl, 72(30-31):297–303.
[Burg et al., 2000] Burg, K. J. L., Jr., W. D. H., Culberson, C. R., Beiler, R. J., Greene,
K. G., Loebsack, A. B., Roland, W. D., Eiselt, P., Mooney, D. J., and Halberstadt,
C. R. (2000). Comparative study of seeding methods for three-dimensional polymeric
scaffolds. J Biomed Mater Res, 51 (4):642–649.
[Carrier et al., 1999] Carrier, R. L., Papadaki, M., Rupnick, M., Schoen, F. J., Bursac,
N., Langer, R., Freed, L. E., and Vunjak-Novakovic, G. (1999). Cardiac tissue engi-
neering: Cell seeding, cultivation parameters, and tissue construct characterization.
Biotechnol Bioeng, 64(5):580–589.
[Carrier et al., 2002] Carrier, R. L., Rupnick, M., Langer, R., Schoen, F. J., Freed, L. E.,
and Vunjak-Novakovic, G. (2002). Perfusion improves tissue architecture of engineered
cardiac muscle. Tissue Eng, 8(2):175–188.
[Dar et al., 2002] Dar, A., Shachar, M., Leor, J., and Cohen, S. (2002). Optimization
of cardiac cell seeding and distribution in 3D porous alginate scaffolds. Biotechnol
Bioeng, 80(3):305–312.
[Davisson et al., 2002] Davisson, T., Sah, R. L., and Ratcliffe, A. (2002). Perfusion
increases cell content and matrix synthesis in chondrocyte three-dimensional cultures.
Tissue Eng, 8(5):807–816.
Literature 133
[Davoren et al., 2006] Davoren, M., Herzog, E., Casey, A., Cottineau, B., Chambers,
G., Byrne, H. J., and Lyng, F. M. (2006). In vitro toxicity evaluation of single walled
carbon nanotubes on human A549 lung cells. Toxicol In Vitro, 21(3):438–448.
[Deshpande and Heinzle, 2004] Deshpande, R. R. and Heinzle, E. (2004). On-line oxy-
gen uptake rate and culture viability measurement of animal cell culture using mi-
croplates with integrated oxygen sensors. Biotechnol Lett, 26(9):763–767.
[Driemel, 2010] Driemel, G. (2010). Modifizierung keramischer Aluminiumoxidober-
flächen zur Wachstumsoptimierung adhärenter CHO-K1 Zellen. PhD thesis, Tech-
nische Universität Berlin, School III Process Sciences, in preparation.
[Du et al., 2008] Du, D., Furukawa, K., and Ushida, T. (2008). Oscillatory perfusion
seeding and culturing of osteoblast-like cells on porous beta-tricalcium phosphate
scaffolds. J Biomed Mater Res A, 86A(3):796–803.
[Fassnacht and Poertner, 1999] Fassnacht, D. and Poertner, R. (1999). Experimental
and theoretical considerations on oxygen supply for animal cell growth in fixed-bed
reactors. J Biotechnol, 72:169–184.
[Freed and Vunjak-Novakovic, 1997] Freed, L. E. and Vunjak-Novakovic, G. (1997). Mi-
crogravity tissue engineering. In Vitro Cell Dev Biol Anim, 33(5):381–385.
[Garrn et al., 2004] Garrn, I., Reetz, C., Brandes, N., Kroh, L. W., and Schubert, H.
(2004). Clot-forming: the use of proteins as binders for producing ceramic foams. J
Eur Ceram Soc, 24(3):579–587.
[Genzel et al., 2004] Genzel, Y., Behrendt, I., Konig, S., Sann, H., and Reichl, U. (2004).
Metabolism of MDCK cells during cell growth and influenza virus production in large-
scale microcarrier culture. Vaccine, 22(17-18):2202–2208.
[Goldstein et al., 2001] Goldstein, A. S., Juarez, T. M., Helmke, C. D., Gustin, M. C.,
and Mikos, A. G. (2001). Effect of convection on osteoblastic cell growth and function
in biodegradable polymer foam scaffolds. Biomaterials, 22(11):1279–1288.
[Gonzalez and Tarloff, 2001] Gonzalez, R. J. and Tarloff, J. B. (2001). Evaluation of
hepatic subcellular fractions for Alamar blue and MTT reductase activity. Toxicol In
Vitro, 15:257–259.
134 Literature
[Gooch et al., 2001] Gooch, K. J., Kwon, J. H., Blunk, T., Langer, R., Freed, L. E.,
and Vunjak-Novakovic, G. (2001). Effects of mixing intensity on tissue-engineered
cartilage. Biotechnol Bioeng, 72(4):402–407.
[Goralczyk et al., 2009] Goralczyk, V., Driemel, G., Kroh, L. W., and King, R. (2009).
Zellzahlbestimmung in dreidimensionalen, porösen Gerüststrukturen. CIT Special Is-
sue: Process-Net-Jahrestagung und 27. Jahrestagung der Biotechnologen, 81(8):1271.
[Grampp et al., 1996] Grampp, G. E., Applegate, M. A., and Stephanopoulos, G.
(1996). Cyclic operation of ceramic-matrix animal cell bioreactors for controlled se-
cretion of an endocrine hormone. a comparison of single-pass and recycle modes of
operation. Biotechnol Prog, 12:837–846.
[Habisreuther et al., 2008] Habisreuther, P., Djordjevic, N., and Zarzalis, N. (2008). Nu-
merische Simulation der Mikroströmung in porösen inerten Strukturen. Chemie In-
genieur Technik, 80(3):327–341.
[Hagen, 2005] Hagen, J. (2005). Chemiereaktoren; Nichtideale Reaktoren und Reaktor-
modelle, chapter 7. Wiley.
[Hansen and Emborg, 1994] Hansen, H. A. and Emborg, C. (1994). Influence of am-
monium on growth, metabolism, and productivity of a continuous suspension chinese
hamster ovary cell culture. Biotechnol Prog, 10:121–124.
[Harada et al., 1996] Harada, Y., Wang, J. T., Doppalapudi, V. A., Willis, A. A., Jasty,
M., Harris, W. H., Nagase, M., and Goldring, S. R. (1996). Differential effects of dif-
ferent forms of hydroxyapatite and hydroxyapatite tricalcium phosphate particulates
on human monocyte macrophages in vitro. J Biomed Mater Res, 31(1):19–26.
[Holtorf et al., 2005] Holtorf, H. L., Sheffield, T. L., Ambrose, C. G., Jansen, J. A., and
Mikos, A. G. (2005). Flow perfusion culture of marrow stromal cells seeded on porous
biphasic calcium phosphate ceramics. Ann Biomed Eng, 33(9):1238–1248.
[Holy et al., 2000] Holy, C., Shoichet, M. S., and Davies, J. E. (2000). Engineering
three-dimensional bone tissue in vitro using biodegradable scaffolds: Investigating
initial cell-seeding density and culture period. J Biomed Mater Res, 51 (3):376–382.
[Innocentini et al., 1998] Innocentini, M. D. M., Salvini, V. R., Pandolfelli, V. C., and
Coury, J. R. (1998). The permeability of ceramic foams. www.ceramicbulletin.org, 78
(9).
Literature 135
[Janssen et al., 2006] Janssen, F. W., Hofland, I., van Oorschot, A., Oostra, J., Peters,
H., and van Blitterswijk, C. A. (2006). Online measurement of oxygen consumption by
goat bone marrow stromal cells in a combined cell-seeding and proliferation perfusion
bioreactor. J Biomed Mater Res A, 79(2):338–348.
[Kim et al., 1997] Kim, B.-S., Putnam, A. J., Kulik, T. J., and Mooney, D. J.
(1997). Optimizing seeding and culture methods to engineer smooth muscle tissue
on biodegradable polymer matrices. Biotechnol Bioeng, 57(1):46–54.
[Kitagawa et al., 2006] Kitagawa, T., Yamaoka, T., Iwase, R., and Murakami, A. (2006).
Three-dimensional cell seeding and growth in radial-flow perfusion bioreactor for in
vitro tissue reconstruction. Biotechnol Bioeng, 93(5):947–954.
[Knazek et al., 1972] Knazek, R., Gullino, P., Kohler, P., and Dedrick, R. (1972). Cell
culture on artificial capillaries: an approach to tissue growth in vitro. Science,
178(56):65–66.
[Krajewski et al., 1996] Krajewski, A., Ravaglioli, A., Kirsch, M., Biagini, G., Solmi, R.,
Belmonte, M., Zucchini, C., Gandolfi, M. G., Castaldini, C., Rodriguez, L., Giardino,
R., Mongiorgi, R., Roncari, E., and Orlandi, L. (1996). Ceramic support for cell
cultures. J Mater Sci Mater Med, 7:99–102.
[Ku et al., 1981] Ku, K., Kuo, M. J., Delente, J., Wildi, B. S., and Feder, J. (1981). De-
velopment of a hollow-fiber system for large-scale culture of mammalian-cells. Biotech-
nol Bioeng, 23(1):79–95.
[Leach and Schmidt, 2004] Leach, J. B. and Schmidt, C. E., editors (2004). Pho-
tocrosslinkable Hyaluronic Acid Hydrogels for Tissue Engineering, Mat Res Soc Symp
Proc EXS-1.
[Lee et al., 1991] Lee, D. W., Grace, J. R., Chow, B. K. C., MacGillivray, R. T. A.,
and Kilburn, D. G. (1991). High density cultivation of BSK cells on sintered alumina
ceramic foam support. Cytotechnology, 5:233–241.
[LeGeros, 2002] LeGeros, R. Z. (2002). Properties of osteoconductive biomaterials: Cal-
cium phosphates. Clin Orthop Relat Res, 395:81–98.
[Li et al., 2001] Li, Y., Ma, T., Kniss, D. A., Lasky, L. C., and Yang, S.-T. (2001).
Effects of filtration seeding on cell density, spatial distribution, and proliferation in
nonwoven fibrous matrices. Biotechnol Prog, 17(5):935–944.
136 Literature
[Li et al., 2008] Li, Z., Gunn, J., Chen, M.-H., Cooper, A., and Zhang, M. (2008). On-
site alginate gelation for enhanced cell proliferation and uniform distribution in porous
scaffolds. J Biomed Mater Res A, 86(2):552–559.
[Liao and Cui, 2004] Liao, S. S. and Cui, F. Z. (2004). In vitro and in vivo degradation of
mineralized collagen-based composite scaffold: nanohydroxyapatite/collagen/poly(L-
lactide). Tissue Eng, 10(1-2):73–80.
[Lieber et al., 1976] Lieber, M., Smith, B., Szakal, A., Nelson-Rees, W., and Todaro, G.
(1976). A continuous tumor-cell line from a human lung carcinoma with properties of
type II alveolar epithelial cells. Int J Cancer, 17:62–70.
[Lydersen, 1987] Lydersen, B. K. (1987). Perfusion Cell Culture System Based on Ce-
ramic Matrices, in ”Large Scale Cell Culture Technology“. B. K. Lydersen.
[Lydersen et al., 1985] Lydersen, B. K., Pugh, G. G., Paris, M. S., Sharma, B. P., and
Noll, L. A. (1985). Ceramic matrix for large scale animal cell culture. Nat Biotechnol,
3:63–67.
[Marler et al., 1998] Marler, J. J., Upton, J., Langer, R., and Vacanti, J. P. (1998).
Transplantation of cells in matrices for tissue regeneration. Adv Drug Deliv Rev,
33:165–182.
[Martin and Vermette, 2005] Martin, Y. and Vermette, P. (2005). Bioreactors for tis-
sue mass culture: design, characterization, and recent advances. Biomaterials,
26(35):7481–7503.
[Mehta and Linderman, 2005] Mehta, K. and Linderman, J. J. (2005). Model-based
analysis and design of a microchannel reactor for tissue engineering. Biotechnol Bioeng,
94(3):596–609.
[Meuwly et al., 2007] Meuwly, F., Ruffieux, P.-A., Kadouri, A., and von Stockar, U.
(2007). Packed-bed bioreactors for mammalian cell culture: Bioprocess and biomedical
applications. Biotechnol Adv, 25(1):45–56.
[Möhler et al., 2008] Möhler, L., Bock, A., and Reichl, A. (2008). Segregated mathe-
matical model for growth of anchorage-dependent MDCK cells in microcarrier culture.
Biotechnol Prog, 24:110–119.
Literature 137
[Minuth et al., 2003] Minuth, W. W., Strehl, R., and Schumacher, K. (2003). Zukunfts-
technologie Tissue Engineering. Wiley-VCH.
[Mitsuda et al., 1991] Mitsuda, S., Matsuda, Y., Kobayashi, N., Suzuki, A., Itagaki,
Y., Kumazawe, E., Higashio, K., and Kawanishi, G. (1991). Continuous production
of tissue plasminogen activator (t-PA) by human embryonic lung diploid fibroblast,
IMR-90 cells, using a ceramic bed reactor. Cytotechnology, 6(1):23–31.
[Moutos et al., 2007] Moutos, F. T., Freed, L. E., and Guilak, F. (2007). A biomimetic
three-dimensional woven composite scaffold for functional tissue engineering of carti-
lage. Nat Mater, 6(2):162–167.
[Narula et al., 1998] Narula, P., Xu, J., Kazzaz, J. A., Robbins, C. G., Davis, J. M.,
and Horowitz, S. (1998). Synergistic cytotoxicity from nitric oxide and hyperoxia in
cultured lung cells. Am J Physiol Lung Cell Mol Physiol, 274(3):L411–L416.
[Navarro et al., 2001] Navarro, F. A., Mizuno, S., Huertas, J. C., Glowacki, J., and
Orgill, D. P. (2001). Perfusion of medium improves growth of human oral neomucosal
tissue constructs. Wound Repair Regen, 9(6):507–512.
[Neves et al., 2005] Neves, A. A., Medcalf, N., and Brindle, K. M. (2005). Influence
of stirring-induced mixing on cell proliferation and extracellular matrix deposition in
meniscal cartilage constructs based on polyethylene terephthalate scaffolds. Biomate-
rials, 26:4828–4836.
[O‘Brien et al., 2000] O‘Brien, J., Wilson, I., Orton, T., and Pognan, F. (2000). Investi-
gation of the alamar blue (resazurin) fluorescent dye for the assessment of mammalian
cell cytotoxicity. Eur J Biochem, 276:5421–5426.
[Orlandi et al., 1997] Orlandi, L., Solmi, R., Krajewski, A., Bearzatto, A., Biagini, G.,
Ciccopiedi, E., and Ravaglioli, A. (1997). Cell growth on cordierite: an approach to the
identification of reliable supports for continuous-flow solid-bed reactors. Biomaterials,
18 (14):955–961.
[Ouyang and Yang, 2007] Ouyang, A. and Yang, S.-T. (2007). Effects of mixing intensity
on cell seeding and proliferation in three-dimensional fibrous matrices. Biotechnol
Bioeng, 96 (2):371–380.
138 Literature
[Park and Stephanopoulos, 1993] Park, S. J. and Stephanopoulos, G. (1993). Packed-
bed bioreactor with porous ceramic beads for animal-cell culture. Biotechnol Bioeng,
41(1):25–34.
[Pazzano et al., 2000] Pazzano, D., Mercier, K. A., Moran, J. M., Fong, S. S., DiBiasio,
D. D., Rulfs, J. X., Kohles, S. S., and Bonassar, L. J. (2000). Comparison of chon-
drogensis in static and perfused bioreactor culture. Biotechnol Prog, 16(5):893–896.
[Piret et al., 1991] Piret, J. M., Devens, D. A., and Cooney, C. L. (1991). Nutrient and
metabolite gradients in mammalian cell hollow fiber bioreactors. Can J Chem Eng,
69:421–428.
[Radisic et al., 2003] Radisic, M., Euloth, M., Yang, L., Langer, R., Freed, L. E., and
Vunjak-Novakovic, G. (2003). High-density seeding of myocyte cells for cardiac tissue
engineering. Biotechnol Bioeng, 82(4):403–414.
[Schubert et al., 2004] Schubert, H., Garrn, I., Berthold, A., Knauf, W. U., Reufi, B.,
Fietz, T., and Gross, U. M. (2004). Culture of haematopoietic cells in a 3-D bioreactor
made of Al2O3 or apatite foam. J Mater Sci Mater Med, 15(4):331–334.
[Scudiero et al., 1988] Scudiero, D. A., Shoemaker, R. H., Paull, K. D., Monks, A.,
Tierney, S., Nofziger, T. H., Currens, M. J., Seniff, D., and Boyd, M. R. (1988).
Evaluation of a soluble tetrazolium formazan assay for cell-growth and drug sensitivity
in culture using human and other tumor-cell lines. Cancer Res, 48(17):4827–4833.
[Sodian et al., 2002] Sodian, R., Lemke, T., Fritsche, C., Hoerstrup, S. P., Fu, P.,
Potapov, E. V., Hausmann, H., and Hetzer, R. (2002). Tissue-engineering bioreactors:
A new combined cell-seeding and perfusion system for vascular tissue engineering. Tis-
sue Engineering, 8 (5):863–870.
[Suck et al., 2008] Suck, K., Fischer, M., van Griensven, M., Stahl, F., Scheper, T.,
Kasper, C., Hoffmeister, H., and Behr, L. (2008). Cultivation of MC3T3-E1 cells on
a newly developed material (Sponceram®) using a rotating bed system bioreactor. J
Biomed Mater Res A, 80A(2):268–275.
[Sutherland et al., 1986] Sutherland, R. M., Sordat, B., Bamat, J., Gabbert, H., Bour-
rat, B., and Mueller-Klieser, W. (1986). Oxygenation and differentiation in multicel-
lular spheroids of human colon carcinoma. Cancer Res, 46:5320–5329.
Literature 139
[Suzuki et al., 1994] Suzuki, T., Sato, T., and Kominami, M. (1994). A dense cell re-
tention culture system using a stirred ceramic membrane reactor. Biotechnol Bioeng,
44:1186–1192.
[Tharakan and Chau, 1986] Tharakan, J. P. and Chau, P. C. (1986). A radial flow
hollow fiber bioreactor for the large-scale culture of mammalian-cells. Biotechnol
Bioeng, 28(3):329–342.
[Timmins et al., 2007] Timmins, N. E., Scherberich, A., Frueh, J.-A., Heberer, M., Mar-
tin, I., and Jakob, M. (2007). Three-dimensional cell culture and tissue engineering
in a T-CUP (Tissue Culture Under Perfusion). Tissue Eng, 13(8):2021–2028.
[Voytik-Harbin et al., 1998] Voytik-Harbin, S. L., Brightman, A. O., Waisner, B., and
Lamar, C. H. (1998). Application and evaluation of the alamarblue assay for cell
growth and survival of fibroblasts. In Vitro Cell Dev Biol Anim, 34:239–246.
[Vunjak-Novakovic et al., 1996] Vunjak-Novakovic, G., Freed, L. E., Biron, R. J., and
Langer, R. (1996). Effects of mixing on the composition and morphology of tissue-
engineered cartilage. AIChE J, 42(3):850–860.
[Vunjak-Novakovic et al., 1998] Vunjak-Novakovic, G., Obradovic, B., Martin, I., Bur-
sac, P. M., Langer, R., and Freed, L. E. (1998). Dynamic cell seeding of polymer
scaffolds for cartilage tissue engineering. Biotechnol Prog, 14:193–202.
[Wang et al., 2003] Wang, Y. C., Uemura, T., Dong, R., H., T., Kojima, J., and Tateishi,
T. (2003). Application of perfusion culture system improves in vitro and in vivo
osteogenesis of bone marrow-derived osteoblastic cells in porous ceramic materials.
Tissue Eng, 9(6):1205–1214.
[Wendt et al., 2003] Wendt, D., Marsano, A., Jakob, M., Heberer, M., and Martin, I.
(2003). Oscillating perfusion of cell suspensions through three-dimensional scaffolds
enhances cell seeding efficiency and uniformity. Biotechnol Bioeng, 84 (2):205–214.
[Wendt et al., 2006] Wendt, D., Stroebel, S., Jakob, M., John, G. T., and Martin, I.
(2006). Uniform tissues engineered by seeding and culturing cells in 3d scaffolds
under perfusion at defined oxygen tensions. Biorheology, 43(3):481–488.
[Wickramasinghe, 1975] Wickramasinghe, S. N. (1975). Human Bone Marrow. Black-
well Scientific, Oxford.
140 Literature
[Wu et al., 1992] Wu, P., Ray, N. G., and Shuler, M. L. (1992). A single-cell model for
CHO cells. Ann NY Acad Sci, 665:152–187.
[Xiao et al., 1999] Xiao, Y.-L., Riesle, J., and van Blitterswijk, C. A. (1999). Static and
dynamic fibroblast seeding and cultivation in porous PEO/PBT scaffolds. J Mater
Sci Mater Med, 10:773–777.
[Zhao and Ma, 2005] Zhao, F. and Ma, T. (2005). Perfusion bioreactor system for hu-
man mesenchymal stem cell tissue engineering: Dynamic cell seeding and construct
development. Biotechnol Bioeng, 91 (4):482–493.
