Skin and hair on-a-chip: in vitro skin models versus ex vivo tissue maintenance with dynamic perfusion [original]

Cite this: Lab Chip, 2013, 13, 3555

Skin and hair on-a-chip: in vitro skin models versus ex

vivo tissue maintenance with dynamic perfusion3

Received 19th February 2013,

Accepted 1st May 2013

DOI: 10.1039/c3lc50227a

www.rsc.org/loc

Beren Ataç,{

Ilka Wagner,{

Reyk Horland,

Roland Lauster,

Uwe Marx,

Alexander

G. Tonevitsky,

Reza P. Azar

and Gerd Lindner*

Substantial progress has been achieved over the last few decades in the development of skin equivalents

to model the skin as an organ. However, their static culture still limits the emulation of essential

physiological properties crucial for toxicity testing and compound screening. Here, we describe a

dynamically perfused chip-based bioreactor platform capable of applying variable mechanical shear stress

and extending culture periods. This leads to improvements of culture conditions for integrated in vitro skin

models, ex vivo skin organ cultures and biopsies of single hair follicular units.

Introduction

The majority of current commercially available skin equiva-

lents are based on static culture systems emulating human

epidermis only, or combining epidermis and dermis in so-

called full-thickness skin equivalents (for reviews, see

1,2

). None

of the existing systems comprise functional and (patho-)

physiologically important elements, such as an immune

system, vasculature or skin appendices.

Most of the skin-

related pathologies, such as wound healing, skin tumours,

psoriasis, contact allergies, androgenic alopecia, etc., essen-

tially require the elements mentioned above. Consequently,

reproducible in vitro systems should reflect these circum-

stances.

Further bioengineering is particularly necessary for the

implementation of adipose tissue, hair follicles and a

functional vascular network.

4–7

Notably, adipose-derived stro-

mal cells are not only important for lipid metabolism, but also

have a major impact on the regulation of fibroblast and

keratinocyte proliferation by cytokine secretion.

In addition,

the hair follicle serves as a considerable storage area, takes a

substantial role in skin metabolism, contains multiple stem

cell lineages with regenerative capacity, and constitutes the

major penetration route of topically applied substances to the

skin.

9–11

Native skin is supplied through a capillary network in the

dermis, formed by endothelial cells. Therefore, endothelial

cells have been integrated into the dermal portion of full-

thickness skin substitutes.

12,6,7

However, because of the lack

of shear stress in existing in vitro systems, no mature blood

vessels have been produced and no skin substitute with an

artificial blood stream at micro-scale has been described so

far.

Such an improved skin equivalent needs to be implemen-

ted in a culture system which permits a constant oxygen and

nutrient supply, measures and removes toxic metabolites, and

can be intravitally monitored for sustainability over weeks and

even months. This is necessary when it is used, for example, as

a potential replacement of animal studies for repeated dermal

toxicity assays, as required in OECD (Organisation for

Economic Co-operation and Development) guidelines 410

and 411. We have recently successfully developed a PDMS-

based microfluidic chip system with the dimensions of a

conventional glass microscope slide

13,14

to function as a long-

term dynamic bioreactor for various tissues. In this study, we

used the chip-based system to prolong the maintenance and

testing period of a commercially available skin equivalent (SE)

and improve its nutritional and cellular deposit with ex vivo

subcutaneous tissue (SCT). Furthermore, ex vivo skin and

single hair follicular units (follicular unit extracts – FUEs) were

cultured in the bioreactor platform to extend the static

maintenance periods concerning substance testing.

Materials and methods

Device design and assembly

The bioreactor platform we developed (Fig. 1a), which will

subsequently be referred as the ‘‘MOC’’ (Multi-Organ-Chip),

provides a dynamically perfused micro-channel system com-

Technische Universita

¨t Berlin, Institute of Biotechnology, Berlin, Germany.

E-mail: gerd.lindner@tu-berlin.de; Fax: +49 (0)30-314-27914;

Tel: +49 (0)30-314-27910

Institute of General Pathology and Pathophysiology Russian Academy of Medical

Science, Moscow, Russia

Zentrum fu

¨r moderne Haartransplantation / Centre for Modern Hair

Transplantation, Berlin, Germany

Conflict of interest disclosures: the authors declare no competing financial

interests.

{These authors contributed equally to this work.

Lab on a Chip

PAPER

This journal is ßThe Royal Society of Chemistry 2013 Lab Chip, 2013, 13, 3555–3561 |3555

Published on 01 May 2013. Downloaded by TU Berlin - Universitaetsbibl on 29/03/2016 10:22:02.

View Article Online

View Journal

| View Issue

bining an on-chip micropump (Fig. 1b, red circle) with variable

tissue culture compartments (Fig. 1b, blue circle, Fig. 1c). The

MOC system provides a platform to assess the effect of

dynamic perfusion in comparison to common static culture

conditions of different skin model systems. It is designed to

operate two microfluidic circuits simultaneously. Fig. 1b is a

photo of the MOC showing the positioning of the key elements

at a glance. The layout of the MOC supports the flexible

integration of conventional miniaturised tissue culture for-

mats, such as Transwell1(Corning, Lowell, MA, USA) inserts,

special organotypic matrices, or native normal or diseased ex

vivo biopsies in the tissue culture compartments (Fig. 1c blue

rectangles). The micropump is capable of providing a pulsatile

flow of medium through 500 mm wide 6100 mm high

channels with a pumping volume range of 7–70 ml min

and a

frequency of 0.2–2.5 Hz. We used a frequency of 0.3 Hz in our

experiments, which leads to a flow-rate of approximately 30 ml

min

. SEs and skin biopsies (SBs) were each held in a single

96-well Transwell1insert with a surface area of 14.3 mm

placed in the insert holders (Fig. 1d) to provide an air–liquid

interphase (Fig. 1c), while FUEs were positioned directly into

the insert areas leading to submerged culture conditions

within the medium stream (Fig. 1c and e). For a detailed

description of the MOC assembly procedure, please refer to

the work of Wagner and colleagues in this issue.

Tissue sources and culture conditions

We used EpiDermFT

(Mattek, Ashland, MA, USA) as a full-

thickness in vitro model for the in vitro SE experiments. A 5

mm punch biopsy device (Stusche, Teltow, Germany) was used

to adjust the SE’s size to the Transwell1(Corning, Lowell, MA,

USA) insert. Dulbecco’s Modified Eagle’s Medium (DMEM)-

based maintenance medium (EFT-400-MM, Mattek, Ashland,

MA, USA) was used, according to the manufacturer’s instruc-

tions. A total volume of 500 ml medium was in place, while 200

ml was changed every day during the 9 days of culture for both

MOC-based dynamic and static cultures in air–liquid inter-

phase. We used deep Transwell1holder wells for the static

cultures, with the same medium volume in all parallel

cultures.

Human juvenile prepuce was obtained in compliance with

the relevant laws, with informed consent and ethical approval

from the Ethics Committee of the Charite

´Universita

¨tsmedizin,

Berlin, Germany, after routine circumcisions from paediatric

surgery. The tissue was stored in PBS at +4 uC and processed

within 4 h after surgery. Skins were sterilized in 80% ethanol

for 30 s and cut open. SCT was removed from the skin and

placed underneath the SE, prepared as described above. These

combined tissues were positioned in the Transwell1insert

and transferred to the respective MOC compartment utilising

the same experimental conditions as mentioned above for a

duration of 7 days (Fig. 1d).

Similarly, we used juvenile prepuce for the ex vivo skin

experiments and processed them as described above. Three

donors were used for each MOC-based static and dynamic

experiment in parallel. The skin was separated from the

subcutaneous tissue, punched to 5 mm biopsies, fixed in the

Transwell1insert, and transferred to the MOC tissue segment

in air–liquid interphase. SBs were supplemented for 14 days in

high glucose (4.5 mg ml

) DMEM with 10% foetal calf serum

(FCS), 100 units ml

penicillin, 100 mgml

streptomycin,

and L-glutamine. An amount of 200 ml of the total 500 mlof

medium was exchanged every day.

Occipital and temporal scalp skin FUEs containing mainly

growing anagen VI hair follicles were obtained from disposed

excess skin samples derived from male patients aged between

Fig. 1 Multi-organ-chip (MOC). (a) MOC with built-in micropump providing a

pulsatile flow of medium. (b) MOC upside down; red circle: built-in micropump

region, yellow circle: injection port, blue circles: insert areas compatible with

96well-Transwell1inserts. Black arrow shows the direction of flow. This chip

includes two circuits as mirror images. (c) Insert areas are used separately for

culturing ex vivo SBs and in vitro SEs in Transwells1or FUEs directly in the

stream, as shown in the schematic. (d) An SE in a Transwell1insert, being

placed in an MOC Transwell1support. (e) FUEs in MOC insert area.

3556 |Lab Chip, 2013, 13, 3555–3561 This journal is ßThe Royal Society of Chemistry 2013

Paper Lab on a Chip

Published on 01 May 2013. Downloaded by TU Berlin - Universitaetsbibl on 29/03/2016 10:22:02.

View Article Online

25–55 years undergoing hair transplantation surgery. FUEs

were delivered in Williams’s E medium with 100 units/ml

penicillin and 100 mgml

streptomycin, and processed within

4 h after surgery by placing them directly in the microcircula-

tion of the MOC system (Fig. 1c and e). We used William’s E

medium supplemented with 10% FCS, 100 units/ml penicillin,

100 mgml

streptomycin, 5 mgml

human insulin, 2 mM L-

glutamine, and 5 610

M hydrocortisone hemisuccinate

(Sigma Aldrich, Schnelldorf, Germany) as the culture medium.

An amount of 350 ml of a total of 600 ml was exchanged every

second day during the 14 days of culture.

Sample preparation, histology and imaging

Samples were removed from Transwells1inserts, placed in

Tissue-Tek1OCT

Compound (Sakura, Alphen aan den

Rijn, Netherlands) and subsequently snap frozen for end-point

analysis. The tissue blocks were sectioned with a Leica

CM1950 Cryostat (Wetzlar, Germany) at 8 mm thickness.

We used standard haematoxylin and eosin (H&E) staining to

analyse the sections histologically. Immunofluorescence stain-

ing was applied to characterize for tissue-specific markers.

Briefly, we used a Ki67 (Abcam, Cambridge, UK) – TUNEL

(Apoptag, Millipore, Darmstadt, Germany) double immuno-

visualisation technique, as described before

on cryosections

for the detection of proliferating and apoptotic cells. Double

immunofluorescence staining with cytokeratin 10 (CK10;

Abcam, Cambridge, UK) and cytokeratin 15 (CK15; Abcam,

Cambridge, UK) were used for epidermal markers, while

Collagen IV (ColIV; Sigma Aldrich, Schnelldorf, Germany) and

Tenascin C (TenC; Santa Cruz Biotechnology, Heidelberg,

Germany) were visualised for basement membrane and

dermis. In addition, pan cytokeratin (an antibody mixture

that stains a wide range of different CKs; Sigma Aldrich,

Schnelldorf, Germany) and ColIV (Sigma Aldrich, Schnelldorf,

Germany) double immunofluorescence staining was applied

for hair follicle tissue samples.

Fig. 2 In vitro skin equivalents (Mattek, Ashland, US) cultured for 9 days in MOC or static conditions in comparison to the same SE cultured for 7 days with

subcutaneous tissue (SCT) from prepuce. Haematoxylin and eosin (H&E) staining for histological comparison of the sections (a–c, j–l) and immunofluorescence

staining for epidermal markers Cytokeratin 10 and 15 (d–f, m–o) are applied. Overlapping markers are visualized as yellow. Finally, Ki67 and TUNEL assay for

proliferation and apoptosis (g–i, p–r) are used for comparison of viability of tissues (arrows indicate discontinuous basal layer of epidermis in e, TUNEL positive

apoptotic cells in p and disintegrated tissue in q). Dashed lines mark the border between SE and SCT. Scale bars indicate 100 mm for each picture.

This journal is ßThe Royal Society of Chemistry 2013 Lab Chip, 2013, 13, 3555–3561 |3557

Lab on a Chip Paper

Published on 01 May 2013. Downloaded by TU Berlin - Universitaetsbibl on 29/03/2016 10:22:02.

View Article Online

A Keyence BZ-9000 Microscope with the BZ-II-Viewer soft-

ware was used for all microscopic imaging and the images

were merged using the BZ-II-Analyser software (Keyence, Neu

Isenburg, Germany). Phase contrast and fluorescent images

were adjusted for tone levels, brightness and contrast.

Results and discussion

9-day cultured EpiDermFT skin equivalent

We prolonged the recommended culture period for static

conditions of EpiDermFT SE to a total culture period of 9 days.

We could observe a rearrangement and compression of the

dermal matrix structure in dynamic MOC cultures as cell

nuclei are located closer to each other with denser extracellular

matrix structure compared to day 0 and the control. There

were more viable cells in the SEs cultured in the MOC

compared to the SEs with static culture conditions (Fig. 2a–c).

The different layers of the epidermis were stained by

immunofluorescence showing the expression of the epidermal

markers CK10 for differentiating keratinocytes and CK15 for

undifferentiated keratinocytes. Both were similarly expressed

in static and dynamic MOC cultures. Although epidermal

barrier function seemed to be preserved better in the MOC

culture according to the continuous lining of the CK15 positive

cells at the basal layer of the epidermis, there is discontinuity

in static culture as indicated with arrows (Fig. 2d–f). Hardly

any proliferating and apoptotic cells were seen in the MOC and

day 0 control sections according to Ki67 TUNEL double

staining (Fig. 2g, i), whereas selected apoptotic cells could be

found in the epidermal layer in static cultures (Fig. 2h),

suggesting the onset of degradation of this SE.

7-day cultured EpiDermFT skin equivalent with sub-cutaneous

tissue

We used EpiDermFT

SE in combination with SCT dissected

from a prepuce. SCT is composed mainly of adipocytes,

fibroblasts and macrophages. Considering the support of lipid

metabolism and paracrine effects of adipose-derived cells on

the upper levels of skin, we integrated SCT to EpiDermFT

under the same experimental conditions for 7 days. We

observed more viable cells and a compressed dermal matrix in

the samples with SCT cultured in the MOC compared to static

cultures with and without SCT stained by H&E (Fig. 2j–l). SCT

tissue was well integrated to SE in the MOC (Fig. 2l), while its

integration was poor in the static culture (Fig. 2q, arrows

indicating disintegrated tissue). CK15 positive cells are more

abundant in the MOC culture (Fig. 2o) showing the greatest

similarity to the native tissue (Fig. 3d as reference), even

though there is an increase of undifferentiated keratinocytes

in the static culture with SCT compared to regular static

culture (Fig. 2m, n). In contrast to static cultures, proliferating

cells are observed at the basal layer of the epidermis in

cultures with SCT and even more in the dynamic cultures

(Fig. 2p–r). SCT is expected to have a stimulating effect on

epidermal cells and this effect seemed to be even more

prevalent in the dynamic culture. Considering the absence of

proliferating cells in the epidermis in the cultures without

Fig. 3 Maintenance of ex vivo prepuce 14 days in culture. Day zero conditions of

the tissue are shown in a, d, g, and j. Static cultures are shown in b, e, h, and k,

while dynamically cultured tissue in MOC is indicated with c, f, i and l. H&E

staining is applied for histological comparison of the tissues (a–c).

Immunofluorescence staining for epidermal markers Cytokeratin 10 and 15 (d–

f), basement membrane markers Tenascin (TenC) and Collagen IV (ColIV) (g–i),

finally, Ki67 and TUNEL assay for proliferation and apoptosis (j–l) are used for

the evaluation of the viability of the tissues. Overlapping markers are visualized

as yellow. Scale bars indicate 100 mm for each picture.

3558 |Lab Chip, 2013, 13, 3555–3561 This journal is ßThe Royal Society of Chemistry 2013

Paper Lab on a Chip

Published on 01 May 2013. Downloaded by TU Berlin - Universitaetsbibl on 29/03/2016 10:22:02.

View Article Online

SCT, epidermal integrity is expected to be weakened (Fig. 2h,

p). A massive increase in the amount of apoptotic cells is seen

in the SCT of the static culture. The tissue itself is also poorly

integrated to the SE (Fig. 2q). Taking these results into

account, a higher metabolism of the ex vivo tissue seems to

require dynamic perfusion.

14-day cultured ex vivo prepuce

We evaluated SB maintenance in culture for future substance

testing in comparison to SEs. Histological evaluation by H&E

showed epidermal disruption and dermal reorganisation in

the static cultures (Fig. 3b), whereas MOC-cultured biopsies

showed similar distribution to the control tissues (Fig. 3a, c).

In addition, CK10/CK15 staining showed a disintegrated

epidermis in the static culture compared to day 0 and MOC

culture (Fig. 3d–f). ColIV and TenC, which are both synthe-

sised and secreted by keratinocytes and fibroblasts, were used

as markers for the extracellular basement membrane.

ColIV

is mainly found in the basement membrane of the skin,

including the epidermal-dermal border and the dermal-

endodermal border including blood vessels.

TenC is an

extracellular matrix component known to be important for cell

shape and migration behaviour. It has a pro-migratory role for

epidermal cells around the basement membrane in skin and

has been shown to be upregulated during wound healing,

inflammatory processes and fibrosis.

19,20

For native skin

expression pattern of ColIV and TenC, see Fig. 3g. ColIV

expression was generally preserved in contrast to TenC

(Fig. 3g–i). Static SB culture shows elevated levels of TenC

below the basement membrane, suggesting induced fibrotic

processes (Fig. 3h). Proliferation in the epidermis was

maintained in the MOC according to Ki67 positive cells, while

it was decreased in static culture compared to day 0. TUNEL

positive cells in dermis are more abundant in the static culture

compared to the MOC culture, suggesting a higher amount of

cell loss (Fig. 3j–l).

14-day cultured ex vivo follicular unit extracts

We used the MOC system to culture FUEs as a further step in

the attempt to emulate the biology of the skin and its

appendages. In contrast to the well-established Philpott assay

using single, truncated hair follicles to study hair follicle

biology in vitro,

we cultured complete hair follicular units.

These almost intact philosebaceous units include the perifol-

licular epidermis, dermis and the sebaceous gland(s). We

aimed to prolong the culture period of the ex vivo hair follicles,

taking into account the support of the glands and the

surrounding skin tissue on hair follicle maintenance. H&E

staining showed partial loss of structural integrity and a

decrease in the total number of cell nuclei within the central

and proximal hair follicle after 14 days of culture (Fig. 4a, b),

indicating signs of the ongoing regression phase (catagen) of

the hair follicle. Nevertheless, a pan cyto-keratin and ColIV

double immunostaining revealed an intact appearance of the

Fig. 4 FUEs cultured in the MOC for 14 days. H&E staining for day 0 and day 14 sample (a, b). Immunofluorescence staining for collective CKs (PCK) and ColIV for day 0

and day 14 samples (c, d) TUNEL assay and Ki67 staining shown for bulb region of the hair for day 0 and day 14 sample (e, f) (arrows indicate TUNEL positive cells in

the DP and CTS. Light microscopy images from day 0 and day 14 indicating hairs-shaft elongation in culture (g, h). Scale bars indicate 200 mm for a, b, c and d; 50 mm

for e and f; 300 mm for g and h.

This journal is ßThe Royal Society of Chemistry 2013 Lab Chip, 2013, 13, 3555–3561 |3559

Lab on a Chip Paper

Published on 01 May 2013. Downloaded by TU Berlin - Universitaetsbibl on 29/03/2016 10:22:02.

View Article Online

Loading more pages...