Biomaterials 29 (2008) 290–301 Synthetic sandwich culture of 3D hepatocyte monolayer Yanan Du a,b,1 , Rongbin Han a,b,1 , Feng Wen a,b , Susanne Ng San San a,b , Lei Xia a,b , Thorsten Wohland b,d , Hwa Liang Leo a, , Hanry Yu a,b,c,e,f,g, a Institute of Bioengineering and Nanotechnology, A*STAR, Singapore 138669, Singapore b Graduate Programme in Bioengineering, Graduate School for Integrative Sciences and Engineering, National University of Singapore, Singapore 117597, Singapore c Department of Physiology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore 117597, Singapore d Department of Chemistry, Faculty of Science, National University of Singapore, Singapore 117543, Singapore e Singapore-MIT Alliance, E4-04-10, 4 Engineering Drive 3, Singapore 117576, Singapore f NUS Tissue-Engineering Programme, DSO Labs, National University of Singapore, Singapore 117597, Singapore g Department of Haematology–Oncology, National University Hospital, Singapore 119074, Singapore Received 13 June 2007; accepted 17 September 2007 Available online 26 October 2007 Abstract The sandwich culture of hepatocytes, between double layers of extra-cellular matrix (ECM), is a well-established in vitro model for re-establishing hepatic polarity and maintaining differentiated functions. Applications of the ECM-based sandwich culture are limited by the mass transfer barriers induced by the top gelled ECM layer, complex molecular composition of ECM with batch-to-batch variation and uncontrollable coating of the ECM double layers. We have addressed these limitations of the ECM-based sandwich culture by developing an ‘ECM-free’ synthetic sandwich culture, which is constructed by sandwiching a 3D hepatocyte monolayer between a glycine-arginine-glycine-aspatic acid-serine (GRGDS)-modified polyethylene terephthalate (PET) track-etched membrane (top support) and a galactosylated PET film (bottom substratum). The bioactive top support and bottom substratum in the synthetic sandwich culture substituted for the functionalities of the ECM in the ECM-based sandwich culture with further improvement in mass transfer and optimal material properties. The 3D hepatocyte monolayer in the synthetic sandwich culture exhibited a similar process of hepatic polarity formation, better cell–cell interaction and improved differentiated functions over 14-day culture compared to the hepatocytes in collagen sandwich culture. The novel 3D hepatocyte monolayer sandwich culture using bioactive synthetic materials may readily replace the ECM-based sandwich culture for liver tissue engineering applications, such as drug metabolism/toxicity testing and hepatocyte-based bioreactors. r 2007 Elsevier Ltd. All rights reserved. Keywords: Sandwich culture; Hepatocyte; Synthetic materials; Polarity; RGD peptide; Galactosylation 1. Introduction In vivo, hepatocytes are organized into a polarized epithelium with distinct apical (bile canalicular) and basal (sinusoidal) domains [1]. The basal domain of the hepato- cytes is in contact with a complex extracellular matrix (ECM) containing fibronectin, laminin, collagen I–V, and proteoglycans in the space of Disse [2]. The interactions of hepatocytes with the ECM environment are important for hepatic polarity and differentiated function maintenance [3]. In standard in vitro culture, primary hepatocytes cultured on substrates coated with ECM protein, such as collagen or fibronectin, typically exhibit spreading morphology with deteriorating differentiated functions and nearly no pola- rized structure [4]. This deteriorating process could be rescued by overlaying another ECM layer, such as collagen or basement membrane (Matrigel TM ), which mimics the ECM distribution in the space of Disse. Hepatocyte ARTICLE IN PRESS www.elsevier.com/locate/biomaterials 0142-9612/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2007.09.016 Corresponding author. Tel.: +65 65163466; +65 68247103; fax: +65 68727150. Also for correspondence E-mail addresses: [email protected] (H.L. Leo), [email protected] (H. Yu). 1 These authors contribute equally to this work.
12
Embed
Synthetic sandwich culture of 3D hepatocyte monolayer
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Synthetic sandwich culture of 3D hepatocyte monolayer
Yanan Dua,b,1, Rongbin Hana,b,1, Feng Wena,b, Susanne Ng San Sana,b, Lei Xiaa,b,Thorsten Wohlandb,d, Hwa Liang Leoa,�, Hanry Yua,b,c,e,f,g,��
aInstitute of Bioengineering and Nanotechnology, A*STAR, Singapore 138669, SingaporebGraduate Programme in Bioengineering, Graduate School for Integrative Sciences and Engineering, National University of Singapore,
Singapore 117597, SingaporecDepartment of Physiology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore 117597, Singapore
dDepartment of Chemistry, Faculty of Science, National University of Singapore, Singapore 117543, SingaporeeSingapore-MIT Alliance, E4-04-10, 4 Engineering Drive 3, Singapore 117576, Singapore
fNUS Tissue-Engineering Programme, DSO Labs, National University of Singapore, Singapore 117597, SingaporegDepartment of Haematology–Oncology, National University Hospital, Singapore 119074, Singapore
Received 13 June 2007; accepted 17 September 2007
Available online 26 October 2007
Abstract
The sandwich culture of hepatocytes, between double layers of extra-cellular matrix (ECM), is a well-established in vitro model for
re-establishing hepatic polarity and maintaining differentiated functions. Applications of the ECM-based sandwich culture are limited by
the mass transfer barriers induced by the top gelled ECM layer, complex molecular composition of ECM with batch-to-batch variation
and uncontrollable coating of the ECM double layers. We have addressed these limitations of the ECM-based sandwich culture by
developing an ‘ECM-free’ synthetic sandwich culture, which is constructed by sandwiching a 3D hepatocyte monolayer between a
In vivo, hepatocytes are organized into a polarizedepithelium with distinct apical (bile canalicular) and basal(sinusoidal) domains [1]. The basal domain of the hepato-cytes is in contact with a complex extracellular matrix
e front matter r 2007 Elsevier Ltd. All rights reserved.
(ECM) containing fibronectin, laminin, collagen I–V, andproteoglycans in the space of Disse [2]. The interactions ofhepatocytes with the ECM environment are important forhepatic polarity and differentiated function maintenance [3].In standard in vitro culture, primary hepatocytes cultured onsubstrates coated with ECM protein, such as collagen orfibronectin, typically exhibit spreading morphology withdeteriorating differentiated functions and nearly no pola-rized structure [4]. This deteriorating process could berescued by overlaying another ECM layer, such as collagenor basement membrane (MatrigelTM), which mimics theECM distribution in the space of Disse. Hepatocyte
ARTICLE IN PRESSY. Du et al. / Biomaterials 29 (2008) 290–301 291
sandwich culture between double layers of ECM is anin vitro model with re-established hepatic polarity andstable differentiated functions [3,5,6]. The hepatocytesandwich culture has been adopted in liver physiologystudies [7,8], drug metabolism/toxicity testing [9] andhepatocyte-based bioreactors [10,11]. Further applicationsof the conventional ECM-based sandwich culture werehampered by the complex molecular compositions of theECM with batch to batch variation [12], uncontrollableECM coating, mass transfer barriers induced by the gelledECM-coated top support (hindering the exchange ofnutrients, xenobiotics or biochemical signals with the bulkculture medium), and shedding of the ECM coating fromthe top support during culture. In this study, we haveaddressed these limitations of the ECM-based sandwichculture by developing an ‘ECM-free’ synthetic sandwichculture, in which we replaced the natural ECM withbioactive polymeric materials to achieve improved masstransfer and stable differentiated functions.
A variety of synthetic substrata with bioactive compo-nents, such as cell adhesion peptides: Arg-Gly-Asp (RGD)[13], Tyr-Ile-Gly-Ser-Arg (YIGSR) [14], Gly-Phe-Hyp-Gly-Glu-Arg (GFOGER) [15] or sugar ligands: galactose [16],glucose [17], lactose [18], have been used for cell culture toreplace natural ECM with well-controlled material proper-ties and cellular responses. Previously, we have fabricated agalactosylated polyethylene terephthalate (PET-Gal) filmfor primary rat hepatocyte culture and identified a 3Dhepatocyte monolayer formed on the PET-Gal [19]. The3D hepatocyte monolayer exhibited 3D cellular structureand polarities, enhanced cell–cell interactions and differ-entiated functions compared to the 2D hepatocyte mono-layer on collagen-coated substratum [19]. Here, weestablished a synthetic sandwich culture by overlaying the3D hepatocyte monolayer on the PET-Gal (bottomsubstratum) with a porous PET track-etched (TE) mem-brane (top support). Since the biochemical compositions ofECM play essential roles in regulating hepatocyte mor-phology, polarity and differentiated functions in ECM-based sandwich culture [20–22], we investigated theinfluence of three different top support (galactosylated,GRGDS-modified or non-modified PET TE membrane) onthe hepatocyte morphology, polarity and differentiatedfunctions in the 3D hepatocyte monolayer of the syntheticsandwich culture. The synthetic sandwich culture withGRGDS-modified PET TE membrane (top support)/PET-Gal (bottom substratum) exhibited the optimal perfor-mances, in terms of stabilizing the 3D monolayermorphology, re-establishing hepatocyte polarity and main-taining other differentiated functions.
We compared this GRGDS-modified PET TE mem-brane/PET-Gal synthetic sandwich culture of 3D hepato-cyte monolayer with the collagen sandwich hepatocyteculture. 3D hepatocyte monolayer in the synthetic sand-wich culture exhibited similar dynamic process of polarityformation and biliary excretion, improved mass transfer,enhanced cell–cell interaction, differentiated functions
compared with the hepatocytes in the collagen sandwichculture. This synthetic sandwich culture model can replacethe ECM-based sandwich culture for relevant hepatocyte-based applications such as drug metabolism/toxicity testingand hepatocyte-based bioreactors [7,8].
2. Materials and methods
2.1. Materials
PET TE membranes with thickness of 9 mm, pore density of
3� 107 pores/cm2 and pore diameter of 0.8mm were purchased from
Sterlitech (WA, USA). The galactose ligand, 1-O-(60-aminohexyl)-
D-galactopyranoside (AHG, M.W. 279) was synthesized previously
[23–25]. GRGDS peptide was purchased from Peptides International
(Kentucky, USA). Minusheet carriers were purchased from Minucells and
conjugated goat anti-mouse IgG, Invitrogen, Singapore) at RT for 1 h
and rinsed 3� with PBS before being mounted in FluorSaveTM
(Calbiochem, CA). The samples were imaged with a Fluoview-300
confocal microscope 15 (Olympus, Japan) using a 63� water-immersion
objective (NA1.2).
2.11. Measurement of hepatocyte differentiated functions [32]
All functional data were normalized to 106 cells. A Rat Albumin
ELISA Quantitation Kit (Bethyl, Texas) was used for the measurement of
daily albumin production; urea synthesis of the hepatocyte culture
incubated in culture medium with 2mM NH4Cl for 90min was measured
with Urea Nitrogen Kit (Stanbio, Texas); the 7-ethoxyresorufin-O-
deethylation (EROD) assay was initiated by incubating the hepatocytes
with 39.2mM 7-ethoxyresorufin in culture medium at 37 1C for 4 h. The
amount of resorufin converted by the enzymes was calculated by
measuring the resorufin fluorescence in the incubation medium at
543 nm excitation/570 nm emission against resorufin standards. All the
EROD cytochrome P450 1A detoxification activities were normalized
relative to freshly isolated hepatocytes.
2.12. Statistical analysis
Results were presented by mean7standard deviation (M7S.D.). Each
result was statistically analyzed by the t-test. The values of po0.05 were
considered statistically significant.
ARTICLE IN PRESSY. Du et al. / Biomaterials 29 (2008) 290–301 293
3. Result
3.1. Fabrication and characterization of bioactive PET TE
membranes to construct the synthetic sandwich culture
The synthetic sandwich culture was constructedby a PET-Gal as the bottom substratum and a PETTE membrane (GRGDS-modified or galactosylatedor non-modified) as the top support. The entire sand-wich construct was secured in the Minusheet Carriers(Fig. 1A).
The fabrication and characterization of the PET-Gal(bottom substratum) were described previously [26]. Wefabricated here GRGDS-modified or galactosylated PETTE membranes (top support, Fig. 1B) based on thecommercially available PET TE membrane which isnaturally hydrophilic with carboxylic and hydroxyl groupspresented on the bulk material after the ‘track-etching’treatment. The density of the carboxylic groups presentedin the non-modified PET TE membrane manufactured bySterlitech is 5.870.13 nmol/cm2 as quantified by TBOassay. We further increased the functional carboxylicgroup density of the membrane to 19.972.5 nmol/cm2 byoxidization. XPS C 1s core-level peak components of thenon-modified PET TE membrane (Fig. 2A) consist of thearomatic carbon (C–H, binding energy (BE) of 284.6 eV),carbon singly bonded to oxygen (C–O, BE of 286.2 eV),and carboxyl carbon (O–CQO, BE of 288.6 eV) in anapproximate area ratio of 3.5:1:0.6. The area ratio isslightly different from the chemical structure of PET (withthe ratio of 3:1:1) probably due to the particle bombard-
Oxidation
OHOH COOHCOOH OHOH COOHCOOH
OHOH COOHCOOH OHOH COOHCOOH
OHOH OHOH
OHOH OHOHBlockinBlocking
KMO4+H2SO4
EthanolamineEthanolamineLigandLigandLigandLigand
LigandLigandLigandLigand
Fig. 1. Schematic diagrams of the synthetic sandwich construct (A) for hepatoc
or galactose ligand onto the PET TE membrane (B).
ment and alkaline hydrolysis of the polyester bulk materialduring the ‘track etching’ treatment. The peak componentarea associated with the O–CQO species increases in theoxidized PET TE membrane compared with the non-modified PET TE membrane indicating the oxidation ofthe hydroxyl groups into carboxylic groups, while thearea associated with the C–O species decreases accordingly(Fig. 2B).GRGDS peptide or Gal ligand (AHG) was covalently
conjugated onto the oxidized PET TE membrane activatedby EDC and sulfo-NHS. C 1s core-level spectra of both theGRGDS-modified and galactosylated PET TE membranesreveal changes in the surface chemical composition aftersurface modification (Fig. 2A). Successful conjugation ofGRGDS peptide or Gal ligand onto the oxidized PET TEmembrane was confirmed by the appearance of two newpeak components at the BEs of 287.6 and 285.7 eV,attributable to the OQC–NH and the C–N functionalgroups, respectively, and the substantial decrease inthe O–CQO peak component intensity. The successfulconjugation of GRGDS peptide or Gal ligand was alsoconfirmed by XPS wide scanning spectrum (Fig. 2B).In contrast to non-modified and oxidized PET TEmembranes, a new peak corresponding to N 1s (BE of400 eV) introduced by bioactive ligands appeared in thespectra of GRGDS-modified and galactosylated PET TEmembranes. The final density of the conjugated GRGDSpeptide or Gal ligand on the PET TE membrane quan-tified by RP-HPLC was 0.6270.23 nmol/cm2 or 1.1870.34 nmol/cm2, which showed �3% or �6% surfacefunctionality, respectively.
PET TE membrane astop-support
Galactosylated PET film as bottom-substratum
Primary hepatocytes
Minusheet carrier
COOHCOOH COOHCOOH COOHCOOHCOOHCOOH
COOHCOOH COOHCOOH COOHCOOHCOOHCOOH
COOHCOOH COOHCOOH
COOHCOOH COOHCOOH
EDC/NHSEDC/NHS
chemistrychemistry
LigandLigandLigandLigand
LigandLigandLigandLigand
GRGDS or galactose
ligand conjugation
yte culture and surface modification method to conjugate GRGDS peptide
ARTICLE IN PRESS
Binding Energy (eV)
284282 288286 290
Binding Energy (eV)
284282 288286 290
Binding Energy (eV)
284282 288286 290
Binding Energy (eV)
284282 288286 290
C-H O-C
O=C-O
N-C
HN-C=O
Non-modified PET TE membrane Oxidized PET TE membrane
Galactosylated PET TE membrance
Non-modified PET TE membrane Oxidized PET TE membrane
GRGDS-modified PET TE membrane
GRGDS-modified PET TE membrane
Galactosylated PET TE membrane
6000
5000
4000
3000
2000
1000
0
0 200 400 600 800
Binding Energy (eV)
0 200 400 600 800
Binding Energy (eV)
0 200 400 600 800
Binding Energy (eV)
0 200 400 600 800
Binding Energy (eV)
Inte
nsity (
Arb
.Units)
6000
5000
4000
3000
2000
1000
0
Inte
nsity (
Arb
.Units)
C1sO1s
C1s
C1sC1s
N1sN1s
O1s
O1sO1s
6000
5000
4000
3000
2000
1000
0
Inte
nsity (
Arb
.Units)
5000
4000
3000
2000
1000
0
Inte
nsity (
Arb
.Units)
Fig. 2. XPS C 1s core-level spectra (A) and wide scanning spectra (B) of the non-modified PET TE membrane; the oxidized PET TE membrane; GRGDS-
modified PET TE membrane and galactosylated PET TE membrane.
Y. Du et al. / Biomaterials 29 (2008) 290–301294
ARTICLE IN PRESS
Non-modified Galactosylated GRGDS-modified
18
16
14
12
10
8
6
4
2D3 D5 D7
Culture Period (Day)
D3D1 D5 D7
Culture Period (Day)
Ure
a P
roduction (
µg/m
illio
n c
ells
/h) 1.2
1.0
0.8
0.6
0.4
0.2
Rela
tive E
RO
D A
ctivity
(norm
aliz
ed to fre
shly
-isola
tded h
epato
cyte
s
Fig. 3. Effects of the synthetic sandwich culture with three different top supports (galactosylated, GRGDS-modified or non-modified PET TE membrane)
on the sandwiched hepatocytes: (A) stabilization of the monolayer morphology (first panel) and F-actin distribution (second panel); (B) hepatocyte
differentiated functions in synthetic sandwich culture with K, non-modified; ., GRGDS-modified; m, galactosylated PET TE membrane. Data are
mean7S.D., n ¼ 6. *, po0.05; **, po0.01; NS, not significant.
Y. Du et al. / Biomaterials 29 (2008) 290–301 295
3.2. Synthetic sandwich culture with various top supports
The synthetic sandwich culture was constructed byoverlaying the hepatocytes cultured on the PET-Gal(bottom substratum) with three top supports (galactosy-lated, GRGDS-modified or non-modified PET TE mem-brane). As reported previously [19], hepatocytes culturedon the PET-Gal formed a 3D hepatocyte monolayerbetween day 1 and day 3 after cell seeding (prior tohepatocyte spheroid formation), which exhibited improvedcell polarity, cell–cell interactions, and enhanced differ-entiated functions compared to conventional 2D hepato-cyte monolayer on collagen substratum. We investigatedhere the effects of the 3 top supports on the morphology,cytoskeleton distribution, urea secretion and detoxificationfunctions of the sandwiched 3D hepatocyte monolayer.Top support was overlaid 24 h after seeding hepatocytesonto the PET-Gal when the hepatocytes aggregated intoisland-like clusters [19]. The sandwiched hepatocytes
continuously migrated horizontally; and the island-likeclusters merged into a monolayer in the synthetic sandwichculture with all the 3 top supports. Overlaying hepatocyteswith galactosylated or GRGDS-modified PET TE mem-brane induced within 12 h dramatic re-organization of theF-actin from cytosolic distribution into a cortical distribu-tion especially near the cell–cell contact (reminiscent of3D cell characteristic, [33]); while overlaying with non-modified PET TE membrane did not effectively induce thesimilar F-actin re-organization (Fig. 3A). After 1-weekculture, hepatocyte multi-layers were formed in thesynthetic sandwich culture with the galactosylated andnon-modified PET TE membrane top supports; while thesynthetic sandwich culture with the GRGDS-modified PETTE membrane top support could stabilize the hepatocytemonolayer morphology (Fig. 3A). Hepatocytes in thesynthetic sandwich culture with the GRGDS-modified PETTE membrane top support exhibited higher urea produc-tion and EROD cytochrome P450 1A activity than the
ARTICLE IN PRESS
Table 1
Diffusivity of FITC-dextran of various molecular weights across the GRGDS-modified PET TE membrane [PET] and gelled collagen layer [Collagen]
synthetic sandwich culture with the galactosylated or non-modified PET TE membrane top supports (Fig. 3B). TheGRGDS-modified PET TE membrane (top support)/PET-Gal (bottom substratum) synthetic sandwich culture istherefore further characterized for culturing hepatocytesover a period of 2 weeks in comparison with the conven-tional collagen sandwich culture [7,34].
3.3. Mass transfer in synthetic vs. collagen sandwich
cultures
FITC-dextrans with molecular weights of 9.5, 70, and150 kDa were used to measure the mass transfer across theGRGDS-modified PET TE membrane top support in thesynthetic sandwich culture and the gelled-collagen top layerin the collagen sandwich culture. For FITC-dextrans withall the selected sizes, an approximately two-fold increase inmass transfer was observed across the GRGDS-modifiedPET TE membrane over the gelled-collagen layer (Table 1).The results indicate that the synthetic sandwich culture canachieve better mass transfer between hepatocytes and theculture medium than the collagen sandwich culture.
3.4. Cell morphology and cell–cell interactions in synthetic
vs. collagen sandwich cultures
SEM images of hepatocytes maintained in both sand-wich cultures 48 h after sandwich assembly were analyzedfor cell morphology and cell–cell interactions. 3D hepato-cyte monolayer in the synthetic sandwich form tightlyorganized cell–cell contacts with smooth surface, mimick-ing the cell–cell interaction pattern in 3D hepatocytespheroids formed on PET-Gal (Fig. 4A). In contrast,hepatocytes in the collagen sandwich are generally moreloosely interacting with each other; and 2D hepatocytemonolayer on the collagen substratum exhibits spreadingmorphology with clearly demarcated cell-cell boundaries.We further investigated the cell-cell interactions in thesefour culture conditions by examining the expressions ofa cell–cell adhesion protein E-Cadherin (Fig. 4B). TheE-Cadherin expression level is the highest in the 3Dhepatocyte spheroids on PET-Gal, which is considered asthe ‘gold-standard’ for 3D hepatocyte culture model;followed by the 3D hepatocyte monolayer in the syntheticsandwich culture, which is significantly higher than theE-Cadherin expression level of the hepatocytes in collagensandwich culture. 3D hepatocyte monolayer in synthetic
sandwich culture therefore enables better cell–cell interac-tions than the collagen sandwich culture.
3.5. Polarity formation and biliary excretion in synthetic vs.
collagen sandwich cultures
A key feature of the sandwich culture is its ability tore-establish in vivo-like hepatocyte polarity. In the earlierstage of polarity formation, bile canaliculi are formedbetween the hepatocytes in concert with changes incytoskeleton distribution and localization of bile canaliculitransporter MRP2 into the apical domain [35,36]. Thecytoskeleton distribution in hepatocytes underwent dra-matic changes upon the top support overlaying in boththe synthetic and collagen sandwich cultures: F-actinre-organized to the cell–cell contact region from its initialrandom distribution 12 h after overlaying, which resemblesthe F-actin distribution in vivo [33] (Fig. 5); 24 h afteroverlaying, extensive and contiguous tight junctionsbetween cells have been established with majority of theMRP2 co-localized to the bile canaliculi formed bycontiguous cells, suggesting the preservation of thepolarized phenotype (Fig. 5). Our observations indicatecomparable hepatic polarity formation of the 3D hepato-cyte monolayer in the synthetic sandwich culture as thehepatocytes in the collagen sandwich culture.The establishment of cell polarity and functional activity
of bile canaliculi can be represented by the biliary excretionof hepatocytes, which is an important function of the liverto excrete metabolites and toxins from the body [7]. Weexamined the dynamic changes of hepatocyte biliaryexcretion in both sandwich cultures with a non-fluorescentsubstrate, FDA. FDA enters the cells via passive diffusion;and is hydrolyzed by intracellular esterases into fluoresceinbefore excretion by bile canaliculi transporter (MRP2) [37].The dynamics of FDA excretion in the 3D hepatocytemonolayer in the synthetic sandwich was similar to theobservation from the hepatocytes in the collagen sandwich(Fig. 6). In both sandwich cultures, there was nofluorescein concentrated in bile canaliculi sacs betweenhepatocytes after 12 h overlaying of the top support; thefluorescein secreted into bile canaliculi sacs began toappear after 24 h overlaying and fully developed between48 and 72 h (Fig. 6). The fluorescein localized in the inter-cellular sacs between hepatocytes was quantified by imageprocessing (Fig. 6 and S, see supplementary material). Theresults indicate that the 3D hepatocyte monolayer in the
ARTICLE IN PRESS
Low
mag
High
mag
Synthetic sandwich Collagen sandwich 3D spheroids on PET-Gal 2D monolayer on collagen
SyntheticSandwich
CollagenSandwich
3D spheroidson PET-Gal
2D monolayeron collagen
E-Cadherin
GAPDH
(kD)
120
35
0
20
40
60
80
100
120
140
160
180
200
E-Cadherin GAPDH
Main
Density o
f B
and
synthetic sandwich collagen sandwich PE T -Gal collagen
*
*
*
*
*
Fig. 4. Cell morphology and cell–cell interaction in synthetic vs. collagen sandwich cultures. (A) SEM images of hepatocytes maintained in synthetic and
collagen sandwich culture 48 h after top support overlaying as well as the 3D hepatocyte spheroids on PET-Gal and 2D hepatocyte monolayer on collagen
substratum at the same time point (low magnification of �� 700 at upper panel with scale bar of 40 mm and high magnification of �� 1800 at lower panel
with scale bar of 20mm). (B) Western blot and relative quantification of E-Cadherin and GAPDH expression of the hepatocytes cultured in the synthetic
sandwich culture, collagen sandwich culture, as 3D spheroids on PET-Gal and as 2D monolayer on collagen; GAPDH expression was used as loading
control. Data are mean7S.D., n ¼ 3. *, po0.05.
Y. Du et al. / Biomaterials 29 (2008) 290–301 297
synthetic sandwich exhibit similar extent of biliary excre-tion compared with the hepatocytes in the collagensandwich.
3.6. Maintenance of hepatocyte differentiated functions in
synthetic vs. collagen sandwich cultures
Key representative differentiated functions of hepato-cytes in both sandwich cultures were compared (Fig. 7).Albumin secretion, urea production and 7-ethoxyresorufin-O-deethylation cytochrome P450 1A activity of 3Dhepatocyte monolayer in the synthetic sandwich culturewere significantly higher than that of the hepatocytes in thecollagen sandwich culture over 14 days with the mostdramatic enhancement observed within the first 4–6 days.The improvement in the hepatocyte functional mainte-
nance in the synthetic sandwich culture may be due to thebetter cell–cell interaction of the 3D hepatocyte monolayerand improved mass transfer of nutrients and wastesremoval across the synthetic top support.
4. Discussion
A novel synthetic sandwich culture was developed byoverlaying a 3D hepatocyte monolayer formed on a PET-Gal with a GRGDS-modified PET TE membrane topsupport. This 3D hepatocyte monolayer has been char-acterized previously with improved cellular structure andpolarities, enhanced cell–cell interactions, better differen-tiated functions compared to the hepatocyte monolayer oncollagen-coated substratum. Due to the weak adhesiveforce obtained from the bottom galactosylated substratum
ARTICLE IN PRESS
Mrp2//CD147
before overlay
F-actin 12h
after overlayF-actin before overlay
Mrp2/CD147 12h
after overlay
synthetic
sandwich
collagensandwich
Fig. 5. Polarity formation in synthetic vs. collagen sandwich cultures: representative confocal images of F-actin staining and MRP2/CD147 double-
staining of hepatocytes in both sandwich cultures before and after top support overlaying (co-localization of the MRP2 to the bile canaliculi is marked by
the arrows).
h84h42h21 72h
3.11±0.65
4.17±0.35
7.12±0.26
6.591±1.15 12.99±2.20
11.78±1.49
13.06±3.45
11.11±2.04
Collagen
sandwich
Synthetic
sandwich
Fig. 6. Biliary excretion of hepatocytes in synthetic vs. collagen sandwich cultures: representative confocal images of dynamic changes of fluorescein
excreted by bile canaliculi transporter. The fluorescein localization in the inter-cellular sacs between hepatocytes is quantified as shown by the number at
the corner of each image (using an image processing method stated in the supplementary material).
Y. Du et al. / Biomaterials 29 (2008) 290–301298
as well as cellular contractions, this 3D hepatocytemonolayer will finally transform to 3D spheroid after 3days. The GRGDS-modified PET TE membrane topsupport may act as (1) a mechanical force applied to thehepatocytes from top, which might enhance the cell–sub-stratum interaction and acts as a balance to the cell–cellinteraction to stabilize the monolayer morphology [38];
(2) a physical boundary on top of the hepatocytemonolayer to confine the space, which prohibits themonolayer from folding into multi-layer structure inspheroids; (3) a biochemical support with the immobilizedbioactive components for morphological and functionalimprovement. As the non-modified PET TE membrane topsupport had little effect on stabilizing the hepatocyte
ARTICLE IN PRESS
20
18
16
14
12
10
8
6
4
2D2 D4 D6 D8 D10 D12 D14
Culture Period (Day)
D2 D4 D6 D8 D10 D12 D14
Culture Period (Day)
D2 D4 D6 D8 D10 D12 D14
Culture Period (Day)
Ure
a P
roduction (
µg/m
illio
n c
ells
/h)
35
30
25
20
15
10
5
0Alb
um
in S
ynth
esis
(µg
/mill
ion c
ells
/Day)
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
Rela
tive E
RO
D A
ctivity
(Norm
aliz
ed to fre
sh isola
ted h
epato
cyle
s)
N.S
N.S
N.S
Fig. 7. Hepatocyte functional maintenance in synthetic vs. collagen
sandwich cultures as represented by (A) urea production; (B) albumin
activity relative to freshly isolated hepatocytes. K, synthetic sandwich; m,
collagen sandwich. Data are mean7SD, n ¼ 6. *, po0.05; **, po0.01;
NS, not significant.
Y. Du et al. / Biomaterials 29 (2008) 290–301 299
monolayer and inducing the F-actin re-organization(Fig. 3A), we deduced that the immobilized bioactiveligand (galactose ligand or GRGDS peptide) on the topsupport play an essential role to achieve morphologicaland functional maintenance. It is known that the ligan-
d–receptor interaction between the galactose and asialo-glycoprotein receptor (ASGPR) was relatively weak [39]and hepatocytes cultured in galactosylated substratatended to form multi-cellular spheroids. RGD–integrininteractions have been shown to induce downstreamsignaling leading to the redistribution of the cytoskeleton,formation of focal adhesion complex, and enhancement ofcell–cell interaction [40,41]. Hepatocytes attached to RGD-modified substrata exhibit a spreading morphology asmonolayer, with similar phenotypes as monolayer formedon collagen [42]. When hepatocytes are exposed toGRGDS peptide or galactose ligand on the top supportand galactose ligand on the bottom substratum in thesynthetic sandwich, the synergistic interplay between thesetwo ligand–receptor interactions is expected. GRGDS-modified PET TE membrane (top support)/PET-Gal(bottom substratum) synthetic sandwich culture performedthe best in terms of morphology stabilization, functionalmaintenance and polarity formation. The GRGDS-mod-ified PET TE membrane top support might induce integrin-mediated cell–matrix interactions on the top support; thusprevent the 3D spheroid formation and stabilize the 3Dhepatocyte monolayer morphology. The galactosylatedPET TE membrane top support has a poorer stabilizationeffect on the 3D hepatocyte monolayer, which might becaused by the weaker interaction between the galactose andASGPR.We have also investigated the optimal procedure for
overlaying the top support onto the 3D hepatocytemonolayer culture. Since the hepatocytes on the bottomPET-Gal film formed island-like clusters on day 1 afterseeding and gradually merged into the 3D hepatocytemonolayer on day 2 [19], we overlaid the GRGDS-modified PET TE membrane top support on day 1 andday 2, respectively; and observed no distinct morphologicaldifferences over the 2-week culture (data not shown). Wetherefore overlaid the top support on day 1 to be consistentwith the time of overlaying collagen top layer in thecollagen sandwich. The hepatocytes within the syntheticsandwich are able to migrate laterally and interact witheach other.The synthetic sandwich exhibits several advantages over
the conventional collagen sandwich: (1) minimizing masstransfer barrier caused by the gelled-ECM top layer, whichhinders the exchange of nutrients, metabolites, xenobioticsor biochemical signals with the bulk of the medium;(2) mass transfer properties of the synthetic sandwichculture could be readily controlled by choosing commercialPET TE membranes with pore sizes ranging 0.1 mm-10 mmand densities ranging 105–108 pores/cm2 (the surfacemodification of the PET TE membrane with bioactiveligand will not affect the property of bulk material). Theimproved and controllable mass transfer achieved in thesynthetic sandwich culture would be especially importantfor hepatocyte-based xenobiotics testing [43] and hepato-cyte sandwich culture under perfusion condition in thebioreactor [44]. We expect that the improved mass transfer
ARTICLE IN PRESSY. Du et al. / Biomaterials 29 (2008) 290–301300
in the synthetic sandwich would be maintained during thefirst few days’ culture due to the sparse secretion of ECMby hepatocytes in vitro [45]. As indicated in Fig. 4A, fewECMs were observable in either the synthetic sandwich orcollagen sandwich after 3-day culture, with most of theECMs deposited around the cell surface. (3) 3D hepatocytemonolayer in synthetic sandwich exhibited enhancedcell–cell interaction and better differentiated functionsmaintained for 2 weeks compared to the hepatocytes incollagen sandwich, which may be partially due to thedifferences between the 3D hepatocyte monolayer on thegalactosylated substrata and the 2D hepatocyte monolayeron the collagen substratum before overlaying of the topsupport. The specific galactose–ASGPR interaction mayalso play an active role to induce downstream cell-signalingfor hepatocyte functional improvement; (4) more homo-geneous hepatocyte morphology was observed in thesynthetic sandwich culture than in the collagen sandwichculture, which might be due to the uniformity of thebioactive ligands exposed to the cells in the syntheticsandwich culture since it is not easy to produce uniformcollagen coating on surfaces. The uniformity of hepatocytebehaviors would be important for mechanism studies usinghepatocyte sandwich in vitro cultures, such as the studies ofhepatic transport and biliary clearance responsible for theaccumulation and excretion of a wide variety of drugs [7,8].We did not observe any significant difference in polarityformation and biliary excretion between the synthetic andcollagen sandwich indicating that the nature of thesubstrata may not be critical for hepatic polarities, as alsomentioned by other studies [5,8].
5. Conclusions
We have established an ECM-free synthetic sandwichculture by maintaining a 3D hepatocyte monolayerbetween a GRGDS-modified PET TE membrane (topsupport) and a PET-Gal (bottom substratum). The 3Dhepatocyte monolayer in the synthetic sandwich cultureexhibited similar polarity formation, improved masstransfer, enhanced cell–cell interactions and higher differ-entiated functions compared with the hepatocytes in theconventional collagen sandwich culture. This syntheticsandwich culture can potentially be used as an alternativeto the ECM-based sandwich culture for relevant hepato-cyte-based applications in liver tissue engineering and drugdiscovery.
Acknowledgments
We would like to thank Mr. Talha Arooz and Ms.Tse Kit Yan for the technical support. This work issupported in part by the Institute of Bioengineering andNanotechnology, Biomedical Research Council, Agencyfor Science, Technology and Research (A*STAR) ofSingapore (R185-001-045-305); Ministry of EducationGrant R-185-000-135-112, National Medical Research
Council Grant R-185-000-099-213 and Singapore-MITAlliance Computational and Systems Biology FlagshipProject funding to HYU. YND, RBH and FW are researchscholars of the National University of Singapore; SSN isan A*STAR graduate scholar. We also acknowledgeadditional support to YND by the NUS President’sGraduate Fellowship.
Appendix A. Supplementary material
Supplementary data associated with this article can befound in the online version at doi:10.1016/j.biomaterials.2007.09.016.
References
[1] Dunn JC, Yarmush ML, Koebe HG, Tompkins RG. Hepatocyte
function and extracellular matrix geometry: long-term culture in a
sandwich configuration. Faseb J 1989;3(2):174–7.
[2] Hughes RC, Stamatoglou SC. Adhesive interactions and the
metabolic activity of hepatocytes. J Cell Sci Suppl 1987;8:273–91.
[3] Berthiaume F, Moghe PV, Toner M, Yarmush ML. Effect of
extracellular matrix topology on cell structure, function, and
physiological responsiveness: hepatocytes cultured in a sandwich