Lineage Restriction of Human Hepatic Stem Cells toMature Fates Is Made Efficient by Tissue-Specific

Biomatrix ScaffoldsYunfang Wang,1* Cai-Bin Cui,2* Mitsuo Yamauchi,3 Patricia Miguez,3 Marsha Roach,4 Richard Malavarca,4

M. Joseph Costello,5 Vincenzo Cardinale,1,6 Eliane Wauthier,1 Claire Barbier,1 David A. Gerber,2

Domenico Alvaro,6 and Lola M. Reid1

Current protocols for differentiation of stem cells make use of multiple treatments of solublesignals and/or matrix factors and result typically in partial differentiation to mature cellswith under- or overexpression of adult tissue-specific genes. We developed a strategy for rapidand efficient differentiation of stem cells using substrata of biomatrix scaffolds, tissue-specificextracts enriched in extracellular matrix, and associated growth factors and cytokines, incombination with a serum-free, hormonally defined medium (HDM) tailored for the adultcell type of interest. Biomatrix scaffolds were prepared by a novel, four-step perfusion decel-lularization protocol using conditions designed to keep all collagen types insoluble. The scaf-folds maintained native histology, patent vasculatures, and �1% of the tissue’s proteins but>95% of its collagens, most of the tissue’s collagen-associated matrix components, and physi-ological levels of matrix-bound growth factors and cytokines. Collagens increased fromalmost undetectable levels to >15% of the scaffold’s proteins with the remainder includinglaminins, fibronectins, elastin, nidogen/entactin, proteoglycans, and matrix-bound cytokinesand growth factors in patterns that correlate with histology. Human hepatic stem cells(hHpSCs), seeded onto liver biomatrix scaffolds and in an HDM tailored for adult liver cells,lost stem cell markers and differentiated to mature, functional parenchymal cells in �1week, remaining viable and with stable mature cell phenotypes for more than 8 weeks.Conclusion: Biomatrix scaffolds can be used for biological and pharmaceutical studies oflineage-restricted stem cells, for maintenance of mature cells, and, in the future, for implant-able, vascularized engineered tissues or organs. (HEPATOLOGY 2011;53:293-305)

The ongoing revolution in stem cell research hasmade possible the identification and isolationof stem cell populations including those from

fetal and postnatal tissues.1 The potential of humanhepatic stem cells (hHpSCs) and other stem/progeni-tors for pharmaceutical research, cell-based therapies,

Abbreviations: AFP, a-fetoprotein; ALB, albumin; ASMA, a-smooth muscle actin; BSM, bladder submucosa matrix; CK, cytokeratin; CS-PG, chondroitinsulfate proteoglycan; CXCR4, chemokine (C-X-C motif ) receptor 4; CYP450, cytochrome P450; ECM, extracellular matrix; EpCAM, epithelial cell adhesionmolecule; FN, fibronectin; GAGs, glycosaminoglycans; GC, Glisson’s capsule; Gly, glycine; HDM, hormonally defined medium; hHB, human hepatoblast; hHpSC,human hepatic stem cell; HS-PG, heparan sulfate proteoglycan; Hyl, hydroxylysine; Hyp, hydroxyproline; KM, Kubota’s medium; MACS, magnetically activatedcell sorting; MSC, mesenchymal stem cell; PLA2, phospholipase A2; SIS, small intestinal submucosa; SDC, sodium deoxycholate.From the 1Departments of Cell and Molecular Physiology and Biomedical Engineering and Program in Molecular Biology and Biotechnology, UNC School of

Medicine, Chapel Hill, NC; 2Department of Surgery, UNC School of Medicine, Chapel Hill, NC; 3Department of Periodontology, UNC School of Dentistry, ChapelHill, NC; 4Gigacyte, LLC, Branford, CT; 5Department of Cell and Developmental Biology, UNC School of Medicine, Chapel Hill, NC; and 6Division ofGastroenterology, Department of Clinical Medicine, Polo Pontino, Sapienza University of Rome, Rome, Italy.Received June 28, 2010; accepted September 19, 2010.Supported by a grant from the North Carolina Biotechnology Center (NCBC), Gigacyte, LLC (Branford, CT), Vesta Therapeutics (Bethesda, MD), and from

NIH grants (AA014243, IP30-DK065933), NIDDK Grant (DK34987), and an NCI grant (CA016086). A patent application on biomatrix scaffolds has beensubmitted. V. Cardinale received salary support with a scholarship from Sapienza University of Rome for studies done at UNC. D. Alvaro is supported by MIUR(Italian Minister of University and Research) grants: PRIN #2007, prot. 2007HPT7BA-003.

*These authors contributed equally to this study.Author contributions: Y.W. and C.C.: Conception and design, collection and assembly of data, data analyses and interpretation, article writing; M.Y. and P.M.:

collection and/or assembly of data and analyses on collagen chemistry; M.R. and R.M.: collection and assembly of data on cultures on ground biomatrix scaffolds;M.J.C.: data analysis on the TEM and SEM studies; V.C. and D.A.: collection of data onbiomatrix scaffolds of biliary tree tissue (supplement); E.W. and C.B.:isolation of human hepatic stem cells and assistance with cultures; D.G.: help with some of the editing of article; L.M.R.: conception and design, assembly of data,data analyses and interpretation, article writing and editing, final approval of article and financial support.

293

and tissue engineering relies on being able to isolatethem, propagate them in culture and differentiate themto a functional mature cell fate(s).2 Current methods fordifferentiation of stem cells involve subjecting cells to amix of soluble signals and/or extracellular matrix compo-nents, and the stem cells must be treated with multiplesets of such signals over weeks of time. The adult fateachieved is typical of only partially differentiated cellswith over- or underexpression of specific adult genes.3

Here we demonstrate strategies for rapidly differentiatingstem cells using matrix scaffolds that elicit more efficientand reproducible responses.Extracellular matrix is an extraordinarily complex mix-

ture of molecules that are highly regulated, secreted by,and adjacent to cells on one or more of their surfaces, andlong understood to be critical for determining the mor-phology, growth, and differentiation of attached cells.4,5

Tissue-specific gene expression in cultured cells isimproved by culturing the cells on or embedded in matrixextracts or purified matrix components.6,7 However, indi-vidual matrix components, alone or in combination, areunable to recapitulate a tissue’s complex matrix chemistryand architecture. This is related to the fact that the matrixcomponents are in patterns associated with natural tissuezones and with histological structures such as blood ves-sels. This complexity of the tissue matrix is more readilyachieved by matrix extracts of decellularized tissue.8-10

Matrix extracts found useful for ex vivo maintenance ofcells include amniotic membrane extracts11; Matrigel, anurea extract of a murine embryonal carcinoma12; extracel-lular matrix (ECM), a detergent- or NaOH-extract ofmonolayer cell cultures13,14; and biomatrices, an extractof homogenized tissues.10,15

More recently, decellularized tissues, prepared by colla-genase digestion of a tissue16 or by delipidation followedby distilled water washes,8 have been used to mimic thematrix environment in vivo.17 Even though these proto-cols result in major losses of some matrix components, thedecellularized scaffolds from different tissues or organs,such as small intestinal submucosa (SIS), bladder submu-cosa matrix (BSM),17,18 vascular tissue,19 heart,20 air-way,21 and liver22 have been used successfully in both pre-clinical and clinical applications.23

Here we describe a strategy, focused on collagenchemistry, that is ideal for preparing substrata of tissue

extracts comprised of tissue-specific matrix componentsand factors bound to the matrix. The extracts, referredto as biomatrix scaffolds, are potent differentiation sub-strata especially when used with a serum-free, hormo-nally defined medium (HDM), tailored for either pro-genitors, such as Kubota’s medium (KM) (24), or forthe adult cell type(s) of interest, such as an HDM formature liver cells.25 In prior studies we establishedmethods for identification and isolation of hHpSCs,human hepatoblasts (hHBs), and committed progeni-tor subpopulations from livers of all donor ages andidentified conditions for their clonogenic expan-sion.26,27 In this article we assess the efficacy of bioma-trix scaffolds to differentiate hHpSCs to mature fatesand to maintain mature parenchymal cells as fullyfunctional for long periods of time.

Materials and Methods

The details of the methods are given in the Sup-porting Information online. Here we present only themethods for the preparation of the biomatrix scaffolds.

Procedures for Decellularization. After anesthesiawith ketamine-xylazine, the rat abdominal cavity wasopened and a sleevelet with a cannula was inserted intothe portal vein to perfuse the entire liver. (1) Perfusionwas done with RPMI 1640 for 10 minutes; followed by(2) delipidation with phospholipase A2 (PLA2) com-bined with a gentle detergent, sodium deoxycholate(SDC) for 30-60 minutes until the tissue becomes trans-parent, and the effusion becomes clear; (3) perfusionwith high salt washes (3.4 M NaCl) until the perfusateis negative for proteins by optical density (OD) at 280nm; (4) perfusion with nucleases (DNase, RNase) inRPMI 1640 until the perfusate is negative for nucleicacids by OD 260 (see Supporting Fig. S3); (5) Finalrinse with RPMI 1640 for 2 hours or more.The biomatrix scaffolds were quickly frozen on dry

ice and frozen sections prepared with a Cryostat, placedonto 24-well cell culture plates, sterilized by gammairradiation (5000 rads), and rehydrated in medium for30 minutes before seeding cells. The sections of bioma-trix scaffolds covered �95% of well surface in the 24-well plate. An alternative method for distributing the

Address reprint requests to: Lola M. Reid, 34 Glaxo Building, 101 Mason Farm Road, UNC School of Medicine, Chapel Hill, NC 27599. E-mail: Lola.M.Reid@gmail.com; fax: (919) 966-6112.CopyrightVC 2010 by the American Association for the Study of Liver Diseases.View this article online at wileyonlinelibrary.com.DOI 10.1002/hep.24012Potential conflict of interest: Nothing to reportAdditional Supporting Information may be found in the online version of this article

294 WANG ET AL. HEPATOLOGY, January 2011

biomatrix scaffolds onto culture dishes consisted of pul-verizing it to a powder using a freezer mill filled withliquid nitrogen. The powder acquires the consistency ofpaint at room temperature and can be painted onto anysurface, culture dish, or cloth to be used for attachingcells. Details are given in the Supporting Methods.

ResultsBiomatrix Scaffolds Are Prepared with a Novel

Four-Step Protocol. Biomatrix scaffolds were preparedusing a novel 4-step protocol: (1) gentle delipidation;(2) washes with buffers with salt concentrations at orabove 3.4 M, salt concentrations known to maintainthe collagens in an insoluble state28; (3) nuclease treat-ment to eliminate residual nucleic acids; and (4) rinseswith a basal medium to eliminate the detergent, salt,

and nuclease residues as well as to equilibrate the ma-trix components with the medium (Fig. 1A).The choices of the rinse media or the buffers for

the nucleases can be any of a number of options aslong as the salt concentration and ionic strength aresuch as to maintain the collagens and associated matrixcomponents in an insoluble state. The choice of thedelipidation method is also critical to be effective andyet should be gentle. We chose a combination of so-dium deoxycholate (SDC) and phospholipase A2(PLA2) to rapidly degrade the phosphoglyceridelocated on the cytoplasm membrane and mitochon-drial membrane into lysolecithin, a powerful surfac-tant, which can induce necrosis and cytolysis. The re-active formula is shown in the Supporting Fig. S1.We avoided prolonged exposure of the scaffolds to

the enzymes from the disrupted cells during delipidation

Fig. 1. Rat liver biomatrix scaffold preparation. (A) Four-step decellularization process comprised of perfusion wash, delipidation with PLA2 andSDC, high salt washes, and nuclease treatment for nucleic acid removal. (B-D) Four stages in the preparation of rat biomatrix scaffold. (B) After perfu-sion wash with basal medium for 10 minutes the liver becomes pale; (C) during delipidation, the liver becomes partially transparent under GC; (D)final intact scaffold looks transparent at 40 minutes of perfusion; (E) biomatrix scaffold shown at low magnification. (E1) Visualization of scaffold per-fused with rhodamine-labeled dextran particles demonstrates progressive flow from large vessels to the fine blood vessel branches along the chan-nels without leakage, indicating patent vasculature in scaffolds (see also the Supporting Video). Corresponding hematoxylin and eosin (H&E) stainingof biomatrix scaffold in different stages demonstrated that the histological structures such as blood vessels and lace-like matrix enveloping the paren-chyma are preserved, whereas cells are removed. The normal rat hepatic portal triad structure consisting of the portal vein (PV), hepatic artery (HA),and bile duct (BD) (B1); the matrix fibers becoming apparent as the cells are gradually removed during the decellularization process (C1); decellular-ized portal triad region, compare (B1) with that in (D1); (D1-3) shows that all of the cells removed from the matrix scaffold but mesh structures are pre-served such as the blood vessels, GC, and the lace-like matrix that surrounds muralia of parenchymal cells.

HEPATOLOGY, Vol. 53, No. 1, 2011 WANG ET AL. 295

and the high salt washes because they can greatlydecrease the content of elastin and the content of glycos-aminoglycans (GAGs) such as heparan sulfates (HS),chondroitin sulfates (CS), dermatan sulfates (DS), andheparins (HP), sites at which cytokines and growth fac-tors bind.29 We used soybean trypsin inhibitor and care-ful control of the pH (7.5-8.0) and time (30-60minutes) to limit the activity of the proteases derivedfrom disrupted cells.We perfused the whole tissue through relevant vas-

culature (e.g., portal vein in the liver), enabling us torapidly isolate (within a few hours) a biomatrix scaf-fold with minimal loss of matrix components. The ra-pidity of the isolation is due to the initial step withdetergent that delipidates the tissue within �30-60minutes (not hours or days as in protocols used byothers, see Supporting Table 5). The resulting bioma-trix scaffolds are translucent or white (Fig. 1D). More-over, using this perfusion method we maintained theprimary vasculature channels, portal and hepatic vein,and most of the vascular branches in the liver, whichincreased the decellularization efficiency (Fig. 1E). Flu-orescent rhodamine-labeled dextran particles perfusedthrough the biomatrix scaffolds remained within theremnants of the vasculature, demonstrating that theyare patent (Fig. 1E1). There is a progressive flow ofthe dye from large vessels to the fine blood vesselbranches along the channels without leakage (demon-strated even more dramatically in the SupportingVideo). This fact will be helpful in the future in revas-cularization of scaffolds as a means of preparing engi-neered tissues for either three-dimensional culture and/or for implantation ex vivo.When sectioned, scaffolds retain the histological

structure of the original tissue, including the recogniz-able remnants of major histological entities such asblood vessels, bile ducts, and Glisson’s capsule (GC).Compare Fig. 1B1 and 1D1, in which a section ofliver tissue is contrasted with that of a biomatrix scaf-fold. The matrix remnants of the muralia of parenchy-mal cells consisted of a lace-like network (Fig. 1D1-1D3).Collagen, Collagen-Associated Proteins, and

Bound Cytokines Are Maintained in the BiomatrixScaffolds. The amount of collagens in biomatrix scaf-folds was evaluated by amino acid analysis by methodsused previously.30 Because hydroxyproline (Hyp) isunique to collagens and collagenous proteins, the col-lagen composition relative to total protein wasexpressed as residues of Hyp per 1,000 amino acids.The results demonstrated that collagen contentincreased from almost undetectable levels, i.e., less

than 0.2 residues of Hyp/1,000 in liver, to �13 resi-dues of Hyp/1,000 in biomatrix scaffolds. This indi-cates that delipidation and the high salt washes,described above, did not remove collagens, leavingalmost all of the collagens in the biomatrix scaffolds.Detection of significant levels of hydroxylysine (Hyl),another collagen-associated amino acid, and higher lev-els of glycine (Gly) in biomatrix scaffold supports ourconclusion that collagen is markedly enriched in bio-matrix scaffolds (Fig. 3A; Supporting Fig. S2, Support-ing Table 1).Through immunohistochemical and ultrastructural

studies, we were able to identify in the scaffolds allknown collagen types found in liver in situ includingfibrillar collagens (collagen types I, III, and V, 10-30 nmin diameter for fibrils and 500-3,000 nm for assembledfibers) and beaded filaments (possibly type VI). Thosefibers and filaments are present in the subcapsular con-nective tissue layer lying beneath the mesothelial layer.Although typical structures of basement membraneswere not found along the sinusoids from portal triads tocentral veins, we found collagen type IV and somebound, small fibrils form net-like, porous 3D lattices,serving as scaffolding for the parenchymal cells (Fig. 2).Collagen type I bundles can be viewed as the principalstructure of the scaffolds to which other collagen types,glycoproteins, and proteoglycans are attached. In theSpace of Disse, we found small bundles of collagen typeI and fibers of collagen types III and VI as well as sometype V, which is more abundant near portal triads andcentral veins. Representative immunohistochemistrydata are presented in Fig. 3B, and a summary of matrixcomponents and their locations in normal liver tissueversus those in the biomatrix scaffolds are listed in Fig.4D. Early studies in the development of the protocolsfor biomatrix scaffold preparation indicated that thebulk of the cytoskeletal components are lost in thewashes (data not shown). Still, we assessed the scaffoldsby immunohistochemistry for residues of cytoskeletalcomponents and found no evidence for tubulin, desmin,or actin, trace amounts of cytokeratins 18 and 19, andlow levels of vimentin scattered throughout thescaffolds.The matrix associated with the bile ducts and por-

tions of the hepatic vascular systems (arterial and ve-nous vessels) consists of typical basement membranestructures and so is quite distinctive from the thinlayers of the matrix associated with the vascular struc-tures found in the sinusoids. Laminin, entactin/nido-gen, perlecan, and collagen type IV are found in theportal triad, whereas only perlecan and some collagentype IV are found in the Space of Disse (data not

296 WANG ET AL. HEPATOLOGY, January 2011

shown). Enormous amounts of hydrophobic, wavyelastin are present; it crosslinks together and formssheets and fibers restricted primarily to the subcapsularconnective tissue, portal regions, and arterial walls.

Fibronectins are ubiquitous and prevalent throughoutthe scaffolds and are especially abundant in the Spaceof Disse, where they form either fine filaments orgranular deposits (Figs. 2, 3B).

Fig. 2. Transmission and scanning electron microscopy images of rat liver biomatrix scaffolds. (A) Low magnification of a blood vessel (BV)probably the portal vein, based on the narrow wall (W) thickness compared to the large diameter of the vessel. The undulations or scalloping ofthe vessel (sometimes linked to the internal elastic lamina of an artery) is here probably a partial collapse of the vessel wall. Collagen Type Ifibers (large arrowhead) are numerous and contains cross-sections of individual fibers that do not take up heavy metal stains (white dots, smallarrowheads). (A1) Higher magnification of a vessel wall shows basement membrane (large arrow), amorphous elastin (*), and associated elasticfibers, a rare membrane vesicle remnant (small arrowhead), a collagen Type I banded fiber (arrowhead), and small fibrils (small arrows). Thesmall fibrils are probably fibrillin (Type VI collagen) that associates closely and helps organize Type I collagen. (B) High magnification of Type Icollagen with 64 nm banding pattern (arrows). (C) Low magnification of a vessel with a thin wall (BV) and the wall of a larger vessel (W). (D) Athigher magnification, the large vessel wall (W) is scalloped, consistent with hepatic artery of a portal triad, see (A). Beneath the wall are numer-ous Type I collagen bundles (large arrow) linked by long branching thin, reticular (Type III) collagen fibrils (small arrows). (E) A large bundle ofType I collagen has characteristic parallel fibers (large arrow) associated with a variety of smaller fibers (arrow) and nodular or beaded fibers(arrowhead). (F) 3D meshwork of large/small fibers interlinked in a plane that forms a boundary such as to a liver sinusoid. (F1) Higher magnifi-cation of the meshwork showing a variety of fibers (arrows): Type III collagen (larger diameter straight), elastic fibers, or Type VI collagen.

HEPATOLOGY, Vol. 53, No. 1, 2011 WANG ET AL. 297

Immunohistochemistry indicates that known proteo-glycans in the tissue are preserved in the biomatrixscaffolds (Figs. 3B, 4D). Among heterogeneous proteo-glycans identified, syndecan was found intercalated andcontinuously along sinusoids, and perlecan is morepunctuate in the Space of Disse. The forms of HS-PGs and CS-PGs are present throughout the remnantsof the sinusoids in the biomatrix scaffolds and in pat-terns correlating with known zonation of liver tissue.Proteoglycans and other matrix components are im-

portant reservoirs for cytokines and growth factors thatbind tightly to their GAGs.31 Most growth factors andhormones are found in biomatrix scaffolds at physio-logical concentrations. In Table 1 the data are givenfrom the lysates of rat livers versus rat liver biomatrixscaffolds, and in Supporting Table 2 parallel data arefrom human bile duct tissue versus bile duct biomatrixscaffolds. Interestingly, there were a few examples (e.g.,bFGF) that were strongly enriched in liver biomatrixscaffolds over that found in liver lysates. The growthfactors and cytokines bound are distinct qualitatively

and quantitatively between the scaffolds of liver versusbile duct tissue, implicating either tissue-specificity orspecies-specificity, a conclusion that awaits furtheranalyses on multiple tissues. Alternatively, it may bedue, in part, to the fact that bile duct scaffolds wereprepared, from necessity, by shaking the tissue in buf-fers on a rocker and not by perfusion throughvasculature.Chemistry of the Biomatrix Scaffolds Correlates

with Histology. A significant feature of this new pro-tocol is the retention of the matrix chemistry in pat-terns correlating with hepatic acinar zones 1-3 fromportal triad to central vein and with histological enti-ties such as vascular channels and GC, as shown inFig. 4A-C. The matrix chemistry periportally in zone1 is similar to that found in fetal livers and consists oftype I and type III collagens, laminin, and forms ofCS-PGs. It transitions to a different matrix chemistryin the mid-acinar (zone 2) and pericentral zones (zone3), ending with a very stable matrix with high levels oftype IV collagen and HP-PGs.32

Fig. 3. Chemical analyses of collagens and expression of ECM components in biomatrix scaffolds. (A) The content of three amino acids, allfound in collagens: Hyp, Hyl, and Gly. The numbers represent the residues of each amino acid / 1,000 amino acids. The data indicate the dra-matic increase in the collagen content in the decellularization process going from <0.2% in liver to more than 15% in the biomatrix scaffolds;The Hyp content in the whole liver was too low to be reasonably quantified (less than 0.1 nmol in �500 nmol of total amino acids, i.e., <0.2res/1,000). Thus, in Supporting Table 1 the amount is described as 0. (B) Immunohistochemical staining of matrix molecules in biomatrix scaf-folds shows distribution in liver biomatrix scaffolds of laminin (LAM), heparan sulfate (HS), collagen type III (COL3), and fibronectin (FN) and typ-ical basement membrane proteins in association with remnants of blood vessels. At higher magnification one can observe the primarycomponents of basement membrane, including Type IV collagen (COL4), entactin (Ent; also called nidogen), LAM, and perlecan (Per), a form ofHS-PG in the portion of the scaffolds near the portal triads. See also the details of the collagen chemistry given in Supporting Table 1 and Sup-porting Fig. S2.

298 WANG ET AL. HEPATOLOGY, January 2011

Myriad proteins (e.g., growth factors and hormones,coagulation proteins, various enzymes) are known tobind to the matrix and to be held stably by way ofbinding to the discrete and specific sulfation patternsin the GAGs or to other matrix components.29 Thus,the matrix chemistry transitions from its start point inthe stem cell niche having labile matrix chemistry asso-ciated with high turnover and minimal sulfation to sta-ble matrix chemistries and having increasing amountsof sulfation with progression towards the pericentralzone. We hypothesize that maintenance of the naturalarchitecture and matrix chemistry correlating with his-tology will facilitate recellularization in tissue engineer-ing processes by guiding cells to specific sites on thebiomatrix scaffolds and/or providing the proper mix ofsignals to drive differentiation into mature cells.

Biomatrix Scaffold Can Be Prepared from Differ-ent Tissues and Species. The biomatrix scaffolds canbe prepared from any tissue, normal or diseased, and

from any species. In the supplement we show bioma-trix scaffolds from human pancreas, biliary tree, andduodenum and from rat pancreas (Supporting Figs.S6-S9). Figures 5-7 and Supporting Fig. S5 showeffects of bovine or rat liver biomatrix scaffolds on he-patic cells. In addition, biomatrix scaffolds have beenprepared from human abdominal aorta, iliac vein, andfrom rat and pig intestine (data not shown). Histologi-cal, ultrastructural, and immunohistochemical studieson the biomatrix scaffolds suggested a marked tissuespecificity, but not species specificity, in their structure,chemical composition, and functions (data notshown).

Biomatrix Scaffolds Induced and/or MaintainedDifferentiation of Cells. Plating hHpSCs onto disheswith sections of liver biomatrix scaffolds and in HDMtailored for adult liver cells resulted in essentially100% of the viable cells attached within a few hoursonto biomatrix scaffolds, whether intact or after

Fig. 4. Pattern of ECM components from portal triad to central vein in biomatrix scaffolds. Histological comparison from portal triad (zone 1)to central vein (zone 3) of normal liver (A) and liver biomatrix scaffold (B); both are hematoxylin/eosin stained sections. (C) The model illustrat-ing a stem cell and maturational lineage system in the liver with representative matrix components shown that they form patterns associatedwith the liver zonation. The components are listed in order of abundance from the findings of immunohistochemistry. The known lineage stageswithin human livers begin periportally in zone 1 (around portal triads) and progress in maturation ending with apoptotic cells in zone 3. Theknown matrix chemistry identified in the liver’s stem cell niche is comprised of hyaluronans, type III collagen, a form of laminin that binds toa6b4 integrin, and a weakly sulfated form of CS-PG.49,50 Just outside the stem cell niche are found Type IV collagen, normally sulfated CS-PGsand HS-PGs, and forms of laminin binding to ab1 integrin. HP-PGs have been documented to be located uniquely pericentrally.51,52 (D) The sur-vey of matrix components and their location in liver versus those in biomatrix scaffolds, data summarized from immunohistochemistry findings(N/D ¼ not tested. *Found by others to be exclusively near central veins). Most components of the cytoskeleton are lost during the washes(data not shown), residues of some, but not all, cytoskeletal proteins are present. The scaffolds are devoid of tubulin, desmin, and actin (phal-loidin assays). However, there are trace amounts of cytokeratins scattered randomly in the scaffolds; trace amounts of a-smooth muscle actinaround remnants of blood vessels at the portal triads; and low levels of vimentin throughout.

HEPATOLOGY, Vol. 53, No. 1, 2011 WANG ET AL. 299

cryogenic pulverization. The colonies of cells that ini-tially formed on the sections of scaffolds retained someof their stem cell phenotype, as the cells in the centerof the colonies were able to resist staining with dyes(Supporting Fig. S4) and expressed classic hepatic pro-genitor markers, such as chemokine (C-X-C motif ) re-ceptor 4 (CXCR4) and epithelial cell adhesion mole-cule (EpCAM) (Fig. 5E). They divided once or twiceand then transitioned into cell cycle arrest and into3D cord-like morphologies typical for cultures ofmature parenchymal cells (Figs. 5, 6 for stem cells dif-ferentiation; compare with Fig. 7 and Supporting Fig.S5 for maintenance of mature hepatocytes). TheHDM used did not require all the usual cytokines orgrowth factors because these are present bound to thebiomatrix scaffolds. The transition to growth arrestcorrelated with staining throughout the colonies withviability dyes (Supporting Fig. S4), with loss of expres-sion of EpCAM and CXCR4 (Fig. 5E) and with a

steady increase in the expression of adult-specific hepa-tocytic and cholangiocytic genes such as urea and cyto-chrome P450 3A4 (Fig. 5F).Normal adult rat and human hepatocytes were

plated onto type I collagen versus on biomatrix scaf-folds from rat or bovine livers and into HDM foradult cells. The adult parenchymal cells were able toattach to scaffolds within a few minutes (even in se-rum-free medium) versus within hours on type I colla-gen; remained in growth arrest from the point ofattachment; and remained viable and fully functionalfor more than 8 weeks on scaffolds, versus only about�2 weeks on type I collagen (Fig. 7 and SupportingFig. S5). The levels of functions of the mature livercells on biomatrix scaffolds for weeks proved to be thesame or similar to the findings of others of freshly iso-lated, adult hepatocytes.33 The dramatic distinctionsare that the cultures on type I collagen deterioratedrapidly after 2 weeks, whereas those on biomatrix

Table 1. Analyses of Growth Factors Bound to Liver Biomatrix Scaffolds

Name Cytokine Full Name Rat Livers Rat Biomatrix Scaffolds Percent

bFGF Basic fibroblast growth factor 100.06 394.14 394

EGF Epidermal growth factor 74.81 76.02 102

EGF R Epidermal growth factor receptor 92.81 81.64 88

FGF-4 Fibroblast growth factor-4 15.06 13.21 88

FGF-6 Fibroblast growth factor-6 4.81 3.77 78

FGF-7 Fibroblast growth factor-7 10.06 6.32 63

GCSF Granulocyte-colony stimulating factor 348.06 338.20 97

GDNF Glial-derived neurotrophic factor 81.31 43.59 54

GM-CSF Granulocyte macrophage-colony stimulating factor 133.56 105.38 79

HB-EGF Heparin-binding epidermal growth factor 44.56 38.23 86

IGFBP-1 Insulin-like growth factor binding proteins 1 67.81 70.40 104

IGFBP-3 Insulin-like growth factor binding proteins 3 140.81 201.90 143

IGFBP-4 Insulin-like growth factor binding proteins 4 83.56 58.92 71

IGFBP-6 Insulin-like growth factor binding proteins 6 91.81 72.19 79

IGF-I Insulin-like growth factor-I 1.56 1.98 127

IGF-I SR Insulin-like growth factor-I 7.31 3.51 48

IGF-II Insulin-like growth factor-2 3749.06 3482.52 93

M-CSF Macrophage-colony stimulating factor 170.31 134.68 79

M-CSF R Macrophage colony stimulating factor receptor 70.56 50.47 72

NT-3 Neurotrophin-3 25.56 5.03 20

NT-4 Neurotrophin-4 55.06 43.59 79

PDGF R a Platelet- derived growth factor receptor alpha 10.56 21.11 200

PDGF R b Platelet- derived growth factor receptor beta 113.81 85.46 75

PDGF-AA Platelet- derived growth factor AA 62.06 106.40 171

PDGF-AB Platelet- derived growth factor AB 19.31 19.34 100

PDGF-BB Platelet- derived growth factor BB 9.56 14.23 149

PlGF Phosphatidylinositol glycan anchor biosynthesis, class F 4.81 8.36 174

SCF Stromal cell-derived factor-1 2.06 42.56 2064

SCF R Stromal cell-derived factor receptor 17.06 17.80 104

TGF-a Transforming growth factor alpha 21.31 21.63 102

TGF-b Transforming growth factor-beta 330.31 342.77 104

TGF-b 2 transforming growth factor-beta 2 134.06 152.34 114

TGF-b 3 Transforming growth factor-beta 3 1.06 0.18 17

VEGF Vascular endothelial growth factor 70.56 94.14 133

VEGF R2 Vascular endothelial growth factor receptor 2 13.56 11.93 88

VEGF R3 Vascular endothelial growth factor receptor 3 459.56 46.91 10

300 WANG ET AL. HEPATOLOGY, January 2011

scaffolds remained stable morphologically and func-tionally for as long as the cultures were maintained(Fig. 7 and Supporting Fig. S5).

Discussion

Biomatrix scaffolds contain most of the tissue’sextracellular matrix components and matrix-boundcytokines and growth factors, providing a compositeset of chemical signals that can be used as an insolu-ble, stable scaffolding with an extraordinary ability toinduce hHpSCs to adult liver fates as well as maintainadult cells fully differentiated for weeks. In comparingthe extant types of matrix extracts from decellularizedtissues with that of biomatrix scaffolds (Supporting

Table 5), it is clear that physical, enzymatic, andchemical treatments have substantial effects on thecomposition, mechanical behavior, and host responsesto biological scaffolds derived from the decellulariza-tion of native tissues and organs and, accordingly, haveimportant implications for their in vitro and in vivoapplications. All other existing methods for preparationof substrata or scaffolds remove a large portion of ma-trix components either through use of matrix-degrad-ing enzymes16 or using buffers that dissolve portionsof the matrix.9 Physical methods (e.g., snap freezingand agitation) can work to prepare matrix extractsfrom tissues with a layered structure such as dermis(e.g., SIS, BSM)34 but are not useful for organs withcomplex tissue structures such as liver. By contrast, the

Fig. 5. Characterization of hHpSCs on liver biomatrix scaffolds versus on type I collagen. Phase-contrast images (A-D) show the morphologicchanges of hHpSC colonies derived from the same liver and cultured in serum-free Kubota’s medium and on tissue culture plastic (A), one ofthe conditions for self-replication, versus in the differentiation conditions of the serum-free differentiation medium for liver, and on Type I collagen(B) versus on bovine liver biomatrix scaffolds (C,C1,D). The cultures transitioned to cells by days 7-12 with increased cytoplasmic/nuclear ratioand marked glycogen expression (C) and then to ones with classic polygonal hepatocyte morphology interspersed by clear bile canaliculi (C1), aculture morphology that persisted thereafter, as indicated in the representative culture at day 24 (D). Reverse-transcription polymerase chainreaction (RT-PCR) assays show gene expression changes of hHpSCs under self-replication conditions on culture plastic versus on rat liver bioma-trix scaffolds on day 7 (E). We compared expression of hHpSC markers, including CXCR4 and EpCAM; early hepatocytic genes including CK19,HNF 6, FOXA2, AFP, and low levels of albumin; mature hepatocytic markers including high levels of albumin, transferrin, CYP3A4, tyrosine amino-transferase (TAT), and glucose-6-phoshatase (G6PC), and cholangiocytic genes, including CFTR, gamma glutamyl transpeptidase (GGT1), anionexchange 2 (AE2), and apical sodium-dependent bile acid transporter (ASBT). Biochemical assays measuring urea synthesized in cultures ontype I collagen versus on rat liver biomatrix scaffolds and P450 3A4 activity (F) in cultures on type I collagen versus on biomatrix scaffolds pre-pared from either rat or bovine livers at day 12.

HEPATOLOGY, Vol. 53, No. 1, 2011 WANG ET AL. 301

method for biomatrix scaffolds resulted in loss of mostcellular proteins but preserved essentially all of the col-lagens and collagen-associated components includingthe matrix-bound cytokines and growth factors.Extracellular matrix is embedded in a mosaic lipid

bilayer, which in even the simplest organism is a com-plex, heterogeneous, and dynamic environment. Thedelipidation method is a critical facet of the protocol.The commonly used methods for decellularization oftissues involve ionic detergents such as SDC and so-dium dodecyl sulfate (SDS). SDC is relatively milderthan SDS, tends to cause less disruption to the nativetissue architecture, and is less effective at solubilizingboth cytoplasmic and nuclear cellular membranes.35

There are no reports of tissue decellularization usingSDC alone. Many studies have made use of a harshnonionic detergent (e.g., Triton X-100)36 or zwitter-

ionic detergents (e.g., 3-(3-cholamidopropyl) dimethy-lammonio]-1-propanesulfonate, CHAPS).37 By con-trast, our method of using a combination of SDC andPLA2 delipidated the tissue rapidly and gently.At least 29 types of collagens (I-XXIX) have been

identified with functional roles in cell adhesion, differ-entiation, growth, tissue development, and structuralintegrity.38,39 The major structural component in thematrix, collagens, are known to remain insoluble inhigh salt concentrations and at neutral pH,28,40-42 afinding that is the basis of our strategy in preparationof biomatrix scaffolds. The strategy has added advan-tages that collagens enable preservation of matrix com-ponents bound to them, such as laminins and fibro-nectins (FNs), small leucine-rich proteoglycans (PGs),and GAGs that in turn preserve cytokines, growth fac-tors, or cell surface receptors bound to them.

Fig. 6. Immunofluorescence staining of cells lineage restricted from hHpSCs on biomatrix scaffolds. (A) Stained with hepatic specific marker:albumin (Alb, red) and hHpSC cell surface marker: EpCAM (green). Note that cells plated on biomatrix scaffold do not express EpCAM. Scalebar ¼ 200 lm. (B) Stained with early hepatic marker a-fetoprotein (AFP, red) and with an antibody to human CK19 (green) that at this level ofexpression is indicative of mature cholangiocytes. This antibody to CK19 assays is human-specific and did not stain the residue of rat CK19 inthe scaffolds not seeded with cells (data not shown). The AFP expression is low but still evident at day 7. Scale bar ¼ 200 lm. (C) Stainedwith Alb (red) and hepatic stellate cell marker, a-smooth muscle actin (ASMA, green). The expression of albumin and ASMA is a strong indica-tion that both maturing hepatocytes and stellate cells are present. Scale bar ¼ 100 lm. (D) Stained with functional hepatic protein CYP450(red) and cholangiocytes-specific marker, secretin receptor (SR, green) showing that the maturing hepatocytes and cholangiocytes are functionaland express classic markers for these two cell types. Scale bar ¼ 200 lm.

302 WANG ET AL. HEPATOLOGY, January 2011

Biomatrix scaffolds are unique in their profound abil-ity to induce rapid and consistent differentiation ofstem/progenitor cells, such as hHpSCs, to adult fatesand to maintain those lineage-restricted cells, or tomaintain adult cells plated onto the scaffolds, as viableand fully functional cells for many weeks (>8 weeks).Differentiation of stem cells, such as embryonic stem

(ES) cells, induced pluripotent stem (iPS) cells, or vary-ing forms of mesenchymal stem cells (MSCs) into fullymature liver cell types requires multiple sets of signals(soluble and matrix) presented in stages, with inductionby one set requiring priming to respond to a differentset, and takes many weeks, up to 6 weeks of culture, togenerate cells having the adult liver fate.43 Moreover,lineage restriction of MSCs to liver fates gives inconsis-tent results with adult cells having mixed hepatocyte and

MSC phenotypes.3,44,45 The hepatocyte-like cells fromany of these precursors express some, but never all, ofthe major liver-specific genes, with variability in whichgenes are observed, and with the protein levels for he-patic genes being usually low46 or high for one hepaticgene and negligible for others.3,47,48 For reasonsunknown, the results are different from preparation topreparation. In contrast, differentiation of hHpSCs onbiomatrix scaffolds resulted in essentially all cellsexpressing a classic adult phenotype with urea, albumin,and CYP450 activities at near normal levels within 1 to2 weeks in culture and with stability of that phenotypefor many weeks. We assume that the biomatrix scaffoldscan greatly facilitate differentiation of other stem cellpopulations, such as ES, iPS, and MSCs to an adult liverphenotype, a hypothesis now being tested.

Fig. 7. Stability of fully functional, mature human hepatocytes on biomatrix scaffolds. Adult human hepatocytes plated in the differentiationmedium and onto type I collagen (A,B) versus on bovine liver biomatrix scaffolds (C) that were cryogenically pulverized, dispersed in medium,and allowed to sediment onto the plates. Cells on Type I collagen are fully viable and at their peak of differentiation from 7-12 days (A shownat 7 days); they begin to deteriorate after �2 weeks, and by 20 days (B) they are dead, dying, and nonfunctional. By contrast, those platedonto liver biomatrix scaffolds (C) are functional for at least 8 weeks (we have not assessed them at longer times yet); here shown after 21 daysin culture on pulverized liver biomatrix scaffolds. P450 3A4 assays on cultures of two separate preparations of cryopreserved adult human livercells plated onto biomatrix scaffolds versus on type I collagen and assayed on day 12 (D). The sample ZHep-007 is representative of cryopre-served samples with good attachment after thawing; the sample ZL-013 is representative of those lots that have poor or no attachment afterthawing. Thus, even these poorer quality samples are able to attach to biomatrix scaffolds and remain viable long term. In both samples assayedthe levels of P450s are higher when cultured on liver biomatrix scaffolds. With time on the biomatrix scaffolds, the lots of poorer quality cryopre-served cells will improve; studies are ongoing to assess the extent of this improvement and the time required for it to occur (data not shown).

HEPATOLOGY, Vol. 53, No. 1, 2011 WANG ET AL. 303

The ability to differentiate stem cells on biomatrixscaffolds and in an HDM to achieve mature and func-tional cells and tissues offers considerable opportunitiesfor academic, industrial, and clinical programs ena-bling the use of well-differentiated cell types for everytype of analytical study, and, excitingly, enabling animproved way to generate implantable, revascularizedtissues or organs that might be used for basic researchand clinical programs.Preliminary studies, still ongoing, suggest that scaf-

folds are both efficient inducers of differentiation andalso can dictate fate. If true, it implicates the excitingpotential of being able to define variables dictating fateand deriving them from the microenvironment versusthose entirely within the stem cells.

Acknowledgment: We thank Dr. V. Madden forTEM and SEM processing; Dr. Y. Rong for the rat he-patocyte preparations, and Lucendia English for glass-ware washing and lab management.

References1. Lanza R, Gearhart J, Hogan B, Melton D, Pedersen R, Thomson J,

et al. Handbook of Stem Cells. New York: Elsevier Academic Press;2004.

2. Vacanti JP, Langer R. Tissue engineering: the design and fabrication ofliving replacement devices for surgical reconstruction and transplanta-tion. Lancet 1999;354(Suppl 1):SI32-S34.

3. Snykers S, De Kock J, Rogiers V, Vanhaecke T. In vitro differentiationof embryonic and adult stem cells into hepatocytes: state of the art.Stem Cells 2009;27:577-605.

4. Rhodes JM, Simons M. The extracellular matrix and blood vessel for-mation: not just a scaffold. J Cell Mol Med 2007;11:176-205.

5. Brill S, Holst PA, Zvibel I, Fiorino AS, Sigal SH, Somasundaran U,et al. Extracellular matrix regulation of growth and gene expression inliver cell lineages and hepatomas. In: Arias IM, et al., eds. The Liver:Biology and Pathobiology. Vol. 44, 3rd ed. New York: Raven Press;1994:869-897.

6. Wang Y, Yao H-l, Cui C-B, Wauthier E, Barbier C, Costello MJ, et al.Paracrine signals from mesenchymal cell populations govern expansionand differentiation of human hepatic stem cells to adult liver fates. HE-

PATOLOGY 2010;52:1443-1454.

7. Macdonald JM, Xu A, Kubota H, Lecluyse E, Hamilton G, Liu H,et al. Liver cell culture and lineage biology. In: Atala A, Lanza RP, eds.Methods of Tissue Engineering. London: Academic Press; 2002:151-202.

8. Chun SY, Lim GJ, Kwon TG, Kwak EK, Kim BW, Atala A, et al.Identification and characterization of bioactive factors in bladder sub-mucosa matrix. Biomaterials 2007;28:4251-4256.

9. Huber JE, Spievack A, Simmons-Byrd A, Ringel RL, Badylak S. Extrac-ellular matrix as a scaffold for laryngeal reconstruction. Ann Otol Rhi-nol Laryngol 2003;112:428-433.

10. Rojkind M, Gatmaitan Z, Mackensen S, Giambrone MA, Ponce P,Reid LM. Connective tissue biomatrix: its isolation and utilization forlong-term cultures of normal rat hepatocytes. J Cell Biol 1980;87:255-263.

11. Liotta LA, Lee CW, Morakis DJ. New method for preparing largesurfaces of intact human basement membrane for tumor invasion stud-ies. Cancer Lett 1980;11:141-152.

12. Kleinman HK, McGarvey ML, Hassell JR, Martin GR. Formation of asupramolecular complex is involved in the reconstitution of basementmembrane components. Biochemistry 1983;22:4969-4974.

13. Vlodavsky I, Levi A, Lax I, Fuks Z, Schlessinger J. Induction of cellattachment and morphological differentiation in a pheochromocytomacell line and embryonal sensory cells by the extracellular matrix. DevBiol 1982;93:285-300.

14. Gospodarowicz D, Delgado D, Vlodavsky I. Permissive effect of theextracellular matrix on cell proliferation in vitro. Proc Natl Acad Sci US A 1980;77:4094-4098.

15. Doerr R, Zvibel I, Chiuten D, D’Olimpio J, Reid LM. Clonal growthof tumors on tissue-specific biomatrices and correlation with organ sitespecificity of metastases. Cancer Res 1989;49:384-392.

16. Badylak SF. The extracellular matrix as a scaffold for tissue reconstruc-tion. Semin Cell Dev Biol 2002;13:377-383.

17. Gilbert TW, Wognum S, Joyce EM, Freytes DO, Sacks MS, BadylakSF. Collagen fiber alignment and biaxial mechanical behavior of por-cine urinary bladder-derived extracellular matrix. Biomaterials 2008;16.

18. Lee M, Chang PC, Dunn JC. Evaluation of small intestinal submucosa asscaffolds for intestinal tissue engineering. J Surg Res 2008;147:168-171.

19. Martin ND, Schaner PJ, Tukenko TN, Shapiro IM, Dimatteo CA,Williams TK, et al. In vivo behavior of decellularized vein allografts. JSurg Res 2005;129:17-23.

20. Ott HC, Matthiesen TS, Goh SK, Black LD, Kren SM, Netoff TI,et al. Perfusion-decellularize matrix using nature’s platform to engineera bioartificial heart. Nat Med 2008;14:213-221.

21. Macchianni P, Jungebluth P, Go T, Asnaghi MA, Rees LE, Cogan TA,et al. Clinical transplantation of a tissue-engineered airway. Lancet2008;372:2023-2030.

22. Uygun B, Soto-Gutierrez A, Yagi H, Izamis M, Guzzardi MA, Shul-man C, et al. Organ re-engineering through development of a trans-plantable recellularized liver graft using decellularized liver matrix. NatMed 2010;16:814-820.

23. Franklin ME Jr, Trevino JM, Portillo G, Vela I, Glass JL, Gonzalez JJ.The use of porcine small intestinal submucosa as a prosthetic materialfor laparoscopic hernia repair in infected and potentially contaminatedfields: long-term follow-up. Surg Endosc 2008;22:1941-1946.

24. Kubota H, Reid LM. Clonogenic hepatoblasts, common precursors forhepatocytic and biliary lineages, are lacking classical major histocompat-ibility complex class I antigens. Proc Natl Acad Sci U S A 2000;97:12132-12137.

25. Wauthier E, McClelland R, Turner W, Schmelzer E, Kubota H,Zhang L, et al. Hepatic stem cells and hepatoblasts: identification,isolation and ex vivo maintenance. Methods Cell Biol 2008;86:137-225.

26. Schmelzer E, Zhang L, Bruce A, Wauthier E, Ludlow J, Yao H, et al.Human hepatic stem cells from fetal and postnatal donors. J Exp Med2007;204:1973-1987.

27. Zhang L, Theise N, Chua M, Reid LM. Human hepatic stem cells andhepatoblasts: symmetry between liver development and liver regenera-tion. HEPATOLOGY 2008;48:1598-1607.

28. Miller EJ, Rhodes RK. Preparation and characterization of the differenttypes of collagen. Methods Enzymol 1982;82(Part A):33-64.

29. Yayon A, Klagsbrun M, Esko JD, Leder P, Ornitz DM. Cell surface,heparin-like molecules are required for binding of basic fibroblastgrowth factor to its high affinity receptor. Cell Mol Life Sci 1991;64:841-848.

30. Yamuchi M, Shiba M. Lysine hyddroxylation and cross-linking of colla-gen. Methods Mol Biol 2008;446:95-108.

31. Liu Y, Cai S, Shu XZ, Shelby J, Prestwich GD. Release of basic fibro-blast growth factor from a crosslinked glycosaminoglycan hydrogel pro-motes wound healing. Wound Repair Regen 2007;15:245-251.

32. Martinez-Hernandez A, Delgado FM, Amenta PS. The extracellularmatrix in hepatic regeneration. Localization of collagen types I, III, IV,laminin, and fibronectin. Lab Invest 1991;64:157-166 [erratum LabInvest 1991;65:257].

304 WANG ET AL. HEPATOLOGY, January 2011

33. LeCluyse EL. Human hepatocyte culture systems for the in vitro evalu-ation of cytochrome P450 expression and regulation. Eur J Pharm Sci2001;13:343-368.

34. Brown B, Lindberg K, Reing J, Stolz DB, Badylak SF. The basementmembrane component of biologic scaffolds serived from extracellularmatrix. Tissue Eng 2006;12:519-526.

35. Seddon AM, Curnow P, Booth PJ. Membrane proteins, lipids anddetergents, not just a soap opera. Biochem Biophys Acta 2004;1666:105-117.

36. Ozeki M, Narita Y, Kagami H, Ohmiya N, Itoh A, Hirooka Y, et al.Evaluation of decellularized esophagus as a scaffold for cultured esopha-geal epithelial cells. J Biomed Mat Res A 2006;79:771-778.

37. Rieder E, Kasimir MT, Silberhumer G, Seebacher G, Wolner E, SimonP, et al. Decellularization protocols of porcine heart valves differ impor-tantly in efficiency of cell removal and susceptibility of the matrix todecllularization with human vascular cells. J Thorac Cardiovasc Surg2004;127:399-405.

38. Ricard-Blum S, Ruggiero F, van der Rest M. The collagen superfamily.In: Collagen. Vol. 247. Berlin/Heidelberg: Springer; 2005:35-84.

39. Yurchenco PD, O’Rear JJ. Basement membrane assembly. MethodsEnzymol 1994;245:489-518.

40. Seyer JM, Hutcheson ET, Kang AH. Collagen polymorphism in nor-mal and cirrhotic human liver. J Clin Invest 1977;59:241-248.

41. Traub W, Piez KA. The chemistry and structure of collagen. Adv Pro-tein Chem 1971;25:243-352.

42. Deyl Z, Miksik I, Eckhardt A. Preparation procedures and purityassessment of collagen proteins. J Chromatogr B 2003;790:245-275.

43. D’Amour KA, Bang AG, Eliazer S, Kelly OG, Agulnick AD, SmartNG, et al. Production of pancreatic hormone-expressing endocrine cellsfrom human embryonic stem cells. Nat Biotech 2006;24:1392-1401.

44. Pittenger MF, Mackay AM, Beck SC, Jaiswal RK, Douglas R, MoscaJD, et al. Multilineage potential of adult human mesenchymal stemcells. Science 1999;284:143-147.

45. Kazemnejad S, Allameh A, Soleimani M, Gharehbaghian A, Moham-madi Y, Amirzadeh N, et al. Biochemical and molecular characteriza-tion of hepatocyte-like cells derived from human bone marrowmesenchymal stem cells on a novel three-dimensional biocompatiblenanofibrous scaffold. J Gastroenterol Hepat 2009;24:278-287.

46. Lysy PA, Smets F, Najimi M, Sokal EM. Leukemia inhibitoryfactor contributes to hepatocyte-like differentiation of humanbone marrow mesenchymal stem cells. Differentiation 2008;76:1057-1067.

47. Campard D, Lysy PA, Najimi M, Sokal EM. Native umbilical cordmatrix stem cells express hepatic markers and differentiate into hepato-cyte-like cells. Gastroenterology 2008;134:833-848.

48. Song Z, Cai J, Liu Y, Zhao D, Yong J, Duo S, et al. Efficient genera-tion of hepatocyte-like cells from human induced pluripotent stemcells. Cell Res 2009;19:1233-1242.

49. Hayes A, Tudor D, Nowell M, Caterson B, Hughes C. Unique formsof chondroitin sulfate proteoglycans in stem cell niches. J HistochemCytochem 2007;56:125-138.

50. Couvelard A, Bringuier AF, Dauge MC, Nejjari M, Darai E, BeniflaJL, et al. Expression of integrins during liver organogenesis in humans.HEPATOLOGY 1998;27:839-847.

51. Lyon M, Deakin JA, Gallagher JT. Liver heparan sulfate structure. Anovel molecular design. J Biol Chem 1994;269:11208-11215.

52. Vongchan P, Warda M, Toyoda H, Toida T, Marks RM, Linhardt RJ.Structural characterization of human liver heparan sulfate. BiochimBiophys Acta 2005;1721:1-8.

HEPATOLOGY, Vol. 53, No. 1, 2011 WANG ET AL. 305