IntroductionMouse liver organogenesis is induced by a signal from thecardiac mesoderm at the 7-8 somite stage when the hepaticdiverticulum emerges from the foregut endoderm (Gualdi et al.,1996; Houssaint, 1980; Jung et al., 1999; Zaret, 2000). Theendodermal cells in the hepatic bud proliferate and invadethe septum transversum mesenchyme. While the hepaticparenchyma at an early stage of liver organogenesis consistsmostly of hepatoblasts, postnatal and adult liver parenchymaconsists of mature hepatocytes and biliary epithelial cells(BECs) that form intrahepatic bile ducts. Histochemicalanalysis showed that ductal plates are formed around portalveins during mid to late gestation (Shiojiri, 1984; Shiojiri,1997; Shiojiri et al., 2001). A number of metabolic enzymesexpressed in the adult liver, such as tyrosine aminotransferase(TAT), glucose-6-phosphatase (G6Pase), and carbamoylphosphate synthetase (CPS), become expressed at the lategestation or perinatal stage but not at the mid-gestation stage(Greengard, 1970; Haber et al., 1995). Based on these results,it is believed that hepatoblasts are common progenitors ofmature hepatocytes and BECs and that the lineage isdetermined around the mid-gestation stage. However, the

nature of hepatoblasts and the mechanism underlying thelineage determination are still largely unknown.

Hepatic progenitors are also known to be present in the adultliver (Fausto, 1994; Sell, 1994). Oval cells with the ability todifferentiate into both hepatocytes and BECs appear in theperiportal region following hepatic injury when hepatocyteproliferation is inhibited by a carcinogen, 2-acetylaminofluorene (Sell et al., 1981), or D-galactosamine(Lemire et al., 1991). Rat oval cells have been shownhistochemically to express immature hematopoietic cellmarkers, c-Kit, CD34 and Thy1 (Fujio et al., 1994; Omori etal., 1997; Petersen et al., 1998). In addition, small hepatocytesin adult rat liver are suggested to be hepatic progenitor cells(Mitaka et al., 1999; Tateno et al., 2000). However, it stillremains unclear what roles oval cells and small hepatocytesplay in normal liver development, homeostasis andregeneration. Besides adult liver, hepatic progenitor cells wereshown to exist in pancreas (Rao et al., 1989; Reddy et al., 1991;Zulewski et al., 2001; Tosh et al., 2002;). Surprisingly, recentresults showed that bone marrow cells could differentiate intohepatocytes (Lagasse et al., 2000; Oh et al., 2000; Petersen etal., 1999; Wang et al., 2002). Furthermore, hepatocytes were

1775

Hepatoblasts are common progenitors for hepatocytes andbiliary epithelial cells, although their nature remainslargely unknown. In order to isolate and to characterizehepatoblasts, we searched for cell surface antigensexpressed in mouse fetal hepatic cells by the signalsequence trap method and found that Dlk, also knownas Pref-1, was strongly expressed in fetal liver.Immunohistochemical as well as northern analysisindicated that Dlk was highly expressed in the E10.5 liverbud. The strong expression continued until the E16.5 stageand was significantly downregulated thereafter. Using amonoclonal antibody against Dlk, we isolated Dlk+ cellseither by a fluorescence-activated cell sorter or by anautomatic magnetic cell sorter. Dlk+ cells isolated fromfetal livers expressed albumin and formed colonies whencultured at low density with HGF and EGF for 5 days. Over60% of colonies derived from E14.5 Dlk+ cells containedboth albumin+ and cytokeratin 19+ cells, indicating that

a majority of colony-forming Dlk + cells are able todifferentiate into both hepatocyte and biliary epithelial celllineages. In addition, numerous microvilli were observed byelectronmicroscopic analysis in most of those cultured cells,also indicating differentiation of Dlk+ cells under thiscondition. Furthermore, 7% of the colony-forming Dlk+

cells were not only bipotential but also highly proliferative,forming a large colony containing more than 100 cellsduring 5 days of culture. By transplantation of Dlk+ cellsinto the spleen, donor-derived hepatocytes were found inthe recipient liver, indicating that Dlk+ cells differentiatedinto hepatocytes in vivo. These results indicate that Dlk+cells are hepatoblasts and that Dlk is a useful marker toenrich highly proliferative hepatoblasts from fetal liver.

Key word: Signal sequence trap, Hepatocyte, Stem cell, Fetal liver,Epithelial cell

Summary

Isolation of hepatoblasts based on the expression ofDlk/Pref-1 Naoki Tanimizu 1, Mitsuo Nishikawa 2, Hiroki Saito 3, Tohru Tsujimura 4 and Atsushi Miyajima 1,3,*1Stem Cell Regulation, Kanagawa Academy of Science and Technology (KAST), Teikyo University Biotechnology Research Center, 907 Nogawa,Kawasaki, Kanagawa 216-0001, Japan 2Kirin Pharmaceutical Research Lab, Takasaki, Gunma 370-1295, Japan3Institute of Molecular and Cellular Biosciences, University of Tokyo, 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-0032, Japan4First Department of Pathology, Hyogo College of Medicine, Nishinomiya, Hyogo 663-8501, Japan*Author for correspondence (e-mail: miyajima@ims.u-tokyo.ac.jp)

Accepted 23 January 2003Journal of Cell Science 116, 1775-1786 © 2003 The Company of Biologists Ltddoi:10.1242/jcs.00388

Research Article

1776

induced from ES cells as shown by expression of α-fetoproteinand albumin, and they were engrafted into recipient liver ashepatocytes (Chinzei et al., 2002). However, the mechanismsof the transdifferentiation from bone marrow cells and theinduction of hepatocytes from ES cells have not beeninvestigated. In order to understand the molecular mechanismunderlying these phenomena as well as the in vivo liverorganogenesis, it is necessary to isolate hepatoblasts and toinvestigate their differentiation and proliferation.

The methodology using mAbs against cell surface antigensand a cell sorter has been extensively used to isolate HSCs andcan be applicable for the isolation of the hepatic progenitorcells. In addition to the report that oval cells were isolated asThy1+ cells from regenerating adult liver (Petersen et al.,1998), recently, attempts have been made to purify progenitorsfrom fetal liver based on the expression of cell surface antigensby FACS. Suzuki et al. showed that the CD45–TER119–c-Kit–CD29+CD49f+ and CD45–TER119–c-Kit–c-Met+CD49f+/lo fraction of E13.5 mouse liver contained hepaticprogenitor cells (Suzuki et al., 2000; Suzuki et al., 2002).Kubota et al. showed that the RT1A1–OX18loICAM-1+

fraction of E13 rat fetal liver contained hepatoblasts (Kubotaand Reid, 2000). While these studies demonstrate the power ofcell sorters to enrich hepatoblasts, there are not enough surfaceantigens to identify hepatoblasts. In this study, we used thesignal sequence trap method to identify proteins with a signalsequence in E14.5 mouse fetal liver cells and found that Dlkwas abundantly expressed in fetal liver.

Dlk is a type I membrane protein that has six EGF-likerepeats in its extracellular domain and a short cytoplasmicdomain (Laborda et al., 1993; Smas and Sul, 1993). Theextracellular domain shows homology to Delta, one of theDrosophila melanogasterNotch ligands, but lacks the DSLdomain that is important for binding to Notch. Dlk was foundto be highly expressed in a small lung carcinoma cell line(Laborda et al., 1993) and was also identified as preadipocytefactor-1 (Pref-1) in 3T3-L1 preadipocytes (Smas and Sul,1993). Because Dlk orthologues were identified in human, ratand bovine as well as in mouse independently, many nameswere given to the same molecule; pG2 (Helman et al., 1990),fetal antigen-1 (FA-1) (Jensen et al., 1994), Pref-1 (Fahrenkruget al., 1999; Smas and Sul, 1993), stromal cell derived protein-1 (SCP-1) (GenBank/D16847), zona glomerulosa-specificfactor (ZOG) (Halder et al., 1998), and Dlk (Laborda et al.,1993). Here we show that Dlk is strongly expressed in the fetalliver between E10.5 and E16.5 and that Dlk can be used as amarker to enrich hepatoblasts.

Materials and MethodsMice, cells and antibodiesC57BL/6 mice (Nihon SLC, Japan) were used for all the experiments.Hamster anti-mouse Dlk mAb was prepared as described previously(Kanata et al., 2000). Rabbit anti-mouse cytokeratin 19 (CK19)polyclonal antibody was raised against the C-terminal peptide,HYNNLPTPKAI. Rabbit serum was used for immunohistochemistry.Dulbecco’s modified Eagle’s medium (DMEM) (Nissui, Tokyo,Japan), fetal bovine serum (FBS), liver perfusion medium, and liverdigestion medium (Gibco BRL, Gaithersburg, MD) were used toprepare fetal hepatocytes and for primary culture. Dulbecco’smodified Eagle’s F12 medium (Sigma, St Louis, MO) was used for alow density culture.

Cell preparation and cultureFetal hepatic cells of E14.5 liver were prepared and cultured accordingto the method of Kamiya et al. (Kamiya et al., 1999). Hepatic cellswere suspended in DMEM containing 10% FBS (Gibco BRL), 2 mML-glutamine (Gibco), 1× nonessential amino acid solution (Gibco), 1×insuline/transferrin/selenium (ITS) (Gibco), 50 µg/ml of gentamycin,10–7 M dexamethasone (Dex) (Sigma), and 10 ng/ml of mouseoncostatin M (OSM) and plated on gelatin-coated dishes.

Signal sequence trap The signal sequence trap method using a retroviral vector, pMX-SST,developed by Kojima and Kitamura (Kojima and Kitamura, 1999) wasused to identify cDNA clones encoding secreted and membraneproteins. cDNA was synthesized from poly(A) RNA of E14.5CD45–TER119– hepatic cells using the Timesaver cDNA synthesis kit(Pharmacia, Peapack, NJ) with random hexamer primers. After theaddition of the BstXI adaptor (Invitrogen, Carlsbad, CA), cDNA wasinserted into the BstXI site of the pMX-SST vector. The cDNA libraryused in this study contained 5.0×106 independent clones.

Whole mount in situ hybridizationA fragment of Dlk cDNA was amplified by the reverse-transcriptionpolymerase chain reaction (RT-PCR) with two primers: 5′-ATG CTTCCT GCC TGT GC-3′ and 5′-GCA CGG GCC ACT GGC-3′. ThePCR fragment was subcloned into the pCRII vector (Invitrogen).Sense and antisense single-stranded RNA probes were prepared by invitro transcription using the dioxigenin (DIG) RNA labeling kit(Roche, Basel, Switzerland). E10.5 embryo was fixed in 4% PFA andbleached in 6% hydrogen peroxide. After treatment with proteinaseK, the embryo was hybridized with 0.5 µg/ml sense or antisenseprobes at 65°C overnight. After washing and blocking procedures, theembryos were incubated with alkaline phosphatase (AP)-conjugatedanti-DIG antibodies (Roche) at 4°C for overnight. The signal wasdeveloped in 4-nitroblue tetrazolium chloride/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) solution (Roche).

Northern blotting analysisTotal RNA was extracted from tissues or cultured cells with Trizolreagent (Gibco). After electrophoresis, RNA transferred to a nylonmembrane was hybridized with DIG-labeled antisense probe at 48°Covernight and then incubated with AP-conjugated anti-DIG antibodies(Roche) at room temperature. The signal was developed with CDPstar(Roche).

ImmunochemistryFetal and adult livers were embedded in OCT compound (SakuraFinechemical, Tokyo, Japan). Frozen sections were prepared using aMicrotome cryostat HM 500 (Microm, Walldorf, Germany) andmounted on glass slides coated with MAS (Matsunami glass, Japan).They were then fixed in 4% PFA and incubated with normal goatserum. Primary and secondary antibodies were diluted to 1 µg/ml and5 µg/ml, respectively, in 3% normal goat serum. The samples wereadded with anti-Dlk mAb followed by biotinylated goat anti-hamsterIgG (Vector, Burlingame, CA). The samples were incubated in eachantibody solution at 4°C in a moist chamber. Expression of Dlkwas visualized with the Vectastain ABC kit (Vector) and 3,3-diaminobenzidine tetrahydrochloride (DAB) (Roche). To detectCK19, samples were incubated with 1000-fold dilution of anti-CK19serum followed by AP-conjugated goat anti-rabbit IgG (Vector). Thesignal was developed with NBT/BCIP (Roche).

For double immunofluorescence staining, fetal hepatic cellsmounted on glass slides were fixed in 4% PFA and incubated withanti-Dlk mAb and rabbit anti-albumin polyclonal antibody (Nordic,

Journal of Cell Science 116 (9)

1777Isolation of Dlk/Pref-1+ hepatoblasts

Sweden). Dlk was detected with biotinylated goat anti-hamster IgG(Vector) and FITC-conjugated streptavidin (PharMingen) and albuminwas detected with rhodamine-conjugated goat anti-rabbit IgG(Chemicon, Temecula, CA). The samples were examined under afluorescence microscope, Nikon Eclipse E800 (Nikon, Tokyo, Japan).

Dlk+ cells isolated by AutoMACS were mounted on glass slidesand fixed in 4% PFA. They were incubated with 1 µg/ml anti-albuminor 20 µg/ml rabbit anti-human α-fetoprotein (AFP) antibodies (ICNBiomedicals, Costa Mesa, CA). Both signals were detected byrhodamin-conjugated anti-rabbit IgG (Chemicon).

Flow cytometric analysis of Dlk and other cell surface markersE14.5 fetal hepatic cells were incubated with anti-Dlk mAb and ratmAbs against CD45, TER119 and PECAM-1 (PharMingen, San Jose,CA). After washing with PBS, cells were incubated with FITC-conjugated goat anti-hamster IgG (Vector) and PE-conjugated goatanti-rat IgG (Cedarlane, Ontario, Canada). All these antibodies werediluted 100-fold and used for staining. The samples were then washedwith PBS and mixed with 1 µg/ml propidium iodide (PI) before flowcytometric analysis with a FACScallibur (Becton Dickinson, San Jose,CA).

Isolation of Dlk+ cells from fetal liverDlk+ cells were isolated by an automatic magnetic cell sorter

(AutoMACS) (Miltenyi Biotec, Bergisch Gladbach, Germany). E14.5hepatic cells were incubated with anti-Dlk mAb, biotinylated goatanti-hamster IgG. After a wash with PBS, cells were resuspended inAutoMACS running buffer (1×108 cells/ml of PBS containing 0.5%BSA) and 100 µl/ml of streptavidin-labeled microbeads (MiltenyiBiotec) were added. After a wash with the running buffer, the cellswere loaded onto a magnetic column and Dlk+ cells were eluted fromthe column after the depletion of Dlk– cells. Alternatively, E14.5CD45–TER119– cells were incubated with anti-Dlk mAb and FITC-conjugated goat anti-hamster IgG and sorted into Dlk– and Dlk+

fractions by using a FACSvantage (Becton Dickinson).

RT-PCR analysis Total RNA (1 µg) was used to synthesize cDNA using the First-strandcDNA synthesis kit (Amersham Pharmacia Biotech, Piscataway, NJ)and random hexamer primers. The samples were denatured at 94°Cfor 2 minutes, followed by the thermal cycles; denaturation at 94°Cfor 30 seconds, annealing at the temperature set for each pair ofprimers for 30 seconds, extension at 72°C for 2 minutes. The thermalcycle was repeated 20 times for AFP and albumin, 25 times for Dlkand GAPDH, and 30 times for other genes. The primers used for RT-PCR are shown in Table 1. The primers used for Dlk were the sameas those used for in situ hybridization.

Quantitative PCR analysis was also performed to measure themRNA levels for AFP and albumin by using LightCycler (Roche).

Table 1. Oligonucleotides used in RT-PCRGene name Sequence

AFP Sense 5′-CCA TCC TGC AGA CAC TCC AG-3′Antisense 5′-AAC ACA GCC GGA CCA TTT CTC-3′

Albumin Sense 5′-CAT GAC ACC ATG CCT GCT GAT-3′Antisense 5′-GTG GAT CCC TGG TGG AAA GGC-3′

G6Pase Sense 5′-ACT GGT TCA ACC TCG TCT T-3′Antisense 5′-CGA AAG ATA GCG AGA GTA GA-3′

CPS Sense 5′-ACT GAG AGA TGC TGA CCC TA-3′Antisense 5′-CCT GGA AAT TGG TGA GGA GA-3′

TAT Sense 5′-GAC GAG GAA GGC TTT GTG AG-3′Antisense 5′-TTG TAC TTT CCG GAG TCC AGG G-3′

Tryptophan oxygenase (TO) Sense 5′-ACA ATG AAG AAG ACA GAG C-3′Antisense 5′-TGT AGT CTC CTC CAA AGT TA-3′

Glutamine synthetase (GS) Sense 5′-CCA GGG TGA GAA AGT CCA AGC-3′Antisense 5′-GTT CGT CGC CTG TTT CGT TGA G-3′

γ-Glutamyltranspeptidase (GGT) Sense 5′-ATG AGC TCT GAG TTC TAC GC-3′Antisense 5′-TAG GTA AAA GCT GGT TGT GC-3′

HNF1β Sense 5′-GAA AGC AAC GGG AGA TCC TC-3′Antisense 5′-CCT CCA CTA AGG CCT CCC TC-3′

HNF3β Sense 5′-TCA AGT GTG AGA AGC AAC TG-3′Antisense 5′-GAC GAC ATG AGG TTG TTG AT-3′

HNF4 Sense 5′-TCA AAG CCA TCA TCT TCT TT-3′Antisense 5′-CAG GAG CTT GTA GGA TTC AG-3′

HNF6 Sense 5′-CAG CAC CTC ACG CCC ACC TC-3′Antisense 5′-CAG CCA CTT CCA CAT CCT CCG-3′

CK8 Sense 5′-AGT CTC AGA TCT CAG ACA CG-3′Antisense 5′-CCA TAG GAT GAA CTC AGT CC-3′

CK18 Sense 5′-GGA CCT CAG CAA GAT CAT GGC-3′Antisense 5′-CCA CGA TCT TAC GGG TAG TTG-3′

CK19 Sense 5′-GTC CTA CAG ATT GAC AAT GC-3′Antisense 5′-CAC GCT CTG GAT CTG TGA CAG-3′

Connexine-26 (Cx26) Sense 5′-CCA GAA GGT CCG TAT CGA AG-3′Antisense 5′-GGA CTT TCC TGA GCA ATA CC-3′

Cx32 Sense 5′-CTA TCT GGG TTT GCC ATA AG-3′Antisense 5′-TCT TTA CCT CTT CCA GGT GA-3′

Cx43 Sense 5′-GTG ATG AAC AGT CTG CCT TT-3′Antisense 5′-TGA TGA AGA TGG TTT TCT CC-3′

c-Kit Sense 5′-AAC TTT TCC TGG TTG GCC TT-3′Antisense 5′-GAT AGT CAG CGT CTC CTG GC-3′

GAPDH Sense 5′-ACC ACA GTC CAT GCC ATC AC-3′Antisense 5′-TCC ACC ACC CTG TTG CTG TA-3′

1778

Culture of Dlk+ cells A low density culture was performed to examine the growth potentialof Dlk+ cells. E14.5 Dlk– and Dlk+ cells isolated by AutoMACS werecultured in DMEM F-12 (Sigma) at a density of 1000 and 50cells/cm2, respectively, on 6-well plates coated with type IV collagen(Nitta Gelatin, Osaka, Japan). The medium was supplemented with10% FBS, 1×ITS, 10 mM nicotineamide (Wako, Tokyo, Japan), 0.1µM Dex, and 5 mM L-glutamine. Various combinations of 20 ng/mlepidermal growth factor (EGF) (PeproTech, London, UK),hepatocyte growth factor (HGF) (R&D, Minneapolis, MN), and OSMwere added 18 hours after the initiation of the culture. After 5 daysof culture, cell nuclei were stained with hematoxylin (Muto PureChemicals, Japan) and the cells in each colony were counted. Colonynumbers in three wells were determined in each set of culture andthe experiment was repeated four times. These data were analyzedstatistically using JMP program to obtain standard deviations and P-values.

The expression of albumin and CK19 in colonies was analyzedby immunocytochemistry. The cells cultured on chamber slides(NUNC, Roskilde, Denmark) coated with type IV collagen werefixed in methanol at –20°C for 10 minutes. Alternatively, eachlarge colony formed on 6-well plates was placed with a cloning ringand treated with trypsin. The cells removed from the dish weremounted on glass slides and fixed in methanol. After washing withPBS and blocking with 3% donkey serum (Chemicon) for30 minutes, the cells were incubated with 2 µg/ml goat anti-mouse albumin antibody (Bethyl laboratories, Inc., Mongomery,TX) and rabbit anti-CK19 serum (1000-fold dilution) at 4°Covernight. After they were washed with PBS containing 0.05%Tween 20 (PBST), the samples were incubated with Cy3-conjugateddonkey anti-goat IgG antibody (Rockland, Gilbertsville, PA) andFITC-conjugated donkey anti-rabbit IgG antibody (Rockland) for2 hours at 4°C.

FACSvantage was used for the single cell sorting. Each Dlk+ cellsorted from E14.5 hepatic cells was individually plated in one well ofa 96-well plate coated with type IV collagen. After 5 days of culture,cells were fixed in methanol and the expression of albumin and CK19was examined as described above.

Transmission-electron microscopyUltrastructures of E14.5 Dlk+ cells were examined with atransmission-electron microscope before and after the low densityculture. Purified Dlk+ cells from fetal livers were fixed in 2.5%phosphate-buffered glutaraldehyde for 30 minutes at roomtemperature and in 1% phosphate-buffered OsO4 for 15 minutes atroom temperature. Dlk+ cells cultured for 5 days on 6-well platescoated with type IV collagen were detached from dishes by trypsintreatment and similarly fixed. Then, both cells were dehydrated andembeded in epoxy resin. Thin sections for electron microscopy werecounterstained with uranil acetate and lead citrate, and examined witha JEM-1220 electron microscope (JEOL, Tokyo, Japan).

Cell transplantation Dlk+ cells were isolated from E14.5 fetal livers of GFP transgenicmouse by using AutoMACS. Acute liver injury was induced inrecipient mice by intraperitoneal administration of anti-Fas antibodyJo2 (200 µg/kg; PharMinagen) before transplantation. After 24 hoursof the injection of anti-Fas antibody, 2×105 Dlk+ cells weretransplanted intrasplenically into the anesthesized recipient mice asdescribed in the previous work (Ponder et al., 1991). After 8 or 36weeks, the recipient mice were sacrificed and frozen sections of theirlivers were prepared. Donor-derived GFP+ cells were detected undera fluorescence microscope. The sections including donor-derivedGFP+ cells were used for immunostaining with anti-albuminantibody.

Results Identification of secreted and membrane proteinsexpressed in fetal liver by the signal sequence trap In order to identify surface markers for fetal hepatic cells otherthan blood cells, we applied the signal sequence trap usingthe cDNA library of E14.5 hepatic cells deprived ofCD45+TER119+ hematopoietic cells. Among variousmolecules identified by this approach, serum proteins such asAFP and albumin were most frequently found; 40% of theclones obtained comprised cDNAs for these two serumproteins (Table 2). In addition to these clones, we repeatedlyisolated cDNA fragments encoding Dlk also known as Pref-1,suggesting that Dlk is one of the most abundant membraneproteins expressed in E14.5 fetal liver. Additional knownmembrane and secreted proteins identified were carrierproteins such as vitamin D binding protein and retinol bindingprotein, cytokines such as M-CSF and IGF-II, cytokinereceptors such as interferon and lymphotoxin β receptors,proteases such as HGF-activator and plasminogen, andextracellular matrix proteins such as collagen, vitronectin andnidogen.

Dlk is expressed in fetal liver and downregulated alongwith liver development Dlk/Pref-1 was previously shown to be expressed in E8.5 fetusand in liver, pituitary, lung, vertebra and tongue at E13.5 (Smasand Sul, 1993). In order to examine Dlk expression at the onsetof liver organogenesis, whole-mount in situ hybridization wasperformed using E10.5 embryo. In this experiment, Dlk wasdetected in the liver bud as well as vertebra (Fig. 1). The Dlkexpression in the E10.5 liver bud was also confirmed by RT-PCR (Fig. 2A). To examine the expression of Dlk during laterliver development, Northern blot analysis was performed usingtotal RNA extracted from developing livers. Dlk mRNA wasstrongly expressed in fetal liver between E12.5 and E16.5 (Fig.2B). Its expression was downregulated later in gestation anddisappeared in the neonatal and adult livers (Fig. 2B). Theexpression of Dlk during liver development was also examinedby using primary cultures of fetal hepatocytes. We previouslyshowed that E14.5 fetal hepatocytes are induced to differentiate

Journal of Cell Science 116 (9)

Fig. 1. Expression of Dlk mRNA in E10.5 embryo detected by in situhybridization. E10.5 mouse embryo fixed in 4% PFA was hybridizedwith DIG-labeled sense (A) and antisense (B) probes of Dlk asdescribed in Materials and Methods. After incubation with alkalinephosphatase (AP)-conjugated anti-DIG antibody, the signal wasvisualized by AP activity using BCIP/NBT as a substrate. Dlk wasdetected in the liver bud, which is seen under the heart, as well asvertebra. flb, forelimb bud; h, heart; lb, liver bud.

1779Isolation of Dlk/Pref-1+ hepatoblasts

morphologically and to express various metabolic enzymessuch as TAT, G6Pase and CPS by OSM in the presence of Dexin vitro (Kamiya et al., 1999). Northern blot analysis showedthat Dlk expression gradually disappeared without OSM, andwas more rapidly disappeared in the presence of OSM, acondition that induced expression of CPS and TAT (Fig 2C).Expression of TAT and CPS was induced after birth(Greengard, 1970; Haber et al., 1995), while expression of Dlk

was downregulated after E16.5 (Fig. 2B). It is thus likely thatE14.5 hepatocytes in the primary culture spontaneouslydifferentiated to a later fetal stage and OSM was required forfurther differentiation to the postnatal stage in vitro.

Consistent with the results of in situ hybridization,immunohistochemical examination using anti-Dlk mAb onfrozen sections indicated that Dlk was expressed in endodermalcells in the liver bud at the E10.5 stage, while it was not

Table 2. Selected clones identical to known proteinsClone number* Full length (bp)† Detected length (bp)‡

Secretion proteinSerum proteinAlbumin 32 2050 438-1599α-fetoprotein 31 2009 494-1628α1 microbloguline 8 1234 664-778α1 antitrypsin 2 1334 627α2 antiplasmin 3 2179 1390-1473β2 glycoprotein 1 1158 447Inter α inhibitor H3 2 2775 1828-1831Complement factor I 3 2094 977-1010α2 HS glycoprotein 1 1457 1091C1 inhibitor 1 1788 926ProteasePlasminogen 4 2720 606-825P100 serine protease 4 5135 668-680Factor B 3 2443 801-891HGF activator precursor 1 1968 420Career proteinVitamin D binding protein 2 1431 843-1026Retinol binding protein 3 606 564-597Corticosteroid binding protein 1 1462 747Mama protein (scavenger receptor) 2 2168 618-933Growth factorM-CSF 1 2099 1218Epithelin 3 2100 679-1069Insulin-like growth factor II 1 1435 349IFN receptor soluble isoform 1 1045 595Extracellular matrix proteinCollagen type I 1 631Nidogen 1 3788 855Vitronectin 1 1586 479Type I membrane proteinLigand proteinDlk/Pref-1 22 1589 429-960Adhesion moleculeN-cadherin 1 4321 1181Junctional adhesion molecule 1 2029Carcino embryonic antigen (CEA)-like antigen 1 1786 545ReceptorNotch-2 1 7579 1772Lymphotoxin β receptor 1 2076 1231Eph B4 receptor 1 3733 981OthersSel1L 1 2370 480BP1/6C3 antigen (metalloprotease) 1 3584 600CD13/aminopeptidase 1 3367 552CD73/ecto5′ nucleotidase 1 1731 966Leukemia virus envelope protein 1 1415 879Milk fat globule membrane protein 1 2077 519LAMP2 (lysosomal membrane protein) 3 1700 612-696Type II membrane proteinCD82 (C33/R2/IA4) 1 1657 492ER proteinThyroid hormone binding protein 6 2481 1116-1515ER99 protein 1 2759 479

154*Number of clones encoded same protein.†Number of nucleotide of the full-length mRNA.‡Ranges of cDNA length inserted in pMX-SST.

1780

expressed in the foregut, from which the hepatic diverticulumwas generated (Fig. 3A-D). Dlk was also detected in the E14.5liver (Fig. 3E), but not in the adult liver (Fig. 3F). These resultscollectively indicate that Dlk is expressed in immature livercells from the onset of the liver organogenesis.

Expression of Dlk in fetal hepatocytes The fetal liver parenchyma consists of immature hepatocytes,BECs, and their common progenitors, hepatoblasts. In order toknow which endodermal cells express Dlk, we performeddouble immunostaining analysis of Dlk with either albumin orCK19. First, E14.5 fetal hepatic cells mounted on glass slideswere stained with anti-Dlk mAb and anti-albumin antibody.Since the expression of albumin is induced at the beginning ofliver organogenesis (Jung et al., 1999), albumin was expectedto be detected not only in hepatocytes but also in hepatoblasts.

We found that large hepatic cells expressed both Dlk andalbumin in fetal liver between E12.5 and E18.5 (Fig. 3G-J anddata not shown). We then stained fetal liver sections with anti-Dlk mAb, and anti-CK19 antibody that stains BECs. Althoughthe formation of ductal plates has already started at the E14.5or E15.5 stage (Clotman et al., 2002; Coffinier et al., 2002;Shiojiri, 1997; Shiojiri et al., 2001), the cells comprising theductal structure were not uniformly stained with anti-CK19antibody (data not shown). Therefore, in order to investigatewhether Dlk is expressed in BECs, we used E17.5 liversections in which CK19+ ductal plate structures becameapparent. Double immunostaining showed that CK19+ BECswere Dlk– at E17.5 (Fig. 3K-M). These results suggest that Dlkis expressed in both immature hepatocytes and hepatoblasts,and the expression is downregulated when hepatoblastsdifferentiate into CK19+ BECs.

Isolation of Dlk+ cells using cell sortersAs Dlk is a membrane protein, it might be useful as a surfaceantigen to separate immature hepatocytes from other types offetal hepatic cells such as hematopoietic and endothelial cells.Flow cytometric analysis using anti-Dlk mAb showed thatabout 10% of total hepatic cells were Dlk+ and that they wereclearly separable from Dlk– cells, which mainly consisted

Journal of Cell Science 116 (9)

Fig. 2. Expression of Dlk in immature liver cells in vivo and in vitro.RT-PCR (A) and northern blotting (B,C) were performed to detectDlk mRNA during liver development and fetal hepatocyte primaryculture. (A) Dlk expression was clearly detected in the E10.5 liverbud and the E14.5 liver but not in the neonatal liver shown by RT-PCR using cDNA synthesized from total RNA of livers at each stage.GAPDH expression was also examined to ensure an equal quantityof cDNA used for PCR. (B) Dlk was strongly expressed in fetallivers between E12.5 and E16.5, but it was rapidly downregulated inlater gestation. In neonatal and adult livers, its expression was notdetected. Each lane was loaded with 10 µg of total RNA extractedfrom livers at each stage. GAPDH expression was also examined toensure an equal loading of RNA. (C) Fetal hepatic cells wereprepared from E14.5 fetal liver and cultured on gelatin-coated dishesin the presence or absence of dexamethasone (Dex) and oncostatin M(OSM). TAT and CPS were weakly expressed after 4 days andclearly detected after 7 days of culture with Dex and OSM. Each lanewas loaded with 10 µg of total RNA extracted from cultured cells ateach time point. Note that Dlk was rapidly downregulated in thepresence of Dex and OSM, although downregulation occurredspontaneously without OSM.

Table 3. Gene expression profile of E14.5 Dlk– and Dlk+

cellsDlk– Dlk+

Hepatocyte markerα-fetoprotein – +Albumin – +Glucose-6-phophatase – –Tyrosine aminotransferase – –Carbamoyl phosphate synthetase – –/+Tryptophane oxygenase – –Glutamine synthetase –/+ +Connexine 26 –/+ +Connexine 32 – –/+

Biliary epithelial cell markerCytokeratine 19 – –γ-Glutamyltransferase – +Connexine 43 + +

Liver enriched transcription factorHNF1β – +HNF3β – +HNF4 – +HNF6 – +

Immature cell markerc-Kit + –

Other genes expressed in adult liverCytokeratin 8 – +Cytokeratin 18 –/+ +GAPDH + +

Dlk– and Dlk+ cells were isolated from E14.5 fetal liver with AutoMACS.cDNA was synthesized from 1 µg of total RNA of each cell fraction. Thethermal cycle of PCR was repeated 20 times for AFP and albumin, 25 timesfor GAPDH, and 30 times for other genes. Ten out of 50 µl of PCR productswere separated in 1.5% agarose gel. The band intensity detected after stainingin the EtBr solution is represented in this table. +, PCR products were clearlydetected; –/+, the products were slightly detected; –, the products were notdetected.

1781Isolation of Dlk/Pref-1+ hepatoblasts

of CD45+, TER119+ and PECAM-1+ cells (panspecifichematopoietic, erythroid and endothelial cells, respectively)(Fig 4A). We then separated Dlk– and Dlk+ cells from E14.5fetal liver by AutoMACS. Flow cytometric analysis showedthat both Dlk– and Dlk+ fractions were over 95% pure (datanot shown), and immunostaining revealed that about 95% of

cells in the Dlk+ fraction were AFP+ and albumin+ (Fig. 4B).Furthermore, we employed quantitative RT-PCR analysisto examine the expression of AFP and albumin inCD45–TER119–Dlk– and CD45–TER119–Dlk+ cells, whichwere separated by FACSvantage. Dlk+ cells expressed AFP andalbumin about 60 and 40 times, respectively, higher than Dlk–

Fig. 3.Histochemical analysis of Dlk expression in mouse liver. (A-D) Horizontal (A,B,D) and sagittal (C) sections of frozen E10.5 embryowere stained with anti-Dlk mAb. Dlk was detected in the E10.5 liver bud, but not in gut tubes, heart and forelimb bud (A-C). Highermagnification of the box in B is shown in D. The endodermal cells of the liver bud were stained with anti-Dlk mAb, whereas those of the guttube were not stained. (E,F) E14.5 liver (E) and adult liver (F) were also incubated with anti-Dlk mAb. Dlk was expressed in E14.5 fetal liverbut not in adult liver. (G-J) E14.5 hepatic cells were mounted on glass slides and incubated with anti-Dlk mAb and anti albumin antibody. Cellnuclei were stained with hematoxylin (G). The immunofluorescence staining of Dlk and albumin was visualized with FITC (H) and rhodamine(I), respectively. Large fetal hepatic cells (arrowheads in G) were stained with anti Dlk mAb (green in H) and anti-albumin (red in I). Dlk+ cellswere identical to albumin+ cells (yellow in J). Dlk– cells with large nuclei and less cytoplasm were mostly hematopoietic cells.(K-M) Continuous frozen sections of E17.5 fetal liver were stained with anti-Dlk mAb (K), anti-CK19 antibody (L), and both antibodies (M).Dlk+ cells (brown in K) and CK19+ biliary epithelial cells (blue in L) were completely distinguishable (M). CK19+ bile ducts were visiblearound portal veins as well as ductal plates consisting of double layers of CK19+ cells. bd, bile duct; fg, foregut; flb, forelimb bud; h, heart; lb,liver bud; mg, mid-gut; nt, neural tube; pv, portal vein. Bars, 100 (A-D); 50 µm (E-M).

1782

cells (Fig. 4C). These results indicate that immaturehepatocytes, possibly including hepatoblasts, are enriched inthe Dlk+ fraction.

To further characterize the E14.5 Dlk+ cells isolated byAutoMACS, expression of several genes including hepatocyteand BEC markers was examined by RT-PCR (Table 3). Dlk+

cells expressed strongly AFP and albumin, and significantlyCK8, CK18, Cx26, Cx43, GS and GGT. Although GGT isknown to be expressed in BECs of adult liver, it was previously

reported that rat GGT was expressed also in fetal hepatocytes(Shiojiri et al., 1991; Holic et al., 2000). In contrast to earlyhepatocyte markers, expression of CPS, TAT, G6Pase, TO andCx32, which are known to be expressed in mature hepatocytes,was undetected or barely detectable. Consistent with the resultthat Dlk was downregulated in BECs (Fig. 3K-M), Dlk+ cellsdid not express CK19. In addition to early hepatocyte markergenes, Dlk+ cells expressed liver enriched transcription factors,HNF1β, HNF3β, HNF4 and HNF6, whereas none of them wasexpressed in Dlk– cells. On the contrary, expression of c-Kitwas detected in Dlk– but not Dlk+ cells.

Highly proliferative potential of Dlk+ cellsTo examine the proliferative potential of each sorted cell, wecultured Dlk– and Dlk+ cells at a low density to evaluate clonalgrowth. Dlk– and Dlk+ cells were isolated from E14.5 fetalliver using AutoMACS and were plated at a density of 1000and 50 cells/cm2, respectively, on 6-well plates coated withtype IV collagen. After 5 days of culture, we counted thenumber of large colonies containing over 100 cells, which wereconsidered to be formed from highly proliferative cells. First,we examined various combinations of cytokines, EGF, HGF,and OSM, for growth in vitro (Table 4). As Dlk+ cellsefficiently proliferated with HGF or HGF plus EGF, HGF wasthe most effective growth factor for Dlk+ cells among threecytokines, consistent with the fact that Dlk+ cells expressed c-Met, the receptor for HGF (data not shown). Since the numberof large colonies formed in the presence of HGF plus EGFslightly exceeded that with HGF in this low density culturecondition, we used the combination of HGF plus EGF in thefollowing experiments.

We then evaluated the growth potential of Dlk+ cells in a lowdensity culture with EGF and HGF. The growth of a singleDlk+ cell was followed for 5 days. Dlk+ cells proliferatedexponentially between culture day 2 and 4, and some coloniesstill actively proliferated beyond day 4 and reached 100 cellsat the end of the culture. After 5 days of culture, 24±3% ofinput Dlk+ cells proliferated and formed colonies with varioussizes (Fig. 5). Half of the colony-forming Dlk+ cells formedsmall colonies containing less than 40 cells, while 10% of thecolony-forming Dlk+ cells were highly proliferative. Since thenumber of colonies formed from Dlk– cells was less than 5%of that from Dlk+ cells, colony-forming cells were enriched inthe Dlk+ fraction.

Differentiation potential of Dlk+ cellsDlk is expressed in the E10.5 liver bud in which hepaticparenchyma is believed to consist of hepatoblasts. In order to

Journal of Cell Science 116 (9)

Fig. 4.Flow cytometric analysis of Dlk expression and isolation ofDlk+ cells. (A) E14.5 liver cells were stained with anti-Dlk mAb andanalyzed by FACS. Dlk+ cells were separated from Dlk– cells, whichmainly consisted of hematopoietic and endothelial cells. Dlk+ cellswere about 10% of total E14.5 hepatic cells. (B) Dlk+ cells isolated byAutoMACS were stained with anti-AFP (1) and anti-albuminantibodies (2). The morphologies of Dlk+ cells are shown in the upperpanels. Over 95% of Dlk+ cells were stained with anti-AFP and anti-albumin antibodies (red in lower panel). (C) cDNA was synthesizedfrom total RNA of CD45–TER119–Dlk– and CD45–TER119–Dlk+

cells separated by FACSvantage. Quantitative PCR reaction wasperformed with a LightCycler. The signals for AFP and albumin werenormalized with the GAPDH level and relative expression level (theexpression level in Dlk– cells = 1) are shown. The mRNA levels forAFP and albumin in Dlk+ cells (white bar) were 60 and 40 times,respectively, those in Dlk– cells (black bar).

Table 4. Formation of large colonies containing over 100 cells from Dlk– and Dlk+ cells using several growth factorsControl EGF HGF HGF+EGF OSM OSM+EGF HGF+OSM HGF+OSM+EGF

Dlk− 0.19±0.02 0.28±0.13 0.42±0.04 0.44±0.13 0.28±0.11 0.22±0.05 0.29±0.04 0.23±0.07Dlk+ 3.6±0.5 6.7±1.7 11±2 12±4 4.7±1.3 3.9±1.3 4.7±2.1 8.1±2.7

Dlk– and Dlk+ cells inoculated onto type IV collagen-coated dishes at a density of 1000 and 50 cells/cm2, respectively. Data shown here represent the numberof large colonies derived from 500 cells. The experiment was performed independently four times. The colony number in the presence of HGF+EGF significantlyexceeded that in the other conditions (P<0.01) except in the presence of HGF. Although the difference in colony number between the cells cultured with HGF andHGF+EGF was very small, the number of colonies in the presence of both HGF and EGF exceeded that in the presence of HGF alone in each set of the lowdensity culture. In all the conditions, the number of colonies from Dlk− cells was less than 5% of that from Dlk+ cells. This indicated highly proliferative cellswere enriched in Dlk+ cells.

1783Isolation of Dlk/Pref-1+ hepatoblasts

know whether Dlk+ cells are bipotential at the mid-gestation,we tested expression of albumin and CK19 in colonies formedfrom E14.5 Dlk+ cells. The results showed that over 60% of

colonies derived from E14.5 Dlk+ cells contained bothalbumin+ and CK19+ cells (Fig. 6A-C). As colonies withvarious sizes contained both types of cells, there seemed nodirect correlation between differentiation potential andproliferation potential. In this low density culture condition, thecolonies contained cells which expressed either albumin orCK19 and also those that expressed both. To confirm theexistence of albumin+CK19+ cells, a large colony was pickedfrom the culture dish and the cells mounted on glass slides weresubjected to immunostaining. We did find cells that expressedboth albumin and CK19 (Fig. 6D). These results stronglysuggested that single Dlk+ cells were able to differentiate intoalbumin+, CK19+, and albumin+CK19+ cells. However, therestill remained the possibility that some colonies were derivedfrom multiple cells even in the low density culture. In order toexclude such possibility we performed single cell sorting. Byusing FACSvantage each single Dlk+ cell was sorted and platedin one well of a 96-well plate. After 5 days of culture, colonieswere formed in about 20% of wells inoculated with a singleDlk+ cell. About 10% of those colonies contained over 100cells that consisted of both albumin+ and CK19+ cells (Fig.6E), indicating that a single Dlk+ cell was able to differentiateinto both hepatocyte and BEC lineages. Taken together, it isconcluded that the majority of colony-forming Dlk+ cells atE14.5 are hepatoblasts that display bilineage gene expressionand some of Dlk+ hepatoblasts are highly proliferative.

We also analyzed ultrastructure of Dlk+ cells before andafter 5 days of culture by electron microscope. The majority

0

10

20

30

40

50

60

-20 21-40 41-60 61-80 81-100 101-120121-140141-

Number of cells in one colony

Num

ber

of c

olon

ies

Fig. 5.Colony formation of E14.5 Dlk– and Dlk+ cells. Dlk– andDlk+ cells were isolated from E14.5 fetal liver by AutoMACS andcultured at a density of 1000 and 50 cells/cm2, respectively, on typeIV collagen-coated 6-well plates in the presence of 20 ng/ml of HGFand EGF. The number of cells in each colony was counted after 5days of culture. Data shown here are numbers of colonies formedfrom 500 Dlk– (black bar) and Dlk+ (white bar) cells. 24±3% ofE14.5 Dlk+ cells formed various sizes of colony after 5 days ofculture. Among those colony-forming Dlk+ cells, 10% formed largecolonies containing over 100 cells.

Fig. 6.Colonies formed fromE14.5 Dlk+ cells consist ofalbumin+ and CK19+ cells.(A-C) Dlk+ cells were isolatedfrom E14.5 fetal liver byAutoMACS and cultured onchamber slides coated with typeIV collagen at a density of 50cells/cm2 in the presence of HGFand EGF. After 5 days of culture,cells were fixed and stained withanti-albumin and anti-CK19antibodies. Expression of albuminand CK19 was visualized with Cy-3 and FITC, respectively. Varioussizes of colonies (A, a smallcolony containing about 40 cells;B, a medium colony containingabout 60 cells; C, a large colonycontaining over 100 cells) wereformed in a low density culture,which contained both albumin+

(red in A-1, B-1 and C-1) and CK19+(green in A-2,B-2 and C-2) cells. Both images were merged in A-3,B-3 and C-3. (D) A large colony formed from Dlk+

cells was picked from the culture dish and mountedon a glass slide by cytospin. The cells were stainedwith anti-albumin and anti-CK19 antibodies. Some ofthem expressed both albumin (red in D-1) and CK19(green in D-2). In D-3, both images were merged.(E) A single Dlk+ cell was isolated and inoculated inone well of a 96-well plate coated with type IV

collagen by FACSvantage. A large colony derived from a single Dlk+ cell consisted of albumin+ (red in E-1) and CK19+ (green in E-2) cells.Both images were merged in E-3. Bars, 100 µm.

1784

of E14.5 Dlk+ cells had a round-shaped nucleus, a smallnucleolus, and elongated mitochondria (Fig. 7A). The formertwo characteristics suggest that E14.5 Dlk+ cells are ratherprimitive cells that are not actively synthesizing mRNAs. After5 days of culture, most of the cultured cells exhibited a largenumber of microvilli on their cell surfaces and cleaved nucleus,larger nucleolus and many round-shaped mitochondria in thecytoplasm (Fig. 7B), suggesting that Dlk+ cells differentiatedduring the culture. However, because this culture condition wasfor proliferation rather than maturation, as described above, theGolgi apparatus, lysosomes and accumulation of glycogenwere not apparent in the cytoplasm.

Transplantation of Dlk+ cellsIn order to investigate differentiation potential of E14.5 Dlk+

cells in vivo, we transplanted Dlk+ cells into the spleen ofrecipient mice that were pretreated with anti-Fas antibody toinduce acute liver damage. GFP+ cells were found in recipientliver at 8 weeks after transplantation (data not shown) andGFP+ cells in the recipient liver were more frequently foundafter 36 weeks (Fig. 8A,B). These GFP+ cells expressedalbumin (Fig. 8C), indicating that Dlk+ cells can differentiateto hepatocytes in vivo under these experimental conditions.

By contrast, no GFP+ cells were found to express CK19,suggesting that this protocol did not produce a condition toreplace recipient BECs.

DiscussionCell surface antigens are useful not only for identification butalso for isolation of a particular type of cells and numeroussuch antigens have been identified in hematopoietic cells. Onthe contrary, very few such antigens have been known inhepatic cells and hepatoblasts have not been well defined bythe expression of cell surface antigens, making identification,isolation and characterization of hepatoblasts difficult. Weapplied the signal sequence trap to identify cell surfaceantigens of fetal hepatic cells and found that Dlk is highlyexpressed in fetal immature hepatocytes. Dlk was previouslyshown to be expressed in fetal tissues including liver (Smasand Sul, 1993); however, its expression profile duringdevelopment had been unknown. We found that Dlk wasexpressed in immature hepatocytes as early as E10.5 and itsstrong expression continued until E16.5. In contrast, Dlkexpression was undetectable in neonatal and adult liver.

We demonstrated that Dlk+ cells isolated from E14.5 liverwere able to proliferate to form colonies in 5 days of culturein the presence of HGF and EGF. More than 60% of suchcolonies contained both albumin+ and CK19+ cells. Thus,E14.5 Dlk+ cells are mostly, if not entirely, hepatoblasts.Furthermore, since 10% of colony-forming Dlk+ cells formedlarge colonies that contained over 100 cells, highlyproliferative hepatoblasts are enriched in the Dlk+ population.Since the vast majority of Dlk+ cells expressed albumin, andalmost all the adherent cells were albumin+ after 1 day ofculture, colonies were mostly formed from Dlk+albumin+ cells.We also examined the ability of Dlk+ cells at different stagesof gestation to proliferate and to differentiate. Dlk+ cells atE12.5 and E16.5 stages contained highly proliferative andbipotential hepatoblasts, while E18.5 Dlk+ cells lost highlyproliferative potential. As the expression of Dlk in each cellwas also downregulated at E18.5 (data not shown), the Dlklevel may be correlated with the growth potential of Dlk+ cells.We also investigated the differentiation of Dlk+ cells in vivo bytransplantation of Dlk+ cells into the recipient mice treatedwith anti-Fas antibody, Jo2, which induces apoptosis inhepatocytes. In this model system, we detected donor-derived

hepatocytes but not BECs. The inability to demonstratedifferentiation of hepatic progenitors to BECs seems to bea common problem for in vivo transplantation assaysbecause transplanted progenitor cells differentiated intoonly hepatocytes in many cases using mice as recipientanimals (Suzuki et al., 2000; Lagasse et al., 2000; Chinzeiet al., 2002; Forbes et al., 2002; Malhi et al., 2002). Oneexception is that transplanted CD45–TER119–c-Kit–c-Met+CD49f+/lo cells formed bile-duct like structures;however, such structures were found only in the spleen butnot in liver (Suzuki et al., 2002). Thus, it is possible thatthe protocol we used did not create a condition thatallowed the transplanted cell to replace the recipient BECs.However, the possibility remains that Dlk+ cells at E14.5have lost the ability to become BECs in vivo.

Recent studies indicated that hepatoblasts could beenriched by using FACS: mouse hepatoblasts were

Journal of Cell Science 116 (9)

Fig. 8.Engraftment of Dlk+ cells in recipient liver. E14.5 Dlk+ cellsisolated from GFP transgenic mice were transplanted intrasplenically intorecipient mice injured by intraperitoneal administration of anti-Fasantibody, Jo2. Thirty six weeks after transplantation, frozen sections ofrecipient liver were made. (A) GFP+ cells were detected in liverparenchyma (green). Higher magnification of the box in A is shown in B.(C) The frozen section was stained with anti-albumin antibody. GFP+ cellsexpressed albumin (red in C). (D) Overlay image of GFP, albumin (red),and DAPI (blue).

Fig. 7.Electron microscopic analysis of Dlk+ cells before and afterthe culture. Dlk+ cells isolated from E14.5 fetal liver and the cellscollected from the low density culture were fixed and used forpreparing thin sections for transmission-electron microscopy.(A) E14.5 Dlk+ cells exhibited a round-shaped nucleus, smallnucleolus and elongated mitochondria. (B) Cells cultured at lowdensity showed characteristics distinct from freshly isolated cells. Atypical type of cells is shown. There are many microvilli on thesurface, a cleaved nucleus, larger nucleolus, and many round-shapedmitochondria in the cytoplasm. Bars, 2 µm.

1785Isolation of Dlk/Pref-1+ hepatoblasts

enriched in the CD45–TER119–c-Kit–CD29+CD49f+ fraction(Suzuki et al., 2000) and were further enriched in theCD45–TER119–c-Kit–c-Met+CD49f+/lo fraction (Suzuki et al.,2002), and rat hepatoblasts were present inRT1A1–OX18loICAM-1+ cells (Kubota and Reid, 2000). In thepresent study, we demonstrated that hepatoblasts are enrichedin the Dlk+ population of mouse fetal hepatic cells. Flowcytometric analysis indicated that Dlk+ cells wereCD45–TER119– and cKit–. Although both Dlk+ andCD45–TER119–c-Kit–c-Met+CD49f+/lo cells contained highlyproliferative and bipotential cells, there is a substantialdifference in abundance between Dlk+ cells andCD45–TER119–c-Kit–c-Met+CD49f+/lo cells. Dlk+ cellsconstituted about 10% of E14.5 total fetal hepatic cells, whileCD45–TER119–c-Kit–c-Met+CD49f+/lo constituted only 0.3%of them. The potential for differentiation also highlights asignificant difference between Dlk+ cells and CD45–TER119–c-Kit–CD29+CD49f+ or CD45–TER119–c-Kit–c-Met+CD49f+/lo

cells. For example, a longer incubation time was required forCD45–TER119–c-Kit–c-Met+CD49f+/lo cells to differentiateinto hepatocytes and BECs: 70% of large colonies derived fromDlk+ cells contained albumin+ and CK19+ cells only after 5days of culture (Fig. 6), while 80% of large colonies derivedfrom CD45–TER119–c-Kit–c-Met+CD49f+/lo cells containedboth albumin+ and CK19+ cells only after 21 days of culture.Thus, it is likely that CD45–TER119–c-Kit–c-Met+CD49f+/lo

cells represent more immature cells than Dlk+ cells. RatRT1A1–OX18loICAM-1+ hepatoblasts formed coloniescontaining albumin+ and CK19+ cells after 5 days of culture.Although they were derived from different animals, mouse Dlk+

cells and rat RT1A1–OX18loICAM-1+ cells appear to exhibitsimilar characteristics as a progenitor of hepatocytes and BECs.

The result that Dlk is expressed specifically in fetal liversuggests that Dlk might be implicated in proliferation and/ordifferentiation of hepatocytes. Consistent with this idea, Dlk,also known as Pref-1, was previously shown to be involved indifferentiation of pre-adipotcytes as overexpression of Dlkresulted in inhibition of adipogenesis (Smas et al., 1997; Smasand Sul, 1993). It was also reported that Dlk expressionincreased when proliferation of pancreatic β cells reached amaximum (Carlsson et al., 1997). To test whether Dlk mighthave a role for hepatic differentiation in a manner similarto adipocyte differentiation, we expressed Dlk in the fetalhepatocyte primary culture by using a retrovirus vector.However, expression of hepatic differentiation marker geneswas not altered (data not shown). Another possibility is thatDlk is involved in hematopoiesis. Moore et al. reported thatDlk was expressed in fetal liver stroma cells that were able tosupport hematopoiesis (Moore et al., 1997). In addition, thereare reports that Dlk modulated proliferation of thymocytes(Kanata et al., 2000), fetal liver hematopoietic cells (Ohno etal., 2001) and pre-B cells (Bauer et al., 1998). Interestingly,fetal liver hematopoiesis is most active in mid-gestation whenDlk is strongly expressed, which suggests that Dlk is involvedin hematopoiesis. However, recent studies on Dlk deficientmouse show that the mutant mice are viable without apparentdefects in liver formation and hematopoiesis, while they showgrowth abnormality and altered lipid metabolism (Moon et al.,2002). Thus, roles of Dlk in hematopoiesis as well as liverdevelopment still remain elusive and await furtherinvestigation.

In this study, we successfully isolated Dlk+ cells anddemonstrated that hepatoblasts are abundant in E14.5 Dlk+

cells. Using the present method, 95% pure Dlk+ cells can beeasily isolated from total fetal hepatic cells using AutoMACS.In addition, since Dlk expression in fetal liver is similarbetween mouse and human, the present method may beapplicable for the isolation of human hepatoblasts.

We are grateful to Minoru Tanaka, Akihide Kamiya, KojiNakamura, Hiroyuki Yanai and Hiroko Anzai for technical assistanceand helpful discussion. We thank Takashi Sekiguchi for cell sortingby using FACSvantage and Minoru Kakeda for preparation of themonoclonal antibody against Dlk. This work was supported in part byGrants-in-Aid for Scientific Research and Special Coordination Fundfor Promoting Science and Technology from the Ministry ofEducation, Culture, Sports, Science and Technology, a research grantfrom the Ministry of Health, Labour and Welfare, the JapanGovernment and CREST of Japan Science and Technology.

ReferencesBauer, S. R., Ruiz-Hidalgo, M. J., Rudikoff, E. K., Goldstein, J. and

Laborda, J. (1998). Modulated expression of the epidermal growth factor-like homeotic protein dlk influences stromal-cell-pre-B-cell interactions,stromal cell adipogenesis, and pre-B-cell interleukin-7 requirements. Mol.Cell Biol. 18, 5247-5255.

Carlsson, C., Tornehave, D., Lindberg, K., Galante, P., Billestrup, N.,Michelsen, B., Larsson, L.-I. and Nielsen, J. H. (1997). Growth hormoneand prolactin stimulate the expression of rat preadipocyte factor-1/delta-likeprotein in pancreatic islets: Molecular cloning and expression pattern duringdevelopment and growth of the endocrine pancreas. Endocrinology138,3940-3948.

Chinzei, R., Tanaka, Y., Shimizu-Saito, K., Hara, Y., Kakinuma, S.,Watanabe, M., Teramoto, K., Arii, S., Takase, K., Sato, C. et al. (2002).Embryoid-body cells derived from a mouse embryonic stem cell line showdifferentiation into functional hepatocytes. Hepatology36, 22-29.

Clotman, F., Lannoy, V. J., Reber, M., Cereghini, S., Cassiman, D.,Jacquemin, P., Roskams, T., Rousseau, G. G. and Lemaigre, F. P. (2002).The onecut transcription factor HNF6 is required for normal developmentof the biliary tract. Development129, 1819-1828.

Coffinier, C., Gresh, L., Fiette, L., Tronche, F., Schutz, G., Babinet, C.,Pontoglio, M., Yaniv, M. and Barra, J. (2002). Bile system morphogenesisdefects and liver dysfunction upon targeted deletion of HNF1β.Development129, 1829-1838.

Fahrenkrug, S. C., Freking, B. A. and Smith, T. P. (1999). Genomicorganazation and genetic mapping of the bovine Pref-1 gene. Biochem.Biophys. Res. Commun. 264, 662-667.

Fausto, N. (1994). Liver stem cells. In The Liver: Biology and Pathobiology(ed. I. M. Arias, J. L. Boyer, N. Fausto, W. B. Jakoby, D. A. Schachter andD. A. Shafritz), pp. 1501-1518. New York: Raven Press.

Forbes, S., Vig, P., Poulsom, R., Thomas, H. and Alison, M. (2002). Hepaticstem cells. J. Pathol. 197, 510-518.

Fujio, K., Evarts, R. P., Hu, Z., Marsden, E. R. and Thorgeirsson, S. S.(1994). Expression of stem cell factor and its receptor, c-kit, during liverregeneration from putative stem cells in adult rat. Lab. Invest. 70, 511-516.

Greengard, O. (1970). The developmental formation of enzymes in rat liver.In Biochemical Actions of Hormones(ed. G. Litwack), pp. 53-87. New York:Academic Press.

Gualdi, R., Bossard, P., Zheng, M., Hamada, Y., Coleman, J. R. and Zaret,K. S. (1996). Hepatic specification of the gut endoderm in vitro: cellsignaling and transcriptional control. Genes Dev. 10, 1670-1682.

Haber, B. A., Chin, S., Chuang, E., Buikhuisen, W., Naji, A. and Taub, R.(1995). High levels of glucose-6-phosphatase gene and protein expressionreflect an adaptive response in proliferating liver and diabetes. J. Clin. Invest.95, 832-841.

Halder, S. K., Takemori, H., Hatano, O., Nonaka, Y., Wada, A. andOkamoto, M. (1998). Cloning of a membrane-spanning protein withepidermal growth factor-like repeat motifs from adrenal glomerulosa cells.Endocrinology139, 3316-3328.

Helman, L. J., Sack, N., Plon, S. E. and Israel, M. A. (1990). The sequenceof an adrenal specific human cDNA, pG2. Nucleic Acid Res. 18, 685.

1786

Holic, N., Suzuki, T., Corlu, A., Couchie, D., Chobert, M. N., Guguen-Guillouzo, C. and Laperche, Y. (2000). Differential expression of the ratgamma-glutamyl transpeptidase gene promoters along with differentiationof hepatoblasts into biliary or hepatocytic lineage. Am. J. Pathol. 157, 537-548.

Houssaint, E. (1980). Differentiation of the mouse hepatic primordium. I. Ananalysis of tissue interactions in hepatocyte differentiation. CellDifferentiation9, 269-279.

Jensen, C. H., Krogh, T. N., Hojrup, P., Clausen, P. P., Skjodt, K., Larsson,L.-I., Enghild, J. J. and Teisner, B. (1994). Protein structure of fetalantigen 1. A novel circulating human epidermal-growth-factor-like proteinexpressed in neuroendocrine tumors and its relation of the gene products ofdlk and pG2. Eur. J. Biochem. 225, 83-92.

Jung, J., Zheng, M., Goldfarb, M. and Zaret, K. S. (1999). Initiation ofmammalian liver development from endoderm by fibroblast growth factors.Science284, 1998-2003.

Kamiya, A., Kinoshita, T., Ito, Y., Matsui, T., Morikawa, Y., Senda, E.,Nakashima, K., Taga, T., Yoshida, K., Kishimoto, T. et al. (1999). Fetalliver development requires a paracrine action of oncostatin M through thegp130 signal transducer. EMBO J. 18, 2127-2136.

Kanata, M., Osawa, M., Osawa, M., Sudo, K., Nakauchi, H., Farr, A. G.and Takahama, Y. (2000). A role for Pref-1 and HES-1 in thymocytedevelopment. J. Immunol. 164, 256-264.

Kojima, T. and Kitamura, T. (1999). Retrovirus-mediated signal sequencetrap based on a constitutively active cytokine receptor. Nat. Biotech. 17, 487-490.

Kubota, H. and Reid, L. M. (2000). Clonogenic hepatoblasts, commonprecursors for hepatocytic and biliary lineages, are lacking classical majorhistocompatibility complex class I antigen. Proc. Natl. Acad. Sci. USA97,12132-12137.

Laborda, J., Sausville, E. A., Hoffman, T. and Notario, V. (1993). dlk, aputative mammalian homeotic gene differentially expressed in small celllung carcinoma and neuroendocrine tumor cell line. J. Biol. Chem. 268,3817-3820.

Lagasse, E., Connors, H., Al-Dhalimy, M., Reitsma, M., Dohse, M.,Osborne, L., Wang, X., Finegold, M., Weissman, I. L. and Grompe, M.(2000). Purified hematopoietic stem cells can differentiate into hepatocytesin vivo. Nat. Med. 6, 1229-1234.

Lemire, J. M., Shiojiri, N. and Fausto, N. (1991). Oval cell proliferation andthe origin of small hepatocytes in liver injury induced by D-galactosamine.Am. J. Pathol. 139, 535-552.

Malhi, H., Irani, A. N., Gagandeep, S. and Gupta, S. (2002). Isolation ofhuman progenitor liver epithelial cells with extensive replication capacityand differentiation into mature hepatocytes. J. Cell Sci. 115, 2679-2688.

Mitaka, T., Sato, F., Mizuguchi, T., Yokono, T. and Mochizuki, Y. (1999).Reconstruction of hepatic organoid by rat small hepatocytes and hepaticnonparenchymal cells. Hepatology29, 111-125.

Moon, Y. S., Smas, C. M., Lee, K., Villena, J. A., Kim, K.-H., Yun, E. J.and Sul, H. S. (2002). Mice lacking paternally expressed Pref-1/Dlk1display growth retardation and accelated adiposity. Mol. Cell Biol. 22, 5585-5592.

Moore, K. A., Pytowski, B., Witte, L., Hicklin, D. and Lemischka, I. R.(1997). Hematopoietic activity of a stromal cell transmembrane proteincontaining epidermal growth factor-like repeat motifs. Proc. Natl. Acad. Sci.USA 94, 4011-4016.

Oh, S.-H., Miyazaki, M., Kouchi, H., Inoue, Y., Sakaguchi, M., Tsuji, T.,Shima, N., Higashio, K. and Namba, M. (2000). Hepatocyte growth factorinduces differentiation of adult rat bone marrow cells into hepatocyte lineagein vitro. Biochem. Biophys. Res. Commun. 279, 500-504.

Ohno, N., Izawa, A., Hattori, M., Kageyama, R. and Sudo, T. (2001). dlkinhibits stem cell factor-induced colony formation of murine hematopoieticprogenitors: Hes-1-independent effect. Stem Cells19, 71-79.

Omori, N., Omori, M., Evarts, R. P., Teramoto, T., Miller, M. J., Hoang,

T. N. and Thorgeirsson, S. S. (1997). Partial cloning of rat CD34 cDNAand expression during stem cell-dependent liver regeneration in the adultrat. Hepatology26, 720-727.

Petersen, B. E., Goff, J. P., Greenberger, J. S. and Michalopoulos, G. K.(1998). Hepatic oval cells express the hematopoietic stem cell marker Thy-1 in the rat. Hepatology27, 433-445.

Petersen, B. E., Bowen, W. C., Patrene, K. D., Mars, W. M., Sullivan, A.K., Murase, N., Boggs, S. S., Greenberger, J. S. and Goff, J. P. (1999).Bone marrow as a potential source of hepatic oval cells. Science284, 1168-1170.

Ponder, K. P., Gupta, S., Leland, F., Darlington, G., Finegold, M., DeMayo,J., Ledley, F. D., Chowdhury, J. R. and Woo, S. L. (1991). Mousehepatocytes migrate to the liver parenchyma and function indefinitely afterintrasplenic transplantation. Proc. Natl. Acad. Sci. USA88, 1217-1221.

Rao, M. S., Dwivedi, R. S., Yeldandi, A. V., Subbarao, V., Tan, X. D.,Usman, M. I., Thangada, S., Nemali, M. R., Kumar, S. and Scarpelli, D.G. et al. (1989). Role of periductal and ductular epithelial cells of the adultrat pancreas in pancreatic hepatocyte lineage. A change in the differentiationcommitment. Am. J. Pathol. 134, 1069-1086.

Reddy, J. K., Rao, M. S., Yeldandi, A. V., Tan, X. D. and Dwivedi, R. S.(1991). Pancreatic hepatocytes. An in vivo model for cell lineage in pancreasof adult rat. Dig. Dis. Sci. 36, 502-509.

Sell, S. (1994). Liver stem cells. Mod. Pathol. 7, 105-112.Sell, S., Osborn, K. and Leffert, H. (1981). Autoradiography of “oval cells”

appearing rapidly in the livers of rats fed N-2-fluorenylacetamide in acholine devoid diet. Carcinogenesis2, 7-14.

Shiojiri, N. (1984). The origin of intrahepatic bile duct cells in the mouse. J.Embryol. Exp. Morphol. 79, 25-39.

Shiojiri, N. (1997). Development and differentiation of bile ducts in themammalian liver. Microsc. Res. Tech. 39, 328-335.

Shiojiri, N., Lemire, J. M. and Fausto, N. (1991). Cell lineages and oval cellprogenitors in rat liver development. Cancer Res. 51, 2611-2620.

Shiojiri, N., Inujima, S., Ishikawa, K., Terada, K. and Mori, M. (2001).Cell lineage analysis during liver development using the spfash-heterozygousmouse. Lab. Invest. 81, 17-25.

Smas, C. M. and Sul, H. S. (1993). Pref-1, a protein containing EGF-likerepeats, inhibits adipocyte differentiation. Cell 73, 725-734.

Smas, C. M., Chen, L. and Sul, H. S. (1997). Cleavage of membrane-associated pref-1 generates a soluble inhibitor of adipocyte differentiation.Mol. Cell Biol. 17, 977-988.

Suzuki, A., Zheng, Y.-E., Kondo, R., Kusakabe, M., Takada, Y., Fukao,K., Nakauchi, H. and Taniguchi, H. (2000). Flow-cytometric separationand enrichment of hepatic progenitor cells in the developing mouse liver.Hepatology32, 1230-1239.

Suzuki, A., Zheng, Y., Kaneko, S., Onodera, M., Fukao, K., Nakauchi, H.and Taniguchi, H. (2002). Clonal identification and characterization of self-renewing pluripotent stem cells in the developing liver. J. Cell Biol. 156,173-184.

Tateno, C., Takai-Kajihara, K., Yamasaki, C., Sato, H. and Yoshizato, K.(2000). Heterogeneity of growth potential of adult rat hepatocytes in vitro.Hepatology31, 65-74.

Tosh, D., Shen, C. N. and Slack, J. M. (2002). Differentiated properties ofhepatocytes induced from pancreatic cells. Hepatology36, 534-543.

Wang, X., Montini, E., Al-Dhalimy, M., Lagasse, E., Finegold, M. andGrompe, M. (2002). Kinetics of liver repopulation after bone marrowtransplantation. Am. J. Pathol. 161, 565-574.

Zaret, K. (2000). Liver specification and early morphogenesis. Mech. Dev. 92,83-88.

Zulewski, H., Abraham, E. J., Gerlach, M. J., Daniel, P. B., Moritz, W.,Muller, B., Vallejo, M., Thomas, M. K. and Habener, J. F. (2001).Multipotential nestin-positive stem cells isolated from adult pancreatic isletsdifferentiate ex vivo into pancreatic endocrine, exocrine, and hepaticphenotypes. Diabetes50, 521-533.

Journal of Cell Science 116 (9)