, . 184: 272–278 (1998)

EXPRESSION OF HNF-1á AND HNF-1â IN VARIOUSHISTOLOGICAL DIFFERENTIATIONS OF

HEPATOCELLULAR CARCINOMA

1, 1,* 1,2, 1,3, 4, 4 1

1First Division of Pathology, Kobe University School of Medicine, Kobe, Japan2Second Division of Internal Medicine, Kobe University School of Medicine, Kobe, Japan

3First Division of Surgery, Kobe University School of Medicine, Kobe, Japan4Department of Medical Biochemistry, University of Calgary, Calgary, Canada

SUMMARY

Hepatic nuclear factor 1 (HNF-1) regulates genes in a hepatocyte-specific manner. It has been previously reported that the ratio ofHNF-1á and HNF-1â mRNA is related to histological differentiation hepatocellular carcinoma (HCC). In this study, the expressionlevels of the HNF-1á and HNF-1â proteins were analysed relatively and quantitatively in various histologically differentiated HCC andsurrounding non-cancerous tissues, and HNF-1á binding activity for the AT element of the B domain of the human á-fetoproteinenhancer was examined. Western blot analysis demonstrated that HNF-1á protein was expressed at a higher level in well-differentiatedHCC tissues than in the surrounding non-HCC tissues; on the other hand, the HNF-1á protein was expressed at lower levels inmoderately and poorly differentiated HCCs than in the surrounding non-HCC tissues. The levels of HNF-1â expression inwell-differentiated and poorly differentiated HCCs were similar to and higher than those found in the respective surroundingnon-cancerous portions. In binding assays, HNF-1 binding activity was high in well-differentiated HCC and lower in moderately andpoorly differentiated HCCs. Most well-differentiated HCC cases showed immunohistochemical expression of HNF-1á. These findingsshow that poor histological differentiation of HCC correlates with decreases in the level and activity of HNF-1á proteins. ? 1998 JohnWiley & Sons, Ltd.

J. Pathol. 184: 272–278, 1998.

KEY WORDS—HNF-1á; HNF-1â; differentiation; hepatocellular carcinoma

INTRODUCTION

Transcription factor HNF-1 (hepatic nuclear factor 1)binds and regulates the promoters or enhancers of genesthat are expressed almost exclusively in the liver, such asá- and â-fibrinogen,1 albumin,2,3, á-fetoprotein (AFP),4á-antitrypsin,5 aldolase B, etc.6–9 Two related HNF-1factors, HNF-1á and HNF-1â, form homodimers orheterodimers and bind to these sequences in the liver-specific promoters and enhancers.10,11 HNF-1á andHNF-1â protein contain similar homologous homeo-domains and POU (pit1, oct1, oct2, and unc86)domains, but the two transactivation domains ofHNF-1á in the C-terminal part of the protein are notpresent in HNF-1â.12,13 Analysis of the expression ofHNF-1á and HNF-1â mRNA in hepatoblastoma andhepatocellular carcinoma (HCC) tissues by competitivereverse transcriptional-polymerase chain reaction (RT-PCR) assay showed that the ratio of HNF-1á/HNF-1âmRNA is closely linked to histological differentiation ofHCC.14 The changes in the levels of HNF-1á andHNF-1â protein, however, remained unknown. In this

study, we analysed HNF-1á and HNF-1â expression atthe protein level in relation to cell differentiation in HCCtissues. The amounts of HNF-1á and HNF-1â proteinsand HNF-1á human AFP enhancer binding activity invarious histologically differentiated HCCs, as well as intheir surrounding non-cancerous tissues, were analysedand compared.

MATERIALS AND METHODS

Tissue samples

Primary HCC and surrounding non-cancerous livertissues were obtained from 28 cases of surgical resectionand autopsy. Immediately after separation, part of theHCC and non-cancerous tissue was snap-frozen andstored at "80)C and part was embedded and stainedwith haematoxylin and eosin (H & E) and Masson’strichrome for histological analysis. Histological gradesof HCC were assigned according to the Edmondson andSteiner classification.

Production of human polyclonal antibodies to HNF-1áand HNF-1â

The 3*-terminal fragments of human HNF-1á andHNF-1â cDNA, without homology in their sequences,

*Correspondence to: Yoshitake Hayashi, MD PhD The FirstDivision of Pathology, Kobe University School of Medicine, 7-5-1Kusunoki-cho, chuo-ku, Kobe 650, Japan.Contract grant sponsor: Ministry of Education, Science, Sports and

Culture, Japan; Contract grant numbers: C05807014; B07457050;08877199.

CCC 0022–3417/98/030272–07 $17.50 Received 15 April 1997? 1998 John Wiley & Sons, Ltd. Accepted 22 September 1997

were each inserted into EcoRI sites of vector pGEX-2T(Pharmacia). Glutathione S-transferase (GST) proteinand GST fusion protein containing the COOH-terminalregion of HNF-1á or HNF-1â, respectively, wereobtained from 0·5–2·0 litres of Escherichia coli cultureand purified by binding to glutathione-Sepharose beads(Pharmacia). Rabbits were injected with GST fusionprotein and polyclonal antibodies were affinity-purifiedwith CNBr-activated Sepharose beads (Pharmacia)coupled to the GST–HNF-1á or GST–HNF-1â proteinas described by Noguchi.15

Preparation of nuclear extracts

From 1–3 g of fresh frozen tissue was mincedand washed in 5 volumes of hypotonic buffer [10 mHepes (pH 7·9), 10 m KCl, 0·1 m EDTA, 0·75 mspermidine, 0·15 m spermine, and 0·5 m DTT]. After10 min of incubation on ice, tissues were pelleted. Theywere resuspended in 2 volumes of hypotonic buffercontaining 1 ìg/ml aprotinin and 1 m phenylmethyl-sulphonyl fluoride (PMSF), and homogenized at 10strokes in a glass Dounce homogenizer. The homoge-nate was centrifuged at 10 000 rpm for 30 s. Nuclearproteins were extracted by resuspending the nuclearpellets in ten packed nuclei volumes of lysis buffer[20 m Hepes (pH 7·9), 25 per cent glycerol, 0·5 NaCl,1·5 m MgCl2, 0·2 m EDTA, 0·5 m DTT, 1 ìg/mlaprotinin, and 0·5 m PMSF] and the suspension wasrotated at 4)C for 30 min and then centrifuged at30 000 rpm for 30 min. The supernatant was collectedand solid ammonium sulphate (0·4 g/ml) was added.After rotation for 30 min at 4)C, the precipitatedproteins were collected by centrifugation at 35 000 rpmfor 30 min. The pellets were dissolved in dialysis buffer[20 m Hepes (pH 7·9), 20 per cent glycerol, 0·1 KCl,0·2 m EDTA, 0·5 m DTT, 1 ìg/ml aprotinin, and0·5 m PMSF] and dialysed against 100 volumes ofdialysis buffer, with one change of buffer.

Western blot and densitometric quantitation

Fifty micrograms of nuclear protein was separated on10 per cent SDS-PAGE and the blot was incubatedovernight with 1:500 polyclonal antibody for HNF-1á orHNF-1â; horseradish peroxidase-conjugated rabbit anti-bodies to immunoglobin G were used to visualize boundHNF-1á or HNF-1â with the ECL system (Amersham).The HNF-1á or HNF-1â protein detected was quanti-tated by densitometry with the NIH Image 1.60 softwareprogram.

Reverse transcriptional-polymerase chain reaction(RT-PCR) analysis

Total RNA was extracted with ISOGEN (NipponGene Co.) from frozen tissues according to the manu-facturer’s instructions. Single-strand cDNA was reverse-transcribed. PCR was performed on 2·5 ìl of reversetranscriptional single-strand cDNA from each tissuewith the primers 5*CGTGTTTTTCTACAACTGGTTT3* and 3*AGAACTGGACGGGCTGC5*, which

were predicted to amplify 607 bp HNF-1á and 541 bpHNF-1â fragments in a total volume of 50 ìl. Each ofthe 30 PCR cycles consisted of denaturation at 94)C for1 min, annealing at 50)C for 1 min and extension at 72)Cfor 1 min, with a final extension of 10 min at 72)C. PCRproducts were analysed by electrophoresis on a 1 percent agarose gel.

Binding assay

Twenty micrograms of double-stranded oligonucleo-tide of AFP enhancer domain B (gatcTCCTGATTAATAATTACACTA with added gatc protruding ends) waslabelled with 2 U of Klenow DNA polymerase I, 5 ìl of[á-32P]dGTP (3000 Ci/mmol), and 1 m dGTP mix at37)C for 30 min. The nuclear extract was preincubatedwith 5 mg of poly(dI-dC): poly(dI-dC) in binding buffer[10 m Tris–HCl (pH 7·5), 45 m KCl, 2 m MgCl2,1 m EDTA, 1 m DTT, and 8 per cent glycerol] for10 min on ice and then incubated with 2 ìl of DNA(1#104 cpm) end-labelled with [á-32P]dGTP for 20 minat 24)C. The reaction mixture was electrophoresed on a4 per cent polyacrylamide gel. The gel was dried andautoradiographed at "70)C. For competition assaysand upshift experiments, a 100-fold molar excessof competitor probes and 1 ìl polyclonal antibodyfor HNF-1á was preincubated with nuclear extracts for15 min and 1 h, respectively, prior to the addition ofend-labelled probes.

Immunohistochemistry

Frozen sections were fixed in cold acetone. The de-paraffinized sections were heated for 15 min with 0·01 boiling citric acid buffer (pH 6·0) in a microwave oven.As the primary antibodies, rabbit polyclonal antibodiesfor HNF-1á or HNF-1â diluted at 1:20 were reacted at4)C overnight. The biotinylated anti-rabbit secondaryantibody was incubated for 30 min. The peroxidasecolouring reaction was done in 0·05 mol/l Tris–HCl,65 mg/dl sodium azide, and 0·003 per cent hydrogenperoxide. Control sections were obtained either byomission of the primary antibody in frozen sections, orby incubation with preimmune rabbit immunoglobulins.

RESULTS

Character of the polyclonal antibodies for HNF-1á andHNF-1â

By Western blot, purified polyclonal antibody forHNF-1á reacted with 20 ng and 10 ng of GST–HNF-1áfusion protein (lanes 1 and 2) but not with 20 ng of GSTprotein (lane 3) (Fig. 1A). For identical quantities in thecase of HNF-1â, the reaction was positive with thefusion protein (lanes 4 and 5) but negative with GSTprotein (lane 6) (Fig. 1B).

Western blotting analysis

An 80 kD band of HNF-1á present in the nuclearextract from HCCs and non-HCCs was detected by

273HNF-1á AND HNF-1â IN HISTOLOGICAL DIFFERENTIATIONS OF HEPATOCELLULAR CARCINOMA

? 1998 John Wiley & Sons, Ltd. , . 184: 272–278 (1998)

anti-HNF-1á polyclonal antibody. The level of HNF-1áprotein was higher in well-differentiated HCC than inthe surrounding non-cancerous tissue, whereas it waslower in moderately and poorly differentiated HCCs(Fig. 2A). The level of HNF-1â protein (67 kD), on theother hand, was slightly higher in all types of HCC thanin the surrounding non-cancerous tissues (Fig. 2B). Thelevel of HNF-1 protein was quantitated by densito-metric analysis. As shown in Table I, well-differentiatedHCCs show higher levels of HNF-1á protein indicatedby the HCC/non/HCC ratio of HNF-1á (1), while thelower levels of HNF-1á are shown as an HCC/non-HCCratio between 0·23 and 0·85 in 75 per cent of themoderately differentiated HCCs (seven cases) and as afurther decrease to 0·09–0·65 in all poorly differentiatedHCCs. The levels of HNF-1â protein were slightlyelevated with an HCC/non-HCC ratio between 1·03 and1·57 (data not shown).

Reverse transcriptional-polymerase chain reaction(RT-PCR) analysis

RT-PCR amplified a 607 bp fragment derived fromHNF-1á and a 541 bp fragment derived from HNF-1â,

which were distinguishable on a 1 per cent agarose gel.In one well-differentiated HCC, the band of the PCRproduct from HNF-1á was more intense than thatderived from HNF-1â (Fig. 3A). The intensities of thePCR products derived from HNF-1á and HNF-1â werealmost the same in one moderately differentiated HCC(Fig. 3B). Furthermore, the poorly differentiated HCCshowed a band pattern contrary to that of well-differentiated HCC; the intensity of the HNF-1á bandwas much fainter than that of HNF-1â (Fig. 3C). Theseresults reveal that the decreased HNF-1á from well-differentiated to poorly differentiated HCC by Westernblot correlates with the expression patterns of HNF-1áand HNF-1â mRNA, except in cases 10 and 16(Table I).

Binding analysis

Binding assay was done by using a 32P-labelleddouble-stranded oligonucleotide probe corresponding tothe B domain of the human AFP enhancer to whichHNF-1á or HNF-1â homeodimers as well as HNF-1áand HNF-1â heterodimers bind. Figure 4 shows that theamounts of complexes formed by the probes and nuclearextracts were higher in one well-differentiated HCC thanin the corresponding non-cancerous tissue (Fig. 4A) andlower in moderately and poorly differentiated HCCs(Figs 4B and 4C). The addition of a 100-fold molarexcess of unlabelled competitor probes reduced theintensity of the band; the addition of polyclonal anti-body for HNF-1á to the reaction mixture caused anupshift of, and revealed the presence of HNF-1á in, theDNA–protein complex (Fig. 4D).

Immunohistochemistry

The results of HNF-1á and HNF-1â immunostainingin various histological grades of HCC are summarized inTable II. Eight of 12 cases of well-differentiated HCC

Fig. 1—Character of polyclonal antibodies for HNF-1á and HNF-1âby western blot analysis. (A) Purified polyclonal antibody for HNF-1áreacted with 20 ng and 10 ng GST-HNF-1á fusion protein (Lane 1 and2) and 20 ng GST protein (Lane 3). (B) purified polyclonal antibodyfor HNF-1â reacted with 20 ng and 10 ng GST-HNF-1â fusion protein(Lane 4 and 5) and 20 ng GST protein (Lane 6). Arrows showmolecular weight

Fig. 2—Protein levels of HNF-1á (A) and HNF-1â (B) in the well-differentiated (Well diff.), moderately differentiated(Moderately-diff.), and poorly differentiated (Poorly-diff.) HCCs (T) of cases 6, 16, and 24 as well as their respective non-HCCs(N) by Western blot. Controls: nuclear extracts from the huH1/cl-2 cell line positive for HNF-1á, from kidney positive forHNN-1â, and from the hela cell line as a negative control

274 W. WANG ET AL.

? 1998 John Wiley & Sons, Ltd. , . 184: 272–278 (1998)

Table I—Histological features of a series of HCCs and the levels of HNF-1á protein and binding activity

Case His Western blot (T/N) Binding ActivityHNF-1á/HNF-1â

mRNA

1 Ed I 1·81 H A2 Ed I 2·20 H A3 Ed I 2·03 H A4 Ed I 1·76 N A5 Ed I 1·44 H A6 Ed I 2·14 H A7 Ed I 1·10 N A8 Ed I 2·05 H A9 Ed I 2·12 H A10 Ed I 2·68 H M11 Ed I 2·02 N M12 Ed I 1·38 N A

13 Ed II 0·85 N M14 Ed II 0·73 L M15 Ed II 0·90 L M16 Ed II 0·52 L M17 Ed II 0·38 L B18 Ed II 0·23 L M19 Ed II 0·58 L B20 Ed II 1·10 L M

21 Ed III 0·43 L B22 Ed III 0·60 L B23 Ed III 0·25 L B24 Ed III 0·12 L B25 Ed III 0·65 L B26 Ed III 0·38 L B27 Ed IV 0·26 L B28 Ed IV 0·09 L B

Notice. His. Histological Diagnosis. Ed I, Edmondson I (well-differentiated HCC); Ed II, Edmondson II (moderately-differentiated HCC); Ed III, Edmondson III (poorly-differentiated HCC); Ed IV (anaplastic HCC). Western blots wereanalyzed by NIH Image for densitometric analysis and Western (T/N), ratio of HNF-1á protein in HCC (T) to non HCC(N) is given. Binding Activity, H. higher binding activity of HNF-1á in tumour tissue than non-tumour tissue; N. noalteration of binding activity shown in tumour tissue relative to non-tumour tissue; L. lower binding activity of HNF-1áin tumour tissue than non-tumour tissue. HNF-1á/HNF-1â mRNA, A. more abundant transcripts of HNF-1á thanHNF-1â; M. equal amount of HNF-1á and HNF-1â transcripts; B. more abundant transcripts of HNF-1â than HNF-1á(illustrated in Fig. 4).

Fig. 3—Histological differentiation of HCC and RT-PCR. (A) Well-differentiated HCC (case 6), (B) moderately differentiated HCC (case 16), and(C) poorly differentiated HCC (case 24) (H & E) stain, magnification#200) with their respective amplified PCR products. RT-PCR products fromthe huH1/cl-2, Hep3B, and HUH7 cell lines are positive controls and those from the hela cell line are negative

275HNF-1á AND HNF-1â IN HISTOLOGICAL DIFFERENTIATIONS OF HEPATOCELLULAR CARCINOMA

? 1998 John Wiley & Sons, Ltd. , . 184: 272–278 (1998)

showed positive staining for HNF-1á; one of these eightcases stained positive for both HNF-1á and HNF-1â.Three of eight cases of moderately differentiated HCCstained weakly; one for HNF-1á only and two forHNF-1â. Of the eight poorly differentiated HCCs, twocases positive for HNF-1â were detected.Figure 5A shows one well-differentiated HCC grow-

ing in a trabecular pattern; tumour cells reacted with theantibodies for HNF-1á, and the expressing sites ofHNF-1á antigen were the nuclei of tumour cells. Stain-ing intensity was heterogeneous, varying from cell tocell. Tumour cells in this case did not react with anti-bodies for HNF-1â (not shown). Figure 5B shows weak

staining for HNF-1â in the nuclei of one poorly differ-entiated HCC. No staining for HNF-1á antigens wasdetected (not shown).

DISCUSSION

It is generally accepted that hepatocyte differentiationis linked to the expression of liver-specific proteins andthat the expression patterns are controlled primarilyat their transcription levels. The HNF-1 family playsa dominant role in liver-specific transcription. Theyinteract with critical regulatory sequences of several

Fig. 4—HNF-1 binding activity for the B domain of human AFP enhancer in various histologicaldifferentiations of HCC. Unlabelled competitor probes absent (") and present (+). Nuclear extractfrom (A) well-differentiated HCC (case 6), (B) moderately differentiated HCC (case 16), (C) poorlydifferentiated HCC (case 24) and their respective non-cancerous tissues. (D) The DNA–proteincomplex band was upshifted by addition of HNF-1á antibody (arrows)

Table II—Expression of HNF-1á and HNF-1â in various histological differentiations of HCC

Numbers of caseHNF-1á positivitynumber (rate)

HNF-1â positivitynumber (rate)

Well-differentiated HCC 12 8 (66%) 1 (8·3%)Moderately-differentiated HCC 8 1 (12%) 2 (25%)Poorly-differentiated HCC 8 0 (0%) 2 (25%)

Fig. 5—(A) HNF-1á immunohistochemistry showing nuclear immunostaining in the well-differentiated HCC, where the elevated level ofHNF-1á protein was confirmed by Western blot. (B) HNF-1â immunohistochemistry showing some nuclear staining in the poorlydifferentiated HCC

276 W. WANG ET AL.

? 1998 John Wiley & Sons, Ltd. , . 184: 272–278 (1998)

liver-specific gene promoters and enhancers such asalbumin and AFP, and the single HNF-1 site of thealbumin promoter is sufficient for tissue-specific expres-sion after transfection.16 Early study has shown theimportance of HNF-1á/HNF-1â in determining thehistological differentiation of HCC.14In the present study, the difference in HNF-1á and

HNF-1â transcription factors between HCC and non-cancerous liver tissues and its effect on the differentia-tion of hepatocytic neoplasms was investigated; Westernblot revealed that in relation to the surrounding non-HCC tissue, the level of HNF-1á protein was higher inwell-differentiated HCC than in poorly differentiatedHCC; the levels of HNF-1â protein were similar in well-and poorly differentiated HCC, but higher than in thesurrounding non-cancerous portions. The importance ofthe HNF-1á/HNF-1â mRNA ratio in determining thedifferentiation of hepatocyte neoplasms was also ident-ified by RT-PCR, in that there were more HNF-1á thanHNF-1â transcripts in well-differentiated HCC, butfewer HNF-1á than HNF-1â transcripts in poorly dif-ferentiated HCC. The dedifferentiation of hepatocyteswas characterized by a reduced level of some liver-enriched protein mRNAs,17 leading to our postulationthat the loss of HNF-1á expression is responsible for thededifferentiation from well-differentiated to poorly dif-ferentiated HCC. Western blot showed a remarkablyelevated level of HNF-1á protein in well-differentiatedHCC, which corresponded to previous results ofNorthern blotting showing HNF-1á mRNA over-expression in well-differentiated HCCs.14 It has beenshown in rat liver embryogenesis that HNF-1á isexpressed preferentially during early liver developmentand decreases in expression in adult liver.18 It may beassumed that the high level of HNF-1á protein inwell-differentiated HCC results from hepatocytes at anearly stage of proliferation and differentiation. In bind-ing assay, a decreasing tendency was shown in the assayof HNF-1 binding activity from well-differentiated topoorly differentiated HCCs. Since the Western blottingresults showed little difference in the abundance ofHNF-1â in various histological differentiations of HCC,the differences of HNF-1 binding activity observed inFig. 4 are likely to be due to the different levels ofHNF-1á protein in these tissues. More than 80 per centof these cases manifested a higher binding activity inwell-differentiated HCCs. Nearly all moderately differ-entiated and all poorly differentiated cases showed alower binding activity (Table I). Two cases of well-differentiated HCC (cases 4 and 11) showed higher levelsof HNF-1á protein in cancerous tissues, while HNF-1binding activity in HCCs was similar to that in non-HCC tissue. The same occurred in two cases of moder-ately differentiated HCC (cases 15 and 20), where nosignificant change of HNF-1á proteins was shown, whileHNF-1 binding activity decreased. This may be relatedto the altered function of HNF-1á initiated by mutationin carcinogenesis.HNF-1â is expressed in hepatic endoderm of foregut

during embryonic development10 and precedes HNF-1áexpression. We assume that HNF-1â possibly maintainsits expression, albeit with a transient increase of

HNF-1â protein, accompanied by hepatocyte prolifer-ation in HCC. This hypothesis is compatible with ourobservation that the levels of HNF-1â protein weresimilar in well-differentiated and poorly differentiatedHCC, but slightly higher than in the surroundingnon-cancerous tissues.The data that have been accumulated so far on

changes of hepatic nuclear factors are derived froman experimental model of hepatocarcinogenesis.19–21HNF-1á mRNA decreased following HNF-4 mRNA, afactor transactivating HNF-1á and HNF-1â mRNAshowed unchanged expression in a mouse livertumour.19 The findings in this study of the detailedhistological classification of human HCC indicatethat the amounts of HNF-1á increase when non-cancerous liver develops into well-differentiated HCC,but decreases when HCC progresses from well-differentiated to poorly differentiated. The expressionof HNF-1á changes with oncogenesis and histologicaldifferentiation.

ACKNOWLEDGEMENTS

We are grateful to Dr Kimio Hashimoto (Division ofPathology, Nishi-Kobe Medical Center) and Dr TeruakiOka (Department of Pathology, Tokyo University) forproviding the HCC and liver tissues. This work wassupported in part by Grants-in-Aid for ScientificResearch (C05807014 and B07457050) and ExploratoryResearch (08877199) from the Ministry of Education,Science, Sports and Culture, Japan.

REFERENCES

1. Courtois G, Morgan JG, Campbell LA, Fourel G, Crabtree GR. Inter-action of a liver-specific nuclear factor with the fibrinogen and alpha-antitrypsin promoters. Science 1987; 238: 688–692.

2. Cereghini S, Blumenfeld M, Yaniv M. A liver-specific factor essential foralbumin transcription differs between differentiated and dedifferentiated rathepatoma cells. Genes Dev 1988; 2: 957–974.

2. Lichtsteiner S, Schibler U. A glycosylated liver-specific transcription factorstimulates transcription of albumin gene. Cell 1989; 57: 1179–1187.

4. Feuerman MH, Godbout R, Ingram RS, Tilghman SM. Tissue-specifictranscription of the mouse á-fetoprotein gene promoter is dependent onHNF-1. Mol Cell Biol 1989; 9: 4204–4212.

5. Hardon E, Paonessa G, Frain M, Cortese R. Two distinct factors interactwith the promoter regions of several liver-specific genes. EMBO J 1988; 7:1711–1719.

6. Vaulont S, Puvenat N, Levrat F, Cognet M, Kahn A, Raymondjean M.Proteins binding to the liver-specific pyruvate kinase gene promoter. J MolBiol 1989; 209: 205–219.

7. Costa RH, Lai E, Grayson DR, Darnell JE. The cell-specific enhancer of themouse transthyretin (prealbumin) gene binds a common factor at one siteand a liver-specific factor(s) at two other sites.Mol Cell Biol 1988; 8: 81–90.

8. Tsutsumi K, Ito K, Ishikawa K. Developmental appearance of transcriptionfactors that regulate liver-specific expression of the aldolase B gene. MolCell Biol 1989; 9: 4923–4931.

9. Chang HK, Wang B-Y, Yuh C-H, Wei CL, Ting LP. A liver-specific nuclearfactor interacts with region of the large surface protein gene in humanhepatitis B virus. Mol Cell Biol 1989; 9: 5189–5197.

10. DeSimone V, DeMagistris L, Lazzaro D, et al. LFB3, a heterodimer-forming homeoprotein of the LFB1 family, is expressed in specializedepithelia. EMBO J 1991; 10: 1435–1443.

11. Rey-Campos J, Chouard T, Yaniv M, Cereghini S. vHNF1 is a homeopro-tein that activates transcription and forms heterodimers with HNF1.EMBO J 1991; 10: 1445–1457.

12. Mendel DB, Hansen LP, Graves MK, Conley PB, Crabtree GR. HNF-1áand HNF-1â (vHNF-1) share dimerization and homeodomains, but notactivation domains, and form heterodimers in vitro. Genes Dev 1991; 5:1042–1056.

277HNF-1á AND HNF-1â IN HISTOLOGICAL DIFFERENTIATIONS OF HEPATOCELLULAR CARCINOMA

? 1998 John Wiley & Sons, Ltd. , . 184: 272–278 (1998)

13. Frain M, Swart G, Monaci P, et al. The liver-specific transcription factorLFB1 contains a highly diverged homeobox DNA binding domain. Cell1989; 59: 145–157.

14. Ninomiya T, Hayashi Y, Saijoh K, Ohta K. Expression ratio of hepatocytenuclear factor-1 to variant hepatocyte nuclear factor-1 in differentiation ofhepatocellular carcinoma and hepatoblastoma. J Hepatol 1996; 25: 445–453.

15. Noguchi Role of SH-PTP2, a protein-tyrosine phosphatase with srchomology 2 domain, in insulin-stimulated ras activation. Mol Cell Biol 14:6674–6682.

16. Schorpp M, Kugler W, Wagner U, Ryffel GU. Hepatocyte-specific pro-moter element HP-1 of Xenopus albumin gene interacts with transcriptionalfactors of mammalian hepatocytes. Mol Cell Biol1988; 202: 307–320.

17. Deschatrette J. Weiss MC. Characterization of differentiated and de-differentiated clones from a rat hepatoma. Biochimie 1974; 56: 1603–1611.

18. Nagy P, Bisgaard HC, Thorgeirsson SS. Expression of hepatic transcriptionfactors during liver development and oval cell differentiation. J Cell Biol1994; 125: 223–233.

19. Kalkuhl A, Kaestner K, Buchmann A, Schwarz M. Expression ofhepatocyte-enriched nuclear transcription factors in mouse liver tumors.Carcinogenesis 1996; 17: 609–612.

20. Xanthopoulos KG, Mirkovitch J. Gene regulation in rodent hepatocytesduring development, differentiation and disease. Eur J Biochem 1993; 216:353–360.

21. Flodby P, Liao DZ, Blanck A, Xanthopoulos KGH. Expression of liver-enriched transcription factors C/EBPá C/EBPâ, HNF-1 and HNF-4 inpreneoplastic nodules of hepatocellular carcinoma in rat liver. MolCarcinogenesis 1996; 17: 609–612.

278 W. WANG ET AL.

? 1998 John Wiley & Sons, Ltd. , . 184: 272–278 (1998)