The homeoproteins HNF1 (LFB1/HNF1-A) and vHNF1 (LFB3/HNF1β) interact with an essential control element of a group of liver-specific genes. During development, these putative target genes are initially expressed in the visceral endoderm of the yolk sac and subsequently in fetal liver. To assess the possible involvement of HNF1 and/or vHNF1 as transcriptional regulators in the early steps of visceral endoderm differentiation, we have analyzed the expression pattern of both factors both in vitro during differentiation of murine F9 embryonal carcinoma cells and in vivo during early postim-plantation mouse development. We show here that differentiation of F9 cells into either visceral or parietal endoderm is accompanied by a sharp induction in vHNF1 mRNA and protein. By contrast, only low levels of aberrantly sized HNF1 transcripts, but not DNA-binding protein, are found in F9 cells and its differentiated derivatives. At 6 – 7.5 days of gestation, high levels of vHNF1 mRNA are present in the visceral extraem-bryonic endoderm, which co-localize with transcripts of the transthyretin gene. HNF1 transcripts are first detected in the yolk sac roughly two embryonic days later, after the developmental onset of transcription of target genes. As development proceeds, discrepancies are observed between the level of transcripts of both vHNF1 and HNF1 and their respective nuclear binding proteins, notably in the yolk sac and embryonic kidney. In addition, we show that two alternative spliced isoforms of vHNF1 mRNA, vHNF-A and vHNF1-B, are expressed in both embryonic and adult tissues. Taken together, these data suggest that vHNF1 participates as a regulatory factor in the initial transcriptional activation of the target genes in the visceral endoderm of the yolk sac, whereas the later appearance of HNF1 could be required for maintenance of their expression. Our results also provide evidence of a posttranscriptional level of control of vHNF1 and HNF1 gene expression during development, in addition to the spatial restriction in transcription.

HNF1 and vHNF1 are two closely related homeoproteins that interact with an essential sequence present in the promoter and/or enhancer regions of several liver-specific genes (for a review see De Simone and Cortese, 1991; Lai and Darnell, 1991). Whereas the role of HNF1 in cell-specific transcriptional regulation has been established both in cultured cells and in vitro (Maire et al., 1989; Ryffel et al., 1989; Tronche et al., 1990; Lichtsteiner and Schibler, 1989), that of vHNF1 remains uncertain.

vHNF1 (also called LFB3 and HNF1β) is highly homologous to HNF1 in regions involved in DNA binding, i.e. the N-terminal dimerization domain, the POU-like domain and the divergent or ‘extra-large’ homeodomain. Like HNF1, vHNF1 binds DNA as a dimer and forms heterodimers with HNF1 in vitro and probably in vivo (Rey-Campos et al., 1991; De Simone et al., 1991; Mendel et al., 1991a; Bach et al., 1991). The genes encoding HNF1 and vHNF1 are located on different chromosomes and apparently arose by gene duplication (Bach et al., 1991). Unexpectedly, vHNF1 exhibits a transactivation potential similar to HNF1 when tested in transient cotransfection assays, although the two transactivation domains of HNF1 mapped in the C-terminal part of the protein are not conserved in vHNF1 (Rey-Campos et al., 1991; De Simone et al., 1991). In the adult, high levels of HNF1 transcripts are found in liver, kidney and intestine (Kuo et al., 1990b), while the expression of vHNF1 mRNA is largely confined to kidney, with low levels of message in intestine, lung and liver (Rey-Campos et al., 1991; De Simone et al., 1991; Mendel et al., 1991a).

In situ hybridization analyses of mid to late embryonic stages have indicated that HNF1 and vHNF1 mRNAs are present, albeit at different levels, in polarized epithelia of organs of endodermal and mesodermal origin - kidney, liver and the digestive tract. In both the developing liver and kidney, vHNF1 expression precedes that of HNF1 (De Simone et al., 1991; Ott et al., 1991; Lazzaro et al., 1992). Moreover, HNF1 and vHNF1 transcripts are not always located in the same cell types within these organs (Ott et al., 1991; Lazzaro et al., 1992). At present, however, the relationship between the temporal expression of HNF1 and vHNF1 and developmental activation of putative target genes has not been precisely defined.

Visceral and parietal endoderm derive from the primitive endoderm and constitute one of the first cell lineages to differentiate in the mouse embryo. In rodents, visceral endoderm cells form a layer of polarized epithelial cells at the periphery of the egg cylinder. Organized as an absorptive/secretory epithelium, it is the first structure to provide nutritional and hematopoietic function, before the hemo-chorial placenta is formed (Rugh, 1990). In the early postimplantation stage, visceral endoderm, but not parietal endoderm, synthesizes and secretes a number of serum proteins, shown to be regulated by HNF1 in cultured cells, including alpha-fetoprotein (AFP), albumin, transthyretin (TTR), alpha-1-antitrypsin (A1-AT) (Dziadek and Andrews, 1983; Meehan et al., 1984; Makover et al., 1989). The murine F9 embryonal carcinoma (EC) cells are able to differentiate in vitro toward either visceral or parietal endoderm and therefore mimic features of the endoderm differentiation in the mouse embryo shortly before and after implantation (Hogan et al., 1981; Scott et al., 1984; Soprano et al., 1988; Grover and Adamson, 1986). In the present study, we examine whether vHNF1 and/or HNF1 could be involved in the commitment and differentiation of the visceral endoderm, by performing a detailed analysis of the expression profiles of these homeoproteins both in vitro during differentiation of embryonal carcinoma F9 cells and in vivo in the early postimplantation mouse embryo.

We show here that the most significant change upon differentiation of F9 cells to either visceral or parietal endoderm is a sharp induction of vHNF1 mRNA and protein. In contrast, low levels of aberrantly sized HNF1 message are found in both non-differentiated and differentiated F9 cells. Moreover, no HNF1-binding activity is detected in either F9 or its differentiated derivatives.

In early postimplantation embryos, high levels of vHNF1 transcripts are detected by in situ hybridization in the visceral extraembryonic endoderm at 6 to 7.5 days of gestation. Active vHNF1 binding protein is also detected in 7.5-day embryos. At these stages no significant levels of either HNF1 mRNA or protein are observed.

The results of these analyses support the hypothesis that vHNF1 may participate as a regulatory factor in the early transcriptional activation of the presumed target genes in the visceral endoderm of the yolk sac, while the later appearance of HNF1 could be required for maintenance of their expression. These results are discussed in relation to other transcriptional regulators implicated in hepato-specific gene expression.

Cell lines and culture conditions

The rat hepatoma cells used, H4II and its dedifferentiated derivative H5 (Deschatrette and Weiss, 1974), the somatic hybrid HF1, which causes extinction of the expression of liver-specific functions, and HF1-5, a revertant of HF1 reexpressing liver functions (Deschatrette et al., 1979), were cultured as described (Cereghini et al., 1988). Mouse F9 EC, P19 EC, the trophoblastoma 3TDM-1 and the parietal-like PYS-2 cell-lines were provided by J. F. Nicolas and cultured as described by Nicolas et al. (1976). For differentiation into either ‘parietal’ or ‘primitive endoderm’-like cells, 2.5×105 F9 cells were plated on gelatinized 10 cm tissue culture Petri dishes containing 5×10 −7 M RA (Sigma Chemical Co) (primitive endoderm) or 5×10 −7 M RA + 0.1 mM dcAMP (parietal endoderm). Differentiation of F9 aggregates to visceral endoderm was carried out as described by Scott et al. (1984).

Isolation of cDNA clones encoding mouse HNF1 and vHNF1

Oligo(dT)-primed mouse liver and F9 differentiated visceral cDNA libraries were constructed using the kits ‘cDNA synthesis’ and ‘Lambda-gt-10 cloning system’ from Amersham, as described by the manufacturers. The nonamplified libraries (5×105 plaques) were screened by hybridization in low-stringency conditions with a 511 bp NcoI fragment that contains the entire homeodomain of rat HNF1 (334 to 845 nt residues) essentially as described (Bach et al., 1991). Isolation and sequencing of positive clones was performed as described by Bach et al. (1991). Three full-length mouse vHNF1 clones were obtained, while the longest mouse HNF1 clone lacks 45 nt from the ATG initiator codon. The sequence of this clone was identical to that of mouse HNF1 reported by Kuo et al. (1990b). As in previous screening of a H5 rat hepatoma cDNA library (Rey-Campos et al., 1991), two types of mouse vHNF1 clones were obtained. Most of them contained a 78 nt insertion between the B domain and the homeodomain, while others lack this sequence. The mouse vHNF1 sequence reported by Mendel et al. (1991a) corresponds to the type of clones lacking such insertion, whereas that reported by De Simone et al. (1991) corresponds to those containing this insertion. The coding sequences of our mouse (± insertion) vHNF1 clones were identical to that shown by Mendel et al. (1991a) and differ in two regions with that reported by De Simone et al. (1991). These differences are probably due to sequencing errors, since a human vHNF1 clone (Bach et al., 1991) displays identical amino acid residues in these regions to either our or Mendel et al. (1991a) mouse clones.

Nuclear extracts and gel mobility-shift assays

Nuclear extracts from cell-lines and mouse adult tissues were prepared as described (Cereghini et al., 1988). Mouse embryo extracts were prepared at the indicated times after fertilization, by micro-surgery and pooled. Nuclear extracts from whole embryo or microdissected organs were prepared in the presence of 2.5 μg/ml of BSA, at a microscale level, essentially as described for adult tissues, except that the ammonium sulfate precipitation and dialysis steps were omitted. The amount of nuclear protein from embryo extracts used in the DNA-binding assays ranged from 0.1 to 0.5 μg. Normalization of embryo extracts was performed by assaying the ubiquitous NF-Y binding factor. Gel mobility shift assays were as described (Cereghini et al., 1988). When mentioned, specific polyclonal antibodies at the indicated dilution in PBS + 1 μg/ml of BSA, were preincubated with nuclear extracts at room temperature for 10 minutes, before adding the labeled probe. Double-stranded oligonucleotides encompassing the HNF1 (PE) and the NF-Y binding sites of the rat albumin promoter (Cereghini et al., 1988) and the HNF4 site of the apoCIII promoter (Sladek et al., 1990), kindly provided by M.Zakin, were 32P-labeled with T4 polynucleotide kinase. Total protein extracts from embryonic organs were prepared by tissue homogenization as described for the isolation of nuclei, followed by addition of 1/10 volume of 4 M ammonium sulfate, sonication and centrifugation. Western blots with either total or nuclear protein were carried out essentially as described by Blumenfeld et al. (1991), using 40 –60 μg and 25 μg of total and nuclear protein, respectively.

Antiserum preparation

HNF1 and vHNF1-specific antisera were raised, respectively against residues 306 to 391 of mouse HNF1 and residues 293 to 424 of mouse vHNF1, fused to gluthathione transferase in the pGEX3 bacterial expression vector (Smith and Johnson, 1988). The cDNA fragments were generated using specific primers, following amplification from Bluescribe plasmids containing either mouse HNF1 or mouse vHNF1 cDNAs. The two polyclonal anti-bodies were raised into white Bouscat rabbits. Boost injections were performed each 21 –28 days with 100 μg of fusion protein purified on gluthathione agarose columns (Smith and Johnson, 1988). Sera used in this study were obtained after three boosts.

RNA isolation, analysis and quantitative PCR amplification of cDNA

Total RNA, poly (A)+ RNA preparation and northern blot analysis were carried out as previously described (Cereghini et al., 1990). For PCR analysis of RNA, 1 μg of poly (A)+ RNA prepared from cell lines and mouse tissues was reverse transcribed in a final volume of 80 μl using random hexamers, as described by Montarras et al. (1989). When indicated, 4 μg of total RNA was used to prepare the corresponding cDNAs. The conditions of cDNA synthesis and PCR amplification were essentially as described by Montarras et al. (1989). The PCR reactions were performed in a total volume of 50 μl using either 2.5 μl or 10 μl of the cDNA mixture prepared respectively from poly (A)+ RNA or total RNA, for 15 cycles of 94°C denaturation (30 seconds), 55°C annealing (30 seconds) and 72°C extension (1 minute) in an automated thermal cycler. Due to the preferential amplification of some products, independently of their size, the results of RT-PCR experiments were confirmed with at least two different sets of primers. The primers chosen were located in different exons and control experiments showed that the generation of PCR products were strictly dependent on the synthesis of cDNA and on the presence of the selected 5′- and 3′-primers (not shown). The quality and amount of the RNA samples were previously examined by northern blot, using GAPDH as probe. In addition, specific primers for hypoxanthine phosphorybosyl transferase (HPRT) transcripts were also used as positive control of the corresponding cDNA. PCR primer sequences used in the experiments shown here were:

vHNF1 - 5′: AACGGGAGATCCTCCGACAG;

vHNF1 - 3′: CATGGCAACTGTTACCATGGT;

HNF1 - 5′: TGGTTTCTAAGCTGAGCCAGC;

HNF1 - 3′: GGCTCTTCAATCAGTCCGC;

HPRT - 5′: GATATAATTGACACTGGTAAA (exon 6);

HPRT -3′: TCAAGGGCATATCCAACAACAAAC (exon 8) (see Melton et al., 1984). Unless otherwise indicated, the PCR conditions were such that amplification of cDNAs was linear and dependent on the concentration of the respective RNA : low number of cycles (no more than 15) and transfer of the amplified products to a membrane, followed by hybridization. 18 μl of the PCR reactions (final volume 50 μl) were loaded on agarose gels, transferred to Hybond N (Amersham) membranes and hybridized as described for northern blot analysis. Full-length mouse vHNF1 cDNA and the mouse HNF1 cDNA (see above) were used as hybridization probes. The hprt probe was generated by RT-PCR from mouse liver RNA.

Molecular probes and in situ hybridization procedures

The HNF1 and vHNF1 probes used for in situ hybridization encompassed the same DNA sequences from the carboxy terminal part of the mouse vHNF1 and HNF1 cDNAs that were used to generate the fusion proteins in pGEX3 vectors. The PCR amplified fragments (see above) were cloned into Bluescribe plasmid (Stratagene). AFP (Tyner et al., 1990) and TTR probes (Makover et al., 1989) were kindly provided by S. Thilghman and W. Blanes, respectively. C3H mice embryos were used. Noon of the day when the vaginal plug was detected is considered gestational day 0.5. Preparation of 35S-riboprobes, tissue sections and hybridization conditions were as described (Ott et al., 1991). For each hybridization, antisense and sense riboprobes were used; only background levels of hybridization were detected with sense probes and are not shown.

Induction of vHNF1 mRNA, but not of HNF1 RNA, in differentiated F9 cells

In the presence of RA, F9 EC cells differentiate into endoderm-like cells which, depending on the culture conditions, form either parietal or visceral endoderm. When grown as monolayers in the presence of RA, or RA + dibutyryl cyclic AMP (dbcAMP), they differentiate into parietal endoderm cells (Strickland et al., 1980). However, if F9 cells are treated with RA and allowed to aggregate, most of the cells of the outer surface of the aggregate differentiate into visceral endoderm, characterized by the synthesis of AFP, albumin and transthyretin (Hogan et al., 1981; Grover et al., 1983; Soprano et al., 1988).

We examined the levels of HNF1 and vHNF1 mRNA by northern blot analysis using poly(A)+ RNA from F9 cells differentiated into either parietal or visceral endoderm at various times after differentiation (Fig. 1). As markers of differentiation, we followed both the keratin Endo A and AFP mRNA levels. As expected, Endo A is expressed at very low levels in F9 EC cells and is highly induced in parietal derivatives (Oshima, 1982; Fig. 1A), while AFP mRNA is induced exclusively in visceral cells (Fig. 1B).

Fig. 1.

Induction of vHNF1 mRNA in RA-induced differentiation of F9 cells along parietal and visceral endoderm pathways. F9 EC cells were grown either as monolayer in the presence of 5×10 −7 M RA + 0.1 M cAMP (parietal endoderm pathway : Par) or in suspension in the presence of 7.5×10 −8 M (visceral endoderm pathway : VIS). Cells were harvested for RNA and nuclear protein (Fig. 2) analyses at the times indicated in days above the respective lanes. F9EC corresponds to undifferentiated stem cells. Mouse liver RNA was used as a control for HNF1 mRNA; RNA from the parietal endoderm-like cell line PYS-2 and the trophoblastoma cell line 3TDM-1 were also included. Northern blot analyses using 3 μg of poly (A)+ RNA per lane were performed as described in Materials and methods. Similar results were obtained in several independent differentiation experiments. The sizes in kb of the mRNAs hybridized are shown on the left with pointed arrows. (A) The same filter was consecutively hybridized with the probes indicated on the left of the respective autoradiograms and comprise from top: the full-length mouse vHNF1 cDNA clone; full-length rat GAPDH; a 627 bp NcoI fragment derived from HNF1 cDNA (Bach et al., 1991); Endo A cDNA fragment (kindly provided by P. Duprey). The origin of the extra band observed with the vHNF1 probe in the 3TDM RNA sample is under investigation. Note that less RNA was loaded in lane VIS 10d. (B) Specific induction of AFP mRNA in visceral F9 derivatives. Filter hybridized with a mouse AFP cDNA probe (Tyner et al., 1990).

Fig. 1.

Induction of vHNF1 mRNA in RA-induced differentiation of F9 cells along parietal and visceral endoderm pathways. F9 EC cells were grown either as monolayer in the presence of 5×10 −7 M RA + 0.1 M cAMP (parietal endoderm pathway : Par) or in suspension in the presence of 7.5×10 −8 M (visceral endoderm pathway : VIS). Cells were harvested for RNA and nuclear protein (Fig. 2) analyses at the times indicated in days above the respective lanes. F9EC corresponds to undifferentiated stem cells. Mouse liver RNA was used as a control for HNF1 mRNA; RNA from the parietal endoderm-like cell line PYS-2 and the trophoblastoma cell line 3TDM-1 were also included. Northern blot analyses using 3 μg of poly (A)+ RNA per lane were performed as described in Materials and methods. Similar results were obtained in several independent differentiation experiments. The sizes in kb of the mRNAs hybridized are shown on the left with pointed arrows. (A) The same filter was consecutively hybridized with the probes indicated on the left of the respective autoradiograms and comprise from top: the full-length mouse vHNF1 cDNA clone; full-length rat GAPDH; a 627 bp NcoI fragment derived from HNF1 cDNA (Bach et al., 1991); Endo A cDNA fragment (kindly provided by P. Duprey). The origin of the extra band observed with the vHNF1 probe in the 3TDM RNA sample is under investigation. Note that less RNA was loaded in lane VIS 10d. (B) Specific induction of AFP mRNA in visceral F9 derivatives. Filter hybridized with a mouse AFP cDNA probe (Tyner et al., 1990).

As shown in Fig. 1A, a strong increase in vHNF1 mRNA levels is observed in both visceral and parietal endoderm-type cells. Low levels of HNF1 transcripts are present in both nondifferentiated F9 cells and after differentiation into these two pathways. Moreover, these transcripts display different sizes ranging from 3 to 3.5 kb instead of a unique HNF1 transcript of 3.5 kb, as is observed in liver (Fig. 1A), kidney or intestine (Rey-Campos et al., 1991). The different sizes of HNF1 transcripts are not due to nonspecific mRNA degradation as indicated by the sequential hybridization of the same northern blot with the vHNF1, Endo A and GAPDH probes (Fig. 1A). The origin of these aberrantly sized HNF1 transcripts is presently under investigation. The presence of HNF1 transcripts in nondifferentiated F9 cells is not a particularity of this cell-line as they are also detected in embryonic stem cells (ES), albeit at lower levels than in F9 (see below, Fig. 3).

Fig. 2.

Induction of vHNF1 binding activity upon differentiation of F9 into parietal and visceral cells. Nuclear extracts were prepared from the indicated cell lines and F9 cells treated as described in Fig. 1. Gel electrophoretic mobility shift assays were carried out using as probe double strand (ds) PE oligonucleotide encompassing the HNF1 binding site of the rat albumin promoter (A, C and D) and the albumin CCAAT/NF-Y ds oligonucleotide (B). (A) The DNA-binding activity induced upon differentiation of F9 cells commigrates with authentic vHNF1 protein from H5 dedifferentiated hepatoma cells. 3 μg of mouse liver nuclear protein and 4 and 8 μg of nuclear protein from the cell lines indicated above the lanes were used. Nuclear extracts from mouse liver and H5 cells were used as control for migration of HNF1 homodimers and vHNF1 homodimers, respectively. Arrows indicate the position of HNF1 homodimers and HNF1:vHNF1 heterodimers (see Cereghini et al., 1988; Rey-Campos et al., 1991). (B) NF-Y activity. 4 μg of the indicated nuclear extracts were assayed for the ubiquitous NF-Y binding activity to normalize for the amount and quality of the nuclear extracts. (C) The DNA-binding activity present in differentiated F9, PYS-2 and 3TDM-1 cells exhibit similar sequence specificity to HNF1 and vHNF1 proteins. Binding reactions were performed as in A using 8 μg of the indicated nuclear extracts, in the presence of 40-fold excess of wt and mutant competitor oligonucleotides, which display reduced (mut3, mut4 and DS12) or no affinity (DS34) for HNF1/vHNF1 proteins. No competitor in lanes 1, 7, 13; PE wt in lanes 2, 8, 14 ; mut 3 in lanes 3, 9, 15; mut 4 in lanes 4, 10, 16; DS12 in lanes 5, 11, 17 and DS34 (negative for binding) in lanes 6, 12, 18. Sequence of oligonucleotides are reported in Cereghini et al., 1988. Only the retarded DNA-protein complexes are shown. C3: complex migrating faster than HNF1 (C1) and vHNF1 homodimers (C2); see also D. (D) Binding assays in the presence of HNF1 and vHNF1 specific antisera. 3 μg of nuclear protein from mouse liver and 8 μg of nuclear protein from the indicated cells were incubated with 1 μl of preimmune serum (pi) or the indicated dilutions of HNF1 antiserum (A-H) and vHNF1 antiserum (A-v). Arrowheads indicate positions of supershifts generated by vHNF1 antibodies. Note that the faster migrating complex observed in visceral derivatives and to a lesser extent in 3TDM-1 (indicated as C3 in C) is displaced by vHNF1 antiserum and may be due to either partial degradation or association with another protein.

Fig. 2.

Induction of vHNF1 binding activity upon differentiation of F9 into parietal and visceral cells. Nuclear extracts were prepared from the indicated cell lines and F9 cells treated as described in Fig. 1. Gel electrophoretic mobility shift assays were carried out using as probe double strand (ds) PE oligonucleotide encompassing the HNF1 binding site of the rat albumin promoter (A, C and D) and the albumin CCAAT/NF-Y ds oligonucleotide (B). (A) The DNA-binding activity induced upon differentiation of F9 cells commigrates with authentic vHNF1 protein from H5 dedifferentiated hepatoma cells. 3 μg of mouse liver nuclear protein and 4 and 8 μg of nuclear protein from the cell lines indicated above the lanes were used. Nuclear extracts from mouse liver and H5 cells were used as control for migration of HNF1 homodimers and vHNF1 homodimers, respectively. Arrows indicate the position of HNF1 homodimers and HNF1:vHNF1 heterodimers (see Cereghini et al., 1988; Rey-Campos et al., 1991). (B) NF-Y activity. 4 μg of the indicated nuclear extracts were assayed for the ubiquitous NF-Y binding activity to normalize for the amount and quality of the nuclear extracts. (C) The DNA-binding activity present in differentiated F9, PYS-2 and 3TDM-1 cells exhibit similar sequence specificity to HNF1 and vHNF1 proteins. Binding reactions were performed as in A using 8 μg of the indicated nuclear extracts, in the presence of 40-fold excess of wt and mutant competitor oligonucleotides, which display reduced (mut3, mut4 and DS12) or no affinity (DS34) for HNF1/vHNF1 proteins. No competitor in lanes 1, 7, 13; PE wt in lanes 2, 8, 14 ; mut 3 in lanes 3, 9, 15; mut 4 in lanes 4, 10, 16; DS12 in lanes 5, 11, 17 and DS34 (negative for binding) in lanes 6, 12, 18. Sequence of oligonucleotides are reported in Cereghini et al., 1988. Only the retarded DNA-protein complexes are shown. C3: complex migrating faster than HNF1 (C1) and vHNF1 homodimers (C2); see also D. (D) Binding assays in the presence of HNF1 and vHNF1 specific antisera. 3 μg of nuclear protein from mouse liver and 8 μg of nuclear protein from the indicated cells were incubated with 1 μl of preimmune serum (pi) or the indicated dilutions of HNF1 antiserum (A-H) and vHNF1 antiserum (A-v). Arrowheads indicate positions of supershifts generated by vHNF1 antibodies. Note that the faster migrating complex observed in visceral derivatives and to a lesser extent in 3TDM-1 (indicated as C3 in C) is displaced by vHNF1 antiserum and may be due to either partial degradation or association with another protein.

Fig. 3.

Quantitative RT-PCR analysis of vHNF1-A, vHNF1-B and HNF1 transcript accumulation in embryonic and adult tissues. (A) Schematic representation of functional domains of HNF1, vHNF1-A and vHNF1-B proteins. Boxes represent the coding sequences. Highly conserved sequences within the DNA-binding domains of HNF1 and vHNF1 proteins include the N-terminal dimerization domain (A domain; amino acids 1-32), the POU-like or B-domain and the homeodomain. Also indicated are the 78 nt alternative exon and vHNF1-B isoform; the putative hinge glycine/proline-rich region of HNF1, which is not conserved in vHNF1-A and B, and the two transactivation domains of HNF1 mapped by Nicosia et al. (1990). Non conserved domains between the A and B domains and the C-terminal part of the proteins are shown as empty boxes. (B) Alternative spliced vHNF1 transcripts in F9 differentiated cells, different cultured cells and adult tissues. RT-PCR amplification were carried out using poly (A)+ RNA from P19 undifferentiated and grown in suspension in the presence of 10% DMSO and 1×10−9 M RA for 4 days, F9 cultured as monolayer for 24 hours in the presence of 5×10-7 M RA (F9 RA 24 h RA), differentiated along the parietal and visceral pathways as described in Fig. 1, and from the indicated cell lines and mouse tissues as described in Material and methods. The two PCR products of 596 nt and 518 nt that represent the nonspliced (vHNF1-A) and the spliced (vHNF1-B) RNAs, respectively, the HNF1 amplified product of 564 nt and the hprt product of 195 nt are indicated by arrows. A weak signal for vHNF1 was observed in nondifferentiated F9 cells, which may be due to low levels of spontaneous differentiation. In lanes Par 3d and PYS-2, the amplifications were not linear. (C) Expression pattern of vHNF1 and HNF1 in embryonic stem cells, early embryos and developing organs. RT-PCR was performed using total RNA from ES cells, whole embryos, yolk sac and developing organs at the indicated days of gestation. The signals detected in B and C are not comparable. The reasons for the significantly low levels of hprt amplified product observed in yolk sac samples at 9.5 and 14.5 days are presently unknown and are not due to low levels of RNA because similar levels of GAPDH mRNA were found in these samples (not shown). Numbers 1 to 8 below lanes of hprt amplifications correspond to the same RNA samples ES to YS 9.5 d, as indicated above lanes of vHNF1 amplifications

Fig. 3.

Quantitative RT-PCR analysis of vHNF1-A, vHNF1-B and HNF1 transcript accumulation in embryonic and adult tissues. (A) Schematic representation of functional domains of HNF1, vHNF1-A and vHNF1-B proteins. Boxes represent the coding sequences. Highly conserved sequences within the DNA-binding domains of HNF1 and vHNF1 proteins include the N-terminal dimerization domain (A domain; amino acids 1-32), the POU-like or B-domain and the homeodomain. Also indicated are the 78 nt alternative exon and vHNF1-B isoform; the putative hinge glycine/proline-rich region of HNF1, which is not conserved in vHNF1-A and B, and the two transactivation domains of HNF1 mapped by Nicosia et al. (1990). Non conserved domains between the A and B domains and the C-terminal part of the proteins are shown as empty boxes. (B) Alternative spliced vHNF1 transcripts in F9 differentiated cells, different cultured cells and adult tissues. RT-PCR amplification were carried out using poly (A)+ RNA from P19 undifferentiated and grown in suspension in the presence of 10% DMSO and 1×10−9 M RA for 4 days, F9 cultured as monolayer for 24 hours in the presence of 5×10-7 M RA (F9 RA 24 h RA), differentiated along the parietal and visceral pathways as described in Fig. 1, and from the indicated cell lines and mouse tissues as described in Material and methods. The two PCR products of 596 nt and 518 nt that represent the nonspliced (vHNF1-A) and the spliced (vHNF1-B) RNAs, respectively, the HNF1 amplified product of 564 nt and the hprt product of 195 nt are indicated by arrows. A weak signal for vHNF1 was observed in nondifferentiated F9 cells, which may be due to low levels of spontaneous differentiation. In lanes Par 3d and PYS-2, the amplifications were not linear. (C) Expression pattern of vHNF1 and HNF1 in embryonic stem cells, early embryos and developing organs. RT-PCR was performed using total RNA from ES cells, whole embryos, yolk sac and developing organs at the indicated days of gestation. The signals detected in B and C are not comparable. The reasons for the significantly low levels of hprt amplified product observed in yolk sac samples at 9.5 and 14.5 days are presently unknown and are not due to low levels of RNA because similar levels of GAPDH mRNA were found in these samples (not shown). Numbers 1 to 8 below lanes of hprt amplifications correspond to the same RNA samples ES to YS 9.5 d, as indicated above lanes of vHNF1 amplifications

In addition, two stably differentiated teratocarcinoma cell lines, the parietal endoderm-like cell line PYS-2 and 3TDM-1, a trophoblastoma cell-line, characterized by the presence of endoderm-like and giant trophoblastic-like cells (Nicolas et al., 1976), behave similarly to differentiated F9 cells. Cells of both lines express constitutively high levels of vHNF1 mRNA, but low levels of aberrantly sized HNF1 mRNAs. In contrast, no vHNF1 or HNF1 mRNA is observed in P19 teratocarcinoma cells that undergo neuronal differentiation upon aggregation and treatment with RA (not shown). Thus, the induction of vHNF1 mRNA appears to be associated with the endodermal differentiation character of the cells.

The HNF1 binding activity induced in differentiated F9 cells is composed essentially of vHNF1 homodimers

vHNF1 and HNF1 proteins display an identical DNA-binding sequence specificity, but vHNF1 has a lower molecular weight and exhibits a faster mobility than HNF1 in gel retardation assays (Cereghini et al., 1988; Baumhueter et al., 1988). To assay for the presence of HNF1/vHNF1 DNA-binding activities, gel shift assays were performed using as probe the HNF1 site of the albumin proximal promoter (PE oligonucleotide). Nuclear extracts from H5 cells, which lack HNF1 (Cereghini et al., 1988), were used as a control for the mobility of vHNF1 homodimers.

As shown in Fig. 2A, no complexes were detected in nondifferentiated F9 cells. After RA addition, a complex, which comigrates with that formed by vHNF1 homodimers of H5 nuclear extracts is observed in both parietal and visceral derivatives. As a control, we monitored the ubiquitous NF-Y (CCAAT) binding activity in the same extracts and found roughly the same activity (Fig. 2B).

Cross-competition experiments with wild-type and mutant PE oligonucleotides clearly show that the complex formed in differentiated F9 cells exhibits an identical behavior to that which we previously described for both HNF1 and vHNF1 (Fig. 2C; Cereghini et al., 1988). UV cross-linking experiments showed that the complex detected in both parietal and visceral F9 extracts is formed by a polypeptide of apparent relative molecular mass identical to that of vHNF1 from H5 extracts (i.e. ∼70 –75×103; data not shown).

To confirm further the presence of only vHNF1 homodimers in F9 differentiated cells, we generated specific anti-bodies against the most highly divergent region in the C-terminal part of the mouse HNF1 and vHNF1 proteins (see Material and methods). In no case was the complex generated by either visceral or parietal differentiated F9 cells affected by HNF1 antibodies, while it was totally competed by vHNF1 antibodies (Fig. 2D).

These results show that the PE-binding activity induced upon differentiation of F9 cells is primarily composed of vHNF1 homodimers. Consequently, the HNF1 transcripts present in both undifferentiated and differentiated F9 cells are either not translated into active HNF1 binding protein or the protein is extremely unstable. We did not detect HNF1 protein in either nuclear or total cellular extracts from these cells by either western blot or gel shift assays (data not shown), a result that argues against the presence of either inactive DNA-binding protein(s) or cytoplasmic protein not transported to the nucleus. However, we cannot exclude the presence of trace levels of HNF1 protein that could escape detection by these assays.

We also investigated whether HNF1 and vHNF1 might exhibit different sensitivity to RA. No evidence of HNF1 binding activity was observed in extracts from F9 cells grown in suspension and exposed to RA concentrations ranging from 7.5×10 −8 M to 1×10 −6 M, for either 5, 10 or 15 days, while vHNF1 activity was similarly induced at all of these concentrations and time points (data not shown).

Developmental expression of HNF1 and vHNF1 mRNAs

To verify the developmental relevance of the results obtained in differentiated F9 cells, we examined the time course of transcript accumulation of vHNF1/HNF1 during mouse development by a quantitative reverse transcription-polymerase chain reaction (RT/PCR) assay (see Material and Methods).

With regard to vHNF1, specific primers were designed to verify whether the two distinct vHNF1 cDNA clones previously identified (Rey-Campos et al., 1991) are two spliced forms of vHNF1. As reported for rat vHNF1, two different mouse vHNF1 cDNA clones were also obtained upon screening a cDNA library from F9 differentiated cells; most of them contained a 78-nucleotide (nt) insertion between the POU domain and the homeodomain (referred as vHNF1-A), while roughly one out of four clones lacked this sequence (named vHNF1-B). Otherwise, mouse vHNF1-A and vHNF1-B display identical coding and 3′ non-coding sequences (see Materials and Methods). Since restriction analyses of mouse genomic DNA indicated that vHNF1 is a single copy gene (data not shown), these two classes of clones are likely generated by alternative splicing. To examine this more directly, specific primers, located at both sides of the 78 nt insertion, were used to amplify cDNAs transcribed from RNAs from various cell lines and tissues at different developmental stages.

The results of these analyses clearly indicate that both isoforms are present, but vHNF1-A represents the major mRNA species expressed in the different cell lines as well as in the different embryo stages and adult tissues (Fig. 3B,C). However, the ratio of vHNF1-A/vHNF1-B transcripts is reproducibly higher in adult liver and intestine than in other embryo or adult tissues.

vHNF1 and HNF1 mRNA levels were then examined in whole embryos at 7.5 days post coitum (d.p.c.), the stage when visceral and parietal extraembryonic endoderm are clearly differentiated, as well as at later developmental stages both in the embryo proper and the yolk sac. High levels of vHNF1 transcripts are observed in 7.5 d.p.c. embryos, while at this stage HNF1 transcripts are barely detectable. In the yolk sac, the levels of vHNF1 mRNA remains high throughout gestation from 8.5 to 14.5 days. At 8.5 days of gestation HNF1 mRNA is indeed present in the yolk sac and its levels increase at later developmental stages (Fig. 3C and results not shown). Thus, both HNF1 and vHNF1 mRNAs are expressed in the yolk sac from 8.5 days on, but vHNF1 precedes HNF1 expression.

vHNF1 and TTR transcripts co-localize in the visceral extraembryonic endoderm during early postimplantation mouse development

To localize transcripts in early postimplantation embryos, we performed RNA in situ hybridization analysis of embryo sections. At 6.5 days of gestation, the endoderm cells that remain in contact with the egg cylinder are organized into visceral extraembryonic and visceral embryonic endoderm, surrounding the embryonic and extraembryonic ectoderm, respectively. The primitive endoderm cells, which have migrated onto the mural trophoblast, form the parietal endoderm that consists of a flat layer lining the decidua.

As seen in Fig. 4, high levels of vHNF1 transcripts are detected in the visceral endoderm and in the uterine epithelium. Serial sections showed that vHNF1 transcripts are predominant in the extraembryonic endoderm. The decidua and trophoblast appear negative, indicating that parietal endoderm cells do not express vHNF1.

Fig. 4.

Pattern of vHNF1 mRNA expression at embryonic day 6.5. Sections through the implantation site containing the embryo were hybridized with 35S-labeled antisense vHNF1 RNA probe, whereas the section shown in C was hybridized with HNF1 riboprobe. (A-F) Dark-field views and (a-f) the respective bright-field micrographs. Horizontal bars in A and B indicate the regions enlarged and shown in E, F and G. The embryo sections are sagittal in E and oblique transverse through the visceral endoderm in F. Only one section is shown for the HNF1 probe, but other serial sections were also negative. Significant labeling is observed in the visceral extraembryonic endoderm (vee) and the uterine epithelium (ue). Abbreviations in this and Fig. 5 are: em, embryo; ee, embryonic ectoderm; exe, extraembryonic ectoderm; epc, ectoplacental cone; pac, proamniotic cavity; uc, uterine cavity; ue, uterine epithelium; vee, visceral extraembryonic endoderm.

Fig. 4.

Pattern of vHNF1 mRNA expression at embryonic day 6.5. Sections through the implantation site containing the embryo were hybridized with 35S-labeled antisense vHNF1 RNA probe, whereas the section shown in C was hybridized with HNF1 riboprobe. (A-F) Dark-field views and (a-f) the respective bright-field micrographs. Horizontal bars in A and B indicate the regions enlarged and shown in E, F and G. The embryo sections are sagittal in E and oblique transverse through the visceral endoderm in F. Only one section is shown for the HNF1 probe, but other serial sections were also negative. Significant labeling is observed in the visceral extraembryonic endoderm (vee) and the uterine epithelium (ue). Abbreviations in this and Fig. 5 are: em, embryo; ee, embryonic ectoderm; exe, extraembryonic ectoderm; epc, ectoplacental cone; pac, proamniotic cavity; uc, uterine cavity; ue, uterine epithelium; vee, visceral extraembryonic endoderm.

At 7.5 days of gestation, the egg cylinder continues to develop, the proamniotic cavity enlarges and the mesoderm forms between the posterior embryonic ectoderm and the endoderm. At this stage, vHNF1 transcripts are present in the columnar visceral endoderm, both in the extraembryonic part and in the middle region surrounding the mesoderm. The expression pattern of vHNF1 mRNA correlates spatially and temporally with that of transthyretin (Makover et al., 1989; Fig. 5C). However, it partially overlaps with that of AFP, which is confined to the visceral endoderm cells surrounding the embryonic part of the egg cylinder (Dziadek and Andrews, 1983; see also Fig. 5D,E). The synthesis of AFP in the visceral extraembryonic endoderm appears to be suppressed as a result of a inhibitory influence of the underlying extraembryonic ectoderm (Dziadek, 1978). At these two stages, no hybridization signal for HNF1 mRNA was detected in any embryo or extraembryonic structure, even after longer exposure (40 days; data not shown).

Fig. 5.

Coexpression of vHNF1 and TTR mRNAs at 7.5 days of gestation. Serial sections through a 7.5-day embryo were hybridized with 35S-labeled antisense vHNF1, TTR and AFP riboprobes, as indicated. (A-C) Dark-field views and (a-c) the respective bright-field views. For AFP hybridizations, only the dark-field images are shown (D,E). Coexpression of vHNF1 and TTR is observed in the columnar visceral endoderm (vee) overlying the extraembryonic ectoderm as well as in visceral endoderm cells overlaying the nascent mesoderm (m), whereas expression of AFP is observed in the visceral embryonic endoderm (ve) and, as vHNF1 and TTR, in the visceral endoderm surrounding the mesoderm. Parietal endoderm (pe), which is partially detached from the trophoblast in these sections, does not show significant labelling for either vHNF1, TTR or AFP. Serial sections hybridized with HNF1 probe did not give a positive signal in either embryo or extraembryonic structures (not shown). Additional abbreviations in this figure are: ac, amniotic cavity; ec, exocoelom; m, mesoderm; ve, visceral embryonic endoderm; pe, parietal endoderm; yc, yolk cavity.

Fig. 5.

Coexpression of vHNF1 and TTR mRNAs at 7.5 days of gestation. Serial sections through a 7.5-day embryo were hybridized with 35S-labeled antisense vHNF1, TTR and AFP riboprobes, as indicated. (A-C) Dark-field views and (a-c) the respective bright-field views. For AFP hybridizations, only the dark-field images are shown (D,E). Coexpression of vHNF1 and TTR is observed in the columnar visceral endoderm (vee) overlying the extraembryonic ectoderm as well as in visceral endoderm cells overlaying the nascent mesoderm (m), whereas expression of AFP is observed in the visceral embryonic endoderm (ve) and, as vHNF1 and TTR, in the visceral endoderm surrounding the mesoderm. Parietal endoderm (pe), which is partially detached from the trophoblast in these sections, does not show significant labelling for either vHNF1, TTR or AFP. Serial sections hybridized with HNF1 probe did not give a positive signal in either embryo or extraembryonic structures (not shown). Additional abbreviations in this figure are: ac, amniotic cavity; ec, exocoelom; m, mesoderm; ve, visceral embryonic endoderm; pe, parietal endoderm; yc, yolk cavity.

Thus the initial induction of transthyretin and AFP mRNAs is preceded by or concomitant with the expression of vHNF1 and occurs in the absence of detectable HNF1 expression.

Discrepancy between HNF1 and vHNF1 transcript and protein levels in the yolk sac and embryonic kidney

Since in several instances the presence of HNF1 transcripts is not correlated with the presence of DNA-binding protein (i.e. F9 and ES cells), we decided to examine whether similar non correlations are observed in embryo extracts. To this end, nuclear extracts from early embryos and microdissected developing organs were prepared and examined by gel mobility shift assays with specific antisera.

As shown in Fig. 6A, vHNF1 homodimers are detected in the nuclei of 7.5 d.p.c. whole embryos to 10.5 d.p.c. trunk embryos. The highest vHNF1 DNA-binding activity is present at 7.5 to 8.5 days of gestation, after which a decrease occurs until it becomes barely detectable at 10.5 days. In part, this decrease may reflect a reduction in a restricted number of positive cells relative to total embryo cells. Conversely, HNF1 is not detected in 7.5 to 9.5 d.p.c. embryos and becomes weakly detectable at 10.5 days of gestation. Thus, as expected, through 7.5 to 9.5 days of gestation, the DNA-binding complex detected in embryo nuclear extracts is composed essentially of vHNF1 homodimers.

Fig. 6.

vHNF1 and HNF1 binding activities in nuclear extracts of early embryos and developing organs. (A) Nuclear extracts of early embryos and yolk sacs. Nuclear extracts from the sources and at the days of gestation indicated were analyzed by gel mobility shift assays as described in Fig. 2D. When indicated, 1 μl of HNF1 antiserum (A-H) diluted 1:10 and 1 μl of undiluted vHNF1 antiserum (A-v) were used in the binding assays, prior to addition of labeled PE oligonucleotide. The amount of protein used in the assays of 7.5-day whole embryo, 8.5-day trunk embryo and 8.5-day yolk sac was roughly 8-fold, 3-fold and 4-fold less, respectively, than that used from the other extracts. (B) Nuclear extracts from microdissected developing organs at the indicated days of gestation were analyzed as described in A. (C) Western blot analysis using either total protein extracts or nuclear protein (NE) from 14.5-day, 15.5-day embryonic kidney (EK) and 15.5-day yolk sac (YS) were carried out as described in Material and methods. Nuclear protein from adult kidney and liver (AdK and AdL) are also included. To confirm the specificity of the detected bands, when indicated, 5 μg/ml of the corresponding antigen was included during the incubation with the first antibody. NS indicates a non-specific complex observed with anti-vHNF1 antibodies. The lower molecular weight band specifically recognized by the anti-HNF1 serum in adult liver is under investigation. Numbers shown to the left indicate positions of migration of molecular mass markers.

Fig. 6.

vHNF1 and HNF1 binding activities in nuclear extracts of early embryos and developing organs. (A) Nuclear extracts of early embryos and yolk sacs. Nuclear extracts from the sources and at the days of gestation indicated were analyzed by gel mobility shift assays as described in Fig. 2D. When indicated, 1 μl of HNF1 antiserum (A-H) diluted 1:10 and 1 μl of undiluted vHNF1 antiserum (A-v) were used in the binding assays, prior to addition of labeled PE oligonucleotide. The amount of protein used in the assays of 7.5-day whole embryo, 8.5-day trunk embryo and 8.5-day yolk sac was roughly 8-fold, 3-fold and 4-fold less, respectively, than that used from the other extracts. (B) Nuclear extracts from microdissected developing organs at the indicated days of gestation were analyzed as described in A. (C) Western blot analysis using either total protein extracts or nuclear protein (NE) from 14.5-day, 15.5-day embryonic kidney (EK) and 15.5-day yolk sac (YS) were carried out as described in Material and methods. Nuclear protein from adult kidney and liver (AdK and AdL) are also included. To confirm the specificity of the detected bands, when indicated, 5 μg/ml of the corresponding antigen was included during the incubation with the first antibody. NS indicates a non-specific complex observed with anti-vHNF1 antibodies. The lower molecular weight band specifically recognized by the anti-HNF1 serum in adult liver is under investigation. Numbers shown to the left indicate positions of migration of molecular mass markers.

Analysis of nuclear extracts from yolk sac show that at 8.5 d.p.c., although vHNF1 homodimers represent the major activity, HNF1 protein is also present essentially as HNF1:vHNF1 heterodimers (Fig. 6A). As development proceeds, however, there is a significant decrease of vHNF1 homodimers and a parallel increase of HNF1 homodimers. At 14.5 days of gestation, yolk sac nuclear extracts contain almost exclusively HNF1 homodimers and are qualitatively and quantitatively similar to embryo liver of the same developmental stage. In contrast, no drop in vHNF1 mRNA levels occurs between 8.5 days and 14.5 days, while only a 2-fold increase in the levels of HNF1 transcripts is observed over the same period (compare Fig. 3C with Fig. 6A).

Likewise, in the developing mouse kidney essentially vHNF1 activity is detected at 14.5 to 15.5 days of gestation, while HNF1 transcripts are readily detected in this organ from 14.5 days on, not only by the highly sensitive RT-PCR assays (Fig. 3C), but also by in situ hybridization (Ott et al., 1991). It is also relevent to note that, although similar levels of HNF1 transcripts are present in 14.5 and 15.5 d.p.c. embryonic liver, intestine and kidney, HNF1 binding activity is only detected in liver and intestine. Indeed, the binding activities detected in either embryo liver, intestine or lung are in good agreement with the levels of the respective mRNAs: mainly HNF1 homodimers are present in embryo liver, HNF1 and vHNF1 homodimers and heterodimers in the expected proportions in the developing intestine and exclusively vHNF1 homodimers in embryo lung (compare Fig. 3C with Fig. 6B). It is possible that the observed discrepancy between the HNF1 and vHNF1 binding activities and their transcript levels is due to changes in the affinity and/or in their translocation to the nucleus. To examine this, we performed western blot analysis on total protein extracts from embryo kidney and yolk sac. The results indicate that the levels of protein detected by this assay (Fig. 6C) are roughly in agreement with the levels of the respective binding activities detected by gel shift assays (Fig. 6A,B). Thus, neither the affinity of either HNF1 or vHNF1 nor their translocation to the nucleus appear to be severely perturbed in embryo kidney and yolk sac, respectively. Table 1 summarizes the mRNA levels and binding activities of HNF1 and vHNF1 at different embryonic stages and in developing organs.

Table 1.

Expression of HNF and vHNF1 mRNA and DNA-binding activities in embryonic cells

Expression of HNF and vHNF1 mRNA and DNA-binding activities in embryonic cells
Expression of HNF and vHNF1 mRNA and DNA-binding activities in embryonic cells

Taken together, these results suggest that both vHNF1 and HNF1 may be posttranscriptionally regulated.

Developmental expression of HNF4 protein

At present, there is little information about the earliest appearance of other hepatocyte transcription factors during development (Lai and Darnell, 1991; see also Discussion).

Recent studies indicated that HNF1 and HNF4, a member of the steroid hormone receptor superfamily (Sladek et al., 1991), are regulated in a hierarchy by a higher-order locus (Kuo et al., 1992). HNF4 is a crucial transactivator of HNF1 (Kuo et al., 1992; Tian and Schibler, 1991) and, more importantly, stable transfection of HNF4 cDNA in dedifferentiated hepatoma cells rescues the expression of endogenous HNF1 mRNA and DNA-binding activity (Kuo et al., 1992). HNF1 and HNF4 appear also to cooperate in the transcriptional activation of several of the liver-specific genes, notably A1-antitrypsin and transtherytin, which as mentioned express in the visceral yolk sac (Monaci et al., 1988; Costa et al., 1989).

This prompted us to examine the HNF4 binding activity in the embryo nuclear extracts and define whether a correlation between HNF4 and HNF1 expression is observed.

Several unrelated proteins, in addition to HNF4, are able to bind to the HNF4 binding site present in different promoters (Sladek et al., 1991). We choose as probe the HNF4 sequence present in the Apo CIII promoter. This site has been recently shown to be the target for two other members of the steroid receptor class of transcription factors, the apo AI regulatory binding protein 1 (ARP-1) and COUP-TF (Mietus-Snyder et al., 1992 and references therein). To identify these factors, we included in the gel shift assay polyclonal HNF4 antiserum (kindly provided by M. Zhong and J. Darnell) and polyclonal COUP-TF antiserum (kindly provided by M. J. Tsai) which recognises both COUP-TF and ARP-1, but not HNF4 protein (Mietus-Snyder et al., 1992).

As shown in Fig. 7A, HNF4 binding activity is indeed detected both in the yolk sac and in the embryo proper from 7.5 days. The protein complexes detected in early embryo extracts from 8.5 to 9.5 days of gestation react only with COUP-TF-antibodies (Fig. 7B and results not shown). Note that COUP-TF/ARP-1 also appears to constitute the major DNA-binding activity detected in 14.5 day developing kidney and intestine, two organs that express HNF1 mRNA (see Fig. 3C). In contrast, HNF4 binding activity is predominant in the yolk sac from 8.5 days throughout gestation, in embryonic liver and also in both parietal and visceral F9 derivatives (Fig. 7B).

Fig. 7.

HNF4 and COUP-TF/Arp-1 binding activities in nuclear extracts of early embryos, developing organs and differentiated F9 cells. (A) ‘HNF4’ binding activities in embryonic nuclear extracts. Gel shift assays were carried out using as probe a ds oligonucleotide encompassing the HNF4 recognition sequence present in the apo CIII promoter and the embryonic nuclear extracts shown in Fig. 6. For each extract, binding assays were done in the absence and in the presence of 40-fold molar cold oligonucleotide as competitor. (B) Identification of HNF4 and COUP-TF+Arp-1 binding activities in embryonic and F9 parietal and visceral nuclear extracts. When indicated, nuclear extracts were incubated with 1 μl of HNF4 antiserum (Ab -HNF4) diluted 1:6 and 1 μl of COUP-TF antiserum (Ab-COUP) diluted 1:5, before the addition of the labeled HNF4-apoCIII probe. Thin and thick arrows indicate the supershifts generated by HNF4 and COUP-TF antibodies, respectively. Only extracts in which both antibodies were used are shown; the complex observed in 7.5-day embryo extracts in A is not affected by HNF4 antibody (not shown).

Fig. 7.

HNF4 and COUP-TF/Arp-1 binding activities in nuclear extracts of early embryos, developing organs and differentiated F9 cells. (A) ‘HNF4’ binding activities in embryonic nuclear extracts. Gel shift assays were carried out using as probe a ds oligonucleotide encompassing the HNF4 recognition sequence present in the apo CIII promoter and the embryonic nuclear extracts shown in Fig. 6. For each extract, binding assays were done in the absence and in the presence of 40-fold molar cold oligonucleotide as competitor. (B) Identification of HNF4 and COUP-TF+Arp-1 binding activities in embryonic and F9 parietal and visceral nuclear extracts. When indicated, nuclear extracts were incubated with 1 μl of HNF4 antiserum (Ab -HNF4) diluted 1:6 and 1 μl of COUP-TF antiserum (Ab-COUP) diluted 1:5, before the addition of the labeled HNF4-apoCIII probe. Thin and thick arrows indicate the supershifts generated by HNF4 and COUP-TF antibodies, respectively. Only extracts in which both antibodies were used are shown; the complex observed in 7.5-day embryo extracts in A is not affected by HNF4 antibody (not shown).

Thus, HNF4 protein, like HNF1, shows a delayed appearance relative to vHNF1 and the activation of liver-specific markers in the visceral yolk sac.

The present study was undertaken to examine whether HNF1 and/or vHNF1 are expressed in a temporal and spatial pattern in early postimplantation embryos that is consistent with their potential role as transcriptional regulators of several cell-specific markers. The fact that both HNF1 and vHNF1 possess the same DNA sequence specificity and heterodimerize raised the intriguing question whether each of these homeoproteins display distinct or similar developmental function. We focused our attention on the expression pattern of HNF1 and vHNF1 during the formation of visceral endoderm of the yolk sac.

Our results show that induction of the putative HNF1 target genes in the visceral endoderm is preceded by and/or concomitant with the expression of vHNF1 but not HNF1. similar sequence of events seems to occur in vitro after differentiation of F9 cells: vHNF1 (mRNA and protein), but not HNF1, is highly induced in visceral derivatives. However, no vHNF1 transcripts were detected in the parietal yolk sac by in situ hybridization, while vHNF1 is also present in F9 parietal derivatives. We do not know yet whether vHNF1 is expressed transiently in the parietal endoderm at stages earlier than those examined here.

While this work was in progress, another study reporting the regulation of vHNF1 and HNF1 after differentiation of F9 cells into visceral endoderm was published (Kuo et al., 1991). That study and ours agree with regard to the earlier induction of both vHNF1 mRNA and protein relative to HNF1. However, in contrast to our results, Kuo et al. (1991) showed an increase in HNF1 mRNA of ∼4-fold after 5 days of RA treatment of F9 cells in suspension, which correlates with a concomitant elevation of HNF1 binding activity, as measured by a supershift generated by polyclonal antibodies. At present, we have no explanation for this discrepancy. It is plausible that the F9 cell lines used in each study partially differ in the differentiation state induced.

Whatever the case, in vivo no HNF1 transcripts were detected by in situ hybridization in 6.5 to 7.5 days embryos; a weak signal was only observed by RT-PCR at 7.5 days of gestation. In contrast, high levels of vHNF1 mRNA are already present at 6.5 days of gestation in the visceral extraembryonic endoderm, which co-localizes with TTR and RBP transcripts, first detected at 7 –7.5 days of gestation (Makover et al., 1988, present study : Fig. 5). Altogether the results suggest that vHNF1, but not HNF1, may participate in the initial transcriptional activation of these genes. Maintenance of their expression may require additional expression of HNF1, which appears in the visceral yolk sac roughly 2 embryonic days later (Ott et al., 1991; Blumenfeld et al., 1991). Interestingly, at least with regard to the yolk sac, vHNF1 and HNF1 proteins appear to display a reciprocal regulation during development (decrease of vHNF1 and concomitant increase of HNF1 binding activity), whereas the levels of both mRNAs remain similarly high throughout development.

The sequential expression of vHNF1 followed by HNF1 during kidney organogenesis led to the suggestion that vHNF1 may be critical in the early inductory phase characterized by a heterotypic inductive interaction between the epithelium of the ureter bud and the condensed mesenchymal cells, whereas HNF1 is activated later, during the postinductory phase, in already committed cells (Lazzaro et al., 1992). Since inductive interactions with mesenchyme are also required for endodermal cells to differentiate in hepatocytes (LeDouarin, 1975; Houssaint, 1980), vHNF1 and HNF1 may also execute distinct roles in the inductory and postinductory phases during liver organogenesis. Indeed, vHNF1, but not HNF1 mRNA, is detected in the hepatic precursor cells (Ott et al., 1991), which already express the cell-specific markers AFP and albumin (Schmid and Schulz, 1990; Cascio and Zaret, 1991). HNF1 mRNA is first detected in the liver primordium at a later developmental stage (10.5 days), after the initiation of the hepatic cell lineage (Ott et al., 1991).

It is likely that the transcriptional activation of the various HNF1-dependant functions induced either in the yolk sac or liver are due to the combinatorial action of various factors of both positive and negative action (for a discussion see Kuo et al., 1990a). How is the expression of HNF1/vHNF1 related to other families of transcription factors involved in hepatocyte-specific expression. The earliest expression of C/EBP mRNA was observed in fetal liver at 13 days of gestation (Kuo et al., 1990a), which is around four days after the formation of the liver primordium. Members of the extended HNF3/fork head family are expressed not only in the gut-endoderm-derived tissues, but also in the developing rat brain (Tao and Lai, 1992 and references therein). Although no data are still available about the earliest expression of the different HNF3 genes in the mouse, the emerging picture is that HNF3 is a large family of transcription factors, with distinct subfamilies each implicated in differentiation of different cell lineages (Dirksen and Jamrich, 1992; Grossniklaus et al., 1992).

Finally, our results suggest that HNF4, as HNF1, may appear after initiation of visceral endoderm and hepatic cell-lineage differentiation. HNF4 and HNF1 seem to be related by a transcriptional hierarchy in which HNF4 regulates the expression of HNF1 and, ultimately, both cooperate or synergize to activate transcription of downstream genes (Tian and Schibler, 1991; Kuo et al., 1992 and references therein). The developmental role of HNF4 was further suggested by mutagenesis studies in transgenic mice showing that disruption of either HNF1 or HNF4 binding sites present in the promoter-enhancer region of the A1-antitrypsin gene drastically affect its expression in fetal liver and yolk sac, whereas only the HNF1 mutation has a marked effect in adult liver (Tripodi et al., 1991). Since HNF4 is not regulated by HNF1 (Tian and Schibler, 1991; Kuo et al., 1992), we predicted that its expression was an early event during development. In contrast, no HNF4, but COUP-TF/ARP-1 binding activities were detected in embryo extracts from 7.5 to 9.5 days of gestation. It remains to be established whether COUP-TF and/or ARP-1, although they antagonize the HNF4-transactivation of the ApoCIII promoter in liver and intestinal cells (Mietus-Snyder et al., 1992), may function differently with other promoters or in different cell types.

Two alternative spliced vHNF1 RNA isoforms are present in embryonic and adult tissues

We have shown that vHNF1 is expressed as two alternative spliced isoforms (vHNF1-A and vHNF1-B), which differ by the utilisation of an alternative exon of 78 nt present between the POU-like domain and the homeodomain. vHNF1-A, the transcript containing this extra exon, is predominant both in embryonic and adult tissues. However, the ratio of vHNF1-A/vHNF1-B mRNAs in adult liver and intestine is roughly 3- to 6-fold higher than in other tissues, suggesting that a particular ratio of these two transcripts may be functionally relevant. This pattern of alternative splicing of vHNF1 has been conserved throughout evolution, since both transcripts are also present in adult human liver (Bach et al., 1991 and personal communication).

Alternative splicing appears to be a general mechanism of generating multiple isoforms of transcriptional regulators, which in some cases results in dramatic changes in their function (i.e. a transactivator switched to a repressor) and can modulate the developmental function of a transcription factor in a tissue-specific manner (see Foulkes and Sassone-Corsi, 1992, for a recent review). Transient cotransfection experiments indicate that vHNF1-A and vHNF1-B exhibit similar transactivation potential, suggesting that these two isoforms likely do not possess antagonistic or distinct function (S. Power and P. M. Shaw unpublished observations). Further studies are required to establish whether these two transcripts are equally translated during development and whether in vivo functional differences do exist between vHNF-A and B.

Non-correlation between the developmental levels of HNF1 and vHNF1 transcripts and the respective DNA-binding proteins

In previous studies, we have shown that HNF1 expression in differentiated hepatoma cells is controlled primarily at the transcriptional level (Cereghini et al., 1990). Similarly, the high level of induction of vHNF1 mRNA in F9 visceral or parietal derivatives is due to a similar increase in the rate of transcription (S. Cereghini, unpublished data). Thus, it is likely that the appearance of HNF1 and vHNF1 transcrips in a restricted manner during development, is due to transcriptional activation.

The present results suggest a posttranscriptional level of control of HNF1 and vHNF1 gene expression, in addition to spatially regulated transcription. The first evidence of such additional level of control came from our studies in the F9 system, where we have been unable to detect HNF1 DNA-binding activity or protein in non-differentiated F9 cells and its differentiated derivatives, although HNF1 transcripts were present.

Similarly, during mouse development, we observe quite high levels of HNF1 mRNA in the developing kidney, but little or no active binding protein. Reciprocally, vHNF1 binding activity, while easily detected in the yolk sac at 8.5 d.p.c., becomes undetectable at 14.5 days of gestation, whereas the levels of vHNF1 mRNA are similarly high at both embryonic stages (see Fig. 6 and Table 1). Western blot analyses suggest that the DNA-binding affinity of both HNF1 and vHNF1 and/or their localization to the nucleus appear not to be affected. Moreover, transient transfection studies with either HNF1, vHNF1-A or vHNF1-B expression vectors do not indicate that they are unstable proteins. Thus, the incongruity between transcript accumulation and binding activity in mouse embryo extracts suggests a posttranscriptional level of control of vHNF1 and HNF1.

One attractive hypothesis is either translational activation or repression of vHNF1 and HNF1 transcripts during development. In certain situations, i.e. the yolk sac, the result is a reciprocal expression of each activity, in others only one species is modulated (i.e. HNF1 in developing kidney). Indeed, there are some mammalian regulatory proteins whose expression appears to be controlled posttranscriptionally during development. The appearence of the pituitary-specific transcription GHF-1/pit-1 factor (Dollé et al., 1990; Simmons et al., 1990) and the myogenic factor myogenin (Cusella-De Angelis et al., 1992) is delayed by 1 or 2 days relative to the respective transcripts. The actual mechanisms are still unknown. The expression of different members of the C/EBP family of transcription factors is also controlled posttranscriptionally (Mueller et al., 1990). In particular, the gene encoding the liver enriched transcriptional activator protein LAP is expressed in several tissues, but the protein is considerably more abundant in terminally differentiated parenchymal hepatocytes (Descombes et al., 1990). Additional complexity is achieved by the generation of a repressor form, LIP, by the use of an alternative translational start site (Descombes and Schibler, 1991).

Clearly, further studies, in addition to the subcellular localization of HNF1 and vHNF1 in the developing embryo, are required to elucidate the various mechanisms that mediate the observed discrepancies between transcript accumulation and nuclear binding proteins. Moreover, an accessory factor (DCoH: Dimerization Cofactor of HNF1) that does not bind DNA, but acts to stabilize dimerization of both HNF1 and vHNF1, has been recently characterized (Mendel et al., 1991b). We have observed that in all the cells where vHNF1 is induced or present (i.e. visceral and parietal F9 cells, 3TDM, PYS-2) co-expression of DCoH mRNA occurs (S. Cereghini, unpublished results). Yet, the developmental regulatory activity of either HNF1 or vHNF1 can be modulated by variations in the levels of the DCoH protein. Furthermore, it has been recently reported that nascent in vitro elongated HNF1 transcripts are present in certain adult tissues, such as lung and spleen that contain neither HNF1 mRNA nor protein (Xanthopoulos et al., 1991). Thus, the mechanisms that control the expression of vHNF1, and particularly HNF1, seem unexpectedly complex.

Finally, it is tempting to speculate that the distinct temporal and spatial expression patterns of vHNF1 and HNF1 during development are consistent with the suggestion that they may participate in the regulation of different and sequential steps of the visceral endodermal, hepatic and kidney differentiation.

Separate regulatory mechanisms for the establishment and subsequent maintenance of tissue-specific gene transcription have been suggested in several mammalian differentiation systems (Vogt et al., 1988; Mueller et al., 1990; Cascio and Zaret, 1991). It is possible that vHNF1 and HNF1 proteins are both required to ensure the progression through the terminally differentiated phenotype. It remains to be established, however whether HNF1 can functionally replace vHNF1 during development. Gene disruption may answer this question.

We thank M. C. Weiss for support and valuable discussions, I. Bach for helpful discussions and communicating unpublished results, and J. P Rousset, J. F. Nicolas and P. M. Shaw for critical reading of the manuscript. We are grateful to members of the Department of Molecular Biology for help and advice during realization of this work, particularly H. Condamine, C. Kress, E. Coudrier, D. Montarras, C. Pinset and J. F. Nicolas. We thank M. Zhong and J. Darnell for generous gift of HNF4 antiserum, M. J. Tsai for COUP-TF/ARP-1 antiserum, S. Thilghman and W. Blanes for AFP and TTR probes, respectively, and B. Bourachot and M. Yaniv for help in the initial experiments of differentiation of F9 cells. This work was supported by grants from ARC, INSERM. S. Power held a fellowship from the BRIDGE program from EEC.

Bach
,
I.
,
Mattei
,
M. G.
,
Cereghini
,
S.
and
Yaniv
,
M.
(
1991
).
Two members of an HNF1 homeoprotein family are expressed in human liver
.
Nucleic Acids Res
.
19
,
3553
3559
.
Baumhueter
,
S.
,
Courtois
,
G.
and
Crabtree
,
G. R.
(
1988
).
A variant nuclear protein in dedifferentiated hepatoma cells binds to the same functional sequences in the b fibrinogen gene promoter as HNF-1
.
EMBO J
.
7
,
2485
2493
.
Blumenfeld
,
M.
,
Maury
,
M.
,
Chouard
,
T.
,
Yaniv
,
M.
and
Condamine
,
H.
(
1991
).
Hepatic nuclear factor 1 (HNF1) shows a wider distribution than products of its known target genes in developing mouse
.
Development
113
,
589
599
.
Cascio
,
S.
and
Zaret
,
K. S.
(
1991
).
Hepatocyte differentiation initiates during endodermal-mesenchymal interactions prior to liver formation
.
Development
113
,
217
225
.
Cereghini
,
S.
,
Blumenfeld
,
M.
and
Yaniv
,
M.
(
1988
).
A liver-specific factor essential for albumin transcription differs between differentiated and dedifferentiated rat hepatoma cells
.
Genes Dev
.
2
,
957
974
.
Cereghini
,
S.
,
Yaniv
,
M.
and
Cortese
,
R.
(
1990
).
Hepatocyte dedifferentiation and extinction is accompanied by a block in the synthesis of mRNA coding for the transcription factor HNF1/LFB1
.
EMBO J
.
9
,
2257
2263
.
Costa
,
R. H.
,
Grayson
,
D. R.
and
Darnell Jr
,
J. E.
(
1989
).
Multiple hepatocyte-enriched nuclear factors function in the regulation of transthyretin and a1-antitrypsin genes
.
Mol. Cell. Biol
.
9
,
1415
1425
.
Cusella-De Angelis
,
M. G.
,
Lyons
,
G.
,
Sonino
,
C.
,
De-Angelis
,
L.
,
Vivarelli
,
E.
,
Karmer
,
K.
,
Wrigth
,
W. E.
,
Molinaro
,
M.
,
Bouche
,
M.
,
Buckingham
,
M.
and
Cossu
,
G.
(
1992
).
Myo D, myogen independent differentiation of primordial myoblasts in mouse somites
.
J. Cell Biol
.
116
,
1243
1255
.
Deschatrette
,
J.
and
Weiss
,
M. C.
(
1974
).
Characterization of differentiated and dedifferentiated clones from a rat hepatoma
.
Biochimie
56
,
1603
1611
.
Deschatrette
,
J.
,
Moore
,
E. E.
,
Dubois
,
M.
,
Cassio
,
D.
and
Weiss
,
M. C.
(
1979
).
Dedifferentiated variants of a rat hepatoma : analysis by cell hybridization
.
Somat. Cell Genet
.
5
,
697
718
.
Descombes
,
P.
,
Chojkier
,
M.
,
Lichtsteiner
,
S.
,
Falvey
E.
and
Schibler
,
U.
(
1990
).
LAP, a novel member of the C/EBP gene family, encodes a liver-enriched transcriptional activator protein
.
Genes Dev
.
4
,
1541
1551
.
Descombes
,
P.
and
Schibler
,
U.
(
1991
).
A liver-enriched transcriptional activator protein, LAP, and a transcriptional inhibitory protein, LIP, are translated from the same mRNA
.
Cell
67
,
569
579
.
De Simone
,
V.
,
De Magistris
,
L.
,
Lazzaro
,
D.
,
Gerstner
,
J.
,
Monaci
,
P.
,
Nicosia
,
A.
and
Cortese
,
R.
(
1991
).
LFB3, a heterodimer forming homeoprotein of the LFB1 family, is expressed in specialized epithelia
.
EMBO J
.
10
,
1435
1443
.
De Simone
,
V.
and
Cortese
,
R.
(
1991
).
Transcritional regulation of liver-specific gene expression
.
Curr. Opin. Cell Biol
.
3
,
960
965
.
Dirksen
,
M. L.
and
Jamrich
,
M.
(
1992
).
A novel, activin inducible, blastopore lip-specific gene of Xenopus laevis contains a fork-head DNA-binding domain
.
Genes Dev
.
6
,
599
608
.
Dollé
,
P.
,
Castrillo
,
J. L.
,
Theill
,
L. E.
,
Deerinck
,
T.
,
Ellisman
,
M.
and
Karin
,
M.
(
1990
).
Expression of GHF-1 protein in mouse pituitaries correlates both temporally and spatially with onset of growth hormone gene activity
.
Cell
60
,
809
820
.
Dziadek
,
M.
(
1978
).
Modulation of alphafetoprotein synthesis in the early postimplantation mouse embryo
.
J. Embryol. Exp. Morph
.
46
,
135
146
.
Dziadek
,
M. A.
and
Andrews
,
G. K.
(
1983
).
Tissue specificity of the alpha-fetoprotein messenger expression during mouse embryogenesis
.
EMBO J
.
4
,
549
554
.
Foulkes
,
N. S.
and
Sassone-Corsi
,
P.
(
1992
).
More is better : activators and repressors from the same gene
.
Cell
68
,
411
414
.
Grover
,
A.
,
Oshima
,
R.
and
Adamson
,
E. D.
(
1983
).
Epithelial layer formation in differentiating aggregates of F9 embryonal carcinoma cells
.
J. Cell Biol
.
96
,
1690
1696
.
Grover
,
A.
and
Adamson
,
E. D.
(
1986
).
Evidence for the existence of an early common biochemical pathway in the differentiation of F9 cells into visceral or parietal endoderm: modulation by cyclic AMP
.
Dev. Biol
.
114
,
492
503
.
Grossniklaus
,
U.
,
Pearson
,
R. R.
and
Gehring
,
W.
(
1992
).
The Drosophila sloppy paired locus encodes two proteins involved in segmentation that show homoloy to mammalian transcription factors
.
Genes Dev
.
6
,
1030
1051
.
Hogan
,
B. L. M.
,
Taylor
,
A.
and
Adamson
,
E.
(
1981
).
Cell interations modulate embryonal carcinoma cell differentiation into parietal and visceral endoderm
.
Nature
291
,
235
237
.
Houssaint
,
E.
(
1980
).
Differentiation of the mouse hepatic primordium. I. An analysis of tissue interactions in hepatocyte differentiation
.
Cell Differ
.
9
,
269
279
Kuo
,
C. F.
,
Xanthopoulos
,
K. G.
and
Darnell,
J.E.
Jr.
(
1990a
).
Fetal and adult localization of C/EBP : evidence for combinatorial action of transcription factors in cell-specific gene expression
.
Development
109
,
473
481
.
Kuo
,
C. J.
,
Conley
,
P. B.
,
Hsieh
,
C. L.
,
Francke
,
U.
and
Crabtree
,
G. R.
(
1990b
).
Molecular cloning, functional expression, and chromosomal localization of mouse hepatocyte nuclear factor 1
.
Proc. Natl. Acad. Sci. USA
87
,
9838
9842
.
Kuo
,
C. J.
,
Mendel
,
D. B.
,
Hansen
,
L. P.
and
Crabtree
,
G. R.
(
1991
).
Independent regulation of HNF-1a and HNF-1b by retinoic acid in F9 teratocarcinoma cells
.
EMBO J
.
10
,
2231
2236
.
Kuo
,
C. J.
,
Conley
,
P. B.
,
Chen
,
L.
,
Sladek
,
F. M.
,
Darnell
,
J. E.
Jr.
and
Crabtree
,
G. R.
(
1992
).
A transcriptional hierarchy involved in mammalian cell-type specification
.
Nature
355
,
457
461
.
Lai
,
E.
and
Darnell
Jr.,
J. E.
(
1991
).
Transcriptional control in hepatocytes: a window on development
.
Trends Biochem. Sci
.
16
,
427
430
.
Lazzaro
,
D.
,
De Simone
,
V.
,
De Magistris
,
L.
,
Lehtonen
,
E.
and
Cortese
,
R.
(
1992
).
LFB1 and LFB3 homeoproteins are sequentially expressed during kidney development
.
Development
114
,
469
479
.
LeDouarin
,
N. M.
(
1975
).
An experimental analysis of liver development
.
Med. Biol
.
53
,
427
455
.
Lichtsteiner
,
S.
and
Schibler
,
U.
(
1989
).
A glycosylated liver-specific transcription factor stimulates transcription of the albumin gene
.
Cell
57
,
1179
1187
.
Maire
,
P.
,
Wuarin
,
J.
and
Schibler
,
U.
(
1989
).
The role of cis-acting promoter elements in tissue-specific albumin gene expression
.
Science
244
,
343
346
.
Makover
,
A.
,
Soprano
,
D. R.
,
Wyatt
,
M.
and
Goodman
,
D. S.
(
1989
).
An instu-hybridization study of the localization of retinol-binding protein and tranthyretin messenger RNAs during fetal development in the rat
.
Differentiation
40
,
17
25
.
Meehan
,
R. R
,
Barlow
,
D. P.
,
Hill
,
R. E.
,
Hogan
,
B. L. M.
and
Hastie
,
N. D.
(
1984
).
Pattern of serum protein gene expression in mouse visceral yolk sac and foetal liver
.
EMBO J
.
3
,
1881
1885
.
Melton
,
D. W.
,
Konecki
,
D. S.
,
Brennand
,
J.
and
Caskey
,
C. T.
(
1984
).
Structure, expression and mutation of the hypoxanthine phosphribosyltransferase gene
.
Proc. Natl. Acad. Sci. USA
.
81
,
2147
2151
.
Mendel
,
D. B.
,
Hansen
,
L. P.
,
Graves
,
M. K.
,
Conley
P. B.
and
Crabtree
,
G. R.
(
1991a
).
HNF-1a and HNF-1b (vHNF-1) share dimerization and homeo domains, but not activation domains, and form heterodimers in vitro
.
Genes Dev
.
5
,
1042
1056
.
Mendel
,
D. B.
,
Khavari
,
P. A.
,
Conley
,
P. B.
,
Graves
,
M. K.
,
Hansen,
L.P.
Admon,
A.
and
Crabtree
G. R.
(
1991b
).
Characterization of a cofactor that regulates dimerization of a mammalian homeodomain protein
.
Science
254
,
1762
1767
.
Mietus-Snyder
,
M.
,
Sladek
,
F. M.
,
Ginsburg
,
G. S.
,
Kuo
,
C. F.
,
Ladias
,
J. A. A.
,
Darnell
,
J. E. Jr.
and
Karathanasis
,
S. K.
(
1992
).
Antagonism between apolipoprotein AI regulatory protein 1, Ear3 / COUP-TF, and hepatocyte nuclear factor 4 modulates apolipoprotein CIII gene expression in liver and intestinal cells
.
Mol. Cell. Biol
.
12
,
1708
1718
.
Monaci
,
P.
,
Nicosia
,
A.
and
Cortese
,
R.
(
1988
).
Two different liver-specific factors stimulate in vitro transcription from the human a1-antitrypsin promoter
.
EMBO J
.
7
,
2075
2087
.
Montarras
,
D.
,
Pinset
,
C.
,
Chelly
,
J.
,
Kahn
,
A.
and
Gros
,
F.
(
1989
).
Expression of MyoD1 coincides with terminal differentiation in determined but inducible muscle cells
.
EMBO J
.
8
,
2203
2207
.
Mueller
,
C. R.
,
Maire
,
P.
and
Schibler
,
U.
(
1990
).
DBP, a liver-enriched transcriptional activator, is expressed late in ontogeny and its tissue specificity is determined posttranscriptionally
.
Cell
61
,
279
291
.
Nicolas
,
J. F.
,
Avner
,
P.
,
Gaillard
,
J.
,
Guenet
,
J. L.
,
Jakob
,
H.
and
Jacob
,
F.
(
1976
).
Cell lines derived from teratocarcinomas
.
Cancer Res
.
36
,
4224
4231
.
Nicosia
,
A.
,
Monaci
,
P.
,
Tomei
,
L.
,
De Francesco
,
R.
,
Nuzzo
,
M.
,
Stunnenberg
,
H.
and
Cortese
,
R.
(
1990
).
A myosin-like dimerization helix and an extra-large homeodomain are essential elements of a tripartite DNA binding structure of LFB1
.
Cell
61
,
1225
1236
.
Oshima
,
R. G.
(
1982
).
Developmental expression of murine extraembryonic endodermal cytoskeletal proteins
.
J. Biol. Chem
.
257
,
3414
3421
.
Ott
,
M. O.
,
Rey-Campos
,
J.
,
Cereghini
,
S.
and
Yaniv
,
M.
(
1991
).
VHNF1 is expressed in epithelial cells of distinct embryonic origin during development and precedes HNF1 expression
.
Mech. Dev
.
36
,
47
58
.
Rey-Campos
,
J.
,
Chouard
,
T.
,
Yaniv
,
M.
and
Cereghini
,
S.
(
1991
).
VHNF1 is a homeoprotein that activates transcription and forms heterodimers with HNF1
.
EMBO J
.
10
,
1445
145
.
Rugh
,
R.
(
1990
).
The Mouse: its Reproduction and Development
.
Oxford University Press
.
Oxford
.
Ryffel
,
G. U
,
Kugler
,
W.
,
Wagner
,
U.
, and
Kaling
,
M.
(
1989
).
Liver cell specific transcription in vitro : the promoter elements HP1 and TATA box are necessary and sufficient to generate a liver-specific promoter
.
Nucleic Acids Res
.
17
,
939
953
.
Schmid
,
P.
and
Schulz
,
W. A.
(
1990
).
Coexpresion of the c-MYC protooncogene with a-fetoprotein and albumin in fetal mouse liver
.
Differentiation
45
,
96
102
.
Scott
,
R. W.
,
Vogt
,
T. F.
,
Croke
,
M. E.
and
Thilghman
,
S. M.
(
1984
).
Tissue-specific activation of a cloned alpha-fetoprotein gene during differentiation of a transfected embryonal carcinoma cell line
.
Nature
310
,
562
567
.
Simmons
,
D. M.
,
Voss
,
J. W.
,
Ingraham
,
H. A.
,
Holloway
,
J. M.
,
Broide
,
R. S.
,
Rosenfeld
,
M. G.
and
Swanson
,
L. W.
(
1990
).
Pituitary cell phenotypes involve cell-specific Pit mRNA translation and synergistic interactions with other classes of transcription factors
.
Genes Dev
.
4
,
695
711
.
Sladek
,
F. M.
,
Zhong
,
W.
,
Lai
,
E.
, and
Darnell,
J.E.
Jr.
(
1990
).
Liver-enriched transcription factor HNF-4 is a novel member of the steroid hormone receptor superfamily
.
Genes Dev
.
4
,
2353
2365
.
Smith
,
D. B.
and
Johnson
,
K. S.
(
1988
).
Single-step purification of polypeptides expressed in Escherichia coli as fusions with glutathione S-transferase
.
Gene
67
,
31
40
.
Soprano
,
R. D.
,
Soprano
,
K. J.
,
Wyatt
,
M. L.
and
Goodman
,
W. S.
(
1988
).
Induction of the expression of Retinol Binding Protein and Transthyretin in F9 embryonal carcinoma cells differentiated to embryoid bodies
.
J. Biol. Chem
.
263
,
17897
17900
.
Strickland
,
S.
,
Smith
,
K. K.
and
Marotti
,
K. R.
(
1980
).
The hormonal induction of differentiation in teratocarcinoma stem cells: Generation of parietal endoderm by retinoic acid and dibutyryl cyclic AMP
.
Cell
21
,
347
355
.
Tao
,
W.
and
Lai
,
E.
(
1992
).
Telencephalon-restricted expression of BF-1, a new member of the HNF-3/ fork head gene family, in the developing rat brain
.
Neuron
8
,
957
966
.
Tian
,
J.-M.
and
Schibler
,
U.
(
1991
).
Tissue-specific expression of the gene encoding hepatocyte nuclear factor 1 may involve hepatocyte nuclear factor 4
.
Genes Dev
.,
5
,
2225
2234
.
Tripodi
,
M.
,
Abbott
,
C.
,
Vivian
,
N.
,
Cortese
,
R.
and
Lovell-Badge
,
R.
(
1991
).
Disruption of the LF-A1 and LF-B1 binding sites in the human alpha-1-antitrypsin gene has a differential effect during development in transgenic mice
.
EMBO J
.
10
,
3177
3182
.
Tronche
,
F.
,
Rollier
,
A.
,
Herbomel
,
P.
,
Bach
,
I.
,
Cereghini
,
S.
,
Weiss
,
M. C.
, and
Yaniv
,
M.
(
1990
).
Anatomy of the rat albumin promoter
.
Mol. Biol. Med
.
7
,
173
185
.
Tyner
,
A. L.
,
Godbout
,
R.
,
Compton
,
R. S.
and
Tilghman
,
S. M.
(
1990
).
The ontogeny of a-fetoprotein gene expression in the mouse gastrointestinal tract
.
J. Cell Biol
.
110
,
915
927
.
Vogt
,
T. F.
,
Compton
,
R. S.
,
Scott
,
R. W.
and
Tilghman
,
S. M.
(
1988
).
Differential requirements for cellular enhancers in stem and differentiated cells
.
Nucl. Acids Res
.
16
,
487
500
.
Xanthopoulos
,
K. G.
,
Prezioso
,
V. R.
,
Chen
,
W. S.
,
Sladek
,
F. M.
,
Cortese
R.
, and Darnell,
J. E.
Jr
. (
1991
).
The different tissue transcription patterns of genes for HNF-1, C/EBP, HNF-3, and HNF-4, protein factors that govern liver-specific transcription
.
Proc. Natl. Acad. Sci. USA
88
,
3807
3811
.