ABSTRACT
The tissue-specific transcription factors LFB1 (HNF1) and LFB3 (vHNF1) mainly expressed in liver, kidney and intestine are homeoproteins that interact with the regulatory element HP1. The HP1 sequence constitutes one of the most important cis-acting elements in liver-specifically expressed genes, while its function in other cell types containing LFB1 and LFB3 is not fully understood. In mammals, LFB1 activity is modulated by DCoH, a cofactor that stimulates the LFB1 transactivation significantly. Using the rat cDNA probe, we cloned the corresponding Xenopus sequence XDCoH, encoding a 104 amino acid protein, that is 85% identical to the rat protein. XDCoH enhances the LFB1-dependent transactivation potential in transfection experiments and interacts in vitro directly with LFB1 and its variant form LFB3. The protein is detectable in liver and kidney extracts of adult frogs and in small amounts also in lung and stomach, organs expressing LFB1 and/or LFB3 protein as well.
To investigate the possible involvement of XDCoH in Xenopus development, we analyzed its temporal and spatial expression pattern during early embryogenesis. XDCoH is a maternal factor, although LFB1 is absent in the egg. In early cleavage stages, the protein is detectable in the cytoplasm of each blastomere and enters the nuclei of the cells as early as the zygotic transcription in the Xenopus embryo starts. The amount of XDCoH increases dramatically following neurulation, when the formation of liver, pronephros and other organs takes place. Whole-mount immunostaining demonstrates that, in the developing larvae, XDCoH is localized in the nuclei of the hepatocytes, the gut cells and the pronephric cells, tissues of mesodermal and endodermal origin known to contain LFB1 and LFB3. Surprisingly it is also present in the pigmented epithelium surrounding the eye of the embryo, which is derived from the anterior part of the ectodermal neural plates and lacks LFB1. The tissue distribution of XDCoH during embryogenesis suggests that XDCoH is involved in determination and differentiation of various unrelated cell types. It seems likely that XDCoH interaction is not only essential for the function of LFB1 and LFB3 but also for certain other transcription factors.
INTRODUCTION
Eukaryotic organisms have developed mechanisms to guarantee that gene expression is tightly regulated, thus allowing specific genes to be expressed in different cell types at defined developmental stages. This selective expression of genes is controlled by tissue-specific transcription factors, which act by binding to cis-acting elements present in the promoters of the differentially expressed genes.
Recent data have shown that most of the so-called tissuespecific transcription factors are present in more than one cell type suggesting that the presence or absence of a transcription factor alone is not sufficient for the differentiation of a cell. Thus it is generally assumed that the coordinate activity of different factors determines the differentiation state of a cell. For example, Tronche and Yaniv (1992) argue, in a recent review, that the hepatic phenotype results from the combinatorial expression of several regulatory transcription factors, including members of the LFB1 (HNF1), HNF3 and the C/EBP family as well as HNF4 whereas none of these factors is exclusively expressed in the liver.
A further method by which a limited number of tissuespecific transcription factors establishes the biological complexity of an organism is the ability of transcription factors to form homoand heterodimers with each other, leading to altered DNA-binding specifity or altered transcriptional activity (Johnson and McKnight, 1990). This phenomenon might include the existence of cofactor proteins regulating the dimerization process.
Such a cofactor (DCoH), which stabilizes the dimerization and thereby modulates the transcriptional activity of a transcription factor, has recently been identified for the homeoproteins LFB1 (HNF1) and LFB3 (vHNF1) (Mendel et al., 1991a). These two closely related proteins are highly homologous in regions responsible for DNA-binding including the Nterminal dimerization domain, the POU-like domain and the region containing an extended homeobox (Tronche and Yaniv, 1992). They exclusively bind to their DNA recognition sequence, the HP1 element as homodimers and heterodimers and reveal a similar transactivation potential in transient transfection experiments (De Simone et al., 1991; Mendel et al., 1991b; Bach et al., 1990).
LFB1, a transcription factor originally described to be liver specific, is expessed also in kidney, stomach, intestine and colon (Blumenfeld et al., 1991; Bartowski et al., 1993). However, most of the target genes of LFB1 so far known are liver-specifically expressed genes (Courtois et al., 1987; Hardon et al., 1988; Kugler et al., 1988).
The early expression of LFB1 in Xenopus embryogenesis suggests that LFB1 is involved in determination and differentiation of specific cell types. In fact LFB1 belongs to the first zygotic genes transcribed in early amphibian embryogenesis, as transcripts are detectable at mid-blastula transition, a time point clearly prior to organogenesis (Bartkowski et al., 1993). In mammals, there is evidence for a crucial role of LFB1 in differentiation of hepatocytes, as the presence of LFB1 is positively correlated with the differentiation state of hepatoma cells (Cereghini et al., 1990). Furthermore, the DNA recognition sequence of LFB1, the HP1 element in the albumin promoter, is the exclusive regulatory element conserved between mammals and amphibians (Schorpp et al., 1988). This element, which is sufficient to direct liver-specific transcription in vitro and in transfection experiments, is present in the promoters of a number of liver-specifically expressed genes (Lichtsteiner and Schibler, 1989; Monaci et al., 1988; Ryffel et al., 1989).
Binding assays with the mammalian cofactor DCoH revealed that a heterotetramer consisting of two molecules of LFB1 and of DCoH binds to the HP1 element (Mendel et al., 1991b). This tetramer binds to the DNA element in a more stable manner than the LFB1 dimer. In transfection experiments, the LFB1-dependent transactivation of a reporter gene is strongly enhanced by coexpression of DCoH, whereas the cofactor does not stimulate transcription on its own (Mendel et al., 1991a). It has been proposed that the differential expression of the cofactor in specific cell lines might explain the varying transactivation potential of LFB1 depending on the recipient cell used (Nicosia et al., 1992).
To get more insight into the role of DCoH, we analyzed its expression in the development of Xenopus as, in this amphibian system, the LFB1 expression has been studied in detail (Bartkowski et al., 1993; Zapp et al., 1993). We cloned the Xenopus homologue of DCoH that enabled us to compare the structure and function of the Xenopus and rat proteins. To investigate whether XDCoH might contribute to the LFB1 function during Xenopus embryogenesis, we have analyzed the temporal and spatial expression pattern of XDCoH and we have compared the data with the expression pattern of LFB1.
Our results suggest that, during embryogenesis, XDCoH is not only involved in the LFB1 function, but it may also interact with other transcription factors.
MATERIALS AND METHODS
Isolation and sequencing of the XDCoH cDNA
A randomly primed cDNA library of Xenopus liver was used for hybridization with a 950-bp EcoRI fragment encoding the rat cDNA (generous gift of G. R. Crabtree) containing the entire open reading frame of the rat protein (construction of the cDNA library and the hybridization procedure is described by Bartkowski et al., 1993; except the hybridization and washing temperature of 55°C). The inserts from positive clones were subcloned into the pBluescript II SK+ vector (Stratagene) and sequenced with an ALF sequencer (Pharmacia). Computer analysis of the sequence data was carried out with HIBIO DNASIS. The sequence has been deposited in the EMBL data bank (Accession no. Z37525).
Expression vector, transient transfection and CAT assays
An EcoRI site of the linker (position −42 relative to the start codon) and a DraI site of the cDNA (position +105 relative to the stop codon) were used to insert the XDCoH cDNA into the filled in HindIII site of the expression vector Rc/CMV (Invitrogen).
The neuronal cell line Neuro2A, grown in Dulbecco minimal essential medium (DMEM) with 100 U of penicillin and streptomycin per ml and 10% fetal calf serum was transfected with 5 μg of reporter plasmid (HP1)3-TATA-CAT (containing three copies of LFB1binding site HP1 of the Xenopus 68×103Mr albumin gene in front of the TATA box) and with 10ng-1μg of the expression vectors encoding XLFB1 (Bartkowski et al., 1993), XDCoH or rat DCoH (Mendel et al., 1991a) by the calcium phosphate precipitation method (Schorpp et al., 1988). 50 μg of protein was used for the chloramphenicol acetyltransferase assay (Klein-Hitpass et al., 1986).
Production of XDCoH-specific antiserum
We inserted a 500 bp fragment of the XDCoH cDNA containing the entire open reading frame using an EcoRI site of the linker (position −42 relative to the start codon) and a DraI site of the cDNA (position +105 relative to the stop codon) into an EcoRI/EcoRV restricted pBluescript II SK+. This fragment was isolated using a BamHI and a KpnI site of the polylinker and ligated into the prokaryotic expression vector pQE31 (Diagen) creating a fusion protein with 6 histidine residues at the N terminus. The fusion protein was purified using a nickel/NTA column as recommended by the manufacture (Diagen) and used by Eurogentec to immunize rabbits following their standard protocol. Mice (Balb/c) were immunized with 100 μg protein each and boostered three times with 150 μg to raise polyclonal antisera.
Band shift assays
Extracts from whole embryos and Xenopus liver were made as described by Taylor et al. (1991) with benzamidine (0.5 mM), pepstatin A (25 μg/ml), and leupeptin (25 μg/ml) as protease inhibitors. The band shift (Kugler et al., 1990) was performed in the presence of 0.1 μl preimmune or of XDCoH-specific polyclonal mouse antiserum. To detect LFB1, the monoclonal antibody XAD5 was used (Bartkowski et al., 1993).
Western blot analysis
Aliquots of 20% (w/v) extracts were separated on 15% SDS gels and electrotransferred to a nitrocellulose membrane (Sambrook et al., 1989). The blots were incubated overnight with XDCoH-specific rabbit antiserum in a 2×10−3 dilution. Horse radish peroxidase-conjugated mouse antibodies against rabbit immunglobulin G were used to visualize bound XDCoH with the ECL-system (Amersham).
The XDCoH-specific antiserum was affinity purified using a column with recombinant XDCoH protein covalently coupled to seryl-agarose (4 mg/ml) and eluted with 100 mM glycine pH 2.5. The purified antibodies were used for western blots with extracts of embryos and for whole-mount immunostaining experiments following dialysis against phosphate buffered saline (PBS).
Tissue fixation and immunohistochemistry
Liver tissue of adult frogs was fixed using the periodate-lysineparaformaldehyde method (McLean and Nakane, 1974). The tissue was rinsed sequentially in PBS containing 5% and 15% sucrose over night and for 4 hours, respectivly. Cryosections (8 μm) were mounted on BSA-coated glass slides and rinsed once with PBS. The cryosections were blocked for 1 hour with PBS/10% goat serum at 22°C and incubated overnight with the XDCoH-specific rabbit antiserum (500fold diluted in PBS), rinsed again with PBS and incubated for 2 hours with fluorescein isothiocyanate-conjugated anti-rabbit IgG antibody from mouse at 22°C (Jackson Immunology, 750-fold diluted in PBS).
To prepare a control serum for antigen specifity, the XDCoHspecific antiserum was depleted from XDCoH-specific antibodies using recombinant XDCoH protein bound to a Ni-NTA column (5 mg/ml).
Whole-mount immunostaining
The immunostaining was performed as described by HemmatiBrivanlou and Harland (1989), except for using an alkaline phosphatase coupled anti-rabbit IgG mouse antibody (Boehringer, 300fold diluted) to visualize bound XDCoH-specific rabbit antibodies (200-fold diluted affinity-purified polyclonal antiserum). The alkaline phosphatase reaction was carried out as recommended by Boehringer.
Whole-mount in situ hybridization
We used the method described by Harland (1991), except for performing the alkaline phosphatase reaction overnight at 4°C without shaking, which increases signal-to-backround ratio (Schürer and Grunz, personal communication).
RESULTS
Identification of a cDNA encoding the Xenopus dimerization cofactor of LFB1 (XDCoH)
Using the rat DCoH cDNA (Mendel et al., 1991a) as a probe, we isolated a 1.3 kb cDNA clone from a randomly primed liver Xenopus cDNA library. Sequencing of this cDNA clone revealed an open reading frame coding for a 104 amino acid protein with a predicted relative molecular mass of 12×103. The length of the protein is identical to the value reported for the mammalian DCoH and the amino acid sequences are 85% identical. We therefore conclude that we cloned XDCoH, the Xenopus homologue of the mammalian dimerization cofactor of LFB1. Fig. 1 illustrates that the differences in amino acids between the Xenopus and the rat protein are scattered throughout the entire molecule and that 5 of the 14 different amino acids are conservative exchanges (underlined). The high degree of homology between all parts of the coding region suggests that the complete protein is important for maintaining its function.
Comparison between the amino acid sequences of XDCoH and rat DCoH. Amino acids that differ in the rat sequence from the Xenopus sequence are indicated and conservative changes are underlined.
At the nucleotide sequence level, there is 74% identity within the coding region between the Xenopus and the rat cDNA. In the 3′ untranslated region of the cDNAs, no significant homology was found except for one A/T rich stretch of 12 nucleotides (5′ATAATTTGAATA3′, positions 190–201 and 205–216 relative to the stop codon in the XDCoH and in the rat DCoH sequence, respectively) which is identical in both species.
XDCoH enhances the LFB1-dependent transactivation potential in cotransfection experiments
To test whether XDCoH acts as a cofactor with LFB1, we transfected the expression vectors encoding DCoH or Xenopus LFB1 either alone or together into the neuronal cell line Neuro2A lacking LFB1 and DCoH and measured the activity of the cotransfected reporter construct (HP1)3TATA-CAT, which contains three LFB1-binding sites. Fig. 2 shows low transactivation activity when LFB1 is transfected, whereas coexpression of DCoH either from Xenopus or from rat enhances this activity about fourfold. Transfection of the XDCoH expression vector alone does not activate the reporter compared to the activity obtained with an expression vector without any cDNA. An activation of the LFB1-dependent reporter gene by the Xenopus or the rat cofactor is also observed upon coexpression with rat LFB1 (data not shown).
Stimulation of LFB1 transactivation by XDCoH and rat DCoH. CAT assays were performed with extracts from Neuro2A cells transfected with expression vectors encoding XLFB1, XDCoH and rat DCoH as indicated. (HP1)3-TATA-CAT was used as the reporter construct. The amount of transfected DNA was maintained constant by the addition of Rc/CMV, the expression vector without a cDNA insert. The CAT activity of each extract is indicated as the percentage of acetylated chloramphenicol. Data of two independent experiments are shown.
Stimulation of LFB1 transactivation by XDCoH and rat DCoH. CAT assays were performed with extracts from Neuro2A cells transfected with expression vectors encoding XLFB1, XDCoH and rat DCoH as indicated. (HP1)3-TATA-CAT was used as the reporter construct. The amount of transfected DNA was maintained constant by the addition of Rc/CMV, the expression vector without a cDNA insert. The CAT activity of each extract is indicated as the percentage of acetylated chloramphenicol. Data of two independent experiments are shown.
These data establish that XDCoH acts as a positive cofactor of LFB1. The fact that XDCoH and rat DCoH are interchangeable in this assay shows that the cofactor protein is functionally conserved.
XDCoH and LFB1 are both part of the protein/DNA complex
To investigate DCoH involvement in DNA binding, we tested in a band shift experiment whether the binding of LFB1 to its recognition sequence, the HP1 element, involves simultanous binding of XDCoH. Fig. 3 shows that the LFB1-specific antibody XAD5 upshifts the protein/HP1 complex generated by a Xenopus liver extract, confirming that LFB1 interacts with the HP1 oligonucleotide (lane 4 and 5). Similarly, addition of a XDCoH-specific antiserum results in an upshift of the identical protein/DNA complex, thus proving the simultanous presence of XDCoH and LFB1 protein in the same complex (lane 6). Identical results are obtained when Xenopus extracts from stage 41 embryos (3-day-old larvae) are tested, indicating that the LFB1/DNA complex in embryos also contains XDCoH (lane 1–3).
Simultanous binding of LFB1 and DCoH to HP1 in extracts of Xenopus embryos and of adult liver. Band shift assays were performed using an oligonucleotide containing the LFB1 recognition sequence HP1. Binding assays were done in the presence of the monoclonal antibody XAD5 specific for LFB1 or the polyclonal XDCoH-specific antiserum as indicated.
Simultanous binding of LFB1 and DCoH to HP1 in extracts of Xenopus embryos and of adult liver. Band shift assays were performed using an oligonucleotide containing the LFB1 recognition sequence HP1. Binding assays were done in the presence of the monoclonal antibody XAD5 specific for LFB1 or the polyclonal XDCoH-specific antiserum as indicated.
Using in vitro translated rat LFB3 (vHNF1) in the band shift assay, we could show that LFB3 protein and XDCoH also form a complex with HP1 (data not shown).
XDCoH is expressed tissue specifically
To analyse the tissue distribution of XDCoH, we performed Western blots with extracts derived from various tissues of adult frogs using a XDCoH-specific polyclonal antiserum raised against the recombinant protein. The protein detected in kidney and liver (Fig. 4) has the expected size of 12×103Mr and, after a longer exposure of the blot, there is a faint signal detectable in lung and stomach (data not shown), whereas all the other tissues tested lack XDCoH. As we know from our previous work that the LFB1 expression is high in liver and kidney (Bartkowski et al., 1993), we conclude that XDCoH and LFB1 are colocalized in these tissues. In contrast, intestine expressing LFB1 abundantly (Bartkowski et al., 1993) does not contain XDCoH.
Tissue-specific expression of XDCoH. Aliquots (20 μl) from cleared extracts (20% w/v) of lung (Lu), heart (He), spleen (Sp), kidney (Ki), liver (Li), stomach (St), small and thick intestine (In), muscle (Mu), ovary (Ov), blood cells (Bl), testis (Te) and brain (Br) were separated on a 15% SDS gel and the blot was incubated with a XDCoH-specific antiserum as the primary antibody.
Tissue-specific expression of XDCoH. Aliquots (20 μl) from cleared extracts (20% w/v) of lung (Lu), heart (He), spleen (Sp), kidney (Ki), liver (Li), stomach (St), small and thick intestine (In), muscle (Mu), ovary (Ov), blood cells (Bl), testis (Te) and brain (Br) were separated on a 15% SDS gel and the blot was incubated with a XDCoH-specific antiserum as the primary antibody.
To investigate the cellular localization of XDCoH, we immunostained sections of Xenopus liver with the XDCoHspecific antiserum. Comparing the immunoflourescence (Fig. 5A) and the phase contrast (Fig. 5B) of the same section, it is clear that hepatocytes contain XDCoH whereas the blood cells are completely negative for XDCoH-specific staining. Furthermore the biliary duct (b) is also positive for XDCoH (Fig. 5A,B). This interestingly correlates with the fact that cells of the biliary duct express LFB1 (Weber et al., 1995).
Immuncytochemistry to localize XDCoH in the adult Xenopus liver. Cryostat sections of adult liver were incubated with XDCoH-specific polyclonal antiserum as the first antibody. (A,C) Immunofluorescence pictures showing XDCoH positive hepatocytes. (B) Corresponding phase contrast to A demonstrating the localization of blood cells (bc). Note also the immunostaining of the biliary duct (b). The bar corresponds to 50 μm. Using XDCoH-specific antiserum, that was depleted by binding to recombinant XDCoH protein, gave no staining (data not shown).
Immuncytochemistry to localize XDCoH in the adult Xenopus liver. Cryostat sections of adult liver were incubated with XDCoH-specific polyclonal antiserum as the first antibody. (A,C) Immunofluorescence pictures showing XDCoH positive hepatocytes. (B) Corresponding phase contrast to A demonstrating the localization of blood cells (bc). Note also the immunostaining of the biliary duct (b). The bar corresponds to 50 μm. Using XDCoH-specific antiserum, that was depleted by binding to recombinant XDCoH protein, gave no staining (data not shown).
The higher magnification given in Fig. 5C reveals the localization of XDCoH in the nuclei and unexpectedly in the cytoplasm of the hepatocytes. The localization in the nucleus supports the idea that XDCoH is in hepatocytes involved in the regulation of transcription. In contrast, the cytoplasmatic localization of XDCoH can be interpreted by the recent finding that rat DCoH is a bifunctional protein that possesses in addition to its cofactor activity the activity of the 4a-carbinolamine dehydratase, an enzyme involved in the cytoplasmatic phenylalanine metabolism of hepatocytes (Citron et al., 1992). In an appropriate enzymatic assay with recombinant XDCoH, we proved that XDCoH contains 4a-carbinolamine dehydratase as well (unpublished data in collaboration with G. Johnen and S. Kaufman).
Embryonic expression of XDCoH
To evaluate the significance of XDCoH in embryogenesis, we investigated the occurrence of the protein during embryogenesis by western blot analysis. The antibodies detect at all stages a protein of 12×103Mr, identical in size to the protein found in liver and kidney (Fig. 6). We conclude therefore that XDCoH is expressed maternally at significant levels and gradually decreases until stage 21 (late neurula). Following neurulation, an about tenfold increase of XDCoH is seen, whereas from stage 29 onwards the protein level does not change significantly until stage 45. The time course of XDCoH protein expression differs from the corresponding time course of LFB1, since LFB1 protein is detectable in Western blots from stage 35 onwards and its amount increases about tenfold until stage 45 (Bartkowski et al., 1993).
Expression pattern of XDCoH during Xenopus development. Embryo extracts of different developmental stages (200 μg or 20 μg as indicated) were analyzed by Western blotting using an affinity purified polyclonal XDCoH-specific antiserum as the first antibody. The developmental stages examined were egg (e), 2-cell stage (2) and various stages up to the three days old larvae that were determined as described by Nieuwkoop and Faber (1975). Stage 24 is a very early tail bud stage. The lanes to the right show the Western blot analysis from extracts (50 μg each) of stage 41 larvae dissected into head (h), middle (m), and tail (t).
Expression pattern of XDCoH during Xenopus development. Embryo extracts of different developmental stages (200 μg or 20 μg as indicated) were analyzed by Western blotting using an affinity purified polyclonal XDCoH-specific antiserum as the first antibody. The developmental stages examined were egg (e), 2-cell stage (2) and various stages up to the three days old larvae that were determined as described by Nieuwkoop and Faber (1975). Stage 24 is a very early tail bud stage. The lanes to the right show the Western blot analysis from extracts (50 μg each) of stage 41 larvae dissected into head (h), middle (m), and tail (t).
The analysis from extracts of dissected larvae (swimming larvae, stage 41) revealed that XDCoH is detectable not only in the middle part of the embryo, but in similar amounts also in the head and in low amounts in the tail (Fig. 6, right part). This contrasts to our earlier studies (Bartkowski et al., 1993) demonstrating that LFB1 expression in Xenopus is restricted to the middle part of the dissected embryo, where tissues containing LFB1, the pronephros, the liver and the gut, are developing. The results obtained by western blot demonstrate that XDCoH expression is clearly not restricted to parts of the embryo that contain the organs known to express LFB1 and LFB3 in the adult.
Spatial expression pattern of XDCoH
To localize the XDCoH protein more precisely in the developing larvae, we analyzed the distribution of XDCoH by whole-mount immunostainings of albino embryos of different developmental stages. Fig. 7A shows staining of a 4-cellstage embryo (lower embryo, animal view), demonstrating that the protein is detectable in each of the blastomeres, where it is uniformily distributed in the cytoplasm. Preimmune serum (not shown) or the XDCoHspecific antiserum immunoabsorbed against recombinant XDCoH protein failed to stain the embryos (upper embryo of Fig. 7A) confirming the antigen specifity of the antiserum used. The predominantly cytoplasmatic localization of XDCoH is maintained in the 48-cell stage (Fig. 7B,C). The lateral view of the embryo (Fig. 7B) reveals that there is a gradient of the XDCoH-specific staining decreasing from the smaller animal blastomeres to the larger cells of the vegetal hemisphere. By staining of a stage 8 blastula (Fig. 7D, animal view), it is clearly visible that the signal becomes nuclear restricted in some cells, whereas at the stage 9 blastula (Fig. 7E, lateral view, animal pole to the left) only the nuclei of the bastomeres, animal as well as vegetal, are stained. Fig. 7F shows a stage 11 gastrula that passes through a partial exogastrulation. This embryo demonstrates very well that the nuclei of endodermal cells are positive for XDCoH and, in a higher magnification, the nuclear signal is also visible in the animal cells (data not shown). These analyses prove that maternal XDCoH is a cytoplasmatic protein that becames nuclear at mid blastula transition (stage 8) when the zygotic transcription of the Xenopus embryo starts.
Localization of XDCoH in the Xenopus embryo. Whole-mount staining of different development stages was performed using an affinity-purified polyclonal XDCoH-specific antiserum as the first antibody and an enzyme linked secondary antibody. (A) The upper 4-cell stage embryo is a control for antibody specifity and was incubated with the XDCoH-specific antiserum that was depleted using recombinant XDCoH protein. (B,C) A lateral and animal view of a 48-cell stage embryo (stage 6.5), respectively. (D,E) Blastulae at stage 8 and 9, respectively, and (F) a gastrula (stage 11). (G,H) XDCoHspecific stained organs are the pronephros (p), eye vesicle (e), gut and liver (g, l) and the pigmented epithelium (pe).
Localization of XDCoH in the Xenopus embryo. Whole-mount staining of different development stages was performed using an affinity-purified polyclonal XDCoH-specific antiserum as the first antibody and an enzyme linked secondary antibody. (A) The upper 4-cell stage embryo is a control for antibody specifity and was incubated with the XDCoH-specific antiserum that was depleted using recombinant XDCoH protein. (B,C) A lateral and animal view of a 48-cell stage embryo (stage 6.5), respectively. (D,E) Blastulae at stage 8 and 9, respectively, and (F) a gastrula (stage 11). (G,H) XDCoHspecific stained organs are the pronephros (p), eye vesicle (e), gut and liver (g, l) and the pigmented epithelium (pe).
Analysis of a 25/26 stage embryo (Fig. 7G) reveals that the XDCoH expression now starts to get restricted to the pronephros anlage (p), the primary eye vesicles (e) and the cement gland (cg). In the swimming larvae (stage 37/38), predominantly four organs are specifically stained (Fig. 7H): the pronephros (p), the gut and the liver (g,l) and the pigmented epithelium surrounding the eye (pe). Staining of the gut cells is clearly visible in the the anterior part of the gut (g) as well as in the hind gut (hg).
Using a laser scanner analysis of whole mounts stained with a flourescence secondary antibody, single XDCoH-positive cells of the pronephros (Fig. 8A), the liver (Fig. 8B), the gut (Fig. 8C) as well as of the pigmented epithelium of the eye (Fig. 8D) are detectable. In all cells, XDCoH is predominantly located in the nuclei, suggesting its involvement in transcriptional regulation.
XDCoH is nuclear localized in tissues of a stage 37/38 embryo. To perform a laser scanner analysis the XDCoH-specific whole-mount staining was performed using a flourescent secondary antibody. (A-D) The celluar localization of XDCoH in the tissues identified to express XDCoH in the larvae (compare to Fig. 7H). XDCoH-specific staining is detected in the nuclei of pronephric cells (p), the hepatocytes (h), the gut cells (g) and in the pigmented epithelium of the eye (pe). Staining of a larvae without primary antibodies gave no signal.
XDCoH is nuclear localized in tissues of a stage 37/38 embryo. To perform a laser scanner analysis the XDCoH-specific whole-mount staining was performed using a flourescent secondary antibody. (A-D) The celluar localization of XDCoH in the tissues identified to express XDCoH in the larvae (compare to Fig. 7H). XDCoH-specific staining is detected in the nuclei of pronephric cells (p), the hepatocytes (h), the gut cells (g) and in the pigmented epithelium of the eye (pe). Staining of a larvae without primary antibodies gave no signal.
To verify the specifity of the signals obtained by the antibodies, we analyzed the distribution of the XDCoH mRNA in a stage 37/38 larvae by whole-mount in situ hybridization using an antisense XDCoH RNA (Fig. 9). The result correlates perfectly with the protein data: the mRNA is detectable in the pronephros (p), the gut and the liver (g,l) and the pigmented epithelium (Fig. 9, upper embryo). As demonstrated in the same figure (lower embryo), there is no signal when a sense XDCoH RNA is used as a control.
Whole-mount in situ hybridization of a stage 37/38 embryo using an antisense XDCoH RNA (upper embryo) or a sense XDCoH RNA (lower embryo) as the probe. The pronephros (p), the liver (l) and the gut (g) as well as the pigmented epithelium (pe) of the eye are specifically stained, whereas the hybridization with the sense probe reveals no signal.
Whole-mount in situ hybridization of a stage 37/38 embryo using an antisense XDCoH RNA (upper embryo) or a sense XDCoH RNA (lower embryo) as the probe. The pronephros (p), the liver (l) and the gut (g) as well as the pigmented epithelium (pe) of the eye are specifically stained, whereas the hybridization with the sense probe reveals no signal.
These data prove that XDCoH is expressed in tissues derived from each of the three germ layers in a restricted group of cells indicating a role in the differentiation of various cell types.
DISCUSSION
Evolutionary conservation of DCoH, the dimerization cofactor of the LFB1 family
Using a rat DCoH cDNA probe, we isolated a cDNA clone encoding DCoH from Xenopus. The deduced proteins of the mammalian and the amphibian cDNA clones are identical in size (104 amino acids) and have a high degree of 85% identity at the amino acid level. Since we have shown in transfection experiments that each of the proteins can directly interact with LFB1 from Xenopus and enhance LFB1-mediated transactivation (Fig. 2), we conclude that the rat and the Xenopus DCoH are not only related in the primary sequence but they are also conserved in the structure essential for the function of the protein. The similarity of the proteins in structure and function suggests a common role in the regulation of transcription in such evolutionary divergent organisms as frog and rat. The localization of considerable amounts of XDCoH in the cytoplasm of hepatocytes (Fig. 5) as well as the detection of the 4a-carbinolamine dehydratase activity of the recombinant Xenopus protein (G. Johnen and S. Kaufman, personal communication) establish that the bifunctionality of DCoH occurs in amphibians as well as in mammals.
In our experiments, it turned out that XDCoH can neither bind to DNA on its own (data not shown) nor stimulate HP1dependent transcription in the absence of LFB1 (Fig. 2). Thus, we conclude that the role of XDCoH is to enhance the LFB1 transcriptional potential as described for the mammalian cofactor (Mendel et al., 1991a). This conservation of the function adds to the fact that the DNA-recognition sequence, the HP1 element (Schorpp et al., 1988), and the corresponding transcription factor LFB1 (Bartkowski et al., 1993) are well conserved during evolution. The similarity of the mammalian and the amphibian protein is further demonstrated by the fact that the polyclonal antiserum raised against recombinant XDCoH detects in western blots the homologuous factor from human and rat (data not shown).
As deletion experiments of the rat LFB1 protein have shown that the 213 N-terminal amino acids of LFB1 are sufficient for DCoH binding (Mendel et al., 1991a), it is interesting that this region of LFB1 contains, in addition to the domains involved in dimerization (amino acid 1–33) and the POU-related DNAbinding domain (amino acid 100-184), a third region (amino acid 58–97) highly homologuous between Xenopus and mammals (Bartkowski et al., 1993), which might be involved in DCoH binding. Within this region amino acids 80 to 97 are also conserved in LFB3, the other member of the LFB1 family that interacts with XDCoH (data not shown).
Differential expression of XDCoH may contribute to the complexity of LFB1 function
Analyzing tissues of adult Xenopus with specific antibodies in western blots, we localized XDCoH predominantly in the liver and kidney (Fig. 4), two tissues also containing high levels of LFB1 (Bartkowski et al., 1993). We hardly detect any XDCoH in the stomach (overexposure of Fig. 4) and no XDCoH in intestine, two tissues that we have previously shown to contain LFB1 levels comparable to the liver. This means that in the adults there are LFB1-containing tissues expressing XDCoH (liver, kidney) and others (intestine, stomach) that hardly contain any XDCoH. A similar situation may exist in mammals where high levels of 4a-carbinolamine dehydratase, the enzyme that corresponds to DCoH (Citron et al., 1992), was found in liver and kidney (Davis et al., 1992) while the other LFB1-containing tissues were not tested.
Based on the differential expression of LFB1 and XDCoH, it is an attractive hypothesis that the presence or absence of XDCoH may modulate the activity of LFB1 in a tissue-specific way. Thus XDCoH may for instance contribute to the differential regulatory function of LFB1 in the liver and in the stomach, two tissues that both contain LFB1, but express distinct sets of LFB1-dependent genes.
Developmental expression of XDCoH
In contrast to the situation in adults, in the developing larvae (stage 37/38) all tissues expressing LFB1 also contain XDCoH. Both proteins are simultanously present in the liver, the gut and the pronephros. In contrast, the pigmented epithelium of the embryonic eye contains XDCoH at very high levels (Figs 7H, 8D), whereas no LFB1 protein is present in this anterior region of the embryo (Bartkowski et al., 1993). As band shift analysis using extracts of the head gives no indication for the presence of the Xenopus homolgue of the mammalian LFB3 in this region (Demartis et al., 1994; Weber et al., 1995), we exclude the presence of embryonic XLFB3 in the eye. We thus assume that factors distinct from the LFB1 family cooperate with nuclear XDCoH in this cell type.
Further evidence that XDCoH might have a role independently from LFB1 is provided by our observation that XDCoH is a maternal protein in the egg (Figs 6, 7). The protein is present in the cytoplasm of the very early cleavage stages thus establishing that during further cleavages each of the daughter cells contain XDCoH protein. When the zygotic transcription of the embryo starts at mid-blastula transition, XDCoH moves into the nuclei of the cells, suggesting a function in early transcriptional events (Fig. 7D,E). As at this stage LFB1 is not detectable (Bartkowski et al., 1993), we assume that another maternal transcription factor that cooperates with XDCoH might exist in the egg. It seems unlikely that this factor is LFB3, a protein that is able to interact with XDCoH (seen in band shift experiments, data not shown; Mendel et al., 1991a), as recent data have shown that the LFB3 mRNA in Xenopus is firstly detectable after mid-blastula transition (Demartis et al., 1994).
Although a whole series of maternal transcription factors has been identified in the Xenopus egg, including several protooncogenes (e. g. c-fos (Kindy and Verma, 1990); c-myc (Principaud and Spohr, 1991), c-ets-1 (Stiegler et al., 1991), XrelA1 (Richardson et al., 1994; Bearer, 1994); and B-myb, Bouwmeester et al., 1994), it is difficult to decide which are the most likely candidates for cooperation with DCoH. Assuming that XDCoH can interact only with homeoprotein factors structurally related to LFB1 and LFB3, the Oct-60 (XLPOU-60) transcription factors as members of the POUfamily might be the potential partner (Hinkley et al., 1992; Whitfield et al., 1993).
Quite unexpectally the time course of XDCoH accumulation precedes the increase in LFB1 and in fact reaches a plateau at a very early tailbud stage (stage 24, Fig. 6), whereas the LFB1 protein reaches its maximum at stage 45. It will be important to analyze the accumulation of LFB3 protein in Xenopus and to compare it with the XDCoH protein expression. Recent analysis of the appearance of LFB3 mRNA during Xenopus development (Demartis et al., 1994) revealed that LFB3 mRNA reaches a plateau in the neurula, when the level of LFB1 mRNA is still increasing (Bartkowski et al., 1993).
Although DCoH is a bifunctional protein (Citron et al., 1992), we believe that XDCoH functions as a transcriptional cofactor in the developing embryo, as after mid-blastula transition this protein is predominantly localized in the nuclei (Figs 7, 8). In contrast in the hepatocytes of the adult liver, substantial amounts of XDCoH protein are not only nuclear but also cytoplasmatic (Fig. 5C) implying that XDCoH functions in the liver also as the 4a-carbinolamine dehydratase.
In conclusion our data imply that XDCoH contributes to the activation potential of the members of the LFB1 family in a cell-type-specific manner and that in several cell types XDCoH may even cooperate with transcription factors distinct from the LFB1 family.
ACKNOWLEDGEMENTS
We thank G. R. Crabtree and M. Yaniv for the rat DCoH and LFB3 cDNA clones, respectively, and C. Schürer and H. Grunz for teaching the in situ hybridization technique.We gratefully acknowledge the laser scanner analysis carried out by H. Althoff from Zeiss, Germany and are grateful to L. Klein-Hitpaß, H. Weber and G. Johnen for critical reading of the manuscript. This work was supported by the Deutsche Forschungsgemeinschaft (Ry 5/1-3).