The homeodomain of the transcription factor LF-B1 has a 21 amino acid loop between helix 2 and helix 3

Homeodomain proteins: an update

Here, we provide an update of our review on homeobox genes that we wrote together with Walter Gehring in 1994. Since then, comprehensive surveys of homeobox genes have become possible due to genome sequencing projects. Using the 103 Drosophila homeobox genes as example, we present an updated classification. In animals, there are 16 major classes, ANTP, PRD, PRD-LIKE, POU, HNF, CUT (with four subclasses: ONECUT, CUX, SATB, and CMP), LIM, ZF, CERS, PROS, SIX/SO, plus the TALE superclass with the classes IRO, MKX, TGIF, PBC, and MEIS. In plants, there are 11 major classes, i.e., HD-ZIP (with four subclasses: I to IV), WOX, NDX, PHD, PLINC, LD, DDT, SAWADEE, PINTOX, and the two TALE classes KNOX and BEL. Most of these classes encode additional domains apart from the homeodomain. Numerous insights have been obtained in the last two decades into how homeodomain proteins bind to DNA and increase their specificity by interacting with other proteins to regulate cell- and tissue-specific gene expression. Not only protein-DNA base pair contacts are important for proper target selection; recent experiments also reveal that the shape of the DNA plays a role in specificity. Using selected examples, we highlight different mechanisms of homeodomain protein-DNA interaction. The PRD class of homeobox genes was of special interest to Walter Gehring in the last two decades. The PRD class comprises six families in Bilateria, and tinkers with four different motifs, i.e., the PAIRED domain, the Groucho-interacting motif EH1 (aka Octapeptide or TN), the homeodomain, and the OAR motif. Homologs of the co-repressor protein Groucho are also present in plants (TOPLESS), where they have been shown to interact with small amphipathic motives (EAR), and in yeast (TUP1), where we find an EH1-like motif in MATα2.

Crystal structure of a paired domain-DNA complex at 2.5 A resolution reveals structural basis for Pax developmental mutations

The 2.5 A resolution structure of a cocrystal containing the paired domain from the Drosophila paired (prd) protein and a 15 bp site shows structurally independent N-terminal and C-terminal subdomains. Each of these domains contains a helical region resembling the homeodomain and the Hin recombinase. The N-terminal domain makes extensive DNA contacts, using a novel beta turn motif that binds in the minor groove and a helix-turn-helix unit with a docking arrangement surprisingly similar to that of the lambda repressor. The C-terminal domain is not essential for prd binding and does not contact the optimized site. All known developmental missense mutations in the paired box of mammalian Pax genes map to the N-terminal subdomain, and most of them are found at the protein-DNA interface.

Two independent and interactive DNA-binding subdomains of the Pax6 paired domain are regulated by alternative splicing

Vertebrate Pax proteins share a conserved 128-amino-acid DNA-binding motif, the paired domain. The PAX6 gene, which is mutated in the murine Small eye and human aniridia developmental defects, also encodes a second protein with a 14-amino-acid insertion in the paired domain. This protein, which arises by alternative mRNA splicing, exhibits unique DNA-binding properties. Unlike other paired domains, which bind DNA predominantly by their amino termini, the extended Pax6 paired domain interacts with DNA exclusively through its carboxyl terminus. This property can be stimulated by deletion of 30 amino-terminal residues from the Pax6 or Pax2 paired domains. Thus, the insertion acts as a molecular toggle to unmask the DNA-binding potential of the carboxyl terminus. The functional nonequivalence of the two Pax6 proteins is underscored by a T-->C mutation at position -3 of the alternative splice acceptor site that changes the ratio of the two isoforms and causes a distinct human ocular syndrome.

More potent transcriptional activators or a transdominant inhibitor of the HNF1 homeoprotein family are generated by alternative RNA processing

We report the isolation of cDNAs from human liver encoding several isoforms of the hepatocyte nuclear factor homeoproteins HNF1 and vHNF1 generated by the differential use of polyadenylation sites and by alternative splicing. In the novel isoforms intron sequences that are excised in the previously described forms are translated in the same frame as exon sequences until the first termination codon is encountered. Hence, the newly found isoforms all contain different C-terminal domains. For HNF1 it has been shown that its C-terminal region is responsible for the activation of transcription. In transient transfection assays the two novel HNF1 isoforms, HNF1-B and -C, transactivate 5-fold better than the previously described HNF1 protein (HNF1-A). The newly isolated isoform of vHNF1, designated vHNF1-C, is unable to transactivate and behaves as a transdominant repressor when cotransfected with HNF1-A, -B or -C. All of the different isoforms of HNF1 and vHNF1 can form homo- and heterodimers and their mRNAs are differentially expressed in fetal and adult human liver, kidney and intestine, suggesting distinct roles during development. Our studies show that the transactivation domain of the members of the HNF1 homeoprotein family is organized in modules which can be exchanged to generate either more potent transcriptional activators or a transdominant repressor.

The X-ray structure of an atypical homeodomain present in the rat liver transcription factor LFB1/HNF1 and implications for DNA binding

The transcription factor LFB1/HNF1 from rat liver nuclei is a 628 amino acid protein that functions as a dimer binding to the inverted palindrome GTTAATN-ATTAAC consensus site. We have crystallized a 99 residue protein containing the homeodomain portion of LFB1, and solved its structure using X-ray diffraction data to 2.8 A resolution. The topology and orientation of the helices is essentially the same as that found in the engrailed, MAT alpha 2 and Antennapedia homeodomains, even though the LFB1 homeodomain contains 21 more residues. The 21 residue insertion is found in an extension of helix 2 and consequent lengthening of the connecting loop between helix 2 and helix 3. Comparison with the engrailed homeodomain-DNA complex indicates that the mode of interaction with DNA is similar in both proteins, with a number of conserved contacts in the major groove. The extra 21 residues of the LFB1 homeodomain are not involved in DNA binding. Binding of the LFB1 dimer to a B-DNA palindromic consensus sequence requires either a conformational change of the DNA (presumably bending), or a rearrangement of the subunits relative to the DNA.

Developmental regulation and tissue distribution of the liver transcription factor LFB1 (HNF1) in Xenopus laevis

The transcription factor LFB1 (HNF1) was initially identified as a regulator of liver-specific gene expression in mammals. It interacts with the promoter element HP1, which is functionally conserved between mammals and amphibians, suggesting that a homologous factor, XLFB1, also exists in Xenopus laevis. To study the role of LFB1 in early development, we isolated two groups of cDNAs coding for this factor from a Xenopus liver cDNA library by using a rat LFB1 cDNA probe. A comparison of the primary structures of the Xenopus and mammalian proteins shows that the myosin-like dimerization helix, the POU-A-related domain, the homeo-domain-related region, and the serine/threonine-rich activation domain are conserved between X. laevis and mammals, suggesting that all these features typical for LFB1 are essential for function. Using monoclonal antibodies, we demonstrate that XLFB1 is present not only in the liver but also in the stomach, intestine, colon, and kidney. In an analysis of the expression of XLFB1 in the developing Xenopus embryo, XLFB1 transcripts appear at the gastrula stage. The XLFB1 protein can be identified in regions of the embryo in which the liver diverticulum, stomach, gut, and pronephros are localized. The early appearance of XLFB1 expression during embryogenesis suggests that the tissue-specific transcription factor XLFB1 is involved in the determination and/or differentiation of specific cell types during organogenesis.

Developmental regulation and tissue distribution of the liver transcription factor LFB1 (HNF1) in Xenopus laevis

The transcription factor LFB1 (HNF1) was initially identified as a regulator of liver-specific gene expression in mammals. It interacts with the promoter element HP1, which is functionally conserved between mammals and amphibians, suggesting that a homologous factor, XLFB1, also exists in Xenopus laevis. To study the role of LFB1 in early development, we isolated two groups of cDNAs coding for this factor from a Xenopus liver cDNA library by using a rat LFB1 cDNA probe. A comparison of the primary structures of the Xenopus and mammalian proteins shows that the myosin-like dimerization helix, the POU-A-related domain, the homeo-domain-related region, and the serine/threonine-rich activation domain are conserved between X. laevis and mammals, suggesting that all these features typical for LFB1 are essential for function. Using monoclonal antibodies, we demonstrate that XLFB1 is present not only in the liver but also in the stomach, intestine, colon, and kidney. In an analysis of the expression of XLFB1 in the developing Xenopus embryo, XLFB1 transcripts appear at the gastrula stage. The XLFB1 protein can be identified in regions of the embryo in which the liver diverticulum, stomach, gut, and pronephros are localized. The early appearance of XLFB1 expression during embryogenesis suggests that the tissue-specific transcription factor XLFB1 is involved in the determination and/or differentiation of specific cell types during organogenesis.

Trans-dominant inhibition of transcription activator LFB1

Liver-enriched factor LFB1 (also named HNF1) is a dimeric transcription activator which is essential for the expression of many hepatocyte-specific genes. Here we demonstrate that LFB1 mutants in the POU A-like or in the homeo domains inhibit wild-type DNA binding by forming inactive heterodimeric complexes. Co-transfection of one of these mutants with wild-type LFB1 in HeLa cells eliminated LFB1 DNA binding and transcriptional activities through a trans-dominant mechanism. Expression of the same dominant negative mutant in human hepatoma HepG2 cells only partially inhibited endogenous LFB1 activity, due to stabilization of LFB1 dimers in these cells. Dimer stabilization in hepatoma cells is mediated by a heat-labile association with an 11kD polypeptide, analogous to the DCoH cofactor identified in rat liver by Mendel et al. (1). The property of stabilizing LFB1 dimers is also shared by HeLa cells which produce a HeLa homolog of DCoH. These results demonstrate that LFB1 dimer stabilization as well as the synthesis of 'stabilizing factors' are not restricted to cells expressing LFB1 or other members of its family.

Structure of the gene encoding hepatocyte nuclear factor 1 (HNF1)

Genomic clones have been isolated that cover the entire gene for the transcription factor HNF1 (hepatocyte nuclear factor 1). This protein governs the expression of many genes, synthesized in the liver in a tissue-specific manner. We have determined the intron/exon structure of the HNF1 gene, which is strictly conserved between rat and mouse and estimate that it spans not more than 40kb in the rat genome. Whereas most homeoprotein genes do not contain introns within the homeodomain, HNF1 displays an intron between the regions encoding the second and the third helices. We discuss possible evolutionary mechanisms leading to this homeobox intron/exon pattern.

HNF1, a homeoprotein member of the hepatic transcription regulatory network

Numerous liver specific genes are transcriptionally activated by the binding to their promoter or enhancer of Hepatic Nuclear Factor 1 (HNF1). HNF1 contains a variant homeo-domain and binds to DNA as either a homodimer or a heterodimer with the vHNF1 protein. Surprisingly, HNF1 is not restricted to hepatocytes but is expressed in epithelial cells of several endoderm derived organs and in mesoderm derived kidney tubules. Hence, HNF1 alone can not account for the differentiated state of the hepatic cells. In fact, several other liver-enriched transcription factors have been cloned. The hepatic phenotype could result from the combinatorial expression of these regulators. Possible involvement of these trans-acting factors in liver organogenesis and hepatic differentiation is discussed.

Hepatocyte nuclear factor 1 alpha is expressed in a hamster insulinoma line and transactivates the rat insulin I gene

Systematic mutational analysis previously identified two primary regulatory elements within a minienhancer (-247 to -198) of the rat insulin I promoter that are critical for transcriptional activity. The Far box (-241 to -232) and the FLAT element (-222 to -208) synergistically upregulate transcription and, together, are sufficient to confer tissue-specific and glucose-responsive transcriptional activity on a heterologous promoter. Detailed analysis of the FLAT element further revealed that, in addition to the positive regulatory activity it mediates in tandem with the Far box, it is a site for negative regulatory control. A portion of the FLAT element bears considerable sequence similarity to the consensus binding site for hepatocyte nuclear factor 1 alpha (HNF1 alpha; LF-B1) a liver-enriched homeodomain-containing transcription factor. Here we show that the HNF1-like site within the FLAT element exhibited positive transcriptional activity in both HepG2 and HIT cells and bound similar, but distinguishable, nuclear protein complexes in the respective nuclear extracts. Screening of a hamster insulinoma cDNA library with a PCR-derived probe encompassing the DNA-binding domain of rat HNF1 alpha resulted in isolation of a hamster HNF1 alpha (hHNF1 alpha) cDNA homolog. Specific antiserum identified the HNF1 alpha protein as one component of a specific FLAT-binding complex in HIT nuclear extracts. Expression of the hHNF1 alpha cDNA in COS cells resulted in transactivation of reporter constructs containing multimerized segments of the rat insulin I minienhancer. Thus, HNF1 alpha, one component of a DNA-binding complex involved in transcriptional regulation of the rat insulin I gene, may play a significant role in nonhepatic as well as hepatic gene transcription.

A two-component regulatory system for self/non-self recognition in Ustilago maydis

In U. maydis the multiallelic b locus controls sexual and pathogenic development. In the b locus a gene coding for a regulatory protein had been identified, and it was suggested that the interaction of two b polypeptides specified by different alleles programs sexual development in this fungus. We now demonstrate the existence of a second regulatory gene in the b locus. We term this gene bW and refer to the former as the bE gene. Both genes exist in many alleles. Although unrelated in primary sequence, both genes are similar in their overall organization. The gene products display allele-specific variability in their N-terminal domains, show a high degree of sequence conservation in the C-terminal domains, and contain a homeodomain-related motif. Genetic evidence is provided to show that the pair of bE and bW polypeptides encoded by different b alleles is the key regulatory species.

Properties of the DNA-binding domain of the Saccharomyces cerevisiae STE12 protein

The STE12 protein of the yeast Saccharomyces cerevisiae binds to the pheromone response element (PRE) present in the upstream region of genes whose transcription is induced by pheromone. Using DNase I footprinting assays with bacterially made STE12 fragments, we localized the DNA-binding domain to 164 amino acids near the amino terminus. Footprinting of oligonucleotide-derived sequences containing one PRE, or two PREs in head-to-tail or tail-to-tail orientation, showed that the N-terminal 215 amino acids of STE12 has similar binding affinity to either of the dimer sites and a binding affinity 5- to 10-fold lower for the monomer site. This binding cooperativity was also evident on a fragment from the MFA2 gene, which encodes the a-factor pheromone. On this fragment, the 215-amino-acid STE12 fragment protected both a consensus PRE as well as a degenerate PRE containing an additional residue. Mutation of the degenerate site led to a 5- to 10-fold decrease in binding; mutation of the consensus site led to a 25-fold decrease in binding. The ability of PREs to function as pheromone-inducible upstream activation sequences in yeast correlated with their ability to bind the STE12 domain in vitro. The sequence of the STE12 DNA-binding domain contains similarities to the homeodomain, although it is highly diverged from other known examples of this motif. Moreover, the alignment between STE12 and the homeodomain postulates loops after both the putative helix 1 and helix 2 of the STE12 sequence.

Vertebrate homeobox genes

Recent highlights in vertebrate homeobox gene research include the discovery of new genes with novel expression patterns, observations that peptide growth factors and retinoic acid influence homeobox gene expression, and the generation of mutant phenotypes of embryos homozygous for null mutations. These combined studies reinforce the idea that homeobox genes function near the top of the gene hierarchies controlling vertebrate embryogenesis.

Properties of the DNA-binding domain of the Saccharomyces cerevisiae STE12 protein

The STE12 protein of the yeast Saccharomyces cerevisiae binds to the pheromone response element (PRE) present in the upstream region of genes whose transcription is induced by pheromone. Using DNase I footprinting assays with bacterially made STE12 fragments, we localized the DNA-binding domain to 164 amino acids near the amino terminus. Footprinting of oligonucleotide-derived sequences containing one PRE, or two PREs in head-to-tail or tail-to-tail orientation, showed that the N-terminal 215 amino acids of STE12 has similar binding affinity to either of the dimer sites and a binding affinity 5- to 10-fold lower for the monomer site. This binding cooperativity was also evident on a fragment from the MFA2 gene, which encodes the a-factor pheromone. On this fragment, the 215-amino-acid STE12 fragment protected both a consensus PRE as well as a degenerate PRE containing an additional residue. Mutation of the degenerate site led to a 5- to 10-fold decrease in binding; mutation of the consensus site led to a 25-fold decrease in binding. The ability of PREs to function as pheromone-inducible upstream activation sequences in yeast correlated with their ability to bind the STE12 domain in vitro. The sequence of the STE12 DNA-binding domain contains similarities to the homeodomain, although it is highly diverged from other known examples of this motif. Moreover, the alignment between STE12 and the homeodomain postulates loops after both the putative helix 1 and helix 2 of the STE12 sequence.

Two members of an HNF1 homeoprotein family are expressed in human liver

HNF1 is a transcriptional activator, required for the liver-specific expression of a variety of genes, that binds to DNA as a dimer via the most diverged homeodomain known so far. We were interested to examine whether HNF1 is a unique homeoprotein example or whether it is the prototype of a new subfamily of homeodomain containing proteins. In this work we describe the isolation of a cDNA clone from a human liver library encoding a protein, highly homologous to HNF1 in three regions, including the homeo- and dimerization domains. We show that this protein can heterodimerize with human HNF1 in vitro. Sequence comparison of our clone with a rat variant HNF1 (vHNF1) clone, isolated in parallel in our laboratory from the dedifferentiated H5 hepatoma cell line, identified our cDNA as human vHNF1. vHNF1 is a nuclear protein recognizing the same binding site as HNF1 and previously thought to occur only in dedifferentiated hepatoma cells that fail to express most liver specific genes. Nevertheless, we show by Northern blot analysis that vHNF1 transcripts are present in differentiated human HepG2 hepatoma cells as well as in rat liver and that this transcript level is 10-20 fold lower than that of HNF1. We assigned the vHNF-1 gene to human chromosome 17 and murine chromosome 11. These chromosomal localizations differ from that of the HNF-1 gene indicating that both genes are not clustered on the genome.

HNF-1 alpha and HNF-1 beta (vHNF-1) share dimerization and homeo domains, but not activation domains, and form heterodimers in vitro

HNF-1 alpha (previously referred to as HNF-1, LPB1, and APF) is a vertebrate transcription factor that contains a divergent homeo domain and plays a prominent role in regulating genes that have the common characteristic of being expressed in hepatocytes and a complex group of endodermally and mesodermally derived tissues. HNF-1 alpha is unique among the vertebrate homeo domain-containing proteins in that it dimerizes in the absence of its DNA recognition sequence, suggesting the possibility that the function of HNF-1 alpha may be diversified by forming heterodimers with other related proteins. We report the initial characterization of HNF-1 beta, which is closely related to HNF-1 alpha and is able to form heterodimers with HNF-1 alpha in vitro. Although HNF-1 alpha, but not HNF-1 beta, is expressed in the liver, HNF-1 alpha and HNF-1 beta are coexpressed in the murine Hepa1A cell line and in the mammalian kidney where a subset of hepatocyte genes are expressed. In contrast, exclusive expression of HNF-1 beta is associated with repression of a subset of hepatocyte-specific genes in the dedifferentiated hepatocyte cell line C2, differentiated F9 cells, in somatic hybrids between hepatocytes and fibroblasts, and in the lung. The extent of heterodimerization may be regulated in a tissue-specific way because freely exchangeable heterodimers are formed in Jurkat T cells transfected with HNF-1 alpha and HNF-1 beta, whereas in liver cells stable homodimers are present. These studies define a pair of homeo domain proteins that have the potential to interact to produce an embryologically complex pattern of gene expression.

The ubiquitous transcription factor Oct-1 and the liver-specific factor HNF-1 are both required to activate transcription of a hepatitis B virus promoter

The liver-specific transcription factor HNF-1 activates transcription of several mammalian hepatocyte-specific genes. The hepatitis B virus preS1 promoter shows hepatocyte specificity, which has been ascribed to binding of HNF-1 to a cognate DNA sequence upstream of the TATA box. We show here that there is an adjacent site that binds the ubiquitous transcription factor Oct-1. Both the Oct-1 and HNF-1 sites are necessary for liver-specific transcription of the preS1 promoter, but neither site alone activates transcription. The Oct-1 site is also necessary for activation of the preS1 promoter in HeLa cells, expressing transfected HNF-1. Our results show that while Oct-1 is not restricted to hepatocytes, it nevertheless can play a critical role in the expression of a liver-specific gene.

The ubiquitous transcription factor Oct-1 and the liver-specific factor HNF-1 are both required to activate transcription of a hepatitis B virus promoter

The liver-specific transcription factor HNF-1 activates transcription of several mammalian hepatocyte-specific genes. The hepatitis B virus preS1 promoter shows hepatocyte specificity, which has been ascribed to binding of HNF-1 to a cognate DNA sequence upstream of the TATA box. We show here that there is an adjacent site that binds the ubiquitous transcription factor Oct-1. Both the Oct-1 and HNF-1 sites are necessary for liver-specific transcription of the preS1 promoter, but neither site alone activates transcription. The Oct-1 site is also necessary for activation of the preS1 promoter in HeLa cells, expressing transfected HNF-1. Our results show that while Oct-1 is not restricted to hepatocytes, it nevertheless can play a critical role in the expression of a liver-specific gene.

Molecular cloning, functional expression, and chromosomal localization of mouse hepatocyte nuclear factor 1

The homeodomain-containing transcription factor hepatocyte nuclear factor 1 (HNF-1) most likely plays an essential role during liver organogenesis by transactivating a family of greater than 15 predominantly hepatic genes. We have isolated cDNA clones encoding mouse HNF-1 and expressed them in monkey COS cells and in the human T-cell line Jurkat, producing HNF-1 DNA-binding activity as well as transactivation of reporter constructs containing multimerized HNF-1 binding sites. In addition, the HNF-1 gene was assigned by somatic cell hybrids and recombinant inbred strain mapping to mouse chromosome 5 near Bcd-1 and to human chromosome 12 region q22-qter, revealing a homologous chromosome region in these two species. The presence of HNF-1 mRNA in multiple endodermal tissues (liver, stomach, intestine) suggests that HNF-1 may constitute an early marker for endodermal, rather than hepatocyte, differentiation. Further, that HNF-1 DNA-binding and transcriptional activity can be conferred by transfecting the HNF-1 cDNA into several cell lines indicates that it is sufficient to activate transcription in the context of ubiquitously expressed factors.

A distal dimerization domain is essential for DNA-binding by the atypical HNF1 homeodomain

Hepatic Nuclear Factor 1 (HNF1, also referred to as LFB1, HP1 or APF) is a liver-specific transcription factor required for the expression of many hepatocyte specific genes. We report here the purification of this rat liver nuclear protein and the cloning of its cDNA using a PCR-derived approach. Seven independent clones reveal 3 alternative polyadenylation sites and a unique open reading frame. Both a motif homologous to the homeodomain and a distal dimerization domain are required for specific DNA binding. Sequence comparisons reveal several atypical features at key positions in the segment corresponding to helices III and IV of the Antaennapedia homeodomain as well as a potential 24 amino acid loop in place of the universal turn between helices II and III. Together with its property to dimerize in the presence or absence of DNA, these features place HNF1 as the prototype of a novel subclass of transcription factors distantly related to homeoproteins.

Cloning of human hepatic nuclear factor 1 (HNF1) and chromosomal localization of its gene in man and mouse

HNF1 is a transcription factor that is required for hepatocyte-specific expression of several genes, including albumin and fibrinogen. Rat HNF1-encoding cDNAs have recently been cloned, revealing that this factor is a distant member of the homeoprotein family. We have now isolated HNF1 clones from a human liver cDNA library by using a rat HNF1 cDNA-derived probe. The longest clone, HCL20, contains a sequence corresponding to the intact rat HNF1-coding region followed by a 3' nontranslated region and a poly(A) tail, hence representing an almost full-length HNF1 cDNA. Alignment of the human and rat sequences shows that HNF1 is highly conserved between the two species. The HNF1 gene was mapped by in situ hybridization and by RFLP analysis of interspecific mouse backcrosses to chromosomes 12q24.3 and 5F in human and mouse, respectively, establishing a new segmental homology between these two chromosomes.

HNF-3A, a hepatocyte-enriched transcription factor of novel structure is regulated transcriptionally

Hepatocyte-specific gene expression requires the interaction of many proteins with multiple binding sites in the regulatory regions. HNF-3 is a site found to be important in the maximal hepatocyte-specific expression of several genes. We find that liver nuclear extracts contain three major binding activities for this site, which we call HNF-3A, HNF-3B, and HNF-3C. Purification from rat liver nuclear extracts of HNF-3A and HNF-3C reveals that each activity corresponds to a distinct polypeptide, as determined by SDS-PAGE. Peptide sequence derived from the most abundant species, HNF-3A, was used for synthesizing probes with which to isolate a cDNA clone of this protein. The encoded protein contains 466 amino acids (48.7 kD) and has binding properties identical to those of the purified protein. A 160-amino-acid region that does not resemble the binding domain of any known transcription factor is essential for DNA binding. The mRNA for HNF-3A is present in the rat liver but not in brain, kidney, intestine, or spleen, and the basis for this difference is cell-specific regulation of HNF-3A gene transcription.

A myosin-like dimerization helix and an extra-large homeodomain are essential elements of the tripartite DNA binding structure of LFB1

The transcription activator LFB1 is a major determinant of hepatocyte-specific expression of many genes. To study the mechanisms underlying LFB1 transcriptional selectivity, we have initiated its biochemical characterization. By in vitro complementation assays we have defined two distinct regions required for high levels of transcription, which resemble previously described activation domains. In contrast, the region of LFB1 necessary for DNA binding displays several novel features. The DNA binding domain is tripartite, including a homeodomain of unusual length (81 amino acids) and an N-terminal helix similar to part of myosin. This helical region mediates dimerization, which is shown to be essential for DNA binding.