Table of contents

Mammalian Forkhead homologs: The HNF3 (Foxa1-3) family

The HNF-3 alpha, beta and gamma genes constitute a family of transcription factors that are required for hepatocyte-specific gene expression of a number of genes, e.g. transthyretin, alpha-1 antitrypsin and tyrosine aminotransferase. HNF-3 beta mRNA is expressed in the node at the anterior end of the primitive streak in all three germ layers and is the first gene of this family to be activated. Subsequently, HNF-3 alpha is transcribed in the primitive endoderm in the region of the invaginating foregut and HNF-3 gamma appears upon hindgut differentiation. These genes have different anterior boundaries of mRNA expression in the developing endoderm and transcripts are found in all endoderm-derived structures that differentiate posterior to this boundary. Therefore, these genes define regionalization within the definitive endoderm. Furthermore, differential mRNA expression of HNF-3 alpha and beta is detected in cells of the ventral neural epithelium, chordamesoderm and notochord. In the neural epithelium, expression of HNF-3 alpha and beta mRNA becomes localised to cells of the floor plate. In addition to their characterised requirement for liver-specific gene expression, HNF-3 alpha and beta are required for mesoderm and neural axis formation. HNF-3 beta appears to be the true orthologue of the Drosophila forkhead gene. In adult mouse tissues, transcripts from HNF-3 alpha and beta have been localised to the liver, intestine and lung, whereas HNF-3 gamma is found in the liver, intestine and testis (Monaghan, 1993).

The visceral yolk sac plays a critical role in normal embryogenesis, yet little is known about the specific molecules that regulate its development. Four winged-helix genes (HNF-3alpha, HNF-3beta, HNF-3gamma and HFH-4) are restricted to visceral endoderm. In the absence of HNF-3beta, visceral endoderm forms but the morphogenetic movements by which the embryo becomes enclosed within its yolk sac are disrupted and serum protein gene transcription is greatly reduced. Hedgehog and Bmp genes, which encode signaling molecules known to play multiple roles in embryonic development, are also differentially expressed in the closely apposed yolk sac mesoderm and endoderm layers. It is thought that Indian hedgehog signals from the visceral mesoderm to establish BMP2, BMP4 and BMP6 in the yolk sac mesoderm. All three BMPs may amplify their own transcription by an autoregulatory mechanism and participate in the differentiation of mesodermal cells. In an autocrine role, Indian hedgehog also signals to establish BMP6 in visceral mesoderm. Desert hedgehog may signal back from yolk sac mesoderm to induce BMP6 in visceral endoderm. These results suggest that similar mechanisms may be utilized to mediate inductive interactions in both extraembryonic and embryonic tissues (Farrington, 1997).

Mouse HNF3 beta was recovered from an embryo cDNA library. The gene is first expressed in the anterior portion of the primitive streak at the onset of gastrulation, in a region where definitive endoderm first arises. Its expression persists in axial structures derived from the mouse equivalent of Hensen's node, namely definitive endoderm and notochord, and in the ventral region of the developing neural tube. Expression of the highly related gene, HNF3 alpha, appears to initiate later than HNF3 beta and is first seen in midline endoderm cells. Expression subsequently appears in notochord, ventral neural tube, and gut endoderm in patterns similar to HNF3 beta. HNF3 proteins are detectable in the midgut at 9.5 days of development. At later stages HNF3 mRNAs and protein are expressed strongly in endoderm-derived tissues such as the liver. HNF3 is also the only known hepatocyte-enriched transcription factor present in a highly de-differentiated liver cell line that retains the capacity to redifferentiate to the hepatic phenotype. These studies suggest that HNF3 alpha and HNF3 beta are involved in both the initiation and maintenance of the endodermal lineage (Ang, 1993).

During early embryogenesis, the transcription factor HNF3beta is expressed in visceral and definitive endoderm, node, notochord and floorplate. A targeted mutation in the HNF3beta gene results in the lack of a definitive node and notochord. Furthermore, lack of HNF3beta results in failure of proper primitive streak elongation. To address whether HNF3beta is required in visceral endoderm, tetraploid embryo-ES cell aggregations were used to generate chimeric mouse embryos with wild-type visceral endoderm and homozygous mutant HNF3beta embryonic ectoderm or vice versa. Replacing the visceral endoderm of mutant HNF3beta embryos rescues proper primitive streak elongation and, conversely, mutant visceral endoderm imposes a severe embryonic-extraembryonic constriction on wild-type embryonic ectoderm. Restoration of normal streak morphogenesis is not sufficient to allow formation of the node and notochord in HNF3beta mutant embryos. Thus, these results demonstrate that HNF3beta has two separate roles in primitive streak formation. One is to act within the visceral endoderm to promote proper streak morphogenesis. The second is autonomous to the node and its precursors and involves specification of node and notochord cell fates. HNF3beta mutant embryos rescued for the embryonic-extraembryonic constriction develop further than mutant embryos, allowing examination of later roles for HNF3beta. Such mutant embryos lack foregut and midgut endoderm. In addition, left-right asymmetry is affected in the mutant embryos (Dufort, 1998).

Recent embryological and genetic experiments have suggested that the anterior visceral endoderm and the anterior primitive streak of the early mouse gastrula function as head- and trunk-organizing centers, respectively. HNF3beta and Lim1 are coexpressed in both organizing centers suggesting synergistic roles for these genes in regulating organizer functions and hence axis development in the mouse embryo. To investigate this possibility, compound HNF3beta and Lim1 mutant embryos were generated. An enlarged primitive streak and a lack of axis formation were observed in double mutant but not in single homozygous mutant embryos. Chimera experiments indicate that the primary defect in these double homozygous mutants is due to loss of activity of HNF3beta and Lim1 in the visceral endoderm. Altogether, these data provide evidence that these genes function synergistically to regulate organizer activity of the anterior visceral endoderm. Moreover, double mutant embryos also exhibit defects in mesoderm patterning that are likely due to lack of specification of anterior primitive streak cells (Perea-Gómez, 1999).

The first morphological sign of A-P pattern in the epiblast of the mouse embryo is the site of formation of the primitive streak at the posterior end of the embryo. The genetic pathway that initiates primitive streak formation remains to be elucidated, but expression of T on one side of the epiblast at the onset of gastrulation marks posterior primitive streak cells. In HNF3beta,Lim1 double mutant embryos, T expression in the epiblast is no longer restricted posteriorly, but is instead expressed throughout the epiblast by the mid-streak stage. Thus, A-P polarity of the epiblast is abnormal in HNF3beta,Lim1 embryos and widespread expression of T strongly suggests that mutant epiblast cells are transformed into primitive streak cells. The loss of epiblast cells is confirmed by the absence and reduction of expression of Otx2 and Oct4, respectively. In addition, mid-streak-stage embryos also show ectopic mesoderm formation as demonstrated by the expression of MesP1 and Lefty2. As a consequence of these early patterning defects, ectoderm and neurectoderm cells that are derived from distal and anterior epiblast cells are missing in these embryos at 7.5-7.75 d.p.c. These epiblast defects are not observed in single homozygous HNF3beta and Lim1 mutants. Altogether, these data demonstrate that HNF3beta and Lim1 function synergistically to establish A-P patterning of the epiblast and to restrict primitive streak formation to the posterior side of mouse embryos (Perea-Gómez, 1999).

Posterior neuropore (PNP) closure coincides with the end of gastrulation, marking the end of primary neurulation and primary body axis formation. Secondary neurulation and axis formation involve differentiation of the tail bud mesenchyme. Genetic control of the primary-secondary transition is not understood. A detailed analysis of gene expression in the caudal region of day 10 mouse embryos during primary neuropore closure is reported. Embryos were collected at the 27-32 somite stage, fixed, processed for whole mount in situ hybridization, and subsequently sectioned for a more detailed analysis. Genes selected for study include those involved in the key events of gastrulation and neurulation at earlier stages and more cranial levels. Patterns of expression within the tail bud, neural plate, recently closed neural tube, notochord, hindgut, mesoderm, and surface ectoderm are illustrated and described. Specifically, continuity of expression of the genes Wnt5a, Wnt5b, Evx1, Fgf8, RARgamma, Brachyury, and Hoxb1 from primitive streak and node into subpopulations of the tail bud and caudal axial structures is reported. Within the caudal notochord, developing floorplate, and hindgut, HNF3alpha, HNF3beta, Shh, and Brachyury expression domains correlate directly with known genetic roles and predicted tissue interdependence during induction and differentiation of these structures. The patterns of expression of Wnt5a, Hoxb1, Brachyury, RARgamma, and Evx1, together with observations on proliferation, reveal that the caudal mesoderm is organized at a molecular level into distinct domains delineated by longitudinal and transverse borders before histological differentiation. Expression of Wnt5a in the ventral ectodermal ridge supports previous evidence that this structure is involved in epithelial-mesenchymal interaction. These results provide a foundation for understanding the mechanisms facilitating transition from primary to secondary body axis formation, as well as the factors involved in defective spinal neurulation (Gofflot, 1999).

Nucleosome positioning at genetic regulatory sequences is not well understood. The transcriptional enhancer of the mouse serum albumin gene is active in liver, where regulatory factors occupy their target sites on three nucleosome-like particles designated N1, N2, and N3. The winged helix transcription factor HNF3 binds to two sites near the center of the N1 particle. Dinucleosome templates were created using the albumin enhancer sequence. Site-specific binding of HNF3 protein results in nucleosome positioning in vitro similar to that seen in liver nuclei. Histones move from random positionings to site specific positioning underlying HNF3's binding site. Regarding the positioning mechanism, HNF3 bends DNA slightly toward the protein. If HNF3 binds the side of the DNA as it curves around the core histones, or inside the gyre, as has been proposed for winged helix linker histones, then HNF3 binding might help DNA bend around the histone octamer and thereby stabilize the position of the core particle. More generally, the combination of HNF3, core histones, and DNA sequence may stabilize a DNA distortion that favors the wrapping of the DNA around the histone octamer (Shim, 1998).

The regulation of transcription factors HNF-3alpha and HNF-3beta was studied during the retinoic acid-mediated differentiation of mouse P19 cells. Retinoic acid treatment converts P19 stem cells into neurons and astrocytes; gene expression of both HNF-3alpha and HNF-3beta is activated during this process. HNF-3alpha transcription is detected 2 h after addition of retinoic acid and takes place in the absence of de novo protein synthesis. This suggests that HNF-3alpha is a primary target for retinoic acid action. HNF-3alpha induction displays a biphasic profile; HNF-3alpha mRNA reaches maximal levels at 2 and 6 days postdifferentiation. Additional experiments strongly suggest that the second peak is due to HNF-3alpha induction in postmitotic neurons. In contrast, P19 stem cells do not contain any detectable HNF-3alpha mRNA. The retinoic acid-mediated induction of HNF-3alpha occurs at the level of transcriptional initiation and is conferred by distal promoter sequences. In comparison to HNF-3alpha, HNF-3beta induction is a subsequent event and detectable levels of HNF-3beta mRNA materialize approximately 1 day after addition of retinoic acid to P19 stem cells. Time course studies firmly demonstrate that HNF-3beta mRNA peaks at about 2 days postdifferentiation and then declines to virtually unreadable levels. This temporal pattern is consistent with HNF-3beta being a secondary target for retinoic acid. By analogy to HNF-3alpha, HNF-3beta activation also takes place at the level of transcriptional initiation. Recent studies implicate HNF-3alpha and HNF-3beta in early mammalian neurogenesis (Jacob, 1997).

Within the neural tube of vertebrate embryos, the floor plate plays important roles in ventral pattern formation and axonal guidance. A critical event for floor plate development is the induction of a winged helix transcription factor, Hepatocyte Nuclear Factor-3ß (HNF-3ß). The enhancer for floor plate expression of HNF-3ß is located 3' of the transcription unit and consists of multiple elements. HNF-3ß induction depends on the notochord-derived signal, Sonic hedgehog (Shh, a homolog of Drosophila Hedgehog), that signals througe Gli, a vertebrate homolog of Drosophila Cubitus interruptus. A Gli-binding site is required for the activity of the minimal floor plate enhancer of HNF-3ß in vivo. Three Gli genes (Gli, Gli2 and Gli3) are differentially expressed in the developing neural tube. Gli expression is restricted to the ventral part, while Gli2 and Gli3 are expressed (respectively) throughout the neural tube and dorsally. Strong expression of Gli and Gli2, and weak expression of Gli3 transiently overlap with HNF-3ß at the time of its induction. Consistent with ventrally localized expression, Gli expression can be up-regulated by Shh. Finally, the Gli-binding site acts as a Shh responsive element, and human GLI (but not GLI3) can activate this binding site in tissue culture. Taken together, these findings suggest that Gli, and probably also Gli2, are good candidates for transcriptional activators of the HNF-3ß floor plate enhancer, and the binding site for Gli proteins is a key element for response to Shh signaling. These results also support the idea that Gli/Ci are evolutionarily conserved transcription factors in the Hedgehog signaling pathway (Sasaki, 1997).

The homeobox gene goosecoid (See Drosophila Goosecoid) and the winged-helix gene Hepatic Nuclear Factor-3ß (HNF-3ß) are co-expressed in all three germ layers in the anterior primitive streak and at the rostral end of mouse embryos during gastrulation. In the early mid-streak stage, gsc expression is encompased within the domain of expression of HNF-3ß, while HNF-3ß is expressed by itself in the ectoderm germ layer, in particular in the distal-most region of the primitive streak. Since the two genes are coexpressed, a test was made as to whether they interact. Double-mutant embryos of genotype gsc (-/-);HNF-3ß(+/-) show a new phenotype as early as embryonic days 8.75. There is a dramatic reduction in forebrain size, abnormal branchial arches and defects in heart looping. Analysis of D-V molecular markers demonstrate that dorsal cell fates are expanded ventrally, while ventral cell fates, including optic vesicles in the diencephalon and floor plate cells in the midbrain and hindbrain, are missing in severely affected individuals. Loss of Sonic hedgehog and HNF-3ß expression is observed in the notochord and ventral neural tube of these embryos. These results indicate that gsc and HNF-3ß interact to regulate Shh expression and consequently dorsal-ventral patterning in the neural tube. In the forebrain of the mutant embryos, severe growth defects and absence of optic vesicles could involve loss of expression of fibroblast growth factor-8, in addition to Shh. These results also suggest that interaction between gsc and HNF-3ß regulate other signaling molecules required for proper development of the foregut, branchial arches and heart (Filosa, 1997).

Hepatocyte nuclear factor-3beta (HNF-3beta), a nuclear protein of the winged helix family of transcription factors, is known to play a critical role in the formation of the embryonic node, notochord, and foregut endoderm. HNF-3beta influences the expression of a number of target genes in the respiratory epithelium, activating transcription of thyroid transcription factor-1, surfactant protein-B and clara cell secretory protein. In order to discern the role of HNF-3beta in differentiation and gene expression in the lung, HNF-3beta was expressed in developing respiratory epithelial cells of transgenic mice, under the control of the human surfactant protein C gene promoter. Pulmonary abnormalities are observed in the lungs of fetal mice bearing the HNF-3beta transgene. Differentiation of distal respiratory epithelial cells is arrested in the early pseudoglandular stage. Branching morphogenesis and vasculogenesis are markedly disrupted in association with decreased E-cadherin and vascular endothelial growth factor expression. HNF-3beta limits cellular diversity of developing respiratory epithelium and alters lung morphogenesis in vivo, suggesting that precise temporal-spatial regulation of HNF-3beta expression is critical for respiratory epithelial cell differentiation and lung morphogenesis (Zhou, 1997).

Murine HNF-3 beta mediates cell-specific transcription of genes important for the function of hepatocytes, intestinal and bronchiolar epithelial cells, and pancreatic acinar cells. A liver-enriched transcription factor, HNF-6, is required for HNF-3 beta promoter activity and also recognizes the regulatory region of numerous hepatocyte-specific genes. In this study the yeast one-hybrid system was used to isolate the HNF-6 cDNA, which encodes a cut-homeodomain-containing transcription factor (see Drosophila Cut) that binds with the same specificity as the liver HNF-6 protein. Cotransfection assays demonstrate that HNF-6 activates expression of a reporter gene driven by the HNF-6 binding site from either the HNF-3 beta or transthyretin (TTR) promoter regions. Interspecific backcross analysis was used to determine that the murine Hnf6 gene is located in the middle of mouse chromosome 9. In situ hybridization studies of staged specific embryos demonstrate that HNF-6 and its potential target gene, HNF-3 beta, are coexpressed in the pancreatic and hepatic diverticulum. More detailed analysis of HNF-6 and HNF-3 beta's developmental expression patterns provides evidence of colocalization in hepatocytes, intestinal epithelial, and in the pancreatic ductal epithelial and exocrine acinar cells. The expression patterns of these two transcription factors do not overlap in other endoderm-derived tissues or the neurotube. HNF-6 is also abundantly expressed in the dorsal root ganglia, the marginal layer, and the midbrain. At day 18 of gestation and in the adult pancreas, HNF-6 and HNF-3 beta transcripts colocalize in the exocrine acinar cells, but their expression patterns diverge in other pancreatic epithelium. HNF-6 expression, but not the expression of HNF-3 beta, continues in the pancreatic ductal epithelium, whereas only HNF-3 beta becomes restricted to the endocrine cells of the islets of Langerhans (Rausa, 1997).

The hepatic nuclear factor 3gamma (Hnf3g) is a member of the winged helix gene family of transcription factors and is thought to be involved in anterior-posterior regionalization of the primitive gut. Hnf3g differs from HNF3ß in that the former is expressed exclusively in gut and liver, and not in the notochord and floor plate. In this study, cis-regulatory elements essential for the expression of Hnf3g in vivo have been characterized. To this end, a 170 kb yeast artificial chromosome (YAC) carrying the entire Hnf3g locus was isolated and modified with a lacZ reporter gene. The two mouse lines carrying the unfragmented Hnf3g-lacZ YAC show tissue-specific, copy number-dependent and position-independent expression, proving that 170 kb of the Hnf3g locus contain all elements important in the regulation of Hnf3g. Cis-regulatory elements necessary for expression of Hnf3g were identified in a three-step procedure:

  1. DNase I hypersensitive site mapping was used to delineate important chromatin regions around the gene required for tissue-specific activation of Hnf3g.
  2. Plasmid-derived transgenes and gene targeting of the endogenous Hnf3g gene locus were used to demonstrate that the 3'-flanking region of the gene is necessary and sufficient to direct reporter gene expression in liver, pancreas, stomach and small intestine.
  3. A binding site for HNF-1alpha and beta, factors expressed in organs derived from the endoderm (such as liver, gut and pancreas) was identified in this 3'-enhancer and shown to be crucial for enhancer function in vitro.
Based on its expression pattern it is inferred that HNF-1beta is a likely candidate for directly activating Hnf3g gene expression during development (Hiemisch, 1997).

Mammalian hepatocyte nuclear factor-3 (HNF-3) and the Drosophila homeotic gene fork head proteins are prototypes of an extensive family of cell-specific transcription factors that share homology in the winged helix DNA-binding domain. One of these mammalian family members, HNF-3/fork head homolog-4 (HFH-4), has been isolated by PCR amplification of rodent brain cDNA and it exhibits abundant expression in the adult bronchiolar epithelium. HFH-4 is expressed in both the presumptive and differentiated choroid plexus epithelium, which is responsible for the synthesis and secretion of cerebrospinal fluid (CSF) proteins. HFH-4 is a potent transcriptional activator in cotransfection assays. Several protein sequences are defined in this paper that are important for HFH-4 transcriptional activity. HFH-4 protein recognizes the DNA consensus sequences HWDTGTTTGTTTA or KTTTGTTGTTKTW (where H is not G, W is A or T, D is not C, and K is G or T). This HFH-4 consensus was used to identify potential HFH-4 target genes in the choroid plexus epithelium. These promoter sequences bind to recombinant HFH-4 protein in electrophoretic mobility shift assays. Recombinant HFH-4 forms specific protein-DNA complexes with the promoter regions of the human prothrombin, beta amyloid precursor protein, alpha1-antichymotrypsin, cystic fibrosis transmembrane conductance regulator and rodent alpha2-macroglobulin, growth hormone receptors, and insulin-like growth factor II genes. Putative HFH-4 target genes were identified in the bronchiolar epithelium, including the clara cell secretory protein gene and the HNF-3alpha gene, a winged helix family member involved in the transcriptional regulation of genes in the bronchiolar epithelium. Cotransfection assays show that HFH-4 potentiates expression of the HNF-3alpha and clara cell secretory protein promoter regions (Lim, 1997).

Members of the hepatic nuclear factor 3 (HNF3) family, including HNF3alpha, beta and gamma, play important roles in embryonic development, the establishment of tissue-specific gene expression and the regulation of gene expression in differentiated tissues. The transducin-like Enhancer of split (TLE) proteins, the human homologs of Drosophila Groucho, directly associate with HNF3beta. The CRII region of HNF3beta (a.a. 361-388) is responsible for the interaction with TLE1. The CRII region is conserved between all three mammalian HNF3 proteins and the Drosophila Forkhead protein. The CRII region of HNF3beta does not contain the WRPW motif that is required for the interaction between TLE/Groucho and the Hairy, Hairy/Enhancer of split-like (HES). The CRII region also does not resemble the eh1/GEH motif that is required for the interaction between TLE/Groucho and Engrailed or Goosecoid. The CRII region, however, does have an FNHPF sequence that may serve as a TLE binding site, since it has two aromatic residues separated by a proline. A mammalian two hybrid assay was used to confirm that this interaction occurs in vivo. Overexpression of TLE1 in HepG2 and HeLa cells decreases transactivation mediated through the C-terminal domain of HNF3beta, and Grg5, a naturally occurring dominant negative form of Groucho/TLE, also increases the transcriptional activity of this region of HNF3. These results suggest that TLE proteins could influence the expression of mammalian genes regulated by HNF3 (Wang, 2000).

Protein interactions of Foxa1-3 family members

Transgenic mice expressing the homeobox gene Hoxa5 under the control of Hoxb2 regulatory elements present a growth arrest during weeks two and three of postnatal development, resulting in proportionate dwarfism. These mice present a liver phenotype illustrated by a 12-fold increase in liver insulin-like growth factor binding protein 1 (IGFBP1) mRNA and a 50% decrease in liver insulin-like growth factor 1 (IGF1) mRNA correlated with a 50% decrease in circulating IGF1. The Hoxa5 transgene is expressed in the liver of these mice, leading to an overexpression of total (endogenous plus transgene) Hoxa5 mRNA in this tissue. Several cell lines were used to investigate a possible physiological interaction of Hoxa5 with the main regulator of IGFBP1 promoter activity, the Forkhead box transcription factor FKHR. In HepG2 cells, Hoxa5 has little effect by itself but inhibits the FKHR-dependent activation of the IGFBP1 promoter. In HuF cells, Hoxa5 cooperates with FKHR to dramatically enhance IGFBP1 promoter activity. This context-dependent physiological interaction probably corresponds to the existence of a direct interaction between Hoxa5 and FKHR and FoxA2/HNF3ß, as demonstrated by pull-down experiments achieved either in vitro or after cellular co-expression. In conclusion, it is proposed that the impaired growth observed in this transgenic line relates to a liver phenotype best explained by a direct interaction between Hoxa5 and liver-specific Forkhead box transcription factors, in particular FKHR but also Foxa2/HNF3ß. Because Hoxa5 and homeogenes of the same paralog group are normally expressed in the liver, the present results raise the possibility that homeoproteins, in addition to their established role during early development, regulate systemic physiological functions. (Foucher, 2002).

Table of contents

forkhead: Biological Overview | Regulation | Targets of Activity | Developmental Biology | Effects of Mutation | References

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.