Table of contents

Regulation of Foxa1-3

The expression of a winged-helix transcription factor, Foxa2/HNF3ß, is essential for development of the node and the notochord. The node/notochord enhancer of mouse Foxa2 was examined for sequence motifs conserved across vertebrate species. Foxa2 genes were cloned from chicken and fish, and the respective node/notochord enhancers, that were active in transgenic mouse embryos, were identified. Comparison of the sequences of the enhancers revealed three evolutionally conserved sequence motifs: CS1, CS2 and CS3. Mutational analysis of the mouse enhancer indicates that CS3 is indispensable for gene expression in the node and the notochord, while CS1 and CS2 are required to augment enhancer activity. These motifs do not correspond to the consensus binding sequences of transcription factors known to be involved in node/notochord development (Nishizaki, 2001).

The cell population and the activity of the organizer change during the course of development. The mechanism of mouse node development has been addressed via an analysis of the node/notochord enhancer (NE) of Foxa2. The core element (CE) of the enhancer, which in multimeric form drives gene expression in the node, was identified. The CE is activated in Wnt/ß-catenin-treated P19 cells with a time lag, and this activation is dependent on two separate sequence motifs within the CE. These same motifs are also required for enhancer activity in transgenic embryos. The Tead family of transcription factors (see Drosophila Scalloped) was identified as binding proteins for the 3' motif. Teads and their co-factor YAP65 activate the CE in P19 cells, and binding of Tead to CE is essential for enhancer activity. Inhibition of Tead activity by repressor-modified Tead compromises NE enhancer activation and notochord development in transgenic mouse embryos. Furthermore, manipulation of Tead activity in zebrafish embryos leads to altered expression of foxa2 in the embryonic shield. These results suggest that Tead activates the Foxa2 enhancer core element in the mouse node in cooperation with a second factor that binds to the 5' element, and that a similar mechanism also operates in the zebrafish shield (Sawada, 2005).

Transcriptional targets of Foxa1-3

The axial midline mesoderm and the ventral midline of the neural tube, the floor plate, share the property of being a source of the secreted protein, Sonic hedgehog (Shh), which has the capacity to induce a variety of ventral cell types along the length of the mouse CNS. Three enhancers within the Shh promoter region were identified that direct lacZ expression to distinct regions along the anteroposterior axis including the ventral midline of the spinal cord, hindbrain, rostral midbrain and caudal diencephalon, suggesting that multiple transcriptional regulators are required to initiate Shh gene expression within the CNS. Sequence analysis of the genomic clones responsible for enhancer activity from a variety of organisms, including mouse, chicken and human, have identified highly conserved binding sites for the hepatocyte nuclear factor 3 (Hnf3) family of transcriptional regulators in some, but not all, of the enhancers. A Shh floor plate enhancer (SFPE1) has been identified within a 1.1 kb genome fragment located approximately 8 kb upstream of the Shh transcriptional start site. Intronic enhancers direct reporter gene expression to cranial and spinal cord regions. The location of an Shh brain enhancer (SBE1) was delineated to within a 2.2 kb segment within the second intron. SFPE1 activity is regulated independently of Hnf3 function, while SBE1 and SFPE2, a second intronic floor plate enhancer directing expression to hindbrain and spinal cord, both rely on Hnf3 function for midline but not lateral Shh expression. Moreover, the generation of mutations in the Hnf3-binding sites shows their requirement in certain, but not all, aspects of Shh reporter expression. Taken together, these results support the existence of Hnf3-dependent and -independent mechanisms in the direct activation of Shh transcription within the CNS and axial mesoderm (Epstein, 1999).

Hepatocyte nuclear factors (HNFs) are a heterogeneous class of evolutionarily conserved transcription factors that are required for cellular differentiation and metabolism. Mutations in HNF-1alpha and HNF-4alpha genes impair insulin secretion and cause type 2 diabetes. Regulation of HNF-4/HNF-1 expression by HNF-3alpha and HNF-3beta was studied in embryoid bodies in which one or both HNF-3alpha or HNF-3beta alleles were inactivated. HNF-3beta positively regulates the expression of HNF-4alpha/HNF-1alpha and their downstream targets, implicating a role in diabetes. HNF-3beta is also necessary for expression of HNF-3alpha. In contrast, HNF-3alpha acts as a negative regulator of HNF-4alpha/HNF-1alpha, demonstrating that HNF-3alpha and HNF-3beta have antagonistic transcriptional regulatory functions in vivo. HNF-3alpha does not appear to act as a classic biochemical repressor but rather exerts its negative effect by competing for HNF-3 binding sites with the more efficient activator HNF-3beta. In addition, the HNF-3alpha/HNF-3beta ratio is modulated by the presence of insulin, providing evidence that the HNF network may have important roles in mediating the action of insulin (Duncan, 1998).

The transcription factors of the hepatocyte nuclear factor 3 (HNF3) family, which are active in the liver, are expressed early during endoderm differentiation. Their role in early murine development was examined, specifically, their role in embryonic stem (ES) cells. HNF3alpha or HNF3beta mRNA transcripts are not detected in ES cells before differentiation, and only low levels of HNF3beta mRNA are detected at a late stage in the differentiation of ES cells into embryoid bodies (EB), occurring 20 days after induction of differentiation. To examine the consequences of overexpressing HNF3alpha or -beta in ES cells, the two genes were transferred into these cells and the levels of expression of tissue-specific genes were determined during EB differentiation. Specifically, the expression of albumin, cystic fibrosis transmembrane conductance regulator (CFTR), phosphoenolpyruvate carboxykinase (PEPCK), alpha1-antitrypsin, transthyretin, zeta-globin, and neurofilament 68kd were all examined as markers for different cell lineages. Overexpression of HNF3beta (and to a lesser extent of HNF3alpha) induces the expression of genes associated with endodermal lineage, namely, the genes for CFTR and albumin, but does not induce the expression of genes involved in late endoderm differentiation, such as the genes for PEPCK and alpha1-antitrypsin. Moreover, expression of HNF1beta is highly induced in HNF3-overexpressing cells, while expression of HNF1alpha and HNF4 is only mildly induced in these cells. Therefore, HNF3alpha and -beta seem to be involved in early endoderm differentiation of ES cells and together with other developmental factors are apparently needed for the induction of the endodermal lineage in vivo (Levinson-Dushnik, 1997).

Gene inactivation studies have shown that members of the GATA family of transcription factors are critical for endoderm differentiation in mice, flies and worms, yet how these proteins function in such a conserved developmental context has not been understood. In vivo footprinting of mouse embryonic endoderm cells was used to show that a DNA-binding site for GATA factors is occupied on a liver-specific, transcriptional enhancer of the serum albumin gene. The albumin enhancer is co-occupied with an adjacent binding site for HNF3 in the embryonic gut endoderm, embryonic hepatocytes and adult liver. Increasing amounts of HNF3 leads to increasing amounts of a complex with both factors bound and a depletion of the GATA-4/DNA complex. The appearance of GATA-4 and HNF3 bound to the same DNA in the presence of excess enhancer probe indicates that the factors bind cooperatively. GATA site occupancy occurs in gut endoderm cells at their pluripotent stage: the cells have the potential to initiate tissue development but they have not yet been committed to express albumin or other tissue-specific genes. The GATA-4 isoform accounts for about half of the nuclear GATA-factor-binding activity in the endoderm. GATA site occupancy persists during hepatic development and is necessary for the activity of albumin gene enhancer. Thus, GATA factors in the endoderm are among the first to bind essential regulatory sites in chromatin. Binding occurs prior to activation of gene expression, changes in cell morphology or functional commitment that would indicate differentiation. It is suggested that GATA factors at target sites in chromatin may generally help potentiate gene expression and tissue specification in metazoan endoderm development (Bossard, 1998).

The mammalian homeobox gene pdx-1, which has no known Drosophila homolog, is expressed in pluripotent precursor cells in the dorsal and ventral pancreatic bud and duodenal endoderm, which will produce the pancreas and the rostral duodenum. In the adult, pdr-1 is expressed principally within insulin-secreting pancreatic islet beta cells and cells of the duodenal epithelium. Studies of transgenic mice in which a genomic fragment of the mouse pdx-1 gene from kb -4.5 to +8.2 was used to drive a beta-galactosidase reporter show that the control sequences sufficient for appropriate developmental and adult specific expression are contained within this region. Three nuclease-hypersensitive sites, located between bp -2560 and -1880 (site 1), bp -1330 and -800 (site 2), and bp -260 and +180 (site 3), were identified within the 5'-flanking region of the endogenous pdx-1 gene. Pancreatic beta-cell-specific expression is controlled by sequences within site 1. The activity of the site 1-driven constructs is reduced substantially in beta-cell lines by mutating a hepatocyte nuclear factor 3 (HNF3)-like site located between nucleotides -2007 and -1996. HNF3beta present in islet beta cells binds to this element. Immunohistochemical studies reveal that HNF3beta is present within the nuclei of almost all islet beta cells and subsets of pancreatic acinar cells. Together, these results suggest that HNF3beta, a key regulator of endodermal cell lineage development, plays an essential role in the cell-type-specific transcription of the pdx-1 gene in the pancreas (Wu, 1997).

The floor plate is an organizing center that controls neural differentiation and axonogenesis in the neural tube. The axon guidance molecule Netrin1 is expressed in the floor plate of zebrafish embryos. To elucidate the regulatory mechanisms underlying expression in the floor plate, the netrin1 locus was scanned for regulatory regions and an enhancer was identified that drives expression in the floor plate and hypochord of transgenic embryos. The expression of the transgene is ectopically activated by Cyclops (Nodal) signals but does not respond to Hedgehog signals. Other netrin1 enhancers, which have yet to be identified, control the observed ectopic expression of endogenous netrin1 in response to shh injection. The winged-helix transcription factor foxA2 (also HNF3ß, axial: Drosophila homolog Forkhead) is expressed in the notochord and floor plate. Knock-down of FoxA2 leads to loss of floor plate, while notochord and hypochord development is unaffected, suggesting a specific requirement of FoxA2 in the floor plate. The transgene is ectopically activated by FoxA2, and expression of FoxA2 leads to rescue of floor plate differentiation in mutant embryos that are deficient in Cyclops signalling. Zebrafish and mouse use different signalling systems to specify floor plate. The zebrafish netrin1 regulatory region also drives expression in the floor plate of mouse and chicken embryos. This suggests that components of the regulatory circuits controlling expression in the floor plate are conserved and that FoxA2 -- given its importance for midline development also in the mouse -- may be one such component (Rastegar, 2003).

The MAP1B (Mtap1b) promoter presents two evolutionary conserved overlapping homeoproteins and Hepatocyte nuclear factor 3ß (HNF3ß/Foxa2) cognate binding sites (defining putative homeoprotein/Fox sites, HF1 and HF2). Accordingly, the promoter domain containing HF1 and HF2 is recognized by cerebellum nuclear extracts containing Engrailed and Foxa2 and has regulatory functions in primary cultures of embryonic mesmetencephalic nerve cells. Transfection experiments further demonstrate that Engrailed and Foxa2 interact physiologically in a dose-dependent manner: Foxa2 antagonizes the Engrailed-driven regulation of the MAP1B promoter, and vice versa. This led to an investigation to see if Engrailed and Foxa2 interact directly. Direct interaction was confirmed by pull-down experiments, and the regions participating in this interaction were identified. In Foxa2 the interacting domain is the Forkhead box DNA-binding domain. In Engrailed, two independent interacting domains exist: the homeodomain and a region that includes the Pbx-binding domain. Finally, Foxa2 not only binds Engrailed but also Lim1, Gsc and Hoxa5 homeoproteins and in the four cases Foxa2 binds at least the homeodomain. Based on the involvement of conserved domains in both classes of proteins, it is proposed that the interaction between Forkhead box transcription factors and homeoproteins is a general phenomenon (Foucher, 2003).

Mapping of the interacting domains identified the Forkhead box binding domain in Foxa2 as the only domain interacting with Engrailed, Hoxa5, Lim1, and Gsc and Otx2. Similarly, for all homeoproteins tested, the homeodomain alone binds Foxa2. However, and in contrast with Foxa2, four out of these five homeoproteins contained additional Foxa2-interacting regions: Engrailed, Hoxa5, Gsc (in all three cases in the N-terminal sequence) and Otx2 [in its C-terminal sequence]. A detailed analysis of the interacting domains has been done for Engrailed only and the mapping of the other homeoproteins has been limited to the homeodomain, and its flanking N- and C-terminal regions, at large. In the case of Engrailed, in addition to the homeodomain, a short sequence (amino acids 146-199) overlapping the Pbx-interacting domain also binds Foxa2. This latter domain and the homeodomain bind independently to Foxa2 and the possibility that they interact with different sub-regions of the Forkhead box domain was not investigated. Such an additional non-homeodomain Foxa2 interacting domain was also present in the N-terminal sequences of Hoxa5 and Gsc, but not in Lim1. With the exception of the hexapeptide sequence present in Engrailed and Hoxa5, no further similarities were found between the Foxa2-binding domains identified outside the homeodomain in Engrailed, Hoxa5, Gsc and Otx2. It is thus possible that, in addition to the homeodomain, different homeoproteins have evolved separate Foxa2-binding regions with regulatory functions (Foucher, 2003).

In this context it is interesting that the fragment 146-199 of Engrailed includes the EH2 (homologous to hexapeptide in Hox proteins) and EH3 domains of Engrailed, both of which are implicated in functional interactions with Exd/Pbx homeoproteins. The same observation also holds for Hoxa5, for which the N-terminal sequence containing the hexapeptide sequence binds Foxa2. Both Pbx and Foxa2 might bind Engrailed (or Hoxa5) to form a tripartite complex or, alternatively, that Foxa2 and Pbx binding are mutually exclusive. Also intriguing is the fact that Engrailed and Gsc, as well as different Forkhead box proteins -- including BF1 and Foxa2 -- interact with co-factors of the Groucho/TLE family. Since the Groucho/TLE-interacting domains of Engrailed and Foxa2 have been mapped to the EH1 and CRII domains, respectively (two domains not involved in the Foxa2-Engrailed interaction) it is possible that larger complexes involving Groucho/TLE proteins, Forkhead transcription factors and homeoproteins form in vivo (Foucher, 2003).

The establishment of the floor plate at the ventral midline of the CNS is dependent on an inductive signaling process mediated by the secreted protein Sonic hedgehog (Shh). To understand molecularly how floor plate induction proceeds a Shh-responsive regulatory element was identified that directs transgene reporter expression to the ventral midline of the CNS and notochord in a Shh-like manner and critical cis-acting sequences regulating this element were characterized. Cross-species comparisons narrowed the activity of the Shh floor plate enhancer to an 88-bp sequence within intron 2 of Shh that included highly conserved binding sites matching the consensus for homeodomain, Tbx and Foxa transcription factors. Mutational analysis revealed that the homeodomain and Foxa binding sites are each required for activation of the Shh floor plate enhancer, whereas the Tbx site was required for repression in regions of the CNS where Shh is not normally expressed. Shh enhancer activity is detected in the mouse node from where the floor plate and notochord precursors derive. Shh reporter expression was restricted to the ventral (mesodermal) layer of the node in a pattern similar to endogenous Shh. X-gal-positive cells emerging from the node were only detected in the notochord lineage, suggesting that the floor plate and notochord arise from distinct precursors in the mouse node (Jeong, 2003).

The Shh-dependent pathway resulting in floor plate formation relies on triggering a transcription factor cascade culminating in the stable expression of Shh in the ventral midline of the neural tube. Shh signaling from the notochord activates Gli2, a zinc-finger transcriptional regulator, in the overlying neural plate. Gli2, which is required for floor plate development, is responsible for initiating the transcription of Foxa2 (formerly Hnf3b). Although, the misexpression of Foxa2 in the CNS can under certain conditions result in the ectopic activation of Shh, it remains unclear whether Foxa2 is required to regulate Shh transcription within sites of endogenous expression including the floor plate. Attempts at addressing this question through conventional loss-of-function studies is confounded by the requirement for Foxa2 in node formation, resulting in Foxa2-/- embryos that lack both the notochord and floor plate (Jeong, 2003 and references therein).

Given that vertebrate species show similar patterns of Shh expression in the CNS and that regulatory sequences directing floor plate expression have been localized to intron 2 in mouse, chicken and zebrafish, the premise that conservation of sequence underscores conservation of function was explored. Comparative sequence analysis of the 746-bp fragment of mouse DNA previously attributed with Shh floor plate enhancer 2 (Sfpe2) activity was undertaken with comparable regions from human, chicken and zebrafish using ClustalW algorithms. As expected, alignment of mouse and human sequence showed the highest degree of overall homology at 67%. Conservation of chicken and zebrafish sequences was found on average to be lower when compared to that of mouse, with homologies of 44% and 36%, respectively. On closer inspection however, the 4-way alignment revealed higher homology scores over short stretches of sequence compared to the overall average. Of these short stretches of sequence, three homologous regions corresponding to HR-a (nucleotide position: 60-139), HR-b (184-233) and HR-c (221-308) were remarkable given that all of the 2-by-2 comparisons between mouse and the individual species in question displayed higher homologies than the overall average for that species (Jeong, 2003).

Foxa2 is not sufficient to mediate Sfpe2 function; cooperative interactions with a homeodomain transcription factor are required to direct reporter expression to the floor plate in a Shh-like manner. These results are seemingly inconsistent with previous reports documenting that forced expression of Foxa2 is sufficient to activate Shh transcription. However, additional observations are supportive of this conclusion: Foxa2 is expressed along the length of the floor plate yet the activities of the enhancers regulating Shh are regionalized along the anteroposterior axis of the neural tube; moreover, the sequences mediating Sbe1 and Sfpe2 activity, although both possessing Foxa binding sites, cannot independently direct reporter expression to the floor plate even when multimerized. Therefore, additional transcription factors must be acting in concert with Foxa2 to regulate Shh expression in the floor plate. To reconcile differences between these results and the Foxa2 gain-of-function studies, it is speculated that: (1) Foxa2 may be inducing the expression of the cooperating transcription factor(s); (2) Foxa2 may only be capable of activating Shh transcription where the cooperating transcription factor(s) is/are expressed. Restrictions in where Foxa2 can activate Shh within the neural tube have been described; and (3) ectopic expression of Foxa2 may be activating Shh transcription through enhancers other than Sfpe2 (Jeong, 2003).

How might Foxa2 and the cooperating homeodomain factor interact to regulate Shh transcription? Previous studies have shown that the binding of Foxa proteins to their recognition sites on active enhancers can result in the stabilization of nucleosome position, thus facilitating the binding of additional transcription factors to the enhancer complex. Foxa2 may be functioning in a similar capacity on Sfpe2 by promoting the stable binding of homeodomain proteins such as Nkx6 family members (Jeong, 2003).

The observation that the Shh target genes Ptc and Foxa2 are expressed in the dorsal layer of the node offers further support that the process of floor plate induction begins in the mouse node at early somite stages and doesn't terminate until Shh transcription is activated in the ventral midline of the CNS between 8 to 12 somite stages. In this homeogenetic model of floor plate induction, Shh secreted from the axial mesoderm signals to the overlying neural plate to activate effectors of the Shh signal transduction cascade. A consequence of this vertical signaling step is the initiation of Shh transcription, through the direct binding of Foxa2 and a homeodomain protein to specific enhancer sequences. Given that sequences mediating Sbe1 activity are also required for floor plate expression, it is speculated that additional transcriptional activators are participating in the regulation of Shh expression. Identifying the critical sequences mediating Sbe1 activity and the factors binding to these sites should further elucidate how Shh expression is activated in the floor plate of the mouse spinal cord (Jeong, 2003).

Dynamics of enhancer-promoter communication during differentiation-induced gene activation: HNF3 regulation of HNF4-alpha

The order of recruitment of factors to the HNF-4alpha regulatory regions was followed upon the initial activation of the gene during enterocyte differentiation. An initially independent assembly of regulatory complexes at the proximal promoter and the upstream enhancer regions was followed by the tracking of the entire DNA-protein complex formed on the enhancer along the intervening DNA until it encountered the proximal promoter. This movement correlates with a unidirectional spreading of histone hyperacetylation. Transcription initiation coincides with the formation of a stable enhancer-promoter complex and remodeling of the nucleosome situated at the transcription start site. The results provide experimental evidence for the involvement of a dynamic process culminating in enhancer-promoter communication during long-distance gene activation (Hatzis, 2003).

At the very beginning of the differentiation program (time 0), both the enhancer and the promoter are already occupied by the cognate DNA binding proteins. At this stage at least three basal transcription factors, TFIIA, TFIIB, and TBP, are also detectable at the proximal promoter. This complex may be viewed as a signature structure that creates a so-called poised or committed state to mark the gene for subsequent events. The nucleosomal organization of the HNF-4 regulatory regions also indicates that the gene is in a transcriptionally competent state from the beginning of the program. Unlike in nonexpressing cell lines, where positioning of nucleosomes is random, in CaCo-2 cells the proximal promoter is occupied by an array of positioned nucleosomes, while at the enhancer area two positioned nucleosomes are followed by a nucleosome-free region. In addition, the H3 component of the nucleosomes at the proximal promoter is methylated at the lysine 4 residue, a modification that has been proposed to correspond to an epigenetic mark for active chromatin. Another interesting feature of this state is that the binding site of HNF-3, a so-called pioneer factor that can disrupt higher order chromatin structure, is located at the nucleosome-free region of the HNF-4alpha enhancer. Since HNF-3-mediated disruption of internucleosomal interactions can affect the length of linker DNA, the formation of a nucleosome-free region at the enhancer may well be the result of HNF-3 action at an earlier stage of differentiation. Consistent with the above is the fact that the CaCo-2 cell culture model mimics only terminal enterocyte differentiation, since the line originates from enterocytes that have already passed through early developmental decisions determining lineage commitment (Hatzis, 2003).

In the second temporally separable phase, a selective recruitment of histone acetyltransferases (CBP and P/CAF) and the Brg-1 chromatin remodeling protein to the enhancer (20 hr time point) was observed. This coincided with the first histone H3 and H4 hyperacetylation signals confined to the enhancer region. At the same time, other components of TFIID, TFIIH, the mediator component TRAP-220, and RNA pol-II were recruited to the proximal promoter. Since in the previous step TAF1 and TAF10 were absent from the promoter, it is speculated that the TBP detected at time 0 is not part of the classical TFIID complex. Whether the TFIID detected at the 20 hr time point is generated by a progressive assembly of TAFs onto promoter-bound TBP or by an exchange of a TAF-less or nonclassical TFIID by a TAF-containing TFIID is not known. The key characteristic of this early stage is that all the major components of the general transcription machinery as well as CTD serine 5-phosphorylated RNA pol-II are stably assembled at the promoter, without initiating transcription. This indicates that recruitment of pol-II to the transcription start site is not sufficient for transcription, and other factors or events are also required for its escape from the promoter. In this regard it is important to note that at this period there was no indication for potential enhancer-promoter synergy, suggesting that the assemblies of the complexes of different compositions at the two regions are independent of each other (Hatzis, 2003).

During the ensuing time period (40-80 hr), the immunoprecipitates of all enhancer-associated factors (HNF-1alpha, C/EBPalpha, HNF-3ß, CBP, P/CAF, and Brg-1) contained DNA fragments corresponding to the regions between the enhancer and the promoter. Since at the same time these factors are also found to be associated with the enhancer region, this finding points to the formation of a binary crosslinked DNA complex composed of the enhancer and intermediary region DNA, bridged by the factors associated with them. Importantly, none of the factors recruited to the proximal promoter are detected in the intervening regions at any time during the differentiation program. The potential scenario that multiple molecules of the enhancer-recruited proteins first associate with the enhancer and then, at least part of them, may escape the enhancer DNA and scan freely toward the promoter, while others remain associated with the enhancer, is rather unlikely considering the sequence-specific DNA binding properties of HNF-3ß and C/EBPalpha, which can associate selectively with the enhancer region. Furthermore, in the next step (80-110 hr) the immunoprecipitates of all of the promoter-recruited factors contain enhancer DNA fragments, and immunoprecipitates of all of the enhancer-binding factors contain promoter DNA segments. In other words, if the detection of these two distant DNA sequences in the immunoprecipitates of factors that are recruited to one or the other region were to be interpreted as the result of either independent recruitment or free diffusion from one region to the other, then one would have to assume that general transcription factors or RNA pol-II would suddenly be recruited to a far-upstream location at the time of transcription initiation, a possibility which is hard to conceptualize. The reduction and subsequent disappearance of the ChIP signals of the enhancer-associated factors from the intervening regions at the times of active transcription (80-110 hr) is also inconsistent with the continuous escape of DNA binding factors from the enhancer. The possibility of direct recruitment of RNA pol-II followed by a long-range transfer to the promoter or the activation of a cryptic promoter at the upstream region could also be ruled out, since recruitment of RNA pol-II together with other general transcription factors to the proximal promoter could be observed long before enhancer-promoter complex formation and since no transcript corresponding to upstream sequences could be detected at any time during the differentiation program. Therefore, the direct evidence provided by the continuous ChIP signal observed with the enhancer DNA, together with the above-mentioned considerations, corroborate the claim that the signals detected at the intervening regions correspond to a complex containing enhancer DNA. These observations thus indicate that the entire DNA-protein complex forms on the enhancer tracks along the intervening region toward the promoter, a process that is in agreement with a recently proposed 'facilitated tracking' hypothesis. This model assumes that the enhancer-bound complex tracks via small steps along the chromatin until it encounters the cognate promoter, at which stage a stable looped structure is formed. Important components of the tracking complex are the histone acetyltransferases CBP and P/CAF. These proteins may modify the chromatin as they move along the DNA. The unidirectional spreading of H3 and H4 hyperacetylation from the HNF-4alpha enhancer toward the promoter in a temporally identical manner to CBP and P/CAF occupancy demonstrates that this is indeed the case. Cooperation between histone acetyltransferases and ATP-dependent chromatin remodeling complexes has been proposed to be important for gene activation. The presence of Brg-1, a catalytic subunit of the human SWI/SNF complex, in the tracking complex provides important clues with respect to the dynamics of the process. The coordinated action of the acetylases and Brg-1 should lead to the acetylation of the histone tails of the neighboring nucleosome, which in turn would create a new interaction surface for the bromodomains of Brg-1, CBP, and P/CAF. This would facilitate the propagation of the complex to the next nucleosome, thus creating sequential signals for a stepwise process, powered by the ATP-ase activity of Brg-1. An important characteristic of the HNF-4alpha enhancer tracking is its unidirectional path. Although the results of this work do not provide an answer for the question of how this one-way course is controlled, it is speculated that sequences upstream of the HNF-4alpha enhancer may act as insulators that block the movement of the enhancer complex toward the opposite direction (Hatzis, 2003).

At the proximal promoter, ChIP signals of enhancer-recruited factors (HNF-1alpha, C/EBPalpha, HNF-3ß, CBP, P/CAF, and Brg-1) were first observed at 60 hr of the differentiation program, with an increased intensity at 80 and 110 hr. Concurrently, immunoprecipitates of proximal promoter-associated proteins (HNF-6, TFIIA, TFIIB, TBP, TAF1, TAF10, TFIIH, TRAP-220, and pol-II) contained DNA fragments corresponding to the enhancer region. In the experimental system employed, the simultaneous presence of the two DNA fragments in immunoprecipitates of this variety of factors demonstrates that the two regions come into close proximity to form a higher order complex by looping out the intervening DNA. This stable enhancer-promoter complex formation coincides with promoter hyperacetylation, phosphorylation of pol-II at the serine 2 position of its carboxy-terminal domain, remodeling of the nucleosome located at the transcription start site and finally with the release of pol-II from the promoter (Hatzis, 2003).

In summary, these results on HNF-4alpha enhancer-mediated activation demonstrate a dynamic mechanism, which accounts for many features of long-distance gene regulation that have been described for other genes. The ability to dissect the process to at least four temporally separable steps, all of which could be influenced by physiological signals, further emphasizes the complexity of the pathways that have evolved to regulate differential gene expression (Hatzis, 2003).

Mutation of Foxa1-3

The signals that induce vertebrate neural tissue and pattern it along the anterior-posterior (A-P) axis have been proposed to emanate from Spemann's organizer, which in mammals is a structure termed the node. However, mouse embryos mutant for HNF3 beta lack a morphological node and node derivatives yet undergo neural induction. Gene expression domains occur at their normal A-P axial positions along the mutant neural tubes in an apparently normal temporal manner, including the most anterior and posterior markers. This neural patterning occurs in the absence of expression of known organizer genes, including the neural inducers chordin and noggin. Other potential signaling centers in gastrulating mutant embryos appear to express their normal constellation of putative secreted factors, consistent with the possibility that neural-inducing and -patterning signals emanate from elsewhere or at an earlier time. Nevertheless, it has been found that the node and the anterior primitive streak, from which the node derives, are direct sources of neural-inducing signals, as judged by expression of the early midbrain marker Engrailed, in explant-recombination experiments. Similar experiments show the neural-inducing activity in HNF3 beta mutants to be diffusely distributed. These results indicate that the mammalian organizer is capable of neural induction and patterning of the neural plate, but that maintenance of an organizer-like signaling center is not necessary for either process (Klingensmith, 1999).

Thus the loss of the node and its derivatives does not lead to loss of the ability to form and pattern the neural tube along the A-P axis. Using an extensive series of markers, it has been shown that regionally restricted markers of the nervous system are expressed in the same order and with very similar boundaries of expression in HNF3beta mutant and wild-type embryos. The most anterior markers examined, such as Six3 and Fgf8, show more variability in extent of expression; anterior truncation of expression domains was apparent in a number of embryos. However, in all cases, some expression was observed, indicating that some degree of anterior patterning had occurred. Because HNF3beta mutant embryos die around E9.5, it is not possible to say how well the regionalization of the neural tube would be translated into the morphological structures of the brain. However, it is unlikely that development would be normal, because the embryos show severe defects in D-V patterning of the neural tube in the absence of the notochord. This leads to loss of ventral structures in the brain and spinal cord and would impact on the further morphogenesis of the nervous system throughout. Thus, the loss of the node and its derivatives impacts on later development of the nervous system, but does not seem to adversely affect its initial induction and A-P patterning (Klingensmith, 1999).

Both anterior visceral endoderm and node-associated signals are only transiently expressed and are poorly localized in HNF3beta mutant embryos, suggesting that one of the roles of the node is indeed as an organizer -- organizing and localizing the different sources of signals in the developing embryo. Explant-recombination experiments with mutant and wild-type streak tissue support this. The mature node or its precursor, the anterior of the elongating primitive streak, can induce En expression when combined with naive ectoderm, whereas more posterior primitive streak tissue does not. In contrast, both anterior and posterior streak tissue from HNF3beta mutant embryos has En-inducing capacity, albeit weak. Although the organized node fails to form in HNF3beta mutants, it appears that there is early disperse-inducing activity. This activity, in combination, perhaps, with weak anterior visceral endoderm signals, is sufficient to initiate neural axis development in HNF3beta mutant embryos. Once axis development begins, then other sources of signals for A-P patterning become established and can presumably explain the continued development of more posterior markers in HNF3beta mutant embryos. Members of the FGF and Wnt family as well as RA have been proposed to be important for posterior development. An examination of these factors in HNF3beta mutants reveals that all are still expressed. The activity of these factors is associated with primitive streak and paraxial mesoderm, rather than the node itself, and so the expression of these factors is not ablated in the mutants. However, the domains of expression of these factors are often reduced, suggesting again that the node helps to organize the signaling sources of the embryo, possibly by promoting the convergence-extension needed to elongate the vertebrate body axis and align the different tissues of the developing axis (Klingensmith, 1999).

HNF-3/forkhead homolog 4 (HFH-4), a transcription factor of the winged helix/forkhead family, is expressed in various tissues including lung, brain, oviduct, testis, and embryonic kidney. In order to test whether the temporospatial expression of HFH-4 influences lung morphogenesis, HFH-4 was expressed in lungs of transgenic mice under control of the surfactant protein C (SP-C) promoter. The morphology of the lungs from SP-C/HFH-4 embryos (day 18 postconception) was distinctly abnormal; the severity of the alterations correlates with the level of transgene expression as detected by in situ hybridization. At high levels of expression, HFH-4 alters epithelial cell differentiation and inhibits branching morphogenesis. Atypical cuboidal or columnar cells line the lung periphery of SP-C/HFH-4 transgenic mice. The atypical epithelial cells seen in the SP-C/HFH-4 mice express thyroid transcription factor-1 and hepatocyte nuclear factor 3beta (HNF-3beta). However, surfactant proteins SP-B, SP-C, and Clara cell secretory protein, normally produced by nonciliated epithelial cells in lung parenchyma are lacking. beta-Tubulin IV, a marker of ciliated cells, stains the atypical columnar cells produced by expression of high levels of the SP-C/HFH-4 transgene. Ectopic expression of HFH-4 in developing mouse lung alters epithelial cell differentiation and morphology, restricting the expression of markers typical of nonciliated cells of the distal lung parenchyma (Tichelaar, 1999).

Conditional gene ablation was used to uncover a dramatic role for the winged-helix transcription factor Foxa2 (formerly HNF-3 beta) in pancreatic beta-cell differentiation and metabolism. Mice that lack Foxa2 specifically in beta cells (Foxa2(loxP/loxP); Ins.Cre mice) are severely hypoglycemic and show dysregulated insulin secretion in response to both glucose and amino acids. This inappropriate hypersecretion of insulin in the face of profound hypoglycemia mimics pathophysiological and molecular aspects of familial hyperinsulinism. The two subunits of the beta-cell ATP-sensitive K(+) channel (K(ATP)), the most frequently mutated genes linked to familial hyperinsulinism, have been identified as novel Foxa2 targets in islets. The Foxa2(loxP/loxP); Ins.Cre mice will serve as a unique model to investigate the regulation of insulin secretion by the beta cell and suggest the human FOXA2 as a candidate gene for familial hyperinsulinism (Sund, 2001).

In mouse embryo, the early induction of the head region depends on signals from the anterior visceral endoderm (AVE) and the anterior primitive streak. Subsequently, node derivatives, including anterior definitive endoderm and axial mesendoderm, are thought to play a role in the maintenance and elaboration of anterior neural character. Foxa2 encodes a winged-helix transcription factor expressed in signaling centers required for head development, including the AVE, anterior primitive streak, anterior definitive endoderm, and axial mesendoderm. To address Foxa2 function during formation of the head, use was made of conditional mutants in which Foxa2 function is preserved in extraembryonic tissues during early embryonic stages and inactivated in embryonic tissues after the onset of gastrulation. In Foxa2 conditional mutants, the anterior neural plate and anterior definitive endoderm are initially specified. In contrast, the axial mesendoderm fails to differentiate. At later stages, specification of the anterior neural plate and anterior definitive endoderm is labile. As a result, head truncations are observed in Foxa2 conditional mutants. These results therefore indicate that anterior definitive endoderm alone is not sufficient to maintain anterior head specification and that an interaction between the axial mesendoderm and the anterior definitive endoderm is required for proper specification of the endoderm. Foxa2 therefore plays an integral role in the formation of axial mesendoderm, which is required to maintain the specification of the forebrain and the anterior definitive endoderm (Hallonet, 2002).

The specification of the vertebrate liver is thought to occur in a two-step process, beginning with the establishment of competence within the foregut endoderm for responding to organ-specific signals, followed by the induction of liver-specific genes. On the basis of expression and in vitro studies, it has been proposed that the Foxa transcription factors establish competence by opening compacted chromatin structures within liver-specific target genes. Foxa1 and Foxa2 (forkhead box proteins A1 and A2) are required in concert for hepatic specification in mouse. In embryos deficient for both genes in the foregut endoderm, no liver bud is evident and expression of the hepatoblast marker alpha-fetoprotein (Afp) is lost. Furthermore, Foxa1/Foxa2-deficient endoderm cultured in the presence of exogenous fibroblast growth factor 2 (FGF2) fails to initiate expression of the liver markers albumin and transthyretin. Thus, Foxa1 and Foxa2 are required for the establishment of competence within the foregut endoderm and the onset of hepatogenesis (Lee, 2005).

The role of transcription factors in regulating the development of midbrain dopaminergic (mDA) neurons is intensively studied owing to the involvement of these neurons in diverse neurological disorders. This study demonstrates novel roles for the forkhead/winged helix transcription factors Foxa1 and Foxa2 in the specification and differentiation of mDA neurons by analysing the phenotype of Foxa1 and Foxa2 single- and double-mutant mouse embryos. During specification, Foxa1 and Foxa2 regulate the extent of neurogenesis in mDA progenitors by positively regulating Ngn2 (Neurog2) expression. Subsequently, Foxa1 and Foxa2 regulate the expression of Nurr1 (Nr4a2) and engrailed 1 in immature neurons and the expression of aromatic l-amino acid decarboxylase and tyrosine hydroxylase in mature neurons during early and late differentiation of midbrain dopaminergic neurons. Interestingly, genetic evidence indicates that these functions require different gene dosages of Foxa1 and Foxa2. Altogether, these results demonstrate that Foxa1 and Foxa2 regulate multiple phases of midbrain dopaminergic neuron development in a dosage-dependent manner (Ferri, 2007).

In many regions of the developing CNS, distinct cell types are born at different times. The means by which discrete and stereotyped temporal switches in cellular identities are acquired remains poorly understood. This study has examined how visceral motor neurons (VMNs) and serotonergic neurons, two neuronal subtypes, are sequentially generated from a common progenitor pool in the vertebrate hindbrain. The forkhead transcription factor Foxa2, acting in progenitors, is essential for the transition from VMN to serotonergic neurogenesis. Loss-of-function and gain-of-function experiments indicated that Foxa2 activates the switch through a temporal cross-repressive interaction with paired-like homeobox 2b (Phox2b), the VMN progenitor determinant. This mechanism bears a marked resemblance to the cross-repression between neighboring domains of transcription factors that establish discrete progenitor identities along the spatial axes. Moreover, the subsequent differentiation of central serotonergic neurons required both the suppression of VMN neurogenesis and the induction of downstream intrinsic determinants of serotonergic identity by Foxa2 (Jacob, 2007).

FOXA1, estrogen receptor alpha (ERalpha) and GATA3 independently predict favorable outcome in breast cancer patients, and their expression correlates with a differentiated, luminal tumor subtype. As transcription factors, each functions in the morphogenesis of various organs, with ERalpha and GATA3 being established regulators of mammary gland development. Interdependency between these three factors in breast cancer and normal mammary development has been suggested, but the specific role for FOXA1 is not known. This study reports that Foxa1 deficiency causes a defect in hormone-induced mammary ductal invasion associated with a loss of terminal end bud formation and ERalpha expression. By contrast, Foxa1 null glands maintain GATA3 expression. Unlike ERalpha and GATA3 deficiency, Foxa1 null glands form milk-producing alveoli, indicating that the defect is restricted to expansion of the ductal epithelium, further emphasizing the novel role for FOXA1 in mammary morphogenesis. Using breast cancer cell lines, it was also demonstrated that FOXA1 regulates ERalpha expression, but not GATA3. These data reveal that FOXA1 is necessary for hormonal responsiveness in the developing mammary gland and ERalpha-positive breast cancers, at least in part, through its control of ERalpha expression (Bernardo, 2010).

Fox factors and floor plate development

The secreted ligand Sonic Hedgehog (Shh) organizes the pattern of cellular differentiation in the ventral neural tube. For the five neuronal subtypes, increasing levels and durations of Shh signaling direct progenitors to progressively more ventral identities. This study demonstrates that this mode of action is not applicable to the generation of the most ventral cell type, the nonneuronal floor plate (FP). In chick and mouse embryos, FP specification involves a biphasic response to Shh signaling that controls the dynamic expression of key transcription factors. During gastrulation and early somitogenesis, FP induction depends on high levels of Shh signaling. Subsequently, however, prospective FP cells become refractory to Shh signaling, and this is a prerequisite for the elaboration of their identity. This prompts a revision to the model of graded Shh signaling in the neural tube, and provides insight into how the dynamics of morphogen signaling are deployed to extend the patterning capacity of a single ligand. In addition, evidence is provided supporting a common scheme for FP specification by Shh signaling that reconciles mechanisms of FP development in teleosts and amniotes (Ribes, 2010).

Following initial induction of FP specification, Shh signaling is attenuated in presumptive FP cells. Maintaining Shh signaling at this time converts FP cells to ventral neural progenitors, demonstrating that the down-regulation of signaling is a prerequisite for the elaboration of FP identity. Thus, the specification of the FP and other ventral neuronal progenitors depends on distinct timing and duration of Shh signaling. Consistent with this, it was demonstrated that presumptive FP cells display a dynamic transcriptional code that distinguishes FP precursors from ventral neural progenitors. Evidence is provided that FoxA2 is required for FP specification and the inhibition of p3 fate. Taken together, the data indicate that, in all vertebrates, FP induction takes place in a brief time window during the course of gastrulation, and the extrinsic signals involved in this process regulate FoxA2 expression. The difference between species resides mainly in the relative contribution of each signal. It is therefore tempting to hypothesize that both Shh and Nodal signals were involved in FP specification in the common ancestor of vertebrates. Subsequently, the relative importance of each signal changed during the evolution of individual species. Detailed analysis of the regulatory elements directing expression of FoxA2 in different species should shed further light on this hypothesis (Ribes, 2010).

Foxa1-3 and chromatin

The transcription factor HNF3 and linker histones H1 and H5 possess winged-helix DNA-binding domains, yet HNF3 and other forkhead-related proteins activate genes during development whereas linker histones compact DNA in chromatin and repress gene expression. A comparison has been made of how the two classes of factors interact with chromatin templates; it has been found that HNF3 binds DNA at the side of nucleosome cores, similarly to linker histone. A nucleosome structural binding site for HNF3 is occupied at the albumin transcriptional enhancer in active and potentially active chromatin, but not in inactive chromatin in vivo. While wild-type HNF3 protein does not compact DNA extending from the nucleosome, as does linker histone, site-directed mutants of HNF3 can compact nucleosomal DNA if they contain basic amino acids at positions previously shown to be essential for nucleosomal DNA compaction by linker histones. The results illustrate how transcription factors can possess special nucleosome-binding activities that are not predicted from studies of factor interactions with free DNA (Cirillo, 1998).

The transcription factors HNF3 (FoxA) and GATA-4 are the earliest known to bind the albumin gene enhancer in liver precursor cells in embryos. To understand how they access sites in silent chromatin, nucleosome arrays containing albumin enhancer sequences were assembled and they were compacted with linker histone. HNF3 and GATA-4 but not NF-1, C/EBP, and GAL4-AH, bind their sites in compacted chromatin and opened the local nucleosomal domain in the absence of ATP-dependent enzymes. The ability of HNF3 to open chromatin is mediated by a high affinity DNA binding site and by the C-terminal domain of the protein, which binds histones H3 and H4. Thus, factors that potentiate transcription in development are inherently capable of initiating chromatin opening events (Cirillo, 2002).

How might HNF3 access its binding sites in vivo? Under physiological salt conditions, nucleosome arrays exist in a dynamic equilibrium between folded and compacted states. Additionally, linker histone is rapidly exhanged on chromatin in living cells. HNF3 could exploit both of these properties to initially bind its sites in compacted chromatin. Previously published data indicate that the essential HNF3 binding sites eG and eH flank the dyad axis of the nucleosome particle to which HNF3 binds. This would place HNF3 on the bound particle in the vicinity of where X-ray crystallography data places histones H3 and H4 in the nucleosome. Histones H3 and H4 make internucleosomal contacts which have been implicated in the formation of nucleosomal arrays, and nucleosome arrays lacking the H3/H4 amino-terminal tails fail to fold into a fully compacted state or undergo mitotic chromosome condensation. It is suggested that HNF3 disrupts internucleosomal interactions promoted by H3/H4 tetramers, thereby decompacting the array locally and making it accessible to other proteins. In addition, HNF3 binding helps stabilize the position of an underlying nucleosome, which could affect the length of linker DNA on either side of the N1 particle. Small variations in linker length can have dramatic effects on the compaction of nucleosomes in chromatin. It is therefore suggested that HNF3 disrupts local chromatin structure by a combination of core histone interactions and by inducing changes in the position or orientation of nearby linker regions (Cirillo, 2002).

Foxa2 acts as a co-activator potentiating expression of the Nurr1-induced DA phenotype via epigenetic regulation

Understanding how dopamine (DA) phenotypes are acquired in midbrain DA (mDA) neuron development is important for bioassays and cell replacement therapy for mDA neuron-associated disorders. This study demonstrate a feed-forward mechanism of mDA neuron development involving Nurr1 (Drosophila homolog Hormone receptor-like in 38) and Foxa2. Nurr1 acts as a transcription factor for DA phenotype gene expression. However, Nurr1-mediated DA gene expression was inactivated by forming a protein complex with CoREST, and then recruiting histone deacetylase 1 (Hdac1; Drosophila homolog, Rpd3), an enzyme catalyzing histone deacetylation, to DA gene promoters. Co-expression of Nurr1 and Foxa2 was established in mDA neuron precursor cells by a positive cross-regulatory loop. In the presence of Foxa2, the Nurr1-CoREST interaction was diminished (by competitive formation of the Nurr1-Foxa2 activator complex), and CoREST-Hdac1 proteins were less enriched in DA gene promoters. Consequently, histone 3 acetylation (H3Ac), which is responsible for open chromatin structures, was strikingly increased at DA phenotype gene promoters. These data establish the interplay of Nurr1 and Foxa2 as the crucial determinant for DA phenotype acquisition during mDA neuron development (Yi 2014).

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.