The forkhead gene family, named after the founding gene member in Drosophila, is characterized by a unique DNA-binding domain. This so-called forkhead box encodes a winged-helix DNA-binding motif, the name of which describes the structure of the domain when bound to DNA. The three Fox (forkhead box) group A genes, Foxa1, Foxa2 and Foxa3, are expressed in embryonic endoderm, the germ layer that gives rise to the digestive system, and contribute to the specification of the pancreas and the regulation of glucose homoeostasis. Deletion of the Foxa2 gene in pancreatic β-cells in mice results in a phenotype resembling PHHI (persistent hyperinsulinaemic hypoglycaemia of infancy). Molecular analyses have demonstrated that Foxa2 is an important regulator of the genes encoding Sur1, Kir6.2 and Schad (short chain L-3-hydroxyacyl-CoA dehydrogenase), mutation of which causes PHHI in humans. Foxa1 was shown to be an essential activator of glucagon gene expression in vivo. An additional winged-helix protein, Foxo1, contributes to pancreatic β-cell function by regulating the Pdx1 gene, which is required for pancreatic development in cooperation with Foxa2.
HISTORY OF THE
FORKHEAD GENE FAMILY
The Drosophila forkhead (fkh) gene was identified in a screen for embryonic-lethal mutations nearly 20 years ago, and was later cloned by chromosome walking [1,2]. Expression analysis in Drosophila revealed not only a strict nuclear localization of the fkh protein, but also suggested a function for the gene in embryonic gut development. In the absence of forkhead, head involution is blocked and the terminal domains that normally give rise to the anterior and posterior regions of the gut undergo homoeotic transformations, resulting in the spiked head structures (or ‘forks’) that gave rise to the gene name [1,2]. More detailed phenotype analyses of fkh mutant flies revealed important additional roles for the gene in the developing midgut and salivary glands [1,2]. At the time of discovery, the predicted protein displayed no previously characterized motifs or structural characteristics. Without significant amino acid similarity to any known proteins, forkhead became the founding member of a novel class of nuclear regulatory proteins.
In unrelated studies of regulation of gene expression in the mammalian liver, a new protein, HNF-3A (hepatocyte nuclear factor 3A), was identified based on its ability to bind specific DNA sequences in the promoters of the transthyretin and α1-antitrypsin genes [3,4]. Mutation of these HNF-3 sites resulted in decreased expression of promoter/reporter constructs of both of these important liver genes . Shortly after these discoveries, sequence comparisons revealed a striking similarity between Drosophila forkhead and rat HNF-3A in a central domain, with 129 out of 176 amino acids being identical between the two proteins . Accounting for conservative amino acid substitutions, a 110-amino acid stretch not only attained 92% similarity, but also, remarkably, had DNA-binding capabilities in both species. This conserved region of amino acids was termed ‘forkhead domain’, after the founding member cloned in Drosophila .
Species ranging from yeasts to humans produce DNA-binding proteins containing forkhead domains [6–9]. Using one of these proteins from rat (HNF-3γ), the crystal structure of the forkhead domain bound to a 13 bp piece of DNA was resolved . The forkhead domain binds DNA as a monomer in a variation of the helix–turn–helix motif (HTH, reviewed in ), consisting of three α-helices (H1–3), three β-strands (S1–3), a hydrophobic core and two loops (W1 and W2) which interact with the phosphate backbone of the target DNA molecule . In picturesque terms, the crystal structure of the forkhead domain resembles a butterfly, with a thorax of α-helices and two wing-like loops, and has come to be known as the ‘winged-helix’ motif. Thus the forkhead domain encodes a winged-helix DNA-binding motif.
NEW GENES AND NEW NOMENCLATURE
The first indication of the existence of a family of winged-helix proteins came from Northern blot analysis of rat liver RNA using HNF-3A as the probe. In addition to a 2 kb band corresponding to the HNF-3A cDNA, two larger size bands were discovered . Shortly thereafter, these were shown to correspond to two additional gene products, and the whole group was renamed HNF-3α, HNF-3β and HNF-3γ with molecular masses of 50, 47, and 42 kDa respectively . Although all three proteins share a similar DNA-binding domain, the differences in coding regions, variable affinities for HNF-3 binding sites  and eventual mapping of the mouse and human homologues to different chromosomes [13–16] confirmed that they were, in fact, the products of three separate genes. Comparison of the Drosophila forkhead sequence to each of the HNF-3 members showed the greatest similarity to HNF-3β, where 100 of the 110 amino acids that compose the DNA-binding domain were identical . Since Drosophila forkhead and mouse HNF-3β also share specific early expression patterns, it has been suggested that HNF-3β is the true murine forkhead orthologue .
By 1998, over 100 members of the forkhead gene family had been identified in species ranging from yeasts to humans (reviewed in [18,19]). Although all of these genes share a common DNA-binding domain, the naming and classification systems used by different laboratories studying a variety of organisms became difficult to decipher. Although some genes were referred to as HNF-3/fkh homologues (HFH; ) or fkh-related-domain sequences (FD; [14,21]), other gene names made no reference to the original members of the gene family. In chordates, a new and unified naming system was proposed and adopted by the mouse and human nomenclature committees whereby these genes would be recognized by the symbol Fox (for Forkhead box) followed by a letter to distinguish subclass (of which there are 15) and a number to identify individual proteins . All capital letters pertain to human genes (e.g. FOXC2) and only the first letter is capitalized in mice (e.g. Foxc2); all other chordates have a capitalized first letter and subclass letter (e.g. FoxC2). Following this new nomenclature, the mouse HNF-3α, HNF-3β, and HNF-3γ genes correspond to Foxa1, Foxa2 and Foxa3 respectively, and will be referred to as such for the remainder of this review.
THE FOXA PROTEINS DEMARCATE THE MAMMALIAN GUT
Of the three Foxa genes, Foxa2 can be detected earliest during gastrulation in a small patch of epiblast cells in the anterior portion of the prestreak embryo in both the ectoderm and endoderm [14,17,23]. Embryonic day 6.5 (E6.5) marks the beginning of mesoderm formation which gives rise to the node. Foxa2 is expressed there until approximately E8.0 when a discrete nodal structure is no longer discernable [14,17,23]. Expression is seen in mesoderm and endoderm cells which migrate anteriorly from the node. At the headfold stage (E7.5), Foxa2 is present in the neural plate, notochord and in the foregut endoderm. By E9.5, Foxa2 is strongly expressed in the liver primordium and, at E16.5, endoderm-derived tissues, most notably the pancreas, maintain high expression [14,17,23]. The continuous expression of Foxa2 from early embryonic stages through adulthood suggests a crucial role in both the specification and maintenance of endoderm-derived tissues, including liver, lung and pancreas.
Unlike Foxa2, Foxa1 is not detectable in the early primitive streak stage of development [17,23]. Rather, its expression initiates in the late primitive streak embryo near the anterior part of the primitive streak. Subsequently, intense Foxa1 expression is seen in the invaginating foregut while weak expression is detectable in the notochord and in cartilaginous cells that will eventually differentiate to form the vertebral column . At the early headfold stage, Foxa1 is restricted to the definitive endoderm and the notochord until its integration into the spinal cord . By E9.0, the entire gut region including the liver primordium and the floorplate of the neural tube express Foxa1. The expression of Foxa1 is maintained in organs of endoderm lineage in the adult [12,15].
Foxa3 appears to act later in mouse development than either of the other Foxa genes. RNA in situ hybridization experiments demonstrated weak expression of Foxa3 in the extraembryonic endoderm before E8.0  and stronger expression in cells of the invaginating hindgut near E9.0. By E10.5, Foxa3 expression is weak in early liver cells, but continually increases throughout liver development  with peak expression at E15.5 . In addition to hindgut and liver, Foxa3 is also found in early bone tissue. After E13.0, expression is seen in developing vertebrae, forelimb and hindlimb bones, ribs and bones that compose the skull and jaw, but all Foxa3 expression in bone tissue ceases by E16.0 . Despite the fact that Foxa3 is expressed later and to a lesser extent in the developing embryo, this gene plays a crucial role in maintenance of glucose homoeostasis in the adult , as will be discussed later in this review.
GENE ABLATION STUDIES OF THE
FOXA GENES: FROM EXPRESSION PATTERN TO PHENOTYPE
Although the initial studies documenting embryonic expression patterns of the Foxa genes provided valuable insight into potential roles in development, the in vivo functions were defined by targeted gene deletion. Embryonic stem cells lacking Foxa1 were generated and differentiated to form EBs (embryoid bodies) . Gene expression analysis from these EBs revealed several up-regulated genes, including the genes encoding GLUT2 and L-pyruvate kinase, which, at the time, suggested that Foxa1 was a repressor of glycolytic enzymes .
The first proof that Foxa genes function in the endocrine pancreas in vivo came from gene ablation studies of Foxa1, expression of which is enriched in islets . In 1999, two groups derived Foxa1-deficient mice through homologous recombination in embryonic stem cells [27,28]. Foxa1−/− mice were born in the expected Mendelian ratios and were of comparable size with control littermates. However, mutant pups were severely growth-retarded soon after birth and died anywhere from 2 to 14 days later [27,28]. Although Foxa1 may play a role during embryogenesis as expression studies suggest, gene ablation studies indicate that its more crucial function lies in postnatal life.
Physiological characterization revealed that blood glucose levels in Foxa1-null mice are significantly lower than those of controls [27,28]. Consistent with the severe hypoglycaemia, Foxa1−/− mice also have low plasma insulin levels and high corticosteroid levels, but demonstrate inappropriately low levels of circulating plasma glucagon [27,28]. Although low glucagon can be the result of a decrease in α-cell number, immunostaining suggested that cell lineage allocation of the mutant pancreatic islets was normal [27,28]. Northern blot analysis showed a 70% reduction in preproglucagon mRNA levels and a direct regulatory relationship between Foxa1 and glucagon was established by transfection experiments with a glucagon promoter/luciferase reporter and a Foxa1 expression plasmid. These experiments demonstrated that Foxa1 is a potent activator of the glucagon gene promoter , and proved that this forkhead family gene acts as an essential transcriptional regulator of glucose homoeostasis (Figure 1). Interestingly, the notion that Foxa1 is a repressor of glycolytic genes, as suggested by the studies in EBs above, was not confirmed in vivo, as the mRNA levels for all these genes were unchanged in Foxa1-deficient liver or yolk sac .
Foxa1 and Foxa2 are expressed and/or required at multiple stages of pancreatic differentiation
To examine the role of Foxa3, mice homozygous for a Foxa3-null allele lacking the entire DNA-binding domain were generated through homologous recombination . Foxa3-null mice survived to birth and were born in the expected Mendelian ratios. Growth measurements demonstrated no differences in weight gain compared with littermate controls from birth through adulthood and both male and female mice were found to be fertile . Physiological evaluation revealed no deviations from normal fed blood glucose, triacylglycerols (triglycerides), amino acids or cholesterol levels . However, RNase protection assays revealed an up-regulation of both Foxa1 and Foxa2 in the Foxa3-null livers, suggesting the existence of regulatory networks among gene family members .
Further study of this mouse model revealed a requirement of Foxa3 in the maintenance of proper glucose homoeostasis during periods of prolonged fasting . Although insulin and glucagon secretion in the mutants was comparable with controls, the mRNA level of hepatic Glut2, the glucose transporter on the cell membrane, was decreased by over 60% . Since glucose transport across the cell membrane of hepatocytes occurs, in part, due to facilitated diffusion via Glut2, deficiency of this protein hinders the efflux of newly synthesized glucose from the liver and probably contributes to the fasting hypoglycaemia in Foxa3-null mice.
Foxa2 is not expressed in wild-type embryonic stem cells, but low levels are detected after differentiation of EBs , suggesting a possible role in very early development. A homozygous null Foxa2 mutation engineered through homologous recombination demonstrated the absolute requirement of this gene in early murine development, as its absence led to embryonic lethality by E10–11 [31,32]. Foxa2−/− embryos were smaller than either heterozygote or wild-type littermates, had disorganized somites, failed to form a distinct node and never developed a notochord [31,32]. Since there was no initiation of notochord development, Foxa2 mutant embryos also lacked motor neurons and floorplate cells, the appearance of which are normally induced by the notochord itself . Although initial characterization of Foxa2−/− embryos showed defects of neural tissue, further analysis revealed an additional requirement for this gene in the successful development of non-neural tissue as well.
Pancreatic development in the mouse initiates at approximately E9.5 with the formation of a dorsal pancreatic bulge . Even though Foxa2−/− embryos do not survive long enough to study the role of Foxa2 in pancreatic development, there is evidence that this gene is crucial in early gut formation. Of the three Foxa genes, Foxa2 is detected first in the developing mouse embryo, but Foxa1 precedes Foxa2 in the earliest endodermal cells . Using Foxa1 as a marker, a sheet of endodermal cells could be found on the ventral surface of Foxa2-null embryos, but these cells did not invaginate to form the gut tube . As Foxa2 was required for the formation of the foregut, the next step, that is the differentiation of the foregut-derived organs, including thyroid, lung, liver and pancreas, could not be evaluated in Foxa2-null embryos.
β-CELL-SPECIFIC DELETION ELUCIDATES THE ROLE OF
FOXA2 IN PANCREATIC DEVELOPMENT AND FUNCTION
In order to dissect the function of Foxa2 specifically in the pancreas, independent of its requirement in the developing endoderm, tissue-specific gene ablation was employed. Using loxP sites inserted around exon 3 of the Foxa2 gene by homologous recombination  and a Cre transgene under the control of the rat insulin II promoter (Ins.Cre; [37,38]), mice with a β-cell-specific deletion of Foxa2 were derived.
Foxa2loxP/loxP;Ins.Cre mice survived to birth and were grossly indistinguishable from wild-type littermates on postnatal day 1 (P1), but demonstrated an array of pancreatic abnormalities a few days later . Wild-type mouse islets are characterized by their spherical shape with a thin outer layer of glucagon-secreting α-cells, an inner core of insulin-secreting β-cells, as well as pancreatic polypeptide cells, somatostatin-secreting δ-cells and the recently identified ghrelin-secreting ε-cells scattered throughout [34,39,40]. Foxa2loxP/loxP;Ins.Cre mice specifically lacked Foxa2 protein in approx. 85% of the β-cells and exhibited highly disorganized islets . Although all cell lineages were present in appropriate proportions, many of the islets were amorphous in shape and α-cells were often found in the inner mass of β-cells in a random fashion , indicating a possible role for Foxa2 in islet formation. Physiological examination revealed additional roles for Foxa2 in glucose metabolism. With fed glucose levels barely reaching 25% of littermate controls, Foxa2loxP/loxP;Ins.Cre mice were severely hypoglycaemic and usually died between P9 and P12 . Although plasma insulin levels were comparable with controls, they were inappropriately high given the low blood glucose of Foxa2 mutant mice.
Reverse-transcription PCR analysis using islets isolated from Foxa2loxP/loxP;Ins.Cre mice revealed dependence of several components of the glucose-stimulated insulin secretion pathway on Foxa2. Although several genes, including those encoding Glut2, glucokinase and glutamate dehydrogenase, were not differentially expressed between controls and mutants, genes encoding Sur1 and Kir6.2, the two subunits of the KATP channel, were down-regulated in Foxa2loxP/loxP;Ins.Cre islets by approx. 75% . Genes encoding Sur1 and Kir6.2 are expressed in both α- and β-cells of the pancreas and play a role in the secretion of insulin and glucagon [41–43]. RNA in situ hybridization revealed that the remaining Sur1 and Kir6.2 mRNAs were restricted to the α-cells and that these transcripts were undetectable in the Foxa2-deficient β-cells . Foxa2 regulates the genes encoding Sur1 and Kir6.2 directly, as demonstrated by cotransfections of reporter plasmids with a Foxa2 expression construct .
Because expression of both KATP channel subunits is dependent upon the presence of Foxa2, mice with a β-cell-specific deletion of this transcription factor suffered severe physiological consequences. Perifusion studies using isolated islets revealed defects in insulin secretion in response to both glucose and glyburide, a KATP channel inhibitor, confirming the absence of functional KATP channels in β-cells . Surprisingly, the phenotype of Foxa2loxP/loxP;Ins.Cre mice was much more severe than that of either Sur1−/− or Kir6.2−/− mice [45–47], suggesting the existence of additional Foxa2 targets. Expression profiling via microarray analysis revealed that hadhsc, the gene encoding Schad (short chain L-3-hydroxyacyl-CoA dehydrogenase), is down-regulated in mutant islets by over 3-fold . Schad is a mitochondrial matrix protein that plays a role in fatty acid oxidation, and Schad deficiency results in accumulation of short chain acyl-CoA esters. Such short chain acyl-CoA esters inhibit the outer mitochondrial enzyme CPT1 (carnitine palmitoyl transferase-1) and lead to conversion of long-chain fatty acyl-CoAs into triacylglycerols, diacylglycerol, fatty acids and acylated proteins, all of which have the ability to enhance insulin secretion by means independent of the KATP channel [48,49]. Similar to Foxa2loxP/loxP;Ins.Cre mice, patients with SCHAD deficiencies are hypoglycaemic and hyperinsulinaemic [48,50,51], suggesting not only the importance of fatty acid oxidation in the β-cells for proper glucose homoeostasis, but also a potential regulatory role for Foxa2 in this process.
A ROLE FOR
FOXA2 IN THE DEVELOPMENT OF TYPE II DIABETES IN HUMANS?
The importance of Foxa2 in pancreatic development combined with its function in regulating pancreatic gene transcription discussed above prompted several groups to search for mutations and polymorphisms in the FOXA2 gene in patients with Type II diabetes. A study of 96 diabetic subjects revealed four nucleotide changes within the FOXA2 coding region, two of which were silent and two of which generated missense mutations . However, the frequency of these mutations was very low and the mutant forms were biochemically as active as wild-type FOXA2 proteins . In the same study, mutations in FOXA1 and FOXA3 were also identified, but subsequent in vitro analysis suggested that none of these variations contributed to the risk of Type II diabetes . Given the small size of the population studied, important roles for the FOXA genes in different ethnic populations or in other human metabolic defects remain a possibility.
Another group hypothesized that FOXA2 mutations contribute to MODY (maturity-onset diabetes of the young), a form of Type II diabetes that is inherited in an autosomal-dominant manner before the age of 25 . Each of the six currently recognized forms of MODY is due to mutations in one particular gene [genes encoding HNF-4α, glucokinase, HNF-1α, IPF1 (insulin promoter factor 1), HNF-1β and NeuroD1 respectively]. Although mutations in these genes account for over 80% of MODY cases in the Caucasian population , this is not true for MODY patients of Japanese descent where 80–85% of MODY patients do not have mutations in any of the aforementioned genes . Although eight nucleotide alterations in FOXA2 were identified among the cohort of 68 subjects, only one was a missense mutation not found in any control subjects . One additional study investigated a particular region in the first intron of FOXA2 which contains a trinucleotide repeat polymorphism (TCC)n. Although a total of ten different allele sizes were identified in a group of 112 Japanese Type II diabetics (onset after age 35) and 96 control subjects, the distribution of these alleles was similar between the groups. It was concluded that this FOXA2 polymorphism does not contribute to Type II diabetes with late onset in this population . Although mutations to FOXA2 may, in fact, contribute to diabetes, a common role for FOXA2 in MODY is unlikely.
FOXA2 IS REQUIRED FOR THE DIFFERENTIATION OF MATURE α-CELLS
As described above, the function of Foxa2 in pancreatic β-cells at late gestation has been investigated in great detail, but little is known about the role of Foxa2 in the foregut endoderm at the onset of pancreas development. The Foxa2loxP/loxP;Ins.Cre mice described above are not informative in this regard, as deletion of Foxa2 occurs only in β-cells after E14.5 . To address the role of Foxa2 in the endoderm during early embryogenesis, a new Cre mouse, Foxa3Cre, was derived which recombines floxed DNA sequences at E8.5 when endogenous Foxa3 expression is first detected [17,57]. Utilizing the Foxa3Cre transgene, Foxa2 was deleted in the entire endoderm of the developing gut, including the pancreatic primordium. Foxa2loxP/loxP;Foxa3Cre mice survived to birth, but most died by P3 with a few surviving to P5. Although the mutant pups were born in the appropriate Mendelian ratios, they were smaller in size and appeared dehydrated, despite the fact that milk was present in their stomachs. Additionally, blood glucose levels of Foxa2loxP/loxP;Foxa3Cre mice were dramatically lower than those of control littermates, and this severe hypoglycaemia probably contributed to the early lethality of the mutants.
As a response to hypoglycaemia, α-cells are normally stimulated to secrete glucagon, which signals the release of hepatic glycogen stores to restore blood glucose levels. However, plasma glucagon levels were markedly decreased in Foxa2loxP/loxP;Foxa3Cre mice, indicating a defect in either the production or secretion of this hormone. Further analysis revealed a decrease in both glucagon protein level in the islet, as shown by immunofluorescence, and also a more than 90% reduction in preproglucagon transcript levels in total pancreas .
In the developing pancreas, pancreatic precursor cells are first specified to become endocrine progenitor cells, and then are regulated by the combinatorial effects of transcription factors to differentiate into α-, β-, δ-, pancreatic polypeptide or ε-cells. These cells are deemed ‘mature’ once they initiate secretion of their respective hormones [40,58]. Using PC2 (prohormone convertase 2) as an early marker of the endocrine lineage in combination with antibodies against insulin, glucagon and somatostatin, it was determined that Foxa2 was not required for the initial specification of the α-cell lineage. Instead, the terminal differentiation step and maturation of glucagon-producing α-cells was dependent on Foxa2 (see Figure 1). Thus analysis of Foxa2loxP/loxP;Foxa3Cre mice revealed a novel requirement for Foxa2 in the final steps of α-cell differentiation.
FOXA2 AND PDX1
Pdx1 (pancreatic–duodenal homoeobox 1; also known as IDX1 or STF1 in humans) is a transcription factor expressed mainly in β-cells of adult islets and plays a central role in the developing human and mouse pancreas [59,60]. In the mouse, Pdx1 expression initiates upon commitment of the foregut endoderm to a pancreatic fate; this commitment precedes hormone gene expression . Pdx1 is crucial for pancreatic formation, as mice homozygous for a null mutation in this gene lack a pancreas. Although Pdx1−/− pups survive to birth, they die just a few days into postnatal life . In humans, only a few cases of pancreatic agenesis have been reported [62–66]. Although most of these cases are not correlated with a particular genetic defect, one of the patients described is homozygous for a single nucleotide deletion within the human PDX1 gene, which causes a frameshift in the reading frame and the deletion of the homoeodomain and is therefore probably the main cause for the lack of pancreatic development in this patient .
Investigation into the β-cell-specific expression of Pdx1 in adult islets revealed a regulatory relationship between this transcription factor and insulin gene expression . The role of Pdx1 in mature β-cells was investigated using cell-type-specific gene ablation. The conditional Pdx1 deletion in these mice eventually led to overt diabetes, manifested in the form of increased blood and urine glucose levels . Since Pdx1 has been proposed to positively regulate Glut2  as well as insulin expression , the deficiency of Pdx1 activity probably contributes to the development of diabetes in these animals.
Analysis of the promoter and enhancers that control Pdx1 expression revealed the presence of a conserved binding site for Foxa proteins [69,70], suggesting that Foxa genes might be upstream regulators of Pdx1. Because of the embryonic lethality of Foxa2-null mice, it was not possible to assess the in vivo regulation of Pdx1 by Foxa2 with this model [31,32]. To overcome this limitation, Foxa2loxP/loxP;Ins.Cre mice, which have a β-cell-specific deletion of Foxa2, were analysed [38,71]. In isolated islets from P8 mice, a 68% down-regulation of Pdx1 transcript was observed by quantitative real-time PCR in the mutants compared with control littermates . Western blot analysis using islets isolated from control and Foxa2loxP/loxP;Ins.Cre mice demonstrated a 69% down-regulation of Pdx1 at the protein level . These results provided the first in vivo evidence that Foxa2 lies upstream of Pdx1 in the differentiated β-cell (Figure 1). The partial loss of Pdx1 expression in the absence of Foxa2 raises the possibility that Foxa1 and/or Foxa3 might also contribute to the regulation of Pdx1, a notion that remains to be investigated.
FOXO1: ANOTHER CRUCIAL TRANSCRIPTION FACTOR IN THE PANCREAS LINKING INSULIN SIGNALLING TO β-CELL FUNCTION
Insulin signalling is proposed to feedback on the β-cell and control β-cell proliferation and, concurrently, insulin secretion . Although the precise mechanism(s) of this regulation is still under investigation, mice with a homozygous null mutation in Irs2 (insulin receptor substrate 2 gene) developed β-cell failure and impaired β-cell proliferation [73,74]. Similarly, mice lacking Insr (insulin receptor gene) also demonstrated high plasma glucose levels and insulin insensitivity and died shortly after birth .
The first link between insulin signalling and winged-helix transcription factors was established in the nematode Caenorhabditis elegans. Investigators studying mutations affecting the lifespan of the worm discovered that the inactivation of daf-16, the orthologue of Foxo1, was able to rescue the phenotype caused by mutations in daf-2, the orthologue of the Insr gene [76,77]. Embryos lacking Foxo1 demonstrated incomplete vascular development and did not survive beyond E10.5, indicating a crucial role for this gene in embryonic development . Because of the early lethality of Foxo1−/− animals, the effects of Foxo1 on pancreatic function were studied using alternate mouse models.
To investigate the potential role of Foxo1 in murine insulin secretion, both Insr−/− and Irs2−/− mice were bred to Foxo1 heterozygotes [79,80]. The Foxo1 haploinsufficiency partially reversed the β-cell failure and promoted the proliferation of β-cells on the Irs2−/− background . However, despite the 25% improvement in glucose levels in the Insr−/−;Foxo1+/− mice, the early lethality of these neonates could not be rescued . Taken together, these studies indicate that Foxo1 functions to regulate insulin secretion in a negative fashion.
Additional analysis of Foxo1+/−;Irs2−/− mice revealed elevated Pdx1 levels compared with Irs2-null controls, suggesting that Foxo1 functions as a repressor of Pdx1 expression . The repressive relationship between Foxo1 and Pdx1 in the β-cell is in contrast with that of Foxa2 and Pdx1, as previously addressed in this review. Through immunostaining, electrophoretic mobility shift assays and cotransfection experiments, it was demonstrated that Foxa2 and Foxo1 proteins compete for the same binding site on the Pdx1 promoter to regulate the expression of this gene . A model summarizing this regulatory relationship is shown in Figure 2. Whether the control of Pdx1 transcription by Foxa2 or Foxo1 is a developmentally timed switch or regulated under conditions of insulin resistance remains to be determined. It is clear, however, that Foxa2 and Foxo1 are crucial regulators of glucose homoeostasis in the pancreas (reviewed in ).
Coordinated regulation of
Pdx1 by Foxo1 and Foxa2 in the β-cell
The forkhead (Fox) proteins are expressed in organisms ranging from yeasts to humans. Through transcriptional regulation of their target genes, the Fox proteins expressed in the pancreas play crucial roles in the function of this endocrine organ. Much has been learned about both the expression patterns and requirements for Foxa1, Foxa2, Foxa3 and Foxo1 both in the developing embryo and in adults. Animals with null mutations or tissue-specific deletions of the Foxa genes have refined our understanding of these important transcription factors in the context of gene regulation and maintenance of glucose homoeostasis. Specifically, Foxa2 regulates several genes essential for regulated insulin secretion in pancreatic β-cells, including Kir6.2, Sur1 and Schad, mutations of which cause PHHI. In addition, Foxa2 and another forkhead protein expressed in the β-cell, Foxo1, regulate β-cell proliferation and function in a coordinate fashion through control of the Pdx1/IPF1 locus, which is one of the genes mutated in MODY.
- E6.5 etc.
embryonic day 6.5 etc
hepatocyte nuclear factor 3
maturity-onset diabetes of the young
- P1 etc.
postnatal day 1 etc.
pancreatic–duodenal homoeobox 1
persistent hyperinsulinaemic hypoglycaemia of infancy
short chain L-3-hydroxyacyl-CoA dehydrogenase
Work in our laboratory was supported by NIH (National Institute of Health) grants DK055342, DK056947 and DK049210. We apologize to those colleagues whose work we could not discuss in detail due to space constraints.