HES1 is a novel downstream modifier of the SHH-GLI3 Axis in the development of preaxial polydactyly

Sonic Hedgehog/GLI3 signaling is critical in regulating digit number, such that Gli3-deficiency results in polydactyly and Shh-deficiency leads to digit number reductions. SHH/GLI3 signaling regulates cell cycle factors controlling mesenchymal cell proliferation, while simultaneously regulating Grem1 to coordinate BMP-induced chondrogenesis. SHH/GLI3 signaling also coordinates the expression of additional genes, however their importance in digit formation remain unknown. Utilizing genetic and molecular approaches, we identified HES1 as a downstream modifier of the SHH/GLI signaling axis capable of inducing preaxial polydactyly (PPD), required for Gli3-deficient PPD, and capable of overcoming digit number constraints of Shh-deficiency. Our data indicate that HES1, a direct SHH/GLI signaling target, induces mesenchymal cell proliferation via suppression of Cdkn1b, while inhibiting chondrogenic genes and the anterior autopod boundary regulator, Pax9. These findings establish HES1 as a critical downstream effector of SHH/GLI3 signaling in the development of PPD.

Introduction Development of the vertebrate limb is dependent on two signaling centers, the apical ectodermal ridge (AER) that controls proximal-distal (P-D) outgrowth and the zone of polarizing activity (ZPA), the source of Sonic hedgehog (Shh), which regulates anterior-posterior (A-P) patterning, digit number, and identity [1][2][3][4]. Prior to ZPA establishment within the distal posterior mesenchyme, an early A-P axis is first initiated via the expression of Gli family zinc finger 3 (Gli3) and Aristaless-like 4 (Alx4) within the anterior limb bud mesenchyme. GLI3 restricts the expression of Heart and neural crest derivatives-expressed protein 2 (Hand2) to the posterior mesenchyme, while HAND2 antagonizes Gli3 and Alx4 expression [5]. This antagonistic relationship pre-patterns the mesenchyme allowing for the initiation of Shh within the posterior mesenchyme via the action of Fibroblast Growth Factors (FGFs) secreted from the AER [6,7]. SHH in turn enforces Fgf expression (primarily Fgf8 and Fgf4) within the AER, establishing a positive feedback loop that maintains both signaling centers and keeps cells within the distal mesenchyme or apical zone (AZ) in a proliferative and undifferentiated state prior to chondrogenic differentiation and digit formation [8,9].
During A-P patterning of the limb bud, SHH signals in a posterior to anterior manner from the ZPA to inhibit GLI3 protein processing from the full-length activator (GLI3A) to the truncated repressor (GLI3R), thereby restricting the highest concentrations of GLI3R to the anterior limb bud mesenchyme [10]. SHH-mediated regulation of GLI3 processing is critical for establishing the pentadactylous autopod with proper digit identities. Germline mutations of Gli3 lead to multiple forms of polydactyly, and in humans is the underlying cause for Greig cephalopolysyndactyly syndrome (OMIM: 175700), Pallister-Hall syndrome (OMIM: 146510), postaxial polydactyly types A1 and B (OMIM: 174200), and preaxial polydactyly type IV (OMIM: 174700). Conventional homozygous gene deletion of Gli3 in the mouse, represented by the extra-toes mutation (Gli3 xt/xt ), results in the development of generalized polydactyly (seven or greater digits without obvious identities), while heterozygous deletion or haploinsufficiency leads to the development of a single anterior extra digit or preaxial polydactyly (PPD) [11,12]. Conversely, germline homozygous deletion of Shh results in the development of only a single anterior digit with digit one identity due to the preponderance of GLI3R throughout the limb bud mesenchyme, directly implicating the SHH-GLI3 signaling axis in regulating digit number and identity [5,13,14]. Therefore, a primary function of SHH within the autopod is to counteract the GLI3R-mediated constraints on digit development, and this dominance of GLI3R-mediated regulation of digit number and identity is highlighted by the indistinguishable forms of polydactyly observed in Gli3 xt/xt mutant and Shh -/-; Gli3 xt/xt compound mutant mice [15,16].
Mechanistically, the SHH-GLI3 signaling axis regulates the expression of numerous factors implicated in coordinating digit number and/or identity during limb development. Unbiased gene expression studies have identified a few hundred potential candidates likely to be regulated directly or indirectly by SHH and GLI transcription factors [17,18]. However, many of these genes, such as Hoxd10-13 [19][20][21], Hand2 [22], Alx4 [23], Twist1 [24,25], Fgf4 [26], Fgf8 [26,27], Etv4 [25,28], Etv5 [25,28], Tbx2 [29,30], Tbx3 [30], Gata6 [31], and others, also in turn directly regulate Shh/Gli3 expression and signaling that results in a feedback loop impacting digit development. Recent genetic interaction and functional studies determined that one model by which GLI3R functions to constrain digit number requires the coordinated suppression of downstream cell cycle regulators, Cyclin d1 (Ccnd1) and Cyclin dependent kinases 2, 4, and 6 (Cdk2, Cdk4, Cdk6), and the Bone Morphogenetic Protein (BMP) antagonist, Gremlin (Grem1) (32). To control size of the limb field and digit number, this gene regulation ensures that SHH-GLI3 signaling simultaneously coordinates mesenchymal progenitor cell proliferation and the timing by which BMP signaling induces digit chondrogenesis [32]. Additional factors have been identified that potentially function downstream of the SHH-GLI3 signaling without known feedback regulation, these factors include Paired box 9 (Pax9) and several NOTCH signaling pathway components, Jagged1 (Jag1), Hairy/enhancer-of-split 1 (Hes1), and Hairy/enhancer-of-split related with YRPW motif protein 1 (Hey1) [17,18,33]. Pax9 is expressed in the anterior mesenchyme of the limb bud and when deleted gives rise to a single anterior extra digit without obvious alterations to Shh/Gli3 expression or signaling [34]. In a reciprocal fashion, Jag1, Hey1, and Hes1 are expressed in the posterior distal mesenchyme of the limb bud overlapping the ZPA (Shh expression) and surrounding mesenchyme [33,[35][36][37]. Genetic evidence and gene expression studies suggest that each of these genes may be regulated by SHH-GLI3 signaling, such that deletion of Gli3 results in an anterior expansion of Jag1, Hey1, and Hes1 expression with the concomitant loss of Pax9 within the anterior mesenchyme of the limb bud [33,36]. Further, cis-regulatory analyses identified GLI-consensus DNA binding sites and SHH-GLI3 regulation of Pax9, Jag1, and Hes1 [17,18], however it is unknown whether any of these genes are directly responsible for SHH-GLI3 function in regulating digit number during normal autopod development or leading to the PPD pathology.
Here we define the potential role of HES1 as a novel downstream mediator of SHH/GLI3 signaling in the development of PPD. Using a series of genetic interaction and functional studies, we demonstrate that HES1 is regulated via SHH-GLI3 signaling, is sufficient to induce PPD, is required for Gli3-deficient PPD, and is capable of overcoming the digit number constraints of Shh-deficiency. Mechanistically, we show that HES1 is capable of regulating mesenchymal cell proliferation, delaying the onset of chondrogenesis, and coordinating anterior boundary formation to regulate digit number. Collectively, our data highlight a previously unknown role for HES1 as a modifier of SHH/GLI3 signaling capable of altering digit number.

Hes1 over-expression is sufficient to induce preaxial polydactyly (PPD)
To determine whether Hes1 over-expression within limb bud mesenchyme is sufficient to induce PPD, we developed conditional Hes1 gain-of-function mutant mice that carry the Prx1Cre transgene and R26-Hes1 f/f alleles [38][39][40]. Hes1 is normally expressed within the distal posterior mesenchyme of the limb bud [33,35], however utilizing this model the expression of Hes1 is expanded throughout the limb bud mesenchyme to recapitulate the pattern of Hes1 expression observed within Gli3 xt/+ and Gli3 xt/xt limb buds [33].
Skeletal analyses indicate a PPD phenotype in 91% of Prx1Cre; R26-Hes1 f/f (HES1 GOF) mutant limbs as compared to wild-type (WT) littermate controls (Fig 1A and 1B). X-ray ( Fig  1C) and microCT (Fig 1D) at 2-months of age reveal a sixth digit (red asterisk) anterior in position to digit 1. In approximately 9% of HES1 GOF mutants the sixth digit is an incomplete syndactylous digit still fused to digit one (S1 Fig), represented as 5.5 digits for quantitative purposes (Fig 1B). Most other bones in HES1 GOF mice display shortening and altered morphology [41], including fusions of several carpal/tarsal bones (Figs 1C, 1D and S1). When comparing the length and morphology of the additional anterior digit to other digits of HES1 GOF mice, we determined that at both E16.5 and 2-months of age (Fig 1A, 1C, 1D and 1E) the additional anterior digit more closely resembles that of digits 2-5. These data suggest that it is not simply a duplication of digit 1, but rather the addition of an anterior digit with more proximal digit features.
To determine the regional effects of Hes1 over-expression on digit development, we generated ShhCre; R26-Hes1 f/f mutant and controls. ShhCre activation induces Hes1/Gfp overexpression within posterior mesenchymal cells of the ZPA and their descendants that give rise to digits 4 and 5 (S2A Fig) [42]. Skeletal staining demonstrates that over-expression of Hes1 within the posterior mesenchyme does not induce additional digit formation (S2B Fig), further suggesting that HES1 action within the anterior mesenchyme drives the PPD phenotype in HES1 GOF mice.

Hes1 over-expression within limb bud mesenchyme alters cell cycle kinetics
To determine potential mechanisms underlying HES1-induced PPD, we analyzed mesenchymal cell proliferation utilizing BrdU and phospho-histone H3 (PH3) immunohistochemistry/ immunofluorescent (IHC/IF) staining. HES1 GOF limb buds display a significant increase in BrdU and PH3 positive mesenchymal cells within the apical zone, along the P-D axis, and within the anterior mesenchyme (Figs 2A, 2B, S3A, and S3B). Western analyses for cell cycle regulators indicate a decrease in the cell cycle inhibitor, p27, and an increase in CYCLIN D1 (CCND1) (Fig 2C) within the mesenchyme of HES1 GOF limb buds. Real-time quantitative PCR (qPCR) analyses using RNA isolated from the anterior halves of E11.5 limb buds demonstrate a mild increase in Cdk2, Cdk4, and Cdk6 expression and confirmed the significant decrease in Cdkn1b expression, encoding p27 (Fig 2D). These gene expression changes were not observed in the posterior halves of HES1 GOF limb buds (S4A Fig). Since HES1 primarily acts as a transcriptional repressor [43], we searched for and identified putative HES1 binding sites (one E-box and two N-boxes) within 3kb of the Cdkn1b transcriptional start site (S5A Fig). Chromatin immunoprecipitation (ChIP) using HES1 antibodies on DNA isolated from E11.5 HES1 GOF and WT limb buds demonstrated that HES1 is capable of binding the  (S5A Fig lanes 9-12). This binding was substantially enriched in HES1 GOF limb buds (Fig 2E). To confirm the accuracy of our ChIP data, we also performed a positive control ChIP using HES1 antibodies on DNA isolated from E11.5 HES1 GOF and WT limb buds and demonstrated that HES1 binds to its own promoter sequence known to contain self-regulatory N-boxes (S5D Fig). These data indicate that over-expression of Hes1 potentially enhances mesenchymal cell proliferation during limb development in part via the direct suppression of Cdkn1b.

Hes1 over-expression within limb bud mesenchyme delays chondrogenesis
To further investigate the effects of Hes1 over-expression on limb development, we generated HES1 GOF and controls from E11.5 to E12.5 and performed whole-mount in situ hybridization (WISH), histology, IHC, and qPCR. WISH for Sry-box 9 (Sox9), a transcription factor critical for forming mesenchymal condensations and chondrocyte differentiation [44], demonstrates a modest reduction in Sox9 expression within E11.5 HES1 GOF limb buds (Fig 3A). Alcian blue staining of E12.5 HES1 GOF and WT limb bud sections demonstrates a delay in digit chondrogenesis of HES1 GOF embryos (Fig 3A). IHC analyses for COL2A1 confirm the delay in cartilage formation of HES1 GOF autopods, however proximal limb condensations eventually undergo chondrogenic differentiation leading to the formation of smaller cartilage elements (Fig 3A). qPCR performed on RNA extracted from the anterior halves at E11.5 and E12.5 demonstrate a reduction in Sox9, Sry-box 5 (Sox5), Sry-box 6 (Sox6), and Aggrecan (Acan) expression in HES1 GOF limb buds at each time point (Fig 3B). Consistent with published reports [45], our sequence analyses and ChIP studies demonstrate that HES1 is capable of binding a conserved N-box within the Col2a1 enhancer approximately 1700bp downstream of exon 1 (Figs 3C and S5B lanes 1-8), but not sequences further upstream within exon 1 of Col2a1 (S5B Fig lanes 9-12). This binding is enhanced in HES1 GOF limb buds (Fig 3C). These data indicate that over-expression of Hes1 within limb mesenchyme delays the formation of cartilage condensations and chondrogenesis within the limb skeleton, likely via direct negative regulation of chondrogenic genes such as Col2a1 and potentially others [41,45].

Hes1 over-expression alters several SHH/GLI-associated factors critical for digit number and patterning
To assess HES1 regulation of genes implicated in PPD, we performed WISH and qPCR for known regulators of digit number and patterning, including Pax9, Alx4, Fgf8, and Bmp4. We observed patterns of expression similar to Gli3 xt/+ and Gli3 xt/xt mutant limb buds [32,33,46], including decreased Pax9 expression (white arrow), restricted Alx4 expression to the proximal mesenchyme (red arrow), expanded Fgf8 expression within the AER (yellow arrow), and reduced mesenchymal expression of Bmp4 (black arrows) only within the anterior limb bud (Fig 4A) and confirmed these results via qPCR (Figs 4B and S4B). These data indicate that HES1 overexpression alters the expression of several critical regulators of digit number and patterning, similar to that observed in Gli3-deficient polydactylies.
To determine whether Shh expression and SHH signaling was altered in HES1 GOF mutants, we first performed WISH and qPCR for Shh using whole limb buds. HES1 GOF limb buds display both a mild expansion and enhancement in Shh expression at E11.5 (Fig 5A and  5B). To assess whether the altered Shh expression results in enhanced SHH signaling, we assessed SHH target gene (Ptch1, Gli1, and Grem1) expression and identified no changes when comparing whole limb buds, anterior halves, or posterior halves of HES1 GOF and WT limb buds via WISH or qPCR (Figs 5A, 5B and S4B). Since the processing of GLI3 from its fulllength active form (GLI3A) to its repressor form (GLI3R) serves as a readout of SHH activity, we performed Westerns for GLI3 and demonstrate an unaltered ratio of GLI3R to GLI3A in HES1 GOF limb buds (Fig 5C). These data indicate that over-expression of Hes1 within limb bud mesenchyme functions independent or downstream of SHH/GLI3/GREM1 signaling in regulating digit number.

HES1 functions independent or downstream of SHH/GLI3 and can compensate for Shh-deficiency
To determine whether Hes1 over-expression can compensate for the loss of Shh in regulating digit number, we generated HES1 GOF mutant mice in a Shh conditional loss-of-function (SHH LOF) background (Prx1Cre; R26-Hes1 f/f ; Shh f/f ). Skeletal analyses demonstrate that Prx1Cre; R26-Hes1 f/f ; Shh f/+ embryos develop a PPD phenotype similar to HES1 GOF mutants (approx. 6 digits), while Prx1Cre; Shh f/f mutants develop 1-3 digits as previously indicated (Fig  6A and 6B) [47]. Importantly, Hes1 over-expression within the limb mesenchyme of SHH LOF mice (Prx1Cre; R26-Hes1 f/f ; Shh f/f ) induces the formation of 2-5 digits in nearly 95% of double mutants (Fig 6A and 6B). Greater than 60% of double mutants develop 3 or more digits, while only 12.5% of SHH LOF mutants develop 3 digits (Fig 6A and 6B). More than 56% of SHH LOF embryos develop a single digit, while a mere 5-6% of double mutants exhibit one digit (Fig 6A and 6B). In addition to the partial rescue in digit number, we also observed a rescue of both zeugopod elements (radius and ulna) in~10% of double mutants whereas all SHH LOF mutants develop a single zeugopod element resembling the ulna (Fig 6A). Of note, the majority SHH LOF mice die at or prior to birth, while all Prx1Cre; R26-Hes1 f/f ; Shh f/f double mutants generated to date survive to adulthood. These results provide strong evidence that Hes1 over-expression within Prx1Cre-expressing mesenchymal cells is largely capable of compensating for the loss of Shh in regulating digit number, patterning of the zeugopod, and survival.
To further understand the molecular mechanisms associated with these genetic changes and the corrections to SHH LOF mutant phenotypes by over-expressing Hes1, we performed qPCR for Hes1 and a number of SHH targets, patterning factors, cell cycle, and cell death regulators using RNA collected from E11.5 whole limb buds of the genotypes: (Fig 6C and S6A). Hes1 expression is decreased in SHH LOF limb buds, however over-expression of Hes1 within the limb bud mesenchyme is sufficient to drive Hes1 expression above both SHH LOF and control levels in SHH LOF/HES1 GOF limb buds (Fig 6C). The SHH target genes, Ptch1 and Gli1, are markedly reduced in SHH LOF limb buds, while over-expression of Hes1 in the absence or presence of Shh is incapable of regulating their expression (Figs 6C, 5A, and 5B). The anti-apoptotic factor, Bcl2, is a direct SHH/ GLI target gene [48] and an important regulator of cell death within the limb bud mesenchyme [49]. Bcl2 expression is reduced by nearly 50% in SHH LOF limb buds, while SHH LOF/HES1 GOF displayed normal levels of Bcl2 expression (S6A Fig). These data suggest that Hes1 overexpression aids in reversing the cell death phenotype that occurs in SHH LOF mutant limb buds, however TUNEL staining for apoptotic cells suggests that Hes1 overexpression alone in a WT background does not alter mesenchymal cell death (S6B Fig). Importantly, Pax9 exhibits increased expression within SHH LOF limb buds, however over-expression of Hes1 in the absence of Shh dramatically reverses this effect leading to strong suppression of Pax9 (Fig 6C); also observed in HES1 GOF limb buds (Fig 4B). Cdkn1b, a negative regulator of the cell cycle and a direct target of HES1, shows no change in expression when comparing SHH LOF limb buds to controls, however forced expression of Hes1 in the absence of Shh suppresses Cdkn1b expression (Fig 6C) similar to that observed in HES1 GOF limb buds (Fig 2D). Therefore, Hes1 over-expression within the limb mesenchyme can reverse many phenotypic features observed in SHH LOF mutant mice, without directly affecting canonical SHH/GLI3 signaling.

HES1 and SHH/GLI3 signaling synergize to regulate digit number and gene expression
To study potential cooperative effects of HES1 and SHH/GLI3 signaling in regulating digit number, we developed HES1 GOF mutants in the Gli3 xt/+ (GLI3 HET) background (Prx1Cre; R26-Hes1 f/f ; Gli3xt /+ ) and appropriate controls. While GLI3 HET and HES1 GOF mice exhibit 5.5 digits (syndactylous 6 th digit) or a complete 6 digits, the combination of these alleles induces the formation of more than 6 digits in all Prx1Cre; R26-Hes1 f/f ; Gli3 xt/+ double mutants, thereby producing a more severe PPD phenotype than either single mutant alone (Fig 7A and 7B).
At the molecular level, we also observe an additive effect of these alleles on the expression of specific genes. qPCR utilizing RNA isolated from E11.5 anterior limb buds demonstrates an increase in Hes1 expression in GLI3 HET limb buds, which was further upregulated in a progressive manner in HES1 GOF and Prx1Cre; R26-Hes1 f/f ; Gli3 xt/+ double mutant limb buds (Fig 7C). Concomitantly, both qPCR and WISH analyses demonstrate a progressive downregulation of Pax9 expression (Figs 7C and S7) while Hes1 expression increases (Fig 7C). Due to this dynamic inverse relationship and the critical nature of Pax9 in regulating anterior digit number [34], we performed sequence analyses of the Pax9 promoter and ChIP for HES1. HES1 is capable of binding a specific and conserved E-box within the Pax9 promoter located approximately 8400bp upstream of the transcriptional start site (Figs 7D and S5C lanes 1-8), however, was incapable of binding other upstream regions of the Pax9 promoter (S5C Fig lanes  9-12). This binding was significantly enhanced in HES1 GOF limb buds (Fig 7D). Similarly, we analyzed Alx4 promoter/enhancer sequences and could not identify obvious conserved E- box/N-box binding sites. These data indicate that SHH/GLI3 and HES1 signals cooperate in regulating digit number and may do so in part via the direct HES1-mediated regulation of Pax9 within the limb bud mesenchyme.

Gli3-haploinsufficient PPD
To determine whether HES1 functions downstream of SHH/GLI3 signaling and is required for Gli3-haploinsufficient PPD, we generated conditional homozygous deletions of Hes1 floxed alleles within the limb bud mesenchyme of Gli3 xt/+ mutant mice (Prx1Cre; Hes1 f/f ; Gli3 xt/+ ). Skeletal analyses demonstrate that more than 80% of the Hes1 f/f ; Gli3 xt/+ (GLI3 HET) pups develop a syndactylous polydactyly (5.5 digits) phenotype in this genetic background, while only~47% of the Prx1Cre; Hes1 f/f ; Gli3 xt/+ mutant pups develop the same phenotype (Fig 8A  and 8B). These data indicate that HES1 functions downstream of GLI3 and is largely required for the polydactyly phenotype resulting from genetic ablation of a single Gli3 allele.
To further understand the molecular mechanisms associated with the partial genetic rescue of the Gli3 xt/+ polydactylous phenotype, we performed qPCR for Hes1 using RNA collected from E11.5 limb buds of the following genotypes: Hes1 f/f (WT controls), Prx1Cre; Hes1 f/f (HES1 LOF), Gli3 xt/+ (GLI3 HET), and Prx1Cre; Hes1 f/f ; Gli3 xt/+ (HES1 LOF/GLI3 HET) ( Fig  8C). Hes1 expression decreases in both HES1 LOF and HES1 LOF/GLI3 HET limb buds, while it increases in GLI3 HET limb buds (Figs 8C and 7C). These data suggest that Hes1 is a target of SHH/GLI3 signaling in the regulation of digit number. Previous studies identified GLI factors as potential direct regulators of Hes1 [18,50]. Therefore, we performed sequence analyses of the Hes1 promoter and ChIP utilizing anti-GLI2 and anti-GLI3 antibodies to precipitate chromatin isolated from E11.5 WT limb buds. These data demonstrate that GLI3 and Collectively, these data indicate a negative regulation of Hes1 by GLI3R within the limb bud mesenchyme during limb development.
Since we identified Pax9 and Cdkn1b as potential direct HES1 gene targets during HES1-induced PPD, we assessed the expression of these genes in WT, HES1 LOF, GLI3 HET, and HES1 LOF/GLI3 HET limb buds. The expression of both Pax9 and Cdkn1b are upregulated in HES1 LOF limb buds and downregulated in GLI3 HET limb buds, however HES1 LOF/GLI3 HET rescued mutant limb buds exhibit Pax9 and Cdkn1b expression levels similar to or slightly higher than WT controls (Fig 7C). Additionally, WISH for Pax9 and Ptch1 demonstrate a partial rescue of the spatial expression pattern of Pax9 in the HES1 LOF/GLI3 HET double mutants as compared to GLI3 HET, while the expression of Ptch1 remained expanded throughout much of the mesenchyme of GLI3 HET limb buds in either the presence or absence of Hes1 (S8 Fig). These data suggest that SHH/GLI regulation of digit number may be dependent on HES1 and the HES1 transcriptional regulation of Cdkn1b and Pax9, which control both proliferation of the limb mesenchyme and anterior boundaries of digit formation.

Discussion
Cooperative regulation of mesenchymal cell proliferation and chondrogenic differentiation is critical in regulating the size of the limb field and digit number. As chondrogenesis within the limb skeleton proceeds in a proximal to distal direction, the apical zone mesenchyme of the autopod is influenced by signals from the ZPA (SHH) and AER (FGFs) [1,2,9]. FGFs are required for mesenchymal cell survival and to reinforce Shh expression within the ZPA [7]. SHH in turn regulates mesenchymal cell proliferation by relieving GLI3 repression, resulting in the induction of Cdk6 and Ccnd1, among other positive regulators of the cell cycle [17,18,32]. Additionally, SHH/GLI signaling induces Grem1, which functions to inhibit BMPinduced cell cycle exit and chondrogenesis to maintain apical zone mesenchymal cells in a primitive state until SHH/GLI signaling is down-regulated and BMP signals initiate the condensation of SOX9-expressing mesenchymal progenitors [32,36]. Gli3 deficiency results in enhanced Cdk6, Ccnd1, and Grem1 expression, which promotes expansion of the limb field via increased mesenchymal cell proliferation, delayed BMP-induced cell cycle exit, and delayed chondrogenesis. This allows for the development of additional mesenchymal condensations within the expanded autopod resulting in polydactyly [32]. Highlighting the importance of the tight control of mesenchymal proliferation and differentiation in regulating digit number, Gli3-deficient polydactyly is partially overcome via the conditional removal of Cdk6 or a single copy of Grem1 [32] or exacerbated by deletion of one or more Bmp4 alleles [32,51]. Further demonstrating the exquisite sensitivity of this program, conditional deletion of Bmp4 [52] alone or overexpression of Grem1 [53] within limb bud mesenchyme delays mesenchymal cell cycle exit and the onset of chondrogenesis resulting in a polydactyly phenotype resembling that of Gli3-deficient mice. Here we describe a novel role for HES1 in regulating digit number in the developing autopod downstream of SHH/GLI3 signaling. Our data indicate that GLI3R directly suppresses Hes1 expression in the anterior mesenchyme, while active SHH/GLI signaling may promote Hes1 expression in the distal posterior mesenchyme of the developing limb bud. Mechanistically, HES1 inhibits chondrogenesis and stimulates mesenchymal cell proliferation by directly repressing chondrogenic genes (e.g. Col2a1) and negative regulators of mesenchymal cell proliferation (e.g. Cdkn1b/p27), which appears to be distinct from the positive and direct regulation of cell cycle inducers (e.g. Cdk2, Cdk4, Cdk6) and the activation of a potent inhibitor of BMP-induced cell cycle exit and chondrogenesis (e.g. Grem1) via SHH/ GLI3. Therefore, HES1 mediates unique aspects of SHH/GLI signaling to expand the limb field to the appropriate size via the regulation of both mesenchymal cell proliferation and the onset of chondrogenesis.
In addition to BMP signaling and SHH-induced regulators of the cell cycle, anterior expressed genes such as Alx4 and Pax9 are under the control of SHH/GLI signaling and aid in establishing anterior boundaries for autopod expansion [23,33,34,54]. ALX4 restricts the expression of Shh and 5'Hoxd genes to the distal posterior mesenchyme [23,54], and inhibits cell proliferation in some contexts [55]. PAX9 is a critical regulator of A-P patterning and boundary formation in multiple tissues including the limb bud, odontogenic mesenchyme, and the palatal mesenchyme [34,56,57]. In both of the latter instances, PAX9 regulates the expression of Bmp4 and Msx1, while PAX9 and MSX1 cooperatively and directly regulate the transcriptional activation of Bmp4 [56][57][58]. Both Alx4 and Pax9 are reduced or absent in the limb mesenchyme of Gli3 xt/+ or Gli3 xt/xt polydactylous mice and enhanced throughout the mesenchyme in Shh -/limb buds [23,33,54]. Interestingly, Bmp4 expression correlates with GLI3R levels and is subsequently reduced in the anterior mesenchyme of Gli3 xt/xt limb buds and expressed throughout the mesenchyme of Shh -/limb buds [59]. Since genetic ablation of Pax9 and conditional inactivation of Msx1 and Msx2 within the limb mesenchyme results in preaxial polydactyly phenotypes resembling Gli3 xt/+ mice [34,60], it is likely that Bmp4 expression within the anterior limb mesenchyme is transcriptionally regulated via PAX9-MSX1 in a manner similar to that observed in the odontogenic and palatal mesenchyme [57,58]. While PAX9 regulation of Bmp4 within the anterior mesenchyme provides a likely mechanism by which PAX9 establishes an anterior boundary for digit development via regulation of cell cycle exit and chondrogenesis, the precise molecular mechanism by which SHH/GLI signaling regulates the expression of Pax9 has remained elusive. Our data indicate an important role for HES1 downstream of SHH/GLI3 signaling to directly regulate Pax9 expression, while no direct regulation of Alx4 via HES1 could be identified. In Gli3-deficient mice, as in HES1 GOF mice, HES1 expression is observed broadly within the posterior and anterior limb bud mesenchyme resulting in the direct suppression of Pax9 that in turn reduces the expression of Bmp4. Conversely, in the Shh-deficient limb bud, GLI3R is expressed throughout the limb mesenchyme suppressing Hes1 expression which leads to the subsequent posterior expansion of both Pax9 and Bmp4; contributing factors to Shh-deficient digit reductions. Collectively, all of our data are consistent with, and fit into the greater framework of, digit number regulation as demonstrated at the anatomic and molecular levels by numerous genetic models, including Gli3-deficiency [15,32], Pax9-deficiency [34], Msx1/Msx2-deficiency [60], Bmp4-deficiency [51,52], Grem1-over-expression [53], and Shh-deficiency [4,15,47], among others. Our findings further provide novel insights into SHH/GLI regulation of Hes1, as well as, the ability of HES1 to directly bind the promoters of Cdkn1b and Pax9, regulate their expression, and potentially their roles in coordinating digit number in the backgrounds of Hes1 overexpression, Gli3-deficient polydactyly, and Shh-deficient digit reductions. The SHH/GLI/HES1 signaling axis identified here functions in parallel with the SHH/GLI/GREMLIN axis previously identified as a regulator of digit development [32], highlighting the complexity of integrated signals necessary to coordinate proper pentadactylous digit formation within the developing autopod and/or regulating the PPD pathology (Fig 9).
HES1 belongs to a class of basic Helix-Loop-Helix (bHLH) transcription factors, which bind DNA at N-box or E-box sequences to regulate or modify the expression of numerous target genes during development. The SHH/GLI signaling pathway regulates several bHLH factors important for the transcriptional control of genes coordinating digit number and patterning, including TWIST1, TWIST2, HAND1, and HAND2. Each of these bHLH factors can function as transcriptional activators and/or repressors in either homo-or heterodimeric combinations, allowing them to regulate transcription in both active (direct binding or competitive binding of E/N-boxes) or passive (heterodimerization of activator and repressor bHLH proteins that interferes with normal DNA binding) manners. Alterations in their gene dosage via genetic deletion and/or overexpression may shift the balance and occupancy of bHLH homo-or heterodimers localized to E-boxes (or N-boxes) within a number of gene promoters/enhances important for digit number and patterning. Indeed, Twist1 +/and Hand2overexpressing mice develop PPD phenotypes similar to that observed in HES1 GOF mutant mice [61]. Twist1 is normally expressed throughout the limb bud mesenchyme, while Hand2 and Hes1 are natively restricted to the posterior and/or distal posterior mesenchyme [33,35,61]. Therefore, misexpression of Hes1 within the anterior mesenchyme in HES1 GOF or Gli3-deficient mice [33] may cause an imbalance in bHLH homo-or heterodimer formation with TWIST1, HAND2, or other bHLH factors, while also altering their occupancy at E-box/ N-box sites within the promoters/enhancers of genes regulating mesenchymal cell proliferation and differentiation or anterior boundary formation. Indeed, misexpression of Hes1 in this way significantly alters (or induces) binding of HES1 to conserved bHLH biding sites localized to Cdkn1b, Col2a1, and Pax9 promoters/enhancers, and likely represents a mechanism by which PPD occurs in mutants with altered Hes1 expression. Our data shown here, as well as others assessing bHLH factors in digit number regulation, raise the intriguing possibility that Gli3-or Shh-deficiencies (mutations) leading to digit abnormalities may be overcome by shifting the landscape of bHLH factor expression, interaction, and/or promoter/enhancer occupancy during critical windows of autopod development.
While we have identified HES1 as a new downstream bHLH transcriptional modifier of SHH/GLI signaling in the development of PPD, additional questions remain regarding this complex signaling control of digit development. Hes1 is an established target gene of the NOTCH signaling pathway and the NOTCH ligand, Jagged1, and NOTCH target gene, Hey1, are also regulated via SHH/GLI signaling, raising questions as to whether the NOTCH pathway itself functions downstream or in parallel to SHH/GLI-mediated control of digit number [17,18,33,50]. Similar to our HES1 GOF mutant mice and Gli3-deficient mice, overexpression of the NOTCH intracellular domain (NICD) within the limb bud mesenchyme (Prx1Cre; R26-NICD f/+ ) results in enhanced mesenchymal cell proliferation, however, completely inhibits chondrogenesis as compared to a temporary delay in chondrogenesis observed in HES1 GOF and Gli3-deficient limb buds [11,62]. Alternatively, conditional deletion of Rbpjk floxed alleles within the limb mesenchyme (Prx1Cre; Rbpjk f/f ) accelerates chondrogenesis and only reduces skeletal element or digit size as compared to Shh-deficient digit number reductions [14,62]. Interestingly, autosomal dominant forms of Adams-Oliver Syndrome (OMIM 614814, 616028, 616589) caused by germline loss-of-function mutations in RBPjk, NOTCH1, and the NOTCH ligand, DLL4, commonly present with abnormalities of the hands and feet including fused digits (syndactyly), severe shortening of digits (brachydactyly), and/or complete loss of digits (oligodactyly). These most severe types of digit reductions are likely not observed in our conditional NOTCH pathway loss-of-function mice (Prx1Cre; Rbpjk f/f or Prx1Cre; Hes1 f/f as examples) [41,62] for a number of reasons, including NOTCH target gene redundancy (compensation via upregulation of Hes5 in the Prx1Cre; Hes1 f/f mice) [41], the timing by which the Prx1Cre transgene induces gene deletion, or the limited expression of the transgene to the musculoskeletal limb mesenchyme. Similar differences between germline deletions and conditional deletions induced by the Prx1Cre transgene can be highlighted by the differential digit reductions observed in Prx1Cre; Shh f/f and Shh -/mutant mice [14,47]. SHH/GLI signaling may indeed regulate Hes1 in both NOTCH-dependent and -independent manners, therefore future studies will be required to tease apart whether NOTCH signaling itself is required for Gli3-deficient polydactyly or whether Shh-deficient digit restrictions can be overcome by some form of NOTCH activation.

Ethics statement
All animal research performed in this study was approved by the Institution Animal Care and Use Committee (IACUC) at Duke University under protocol number A068-20-03.

Whole mount skeletal staining and in situ hybridization
Whole embryo skeletal staining for digit number analysis was performed using the protocol as previously described [65]. To look at spatial-temporal expression of various genes we performed WISH on E11.5 limb buds using the previously described protocol [66].

RNA isolation and qPCR
Whole and anterior limb buds were harvested in cold 1× PBS and flash frozen using liquid nitrogen. After genotypes were obtained, mutant and wild type limb buds were pooled and homogenized in Trizol (Invitrogen) using a 25g needle. The RNA was then precipitated using 1-bromo-3-choloropropane (MRC). The aqueous layer was then separated and washed with 70% ethanol and purified using the RNeasy Plus Mini Kit (Qiagen). The RNA was transcribed into cDNA and qPCR was performed using methods described [62]. Gene expression was normalized to Actb followed by calculating relative expression using the 2 -(ΔΔCt) method. Mouse specific primer sequences are listed in S1 Table.

Chromatin immunoprecipitation (ChIP) assay
The ChIP assay was performed on wild type and Hes1 GOF E11.5 limb buds using the MAGnify Chromatin Immunoprecipitation system (Invitrogen) using the protocol described [41]. The HES1 antibody was a generous donation from Dr. Ryoichiro Kageyama (Kyoto University). The antibody was used at a concentration of 5 μg per reaction. Data analysis was performed using PCR primers specifically designed flanking the conserved E-box or Nbox sequences within the Cdkn1b, Col2a1, Pax9, and Hes1 promoter or enhancer regions, as well as, upstream negative control primers. Primers are listed in S1 Table. ChIP quantifications were performed on three gel replicates. The ImageJ software package was utilized to perform densitometry measurements of PCR bands from WT and HES1 GOF limb buds and normalized to input controls for all replicates.

Western blots
Whole and anterior limb buds at E11.5 were isolated from wild type and Hes1 GOF embryos and lysed in RIPA buffer (Invitrogen) containing protease inhibitors. The lysate was then centrifuged, the supernatant containing the protein was obtained and quantified using a BCA system. Protein (15μg) was separated using NuPAGE Novex 8% and 10% Bis-Tris pre-cast gels (Invitrogen) and was transferred to a PVDF membrane using the BIORAD system.

Immunohistochemical and histological staining
Embryos were harvested in cold 1× PBS, fixed in 4% paraformaldehyde (PFA), the limbs were dissected and then hand processed. Limbs were embedded in OCT for frozen sectioning and paraffin for standard microtomy and IHC or histologic staining. Frozen sections were cut at 10μm, while paraffin sections were cut at 5μm. To analyze the general cellular morphology, standard histological staining using ABH-OG was performed. IHC was performed on paraffin sections using the VectaStain ABC kits and developed with ImmPACT DAB (Vector Labs). Primary antibodies against the following proteins were used for IHC analyses: ACAN (1:200; Chemicon), COL2A1 (1:100; Thermo Scientific), SOX9 (1:100; Santa Cruz Biotechnology), and PH3 (1:200; Cell Signaling). Immunohistochemistry for BrdU (Invitrogen) was performed as previously described [62].

Quantification and Statistical analysis
All statistics were performed in GraphPad Prism 6 software. Statistical significance was determined by an unpaired 2-tailed Student's t-test or one-way ANOVA. All quantifications are represented as mean ± standard deviation. P values of less than 0.05 are considered statistically significant. All experiments were performed with N�3 biological replicates which represent 3 or more age matched separate littermate animals. Biological replicates (N) and statistical analyses are noted within all figure legends.