|Home | About | Journals | Submit | Contact Us | Français|
During central nervous system development the timing of progenitor differentiation must be precisely controlled to generate the proper number and complement of neuronal cell types. Proneural basic helix-loop-helix (bHLH) transcription factors play a central role in regulating neurogenesis, and thus the timing of their expression must be regulated to ensure that they act at the appropriate developmental time. In the developing retina, the expression of the bHLH factor Ath5 is controlled by multiple signals in early retinal progenitors, although less is known about how these signals are coordinated to ensure correct spatial and temporal pattern of gene expression. Here we identify a key distal Xath5 enhancer and show that this enhancer regulates the early phase of Xath5 expression, while the proximal enhancer we previously identified acts later. The distal enhancer responds to Pax6, a key patterning factor in the optic vesicle, while FGF signaling regulates Xath5 expression through sequences outside of this region. In addition, we have identified an inhibitory element adjacent to the conserved distal enhancer region that is required to prevent premature initiation of expression in the retina. This temporal regulation of Xath5 gene expression is comparable to proneural gene regulation in Drosophila, whereby separate enhancers regulate different temporal phases of expression.
During nervous system development, neurons are born in a precise temporal order, and the timing of the generation of different classes of neurons is critical for the correct complement of cells to form. For example, in the developing retina there are seven major cell types that are born in a conserved order over developmental time and the differentiation status and birthdate of these cells are tightly correlated (Livesey and Cepko, 2001). Members of the basic helix-loop-helix (bHLH) transcription factor family have been shown to play a role in specification and differentiation of specific neuronal cell types throughout the developing nervous system (Guillemot, 1999). Although individual bHLH factors are required for the development of different neuronal classes, the neuronal cell type that these bHLH factors specify depends in part upon the timing of their expression as well as other regulatory factors that are co-expressed (Moore et al., 2002; Kageyama et al., 1997). Thus, the expression of proneural bHLH transcription factors must be spatially and temporally regulated to ensure that the appropriate complement of neurons is generated as the nervous system develops.
Ath5 is a bHLH transcription factor that has been shown to have a crucial role in regulating neurogenesis in the retina (Vetter and Brown, 2001). Ath5 expression in the developing retina is conserved across vertebrate species (Kanekar et al., 1997; Brown et al., 1998, Masai et al., 2000; Matter-Sadzinski et al., 2001), and in Xenopus expression begins in progenitors just prior to the onset of neurogenesis and is later down-regulated as cells begin to differentiate (Kanekar et al., 1997). Ath5 is required for the genesis of retinal ganglion cells (RGCs), which are the first born retinal cell type, since Ath5 mutants in both mouse and zebrafish fail to generate RGCs (Brown et al., 2001; Wang etal., 2001; Kay et al., 2001). This is comparable to the function of the Drosophila ortholog, atonal, which is required for the genesis of R8 cells, the first cell to be specified in the Drosophila eye (Jarman et al., 1994).
Because of its central role in regulating RGC genesis, the mechanisms regulating Ath5 expression have been investigated in several vertebrate species and it is clear that multiple intrinsic factors and extrinsic signals are required. bHLH factors themselves play a role in regulating Ath5 expression through auto- and cross-regulation (Matter-Sadzinski et al., 2001; Hutcheson et al., 2005). In chick, Cath5 and Ngn2 bind to the Cath5 cis-regulatory region and regulate its expression (Skowronska-Krawczyk et al., 2004), while in Xenopus the activity of the proximal promoter is dependent upon two highly conserved E-boxes and ectopic expression can be activated by multiple proneural genes, including Xath5, NeuroD and XNgnR1 (Hutcheson et al., 2005). In mouse, there is no evidence for autoregulation of Math5 expression (Brown et al., 2001; Wang et al., 2001; Hutcheson et al., 2005), but expression does depend upon Pax6 function (Brown et al., 1998; Marquardt et al., 2001). Extrinsic signaling is also critical since FGF signaling is required for the initiation of Ath5 expression in zebrafish and chick (Martinez-Morales et al., 2005).
Although multiple signals have been implicated in regulating Ath5 expression, less is known of how these signals are coordinated to ensure correct spatial and temporal patterns of gene expression. Proneural gene regulation has been best characterized in Drosophila, where a model has been proposed suggesting that within a given spatial domain bHLH genes are generally controlled by separate enhancers to regulate different temporal phases of gene expression, thus driving progression of progenitors towardsneuronal commitment (Gilbert and Simpson, 2003). For example, atonal expression in the developing fly eye is initiated by factors such as Pax6 and Hh that act through a 3′ enhancer, and subsequently there is refinement and upregulation of atonal expression through a 5′ autoregulatory enhancer (Sun et al., 1998; Zhang et al., 2006). Drosophila scute is similarly regulated through distinct regulatory elements that control initiation of expression in response to prepattern factors within proneural clusters (Garcia-Garcia et al., 1999; Rodriguez et al., 1990), followed by refinement and maintenance of expression within sense organ precursors (Culi and Modolell, 1998; Gilbert and Simpson, 2003). For vertebrate proneural genes, enhancers have been identified that regulate expression in different regions of the embryo (Blader et al., 2003; Scardigili et al., 2001), but not different temporal windows within the same domain of expression, raising the question of whether temporal regulation of different phases of expression occurs in vertebrates. Here we identify a key distal Xath5 enhancer and show that this enhancer regulates the early phase of Xath5 expression within the developing eye, while the proximal enhancer previously identified (Hutcheson et al., 2005) acts later. The distal enhancer responds to key patterning factors in the optic vesicle including Pax6, while FGF signaling acts through enhancer sequences outside of both the conserved proximal and distal regions to regulate Ath5 expression. In addition, we have identified an inhibitory element adjacent to the conserved distal enhancer region that is required to prevent premature initiation of expression in the retina. Thus the control of Xath5 gene expression provides a striking example in vertebrates for temporal regulation of proneural gene expression through distinct regulatory elements.
The 3.3 kb of 5′ Xath5 genomic sequence (pG1 3.3 kb) was previously cloned into the promoterless GFP reporter construct pG1 (Hutcheson et al., 2005). All deletion constructs were generated by PCR amplification and cloned into the pG1-TATAA construct, which is pG1 containing the Xath5 minimal promoter sequence (Hutcheson et al., 2005). Constructs with point mutations were made by site-directed mutagenesis using PCR amplification (Weiner et al., 1995). All E-boxes were mutated from CANNTG to ATTnTG (Helms et al., 2000). The Pax6 sites were mutated from ttgtccTGGgattaTCaagcctcatttcac to ttgtccAAAgattaAAaagcctcatttcac (as denoted in Fig. 4A). All constructs were verified by sequencing.
Transgenic Xenopus embryos were generated as previously described (Kroll and Amaya, 1996; Hutcheson and Vetter, 2002). Whole mount antibody staining was performed on all embryos using a polyclonal anti-GFP antibody (Torrey Pines) at 1:500 and an Alexa Fluor 488-conjugated secondary (Molecular Probes) at 1:1000 unless otherwise noted.
Embryos were processed for in situ hybridization (ISH) as previously described (Kanekar et al., 1997) with the following probes: Xath5 (Kanekar et al., 1997), Rx1 (Mathers et al., 1997), X-Ngnr1 (Ma et al., 1996), Sox2 (Mizuseki et al., 1998), Xpax6 (Hirsch and Harris, 1997), XER81 (Chen et al., 1999), Vsx1 (D’Autilia et al., 2006) and XNeuroD (Lee et al., 1995). Where indicated, mRNA was injected into one dorsal animal blastomere at the 8 cell stage in the following amounts: 600 pg dnPax6 (Chow et al., 1999). 500 pg XFD (Amaya et al., 1991), 500 pg mCherryCAAX (a gift from Kristen Kwan), 1 ng dsRed (Clontech), 50 pg β galactosidase.
Sequence analysis was performed using VISTA (http://genome.lbl.gov/vista/index.shtml) (Frazer et al., 2004, Dubchak et al., 2000) with default parameters. Candidate transcription factor binding sites were identified using the Transfac MATCH program cutoffs of 0.75 (core score) and 0.70 (matrix score) (http://www.gene-regulation.com/pub/databases.html#transfac) (Matys et al., 2003). Sequence alignments were performed by ClustalW (v1.4) using MacVector (v7.9) with default parameters, using 5′ genomic sequence from 4 species: human HATH5 (Accession #AF418922 chr10:69,663,273-69,663,312; UCSC Genome Browser), mouse MATH5 (Accession #AF418923, chr10:62,561,153-62,561,194; UCSC Genome Browser, July, 2007 Assembly), chick CATH5 (Accession #NW_001471715 chr6:11,644,145-11,644,185; UCSC Genome Browser), and Xenopus laevis XATH5 (Accession # EU202949).
An insert containing the X5distal 152 bp multimer, Xath5 TATAA and GFP was cut out of the pG1 vector using MluI and XbaI and cloned into the ISceI-pBSII SK+ vector (gift from Jochen Wittbrodt). I-SceI-assisted transgenesis (Thermes et al., 2002, Grabher et al., 2003) was used to generate a stable transgenic fish line : Tg(X5distal 152 bp mult:GFP) zc38.
Wild type or zc38/+ fish were dechorionated and incubated in 20 μM SU5402 (Calbiochem) in E3 from 14 hpf to 28 hpf. The fish were processed for ISH as described in Suli et al., 2006. Whole mount antibody staining was performed using an anti-GFP antibody (Molecular Probes) at 1:400 and an Alexa Fluor 488-conjugated secondary (Molecular Probes) at 1:100 as described in Suli et al., 2006.
The anti-Pax6 antibody used for the chromatin immunoprecipitation (ChIP) experiments (Covance PRB-278P) was generated against a highly conserved C-terminal peptide from the mouse Pax-6 protein, and has been used for both western blot and ChIP in several previous studies (Samaras et al., 2002; Martin et al., 2004; Grinchuk et al., 2005). We verified that the antibody could recognize Xenopus Pax6 protein by western blot following in vitro transcription and translation (TNT Kit; Promega) of Xenopus laevis Pax6 cDNA (Hirsch and Harris, 1997; data not shown).
ChIP was performed using standard protocols (Wells et al., 2002; Boyer et al., 2005). Briefly, stage 28 pG1X5 3.3 kb transgenic Xenopus embryos were fixed in cross-linking buffer containing 1% formaldehyde. Heads were then dissected and the tissue was homogenized using a Dounce homogenizer. Nuclei were isolated and sonicated in a water bath sonicator (Diagenode Bioruptor) for 12 minutes (30 seconds on/30 seconds off) at “High” power setting. Sonicates were incubated with either control rabbit IgG (Sigma) or rabbit anti-Pax6 antibody (Covance PRB-278P). DNA was recovered following reversal of crosslinking, purified, and used as template for real-time PCR. Real-time PCR was performed in a 25-μL amplification mixture containing 1 μL of ChIP product pulled down by either Pax6 antibody or IgG control, 12.5 μL of 2x PCR master mix (SYBR Green; Applied Biosystems, Foster City, CA), and 100 nM forward and reverse primers, respectively (5′-ATTTCACCCAGGGCGGATT -3′, 5′-TTCCTTGAACTCGAACCATGTG -3′ enhancer region; 5′-CGGCCACAAGTTGGAATACA -3′, 5′-TTGGCTTTGATGCCGTTCTT -3′ for GFP control). The PCR conditions included a polymerase activation step at 95°C for 10 minutes followed by 40 cycles of 95°C for 15 seconds and 60°C for 60 seconds and run on a sequence detector (model 7500; Applied Biosystems). The CT of enhancer in the ChIP products was normalized by the CT of GFP control, and the fold difference is calculated by 2^(−ΔCT), ΔCT being the cycle number difference between enhancer region versus GFP. Real-time PCR was performed in triplicate for each ChIP product and primer pair and the average CT value and standard deviation were calculated from the triplicate reactions. The same trend was obtained from two independent experiments. The statistical significance of relative differences in ChIP product abundance was determined by Student’s group t-test.
In our previous study we identified a proximal conserved cis-regulatory region that was sufficient to promote retina-specific expression of a GFP transgene (Hutcheson et al., 2005). We also found that 3kb of sequence 5′ to this proximal region promoted retinal expression, suggesting the existence of an additional cis-regulatory enhancer. To more precisely map this enhancer we searched for conserved sequence within the distal 3 kb region by aligning sequence from X. laevis with that from a related Xenopus species, X. tropicalis, and found an approximately 1 kb region that is highly conserved between the two species (Fig. 1A).
To test for regulatory activity within this 1 kb of conserved sequence we cloned it into a vector that contains GFP (pG1) and the minimal Xath5 promoter that alone does not promote GFP expression (Hutcheson et al., 2005). We used this construct to generate transgenic Xenopus embryos (Kroll and Amaya., 1996) and scored for GFP expression at stage 32/33, when Xath5 is normally strongly expressed (Kanekar et al., 1997). With this nuclear transfer method, 25–60% of embryos that cleave and gastrulate normally will be transgenic and express the transgene in a reproducible pattern (Kroll and Amaya, 1996). We found that this 1 kb sequence was sufficient to promote retina, olfactory, and pineal expression (48%, 89/187) in a pattern that closely mimicked endogenous Xath5 expression (Fig. 1 B, C; Kanekar et al. 1997). To determine more precisely where the regulatory sequences lie within this fragment, we generated a series of nested deletions and tested them for their ability to promote transgene expression (Fig. 1B). We identified a minimal 152 bp distal enhancer that was sufficient to promote retina, olfactory, and pineal expression, with stronger GFP expression when this enhancer was duplicated (35%, 33/93, Fig. 1 B, D). However, a smaller 73 bp fragment did not promote retinal expression (0/37) but was only sufficient to promote transgene expression in the olfactory placodes and pineal gland (15/37, 40.5%), suggesting that there may be specific regulation of Xath5 expression within these tissues that can be uncoupled from retinal expression (Fig. 1 B, E).
To test whether the conserved 152 bp distal enhancer was functional in other vertebrate species we generated a zebrafish transgenic line using the pG1 transgene with two copies of the Xenopus 152 bp enhancer driving GFP expression, Tg(X5distal 152 bp mult:GFP) zc38. We observed transgene expression in the developing zebrafish retina as early as 15 hours postfertilization (hpf) (data not shown), with stronger expression developing from 18–24 hpf and persisting through 72hpf (Fig. 1F–H; Supplemental Fig. 1; data not shown). Notably, the onset of transgene expression preceded endogenous ath5 (formally, atoh7 - Zebrafish Information Network) expression, which commences in the zebrafish retina at 25 hpf (Masai et al., 2000). Since the 152 bp distal Xath5 enhancer could promote retina specific expression in both zebrafish and Xenopus, it suggests that at least some of the mechanisms regulating expression from this enhancer are conserved.
Since the 152bp element yielded early expression in zebrafish, prior to endogenous ath5 expression, we investigated whether this enhancer promotes early expression in Xenopus, perhaps contributing to the temporal regulation of Xath5 expression. We previously showed that the 3.3 kb Xath5 5′ cis-regulatory sequence initiates transgene expression in the retina at stage 24, coincident with the onset of endogenous Xath5 expression (Hutcheson et al., 2005). Consistent with this, we could not detect GFP transgene expression driven by either the 3.3 kb region or just the proximal enhancer element before stage 24 (Fig. 2A,B, I; data not shown). However, we observed transgene expression driven by the distal 152 bp enhancer in the optic vesicle as early as stage 22 – 2 hours prior to the initiation of endogenous Xath5 expression (31/109, Fig. 2E, F, I). To determine whether the proximal element normally constrains expression until stage 24, we combined the proximal and 152 bp distal enhancer elements but still observed transgene expression in the optic vesicle at stage 22 (46/118, data not shown), similar to the 152bp enhancer alone.
Interestingly, a transgene containing a longer distal fragment spanning the 1kb region that is conserved between X. laevis and X. tropicalis was not expressed prematurely, since GFP expression was not evident in the optic vesicle at stage 22 (0/55; Fig. 2C,I), but was clearly present at later stages (see Fig. 2D, I). This suggests that this region may contain a modulatory element that is capable of suppressing precocious distal enhancer activity, which would normally function to ensure appropriate temporal expression of Xath5. To more precisely map where the putative inhibitory element lies within the distal region, we tested a series of deletion constructs and found that constructs that included an additional 100 bp 5′ to the 152 bp distal enhancer did not show premature (stage 22) transgene expression (Fig. 2G–I). Thus, the distal enhancer can initiate Xath5 gene expression in the optic vesicle, but the timing of its activity is modulated by adjacent cis-regulatory sequence.
Once we had defined a key Ath5 enhancer, we sought to determine whether its activity is controlled by factors known to regulate endogenous Ath5 expression. bHLH transcription factors are known to both cross-regulate and auto-regulate expression (Helms et al., 2000; Skowronska-Krawczyk et al., 2004; Matter-Sadzinski et al., 2001; Gilbert and Simpson, 2003) and we previously found evidence that this was true for the proximal Xath5 cis-regulatory region (Hutcheson et al., 2005). Therefore, we investigated whether bHLH factors are also involved in regulating Xath5 expression through the 152 bp distal enhancer. Cross-species sequence comparison of mouse, human, chick and Xenopus identified 4 conserved E-boxes (CANNTG) within this minimal enhancer (Fig. 3A). Two of these E-boxes, E3 and E4, had been previously identified by us and shown not to be required for expression in the context of the 3.3kb cis-regulatory region (Hutcheson et al., 2005). The other two E-boxes, E5, and E6, are conserved only between X. laevis and X. tropicalis. Zebrafish Ath5 contains a divergent distal enhancer sequence that was not included in this alignment (Accession #AL627094), although two E-boxes are present. To test whether E-boxes are required for distal enhancer expression we mutated either E3 and E4 (Fig. 3B, C), or all four E-boxes within the context of the distal enhancer (Figure 3B′,D). Even when all four E-boxes were mutated, the 152 bp distal enhancer was still able to promote transgene expression in the retina, although there was often ectopic expression in other CNS regions as well (Fig. 3D). Therefore, E-boxes are not required for distal enhancer expression, although they may function to limit expression in regions outside of the developing eye.
In mouse, Pax6 plays a role in regulating Math5 expression (Brown et al., 1998; Marquardt et al., 2001). To determine whether Pax6 is required for ath5 expression in Xenopus we injected mRNA for dominant negative Pax6 (dnPax6) (Chow et al., 1999) and mCherryCAAX, to mark the injected side, into one dorsal blastomere of 8 cell stage Xenopus embryos and assayed for Xath5 expression via in situ hybridization. We found that Xath5 expression was dramatically decreased on the injected side of 15/31 embryos (Fig. 4A–B). Xath5 expression is not lost due to a failure of eye formation since expression of the retinal progenitor markers Rx (Fig. 4 E–F; 0/31 decreased), Vsx1 (Fig. 4 C–D; 2/22 decreased) and Sox2 (0/15 decreased; data not shown) was unchanged. This suggests that Pax6 may act upstream of Xath5 to regulate its expression.
Since Pax6 is required for Xath5 expression we tested whether Pax6 alone is sufficient to induce Xath5 expression. We injected mRNA for Pax6 and GFP, to mark the injected side, into one dorsal animal blastomere of 8 cell stage embryos and assayed for Xath5 expression via in situ hybridization. There was no induction of ectopic Xath5 expression on the injected side (Fig. 4 G–H; 0/28 with ectopic expression). We similarly overexpressed Pax6 by mRNA injection into pG1X5 3.3 kb transgenic embryos derived by fertilizing eggs from an adult transgenic female, this time using Cherry-CAAX mRNA to mark the injected side. We found no ectopic transgene expression (Fig. 4 I–J; 0/68 with ectopic expression), and thus conclude that Pax6 alone is not sufficient to promote Xath5 expression.
While Pax6 is not sufficient to induce Xath5 expression it is required for Xath5 expression, and sequence analysis of the 3.3 kb 5′ cis-regulatory sequence using Transfac (http://www.biobase-international.com) MATCH (Matrix Search Transcription Factor Binding Site) identified two overlapping Pax6 paired domain binding sites, P1 and P2 which are located within the 152 bp distal enhancer (Fig. 5A). These are the only two Pax6 sites identified within the 3.3kb cis-regulatory sequence. P1 aligns with the Pax6_01 matrix (0.832 core match, 0.812 matrix match; Epstein et al., 1994), while P2 aligns with the Pax6_Q2 matrix from the Transfac database (0.861 core match, 0.778 matrix match; Duncan et al., 1998, Roth et al., 1991, Sander et al., 1997, Zhou et al., 2000). Pax6 may thus regulate Xath5 expression directly through a conserved binding site in this distal enhancer region.
To determine whether the distal enhancer Pax6 sites we identified are required for transgene expression in Xenopus, we generated a construct in which the core nucleotides of each Pax6 site are mutated in the context of the 152 bp distal enhancer (Fig. 5A). For the P1 site mutation we mutated the conserved paired domain binding consensus site (Epstein et al., 1994). This mutation affects two nucleotides in the core that directly contact the N-terminal helical domain of Pax6 (Xu et al., 1999). For the P2 site mutation, we mutated the 3 nucleotides that make up the conserved core of a Pax6_Q2 binding site (Duncan et al., 1998, Roth et al., 1991, Sander et al., 1997, Zhou et al., 2000). First we tested the P2 mutation alone and found that there was only very weak but specific retina expression at stage 24 (4/49; Fig. 5B). Next we mutated both Pax6 sites and found only very weak transgene expression in a few cells in the retina at stage 24 (very weak expression in 7/215 embryos; Fig. 5 B, D, compare with C), and at stage 33 (0/54; data not shown). This suggests that regulation of Ath5 expression by Pax6 is conserved in Xenopus and that both Pax6 sites in the 152 bp distal enhancer are required to mediate this effect.
Finally, we tested whether Pax6 binds to the distal enhancer region in vivo by performing chromatin immunoprecipitation (ChIP). To enrich for the distal enhancer target sequence we used tissue from pG1X5 3.3 kb transgenic Xenopus embryos generated by fertilizing eggs from an adult transgenic female. Embryos were fixed at stage 28 and chromatin was prepared from isolated head tissue. We precipitated using an anti-Pax6 antibody that has been used in previous ChIP studies (Samaras et al., 2002; Martin et al., 2004; Grinchuk et al., 2005) and amplified the co-precipitated DNA with primers specific to the region containing the two overlapping Pax6 sites we identified (see Fig. 3A), which are the only Pax6 sites present within the Xath5 cis-regulatory region. We found that the Xath5 enhancer was precipitated by anti-Pax6, but not control, antibody (Fig. 5E). The GFP coding region of the transgene was not precipitated. We conclude that Pax6 directly regulates Xath5 expression by binding to the conserved Pax6 sites within the distal enhancer.
In addition to intrinsic factors, Ath5 is also regulated by extrinsic signaling pathways. While it has been shown that FGF signaling is required for Ath5 expression in zebrafish and chick, little is known about the mechanism(s) involved (Martinez-Morales et al., 2005). We first wanted to determine whether regulation of Ath5 expression by FGF signaling is conserved in Xenopus laevis. We blocked FGF signaling in the Xenopus retina using the dominant negative FGF receptor XFD, comprising the FGF receptor extracellular and transmembrane domains but lacking the intracellular transactivation domain (Amaya et al., 1991). To block FGF signaling specifically in the retina at a time prior to Xath5 expression but after formation of the eye field we used the human frizzled 5 (Fzd5) enhancer (N. Marsh-Armstrong, unpublished) to drive XFD expression in retinal progenitors. Use of the Fzd5 enhancer eliminated potential effects that blocking FGF signaling might have on gastrulation movements and/or dorsal mesoderm formation since it initiates expression specifically in the optic vesicle beginning at stage 18 (N. Marsh-Armstrong, unpublished). We then generated transgenic embryos carrying the Fz5-XFD construct and assayed for gene expression at stage 28, when proneural genes are robustly expressed (Kanekar et al., 1997, Lee et al., 1995). We found that Xath5 expression was dramatically reduced or absent in 60% of embryos (24/42), which is consistent with our standard transgenic rate of 25–60% (Fig. 6A vs. B). In addition, we found that expression of the bHLH factors XNeuroD (16/43; Fig. 6C vs. D) and XNgnR1 (22/53; data not shown) was also dramatically reduced, indicating that FGF signaling not only plays a role in regulating Xath5 expression but may serve a more global role in regulating proneural gene expression in the developing retina.
To determine whether FGF signaling acts to regulate Xath5 expression by affecting the expression of factors upstream of Xath5, we assayed stage 28 Fz5-XFD transgenic embryos for factors that are expressed prior to Xath5 in the retina. We found that the expression of the retinal progenitor markers Rx (Fig. 6E vs. F), Pax6, and Vsx1 was not changed in Fz5-XFD transgenic embryos (data not shown). The expression of Sox2, which is necessary for neural competence and proneural gene expression (Van Raay et al., 2005) was also not affected (Fig. 6G vs. H).
One way that FGF signaling may be mediating its effects on Xath5 expression is through the Pea3 family of ETS-domain transcription factors. Members of this family are activated in response to FGF signaling through the RAS-MAPK pathway in multiple tissues (Janknecht, 1996,O’Hagan et al., 1996, McCabe et al., 2006). In Xenopus, Er81 is the only family member identified, and its expression can be activated by FGF signaling in animal cap explants (Chen et al., 1999). Further, Xer81 is expressed in the optic vesicle beginning at stage 22 (Chen et al., 1999), just prior to the onset of Xath5 expression (Kanekar et al., 1997). Little is known about the role of Er81 in retinal development, and it remains to be shown whether its retinal expression is regulated by FGF signaling. To determine whether Xer81 expression in the retina is dependent upon FGF signaling we generated Fz5-XFD transgenic embryos and scored them for retinal Xer81 expression. We found that Xer81 expression was decreased in 24/55 (44%) of transgenic embryos, which is consistent with our standard transgenic rate (Fig. 6J compare to I). This demonstrates that FGF signaling has been blocked in Fz5-XFD transgenic embryos and suggests that FGF signaling may act through Xer81 to regulate Xath5 expression in the retina.
To determine whether FGF signaling is regulating Xath5 expression through its 5′ regulatory sequence, we blocked FGF signaling in retinal progenitors in a pG1Xath5 3.3kb transgenic frog line (Hutcheson et al., 2005). We fertilized eggs from a pG1Xath5 3.3kb transgenic female, collected cleavage stage embryos and injected one dorsal animal blastomere at the 8 cell stage with mRNA for XFD and mCherryCAAX to mark the injected side. We found that transgene expression was dramatically reduced or absent on the injected side in 93% (14/15) of the XFD injected embryos (Fig. 7A–B). In addition, we found that while overexpression of XFD at the 8 cell stage inhibits Xath5 expression on the injected side, it does not affect the expression of Sox2 or Rx (data not shown) so that, the lack of transgene expression in the pG1Xath5-3.3kb transgenic embryos is not due to a lack of progenitor cells. Thus FGF signaling clearly has a role in regulating Xath5 expression and likely acts through sequences located within the 3.3 kb 5′ cis regulatory region. To test whether FGF acts on the 152 bp distal enhancer, we used SU5402, a drug shown to specifically inhibit FGF signaling in both frog and fish (Mohammadi et al., 1997; Moreno and Kintner, 2004). We found that in our experiments the drug inhibited FGF signaling, as assayed by decreased endogenous Ath5 expression, more reliably in fish than in frogs. Thus, we crossed wild type zebrafish with zc38 transgenic zebrafish (yielding 50% transgenic offspring, see Fig. 1F–H), then incubated the embryos in SU5402. The zebrafish were treated from 14hpf, just prior to transgene expression, until 28hpf, then collected and analyzed for endogenous ath5 expression by in situ hybridization, or GFP expression as a reporter of transgene activity. Since SU5402 treatment blocks FGF signaling throughout the embryo, the treated zebrafish had smaller heads than controls (Fig. 7 D, F compare to C, E). Nonetheless, we found that while SU5402 treatment blocked endogenous ath5 expression in embryos (Fig. 7 C, D; 1/104 SU5402 treated embryos expressed ath5 vs. 38/41 DMSO treated) it had no effect on transgene expression (Fig. 7 E, F; 27/42 GFP positive SU5402 treated embryos vs. 20/47 DMSO treated). Thus, while FGF signaling clearly plays a role in regulating Ath5 expression, it does not appear to do so through the 152 bp distal enhancer.
We have found that within the upstream cis-regulatory region of Xath5 there are two conserved sequences, a proximal enhancer and a distal enhancer, that are each sufficient to promote retinal transgene expression in transgenic Xenopus embryos when fused to a minimal promoter (this study and Hutcheson et al., 2005). However, although they share similar tissue specificity, they appear to initiate expression at different stages of eye development, suggesting that they do not simply act redundantly. We found that the distal enhancer initiated expression at early optic vesicle stages in both zebrafish and Xenopus, well before endogenous Ath5 expression. Furthermore, although conserved E-boxes were present within the distal enhancer, retinal expression was not altered when these sites were mutated, demonstrating that this enhancer may be the bHLH-independent enhancer we previously showed was present in the 3 kb distal region (Hutcheson et al., 2005). In contrast, the proximal enhancer we described previously closely mimics endogenous Ath5 expression, and likely plays a role in auto- or cross-regulation, since its activity was dependent upon two highly conserved E-boxes (Hutcheson et al., 2005) and expression from this enhancer does not initiate until after endogenous Xath5 expression (M. Willardsen unpublished observation).
Previous studies of vertebrate proneural gene regulation have described modular enhancer elements that regulate different spatial domains of bHLH expression (Verma-Kurvari et al., 1998; Helms et al., 2000; Scardigli et al., 2001; Blader et al., 2003; Nakada et al., 2004). However, the temporal regulation of vertebrate proneural bHLH gene expression is less well understood. In Drosophila, specific enhancers regulate different temporal phases of scute (sc) and atonal (ato) gene expression (Skeath et al., 1994; Baker et al., 1996; Sun et al., 1998). For example, for ato a 3′ enhancer initiates expression in cells anterior to the morphogenetic furrow, while a 5′ enhancer maintains expression in intermediate clusters and prospective R8 cells (Sun et al., 1998). Our work shows that Xath5 fits a more general model for proneural gene regulation involving two phases: an early phase in which gene expression is initiated by an enhancer, such as the distal enhancer, that responds to upstream regulators, followed by a later phase in which a separate enhancer(s), in this case the proximal enhancer, functions to maintain expression, generally through autoregulation. Similarly, separate enhancers have been shown to regulate different temporal phases of myogenic bHLH gene expression during muscle development (Chen et al., 2001; Carvajal et al., 2001). However, this is the first demonstration in vertebrates of separate enhancers controlling different temporal phases of proneural gene expression, suggesting that the temporal regulation of proneural gene expression is in some cases conserved from invertebrates to vertebrates.
There appears to be added complexity to this basic model of Xath5 gene regulation, since additional sequences are required to ensure appropriate onset of Xath5 gene expression. The inclusion of 100 bp of 5′ Xath5 genomic sequence prevents precocious transgene expression, suggesting that the 100 bp contains an element that ensures proper timing of Xath5 expression. Once again there are striking parallels with the regulation of Drosophila ato expression, where a 1.2 kb inhibitory element has been identified in the 3′ cis-regulatory region and similarly shown to prevent premature expression (Zhang et al., 2006). The precise control of the onset of Ath5 expression is crucial since premature expression results in decreased eye size and cell number, and disrupts the complement of cell types within the retina (Kanekar et al., 1997; Brown et al., 1998, Moore et al., 2001). In chick, Cash1 expression precedes Cath5 expression and prevents premature expression (Matter-Sadzinski et al., 2001). In Xenopus, Xash1 and Xash3 precede Xath5 in retinal progenitors (Kanekar et al., 1997), but there are no E-boxes within the 100bp inhibitory fragment, suggesting that these proneural factors are not directly inhibiting activity of this modulatory region. Future studies will focus on defining the signals that act through this 100bp sequence.
How generally conserved is the regulation of Ath5 gene expression? There is a high degree of sequence conservation across vertebrates in the distal enhancer region, and it promotes retinal transgene expression in both Xenopus and zebrafish. In addition, the corresponding region of the mouse Ath5 gene also promotes retinal transgene expression in Xenopus (Riesenberg et al., in press). Together this suggests that the signals that initiate Ath5 expression through this enhancer region are likely to be conserved. This is in contrast to the proximal enhancer region, which is also highly conserved, but promotes retinal transgene expression in Xenopus but not mouse or zebrafish (Hutcheson et al., 2005; data not shown). Furthermore, there is no evidence that Math5 regulates its own expression during mouse retinal development (Brown et al., 2001; Wang et al., 2001; Hutcheson et al., 2005). Thus, although aspects of Ath5 gene regulation may be shared, there may be unique mechanisms to accommodate the features of eye development and retinal neurogenesis for a given species.
Which patterning factors act through the distal enhancer to initiate Xath5 expression? Pax6 is an essential regulator of eye development (Gehring and Ikeo, 1999) and is required for Ath5 expression in mouse (Brown et al., 1998; Marquardt et al., 2001). Here we show that conserved Pax6 binding sites within the distal enhancer are required for distal enhancer activity, and that Pax6 directly binds to the region containing these sites in vivo. In addition, we found that while Pax6 activity is not sufficient to induce Xath5 expression, it is required for endogenous Xath5 expression. Parallel studies in mouse have shown that a similar distal Math5 enhancer contains a conserved Pax6 site that binds to Pax6 in gel shift assays and is sufficient to promote retinal transgene expression in frog (Riesenberg et al., in press). In Drosophila, the 3′ Ato enhancer, which is involved in initiating ato expression is regulated directly by Pax6 (Zhang et al., 2006). Pax6 has also been shown to regulate Ngn1 and Ngn2 expression in multiple tissues in mouse and zebrafish (Blader et al., 2003; Toresson et al., 2000). In addition, although the distal enhancer sequence in zebrafish is somewhat divergent, there is still a Pax6 site present. Thus Pax6 appears to play a highly conserved role as an essential patterning factor required for initiation of proneural gene expression during eye development.
FGFs are known to be important for Ath5 expression in zebrafish and chick (Martinez-Morales et al., 2005). We have shown that FGF signaling is required for endogenous Xath5 expression but that this effect is not likely mediated through the 152bp distal enhancer. We show that inhibition of FGF signaling is not only required for Xath5 expression but also for the expression of other bHLH factors indicating that FGF signaling may be a more general mechanism by which proneural gene expression is initiated during eye development. This may be mediated via the Pea3 family member ER81. In the developing chick retina both Pea3 and Cath5 expression are regulated by FGF signaling (McCabe et al., 2006). In Xenopus, ER81 expression can be activated by FGF signaling in animal cap explants, while its endogenous expression in the marginal zone can be blocked by dominant-negative FGF receptor (XFD) (Chen et al., 1999, Munchberg and Steinbeisser, 1999). We have shown that ER81 expression in the Xenopus retina is dependent upon FGF signaling, making it an intriguing candidate as a regulator of Xath5 expression in the retina. Future experiments will focus on defining the mechanisms through which FGF signaling is acting to regulate Ath5 gene expression.
We thank Michael Spencer for assistance with sequence analysis and mutation of the Pax6 sites; Loni DeFriez, Erin Callahan, and Guoying Wang for assistance with the Xenopus transgenic procedure, Kristen Kwan for the mCherryCAAX-pCS2 construct and all our many colleagues for plasmids, advice and discussion. We thank Bill Harris and Joe Yost for generously providing SU5402. This work was supported by a University of Utah graduate research fellowship to MIW, NIH grant EY13612 to NLB, NIH grant R01 EY016097 to NMA, NIH grant EY12873 to CBC, NIH grant EY15480 to HME, NIH grant EY179932 to KBM, and NIH grant EY12274 to MLV.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.