A novel function for Egr4 in posterior hindbrain development

Segmentation of the vertebrate hindbrain is an evolutionarily conserved process. Here, we identify the transcription factor early growth response 4 (egr4) as a novel regulator of posterior hindbrain development in Xenopus. egr4 is specifically and transiently expressed in rhombomeres 5 and 6 (r5/r6), and Egr4 knockdown causes a loss of mafb/kreisler and krox20/egr2 expression in r5/r6 and r5, respectively. This phenotype can be fully rescued by injection of frog or mouse Egr4 mRNA. Moreover Egr4-depleted embryos exhibit a specific loss of the neural crest stream adjacent to r5, and have inner ear defects. While the homeodomain protein vHnf1/Hnf1b directly activates Mafb and Krox20 expression in the mouse hindbrain to specify r5, we show that in Xenopus this process is indirect through the activation of Egr4. We provide evidence that rearrangements in the regulatory sequences around egr4 and mafb genes may account for this difference.

D evelopment of the vertebrate hindbrain along the anteroposterior axis is an evolutionarily conserved process. It involves the transient formation of seven to eight transversal swellings known as rhombomeres (r). These territories are cell lineage restricted compartments and constitute developmental units for cranial motor nerves and neural crest cells. In most vertebrates, cranial neural crest cells originate from r2, r4 and r6 and migrate as three streams that populate the 1st, 2nd and 3rd branchial arches, respectively. Neural crest cells give rise to the ganglia of cranial nerves V, VII/VIII and IX, and to craniofacial skeletal elements unique for each branchial arch. Fewer neural crest cells are produced from r3 and r5; these cells typically migrate rostrally and caudally to join the adjacent streams [1][2][3] . This conserved pattern of neural crest cell migration directly depends on the proper segmentation of the hindbrain. Caudal hindbrain patterning is also critical for the specification and development of the inner ear which arises from the ectoderm adjacent to r4/r5 4

.
The gene regulatory network underlying hindbrain segmentation involves a complex interplay of transcription factors and signaling molecules leading to the regional expression of different combination of genes, which imparts a unique molecular identity to each rhombomere 5 . Several of these genes encode transcription factors, such as Krox20/Egr2, MafB/Kreisler/Valentino, Hnf1b/vHnf1 and members of the Hox family. In the posterior hindbrain, the zinc finger transcription factor Krox20 is expressed in the prospective r3 and r5 6,7 . The basic leucine zipper Mafb is expressed in r5 and r6, and controls the specification of these two rhombomeres, and activates Krox20 in r5 8,9 . The homeodomain transcription factor vHNF1 (variant hepatocyte nuclear factor 1) cooperates with FGF to promote caudal hindbrain identity, by activating Mafb and Krox20 expression in r5/r6 and r5, respectively 10,11 . Hoxb1 is expressed in the neural plate with an anterior limit at the prospective r3/r4 boundary, and later is only maintained in r4, where it acts with other Hox factors to specify r4 identity 12 . The regulatory cascade underlying rhombomere specification is highly conserved among vertebrates, and only minor variations in this blueprint have been reported.
Here we describe the function of the transcription factor early growth response 4 (Egr4), a novel regulator of hindbrain development in Xenopus. Egr4 is transiently expressed in r5/r6 at neurula stages. Using loss-and gainof-function approaches we demonstrate that Egr4 is required downstream of Hnf1b to activate mafb and krox20 expression and promote r5 identity in the posterior hindbrain. This is the first report describing a role for Egr4 in the vertebrate nervous system. Xenopus laevis egr4 possesses an open reading frame encoding 486 amino acids. At the amino acid level, Xenopus laevis Egr4 shares 40% identity with human EGR4 (accession # NP_0011956), 44% identity with mouse Egr4 (accession # NP_065621), 50% identity with chicken Egr4 (accession # SP_420888) and 89% identity with Xenopus tropicalis Egr4 (accession # XP_004911346). By whole mount in situ hybridization egr4 is detected in a single stripe across the neural plate, extending into the neural plate border where egr4 overlaps with pax3 and zic1 expression domains (Fig 1a). To confirm that egr4 is a target of Pax3 and Zic1, we performed perturbation experiments in the embryo using pax3-and zic1-specific morpholino antisense oligonucleotides (MOs) [14][15][16] . Pax3 or Zic1 knockdown resulted in a reduction or loss of egr4 expression in most embryos (Fig 1b-c), suggesting that both factors are independently required for egr4 expression. In animal cap explants from embryos injected at the 2-cell stage (Fig 1d), expression of pax3GR and/or zic1GR (the hormone-inducible versions of pax3 and zic1, fused to the human glucocorticoid receptor ligand-binding domain) showed that Pax3 and Zic1 are sufficient to synergistically activate egr4 (Fig 1e). Moreover, egr4 expression in pax3/zic1-injected explants was dramatically reduced when the protein synthesis inhibitor cyclohexamide (CHX) was added to the culture medium, indicating that the induction of egr4 by Pax3 and Zic1 is indirect (Fig 1e).
Egr4 is specifically and transiently expressed in the posterior hindbrain. In order to more precisely map the expression domain of egr4, we performed in situ hybridization on embryos at different stages using digoxigenin-labeled RNA probes for a number of genes expressed in similar region of the ectoderm, including en2 (midbrain-hindbrain boundary 17 ), krox20/egr2 (hindbrain r3/r5 18 ), fgf3 (hindbrain r4 19 ) mafb/kreisler (hindbrain r5/r6 20 ), hnf1b/vHnf1b (hindbrain r5/r6/r7 21 ) and snail2 (neural crest 22 ). The probes were used alone or in combination as indicated (Fig 2a). egr4 expression is initiated at stage 13 as a single domain across the neural pate. This expression progressively increases to reach a maximum around stage 14-15, and then slowly decreases to become undetectable around stage 19/20 (Fig 2a; not shown). This expression profile was independently confirmed by qPCR (Fig 2b). Spatially, there is no overlap between egr4 and en2 expression domain, however, egr4 is co-expressed with krox20 in r5 (Fig 2a). Temporally, the escalation of egr4 expression precedes krox20 and mafb expression as shown by qPCR (Fig 2b). We also performed two-color in situ hybridization using snail2, fgf3 and mafb in combination with egr4 (Fig 2c). The most lateral domain of egr4 overlaps with snail2 demonstrating that egr4 is expressed in a subpopulation of neural crest progenitors. In addition, egr4 expression domain in the hindbrain lays immediately posterior to fgf3 expression domain in r4, and overlaps completely with mafb in r5/r6 (Fig 2c). Taken together, these results indicate that egr4 is specifically and transiently expressed in r5/r6, and its expression escalates ahead of the expression of krox20 and mafb expression in this region.
Egr4 is required for mafb and krox20 expression in the posterior hindbrain. To determine whether Egr4 function is required for posterior hindbrain development we performed Egr4 knockdown in developing embryos using MOs. The first MO (erg4MO1) was designed to specifically interfere with translation of egr4 mRNA. This was confirmed in an in vitro transcription/translation assay in which erg4MO1 blocked Egr4 protein production (Fig 3a; Supplementary Figure 1). Unilateral injection of egr4MO1 at the 2cell stage led to a reduction of krox20 expression in r5 at stage 15, without affecting krox20 expression in r3; however, r3 appeared  shifted posteriorly in some embryos (Fig. 3d). egr4 knockdown also strongly inhibited the expression of mafb in r5/r6 (Fig. 3e). In morphant embryos the loss of mafb expression in r5/r6 was associated with a reduction of hoxa3 expression, a direct target of Mafb 23 (Fig 3f). The loss of mafb and krox20 in morphant embryos could be rescued by injection of mRNAs encoding the hormone inducible version of Xenopus egr4 (egr4GR) or mouse Egr4 (mEgr4GR) (Fig. 3e, g). To confirm the specificity of the knockdown phenotype, we used a second MO (egr4MO2) that specifically interfered with egr4 pre-mRNA splicing by targeting the splice donor site of exon 1 (Fig 3b), resulting in the production of a transcript of higher size, due to intron retention (Fig 3c). The phenotype of egr4MO2-injected embryos was indistinguishable from the phenotype generated by injection of the translation blocking MO, with strong inhibition of mafb and krox20 expression in r5/r6 and r5, respectively (Fig 3h). Furthermore, coinjection of the two MOs at low doses showed an additive effect on the repression of mafb expression (Supplementary Figure 2). The similarity of the two MO phenotypes, and the ability to rescue krox20 and mafb expression in morphant embryos vouch for a specific requirement of Egr4 in posterior hindbrain development.
To evaluate the consequences of the loss of krox20 and mafb on other regions of the hindbrain, we analyzed the expression of hoxb1, fgf3 and hnf1b in morphant embryos. We found that the expression domain of hoxb1 and fgf3, normally confined to r4, was expanded posteriorly (Fig 3i), consistent with a loss of r5 identity. The expression of hnf1b was largely unaffected, although slightly shifted posteriorly (Fig 3j), suggesting that Egr4 is functioning downstream of hnf1b. The transcription factor Meis3 is expressed in r2-r4 24 and has been shown to be acting downstream of Pax3 and Zic1 25 . meis3 expression level was unchanged in egr4-depleted embryos, however there was a posterior shift of its anterior boundary of expression, in similar region of the ectoderm, including en2 (midbrain-hindbrain boundary), krox20/egr2 (hindbrain r3/r5), mafb/kreisler (hindbrain r5/r6) and hnf1b/vHnf1b (hindbrain r5/r6/r7). The probes were used alone or in combination as indicated. Scale bar, 300 mm. (b) Temporal dynamics of the expression of egr4, krox20, mafb, and hnf1b analyzed by qPCR in whole embryos at the indicated stages. The values were normalized to odc1. (c) Two-color in situ hybridization for egr4/snail2, egr4/fgf3, and egr4/mafb at the neurula stage. The expression of egr4 at the neural plate border overlaps with the neural crest marker, snail2 (arrows). egr4 is detected immediately adjacent and posterior to fgf3 expression domain in r4. egr4 and mafb have overlapping expression domains in r5/r6. Scale bar, 300 mm.
www.nature.com/scientificreports SCIENTIFIC REPORTS | 5 : 7750 | DOI: 10.1038/srep07750 similar to the one observed for hnf1b (Fig 3j). We also analyzed the regulation of egr4 expression by Meis3. Meis3 MO-mediated knockdown resulted in a severe reduction of egr4 and mafb expression in r5/r6 (Fig 3k), suggesting that Meis3 is acting upstream of Egr4 and Mafb. This result is consistent with a previous study showing that Meis3 regulate cell fates in the hindbrain in a non-cell autonomous manner 24 .
Egr4-depleted embryos have neural crest and inner ear defects. Because egr4 expression at the neural plate border overlaps with the neural crest territory (Fig 2c), we analyzed the expression of two early neural crest transcriptional regulators, snail2 and sox9, in morphant embryos. We found that both genes were largely unaffected in Egr4depleted embryos (Fig 3l), suggesting that egr4 knockdown did not interfere with neural crest specification. In Xenopus, neural crest cells at hindbrain level migrate towards the branchial arches into four streams, the mandibular, hyoid, anterior and posterior branchial neural crest 26 . The stream of neural crest expressing egr4, completely overlaps with krox20-positive neural crest cells (Fig 2a), which corresponds to the anterior branchial stream that travels caudal to the otic vesicle (centered onto r4 in Xenopus 27 ) into the third branchial arch 28 . In egr4-depleted embryos this stream of neural crest cells was specifically reduced or lost, as revealed by the expression of krox20, sox10 and tfap2a at the tailbud stage (Fig 3m). Interestingly, even in morphant embryos retaining some krox20 expression in r5, the stream of neural crest cells adjacent to this axial level was often missing, suggesting that both domains are regulated by different levels of egr4. The Mafb mouse mutant Kreisler, and its counterpart in fish valentino, which exhibit mispatterning of hindbrain caudal to r3, have severe inner ear defects 23,29,30 . Since egr4-depleted embryos fail to express mafb, we analyzed otic vesicle formation in egr4 morphant embryos by histology. We found that a majority of these embryos had reduced or missing otic vesicles on the injected side (Fig 3n). We posit that the neural crest and the inner ear phenotypes observed in egr4-depleted embryos could be the result of defects in posterior hindbrain patterning.
Hnf1b activates egr4 expression in r5/r6. Hnf1b/vHnf1 is a major regulator of posterior hindbrain development where it participates in the formation of the boundary between r4 and r5 and establishes r5 and r6 identities. More specifically, Hnf1b represses r4 identity in r5/ r6, and synergizes with r4-derived Fgf3 signal to drive mafb expression in r5/r6, and cooperate with Mafb for the direct transcriptional activation of krox20 in r5 10,11 . To determine the position of Egr4 in this regulatory cascade we performed gain-offunction experiments by injection of mRNAs encoding a hormone inducible version of hnf1b (hnf1bGR). Unilateral injection of hnf1bGR mRNAs at the 2-cell stage caused an anterior expansion of egr4 and mafb expression domains, associated with a downregulation of hoxb1 and fgf3 (Fig 4a), indicative of a loss of r4 identity. The expression of krox20 was reduced in these embryos (Fig 4a), since r4-derived Fgf signals are required to maintain krox20 expression in r5 31,32 . These results suggest that egr4 is activated by Hnf1b to establish r5 identity. Consistent with this view, egr4 gain-of-function by injection of Xenopus egr4 (egr4GR) or mouse Egr4 (mEgr4GR) mRNAs caused a dramatic expansion of krox20 and mafb expression domains, often extending to the entire injected side (Fig 4b-c). The expression of hnf1b, hoxb1, and fgf3 was strongly repressed in these embryos (Fig 4b-c).
The repression of hnf1b in Egr4GR-injected embryos suggested the existence of a negative feedback loop in the regulation of hnf1b expression in r5/r6 that may involve Egr4 directly or other factors activated by Egr4. To test this possibility, embryos were injected with mRNAs encoding a hormone inducible version of mouse Krox20 (mKrox20GR) or Xenopus mafb (mafbGR), and in both cases we observed a strong inhibition of hnfb1 expression as well as a loss of egr4 on the injected side (Fig 4d), indicating that Krox20 and Mafb are the likely factors repressing hnf1b expression in r5/r6. In most vertebrates, Hnf1b is initially expressed in the posterior hindbrain with an anterior limit at the r4/r5 boundary [33][34][35] . Later, as Krox20 is activated in r5 this anterior limit progressively retracts posteriorly, and hnf1b is eventually excluded from r5 and r6 21,[33][34][35] . We performed two-color in situ hybridization to assess the spatio-temporal expression of egr4 and hnf1b. At stage 13, egr4 overlaps with hnf1b expression domain anteriorly, and as development proceeds, the anterior limit of hnf1b is shifted posteriorly and segregates from egr4 expression domain (Fig 4e).
Egr4 directly regulates mafb expression. Our results suggest that Hnf1b induces mafb and krox20 expression indirectly through the activation of Egr4 (Fig 5a). This is in contrast to what has been described in the mouse in which Hnf1b has been shown to directly activate Mafb expression 34,36 (Fig 5b). To determine whether the activation of mafb by Hnf1b is primarily mediated through Egr4 in Xenopus, we evaluated the ability of hnf1bGR to activate mafb in the absence of Egr4 function. Co-injection of egr4MO2 with hnf1bGR mRNA completely blocked endogenous as well as ectopic mafb expression by Hnf1b (Fig. 5c). These results support the view that Hnf1b activates mafb via Egr4 in Xenopus.
In the mouse, Hnf1b directly binds with an evolutionally conserved region of DNA upstream of the Mafb gene, known as the S5 enhancer 34,36 . This region also contains three Cdx1 binding sites. In the hindbrain Cdx1 has an anterior limit of expression at the prospective r6/r7 boundary, and Cdx1 represses Mafb expression in this region, leading to the characteristic r5/r6 expression of Mafb 36 . The S5 enhancer has been mapped to a region 20 Kb upstream of the mouse Mafb gene. Sequence alignments indicate that this regulatory module is well conserved in human and chicken. A similar region could not be found around Xenopus tropicalis mafb gene using the UCSC Gene browser (Fig 5d,e). However, a region with similar characteristics, containing two putative Hnf1b binding sites and a single putative Cdx1 binding site, was identified 6Kb downstream of Xenopus tropicalis egr4 gene (Fig 5f,g), here referred as a putative S5 enhancer (pS5). To determine whether this regulatory module can drive egr4 expression, pS5 was cloned in forward (pS5F) and reverse (pS5R) orientation into ptkEGFP vector containing a basal thymidine kinase promoter driving enhanced GFP expression 37 . Both constructs, pS5F-ptkEGFP and pS5R-ptkEGFP, were co-injected with hnf1bGR mRNA at the 2-cell stage and immediately treated with dexamethasone. Embryos were cultured for 6 hours and analyzed by qPCR (Fig 5h). We observed a significant up-regulation of EGFP transcripts in embryos co-injected with hnf1bGR mRNA and pS5F-ptkEGFP (4.6 fold) or pS5R-ptkEGFP (10.2 fold) as compared to embryos injected with each vector alone (Fig 5i). These results indicate that Hnf1b may directly activate egr4 expression via the pS5 enhancer.

Discussion
Our findings reveal a novel function for Egr4 in posterior hindbrain development. Egr4 is specifically and transiently activated in r5/r6 by Hnf1b, and is required for mafb and krox20 expression in r5/r6 and r5, respectively (Fig 6). In addition, Egr4 knockdown leads to a specific loss of the anterior branchial neural crest stream that normally travels to the third branchial arch, and inner ear defects. While a recent study has implicated Egr4 in the differentiation of the brain primordia during planarian regeneration 38 , our study is the first report describing a role for Egr4 in the vertebrate nervous system.
Using a combination of gain-and loss-of-function approaches we provide evidence that Egr4 is part of a conserved regulatory cascade, involving Hnf1b, Mafb and Krox20, essential to specify r5 identity in the hindbrain. Hnf1b/vHnf1 is a major regulator of posterior hindbrain development. It is believed to repress r4 identity in r5/r6, and to synergize with r4-derived FGF signals to directly drive mafb expression in r5/r6. Hnf1b eventually cooperates with Mafb for the direct transcriptional activation of krox20 in r5 10,11 . Our gain-of-function studies indicate that Hnf1b can promote egr4 and mafb expression. Moreover, ectopic expression of egr4 is sufficient to activate mafb and krox20 in the embryo, suggesting that Hnf1b may control the expression of these factors indirectly in Xenopus, through the activation of Egr4. Consistent with this possibility, hnf1bGR is unable to rescue mafb expression in Egr4-depleted embryos. This is in contrast (e) Two-color in situ hybriidization for egr4/hnf1b and krox20/hnf1b. Overtime, the anterior limit of hnf1b retracts posteriorly and segregates from the r5 egr4 and krox20 expression domains. Dorsal views, anterior to top. In all panels scale bars, 300 mm.

www.nature.com/scientificreports
to what has been reported in other species in which Hnf1b directly activates Mafb and Krox20 expression 11,21,36,39 . Once activated by Hnf1b in r5/r6, Egr4 induces the expression of mafb1 and krox20, which in turns act as negative regulators of Hnf1b to restrict its expression to a level posterior to r6. Accordingly, ectopic expression of Krox20 or Mafb represses hnf1b expression in the caudal hindbrain. The inhibitory activity of Krox20 in restricting Hnf1b expression posteriorly has been well described in other species [33][34][35] . This negative feedback loop may also account for the transient expression of egr4 in r5/r6 in Xenopus, which appears to require continued Hnf1b input for its maintenance. The expression of Krox20 and Mafb1 is maintained primarily through an autoregulatory mechanism 36,39,40 (Fig 6).
Egr4 belongs to the early growth response factor family of transcription factors. Members of this family were initially identified as genes rapidly induced by growth factors. A defining feature of Egr proteins is a highly conserved DNA-binding domain composed of three zinc finger motifs that bind a 9-bp response element to initiate transcriptional response 41 . Egr2/Krox20, the most extensively studied member of this family, is essential for the development of  r3/r5 by directly activating Hoxa2 and Hoxb2 in r3/r5 and Hoxb3 in r5 [42][43][44] . Egr2/Krox20 mutant mouse embryos also have peripheral nerve myelination defects 45 . Egr1/Krox24 plays a critical role in regulating the luteinizing hormone (LH) b gene expression; Egr1-deficient mice have low levels of LHb in the pituitary and low levels of LH in serum, and this appears to primarily affect female fertility 46,47 . Egr3 is required for muscle spindle morphogenesis and normal proprioception in mice 48 .
Egr4 is detected at low levels in testicular germ cells and Egr4deficient mice are characterized by a defect in germ cell maturation, leading to male infertility 49 . Egr4 is also expressed at high level in postnatal rat brain 50 , and at E13.5 in the mouse spinal cord 51 . Beside this report, there is no detailed description of Egr4 developmental expression in the mouse. Egr4 mutant mouse embryos exhibit no obvious phenotypic abnormalities in the nervous system. However, embryos have not been examined specifically for defects in the developing brain. It is possible that the effects may be transient or subtle. Alternatively, Egr4 may not be an essential component of the regulatory cascade controlling r5 identity in mammals. Interestingly, mouse Egr4 can fully compensate for the loss of Egr4 in frog embryos, and in overexpression studies its function is indistinguishable from that of Xenopus Egr4. Therefore, the difference in the regulation of r5 identity between frog and mouse cannot be explained by intrinsic functional differences in the Egr4 proteins.
In the mouse, Hnf1b regulates Mafb expression through the S5 enhancer 36 , a region highly conserved in chicken and human. This enhancer could not be identified around Xenopus tropicalis mafb gene. However, we found that a regulatory module similar to the S5 enhancer, containing two Hnf1b binding sites, was present downstream of the egr4 gene in Xenopus tropicalis, a region also conserved in zebrafish. Using a reporter assay in the embryo we were able to show that Hnf1b can activate this putative enhancer. We propose that rearrangements in the regulatory regions around egr4 and mafb genes may account for the evolutionary changes in the importance of Egr4 in establishing r5 identity in the hindbrain of amniotes versus anamniotes.

Experimental Procedures
Plasmid constructs and morpholino antisense oligonucleotides. The full-length Xenopus laevis egr4 was obtained by RACE-PCR using SMARTer TM RACE cDNA amplification Kit (Clontech, Mountain View CA) according to the manufacturer's instructions. Briefly, a first-strand cDNA was synthesized from 1 mg of total RNA isolated from neurula stage embryos. Two sets of gene-specific primers (GSP), 59-RACE-egr4 and 39-RACE-egr4 (59-AGCAAGGG-CTGTGGCGACTTGACTGT CT-39 and 59-AGCCGAAGCGTCA-AAGTGGGAAACGCGA-39) were designed based on EST sequences (Unigene ID Xl.53808). The RACE-PCR conditions were as follows: 5 cycles at 94uC for 30 sec and at 72uC for 3 min; 5 cycles at 94uC for 30 sec, at 70uC for 30 sec, and 72uC for 3 min; 27 cycles at 94uC for 30 sec, at 68uC for 30 sec, and 72uC for 3 min. 59-RACE and 39-RACE PCR products were cloned into pGEM-T Easy vector (Promega, Madison WI) and sequenced. The sequence of Xenopus egr4 has been deposited into GeneBank (Accession # KF957598).
Embryos, injections and explants culture. Xenopus laevis embryos were staged as previously described 55 and raised in 0.1X NAM (Normal Amphibian Medium 56 ;). This study was performed in accordance with the recommendations of the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The animal protocol (# 120311) was approved by the Institutional Animal Care and Use Committee of New York University. Embryos were injected in one blastomere at the 2-cell stage and analyzed by in situ hybridization at stage 15. MOs (40-50 ng) and synthetic mRNA (100 pg) were injected together with 500 pg of b-galactosidase mRNA as a lineage tracer. Embryos injected with mRNA encoding the hormone inducible version of Xenopus egr4 (egr4GR), hnf1b (hnf1bGR) and mafb (mafbGR), and mouse Egr4 (mEgr4GR) and Krox20 (mKrox20) embryos were treated at stage 12.5 with 10 mM dexamethasone (DEX; Sigma-Aldrich, St Louis MO) and collected at stage 15/17. For animal cap explant experiments, both blastomeres at the 2-cell stage were injected in the animal pole region 100 pg of mRNAs encoding pax3GR and zic1GR, explants were dissected at the late blastula stage and immediately cultured for several hours in NAM 0.5X plus 10 mM of dexamethasone. In some experiments, 10 mg/ml cycloheximide (CHX; Sigma-Aldrich, St Louis MO) was added to the culture medium to block protein synthesis, in which case explants were pretreated with cycloheximide for 30 min before dexamethasone treatment 53 . The explants were subsequently analyzed by quantitative PCR (qPCR).
Lineage tracing, in situ hybridization and histology. Xenopus embryos were fixed in MEMFA and processed for Red-Gal (Research Organics, Clevelan OH) staining to visualize the lineage tracer (b-galactosidase) prior to in situ hybridization. Whole-mount in situ hybridization was performed as previously described 57 Table 1 and the Power SYBR Green PCR Master Mix (Invitrogen, Grand Island NY). The reaction mixture consisted of 10 ml of Power SYBR Green PCR Master Mix, 200 nM primers, and 2 ml of cDNA in a total volume of 10 ml. The PCR conditions were as follows: incubation at 50uC for 2 min; activation at 95uC for 10 min; 40 cycles at 95uC for 10 s and at 60uC for 30 s; melt curve at 95uC for 15 s, 55uC for 15 s, and at 95uC for 15 s. The DDCT method was used to quantify the qPCR results. Each reaction included a standard curve of serial dilution points (in 10-fold increments) of test cDNA. In each case, ornithine decarboxylase 1 (odc1) or elongation factor 1a (ef1a) was used for normalization.
Isolation of conserved regions. Using the UCSC Gene browser, we identified regions in Xenopus tropicalis genome containing two putative Hnf1b binding sites and a cdx1 binding site in the vicinity of Egr4 coding region (pS5). Using PCR, we amplified this putative regulatory region from Xenopus tropicalis genomic DNA. The pS5 module was cloned in forward and reverse orientation upstream of ptkEGFP vector that contains the Herpes simplex virus thymidine kinase basal promoter driving enhanced GFP expression 37 . These constructs are referred as pS5F-ptkEGFP and pS5R-ptkEGFP, respectively.