Downstream of FGF during mesoderm formation in Xenopus: The roles of Elk-1 and Egr-1

Developmental Biology 336 (2009) 313–326

Contents lists available at ScienceDirect

Developmental Biology j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / d e v e l o p m e n t a l b i o l o g y

Genomes & Developmental Control

Downstream of FGF during mesoderm formation in Xenopus: The roles of Elk-1 and Egr-1 Oliver Nentwich a, Kevin S. Dingwell a,b, A. Nordheim c, J.C. Smith a,b,⁎ a b c

Wellcome Trust /Cancer Research UK Gurdon Institute and Department of Zoology, University of Cambridge, Tennis Court Road, Cambridge CB2 1QN, UK Medical Research Council National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 1AA, UK Interfakultäres Institut für Zellbiologie, Abteilung Molekularbiologie, Auf der Morgenstelle 15 (Verfügungsgebäude), 72076 Tübingen, Germany

a r t i c l e

i n f o

Article history: Received for publication 18 December 2008 Revised 19 September 2009 Accepted 21 September 2009 Available online 30 September 2009 Keywords: Xenopus FGF Elk-1 Egr-1 Mesoderm Xbra Activin MAPK

a b s t r a c t Signalling by members of the FGF family is required for induction and maintenance of the mesoderm during amphibian development. One of the downstream effectors of FGF is the SRF-interacting Ets family member Elk-1, which, after phosphorylation by MAP kinase, activates the expression of immediate–early genes. Here, we show that Xenopus Elk-1 is phosphorylated in response to FGF signalling in a dynamic pattern throughout the embryo. Loss of XElk-1 function causes reduced expression of Xbra at neurula stages, followed by a failure to form notochord and muscle and then the partial loss of trunk structures. One of the genes regulated by XElk-1 is XEgr-1, which encodes a zinc finger transcription factor: we show that phosphorylated XElk-1 forms a complex with XSRF that binds to the XEgr-1 promoter. Superficially, Xenopus tropicalis embryos with reduced levels of XEgr-1 resemble those lacking XElk-1, but to our surprise, levels of Xbra are elevated at late gastrula stages in such embryos, and over-expression of XEgr-1 causes the down-regulation of Xbra both in whole embryos and in animal pole regions treated with activin or FGF. In contrast, the myogenic regulatory factor XMyoD is activated by XEgr-1 in a direct manner. We discuss these counterintuitive results in terms of the genetic regulatory network to which XEgr-1 contributes. Crown Copyright © 2009 Published by Elsevier Inc. All rights reserved.

Introduction The mesoderm of the amphibian embryo is induced by signals derived from the vegetal hemisphere of the embryo during blastula and gastrula stages (Smith, 1995). Current thinking has it that mesoderm-inducing molecules are members of the transforming growth factor type β (TGF-β) family and that mesodermal gene expression is maintained by members of the fibroblast growth factor (FGF) family (Heasman, 1997; Heasman, 2006). For example, maintenance of Xbra expression in Xenopus is thought to occur through a loop in which FGF signalling activates the expression of Xbra and Xbra then directly activates expression of the FGF family member eFGF (Isaacs et al., 1994; Kroll and Amaya, 1996; SchulteMerker and Smith, 1995). Recent experiments suggest that FGF signalling is also necessary for the establishment of Xbra expression (Fletcher and Harland, 2008), but the questions remain: how is this simple autocatalytic feed-forward loop kept in check, and what other components play a role in modulating mesoderm formation in response to FGF? The answers to these questions requires identification of other components of the FGF signal transduction pathway and elucidation of the genetic regulatory network (Alon, 2007; Davidson ⁎ Corresponding author. Medical Research Council National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 1AA, UK. E-mail address: [email protected] (J.C. Smith).

and Erwin, 2006; Levine and Davidson, 2005) that is initiated by FGF signalling. FGF family members signal through receptor tyrosine kinases. The receptors go on to activate several signal transduction pathways, prominent amongst which is the MAP kinase pathway, which includes molecules such as Ras, Raf, Mek and Erk, all of which, in Xenopus, display loss-of-function phenotypes similar to that of FGF (Gotoh et al., 1995; LaBonne et al., 1995; Umbhauer et al., 1995). One target of MAP kinase is the SRF-interacting Ets transcription factor Elk-1 (Gille et al., 1995; Hipskind et al., 1994; Sharrocks, 1995), which activates immediate–early genes such as Egr-1, whose product is a transcription factor that can act both as a repressor and as an activator of transcription (Bahouth et al., 2002; Chapman and Perkins, 2000; Lemaire et al., 1990; Wang et al., 2005). Although a dominant-negative derivative of human Elk-1 is known to block the induction of XEgr-1 by FGF in isolated animal pole regions (Panitz et al., 1998), no Xenopus homologue of Elk-1 has yet been isolated and its function has not been analyzed. Thus, in the first part of this paper we show that Xenopus Elk-1 (XElk-1) is expressed ubiquitously in the early Xenopus embryo, but that the phosphorylated form of the protein is enriched in the dorsal marginal zone. Interference with XElk-1 activity causes embryos to resemble those in which FGF signalling has been disrupted; in particular, they are truncated posteriorly and muscle and notochord differentiation are compromised. In addition, expression of XEgr-1 is down-regulated.

0012-1606/$ – see front matter. Crown Copyright © 2009 Published by Elsevier Inc. All rights reserved. doi:10.1016/j.ydbio.2009.09.039

314

O. Nentwich et al. / Developmental Biology 336 (2009) 313–326

XEgr-1 is activated directly by a ternary complex of XElk-1 and XSRF and it is expressed, like Xbra (Smith et al., 1991), throughout the marginal zone of the embryo (Panitz et al., 1998). Surprisingly, although XEgr-1 and Xbra are co-expressed, over-expression of XEgr-1 inhibits both the endogenous expression of Xbra and its expression in animal pole regions in response to activin or FGF. However, XEgr-1 is not a general repressor of mesoderm formation because it can activate, in a direct fashion, the expression of the myogenic regulatory gene XMyoD. We discuss our results in terms of a genetic regulatory network consisting of modified incoherent feed-forward loops (Alon, 2007) involving eFGF, XEgr-1, Xbra and XMyoD.

Xenopus tropicalis experiments were carried out as described (Rana et al., 2006). Antisense morpholino oligonucleotides (MOs) were from GeneTools (Philomath, OR, USA). They were dissolved in distilled water and stored at −80°C. Sequences were as follows: XtElk-1 MO: 5′-TCCATGACTGCGGGAGCAAAGAGAC-3′; XtEgr-1 MO1: 5′-CCCTAAGGGTGAATGGTGCTGCCGC-3′; XtEgr-1 MO2: 5′-GCTTGGCACAGTGAGGGGAGACAGG-3′; Control MO 5′-CCTCTTACCTCAGTTACAATTTATA-3′. Animal cap assays and growth factor treatments were carried out as described (Smith, 1993). Explants in which protein synthesis was inhibited were pre-treated for 20 min in 10 μM cycloheximide before application of additional factors (2 μM dexamethasone; 25 U/ml activin).

Materials and methods Constructs Xenopus embryo manipulations Xenopus laevis embryos were obtained by in vitro fertilization (Smith, 1993) and staged according to Nieuwkoop and Faber (1975).

All PCR amplifications for plasmid construction were carried out using Pfu DNA polymerase and confirmed by sequencing. The coding sequence of XElk-1 was isolated by RT-PCR using cDNA from embryos

Fig. 1. Spatiotemporal distribution of XElk-1 mRNA and protein. (A) Schematic representation of XElk-1 functional domains. See text for details. (B) Temporal expression of XElk-1; transcripts are maternally expressed and are present throughout early development. These data are representative of two independent experiments. Error bars show standard deviations of technical replicates of a single experiment. (C) Spatial distribution of XElk-1 mRNA in transverse sections of an albino embryo at the neurula stage reveal XElk-1 expression in notochord and somitic mesoderm. (D) Comparison of the amino acid sequences of human (h), Xenopus laevis (Xl) and Xenopus tropicalis (Xt) Elk-1 in the vicinity of human serine 383, which is phosphorylated by MAP kinase. (E) Western blot analysis of XElk-1 phosphorylation in response to FGF. Animal caps derived from embryos that had been injected with 400 pg myc-tagged XElk-1 mRNA were treated with 200 ng/ml bFGF for the indicated times and then subjected to western blotting using an antibody specific for phosphorylated XElk-1 (α-pElk-1) or a pan-Elk-1 antibody (α-Elk-1). Levels of phosphorylated XElk-1 are high at time zero, perhaps as a response to dissection (Christen and Slack, 1999; Krain and Nordheim, 1999) but decline by 45 min unless bFGF is present. (F–J) Immunocytochemical analysis of phosphorylated XElk-1 protein. (F, G) Sagittal sections of Xenopus embryos at the early gastrula stage. Control embryos incubated in the absence of primary antibody show only non-nuclear staining in cells of the yolky vegetal hemisphere. Use of an antibody directed against phosphorylated XElk-1 protein reveals staining in nuclei of the animal hemisphere and of the ventral and dorsal mesoderm, especially in the vicinity of the dorsal blastopore lip (arrow, G). (H–J) Whole mount immunocytochemistry showing the distribution of phosphorylated XElk-1 at gastrula (H), neurula (I) and tailbud (J) stages. pXElk-1 is enriched in the dorsal midline and extends towards the presumptive forebrain as gastrulation proceeds (arrows, H, I). At the early tailbud stage high levels of pXElk-1 are present in the presomitic mesoderm (arrow, J).

O. Nentwich et al. / Developmental Biology 336 (2009) 313–326

at stage 12. Primers were forward: 5′-TCTAGAATGGACCCCGCTGGG-3′; reverse: 5′-GAATTCATTATGGCTTTTGAAGTCCAGG-3′. The resulting PCR fragment was cloned into the EcoRV and SpeI sites of pT7TS. XElk-1mt was created by PCR using 5′-GGACTAGTTCAGGTCCTCCTCGGAGATCAGCTTCTGCTCTGGCTTTTGAAGCCCAGG-3′ as reverse primer. XElk-1-EnR was cloned by excising a BbsI/EcoRI fragment from pT7TSXElk-1 and inserting a ClaI/XbaI fragment from Xbra-EnR that had previously been modified by insertion of 5 copies of the myc epitope. ΔEts-EnR was constructed by deleting a BlpI/MspAII fragment from XElk-1-EnR. All XElk-1 constructs were linearized with XbaI and mRNA synthesis was carried out using T7 RNA polymerase. XEgr-1-GR was constructed by amplifying the XEgr-1 open reading frame using forward primer 5′-TTTACTAGTCTCCCTAGGGATTCCCGAGA-3′ and reverse primer 5′-AAAAAAGCGGCCGCGCAAATCTCAATTGTCCTTGG-3′. To create ΔXEgr-1-GR, the reverse primer was 5′-AAAAGCGGCCGCGCAACTCAGAGGGGGGCTCT-3′. The resulting fragments were cloned into the SpeI/NotI sites of a modified pSP64TGR (Tada et al., 1997). mRNA encoding SRF-VP16 (Dalton and Treisman, 1992) was transcribed from pCS2+SRF-VP16 (Panitz et al., 1998). A constitutively active (ca) form of Elk-1 (VP16-Elk1) (Hill et al., 1994) was transcribed from pT7ELKVP16. Chromatin immunoprecipitation Chromatin immunoprecipitation was performed as described (Schratt et al., 2004). Two hundred animal caps per sample were cross-linked in 0.5× MMR, 1% formaldehyde for 15 min. Sonication was carried out in a volume of 3 ml at 4°C using a microtip and a Misonix Sonicator 3000. Sonication buffer consisted of 21 mM Tris/ HCl pH 8.0, 3.6 mM EDTA, 0.3% SDS, 0.36 mM EGTA, 3 μg/ml Aprotinin, 2 μM Pepstatin and 1 mM PMSF. Anti-SRF (Santa Cruz G-20) and antipS383Elk-1 (NEB/CST 9186) antibodies were used at dilutions of 1:500. Incubations were carried out overnight. The equivalent of 25 explants were used in PCR reactions. Primers used to detect XEgr-1 genomic sequence (AF250346) by quantitative PCR were as follows. XEgr-1 CArG I forward: 5′-GCTCCAGCACCTCATCAGC-3′; reverse: 5′-AGCAGCCATGGATTCTACCG-3′. XEgr-1 CArG II forward: 5′-CTGGTTCCGAAGGGTTTGC-3′; reverse: 5′-CGTAGCCTTCAATCTCCTCCC-3′. Primers specific for the XeFGF T-box site (AF078081) were forward: 5′-AGGCAGAAGTATCACAGCAGT-3′; reverse: 5′-AATGGGCTCCGTCAGAGCAGA-3′. Whole mount RNA in situ hybridization embryo sections and antibody staining In situ hybridization was performed as described (Harland, 1991; Steinbach et al., 1998). Plasmids for synthesis of digoxigenin- or FITC-

Fig. 2. Impairment of embryonic development by interference with Elk-1 activity. (A) The XElk-1 constructs used in these experiments. Top: myc-tagged XElk-1; middle: myctagged XElk-1-EnR; bottom: ΔEtsXElk-1-EnR. (B) Injection of XElk-1-myc mRNA (0.5 ng into each cell of the two-cell stage; total 1 ng) has no effect on early Xenopus development. (C, D) A non-tagged version of XElk-1 also has no effect on development. Compare C (uninjected) with D (injected). (E–G) Interference with XElk-1 activity by means of a dominant-interfering construct disrupts Xenopus development. (E) Uninjected embryos. (F) Embryos injected with 750 pg RNA encoding XElk-1-EnR; such embryos form a dorsally curved trunk and open neural folds. (G) Embryos injected with 750 pg RNA encoding ΔEtsXElk-1-EnR appear normal. (H–J) Trunk and anterior defects caused by XElk-1-EnR (I) are rescued, at least partially, by co-expression of X-Elk1 mRNA (J). (K–N) Notochord (marked by MZ15) and muscle (marked by 12/101) differentiation is diminished in embryos in which XElk-1 activity is perturbed. Antibody staining was carried out on 50-μm sections of stage 35 embryos previously injected with 750 pg RNA encoding XElk-1-EnR. Little notochord can be detected in embryos expressing XElk-1-EnR (L; compare with control embryo in K). Muscle differentiation is also greatly inhibited (N; compare with control embryo in M). (O, P) Xbra expression is also greatly inhibited in embryos injected with 750 pg RNA encoding XElk-1-EnR (O; compare with control embryos in P). All results were obtained in three independent experiments with over 200 embryos examined per treatment, except for data in panels C and D, which were derived from two experiments with 60 embryos in each case.

315

labelled antisense RNA probes used were pSP72-Xbra (Smith et al., 1991), pBKCMV-XEgr-1 (Panitz et al., 1998) and pCS2+XMyoDb (Wittenberger et al., 1999). Immunocytochemistry was carried out on whole embryos using polyclonal antibodies against phospho-Ser383Elk-1 (NEB/CST 9181) with alkaline phosphatase coupled secondary antibodies and BM-Purple substrate (Roche). The monoclonal antibodies 12/101 (Kintner and Brockes, 1984) and MZ15 (Smith and

316

O. Nentwich et al. / Developmental Biology 336 (2009) 313–326

Watt, 1985; Zanetti et al., 1985) were used as described (Fainsod et al., 1995), employing an HRP coupled secondary antibody with DAB substrate (Sigma). Western blot analysis Embryos were lysed in lysis buffer as described (Zetser et al., 2001). Western blotting was carried out following polyacrylamide gel electrophoresis and used anti-Elk-1 (NEB/CST 9182), anti-pS383Elk-1 (NEB/CST 9181), anti-diphosphorylated ERK 1 and 2 (Sigma M8159) and anti-pan ERK 1 and 2 (BD Transduction Lab 610123) antibodies. Electrophoretic mobility shift assay (EMSA) EMSA was performed as described (Weinhold et al., 2000). In vitro translated protein was generated using the TnT coupled in vitro transcription translation system (Promega). Proteins were synthesized from pSR64T-XSRF (Mohun et al., 1991) and pT7TS-XElk-1 templates using SP6 RNA polymerase. Oligonucleotides for EMSA were as described (Panitz et al., 1998).

reverse: 5′-AACTTCCTATGGCAGAAATGGC-3′; XmAct forward: 5′-GCTGACAGAATGCAGAAG-3′; reverse: 5′-TTGCTTGGAGGAGTGTGT-3′. XMyoDb forward: 5′-AACTGCTCCGATGGCATGATGGATTA-3′; reverse: 5′-ATTGCTGGGAGAAGGGATGGTGATTA-3′. Xmyf5 forward: 5′CCCTCAATGGTCTGGAAGAA-3′; reverse: 5′-CGGGGTGATAGAGTCTGGAA-3′. XpMes1 forward: 5′-GATTCTGCAGGAGCTGAGGAC-3′; reverse: 5′-GCATGGCAGGGGTACACAGAC-3′. Quantitative PCR was carried out using Lightcycler 480 Mastermix (Roche) according to the manufacturer's instructions. Samples were run in 384-well plates in Lightcycler 480 (Roche) on standard run templates. Primers were as follows. Histone H4: see above. Xbra forward: 5′-GAATGAGCTCCAGGCTGGC-3′; reverse: 5′-TCATCTCGTTGGTGAGCTCCT-3′. XEgr-1 forward: 5′-ACCATCAAGGCCTATGCAAC-3′; reverse: 5′-GCAACCATATGGCCTCTCAT-3′. XMyoD forward: 5′-GCTGGTTGCTGAATTTCCAT-3′; reverse: 5′-TCAACACAACATTGGCAGGT-3′. XtBra forward: 5′-AGACATCTTGGATGAGGG-3′; reverse: 5′-GAAGGGTACTGACTTGAG-3′. XtODC forward: 5′-GCCATCGTGAAGACTCTCTCCC-3′; reverse: 5′-TTCGGGTGATTCCTTGCCAC-3′. XtMyoD forward: 5′-AGGTCAATGAGGCGTTTGAG-3′; reverse: 5′-CAGAGTCCCCGCTATAATGC-3′. Results

Quantitative and semi-quantitative PCR and RT-PCR Identification of XElk-1 and spatiotemporal localization of pXElk-1 RNA preparation, cDNA synthesis and semi-quantitative RT-PCR were performed as described (Steinbach and Rupp, 1999). Primers were as follows. Goosecoid forward: 5′-TTCACCGATGAACAACTGGA-3′; reverse: 5′-TTCCACTTTTGGGCATTTTC-3′; Histone H4 forward: 5′CGGGATAACATTCAGGGTATCACT-3′; reverse: 5′-ATCCATGGCGGTAACTGTCTTCCT-3′. XElk-1 forward: 5′-CTGTGATTGGCCAGGA-3′;

In an effort to understand the genetic regulatory network that is initiated and maintained by FGF signalling, we first derived a cDNA encoding Xenopus laevis Elk-1. Comparison of the amino acid sequence of XElk-1 with its mouse and human orthologues shows that the four known functional domains of this protein are conserved

Fig. 3. An antisense morpholino oligonucleotide targeted against X. tropicalis Elk-1 has less effect on gene expression than does XElk-1-EnR but does interfere with muscle and notochord differentiation. (A, C) Embryos injected with a control morpholino oligonucleotide showing expression of Xbra (A) and XtEgr-1 (C). (B, D) Embryos injected with an antisense morpholino oligonucleotide directed against XtElk-1. Note delay in blastopore closure caused by the MO (compare A and B). Expression of Xbra is only slightly affected by XtElk-1 MO (B) but expression of XtEgr-1 is more significantly reduced. (E, F) Expression levels of XtMyoD are substantially decreased by XtElk-1 MO. (E) An embryo injected with a control MO; (F) an embryo injected with an antisense morpholino oligonucleotide directed against XtElk-1. (G, J) Control embryos at tailbud stages stained with 12/101 (G) or MZ15 (J). (H, K) Embryos injected with an antisense morpholino oligonucleotide targeted against XtElk-1. Note the reduction in muscle (H) and notochord (K) differentiation. (I, L) Rescue of the effects of the MO targeted against XtElk-1. Muscle (I) and notochord (L) formation are both restored in embryos injected with XElk-1 mRNA as well as XtElk-1 MO. Results in G-L were obtained in three independent experiments. Eighty-five embryos were injected with our control MO, of which 90% appeared normal; 109 were injected with XtElk-1 MO, of which 13% appeared normal; and 118 were injected with XtElk-1 MO together with XElk-1-myc mRNA, of which 79% appeared normal.

O. Nentwich et al. / Developmental Biology 336 (2009) 313–326

317

Fig. 4. Binding of XElk-1 to its putative target sequence and analysis of XEgr-1 expression in embryos in which XElk-1 function is perturbed. (A) Binding of XElk-1 to its putative binding sites in the promoters of XEgr-1 and hc-fos. 32P-labelled double-stranded DNA probes (30,000 cpm) were incubated with in vitro translated proteins. XElk-1 is unable to bind to DNA on its own (lanes 2 and 7) but forms a ternary complex with XSRF (lanes 4 and 9). This complex can be super-shifted using antibodies recognising Elk-1 (lanes 5, 10). Lanes 1 and 6 were loaded with mock lysates. (B) Positions of the CArG boxes in the XEgr-1 promoter. Arrows indicate the positions of PCR primers for the CArG boxes and for the XEgr-1 open reading frame (ORF). (C) Chromatin immunoprecipitation of endogenous pXElk-1 bound to the XEgr-1 SRE. Chromatin derived from untreated animal caps (first four lanes) or from animal caps treated with 200 ng/ml bFGF (second four lanes) was immunoprecipitated using the indicated antibodies. Samples containing 10% of non-immunoprecipitated sample material served as input control for sample integrity and PCR (first and fifth lanes). Locations of primers are indicated in (B), with primers recognising sequence around the XeFGF T-Box, which lacks SRF/elk binding sites, serving as a negative control. The weak α-phospho Elk-1 band on the CArG1 element is likely to derive from the activation of the MAP kinase pathway that occurs following dissection of animal pole regions (Krain and Nordheim, 1999; LaBonne and Whitman, 1997). Bands in the ORF lanes are background. (D) The structure of SRF-VP16. (E) SRF-VP16 increases XEgr-1 expression at the early gastrula stage. Embryos at the four-cell stage were injected in a single blastomere with mRNA encoding nuclear β-Gal together with mRNA encoding SRF-VP16, and specimens were processed for in situ hybridization at the early gastrula stage. Development time was adjusted such that ectopic expression of XEgr-1 (blue) appeared weak enough to allow visualisation of LacZ (red). Note that XEgr-1 is activated in a non-cell-autonomous fashion. (F) The structures of XElk-1-EnR and of XElk-1ΔEts-EnR-Myc. (G, H) Dominant interfering Elk-1 blocks expression of XEgr-1. Embryos at the four-cell stage were injected in a single blastomere with mRNA (250 pg) encoding XElk-1-EnR (G) or XElk-1ΔEts-EnR-Myc (H), together with mRNA encoding LacZ. They were fixed at the early gastrula stage and analyzed by in situ hybridization. LacZ staining appears red. Note that XElk-1-EnR (G), but not XElk-1ΔEts-EnR-Myc (H) causes down-regulation of XEgr1 in a non-cell-autonomous manner. Xbra is also down-regulated in such experiments (I, J).

318

O. Nentwich et al. / Developmental Biology 336 (2009) 313–326

O. Nentwich et al. / Developmental Biology 336 (2009) 313–326

(Fig. 1A). The N-terminal Ets domain (A-box: amino acids 1–90) mediates DNA binding while the B-box (amino acids 130–153) interacts with SRF. Transactivation and interaction with MAP kinase occur at the D-box (amino acids 299–311) and C-box (amino acids 332–382). RT-PCR analysis (Fig. 1B) reveals that XElk-1 is expressed maternally and that transcripts are present throughout early Xenopus development. In situ hybridization shows that this expression occurs throughout the embryo at gastrula and early neurula stages (data not shown) but is enriched in dorsal mesodermal structures by the late neurula stage (Fig. 1C). The induction of immediate–early gene expression by human Elk-1 involves the phosphorylation of serine 383 by the MAP kinase Erk-1 (Janknecht et al., 1993). To ask whether the same is true of the orthologous serine 366 of Xenopus Elk-1 (Fig. 1D), animal pole regions derived from embryos previously injected with RNA encoding a myc-tagged form of XElk-1 (XElk-1-myc; Fig. 2A) were treated with bFGF. Protein extracts were analyzed by western blotting using an antibody specific for phosphorylated Elk-1. Fig. 1E shows that bFGF treatment causes significant enrichment of levels of phosphorylated XElk-1 within 45 min. Similar results were obtained using an untagged version of Xelk-1 (Fig. S1). In contrast to Erk-1 (Fig. S1), we note that a significant proportion of Elk-1 is phosphorylated in the absence of FGF (Figs. 1E and S1). Consistent with this observation, immunocytochemical analysis detects phosphorylated Elk-1 in the animal pole region of the Xenopus embryo (Figs. 1F and G), where levels of FGF signalling are low (Christen and Slack, 1999; Cordenonsi et al., 2007; LaBonne et al., 1995; Schohl and Fagotto, 2002). The highest levels of the phosphorylated form of XElk-1 (pXElk-1) are, however, present in the dorsal marginal zone of the early gastrula (Fig. 1G). By the late gastrula stage pXElk-1 is enriched in the dorsal midline (Figs. 1H and I). In early tailbud embryos XElk-1 phosphorylation expands from the closed blastopore towards the presumptive forebrain and into the presomitic mesoderm (Fig. 1J), overlapping with domains of MAP kinase activity (Christen and Slack, 1999; Curran and Grainger, 2000). The spatial and temporal expression patterns of pXElk-1 differ from those of XElk-1 mRNA, suggesting that the activity of XElk-1 is regulated post-translationally. Disruption of mesodermal structures caused by XElk-1 loss of function Over-expression of neither myc-tagged (Figs. 2A and B) nor untagged (Figs. 2C and D) XElk-1 has any effect on early development of the Xenopus embryo, consistent with the idea that its activity is regulated post-translationally. In an effort to inhibit the function of XElk-1, we first created a form of the protein in which the C-terminal transactivation domain is replaced by the repressor domain of the Drosophila engrailed protein (Fig. 2A; XElk-1-EnR) (Taylor et al., 1996). As a control, we also created a version of this protein in which the Ets domain, which binds DNA, is deleted (Fig. 2A; XElk-1ΔEtsEnR). Both proteins were tagged with a myc epitope, and whole

319

mount immunocytochemistry indicated that the two are expressed at similar levels (data not shown). The phenotypes of embryos injected with RNA encoding XElk-1-EnR resembled those caused by loss of function of FGF (Amaya et al., 1993) or of components of the FGF signal transduction pathway such as Ras, Raf and Erk (LaBonne et al., 1995; LaBonne and Whitman, 1994; Umbhauer et al., 1995; Whitman and Melton, 1992). Thus, embryos displayed gastrulation defects (not shown) and a shortening of the body axis (Figs. 2F and I). No defects were observed following injection of RNA encoding XElk-1ΔEts-EnR (Fig. 2G), and the phenotype was partially rescued by co-injection of RNA encoding wild-type XElk1 (Figs. 2H–J), arguing that the observed effects are specific. Muscle and notochord differentiation was defective (Figs. 2K–N), and expression of Xbra in embryos expressing XElk-1-EnR declined during neurula stages (Figs. 2O and P), consistent with the observation that FGF signalling is required for the maintenance of Xbra. This inhibition of Xbra expression begins at the early gastrula stage (see Figs. 4I and J). To confirm the idea that XElk-1 causes developmental defects that resemble those caused by loss of function of FGF, we turned to antisense morpholino oligonucleotides (MOs). The genome of the pseudotetraploid species Xenopus laevis contains at least two XElk-1 alleles (Nentwich and Nordheim, unpublished observation), so to simplify our analysis we designed an MO directed against the translational start site of Elk-1 derived from the diploid species X. tropicalis (XtElk-1; see Materials and methods). Injection of this MO caused a delay in blastopore closure at the early gastrula stage and a slight decrease in Xbra and XtEgr-1 expression (Figs. 3A–D). At the late gastrula stage, expression of XtMyoD was significantly reduced (Figs. 3E and F). At later stages our XtElk-1 antisense morpholino oligonucleotide caused truncated embryos to form and a disruption of both muscle (Fig. 3H) and notochord development (Fig. 3K). Our results suggest that this phenotype is specific, because it can be partially rescued by co-injection of XElk-1 mRNA which lacks the morpholino binding site (Figs. 3I and L). For example, in the case of the notochord, the poorly defined and fragmented MZ15 staining caused by XtElk-1 MO is clearly rescued by co-injection of XElk-1 mRNA (Figs. 3K and L). The MO phenotype is weaker than that caused by injection of XElk-1-EnR RNA, perhaps because the knock down is incomplete or perhaps because of persistence of maternal XElk-1 protein (Eisen and Smith, 2008). It is unlikely that the weaker phenotype is due to another SRF-interacting factor such as XSap-1 (Nentwich et al., 2001) substituting for XElk-1 because we do not observe synergistic effects following co-injection of MOs targeted against XtElk-1 and X. tropicalis Sap-1 (data not shown). XElk-1 activates XEgr-1 directly In previous work we have suggested that a Xenopus homologue of Elk-1 might activate the zinc finger-containing transcription factor XEgr-1, an immediate–early gene expressed in the marginal zone of

Fig. 5. Effects of gain and loss of XEgr-1 function on early Xenopus development. (A–D) XEgr-1 over-expression perturbs transcription of Xbra in the marginal zone. Embryos were injected at the one-cell stage with the indicated doses of XEgr-1 mRNA. They were cultured to stage 11 and then analyzed by in situ hybridization for expression of Xbra. As levels of XEgr-1 increase, transcription of Xbra decreases. (E–G) Similar experiments reveal that XMyoD is up-regulated in such embryos. (E) An embryo showing the combined endogenous expression of both Xbra and XMyoD. (F) An embryo previously injected with XEgr-1 mRNA into one blastomere at the four-cell stage showing down-regulation of Xbra. (G) An embryo previously injected with XEgr-1 mRNA into one blastomere at the four-cell stage showing the combined expression of Xbra and XMyoD. Note the down-regulation of Xbra and XMyoD in the marginal zone and the ectopic expression of XMyoD in the animal hemisphere. (H–K) Embryos injected with 1 ng XEgr-1 mRNA were allowed to develop to stage 26, when formation of muscle and notochord was analyzed using antibodies 12/101 and MZ15, respectively. Over-expression of XEgr-1 causes reduction in both tissues. (L–S) Loss of Xenopus tropicalis Egr-1 activity causes elevated levels of Xtbra. X. tropicalis embryos were injected with 30 ng XtEgr-1 MO2 and analyzed for expression of Xtbra throughout gastrula and neurula stages. Marginal zone expression of Xtbra is unaffected during gastrula stages (L–N, P–R) but by stage 15 Xtbra is higher in embryos lacking XtEgr-1 than in control embryos (O, S). Loss of X. tropicalis Egr-1 activity causes down-regulation of XtMyoD at the late gastrula stage. Expression of XtMyoD is normal in uninjected embryos (T) and embryos injected with a control morpholino oligonucleotide (U) but is down-regulated in embryos injected with antisense morpholino oligonucleotides directed against X. tropicalis Egr-1 (V, W). (X) Inhibition of X. tropicalis Egr-1 function causes down-regulation of XtMyoD and up-regulation of Xtbra. Embryos were left uninjected (Uninj), were injected with a control morpholino oligonucleotide (Con) or were injected with antisense morpholino oligonucleotides directed against X. tropicalis Egr-1 (MO1, MO2). They were assayed for expression of XtMyoD at the neurula stage. Note the down-regulation of XtMyoD in embryos in which XtEgr-1 function is compromised. Expression of Xtbra is up-regulated in embryos injected with XtEgr-1 MO2 (which yields a stronger phenotype than MO1). (Y–EE) Loss of XEgr-1 function causes trunk defects and perturbs notochord and muscle formation. X. tropicalis embryos were injected with 30 ng control MO (Y, BB, DD), 30 ng XtEgr-1 MO1 (Z) or 20 ng XtEgr-1 MO2 (AA, CC, EE) and cultured to stage 26. Embryos lacking XtEgr-1 display defects in trunk development, which are particularly marked in embryos injected with XtEgr-1 MO2. Muscle (BB, CC) and notochord (DD, EE) differentiation are perturbed in such embryos. The figure shows representative results from three independent experiments: n N 70.

320

O. Nentwich et al. / Developmental Biology 336 (2009) 313–326

the early Xenopus embryo (Panitz et al., 1998; Saka et al., 2000). The human Egr-1 gene is a target of MAP kinase, SRF and Elk signalling and activation of Xenopus Egr-1 by Activin, BMP-4 and XeFGF is inhibited by a truncated form of Elk-1 that lacks the C and D boxes (Panitz et al., 1998). The XEgr-1 promoter contains conserved SRF/Elk binding sites (Panitz et al., 1998) and we show here by means of electrophoretic mobility shift assays that in the presence of XSRF, in vitro translated XElk-1 can bind the XEgr-1 serum response element (SRE; Fig. 4A). As observed with human Elk-1 (Hill et al., 1993) XElk-1 cannot bind the XEgr-1 SRE in the absence of XSRF (Fig. 4A). Chromatin immunoprecipitation experiments using antibodies specific for SRF and for phosphorylated Elk-1 reveal that pXElk-1 interacts with the XEgr-1 SRE in FGF-treated animal caps whereas in untreated explants no pXElk-1 can be precipitated from its target sequence (Fig. 4C lanes 4 and 8). In contrast, XSRF containing complexes can be precipitated from the XEgr-1 SRE in both FGFtreated and -untreated animal caps, indicating that XSRF occupies the XEgr-1 SRE in a constitutive manner (Fig. 4C lanes 3 and 7). One interpretation of these experiments is that, as in primary fibroblasts (Li et al., 2003), a complex of XElk-1 and SRF binds its target sequence in a constitutive manner, and that the complex is activated by phosphorylation of XElk-1. Unfortunately, we have not been able to confirm this model because we have been unable to precipitate XElk-1 protein from the XEgr-1 SRE using the a pan-Elk-1 antibody. These results indicate that XSRF and XElk-1 regulate the expression of XEgr-1 in the early mesoderm. Consistent with this idea, XEgr-1 is first expressed in the dorsal blastopore lip, where levels of pXElk-1 are highest (Fig. 1G). Furthermore, a constitutively active version of SRF in which the viral VP16 transactivation domain replaces the C-terminus of hSRF (SRF-VP16; Fig. 4D) enhances XEgr-1 expression in an apparently non-cell-autonomous fashion (Fig. 4E). We also find, consistent with the observation that a truncated form of human Elk1 prevents the activation of XEgr-1 by XeFGF (Panitz et al., 1998), that XElk-1-EnR but not XElk-1ΔEts-EnR (Fig. 4F) abolishes expression of XEgr-1 during gastrula stages (Figs. 4G and H). Such embryos show a delay in blastopore closure and, interestingly, greater than 90% show down-regulation of XEgr-1 in a non-cell-autonomous fashion (Fig. 4G), suggesting there is an additional route by which XElk-1 regulates XEgr-1. Similar results are observed in embryos injected with SRF-EnR (data not shown). As described above, we also note that XElk-1-EnR inhibits the early expression of Xbra (Figs. 4I and J). XEgr-1 is a negative regulator of Xbra expression Our results indicate that XElk-1 acts downstream of the FGF receptor to regulate the expression of XEgr-1 in a direct manner. To

Fig. 6. XEgr-1 represses gene activation in response to activin and FGF. (A) Induction of Goosecoid (Gsc) in response to activin is reduced in animal caps derived from embryos previously injected with RNA encoding XEgr-1. Animal pole regions were dissected at stage 8 and 9 and cultured to stage 11 before being assayed by PCR for expression of Gsc and Histone H4. (B) A hormone-inducible version of XEgr-1 was constructed by fusing the open reading frame of XEgr-1 to the ligand binding domain of the glucocorticoid receptor. An XEgr-1-GR fusion lacking the DNA binding and transactivation domains of XEgr-1 (ΔXEgr-1-GR) served as a negative control. (C) In vitro translation of mRNAs encoding XEgr-1-GR and ΔXEgr-1-GR demonstrates that they are translated with similar efficiency. (D) XEgr-1 is a direct repressor of FGF-induced Xbra expression. Animal pole regions derived from embryos previously injected with RNA encoding either XEgr-1-GR or ΔXEgr-1-GR were treated at stage 9 with bFGF (200 ng/ml), Dexamethasone (Dex: 2 μM) or Cycloheximide (CHX: 10 μM) as indicated. They were then analyzed for Xbra expression at stage 11.5. Xbra is activated by bFGF in tissue expressing ΔXEgr-1-GR in the absence or presence of CHX and Dex, both individually and in combination. In contrast, Dex treatment of animal pole regions expressing XEgr1GR blocks the activation of Xbra even in presence of CHX. (E) Animal caps were treated as in panel D but with 16 U/ml activin for 2 h instead of bFGF. In tissue expressing ΔXEgr-1-GR, high levels of Xbra are induced by activin only if CHX is absent, but significant expression does occur in the presence of Dex and CHX (final pair of samples), and this expression is inhibited in animal pole regions expressing XEgr-1-GR. Note weak induction by cycloheximide alone in the third pair of samples.

investigate the function of XEgr-1 during Xenopus development, we first carried out experiments in whole embryos. To our surprise, because misexpression of FGF causes ectopic expression of Xbra (Smith et al., 1991), over-expression of XEgr-1 caused downregulation of Xbra at gastrula stages (Figs. 5A–D), with such embryos also showing up-regulation of XMyoD (Figs. 5E–G). By later stages there seemed to have been some recovery, because although expression of the muscle-specific marker 12/101 was reduced, the notochord-specific MZ15 was only slightly affected (Figs. 5H–K). Consistent with the observation that over-expression of XEgr-1 down-regulates expression of Xbra in whole embryos, loss of XEgr-1 function causes a slight up-regulation of Xbra. In these experiments, two non-overlapping MOs targeting the translation start site of X. tropicalis Egr-1 (XtEgr-1) mRNA were injected individually into X. tropicalis embryos at the one-cell stage, and the embryos were examined by in situ hybridization at stages 10.5, 11, 12 and 15

O. Nentwich et al. / Developmental Biology 336 (2009) 313–326

321

Fig. 7. Direct regulation of XMyoD by XEgr-1. (A) Direct activation of XMyoD by XEgr-1. Animal pole regions derived from embryos injected with RNA encoding ΔXEgr-1-GR or XEgr1-GR were treated at stage 8 with Dex or CHX or both, as indicated. MyoD expression was assayed by quantitative RT-PCR at the equivalent of stage 14. Note that XEgr-1 can activate MyoD even in the presence of CHX (final pair of samples). (B) XEgr-1 can induce premature expression of XMyoD, but not of muscle-specific actin, Xmyf5 or Xpmes-1. Animal pole regions derived from embryos injected with RNA encoding XEgr-1-GR, ΔXEgr-1-GR (ΔGR) or Xbra-GR, as indicated, were treated with Dex at stage 8 and assayed for expression of the indicated genes by semi-quantitative RT-PCR at the indicated stages. For comparison, animal pole regions derived from uninjected embryos were treated with activin at stage 8, and RNA samples derived from whole embryos (WE) were also assayed. Note low maternal levels of XMyoD at stage 8.5, but also that this gene is activated prematurely in samples expressing XEgr-1-GR and treated with Dex. Premature activation does not occur following activin treatment, nor in samples expressing Xbra-GR, nor is it observed for the other genes studied. The slightly premature activation of muscle-specific actin, at stage 12, is likely to be an indirect effect caused by the earlier activation of XMyoD. (C) XMyoD is activated by exposure to XEgr-1 as late as stage 12, by which time competence to respond to activin has been lost. Animal pole regions were treated as in (A), except that treatment with activin or Dex started at stage 8, 10 or 12, as indicated. (D–F) Ectopic activation of XEgr-1 causes misexpression of XMyoD in whole embryos. Xenopus embryos were injected at the one-cell stage with RNA encoding ΔXEgr-1-GR (D) or XEgr-1-GR (E, F) and left untreated (E) or were treated with Dexamethasone (D, F) from stage 12. They were cultured to stage 22 and assayed for expression of XMyoD by in situ hybridization. Note that Dex treatment of embryos expressing XEgr-1-GR causes ectopic activation of XMyoD (F). (G) Activation of XMyoD by Xbra requires XEgr-1 activity. Xenopus tropicalis animal caps were derived from embryos previously injected with mRNA encoding with Xbra-GR, ΔXEgr-1-GR or XEgr-1GR as indicated, as well as XtEgr-1 MO1 or 2 as indicated. Animal caps were treated with Dexamethasone at stage 8.5 and cultured to the equivalent of stage 14 before being assayed for expression of XtMyoD. Note that Xbra-GR activation induces XtMyoD (second lane) and that this activation requires XtEGR-1 activity (third and fourth lanes). Co-expression of XEgr-1GR rescues XtMyoD activation (sixth and seventh lanes).

322

O. Nentwich et al. / Developmental Biology 336 (2009) 313–326

(Figs. 5L–S and data not shown). Of the two MOs, XtEgr-1 MO2 was the more effective (see Figs. 5T–AA), and use of this reagent revealed that although there was little difference in levels of Xbra at earlier stages (Figs. 5L–N and P–R), by stage 15 expression of Xbra was elevated in embryos in which XEgr-1 function was inhibited, particularly in the posterior region of the embryo (Figs. 5O and S). This impression was confirmed by analysis of Xtbra expression by quantitative RT-PCR (Fig. 5X), and in situ hybridization and quantitative RT-PCR experiments also demonstrated that XtmyoD is down-regulated in such embryos (Figs. 5T–X). At later stages, loss of XEgr-1 function caused anterior, dorsal and axial structures to develop poorly, with impairment of both muscle and notochord differentiation (Figs. 5Y–EE). These later defects are likely to be indirect consequences of the earlier disruption of gene expression. Together, these experiments show that loss of XtEgr-1 function causes the up-regulation of Xtbra and the down-regulation of XtMyoD. We went on to investigate the interactions between these three genes by combined gain- and loss-of-function studies. XEgr-1 is a direct repressor of Xbra The experiments described above indicate that XEgr-1 is a negative regulator of Xbra in the intact Xenopus embryo. Similar results were obtained in experiments using isolated animal pole regions: XEgr-1 caused a significant decrease in the expression of Xbra (data not shown, but see below), and of other genes such as goosecoid (Fig. 6A) and Xvent-1 (data not shown), in response to both activin and FGF. The gastrulation-like movements that occur in response to activin were also inhibited by XEgr-1 (data not shown). To ask whether XEgr-1 exerts these effects in a direct manner, we fused the XEgr-1 open reading frame to the ligand binding domain of the human glucocorticoid receptor to create XEgr-1-GR (Fig. 6B). Fusion proteins of this sort remain inactive unless dexamethasone is added to the medium (Kolm and Sive, 1995; Tada et al., 1997). A derivative of XEgr-1-GR lacking the DNA binding zinc finger domain and part of the transactivation domain was used as a control in these experiments (Fig. 6B). This construct is expressed at similar levels to XEgr-1-GR in in vitro transcription/translation reactions (Fig. 6C). Experiments were performed in which animal pole regions were derived from control-injected embryos or embryos that had previously been injected with RNA encoding XEgr-1-GR and were then treated with combinations of activin, bFGF, cycloheximide and dexamethasone. Such experiments demonstrated that XEgr-1-GR can repress induction of Xbra by FGF (Fig. 6D) and by activin (Fig. 6E), even in the presence of cycloheximide. Together, these observations suggest that XEgr-1 acts directly on the Xbra promoter to repress its expression. XEgr-1 is a direct activator of XMyoD The results described above indicate that XEgr-1 represses genes such as Xbra that are expressed in the mesoderm of the Xenopus embryo. In additional experiments, however, we observe that XEgr-1 can activate the expression of XMyoD in isolated animal pole regions and that this activation occurs in a direct fashion (Fig. 7A). In doing so, XEgr-1 acts downstream of Xelk-1 because although XEgr-1 MO2 does not prevent a constitutively active (ca) form of Elk-1 (Hill et al., 1994) from inducing transcription of XtEgr-1 itself in isolated X. tropicalis animal pole regions (Fig. 8A), it does inhibit activation of XtMyoD (Fig. 8B). Remarkably, XEgr-1 can cause premature activation of XMyoD at mid blastula stage 8.5, before the gene is transcribed in the normal embryo (Fig. 7B). This observation stands in contrast to many other experiments in Xenopus, where the time of administration of a stimulus does not affect the stage at which the response occurs (Cooke and Smith, 1990). For example, activin treatment of animal

Fig. 8. Elk-1 induces expression of XMyoD via XEgr-1. Animal pole explants were derived from Xenopus tropicalis embryos previously injected with mRNA (500 pg) encoding a constitutively active (ca) form of Elk-1 together with control morpholino or XEgr-1 MO2. Animal caps were cultured to the equivalent of stage 14 when they were assayed for expression of XtEgr-1 or XtMyoD. caElk-1 activates expression of XtEgr-1, and this activation is not inhibited by XtEgr-1 MO2 (A). Activation of XtMyoD, however, is substantially inhibited in animal pole regions in which XtEgr-1 activity is diminished (B).

pole regions causes the onset of gastrulation movements at stage 10.5 (Symes and Smith, 1987), as well as XMyoD transcription (Steinbach et al., 1998), irrespective of the stage at which activin was administered. Furthermore, the onset of cardiac actin gene expression in animal pole regions always occurs around stage 12 irrespective of the stage at which they were juxtaposed with vegetal pole tissue (Gurdon et al., 1985). Another timing phenomenon in Xenopus concerns the acquisition and loss of competence: the ability of cells to respond to a stimulus (Dale et al., 1985; Grainger and Gurdon, 1989; Green and Smith, 1990). The ability of the XMyoD gene in wild-type animal caps to respond to activin is lost by stage 10 (Steinbach et al., 1997), but we observe that the ability to respond to XEgr-1 persists at least until stage 12 (Fig. 7C). These results further emphasise that XEgr-1 is a direct regulator of MyoD, as do the observations that MyoD expression (Figs. 5T–X) and myogenesis (Figs. 5BB and CC) are down-regulated in embryos injected with MOs targeted against XEgr-1 and that overexpression of XEgr-1 causes ectopic XMyoD expression (Figs. 7D–F). Although misexpression of XEgr-1 can activate XMyoD and subsequently muscle-specific actin (Fig. 7B), it fails to induce other myogenic regulatory factors such as Xmyf5 and it also fails to activate pMesogenin-1, which regulates paraxial mesoderm formation. However, these genes do respond to Xbra and Activin (Fig. 7B and data not shown), as described previously (Lin et al., 2003 and data not shown). Xbra can also activate expression of XMyoD, but in contrast to XEgr-1 it cannot do so prematurely (Fig. 7B) and indeed it requires XEgr-1 function (Fig. 7G). Xbra may, however, play some direct role in activating XMyoD (data not shown) and indeed a zebrafish orthologue of Xbra, Ntl, interacts with the MyoD regulatory region (Morley et al., 2009). In the Discussion of this paper we ask how our results can be integrated into a model for the action of FGF in mesoderm formation in the early Xenopus embryo. Discussion The body plan of the early amphibian embryo is established during late blastula and gastrula stages in response to signalling by members of the TGF-β, FGF, BMP and Wnt families of signalling molecules. The

O. Nentwich et al. / Developmental Biology 336 (2009) 313–326

different signalling pathways establish a complex genetic regulatory network involving positive and negative autoregulation and coherent and incoherent feed-forward loops (Alon, 2007; Casey et al., 1998; Collavin and Kirschner, 2003; Loose and Patient, 2004; Pera et al., 2003; Steinbach et al., 1998; Tanegashima et al., 2000). Although the network is initiated by several distinct signalling molecules, the present work emphasises that there is significant cross talk between their signal transduction pathways. Furthermore, detailed analysis of the network described here reveals some surprising and counterin-

323

tuitive results that shed light upon the dynamics of mesoderm formation. Towards a genetic regulatory network for mesoderm formation The results obtained in the course of these experiments, together with data derived from the work of others, are summarised in Fig. 9A. Networks of this sort are necessarily incomplete, but they provide a valuable starting point for discussion and for designing future

Fig. 9. The positions of XElk-1, XSRF, XEgr-1, Xbra and XMyoD in the genetic regulatory network activated by FGF in the Xenopus embryo. See text for details.

324

O. Nentwich et al. / Developmental Biology 336 (2009) 313–326

experiments. Thus, TGF-β signals derived from the vegetal hemisphere of the embryo cause the nuclear translocation of a Smad2/ Smad4 heteromer that interacts with specific co-activators to induce the expression of genes such as Xbra and goosecoid. Goosecoid encodes a transcriptional repressor, whose targets include Xbra. Xbra encodes a transcriptional activator, amongst whose targets are eFGF (Casey et al., 1998), although eFGF is also a direct target of TGF-β signals (Fisher et al., 2002). Activation of eFGF by Xbra occurs through the two regulatory elements 5′-TCACACCT-3′ and 5′-CCACACCT-3′ (Casey et al., 1998). Once activated by Xbra, eFGF participates in a positive autoregulatory loop in which its secreted product induces expression of Xbra through the MAP kinase pathway (and perhaps through Elk-1; see Fig. 2P and Fig. 4I) (Gotoh et al., 1995; LaBonne et al., 1995; Umbhauer et al., 1995). XEgr-1: downstream of FGF and SRF Another target of FGF signalling is XEgr-1. FGF causes the phosphorylation of XElk-1 on serine 366 (Figs. 1D and E), and phosphorylated XElk-1 then joins XSRF on the XEgr-1 promoter to form a ternary complex that activates XEgr-1 expression (Fig. 4). Although the role of Egr-1 in Xenopus has been little studied, overexpression of SRF is known to interfere with mesoderm formation, while its inhibition expands it (Panitz et al., 1998; Yun et al., 2007). XEgr-1 is activated by SRF, so our observation that over-expression of XEgr-1 inhibits expression of Xbra (Figs. 5A–D), a gene expressed throughout the marginal zone, is consistent with these results, as is the observation that inhibition of XEgr-1 activity elevates Xbra (Figs. 5L–S, X). Similarly, over-expression of SRF inhibits muscle differentiation in the Xenopus embryo (Yun et al., 2007), and the same phenotype is observed following over-expression of XEgr-1 (Figs. 5H and I). But as we discuss in more detail below, this cannot be the whole story. FGF acts as a mesoderm-inducing factor and activates expression of Xbra, so why should XEgr-1, one of its direct targets, go on to inhibit Xbra? And if over-expression of XEgr-1 prevents proper muscle differentiation in Xenopus, why should its loss function have the same effect (Figs. 5BB and CC)? XEgr-1: a key regulator of mesoderm formation Our results show that XEgr-1 has two distinct effects. First, it is a direct activator of XMyoD (Fig. 7A), and thereby an indirect activator of muscle-specific actin. The direct nature of its activity is emphasised by two surprising results. First, use of a hormoneinducible version of XEgr-1 shows that it can cause the premature activation of XMyoD (Fig. 7B). And second, use of the same construct shows that it is able to activate XMyoD at stages when cells are no longer competent to respond to the mesoderm-inducing factor activin (Fig. 7C). Loss of XEgr-1 function in X. tropicalis reduces XMyoD expression (Figs. 5T–X) while over-expression activates it (Fig. 7), emphasising the primacy of XEgr-1 in regulating XMyoD and controlling muscle formation during normal Xenopus development. The ability of XEgr-1 to override the mechanisms that ensure the proper timing of gene expression in the Xenopus embryo is unusual and is reminiscent of the effect of Smarcd, whose over-expression in the zebrafish embryo causes the premature activation of MyoD and Myf5 (Ochi et al., 2008). Smarcd3 is a subunit of the SWI/SNF chromatin-remodelling complex. Intriguingly, it interacts with Ntl, the zebrafish orthologue of Xbra and requires Ntl for its activity (Ochi et al., 2008). This observation is of note when considered alongside the second activity of XEgr-1: the down-regulation of genes such as goosecoid, Xvent-1 and Xbra, which, in at least the case of Xbra, appears to occur in a direct manner. This repression of Xbra by XEgr-1, together with the fact that XEgr-1 can induce XMyoD at stage 8.5,

before significant transcription of Xbra occurs (Piepenburg et al., 2004), suggests that XEgr-1 does not require Xbra to activate XMyoD. Egr-1 was originally characterised as an activator of gene expression (Lemaire et al., 1990), but more recently it has also been shown to act as a repressor (Bahouth et al., 2002; Chapman and Perkins, 2000; Wang et al., 2005). We do not yet know the mechanism by which XEgr-1 regulates its targets in Xenopus, although the X. tropicalis MyoD promoter contains the sequence 5′-GCGGGGGCT-3′ with which human Egr-1 interacts to activate the expression of apolipoptotein A1 (Kilbourne et al., 1995). We have detected no such binding site in the Xenopus laevis (Latinkic et al., 1997) or X. tropicalis Brachyury promoter and do not yet know how XEgr-1 represses Xbra transcription. Like XEgr-1, XMyoD and Xbra are expressed in the prospective mesoderm and all three are activated by FGF signalling (Fisher et al., 2002; Smith et al., 1991). The ability of XEgr-1 simultaneously to activate XMyoD and to repress Xbra is thus rather counterintuitive. It may be informative, however, to discuss them in terms of two genetic circuits, both of which resemble incoherent feed-forward loops (Alon, 2007; Mangan et al., 2006). In the first motif (Fig. 9B), eFGF induces the expression of both XEgr-1 and Xbra. Xbra then induces expression of eFGF, but its transcription is down-regulated by XEgr-1. This motif resembles an incoherent type 1 feed-forward loop (Alon, 2007), with an additional component in which Xbra activates eFGF. Two characteristics of the incoherent type 1 feedforward loop are that it can promote peaks of transcription following the application of an inducer and also that it can accelerate the transcriptional response of the network (Alon, 2007; Mangan et al., 2006). The Xenopus embryo develops at 20–24 °C, at which temperatures reaction rates may be one third of those at 37°C; such genetic circuits may therefore be important in ensuring rapid responses to inductive interactions. The second motif (Fig. 9C) emphasises that XEgr-1 acts both as an activator of XMyoD and as a repressor of Xbra. This motif also includes an autocatalytic loop in which XMyoD regulates its own expression (Steinbach et al., 1998) and a negative feedback loop in which Xbra activates expression of XEgr-1 via induction of XeFGF and XEgr-1 then represses Xbra. Under a scheme such as this Xbra is likely to contribute little to the steady state expression level of XMyoD but may be involved, albeit indirectly, in its activation (Alon, 2007). Consistent with this suggestion, muscle development is only slightly impaired in embryos in which Brachyury function is inhibited (Conlon et al., 1996), and in zebrafish ntl embryos the later phase of MyoD expression appears normal, while the early phase is absent or delayed (Weinberg et al., 1996). Conclusions Our results illustrate the complexity of mesoderm formation in the early amphibian embryo and emphasise the importance of understanding the regulatory networks that underlie early developmental decisions. Results that are apparently counterintuitive might be explained by considering temporal aspects of development and the contributions of different proteins to the initial activation of genes and to the steady state situation. Acknowledgments We thank Linda Bross for initial experiments. This work was supported by the DGF (SFB446/TP7) and by a Wellcome Trust grant awarded to JCS. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ydbio.2009.09.039.

O. Nentwich et al. / Developmental Biology 336 (2009) 313–326

References Alon, U., 2007. Introduction to systems biology: design principles of biological circuits. Chapman and Hall. Boca Raton, London, New York. Amaya, E., Musci, T.J., Kirschner, M.W., 1993. FGF signalling in the early specification of mesoderm in Xenopus. Development 118, 477–487. Bahouth, S.W., Beauchamp, M.J., Vu, K.N., 2002. Reciprocal regulation of beta(1)adrenergic receptor gene transcription by Sp1 and early growth response gene 1: induction of EGR-1 inhibits the expression of the beta(1)-adrenergic receptor gene. Mol. Pharmacol. 61, 379–390. Casey, E.S., O'Reilly, M.A., Conlon, F.L., Smith, J.C., 1998. The T-box transcription factor Brachyury regulates expression of eFGF through binding to a non-palindromic response element. Development 125, 3887–3894. Chapman, N.R., Perkins, N.D., 2000. Inhibition of the RelA(p65) NF-kappaB subunit by Egr-1. J. Biol. Chem. 275, 4719–4725. Christen, B., Slack, J.M., 1999. Spatial response to fibroblast growth factor signalling in Xenopus embryos. Development 126, 119–125. Collavin, L., Kirschner, M.W., 2003. The secreted Frizzled-related protein Sizzled functions as a negative feedback regulator of extreme ventral mesoderm. Development 130, 805–816. Conlon, F.L., Sedgwick, S.G., Weston, K.M., Smith, J.C., 1996. Inhibition of Xbra transcription activation causes defects in mesodermal patterning and reveals autoregulation of Xbra in dorsal mesoderm. Development 122, 2427–2435. Cooke, J., Smith, J.C., 1990. Measurement of developmental time by cells of early embryos. Cell 60, 891–894. Cordenonsi, M., Montagner, M., Adorno, M., Zacchigna, L., Martello, G., Mamidi, A., Soligo, S., Dupont, S., Piccolo, S., 2007. Integration of TGF-beta and Ras/MAPK signaling through p53 phosphorylation. Science 315, 840–843. Curran, K.L., Grainger, R.M., 2000. Expression of activated MAP kinase in Xenopus laevis embryos: evaluating the roles of FGF and other signaling pathways in early induction and patterning. Dev. Biol. 228, 41–56. Dale, L., Smith, J.C., Slack, J.M.W., 1985. Mesoderm induction in Xenopus laevis: a quantitative study using a cell lineage label and tissue-specific antibodies. J. Embryol. Exp. Morphol. 89, 289–312. Dalton, S., Treisman, R., 1992. Characterization of SAP-1, a protein recruited by serum response factor to the c-fos serum response element. Cell 68, 597–612. Davidson, E.H., Erwin, D.H., 2006. Gene regulatory networks and the evolution of animal body plans. Science 311, 796–800. Eisen, J.S., Smith, J.C., 2008. Controlling morpholino experiments: don't stop making antisense. Development 135, 1735–1743. Fainsod, A., Steinbeisser, H., De Robertis, E.M., 1995. On the function of BMP-4 in patterning the marginal zone of the Xenopus embryo. EMBO J. 13, 5015–5025. Fisher, M.E., Isaacs, H.V., Pownall, M.E., 2002. eFGF is required for activation of XmyoD expression in the myogenic cell lineage of Xenopus laevis. Development 129, 1307–1315. Fletcher, R.B., Harland, R.M., 2008. The role of FGF signaling in the establishment and maintenance of mesodermal gene expression in Xenopus. Dev. Dyn. 237, 1243–1254. Gille, H., Kortenjann, M., Thomae, O., Moomaw, C., Slaughter, C., Cobb, M.H., Shaw, P.E., 1995. ERK phosphorylation potentiates Elk-1-mediated ternary complex formation and transactivation. EMBO J. 14, 951–962. Gotoh, Y., Masuyama, N., Suzuki, A., Ueno, N., Nishida, E., 1995. Involvement of the MAP kinase cascade in Xenopus mesoderm induction. EMBO J. 14, 2491–2498. Grainger, R.M., Gurdon, J.B., 1989. Loss of competence in amphibian induction can take place in single nondividing cells. Proc. Natl. Acad. Sci. U. S. A. 86, 1900–1904. Green, J.B.A., Smith, J.C., 1990. Graded changes in dose of a Xenopus activin A homologue elicit stepwise transitions in embryonic cell fate. Nature 347, 391–394. Gurdon, J.B., Fairman, S., Mohun, T.J., Brennan, S., 1985. Activation of muscle-specific actin genes in Xenopus development by an induction between animal and vegetal cells of a blastula. Cell 41, 913–922. Harland, R.M., 1991. In situ hybridization: an improved whole mount method for Xenopus embryos. Meth. Cell. Biol. 36, 675–685. Heasman, J., 1997. Patterning the Xenopus blastula. Development 124, 4179–4191. Heasman, J., 2006. Patterning the early Xenopus embryo. Development 133, 1205–1217. Hill, C.S., Marais, R., John, S., Wynne, J., Dalton, S., Treisman, R., 1993. Functional analysis of growth factor-responsive transcription factor complex. Cell 73, 395–406. Hill, C.S., Wynne, J., Treisman, R., 1994. Serum-regulated transcription by serum response factor (SRF) a novel role for the DNA-binding domain. EMBO J. 13, 5421–5432. Hipskind, R.A., Baccarini, M., Nordheim, A., 1994. Transient activation of RAF-1, MEK, and ERK2 coincides kinetically with ternary complex factor phosphorylation and immediate–early gene promoter activity in vivo. Mol. Cell. Biol. 14, 6219–6231. Isaacs, H.V., Pownall, M.E., Slack, J.M.W., 1994. eFGF regulates Xbra expression during Xenopus gastrulation. EMBO J. 13, 4469–4481. Janknecht, R., Ernst, W.H., Pingoud, V., Nordheim, A., 1993. Activation of ternary complex factor Elk-1 by MAP kinases. EMBO J. 12, 5097–5104. Kilbourne, E.J., Widom, R., Harnish, D.C., Malik, S., Karathanasis, S.K., 1995. Involvement of early growth response factor Egr-1 in apolipoprotein AI gene transcription. J. Biol. Chem. 270, 7004–7010. Kintner, C.R., Brockes, J.P., 1984. Monoclonal antibodies recognise blastemal cells derived from differentiating muscle in newt limb regeneration. Nature 308, 67–69. Kolm, P.J., Sive, H.L., 1995. Efficient hormone-inducible protein function in Xenopus laevis. Dev. Biol. 171, 267–272. Krain, B., Nordheim, A., 1999. Artefactual gene induction during preparation of Xenopus laevis animal cap explants. Int. J. Dev. Biol. 43, 563–566.

325

Kroll, K.L., Amaya, E., 1996. Transgenic Xenopus embryos from sperm nuclear transplantations reveal FGF signalling requirements during gastrulation. Development 122, 3173–3183. LaBonne, C., Whitman, M., 1994. Mesoderm induction by activin requires FGF-mediated intracellular signals. Development 120, 463–472. LaBonne, C., Whitman, M., 1997. Localization of MAP kinase activity in early Xenopus embryos: implications for endogenous FGF signaling. Dev. Biol. 183, 9–20. LaBonne, C., Burke, B., Whitman, M., 1995. Role of MAP kinase in mesoderm induction and axial patterning during Xenopus development. Development 121, 1475–1486. Latinkic, B.V., Umbhauer, M., Neal, K., Lerchner, W., Smith, J.C., Cunliffe, V., 1997. The Xenopus Brachyury promoter is activated by FGF and low concentrations of activin and suppressed by high concentrations of activin and by paired-type homeodomain proteins. Genes Dev. 11, 3265–3276. Lemaire, P., Vesque, C., Schmitt, J., Stunnenberg, H., Frank, R., Charnay, P., 1990. The serum-inducible mouse gene Krox-24 encodes a sequence-specific transcriptional activator. Mol. Cell. Biol. 10, 3456–3467. Levine, M., Davidson, E.H., 2005. Gene regulatory networks for development. Proc. Natl. Acad. Sci. U. S. A. 102, 4936–4942. Li, Q.J., Yang, S.H., Maeda, Y., Sladek, F.M., Sharrocks, A.D., Martins-Green, M., 2003. MAP kinase phosphorylation-dependent activation of Elk-1 leads to activation of the co-activator p300. EMBO J. 22, 281–291. Lin, G.F., Geng, X., Chen, Y., Qu, B., Wang, F., Hu, R., Ding, X., 2003. T-box binding site mediates the dorsal activation of myf-5 in Xenopus gastrula embryos. Dev. Dyn. 226, 51–58. Loose, M., Patient, R., 2004. A genetic regulatory network for Xenopus mesendoderm formation. Dev. Biol. 271, 467–478. Mangan, S., Itzkovitz, S., Zaslaver, A., Alon, U., 2006. The incoherent feed-forward loop accelerates the response-time of the gal system of Escherichia coli. J. Mol. Biol. 356, 1073–1081. Mohun, T.J., Chambers, A.E., Towers, N., Taylor, M.V., 1991. Expression of genes encoding the transcription factor SRF during early development of Xenopus laevis: identification of a CArG box-binding activity as SRF. EMBO J. 10, 933–940. Morley, R.H., Lachani, K., Keefe, D., Gilchrist, M.J., Flicek, P., Smith, J.C., Wardle, F.C., 2009. A gene regulatory network directed by zebrafish No Tail accounts for its roles in mesoderm formation. Proc. Natl. Acad. Sci. U. S. A. 106, 3829–3834. Nentwich, O., Munchberg, F.E., Frommer, G., Nordheim, A., 2001. Tissue-specific expression of the Ets gene Xsap-1 during Xenopus laevis development. Mech. Dev. 109, 433–436. Nieuwkoop, P.D., Faber, J., 1975. Normal Table of Xenopus laevis (Daudin). Amsterdam, North Holland. Ochi, H., Hans, S., Westerfield, M., 2008. Smarcd3 regulates the timing of zebrafish myogenesis onset. J. Biol. Chem. 283, 3529–3536. Panitz, F., Krain, B., Hollemann, T., Nordheim, A., Pieler, T., 1998. The Spemann organizer-expressed zinc finger gene Xegr-1 responds to the MAP kinase/Ets-SRF signal transduction pathway. EMBO J. 17, 4414–4425. Pera, E.M., Ikeda, A., Eivers, E., De Robertis, E.M., 2003. Integration of IGF, FGF, and anti-BMP signals via Smad1 phosphorylation in neural induction. Genes Dev. 17, 3023–3028. Piepenburg, O., Grimmer, D., Williams, P.H., Smith, J.C., 2004. Activin redux: specification of mesodermal pattern in Xenopus by graded concentrations of endogenous activin B. Development 131, 4977–4986. Rana, A.A., Collart, C., Gilchrist, M.J., Smith, J.C., 2006. Defining synphenotype groups in Xenopus tropicalis by use of antisense morpholino oligonucleotides. PLoS Genet. e193, 2. Saka, Y., Tada, M., Smith, J.C., 2000. A screen for targets of the Xenopus T-box gene Xbra. Mech. Dev. 93, 27–39. Schohl, A., Fagotto, F., 2002. Beta-catenin, MAPK and Smad signaling during early Xenopus development. Development 129, 37–52. Schratt, G., Philippar, U., Hockemeyer, D., Schwarz, H., Alberti, S., Nordheim, A., 2004. SRF regulates Bcl-2 expression and promotes cell survival during murine embryonic development. Embo J. 23, 1834–1844. Schulte-Merker, S., Smith, J.C., 1995. Mesoderm formation in response to Brachyury requires FGF signalling. Curr. Biol. 5, 62–67. Sharrocks, A.D., 1995. ERK2/p42 MAP kinase stimulates both autonomous and SRFdependent DNA binding by Elk-1. FEBS Lett. 368, 77–80. Smith, J.C., 1993. Purifying and assaying mesoderm-inducing factors from vertebrate embryos. In: Hartley, D. (Ed.), Cellular Interactions in Development—A Practical Approach. Oxford Univ. Press, Oxford, pp. 181–204. Smith, J.C., 1995. Mesoderm-inducing factors and mesodermal patterning. Curr. Opin. Cell Biol. 7, 856–861. Smith, J.C., Watt, F.M., 1985. Biochemical specificity of Xenopus notochord. Differentiation 29, 109–115. Smith, J.C., Price, B.M., Green, J.B., Weigel, D., Herrmann, B.G., 1991. Expression of a Xenopus homolog of Brachyury (T) is an immediate–early response to mesoderm induction. Cell 67, 79–87. Steinbach, O.C., Rupp, R.A., 1999. Quantitative analysis of mRNA levels in Xenopus embryos by reverse transcriptase-polymerase chain reaction (RT-PCR). Methods Mol. Biol. 127, 41–56. Steinbach, O.C., Wolffe, A.P., Rupp, R.A., 1997. Somatic linker histones cause loss of mesodermal competence in Xenopus. Nature 389, 395–399. Steinbach, O.C., Ulshofer, A., Authaler, A., Rupp, R.A., 1998. Temporal restriction of MyoD induction and autocatalysis during Xenopus mesoderm formation. Dev. Biol. 202, 280–292. Symes, K., Smith, J.C., 1987. Gastrulation movements provide an early marker of mesoderm induction in Xenopus. Development 101, 339–349.

326

O. Nentwich et al. / Developmental Biology 336 (2009) 313–326

Tada, M., O'Reilly, M.A.J., Smith, J.C., 1997. Analysis of competence and of Brachyury autoinduction by use of hormone-inducible Xbra. Development 124, 2225–2234. Tanegashima, K., Yokota, C., Takahashi, S., Asashima, M., 2000. Expression cloning of Xantivin, a Xenopus lefty/antivin-related gene, involved in the regulation of activin signaling during mesoderm induction. Mech. Dev. 99, 3–14. Taylor, D., Badiani, P., Weston, K., 1996. A dominant interfering Myb mutant causes apoptosis in T cells. Genes Dev. 10, 2732–2744. Umbhauer, M., Marshall, C.J., Mason, C.S., Old, R.W., Smith, J.C., 1995. Mesoderm induction in Xenopus caused by activation of MAP kinase. Nature 376, 58–62. Wang, C., Dostanic, S., Servant, N., Chalifour, L.E., 2005. Egr-1 negatively regulates expression of the sodium-calcium exchanger-1 in cardiomyocytes in vitro and in vivo. Cardiovasc. Res. 65, 187–194. Weinberg, E.S., Allende, M.L., Kelly, C.S., Abdelhamid, A., Murakami, T., Andermann, P., Doerre, O.G., Grunwald, D.J., Riggleman, B., 1996. Developmental regulation of zebrafish MyoD in wild-type, no tail and spadetail embryos. Development 122, 271–280.

Weinhold, B., Schratt, G., Arsenian, S., Berger, J., Kamino, K., Schwarz, H., Ruther, U., Nordheim, A., 2000. Srf(-/-) ES cells display non-cell-autonomous impairment in mesodermal differentiation. EMBO J. 19, 5835–5844. Whitman, M., Melton, D.A., 1992. Involvement of p21 ras in Xenopus mesoderm induction. Nature 357, 252–255. Wittenberger, T., Steinbach, O.C., Authaler, A., Kopan, R., Rupp, R.A., 1999. MyoD stimulates delta-1 transcription and triggers notch signaling in the Xenopus gastrula. EMBO J. 18, 1915–1922. Yun, C.H., Choi, S.C., Park, E., Kim, S.J., Chung, A.S., Lee, H.K., Lee, H.J., Han, J.K., 2007. Negative regulation of Activin/Nodal signaling by SRF during Xenopus gastrulation. Development 134, 769–777. Zanetti, M., Ratcliffe, A., Watt, F.M., 1985. Two subpopulations of differentiated chondrocytes identified with a monoclonal antibody to keratan sulfate. J. Cell Biol. 101, 53–59. Zetser, A., Frank, D., Bengal, E., 2001. MAP kinase converts MyoD into an instructive muscle differentiation factor in Xenopus. Dev. Biol. 240, 168–181.