A dual requirement for Iroquois genes during Xenopus kidney development

Studies performed in different vertebrates indicate that most of the genes necessary for pronephros formation in Xenopus are also crucial for the formation of more complex mammalian metanephros or adult kidneys(Carroll et al., 1999; Dressler, 2006; Ryffel, 2003). Moreover, these similarities at the molecular level correlate with physiological homologies. Thus, the tubules of all nephrons have similar subdivisions along the anteroposterior axis with analogous distribution of transporters of small molecules and ions along this axis(Reggiani et al., 2007; Wingert et al., 2007; Zhou and Vize, 2004). This fact, and the accessibility of Xenopus to genetic manipulation, makes this animal an excellent model system with which to study kidney development. In Xenopus, the specification of the pronephric anlage occurs in the late gastrula/early neurula (stage 12). However, the first sign of pronephric morphogenesis, the thickening of the lateral mesoderm, is detected 10 hours later, in the late neurula (stage 20/21). At the tailbud stage (stage 25-30),differentiation of the three basic segments occurs: the corpuscle, the tubules and the duct. This is followed by the final maturation of the organ, which is associated with the physiological specialization of the pronephric tubules along the proximal-distal axis, as observed by the differential activation of several genes encoding different transport proteins(Carroll et al., 1999; Reggiani et al., 2007; Ryffel, 2003; Zhou and Vize, 2004).

The Iroquois (Irx) genes encode homeoproteins conserved during evolution with multiple functions during animal development(Gómez-Skarmeta and Modolell,2002). Their role during patterning of the vertebrate nervous system has been studied in detail(Gómez-Skarmeta and Modolell,2002). By contrast, their participation in the development of other organs is less well understood. Recently, it has been reported that Xenopus Irx1, Irx2 and Irx3 are expressed from the tailbud stage in the intermediate tubule segment of the pronephros, immediately prior to regionalization of the proximal-distal axis(Reggiani et al., 2007). This study further showed that Irx3, but not Irx1 or Irx2, is required for development of this region(Reggiani et al., 2007).

We report that Xenopus Irx1, Irx2 and Irx3 are also expressed in the pronephric territory during earlier mid-neurula stages. Morpholino loss-of-function analyses, together with misexpression of inducible forms of wild-type and dominant-negative Irx proteins, reveal a two-step requirement for Irx1 and Irx3 during kidney development. Initially, Irx1 and Irx3 maintain the identity of the pronephric territory and define its size. This seems to occur, at least in part, through positive regulation of Bmp signalling. Later, both Irx genes are required for the formation of the intermediate tubule segment of the pronephros, as reported only for Irx3(Reggiani et al., 2007). In addition, we show that both genes are regulated by retinoic acid, which is known to be necessary to activate early kidney genes and to participate in the segmentation of the pronephros in the proximal-distal axis(Cartry et al., 2006; Wingert et al., 2007).

Constructs were made using Irx alleles from sequences AJ001834, AJ001835,AF027175, AF338157 and AF338158. MT-Irx1, MT-Irx2 and MT-Irx3 have been described previously(de la Calle-Mustienes et al.,2002). To generate MT-Irx4 and MT-Irx5, a fragment from the 5′ region of each cDNA, including unique sites within the ORF (KpnI in Irx4 and ClaI in Irx5),were PCR amplified. These sites allowed us to fuse the PCR fragments to the rest of each cDNA. 5′ primers contained a XhoI (Irx4)or EcoRI (Irx5) site, allowing us to clone these PCRs fragments in frame within the pCS2-MT plasmid(Turner and Weintraub, 1994)at XhoI (MT-Irx4) or between EcoRI and XhoI (MT-Irx5). Primers used were: Irx4,5′-ccctcgagATGTCATATCCTCAGTTTGGC-3′ and 5′-GCTCCCATCCATGGTACCATACC-3′; Irx5,5′-gggaattcaCATGTCCTATCCGCAGGGC-3′ and 5′-ATCCCCTGCATCTCCATC-3′. Bold nucleotides indicate restriction sites used for cloning procedures; capitals indicate sequences present in the cDNAs.

To generate Irx-MT constructs, a fragment from the 3′ region of each cDNA, including unique sites within the ORF (SacI in Irx1, PstI in Irx2, SacII in Irx3, KpnI in Irx4 and XbaI in Irx5 cDNAs), were PCR amplified. For Irx1, Irx2, Irx3 and Irx5, the fusion to the rest of each cDNA in pCS2-MT was carried out as follows: EcoRI/SacI, PstI, SacII or XbaI fragments that contain the 5′ cDNA regions of Irx1, Irx2, Irx3 or Irx5,respectively, were cloned into pBluescript. These fragments were excised with HindIII, SacI, PstI, SacII or XbaI and ligated with the PCR fragment in pCS2-MT between the HindIII and ClaI (Irx1) or BamHI(Irx2, Irx3 and Irx5) sites. The full Irx ORF in frame with the Myc tag was then transferred into the EcoRI site of pCS2+. For Irx4, a ClaI/KpnI fragment containing the 5′ cDNA was ligated with the corresponding 3′ PCR fragment, expanding the 3′cDNA into the ClaI site of pCS2-MT. Primers used were: for Irx1, 5′-GCAACAAGCCCAGATGG-3′ and 5′-ccaatcgatGGCAGAGGGAAGTGCTG-3′; for Irx2,5′-GCCGACCATCTTTGCG-3′ and 5′-ggggatccTGGGTATGGTTGTACTCC-3′; for Irx3,5′-CACAGCCCCATGTTCTGG-3′ and 5′-ggggatccGGATGAGGATAAAGCGGA-3′; for Irx4,5′-CCATGGTACCTACCCTCG-3′ and 5′-ccaatcgatAGCAAGATGTTCTGTTCCT-3′; for Irx5,5′-CTTCTCCATCTAGATCTCC-3′ and 5′-ggggatccAATGCTAGACATACCTTTCTTC-3′. Bold nucleotides indicate restriction sites used for cloning procedures; capitals indicate sequences present in the cDNAs.

We first generated Irx-MT-GR derivatives by cloning the GR domain within the XhoI and XbaI fragment located 3′of the Irx-MT in the pCS2 Irx-MT vectors. The hormone-inducible GR domain was obtained by PCR using the oligonucleotides 5′-cccctcgagATCCCCTCTGAAAATCC-3′ and 5′-ctctagaCACTTTTGATGAAACAGAAG-3′ from a MyoD-GR plasmid kindly donated by H. Sive. To make the chimeric mRNAs of these constructs insensitive to the MOs, we introduced a MT 5′ by fusing these constructs with their corresponding MT-Irx as follows: for Irx1, a 5′ EcoRI-SacI fragment from MT-Irx1 was ligated to a 3′ SacI-NotI fragment from MT-Irx1-GR in pCS2-MT; for Irx2, a 5′ EcoRI-ApaI fragment from MT-Irx2 was ligated to a 3′ ApaI-NotI fragment from MT-Irx2-GR in pCS2-MT; for Irx3, a 5′ EcoRI-SacII fragment from MT-Irx3 was ligated to a 3′ SacII-NotI fragment from MT-Irx3-GR in pCS2-MT.

Antisense RNA probes were prepared from cDNAs using digoxigenin or fluorescein labels (Roche). Xenopus specimens were prepared,hybridized and stained as described(Harland, 1991). X-Gal staining was performed accordingly to Coffman et al.(Coffman et al., 1993). Double fluorescent in situ hybridization was performed as previously described(Zhou and Vize, 2004). Antibody staining was performed as previously described(Gómez-Skarmeta et al.,2001). Monoclonal antibodies 3G8 and 4A6 were kindly provided by E. Jones. The monoclonal antibody 12/101, generated by J. P. Brockes, was obtained from the Developmental Studies Hybridoma Bank (NICHD and The University of Iowa, Department of Biological Science, Iowa City, IA 52242).

DNAs were linearized and transcribed as described(Harland and Weintraub, 1985)with GTP cap analogue (New England Biolabs). SP6, T3 or T7 RNA polymerases were used. After DNAse treatment, RNA was extracted with phenol-chloroform,column purified and precipitated with ethanol. mRNAs for injection were resuspended in water. Synthetic mRNAs or MOs were injected into V2.2 blastomeres with 2-4 nl solutions. The following morpholinos were used in this study: MOIrx1, 5′-CATGTCTCTCCGGCAGGGAAATCGC-3′;MOIrx2, 5′-AGGTAACCCTGAGGATAGGACATGG-3′; MOIrx3,5′-CTGTGGGAAGGACATGGTGCAGCCG-3′; MOIrx3.2,5′-AGCTGTGGGAAGGACATGGTGCAGC-3′; MOIrx4,5′-GTAGCCAAACTGAGGATATGACATT-3′; and MOIrx5,5′-CAAGTAGCCCTGCGGATAGGACATG-3′. MOIrx1 and MOIrx5 are 100% homologous to the Irx1 and Irx5alleles used in this study. The second Irx5 and Irx1 alleles contain 1 and 2 sequence mismatches, respectively, with their corresponding MOs. The other Irx MOs have 100% homology with all their corresponding Irx alleles. In the MO- or mRNA-injected embryos, in images taken at the same magnification, we used the histogram function of Photoshop to compare the size of the area expressing different markers in the injected versus the uninjected sides of the same embryo. Grafts were performed as previously described(Gómez-Skarmeta et al.,1999).

We examined the expression of the full complement of Xenopus Irx genes during pronephros development. Irx1 and Irx2 were largely co-expressed during pronephric development from late neurula onwards(Fig. 1A-H). Irx3 was temporally and spatially expressed in similar, but not identical, territories(Fig. 1I-L). Irx4 was only found in the pronephros at tadpole stage, Irx5 was never detected in the kidney territory (see Fig. S1 in the supplementary material)and Irx6 was not expressed at these stages(de la Calle-Mustienes et al.,2005). Irx1, Irx2 and Irx3 were initially expressed at mid neurula stages in the pronephric anlage(Fig. 1A,E,I). Double staining with the early pronephric markers Lim1 and Pax8(Carroll and Vize, 1999; Heller and Brandli, 1999)confirmed that Irx1 and Irx3 were expressed in the pronephric field (insets in Fig. 1A,I; Fig. 1M,Q). Moreover, although the Irx3 pronephric domain was broad and encompasses most, if not all, of the Lim1 territory, Irx1was confined to the dorsal area of the Lim1 field. At late neurula stage, Irx1 was still confined to the dorsal pronephric domain(Fig. 1B,N). By contrast, Irx3 became now restricted to the ventral pronephric territory(Fig. 1J,R). During tailbud stages, Irx1 expression became localized in the intermediate tubule,as judged by double staining with Pax8(Fig. 1C, inset) or with the proximal tubule marker Sglt1k(Fig. 1O)(Reggiani et al., 2007; Zhou and Vize, 2004). At this stage, Irx3 also became confined to the intermediate tubule, but its expression extended into the distal area of the proximal tubule and into the distal tubule, as determined by double staining with Pax8(Fig. 1K, inset) or with the intermediate tubule marker Nkcc2(Fig. 1S)(Reggiani et al., 2007; Zhou and Vize, 2004). The expression patterns of Irx1 and Irx3 were maintained at tadpole stages (Fig. 1D,L,P,T). In addition, late Irx1 expression was also observed in migrating ventral mesoderm (Fig. 1D, blue arrowhead). We conclude that Irx genes have dynamic patterns of expression during pronephros development and that their expression in the pronephric territory starts much earlier than recently reported(Reggiani et al., 2007).

To examine the requirement for Irx genes during Xenopus pronephros development, we interfered with the activity of each Irx mRNA by injecting specific translation-blocking morpholinos (MOs). As Xenopus laevis is pseudotretaploid, we identified all ESTs available in the database for each Irx gene and found one allele for Irx4 and two alleles for Irx1,Irx2, Irx3 and Irx5. We designed specific Irx MOs that block translation from both Irx alleles when present in the genome. The specificity of these MOs is shown in Fig. S2 in the supplementary material. Embryos injected with any of these Irx MOs showed different degrees of neural defects(E.R.-S., P.A. and J.L.G.-S., unpublished) indicating that they are effective in blocking the activity of their respective Irx genes. To reduce the MO effects on off-target tissues, in all experiments, we targeted the pronephric anlage by injecting the V2.2 blastomere of 8-16-cell stage embryos. We then evaluated the effect these injections on the early renal markers Lim1and Pax8. We considered that an embryo had an altered pronephros when the area expressing the corresponding marker on the injected side varied by more than 20% relative to the uninjected control side. Differences greater that 20% between the two pronephros of a single embryo were very rarely observed in non-injected embryos or in embryos injected with a control MO(<2%, n=83). Injection of 8 ng of MOs against Irx1 or Irx3 (Fig. 2), but not Irx2, Irx4, Irx5 or a control MO (Fig. S3 in the supplementary material; not shown), caused renal defects. Thus, at mid-late neurula stage,the territory expressing Lim1 or Pax8 was reduced in most Irx1 or Irx3 morphant embryos(Fig. 2A,B,I,J, and not shown). The average of pronephros size reduction was around 40-50%, and was observed in 57% (n=159) and 83% or (n=74) of the MOIrx1- and MOIrx3-injected embryos, respectively. As muscles are a source of signals that influence kidney development(Seufert et al., 1999), we determined whether, in the Irx morphant embryos, muscle development was altered. By staining with the muscle-specific antibody 12/101, we found that this was not the case (Fig. 2A,B,I,J). We also monitored the effect of injecting MOIrx1 or MOIrx3 on the expression of genes expressed at tadpole stages, during the maturation of the pronephros. All three genes examined, Sglt1k, Nkcc2 and Nbcc1 [which are expressed in proximal, intermediate and distal tubule, respectively(Reggiani et al., 2007; Zhou and Vize, 2004)], were downregulated in MOIrx1 (38-47%, n=34-39) or MOIrx3(51-96%, n=24-41) morphants (Fig. 2C-H,K-P). The reduction of Irx function did not significant altered the rate of cell proliferation or cell death in the kidney territory(not shown).

We next examined the effect of knocking-down simultaneously both Irx1 and Irx3 (with a mix containing 3-4ng of each MO). Co-injection of Irx1 and Irx3 MOs at half doses caused phenotypes similar to individual MO injections at double concentration. The simultaneous impairment of Irx1 and Irx3 resulted in the loss of Lim1 (58%, n=66) and Pax8 (89%, n=57) expression only at late neurula stage, coinciding with the onset of expression of Irx1 and Irx3, but it did not affect the expression of these genes at early neurula (Fig. 3A-F). In these injected embryos, the differentiated phronephric structures at tadpole stages,but not the somitic muscles, were severely reduced, as determined by triple staining with the antibodies 3G8 and 4A6, which label the tubules and the duct, respectively (Vize et al.,1995), and 12/101 (Fig. 3G-J). Interestingly, in these Irx-depleted embryos, the number of ventral muscle fibres was reduced (Fig. 3G,H, blue arrowheads). This suggests that the expression of Irx1 in ventral migrating muscle cells may be required for proper development of these muscles. Sections through these injected embryos suggested that cells that lose their kidney fate are likely to end up as fibroblasts, as an increased number of cells with fibroblast shape are detected in the Irx depleted side(Fig. 3J).

The autonomous requirement of Irx1 and Irx3 for pronephros formation was further supported by a transplantation experiment. A graft of lateral plate from a late gastrula embryo co-injected with Irx1MO, Irx3MO and GFP mRNA was transplanted to the equivalent area of a wild-type host. In the transplanted embryo, the expression of Lim1, but not that of neural or muscle markers(Sox2 and 12/101, respectively), was impaired (see Fig. S4 in the supplementary material) (66%, n=6). This was not observed after transplantation of a control graft from an embryo injected with only GFP mRNA (100%, n=7; not shown). Finally, we also monitored the expression of several additional kidney markers (Osr2,Nhf1β, Wnt4 and Wt1) in double Irx1 and Irx3 morphant embryos. Expression of all genes was reduced (see Fig. S5 in the supplementary material). Together, our results indicate that Irx1 and Irx3 are activated following the specification of the kidney anlage and are autonomously required for the maintenance of this specification.

Our results prompted us to test the effect of the misexpression of Irx genes. We first generated hormone-inducible forms of the Irx1, Irx2and Irx3 proteins (MT-Irx-GR) that are insensitive to the MOs (see Materials and methods). These constructs allowed the induction of Irx function after gastrulation, thus eliminating possible earlier effects of Irx genes on mesoderm formation (Glavic et al.,2001). All three MT-Irx-GR proteins behaved similarly in overexpression studies (see below and not shown). Consistent with a requirement for Irx genes during pronephric development, overexpression of MT-Irx1-GR or MT-Irx3-GR mRNAs, upon addition of dexamethasone (Dex) at stage 14, triggered a ventral expansion of Lim1 and Pax8 (Fig. 4A-H; 60% of the embryos showing enlarged pronephros, n=132). In most embryos, the pronephros at the injected side was about 50% larger than the pronephros at the control non-injected side. This expansion enlarged the differentiated kidney tissue, as determined by staining with the tubules antibody (3G8) (Fig. 4Q-T). We next determined the ability of these MT-Irx-GR constructs to rescue the defects observed in Irx morphant embryos. Although interference with Irx1 and Irx3 function with a mix of MOs caused downregulation of Lim1 and Pax8(Fig. 4I-L; Fig. 3C-F), co-injection of Irx MOs with MT-Irx1-GR or MT-Irx3-GR mRNAs rescued the expression of these genes, but only upon hormone addition at early neurula stage (Fig. 4M-P and not shown;15% reduced and 55% enlarged pronephros, n=100). These results indicate that the pronephric expression of Irx genes is required to maintain the transcription of the key renal genes Lim1 and Pax8, and to define the size of the pronephric anlage.

To cast light on the way Irx proteins function during these processes, we overexpressed different hormone-inducible constructs with the homeodomain (HD)of Irx1 alone or fused to either the E1A activator or the Engrailed (EnR)repression domains (HD-GR, HD-GR-E1A and HD-GR-EnR). It is known that, during neural development, Irx proteins act as transcriptional repressors, as overexpression of wild-type Irx proteins or HD-GR-EnR fusions cause similar phenotypes, whereas HD-GR and HD-GR-E1A behave as dominant-negative molecules(Gómez-Skarmeta et al.,2001). By contrast, during kidney development, overexpression of HD-GR-E1A mRNA (Fig. 5A-H) mimicked the ventral expansion of Lim1 (64% of the embryos with enlarged pronephros, n=48) caused by wild-type Irx mRNAs(Fig. 4). Conversely, the overexpression of HD-GR (Fig. 5I-P) or HD-GR-EnR (inset in Fig. 5I, M, and not shown)mRNAs promoted downregulation of Lim1 (88% of the embryos showing reduced pronephros, n=54). Therefore, during pronephros development,HD-GR and HD-GR-EnR proteins behave as dominant-negative molecules that interfere with Irx function. Similar results were found when Pax8expression was examined (not shown). In addition, by providing Dex at different stages of development, we also found that the requirement for Irx function during pronephros development occurred around mid neurula (stages 15-17). Overexpression of Irx proteins at later stages (20-22) had little effect on Lim1 and Pax8(Fig. 5). These data suggest that Irx proteins act as transcriptional activators to maintain the kidney anlage properly and to define the size of this territory, and confirm that they are initially required before pronephros morphogenesis takes place.

The Bmp pathway is implicated at several steps during vertebrate kidney development (Cain et al.,2008). A recent report showed that blocking this pathway during Xenopus neurula stages impaired pronephros formation(Bracken et al., 2008). These results resemble those observed by reducing Irx function. As Irx genes modulate Bmp signalling during neural development(Gómez-Skarmeta et al.,2001), a similar situation may occur during kidney development. To evaluate this, we increased Bmp signalling at neurula stages, by overexpressing an inducible Smad1 construct (Smad1GR) in the absence of Irx1 and Irx3 activity. Induction of Smad1GR at stage 14 caused an expansion of kidney territory (51%, n=35), as determined by Pax8 expression (Fig. 6A,B). In the absence of Dex, embryos co-injected with Smad1GR mRNA and Irx1 and Irx3 MOs showed downregulation of Pax8 (Fig. 6C,D; 41%, n=51). This phenotype is partially rescued by Dex treatment at stage 14 (Fig. 6E,F; 21% with Pax8 downregulated, n=48). Thus,part of Irx function seems to be to positively modulate the Bmp pathway. Bmp7 is expressed and required for kidney development(Dudley et al., 1995; Luo et al., 1995; Wang et al., 1997). We examined whether its expression depended on Irx activity. Indeed, as for other kidney markers, in double Irx1 and Irx3 morphant embryos Bmp7 expression was downregulated(Fig. 6G,H).

Irx3, but not Irx1, has been shown to be required for formation of the intermediate tubule segment of the pronephros(Reggiani et al., 2007). We examined whether Irx1 was also required in this late process. Injection of high doses of Irx MOs downregulated all proximal-distal genes, probably because of the early requirement of Irx genes for maintaining the kidney anlage, thus preventing examination of later Irx functions. To try to overcome this problem, we partially reduced Irx function by injecting Irx1 and Irx3 MOs at lower doses. We complement these experiments injecting a second Irx3 MO (MOIrx3.2) that, in a previous report, was unable to reveal an early Irx3 requirement(Reggiani et al., 2007). Blastomere injection of 4 ng of Irx1 MO or either one of the two Irx3 MOs caused little effect on Lim1 and Pax8expression in late neurula or tailbud stages (not shown). By contrast, at tadpole stage, embryos injected with Irx1 or with any of the Irx3 MOs showed downregulation of the proximal domain of Nkcc2 (40-50%, n=19-26)(Fig. 7C,D,G,H,K,L). This was also accompanied by a reduction in the distal expression of Sglt1k in the MOIrx3, but not in the MOIrx1, injected embryos(Fig. 7A,B,E,F,I,J). These results are consistent with the expression domains of Irx1 and Irx3, and suggest that both genes are required for proximal-distal patterning, as was already reported for Irx3(Reggiani et al., 2007). We also tried to address why, in the previous report(Reggiani et al., 2007), an early requirement for Irx3 was not detected. For that, we injected one of the Irx3 MO they used (MOIrx3.2) at higher (8 ng)doses. At this concentration, most embryos injected with our Irx3 MO were healthy and showed the strong reduction of all segment markers shown in Fig. 2. By contrast, the majority of the embryos injected with 8 ng of MOIrx3.2 were malformed. These malformed embryos also lost most markers(Fig. 7M-P). This might explain the discrepancy if those embryos with stronger phenotypes and malformations were not taken into account in the previous report(Reggiani et al., 2007).

To further examine the Irx requirement in pronephric proximal-distal patterning, we injected inducible wild-type or dominant-negative Irx constructs, activated them at different developmental stages and examined their effect on pronephric proximal-distal patterning. A dominant-negative Irx construct induced at stage 13 downregulated Sglt1k, Nkcc2 and Gata3 (84-90%, n=19-20)(Fig. 8A-D), markers of proximal tubule, intermediate tubule and duct, respectively(Reggiani et al., 2007; Wingert et al., 2007; Zhou and Vize, 2004). By contrast, the same construct induced at stage 27 downregulated Nkcc2(59%, n=17), whereas Sglt1k and Gata3 were not affected (Fig. 8E-H). These results are consistent with an early and a late requirement for Irx function. A further confirmation of this dual function was obtained by examining these proximal-distal markers in embryos injected with Dex-inducible Irx1 or Irx3 proteins. Incubation of injected embryos with Dex from stage 13 caused ectopic patches of Sglt1K (38-47%, n=15-24)(Fig. 8I,J,Q,R) and an enlarged Nkcc2 domain (52-60%, n=17-24)(Fig. 8K,L,S,T). No clear effect was found on Gata3 expression(Fig. 8I,J,Q,R). Addition of hormone at stage 27, expanded the Nkcc2 expression domain (50-60%, n=14-22) (Fig. 8O,P,W,X), but did not affect Sglt1K expression(Fig. 8M,N,U,V).

Retinoic acid (RA) is a requisite for the activation of early kidney genes and for the late segmentation of the pronephros(Cartry et al., 2006; Wingert et al., 2007). Therefore, RA may regulate Irx genes during kidney development. To test this,we examined Irx expression in embryos injected with 100 pg of mRNAs encoding either a dominant-negative (RAR-DN) or a constitutively active (RAR-Vp16) RA receptor (Blumberg et al.,1997). Embryos with reduced or increased RA signalling showed down or upregulation, respectively, of the expression of Irx genes in the kidney(Fig. 9A-H). Thus, RA positively regulates Irx genes in the pronephros. We then determined when RA signalling is required for kidney Irx expression. To this end, we incubated Xenopus embryos at different developmental stages (12.5, 15 or 25)for 1 hour with a control solution (DMSO), with 4-diethylaminobenzaldehyde(DEAB 30 μM), the chemical inhibitor of the RA producing enzyme Raldh2 or with RA (10 μM). Reducing or increasing RA signalling, down or upregulated,respectively, the expression of both genes, but only when the drug treatments were done at late gastrula stage (12.5)(Fig. 9I-Y).

The development of the pronephros can be subdivided in three major steps:specification of the anlage, morphogenesis of the nephron and the generation of different proximal-distal territories. Here, we show that the homeodomain genes Irx1 and Irx3 play essential functions in two of these steps: maintenance of the specified anlage and segmentation of the nephron.

Irx1 and Irx3 show dynamic expression patterns during pronephros development. We find that Irx genes are initially expressed in the pronephric territory at neurula stage. This occurs after the initial specification of this territory by Ors, Lim1 and Pax8 genes at gastrula stage, but before morphological or molecular signs of kidney morphogenesis at late neurula-early tailbud sages. Irx1 is initially activated in the dorsal pronephric territory. By contrast, Irx3 is initially expressed in most of the kidney anlage, but it becomes confined to the ventral pronephros territory. This dorsal-ventral subdivision of the prospective kidney may reflect the initial subdivision of the pronephric territory by Notch signalling into a dorsal region that will generate glomus and proximal tubule, and a ventral domain that will give rise to distal tubule and duct (McLaughlin et al.,2000; Taelman et al.,2006). However, we have not detected an alteration of Irx1 or Irx3 expression by manipulating Notch signalling(not shown), which suggests that Notch does not regulate Irx genes. Interestingly, the expression of Irx genes slightly precedes the onset of expression of Dl1 and Notch1, which is compatible with Irx genes participating in the regulation of Notch signalling in the pronephros. Indeed, Irx genes play a pivotal role in the regulation of Notch signalling during Xenopus neural crest formation(Glavic et al., 2004) and during Drosophila eye development(Domínguez and de Celis,1998). The relationship between Irx genes and Notch signalling is currently under investigation.

Consistent with their initial activation in the pronephric field during neurulation, Irx function is dispensable for the initial activation of the early kidney determinants Ors1, Ors2, Pax8 and Lim1 that occurs at late gastrula (Carroll and Vize,1999; Heller and Brandli,1999; Tena et al.,2007). However, depletion of Irx1 or Irx3impairs the expression of all kidney genes examined at tailbud and tadpole stages. This Irx function seems to be autonomous, as downregulation of kidney genes occurs without affecting neural or other mesodermal tissues. Consistent with a requirement for Irx genes before pronephros morphogenesis,time-controlled loss or gain of Irx function during neurula, but not during tailbud stages, reduces or expands the pronephric field, respectively. Interestingly, gain of Irx function expands but does not promote ectopic expression of Xlim1 and Pax8, as found when Osr genes or Pax8 and Xlim1 are overexpressed(Carroll and Vize, 1999; Tena et al., 2007). This suggests that Irx genes alone are unable to trigger the kidney program. Our results indicate that Irx genes are expressed and required before appearance of any sign of pronephros morphogenesis. This early Irx gene requirement for kidney development is likely to be conserved in other vertebrates as these genes are also expressed in the early kidney anlage of other vertebrates(Houweling et al., 2001; Lecaudey et al., 2005).

During neural development, Irx proteins act as repressors and downregulate Bmp signalling to allow neural plate formation(Gómez-Skarmeta et al.,2001; Itoh et al.,2002). In this work we show that Irx proteins act as activators during kidney formation. Thus, one possible mechanism of action of Irx proteins could be to upregulate Bmp signalling, which is known to participate at many steps during vertebrate kidney formation(Cain et al., 2008). Consistent with this idea, the reduction of Bmp signalling during Xenopusneurula stages causes defects similar to those produced by Irx gene impairment(Bracken et al., 2008). Furthermore, we show that increasing Bmp signalling partially rescue the kidney defects observed in Irx morphant and that Bmp7 expression is downregulated in Irx-deficient embryos. Further experiments are required to determine more precisely the interaction between the Bmp pathway and Irx genes.

Recently, it has been reported that, within the pronephros field, Irx genes are initially expressed at tailbud stages and that only Irx3 is required for proximal-distal pronephric patterning(Reggiani et al., 2007). As indicated above, we detect an earlier expression (at neurula stage) of Irx1, Irx2 and Irx3 in that territory, and a requirement for both Irx1 and Irx3 for the proper development of the kidney territory before pronephric morphogenesis. As it was possible that the late Irx3 described function might be an indirect effect of the earlier Irx requirement, we have re-examined the participation of Irx genes in proximal-distal patterning of the pronephros. Although MOs are very useful reagents to reduce gene activity, their injections into blastomeres may impair gene function from early stages and make it difficult to recognize a late requirement. To try to uncouple early and late Irx requirements, we injected low doses of Irx MOs. In these hypomorphic conditions, embryos did not show early kidney phenotypes but the late pronephric segmentation was affected. This suggests that both genes are required for this late process but does not exclude that this could be an indirect consequence of an early requirement for Irx genes. To determine Irx protein function unambiguously during proximal-distal pronephric patterning, we have used conditional loss- and gain-of-function of Irx genes. Overexpression of an inducible dominant-negative construct demonstrates that early impairment of Irx gene activity downregulates all proximal-distal markers examined. By contrast, late impairment of Irx activity prevents only intermediate tubule formation. We confirmed this dual Irx protein function by overexpressing Irx1 or Irx3 at neurula or tailbud stages. Early Irx activation expands or promotes ectopic expression of different segment markers, whereas late overexpression expands only the intermediate tubule marker Nkcc2. Thus, our study reveals an earlier Irx gene requirement for most of the phronephric field and also, in agreement with Reggiani et al.(Reggiani et al., 2007), a late requirement, although in contrast to this report, our data support the necessity for both Irx1 and Irx3 for proximal-distal patterning.

RA is required for the activation of the early kidney genes Lim1and Pax8 (Cartry et al.,2006), and for the late regionalization of the pronephros(Wingert et al., 2007). We observe that the expansion of the kidney field associated with overexpression of Irx genes are similar to that found upon increasing retinoic acid activity(Cartry et al., 2006). This suggests a possible link between Irx genes and RA signalling. Indeed, we find that both Irx1 and Irx3 are activated by RA signalling. This is consistent with the fact that the pronephric expression of the gene encoding the RA producing enzyme Raldh2 and the RA receptor RARα precedes that of the Irx genes. Thus, RA probably lies upstream of Irx genes during pronephros development. We also find that RA is necessary for Irx gene expression at late gastrula/early neurula stages, but not later. This developmental period is when RA is required for the activation of the early kidney genes Lim1 and Pax8(Cartry et al., 2006), and it is well before Irx genes are initially expressed. Thus, the RA effect on Irx gene expression is likely to be indirect, probably through Lim1 and Pax8. In addition, as we do not detect alteration of the Irxexpression patterns when RA signalling is modulated just before proximal-distal patterning, RA influence on proximal-distal patterning(Wingert et al., 2007) is likely to occur as an indirect consequence of its effect on early kidney genes such as Lim1 and Pax8(Cartry et al., 2006).

In a functional survey of the enhancer activity of highly conserved non-coding elements present in the IrxB complex(de la Calle-Mustienes et al.,2005) we identified two ultraconserved non-coding regions (UCRs)that activate expression in the pronephros, as determined in Xenopustransgenic experiments (de la Calle-Mustienes et al., 2005). Each UCR is located in each Irx gene cluster in relatively close proximity to Irx1 and Irx3. These regulatory regions are likely to contribute to the regulation of Irx genes during kidney development by early kidney specification genes. The detailed analyses of these regions should help unravel the molecular mechanisms that control Irx gene expression during pronephros formation and to define the genetic cascade that operates during this process.

We are most grateful to E. Amaya, A. Brändli, E. Bellefroid, B. Blumberg, A. Fainsod, P. A. Krieg, D. Kimelman, N. Papalopulu, H. Sive, M. Taira, D. Turner, R. Vignali and P. Vize for reagents. We also thank N. Ueno and the NIBB/NIG Xenopus laevis EST project for the Mochii clone XL056l08. We especially thank F. Casares, P. Lemaire, John Pearson and J. Modolell for helpful criticisms. This work was supported by grants from the Spanish Ministry of Education and Science (BFU2004-00310, BFU2007-60042/BMC,CSD2007-00008) and Junta de Andalucía (Proyecto de Excelencia 00260) to J.L.G.-S., and a Marie Curie Reintegration Grant (ERG-014806) and a UPO Grant(APP2D06060) to P.A.E.R.-S. and P.A. are I3P fellows from the CSIC.