Kermit 2/XGIPC, an IGF1 receptor interacting protein, is required for IGF signaling in Xenopus eye development

GIPC (GAIP interacting protein, C terminus) is a PDZ-domain-containing protein initially identified by virtue of its interaction with RGS-GAIP(regulator of G protein signaling-GTPase activating protein for Gαi)(De Vries et al., 1998b). The list of identified binding partners for GIPC includes numerous membrane proteins (Katoh, 2002), such as the transmembrane semaphorin M-SemF(Wang et al., 1999),neuropilin 1 (Cai and Reed,1999), the Wnt receptor frizzled 3(Tan et al., 2001), theβ ₁-adrenergic receptor (Hu et al., 2003), the TGFβ-type III receptor(Blobe et al., 2001),melanosomal membrane protein gp75 (tyrosinase-related protein 1)(Liu et al., 2001), the tyrosine kinase receptor TrkA (Lou et al.,2001), intergrin α subunits(El Mourabit et al., 2002; Tani and Mercurio, 2001),insulin-like growth factor 1 receptor (IGF1R)(Booth et al., 2002; Ligensa et al., 2001), human lutropin receptor (Hirakawa et al.,2003), myosin VI (Dance et al., 2004; Hasson,2003), dopamine D2 and D3 receptors(Jeanneteau et al., 2004a; Jeanneteau et al., 2004b), a myeloid cell-surface marker CD93 (Bohlson et al., 2005) and human papillomavirus type 18 E6 protein(Favre-Bonvin et al., 2005). Although the list of GIPC-binding partners is long, the endogenous function of GIPC and the physiological significance of these associations are much less studied. Loss-of-function studies published so far have been primarily in ex vivo or in vitro systems, such as Xenopus ectodermal explants(Tan et al., 2001) and cultured cell lines (Favre-Bonvin et al.,2005; Hirakawa et al.,2003); the in vivo role of GIPC in embryonic development has not yet been examined.

Previously, we identified a Xenopus homolog of GIPC, kermit(Tan et al., 2001), that is 72% identical to mammalian GIPC (De Vries et al., 1998b). Knockdown of kermit using antisense morpholino oligonucleotides blocked neural crest induction by Xenopus frizzled 3 in ectodermal (animal cap) explants (Tan et al., 2001), but did not inhibit neural crest formation in whole embryos. We reasoned that this negative result may indicate the presence of a compensating or redundant activity in whole embryos that is absent in animal cap explants. We therefore identified a second kermit gene (67% amino acid identity with kermit 1) in Xenopus early embryos to study the redundancy between kermit and kermit2. Unexpectedly, we discovered a novel function for kermit 2 in IGF signaling that does not overlap with kermit 1.

Kermit 2 is identical to XGIPC, identified in a yeast two hybrid screen for IGF1 receptor binding proteins in Xenopus oocytes(Booth et al., 2002). XGIPC/kermit 2 binds to the cytoplasmic domain of the Xenopus IGF1R and this interaction appears to require the PDZ domain of XGIPC. Overexpression of C-terminal truncation mutants of XGIPC that retain the PDZ domain blocks insulin-induced Xenopus MAP kinase activation and oocyte maturation. Human GIPC was also identified as a binding partner for human IGF1R (Ligensa et al.,2001).

IGF signaling has been recently implicated in neural induction in Xenopus and zebrafish embryos(Eivers et al., 2004; Pera et al., 2003; Pera et al., 2001; Richard-Parpaillon et al.,2002). Ectopic expression of IGF in dorsal cells leads to the induction of ectopic eyes and eye expansion and, when expressed in ventral cells, induces secondary head-like structures(Pera et al., 2001; Richard-Parpaillon et al.,2002). Inhibition of IGF signaling by either dominant-negative IGF1R or IGF1R depletion reduces head structures(Pera et al., 2001; Richard-Parpaillon et al.,2002). As kermit 2/XGIPC physically interacts with XIGF1R, we examined the involvement of kermit 2/XGIPC in IGF signaling in Xenopus embryonic development.

We report here that kermit2/XGIPC is expressed throughout Xenopus early embryonic development and is localized to the anterior region during neurula stage in a pattern highly similar to XIGF1R(Richard-Parpaillon et al.,2002). Knockdown of kermit 2 leads to embryos with reduced anterior structures, specifically reduction in the presumptive eye region. Furthermore, depletion of kermit 2 and expression of dominant-negative XIGF1R synergistically inhibit eye development. Kermit 2 is required for IGF1 induced eye formation in whole embryos and for the induction of eye molecular markers in ectodermal explants. Finally, we present evidence that kermit 2 is required to maintain IGF/PI3 kinase-dependent activation of AKT. These results indicate that kermit 2/XGIPC is required for IGF signaling in Xenopus eye development likely through its interaction with the IGF1 receptor.

kermit2 was identified through a Blast search of the Xenopus EST database using the kermit sequence(Tan et al., 2001). The GenBank accession number for kermit2 is BC089139. kermit25′UTR sequence was identified from a stage 10 Xenopus5′RACE cDNA library. Capped mRNAs were synthesized by in vitro transcription of plasmids encoding XIGF1R(Richard-Parpaillon et al.,2002), DN-IGFR (Pera et al.,2001), XIGF1 (Pera et al., 2001), p110^*(Carballada et al., 2001),kermit 2/pCS2, kermit 2-GFP/pCS2 and 5′UTR-kermit 2-GFP/pCS2 (described below) using SP6 mMessage mMachine kits (Ambion). Kermit 2/pCS2 was cloned by RT-PCR amplifying stage 1 Xenopus RNA using primers (forward,AGGGATCCATGCCTCTGGGATTGCGCGTA; reverse, AGTCTAGATTAAAAGCGTCCTTGTTTAGC) with engineered BamHI and XbaI sites and cloned directionally into pCS2+. Kermit 2-GFP/pCS2 was generated by PCR using kermit 2/pCS2 as the template and cloned into BamHI/EcoRI sites of pCS2-EGFP. 5′UTR-kermit 2-GFP/pCS2 was cloned by PCR amplifying using primers(upstream, CGGGATCCCAGGACTAAGGAACAGGAGGCAGG; downstream,AGGAATTCAAAGCGTCCTTGTTTAGCATC) with engineered BamHI and EcoRI sites and cloned directionally into pCS2-EGFP.

RNA isolation and RT-PCR methods are described elsewhere(Yang et al., 2002). Primers for EF-1α (Yang et al.,2002), ODC (Yang et al., 2002), Otx2 (Cox and Hemmati-Brivanlou, 1995), En2(Hemmati-Brivanlou et al.,1994), muscle actin(Hemmati-Brivanlou and Melton,1994), XAG (Blitz and Cho, 1995) and Xrx(Mathers et al., 1997) have been described previously. Primers were designed to detect kermit2(upstream, ATGCCTCTGGGATTGCGCGTAAAG; downstream, TTTAACTTTGTCAATCGTGCTCTC), HoxD1 (upstream, CACTTCTTGCGGGGATGTTT; downstream,AGAGTCCTGTAGCTCAGCTG), Pax6 (by N. Hirsch and W. Harris; upstream,AGTGTCCTCATTCACATCG; downstream, AGTACTGAGACATGTCAGG) and Vent1(upstream, GCATCTCCTTGGCATATTTGG; downstream, TTCCCTTCAGCATGGTTCAAC).

Embryos were co-injected with RNA encoding β-galactosidase with a nuclear localization motif and control morpholino or kermit 2 morpholino(5′-AGAGGCATCTTTCTTTCAGCGAAGG-3′). β-Galactosidase activity was visualized in embryos with Red-Gal (Research organics). Whole-mount in situ hybridization was performed as described(Deardorff et al., 1998). Antisense probes detected chordin(Sasai et al., 1994),gooscoid (Blumberg et al.,1991), Bf1(Papalopulu and Kintner,1996), Otx2 (Lamb et al., 1993), Pax6(Hirsch and Harris, 1997) and Xrx (Mathers et al.,1997). For sense probes, kermit 2/pCS2 was linearized by NotI and transcribed by Sp6 polymerase. For antisense probes, kermit 2/pCS2 was linearized by BamHI and transcribed by T7 polymerase.

Eggs were obtained from Xenopus females, in vitro fertilized and microinjected as described (Deardorff et al., 1998). Each injected blastomere received 10 nl of RNA or morpholino. For unilateral injections, morpholino and/or mRNA were injected into one dorsal blastomere at the four-cell stage. Animal pole explants assays were performed as described previously(Deardorff et al., 2001). For immunoprecipitation, Xenopus embryos were injected in the animal pole at the one-cell stage, cultured to stage 10 and lysed in embryo lysis buffer[20 mM Tris (pH 7.5), 140 mM NaCl, 10% glycerol, 1 mM DTT, 2 mM sodium vanadate, 25 mM NaF, 1% Nonidet P-40 and protease inhibitor cocktail for mammalian cells (Sigma)]. Anti-GFP monoclonal antibody (1 μl) (Roche, 1 814 460) was incubated with 250 μl of cleared lysate overnight at 4°C,collected on 30 μl protein G agarose beads (Invitrogen, 15920-010) for 1 hour, washed three times with PBS, and analyzed by western blot using IGF1R antibody (Cell Signaling, 3022). AKT, P-AKT, MAPK, P-MAPK and β-tubulin antibodies are from Sigma (p2482), Cell Signaling (9271), Cell Signaling(9102), Sigma (M9692) and BD Biosciences (556321), respectively.

Phosphorylated histone H3 staining was performed as described(Bellmeyer et al., 2003). TUNEL staining protocol was from www.xenbase.org

To characterize the temporal pattern of kermit2 expression, we extracted RNA from multiple stages of embryogenesis and performed RT-PCR. kermit2 was detected maternally (four-cell embryos) and throughout early embryonic development (Fig. 1A). kermit2 is ubiquitously expressed in early and mid-neurula stage embryos (Fig. 1B and data not shown). As neurulation progresses, kermit2 expression is restricted to the anterior region(Fig. 1B,D), including the cement gland and neural plate border, adjacent to and/or overlapping the presumptive eye region (Fig. 1D). The expression pattern of kermit2 is very similar to XIGF1R at this stage(Richard-Parpaillon et al.,2002). By tailbud stages, kermit2 expression becomes further restricted, with strong expression in the cement gland, otic vesicles(Fig. 1E), pronephros,branchial arches (Fig. 1F), and cranial nerves V, VII and IX (data not shown).

To study the endogenous function of kermit 2/XGIPC in Xenopusembryos, we depleted kermit 2 using morpholino antisense oligonucleotides. The kermit 2 morpholino blocked translation of kermit 2-GFP mRNA containing the endogenous 5′UTR but had no effect on kermit 2-GFP mRNA lacking the 5′UTR(Fig. 2A). Unilateral injection(into one dorsal blastomere at the four-cell stage) of kermit 2 morpholino strongly inhibited anterior development, especially eye development, in a dose-dependent manner (Fig. 2B;data not shown). Although there was a distribution of phenotypes, 60%(n=38) of embryos receiving 40 ng of kermit 2 morpholino developed without any detectable eyes on the injected side and 32% had small eyes(Fig. 2B). Co-injection of kermit2-GFP mRNA lacking the morpholino target sequence restored eye formation in 72% (n=43) of kermit2-depleted embryos, and of these, 28% appeared completely normal (Fig. 2B). kermit2-GFP mRNA alone did not cause apparent developmental defects. In addition, bilateral injection of kermit 2 morpholino caused absent or miniscule eyes in 83% (n=73) of embryos, and this effect was reversed in 60% (n=85) of embryos co-injected with the kermit2-GFP rescuing mRNA, although bilateral injections were also associated with more severe trunk defects that were not completely rescued by kermit2-GFP (data not shown). These results suggest that the reduced eye formation phenotype is specifically due to loss of kermit 2.

To characterize further the kermit 2 loss-of-function phenotype, we analyzed molecular markers at various stages by in situ hybridization. Kermit 2 depletion disrupted anterior development, which could indicate direct inhibition of anterior neural development or could alternatively indicate mild ventralization. To examine whether kermit 2 plays a role in dorsal development, we examined the expression of the dorsal organizer genes chordin and goosecoid. Kermit 2 or control morpholinos were injected into two dorsal animal blastomeres of four-cell/eight-cell embryos,and chordin or goosecoid expression was assessed at the early gastrula stage (Fig. 3A-D). Expression of chordin and goosecoid was not affected by depletion of kermit 2, suggesting that loss of kermit 2 does not disrupt early dorsal specification. We then examined the expression of the anterior neural markers Bf1 (a forebrain marker), Otx2(expressed in the forebrain, eyes and anterior midbrain) and Pax6 (a forebrain and eye marker) at the neurula stage (stage 20). To our surprise,unilateral injection of kermit 2 morpholino into dorsoanimal blastomeres did not obviously affect Bf1 expression (80% no change, n=45, Fig. 3F). Furthermore Pax6 and Otx2 expression were reduced only in the presumptive eye domain (Pax6, 93% eye reduction, n=75; Otx2, 81% eye reduction, n=32, Fig. 3H,J). Pax6expression in the spinal cord was not affected. Consistent with an apparently eye-specific phenotype, the expression of the eye-specific markers Xrx (85% reduction, n=32, Fig. 3L) and Xath5(data not shown) was also strongly reduced by kermit 2 depletion. This eye-specific phenotype is probably due to loss of kermit 2, as co-injection of kermit2 mRNA lacking the 5′UTR partially recovered Xrxexpression in 73% (n=26, Fig. 3S, compared with 3R and 3L) of kermit 2-depleted embryos. We also analyzed the expression of Xrx and Pax6 at earlier stages of development (early neurula stage, stage 14) and found that depletion of kermit 2 did not have an obvious effect on Xrx (74% no change, n=49, Fig. 3N) or Pax6 (84% no change, n=50, Fig. 3P) expression at this stage, suggesting that kermit 2 is required for the maintenance, but not the initiation, of eye formation.

Kermit 2, despite its high amino acid similarity with Kermit, does not bind to the Kermit interaction partner, frizzled 3 (unpublished data). However,kermit 2/XGIPC was identified as an XIGF1R-binding protein by Booth et al. from a yeast two-hybrid screen (Booth et al., 2002). To study the role of kermit 2/XGIPC in the IGF signaling pathway, we confirmed that kermit 2/XGIPC interacts with the IGF1R in Xenopus embryos. Full-length XIGF1R coimmunoprecipitated with GFP-tagged kermit 2 expressed in gastrula-stage embryos(Fig. 4A), suggesting that kermit 2/XGIPC physically interacts with the XIGF1R in Xenopus.

Inhibition of IGF signaling by a dominant-negative XIGF1R (DN-IGFR) reduces the size of anterior structures, including eyes, in Xenopus embryos(Pera et al., 2001). To test for a functional interaction between kermit 2 and IGF1R, we used doses of the kermit 2 morpholino and DN-IGFR mRNA that alone only mildly reduce eye formation (Fig. 4B). When co-injected, kermit 2 morpholino and DN-IGFR led to embryos with strongly reduced eyes (Fig. 4B),suggesting a synergistic effect between the two. Furthermore, although these levels of kermit 2 morpholino or DN-IGFR alone minimally affected Pax6 expression (72.5%, n=24; 100%, n=27,respectively), co-injection strongly reduced Pax6 expression in the presumptive eye in 82% (n=22) of the embryos(Fig. 4C). Taken together,these data indicate that kermit 2/XGIPC could be involved in the IGF pathway,potentially functioning at the receptor level.

In order to test the requirement of kermit 2/XGIPC for IGF signaling, we examined whether depletion of kermit 2 inhibits IGF function in embryos. Dorsal overexpression of IGF1 mRNA has been shown to induce ectopic eyes and overgrowth of endogenous eyes(Pera et al., 2001; Richard-Parpaillon et al.,2002). We found similarly that dorsal injection of XIGF1mRNA caused expanded eyes in 64% (n=22) of the embryos. Depletion of kermit 2 dramatically inhibited the XIGF1 induced eye phenotype and led to embryos (73%, n=20) with small eyes or no eyes(Fig. 5A). Kermit 2 depletion alone reduced eye formation (similar to Fig. 2). Dorsal injection of IGF1 mRNA also expanded cement gland formation in embryos, but this phenotype was not affected by kermit 2 depletion.

Kermit 2 is also required for IGF induced expression of eye markers in animal cap explants. As reported (Pera et al., 2001; Richard-Parpaillon et al., 2002), expression of IGF1 alone in animal caps induces anterior neural markers (Fig. 5B, lane 4), including XAG (a cement gland marker), Otx2 (a marker for the forebrain and anterior midbrain), Pax6 (a marker for eyes and the forebrain) and Xrx (an eye marker), but not the midbrain-hindbrain boundary marker En2 or the mesodermal marker muscle actin. Kermit 2 depletion strongly reduced the induction of Xrx, mildly reduced Pax6 induction, and did not affect the induction of the anterior neural markers XAG and Otx2 (lane 6). Co-injection of control morpholino did not affect any of the IGF1-induced markers (lane 5). These results suggest that kermit 2/XGIPC is required for IGF1-induced eye formation in Xenopus.

As shown above, kermit 2 is required for IGF induced embryonic phenotypes,but whether it is required for the activation of downstream IGF signaling is still unknown. The two main intracellular pathways activated by IGF/IGFR are the PI3 kinase/AKT pathway and the MAP kinase pathway(Oldham and Hafen, 2003). We first examined the requirement of kermit 2/XGIPC for IGF signaling in Xenopus oocytes. Stage VI oocytes have low levels of background activation of AKT or MAP kinase, but respond robustly to exogenous IGF1. We injected control morpholino or kermit 2 morpholino into oocytes and cultured these oocytes for 48 hours to deplete endogenous kermit 2. We then treated these oocytes with IGF1 protein for 30 minutes or overnight (18 hours). IGF1 induced phosphorylation of AKT (Ser 473) and MAPK (e.g. Thr183 and Tyr185 in ERK2) at sites associated with activation as documented previously(Alessi et al., 1996; Payne et al., 1991; Stephens et al., 1998; Stokoe et al., 1997; Yung et al., 1997). Kermit 2 depletion blocked phosphorylation/activation of AKT induced by long-term exposure to IGF1, but not by short-term exposure (30 minutes)(Fig. 6A). MAP kinase phosphorylation induced by IGF1 protein was not affected by depletion of kermit 2 at any time points (Fig. 6A). The control morpholino did not affect IGF1-induced AKT or MAP kinase phosphorylation/activation. These data suggest that kermit 2/XGIPC is required to maintain activation of AKT induced by IGF1, but is not required for the initiation of signaling.

This analysis was extended to ectodermal (animal cap) explants. Compared with oocytes, animal caps have high levels of endogenous AKT and MAP kinase phosphorylation/activation. We first tested the effect of kermit 2 morpholino on the endogenous phosphorylation of AKT and MAP kinase. Control morpholino or kermit 2 morpholino was injected into fertilized eggs. Animal caps were dissected at the late blastula stage (stage 9) and cultured for 2 hours (stage 10/10.5) or overnight (stage 20). Similar to the result in oocytes, kermit 2 depletion inhibited AKT phosphorylation/activation after overnight incubation,but not after 2 hours incubation; kermit 2 morpholino had no effect on MAP kinase phosphorylation (Fig. 6B). To test whether kermit 2 is required for AKT phosphorylation/activation by exogenous IGF1, embryos were injected with IGF1 mRNA with or without kermit 2 morpholino, animal caps were explanted at the blastula stage and cultured until late neurula stage (stage 20). We found that kermit 2 depletion blocked phosphorylation/activation of AKT by overexpressed IGF1 (Fig. 6C, compare lane 5 with lane 4). (Kermit 2 morpholino did not interfere with translation of injected mRNAs, as GFP mRNA co-injected with IGF1 mRNA was expressed at equal levels with or without kermit 2 morpholino co-injection.) Taken together, the results in oocytes and animal cap explants suggest that kermit 2/XGIPC is required for maintaining IGF induced AKT activation, but not for MAPK activation.

As kermit 2 depletion strongly inhibits the AKT phosphorylation/activation induced by IGF1, we examined the physiological relevance of this inhibition by rescuing the kermit 2 loss-of-function phenotype using a constitutively active p110 subunit (p110^*) of phosphatidylinositol-3′ kinase (PI3 kinase) (Carballada et al.,2001), a lipid kinase upstream of AKT and downstream of IGF1R. Expression of p110^* partially rescues the kermit 2 depletion phenotype (Fig. 7B). In a typical experiment, 96% (n=24) of the embryos injected dorsally with the kermit 2 morpholino alone had either miniscule or no apparent eyes,whereas only 26% (n=20) of embryos co-injected with p110^*showed this phenotype, and 74% developed small but normal-appearing eyes. Dorsal expression of p110^* itself did not cause apparent developmental defects in embryos. p110^* was confirmed to be active as it induced strong AKT phosphorylation in gastrula-stage animal caps(Fig. 7A). In addition to the morphological rescue, p110^* also partially recovers the expression of Xrx, an eye marker, in kermit 2-depleted embryos. Compared with control embryos, most kermit 2-depleted embryos had strongly reduced Xrx expression (63%, n=39; Fig. 7D, middle panel) at the neurula stage (stage 19), whereas the majority of embryos co-injected with p110^* showed only a mild reduction of Xrx expression (73%, n=53; Fig. 7D, right panel). These rescue results suggest that the kermit 2 loss-of-function phenotype is at least partially due to the inhibition of the PI3 kinase/AKT pathway.

The IGF pathway affects multiple cellular processes, including cell survival and cell proliferation. We examined whether kermit 2/XGIPC is involved in these processes. We injected control morpholino or kermit 2 morpholino into the right side of embryos (leaving the left side as the control) and followed apoptosis by TUNEL analysis. Seventy-two percent(n=61) of kermit 2-depleted embryos demonstrate a substantial increase in the number of TUNEL-positive nuclei on the injected side compared with the control side (Fig. 8B). This effect was dose dependent (data not shown) and no difference was detected with control morpholino at equivalent doses(Fig. 8A). Expression of kermit2 mRNA without the 5′UTR significantly reduced TUNEL-positive nuclei in 83% (n=66) of kermit 2-depleted embryos(Fig. 8C), which indicates that the elevated cell death was specifically due to loss of kermit 2. Expression of the same amount of nβ-gal mRNA did not reduce apoptosis in kermit 2-depleted embryos (data not shown). We also assessed whether kermit 2 regulates cell proliferation. Embryos unilaterally injected with kermit 2 morpholino were immunostained with an antibody against phosphorylated histone H3, which specifically recognizes mitotic chromosomes. We did not observe an obvious change in the number of mitotic cells on the injected side compared with the control side in early neurula stage (stage 14)and late neurula stage (stage 22) embryos(Fig. 8E,F). These results indicate that kermit 2/XGIPC is required for cell survival, but do not support a requirement for kermit 2 in proliferation in Xenopus embryos. As kermit 2 is involved in IGF signaling, we further tested whether inhibition of the IGF pathway increases cell death in Xenopus embryos. Indeed,DN-IGFR, like kermit 2 depletion, markedly increased the number of TUNEL-positive nuclei (76%, n=29; Fig. 8D) when expressed in embryos. Activation of the MAPK branch of IGF signaling leads to transcriptional responses associated with cell proliferation, whereas activation of the PI3K/AKT branch regulates the activity of BCL2 family members and anti-apoptotic responses(O'Connor, 2003; Oldham and Hafen, 2003). Therefore, these data are consistent with a requirement for kermit 2 in IGF-induced AKT activation. Finally, to determine whether the kermit 2 depletion phenotype is due to apoptosis of cells within the eye-field, we examined whether inhibition of apoptosis restores eye formation in kermit 2-depleted embryos. Kermit 2 morpholino injection alone strongly reduced Pax6 expression in the presumptive eye in 75% (n=24, Fig. 8H) of embryos. Co-expression of bcl2 rescued this phenotype and restored Pax6 expression in the presumptive eye in almost 72% (n=25, Fig. 8I) of embryos, as well as restoring eye development to normal or near normal morphology in a majority of sibling embryos allowed to develop to later stages (data not shown). This result suggests that the kermit 2 phenotype is at least partially due to elevated cell death in the presumptive eye region.

We report here the identification of the second Xenopus GIPC homolog, kermit 2. Despite its high amino acid similarity with the first Xenopus GIPC homolog Kermit, kermit 2 has a different function in Xenopus embryos. Loss-of-function and molecular marker analyses indicate that kermit 2 is specifically required for Xenopus eye development. We have confirmed the physical interaction of kermit 2/XGIPC with XIGF1R in Xenopus and further established that kermit 2 is required for IGF-dependent eye development. In addition, kermit 2 is required for maintaining IGF induced AKT phosphorylation/activation and a constitutively active PI3 kinase partially rescues kermit 2 loss of function. These results provide the first demonstration of the in vivo role of GIPC in embryonic development and also show that kermit 2/XGIPC is a novel component of the IGF pathway, potentially acting through the modulation of IGF1R function.

Inhibition of IGF signaling in Xenopus by either dominant-negative IGF receptor or IGF receptor depletion disrupts anterior neural development broadly (Pera et al., 2001; Richard-Parpaillon et al.,2002), yet loss of kermit 2/XGIPC specifically inhibits eye formation without apparent effect on other aspects of anterior neural development. At least two mechanisms could explain the morerestricted requirement for kermit 2/XGPIC. First, depletion of kermit 2 blocks the activation of AKT, but not of MAPK. It is likely that the IGF/PI3K/AKT branch is specifically involved in eye formation, while the IGF/MAPK branch regulates anterior neural development more broadly. Indeed, IGF inhibits SMAD1 activity and induces neural differentiation via MAPK(Pera et al., 2003; Kuroda et al., 2005). Second,eye development could have a distinct temporal requirement for IGF signaling. For example, the specification of anterior neural tissue in general may only require an early exposure to IGF, whereas proper eye formation may require continual activation of IGF signaling. As kermit 2/XGIPC is required for the maintenance, but not the initiation of IGF signaling, depletion of kermit 2/XGIPC may inhibit eye formation only at late neurula stages, without disrupting general neural specification.

We also found that depletion of kermit 2 dramatically increases cell death in embryos, but does not reduce cell proliferation. This result is consistent with the data that kermit 2 depletion blocks IGF induced AKT activation, not MAPK activation. However, the increased cell death in kermit 2-depleted embryos is not restricted to the presumptive eye field. Thus, the extent to which increased cell death contributes to the eye phenotype remains to be established.

As kermit 2/XGIPC functions downstream of the IGF ligand and upstream of PI3 kinase and physically associates with the XIGF1R, kermit 2 probably regulates IGF signaling through modulation of the receptor. In principle,kermit 2 could regulate the stability, activity or subcellular localization of the receptor. However, kermit 2 depletion did not obviously reduce the level of XIGF1R under conditions that reduced AKT activation (data not shown),arguing against kermit 2 regulation of IGF1R stability. Furthermore, kermit 2 depletion did not significantly affect the level of IGF-induced MAPK phosphorylation, arguing against a role for kermit 2 in regulating overall stability or activity of the IGF1R. Kermit 2 depletion also did not interfere with the early (30 minute) phosphorylation of AKT in response to IGF,suggesting that coupling to downstream signaling, at least initially, is intact.

Alternatively, mammalian GIPC is involved in the regulation of endocytic trafficking, and kermit 2/XGIPC may similarly regulate the subcellular localization of the IGF receptor. A function for GIPC in endocytic trafficking was first proposed based on its localization to endocytic vesicles(Dance et al., 2004; De Vries et al., 1998a; De Vries et al., 1998b; Jeanneteau et al., 2004a; Lou et al., 2002; Lou et al., 2001). For example, GIPC is enriched at clathrin-rich invaginations and the endocytic compartments found between microvilli in proximal tubule kidney cells(Lou et al., 2002). A role for GIPC in endocytosis is also supported by functional studies in cultured cells(Hirakawa et al., 2003; Jeanneteau et al., 2004a). Hirakawa et al. used RNAi to knockdown GIPC in 293 cells and found that GIPC is partially responsible for maintaining a relatively constant level of the human lutropin receptor (LHR) at the cell surface during ligand-induced internalization and for recycling of the ligand (CG)(Hirakawa et al., 2003). In a similar manner, kermit 2 could be required for the recycling of the IGF1R to the plasma membrane after ligand-dependent internalization. A failure to recycle internalized IGF1R could explain why depletion of kermit 2 reduces AKT phosphorylation after prolonged exposure to IGF, but not after short-term exposure. A defect in recycling of IGF1R would not be expected to disrupt MAP kinase phosphorylation, as MAP kinase can be activated by receptors that have been internalized in association with β-arrestin and component kinases of the MAPK cascade (DeFea et al.,2000; Lefkowitz and Shenoy,2005; Lin et al.,1998; Luttrell et al.,2001; Tohgo et al.,2003), while AKT and PI3 kinase have not been reported to be activated by internalized receptors within endosomes. However, we have so far been unable to demonstrate an effect of kermit 2 depletion on IGF1R trafficking, and this will remain a focus of future studies.

In summary, we have shown that kermit 2/XGIPC is required for eye development in Xenopus embryos. Kermit 2/XGIPC physically and functionally interacts with the IGF1R and is required for IGF signaling in anterior neural development specifically in eye formation. Expression of molecular markers of eye development is induced but not maintained in kermit 2-depleted embryos, and phosphorylation of AKT is similarly induced but not maintained after prolonged exposure to IGF when kermit 2 is depleted. Eye development can be partially rescued in kermit 2-depleted embryos by an active PI3 kinase. Based on these observations, we propose that kermit 2/XGIPC,through the modulation of IGF1R, mediates endogenous IGF signaling in Xenopus eye formation.

We thank Eddy M. De Robertis, Laurent Richard-Parpaillon, Patrick Lemaire,Monica Vetter, Vijayasaradhi Setaluri, Carole LaBonne, Dan Kessler and Jean-Pierre Saint-Jeannet for plasmids and reagents. We thank Dan Kessler,Jean-Pierre Saint-Jeannet, Mary Mullins, Jonathan Epstein, Tom Kadesch and Jing Yang for helpful discussions.