Research ArticleDevelopmental Biology

The Kinase SGK1 in the Endoderm and Mesoderm Promotes Ectodermal Survival by Down-Regulating Components of the Death-Inducing Signaling Complex

Science Signaling  18 Jan 2011:
Vol. 4, Issue 156, pp. ra2
DOI: 10.1126/scisignal.2001211

Abstract

A balance between cell survival and apoptosis is essential for animal development. Although proper development involves multiple interactions between germ layers, little is known about the intercellular and intertissue signaling pathways that promote cell survival in neighboring or distant germ layers. We found that serum- and glucocorticoid-inducible kinase 1 (SGK1) promoted ectodermal cell survival during early Xenopus embryogenesis through a non–cell-autonomous mechanism. Dorsal depletion of SGK1 in Xenopus embryos resulted in shortened axes and reduced head structures with defective eyes, and ventral depletion led to defective tail morphologies. Although the gene encoding SGK1 was mainly expressed in the endoderm and dorsal mesoderm, knockdown of SGK1 caused excessive apoptosis in the ectoderm. SGK1-depleted ectodermal explants showed little or no apoptosis, suggesting non–cell-autonomous effects of SGK1 on ectodermal cells. Microarray analysis revealed that SGK1 knockdown increased the expression of genes encoding FADD (Fas-associated death domain protein) and caspase-10, components of the death-inducing signaling complex (DISC). Inhibition of DISC function suppressed excessive apoptosis in SGK1-knockdown embryos. SGK1 acted through the transcription factor nuclear factor κB (NF-κB) to stimulate production of bone morphogenetic protein 7 (BMP7), and overexpression of BMP7 in SGK1-knockdown embryos reduced the abundance of DISC components. We show that phosphoinositide 3-kinase (PI3K) functioned upstream of SGK1, thus revealing an endodermal and mesodermal pathway from PI3K to SGK1 to NF-κB that produces BMP7, which promotes ectodermal survival by decreasing DISC function.

Introduction

Apoptosis, a form of programmed cell death, is required for forming or deleting structures, controlling cell numbers, and eliminating abnormal, misplaced, or nonfunctional cells during animal development (1). The central apoptotic machinery driven by caspases is an intracellular controller for apoptotic cell death. Spatiotemporal regulation of this intracellular machinery throughout embryogenesis is important for normal development of multicellular organisms. Indeed, misregulation of the apoptotic machinery in vertebrate embryos results in abnormal development of the nervous system, notochord, and heart (24). However, an intercellular control mechanism for this intracellular machinery during developmental apoptosis remains to be elucidated.

The establishment of the vertebrate body plan begins with the formation of three germ layers: ectoderm, mesoderm, and endoderm. Although proper development involves multiple interactions between germ layers (5), intercellular or intertissue signaling pathways that promote cell survival in neighboring or distant germ layers during development are unclear. Here, we have identified an evolutionarily conserved serine-threonine kinase SGK1 (serum- and glucocorticoid-inducible kinase 1) (6, 7) as an endoderm- or dorsal mesoderm–resident signaling molecule that promotes survival of ectodermal cells.

SGK1 was isolated as a product of an immediate-early gene induced by serum and glucocorticoids in rat mammary tumor cells (8, 9). It belongs to the AGC family of protein kinases, which also includes protein kinase A (PKA), Akt, protein kinase C (PKC), p90 ribosomal S6 kinase (p90RSK), and p70 ribosomal S6 kinase (p70S6K). Within the AGC family, SGK is closely related to Akt, which is activated in response to growth and survival factors and plays a critical role in promoting cell survival (10). Like Akt, SGK1 is activated by phosphorylation at a threonine residue within the T loop of the kinase domain and a serine residue within the C-terminal hydrophobic motif, downstream of the phosphoinositide 3-kinase (PI3K) signaling pathway, which is activated by insulin and other growth factors (11, 12). SGK1 promotes the survival of cultured cells, such as MCF-7 breast cancer cells (13), primary cultured cerebellar granule neurons (14), and NmuMg mammary epithelial cells (15). However, an in vivo requirement of SGK1 for cell survival and its role in early development have not been reported. We investigated the in vivo role of SGK1 in Xenopus laevis, a model vertebrate for the analysis of gene function during development, and identified an inter–germ layer survival signaling pathway involving SGK1.

Results

xSGK1 is mainly expressed in endoderm and dorsal mesoderm during Xenopus development

A search of a public database revealed that two pseudoalleles of SGK1, which share 95.0% nucleotide identity and 97.0% amino acid identity, are present in X. laevis (fig. S1A). We named them xSGK1a (GenBank accession number BC073077) and xSGK1b (GenBank accession number BC074305), respectively. Both xSGK1a and xSGK1b consist of 434 amino acids, and they are 82.6% and 81.4% identical to human SGK1, respectively (fig. S1A). Reverse transcription polymerase chain reaction (RT-PCR) analysis with a pair of primers designed to detect both isoforms of xSGK1 showed that xSGK1 is expressed maternally and throughout early embryogenesis (Fig. 1A). Whole-mount in situ hybridization analysis revealed that xSGK1 is expressed in the vegetal hemisphere at the blastula stage, in the endoderm and the dorsal mesoderm at the gastrula stages, and in the endoderm at the neurula stage (Fig. 1B and fig. S1B). At the tailbud stage, we observed xSGK1 expression mainly in the pineal gland and the dorsal fin (Fig. 1B and fig. S1B). Thus, xSGK1 exhibits a dynamic expression pattern during embryogenesis.

Fig. 1

Essential roles of xSGK1 in Xenopus development. (A) Temporal expression pattern of xSGK1 in Xenopus development. Total RNA isolated from embryos at the indicated stages was subjected to RT-PCR analysis. Xenopus embryonic ornithine decarboxylase (XeODC) was used as an RNA loading control. Xbra was also examined. (B) Whole-mount in situ hybridization against xSGK1 at the indicated stages. Stage 10.5 and 12 embryos are photographed with dorsal at the top. Medial sagittal section of the stage 12 embryo is also shown. The arrowhead indicates the dorsal blastopore lip. Stage 15 and 25 embryos are with anterior to the left. No detectable signal was seen with the sense probe (fig. S1B). (C) Two antisense MOs for xSGK1 (xSGK1-MO1 and xSGK1-MO2). The bars denote the targeted sites recognized by xSGK1-MOs. (D) xSGK1-MOs block translation of injected xSGK1 mRNAs. Indicated sets of MO (20 ng) and mRNA (1.0 ng) were injected into Xenopus embryos, and the protein abundance was examined by immunoblotting. We used HA-tagged xWnt8 as a control. (E and F) Effects of xSGK1-MOs on whole embryos. Control MO (60 ng) or xSGK1-MOs (30 ng each of MO1 and MO2) were injected into the indicated region at the four-cell stage. Dorsal injection of xSGK1-MOs caused shortened body axes and reduced head structures, as well as defective eyes [58%; (E), upper right, and (F), upper] or only defective eyes [33%; (E), upper right, and (F), upper]. Ventral injection of xSGK1-MOs caused a defective tail [88%; (E), lower right, and (F), lower].

xSGK1 is essential for proper formation of head, eye, and tail

To assess the role of xSGK1 in early Xenopus development by a loss-of-function approach, we designed two antisense morpholino oligonucleotides (MOs) against xSGK1 as follows: xSGK1-MO1 for xSGK1a and xSGK1-MO2 for xSGK1b (Fig. 1C). To determine MO efficacy, we injected xSGK1-MOs with 5′UTR-xSGK1a-Myc or 5′UTR-xSGK1b-Myc messenger RNA (mRNA), which consists of the 5′UTR (5′ untranslated region) and the coding region of xSGK1a or xSGK1b with a Myc tag at the C terminus. Immunoblot analysis of embryo lysates revealed that xSGK1-MO1 markedly reduced the protein abundance of xSGK1a and xSGK1b, and xSGK1-MO2 markedly reduced that of xSGK1b (Fig. 1D). Because there is only one mismatch between xSGK1a and xSGK1b in the region targeted by xSGK1-MO1 (Fig. 1C), it is not surprising that xSGK1-MO1 markedly reduced the abundance of xSGK1b. Neither of the MOs affected the abundance of Wnt8 (Fig. 1D, lower panels), indicating that xSGK1-MOs were specific. We injected control MO or xSGK1-MOs into the dorsal or ventral blastomeres of the four-cell–stage embryos and allowed the animals to develop. At tadpole stages, embryos dorsally injected with xSGK1-MOs exhibited shortened body axes and reduced head structures, as well as defective eyes (Fig. 1, E, upper right, and F, upper). Some showed the milder phenotype of only defective eyes (Fig. 1, E and F). Embryos ventrally injected with xSGK1-MOs showed a defective tail development (Fig. 1, E, lower right, and F, lower). Control MO had no effect on development (Fig. 1E, left). These results demonstrate that xSGK1 is required for both dorsal and ventral development.

Knockdown of xSGK1 causes excessive apoptosis in ectoderm

Previous reports using cultured cells showed that SGK1 promotes cell survival (1317). Therefore, we examined whether knockdown of xSGK1 induced apoptosis in Xenopus embryos. At the four-cell stage, we injected control MO or xSGK1-MOs into dorsal or ventral marginal zones, and at later stages, we subjected the embryos to whole-mount TUNEL (terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling) staining (Fig. 2, A and B). Embryos injected with xSGK1-MOs dorsally or ventrally appeared normal until the mid-gastrula stage (stage 11), as was the case for control MO (Fig. 2B). Cell death was observed in the xSGK1-MO–injected embryos at stage 13 within the dorsal and ventral ectodermal regions, respectively, with twice as much TUNEL staining in the ventrally injected embryos as in the dorsally injected embryos (Fig. 2, A and B). At stage 13, cell death was not observed in control MO–injected embryos. At the neurula stages (stages 15 and 17), control MO–injected embryos showed no dorsal TUNEL staining and a low amount of ventral TUNEL staining (Fig. 2, A and B). By stages 15 and 17, most of the embryos dorsally injected with xSGK1-MOs had TUNEL-stained cells in the neural ectoderm. All embryos ventrally injected with xSGK1-MOs had TUNEL-stained cells in the epidermal ectoderm. These results suggest that xSGK1 is required for ectodermal cell survival in both dorsal and ventral development.

Fig. 2

xSGK1 promotes the survival of ectodermal cells in a non–cell-autonomous manner. (A) xSGK1-MOs induce excessive apoptosis. Control MO (60 ng) or xSGK1-MOs (30 ng each of MO1 and MO2) were injected into the indicated region of four-cell–stage embryos, and injected embryos were subjected to whole-mount TUNEL staining at the indicated stages. Positive cells are dark blue. Anterior is upward (embryos labeled “Dorsal” or “Ventral”) and to the left (embryos labeled “Lateral”).The number of embryos tested was as follows: n = 16 (St 11), n = 19 (St 13), n = 26 (St 15), n = 21 (St 17) for dorsal injection of control MO; n = 21 (St 11), n = 21 (St 13), n = 27 (St 15), n = 17 (St 17) for dorsal injection of xSGK1-MOs; n = 20 (St 11), n = 28 (St 13), n = 25 (St 15), n = 15 (St 17) for ventral injection of control MO; n = 23 (St 11), n = 16 (St 13), n = 17 (St 15), n = 20 (St 17) for ventral injection of xSGK1-MOs. (B) Quantification of the results shown in (A). (C) xSGK1-depleted ectodermal explants show little or no apoptosis. Control MO (80 ng) or xSGK1-MOs (40 ng each of MO1 and MO2) were injected into the animal pole of four-cell–stage embryos. Ectodermal explants (animal caps) were excised at stage 9, cultured until sibling embryos reached stage 17, and then subjected to TUNEL staining. (D) xSGK1 knockdown in the prospective dorsal endoderm and mesoderm causes apoptosis in the dorsal ectoderm. Control MO (60 ng) or xSGK1-MOs (30 ng each of MO1 and MO2) were injected into two dorsal vegetal blastomeres of eight-cell–stage embryos, and injected embryos were subjected to whole-mount TUNEL staining at stage 13. Five of nine xSGK1-MO–injected embryos showed excessive apoptosis. (E) Rescue experiments show that the phenotypes in xSGK1-knockdown embryos are specific and that the morphological defects are due to increased apoptosis. Myc-xSGK1b is an xSGK1-MO–resistant construct. Indicated sets of MO (20 ng) and Myc-xSGK1b mRNA (1.0 ng) were injected and the protein abundance was examined by immunoblotting. We used HA-tagged xWnt8 as a control. (F) Excess apoptosis induced by xSGK1-MOs is rescued by ectopic expression of xSGK1 or hBcl-2. xSGK1-MOs (20 ng each of MO1 and MO2) were injected into the right blastomeres at the four-cell stage, and injected embryos were subjected to TUNEL staining at stage 15. For rescue experiments, Myc-xSGK1b mRNA (1.0 ng) or hBcl-2 mRNA (170 pg) was coinjected with xSGK1-MOs. The TUNEL-stained embryos were classified into three groups according to the amount of apoptosis and quantified in the right bar graphs: (−), no TUNEL-positive staining; (+), moderate staining; (++), strong staining. (G) A defective eye caused by xSGK1-MOs is rescued by ectopic expression of xSGK1 or hBcl-2. xSGK1-MOs (10 ng each of MO1 and MO2) were injected into the dorsal right blastomere at the four-cell stage. For rescue experiments, Myc-xSGK1b mRNA (750 pg) or hBcl-2 mRNA (170 pg) was coinjected with xSGK1-MOs. The obtained phenotypes were classified into four groups according to the eye size and quantified in the lower bar graphs: class 1, normal or only slightly smaller; class 2, smaller; class 3, much smaller; class 4, none.

Expression of xSGK1 was high in the endoderm and dorsal mesoderm, whereas its expression in the ectoderm was low (Fig. 1B). To examine whether the modest expression of xSGK1 in the ectoderm contributed to ectodermal cell survival in a cell-autonomous manner, we used ectodermal explant assays, in which ectodermal regions were cultured separately from mesodermal and endodermal regions. xSGK1-MOs were injected into animal poles of four-cell–stage embryos, and then ectodermal regions were excised at stage 9 (Fig. 2C, left illustration). Explants were cultured until stage 17, when the excessive apoptosis phenotype was apparent in xSGK1-depleted whole embryos (Fig. 2, A and B). Both control and xSGK1-depleted ectodermal explants showed little or no apoptosis (Fig. 2C, right panels), which suggests that the excess apoptosis phenotype observed in xSGK1 morphants was not due to the depletion of xSGK1 in the ectoderm or nonspecific toxic effects of xSGK1-MOs. Therefore, we hypothesized that endodermal or mesodermal xSGK1 contributed to ectodermal cell survival in a non–cell-autonomous manner. To test this hypothesis, we injected xSGK1-MOs into two dorsovegetal blastomeres of eight-cell–stage embryos, from which dorsal endodermal and dorsal mesodermal cells, but not dorsal ectodermal cells, are mainly derived (Fig. 2D, left illustration). Dorsovegetal injection of xSGK1-MOs induced the excessive apoptosis phenotype in the dorsal ectoderm (Fig. 2D, right panels), supporting the idea that the depletion of xSGK1 in the endoderm and mesoderm was sufficient for increased apoptosis in the ectoderm. Together, these results suggest that xSGK1 ensures proper embryonic development by protecting ectodermal cells from apoptosis non–cell-autonomously.

To confirm the specificity of xSGK1-MOs, we performed a series of rescue experiments with an N-terminally Myc-tagged xSGK1b (Myc-xSGK1b), which contains the full-length coding sequence but not the 5′UTR. Neither xSGK1-MO1 nor xSGK1-MO2 reduced the abundance of Myc-xSGK1b (Fig. 2E), indicating that Myc-xSGK1b is an xSGK1-MO–resistant construct. We examined whether coinjection of Myc-xSGK1b suppressed excess apoptosis induced by xSGK1-MOs. We classified the injected neurulae into three groups according to the amount of apoptosis: (−), no TUNEL-positive staining; (+), moderate staining; (++), strong staining (Fig. 2F, right graphs). Whereas 46% of the embryos injected with xSGK1-MOs showed strong staining (++) (Fig. 2F, upper), only 21% of the embryos coinjected with Myc-xSGK1b exhibited much apoptosis (Fig. 2F, middle). We also examined whether coinjection of Myc-xSGK1b mRNA rescued the eye defect caused by xSGK1-MOs by injecting control MO, xSGK1-MOs, or xSGK1-MOs with Myc-xSGK1b into the dorsal right blastomere of four-cell–stage embryos. Whereas all embryos injected with control MO had normal eyes, 79% of the embryos injected with xSGK1-MOs showed severe defects, and coinjection of Myc-xSGK1b partially rescued the defective eye phenotype (Fig. 2G); these findings suggest that the defective eye phenotype is also a specific response to the depletion of xSGK1 protein.

Morphological defects in xSGK1-depleted embryos are rescued by the inhibition of apoptosis

To test whether the defective eye phenotype in xSGK1 morphants was due to increased apoptosis, we performed coinjection experiments with mRNA for the antiapoptotic protein Bcl-2. Coinjection of human Bcl-2 mRNA (hBcl-2) suppressed the amount of strong TUNEL-positive staining in xSGK1-knockdown embryos (Fig. 2F, lower) and partially restored eye formation (Fig. 2G, rightmost). Thus, increased apoptosis contributes to the morphological defects in xSGK1-knockdown embryos.

xSGK1 promotes survival of ectodermal cells by decreasing the abundance of DISC components

To identify the mechanism by which xSGK1 knockdown resulted in excessive apoptosis, we searched for genes that could be responsible for the apoptosis induced by the xSGK1-MOs by performing a genome-wide analysis with the Affymetrix GeneChip, an X. laevis genome array, which contains more than 14,400 transcripts (15,503 probe sets). We injected embryos radially with control MO or xSGK1-MOs at the four-cell stage and harvested the embryos at stage 12 for RNA preparation. [This is before excess apoptosis is observed in xSGK1 morphants (Fig. 2B).] Of the 15,503 probe sets, 71 belong to the “apoptosis” category according to the gene annotations provided by NetAffx (http://www.affymetrix.com/analysis/index.affx). Of these 71 probe sets, we focused on 22 (Fig. 3A) that were classified into proapoptotic genes according to the Web resource List of Apoptosis Regulators (http://www-personal.umich.edu/%7Eino/List/AList.html). The top three genes that were up-regulated in xSGK1-knockdown embryos encoded the Xenopus orthologs of FADD (Fas-associated death domain protein) (xFADD), caspase-10 (Xcaspase-10), and TRADD (tumor necrosis factor receptor–associated death domain) (xTRADD) (Fig. 3A), which constitute death-inducing signaling complex (DISC). DISC plays a crucial role in the activation of the caspase cascade and apoptosis (18). Quantitative RT-PCR (qRT-PCR) assays showed that the mRNAs for xFADD and Xcaspase-10 were increased in xSGK1-knockdown embryos (Fig. 3B). Thus, these results demonstrate that xSGK1 suppressed the expression of the genes encoding xFADD and Xcaspase-10 during early embryogenesis.

Fig. 3

xSGK1 promotes cell survival by down-regulating DISC components. (A) Expression of proapoptotic genes in xSGK1-MO–injected embryos relative to control MO–injected embryos (1.00). (B) qRT-PCR analyses for xFADD and Xcaspase-10. Embryos were radially injected with control MO (80 ng) or xSGK1-MOs (40 ng each of MO1 and MO2) at the four-cell stage and harvested at stage 12. The mRNA expression was analyzed by qRT-PCR analyses (results of two experiments shown). (C) Excess apoptosis in xSGK1-knockdown embryos is rescued by ectopic expression of a dominant-negative form of xFADD (dnxFADD). Embryos were injected with indicated combinations of MOs and mRNA and subjected to TUNEL staining. The TUNEL-stained embryos were classified into three groups according to the criteria defined in Fig. 2F.

Because Xenopus embryos injected with mRNA for xFADD or Xcaspase-10 showed excessive apoptosis (fig. S2), we hypothesized that the enhancement of apoptosis in xSGK1-knockdown embryos was the result of an increase in the abundance of the DISC components. Consistent with this hypothesis, expression of a dominant-negative form of xFADD (dnxFADD), which lacks a portion of its N-terminal death effector domain (DED) and inhibits DISC-mediated apoptosis (19, 20), suppressed xSGK1-MO–induced excess apoptosis in a dose-dependent manner (Fig. 3C). Thus, the increase in the abundance of DISC components contributes to the induction of apoptosis in xSGK1-knockdown embryos.

Our microarray analyses also revealed that injection of xSGK1-MOs increased the mRNA for the Xenopus ortholog of p53 (Xp53) (Fig. 3A). Although p53 is a tumor suppressor protein that has a primary role in the intrinsic (mitochondrial) apoptotic pathway, it also contributes to the extrinsic apoptotic pathway that is mediated by death receptors and DISC components (21). In situ hybridization assays for Xp53 showed that the injection of xSGK1-MOs expanded the area in the ectoderm that was positive for Xp53 (fig. S3A), which suggests that an increase in the abundance of Xp53 may also contribute to the induction of apoptosis in xSGK1-knockdown embryos. This idea is supported by our finding that xSGK1-MO–induced apoptosis was rescued by coinjection of hBcl-2 mRNA (an inhibitor of the intrinsic apoptotic pathway) (Fig. 2F). Thus, xSGK1 may promote cell survival in the ectoderm through multiple pathways involving DISC and Xp53.

xBMP7 mediates the suppressive action of xSGK1 on expression of genes encoding DISC components

We sought to identify the molecular mechanism underlying the increase in expression of the genes encoding xFADD and Xcaspase-10 in xSGK1-knockdown embryos. To determine whether common cis-regulatory elements exist within the putative promoter regions of the two genes, we isolated ~2.7 kilo–base pairs (kbp) of the 5′UTRs of the genes encoding xFADD and Xcaspase-10 by splinkerette PCR, a method for the amplification of unknown genomic sequences adjacent to a known sequence (22). A search of the TRANSFAC or NSITE database identified cis-regulatory elements for transforming growth factor–β (TGF-β) and bone morphogenetic protein (BMP)–responsive Smad-binding elements (SBEs) in the 5′UTR regions of the two genes (Fig. 4A); these findings suggest that TGF-β or BMP signaling may be involved in the regulation of the expression of these DISC component genes by xSGK1.

Fig. 4

xBMP7 functions downstream of PI3K–xSGK1–NF-κB to down-regulate DISC components. (A) Gene cloning analysis of xFADD and Xcaspase-10 promoters reveals that each contains three Smad-binding elements (SBEs) (xFADD: −19, −56, and −844 bp from ATG; Xcaspase-10: +117, −540, and −2252 bp from ATG). ORF, open reading frame. (B) The abundance of transcript for xBMP7 is decreased in xSGK1-depleted embryos. qRT-PCR analyses for xBMP7 are shown. The samples are the same as those in Fig. 3B; results of two experiments are shown. (C) The abundance of transcripts for xFADD and Xcaspase-10 is decreased by exogenous expression of xBMP7. Embryos were radially injected with xBMP7 mRNA (400 pg) at the four-cell stage and harvested at stage 13. qRT-PCR analyses for xFADD and Xcaspase-10 are shown. Error bars indicate SEM (n = 4). (D) xBMP7 rescues the increased expression of DISC components caused by xSGK1 knockdown. Control MO (80 ng) or xSGK1-MOs (40 ng each of MO1 and MO2) with or without xBMP7 mRNA (200 or 400 pg) were injected into four-cell–stage embryos. Injected embryos were harvested at stage 13. The mRNA expression was analyzed by qRT-PCR. Error bars indicate SEM (n = 6). (E) Gene cloning analysis reveals that the xBMP7 gene promoter contains two NF-κB binding sites (−144 and −1267 bp from ATG). (F) xBMP7 is a direct transcriptional target of NF-κB. xRelA mRNA (400 or 800 pg) was injected into four-cell–stage embryos. Injected embryos were harvested at stage 13. Two qRT-PCR experiments for xBMP7 are shown. (G) ChIP analysis of the xBMP7 gene promoter using stage 10.5 embryos injected with HA-xRelA mRNA at four-cell stage. Two pairs of PCR primers were designed to flank the two NF-κB consensus sites [primer 1 for the distal site (−1267 bp from ATG), primer 2 for the proximal site (−144 bp from ATG)]. The quantified amounts of the precipitates are also shown. The relative band intensity of the precipitates is normalized to that of genomic DNA input. Error bars indicate SEM (n = 5). *P < 0.05 by unpaired t test. (H) xSGK1 enhances the transcription from human BMP7 promoter. The reporter construct (hBMP7-Luc; 25 ng) was transfected with or without xSGK1 S425D (125, 312.5, or 625 ng) into HEK293 cells. After 24 hours, the cells were harvested, and the lysates were subjected to a luciferase assay (results of two experiments are shown). (I) Transcripts for xBMP7 and NF-κB signaling components are expressed in the endoderm of Xenopus embryos. Animal ectoderm and vegetal endoderm were dissected at stage 10.5 as shown. Total RNA isolated from the dissected tissues was subjected to RT-PCR analysis. The endoderm marker xSox17α and the mesoderm marker Xbra were also analyzed to ensure the accuracy of dissections. XeODC was used as an RNA loading control. (J) The expression of direct transcriptional targets of NF-κB is decreased in xSGK1-knockdown embryos. qRT-PCR analyses for xSlug, xSnail, and xTwist are shown. Error bars indicate SEM (n = 4). (K) PI3K, a potential activator of xSGK1, decreases the abundance of DISC components through xBMP7 during embryogenesis. Inhibition of PI3K results in gene expression changes similar to those observed in SGK1-knockdown embryos. Embryos at stage 9 were treated with the PI3K inhibitor LY294002 at 50 or 100 μM and harvested at stage 13. The mRNA expression was analyzed by qRT-PCR analyses. Error bars indicate SEM (n = 4). (L) xBMP7 rescues the increased expression of DISC components caused by the inhibition of PI3K activity. xBMP7 mRNA (400 pg) was injected into four-cell–stage embryos, and then injected embryos were treated with 50 μM LY294002 at stage 9 and harvested at stage 13. The mRNA expression was analyzed by qRT-PCR. Error bars indicate SEM (n = 4). (M) Exogenous expression of an active form of SGK1 partially rescues gene expression changes caused by the inhibition of PI3K activity. xSGK1 S425D mRNA (50 pg) was injected into two–cell stage embryos, and injected embryos were then treated with 50 μM LY294002 at stage 9 and harvested at stage 13. The mRNA expression was analyzed by qRT-PCR (the results of two experiments are shown).

Consistent with this hypothesis, our microarray data showed that a gene encoding the Xenopus ortholog of BMP7 (xBMP7) was one of the down-regulated genes in xSGK1-depleted embryos and we confirmed this decrease in xSGK1-knockdown embryos by qRT-PCR (Fig. 4B). By contrast, the expression of other BMP family genes was not markedly changed in xSGK1-knockdown embryos (BMP2, factor of 1.10 change; BMP4, factor of 0.84 change; relative to expression in control MO–injected embryos). Because mammalian BMP7 plays an important role in cell survival of podocytes (23) and prostate cancer cells (24), we considered the possibility that xBMP7 mediated the antiapoptotic action of xSGK1 by reducing the abundance of DISC components. Stage 13 embryos injected with xBMP7 mRNA had decreased abundance of transcripts for xFADD and Xcaspase-10 (Fig. 4C). Moreover, the increased abundance of transcripts for xFADD and Xcaspase-10 in embryos injected with xSGK1-MOs was suppressed by coexpression of xBMP7 (Fig. 4D), which suggests that xBMP7 functions downstream of xSGK1 to inhibit expression of genes encoding DISC components. In contrast, the increased abundance of the transcript for Xp53 in the xSGK1-knockdown embryos was not suppressed by xBMP7 (fig. S3B), suggesting that the mechanism underlying the increase in the expression of Xp53 is independent of xBMP7. These results suggest that xSGK1 reduces the abundance of DISC components to promote survival of neighboring or distant cells, at least in part, through increasing expression of the gene encoding xBMP7, which acts as a secreted signal.

xBMP7 is a transcriptional target of NF-κB

To identify a transcription factor responsible for xBMP7 expression, we isolated ~2.8 kbp of the 5′UTR of the xBMP7 gene by splinkerette PCR and scanned the region with the TRANSFAC and NSITE databases to identify putative cis-regulatory elements. Among the computationally predicted elements, we focused on nuclear factor κB (NF-κB) binding sites (Fig. 4E), because NF-κB is a survival-promoting transcription factor that has been implicated in antiapoptotic signaling mediated by mammalian SGK1 (16, 25). Furthermore, expression patterns of the genes encoding the NF-κB subunits overlap that of the gene encoding xBMP7 in early Xenopus embryos (2630). To investigate whether xBMP7 is a transcriptional target gene of NF-κB, we injected the mRNA encoding a Xenopus ortholog of the NF-κB subunit RelA (also known as p65; xRelA) into Xenopus embryos and found that the abundance of the mRNA for xBMP7 was increased by overexpression of xRelA (Fig. 4F).

To further validate xBMP7 as an NF-κB target gene, we performed chromatin immunoprecipitation (ChIP) assays. Four-cell–stage embryos were injected with hemagglutinin (HA)–tagged xRelA, and chromatin extracts were subjected to immunoprecipitation with either a control immunoglobulin G (IgG) or an antibody that recognizes HA. The results showed a specific enrichment of the proximal NF-κB binding site (located at −144 bp position from ATG) in the HA antibody precipitates, but not in the control IgG precipitates (Fig. 4G, right panels). In contrast, we detected little enrichment for the distal NF-κB binding site located at −1267 bp position (Fig. 4G, left panels). These data suggest that xRelA physically associates with the proximal NF-κB binding site in the xBMP7 promoter to enhance transcription in Xenopus embryos.

To confirm that SGK1 stimulated BMP7 expression, we performed luciferase assays with the reporter plasmid hBMP7-Luc, encoding luciferase under the control of the human BMP7 promoter, which has one NF-κB binding site (located at −2525 bp position from ATG). In normal human embryonic kidney (HEK) 293 cells, forced expression of an active form of SGK1 (xSGK1 S425D) stimulated the BMP7 promoter activity in a dose-dependent manner (Fig. 4H), which is consistent with BMP7 as a transcriptional target of the SGK1–NF-κB signaling pathway in human cells.

We also examined the spatial expression patterns of transcripts for several NF-κB signaling components: IκB (inhibitor of NF-κB) kinase α (IKKα), IκB kinase β (IKKβ), and the histone acetyltransferase p300, all of which have been described as potential SGK1 substrates mediating SGK1-induced NF-κB activation (16, 25). RT-PCR assays with microsectioning showed that transcripts for the Xenopus orthologs of IKKα, IKKβ, and p300 (xIKKα, xIKKβ, and xp300, respectively) were present in both animal and vegetal regions (Fig. 4I), where they could serve as direct substrates of xSGK1 in the activation of NF-κB in Xenopus embryos. We then examined the requirement of SGK1 in NF-κB–driven gene expression. The transcriptional factors Slug, Snail, and Twist are direct transcriptional targets of NF-κB in Xenopus embryos (27, 28). The mRNA abundance of these NF-κB targets was decreased in xSGK1-MO–injected embryos (Fig. 4J), indicating that NF-κB functions downstream of SGK1. Collectively, these results suggest that NF-κB may contribute to the induction of xBMP7 expression to mediate the antiapoptotic action of xSGK1.

PI3K down-regulates DISC components through xBMP7

Studies with cultured cells have suggested that SGK1 is a downstream effector of PI3K (11, 12). To determine whether PI3K acts upstream of xSGK1 in vivo, we examined the effect of the PI3K inhibitor LY294002 on the expression of the genes encoding xFADD and Xcaspase-10 and found that both of these were increased, whereas expression of the gene encoding xBMP7 was decreased (Fig. 4K). Moreover, the increase in the transcripts for xFADD and Xcaspase-10 in LY294002-treated embryos was suppressed by exogenous expression of xBMP7, as was the decrease in the abundance of the transcripts for xFADD and Xcaspase-10 in xSGK1-knockdown embryos (compare Fig. 4, D and L). Finally, we found that injection of mRNA for xSGK1 S425D partially rescued both the increase in the transcripts for xFADD and Xcaspase-10 and the decrease in the transcript for xBMP7 in LY294002-treated embryos (Fig. 4M). These results are consistent with the model that a PI3K-SGK1 cascade functions in Xenopus embryos to reduce the abundance of DISC components through production of xBMP7.

Discussion

This study shows that SGK1 is required for cell survival in the context of a whole organism. Our analyses show that an endoderm- or dorsal mesoderm–resident signaling molecule SGK1 protects ectodermal cells from apoptosis by stimulating the production of xBMP7, which inhibits the expression of genes encoding components of DISC. We provide evidence consistent with a model in which xSGK1 is downstream of PI3K and the xSGK1-induced stimulation of xBMP7 expression is mediated by NF-κB–dependent transcription. Because NF-κB signaling components IKKα, IKKβ, and p300 have been proposed as direct substrates of mammalian SGK1 (16, 25), and the spatial expression pattern of the gene encoding xSGK1 overlaps those of the genes encoding these NF-κB signaling components, as well as that encoding xBMP7 (Fig. 4I) (29, 30), we speculate that in the endoderm or dorsal mesoderm, PI3K-dependent activation of xSGK1 stimulates NF-κB signaling, leading to production of xBMP7. The secreted xBMP7 acts on neighboring or distant ectodermal cells to repress the expression of genes encoding DISC components in the ectoderm (Fig. 5). Thus, this proposed signaling axis is an example of an intercellular signaling cascade that ensures cell survival across germ layers during amphibian embryonic development.

Fig. 5

A proposed model for the survival signaling action of xSGK1 in Xenopus embryos. The net effect of xSGK1 signaling is inhibition of apoptosis due to repression of genes encoding the DISC components FADD and caspase-10.

Apoptosis signaling can be divided into two distinct pathways: the extrinsic pathway, which is mediated by death receptors and DISC components, and the intrinsic pathway, which is triggered by cytochrome c release from mitochondria (18, 31). These two pathways are not totally independent, because the intrinsic pathway acts downstream of the extrinsic pathway in some types of cells (32, 33). Although SGK1 is known to play a role in the extrinsic pathway in kidney cancer cells (34), we show the involvement of SGK1 in the extrinsic pathway in the context of development. We found that apoptosis induced by xSGK1-MOs was rescued by coinjection of hBcl-2 (an inhibitor of the intrinsic pathway), as well as that of dnxFADD (an inhibitor of the extrinsic pathway) (Figs. 2, F and G, and 3C). Loss of xSGK1 also increased the expression of the gene encoding Xp53 (an activator of the intrinsic pathway), presumably through a pathway independent of xBMP7 (Fig. 3A and fig. S3, A and B). Thus, xSGK1 may inhibit both apoptotic pathways through xBMP7-dependent and -independent mechanisms.

Studies with cultured mammalian cells showed that p53 induces an increase and a decrease in BMP7 mRNA abundance in a breast cancer cell line, MCF-7 (35), and in primary neuronal cultures (36), respectively. However, we found that in Xenopus embryos, SGK1 functions to decrease abundance of the transcript for Xp53 through a pathway independent of BMP7 (fig. S3B), and we found no p53-responsive element in the promoter region of xBMP7. Thus, the link between BMP7 and p53 may not be conserved across species.

SGK1-depleted mice do not show a clear phenotype that can be attributed to developmental defects (37). This apparent discrepancy between Xenopus and mouse could be explained by the presence of different SGK family members in mammals, which might be almost absent from Xenopus. We searched two databases [Gurdon Institute Xenopus laevis EST Database and National Center for Biotechnology Information (NCBI) database] for Xenopus expressed sequence tag (EST) sequences homologous to mammalian SGK family members (SGK1, SGK2, and SGK3). We identified 155 Xenopus ESTs corresponding to SGK1, whereas there were no Xenopus ESTs corresponding to SGK2 and only three ESTs corresponding to SGK3. This indicates that SGK1 is dominantly expressed among SGK family members in Xenopus. In mammals, SGK2 and SGK3 are likely to function redundantly with SGK1 during embryogenesis. Thus, SGK1-depleted mice would not show a severe developmental defect.

The initiating signal that triggers the embryonic PI3K-SGK1 survival pathway remains to be identified. At least three independent pathways activate PI3K and all start with ligand binding to receptor tyrosine kinases (38). We anticipate that one or more growth factors localized in the endoderm and mesoderm may be the initiating signal.

Materials and Methods

Molecular cloning and plasmid construction

Isoforms of Xenopus SGK1 (xSGK1a, accession number BC073077; xSGK1b, accession number BC074305) were already deposited in GenBank. Therefore, we designed primers on the basis of the sequences. We performed PCR with complementary DNAs (cDNAs) derived from embryos at stage 11 for xSGK1a and stage 22 for xSGK1b. The amplified entire coding sequences were cloned into the expression vector. A dominant-negative form of xFADD (dnxFADD) was constructed as described (20).

Embryonic manipulation, whole-mount in situ hybridization, and RT-PCR

In vitro fertilization, microinjection, whole-mount in situ hybridization, and RT-PCR were performed as described (39). The amounts of injected mRNAs or antisense MOs are described in the text and figure legends. LY294002 (Cell Signaling Technology) was used at 50 or 100 μM. The digoxigenin-labeled probe for whole-mount in situ hybridization was synthesized from cDNA corresponding to the coding region of xSGK1a. For qRT-PCR analysis, we used 7300 Real Time PCR System (Applied Biosystems) with SYBR Green PCR Master Mix (Qiagen). EF1α was used as a loading control, and all values were normalized to the EF1α level. The sequences of primer pairs used in RT-PCR are listed in table S1.

Cell culture, transfection, and luciferase assay

HEK293 cells were cultured on collagen-coated plates in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal calf serum. The isolated human BMP7 promoter was ligated into pGL3 luciferase reporter vector (hBMP7-Luc) (Promega). Cells were transfected with the hBMP7-Luc reporter plasmids along with phRL-TK (Promega) and the test plasmid encoding xSGK1 S425D using LipofectAMINE 2000 transfection reagent (Invitrogen). Luciferase activity in cell lysates was measured with the Dual-Luciferase Reporter Assay System (Promega) in GloMax-Multi+ Detection System (Promega). We normalized the relative reporter activity to Renilla luciferase activity expressed by phRL-TK.

TUNEL assay

TUNEL staining was performed as described (40). Embryos were stained with BM purple (Roche).

Immunoblotting

Embryos were lysed in a buffer consisting of 20 mM Hepes (pH 7.2), 0.25 M sucrose, 0.1 M NaCl, 2.5 mM MgCl2, 10 mM NaF, 10 mM EGTA, 10 mM β-glycerophosphate, 1 mM vanadate, 1 mM phenylmethylsulfonyl fluoride, 0.5% aprotinin, and 1 mM dithiothreitol and then centrifuged. The supernatant was used for immunoblotting with antibody against Myc (A-14; Santa Cruz Biotechnology) or antibody against HA (Y-11; Santa Cruz Biotechnology).

Morpholino oligonucleotides

Antisense MOs were obtained from Gene Tools Inc. The MO sequences were as follows: xSGK1-MO1, 5′-GCCGTTTCTGTCTTTACAGTCATGG-3′; xSGK1-MO2, 5′-AGCCGTCTCTGTCTTTACAGTCATG-3′; a standard control MO (control MO), 5′-CCTCTTACCTCAGTTACAATTTATA-3′. Sequences complementary to the predicted start codon are underlined. Both xSGK1-MO1 and xSGK1-MO2 were simultaneously injected in all knockdown experiments.

Microarray analysis

For microarray analysis, we performed two independent experiments. For each microarray experiment, we radially injected control MO (80 ng) or xSGK1-MOs (40 ng each for MO1 and MO2) into four-cell embryos, cultured the embryos until stage 12, extracted total RNAs with Trizol (Invitrogen), and purified them on RNeasy columns (Qiagen). Synthesis of cDNA, in vitro transcription, and biotin labeling of complementary RNA and hybridization to the X. laevis genome array (Affymetrix) were performed according to the Affymetrix protocol (Two-Cycle Target Labeling Assays). Hybridized arrays were scanned with an Affymetrix GeneChip Scanner. Scanned chip images were analyzed with GeneChip Operating Software v.1.4 (GCOS).

Microarray data analysis

Images from scanned chips were processed with the default settings of GCOS and individually scaled to an average target signal of 500. To identify the genes listed in Fig. 3A, we first searched genes that were annotated as “apoptosis” in NetAffx (http://www.affymetrix.com/analysis/index.affx), and then we isolated proapoptotic genes from the annotated genes with reference to the database List of Apoptosis Regulators (http://www-personal.umich.edu/%7Eino/List/AList.html).

Isolation and characterization of Xenopus FADD, caspase-10, and BMP7 promoters

Genomic DNA of X. laevis was isolated from tadpoles at stage 45 with DNeasy Blood and Tissue Kit (Qiagen) according to the manufacturer’s protocol. The 5′UTRs of the xBMP7, xFADD, and Xcaspase-10 genes were isolated by splinkerette PCR walking according to the original paper (22) and the protocol on the Cnidaria Home Page (http://www.biochem.uci.edu/Steele/Splinkerette_page.html). Cis-regulatory elements in the promoter regions were searched with TRANSFAC (41) and NSITE (http://linux1.softberry.com/berry.phtml).

ChIP

ChIP was performed as described (42). Fifty embryos were injected with HA-xRelA mRNA at four-cell stage. At stage 10.5, injected embryos were fixed and cross-linked with formaldehyde for 60 min. The cross-linked samples were sonicated in 650 μl of RIPA (radioimmunoprecipitation assay) buffer [50 mM tris-HCl (pH 7.4), 1% NP-40, 0.25% sodium deoxycholate, 150 mM NaCl, 1 mM EDTA, 0.1% SDS, 0.5 mM dithiothreitol (DTT), 5 mM sodium butyrate, and Protease Inhibitor Cocktail (P8340; Sigma)]. Fragmented chromatin was immunoprecipitated with 1 μg of rabbit IgG (Invitrogen) or anti-HA antibody (Y-11; Santa Cruz Biotechnology). Precipitated DNA was washed, purified, and used for PCR. Primer sequences are available in table S1.

Supplementary Materials

www.sciencesignaling.org/cgi/content/full/4/156/ra2/DC1

Fig. S1. Characterization of xSGK1.

Fig. S2. Overexpression of xFADD or Xcaspase-10 causes excessive apoptosis.

Fig. S3. Loss of xSGK1 increases the expression of Xp53 presumably through a pathway independent of xBMP7.

Table S1. Primer sequences.

References and Notes

  1. Acknowledgments: We thank T. Araki, T. Suzuki, and K. Miyatake for technical advice and helpful discussion. Author contributions: T.E. and M.K. conceived and designed the study. T.E. performed the experiments. T.E., K.S., and T.Y. analyzed the microarray data. T.E., M.K., and E.N. wrote the manuscript. E.N. supervised the project. Funding: This work was supported by grants from the Ministry of Education, Culture, Sports, Science, and Technology of Japan (to E.N. and M.K.) and Core Research for Evolutional Science and Technology, Japan Science and Technology (to E.N.). T.E. is a research fellow of the Japan Society for the Promotion of Science. Competing interests: The authors declare that they have no competing interests. Accession number: The microarray data have been deposited in the Gene Expression Omnibus with accession code GSE26381.
View Abstract

Navigate This Article