Serum and glucocorticoid regulated kinase 1 (SGK1) has been identified as a key regulatory protein that controls a diverse set of cellular processes including sodium (Na+) homeostasis, osmoregulation, cell survival, and cell proliferation. Two other SGK isoforms, SGK2 and SGK3, have been identified, which differ most markedly from SGK1 in their NH2-terminal domains. We found that SGK1 and SGK3 are potent stimulators of epithelial Na+ channel (ENaC)-dependent Na+ transport, while SGK2, which has a short NH2 terminus, is a weak stimulator of ENaC. Further characterization of the role of the SGK1 NH2 terminus revealed that its deletion does not affect in vitro kinase activity but profoundly limits the ability of SGK1 either to stimulate ENaC-dependent Na+ transport or inhibit Forkhead-dependent gene transcription. The NH2 terminus of SGK1, which shares sequence homology with the phosphoinositide 3-phosphate [PI(3)P] binding domain of SGK3, binds phosphoinositides in protein lipid overlay assays, interacting specifically with PI(3)P, PI(4)P, and PI(5)P, but not with PI(3,4,5)P3. Moreover, a point mutation that reduces phosphoinositide binding to the NH2 terminus also reduces SGK1 effects on Na+ transport and Forkhead activity. These data suggest that the NH2 terminus, although not required for PI 3-kinase-dependent modulation of SGK1 catalytic activity, is required for multiple SGK1 functions, including stimulation of ENaC and inhibition of the proapoptotic Forkhead transcription factor. Together, these observations support the idea that the NH2-terminal domain acts downstream of PI 3-kinase-dependent activation to target the kinase to specific cellular compartments and/or substrates, possibly through its interactions with a subset of phosphoinositides.
- epithelial sodium channel
- kinase activity
renal tubular cells must integrate a variety of hormonal and nonhormonal signals to achieve coordinated responses to changes in the environment that are appropriate within a given physiological context. One key regulatory protein that integrates multiple environmental cues in mammalian cells is the serine/threonine kinase, serum and glucocorticoid-regulated kinase 1 (SGK1). Several hormonal, growth factor, and cell stress pathways converge on SGK1 to control diverse cellular processes including sodium (Na+) homeostasis (14, 40), osmoregulation (45, 57), cell survival (35), and cell proliferation (13). In particular, the importance of SGK1 in mediating hormonal and osmotic effects on Na+ transport in renal collecting duct cells is well established (14, 40, 45, 58). SGK1 phosphorylates Nedd4-2 (neural precursor cell-expressed, developmentally downregulated protein 4-2) at three RXRXX(S/T) phosphorylation motifs (18, 46) and promotes an inhibitory interaction of this ubiquitin ligase with 14-3-3 proteins (6, 28, 36). Nedd4-2-mediated ubiquitination enhances removal from the plasma membrane and degradation of the epithelial Na+ channel (ENaC) (1, 29), while SGK1 diminishes removal, at least in part, by inhibiting Nedd4-2 (18, 46).
SGK1 belongs to the AGC family of protein kinases, which, in addition to the SGK kinases, includes the PKB/Akt kinases, p70 ribosomal S6 kinase (S6K), and atypical forms of protein kinase C (PKC). In particular, the SGK and Akt family of kinases share considerable sequence homology and a number of functional similarities, including growth factor-mediated activation via the phosphatidylinositol 3-kinase (PI3-kinase) signaling pathway and common peptide substrates (33, 42). At the primary sequence level, their similarities are most notable in their catalytic domains and in two regulatory sequences surrounding Thr-256 and Ser-422, which lie in the activation loop and hydrophobic motif of SGK1, respectively, and are phosphorylation targets of PI3-kinase-dependent kinases. Phosphorylation of both regulatory sites is required for activation of catalytic function of these kinases and appears to be functionally indistinguishable among these kinase family members (33, 42).
The NH2-terminal domains of these kinases, on the other hand, diverge markedly: PKB/Akt kinases contain an NH2-terminal pleckstrin homology (PH) domain, which selectively binds PI(3,4,5)P3 and PI(3,4)P2 and directs PKB/Akt to the plasma membrane (2–4, 48, 49). SGK1's closest relative, SGK3, bears an NH2-terminal Phox homology (PX) domain that targets it to endosomal membranes primarily via interaction with PI(3)P (55, 60), while the NH2-terminal region of SGK2 is short and lacks homology to any known interaction motif. SGK1 has an NH2-terminal region, which is intermediate in length between SGK2 and SGK3, and contains a short sequence that bears significant homology to the SGK3 PX domain (41). Of the several subcellular compartments in which SGK1 has been detected, a role for the NH2-terminal region in mediating its localization has been demonstrated for plasma membrane (11), mitochondria (22), and endoplasmic reticulum (5, 9); however, whether phosphoinositides are implicated in its localization is not known. Indeed, although this region has been suggested to represent a phosphoinositide interaction motif (41), this has not been demonstrated.
In addition to its role in directing SGK1 to various cellular compartments, the NH2-terminal domain has been shown to play an important role in SGK1 protein stability by triggering its ubiquitination and proteasome-mediated degradation (5, 9, 11, 46). Consequently, deletion of the NH2-terminal domain greatly enhances SGK1 expression (8, 11, 33, 39) and, unless the NH2-terminal region is implicated in another aspect of SGK1 function, would presumably increase SGK1 kinase activity and physiological functions. In the course of experiments directed at examining SGK isoform-specific functions, we compared the functional activities of SGK1, 2, and 3 in stimulating ENaC. These experiments demonstrated that SGK2 has substantially lower ENaC stimulatory activity than SGK1 or SGK3. The near lack of an NH2-terminal domain in SGK2, coupled with the similar NH2-terminal domains between SGK1 and SGK3, highlighted the importance of this region in kinase function. We therefore went on to characterize the role of this domain in SGK1 function. We found that the NH2-terminus is critical for the ability of SGK1 to stimulate ENaC and inhibit the proapoptotic transcription factor Forkhead, another physiological target of SGK1, but is not required for catalytic activity. Indeed, a point mutation that mimics PI 3-kinase-dependent phosphorylation confers constitutive catalytic activity to both full-length and NH2 terminally deleted SGK1; however, such a mutant is able to stimulate ENaC or inhibit Forkhead only if the NH2-terminal domain is intact. We further demonstrate that the NH2 terminus contains a phosphoinositide-binding domain that can bind to a subset of phosphoinositides that is distinct from the binding preferences of either the PH domain of PKB/Akt or the PX domain of SGK3. Taken together, these data suggest that the NH2-terminal domain is important for post PI 3-kinase-dependent activation events in SGK1 function, in part by interacting with a subset of phosphoinositides that are found within specific subcellular compartments.
MATERIALS AND METHODS
Xenopus laevis oocyte coexpression assay.
X. laevis α-, β-, γ-FLAG-ENaC psDEasy expression vectors were generously provided by Drs. B. Rossier and D. Firsov; mouse SGK1, SGK2, and SGK3 were subcloned into pMO expression vectors (gift from Dr. L. Jan). Capped RNAs (cRNAs) were synthesized as described previously (14). Stage V-VI oocytes were injected with cRNAs containing X. laevis α-, β-, γ-FLAG-ENaC (3.0 ng each), X. laevis Nedd4-2 (1.5–3.0 ng), and wild-type mouse SGK1, SGK2, or SGK3 (5.0 ng).
To ensure uniform expression of enzymatically active SGK1 in oocytes, subsequent experiments involving mouse SGK1 were performed with a constitutively active form of SGK1 (CA-SGK1), which has an S422D mutation and does not require PI 3-kinase for activation (33, 42). Depending on the experimental condition, oocytes were injected with cRNAs (0.025–5.0 ng) containing either full-length mouse SGK1/S422D-FLAG (CA-SGK1), ΔN-[deletion of amino acids (aa) 1-85]-SGK1/S422D-FLAG (CA-SGK1/ΔN), or SGK1/R31A/S422D-FLAG (CA-SGK1/R31A) in conjunction with X. laevis α-, β-, γ-FLAG-ENaC and X. laevis Nedd4-2. Based on the known differences in protein stability between wild-type SGK1 and mutants that do not associate with membrane fractions, including ΔN-SGK1 (11, 34, 61), titrated amounts of the different CA-SGK1 constructs were injected to yield equivalent levels of expression as assessed by Western blotting. After injection, the oocytes were incubated in a low Na+ Barth's solution and current measurements were performed as described previously (14). The specific ENaC-mediated current (Iphe) was defined as the difference between currents obtained in the presence or absence of 2 μM phenamil (Sigma, St. Louis, MO). Results were repeated with at least three independent experiments.
Expression of SGK in X. laevis oocytes.
Groups of 10–15 oocytes that underwent current measurements were transferred to Eppendorf tubes. Twenty microliters per oocyte of homogenization buffer (20 mM Tris·HCl, pH 7.6, 100 mM NaCl, 2% NP-40 with protease inhibitors) were added to the oocytes, and the oocytes were lysed by repeated vortexing and pipeting. Cellular debris was pelleted at 3,600 g for 10 min, and the supernatant was centrifuged at 3,600 g twice to remove additional debris. Floating yolk was removed with a cotton-tipped swab. All centrifugation steps were performed at 4°C.
One hundred micrograms of protein from oocyte lysates were resolved by 10% SDS-PAGE and transferred to nitrocellulose membranes (Hybond-C Extra, Amersham Biosciences, Piscataway, NJ). The membranes were immunostained with polyclonal anti-SGK rabbit antiserum (raised against an SGK1 COOH-terminal epitope, Cell Signaling, Danvers, MA) diluted 1:100 for detection of SGK-immunoreactive protein. Membranes were incubated with secondary anti-rabbit IgG horseradish peroxidase conjugate (Amersham Biosciences) diluted 1:5,000 and processed as described (58).
Expression of SGK in HEK 293 cells for in vitro kinase assay.
HEK 293 cells were maintained at 37°C in culture medium containing DMEM, 10% fetal bovine serum, and 1% penicillin/streptomycin. Cells were plated on 6- cm dishes and were transfected with CA-SGK1, CA-SGK1/ΔN, or CA-SGK1/R31A on the following day. As in oocytes, amounts of transfected SGK1 for the different constructs were titrated to yield equivalent levels of expression as assessed by Western blotting.
Protein lysates of HEK 293 cells expressing each of the different CA-SGK1 constructs were immunoprecipitated with EZview red anti-FLAG M2 affinity gel beads (Sigma). The different CA-SGK1 constructs were subsequently eluted from anti-FLAG gel beads with 3X-FLAG peptide (Sigma) at 300 μg/ml to yield free kinase. 3X-FLAG peptide was removed, and the different CA-SGK1 constructs were concentrated through the use of size exclusion spin columns (Microcon YM-30, Millipore, Billerica, MA). Equal amounts of eluted CA-SGK1 constructs were verified by Western blotting before evaluation of kinase activity. Eluted CA-SGK1 constructs were stored in aliquots at −80°C.
In vitro kinase assays were performed using SGKtide (Upstate Biotechnology, Lake Placid, NY) as an SGK1-specific phosphorylation substrate in the presence of [γ-32P] ATP, according to the manufacturer's protocol and as described (42). The reaction product was blotted on P81 phosphocellulose squares (Whatman, Clifton, NJ), washed, and quantified by scintillation counting.
Protein lipid overlay assay.
Mouse SGK3 [1-149 aa], SGK1 [1-85 aa], and SGK1/R31A [1-85 aa] NH2-terminal regions upstream of the respective catalytic domains were expressed in Escherichia coli BL21 as 6X-His fusion proteins using the pET 15b expression vector (Novagen, Madison, WI) and purified on nickel-charged resin (Qiagen, Valencia, CA). Protein lipid overlay assays were performed using “PIP strips” or “PIP arrays” (Echelon, Salt Lake City, UT), which are nitrocellulose membranes spotted with a diverse array of phospholipids designed to detect specific protein-phospholipid interactions. Membranes were incubated with His-tagged protein and retained protein was detected with anti-His Ab (Novagen), as described by the manufacturer's protocol and previously (55). Similar results were obtained in three independent experiments.
Transient transfection luciferase assay.
The Forkhead-responsive luciferase reporter plasmid (FHRE-Luc) was kindly provided by Dr. M. Greenberg (Harvard Medical School, Boston, MA). For transient transfections, low-passage HEK293 cells were seeded 24 h before transfection at a density of 2 × 105 cells/dish in 6-cm dishes. Transient transfections were performed using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. Cells were transfected with 50 ng of FHRE-Luc and depending on the experimental condition, 1,000 ng of CA-SGK1, 10–20 ng of CA-SGK1/ΔN, or 400–600 ng of CA-SGK1/R31A were titrated to give equivalent expression of the different mutants by immunoblot. Transfection for each experimental condition was performed in triplicate. The day after transfection, culture medium was changed and, 10 h later, was replaced with serum-free medium. Cells were harvested the following day, and protein lysates were assessed for luciferase activity using Promega Luciferin Reagent (Promega, Madison, WI) and normalized to total protein levels, which were measured by adding 100 μl of Bio-Rad protein assay dye (Bio-Rad, Hercules, CA) to 1 μl of lysate followed by measuring in a microplate reader. Similar results were obtained in three independent experiments.
Statistical analyses for comparisons of Iphe in X. laevis oocytes and luciferase activity (normalized to total protein levels) in HEK 293 cells were performed using one-way ANOVA with Bonferroni's adjustment. Differences were considered to be significant at P values <0.05.
SGK1 and SGK3, but not SGK2, stimulate ENaC in X. laevis oocytes.
Previous work established that SGK1 increases ENaC-mediated Na+ current by inhibiting the ubiquitin ligase Nedd4-2 (18, 46). Little is known about SGK isoform specificity, and our initial goal in the present study was to compare SGK isoform activities in the context of ENaC activation. Toward this end, we expressed ENaC and Nedd4-2, in combination with SGK1, 2, or 3, in X. laevis oocytes and measured ENaC-mediated Na+ current (Iphe) by two-electrode voltage clamp (Fig. 1). The amount of cRNA injected for each of the SGK isoforms was titrated in a series of pilot experiments to give comparable levels of protein expression (not shown). As shown previously (6), Nedd4-2 markedly inhibited ENaC current, an effect that was largely reversed by SGK1 (Fig. 1, compare second and third bars). SGK3 also markedly increased Na+ current, although not to the same extent as SGK1. In contrast, SGK2 had a negligible effect on Iphe when expressed at equivalent levels. At high levels of expression, SGK2 did significantly stimulate ENaC, but at all levels of expression, this effect remained substantially weaker than that of SGK1 or SGK3 (not shown). One earlier report found that SGK1, 2, and 3 all stimulate ENaC in X. laevis oocytes when equal amounts of cRNA are injected (23); however, high concentrations of cRNA were injected in this study, protein expression levels were not determined, and it seems likely that SGK2, in particular, was expressed at substantially higher levels than SGK1.
The SGK isoforms share ∼80% homology in their kinase domains, 50% homology in their COOH-terminal domains, but <10% homology in their NH2-terminal domains (Fig. 2A), which in SGK2 is quite short and bears no resemblance to any known modular domain. The NH2 terminus of SGK1 contains 58 amino acids (red box, Fig. 2B) that share 36% homology with the NH2 terminus of SGK3, which contains the PX domain. These data, in conjunction with an earlier report showing that SGK1 deleted of its NH2-terminal domain had reduced ability to stimulate ENaC current (27), suggested that the NH2-terminus is a critical determinant of SGK isoform-specific stimulation of ENaC. We were therefore interested in further examining the role of the NH2 terminus in SGK1 activation of ENaC.
SGK1 effects on specific physiological endpoints require that it first be activated and then be targeted to effector substrates in appropriate subcellular compartments. SGK1 catalytic activity requires PI 3-kinase-dependent phosphorylation at Ser-422 followed by non-PI 3-kinase-dependent phosphorylation at Thr-256 (8, 33, 42). Since previous work suggested that the NH2-terminal domain is not required for PI 3-kinase-dependent activation of SGK1 (8), we were interested in examining the role of the NH2 terminus in directing postactivation events in SGK stimulation of ENaC. Substitution of aspartate for serine at residue 422 (SGK1/S422D) effectively mimics PI 3-kinase-dependent phosphorylation and confers constitutive catalytic activity (8, 33, 42). To bypass this activation step and obtain uniform SGK1 activity independent of PI 3-kinase, we used the S422D mutant (CA-SGK1) for our subsequent experiments.
We first verified that deletion of the NH2-terminal domain did not affect the catalytic activity of CA-SGK1. FLAG epitope-tagged full-length CA-SGK1 and NH2-terminal deletion derivative (CA-SGK1/ΔN) were expressed in HEK-293 cells and purified by immunoprecipitation from whole cell lysates. In vitro kinase assays were performed with equivalent concentrations of the two kinases using the peptide substrate SGKtide (42). As shown in Fig. 3A, comparable amounts of labeled SGKtide were produced at all concentrations tested. We next compared the effect of CA-SGK1 and CA-SGK1/ΔN on ENaC activity in a X. laevis oocyte coexpression assay. Consistent with our hypothesis, CA-SGK1/ΔN had substantially decreased ability to stimulate ENaC current at all levels of expression (Fig. 3B), in marked contrast to its preserved kinase activity. It should be noted that the lowest amount of injected cRNA for CA-SGK1/ΔN gave equivalent protein expression to the highest amount of injected cRNA for CA-SGK1 within oocytes, consistent with previous reports that deletion of the NH2 terminus of SGK1 results in greater protein stability (11, 34, 61). Thus the NH2 terminus of SGK1 harbors a domain that is required for ENaC stimulation but is not required for catalytic activity. Furthermore, this domain appears to be required for post PI 3-kinase-dependent activation events in SGK1 function.
NH2 terminus of SGK1 contains a phosphoinositide interaction domain.
Closer examination of the SGK1 NH2 terminus showed that there is a stretch of 18 amino acids within this region (residue 27–44) that shares 61% homology with the corresponding region of the SGK3 PX domain (residue 86–103), as noted previously (41). PX domains were originally identified in the P40phox and P47phox subunits of NADPH oxidase (43) and are found in numerous proteins involved in membrane traffic, including sorting nexins (SNX1–4, 6, 9, 15, 17) and yeast Vps proteins (Vam7p, Mvp1p) (44). The PX domain of SGK3 binds to PI(3)P, which localizes it to early endosomes (55). These observations suggested the possibility that the NH2-terminal region of SGK1, like that of SGK3, harbors a phosphoinositide binding domain.
We therefore examined directly the phosphoinositide interactions of the SGK1 NH2 terminus and compared it with those of the SGK3 PX domain using a protein lipid overlay assay (Fig. 4) (19–21). As shown in Fig. 4A, the SGK3 PX domain bound specifically only to PI(3)P, as described previously (55). Under the same conditions, the SGK1 NH2 terminus interacted specifically with the monophosphorylated phosphoinositide species, PI(3)P, PI(4)P, PI(5)P, as well as PS (phosphatidylserine). Substantially weaker signal was observed with PI(3,5)P2 and PI(4,5)P2 (Fig. 4B, left). Notably, binding to PI(3,4,5)P3, which preferentially binds to the PH domain of PKB/Akt and is thought to be the principal phosphoinositide involved in class I PI 3-kinase-dependent signaling, was undetectable. Overlay assays using membranes with a range of phosphoinositide concentrations further confirmed that the SGK1 NH2 terminus interacted preferentially with monophosphorylated phosphoinositide species and PS but not with PI(3,4,5)P3 (Fig. 4B, right). We conclude that the NH2-terminal 85 residues of SGK1 contain a phosphoinositide-binding domain, which has sequence similarity and partially overlapping phosphoinositide binding specificity with the SGK3 PX domain.
Characterization of an NH2-terminal domain mutant with reduced binding to phosphoinositides.
Significant sequence homology between the NH2 terminus of SGK1 and the PX domains of SGK3 and P40phox, whose crystal structures have been solved (10, 59), starts at residue 27 of SGK1 (residue 86 of SGK3, residue 101 of P40phox) (see within red box, Fig. 2B). In the known PX domain structures, this region (corresponding to aa 27–44 of SGK1) forms an α-helix that provides the floor of the phosphoinositide-binding pocket and possesses an arginine (R90 of SGK3, equivalent to R31 of SGK1) involved in the coordination of bound ligand (10). In the case of P40phox, the analogous residue (R105) interacts with the 4′-OH group of the inositol ring of the phosphoinositide head group of PI(3)P (10). Mutations of this residue [e.g., R105A in P40phox (10), R90L in P47phox (30), or R90A in SGK3 (60)] markedly reduce phosphoinositide binding.
Modeling of the SGK1 structure surrounding R31 based on the solved structures of the P40phox and SGK3 PX domains suggested that R31 can also potentially form a H-bond with the 4′-OH group of the inositol ring of PI(3)P (Fig. 2C). We therefore tested the ability of the SGK1 NH2 terminus, mutated at this position (SGK1 [1-85 aa]/R31A), to bind to phosphoinositides (Fig. 4C). Consistent with its predicted role in phosphoinositide binding, SGK1 [1-85 aa]/R31A showed decreased binding to all phosphoinositide species. Notably, however, significant residual binding was detected, particularly to PI(3)P (Fig. 4C). This was more apparent with longer film exposure times (not shown).
SGK1/R31A has reduced ENaC-stimulatory activity.
We next examined the functional consequences of the R31A mutation on SGK1 regulation of ENaC. First, we assessed whether CA-SGK1 and CA-SGK1/R31A have similar kinase activities. Indeed, CA-SGK1 and CA-SGK1/R31A showed equivalent kinase activities at equivalent molar input levels, indicating that, as with deletion of the NH2-terminal domain, the R31A mutation does not affect SGK1 in vitro kinase activity (Fig. 5A). Next, we examined the ability of SGK1/R31A to stimulate ENaC. X. laevis oocytes were coinjected with ENaC and Nedd4-2 cRNA, in combination with CA-SGK1 or CA-SGK1/R31A, and ENaC current was determined (Fig. 5B). cRNA for CA-SGK1 and CA-SGK1/R31A was titrated to give equivalent levels of expression. Under these conditions, CA-SGK1/R31A had decreased ability to stimulate ENaC current, which was particularly apparent at low levels of expression (Fig. 5B); at high levels of expression, the ENaC-stimulatory activity of CA-SGK1/R31A approached that of CA-SGK1. Thus this mutation significantly impaired the ENaC-stimulatory activity of CA-SGK1, although not as profoundly as did complete deletion of the NH2 terminus.
An intact NH2-terminal domain within SGK1 is required for full inhibition of Forkhead-dependent transcription.
SGK1 is a multifunctional kinase; in addition to its well-characterized effects on Na+ transport, it is also known to have effects on cell survival (35), apoptosis (39), and cell cycle progression (13, 26), in part by phosphorylating and inactivating the proapoptotic Forkhead box transcription factor FKHRL1 (12). To evaluate the potential role of the NH2 terminus in mediating SGK1 regulation of Forkhead-dependent gene transcription, we cotransfected HEK-293 cells with a Forkhead response element luciferase reporter gene [FHRE-Luc (12)] and either CA-SGK1 or one of the phosphoinositide binding-deficient CA-SGK1 mutants. Similar to our X. laevis oocyte coexpression experiments, amounts of transfected DNA were titrated to give equivalent levels of protein expression. As shown in Fig. 6A, CA-SGK1 resulted in ∼70% inhibition of Forkhead reporter activity. In contrast, the CA-SGK1/ΔN mutant failed to inhibit reporter activity (Fig. 6A), whereas the CA-SGK1/R31A mutant had intermediate activity (Fig. 6B), similar to what was observed for ENaC regulation. Taken together, these results strongly suggest that the NH2 terminus is an important determinant of SGK1 action in Forkhead inhibition as well as in ENaC stimulation, potentially by serving as a site for phosphoinositide binding.
NH2 terminus as a determinant of SGK1 activities and of isoform specificity.
The three SGK isoforms have substantial sequence homology in their catalytic and COOH-terminal domains, common substrate preferences in vitro, and similar PI 3-kinase dependence for catalytic activity. SGK1 and SGK3 also share significant homology in a short stretch of amino acids (residue 27–44 for SGK1, residue 86–103 for SGK3) within their NH2 termini, which in SGK3 comprises part of the PI(3)P binding PX domain (55). The SGK2 NH2 terminus, in contrast, is short and bears no resemblance to any known interaction module. Initial experiments showing that SGK1 and SGK3 can stimulate ENaC when expressed at comparable levels in the presence of Nedd4-2 (Fig. 1) suggested the possibility that the NH2-terminal regions of these two kinases contain functionally important selectivity determinants that are not shared by SGK2. It is notable that in a prior report, all three SGK isoforms stimulated ENaC when expressed at high levels in X. laevis oocytes in the absence of Nedd4-2 (23). Our present data show that under conditions of lower levels of kinase expression in the presence of Nedd4-2, SGK2 is substantially less efficacious than SGK1 or SGK3 at stimulating ENaC current (Fig. 1). At higher expression levels, SGK2 did significantly stimulate ENaC, likely because normal determinants of isoform specificity were overridden.
The foregoing data pointed to the NH2 terminus as an important determinant of SGK function. By using a constitutively active form of SGK1, we examined the role of the NH2 terminus in influencing events post-PI 3-kinase activation, an approach used previously for assessment of phosphorylation of physiological SGK1 substrates, as well as for substrates of other SGK and PKB/Akt kinases (6, 12, 38). Under these conditions, the in vitro kinase activities of CA-SGK1 and CA-SGK1/ΔN were indistinguishable (Figs. 3A and 5A), and yet the kinases differed markedly in their abilities either to stimulate ENaC or inhibit FKHRL1 (Figs. 3B, 5B, and 6). While two previous studies examined the effect of NH2-terminal deletion on SGK1-mediated ENaC activity (27) and anti-apoptosis (39), this is the first study to separate the effects of NH2-terminal deletion on SGK1 kinase activity from effects on its function. Our data strongly support the conclusion that the NH2 terminus is a functionally important domain, which acts downstream of PI 3-kinase-dependent activation to target the activated kinase to a distinct set of protein substrates involved in the regulation of ENaC, FKHRL1, and possibly other cellular processes.
Role of phosphoinositide binding to the NH2 terminus in determining SGK1 action.
One potential mechanism for how the NH2 terminus might control targeting is through its ability to bind to phosphoinositides. Phosphoinositides serve as second messengers and integrators of signal transduction by helping to ensure that a specific action (such as phosphorylation of a particular substrate by a protein kinase) occurs in the appropriate cellular compartment and under the appropriate cellular conditions (44). The NH2-terminal 85 amino acids of SGK1, which contain a region with high homology to the PX domain, bound specifically to PI(3)P, PI(4)P, PI(5)P, and PS with approximately equal affinity and to PI(3,5)P2 and PI(4,5)P2 with lower affinity. It notably does not bind to PI(3,4,5)P3, the principal phosphoinositide implicated in PI 3-kinase-dependent activation of Akt. The limited sequence similarity to PX domains and shared binding to PI(3)P (Fig. 4) suggested the SGK1 NH2 terminus is “Phox-like” with some significant differences. Most notably, the far NH2 terminus of SGK1 diverges markedly from the PX domain consensus, lacking residues known to comprise β-pleated sheets and an α-helix implicated in PX domain-phosphoinositide binding (10, 31, 59). Instead, the SGK1 NH2 terminus has a distinct 27 amino acid stretch, which is relatively rich in basic amino acids and could in principle form an amphipathic α-helix that contributes to SGK1's distinct pattern of phosphoinositide binding (orange box, Fig. 2B).
The marked defect in both ENaC stimulation and Forkhead inhibition by NH2 terminally deleted SGK1 strongly supports the importance of this domain in SGK1 function. The possibility that the phosphoinositide-binding capacity of this domain is functionally important is supported by the observation that a specific point mutation R31A, which disrupts lipid binding to the NH2 terminus (Fig. 4), significantly decreased modulation of both ENaC-mediated Na+ transport and Forkhead-driven gene transcription by a constitutively active form of SGK1 (Figs. 5 and 6). These results support the interesting idea that phosphoinositides influence processes downstream of PI 3-kinase-dependent activation, perhaps by controlling SGK1 targeting to subcellular compartments that contain SGK1 substrates such as Nedd4-2 or FKHRL1. It is notable that the R31A mutation did not appear to affect ENaC-stimulatory activity as profoundly as it did the ability of SGK1 to bind phosphoinositides (Figs. 4C and 5B), nor did it have as profound an effect on ENaC-stimulatory activity as deletion of the NH2 terminus. Although this may have been due to residual phosphoinositide binding by the R31A mutant, it is possible that the NH2 terminus of SGK1 contains additional motifs, which are important for both ENaC stimulation and FKHRL1 inhibition. More work examining the function of refined deletion mutants and additional point mutants will be required to fully address this question.
Although our present data do not address the role of the SGK1 NH2 terminus in steps before PI 3-kinase activation, previous evidence supports the idea that, in contrast to PKB/Akt, SGK1 binds directly to the “PIF” binding pocket of PDK1 and is subsequently activated without itself binding to phosphoinositides (8). As noted above, together with our present data, this suggests that phosphoinositide binding to SGK1 is only required for SGK1 action downstream of PI 3-kinase-dependent activation and that phosphoinositides generated by PI 3-kinase may not be implicated in this phase of SGK1 function.
Role of phosphoinositides in SGK1 regulation of diverse cellular processes.
In view of the functional similarities between SGK1 and SGK3, as well as their shared interactions with PI(3)P, our present view is that PI(3)P binding to SGK1 is important for SGK1 regulation of ENaC. It is notable that PI(3)P directs various proteins, including SGK3, to the early endosome (16, 17, 44, 55); similarly, it may direct SGK1 to this compartment. Previous reports support a role for SGK1 and Nedd4-2 in reciprocally modulating ENaC trafficking events in the endocytic pathway (7, 18, 32), although localization of SGK1 along this pathway remains to be demonstrated.
SGK1 has been found in multiple subcellular compartments, including nucleus (13, 35, 37, 42), cytoplasm (11, 13, 35, 42), plasma membrane (11), endoplasmic reticulum (5, 9), and mitochondria (22), which likely reflect the multiple functions SGK1 can perform within a given physiological context. SGK1 has a nuclear localization sequence (NLS) located within its kinase domain, yet our findings demonstrate that disruption of the NH2-terminal lipid binding domain weakens the suppressive effect of SGK1 on Forkhead-driven gene transcription (Fig. 6). This suggests that the NLS is not sufficient for directing SGK1 repression of FKHRL1 and is consistent with the possibility that nuclear phosphoinositides, including PI(3)P (24), PI(4)P (15), or PI(5)P (15, 25), are implicated in SGK1 nuclear localization under the appropriate conditions. With regard to SGK1 localization to endoplasmic reticulum, recent studies have implicated a stretch of hydrophobic amino acids within the NH2-terminus [residues 19–24 (9) or 18–30 (5)] that is important for targeting SGK1 to this compartment and for ubiquitination and proteasomal degradation. Another recent study identified overlapping residues (17–32) as important for targeting SGK1 to the mitochondria and for proteasomal degradation (22). While the mechanisms by which the NH2 terminus directs SGK1 to the endoplasmic reticulum or mitochondria remain unclear, it is notable that the residues implicated in SGK1 subcellular targeting overlap with the lipid binding domain; protein-lipid interactions provide a compelling mechanism for SGK1 recruitment to these various compartments. For instance, in yeast and mammalian cells, the major site for synthesis of PS is the endoplasmic reticulum; after its synthesis, PS can be transported to the plasma membrane, mitochondria, Golgi apparatus, and endosomes (47, 50–54, 56). We propose that the more promiscuous phosphoinositide binding profile of the SGK1 NH2 terminus might contribute to its regulation of diverse cellular processes by directing it to these multiple compartments in a context-dependent fashion.
This work was supported by grants from National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases [K08-DK-073487 (to A. C. Pao), P30-DK-63720 (to D. Pearce), and R01-DK-56695 (to D. Pearce)] and from the American Heart Association Western States Postdoctoral Fellowship (0225080Y to J. A. McCormick).
Present address of J. A. McCormick: Div. of Nephrology and Hypertension, Dept. of Medicine, Oregon Health and Science University, Portland, OR 97239.
We are grateful to K. Ashrafi for helpful comments on the manuscript.
↵* A. C. Pao and J. A. McCormick contributed equally to this work.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society