An amphipathic helix targets serum and glucocorticoid-induced kinase 1 to the endoplasmic reticulum-associated ubiquitin-conjugation machinery

Abstract

Serum- and glucocorticoid-induced kinase 1 (Sgk1) regulates many ion channels and transporters in epithelial cells and promotes cell survival under stress conditions. In this study we demonstrate that Sgk1 is a short-lived protein regulated by the endoplasmic reticulum (ER)-associated degradation system and subcellular localization to the ER. We identified a hydrophobic motif (residues 18–30) as the signal for ER localization and rapid degradation by the ubiquitin (Ub)/proteasome pathway in both yeast and mammalian cells. Deletion or reduction of hydrophobicity of the motif redistributes Sgk1 to the cytosol and nucleus and markedly increases its half-life. We determined that the Ub-conjugating UBC6 and UBC7 and the Ub ligase HRD1 are the ER-associated Ub enzymes that mediate degradation of Sgk1; thus, Sgk1 has been identified as a cytosolic substrate for mammalian HRD1. Compartmentalization of Sgk1 controls the functional and spatial specificities of Sgk1-mediated signaling pathways, whereas rapid protein turnover provides a means to rapidly adjust Sgk1 abundance in response to different hormonal and external stimuli that increase Sgk1 gene transcription.

Serum- and glucocorticoid-induced kinase 1 (Sgk1) is a S/T protein kinase that belongs to the AGC (cAMP-dependent, cGMP-dependent, and protein kinase C) kinase family. Sgk1 is expressed in many cell types and participates in numerous cellular processes. Sgk1's best-characterized function is in the regulation of ion channels and transporters (1) as illustrated by the phenotype of Sgk1 null mice. The kidneys of these mice waste sodium under a low-salt diet owing to decreased activity of the epithelial sodium channel, ENaC (2, 3). However, additional functions are expected because Sgk1 expression is under acute transcriptional control by serum, nuclear receptors (glucocorticoids and aldosterone), surface receptors, and cellular stress (4, 5).

The primary sequence of Sgk1 predicts a soluble protein without any known membrane-binding motif. Three domains are distinguished in the structure of this family of kinases. The N terminus is the most variable and plays a regulatory function. The N terminus of Sgk1 is short and does not contain a canonical membrane-binding domain, whereas in the related kinases the N terminus encodes a pleckstrin homology domain in thyoma vical proto-oncogene (Akt/PKB) (6) or a phox homology domain in Sgk3 (7). These domains bind phosphoinositides to direct localization of the kinases to membrane compartments. The catalytic domain is in the middle of the protein, and the C terminus contains a conserved hydrophobic motif that is essential for the activation of all members of the family.

Activation of Sgk1 requires two phosphorylation events carried out by different upstream kinases. Ser-422 is phosphorylated first by phosphoinositide-dependent protein kinase 2 (PDK2) (8). The molecular identity of PDK2 has been determined only for Akt, for which the mTOR/raptor complex fulfills this role (9); however, it is possible that more than one kinase may function as PDK2 (10). Phosphorylation of Ser-422 or substitution by an acidic residue creates a docking site to recruit PDK1, which phosphorylates Thr-256 to render Sgk1 active (11, 12).

There is evidence that Sgk1 might also be regulated at the posttranscriptional level by rapid proteolysis. Many tissues express abundant SGK1 mRNA, whereas the protein is barely detectable (13). Moreover, it has been found that Sgk1 is short-lived in transfected cells and can be ubiquitinated (14). Ubiquitination of proteins proceeds by the sequential action of three types of enzymes. E1 activates ubiquitin (Ub), E2 transfers Ub from E1 to the amino group of a lysine residue of the target protein, and E3, or Ub-protein ligase, is responsible for substrate recognition (15).

In this study we demonstrate that Sgk1 is ubiquitinated and degraded at the ER membrane by the action of the ER-associated Ub-conjugating enzymes UBC6 and UBC7 (E2) and the ligase (E3) HRD1 (synoviolin) (16). A hydrophobic α-helix located within the N terminus of Sgk1 serves as a signal for targeting the protein to the ER for ubiquitination and subsequent degradation. These findings reveal that subcellular localization is a key factor controlling the metabolic stability of Sgk1, and rapid Sgk1 turnover is mediated by a transmembrane Ub ligase of the ER.

Results

Distinct Protein Stability of Sgk1 Compared with Closely Related Members of the AGC Family.

We examined the half-life (t_1/2) of protein kinases of the AGC family: Sgk1, Sgk2, and Akt, by pulse–chase experiments of transfected CHO cells (Fig. 1 A and B). The t_1/2 of Sgk1 was 28 min, the t_1/2 of Sgk2 was 180 min, and the t_1/2 of Akt1 was 120 min. These experiments demonstrate that Sgk1 is rapidly degraded, exhibiting a significantly shorter t_1/2 than the related kinases when expressed in the same cell line. As the first 60 residues of Sgk1 are absent in Sgk2, we tested whether this region determines the difference in protein stability. The degradation rates of a series of N-terminal deletions of Sgk1 were measured. The t_1/2 determined from densitometric analysis had the following values: 30 min for Δ17, 120 min for Δ33, 120 min for Δ45, and 180 min for Δ60 (Fig. 1 C and D). Longer N-terminal deletions up to 80 residues did not further change the t_1/2. Therefore, the segment between residues 17 and 33 is a key determinant of Sgk1 stability. To ascertain whether this segment is sufficient for rapid degradation, we transferred the first 60 residues of Sgk1 to the N terminus of GST in CHO cells. The t_1/2 of 60Sgk1-GST fusion protein was indeed significantly shortened relative to GST itself (t_1/2 of 28 and 200 min, respectively; Fig. 1 E and F). Hence, a transferable degradation signal is contained in the N-terminal domain of Sgk1.

Fig. 1.

Stability of Sgk1 is determined by the N-terminal domain. t_1/2 and densitometric analysis determined by pulse–chase experiments of transfected CHO cells. (A and B) Mouse Sgk1, Sgk2, and Akt1. (C and D) Progressively N-terminal truncations of Sgk1. (E and F) GST and the fusion protein 60Sgk1-GST. Data points are the average of four experiments ± SD. Lines represent the fit of the data to a single exponential.

Characterization of the Degradation Signal of Sgk1.

Analysis of the N terminus of Sgk1 revealed a highly hydrophobic domain encompassing residues R18–Q30, with a Kyte–Doolittle hydrophobicity score of 3.4 by using PEPWINDOW software (http://bioweb.pasteur.fr/seqanal/interfaces/pepwindow.html). Further protein sequence analysis to predict the secondary structure and distribution of hydrophobic residues in this region indicated that amino acids R18–Q30 probably form an amphipathic α-helix (Fig. 2A). Substitution of alanines (A) in positions 22 and 26 on the hydrophilic side of the helix with aspartates (D) in the 60Sgk1-GST construct the hydrophobicity score decreased to 2.0 but did not change the rate of degradation. In contrast, when two additional D residues were introduced in the hydrophobic side of the helix (V21 and F27), the hydrophobicity dropped to a score of 1.0 and Sgk1 was strongly stabilized. Fig. 2B shows pulse–chase analyses of GST, destabilization by fusion with the N terminus of Sgk1 (60Sgk1-GST), and return to the original long half-life of GST after disruption of the amphipathic α-helix (60Sgk14D-GST).

Fig. 2.

Effects of the Sgk1 degron on stability and localization of 60Sgk1-GST fusion protein. (A) Helical wheel representation of Sgk1 residues R18–Q30. (Left) The WT sequence. (Right) The change in hydrophobicity pattern produced by four Asp (D) mutations at positions 21, 22, 26, and 27 of the α-helix. Gray and white circles represent hydrophobic and hydrophilic residues. (B) Pulse–chase of CHO cells transfected with WT GST, 60Sgk1-GST, or 60Sgk14D-GST. (C) Immunofluorescence of CHO cells transfected with the constructs indicated in B and labeled with anti-GST antibody and Hoescht dye. (1 and 2) GST. (3 and 4) 60Sgk1-GST. (5 and 6) 60Sgk14D-GST. (Magnification: C, ×60.)

The cellular distribution of the three previous constructs indicated that GST localizes throughout the cytosol and nucleus (Fig. 2C 1 and 2), whereas 60Sgk1-GST exhibits a reticular and perinuclear distribution (Fig. 2C 3 and 4). Mutations of the amphipathic α-helix in 60Sgk14D-GST reverted the distribution to the pattern of GST alone (Fig. 2C 5 and 6). Therefore, the amphipathic helix formed by residues R18–Q30 constitutes a key determinant for Sgk1 degradation and localization to a perinuclear compartment.

The compartment where Sgk1 localizes was defined by cotransfecting CHO cells with WT or Δ60Sgk1 and calnexin-V5. Fig. 3 shows colocalization of Sgk1 with calnexin, whereas Δ60Sgk1 distributes homogeneously over the whole cell, including cytosol and nucleus. Therefore, Sgk1 associates to the ER membrane.

Fig. 3.

Immunolocalization of WT Sgk1 and Δ60Sgk1 in transfected CHO. Cofocal images of CHO cells cotransfected with Sgk1 and Calnexin-V5 (Upper) or Δ60Sgk1 and Calnexin-V5 (Lower). (Magnification: ×60.)

The ER-Associated Ub Proteasomal Machinery Mediates Sgk1 Rapid Degradation.

When the proteasome was blocked with lactacystin in CHO and M1 cells we observed a marked increase in Sgk1 t_1/2, from 28 min to 3 h (Fig. 4A) and accumulation of ubiquitinated species migrating more slowly in the gel than unmodified Sgk1, consistent with polyubiquitination of Sgk1 (Fig. 4B). To identify the ubiquitination enzymes involved in degradation of Sgk1, we first examined Sgk1 degradation in the yeast Saccharomyces cerevisiae, where genetic manipulation of the Ub system is easier than in mammalian cells. Cycloheximide chase experiments yielded Sgk1 t_1/2 of 41 min in WT cells (Fig. 5A). Little, if any, degradation was observed for Δ60-Sgk1 (data not shown).

Fig. 4.

Sgk1 degradation by the Ub-dependent pathway. (A) Pulse–chase of Sgk1 in transfected CHO cells with or without lactacystin (10 μM) for 12 h. (B) Western blot of ubiquitinated Sgk1 in M1 cells and +/− transfected CHO cells with Sgk1 with or without treatment with lactacystin. IP, immunoprecipitation; IB, immunoblot.

Fig. 5.

Ub-conjugating enzymes and Ub ligases involved in Sgk1 degradation in Saccharomyces cerevisiae. (A) (Left) Cyclohexamide (CHX) chase of Sgk1 in transformed yeast strains: WT, Ubc4/5 and Ubc6/7 null mutants. (Right) Calculated t_1/2 of Sgk1 in the indicated yeast strains. (B) (Left) [³⁵S]Methionine pulse–chase experiments of Sgk1 in WT and Doa10/Hrd1 double mutant strains. (Right) Calculated t_1/2 of Sgk1 from the corresponding experiments. (C) Immunostaining of spheroplasts from doa10D mutant yeast transformed with Sgk1-HA and Δ60Sgk1-HA stained with monoclonal HA and DAPI. (Scale bars: 4 μm.)

S. cerevisiae has 11 different E2 (Ubc) enzymes. We screened strains lacking specific members of this family, beginning with a mutant deleted for Ubc4 and Ubc5, two closely related cytosolic E2s that participate in many different ubiquitination pathways. Sgk1 exhibited a t_1/2 of ≈50 min in the ubc4Δubc5Δ double mutant, only slightly longer than in WT cells. In contrast, in cells lacking two E2s associated with the ER membrane, Ubc6 and Ubc7, the t_1/2 of Sgk1 increased to 135 min. These results indicate that Ubc6 and Ubc7 make a major contribution to Sgk1 degradation in yeast.

Ubc6 and Ubc7 function with two E3s, Hrd1 and Doa10, at the ER membrane. We therefore transformed Sgk1 into yeast strains lacking either or both of these two transmembrane E3 ligases. The t_1/2 of Sgk1 was ≈50 min in the WT strain and >2 h in the doa10Δhrd1Δ double mutant (Fig. 5B). doa10Δ and hrd1Δ single mutants stabilized Sgk1 almost as much as the double mutant (Fig. 8, which is published as supporting information on the PNAS web site). Hence, specific ER-associated E2 and E3 enzymes mediate degradation of Sgk1 in yeast. We also found that Sgk1 localized to the yeast ER and Δ60-Sgk1 was instead diffusely distributed throughout the cell (Fig. 5C).

To determine whether an ER-linked Ub-dependent degradation pathway is also responsible for Sgk1 in mammalian cells, we first cotransfected CHO cells with plasmids expressing Sgk1 and either mouse UBC6 or UBC7 (Fig. 6A). Overexpression of either of these E2s decreased Sgk1 t_1/2 to ≈17 min. In contrast, coexpression of Δ60Sgk1 with UBC6 or UBC7 left the t_1/2 unchanged at ≈180 min (Fig. 9, which is published as supporting information on the PNAS web site).

Fig. 6.

Ub-conjugating enzymes and Ub ligases involved in Sgk1 degradation in mammalian cells. (A) Pulse–chase experiments of Sgk1 in CHO transfected with Sgk1 alone or with UBC6 or UBC7. Similar experiments in cells cotransfected with three different ER-anchored E3 ligases and corresponding inactive mutants. IP, immunoprecipitation. (B) gp78 or gp78R2M. (C) TEB4 and TEB4_C9S. (D) HRD1 or HRD1_C291S. (E–G) Graphical representation of the fraction of initial protein at three different time points. Each data point represents the average of two to three measurements. (H) Pulse–chase of Sgk1 cotransfected with scrambled siRNA, siRNA of TEB4, or HRD1 in HEK-293 cells. (I) RT-PCR of endogenous TEB4 and HRD1 in HEK cells transfected with siRNAs specific for the indicated E3 ligases. Controls are mock-transfected cells and scrambled siRNA. (J) Western blot of endogenous TEB4 in HEK cells transfected with siRNA scrambled and siRNA specific for human TEB4. IB, immunoblot.

The effect of overexpressing the mammalian E3 ligases known to reside in the ER membrane [gp78 (17), TEB4 (18), and HRD1 (16)] was examined by cotransfecting either WT or inactive versions of each of these E3s together with Sgk1. The E3s can be inactivated by mutation of one of the zinc-coordinating Cys residues in the RING-related domain. gp78, TEB4, or their inactive derivatives did not have any effect on Sgk1 degradation (Fig. 6 B–F). In contrast, HRD1 overexpression decreased the t_1/2 of Sgk1 to 16 min, whereas overexpression of inactive HRD1 exhibited a dominant-negative effect, increasing the t_1/2 of Sgk1 to 80 min (Fig. 6 D–G). To verify the results obtained by overexpression of E3s, we conducted the reversed experiment, knocking down expression of endogenous human HRD1 and TEB4 (the mammalian homolog of yeast Doa10) in HEK-293 cells cotransfected with Sgk1 and sequence-specific siRNAs against HRD1 or TEB4. Marked stabilization of Sgk1 protein was observed with siRNA against HRD1 but not with siRNA against TEB4 or a scrambled siRNA (Fig. 6H). Additionally, the specificity of the siRNAs was demonstrated by a significant decrease in HRD1 and TEB4 mRNAs (RT-PCR; Fig. 6I) by the cognate siRNAs. Moreover, even though TEB4 mRNA was markedly decreased by the TEB4 siRNA, the t_1/2 of Sgk1 was not affected (Fig. 6H), whereas suppression of HRD1 mRNA stabilized Sgk1. Under the same conditions, TEB4-specific siRNA markedly diminished expression of endogenous TEB4 protein level (Fig. 6J). Thus, HRD1 and UBC6/UBC7, all of which are associated with the ER membrane, mediate ubiquitination and degradation of Sgk1 in mammalian cells.

Degradation and Localization of Endogenous Sgk1 in M1 Cells.

The t_1/2 of endogenous Sgk1 in M1 cells was found to be 20 min, slightly shorter than the value obtained in transfected cells (Fig. 7 A and B); the degradation is also mediated by endogenous HRD1 as demonstrated by down-regulation of mouse HRD1 expression by three different siRNAs (Fig. 7 C and D). As in transfected CHO cells, endogenous Sgk1 in M1 cells localizes at the ER (Fig. 7E). Together, these results prove that endogenous Sgk1 is rapidly degraded at the ER by the HRD1.

Fig. 7.

Degradation and localization of endogenous Sgk1 in M1 cells. (A) Autoradiography of representative pulse–chase experiment of endogenous Sgk1 (≈1 month of exposure). (B) Graphic of t_1/2 of endogenous Sgk1. (C) Autoradiography of immunoprecipitated Sgk1 in M1 cells nontransfected and transfected with scrambled siRNA and three different oligos specific for mouse HRD1 (4-day exposure). (D) Average of protein abundance of former experiments (n = 3). (E) Confocal images of M1 cells showing colocalization of endogenous Sgk1 with transfected calnexin. Peptide panel indicates preincubation of anti-Sgk1 antibody with cognate peptide. (Magnification: E, ×60.)

Other Potential Pathways for Sgk1 Turnover.

A small fraction of Sgk1 remains in the cytosol after separation of the microsomal fraction by differential centrifugation (Fig. 10A, which is published as supporting information on the PNAS web site). To define the pathway of degradation of the cytosolic component of Sgk1, we examined the effect of the soluble E3 ligase Nedd4-2 (a HECT domain E3), which has been previously shown to interact with Sgk1 via a PPFY motif located in the catalytic domain of the kinase (19, 20). We cotransfected with Nedd4-2/V5 and WT Sgk1 or Sgk1-PY, which has the PPFY motif mutated to LPFR to prevent interaction with Nedd4-2. Pulse–chase experiments showed no change in Sgk1 t_1/2 (28 min ± Nedd4-2) and no change for Sgk1-PY (Fig. 10B). Moreover, mutants with different levels of enzymatic activity: Sgk1_S422D (constitutively active), Sgk1_S422A (not phosphorylatable), and Sgk1_K127A (kinase dead) (21) did not change the t_1/2 of Sgk1. Full activation of Sgk1 was ensured by contransfection of PDK1-V5 (22) (Fig. 10D).

We also investigated the effect of Ca²⁺-activated proteases. Calpains and other Ca²⁺-activated proteases degrade soluble and membrane-bound proteins containing PEST sequences (23–25). We identified a PEST sequence motif (PEST score + 6.7, significant ≥ +5) in the N terminus of Sgk1 encompassed by residues H₆₂ to H₉₂ (http://bioweb.pasteur.fr/seqanal/interfaces/pestfind-simple.html). Neither Ca²⁺ nor deletion of PEST sequence in Sgk1 had significant effect on the t_1/2 of Sgk1 (Fig. 11, which is published as supporting information on the PNAS web site). Therefore, degradation of the cytosolic fraction of Sgk1 is not increased by expression of Nedd4-2, the state of activation of Sgk1, the PEST sequence, or Ca²⁺-dependent pathways.

Discussion

Sgk1 Is a Novel Substrate of HRD1.

We have demonstrated that Sgk1 concentrates at the ER where it is targeted for degradation by the ER-associated Ub-proteasome pathway. A transmembrane E3 Ub ligase of the ER, HRD1, is required for Sgk1 degradation. ER localization and degradation depend on a short amphipathic helix in the N-terminal domain of Sgk1.

The ER-associated degradation (ERAD) system is a component of ER protein quality control (26). Misfolded or misassembled membrane or luminal proteins, and some normal but naturally short-lived ER-resident proteins, are polyubiquitinated at the ER membrane, retrotranslocated to the cytosol, and degraded by the proteasome. Certain soluble cytoplasmic or nuclear proteins can also reach the ER and become substrates for the ERAD system. An example of this is the yeast transcriptional repressor Matα2, which is ubiquitinated at the ER/nuclear envelope membrane by the Ubc6 and Ubc7 E2s (27) and the ER transmembrane E3 Doa10 (22). In mammalian cells, the cytosolic TRAF2 signaling protein relocates to the ER along with the cIAP1 E3, where cIAP1 functions with UBC6 to ubiquitinate TRAF2 (28). Here we identify HRD1 as the major E3 involved in Sgk1 ubiquitination in mammalian cells and UBC6 and UBC7 as the likely E2s involved. When expressed in yeast, Sgk1 also localizes to the ER and is degraded by the ERAD system.

For the Matα2 substrate, the first 62 residues of the protein are sufficient for rapid degradation by Doa10; this segment, called the Deg1 degron, includes an amphipathic helix essential for substrate ubiquitination (29). The degron or degradation signal of Sgk1 by the ERAD pathway is also encoded in its N-terminal domain, specifically, a 13-residue amphipathic helix, which is consistent with findings by Gilon et al. (30) that highly hydrophobic regions serve as degrons for proteins channeled to the Ubc6/Ubc7 degradation pathway in yeast. Whether the amphipathic helix is directly recognized by HRD1 or an intermediary protein cannot be distinguished at this point. Nevertheless, these results are consistent with the partitioning of a significant portion of Sgk1 to microsomes as we previously demonstrated (31).

Sgk1 is a cytosolic in vivo substrate for HRD1. This finding highlights an unanticipated difference between the yeast and mammalian ERAD systems. In yeast, all substrates reported to date for Hrd1 are either membrane or luminal proteins of the ER; almost all of the aberrant membrane protein substrates of Hrd1 carry luminal lesions. Doa10, in contrast, can recognize cytosolic proteins and ER membrane proteins, and the latter usually bear cytosol mutations. Therefore, it was proposed that the yeast ERAD system was functionally divided into two branches: the ERAD-L (luminal) and ERAD-C (cytosolic) pathways with the primary E3s in each pathway being Hrd1 and Doa10, respectively (32). However, Sgk1 expressed in yeast is substrate for both E3s. In mammalian cells, Sgk1 degradation requires HRD1 but not TEB4, the mammalian ortholog of Doa10.

Functional Implications of Sgk1 Rapid Degradation and ER Localization.

We demonstrate that endogenous or overexpressed Sgk1 is a short-lived protein, explaining the difficulties in detecting Sgk1 in most tissues and the simultaneously abundant mRNA and sparse protein levels observed in vivo (13). Uniquely among kinases of the AGC family, Sgk1 is regulated to a great extent by changes in transcriptional levels. Different factors and external cues increase Sgk1 transcription (33). The different stimuli converge on a single level of regulation, diminishing the temporal and stimulus-dependent specificity of the response. However, as we show here, a degron signal in Sgk1 maintains low basal abundance of expression and prevents accumulation of Sgk1 to very high levels even when more than one transcriptional activator acts on a tissue. Thus, rapid degradation and the localization of Sgk1 to the ER confer functional and spatial specificity to the signaling pathway(s) mediated by Sgk1.

In contrast to a previous report of localization and ubiquitination of Sgk1 at the plasma membrane (14), we demonstrate that Sgk1 localizes to the ER where it is ubiquitinated and degraded by the ER-associated Ub-proteasomal machinery. Localization of Sgk1 predominantly to the ER in resting cells has additional implications for Sgk1 function. Targeting of Sgk1 to the ER by its degron is the default pathway in mammalian cells and yeast. This process is not altered by phosphorylation of T256 and S422 or the enzymatic activity of Sgk1. However, in certain cell types, Sgk1 shuttles between the nucleus and cytoplasm as a function of the cell cycle (34) or hormonal stimulation (35). In epithelial cells, it has been suggested that Sgk1 migrates to the plasma membrane to modify the activity of many ion channels and transporters (5). As yet, the molecular events that determine translocation of Sgk1 to the nucleus or the plasma membrane have not been defined, but it is conceivable that certain signals may prevent Sgk1 from reaching the ER and being degraded. For instance, a binding protein(s) might mask the N-terminal degron, keeping Sgk1 from relocating to the ER.

Additional roles for the ER localization of Sgk1 are possible. Sgk1 might associate with a scaffold protein at the ER membrane that brings together Sgk1 with upstream activators and downstream effectors to increase the efficiency and specificity of the Sgk1-mediated signaling pathway. In this regard, the mammalian target of rapamycin, mTOR, recently identified as a PDK2 that phosphorylates the hydrophobic motif of Akt (36), also localizes to the cytosolic side of the ER (37). Akt and Sgk1 both respond to the common upstream stimuli of insulin and increased phosphoinositide-3-phosphate levels, which activate PDK2. But whereas Akt subsequently translocates to the plasma membrane by binding phosphoinositide-3-phosphate to its pleckstrin homology domain (6), Sgk1 remains associated with the ER. Hence, upon insulin stimulation, Akt phosphorylates targets near the plasma membrane, whereas Sgk1 may phosphorylate proteins near the ER membrane. Our findings indicate that compartmentalization represents a mechanism of segregating the signaling pathways of these closely related kinases. This helps ensure their functional specificity even though these two kinases recognize similar phosphorylation motifs when tested by in vitro assays.

Experimental Procedures

Plasmids and Constructs.

Mouse Sgk1 cDNA was amplified by RT-PCR with an HA tag at the C terminus and cloned into pcDNA3.1 TOPO (Invitrogen). N-terminal truncated Sgk1 lacking the first 17, 33, 45, 60, or 80 residues was obtained by PCR. Point mutations and deletion of a PEST sequence were made with a Quikchange mutagenesis kit (Stratagene). For experiments in yeast, full-length and truncated forms of Sgk1 were subcloned into pCu416CUP1 vector (38). All constructs were sequenced at the Keck Facility at Yale University. Full-length GST cDNA was subcloned into pcDNA3.1. The first 60 residues of Sgk1 were fused to the N terminus of GST to make 60Sgk1-GST. Mouse Sgk2, calnexin, Nedd4-2, PDK1, TEB4, gp78, UBC6, and UBC7 cDNAs were amplified by RT-PCR and cloned into pcDNA3.1-TOPO. HA epitope was fused at the 5’ end of Akt1, Sgk2, and gp78 cDNA. Calnexin, Nedd4-2, and PDK1 were in-frame with the V5 epitope of the pcDNA3.1. cDNAs of humam TEB4 and gp78 were provided by A. M. Weissman (National Institutes of Health, Bethesda, MD); human HRD1 was provided by Yuval Reiss (Rehovot, Israel); and Akt1 was provided by Anton Bennett (Yale University School of Medicine).

Cell Transfection/Transformation.

Mammalian cell cultures and transfections.

M1 cells were propagated in ATCC complete medium, CHO in α-minimal essential medium supplemented with 10% FBS. Cells were transfected with the indicated plasmids by Lipofectamin-2000 (Invitrogen). Where endogenous human Ub ligases were knocked down we used HEK-293 cells cotransfected with Sgk1 and various siRNAs: Ub ligase HRD1, TEB4, or scrambled. The sense sequences of the specific siRNAs were: TEB4 (TTAAGAGTGTGCTGCCTAA) and HRD1 (CGTTCCTGGTACGCCGTCA). For endogenous mouse HRD1 the siRNAs were: ID 177914, ID 177913, and 83279, and scrambled, control 1 siRNA, all from Ambion, Austin, TX.

Yeast strains and transformations.

Yeast strains were as described (22). Cells were grown in rich (yeast extract/peptone/dextrose) and Synthetic Defined (SD) media and transformed with pCu416CUP1-Sgk1 by the one-step transformation method.

Pulse–Chase Experiments.

Mammalian cells.

CHO or HEK-293 cells transiently transfected were washed with DMEM serum-free medium without methionine and cysteine, labeled with 150 μCi/ml of an Express Cell Labeling Mix (PerkinElmer Life Sciences) for 30 min, and chased in medium with 10-fold molar excess of both methionine/cysteine and 0.1 mg/ml of cyclohexamide. Cells were lysed (150 mM NaCl, 5 mM EDTA, 50 mM Hepes, pH 7.5, and 1.0% Triton X-100) and immunoprecipitated with anti-Sgk_ct antibody (31) and protein A beads (Sigma). Beads were washed with lysis buffer and eluted with SDS/PAGE sample buffer followed by electrophoresis. Gels were exposed to x-ray film and analyzed by densitometry with Bio-Rad G800 and QuantityOne software.

Yeast.

Yeast strains transformed with pCu416CUP1-Sgk1 were grown in selective medium (SD-URA) to midlog phase (OD₆₀₀ ≈1.0) at 30°C. Cells were washed in Synthetic Defined (SD) with 2% glucose and without methionine and were labeled with 200 μCi of [³⁵S]methionine (19). Afterward, cells were resuspended in chase medium with 1 mg/ml of methionine and 0.3 mg/ml of cycloheximide. For each chase time point, cells were lysed (2% SDS/90 mM Hepes, pH 7.5/30 mM DTT) and boiled. Lysates were diluted in 1% Triton X-100 lysis buffer with protease inhibitors (Complete; Roche, Manaheim, Germany). Radioactivity counts were determined from a trichloroacetic acid (TCA)-insoluble fraction. Equal amounts of TCA-insoluble ³⁵S cpm were immunoprecipitated with anti-Sgk1_ct antibody. Gels were exposed to a Phosphoimager for quantification.

Cycloheximide chase.

Yeast cultures were concentrated to 2.5 OD₆₀₀ per ml of Synthetic Defined (SD) medium. After addition of cycloheximide (0.3 mg/ml), aliquots were taken at different time points, treated with 0.1 M NaOH, and centrifuged. Pellets were resuspended in SDS sample buffer and used for immunoblot analysis.

Western Blot Analysis.

After electrophoresis in 10% SDS/PAGE samples were transferred to Immobilon-P membranes (Millipore), blocked, and probed with primary antibodies: anti-HA-HRP, monoclonal anti-myc, anti-tubulin, or Ub (P4D1) all from Santa Cruz Biotechnology followed by incubation with anti-mouse IgG conjugated with HRP (Chemicon). Endogenous TEB4 was detected with an antibody developed by L.W. Signals were developed with ECL⁺ (Amersham Pharmacia), and blots were exposed to BioMax MR Film (Eastman Kodak).

Immunofluorescence Microscopy.

Cells were fixed with 4% formaldehyde, permeabilized (0.5% Triton X-100), and blocked with 10% goat serum. Primary antibodies were added to the blocking solution and incubated for 1 h. Cells were washed, and secondary antibodies Alexa Fluor-488 goat anti-rabbit IgG or Alexa Fluor-594 goat anti-mouse IgG (Molecular Probes) were added at 1:400 dilution. Slides were covered with DAPI mounting solution (Vectashield; Vector Laboratories). Sgk1 in M1 cells transfected with calnexin-V5 and pretreated with dexamethasone was detected with tyramide signal amplification, TSA Fluorescence System (NEN Life Science Products). Cells were examined with a Zeiss IE-25 LSM510 Meta confocal microscope.

Abbreviations

Sgk1

serum- and glucocorticoid-induced kinase 1

ER

endoplasmic reticulum

ERAD

ER-associated degradation

PDK

phosphoinositide-dependent protein kinase

Ub

ubiquitin