A Phosphorylation Cascade Controls the Degradation of Active SREBP1*

Sterol regulatory element-binding proteins (SREBPs) are a family of transcription factors that regulates cholesterol and lipid metabolism. The active forms of these transcription factors are targeted by a number of post-translational modifications, including phosphorylation. Phosphorylation of Thr-426 and Ser-430 in SREBP1a creates a docking site for the ubiquitin ligase Fbw7, resulting in the degradation of the transcription factor. Here, we identify a novel phosphorylation site in SREBP1a, Ser-434, which regulates the Fbw7-dependent degradation of SREBP1. We demonstrate that both SREBP1a and SREBP1c are phosphorylated on this residue (Ser-410 in SREBP1c). Importantly, we demonstrate that the mature form of endogenous SREBP1 is phosphorylated on Ser-434. Glycogen synthase kinase-3 phosphorylates Ser-434, and the phosphorylation of this residue is attenuated in response to insulin signaling. Interestingly, phosphorylation of Ser-434 promotes the glycogen synthase kinase-3-dependent phosphorylation of Thr-426 and Ser-430 and destabilizes SREBP1. Consequently, mutation of Ser-434 blocks the interaction between SREBP1 and Fbw7 and attenuates Fbw7-dependent degradation of SREBP1. Importantly, insulin fails to enhance the levels of mature SREBP1 in cells lacking Fbw7. Thus, the degradation of mature SREBP1 is controlled by cross-talk between multiple phosphorylated residues in its C-terminal domain and the phosphorylation of Ser-434 could function as a molecular switch to control these processes.

DNA binding, and mutation of Ser-434 blocks the phosphorylation of Thr-426 and Ser-430 during this process, suggesting that the phosphorylation of Ser-434 could control the degradation of active SREBP1 molecules. Thus, the degradation of mature SREBP1 is controlled by cross-talk between multiple phosphorylated residues in its C-terminal domain, and the phosphorylation of Ser-434 could function as a molecular switch to control these processes.

EXPERIMENTAL PROCEDURES
Cell Culture-All tissue culture media and antibiotics were obtained from Invitrogen and Sigma. HEK293, HEK293T, HepG2, U2OS, and HeLa cells were from ATCC. Fbw7-positive and Fbw7-negative HCT116 cells were provided by B. Vogelstein (31).
Generation of Phosphorylation-specific SREBP1 Antibodies-Synthetic phosphopeptides corresponding to residues 431-437 (Ser-434 phosphorylated) in human SREBP1a were coupled to keyhole limpet hemocyanin before being injected into rabbits. The phosphopeptides and the corresponding nonphosphorylated peptides, as well as phospho-Ser, were coupled to Sulfolink (Pierce) and used as affinity matrices to purify the antibodies from rabbit sera (9,11).
Immunoprecipitations and Immunoblotting-Cells were lysed in buffer A (50 mM HEPES (pH 7.2), 150 mM NaCl, 1 mM EDTA, 20 mM NaF, 2 mM sodium orthovanadate, 10 mM ␤-glycerophosphate, 1% (w/v) Triton X-100, 10% (w/v) glycerol, 1 mM phenylmethylsulfonyl fluoride, 10 mM sodium butyrate, 1% aprotinin, 0.1% SDS, and 0.5% sodium deoxycholate) and cleared by centrifugation. For co-immunoprecipitations, cell lysates were prepared in the absence of SDS and sodium deoxycholate. Cell lysates and immunoprecipitates were resolved by SDS-PAGE and transferred to nitrocellulose membranes (Millipore). To ensure that equal amounts of protein were loaded in each well, the levels of ␣-tubulin in the samples were estimated by Western blotting.
Determination of Protein Half-life-Cells were treated with cycloheximide to stop protein synthesis and incubated for the indicated times. Total cell lysates were prepared, and the proteins were separated by SDS-PAGE and transferred to nitrocellulose membranes. SREBP1 was visualized by Western blotting followed by quantitation on a charge-coupled device camera (Fuji) and image analysis software (Aida Image Analyzer 3.10).
Luciferase and ␤-Galactosidase Assays-Cells were transiently transfected with the indicated promoter-reporter genes in the absence or presence of expression vectors for SREBP1, either wild-type or the indicated mutants. After 36 h, luciferase activities were determined in duplicate samples as described by the manufacturer (Promega, Madison, WI). The pCH110 vector encoding the ␤-galactosidase gene under the control of the SV40 promoter (Amersham Biosciences) was used as an internal control for transfection efficiency. Luciferase values (relative light units) were calculated by dividing the luciferase activity by the ␤-galactosidase activity. The data represent the average Ϯ S.D. of three independent experiments performed in duplicates.
Reverse Transcription-PCR Assays-RNA was extracted with TRIzol reagent (Invitrogen). Total RNA was subjected to reverse transcription with oligo(dT), followed by PCR with target-specific primers. The PCR reactions, using Invitrogen High Fidelity DNA polymerase, were optimized for the individual target genes. The PCR programs and primer sequences for the human LDL receptor, HMG-CoA synthase, fatty acid synthase, and glyceraldehyde-3-phosphate dehydrogenase genes are available on request.

GSK-3 Phosphorylates Ser-434 in SREBP1-
We recently identified two phosphorylated residues in the C terminus of mature SREBP1a, Thr-426 and Ser-430 (9). GSK-3 is involved in the phosphorylation of both Thr-426 and Ser-430. However, kinase assays with recombinant GSK-3␤ and SREBP1a followed by phosphopeptide mapping indicated that the C terminus of mature SREBP1a contained additional residues phosphorylated by GSK-3 (data not shown). The C-terminal domains of mature SREBP1a and SREBP1c are identical. Data base searches suggested that Ser-434 in SREBP1a (Ser-410 in SREBP1c) could be phosphorylated by GSK-3. To determine if SREBP1a was phosphorylated on Ser-434, we generated a phospho-Ser-434-specific anti-SREBP1 antibody (pS434). This antibody recognized wild-type SREBP1a following expression in HEK293T cells, whereas it failed to recognize the S434A mutant (Fig. 1A). Wild-type SREBP1c was also recognized by the antibody, whereas the S410A mutant was not (Fig. 1A), demonstrating that both isoforms of SREBP1 are phosphorylated on this serine residue. In addition, the recognition of wild-type SREBP1a was competed with a peptide containing phospho-Ser-434, but not with peptides in which Thr-426 or Ser-430 was phosphorylated (supplemental Fig. S1), supporting the specificity of the antibody.
Our phosphopeptide mapping suggested that Ser-434 could be phosphorylated by GSK-3␤ in vitro. Indeed, in vitro kinase assays demonstrated that GSK-3␤ phosphorylates Ser-434 in SREBP1a (Fig. 1B). To test whether endogenous GSK-3 could phosphorylate Ser-434 in endogenous SREBP1, HeLa cells were treated with lithium, a pharmacologic GSK-3 inhibitor. The pS434 antibody only recognized the mature form of SREBP1 and not the membrane-associated precursor, suggesting that phosphorylation of Ser-434 is specific for the nuclear form of the protein. Inhibition of GSK-3 reduced the phosphorylation of Ser-434 (Fig. 1C), indicating that this residue is phosphorylated by GSK-3 in vivo. Lithium also reduced the phosphorylation of Ser-434 in HepG2 (Fig. 1D). Similar results were obtained with the GSK-3 inhibitor SB 415286 (Fig. 1E). The reduction in Ser-434 phosphorylation was specific for GSK-3 inhibitors, because inhibitors of mitogen-activated protein kinase (UO126) and cyclin-dependent kinases (roscovitine) failed to affect the phosphorylation of this residue in HeLa cells (Fig. 1F).
Interestingly, prolonged treatment of HeLa cells with lithium and SB 415286 increased the steady-state levels of mature SREBP1 ( Fig. 2A), supporting the notion that GSK-3-dependent phosphorylation of mature SREBP1 promotes its degradation (9). In support of this possibility, we found that nuclear SREBP1 phosphorylated on Ser-434 accumulated in HeLa cells treated with the proteasome inhibitor MG-132 (Fig. 2B), indicating that nuclear SREBP1 phosphorylated on this residue is rapidly turned over by proteasome-mediated degradation. Similar results were also obtained in HepG2 cells (supplemental Fig. S2). Furthermore, siRNA-mediated inactivation of GSK-3 reduced the phosphorylation of Ser-434 in endogenous SREBP1 in HeLa cells (Fig. 2C), as well as HepG2 and U2OS cells (supplemental Fig. S2), confirming that GSK-3 contributes to the phosphorylation of Ser-434 in SREBP1. siRNA-mediated inactivation of both GSK-3␣ and -␤ in HeLa cells resulted in a small reduction of the levels of mature SREBP1 (Fig. 2C). This could be explained, at least in part, by a significant reduction in the expression of the precursor form of SREBP1 in response to prolonged inactivation of GSK-3 (data not shown). Growth factors, including insulin, negatively regulate the activity of GSK-3 through Akt-mediated phosphorylation. Thus, we speculated that insulin-dependent inhibition of GSK-3 could attenuate the phosphorylation of Ser-434 in SREBP1. As illustrated in Fig. 2D, a short treatment of HepG2 cells with insulin enhanced the inhibitory phosphorylation of GSK-3. Consequently, the phosphorylation of SREBP1 on Ser-434 was reduced in response to insulin treatment, resulting in the accumulation of nuclear SREBP1. These data indicate that the phosphorylation of Ser-434 in SREBP1 requires endogenous GSK-3 activity and that these processes are regulated by insulin signaling.
Phosphorylation of Ser-434 Destabilizes SREBP1-We have previously demonstrated that the mature form of SREBP1 is destabilized following GSK-3-dependent phosphorylation of Thr-426 and Ser-430 (9). To determine if phosphorylation of Ser-434 could be involved in regulating the stability of SREBP1, cells were transfected with mature SREBP1a, either wild-type or the phosphorylation-deficient S434A mutant. The steadystate levels of the S434A mutant were enhanced compared with the wild-type protein (Fig. 3A), indicating that phosphorylation of Ser-434 destabilizes SREBP1. Similar results were obtained when the corresponding residue (Ser-410) was mutated in SREBP1c ( Fig. 3A), suggesting that phosphorylation of this residue regulates the stability of both SREBP1 isoforms. The S434A mutant of SREBP1a was also stabilized in HEK293, HeLa, and HepG2 cells (supplemental Fig. S3). Interestingly, the steady-state levels of a phosphorylation-mimetic mutant of SREBP1a (S434D) were slightly reduced compared with the wild-type protein (Fig. 3B), further supporting the notion that phosphorylation of Ser-434 destabilizes SREBP1. The mRNA levels of the constructs in Fig. 3 (A and B) were similar (supplemental Fig. S4), suggesting that the expression of these mutants was affected at the protein level. To confirm that the enhanced abundance of the S434A mutant resulted from reduced degradation, we measured the half-life of mature SREBP1a, either wild-type or the S434A mutant, in transfected HEK293T cells.
As illustrated in Fig. 3C, the turnover of the mutant protein was reduced compared with wild-type SREBP1a.
The accumulation of the S434A mutant should lead to an enhanced expression of SREBP target genes. To test this hypothesis, HepG2 cells were transfected with an SREBP-responsive promoter-reporter gene in the absence or presence of mature SREBP1a, either wild-type, S434A, or S434D. In support of our hypothesis, the transcriptional activity of the S434A mutant was enhanced, whereas the activity of the S434D mutant was reduced compared with the wild-type protein (Fig.  3D). Mutation of the corresponding residue in SREBP1c (S410) also enhanced its transcriptional activity (Fig. 3D), indicating that the phosphorylation of Ser-434 and its effect on the stability of SREBP1 influences the biological function of SREBP1. This possibility was supported when we analyzed the expression of endogenous target genes in HEK293 cells transfected with mature SREBP1a. The expression of both the LDL receptor and HMG-CoA synthase genes was higher in cells expressing the S434A mutant compared with cells expressing wildtype SREBP1a (Fig. 3E).
If our hypothesis was correct, activation of GSK-3 should induce the degradation of mature SREBP1. To test this, cells were treated with wortmannin and LY294002, two phosphatidylinositol 3-kinase (PI3K) inhibitors. Both inhibitors reduced the inhibitory phosphorylation of GSK-3 and resulted in reduced steady-state levels of mature SREBP1 (Fig. 4A). This effect was specific for the PI3K inhibitors, because the mTor inhibitor rapamycin failed to affect SREBP1 levels. Interestingly, both PI3K inhibitors failed to affect the levels of mature SREBP1 in the presence the proteasome inhibitor MG-132 ( Fig.  4B), suggesting that activation of GSK-3 results in enhanced degradation of mature SREBP1. To test this hypothesis, HepG2 cells were transfected with mature SREBP1a, either wild-type, S434A, or the double mutant T426A/S430A and treated with LY294002. As seen in Fig. 4C, LY294002 reduced the levels of wild-type SREBP1a, whereas the degradation of the S434A mutant was attenuated and the double mutant was resistant to LY294002 treatment, suggesting that phosphorylation of Ser-434 and the phosphodegron in SREBP1 is important for this effect. As expected, inhibition of PI3K in HepG2 cells attenuated the accumulation of mature SREBP1 in response to a short pulse of insulin (Fig. 4D). Again, this effect was specific for PI3K, because rapamycin only had a marginal effect, suggesting that PI3K signaling, but not mTor, is important for the acute response to insulin signaling in these cells.
Ser-434 Regulates the Phosphorylation of Thr-426 and Ser-430-Our data indicate that phosphorylation of Ser-434 promotes the degradation of SREBP1. The active form of SREBP1 is targeted for degradation following phosphorylation of Thr-426 and Ser-430. One possibility was therefore that the phosphorylation of Ser-434 could influence the subsequent phosphorylation of Thr-426 and Ser-430. To test this possibility, HEK293T cells were transfected with mature SREBP1a, either wild-type or the T426A, S430A, or S434A mutants, and the phosphorylation of Thr-426, Ser-430, and Ser-434 was determined following immunoprecipitation of the various proteins. As illustrated in Fig. 5A, all three residues were phosphorylated when wild-type SREBP1a was expressed in cells. Muta- tion of Thr-426 did not only block the phosphorylation of this residue but also significantly reduced the phosphorylation of Ser-430, whereas the phosphorylation of Ser-434 was only somewhat reduced in the T426A mutant. Mutation of Ser-430 attenuated the phosphorylation of Thr-426 and Ser-434. Importantly, mutation of Ser-434 drastically reduced the phosphorylation of Ser-430 and also resulted in a significant attenuation of the phosphorylation of Thr-426. Thus, both Thr-426 and Ser-434 regulate the phosphorylation of Ser-430, and Ser-434 regulates the phosphorylation of both Thr-426 and Ser-430. Therefore, phosphorylation of Ser-434 could potentially regulate the phosphorylation of the phosphodegron in SREBP1, thereby affecting the stability of mature SREBP1. This possibility is in agreement with our observation that the S434A mutant of SREBP1 is stabilized (Fig. 3, A-C).
The results in Fig. 5A suggest that Ser-434 regulates the GSK-3-dependent phosphorylation of Thr-426 and Ser-430. To test this possibility, we used recombinant SREBP1, either wild-type or the S434A mutant, in kinase assays with recombinant GSK-3␤. As seen in Fig. 5B, the GSK-3␤-dependent phosphorylation of Thr-426 was significantly reduced, and the phosphorylation of Ser-430 completely lost in the S434A mutant, confirming that Ser-434 plays an important role in the phosphorylation of both these residues. Interestingly, mutation of Ser-430 reduced the GSK-3-dependent phosphorylation of Thr-426 (Fig. 5C), suggesting that Ser-430 is a priming site for the phosphorylation of Thr-426. SREBP1 has a serine residue (Ser-438 in SREBP1a) four residues downstream of Ser-434 that could function as a priming site for the phosphorylation of Ser-434. However, mutation of this residue (S438A) failed to affect the phosphorylation of Ser-434, Ser-430, or Thr-426 and failed to affect the steady-state levels of transfected mature SREBP1a (supplemental Fig. S5). Taken together, our results suggest that the GSK-3-dependent phosphorylation of Thr-426 is dependent on the phosphorylation of Ser-430, which in turn is dependent on the phosphorylation of Ser-434.
We recently found that GSK-3␤ is recruited to SREBP target genes and that the GSK-3␤-mediated phosphorylation of Thr-426 and Ser-430 in SREBP1 is enhanced in response to DNA binding (12). To determine if the phosphorylation of Ser-434 is enhanced in response to DNA binding, we performed in vitro kinase assays using recombinant mature SREBP1a and HeLa nuclear extracts in the absence and presence of a promoter fragment containing SREBP binding sites. Similar to Thr-426 and Ser-430, the phosphorylation of Ser-434 was enhanced in response to DNA binding (Fig. 5D). Interestingly, the DNA binding-dependent phosphorylation of Ser-434 was attenuated in the presence of lithium (Fig. 5D), suggesting that GSK-3 contributes to the phosphorylation of Ser-434 in response to DNA binding. This effect was specific for lithium and was not seen with other kinase inhibitors tested (supplemental Fig. S6). Our earlier results suggested that Ser-434 regulates the GSK-3-dependent phosphorylation of Thr-426 and Ser-430. To test if this was also true in response to DNA binding, the kinase assays were repeated with recombinant SREBP1, either wild-type or the S434A mutant, in the absence or presence of DNA. As seen in Fig. 5E, the phosphorylation of both Thr-426 and Ser-430 in response to DNA binding was almost completely lost in the S434A mutant, confirming that Ser-434 plays an important role in the phosphorylation of both these residues.
To further define the role of Ser-434 in GSK-3-dependent phosphorylation of SREBP1, we used synthetic peptides corresponding to the region surrounding Ser-434 in SREBP1 in peptide pulldown assays with extracts from HeLa cells. GSK-3␣ and -␤ were recruited to the non-phosphorylated peptide, and the interaction was enhanced when the peptide was phosphorylated on Ser-434, whereas the interaction was unaffected in the S434A mutant peptide (Fig. 5F). Interestingly, phosphorylation of Ser-430 greatly enhanced the recruitment of GSK-3 to the peptide (Fig. 5G), indicating that the interaction between GSK-3 and SREBP1 is enhanced in response to phosphorylation of both Ser-430 and Ser-434. Taken together, our results demonstrate that Ser-434 in the mature form of SREBP1a, as well as the corresponding residue in SREBP1c, is phosphorylated by GSK-3, both in vitro and in vivo. The phosphorylation of Ser-434 regulates the stability of SREBP1 by promoting its degradation. In addition, the phosphorylation of Ser-434 is important for the subsequent phosphorylation of Thr-426 and Ser-430 in the phosphodegron in SREBP1.
Phosphorylation of Ser-434 Regulates the Fbw7-dependent Degradation of SREBP1-The interaction between Fbw7 and SREBP1 is dependent on the phosphorylation of Thr-426 and Ser-430 (9). Our data suggest that the phosphorylation of Thr-426 and Ser-430 is dependent on Ser-434, suggesting that the interaction between SREBP1 and Fbw7 could be regulated by Ser-434. To address this issue, we performed coimmunoprecipitation assays between Fbw7 and SREBP1a, either wild-type or the S434A mutant. Wild-type SREBP1 interacted with Fbw7, whereas the S434A mutant failed to interact (Fig. 6A). These results suggest that the phosphorylation of Ser-434 could control the interaction between SREBP1 and Fbw7 by regulating the phosphorylation of Thr-426 and Ser-430 in the phosphodegron. To further define the role of Ser-434 phosphorylation in the interaction between SREBP1 and Fbw7, we used synthetic peptides corresponding to the phosphodegron in SREBP1 in peptide pulldown assays with Fbw7␣. As demonstrated earlier (9), the binding of Fbw7␣ to the peptide was strongly dependent on the phosphorylation of Thr-426 and Ser-430 (supplemental Fig. S7). Interestingly, phosphorylation of Ser-434 did not affect the binding of Fbw7 to the peptide, supporting the notion that phosphorylation of Ser-434 regulates the interaction between SREBP1 and Fbw7 in vivo by controlling the phosphorylation of Thr-426 and Ser-430.
Thus, our results indicate that SREBP molecules that are not phosphorylated on Ser-434, such as the S434A mutant, would be poor substrates for Fbw7-mediated degradation in vivo. To test this hypothesis, HEK293T cells were transfected with mature SREBP1a, either wild-type or the S434A mutant, in the absence or presence of increasing amounts of Fbw7␣. In support of our hypothesis, the S434A protein was less sensitive to Fbw7␣-mediated degradation compared with the wild-type protein (Fig. 6B). In addition, transfected wild-type mature SREBP1c was stabilized in response to shRNA-mediated inactivation of Fbw7 in HepG2 cells, whereas the S410A mutant was stable under normal conditions and insensitive to inactivation of Fbw7 (Fig. 6C). These results are in agreement with our observation that the Ser-434 mutant accumulates when expressed in cells and suggest that phosphorylation of Ser-434 could be important for Fbw7-dependent degradation of mature SREBP1. We have earlier demonstrated that mature SREBP1 accumulates in Fbw7-negative cells and that the protein is highly phosphorylated on both Thr-426 and Ser-430, suggest- ing that SREBP1 molecules phosphorylated on these residues are rapidly degraded by an Fbw7-dependent mechanism under normal conditions. We found that mature SREBP1 in Fbw7negative cells was highly phosphorylated on Ser-434 (Fig. 6D), supporting the notion that phosphorylation of Ser-434 plays an important role in Fbw7-dependent degradation of SREBP1. In support of this possibility, mature SREBP1 phosphorylated on Ser-434 also accumulated in response to siRNA-mediated inactivation of Fbw7 in HeLa cells (Fig. 6E), as well as HepG2 and U2OS cells (supplemental Fig. S8).
Fbw7 Controls Insulin-dependent Regulation of Mature SREBP1-The phosphorylation of SREBP1 on Thr-426, Ser-430, and Ser-434 is reduced in response to insulin treatment (Fig. 2D) (9). Thus, we speculated that the accumulation of nuclear SREBP1 in response to insulin could be regulated by GSK3-dependent phosphorylation and proteasome-dependent degradation of mature SREBP1. To test this hypothesis, HepG2 cells were treated with insulin in the absence or presence of lithium or the proteasome inhibitor MG132. Under normal conditions, insulin treatment resulted in the accumulation of mature SREBP1 (Fig. 7A). However, inhibition of GSK3 or the proteasome resulted in the accumulation of mature SREBP1, and insulin was unable to further enhance the levels of mature SREBP1 under these conditions. To test if Fbw7 could be important for the insulin-dependent accumulation of mature SREBP1, Fbw7-positive and -negative HCT116 cells were treated in the absence or presence of insulin. As shown in Fig.  7B, insulin enhanced the levels of mature SREBP1 in Fbw7positive cells, whereas the levels of mature SREBP1 were high in Fbw7-negative cells and non-responsive to insulin treatment. Similar results were also obtained in response to siRNA-mediated inactivation of Fbw7 in HepG2 cells (Fig. 7B, right), suggesting that Fbw7 could play a role in insulin-dependent regulation of mature SREBP1. To test this possibility, HepG2 cells were transfected with wild-type mature SREBP1c in the absence or presence of Fbw7 shRNA, followed by insulin treat- The levels and phosphorylation (pT426 and pS430) of SREBP1a were determined by Western blotting. C, recombinant His 6 -SREBP1a, either wild-type or the S430A mutant, was incubated with GSK-3␤. The levels and phosphorylation (pT426) of SREBP1a were determined by Western blotting. D, recombinant His 6 -SREBP1a was incubated with HeLa nuclear extract in the absence or presence a plasmid DNA template containing the proximal region of the HMG-CoA synthase promoter and in the presence of NaCl or LiCl (20 mM). The levels and phosphorylation (pT426, pS430, and pS434) of SREBP1a were determined by Western blotting. E, recombinant His 6 -SREBP1a, either wild-type or the S434A mutant, was incubated with HeLa nuclear extract in the absence or presence of a plasmid DNA template containing the proximal region of the HMG-CoA synthase promoter. The levels and phosphorylation (pT426, pS430, and pS434) of SREBP1a were determined by Western blotting. F, WCE from HeLa cells were used in peptide pulldown assays, using three separate peptides corresponding to residues 422-442 of human SREBP1a, either unphosphorylated (Ref), the same peptide phosphorylated on Ser-434 (pS434), or the same peptide containing the S434A mutation. The bound proteins were subjected to SDS-PAGE and Western blotting using 20% of input as control. G, WCE from HeLa cells were used in peptide pulldown assays, using two separate peptides corresponding to residues 422-442 of human SREBP1a, either unphosphorylated (Ref) or the same peptide phosphorylated on Ser-430 (pS430). The bound proteins were subjected to SDS-PAGE and Western blotting using 20% of input as control. FEBRUARY 27, 2009 • VOLUME 284 • NUMBER 9 ment. As illustrated in Fig. 7C, insulin enhanced the levels of SREBP1c in the presence of control shRNA. Inactivation of Fbw7 enhanced the levels of transfected mature SREBP1c, and no further increase was observed in response to insulin treatment. To determine if phosphorylation of Ser-434/410 could regulate the accumulation of mature SREBP1 in response to insulin, HepG2 cells were transfected with mature SREBP1c, either wild-type or the S410A mutant. The levels of wild-type SREBP1c were enhanced in response to insulin treatment, whereas the levels of the S410A mutant were high in nonstimulated cells and failed to respond to insulin (Fig. 7D), suggesting that the phosphorylation of Ser-434/410 could be a target for insulin-dependent regulation of mature SREBP1.

Phosphorylation-dependent Degradation of SREBP1
Our results suggest that Fbw7 could affect insulin-dependent regulation of mature SREBP1. To determine if Fbw7 could regulate SREBP-dependent transcription, HepG2 cells were transfected with three different SREBP-dependent promoter-reporter genes and either control or Fbw7 shRNA. As expected, shRNA-mediated inactivation of Fbw7 enhanced the expression of all three promoter-reporter genes in the absence of insulin treatment (Fig. 7E). Importantly, inactivation of Fbw7 attenuated the insulin-dependent activation of all three promoter-reporters, suggesting that Fbw7 could regulate SREBP1 function in response to insulin signaling. This possibility was supported by our observation that the insulin-dependent induction of several SREBP target genes was attenuated in Fbw7-negative HCT116 cells (Fig. 7F). Again, the expression of all three target genes in the absence of insulin was enhanced in response to inactivation of Fbw7.
Taken together, our results demonstrate that Ser-434 in the C terminus of mature SREBP1 is phosphorylated by GSK-3. In addition, our results demonstrate that the phosphorylation of Ser-434 in SREBP1 is required for the GSK-3-dependent phosphorylation of Thr-426 and Ser-430 in its phosphodegron. Consequently, the phosphorylation of Ser-434 regulates the interaction between the phosphodegron in SREBP1 and the Fbw7 ubiquitin ligase, thereby controlling the degradation of mature SREBP1. The phosphorylation of Ser-434 is also enhanced in response to DNA binding, and mutation of Ser-434 blocks the phosphorylation of Thr-426 and Ser-430 during this process, suggesting that the phosphorylation of Ser-434 could control the degradation of active SREBP1 molecules. In addition, our results suggest that the ubiquitin ligase Fbw7 could play an important role in insulindependent regulation of nuclear SREBP1. Thus, the degradation of mature SREBP1 is controlled by a phosphorylation cascade, and the phosphorylation of Ser-434 could function as a molecular switch to control these processes.

DISCUSSION
The mature form of the transcription factor SREBP1 is degraded by the ubiquitin-proteasome system following phosphorylation of two residues in its C-terminal domain, Thr-426 and Ser-430 in SREBP1a (Thr-402 and Ser-406 in SREBP1c) (9,12). Phosphorylation of these residues creates a docking site for the ubiquitin ligase SCF Fbw7 , which catalyzes the polyubiquitination of mature SREBP1. Both residues are phosphorylated by GSK-3␤ in vitro, and inactivation of GSK-3 in cells attenuates the phosphorylation of both residues, suggesting that GSK-3 contributes to the phosphorylation of the SREBP1 phosphodegron in vivo. Here we identify an additional phosphorylated residue in the C-terminal domain of mature SREBP1a, Ser-434 (S410 in SREBP1c). Phosphorylation of Ser-434 in SREBP1a promotes the phosphorylation of its phosphodegron, thereby enhancing the recruitment of Fbw7 and Fbw7-dependent degradation of SREBP1. Thus, the degradation of mature SREBP1 is controlled by cross-talk between multiple phosphorylated Ser and Thr residues in its C-terminal domain. In vitro kinase assays with recombinant SREBP1a and GSK-3␤ followed by phosphopeptide mapping indicated that Ser-434 could be phosphorylated by GSK-3␤ in vitro. This notion was confirmed using phospho-S434-specific antibodies. These antibodies also recognized endogenous SREBP1, and the phosphorylation of Ser-434 was reduced following inhibition of GSK-3 with lithium. The phosphorylation of Ser-434 was also reduced following siRNA-mediated inactivation of GSK-3, providing support for a role of GSK-3 in the phosphorylation of Ser-434. GSK-3 is inactivated in response to insulin signaling (33). The amount of mature SREBP1 in HepG2 cells was enhanced in response to insulin treatment, whereas the phosphorylation of Ser-434 was reduced. These results suggest that the SREBP1 molecules that accumulate in response to insulin treatment are not phosphorylated on Ser-434 and provide indirect support for a physiological role of GSK-3 in the phosphorylation of Ser-434. Inactivation of the ubiquitin ligase Fbw7 results in the accumulation of mature SREBP1. Interestingly, insulin was unable to enhance the levels of mature SREBP1 in cells lacking Fbw7, suggesting that inactivation of GSK-3, and thereby Fbw7-dependent degradation of mature SREBP1, could be an important target for insulin signaling. Insulin is known to enhance the expression of the SREBP1c gene and to promote the cleavage of the precursor form of SREBP1 (7,34). We propose that inhibition of the phosphorylation-dependent degradation of the active transcription factor by SCF Fbw7 could be an early target of insulin signaling, thereby contributing to a feed-forward mechanism resulting in enhanced expression of SREBP target genes. Taken together, the data in this report indicate that GSK-3 is involved in the phosphorylation of Ser-434 in mature SREBP1. However, our results do not exclude the involvement of other kinases, and a more detailed analysis of the kinases targeting Ser-434 in vivo is warranted.
Treatment of cells with lithium not only attenuated the phosphorylation of Ser-434, but also enhanced the amount of nuclear SREBP1. In addition, SREBP1 molecules phosphoryla-ted upon Ser-434 accumulated in response to inhibition of the proteasome. Taken together, these results suggested that phosphorylation of Ser-434 could destabilize SREBP1. This notion was supported by the fact that the S434A mutant of mature SREBP1 was stable, whereas the corresponding phospho-mimetic S434D mutant was unstable. The degradation of mature SREBP1 is, at least in part, regulated by SCF Fbw7 (9), a ubiquitin ligase that also targets cyclin E, c-Myc, and c-Jun (19). Fbw7mediated ubiquitination of SREBP1 is dependent on the phosphorylation of Thr-426 and Ser-430 in its phosphodegron. Interestingly, mutation of Ser-434 attenuated the phosphorylation of Thr-426 and Ser-430 in vivo. Although phosphorylation of Ser-434 did not contribute directly to the interaction with Fbw7, the S434A mutant of SREBP1a interacted poorly with the ubiquitin ligase following expression in cells. Consequently, the S434A mutant was less sensitive to Fbw7-dependent degradation compared with wild-type SREBP1. Taken together, our results suggest that Ser-434 regulates the phosphorylation of the phosphodegron in SREBP1, thereby regulating its Fbw7-dependent degradation. SREBP2 is also degraded by Fbw7 in a phosphorylation-dependent manner (9). SREBP2 also has a Ser residue at the position corresponding to Ser-434 in SREBP1a (Ser-440 in SREBP2). However, Ser-440 in SREBP2 is not followed by a Pro residue, and mutation of Ser-440 results in destabilization of the protein (data not shown), indicating that the mechanisms regulating the phosphorylation and degradation of SREBP1 and -2 could differ.
As indicated above, mutation of Ser-434 attenuated the phosphorylation of Thr-426 and Ser-430 in transfected cells. Interestingly, the phosphorylation of Thr-426 and Ser-430 by recombinant GSK-3␤ in vitro was significantly reduced in the S434A mutant, suggesting that Ser-434 directly enhances GSK-3-dependent phosphorylation of Thr-426 and Ser-430. At the same time, we could demonstrate that the GSK-3-dependent phosphorylation of Thr-426 was dependent on the phosphorylation of Ser-430. Using peptide pulldown assays, we found that phosphorylation of Ser-434 enhanced the interaction between GSK-3 and a peptide containing Thr-426 and Ser-430, and that the phosphorylation of Ser-430 further enhanced this interaction. Phosphorylation of many GSK-3 substrates requires a priming phosphorylation four residues C-terminal to the GSK-3 phosphorylation site (33,35,36). We propose that the phosphorylates Ser-434 in the C terminus of SREBP1, thereby creating a binding site for GSK-3, which results in enhanced GSK-3-dependent phosphorylation of Ser-430. Phosphorylation of Ser-430 generates a high affinity binding site for GSK-3, which enhances the GSK-3-dependent phosphorylation of Thr-426. Phosphorylation of Thr-426 and Ser-430 in turn creates a binding site for the ubiquitin ligase SCF Fbw7 , resulting in the polyubiquitination and degradation of mature SREBP1. The phosphorylation of all three residues is attenuated by insulin signaling, thereby contributing to the insulin-dependent accumulation of mature SREBP1. Thus, the degradation of mature SREBP1 is controlled by cross-talk between multiple phosphorylated residues in its C-terminal domain, and the phosphorylation of Ser-434 could function as a molecular switch to control these processes. B, Fbw7 could regulate SREBP1 on different levels. Fbw7 negatively regulates mature SREBP1 by targeting it for proteasome-mediated degradation following phosphorylation of its phosphodegron. In addition, Fbw7 also targets mTor for degradation (39). mTor has been reported to promote the activation of premature SREBP1 (40). Thus, Fbw7 could be a negative regulator of both the activation of SREBP, as well as the function of the active transcription factor. phosphorylation of Ser-434 could be such a priming phosphorylation event for the subsequent phosphorylation of Ser-430. Furthermore, once Ser-430 is phosphorylated, it serves as a priming site for GSK-3-dependent phosphorylation of Thr-426. In turn, phosphorylation of Thr-426 and Ser-430 creates a high affinity docking site for Fbw7, resulting in rapid degradation of the protein (Fig. 8A). The synergistic phosphorylation of Ser-434, Ser-430, and Thr-426 sensitizes SREBP1 toward changes in GSK-3 activity. In addition, this mechanism of phosphorylation increases the probability that individual SREBP1 molecules are phosphorylated on both Ser-430 and Thr-426, thereby making these molecules more susceptible to degradation. GSK-3 substrates that require prior phosphorylation are usually targeted by a separate kinase, a so-called priming kinase (33,35,36). However, there are examples of GSK-3 substrates that do not require priming phosphorylation (33,37,38). In the case of SREBP1, our data suggest that GSK-3 could function as its own priming kinase by phosphorylating Ser-434/ Ser-410, at least in vitro. This is supported by our observation that mutation of Ser-438 in SREBP1a failed to affect the phosphorylation of Ser-434, Ser-430, and Thr-426, and failed to affect the stability of mature SREBP1a. However, further studies of the kinases involved in these processes in response to different signaling pathways are warranted.
Taken together, our results demonstrate that Fbw7 regulates the stability of active SREBP1. However, it is possible that Fbw7 regulates SREBP1 in other ways as well. It was recently demonstrated that Fbw7 targets mTor for degradation (39). Interestingly, it has been suggested that mTor is a positive regulator of the activation of premature SREBP1 (40). Thus, Fbw7 could regulate both the mature and premature forms of SREBP1 (Fig.  8B). However, inhibition of mTor by rapamycin failed to affect the insulin-dependent regulation of SREBP1 in HepG2 cells. We only treated cells with insulin for short times in our experiments, suggesting that mTor does not play an important role in the acute response to activation of the insulin signaling pathway. However, the role of mTor, as well as the Fbw7-dependent degradation of mTor, in the regulation of SREBP1 will be an important subject for future studies.