S6K1-mediated phosphorylation of PDK1 impairs AKT kinase activity and oncogenic functions

Abstract

Functioning as a master kinase, 3-phosphoinositide-dependent protein kinase 1 (PDK1) plays a fundamental role in phosphorylating and activating protein kinases A, B and C (AGC) family kinases, including AKT. However, upstream regulation of PDK1 remains largely elusive. Here we report that ribosomal protein S6 kinase beta 1 (S6K1), a member of AGC kinases and downstream target of mechanistic target of rapamycin complex 1 (mTORC1), directly phosphorylates PDK1 at its pleckstrin homology (PH) domain, and impairs PDK1 interaction with and activation of AKT. Mechanistically, S6K1-mediated phosphorylation of PDK1 augments its interaction with 14-3-3 adaptor protein and homo-dimerization, subsequently dissociating PDK1 from phosphatidylinositol 3,4,5 triphosphate (PIP₃) and retarding its interaction with AKT. Pathologically, tumor patient-associated PDK1 mutations, either attenuating S6K1-mediated PDK1 phosphorylation or impairing PDK1 interaction with 14-3-3, result in elevated AKT kinase activity and oncogenic functions. Taken together, our findings not only unravel a delicate feedback regulation of AKT signaling via S6K1-mediated PDK1 phosphorylation, but also highlight the potential strategy to combat mutant PDK1-driven cancers.

Fig. 1. S6K directly phosphorylates PDK1 on Ser549.

a A schematic presentation of the evolutionarily conserved putative AGC kinase phosphorylation consensus in PDK1. b IB analysis of WCL and immunoprecipitates (IP) derived from HEK293T cells transfected with Flag-PDK1 and the indicated HA-tagged constitutive active AGC family kinases. c DLD1 cells were serum-starved for 24 h and then collected after treated with Rapamycin (20 nM), S6K1-I (10 μM), LY294002 (10 μM), BKM120 (10 μM) for 1 h before being stimulated with insulin (100 nM) for 30 min. The resulting cells were subjected for IP and IB analysis. Indicated protein were quantified. (mean ± SD, n = 3), **P < 0.01. d 293 T cells were transfected with increasing dose of HA-S6K-R3A and subjected for IP and IB analysis. e IB analysis of WCL and IP products derived from 293 T cells transfected with Flag-PDK1 and HA-S6K1-R3A, where indicated, cells were treated with Rapamycin (20 nM) and S6K1-I (10 μM) for 1 hr. f IB analysis of WCL and IP products derived from S6k1/2^−/− MEFs transfected with indicated constructed, and were serum-starved for 24 h and then collected after insulin (100 nM) stimulation for 30 min. g, h IB analysis of IP products and WCL derived from DLD1 cells infected with shRNA against S6K1 g or 293 T cells transfected with indicated constructs h. Indicated protein were quantified. (mean ± SD, n = 3, P = 0.001, 0.435, 0.0009, 0.157). i In vitro kinase assays were performed with recombinant bacterially purified PDK1 protein as substrates and IP S6K1 protein from 293 T cells was used as the source of kinase. ³²P isotope-ATP was used to detect the phosphorylated PDK1. Similar results were obtained in n ≥ 3 independent experiments in b, f, i. Statistical significance was determined by two-tailed Student’s t-test in c, g, h. N.S > 0.05, *P < 0.05,**P < 0.01. Source data are provided in Source Data files. EV, empty vector. WCL, whole cell lysate. IP, immunoprecipitation. Scr, scramble. WT, wild type.

Fig. 2. Phosphorylation of PDK1 inhibits AKT kinase activity and oncogenic functions.

a IB analysis of DLD1-PDK1 knockout cells stably infected with the indicated constructs. Indicated protein were quantified. (mean ± SD, n = 3, P = 0.004, 0.003). b IB analysis of WCL derived from DLD1-PDK1 knockout cells transfected with the indicated Flag-PDK1 constructs that were serum-starved for 12 h and then treated with the insulin (100 nM) for the indicated time periods before collection for IB analysis. c pT308-AKT/AKT1 ratio in b were calculated. d, e Cells generated in a were subjected to colony formation and soft agar assays. (mean ± SD, n = 3, P = 0.003, 0.003, 0.005, 0.002). f, g, h Cells generated in a were subjected to mouse xenograft assays. Tumor sizes were monitored (f P = 0.0002), and dissected tumors were weighed and calculated (g, h, P = 0.0042, 0.0059). The Error bars in the f, h are mean plus or minus SD, n = 5 mice. *P < 0.05, **P < 0.01. i IB analysis of WCL derived from dissected tumor tissues. j pT308-AKT/AKT1 ratio in (i) were calculated. (mean ± SD, n = 3, P = 0.0033, 0.0018), k, l Graphic representation of pS6 staining derived from tumor tissues k, which were further normalized and quantified l. (mean ± SD, n = 5, P = 6.41E-06, 2.29E-05) (t test), scale bar, 50 μm. Similar results were obtained in n ≥ 3 independent experiments in b. Statistical significance was determined by two-tailed Student’s t-test in a, e, h. j, l and by two-way ANOVA in f. *P < 0.05,**P < 0.01. Source data are provided in Source Data files. EV, empty vector. WT, wild type.

Fig. 3. S6K1-mediated phosphorylation of PDK1 inhibits PDK1 membrane localization and PIP3 binding.

a The indicated Flag-PDK1 constructs with/without HA-S6K1-R3A were transfected into 293 T cells and anti-Flag immunoprecipitation was recovered as the kinase source to phosphorylate insect cells purified His-AKT1 in vitro. pT308-AKT/AKT1 ratio were calculated. (mean ± SD, n = 3). b IB analysis of WCL and GST-pulldown derived from 293 T cells transfected with GST-PDK1, Flag-AKT1 and increasing dose of HA-S6K1-R3A. pS549-PDK1/PDK1 ration were calculated. (mean ± SD, n = 3, P = 0.0001, 0.0002, 7.82E-10). c IB analysis of WCL and IP products derived from 293 T cells transfected with Flag-PDK1(WT, S549A, S549D) and HA-AKT1. Indicated protein were quantified. (mean ± SD, n = 3, P = 0.005, 0.004). d IB analysis of PIP₃ pull-down products and WCL derived from 293 T cells transfected with the indicated constructs. Where indicated, empty beads (EV) serve as a negative control. e IB analysis of PIP₃ pull-down products and WCL derived from HeLa cells transfected with Flag-PDK1 that were serum-starved for 24 h and then collected after insulin (100 nM) stimulation for 30 min, where indicated, the kinase inhibitors were added. f IB analysis of cell fractionations separated from 293 T cells transfected with indicated constructs. PDK1/Tubulin or AIF ratio were calculated. (mean ± SD, n = 3, P = 2.98E-06, 0.002). g Representative immunofluorescence images of 293 T cells transfected with indicated constructs, scale bar, 10 μm. Mean PDK1 fluorescence intensity at plasma membrane relative cytosol was determined, data represent mean ± SD, P = 0.008, 0.03. Greater than 60 cells were analyzed from 3 independent experiments. Similar results were obtained in n ≥ 3 independent experiments in d, e. Statistical significance was determined by two-tailed Student’s t-test in b, c, f. g N.S > 0.05, *P < 0.05,**P < 0.01. Source data are provided in Source Data files. EV, empty vector. WCL, whole cell lysate. IP, immunoprecipitation. WT, wild type. PD, pulldown. Cyto, cytoplasm. Memb, membrane.

Fig. 4. 14-3-3 binds phosphorylated PDK1 and dissociates it from the cytoplasm membrane.

a IB analysis of WCL and IP products derived from 293 T cells transfected with Flag-PDK1 and the indicated various HA-tagged 14-3-3 constructs. b, c DLD1 cell lysates were subjected to IP with control IgG, anti-PDK1 or 14-3-3γ antibodies for IB analysis. d A schematic presentation of the evolutionarily conserved putative AGC kinase phosphorylation consensus in PDK1 and 14-3-3 recognized consensus. e IB analysis of WCL and IP products derived from 293 T cells transfected with indicated constructs. f IB analysis of WCL and IP products derived from 293 T cells transfected with HA-14-3-3γ and indicated constructs. g IB analysis of PIP₃ pull-down products and WCL derived from 293 T cells transfected with the indicated constructs. h IB analysis of WCL and GST-pulldown derived from 293 T cells transfected with the indicated constructs. i IB analysis of WCL derived from 293 T cells transfected with the indicated Flag-PDK1 constructs that were serum-starved for 12 h and then treated with insulin (100 nM) for the indicated time periods before collection for IB analysis. pT308-AKT/AKT1 ratio were calculated. (mean ± SD, n = 3, P = 0.0003). j IB analysis of the indicated fractionations derived from the gel filtration experiment with 293 T cells transfected with Flag-PDK1. k IB analysis of WCL and GST-pull-down derived from 293 T control or 14-3-3γ knockdown cells transfected with GST-PDK1 and Flag-PDK1. l IB analysis of WCL derived from 293 T control or 14-3-3γ knockdown cells. m IB analysis of WCL derived from 293 T control or 14-3-3γ knockdown cells that were serum-starved for 12 h and then treated with insulin (100 nM) for the indicated time periods before collection for IB analysis. pT308-AKT/AKT1 ratio were calculated. (mean ± SD, n = 3, P = 0.0001). n Proposed model for 14-3-3 involved in the PDK1/AKT pathway regulation. Red arrows indicate positive regulation, and blue arrows indicate negative regulation. Similar results were obtained in n ≥ 3 independent experiments in a, b, c, e, f, g, h. Statistical significance was determined by two-way ANOVA in i, m. **P < 0.01. Source data are provided in Source Data files. EV, empty vector. WCL, whole cell lysate. IP, immunoprecipitation. WT, wild type. PD, pulldown. Scr, scramble.

Fig. 5. Patients-associated mutations of PDK1 confer to PDK1 membrane location and AKT kinase activation.

a A schematic presentation of the patients-associated PDK1 mutations occurred around S6K1-mediated PDK1 phosphorylation or 14-3-3γ binding region. b IB analysis of WCL and IP products derived from 293 T cells transfected with the indicated constructs. c IB analysis of WCL and GST-pull-down derived from 293 T cells transfected with GST-14-3-3γ and the indicated constructs. d IB analysis of PIP₃ pull-down products and WCL derived from 293 T cells transfected with Flag-PDK1(WT, R544K, R546P, S549N, P551A, P551L). e IB analysis of cell fractionations separated from 293 T cells transfected with the indicated constructs. f PDK1/Tubulin or AIF ratio in e were calculated, (mean ± SD, n = 3), *P < 0.05. g Representative immunofluorescence images of 293 T cells transfected with the indicated constructs, scale bar, 10 μm. h Mean PDK1 fluorescence intensity at plasma membrane relative cytosol was determined, data represent mean ± SD, P = 0.022, 0.036, 0.011, 0.035, 0.026, 0.030. Greater than 60 cells were analyzed from 3 independent experiments. i IB analysis of WCL and IP products derived from 293 T cells transfected with HA-AKT1 and the indicated construct. j AKT1/PDK1 ratio in i were calculated, (mean ± SD, n = 3), *P < 0.05. k IB analysis of WCL derived from 293T-PDK1 knockout cells transfected with the indicated constructs. l pT308-AKT/AKT1 ratio in k were calculated, (mean ± SD, n = 3), *P < 0.05, **P < 0.01. Similar results were obtained in n ≥ 3 independent experiments in b, c, d. Statistical significance was determined by two-tailed Student’s t-test in f, h, j, l. *P < 0.05, **P < 0.01. Source data are provided in Source Data files. EV, empty vector. WCL, whole cell lysate. IP, immunoprecipitation. WT, wild type. PD, pulldown. Cyto, cytoplasm. Memb, membrane.

Fig. 6. Patient-derived mutations of PDK1 promote AKT1 activation and oncogenic functions.

a IB analysis of DLD1-PDK1 knockout cells stably infected with the indicated constructs. pT308-AKT/AKT1 ratio were calculated. (mean ± SD, n = 3), *P < 0.05, **P < 0.01. b, c Cells generated in a were subjected to colony formation and soft agar assays. (mean ± SD, n = 3) **P < 0.01. d, e Cells generated in a were subjected to mouse xenograft assays. Tumor sizes were monitored (d, P = 0.0045), and dissected tumors were weighed (e, P = 0.013, 0.008, 0.002, 0.003, 0.0003, 0.003). The Error bars in the d, e are mean plus or minus SD, n = 5 mice. *P < 0.05, **P < 0.01. f IB analysis of WCL derived from dissected tumor tissues. g pT308-AKT/AKT1 ratio in f were calculated, (mean ± SD, n = 3), *P < 0.05, **P < 0.01. Statistical significance was determined by two-tailed Student’s t-test in a, c, e, g and two-way ANOVA in d. *P < 0.05, **P < 0.01. Source data are provided in Source Data files. EV, empty vector. WCL, whole cell lysate. WT, wild type.

Fig. 7. Models depicting the negative regulation of PDK1 by S6K1 under physiological and pathological conditions.

a The canonical activation of PI3K-AKT-mTOR pathway toward growth factors, such as Insulin or EGF treatment. b The negative feedback regulations of AKT kinase by S6K1-mediated phosphorylation of SIN1, PDK1 as well as IRS1. Among which, phosphorylation of IRS1 decreases PI3K-mediated PIP₃ generation; phosphorylation of SIN1 dissociates mTORC2 from membrane location; phosphorylation of PDK1 recruits 14-3-3 and dissociates from membrane location. These pathways together tightly negatively control growth factors-induced constant activation of AKT kinase. c Patient-associated PDK1 mutations could block S6K-mediated PDK1 phosphorylation (Type I mutations) or 14-3-3-mediated PDK1 membrane dissociation (Type II mutations), leading to AKT constitutive activation and oncogenic roles, such as accelerating cell proliferation and anti-apoptosis. Red and black arrows indicate positive and negative regulation, respectively. Solid and dotted lines indicate direct and indirect (multistep) regulation, respectively.