HSF1 critically attunes proteotoxic stress sensing by mTORC1 to combat stress and promote growth

Abstract

To cope with proteotoxic stress, cells attenuate protein synthesis. However, the precise mechanisms underlying this fundamental adaptation remain poorly defined. Here we report that mTORC1 acts as an immediate cellular sensor of proteotoxic stress. Surprisingly, the multifaceted stress-responsive kinase JNK constitutively associates with mTORC1 under normal growth conditions. On activation by proteotoxic stress, JNK phosphorylates both RAPTOR at S863 and mTOR at S567, causing partial disintegration of mTORC1 and subsequent translation inhibition. Importantly, HSF1, the central player in the proteotoxic stress response (PSR), preserves mTORC1 integrity and function by inactivating JNK, independently of its canonical transcriptional action. Thereby, HSF1 translationally augments the PSR. Beyond promoting stress resistance, this intricate HSF1-JNK-mTORC1 interplay, strikingly, regulates cell, organ and body sizes. Thus, these results illuminate a unifying mechanism that controls stress adaptation and growth.

Figure 1. Proteotoxic stress activates JNK signaling but suppresses mTORC1 activity

(a) Profiling HS-induced alterations in major signal transduction pathways. Following HS at 45°C for 30 min, lysates of HEK293T cells were incubated with phospho-antibody arrays (left). Results are presented as fold changes (mean of 2 technical replicates from one experiment). (b)-(d) HEK293T cells were heat shocked at 45°C for 30 min (b) or treated with other stressors (500 nM MG132, 10 μM tubastatin, 5 mM azetidine, 40 μM VER155008, or 20 μM Pifithrin-μ) for 6 hr (c). DMSO was a solvent control. Fold changes in protein phosphorylation were presented as a heat map (d). (e) and (f) HEK293T cells were treated with 500 nM MG132 for indicated times, with or without pre-treatment with 3 μM JNK-IN-8 for 60 min (mean of 3 wells of cells per time point per experiment, and this experiment was repeated twice). p-JNK and p-S6K proteins were quantitated by sandwich ELISA (Cell Signaling Technology). (g) HEK293T cells were co-transfected with AP1-secreted embryonic alkaline phosphatase (SEAP) and CMV-Gaussia luciferase (GLuc) reporter plasmids. After 24 hr cells were treated with stressors as described in (b) and (c). Reporter activities were measured 24 hr later and SEAP activities were normalized against GLuc activities (mean of 5 wells of cells per group per experiment, and this experiment was repeated twice). (h) HEK293T cells were treated with 500 nM MG132 for indicated time and endogenous RAPTOR-mTOR interactions were examined by coIP. WCL: whole cell lysate. (i) After treatment with 500 nM MG132 or 1 μM AZD8055 for 4 hr, mTOR complexes were precipitated from HEK293T cells. mTORC1 kinase activities were measured in vitro using recombinant human His-EIF4EBP1 proteins as the substrate. Phosphorylation of 4EBP1 was detected by immunoblotting. Uncropped images of blots are shown in Supplementary Fig. 8. Source data for Fig. 1a, e, f, g can be found in Supplementary Table 1.

Figure 2. JNK negatively regulates mTORC1, translation, and cell size

(a) HEK293T cells were stably transduced with lentiviral scramble or two combinations of JNK1/2-targeting shRNAs, A (shJNK1_2 and shJNK2_1) and B (shJNK1_4 and shJNK2_2). Transduced cells were treated with DMSO or 500 nM MG132 for 4 hr. (b) Three independent lines of primary MEFs were prepared for each genotype. (c) and (d) Primary Jnk1^-/- or Jnk2^-/- MEFs were treated with 200 nM MG132 for 6 hr. While JNK1 antibodies only recognized the p46 isoform, JNK2 antibodies recognized both p54 and p46 isoforms. (e) HEK293T cells were transfected with indicated plasmids for 48 hr. The JNK1^CA plasmid encodes a fusion protein between MKK7 and JNK1A1. (f) and (g) Following transfection with indicated siRNAs or plasmids, HEK293T cells were labeled with 6-FAM-dc-puromycin for 30 min and analyzed by flow cytometry. (h)-(j) Sizes of transfected HeLa cells (h and i) and primary Jnk-deficient MEFs (j) were measured by a Multisizer™ 3 Coulter Counter. JNK manipulation causes statistically significant changes in cell size distribution (Kolmogorov-Smirnov test, p<0.001). (k) Freshly prepared single liver cell suspensions were immediately analyzed by flow cytometry. Results were normalized against wild-type controls (mean±SD, n=3 or 4 mice per genotype, One-way ANOVA). (l) mTORC1 signaling in livers of 6-week-old male mice was immunoblotted. (m) Two hours before harvesting tissues, mice were i.p. injected with 1mg puromycin. The levels of puromycin-labeled proteins in livers were quantitated by ELISA (mean±SD, n=3 mice per genotype, unpaired two-tailed Student's t test). (n)-(p) Whole-body weight and composition were measured in 6-week-old female mice (mean±SD, n=5, 6, or 9 mice per genotype, One-way ANOVA). Statistics source data for 2k, m can be found in Supplementary Table 1.

Figure 3. JNK physically associates with mTORC1

(a) Following IP of endogenous JNK1/2 from HEK293T cells treated with and without 500 nM MG132 for 4 hr, mTORC1 components were immunoblotted. (b) Following IP of endogenous RICTOR and RAPTOR from HEK293T cells, mTOR and JNK1 proteins were immunoblotted. (c) and (d) Endogenous JNK1-mTOR and JNK1-RAPTOR interactions were detected by PLA in HeLa cells with and without 500 nM MG132 treatment for 4 hr. Mouse anti-JNK1 (JM2671) and rabbit anti-mTOR (7C10) were used for JNK1-mTOR PLA. Rabbit anti-JNK1/3 (C-17) and mouse anti-RAPTOR (1H6.2) were used for JNK1-RAPTOR PLA. Scale bars: 10 μm. Images are representative of at least 3 independent experiments. (e) HEK293T cells were deprived of glucose for 4 hr. Following IP of endogenous RAPTOR, AMPKα and JNK1 were immunoblotted. (f) Following co-transfection of FLAG-mTOR, FLAG-RAPTOR, and FLAG-JNK1 plasmids into HEK293T cells, RAPTOR was precipitated using anti-RAPTOR antibodies and co-precipitated mTOR and JNK1 were immunoblotted under both no-reducing and reducing conditions using anti-FLAG antibodies. FLAG signals were quantitated by ImageJ software (mean±SD, n=3 independent experiments). β-ME: 2-mercaptoethanol. HC: heavy chain. (g) HEK293T cells stably expressing RAPTOR-targeting shRNAs were co-transfected with indicated plasmids. Following IP of HA-tagged RAPTOR, total phosphorylation of RAPTOR was immunoblotted using anti-phosphoserine/threonine antibodies. RAPTOR Ser863 phosphorylation was directly immunoblotted using phospho-specific antibodies. (h) HEK293T cells stably expressing mTOR-targeting shRNAs were co-transfected with indicated plasmids. Following IP of Myc-tagged mTOR, total and Ser567-specific phosphorylation of mTOR were immunoblotted using anti-phosphoserine/threonine and anti-phosphoSer567 antibodies, respectively. (i) and (j) Exogenously expressed RAPTOR or mTOR was precipitated from HEK293T cells, and further incubated with 100 ng recombinant GST or GST-JNK1 proteins in vitro. Total and serine-specific phosphorylation of precipitated RAPTOR and mTOR were immunoblotted. (k) and (l) HEK293T cells depleted of endogenous RAPTOR or mTOR due to stable shRNA expression were transfected with indicated plasmids and treated with DMSO or 500 nM MG132 for 4 hr. Following IP of exogenously expressed RAPTOR or mTOR, serine-specific phosphorylation and RAPTOR-mTOR interactions were immunoblotted. Activities of precipitated mTORC1 were measured in vitro using recombinant His-EIF4EBP1 proteins. (m) Schematic depiction of proposed JNK-mTORC1 interactions.

Figure 4. HSF1 maintains mTORC1 activity and integrity through inactivation and sequestration of JNK

(a) Immortalized Rosa26-CreER^T2; Hsf1^fl/fl MEFs were incubated with 1 μM 4-OHT for 3 days to deplete Hsf1. Following treatments with DMSO (DO), 200 nM MG132 (MG), 10 μM tubastatin (Tub), 2.5 mM azetidine (AZE), or 200 nM 17-DMAG (DG) for 8 hr, JNK and mTORC1 signaling were immunoblotted. EtOH: ethanol; 4-OHT: 4-hydroxytamoxifen. (b) and (c) JNK and mTORC1 signaling were detected in Hsf1^+/+ and Hsf1^-/- mouse livers and brains, 2 mice per genotype. (d) HEK293T cells stably expressing JNK1/2-targeting shRNAs were further transfected with either scramble or HSF1-targeting shRNAs. JNK and mTORC1 signaling were immunoblotted. (e) Following treatment with 3μM JNK-IN-8 for 60 min, endogenous JNK1-RAPTOR-mTOR interactions were detected by coIP in HEK293T cells transfected with either scramble or two independent HSF1-targeting shRNAs (hA9 and hA6). (f) Following IP of endogenous JNK1 or RAPTOR proteins from HEK293T cells, co-precipitated RAPTOR, JNK1, or HSF1 proteins were immunoblotted.

Figure 5. HSF1 suppresses JNK and activates mTORC1, independently of its transcriptional action

(a) HEK293T cells were transfected with either scramble or HSF1-targetting (hA6) shRNAs for 4 days. Prior to immunoblotting, these transfected HEK293T cells were treated with either DMSO (DO) for 12 hr, 3 μM JNK-IN-8 (JIN8) for 60 min, 200 nM 17-DMAG (DG) for 12 hr, or combined JNK-IN-8 (JIN8) and 17-DMAG (DG) for 12 hr. (b) Immortalized Rosa26-CreERT2; Hsf1^fl/fl MEFs were incubated with 1 μM 4-OHT for 3 days to deplete Hsf1. These MEFs were treated with various inhibitors as described in (a). (c) Transcriptional activities of HSF1^WT, HSF1^1-379, and HSF1^AD were measured in HEK293T cells co-transfected with heat shock element (HSE)-SEAP and CMV-GLuc reporter plasmids (mean of6 wells of cells per group per experiment, and this experiment was repeated twice). (d) JNK activities were measured in HEK293T cells using the dual AP1-SEAP and CMV-GLuc reporter system following transfection with HSF1^WT, HSF1^1-379, and HSF1^AD plasmids (meanof 6 wells of cells per group per experiment, and this experiment was repeated twice). JNK1^CA served as a positive control. (e) HEK293T cells were transfected with LacZ, FLAG-HSF1^WT, FLAG-HSF1^1-379, or FLAG-HSF1^AD plasmid and treated with and without 500 nM MG132 for 4 hr. JNK and mTORC1 signaling were immunoblotted. (f) HEK293T cells transfected with indicated plasmids were treated with and without 200 nM 17-DMAG for 12 hr prior to immunoblotting. Statistics source data for 5c, d can be found in Supplementary Table 1.

Figure 6. mTORC1 translationally augments the PSR and promotes resistance to proteotoxic stress

(a) Primary Tsc1^fl/fl MEFs were transduced with adenoviral GFP or Cre particles. Protein levels of HSF1 and HSPs were detected by immunoblotting. (b) Following transduction with adenoviral GFP or Cre particles, primary Tsc1^fl/fl MEFs were labeled with 100 nM biotin-dc-puromycin for 3 hr, and lysates were incubated in ELISA plates coated with either normal IgG or anti-HSF1 antibodies (mean of3 wells of cells per group per experiment, and this experiment was repeated twice). HEK293T cells transfected with HSF1 served as a positive control. (c) Following transduction with adenoviral GFP or Cre particles, primary Tsc1^fl/fl MEFs were treated with and without MG132 overnight. Viable cells were quantitated using CellTiter-Blue^® reagents (mean of6 wells of cells per group per experiment, and this experiment was repeated twice). (d) and (e) Three days after transduction with adenoviral GFP or Cre particles, primary Tsc1^fl/fl MEFs were treated with and without 200 nM MG132 for 4 hr. Total and polysome-associated RNAs were extracted, and mRNA levels of Hsf1 and Hsps were quantitated by qRT-PCR (mean of 3 dishes of cells per group per experiment, and these experiments were repeated three times). βActin served as the internal control. (f) HSF1 and HSP levels were detected by immunoblotting in HEK293T cells stably expressing RAPTOR- or mTOR-targeting shRNAs. (g) Hsf1^+/+ MEFs were treated with 200 nM MG132 and/or 1 μM AZD8055 overnight. HSP, Lys48-specific ubiquitination, and cleaved caspase 3 were immunoblotted. Source data for Fig. 6b, c, d, e, can be found in Supplementary Table 1.

Figure 7. HSF1 positively regulates cell, organ, and body sizes through suppression of JNK

(a) and (b) Following transfection of indicated plasmids or co-transfection of indicated shRNAs and siRNAs, sizes of HEK293T cells were measured by a Multisizer™ 3 Coulter Counter. Changes in cell size distribution are statistically significant (Kolmogorov-Smirnov test, p<0.001). (c) and (d) Single liver cell suspensions were freshly prepared from 6-week-old male mice and cell sizes were measured as described in Fig. 2l (mean±SD, n= 4 or 5 mice per genotype, One-way ANOVA). (e) and (f) Livers of 6-week-old male mice were weighed (mean±SD, n=9, 12, or 13 mice per genotype, One-way ANOVA). (g) JNK and mTORC1 signaling were detected by immunoblotting in the same liver tissues described in (f). (h) and (i) Brains of 6-week-old male mice were weighed (mean±SD, n=4 mice per genotype, One-way ANOVA). (j) and (k) Body weights of 6-week-old male mice were measured (mean±SD, n=9, 20, 23, or 24 mice per genotype, One-way ANOVA). (l) and (m) Whole-body composition of 6-week-old male mice was measured (mean±SD, n=9, 10, or 13 mice per genotype, One-way ANOVA). Statistics source data for Fig. 7d, i can be found in Supplementary Table 1.

Figure 8. HSF1-JNK interactions regulate liver growth and proliferation

(a) and (b) Liver weights of 6-week-old male mice (mean±SD, n=7, 14, 15, or 25 mice per genotype, One-way ANOVA). (c) NMR-measured liver lean mass of 6-week-old male mice (mean±SD, n=5, 12, 14, or 21 mice per genotype, One-way ANOVA). (d) Size was measured using single liver cell suspensions from 6-week-old male mice as in Fig. 2l (mean±SEM, n=4 mice per genotype, One-way ANOVA). (e) Livers of 6-week-old male mice were immunoblotted. (f) Liver translation rates were measured in 6-week-old male mice as in Fig. 2n (mean, n= 2 mice per genotype, One-way ANOVA). (g) Following i.p. injection of 2 mg/mouse BrdU for 2 hr, single liver cell suspensions were co-stained with propidium iodide and rat anti-BrdU antibody (mean±SD, n=4 mice per genotype, One-way ANOVA). (h) Caspase 3 activity was quantitated in 6-week-old male mouse livers using DEVD-R110 (mean±SD, n= 3 mice per genotype, One-way ANOVA). (i) Immortalized Hsf1^-/- MEFs were co-cultured with primary Jnk1 MEFs (1:1) for 48hr, and co-stained with rabbit anti-Ki-67 and mouse anti-HSF1 antibodies. Ki-67 levels of HSF1-negative cells were compared (mean±SD, n=3 independent experiments, One-way ANOVA). (j) mRNAs in 6-week-old male mouse livers were quantitated by qRT-PCR, 2 mice per genotype. Fold changes were presented as a heat map. (k) Endogenous HSF1 DNA-binding was detected by PLA as described previously and quantitated by flow cytometry (mean±SD, n=3 independent experiments, One-way ANOVA). (l) Following shRNA transfection, HGF mRNAs were quantitated in HEK293T cells stably expressing JNK1/2-targeting shRNAs (mean, n=3 wells of cells per group per experiment, repeated three times). (m) HSF1 binding to HGF promoter was quantitated by chromatin IP as described previously in HEK293T cells with stable JNK1/2 knockdown (mean, n=3 dishes of cells per group per experiment, repeated twice). (n) Liver cryosections were co-immunostained with anti-p-HSF1 Ser326, anti-p-c-MET Y1234/1235, and anti-Cre antibodies. Scale bars: 10 μm. Images are representative of 3 independent experiments. (o) Schematic depiction of the non-cell-autonomous interaction between Hsf1-deficent hepatocytes and non-parenchymal cells with activated HSF1. Statistics source data for Fig. 8d, f-i, k-m can be found in Supplementary Table 1.