EPRS is a critical mTORC1-S6K1 effector that influences adiposity in mice

Abstract

Metabolic pathways that contribute to adiposity and ageing are activated by the mammalian target of rapamycin complex 1 (mTORC1) and p70 ribosomal protein S6 kinase 1 (S6K1) axis. However, known mTORC1-S6K1 targets do not account for observed loss-of-function phenotypes, suggesting that there are additional downstream effectors of this pathway. Here we identify glutamyl-prolyl-tRNA synthetase (EPRS) as an mTORC1-S6K1 target that contributes to adiposity and ageing. Phosphorylation of EPRS at Ser999 by mTORC1-S6K1 induces its release from the aminoacyl tRNA multisynthetase complex, which is required for execution of noncanonical functions of EPRS beyond protein synthesis. To investigate the physiological function of EPRS phosphorylation, we generated Eprs knock-in mice bearing phospho-deficient Ser999-to-Ala (S999A) and phospho-mimetic (S999D) mutations. Homozygous S999A mice exhibited low body weight, reduced adipose tissue mass, and increased lifespan, similar to S6K1-deficient mice and mice with adipocyte-specific deficiency of raptor, an mTORC1 constituent. Substitution of the Eprs^S999D allele in S6K1-deficient mice normalized body mass and adiposity, indicating that EPRS phosphorylation mediates S6K1-dependent metabolic responses. In adipocytes, insulin stimulated S6K1-dependent EPRS phosphorylation and release from the multisynthetase complex. Interaction screening revealed that phospho-EPRS binds SLC27A1 (that is, fatty acid transport protein 1, FATP1), inducing its translocation to the plasma membrane and long-chain fatty acid uptake. Thus, EPRS and FATP1 are terminal mTORC1-S6K1 axis effectors that are critical for metabolic phenotypes.

Extended Data Figure 1. Identification of S6K1 as EPRS Ser⁹⁹⁹ kinase

a, Screening of AGC kinase group members for phosphorylation of EPRS Ser⁹⁹⁹ by immunocomplex kinase assay and [γ-³²P]ATP-labeling with S886A linker target in U937 cells. Kinase activity using kinase-specific substrate is shown (bottom; mean ± SEM, n = 3). b, Specificity of S6K1 for Ser⁹⁹⁹ phosphorylation, determined by ³²P incorporation in wild-type (WT), S999A and S886 linker. c, siRNA targeting S6K1 inhibits IFN-γ-stimulated EPRS phosphorylation in U937 cells determined by ³²P-labeling (mean ± SEM, n = 3). d, Active recombinant kinases used for in vitro phosphorylation of linker bearing S886A (Ser⁹⁹⁹ P-acceptor) mutation shows site-specific phosphorylation by S6K1. e, Raptor not rictor is required for Ser⁹⁹⁹ phosphorylation. f, siRNA targeting the S6K1 3’UTR inhibits S6K1 expression and phosphorylation of EPRS Ser⁹⁹⁹, but not Ser⁸⁸⁶ (mean ± SEM, n = 3). g, Phosphorylation of S6K1 Thr³⁸⁹ is required for phosphorylation of EPRS. Cells were co-transfected with siRNA targeting the 3’UTR to knock down endogenous S6K1, and with myc-tagged wild-type or mutant S6K1 cDNA; IFN-γ-stimulated EPRS phosphorylation determined by ³²P-labeling (mean ± SEM, n = 3). h, Cells treated as in (e) but followed by reciprocal co-immunoprecipitation.

Extended Data Figure 2. Gene targeting and generation of EPRS^A/A and EPRS^D/D knock-in mice, and their body weight phenotypes

a, Knock-in mice bearing Ser⁹⁹⁹-to-Ala and -Asp mutations were generated by homologous recombination. b, Validation of wild-type (EPRS^S/S) and EPRS knock-in mice by PCR genotyping (top) and sequence analysis (bottom). c, Genotype analysis of littermates. Total of 1410 and 644 progeny from interbreeding EPRS^S/A and EPRS^S/D mice, respectively, were used at postnatal days P0, P4, and P21. d, Growth curves of EPRS^A/A (mean ± SEM, n = 10/group; P < 0.0001, 2-way ANOVA) and EPRS^D/D (n = 10/group) female mice. e, Representative images (left) and weights (right) of wild-type (EPRS^S/S; S/S) and EPRS^A/A (A/A) mice on embryonic day E16.5 and post-embryonic development stages. Data shown are mean ± SEM; n = 11 for E16.5 embryos, n = 10 for P0 and P10 mice, and n = 14 for 3, 12, 20, 30, and 50-week mice. f, Representative images (left) and weights (right) of 50-week S/S and EPRS^D/D (D/D) mice (mean ± SEM; n = 10/group).

Extended Data Figure 3. Lifespan analysis of EPRS^S/S, EPRS^A/A and EPRS^D/D mice monitored from weaning (>21 days)

a, Youngest and oldest 10% are the mean lifespan of the shortest- and longest-living 10% mice. Numbers of days are represented to nearest full day. Median and mean ± SEM are shown. b, Cox proportional hazard (CPH) regression analyses of EPRS^S/S and EPRS^A/A mice shows genotype as the most significant predictor of increased longevity. Longevity relative to survival days (i.e., age at death) was analyzed in pooled mice by CPH regression model. The 4 independent variables genotype, date of birth (DOB), parental ID (PID), and gender were replaced with a set of category variables. In case of genotype and gender, category represents their presence or absence. DOB and PID data were divided into multiple categories as described in the supplementary methods. Independent variables were fitted into the CPH model individually (univariate) or simultaneously (multivariate). Shown are: β(the unstandardized regression coefficient) with standard error (SE), the degrees of freedom (df), and the significance for each model fit. Exp(β) for the covariate of interest is the predicted change in hazard ratio for a unit increase in the predictor, and its 95% confidence interval (CI). c, Kaplan-Meier survival curves show no change in lifespan of male, female, or combined EPRS^D/D mice. Male (MC χ² = 0.003, P = 0.956; GBW χ² = 0.001, P = 0.972), female (MC χ² = 0.158, P = 0. 0.691; GBW χ²= 0.206, P = 0.650), and gender-combined (MC χ² = 0.079, P = 0. 0.779; GBW χ²= 0.076, P = 0.783). d and e, Survival and CPH regression analyses of EPRS^S/S and EPRS^D/D mice as described above in a and b.

Extended Data Figure 4. Adipose tissue deposition and lipolysis in EPRS^A/A and EPRS^D/D mice

a, Length of mice was measured from head to beginning of tail using a digital caliper (Fisherbrand Traceable). Data shown are mean ± SEM, n = 15 for 20-week male mice. b, Ventral view of wild-type and EPRS^A/A mice abdominal cavity. c, Weights of adipose and non-adipose tissues from 20-week male EPRS^D/D and control mice (mean ± SEM, n = 14/group, P value from unpaired t-test). d, Scanning electron micrographs of EWAT in 20-week male EPRS^S/S, EPRS^A/A, and EPRS^D/D mice. e, Total adipocyte cell number in EWAT of EPRS^A/A knock-in and wild-type mice. Data represent mean ± SEM, n = 5/group. f, Elevated lipolysis in adipocytes from EPRS^A/A, but not EPRS^D/D, mice (mean ± SEM; n = 6/group). g, Elevated β-oxidation in WAT explants from EPRS^A/A mice as determined by release of ¹⁴CO₂ from ¹⁴C-oleic acid (mean ± SEM; n = 5/group). h, Serum levels of insulin, glucose, triglycerides (TG) and free fatty acids (FFA) in 12-h fasted and 1-h post-prandial (fed) 16-week old male mice (mean ± SEM, n = 10/group, *P < 0.05, unpaired t-test). i, Growth curves of EPRS^A/A and EPRS^D/D mice (males, n = 12/group, mean ± SEM, P < 0.001, 2-way ANOVA) started at 6 wk on an unrestricted high-fat diet (HFD, Harlan Teklad TD.06414) deriving 18, 60, and 21 kcal% from protein, fat, and carbohydrate, respectively. j, Phosphorylation of EPRS Ser⁹⁹⁹ in WAT from EPRS^A/A and EPRS^D/D mice after HFD feeding for 24 wk.

Extended Data Figure 5. Adiposity and lifespan of S6K1^−/− mice

a, PCR genotyping of wild-type (S6K1^+/+), heterozygous (S6K1^+/–) and homozygous (S6K1^−/−) mice. b, Immunoblot analysis of S6K1, EPRS, and FATP1 in S6K1^−/− mice. c, Weight of S6K1^−/− mice at embryonic day E16.5, and at postnatal days P0 and P10. d, Kaplan-Meier survival curves shows increased longevity in male, female or combined S6K1^−/− mice. Male (n = 29/group; MC χ² = 4.919, P = 0.027; GBW χ² = 4.660, P = 0.031), female (n = 28 for S6K1^+/+ and n = 26 for S6K1^−/−; MC χ² = 7.927, P = 0.005; GBW χ²= 7.277, P = 0.007), and gender-combined (n = 26 for S6K1^+/+ and n = 55 for S6K1^−/−; MC χ² = 11.78, P = 0.0006; GBW χ²= 11.01, P = 0.0009). e and f, Lifespan and CPH regression analyses of S6K1^+/+ and S6K1^−/− mice as described above in Extended Data Fig. 3a,b.

Extended Data Figure 6. Glucose metabolism, food intake and fecal lipid excretion in EPRS^A/A and EPRS^D/D mice

a, Glucose tolerance test (GTT) in 112-day-old EPRS^S/S, EPRS^A/A, and EPRS^D/D mice (mean ± SEM, n = 10/group). b, Insulin tolerance test (ITT) on mice as in (a; mean ± SEM, n = 10/group). c,d, Same as a,b but using ~600-day-old mice (mean ± SEM, n = 9/group). e-g, Determination of food intake as g/mouse/day (left) or g/body weight (g)/day (right in male (e) and female (f) EPRS^A/A mice, and in male EPRS^D/D mice (g). All values represent mean ± SEM, n = 14/group. h, Fecal lipid excretion in EPRS^S/S, EPRS^A/A, and EPRS^D/D mice (mean ± SEM, n = 6/group). i, Serum ketone body (β-hydroxybutyrate) level in 6-h fasted mice.

Extended Data Figure 7. Energy metabolism in EPRS^A/A and EPRS^D/D mice

a,b, Determination of VO₂ (left) and VCO₂ (right) in EPRS^S/S and EPRS^A/A male mice over a 24-hour period. c,d, Respiratory exchange ratio (RER) (left) and heat generation (right) in 12-h light and dark cycles were determined (mean ± SEM, n = 6/group). e-h, Same as (a-d) but comparing EPRS^S/S and EPRS^D/D male mice (mean ± SEM, n = 6/group). i, Determination of VO₂ in female EPRS^S/S, EPRS^A/A, and EPRS^D/D mice (mean ± SEM, n = 3/group).

Extended Data Figure 8. Absence of GAIT pathway in adipocytes and inflammatory response in EPRS^A/A mice

a, Total protein synthesis determined by [³⁵S]Met/Cys labeling (left), and by incorporation of [¹⁴C]Glu and [¹⁴C]Pro into TCA-precipitated proteins in adipocytes from EPRS^S/S, EPRS^A/A, and EPRS^D/D mice. b, Effect of siRNA-mediated knockdown of S6K1, raptor, and rictor on phosphorylation of EPRS Ser⁹⁹⁹ in differentiated mouse 3T3-L1 adipocytes in presence of 100 nM insulin. c, Effect of IFN-γ and insulin on phosphorylation of EPRS Ser⁹⁹⁹ in differentiated 3T3-L1 adipocytes or mouse primary adipocytes determined using phospho-specific EPRS antibody. d, GAIT complex formation in IFN-γ-stimulated U937 cells and insulin-stimulated 3T3-L1 adipocytes by immunoprecipitation with anti-EPRS antibody and immunoblot with antibodies against GAIT complex constituents. Cytosolic lysates from IFN-γ-treated U937 cells served as positive control. e, Determination of GAIT pathway activity in IFN-γ-stimulated U937 cells and insulin-stimulated 3T3-L1 adipocytes by in vitro translation of a control (T7 gene 10) and GAIT element bearing (Luc-ceruloplasmin (Cp) GAIT element) reporter RNAs. f, White blood cells (WBC) counts in blood from EPRS^S/S and EPRS^A/A mice by Advia hematology system (LUC, large unstained cells). g, Determination of cytokine levels in serum from EPRS^S/S and EPRS^A/A mice. Mouse cytokine antibody arrays were incubated with serum (100 μg, protein, right). Pixel intensity was determined by densitometry (right; mean ± SEM, 2 mice/group). h, Immunoblot analysis of selected cytokines in serum from EPRS^S/S and EPRS^A/A mice.

Extended Data Figure 9. Tissue-specificity of insulin-stimulated LCFA uptake and EPRS-FATP1 interaction

a, [¹⁴C]oleate uptake determined in insulin-stimulated hepatocytes, cardiac cells, soleus muscle strips, and BMDM from EPRS^S/S, EPRS^A/A, and EPRS^D/D mice (mean ± SEM, n = 6 mice/group). b, Insulin-stimulated EPRS Ser⁹⁹⁹ phosphorylation (top), [¹⁴C]oleate uptake (middle), and [¹⁴C]2-deoxy-D-glucose (DG) uptake (bottom) in adipocytes from white adipose tissue of S6K1^+/+ and S6K1^−/− mice (mean ± SEM, n = 5 mice/group). c, Efficiency of EPRS and FATP1 knockdown in 3T3-L1 adipocytes by siRNA targeting each mRNA alone and in combination, as determined by densitometry (NIH image J) of immunoblots shown in Fig. 3g (mean ± SEM, n = 4 experiments). d, Insulin-induced EPRS Ser⁹⁹⁹ phosphorylation, interaction with FATP1, and [¹⁴C]oleate uptake in human adipocytes (mean ± SEM, n = 3 experiments in duplicate). e, Co-immunoprecipitation experiment to determine FATP1 binding to EPRS in lysates from multiple tissues as indicated. f, EPRS and FATP1 expression in male and female S6K1^−/− mice (mean ± SEM, n = 3 mice/group). g, Lack of interaction of EPRS and FATP1 in insulin-stimulated adipocytes of S6K1^−/− mice.

Extended Data Figure 10. Hepatic lipids and FATP1/EPRS membrane localization in EPRS^A/A and EPRS^D/D mice

a, Optimum cutting temperature (OCT) compound-fixed liver slices from EPRS^S/S, EPRS^A/A, and EPRS^D/D mice were stained with H&E (top) or oil red O (bottom), and the latter quantitated by densitometry (right; mean ± SEM, n = 3 mice/group). b, Determination of liver triglycerides in wild-type and mutant mice (mean ± SEM, n = 5 mice/group). c, Insulin-inducible membrane localization of EPRS and FATP1 in adipocytes from wild-type and mutant mice. d, Membrane fractionation shows EPRS and FATP1 co-localizing in plasma membrane (mean ± SEM, n = 3 experiments).

Figure 1. mTORC1-S6K1 phosphorylates EPRS at Ser⁹⁹⁹

a, Schematic of IFN-γ-induced EPRS phosphorylation. Inset, pathway responsible for EPRS Ser⁹⁹⁹ phosphorylation. b, Immunocomplex kinase assay shows S6K1 phosphorylates Ser⁹⁹⁹ in EPRS linker bearing S886A mutation. Kinase activity determined with kinase-specific peptides (mean ± SEM, n = 3). c, S6K1 is required for Ser⁹⁹⁹ phosphorylation in 4-h IFN-γ-treated, siRNA-targeted PBMs. d, S6K1 is the primary S6K isoform that phosphorylates Ser⁹⁹⁹ determined using BMDM of S6K1^−/−, S6K2^−/−, and double-knockout mice. e, IFN-γ induces S6K1-EPRS interaction determined by co-immunoprecipitation. f, Recombinant full-length S6K1 directly phosphorylates Ser⁹⁹⁹ determined by [γ-³²P]ATP labeling and by phospho-specific antibodies. g, Rapamycin (10 nM) blocks S6K1 activation and Ser⁹⁹⁹ phosphorylation in U937 cells.

Figure 2. Reduced adiposity and extended lifespan of EPRS^A/A mice

a, Growth curves of EPRS^A/A (males, mean ± SEM, n = 14/group; P < 0.0001, 2-way ANOVA) and EPRS^D/D (n = 10/group) mice. b, Kaplan-Meier survival curves for male (n = 52/group; log rank Mantel-cox (MC) χ² = 6.75, P = 0.009; Gehan-Breslow-Wilcoxon (GBW) χ² = 6.24, P = 0.012), female (n = 54/group; MC χ² = 7.92, P = 0. 0.005; GBW χ²= 8.91, P = 0.003), and gender-combined (n = 106/group; MC χ² = 15.28, P ≤ 0. 0.0001; GBW χ²= 14.96, P = 0.0001) EPRS^S/S and EPRS^A/A mice. c, Weights of white adipose tissues (WAT) (epididymal, EWAT; inguinal, IWAT; mesenteric, MWAT; retroperitoneal, RWAT) and interscapular brown adipose tissue (IBAT), from 20-week male mice; excised calf muscle comprised gastrocnemicus, soleus and plantaris from fascia (mean ± SEM, n = 14/group, P value from unpaired t-test). d, Histological micrographs of EWAT in 20-week male wild-type (EPRS^S/S), EPRS^A/A, and EPRS^D/D mice. e, Growth of S6K1-null (EPRS^S/S/S6K1^−/−) and wild-type mice (EPRS^S/S/S6K1^+/+), and S6K1-null mice bearing S999D allele (EPRS^D/D/S6K1^−/−) (mean ± SEM, n = 10 males/group; P < 0.0001, 2-way ANOVA). Representative 20-week male mice (right). f, Tissues weights of 20-week mice of genotypes described in (f) (mean ± SEM, n = 10 males/group).

Figure 3. Insulin induces EPRS-FATP1 interaction and LCFA uptake

a, Ser⁹⁹⁹ phosphorylation in insulin-treated adipocytes by ³²P-labeling and by immunoblot with anti-phospho-Ser and -phospho-Ser⁹⁹⁹ antibodies. b, Insulin-induced LCFA uptake is reduced in EPRS^A/A but not EPRS^D/D adipocytes (top, mean ± SEM, n = 10 mice/group). Insulin-induced deoxyglucose uptake is similar in all mice types (bottom, mean ± SEM, n = 6 mice/group). c, EPRS interaction screen of adipose tissue fatty acid binding proteins/transporters in insulin-treated 3T3-L1 adipocytes. d, Verification of EPRS-FATP1 interaction by two-way co-immunoprecipitation, and inhibition by S6K1 knockdown or rapamycin. e, S6K1 knockdown and mTORC1 inhibition block insulin-induced uptake of bodipy-dodecanoate (top) and [¹⁴C]oleate (bottom) (mean ± SEM, n = 4 experiments in duplicate). f, siRNA-mediated knockdown of EPRS, FATP1, or both (top) inhibits LCFA uptake (bottom; mean ± SEM, n = 4 experiments in duplicate) without affecting protein synthesis. g, Insulin-inducible EPRS-FATP1 interaction is significantly reduced in EPRS^A/A adipocytes (mean ± SEM, n = 10 mice/group).

Figure 4. EPRS linker binds FATP1 and induces membrane translocation

a, Insulin induces EPRS linker phosphorylation, binding to FATP1, and LCFA uptake. Endogenous EPRS in 3T3-L1 adipocytes was knocked down by 3′UTR-specific siRNA and cells transfected with Flag-tagged EPRS domains (ERS, PRS, and linker), and insulin-stimulated phosphorylation determined by ³²P-labeling (left). Binding of EPRS domains to FATP1 was detected by co-immunoprecipitation, and LCFA uptake measured (mean ± SEM, n = 4 experiments in duplicate) (right). b, Expression of phospho-mimetic S999D linker facilitates interaction with FATP1 and fatty acid uptake even without insulin (mean ± SEM, n = 3 experiments in duplicate). c, Insulin induces translocation of Ser⁹⁹⁹-phosphorylated EPRS and FATP1 to membranes. Anti-GAPDH and -Na⁺/K⁺ ATPase antibodies verified cytosolic and membrane isolation. d, Effect of knockdown of EPRS (left) and FATP1 (targeting the coding sequence [CDS], right) on translocation to membrane. e, Schematic of mTORC1-S6K1 activation of EPRS and FATP1-mediated LCFA uptake in adipocytes.