Wild-type (WT) and SOCS3-flox (SOCS3f/f) mice were obtained from Jackson Laboratories (Bar Harbor, ME). Cardiac-specific knockout of SOCS3 (SOCS3cko) mice were generated by mating SOCS3f/f mice with mice expressing Cre recombinase under the α-myosin heavy chain (α-MHC) promoter as described previously (Oba et al., 2012). All animals were C57BL/6J background. To induce Cre-dependent recombination, tamoxifen (20 mg/kg body weight, Sigma-Aldrich) was injected intraperitoneally for 5 days over a 3-week duration before experiments. SOCS3f/f mice were used as a control for SOCS3cko mice. Male mice (aged 8–10 weeks) were maintained in a pathogen-free facility at the Laboratory Animal Center at Dalian Medical University. All procedures were performed in accordance with protocols outlined in the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH publication No. 85-23, revised 1996) and approved by the Committee on the Ethics of Animal Experiments of Dalian Medical University, as described in our previous study (Xie et al., 2018).

Male SOCS3f/f and SOCS3cko mice were anesthetized with isoflurane and subjected to pressure overload induced by transverse aortic constriction (TAC) for 4 weeks, as previously described (Li et al., 2007; Xie et al., 2019). Sham-treated mice underwent the same operation without aortic constriction. After recovering from surgery, mice were injected intraperitoneally with 4-PBA (20 mg/kg/day) or vehicle (dimethyl sulfoxide, DMSO) daily for 4 weeks. 4-PBA was first dissolved in DMSO, then diluted in 0.9% NaCl (Metz et al., 1977).

Heart samples were quickly dissected out and rinsed with cool sterile saline, and then fixed in 10% paraformaldehyde, embedded in paraffin, and cut into 5-μm-thick sections for histological analysis. Heart sections were stained with hematoxylin and eosin (H&E), wheat germ agglutinin (WGA) and Masson's trichrome as previously described (Wang L. et al., 2018; Xie et al., 2018). All digital images were taken at ×100 or ×200 magnification of 15–20 random fields from each heart sample. Analysis of myocyte cross-sectional area was calculated by measuring 150 to 200 cells per slide. The areas of myocardial fibrosis were evaluated by Image Pro Plus 3.0 (Nikon, Japan).

Primary experimental procedures for proteomic analysis included protein extraction, trypsin digestion, high-performance liquid chromatography fractionation, liquid chromatography with tandem mass spectrometry, and data analysis supported by Jingjie PTM BioLabs (Hangzhou, China).

All data are expressed as mean ± SEM. All statistical analyses were performed with SPSS 16.0 software (IBM, Armonk, NY). For two-group comparisons, we performed Student's t-test. For comparison of multiple groups, significance was determined using 1-way or 2-way ANOVA with Tukey's post-hoc test. P < 0.05 was considered statistically significant.

To identify which SOCS family members are essential for cardiac hypertrophy and dysfunction, we first evaluated expression of endogenous SOCS members after stimulation with hypertrophic agonists. Quantitative real-time PCR analysis showed that among the eight SOCS family members, only SOCS3 was significantly upregulated at week 2 (the hypertrophic stage) and decreased at week 4 (the HF stage) after TAC (Figure 1A). This change in SOCS3 protein level was further validated by immunoblotting (IB) of the same hearts at different time points (Figure 1B). SOCS3 expression was also appreciably increased in neonatal rat cardiomyocytes (NRCMs) in response to phenylephrine (PE, 100 μmol/L) stimulation for 12–48 h, but was decreased at 72 h (Figure 1C). However, SOCS3 expression was not altered in neonatal rat cardiac fibroblasts after PE stimulation at different time points (Supplementary Figure 1A).

We next investigated the role of SOCS3 in cardiac hypertrophy in vitro. NRCMs were infected with an adenovirus overexpressing SOCS3 (Ad-SOCS3) or empty vector with green fluorescent protein (GFP, Ad-GFP), and treated with PE (100 μM) for 72 h. SOCS3 overexpression (increased by ~2-fold, Supplementary Figure 1B) significantly inhibited PE-induced increases of cardiomyocyte size and expression of the hypertrophic markers atrial natriuretic factor (ANF) and brain natriuretic peptide (BNP) (Figures 1D,E). Conversely, SOCS3 knockdown by siRNAs (decreased by ~50%, Supplementary Figure 1C) enhanced PE-induced hypertrophic responses compared with siRNA-control (Figures 1F,G). Accordingly, SOCS3 overexpression remarkably inhibited activation of the downstream targets gp130, p-JAK2, and p-STAT3 compared with the Ad-GFP control in NRCMs after PE treatment (Figure 1H). Overall, these results indicated that SOCS3 exerts an antihypertrophic role in vitro.

To examine the in vivo pathophysiological role of SOCS3 in the heart, we increased SOCS3 expression in WT hearts by injecting a rAAV9 expressing SOCS3 (rAAV9-SOCS3) or ZsGreen (rAAV9-ZsGreen, a negative control). Transfection efficiency and expression of SOCS3 were confirmed in hearts by fluorescence microscopy of ZsGreen (Supplementary Figure 2A), and immunoblotting analysis indicated a 2.3-fold increase of SOCS3 expression compared with control (Supplementary Figure 2B). Four weeks after TAC operation, echocardiographic assessment revealed that TAC significantly impaired contractile function, as reflected by decreased ejection fraction (EF%) and fractional shortening (FS%) in rAAV9-ZsGreen-injected mice compared with Sham groups, whereas rAAV9-SOCS3-injected mice recovered cardiac dysfunction similar to or better than Sham mice (Figure 2A). Furthermore, TAC-induced decompensation of hypertrophy, as indicated by increases in heart size, ratios of heart weight/body weight (HW/BW) and heart weight/tibia length (HW/TL), cross-sectional area of myocytes, and fibrotic area in rAAV9-ZsGreen-injected mice, were also significantly abrogated in rAAV9-SOCS3-injected mice (Figures 2B–D). Accordingly, mRNA expressions of ANF, BNP, and β-myosin heavy chain (β-MHC), collagen I, and collagen III were markedly reduced in rAAV9-SOCS3-infected mice compared with rAAV9-ZsGreen-infected animals after TAC (Supplementary Figures 2C,D). Moreover, SOCS3 overexpression markedly reduced TAC-induced cardiomyocyte apoptosis, as indicated by the number of TUNEL-positive nuclei in WT hearts compared with rAAV9-ZsGreen-infected hearts (Figure 2E). In addition, gp130, p-JAK2, and p-STAT3 protein levels were consistently downregulated in rAAV9-SOCS3-injected mice compared with rAAV9-ZsGreen-injected mice (Figures 2F,G). These results suggested that cardiac overexpression of SOCS3 enabled improvement of TAC-induced cardiac hypertrophy and dysfunction.

To precisely ascertain whether loss of SOCS3 in cardiomyocytes predisposes mice to HF, SOCS3f/f mice were bred with α-MHC-Cre mice to generate cardiomyocyte-specific SOCS3-knockout mice (SOCS3cko). Specific deletion of SOCS3 in cardiomyocytes was confirmed by immunoblotting analysis, as previously described (Yajima et al., 2011). Wild-type (SOCS3f/f) and SOCS3cko mice were subjected to Sham or TAC operation for 4 weeks. SOCS3f/f mice exhibited characteristics of HF, as indicated by significant reductions in EF% and FS% compared with Sham groups, and this effect was aggravated in SOCS3cko mice (Figure 3A). Moreover, SOCS3cko mice developed severe cardiac hypertrophy and fibrosis, as indicated by increased heart size, HW/TL and LW/TL ratios, cross-sectional area of myocytes, and interstitial collagen deposition compared with SOCS3f/f mice following TAC (Figures 3B–D). Similarly, the mRNA levels of ANF, BNP, β-MHC, collagen I, and collagen III were markedly upregulated in SOCS3cko mice compared with SOCS3f/f animals after TAC (Figures 3E,F). Loss of SOCS3 also induced cardiomyocyte apoptosis (increased number of TUNEL-positive nuclei) in SOCS3cko hearts compared with SOCS3f/f hearts after TAC (Figure 3G). Finally, the protein levels of gp130, p-JAK2, and p-STAT3 were significantly upregulated in SOCS3cko mice compared with SOCS3f/f mice after TAC (Figure 3H). There was no difference in these pathological parameters between the two groups after Sham operation (Figures 3A–H). These results demonstrated that SOCS3cko mice were more susceptible to TAC-induced hypertrophic remodeling and dysfunction.

Although ablation of SOCS3 in cardiomyocytes resulted in activation of gp130/JAK/STAT3 signaling leading to hypertrophic remodeling after long-term TAC (Yasukawa et al., 2001; Yajima et al., 2011), the underlying regulatory mechanism remains largely unknown. To identify novel targets or pathways involved in hypertrophic remodeling in SOCS3cko mice subjected to TAC, we performed proteomic analysis using an iTRAQ-based strategy (data are available via ProteomeXchange with identifier PXD014946). A total of 4,482 proteins were identified from these pairs of heart tissues, among which 3,622 proteins had quantitative changes (Figure 4A). Differentially expressed proteins were selected by filtering with an average cut-off of 1.5-fold change in expression and p-value ≤ 0.05 when comparing TAC-treated heart samples with their corresponding Sham tissues (Figure 4A). A total of 534 proteins qualified as differentially expressed between SOCS3f/f and SOCS3cko mice, including 297 upregulated and 237 downregulated proteins. Analysis of Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment revealed that upregulated proteins in SOCS3cko mice were enriched for protein digestion and absorption, phagosomes and lysosomes, and protein processing in the endoplasmic reticulum (ER) (Figure 4B). Surprisingly, differentially expressed proteins that were downregulated were mostly localized to metabolic pathways and mitochondria. Indeed, 68 of 69 mitochondrial-related proteins were downregulated in SOCS3cko mice after TAC, including proteins linked to mitochondria-related functions such as oxidative phosphorylation, the citrate cycle, carbon metabolism, fatty acid elongation and degradation, pyruvate metabolism, biosynthesis of unsaturated fatty acids, glycolysis, and gluconeogenesis (Figure 4C). These results suggested that increased activation of ER stress and autophagy, and decreased mitochondrial function may play a critical role in cardiac dysfunction of SOCS3cko mice.

We next focused how SOCS3 deletion stimulated activation of ER stress and autophagy in cardiomyocytes in response to TAC stress. We detected several markers of ER stress and mitochondrial autophagy (also known as mitophagy) in hearts of SOCS3f/f and SOCS3cko mice after TAC. Immunoblotting analysis showed that protein levels of GRP78 (also known as BiP and HSPA5), phosphorylated dsRNA-activated protein kinase-like ER kinase (p-PERK, Thr-982), p-elF2α, cleaved ATF6 (p50 fragment), phosphorylated inositol-requiring kinase 1 (p-IRE1, Ser724), CHOP, Parkin, Nix, ATG6 (Beclin 1), and ATG5 proteins were significantly increased in SOCS3cko mice compared with SOCS3f/f animals after TAC (Figure 4D), indicating that SOCS3 is involved in controlling GRP78-mediated ER stress-mitophagy pathway in the heart after TAC.

To further confirm the effect of SOCS3 deletion on mitophagy in vitro, NRCMs were transfected with small interfering RNA (siRNA) against SOCS3 (siRNA-SOCS3) or scrambled control (siRNA-control), and then stained with Mtphagy Dye and Lyso Dye after 24 h of PE or vehicle treatment. Consistent with immunoblotting results, PE stimulation significantly increased induction of mitophagy (red) and mitochondrial autophagosomes (green) in siRNA-control-treated cells, but was further enhanced by siRNA-SOCS3 (Supplementary Figure 3A). Moreover, PE-induced increase of mitochondrial superoxide [as indicated by MitoSOX (red) and MitoTracker Green (green)] and reduction of mitochondrial membrane potential (ΔΨm; stained by MitoProbe JC-1) in cardiomyocytes transfected with siRNA-control was accelerated by transfection of siRNA-SOCS3 (Supplementary Figures 3B,C). Collectively, these results suggested that SOCS3 ablation in cardiomyocytes resulted in marked activation of ER stress and mitophagy leading to mitochondrial dysfunction, which may contribute to progression of cardiomyocyte apoptosis, hypertrophy, and HF.

GRP78 protein is a marker of ER stress and major ER chaperone that controls the activation of transmembrane ER stress sensors. Prompted by our results showing that GRP78 protein (Figure 4D), but not mRNA (Supplementary Figure 2E), was markedly upregulated in SOCS3cko mice compared with SOCS3f/f controls after TAC, we first examined whether SOCS3 associated with GRP78 protein in NRCMs. Co-immunoprecipitation (Co-IP) assays showed that endogenous SOCS3 protein was precipitated by an anti-GRP78 antibody, but not by a non-specific IgG control (Figure 5A). The interaction between SOCS3 and GRP78 was confirmed by an in vitro glutathione-S-transferase (GST) pull-down assay (Figure 5B). Furthermore, an immunoprecipitation (IP) assay was performed in human embryonic kidney (HEK) 293T cells transfected with Myc-SOCS3 and Flag-GRP78. We detected Myc-SOCS3 in the Flag-GRP78 immune complex, whereas no Myc-SOCS3 was found in controls (Figure 5C), indicating that SOCS3 interacted directly with GRP78.

To examine whether SOCS3 regulates GRP78 ubiquitination as an E3 ligase, we con-transfected HEK293T cells with plasmids encoding Myc-ubiquitin, Flag-GRP78, and either GFP-tagged wild-type (WT) SOCS3 or a catalytically inactive mutant (ΔSB: deletion of SB domain). SOCS3 overexpression (WT) significantly enhanced GRP78 ubiquitination, especially Lys48-linked polyubiquitination, whereas this effect was markedly abrogated in cells transfected with SOCS3 (ΔSB) plasmid (Figure 5D). Moreover, upon examining the effect of endogenous SOCS3 on GRP78 ubiquitination in mice, we found that TAC markedly increased GRP78 ubiquitination in SOCS3f/f mice compared with Sham groups, but this effect was remarkably attenuated in SOCS3cko hearts (Figure 5E). Conversely, the TAC-induced response was further enhanced in rAAV9-SOCS3-injected mice compared with rAAV9-ZsGreen control after TAC operation (Figure 5F).

Next, we examined the involvement of the proteasome in SOCS3-mediated degradation of GRP78, a pulse-chase assay was performed in NRCMs using cycloheximide (CHX, a eukaryote protein synthesis inhibitor) with or without MG-132 (a proteasome inhibitor). We discovered that knockdown of SOCS3 by siRNA markedly prolonged the half-life of GRP78 protein compared with the siRNA-control (Figure 5G). Conversely, SOCS3 overexpression with Ad-SOCS3 yielded GRP78 protein with a short half-life compared with Ad-GFP control, but this reduction was completely reversed by MG-132 treatment (Figure 5H), suggesting that SOCS3 promotes GRP78 degradation via the proteasome.

To assess whether GRP78 mediates cardiac hypertrophy in SOCS3cko mice after TAC, we injected SOCS3f/f and SOCS3cko mice with rAAV9-siRNA to knock down endogenous GRP78 expression. After 3 weeks of injection, all mice were then subjected to TAC for 4 additional weeks. Injection of AAV9-siGRP78 significantly downregulated GRP78 protein levels in the heart by about 45–50% compared with rAAV9-siControl (Figure 6D). Moreover, consistent with results described above (Figure 3), after 4 weeks of TAC, SOCS3cko mice showed marked cardiac dysfunction (reduced EF% and FS%), hypertrophy (increased heart size, ratios of HW/BW and HW/TL and cross-sectional areas of myocytes), and interstitial fibrosis, as well as upregulation of ANF, BNP, collagen I, and collagen III mRNA expression compared with SOCS3f/f mice after injection of rAAV9-siControl (Figures 6A–C; Supplementary Figures 4A,B, lane 3 vs. 1). Conversely, these deleterious effects were markedly reversed in SOCS3cko mice injected with rAAV9-siGRP78 (Figures 6A–C; Supplementary Figures 4A,B, lane 4 vs. 3). Similarly, rAAV9-siGRP78 injection of SOCS3f/f mice also improved TAC-induced cardiac dysfunction and hypertrophic responses compared with rAAV9-siControls (Figures 6A–C; Supplementary Figures 4A,B, lane 2 vs. 1). Correspondingly, GRP78 knockdown in SOCS3cko or SOCS3f/f mice reduced protein levels of p-PERK, cleaved ATF6, CHOP, and Parkin compared with rAAV9-siControl mice after TAC (Figure 6D). Together, these in vivo findings suggested that SOCS3 ablation aggravated cardiac hypertrophy and dysfunction by enhancing GRP78 and its downstream effectors.

To further verify the involvement of GRP78-mediated activation of ER stress in TAC-induced cardiac hypertrophy in SOCS3cko mice, SOCS3f/f and SOCS3cko mice were administered 4-PBA, a reported inhibitor of ER stress that can attenuate pressure overload-induced hypertrophy (Luo et al., 2015). As expected, 4-PBA treatment reduced GRP78 expression in both SOCS3f/f and SOCS3cko mice after TAC (Supplementary Figures 5E,F). Consistent with observations from siRNA-GRP78 knockout experiments (Figure 6), TAC-induced increases in cardiac dysfunction (reduced EF% and FS%), hypertrophy (increased heart size, ratios of HW/BW and HW/TL and cross-sectional areas of myocytes), and interstitial fibrosis were enhanced in SOCS3cko mice compared with SOCS3f/f mice (Supplementary Figures 5A–D, lane 2 vs. 1), but were significantly reduced in SOCS3cko mice treated with 4-PBA compared with vehicle (Supplementary Figures 5A–D, lane 4 vs. 2). Furthermore, 4-PBA administration in SOCS3f/f mice also showed marked cardioprotection compared with vehicle control after TAC (Supplementary Figures 5A–D, lane 3 vs. 1). The preventive effect of 4-PBA treatment on activation of ER stress and autophagy markers (p-PERK, cleaved ATF6, CHOP, and Parkin) was further confirmed in SOCS3cko and SOCS3f/f mice (Supplementary Figures 5E,F). Thus, these in vivo observations confirmed that the prohypertrophic effect of SOCS3 ablation resulted from activation of ER stress.