Hepatic ChREBP Drives Cardiac Remodeling via ApoM Non-transcriptional Repression Authors: Shuang Zhang, Zhenzhen Zhang, Zihan Ma, Wen Wu, Lu Tang, Yanlu Han, Songning Chen, Tengteng Yan, Ye Chen, Junwu Liu, Dongdong Jian, Ji’e Yang, Likun Ma, Zequn Yin, Houzao Chen, Baofa Sun, Deling Kong, Junbo Ge, Yajun Duan Published: 2026 ABSTRACT BACKGROUND:Pathological cardiac remodeling is a hallmark of numerous cardiovascular diseases and develops into heart failure. As a systemic disease, effective treatments for cardiac remodeling from a tissue crosstalk perspective are still significantly unmet.METHODS:Hepatocyte-specific ChREBP (carbohydrate response element binding protein) knockout (KO) or overexpressed mice, global ApoM (apolipoprotein M) KO and adeno-associated virus-mediated hepatic ApoM knockdown or overexpressed mice, cardiomyocyte-specific ChREBP overexpressed mice, as well as S1PR1 (sphingosine-1-phosphate receptor 1) knockdown mice were used in isoproterenol- or transverse aortic constriction-induced cardiac remodeling models. RNA sequencing and liquid chromatography - tandem mass spectrometry (LC-MS/MS) analysis were used to detect changed pathways and the interaction between ChREBP and SURF4 (surfeit 4).RESULTS:We found increased ChREBP expression in the liver, but not in the heart, especially in the cytosol of hepatocytes, but not the nucleus, in isoproterenol- or transverse aortic constriction-induced mice. Hepatocyte-specific ChREBP deficiency protected against isoproterenol or transverse aortic constriction-induced cardiac remodeling. Mechanistically, hepatocyte ChREBP deficiency increased ApoM expression in the liver and its secretion, not by transcriptional regulation of ApoM, but by increasing its secretion through the release of SURF4. ApoM overexpression in the liver or ApoM-containing high-density lipoprotein (HDL) injection can both ameliorate cardiac remodeling through the S1P (sphingosine-1-phosphate)/S1PR1 pathway in the heart.CONCLUSIONS:This work identifies the hepatic ChREBP-SURF4-ApoM axis as a critical pathway in cardiac remodeling, and induction of hepatic ApoM secretion constitutes a new promising approach for treating cardiac remodeling and heart failure. Novelty and Significance What Is Known? Cardiac remodeling drives heart failure and cardiovascular mortality. Cardiac remodeling drives heart failure and cardiovascular mortality. Hepatic ChREBP (carbohydrate response element binding protein) regulates glucose and lipid metabolism via nuclear translocation and ChoRE (carbohydrate response element) binding. Hepatic ChREBP (carbohydrate response element binding protein) regulates glucose and lipid metabolism via nuclear translocation and ChoRE (carbohydrate response element) binding. Liver-secreted ApoM (apolipoprotein M) protects against cardiovascular diseases. Liver-secreted ApoM (apolipoprotein M) protects against cardiovascular diseases. What New Information Does This Article Contribute? Liver-specific (not cardiac) ChREBP upregulation drives cardiac remodeling and acts via cytoplasmic accumulation, not nuclear translocation. Liver-specific (not cardiac) ChREBP upregulation drives cardiac remodeling and acts via cytoplasmic accumulation, not nuclear translocation. Cytosolic ChREBP binds SURF4 (surfeit 4), blocking ApoM secretion. Cytosolic ChREBP binds SURF4 (surfeit 4), blocking ApoM secretion. Hepatic ChREBP deficiency restores ApoM secretion and improves cardiac function via S1P (sphingosine-1-phosphate)/S1PR1 (sphingosine-1-phosphate receptor 1). Hepatic ChREBP deficiency restores ApoM secretion and improves cardiac function via S1P (sphingosine-1-phosphate)/S1PR1 (sphingosine-1-phosphate receptor 1). Pathological cardiac remodeling is a major risk factor for heart failure, hepatic transcription factor ChREBP typically regulates glucose and lipid metabolism by translocating into the nucleus to exert transcriptional regulatory functions. Furthermore, the liver-secreted protein ApoM exerts cardiovascular protective effects, and inter-organ crosstalk plays a role in systemic homeostasis. This study reveals that in response to cardiac remodeling, ChREBP is specifically upregulated in the hepatocyte cytoplasm, where it interacts with the ER protein SURF4 to inhibit ApoM secretion. We identify this liver-heart axis via hepatic ChREBP-SURF4-ApoM pathway as a novel mechanistic basis for ameliorating cardiac remodeling, offering a promising therapeutic strategy for heart failure. Meet the First Author, see p e000770 Pathological cardiac hypertrophy, characterized by myocardial thickening, is an independent risk factor for cardiovascular disease.1,2 Compensatory cardiac enlargement involves biochemical, molecular, structural, and metabolic adaptations. Chronic stress or disease eventually leads to ventricular dilation, impaired contractile function, and ultimately heart failure.3 Although hypertrophy initially serves as a compensatory mechanism, it often marks the onset of myocardial remodeling clinically. Hepatic ChREBP (carbohydrate response element binding protein; barely detectable in the heart) is a key transcriptional regulator of de novo lipogenesis, glucose homeostasis, and insulin sensitivity.4 In response to glucose metabolites, it translocates to the nucleus and activates target genes via ChoRE (carbohydrate response element).5,6 Several hepatokines, including fibroblast growth factor 21 (FGF21),7 ANGPTL3 (angiopoietin-like 3),8 ANGPTL8, HGFAC (hepatocyte growth factor activator),9 and S100A6 (S100 calcium-binding protein A6),10 are direct ChREBP targets.11 These secreted proteins enter the circulation to modulate metabolic homeostasis in distant tissues. ApoM (apolipoprotein M), a liver-secreted lipocalin,12 exerts antioxidative and antiatherogenic effects,13 modulates inflammation,14 protects the endothelium,15 and enhances cell survival.16 Reduced circulating ApoM levels correlate with higher heart failure mortality.17 ApoM deficiency aggravates diabetes-induced heart failure with preserved ejection fraction.18 Its expression is transcriptionally upregulated by HNF-1α (hepatic nuclear factor-1α),19 HNF-4α,20 and LRH-1 (liver receptor homolog-1),21 but downregulated by liver LXR (X receptor), retinoid X receptor (RXR), farnesoid X receptor (FXR), and small heterodimer partner (SHP).22 However, the mechanisms governing ApoM regulation and its signaling pathways remain unclear. Interorgan crosstalk has recently garnered increasing attention, as cross-tissue communication plays a pivotal role in coordinating organ function, maintaining whole-body homeostasis, and mediating disease adaptation.23 Although certain hepatic factors, such as IL-6 (interleukin-6) and FXI (coagulation factor XI), have been implicated in heart failure and myocardial infarction,24,25 the roles of many others remain to be comprehensively investigated. In this study, we identified ChREBP as the hepatic (but not cardiac) transcription factor most associated with altered pathways in cardiac remodeling. Unlike previous reports, hepatic ChREBP showed cytoplasmic enrichment rather than nuclear translocation, with pathway disruptions in protein transport and secretion. Through in vivo, in vitro, and clinical studies, we demonstrated that cytoplasmic ChREBP interacts with the endoplasmic reticulum (ER) protein SURF4 (surfeit 4), impairing protein secretion, reducing ApoM, and promoting cardiac remodeling progression. Methods Data Availability The detailed methods used in this study are provided in the Supplemental Material and Major Resources Table. The raw RNA sequencing data generated and analyzed in this study have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus database (GSE337074). All other relevant data are available from the corresponding author upon reasonable request. Mice Models ISO- or TAC-Induced Model To establish a mouse model of cardiac remodeling, 8-week-old male mice were administered subcutaneous injections of isoproterenol (ISO) at a dose of 10 mg/kg per day.26 Mice were then sacrificed at week 4. Control mice received subcutaneous injections of normal saline following the same schedule. Transverse aortic constriction (TAC) surgery was conducted on 8-week-old male mice.27 Mice were anesthetized with sodium pentobarbital (50 mg/kg, i.p.), and a small incision was made at the left second intercostal space. The thoracic aorta was ligated with a 7-0 silk suture using a 27-G needle (for body weight 22–28 g) or 26-G needle (>28 g), which was then promptly removed. After 4 weeks, ligation efficacy was confirmed by echocardiography and histopathology. Sham mice underwent the same procedure without ligation. Blood Pressure Measurement The blood pressure of awake mice was evaluated using a small animal noninvasive blood pressure measuring instrument (Zhongshi Technology, NIBP-600). Mice were placed in a mouse immobilizer and measurements were recorded under steady-state conditions until 3 stable consecutive measurements were obtained. Echocardiographic Test Transthoracic echocardiography was performed using the VINNO 6 LAB system (VINNO, China). Mice were anesthetized with oxygen and isoflurane (1%–2%), and left ventricular ejection fraction (LVEF) and other systolic function indicators were obtained by a horizontal short-axis M-mode scan of the ventricles. Diastolic function at the mitral valve level was measured using pulsed-wave and tissue Doppler. Parameters collected included shortened left ventricular fraction (LVFS), LVEF, peak transmitral velocity in early diastole (E), peak transmitral velocity in late diastole (A), peak tissue Doppler velocity in early diastolic mitral annular myocardial diastolic velocity (e′) and relative wall thickness (RWT). All measurements were made by 2 independent observers blinded to group allocation and clinical data. Exercise Exhaustion Test Locomotor capacity was assessed using a small animal treadmill (ZS-PT-III, Zhongshi Technology). Mice ran at a warm-up speed of 5 m/min for 4 minutes, followed by 14 m/min for 2 minutes, after which the speed was increased by 2 m/min every 2 minutes until exhaustion. Exhaustion was defined as failure to resume running within 10 seconds after contact with the electric stimulation grid. Running time and distance were recorded at exhaustion. All mice underwent pretest training. Statistical Analysis Data are presented as mean ± SD and were analyzed using GraphPad Prism (v10.5.0) or R 4.5.1. For n < 10, nonparametric tests were used: Mann-Whitney U test for 2 groups, Kruskal-Wallis test + Dunn test for single-factor multigroup, and aligned rank transform + Tukey test for 2-factor designs. For n ≥ 10, normality and homoscedasticity were assessed using the Shapiro-Wilk and Brown-Forsythe tests. Two-group comparisons, an unpaired t test (equal variances), Welch t test (unequal variances), or Mann-Whitney U test (nonnormal data) was used. For single-factor multigroup comparisons, 1-way ANOVA + Tukey test (normal data, equal variances), Brown-Forsythe and Welch ANOVA + Dunnett T3 test (normal data, unequal variances), or Kruskal-Wallis + Dunn test (nonnormal data or small samples) was used. FDR correction was applied to omics data; within-experiment post hoc adjustments were used for wet-laboratory assays without global correction across experiments. Statistical significance was defined as P < 0.05. Please see the Major Resources Table in the Supplemental Material. Results Differential Increase of ChREBP Expression Observed in Liver But Not in Heart in the Pathological Cardiac Remodeling Mice Model As pathological hypertrophy can lead to heart failure, understanding this phase is of paramount importance for therapeutic development. First, we established ISO-induced models for 1, 2, and 4 weeks (Figure 1A). No remodeling was observed after 1 or 2 weeks (Figure S1A through S1I, S1J through S1R), whereas 4‑week treatment induced cardiac dysfunction, as evidenced by decreases in LVEF and LVFS, increases in RWT, E/A, and E/e′ ratios (Figure 1B and 1C), increased heart weight-to-tibial length (HW/TL), heart weight-to-body weight (HW/BW), and lung weight-to-body weight (LW/BW) ratios (Figure 1D), reduced running distance, elevated systolic blood pressure (SBP) and diastolic blood pressure (DBP; Figure S2A and S2B), higher serum creatine kinase (CK), LDH (lactate dehydrogenase), HBDH (hydroxybutyrate dehydrogenase), and AST (aspartate aminotransferase) (Figure S2C), and cardiomyocyte hypertrophy with fibrosis (Figure 1E and Figure S2D), confirming established myocardial remodeling at this time point. ChREBP (carbohydrate response element binding protein) is upregulated and localized in cytoplasm of hepatocytes in pathological cardiac remodeling. A, The schedule of in vivo experiment. B, Representative M-mode echocardiograms and Doppler flow images of mice. C, Quantitative analysis of left ventricular ejection fraction (LVEF), shortened left ventricular fraction (LVFS), relative wall thickness (RWT), E/A (early diastolic transmitral flow velocity to late diastolic transmitral flow velocity) and E/e′ (early diastolic transmitral flow velocity to early diastolic mitral annular velocity) ratios. D, Heart weight-to-tibial length (HW/TL), heart weight-to-body weight (HW/BW), lung weight-to-body weight (LW/BW) ratios. E, Wheat germ agglutinin (WGA) staining and cardiomyocyte cross-sectional area. F, Enrichment of representative gene ontology (GO) terms (P < 0.05 and FDR < 0.25) by gene set enrichment analysis (GSEA) in the isoproterenol (ISO) group compared with control group. Ridge plot displays the distribution of core enriched gene sets. Ridge color: −log10(P value). Point color: normalized enrichment score (NES). G, Differential expression of transcription factors (TFs) in liver tissues from ISO-induced models vs controls. Left: Heatmap showing Z-score normalized expression of TFs. Right: Dot color denotes log2FC and size corresponds to −log10(FDR). H, Western blot of protein expression of transcription factors. I, Western blot of ChREBP expression in cytoplasm and nucleus. J, Immunofluorescence of ChREBP (red) and Alb (green) with colocalization quantification. K, Western blots and quantification of hepatic ChREBP levels at 1, 2, and 4 weeks post-ISO induction. L, Scatter plots showing significant negative correlations between hepatic ChREBP integrated optical density (IOD) with LVEF and LVFS. Statistical significance was determined by Pearson correlation analysis. All quantitative data are expressed as mean ± SD. P values were determined by: Mann-Whitney U test for C through E. P < 0.05 was considered statistically significant. FC indicates fold change. We also performed TAC surgery for 1, 2, and 4 weeks (Figure S3A). After 1 week, no significant cardiac morphological or functional changes were detected (Figure S3B through S3J). After 2 and 4 weeks, mice exhibited marked cardiac dysfunction (decreased LVEF and LVFS, increased RWT, and elevated E/A and E/e′ ratios; Figures S4A through S4C and S5), reduced running endurance, elevated SBP and DBP, increased HW/TL, HW/BW, and LW/BW ratios (Figure S4E and S4F), and higher serum CK, LDH, HBDH, and AST levels (Figure S4G). Histopathology revealed cardiomyocyte hypertrophy and myocardial fibrosis (Figure S4H and S4I). The ISO model recapitulates circulating factor-driven metabolic disturbances, providing a physiological platform for liver-heart axis studies. Our study shows that 4-week ISO induction triggers significant myocardial remodeling, a stage with defined pathology and clinical relevance that serves as a critical window for interrogating organ crosstalk before heart failure. Thus, we used this model to investigate the mechanistic role of hepatic ChREBP in myocardial remodeling. Liver RNA sequencing from 4-week ISO mice with Gene Set Enrichment Analysis (GSEA) positively enriched in unsaturated fatty acid metabolic process, fatty acid metabolic process, negatively enriched lipid oxidation, cellular response to lipid, and signaling receptor activator activity (Figure 1F). Among the hepatic transcription factors screened, including Mlxipl (encoding ChREBP28), Nr1h4 (encoding FXR/farnesoid X receptor), Pparg (encoding PPARγ/peroxisome proliferator-activated receptor gamma), Hnf1a (encoding HNF1α/hepatocyte nuclear factor 1 alpha), Nr1h3 (encoding LXRα/liver X receptor alpha), Hnf4a (encoding HNF4α), and Nr5a2 (encoding LRH1/liver receptor homolog 1), Mlxipl showed the strongest and most significant upregulation (Figure 1G). Among the examined transcription factors, only ChREBP protein was significantly elevated in ISO-treated livers (Figure 1H). Unlike its canonical nuclear translocation,29,30 ISO and TAC induced cytoplasmic ChREBP accumulation with negligible nuclear increase (Figure 1I and 1J; Figure S6A). Hepatic ChREBP upregulation occurred as early as 2 weeks and became marked at 4 weeks in both ISO and TAC models (Figure 1K; Figure S6B), indicating that this change precedes myocardial remodeling and is driven by pressure overload rather than ISO directly. GSEA of liver RNA sequencing revealed enriched GO processes in ISO-treated mice, with Mlxipl as a shared core gene, suggesting its regulatory role in metabolic and cellular responses to ISO (Figure S7A). Furthermore, correlation analysis revealed that Mlxipl expression negatively correlated with several cellular processes, including protein localization to the plasma membrane, Golgi-to-lysosome transport, receptor complex, and regulation of receptor binding (Figure S7B), suggesting its involvement in altered intracellular trafficking and receptor-related biology. Clinically, we measured hepatic ChREBP expression by immunohistochemistry in peritumoral liver samples from patients with available LVEF and LVFS data (Figure S8) and found that ChREBP negatively correlated with both LVEF and LVFS (Figure 1L), further supporting the negative association between hepatic ChREBP and cardiac function. Immunofluorescence showed that ChREBP did not colocalize with oval cells, endothelial cells, hepatic stellate cells, or cholangiocytes in the liver, nor with cardiomyocytes, endothelial cells, or myofibroblasts in the heart, and cardiac ChREBP levels were low basally (Figure S9A through S9D). These findings identify hepatocyte ChREBP as a key target in pathological cardiac remodeling, acting independently of its canonical transcriptional function. Hepatocyte-Specific ChREBP Deficiency Ameliorated ISO- and TAC-Induced Pathological Cardiac Remodeling in Mice Based on these findings, we generated L‑ChREBP KO mice and established an ISO model (control, ISO, KO, KISO; Figure 2A). KO alone had no cardiac effect, but under ISO, KISO restored LVEF, LVFS, RWT, E/A, E/e′, running distance, SBP, and DBP to control levels (Figure 2B through 2F), normalized HW/TL and HW/BW (Figure 2G), prevented ISO-induced elevations of CK, LDH, HBDH, and AST (Figure 2H), and attenuated cardiac hypertrophy, fibrosis, and cardiomyocyte enlargement (Figure 2I through 2K). In addition, KO reduced systemic inflammation (Figure 2L) and reversed ISO‑induced increases in atrial natriuretic peptide (ANP), brain natriuretic peptide (BNP), β-myosin heavy chain (β‑MHC), and α smooth muscle actin (α‑SMA; Figure 2M). Hepatocyte-specific ChREBP (carbohydrate response element-binding protein) knockout ameliorates isoproterenol (ISO)-induced cardiac remodeling and dysfunction. A, The schedule of in vivo experiment. B, Representative M-mode echocardiograms and Doppler flow images. C, Quantitative analysis of left ventricular ejection fraction (LVEF), shortened left ventricular fraction (LVFS), relative wall thickness (RWT). D, E/A (early diastolic transmitral flow velocity to late diastolic transmitral flow velocity) and E/e′ (early diastolic transmitral flow velocity to early diastolic mitral annular velocity) ratios. E, Running distance. F, Systolic blood pressure (SBP) and diastolic blood pressure (DBP) levels. G, Heart weight-to-tibial length (HW/TL) and heart weight-to-body weight (HW/BW) ratios. H, Serum levels of CK (creatine kinase), LDH (lactate dehydrogenase), HBDH (hydroxybutyrate dehydrogenase), and AST (aspartate aminotransferase). I, Gross heart images and hematoxylin and eosin staining. J, Sirius Red and wheat germ agglutinin (WGA) staining. K, Quantification of collagen fibers and cardiomyocyte cross-sectional area. L, Serum tumor necrosis factor-α (TNF-α) and interleukin-1β (IL-1β) levels measured by ELISA. M, Western blot of atrial natriuretic peptide (ANP), brain natriuretic peptide (BNP), β-myosin heavy chain (β-MHC), and α smooth muscle actin (α-SMA) expression levels in mice heart. n = 6. All quantitative data are expressed as mean ± SD. P values were determined by: Aligned rank transform (ART) test with Tukey post hoc test for C through L. P < 0.05 was considered statistically significant. To test whether hepatic ChREBP overexpression worsens remodeling, WT mice received adeno-associated virus 8 (AAV8)-TBG-ChREBP followed by ISO (Figure S10A). Overexpression alone had no baseline effect, but under ISO, it exacerbated LVEF and LVFS decline and loss of running endurance, increased SBP and DBP, and moderately worsened hypertrophy and fibrosis (Figure S10B through S10E, S10H, and S10I), with no significant changes in RWT, E/A, E/e′, HW/TL, HW/BW, LW/BW, or serum CK, LDH, HBDH, and AST (Figure S10C, S10F, and S10G). These data indicate that hepatic ChREBP upregulation alone is insufficient to trigger remodeling, but under ISO, its removal is cardioprotective, whereas its overexpression adds limited further exacerbation. To investigate cardiac ChREBP function despite its low basal expression, we generated cardiac-specific ChREBP overexpression mice via AAV9-cTNT-ChREBP, with or without ISO challenge (Figure S11A). No cardiac functional changes were observed (Figure S11B through S11I), confirming that hepatic, not cardiac, ChREBP drives myocardial remodeling pathogenesis. We extended our analysis to TAC‑induced cardiac remodeling in L‑ChREBP KO mice (Figure S12A). Consistent with the ISO results, ChREBP knockout significantly attenuated TAC-induced declines in LVEF and LVFS; attenuated the increases in RWT, E/A, and E/e′; restored running distance; attenuated the elevation of SBP and DBP; reduced cardiac hypertrophy indices; and lowered serum cardiac enzymes (Figure S12B through S12H). Moreover, cardiomyocyte hypertrophy, fibrosis (Figure S12I through S12K), serum TNF-α and interleukin-1β (IL-1β), and cardiac hypertrophy markers ANP, BNP, β-MHC, and α-SMA were all restored to control levels (Figure S12L and S12M). In conclusion, hepatocyte ChREBP knockout consistently attenuated cardiac hypertrophy, fibrosis, and systemic inflammation, supporting its therapeutic potential in cardiac remodeling. ApoM May Be the Key Hepatic Secretory Protein Linking Hepatic ChREBP to Cardiac Remodeling To elucidate the molecular mechanisms by which hepatic ChREBP regulates cardiac remodeling, we performed RNA sequencing analysis on mice liver tissues in Figure 2A. To identify potential target genes, we included all differentially expressed genes with FDR < 0.05 and |log2FC| > 1 from both the ISO versus control and KISO versus ISO comparisons (Table S1). The 2-set overlap was cross-referenced with 2 protein databases: (1) mice liver-heart interaction protein set25 (Table S2); and (2) human secreted proteins released from the liver31,32 (Table S3; Figure 3A). This screening for liver-derived secreted proteins affected by hepatic ChREBP knockout in cardiac remodeling and conserved across mice and humans identified only 1 candidate, ApoM, which was significantly upregulated in the KISO versus ISO groups (Figure 3B). As a liver-specific S1P (sphingosine-1-phosphate)-binding high-density lipoprotein (HDL) component, ApoM has been shown to protect against cardiovascular events.13,14 We found that, in both ISO- and TAC-induced models, serum and hepatic ApoM levels were significantly suppressed, but this effect was reversed in L-ChREBP KO mice (Figure 3C and 3D), indicating an inverse relationship between hepatic ChREBP and ApoM secretion and expression. We next analyzed the direct relationship between ApoM and cardiac indexes in ISO- and TAC-induced mice models. We observed a significant increase in serum BNP levels, a decrease in ApoM levels (Figure 3E and 3F), and a marked decrease in LVEF (Figure 2C; Figure S12C). Liver-specific derived ApoM (apolipoprotein M) is related to cardiac function and is downregulated in the pathological cardiac remodeling mice. A, Venn of candidate genes: overlap of up/down differentially expressed genes (DEGs) false discovery rate (FDR) < 0.05, |log2FC| > 1 from isoproterenol (ISO)/control and KISO (knockout with isoproterenol)/ISO with liver‑secreted protein and liver-heart mediator databases. B, Volcano plot of DEGs in KISO vs ISO: pink (up, FDR < 0.05, FC > 2), blue (down, FDR < 0.05, FC < 0.5), gray (ns). C, Serum ApoM levels measured by ELISA. D, Western blot analysis of ApoM expression in mice liver. E and F, Serum brain natriuretic peptide (BNP) and ApoM levels. G and H, Spearman correlations of circulating ApoM with left ventricular ejection fraction (LVEF) and BNP. I, Pearson correlations of hepatic ApoM IOD with LVEF and shortened left ventricular fraction (LVFS). J, ELISA quantification of serum ApoM concentrations in control subjects and patients with myocardial hypertrophy, and the corresponding LVEF, LVFS and NT-proBNP (N-terminal pro-B-type natriuretic peptide) were measured. All quantitative data are expressed as mean ± SD, P values were determined by: Aligned rank transform (ART) test with Tukey post hoc test for C. Mann-Whitney U test for E (BNP) and J (LVEF, LVFS, and NT-proBNP); Unpaired t test with Welch correction for E (ApoM), F (BNP) and J (ApoM); 2-tailed Student t test for F (ApoM); Spearman correlation analysis for G and H. P < 0.05 was considered statistically significant. ns indicates not significant. To determine whether ApoM changes during early cardiac hypertrophy, we measured its circulating levels in ISO- and TAC-induced mice models at 1, 2, and 4 weeks. A significant decrease was observed at both 2 and 4 weeks (Figure S13A and S13B). At 4 weeks, serum ApoM positively correlated with LVEF and negatively with BNP in control and ISO- or TAC-induced mice (Figure 3G and 3H). Clinically, hepatic ApoM expression levels positively correlated with LVEF and LVFS levels in humans (Figure 3I; Figure S14). Consistently, serum ApoM levels in patients with myocardial hypertrophy were significantly lower than those in the control group, along with decreased LVEF and LVFS and increased NT-proBNP levels (Figure 3J). These findings link hepatic ChREBP to ApoM expression and secretion and reveal a significant correlation between ApoM and cardiac function. Hepatic ApoM Mediates ChREBP Knockout-Induced Cardioprotection and Its Overexpression Alleviates Cardiac Remodeling To test whether hepatic ApoM mediates the protection of ChREBP knockout, we knocked down ApoM in L-ChREBP KO mice (AAV8-TBG-shApoM) followed by ISO treatment (Figure 4A). ApoM knockdown reversed all KO-mediated benefits, including LVEF, LVFS, RWT, E/A, E/e′, running distance, SBP, DBP, HW/TL, HW/BW, LW/BW, serum CK, LDH, HBDH, and AST, and aggravated hypertrophy/fibrosis with restored ANP, BNP, β-MHC, and α-SMA (Figure 4B through 4L). These results indicate that hepatic ApoM is a key mediator by which ChREBP knockout alleviates cardiac remodeling, and that ApoM suppression abolishes this protection. Hepatocyte-specific ApoM (apolipoprotein M) knockdown abrogated the protective effects of hepatocyte ChREBP (carbohydrate response element binding protein) knockout against isoproterenol (ISO)-induced pathological cardiac remodeling in mice. A, The schedule of in vivo experiment. B, Representative M-mode echocardiograms and Doppler flow images. C, Quantitative analysis of left ventricular ejection fraction (LVEF), shortened left ventricular fraction (LVFS), relative wall thickness (RWT). D, E/A (early diastolic transmitral flow velocity to late diastolic transmitral flow velocity ratio) and E/e′ (early diastolic transmitral flow velocity to early diastolic mitral annular velocity) ratios. E, Running distance. F, Systolic blood pressure (SBP) and diastolic blood pressure (DBP) levels. G, Heart weight-to-tibial length (HW/TL), heart weight-to-body weight (HW/BW), and lung weight-to-body weight (LW/BW) ratios. H, Serum levels of creatine kinase (CK), LDH (lactate dehydrogenase), HBDH (hydroxybutyrate dehydrogenase), and AST (aspartate aminotransferase). I, Wheat germ agglutinin (WGA) staining and Sirius Red staining. J, Quantification of cardiomyocyte area and fibrosis. K, Gross heart images and hematoxylin and eosin staining. L, Western blot analysis of atrial natriuretic peptide (ANP), brain natriuretic peptide (BNP), β-myosin heavy chain (β-MHC), and α smooth muscle actin (α-SMA) expression levels in mice heart. n = 6. All quantitative data are expressed as mean ± SD. P values were determined by: Kruskal-Wallis test followed by Dunn multiple comparisons test for C through J. P < 0.05 was considered statistically significant. To test the direct role of hepatic ApoM, mice received AAV8-TBG-ApoM followed by ISO treatment (Figure 5A). ApoM overexpression significantly reversed ISO-induced cardiac dysfunction, as shown by improved LVEF and LVFS, reduced RWT, E/A, and E/e′, increased running distance, lower SBP and DBP, decreased HW/TL, HW/BW, and LW/BW, reduced serum CK, LDH, HBDH, and AST, attenuated hypertrophy and fibrosis, and suppressed ANP, BNP, β-MHC, and α-SMA (Figure 5B through 5K). Thus, hepatic ApoM overexpression alone ameliorates cardiac abnormalities, suggesting ApoM enhancement as a potential therapy for cardiac remodeling. Hepatocyte-specific overexpression of ApoM (apolipoprotein M) alone significantly ameliorates pathological phenotypes associated with cardiac remodeling in mice. A, The schedule of in vivo experiment. B, Representative M-mode echocardiograms and Doppler flow images. C, Quantitative analysis of left ventricular ejection fraction (LVEF), shortened left ventricular fraction (LVFS), relative wall thickness (RWT). D, E/A (early diastolic transmitral flow velocity to late diastolic transmitral flow velocity) and E/e′ (early diastolic transmitral flow velocity to early diastolic mitral annular velocity) ratios. E, Running distance. F, Systolic blood pressure (SBP) and diastolic blood pressure (DBP) levels. G, Heart weight-to-tibial length (HW/TL), heart weight-to-body weight (HW/BW), and lung weight-to-body weight (LW/BW). H, Serum levels of creatine kinase (CK), LDH (lactate dehydrogenase), HBDH (hydroxybutyrate dehydrogenase), and AST (aspartate aminotransferase). I, Gross heart images, hematoxylin and eosin, Sirius Red and wheat germ agglutinin (WGA) staining. J, Quantification of fibrosis and cardiomyocyte area. K, Western blot analysis of atrial natriuretic peptide (ANP), brain natriuretic peptide (BNP), β-myosin heavy chain (β-MHC), and α smooth muscle actin (α-SMA), and ApoM expression levels. n = 6. All quantitative data are expressed as mean ± SD. P values were determined by: Mann-Whitney U test for C through J. P < 0.05 was considered statistically significant. Protein Interaction Between ChREBP and SURF4 Inhibits SURF4-Mediated Protein Transport and Secretion of ApoM To elucidate how cytosolic ChREBP inhibits ApoM expression and secretion, we performed gene ontology (GO) enrichment analysis on differentially expressed genes from KISO versus ISO liver transcriptomes. Several enriched GO terms related to lipoprotein and lipid metabolism were observed, suggesting that ChREBP may influence lipoprotein-related processes under ISO (Figure 6A). Combined with our finding that ISO induces cytoplasmic ChREBP accumulation (Figure 1), these results suggest that cytosolic ChREBP may regulate ApoM transport and secretion. To further characterize the ChREBP-interacting proteome, co-IP coupled with LC-MS/MS analysis was performed in HEK293T cells overexpressing ChREBP (Figure 6B). Comparative analysis identified 81 candidate ChREBP-related proteins that were specifically detected in the ChREBP-overexpressing group but absent in the IgG group (Table S4). Among these, SURF4 ranked second by normalized intensity, consistent with its known role in apolipoprotein secretion,33–35 and its interaction with ChREBP was validated by co-IP in primary hepatocytes and HEK293T cells (Figure 6C). ChREBP (carbohydrate response element binding protein) and SURF4 (surfeit 4) exist in a protein complex and inhibit ApoM (apolipoprotein M) secretion. A, GO terms enriched (|log2FC| > 0.585, false discovery rate (FDR) < 0.05) in liver of KISO (knockout with isoproterenol) vs isoproterenol (ISO). Significance of gene ontology (GO) term enrichment was determined by FDR < 0.05. B, Liquid chromatography - tandem mass spectrometry (LC-MS/MS) of ChREBP immune-precipitates from human embryonic kidney 293T (HEK293T) cells. C, Co-IP assays confirming the physical interaction between ChREBP and SURF4 in both mice primary hepatocytes and HEK293T cells. D and G, Phalloidin staining and cardiomyocyte area quantification. E and H, Western blot of atrial natriuretic peptide (ANP), brain natriuretic peptide (BNP), β-myosin heavy chain (β-MHC), and α smooth muscle actin (α-SMA), and ApoM expression in cardiomyocytes overexpressing ChREBP and SURF4. F and I, Concentration of secreted ApoM measured by ELISA. J, Co-IP of ChREBP-SURF4 in liver after ISO or transverse aortic constriction (TAC). K, Predicted structural model of the ChREBP-SURF4 complex. L. Mapping the ChREBP-SURF4 bind interface using individual point mutations (Mut1-Mut8); Mut7 (K229-A) significantly disrupts the binding. M, Phalloidin staining. N, Cardiomyocyte area quantification and ELISA analysis of ApoM levels. O, Schematic diagram. All quantitative data are expressed as mean ± SD, P values were determined by: 1-way ANOVA followed by Tukey multiple comparisons test for D, G, and M; Kruskal-Wallis test followed by Dunn multiple comparisons test for F, I, and N. P < 0.05 was considered statistically significant. Using a hepatocyte-cardiomyocyte co-culture system, we overexpressed ChREBP or ChREBP + SURF4 specifically in hepatocytes and assessed cardiomyocyte morphology and functional markers under basal and ISO conditions. Without ISO, neither overexpression affected cardiomyocytes (Figure 6D and 6E). However, ChREBP overexpression in hepatocytes moderately suppressed ApoM secretion, an effect that was reversed by SURF4 overexpression (Figure 6F). Under ISO, ChREBP overexpression worsened hypertrophy and reduced ApoM, although these effects were less pronounced due to ISO-induced baseline shifts. SURF4 co-overexpression reversed these effects on cardiomyocyte size, hypertrophic markers (Figure 6G and 6H), and ApoM (Figure 6I). ISO and TAC both enhanced hepatic ChREBP-SURF4 interaction (Figure 6J), suggesting SURF4 as a key limiting factor in ChREBP-mediated ApoM suppression under pathological stress. To elucidate the molecular basis of the interaction between ChREBP and SURF4, we used computational structural biology approaches. Molecular docking predicted a ChREBP-SURF4 binding interface (Figure 6K), and top-ranked models identified 8 candidate residues (R43, Q47, E50, Q51, N60, R119, K229, D238) for alanine mutagenesis (Figure S15A and S15B; Table S5). Only the K229A mutation significantly disrupted binding (Figure 6L). In hepatocyte-cardiomyocyte co-cultures with ISO, wild-type SURF4 limited cardiomyocyte enlargement and increased ApoM levels, whereas the K229A mutant further enhanced these effects (Figure 6M and 6N), indicating that ISO-induced ChREBP restricts endogenous SURF4 function, which is rescued by SURF4 overexpression. K229A SURF4, free from ChREBP binding, effectively supports ApoM transport and better ameliorates hypertrophy (Figure 6O). Thus, in cardiac remodeling, cytoplasmic ChREBP binds SURF4, inhibits its transport function, and impairs ApoM secretion. Liver-Derived ApoM Ameliorates Cardiac Remodeling by Targeting S1P/S1PR1 Pathway in Heart Studies have reported that ApoM binds to S1P and facilitates its transport as a component of HDL particles (Figure 7A). To investigate whether ApoM’s ameliorative effect on cardiac remodeling involves S1P/S1PRs, we measured serum S1P levels in ApoM-overexpressing ISO mice (Figure 5A) and L-ChREBP KO ISO/TAC mice (Figure 2A; Figure S12A). Serum S1P levels were significantly inhibited in ISO- and TAC-induced mice and in contrast, they were elevated in L-ChREBP KO mice and ApoM-overexpressing mice (Figure 7B). Among S1P receptors, only S1PR1 (sphingosine-1-phosphate receptor 1) was downregulated in remodeling models and rescued by ChREBP deletion or ApoM overexpression, with no changes in S1PR2 or S1PR3 (Figure 7C, S16A, B). ApoM overexpression did not alter HDL and LDL levels (Figure S17), indicating a functional enhancement of HDL cargo capacity without changing particle number. Liver-derived ApoM (apolipoprotein M) ameliorates cardiac remodeling by targeting the S1P (sphingosine-1-phosphate)/S1PR1 (sphingosine-1-phosphate receptor 1) axis. A, Schematic diagram of the ApoM-S1P pathway. B, Serum S1P levels measured by ELISA. C, Western blot of S1PR1, S1PR2, and S1PR3 expression. D, Phalloidin staining and cardiomyocyte area quantification. E, Western blot of S1P receptors and hypertrophic markers following treatment with isoproterenol (ISO), ApoM, and specific inhibitors for S1PR1 (W146), S1PR2 (JTE-013), and S1PR3 (CAY10444) in mice primary cardiomyocytes. F, The schedule of in vivo experiment. G, Representative M-mode echocardiograms and Doppler flow images. H, Quantitative analysis of left ventricular ejection fraction (LVEF), shortened left ventricular fraction (LVFS), relative wall thickness (RWT), E/A (early diastolic transmitral flow velocity to late diastolic transmitral flow velocity), and E/e′ (early diastolic transmitral flow velocity to early diastolic mitral annular velocity) ratios, Running distance, systolic blood pressure (SBP), diastolic blood pressure (DBP), heart weight-to-tibial length (HW/TL), heart weight-to-body weight (HW/BW), lung weight-to-body weight (LW/BW) ratios, and serum levels of creatine kinase (CK), LDH (lactate dehydrogenase), HBDH (hydroxybutyrate dehydrogenase), AST (aspartate aminotransferase) in mice. I, Wheat germ agglutinin (WGA) staining and cardiomyocyte area. J, Sirius Red staining and fibrosis area. n = 6. All quantitative data are expressed as mean ± SD, P values were determined by: Aligned rank transform (ART) test with Tukey post hoc test for B; Mann-Whitney U test for B; 1-way ANOVA followed by Tukey multiple comparisons test for D; Mann-Whitney U test for H through J. P < 0.05 was considered statistically significant. To clarify the roles of S1PR1, S1PR2, and S1PR3, we treated mice primary (Figure 7D and 7E) and HL-1 (Figure S18A and S18B) cardiomyocytes with specific inhibitors under ISO plus ApoM overexpression. S1PR1 inhibition (W146) abolished the protective effects of ApoM overexpression against hypertrophy and suppressed marker suppression (ANP, BNP, β-MHC, α-SMA), whereas S1PR2 and S1PR3 inhibitors (JTE-013/CAY10444) had no effect (Figure 7D and 7E; Figure S18A and S18B). S1PR1 is expressed in cardiomyocytes, endothelial cells, and fibroblasts.36–38 Using endothelial-(AAV9-CDH5-shS1PR1) and cardiomyocyte-specific (AAV9-cTNT-shS1PR1) knockdown combined with hepatic ApoM overexpression (AAV8-TBG-ApoM) in ISO-treated mice (Figure 7F), we found that simultaneous S1PR1 silencing abolished the protective effects of ApoM, as evidenced by worsened LVEF, LVFS, RWT, E/A, E/e’, running distance, SBP, DBP, HW/TL, HW/BW, LW/BW, serum CK, LDH, HBDH, and AST, as well as aggravated hypertrophy and fibrosis (Figure 7G through 7J). To evaluate the direct role of S1PR1 in myocardial remodeling in vivo, ISO-induced mice were treated with S1PR1 agonists FTY720 or ozanimod to determine whether they conferred ApoM-like cardioprotective effects (Figure S19A and S19B). Both agonists significantly improved cardiac function (increased LVEF, LVFS, and running endurance; decreased RWT, E/A, E/e′, SBP, DBP, HW/TL, HW/BW, LW/BW, serum CK, LDH, HBDH, and AST) and attenuated cardiomyocyte hypertrophy and fibrosis (Figure S19C through S19N). In conclusion, simultaneous S1PR1 knockdown in cardiomyocytes and endothelial cells provides compelling evidence that ApoM exerts cardioprotection against cardiac remodeling specifically through S1PR1, underscoring the critical importance of the ApoM/S1P/S1PR1 axis. HDL Containing ApoM Directly Ameliorates Pathological Remodeling in Mice The HDL paradox states that function, not high-density lipoprotein cholesterol (HDL-C) level, determines cardiovascular protection. Since ApoM mediates this effect but is present in only ≈5% of HDL, limited ApoM may explain the failure of HDL-C elevation. We injected HDL with (WT‑derived) or without (ApoM KO‑derived) ApoM into ISO‑induced WT and ApoM KO mice (Figure 8A). HDL+ApoM significantly reversed ISO‑induced declines in LVEF, LVFS, and running distance, and increases in RWT, E/A, E/e′, SBP, DBP, HW/TL, HW/BW, LW/BW, serum CK, LDH, HBDH, AST, as well as cardiomyocyte hypertrophy and fibrosis, in both genotypes (Figure 8B through 8H). ApoM KO did not aggravate ISO remodeling beyond WT, indicating sufficient ApoM suppression by ISO. However, in ApoM KO mice, HDL+ApoM (versus HDL−ApoM) rescued all parameters, demonstrating that supplementation with ApoM‑containing HDL improves remodeling independently of endogenous ApoM. Thus, during myocardial remodeling, reduced ApoM renders HDL−ApoM ineffective, whereas HDL+ApoM provides both HDL and ApoM to ameliorate remodeling, offering a promising therapeutic strategy (Figure 8I). ApoM (Apolipoprotein M)-containing high-density lipoprotein (HDL) can alleviate cardiac remodeling. A, Schedule of in vivo experiment. B, Representative M-mode echocardiograms and Doppler flow images. C, Quantitative analysis of left ventricular ejection fraction (LVEF), shortened left ventricular fraction (LVFS), relative wall thickness (RWT), E/A (early diastolic transmitral flow velocity to late diastolic transmitral flow velocity) ratio, E/e′ (early diastolic transmitral flow velocity to early diastolic mitral annular velocity) ratio and running distance. D, Systolic blood pressure (SBP) and diastolic blood pressure (DBP) levels in mice. E, Heart weight-to-tibial length (HW/TL), heart weight-to-body weight (HW/BW), and lung weight-to-body weight (LW/BW). F, Serum levels of creatine kinase (CK), LDH (lactate dehydrogenase), HBDH (hydroxybutyrate dehydrogenase), and AST (aspartate aminotransferase). G, Wheat germ agglutinin (WGA) staining and cardiomyocyte area. H, Sirius Red staining and fibrosis area. I, Schematic summary showing that hepatic ApoM-enriched HDL, rather than ApoM-deficient HDL, effectively inhibits the development of cardiac hypertrophy. n = 6. All quantitative data are expressed as mean ± SD. P values were determined by: Aligned rank transform (ART) test with Tukey post hoc test for C through H. P < 0.05 was considered statistically significant. Discussion This study demonstrates that hepatic (not cardiac) ChREBP upregulation drives cardiac remodeling; hepatocyte ChREBP knockout improves remodeling by enhancing ApoM secretion, as cytoplasmic ChREBP binds SURF4 to inhibit ApoM export independently of transcription. Hepatic ApoM overexpression rescues remodeling via HDL/S1P/S1PR1, highlighting therapeutic targeting of this axis. Liver-derived ApoM, secreted as an S1P/HDL complex, is protective in hypertension, atherosclerosis, and cardiomyopathy.39 Although HDL is cardioprotective,40 its levels are not always inversely linked to risk,41 and ApoM dysfunction may explain this failure. ApoM carries about 60% to 70% of plasma S1P, mediates vascular protection,42 and maintains HDL stability. S1PR1 agonists FTY720 and ozanimod confirmed protection in ISO-induced remodeling (Figure S19) but caused side effects (lymphopenia, bradycardia, immunosuppression),43–46 limiting their use. Upstream ApoM targeting to enhance S1P delivery may offer safer S1PR1 activation. The ApoM-S1P-HDL axis may have better predictive value than HDL concentration and warrants exploration as a therapeutic target in chronic heart failure. Our previous work established that hepatic ChREBP regulates glucose/lipid metabolism and insulin sensitivity via RTN4B (Nogo-B),47 and NADH/NAD+ (reduced nicotinamide adenine dinucleotide/nicotinamide adenine dinucleotide) ratio has been shown to modulate ChREBP activity.48 Cardiac ChREBP expression is extremely low, and its overexpression produced no remodeling effect, reinforcing the hepatic origin and liver-heart axis, although the mechanisms driving its hepatic upregulation and cytoplasmic accumulation under cardiac stress remain unclear. Although ChREBP is activated by carbohydrates and synergizes with SREBP1 in some tissues,49 this synergy is cell-type dependent; cardiac SREBP1 likely acts independently, given the low cardiac ChREBP expression and posttranslational restraints, and hepatic ChREBP affects the heart only indirectly via circulating lipids. Therefore, individuals under cardiac pressure overload should avoid excessive short-term carbohydrate intake, as additional ChREBP activation under already dysregulated redox conditions may impose a second hit on the heart and worsen injury. Previous studies have not addressed ChREBP in protein secretion, focusing instead on its transcriptional activation of target genes via ChoRE. Here we show that cardiac stress induces cytoplasmic accumulation of hepatic ChREBP, which binds to SURF4, an ER protein involved in ApoB secretion, at ER exit sites.33,34,50 Furthermore, impaired ER handling triggers ER stress and NADH/NAD+ imbalance, which further upregulate ChREBP. Consequently, this establishes a vicious cycle that progressively compromises liver function and disrupts metabolic homeostasis in the liver, heart, and circulation. This study has several limitations. While our study focused on hepatocytic ChREBP, other hepatic secreted factors and nonparenchymal cells may also participate. The lack of hepatocyte-specific ChREBP and ApoM double-KO mice, the lower efficiency of AAV-mediated knockdown, and the use of short-term models limit our conclusions; long-term effects and additional hepatic mediators warrant further investigation. The ChREBP-SURF4 pathway requires further characterization across tissues, cargo types, and disease dynamics. Our human analyses relied on peritumoral liver tissues, which may be confounded by inflammation, fibrosis, and metabolic reprogramming arising from the tumor microenvironment. Additionally, our clinical adjustments were limited to age and sex, leaving residual confounding possible. Therefore, the observed correlations should be interpreted as exploratory and associative, not causal. Independent validation in noncancer liver cohorts is warranted. In conclusion, our study indicates that hepatic ApoM protects against cardiac remodeling via HDL/S1P/S1PR1 signaling. Cardiac stress induces hepatic ChREBP, which binds SURF4 to block ApoM secretion, exacerbating cardiac hypertrophy and dysfunction. This liver-heart axis highlights ApoM as a potential therapeutic target for heart failure. ARTICLE INFORMATION Acknowledgments The authors thank Professor Yang Cao (University of Science and Technology of China) for the support with ultrasound technology and liver-secreted proteins; Professor Liang Zhang (Tianjin University) for assistance with molecular docking analysis; and the Clinical Medical Research Center of the Third Affiliated Hospital of Soochow University for providing ApoM (apolipoprotein M) KO mice. Author Contributions S. Zhang and Y. Duan contributed to conceptualization and methodology; S. Zhang, Z. Zhang, Z. Ma, Y. Han, T. Yan, W. Wu, L. Tang, S. Chen, J. Liu, and Z. Yin contributed to validation, visualization and investigation; S. Zhang, Z. Zhang, Z. Ma, Y. Han, and T. Yan conducted data curation; Y. Chen, D. Jian, J. Yang, and L. Ma provided resources; B. Sun contributed to software; S. Zhang, D. Kong, J. Ge, and Y. Duan performed supervision and project administration; S. Zhang, J. Ge, and Y. Duan were responsible for funding acquisition; S. Zhang, Z. Zhang and Z. Ma contributed to writing the original draft and formal analysis; S. Zhang, H. Chen, and Y. Duan performed writing—review and editing. Disclosures None. Supplemental Material Expanded Materials and Methods Tables S1–S8 Figures S1–S20 Results of Statistical Analysis of Band Density for Western Blot Major Resources Table Uncropped Western Blots References 51–58 Raw Data Set Supplementary Material