SNX3 mediates heart failure by interacting with HMGB1 and subsequently facilitating its nuclear-cytoplasmic translocation

Abstract

Sorting nexins (SNXs) as the key regulators of sorting cargo proteins are involved in diverse diseases. SNXs can form the specific reverse vesicle transport complex (SNXs-retromer) with vacuolar protein sortings (VPSs) to sort and modulate recovery and degradation of cargo proteins. Our previous study has shown that SNX3-retromer promotes both STAT3 activation and nuclear translocation in cardiomyocytes, suggesting that SNX3 might be a critical regulator in the heart. In this study we investigated the role of SNX3 in the development of pathological cardiac hypertrophy and heart failure. We generated abdominal aortic constriction (AAC) rat model and transverse aortic constriction (TAC) mouse model; hypertrophic neonatal rat cardiomyocytes (NRCMs) were induced by exposure to isoproterenol (10 μM). We showed that the expression of SNX3 was significantly upregulated in ISO-treated NRCMs and in the failing heart of AAC rats. Overexpression of SNX3 by intramyocardial injection of Ad-SNX3 induced heart failure in rats, and increased the susceptibility of NRCMs to ISO-induced myocardial injury in vitro. In contrast, conditional knockout of SNX3 in cardiac tissue in mice rescued the detrimental heart function in TAC mice, and knockdown of SNX3 protected against ISO-induced injury in NRCMs and AAC rats. We then conducted immunoprecipitation-based mass spectrometry and localized surface plasmon resonance, and demonstrated a direct interaction between SNX3-retromer and high mobility group box 1 (HMGB1), which mediated the efflux of nuclear HMGB1. Moreover, overexpression of HMGB1 in NRCMs inhibited the pro-hypertrophic effects of SNX3, whereas knockdown of HMGB1 abolished the protective effect of SNX3-deficiency. These results suggest that HMGB1 might be a direct cargo protein of SNX3-retromer, and its interaction with SNX3 promotes its efflux from the nucleus, leading to the pathological development of heart failure.

Introduction

The heart adapting to various stimulating factors by increasing cardiac muscle mass is defined as cardiac hypertrophy [1]. It is characterized by increased cell surface area, reactivated fetal genes, altered intracellular proteins and reconstructed cytoskeleton [2, 3]. Pathological cardiac hypertrophy resulting from continuous various types of stimulation eventually progresses to heart failure (HF), which is a transition stage of compensatory hypertrophy to the decompensatory phase [1, 4]. HF is the end-stage of various cardiovascular diseases and the leading cause of mortality. However, the molecular mechanism of cardiac hypertrophy and HF is still not very clear yet.

Sorting nexins (SNXs) can form the specific reverse vesicle transport complex (SNXs-retromer) with vacuolar protein sortings (VPSs) to sort and modulate recovery and degradation of cargo proteins [5, 6]. SNXs are involved in the pathogenesis of diverse diseases and serve as the key regulators for sorting cargo proteins. Among the SNX family members, SNX3 is the simplest subtype that it contains only the basic phox (PX) domain and directly combines with VPS26-VPS35-VPS29 to form the SNX3-retromer complex. It mediates the transport of some cargo proteins to the trans-Golgi network (TGN), cell nucleus, or plasma membrane (PM) [7–9]. Notably, SNX3 is indispensable for the secretion of the Wnt protein, which is involved in a signaling pathway that mediates cardiac hypertrophy [10, 11]. A previous study revealed that mRNA of SNX13 is highly expressed in both normal and end-stage failing human hearts [12]. However, the association between SNX3 and heart disease has not been fully explored. Recently, we found that SNX3 mediates hypertrophy and HF via promoting activation of signal transducer and activator of transcription 3 (STAT3) and its nuclear translocation [13], suggesting that SNX3 is a key transport protein in cardiomyocytes.

High mobility group box 1 (HMGB1) is a nonhistone nuclear factor involved in transcription regulation, DNA replication and repair, and nucleosome assembly [14]. Studies have reported that abnormal subcellular localization of HMGB1 is the pathological basis for the development of cardiac hypertrophy and HF. In cardiomyocytes, extracellular HMGB1 is a factor that mediates cardiac hypertrophy induced by pressure overload, whereas nuclear HMGB1 prevents cardiac hypertrophy and HF [15, 16]. Clinical studies have revealed that nuclear HMGB1 is significantly lower in the hearts of HF patients than in those of healthy subjects [15]. Furthermore, cardiac-specific overexpression of HMGB1 (HMGB1-TG) restores cardiac function compared with wild-type (WT) mice [15]. These findings indicate that nuclear HMGB1 plays an important role in protecting against cardiac hypertrophy. Our previous study also reported that the nuclear export of HMGB1 induced by poly ADP-ribose polymerase-1 (PARP1) was implicated in pathological myocardial hypertrophy [17]. However, the transportation mechanism has not been fully elucidated, and whether transport proteins mediate the translocation process is still unknown. In this study, we investigate whether SNX3 mediates the cytoplasmic transport of HMGB1 in cardiomyocytes, and uncover the potential role in the development of pathological cardiac hypertrophy and heart failure.

Materials and methods

Animals

Male Sprague-Dawley (SD) rats (weighing 200 ±  20 g) and one to three day-old SD rats were obtained from the Experimental Animal Center of Sun Yat-sen University (Guangzhou, China). Cardiac-specific SNX3 knockout mice were generated as described previously [13]. Briefly, SNX3-floxed mice and SNX3 transgenic mice were constructed in the Shanghai Model Organisms Center. SNX3^flox/+ mice were self-crossed to generate homozygous SNX3^flox/flox mice, which were crossed with C57BL/6J-Myh6^{em1(IRES-Cre)Smoc} mice (Shanghai Model Organisms Center, Stock No. NM-KI-00083. MGI ID: 97255) to generate SNX3^flox/flox-Myh6-Cre⁺ (SNX3-cKO) mice and SNX3^flox/flox littermate control (N-KO) mice. The mice were genotyped by PCR and further confirmed by Western blot analysis. All of the animal procedures were conducted under the institutional guidelines of the Animal Care and Use Committee and were approved by the Research Ethics Committee of Sun Yat-sen University. The experimental animals were housed, bred, and maintained in the specific pathogen-free (SPF) facility of the Experimental Animal Center of Sun Yat-sen University.

Recombinant adenoviral constructs and virus infection in vivo

Recombinant adenoviral vectors expressing rat SNX3 cDNA (Ad-SNX3, with Flag-tag) or short hairpin (sh)-SNX3 sequence (sh-SNX3, with Flag-tag), and control vectors, were constructed and purified as previously described [13]. For the gain-of-function experiment, 24 rats were randomly assigned to the sham or Ad-SNX3 group. Rats were anesthetized with sodium pentobarbital (30 mg/kg, Merck, Darmstadt, Germany) by intraperitoneal injection, followed by endotracheal intubation and heart exposure. Then multi-point (4–6 sites) injection of adenovirus (1 × 10¹⁰ PFU in 200 μL) into the left ventricular wall was performed to achieve SNX3 overexpression. The rats in sham group underwent a similar procedure in which the adenoviral vector was injected. For the loss-of-function experiments, 48 rats were randomly assigned to 4 groups including Sham+sh-NC, AAC+sh-NC, Sham + sh-SNX3 and AAC + sh-SNX3 groups. AAC surgery was conducted as described previously [18]. In brief, the abdominal aorta above the kidneys was exposed through a midline abdominal incision and constricted with a blunted 22-gauge needle using a 5-0 silk suture at 4–5 mm above the suprarenal artery. Subsequently, the needle was removed. The rats in sham group underwent an analogical wound but the abdominal aorta was not banded. Four weeks after AAC surgery, the rats were anesthetized with sodium pentobarbital (30 mg/kg) again, and sh-SNX3 (1 × 10¹⁰ PFU in 200 μL) was injected into 4–6 sites of the left ventricular wall, followed by chest closure under standard procedure. Eight weeks later, the rats were subjected to echocardiography assessment and then sacrificed.

Transverse aortic constriction (TAC) surgery

Wild-type and SNX3-cKO male mice at 8-week age were randomly separated into the sham and TAC groups. After being anesthetized with sodium pentobarbital (50 mg/kg) by intraperitoneal injection, the mice were fixed in the supine position and intubated with a 20-G catheter. The volume ventilator was performed with a tidal volume of 200 μL, 120 breaths/min, and 95% oxygen. Mouse skin was prepared by shaving and disinfection with 2% iodine. The chest cavity was opened via an incision in the left second intercostal space. The transverse aorta was dissected, and then a 6-0 silk suture was passed around the transverse aorta and tightened against a 27-G needle which was removed immediately to release the lumen. The chest cavity was sutured layer by layer, and the mice were observed until recovery. The same surgical procedure was used for the mice in the sham group, except that the transverse aorta was not ligated with a suture. Four weeks later, the mice were subjected to echocardiography assessment and then sacrificed.

Echocardiography analysis

Cardiac function of rats and mice was examined through echocardiography by Vevo 2100 Imaging System (VisualSonics, Toronto, Canada). Rats or mice were anesthetized with 4% or 2% (v/v) isoflurane (#R510-22, RWD, Shenzhen, China) in the anesthetic chamber and then transferred to the platform for echocardiography measurement with 2% or 1% isoflurane maintenance. Cardiac ventricular structure and systolic function were examined in two-dimensional-guided M-mode to record parasternal short-axis fields. Cardiac function parameters included left ventricular (LV) anterior wall thickness (LVAW), LV posterior wall thickness (LVPW), LV internal diameter (LVID), interventricular septum (IVS), and LV volume (LVV) at the end-diastolic and end-systolic stages. The ejection fraction (EF), fractional shortening (FS), cardiac output (CO), and stroke volume were also calculated.

Histological and immunohistochemistry (IHC) analysis

Hearts from rats and mice were fixed in 4% paraformaldehyde (Solarbio, Beijing, China) overnight and then embedded in paraffin for sectioning. Sections were stained with wheat germ agglutinin (WGA) or hematoxylin-eosin (H&E) for morphological analysis, Masson’s or picric sirius red (PSR) for collagen sediment assessments, respectively. For the IHC experiments, paraffin sections of myocardial tissue at a thickness of 5-mm were incubated with anti-SNX3 antibody (#K009182P, Solarbio, Beijing, China) or anti-HMGB1 antibody (#6893, Cell Signaling Technology, Beverly, MA, USA) overnight at 4 °C. Then, the samples were incubated with CoraLite488/594-conjugated anti-rabbit IgG (Proteintech, Wuhan, China), followed by counterstaining with DAPI (#D1306, Invitrogen, Carlsbad, CA, USA). Photomicrographs were captured with an EVOS FL Auto Cell Imaging System (Thermo Fisher Scientific, Waltham, MA, USA).

Culture of neonatal rat cardiomyocytes (NRCMs)

NRCMs were isolated from the hearts of 1 to 3-day-old SD rats as described previously [19]. Briefly, the ventricles were cut into 1 mm³ pieces and digested with 0.08% trypsin (#T4799, Sigma-Aldrich, St. Louis, MO, USA) at 37 °C for 9–12 times. The collected cells were suspended in Dulbecco’s modified Eagle’s medium (#12800-017, DMEM, Gibco, Grand Island, NY, USA) supplemented with 10% FBS (#A3160801, Gibco, Grand Island, NY, USA). After 1 h of incubation in a flask, the NRCMs were separated from the adherent fibroblasts. The purified NRCMs were seeded into dishes or plates, and cultured in DMEM containing 10% FBS and 100 μM 5-bromodeoxyuridine (5-BrdU) in an incubator at 37 °C with 5% CO₂.

Plasmid transfection, RNA interference, and adenovirus infection in NRCMs

The plasmid pcDNA3.1-6Flag-HMGB1 was constructed by inserting full-length cDNA of rat HMGB1 (NM_012963.4) into the pcDNA3.1 vector. Small RNAs (siRNAs) for HMGB1 and the negative control were purchased from RiboBio (Guangzhou, China), the sequences of which are listed in Supplementary Table S2. Transient transfection of HMGB1 plasmid and siRNAs in NRCMs was conducted with Lipofectamine 3000 reagent (#11668030, Invitrogen, Carlsbad, CA, USA) per the manufacturer’s instructions. Successful overexpression and knockdown of the HMGB1 gene at 48 h after transfection were determined by PCR and Western blot analysis, respectively (Supplementary Fig. S1, S2). For the overexpression or knockdown of SNX3 in NRCMs, Ad-SNX3 or sh-SNX3 was used to infect NRCMs at a multiplicity of infection (MOI) of 20 particles per 5 cells.

Quantitative real-time polymerase chain reaction (RT-qPCR)

Total RNA from cultured NRCMs was extracted using TRIzol reagent (#9109, Takara, Kyoto, Japan), and the concentration was measured with the Nanodrop 2000 fluorospectrometer (Thermo Fisher Scientific, Waltham, MA, USA). Reverse transcription (RT) of extracted RNA was performed using the RT kit (#K1622, Thermo Fisher Scientific, Waltham, MA, USA), and the PCR reaction was operated with SYBR-Green quantitative PCR kit (#0810, Toyobo, Osaka, Japan), followed by processing in the iCycler iQ system (Bio-Rad, Hercules, CA, USA) as previously described [20]. The β-actin gene served as the endogenous control. All PCR assays were performed in triplicate. The data were analyzed using the 2^−ΔΔCT method. The primers were synthesized by Sangon (Shanghai, China), and are listed in Supplementary Table S1.

Western blot

RIPA lysis buffer (#P0013B, Beyotime Biotechnology, Shanghai, China) with protease and phosphatase inhibitor cocktail (#04693159001, Roche, Basel, Switzerland) and Nucleus Extract Kit (#40010, Active Motif, Carlsbad, CA, USA) were used to extract total protein and nuclear protein, respectively. Western blot analysis was performed following a procedure as described previously [21]. Primary antibodies against HMGB1 (#10829-1-AP), SNX3 (#10772-1-AP), VPS26 (#15915-1-AP) and VPS35 (#10236-1-AP) were all obtained from Proteintech Group (Wuhan, China) and diluted at a concentration of 1:1000. Antibodies against α-tubulin (#sc-32293) and Lamin B1 (#sc-374015) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA) and were also diluted at a concentration of 1:1000. The enzyme horseradish peroxidase (HRP)-conjugated secondary antibodies diluted in 1:10,000 (#7074 and #7076, Cell Signaling Technology, Beverly, MA, USA) were applied to chemiluminescence detection, and the protein band intensities were quantified with LabWorks software (Bio-Rad, Hercules, CA, USA).

Co-immunoprecipitation (co-IP) analysis

Primary antibodies for co-IP were diluted as follows: anti-HMGB1 (rabbit, 1:100), anti-SNX3 (rabbit, 1:100), anti-VPS26 (rabbit, 1:100) and anti-VPS35 (rabbit, 1:100) and were all purchased from Proteintech Group (Wuhan, China). NRCMs treated with or without 10 μM isoproterenol (ISO, Tokyo Chemical Industry, Tokyo, Japan) were harvested with IP lysis buffer. Total protein (approximately 200 μg to 400 μg) was incubated at 4 °C with the corresponding primary antibodies overnight. Protein A/G-agarose beads (#20421, Pierce, Rockford, IL, USA) were then added into the mixture and co-incubated for another 4 h at 4 °C. After incubation, the samples were washed twice with IP buffers I, II and III in sequence. The samples were then lysed in loading buffer and boiled for 5 min. The corresponding IgG was used as the negative control. Western blot analysis was used to analyze the immunoprecipitated proteins. LabWorks software (Bio-Rad, Hercules, CA, USA) was used to analyze the protein band intensities.

Determination of the cell surface area

NRCMs were incubated in DMEM in 48-well plates or confocal dishes. The NRCMs were washed three times with phosphate buffer solution (PBS) after the DMEM was removed. NRCMs were then fixed with 4% paraformaldehyde at room temperature for 10 min, followed by permeabilization with 0.3% Triton X-100 for 10 min. Cells were incubated with rhodamine-phalloidin (#MAN0001777, Invitrogen, Carlsbad, CA, USA) for 30 min in the dark. Then the rhodamine-phalloidin mixture was removed and the cells were washed three times with PBS, followed by nuclear staining with DAPI for 5 min in the dark. The cells were photographed via a high-resolution screening system (Arrayscan VTi 600 Plus, Thermo Fisher Scientific, Waltham, MA, USA). A total of 50 fields in each group in each experiment (repeated in triplicate) were selected, and the cell surface area was determined via the built-in image analysis software.

Immunofluorescence (IF) assay

NRCMs were first fixed with 4% paraformaldehyde for 30 min, followed by three washes with PBS. Then, the cells were permeabilized with 0.3% Triton X-100 for 10 min in the dark. Subsequently, cells were incubated with 10% goat serum at room temperature for 2 h, and then incubated with primary antibodies overnight at 4 °C. The next day, the cells were further incubated with the fluorescence-conjugated secondary antibodies (#SA00013-2 and #SA00013-4, Proteintech, Wuhan, China) in the dark for 1 h. The nuclei were stained with DAPI, and the cells were observed via an EVOS FL Auto Imaging System (Life Technologies, Carlsbad, CA, USA).

Localized surface plasmon resonance (LSPR) assays

LSPR assays were conducted to detect whether there was direct interaction between HMGB1 and SNX3-retromer with an OpenSPR system (Nicoya Lifesciences, Kitchener, Canada). Firstly, the recombinant protein HMGB1 (#ATMP00116HU, AtaGenix, Wuhan, China) served as the ligand and was immobilized on a gold nanoparticle sensor chip. Then different concentrations of the recombinant proteins SNX3, VPS26 and VPS35 [13], respectively, were added to the sensor chip in running buffer (® ltered PBS) at a constant flow rate of 20 μL/min passing over the sensor (approximately 5 min). The quality of the chip surface changed when interaction occurred, and the signal was recorded. The main kinetic parameters include the dissociation constant (k_d), association constant (k_a), and binding constant (K_D). Of note, the K_D value is the ratio of k_d and k_a, and represents the equilibrium dissociation constant which is expressed in M (mol/L) units in this study. The smaller the K_D value is, the greater the binding affinity between proteins. The kinetic parameters were calculated on the basis of the fitting curve generated by the built-in software using the 1:1 binding model for determining the binding strength between two proteins.

Statistical analysis

GraphPad Prism 8.0 (GraphPad, San Diego, CA, USA) was used for the statistical analyses. All the data are expressed as mean ± standard error of the mean (SEM). Student’s t-test was performed to determine differences between two groups. One-way ANOVA analysis followed by the Bonferroni post hoc correction was conducted to examine differences among multiple groups. In all the cases, P value < 0.05 was considered to indicate statistical significance.

Results

SNX3 expression was up-regulated in ISO-treated NRCMs

ISO is a derivative of adrenaline that has been widely used as a hypertrophic stimulus [22]. In this study, NRCMs were exposed to 10 μM ISO for 12 h, and the cell surface area and mRNA levels of atrial natriuretic peptide (ANF), brain natriuretic peptide (BNP), and β-myosin heavy chain (β-MHC), which are the biomarkers of hypertrophic responses, were significantly increased (Fig. 1a, b). To explore the potential role of SNX3, we analyzed its mRNA and protein levels in NRCMs. Interestingly, SNX3 mRNA and protein levels were both increased in ISO-treated NRCMs (Fig. 1c, d), which indicated that SNX3 might be implicated in the pathogenesis of cardiac hypertrophy. Furthermore, we also found that ISO did not significantly influence the expression of HMGB1 (Fig. 1e), but could affect its subcellular distribution. As shown in Fig. 1f–h, HMGB1 shuttled from the nucleus to the cytoplasm at 6 h after ISO treatment, which was consistent with our previous results [17]. Taken together, these data indicate that SNX3 up-regulation and HMGB1 translocation might be associated with the development of ISO-induced cardiac hypertrophy.

Fig. 1. ISO promoted SNX3 expression and HMGB1 nuclear export in NRCMs.

NRCMs were exposed to 10 μM ISO for the indicated time. a Cell surface area was measured by an automatic high content system. Scale bar: 50 μm. b The mRNA levels of the hypertrophic biomarkers ANF, BNP, and β-MHC were determined via RT-qPCR. c mRNA level of SNX3 was determined by RT-qPCR. d SNX3 protein expression was detected by Western blot analysis. e HMGB1 protein expression at indicated time points was examined by Western blot analysis. f The protein levels of HMGB1 in cytoplasm and nucleus were detected by Western blot analysis. g, h IF assay was performed to observe the subcellular distribution of HMGB1, and the fluorescence was quantified. The results were normalized to α-tubulin or Lamin B1, and the data are presented as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001 vs. the control group, n = 3.

SNX3 induced heart failure in vivo

To further investigate the role of SNX3 in HF, we overexpressed SNX3 in SD rat hearts through intramyocardial administration of Ad-SNX3. H&E and WGA staining revealed larger myocardial cells in the hearts of the rats with SNX3 overexpression (Fig. 2a–c, f). Masson and PSR staining showed significant collagen fibrosis (Fig. 2d, e). Consistently, echocardiographic analysis revealed that SNX3 overexpression significantly decreased cardiac function, as indicated by reduced EF, FS, stroke volume, CO, LVIDs/LVIDd, and LVVs, as well as increased IVSs (Fig. 2h–o). In contrast, LVPW at the end-diastolic and end-systolic stages were not affected (Fig. 2p). Moreover, increased ratio of heart weight to tibial length (HW/TL) (Fig. 2g) and ANF, β-MHC protein expression (Fig. 2q, r) in Ad-SNX3-treated SD rat hearts also suggested the pro-hypertrophic role of SNX3 in the heart.

Fig. 2. SNX3 induced heart failure in vivo.

The recombinant adenovirus vector encoding SNX3 (Ad-SNX3) was transferred into SD rat hearts by intramyocardial injection, while the empty vector carrying the Flag-tag was injected into the control group animals. Echocardiography parameters and heart gross were presented. a Gross observation of heart morphology. b, c H&E stained sections of the whole heart and left ventricle. d, e Masson and PSR-stained transverse sections of the left ventricle. f The left ventricle sections of rat hearts were stained with malt germ agglutinin (WGA). g Heart weight/tibia length (HW/TL) was performed in postmortem measurements. h Representative echocardiographic images of the two groups. i–p EF, FS, stroke volume, CO, LVIDd, LVIDs, LVVs, LVVd, IVSs, IVSd, LVPWs, and LVPWd were detected and calculated. q, r The protein levels of ANF, β-MHC, and SNX3 were detected by Western blot with quantitative analysis. Scale bars: 2 mm (b), 100 μm (c–e) or 50 μm (f). Data are presented as mean ± SEM, n = 12 (g–p) or 3 (q, r). *P < 0.05, **P < 0.01, ***P < 0.001 vs. the control group.

SNX3 knockdown attenuated AAC-induced HF in rats

The effect of SNX3 was further investigated via a loss-of-function approach in rats. Abdominal aorta ligation was performed on SD rats to induce AAC, whereas the control groups received sham surgery without artery ligation. Then rats in each group were injected with sh-SNX3 or empty vector at the ventricular wall by multi-point administration. Pathological analysis showed that 8 weeks of AAC induced cardiomyocyte enlargement, increased thickness of the ventricular wall and collagen synthesis (Fig. 3a–f), but these effects were inhibited by knockdown of SNX3. Echocardiography results showed a significant decrease in EF, FS, stroke volume, CO, and an increase in HW/TL, LVID, IVS, LVV and LVPW both at end-diastole and at end-systole in the AAC group compared with the sham group (Fig. 3h–p). Interestingly, these parameters were significantly reversed by cardiac knockdown of SNX3. In addition, SNX3 knockdown reversed the increase in HW/TL and hypertrophic biomarkers that were induced by ACC (Fig. 3g, q, r). Of note, SNX3 knockdown in rats with sham surgery had no significant influence on heart function.

Fig. 3. Adenovirus-mediated SNX3 depletion ameliorated heart failure caused by AAC.

SD rats were subjected to AAC, whereas the control groups received sham surgery without artery ligation. Then, the rats received adenoviral vector encoding SNX3 interference sequence (sh-SNX3), or empty vector (sh-NC) via intramyocardial injection. a Gross observation of heart morphology. b, c H&E stained sections of the whole heart and left ventricle. d Masson-stained transverse sections of the left ventricle. e Left ventricle sections of rat hearts were stained with malt germ agglutinin (WGA). f Cardiomyocyte size of LV. g HW/TL was performed in postmortem measurements. h Representative echocardiographic images of each group. i–p EF, FS, stroke volume, CO, LVIDd, LVIDs, LVVs, LVVd, IVSs, IVSd, LVPWs, and LVPWd were all detected and calculated. q, r The protein levels of ANF, β-MHC and SNX3 were detected by Western blot with quantitative analysis. Scale bars: 2 mm (b), 100 μm (c, d) or 50 μm (e). Results are presented as mean ± SEM. *P < 0.05, ***P < 0.001 vs. the sham surgery plus control empty vector group; ^#P < 0.05, ^##P < 0.01, ^###P < 0.001 vs. the AAC plus control empty vector group, n = 12 (g–p) or 3 (q, r).

Cardiac-specific knockout of SNX3 alleviated TAC-induced HF in mice

To better delineate the importance of cardiac SNX3 expression in the development of chronic heart failure, we performed TAC surgery, a well-characterized heart failure mouse model, with SNX3-cKO mice. The cardiac-specific deletion of SNX3 in SNX3-cKO mice was mediated by hybridization of SNX3^flox/flox mice with transgenic mice expressing Cre recombinase under the control of the α-MHC promoter (C57BL/6J-Myh6^{em1(IRES-Cre)Smoc} mice). Compared with the control (N-KO) sham group, cardiac deficiency of SNX3 (without TAC surgery) did not affect the cardiac morphology and function, whereas TAC for 8 weeks significantly induced cardiac hypertrophy and heart failure in N-KO mice (Fig. 4). Interestingly, SNX3-cKO mice were more resistant to TAC-induced heart failure, as evidenced by preserved cardiac function after TAC. As shown in Fig. 4a–h, deficiency of cardiac SNX3 expression reduced the cardiomyocyte size and cardiac collagen deposition induced by TAC. With echocardiography (Fig. 4i), we found that TAC led to obviously decreased EF (Fig. 4l), FS (Fig. 4m), LVAW (Fig. 4n) and LVPW (Fig. 4o), and elevated LVID (Fig. 4p) and LVV (Fig. 4q) both at end-diastole and at end-systole in N-KO mice, but the effect was significantly reversed in SNX3-cKO mice. Consistently, HW/BW (Fig. 4j), HW/TL (Fig. 4k), as well as protein expression of β-MHC and ANF (Fig. 4r, s) were also mitigated in SNX3-cKO mice subjected to TAC.

Fig. 4. Cardiac-specific knockout of SNX3 rescued heart failure induced by TAC.

N-KO and SNX3-cKO mice at 8-week age underwent TAC, whereas the control groups received sham surgery without artery ligation. a Gross observation of heart morphology. b, c H&E stained sections of the whole heart and left ventricle. d, e Masson and PSR-stained transverse sections of the left ventricle. f The left ventricle sections of rat hearts were stained with malt germ agglutinin (WGA). g, h Interstitial collagen volume and cardiomyocyte size of LV. i Representative echocardiographic images of each group. j The ratio of whole-heart weight (HW) to body weight (BW). k HW/TL was determined in postmortem measurements. l–q EF, FS, LVAWs, LVAWd, LVPWs, LVPWd, LVIDs, LVIDd, LVVs, and LVVd were all detected and calculated. r, s The protein levels of ANF, β-MHC and SNX3 were detected by Western blot with quantitative analysis. Scale bars: 2 mm (b) or 50 μm (c–f). Results are presented as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001 vs. the sham surgery plus N-KO group; ^#P < 0.05, ^##P < 0.01, ^###P < 0.001 vs. the TAC plus N-KO group, n = 8 (g–i, l–q) or 9–11 (j, k) or 3 (r, s).

The pro-hypertrophic effect of SNX3 was mediated by HMGB1

In the in-vitro study, SNX3 overexpression not only triggered hypertrophic response, but also aggravated ISO-induced hypertrophy in NRCMs, as indicated by increased cell surface area (Fig. 5a) and elevated expression of hypertrophic marker genes (Fig. 5b, c). In contrast, SNX3 knockdown significantly attenuated ISO-induced hypertrophic responses (Fig. 5d–f). HMGB1 is reported to be implicated in cardiac hypertrophy and HF due to its nuclear and cytoplasmic distribution in heart tissues [15, 23], and HMGB1 in the nucleus serves as an anti-hypertrophy factor [16, 23]. Herein, we confirmed that overexpression of HMGB1 alleviated ISO-induced hypertrophic responses (Supplementary Fig. S1), whereas siRNA-mediated down-regulation of HMGB1 aggravated ISO-induced hypertrophic responses (Supplementary Fig. S2). To investigate whether HMGB1 was involved in the pro-hypertrophic effect of SNX3, HMGB1 co-knockdown and co-overexpression with SNX3 was conducted in NRCMs, respectively. As shown in Fig. 5d–f, knockdown of HMGB1 significantly counteracted the protective effect induced by SNX3 silencing. Consistently, increased HMGB1 expression in NRCMs suppressed the hypertrophic response induced by SNX3 and ISO (Fig. 5a–c). These data suggested that the pro-hypertrophic action of SNX3 might be mediated by HMGB1.

Fig. 5. SNX3 aggravated ISO-induced cardiomyocyte hypertrophy and the effect was mediated by HMGB1.

a–c NRCMs were transfected with pcDNA3.1-HMGB1 plasmid or pcDNA3.1 vector with Lipofectamine 3000 reagent in Opti-MEM. Six hours later, the NRCMs were cultured in DMEM supplemented with 10% FBS and infected with Ad-SNX3 or Ad-flag control. Then, the cells were treated with or without 10 μM ISO at 24 h and incubated for another 24 h. d–f Similarly, NRCMs were transfected with siRNA of HMGB1 or NC siRNA, followed by infection with sh-SNX3 or sh-NC control 6 h later. Then, the cells were treated with or without 10 μM ISO for 24 h incubation. The cell surface area (a, d), mRNA levels (b, e) and protein levels (c, f) of hypertrophic marker genes were determined. Scale bar: 50 μm. Results are presented as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001 vs. control empty vector group; ^#P < 0.05, ^##P < 0.01, ^###P < 0.001 vs. the control empty vector plus ISO treatment group; ^&P < 0.05, ^&&P < 0.01, ^&&&P < 0.001 vs. the adenovirus encoding SNX3 plus ISO treatment group, n = 3.

SNX3-retromer bound with HMGB1 as its cargo protein

Given that SNX3-retromer complex sorts cargo proteins in intracellular tracking, we further explored whether SNX3-retromer participated in HMGB1 transport. Interestingly, IP-MS revealed the potential direct interaction between SNX3-retromer and HMGB1 (Fig. 6a). Further co-IP results confirmed that HMGB1 could interact with SNX3, VPS26 and VPS35, respectively (Fig. 6b). IF assays also elucidated the co-localization of HMGB1 and SNX3-retromer (Fig. 6c, e, g). Furthermore, LSPR assay revealed that the binding constants (K_D values) of HMGB1 interacting with SNX3, VPS26 and VPS35 were 14.3 nM, 7.03 μM, and 7.94 nM, respectively (Fig. 6d, f, h), suggesting potential interactions between HMGB1 and the recombinant proteins [24]. However, the binding affinity of VPS26 for HMGB1 was lower than that of either SNX3 or VSP35. Taken together, these findings suggest that HMGB1 can bind to SNX3 retromer in ISO-treated NRCMs.

Fig. 6. HMGB1 directly interacts with SNX3-retromer and serves as a cargo protein of the complex.

a IP-MS was performed to screen possible cargo proteins of the SNX3-retromer complex. b Co-IP was performed to determine the interaction between HMGB1 and the SNX3-retromer complex in NRCMs transfected with HMGB1 plasmid with or without ISO treatment. c, e, g IF assay was conducted to detect the co-localization of HMGB1 and the SNX3-retromer complex. Scale bar: 20 μm. d, f, h Proteins were purified, and LSPR assay was applied to probe the direct interaction between HMGB1 and the SNX3-retromer complex.

SNX3-retromer promoted nuclear export of HMGB1

To better understand the mechanism by which HMGB1 mediated the pro-hypertrophic effect of SNX3, we studied the role of SNX3 in regulating HMGB1. Since HMGB1 translocation in cardiomyocytes was critically implicated in cardiac hypertrophy, we explored the potential mechanism by which SNX3 regulates HMGB1 transport. Interestingly, overexpression of SNX3 evidently promoted the nuclear-cytoplasmic translocation of HMGB1, as indicated by decreased HMGB1 protein in the nucleus but increased HMGB1 protein in the cytoplasm (Fig. 7a). Consistently, SNX3 knockdown in NRCMs inhibited ISO-induced HMGB1 shuttling (Fig. 7c). The translocation effect of HMGB1 caused by SNX3 was further confirmed by IF assay (Fig. 7b, d). In addition, IHC and IF assay of left ventricular sections also showed that adenovirus-mediated SNX3 overexpression caused less HMGB1 protein maintenance in the nucleus (Fig. 7e, f, g). Notably, the mRNA level (Supplementary Fig. S2b), protein level (Supplementary Fig. S2c, d, f), and extracellular secretion of HMGB1 (Supplementary Fig. S2e) were not affected by SNX3 in NRCMs.

Fig. 7. SNX3 inhibited HMGB1 accumulation in the nucleus both in vitro and in vivo.

NRCMs were infected with Ad-SNX3 for SNX3 overexpression, whereas SNX3 depletion was mediated by sh-SNX3 infection, followed by incubation with 10 μM ISO for 12 h subsequently. a, c Western blot analysis was conducted to measure the nuclear and cytoplasmic HMGB1. b, d IF assay was performed to observe subcellular distribution of HMGB1. Results are presented as mean ± SEM. *P < 0.05, **P < 0.01 vs. empty vector group; ^#P < 0.05, ^##P < 0.01 vs. the empty vector plus ISO treatment group, n = 3. SD rats were subjected to the intramyocardial injection with Ad-SNX3, whereas control group received the empty vector. e Results from the IHC assay of the SD heart sections are shown. f, g IF assays of both HMGB1 and SNX3 in the heart sections from SD rats. Scale bars: 50 μm (b, d) or 40 μm (e–g).

Discussion

Subcellular localization of HMGB1 has been widely explored in cardiac hypertrophy [25–27]. However, whether the transport process is mediated by specific transport proteins is still unknown. Our study aimed to investigate the role of SNX3 in cardiac hypertrophy and heart failure as well as its regulatory effect on HMGB1 in cardiomyocytes. Firstly, we found that SNX3 expression was increased after ISO treatment. Then the gain-of-function in vivo and in vitro revealed that SNX3 induced myocardial injury. In contrast, SNX3 deficiency not only attenuated AAC- or TAC-induced myocardial injury and heart dysfunction, but also inhibited ISO-stimulated myocyte hypertrophy. Of note, compared with the negative sham group, SNX3 overexpression in the rat myocardium for one month significantly reduced LVIDs and LVIDd (Fig. 2m), but these parameters were not influenced by knockdown of SNX3 (Figs. 3m, 4p). Instead, 8 weeks of AAC induced elevated LVIDs and LVIDd, which could be reversed by knockdown of SNX3 (Figs. 3m, 4p), indicating that SNX3 deficiency might affect only the pathological progression of heart failure in AAC rats, but not healthy hearts.

A previous study suggested that SNX3 may be a potential factor in regulating cardiac function, with significantly higher expression than other subtypes except SNX1 [12]. It is reported that SNXs participate in the disease process by sorting and transporting certain signaling molecules [28, 29]. For example, SNX27 mediates the binding and retrieval of β2 adrenal receptors (β2-AR) to the retromer complex, which was involved in several disease process [30]. In addition, SNX13 depletion mediates heart failure through degradative sorting of apoptosis repressors [12]. Furthermore, SNX3 has been shown induce heart dysfunction via promoting STAT3 activation and nuclear translocation [13]. However, whether SNX3 mediates other cargos in cardiac hypertrophy or heart failure is still unknown. In this study, we confirmed that SNX3 was the potential transport protein for HMGB1 in cardiomyocytes, as evidenced by the specific interaction between SNX3-retromer and HMGB1, which might promote the nuclear export of HMGB1 both in vivo and in vitro.

SNX proteins and VPS26-VPS35-VPS29 heterotrimers are indispensable for cargo recognition, sorting and binding. It has been reported that not all SNXs interact with the retromer complex, indicating diverse roles of SNX family members [28]. The SNX3-retromer complex comprises SNX3, VPS26 and VPS35, which is recognized as a T-shaped architecture [7] and plays a key role in sorting cargoes, and SNX3 activity is required for Wingless/Wnt secretion and iron uptake mediated by transferrin [11, 31]. In addition, SNX3 inhibits the transport of epidermal growth factor receptor (EGFR) into lysosomes [32]. In this study, we identified HMGB1 as a new cargo of the SNX3-retromer complex in cardiomyocytes.

To date, SNXs that mediate the nuclear transport of intracellular cargoes have rarely been reported. Joyal et al. found that SNX11 mediated factor II receptor-like 1 (F2rl1) shuttling from PM to nucleus in retinal ganglion cells [33]. Besides, SNX3 is also found to mediate the transport of STAT3 to the nucleus [13]. Herein, by studying the association between SNX3 and HMGB1, we further confirmed the role of SNX3 in mediating protein transport, which also contributes to the development of cardiac hypertrophy and HF. Further study is still required to elucidate the detailed dynamic process and potential mechanism of SNX3-associated nuclear export of HMGB1.

In conclusion, our study revealed that SNX3 induced cardiomyocyte hypertrophy and heart failure, and knockdown of SNX3 significantly inhibited ISO-induced hypertrophy and AAC-induced heart failure. In addition, cardiac-specific deletion of SNX3 critically improved heart function in TAC mice. The pro-hypertrophic effect of SNX3 was at least partly mediated by the nuclear-cytoplasmic translocation of HMGB1, which might be associated with the interaction between SNX3-retromer and HMGB1. These results suggest that SNX3 could be a potential therapeutic target for cardiac hypertrophy and HF.

Author contributions

JL and PQL conceived the study and revised the manuscript. HL and MXP designed and performed majority of the experiments and data analyses. RXY, JXC, YMW, PXW, YHH and DYP performed several in-vitro experiments. HL, MXP and JL wrote the manuscript. PQL provided scientific advice. All authors read and approved the manuscript.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains Supplementary Material available at 10.1038/s41401-024-01436-z.