1. Human Biology and Medicine
Download icon

Intermittent hypoxia mediated by TSP1 dependent on STAT3 induces cardiac fibroblast activation and cardiac fibrosis

  1. Qiankun Bao  Is a corresponding author
  2. Bangying Zhang
  3. Ya Suo
  4. Chen Liu
  5. Qian Yang
  6. Kai Zhang
  7. Ming Yuan
  8. Meng Yuan
  9. Yue Zhang
  10. Guangping Li  Is a corresponding author
  1. The Second Hospital of Tianjin Medical University, China
  2. Peking University People's Hospital, China
Research Article
  • Cited 0
  • Views 493
  • Annotations
Cite this article as: eLife 2020;9:e49923 doi: 10.7554/eLife.49923

Abstract

Intermittent hypoxia (IH) is the predominant pathophysiological disturbance in obstructive sleep apnea (OSA), known to be independently associated with cardiovascular diseases. However, the effect of IH on cardiac fibrosis and molecular events involved in this process are unclear. Here, we tested IH in angiotensin II (Ang II)-induced cardiac fibrosis and signaling linked to fibroblast activation. IH triggered cardiac fibrosis and aggravated Ang II-induced cardiac dysfunction in mice. Plasma thrombospondin-1 (TSP1) content was upregulated in both IH-exposed mice and OSA patients. Moreover, both in vivo and in vitro results showed IH-induced cardiac fibroblast activation and increased TSP1 expression in cardiac fibroblasts. Mechanistically, phosphorylation of STAT3 at Tyr705 mediated the IH-induced TSP1 expression and fibroblast activation. Finally, STAT3 inhibitor S3I-201 or AAV9 carrying a periostin promoter driving the expression of shRNA targeting Stat3 significantly attenuated the synergistic effects of IH and Ang II on cardiac fibrosis in mice. This work suggests a potential therapeutic strategy for OSA-related fibrotic heart disease.

Introduction

The prevalence of sleep apnea is high in patients with cardiovascular disease, which could be caused by multiple risk factors including respiratory instability, obesity and upper airway dysfunction (Javaheri et al., 2017; Kasai et al., 2012). Sleep apnea, classified as obstructive sleep apnea (OSA) and central sleep apnea, affects or covaries with numerous health outcomes and physiological processes, particularly cardiovascular disease (Javaheri et al., 2017; Kasai et al., 2012; Somers et al., 2008). A recent prospective study of OSA found it associated with increased risk of incident heart failure in community-dwelling middle-aged and older men (Gottlieb et al., 2010). OSA is associated with a profile of perturbations that include sympathetic nerve activity, metabolic dysregulation, inflammation, oxidative stress, vascular endothelial dysfunction, and intermittent hypoxia (IH), all of which are critical for the pathogenesis of coronary heart disease, hypertension, atrial fibrillation, and heart failure (Javaheri et al., 2017; Kasai et al., 2012). IH is a hallmark of OSA and initiates several mechanisms that lead to cardiac fibrosis and cardiac dysfunction (Baguet et al., 2012; Wei et al., 2016).

Cardiac fibrosis is characterized by excessive deposition of extracellular matrix proteins by cardiac fibroblasts (CFs), which respond to pathological stress and environmental stimuli by transforming into myofibroblasts. Activated CFs express elevated levels of various proinflammatory and profibrotic factors causing fibroblast proliferation, migration and scar formation (Travers et al., 2016). The persistence of myofibroblasts eventually results in adverse changes in ventricular structure and compliance, thereby leading to heart failure (Baguet et al., 2012; Travers et al., 2016). However, the effect and mechanism of CF activation in IH-induced cardiac remodeling remain unclear.

One central cytokine linked to cardiac fibrosis is transforming growth factor β (TGFβ), which is highly regulated at the level of activation (Travers et al., 2016; Meng et al., 2016). The activation of latent pro-TGFβ requires thrombospondin-1 (TSP1, encoded by Thbs1 gene), which is a matricellular glycoprotein and can be secreted by various cell types, to remove its latency-associated propeptide (Meng et al., 2016; Crawford et al., 1998; Adams and Lawler, 2011). Myocardial TSP1 expression was increased in a mouse model of pressure overload because of transverse aortic constriction (Xia et al., 2011), and blocking TSP1-dependent TGFβ activation prevented cardiac fibrosis progression and improved cardiac function (Belmadani et al., 2007). However, the role and underlying mechanism of TSP1 in IH-induced CF activation and cardiac fibrosis remain to be elucidated.

As a member of the signal transducer and activator of transcription (STAT) protein family, STAT3 was originally identified as an interleukin-6–activated transcription factor. It can also be phosphorylated by receptor-associated Janus kinase (JAK) in response to growth factor and hemodynamic stress, thus acting as a regulator in fundamental cellular processes including inflammation, cell growth, proliferation, differentiation, migration, and apoptosis (Wei et al., 2003; Chakraborty et al., 2017; He et al., 2018). Emerging evidence demonstrates that STAT3 signaling is hyperactivated in fibrotic diseases, which may be an important molecular checkpoint for tissue fibrosis (Chakraborty et al., 2017; Su et al., 2017). Recent study demonstrated that STAT3 can drive TSP1 expression in astrocytes (Tyzack et al., 2014). Given the integrated function of STAT3 activation in inflammation and fibrosis, we hypothesized that IH-induced STAT3 activation might play a crucial role in CF activation and cardiac fibrosis by increasing TSP1 expression.

In the present study, we investigated the effect of IH exposure on cardiac fibrosis in response to angiotensin II (Ang II) in mice and the potential underlying mechanism. TSP1 expression induced by IH in CFs, mediated by phosphorylation of STAT3 at Tyr705, was involved in CF activation and cardiac fibrosis. Pharmacological or genetic inhibition of STAT3 restrained IH-induced CF activation and cardiac fibrosis and ameliorated IH-induced cardiac dysfunction.

Results

IH induced cardiac fibrosis and aggravated Ang II-induced cardiac dysfunction in mice

Most respiratory events of patients with OSA result in desaturation–reoxygenation sequences that cause IH (Baguet et al., 2012). To investigate IH exposure to cardiac function, we housed mice under IH or normoxia for 28 days (Figure 1A). Hypoxia in heart tissue was evaluated by using pimonidazole (Figure 1—figure supplement 1A). IH exposure slightly increased the ratio of heart weight to tibial length (Figure 1B). Echocardiography analysis revealed a moderate decrease in ejection fraction (EF) and fractional shortening (FS) with IH as compared with normoxia (Figure 1C–D). Furthermore, Masson and Sirius red staining demonstrated mildly larger fibrosis area in the heart of mice after IH exposure (Figure 1E–F).

Figure 1 with 1 supplement see all
Intermittent hypoxia (IH) induces cardiac fibrosis and aggravates pathological cardiac dysfunction by promoting fibroblast activation in myocardial interstitium.

(A) C57BL/6 mice were housed under normoxia or IH with or without infusion of angiotensin II (Ang II) for 28 days. (B) Ratio of heart weight to tibial length of mice in each group. (C, D) Ejection fraction (EF) and fractional shortening (FS) of mice quantified by echocardiography. (E) Representative images of Masson and Sirius red staining of left ventricle. Scale bar, 100 μm. (F) Quantification of fibrotic area in (E). (G) Representative confocal microscopy images of immunofluorescence staining for α-SMA, vimentin (VIME, fibroblast marker), collagen I (Col1), periostin and DAPI. Scale bar, 20 μm. (H) Quantification of co-localization of α-SMA and VIME, Col1 and periostin fluorescence intensity in (G). (I) Col1, periostin, and α-SMA protein level in left ventricle detected by western blot analysis. (J) Quantification of Col1, periostin and α-SMA protein level in (I). (K) Quantification of Acta2, col1a1 and Postn mRNA expression. Data are mean ± SEM, n = 6 mice per group, *p<0.05, 2-way ANOVA with Bonferroni post-test. Data used for quantitative analyses as well as the numerical data that are represented in graphs are available in Figure 1—source data 1.

To investigate the effect of IH associated with cardiac function after injury, we used the mouse model of Ang II-induced cardiac hypertrophy to represent cardiac fibrotic responses. Consistent with previous studies (Su et al., 2017; Schafer et al., 2017), Ang II treatment induced cardiac fibrosis with increasing the ratio of heart weight to tibial length and fibrosis area and impaired EF and FS (Figure 1B–D). Strikingly, IH exposure further impaired cardiac function, with increasing the ratio of heart weight to tibial length and decreasing EF and FS (Figure 1B–D). IH exposure significantly enlarged the fibrotic area of heart as well (Figure 1E–F). IH exposure had little effect on myocyte size or ratio of apoptotic cells in the left ventricle with or without Ang II challenge (Figure 1—figure supplement 1B–C). These results indicated that IH could induce cardiac fibrosis both at the basal level and in response to Ang II and also aggravated Ang II-induced cardiac dysfunction.

IH induced fibroblast activation in myocardial interstitium

CFs are now recognized for their fundamental contributions to the heart’s response to various forms of injury (Travers et al., 2016). To investigate the effect of IH on fibroblast activation, we analyzed the expression of α-SMA, collagen I and periostin in the left ventricle of mice after IH exposure. First, immunofluorescence staining showed strongly α-SMA signaling in vimentin-positive CFs in the IH group. IH-induced α-SMA expression in CFs was further enhanced under the Ang II-induced cardiac pathologic condition (Figure 1G–H). Meanwhile, the expression of collagen I and periostin showed a similar trend as α-SMA expression (Figure 1G–H). In addition, IH exposure slightly increased the mRNA and protein levels of α-SMA, collagen I and periostin (Figure 1I–K) in the mouse left ventricle and significantly elevated the levels in the left ventricle of Ang II-infused mice (Figure 1I–K). These results suggest that IH can activate CFs at the basal level and under the Ang II-induced injury condition.

IH exposure promoted TSP1 expression in CFs

To test whether TSP1 plays an important role in the progression of cardiac fibrosis induced by IH, we next detected TSP1 expression in mice after IH exposure. As compared with normoxia, IH exposure significantly increased plasma TSP1 content and mRNA level in the mouse left ventricle, especially in Ang II-infused mice (Figure 2A–B). The protein level of TSP1 in the left ventricle showed a similar trend to that of collagen I and periostin (Figure 2C–D). Given that TGFβ and its downstream Smad signaling critically modulate the fibroblast phenotype (Travers et al., 2016; Cucoranu et al., 2005), we assessed Smad2/3 phosphorylation in the mouse left ventricle. The phosphorylation of Smad2/3 was slightly increased after IH exposure but greatly increased after IH exposure in the Ang II-induced injury condition (Figure 2C–D), which is consistent with TSP1 expression. Furthermore, immunofluorescence staining showed elevated signaling of TSP1 in the left ventricle after IH, especially in the Ang II-induced condition (Figure 2E–F). In addition, the expression of TSP1 in cells positive for Thy1 (an inclusive surface protein given its association with CFs [Ali et al., 2014; Hudon-David et al., 2007]) and negative for CD11b (macrophage marker), CD31 (endothelial cell marker), CD45 and Ter119 (hematopoietic cell marker) in the heart, detected by flow cytometry, was significantly higher after IH exposure (Figure 2G), which indicated that IH exposure could induce TSP1 expression in CFs.

Increased thomsbospondin-1 (TSP1) expression in mice after IH exposure.

(A) C57BL/6 mice were housed under normoxia or IH with or without infusion of Ang II for 28 days. Plasma concentration of TSP1 in mice detected by ELISA. (B) Thbs1 mRNA level in left ventricle of mice exposed to IH with or without infusion of Ang II for 14 days quantified by RT-PCR. (C) C57BL/6 mice were housed under normoxia or IH with or without infusion of Ang II for 28 days. The protein levels of TSP1, phosphorylated Smad2/3 (p-Smad2/3), and total Smad2/3 (t-Smad2/3) in left ventricle of mice detected by western blot analysis. (D) Quantification of TSP1 and p-Smad2/3 in (C). (E) Representative confocal microscopy images of immunofluorescence staining for TSP1 and DAPI. Scale bar, 20 μm. (F) Quantification of TSP1 fluorescent intensity in (E). Data are mean ± SEM, n = 6 mice per group, *p<0.05, 2-way ANOVA with Bonferroni post-test. (G) Quantification of number of cells negative for CD45, CD31, CD11b, and Ter119 and positive for Thy1 from left ventricle tissue digestion, that stained positive for intracellular TSP1 by flow cytometry. Data are mean ± SEM, n = 6, *p<0.05, unpaired 2-tail t test. (H) Plasma concentration of TSP1 in healthy individuals and patients with obstructive sleep apnea (OSA) detected by ELISA. Data are mean ± SEM, n = 21, *p<0.05, unpaired 2-tail t test. Data used for quantitative analyses as well as the numerical data that are represented in graphs are available in Figure 2—source data 1.

To identify the potential relevance of TSP1 to OSA, we measured the concentration of TSP1 in plasma samples from 21 patients with OSA and 21 healthy subjects. TSP1 content was significantly increased in patients with OSA as compared with healthy individuals (Figure 2H). Together, these data suggest that TSP1 level was elevated in both mice and humans after IH exposure and might contribute to CF activation and cardiac fibrosis.

IH induced primary CF (PCF) activation via TSP1

Next, to confirm the effect of IH on PCF activation in vitro, we used immunofluorescence staining of α-SMA in mouse PCFs (mPCFs) with or without IH exposure. As compared with normoxia, IH increased α-SMA expression in PCFs and led to an activated phenotype of fibroblast cells (Figure 3A–B). In addition, IH exposure increased the proliferation and contractility of PCFs (Figure 3C–D). We tested whether TSP1 plays an important role in fibroblast activation and found the mRNA expression of Thbs1 significantly increased by IH as compared with normoxia in mPCFs (Figure 3E). In addition, IH significantly increased the protein level of TSP1 beginning from 2 hr and sustained until 8 hr in mPCFs (Figure 3F–G). Furthermore, shRNA lentivirus used to produce TSP1 deficiency in PCFs attenuated IH-induced mRNA and protein levels of α-SMA, collagen I and periostin (Figure 3H–K). TSP1 deficiency in PCFs also blocked inflammatory gene Tnfa expression (Figure 3I) induced by IH exposure. Thus, our results suggested that IH induced PCF activation through upregulating TSP1.

IH induced cardiac fibroblast (CF) activation via TSP1.

(A) Mouse primary CFs (mPCFs) were cultured under normoxia or IH for 24 hr. Representative confocal microscopy images of immunofluorescence stained for α-SMA, phalloidin and DAPI. Scale bar, 20 μm. (B) α-SMA fluorescence intensity quantification. (C) Ratio of Ki67 positive cells to total mPCFs after IH exposure for 24 hr. Data are mean ± SEM, n = 5 independent experiments, *p<0.05, unpaired 2-tail t test. (D) Quantification of gel contraction by mPCFs after IH exposure for 24 hr. Data are mean ± SEM, n = 4 independent experiments, *p<0.05, unpaired 2-tail t test. (E) Thbs1 mRNA levels detected by RT-PCR. Data are mean ± SEM, n = 5 independent experiments, *p<0.05, unpaired 2-tail t test. (F) Expression of TSP1 in mPCFs exposed to IH for the indicated time detected by western blot analysis. (G) Quantification of TSP1 in (F). Data are mean ± SEM, n = 4 independent experiments, *p<0.05, 1-way ANOVA with Bonferroni post-test. (H–K) mPCFs were infected with lentivirus to deliver Thbs1 shRNA (LV-shThbs1) or control shRNA (LV-shCtrl) for 48 hr, then cultured under normoxia or IH for 24 hr. (H–I) Thbs1, Acta2, Col1a1, Postn and Tnfa mRNA levels detected by RT-PCR. Data are mean ± SEM, n = 5 independent experiments, *p<0.05, 2-way ANOVA with Bonferroni post-test. (J) Protein level of TSP1, collagen I (Col1), periostin and α-SMA detected by western blot analysis. (K) Quantification of TSP1, Col1, periostin and α-SMA in J). Data are mean ± SEM, n = 4 independent experiments, *p<0.05, 2-way ANOVA with Bonferroni post-test. Data used for quantitative analyses as well as the numerical data that are represented in graphs are available in Figure 3—source data 1.

IH upregulated phosphorylation of STAT3 at Tyr705 site both in vivo and in vitro

TSP1 was reported to be transcriptionally regulated by STAT3 in astrocytes (Tyzack et al., 2014), and emerging evidences has indicated that STAT3 involved in fibrosis diseases (Chakraborty et al., 2017; Wang et al., 2018). Here, we hypothesized that STAT3 might play a vital role in IH-induced TSP1 expression and cardiac fibrosis. Since the phosphorylation status of STAT3 is related to its transcriptional activity (Darnell et al., 1994; Ihle, 1995), we determined the phosphorylation of STAT3 at multiple sites as well as the phosphorylation of JAKs in mPCFs under IH. STAT3 phosphorylation at Tyr705 was significantly increased by IH from 0.5 to 2 hr, but STAT3 phosphorylation at Ser727 or total STAT3 was not altered by IH exposure (Figure 4A–B). Meanwhile, IH exposure significantly induced the phosphorylation of JAK2 at Tyr1008 without affecting that of JAK1 or JAK3 (Figure 4A–B). A JAK2 inhibitor, TG101209, abolished the phosphorylation of STAT3 at Tyr705 induced by IH (Figure 4C–D), so IH-induced STAT3 phosphorylation might be mediated by JAK2. Phosphorylation of STAT3 at Tyr705 was reported to be crucial to STAT3 nuclear translocation and transcriptional activity (Darnell et al., 1994; Ihle, 1995), so we next analyzed the STAT3 subcellular localization. STAT3 nuclear translocation was significantly increased by IH on immunofluorescent staining (Figure 4E–F). Moreover, in the mouse left ventricle after IH exposure, consistent with results in vitro, IH exposure induced phosphorylation of STAT3 at Tyr705 but not Ser727 (Figure 4G–H). Also, IH exposure or together with Ang II markedly increased STAT3 nuclear localization in vivo (Figure 4I). Thus, IH induced STAT3 activation via JAK2-mediated STAT3 phosphorylation at Tyr705.

IH induced STAT3 signaling activation in CFs.

(A) mPCFs were cultured and exposed to IH for the indicated time. The protein levels of phosphorylated JAK1 at Tyr1034/1035 (p-JAK1Y1034/1035), total JAK1 (t-JAK1), p-JAK2 at Tyr1008 (p-JAK2Y1008), total JAK2 (t-JAK2), p-JAK3 at Tyr980/981 (p-JAK3Y980/981), total JAK3 (t-JAK3), p-STAT3 at Tyr705 (p-STAT3Y705) and Ser727 (p-STAT3S727) and total STAT3 (t-STAT3) detected by western blot analysis. (B) Quantification of indicated protein levels in (A). Data are mean ± SEM, n = 4 independent experiments, *p<0.05, 1-way ANOVA with Bonferroni post-test. (C–D) mPCFs were cultured and exposed to IH for 1 hr with or without TG101209 (2 μM). Indicated protein levels were analyzed (D). (E) mPCFs were exposed to IH for 1 hr. Representative confocal microscopy images of immunofluorescence staining for STAT3 and DAPI. Scale bar, 20 μm. (F) Quantification of subcellular localization of STAT3 in (E). Data are mean ± SEM, n = 5 independent experiments, *p<0.05, unpaired 2-tail t test. (G) The protein levels of p-STAT3Y705 and p-STAT3S727 in ventricles of mice with or without IH detected by western blot analysis. (H) Quantification of p-STAT3Y705, p-STAT3S727 and t-STAT3 in (E). Data are mean ± SEM, n = 6 mice per group, *p<0.05, unpaired 2-tail t test. (I) Representative confocal microscopy images of immunofluorescence staining for STAT3, periostin, and DAPI in mice exposed IH with or without Ang II infusion for 4 weeks. The white arrows indicate periostin-positive cells with nuclear localization of STAT3. Scale bar, 10 μm. Data used for quantitative analyses as well as the numerical data that are represented in graphs are available in Figure 4—source data 1.

STAT3 silencing or inhibition blunted IH-induced TSP1 expression and CF activation

We next explored the role of STAT3 in IH-induced TSP1 expression and CF activation. Knockdown of STAT3 with siRNA almost completely abolished IH-induced TSP1 expression both in mRNA and protein level (Figure 5A–C). Simultaneously, IH-induced α-SMA, collagen I and periostin expression was blocked by STAT3 knockdown at both the mRNA and protein level (Figure 5A–C). To further assess the importance of STAT3 transcriptional activity, we treated PCFs with S3I-201, a selective STAT3 inhibitor. S3I-201 barely showed effect on the basal level of TSP1, α-SMA, collagen I and periostin expression but significantly attenuated IH-induced expression of TSP1, α-SMA, collagen I and periostin at both the mRNA and protein level (Figure 5D–F). Quantitative chromatin immunoprecipitation (ChIP) assay proved that IH enhanced the enrichment of STAT3 at TSP1 promoter in PCFs as compared with the control (Figure 5—figure supplement 1A). Taken together, our results indicated a predominant role of STAT3 in IH-induced CF activation.

Figure 5 with 1 supplement see all
Silencing and inactivating STAT3 downregulated TSP1 expression and attenuated IH-induced CF activation.

(A–C) mPCFs were transduced with STAT3 siRNA (siSTAT3) or control siRNA (siCtrl) for 24 hr, then cultured under normoxia or IH for 24 hr. (A) Thbs1, Acta2, Col1a1, Postn, Stat3 mRNA levels were quantified by RT-PCR. Data are mean ± SEM, n = 5 independent experiments, *p<0.05, 2-way ANOVA with Bonferroni post-test. (B) The protein levels of TSP1, collagen I (Col1), periostin, STAT3 and α-SMA detected by western blot analysis. (C) Quantification of TSP1, Col1, periostin, STAT3 and α-SMA protein level in (B). Data are mean ± SEM, n = 4 independent experiments, *p<0.05, 2-way ANOVA with Bonferroni post-test. (D–F) mPCFs were cultured under normoxia and IH for 24 hr with or without S3I-201 (100 ng/mL). (D) Thbs1, Acta2, Col1a1, and Postn, mRNA levels detected by RT-PCR and quantified. Data are mean ± SEM, n = 5 independent experiments, *p<0.05, 2-way ANOVA with Bonferroni post-test. (E) The protein levels of TSP1, Col1, periostin, and α-SMA detected by western blot analysis. (F) Quantification of TSP1, Col1, periostin, and α-SMA protein level in (E). Data are mean ± SEM, n = 4 independent experiments, *p<0.05, 2-way ANOVA with Bonferroni post-test. Data used for quantitative analyses as well as the numerical data that are represented in graphs are available in Figure 5—source data 1.

Pharmacological or genetic inhibition of STAT3 ameliorates IH-induced cardiac dysfunction and fibrosis

As we found that STAT3-induced TSP1 contributed to IH-induced CF activation, we injected S3I-201 into IH-exposed mice with or without Ang II infusion for 28 days (Figure 6A). First, echocardiography analysis showed that S3I-201 promoted significant recovery of ratio of heart weight to tibial length (Figure 6B) and EF after IH exposure (Figure 6C). Moreover, on Masson staining, S3I-201 reduced IH-induced fibrosis under both basal and Ang II infusion conditions (Figure 6D–E). In addition, immunofluorescence staining showed significantly decreased TSP1, collagen I, periostin and α-SMA expression in the mouse left ventricle after S3I-201 treatment under both basal and Ang II infusion conditions (Figure 6D–E). The expression of TSP1, collagen I, periostin and α-SMA, as well as phosphorylation of STAT3 at Tyr705 site in the left ventricle showed similar trends (Figure 6F–G).

Figure 6 with 1 supplement see all
STAT3 inhibitor ameliorated IH-induced cardiac fibrosis.

(A) C57BL/6 mice exposed to IH with or without infusion of Ang II were treated with S3I-201 or vehicle. (B) Ratio of heart weight to tibial length of mice. (C) Quantification of ejection fraction (EF) after echocardiography of mice. (D) Representative Masson staining of left ventricle of mice. Immunofluorescence staining of left ventricle for TSP1, periostin, Collagen I (Col1), α-SMA, VIME and DAPI. Scale bar, 100 μm. (E) Quantification of fibrotic area in Masson-stained slides (D), quantification of TSP1, periostin, and Col1 fluorescent intensity in (D), and quantification of co-localization of α-SMA and VIME in (D). (F) Protein levels of TSP1, Col1, periostin, p-STAT3Y705, t-STAT3 and α-SMA in left ventricle of mice detected by western blot analysis. (G) Quantification of TSP1, Col1, periostin, p-STAT3Y705 and α-SMA in (F) Data are mean ± SEM, n = 6 mice per group, *p<0.05, 2-way ANOVA with Bonferroni post-test. (H) Schematic diagram depicting the key findings of this study. IH induced cardiac fibrosis via a STAT3/TSP1/Smad pathway. Data used for quantitative analyses as well as the numerical data that are represented in graphs are available in Figure 6—source data 1.

To further investigate whether S3I-201 could reverse the damage already induced by IH, we treated mice with S3I-201 for 2 weeks after exposure to IH and Ang II for 28 days (Figure 6—figure supplement 1A). The ratio of heart weight to tibial length, EF and fibrotic area in mice were comparable between groups (Figure 6—figure supplement 1B–E). In addition, because systemic S3I-201 administration might protect the heart via direct and indirect mechanisms, we determined the effect of AAV9-periostin promoter-shStat3 on IH-induced fibrosis (Figure 6—figure supplement 1F). Evidence from flow cytometry confirmed the knockdown of STAT3 in CFs (Figure 6—figure supplement 1G). Similar to that observed in animals subjected to S3I-201 treatment, AAV9-periostin promoter-shStat3 promoted a significant recovery of ratio of heart weight to tibial length and EF after IH exposure (Figure 6—figure supplement 1H–I), and reduced IH-induced fibrosis under both basal and Ang II infusion conditions (Figure 6—figure supplement 1J–K). These results suggest that pharmacological or genetic inhibition of STAT3 might be a potential therapeutic strategy for cardiac fibrosis induced by IH.

Discussion

In the last decades, we have extensive evidence for a causal link between OSA and cardiovascular disease (Gozal and Kheirandish-Gozal, 2008). Here, we provide definitive evidence of elevated plasma TSP1 level in both humans with OSA and mice exposed to IH. Moreover, we found that TSP1 activated TGFβ signaling, which subsequently promoted the transformation of CFs to myofibroblasts, dependent on STAT3 Tyr705 phosphorylation. Finally, pharmacological inhibition of STAT3 with S3I-201 or AAV9-periostin promoter-shStat3 significantly attenuated IH-induced cardiac fibrosis under both basal and Ang II infusion conditions (Figure 6H).

TSP1 belongs to the thrombospondin family, a conserved family of extracellular, oligomeric, multidomain, calcium-binding glycoproteins that can be secreted by various cell types (Crawford et al., 1998; Adams and Lawler, 2011; Hugo et al., 1998). Increased TSP1 levels are associated with many kinds of cardiovascular diseases, including pulmonary hypertension (PH), idiopathic interstitial pneumonia and aging (Ide et al., 2008; van Almen et al., 2011; Kaiser et al., 2016; Kumar et al., 2017; Rogers et al., 2017). In individuals with end-stage PH, both protein and mRNA levels of TSP1 are elevated in the lung parenchyma and pulmonary artery (Rogers et al., 2017). Furthermore, TSP1 blockade protected against Schistosoma- and hypoxia-induced PH (Kumar et al., 2017). In our study, TSP1 protein expression was significantly elevated in patients with OSA as compared with healthy controls. Strikingly, OSA incidence is high in patients with PH, idiopathic pulmonary fibrosis, and other cardiac diseases, which suggests that TSP1 might be the common pathologic mechanism in OSA-associated cardiovascular diseases. Therefore, TSP1 might be a biomarker for diagnosis of OSA-related cardiac dysfunction, and reducing IH-upregulated TSP1 expression might be crucial to protect the cardiac function of patients with sleep apnea.

TGFβ, perhaps the most extensively studied mediator of fibroblast activation and having the greatest role in pathological fibrosis (Travers et al., 2016; Leask and Abraham, 2004), can be activated by TSP1 binding to a defined site on latency-associated propeptide and inducing a conformational change in the latent complex (Murphy-Ullrich and Poczatek, 2000; Ribeiro et al., 1999). The kinetics of its gene expression, along with its independence from de novo protein synthesis, led to classifying TSP1 as an immediate early-response gene (Majack et al., 1987; Patel et al., 1997). For example,in vascular smooth muscle cells, the mRNA level of TSP1 was induced rapidly within 15 min by platelet-derived growth factor (Majack et al., 1987). Consistently, we found that IH upregulated Thbs1 continuously from 2 to 8 hr. However, IH-induced TSP1 protein expression in PCFs peaked at 2 hr and decreased from 4 to 8 hr. The concentration of TSP1 in culture medium was still increased from 4 to 8 hr, which suggests that IH might promote TSP1 secretion as well. However, the mechanism underlying TSP1 secretion needs further elucidation.

STAT3, as a multifaceted molecule, plays a central role in organic fibrogenesis and cardiac hypertrophy (Su et al., 2017). Hypoxic stress can induce phosphorylation of STAT3 and its simultaneous nuclear translocation (Noman et al., 2009). Here we found that STAT3, as a TSP1 transcription factor, was phosphorylated at Tyr705 and translocated into the nucleus after IH exposure, then promoted fibroblast activation. In addition, STAT3-null mice showed severe fibrosis with age (Jacoby et al., 2003), but in transaortic constriction-induced cardiac remodeling, selective inhibition of STAT3 by S3I-201 significantly improved EF and decreased the left ventricle chamber dilation (Unudurthi et al., 2018). Consistently, we found that S3I-201 treatment reduced the pathologic upregulation of TSP1 and ameliorated the cardiac fibrosis of IH-exposed mice with or without Ang II infusion. Importantly, these results were further confirmed by using AAV9-periostin promoter-shStat3. Direct blockade of TSP1/TGFβ signaling is a major therapeutic challenge because of its ubiquitous nature and localized signaling effects (Kumar et al., 2017). Our results indicated that targeting the pathologic activation of STAT3 might hold promise as a potentially safe approach in preventing IH-induced cardiac fibrosis.

In summary, our study demonstrated that IH upregulated the expression of TSP1 in both mice and humans, and TGFβ pathway activation induced by JAK2/STAT3/TSP1 signaling played a vital role in IH-induced fibroblast activation and cardiac fibrosis. Pharmacological or genetic inhibition of STAT3 might be a potential therapeutic strategy in managing several diverse yet intertwined human pathologies related to IH.

Materials and methods

Key resources table
Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional
information
Antibodyanti-thrombospondin-1 (Rabbit monoclonal)Cell Signaling TechnologyCat# 37879WB (1:1000)
Antibodyanti-phospho-SMAD2/3 (Rabbit monoclonal)Cell Signaling TechnologyCat# 8828WB (1:1000)
Antibodyanti-SMAD2/3 (Rabbit monoclonal)Cell Signaling TechnologyCat# 8685WB (1:1000)
Antibodyanti-phospho-JAK1 (Rabbit monoclonal)Cell Signaling TechnologyCat# 74129WB (1:1000)
Antibodyanti-phospho-JAK2 (Rabbit monoclonal)Cell Signaling TechnologyCat# 8082WB (1:1000)
Antibodyanti-phospho-JAK3 (Rabbit monoclonal)Cell Signaling TechnologyCat# 5031WB (1:1000)
Antibodyanti-JAK1 (Rabbit monoclonal)Cell Signaling TechnologyCat# 3344WB (1:1000)
Antibodyanti-JAK2 (Rabbit monoclonal)Cell Signaling TechnologyCat# 3230WB (1:1000)
Antibodyanti-JAK3 (Rabbit monoclonal)Cell Signaling TechnologyCat# 8827WB (1:1000)
Antibodyanti-phospho-STAT3(Tyr705) (Rabbit monoclonal)Cell Signaling TechnologyCat# 9145WB (1:1000)
Antibodyanti-phospho-STAT3(Ser727) (Rabbit polyclonal)Cell Signaling TechnologyCat# 9134WB (1:1000)
Antibodyanti-STAT3 (Rabbit monoclonal)Cell Signaling TechnologyCat# 12640WB (1:1000)
ChIP (1:50)
Antibodyanti-STAT3 (Mouse monoclonal)Cell Signaling TechnologyCat# 9139IF (1:100)
FACS (1 μL per test)
AntibodyNormal Rabbit IgGCell Signaling TechnologyCat# 2729ChIP (1 μg per test)
Antibodyanti-thrombospondin antibody (Mouse monoclonal)AbcamCat# ab1823IF (1:50)
FACS (1 μL per test)
Antibodyanti-αSMA (Mouse monoclonal)AbcamCat# ab7817WB (1:1000)
IF (1:100)
Antibodyanti-Collagen I antibody (Mouse monoclonal)AbcamCat# ab6308WB (1:1000)
IF (1:100)
Antibodyanti-periostin (Rabbit polyclonal)AbcamCat# ab14041WB (1:1000)
IF (1:100)
Antibodyanti-vimentin (Rabbit monoclonal)AbcamCat# ab92547IF (1:500)
Antibodyanti-GAPDH (Mouse monoclonal)ProteintechCat# 60004WB (1:1000)
Antibodyanti-Ki67 (Rabbit monoclonal)HuaAn Biotechnology CoCat# ET1609-34IF (1:100)
Antibodyanti-mouse CD45-PE Cy7 (Rat monoclonal)BioLegendCat# 103114FACS (1 μL per test)
Antibodyanti-mouse CD31-PE Cy7 (Rat monoclonal)BioLegendCat# 102523FACS (1 μL per test)
Antibodyanti-mouse CD11b-PE Cy7 (Rat monoclonal)BioLegendCat# 101215FACS (1 μL per test)
Antibodyanti-mouse Ter119-PE Cy7 (Rat monoclonal)BioLegendCat# 116221FACS (1 μL per test)
Antibodyanti-mouse thy1- Alexa Fluor 488 (Mouse monoclonal)BioLegendCat# 202505FACS (1 μL per test)
Antibodyanti-mouse thy1- Alexa Fluor 488 (Rat monoclonal)BioLegendCat# 105315FACS (1 μL per test)
AntibodyPE RAT anti-mouse IgG1BD BiosciencesCat# 550083FACS (1 μL per test)
AntibodyAlex 594-conjugated goat anti-mouse antibodyThermo Fisher ScientificCat# A-11005IF (1:200)
AntibodyAlex 488-conjugated goat anti-rabbit antibodyThermo Fisher ScientificCat# A-11008IF (1:200)
Sequence-based reagentStat3 siRNASanta Cruz BiotechnologyCat# sc-29494
Sequence-based reagentControl siRNASanta Cruz BiotechnologyCat# sc-37007
Sequence-based reagentMouse ActaThis paperN/ASequences in Supplementary file 1
Sequence-based reagentMouse Col1a1This paperN/ASequences in Supplementary file 1
Sequence-based reagentMouse PostnThis paperN/ASequences in Supplementary file 1
Sequence-based reagentMouse Thbs1This paperN/ASequences inSupplementary file 1
Sequence-based reagentMouse TnfaThis paperN/ASequences in Supplementary file 1
Sequence-based reagentMouse Stat3This paperN/ASequences in Supplementary file 1
Sequence-based reagentMouse 18SThis paperN/ASequences in Supplementary file 1
Peptide, recombinant proteinAngiotensin IIAbcamCat# ab120183
Commercial assay or kitTSP1 ELISA KIT(Human)R and D SystemsCat# DTSP10
Commercial assay or kitTSP1 ELISA KIT(Mouse)Cloud-Clone Corp.Cat# SEA611Mu
Commercial assay or kitTUNEL staining kitKeyGEN BioTECHCat# KGA7073-1
Commercial assay or kitSimpleChIP Enzymatic Chromatin IP KitCell Signaling TechnologyCat# 9003
Commercial assay or kitEastep Super Total RNA Extraction KitPromega CorporationCat# LS1040
Chemical compound, drugS3I-201Santa Cruz BiotechnologyCat# sc-204304
Chemical compound, drugTG101209MedChem ExpressCat# HY-10410
Software, algorithmPrism version 7.0GraphPad Software Inchttps://www.graphpad.com/scientific-software/prism/
Software, algorithmImageJ version 1.52aNIHhttps://imagej.nih.gov/ij/
Software, algorithmFlowJo version 10Tree Star Inchttps://www.flowjo.com/solutions/flowjo/downloads

Animals

Male C57BL/6J mice at 8–10 weeks of age were used. All mice were housed in a controlled environment (20 ± 2°C, 12 hr/12 hr light/dark cycle) and maintained on a standard chow diet with free access to water. Intermittent hypoxia (IH) was induced as described previously by using an automated system to control ambient oxygen concentration (Savransky et al., 2007; Chen et al., 2005). Briefly, O2 concentration was decreased to 4–6% approximately every 60 s. IH mice were exposed to IH for 8 hr/day in the light time for 28 days. In the pathologic cardiac fibrosis model, mice were subcutaneously implanted with an osmotic minipump (2004 model, Alzet, CA) containing Ang II (1 mg/kg/day) in saline (0.9% w/v) or an identical volume of saline. After the surgery, all mice were exposed to IH for 28 days. In rescue experiments, S3I-201 (5 mg/kg) was given by intraperitoneal injection every 2 days during or after the IH exposure.

Adeno-associated virus (AAV) construction and infection

Request a detailed protocol

The AAV used in this study was constructed as previously reported (Piras et al., 2016). Briefly, AAV9 carrying a periostin promoter driving the expression of shRNA targeting Stat3 (AAV9-periostin promoter-shStat3) was constructed by Shanghai Genechem Co. (Shanghai). The sequences of the shRNAs were for shStat3, GTCACACAGATGAACTTGGTCTTCAGGT and GCATCAATCCTGTGGTATA. The periostin AAV9-periostin promoter-shStat3 or AAV9-periostin promoter-shScramble (1.5 × 1011 v.g) was injected into the tail vein of mice. At 1 week post-injection, mice were exposed to IH with or without Ang II infusion.

Echocardiography

Request a detailed protocol

Trans-thoracic echocardiography was performed on all mice by using a Vevo 2100 system with a MS400 linear array transducer (VisualSonics, ON, Canada) as previously reported (Zhang et al., 2018). Briefly, mice were anesthetized with 2% isoflurane and kept warm on a heated platform (37°C). The chest hairs were removed by using depilatory cream, and a layer of acoustic coupling gel was applied to the thorax. An average of 10 cardiac cycles of standard 2-D and m-mode short axis at mid-papillary muscle level were analyzed. Left ventricular ejection fraction and dimensions were calculated by using a modified Quinone method.

Histology

Request a detailed protocol

Tissues were fixed in 10% neutral-buffered formalin for 24 hr at room temperature and embedded in paraffin. Hearts were sectioned at 5 μm for staining. Collagen deposition was stained with Masson’s trichrome (Sigma-Aldrich, MO) and Sirius red (Solarbio Life Sciences, Beijing) according to the manufacturer’s instructions. Images of sections were captured under an Olympus inverted microscope (IX53, Tokyo) and fibrotic areas were semi-quantitatively determined by using ImageJ 1.52.

For immunofluorescence staining, harvested hearts were fixed and embedded in OCT (VWR, PA) and sectioned at 5 μm. Then, sections were stained with primary antibodies for TSP1, Collagen I, periostin, α-SMA and vimentin overnight at 4°C. Alex 488-conjugated goat anti-rabbit and Alex 594-conjugated goat anti-mouse antibodies were used as secondary antibodies. Nuclei were stained with DAPI. TUNEL staining kit (KGA7073-1, KeyGEN BioTECH, Nanjing) was used for analyzing apoptotic cells in cardiac tissue. Images were acquired under an Olympus inverted microscope (IX81, Tokyo). Fluorescent intensity was quantified by using ImageJ. Colocalization of α-SMA and vimentin was quantified by using the Image J with colocalization plugin. Briefly, the plugin initially generates an 8-bit image with only the colocalized points, then it combines the three 8-bit images in an RGB image. Two points are considered as colocalized if their respective intensities are strictly higher than the threshold of their channels.

RT–qPCR

Request a detailed protocol

RNA was extracted by using the Eastep Super Total RNA Extraction Kit (LS1040, Promega, WI). cDNA was synthesized by using the iScript cDNA Synthesis Kit (Bio-Rad, CA). Quantitative RT-PCR was performed with gene-specific primers shown in Supplementary file 1.

Western blot analysis

Request a detailed protocol

Proteins were isolated from snap-frozen heart tissue and cultured cardiac fibroblasts (CFs), which were extracted in RIPA solution with a protease inhibitor cocktail (#4693132001) and PhosSTOP (#04906845001, both Roche, IN). Proteins were quantified by using the BCA Protein Assay Kit (Thermo Fisher Scientific, MA). Then, 20 μg each protein was separated on SDS-PAGE and electrotransferred onto PVDF membranes, blocked with TBST containing 5% bovine serum albumin, and blots on membranes incubated with antigen and antibody complexes were detected by an ECL protocol with horseradish peroxidase-conjugated IgG as secondary antibodies. Immunoblots were quantified by using ImageJ 1.52.

Isolation and culture of primary CFs (PCFs)

Request a detailed protocol

Isolation of mouse primary CFs (mPCFs) was performed as previously reported (Su et al., 2017). Briefly, mouse hearts from freshly euthanized C57BL/6 mice were harvested and minced to 1 mm (Somers et al., 2008) in cold phosphate buffered saline. Minced tissue was subsequently digested with buffer containing collagenase II (#V900892, Sigma-Aldrich, MO) and trypsin (Solarbio life Sciences, Beijing) under constant stirring at 37°C for 60 min. The supernatants were spun to collect cells. Then cells resuspended in DMEM/F12 (#11320033, Gibco, MA) were plated into dishes and incubated for 2 hr. Supernatant was discarded and dishes were replenished with fresh medium. mPCFs were incubated at 37°C in a humidified atmosphere of 5% CO2 and grown to 70–80% confluence. Cells at passages 2 to 3 were used in experiments. Cells exposed to IH were maintained in a hypoxia chamber (5% CO2; balance N2 and O2 from 5% to 21%) for the indicated times.

Flow cytometry

Request a detailed protocol

Flow cytometry was performed as previously reported (Ali et al., 2014). Digested cells from mouse hearts were suspended in staining buffer containing the relevant surface marker antibodies (PE/Cy7 anti-mouse CD45, CD31, CD11b, Ter119, and Alexa Fluor 488 anti-mouse Thy1) in the dark for 30 min at room temperature. Then, for intracellular staining, cells were fixed with cold methanol for 10 min at −20°C, then incubated with a stain buffer containing TSP1 or STAT3 antibody. After three washing steps, a secondary antibody conjugated to PE was added. The cells were then washed and ready for analysis.

Cell immunofluorescence staining

Request a detailed protocol

Mouse primary CFs were fixed with 4% paraformaldehyde and permeabilized with 0.25% Triton X-100. Then cells were stained with antibodies against α-SMA (1:100), STAT3 (1:100) or Ki67 (1:100). Alexa Fluor antibodies (1:200) were used as secondary antibodies. Stress fibers were stained with rhodamine phalloidin (Thermo Fisher Scientific, MA). Nuclei were stained with DAPI. Images were acquired under an Olympus laser scanning microscope (IX81, Tokyo).

Collagen gel contraction assay

Request a detailed protocol

The collagen contraction assay was performed as previously reported (Pincha et al., 2018). Mouse primary CFs were embedded in collagen gel containing collagen type I from rat tail (#08–115, Sigma-Aldrich, MO) and culture medium. Contraction was measured from the gel images after 24 hr.

Lentivirus construction and infection

Request a detailed protocol

Lentiviruses carrying short hairpin (shRNA) for Thbs1 (LV-shThbs1) and non-specific shRNA (LV-shCtrl) were constructed by Shanghai Genechem Co. (Shanghai). CFs were infected with lentivirus at multiplicity of infection (MOI) 10, and no detectable cellular toxicity was observed.

ELISA

Request a detailed protocol

The secretion of TSP1 in mice or human plasma was quantified by using an ELISA kit (SEA611Mu, Cloud-Clone Corp., TX and DTSP10 R and D Systems, MN) according to the manufacturer’s instructions.

siRNA transfection

Request a detailed protocol

Mouse primary CFs were seeded on plates and cultured to 80% confluence, then transfected with Stat3 siRNA (#sc-29494) or control siRNA (#sc-37007, both Santa Cruz Biotechnology, CA) by using Lipofectamine RNAi MAX transfection reagent (Thermo Fisher Scientific, MA) for 24 hr.

Chromatin immunoprecipitation (ChIP) assay

Request a detailed protocol

ChIP was performed as previously reported (He et al., 2018) with SimpleChIP Enzymatic Chromatin IP Kits (Cell Signaling Technology, MA) according to the manufacturer’s instructions. Briefly, mPCFs were crosslinked with 1% formaldehyde followed by quenching with glycine for 5 min. Cell lysates were digested by micrococcal nuclease, sonicated, and proteins were immunoprecipitated with antibody to STAT3 or rabbit IgG as a control. After complete washing, immunoprecipitated DNA was eluted with elution buffer and reverse-crosslinked overnight at 65°C. DNA was purified and quantified by quantitative RT-PCR. Enrichment was calculated relative to input. Primers for ChIP-qPCR were forward, 5’- TGGCTTCCTCTGTGGTCTCT-3’, and reverse, 5’-GTCAAGGTCATGGGATGGTC-3’.

Statistical analysis

Request a detailed protocol

Sample sizes were designed with adequate power according to the literature and our previous studies. Data are presented as mean ± standard error of the mean (SEM). Statistical analysis involved use of GraphPad Prism 7 v7.04 with two-tailed, unpaired Student’s t test or one-way or two-way ANOVA with Bonferroni multiple comparison post-test, as appropriate. The criterion for statistical significance was p<0.05.

References

  1. 1
    The thrombospondins
    1. JC Adams
    2. J Lawler
    (2011)
    Cold Spring Harbor Perspectives in Biology 3:a009712.
    https://doi.org/10.1101/cshperspect.a009712
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
    Induction of thrombospondin messenger RNA levels occurs as an immediate primary response to platelet-derived growth factor
    1. RA Majack
    2. J Mildbrandt
    3. VM Dixit
    (1987)
    The Journal of Biological Chemistry 262:8821–8825.
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44

Decision letter

  1. Richard P Harvey
    Reviewing Editor; Victor Chang Cardiac Research Institute, Australia
  2. Jonathan A Cooper
    Senior Editor; Fred Hutchinson Cancer Research Center, United States
  3. Barry Fine
    Reviewer; Columbia University, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

The paper, which is much improved after extensive revision in response to reviewers' comments, provides a detailed dissection of the impact of intermittent hypoxia (IH), as found in sleep apnea, on fibotic signalling through activation of JAK2, STAT3 and TSP1, leading to expression of SMA, Col1 and periostin reflecting fibroblast activation. A nice feature of this study is that the relatively mild effects of short term IH alone, significantly aggravate the pro-fibrotic effects of Ang II, allowing greater dynamic range for demonstration of its molecular and functional impact on the heart. Significantly, inhibition of STAT3 signaling reverses this impact, pointing to a possible new therapeutic approach for sleep apnea. The study effectively combines in vivo and in vitro evidence for molecular mechanism.

Decision letter after peer review:

Thank you for submitting your article "Intermittent hypoxia mediated by TSP1 dependent on STAT3 induces cardiac fibroblast activation and cardiac fibrosis" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Jonathan Cooper as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Barry Fine (Reviewer #2). The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

The study proposes a novel mechanism for the induction of cardiac fibrosis in patients suffering obstructive sleep apnea, which involves body-wide exposure to intermittent hypoxia (IH). A mouse model was established of IH in the context of a pro-fibrotic stimulus provided by Angiotenin II (Ang II). Cardiac functional outputs were explored in the various treatment groups comparing normoxia and IH. Developed fibrosis (sirius red staining) and induction of signaling pathways (western blotting, qRT-PCR, IF) were also explored. Through a series of in vivo and in vitro studies, the authors find that IH enhances the cardiac fibrosis, HW/BW changes and compromised cardiac outputs induced by Ang II. IH alone had weaker and non-significant effects. They define a pathway for the IH/Ang II interaction involving activation of the transcription factor STAT3 (through phosphorylation at position Y705), leading to induction of TSP1 (an inducer of latent TGFβ) and TGFβ/Smad2/3 signaling pathway, the main known inducer of cardiac fibrosis in different injury situations. siRNA and chemical inhibitors of STAT3 and TSP1 were used in vivo and/or in vivo to confirm the pathway links. Inhibition of STAT3 using S3I-201 significantly restored cardiac ejection fraction and reduced scar size, accompanied by changes in the proportion SMA+ fibroblasts, STAT3 activating phosphorylation, TSP1 protein levels.

Essential revisions:

The paper is extensive and novel, and the data are generally clean with a clear message. However the reviewers have identified a number of significant issues that need to be addressed before publication in eLife. In particular, the characterisation of fibrosis using just SMA staining or gene expression in vitro and in vivo is insufficient (SMA is a poor marker of myofibroblasts), and the rapid induction of TSP1 and SMA in vitro within 2 hours begs the question of the validity of the mechanism of activation of fibrosis promoted. We would specifically like to see data on the mechanism of activation of STAT3 and via which JAKs if possible, stronger evidence for STAT3 activation in vivo, and determination that TSP1 is a direct target of STAT3. Strengthening the assessment of fibrosis and accompanying inflammation could be achieved by analysis of collagen and other ECM components directly, and inflammatory markers. Use of genetic markers such as Postn-Cre would be desirable but not essential. Also, many of the other questions and comments should be addressed, although we don't consider human data to be an essential revision.

1) The concept of Stat3-TSP1-TGFβ signaling pathway in cardiac fibrosis activation is interesting. However, the traditional signaling involved in ischemia/hypoxia mediated cellular signaling often leads to cellular injury and death. This study offers a potential opportunity to illustrate if Stat3 activation in cardiac fibroblasts is a specific and unique signaling event for IH vs. ischemia/hypoxia injury. It is also important to investigate the molecular mechanism involved in Stat3 activation under IH and Ang II stimulation. Which JAK molecule is involved? How was Stat3 selected to study among other possible signaling pathways?

2) Cellular specificity of the Stat3 signaling in cardiac fibrosis should be better demonstrated. Stat3 mediated signaling has a significant impact on cardiomyocyte hypertrophy and survival. Much of the in vivo characterization, including Stat3 measurement or inhibitor treatment, can potentially be contributed by myocytes rather than fibroblasts. Their interaction could also be an important contributor to the hypertrophic remodeling and contractile dysfunction. Therefore, authors should recognize the limitation of these data and their in vivo experimental approaches. They should carefully determine the myocyte morphology, apoptotic status and function in conjunction with fibrosis analysis. It is also important to recognize that the IH effect is global to all tissues. Overall, the lack of fibroblast specific manipulation of Stat3 or TSP1 in vivo appears to be a significant limitation of the current study.

3) In characterization of the relevance of TSP1 induction among OSA patients, there was no data demonstrating potential spectrum of cardiac function among the OSA patients (their Echo parameters) which could be a major support for the functional relevance of the elevated plasma TSP1 in the cardiovascular complications among OSA patients.

4) The use of human serum samples is quite intriguing and hypothesis generating and unfortunately not very well developed. Clinical relevance and human applicability is always suspect in terms of mouse models so it is quite refreshing that it appears there is some signal of TSP1 in human serum – though this is not well fleshed out by further experiments. The paper, specifically Figures 1 and 2, are very reliant on finding a phenotype that reaches significance thresholds by overlaying with a model of hypertrophy using angiotensin II. The interaction between OSA and hypertension and heart failure is a very hot topic and I think that the authors here really missed a significant opportunity to make a clinical leap. They showed conclusively an additive interaction between IH and angiotensin. What would have really elevated this paper into clinical relevance is treating their mouse model somewhere with an ARB to see if they can mitigate either some of the signaling or mouse phenotype. That could have then been further investigated in their human samples, specifically the TSP1 levels stratified by whether the patient is on an ACEi or ARB. This could have been a real nidus for a clinical trial in terms of treating OSA induced hypertension with RAAS blockade.

5) The authors use a published intermittent hypoxia model but do not show any indication that the hearts are hypoxic. Based on the mechanism proposed, it is important to distinguish between a direct impact of hypoxia on cardiac fibroblasts (CF) as implied from in vitro data, from myriad other mechanisms that may impact cardiac function including kidney dysfunction and inflammation, to name just two. Showing hypoxia with a hypoxia probe such as pimonidazole is important.

6) The authors have superficially evaluated the impact of IH and Ang II by evaluating only α-SMA expression as a readout of myofibroblast activity. The authors do not evaluate collagen or other ECM genes and do not evaluate the contribution of CF proliferation to the phenotype. Do CF proliferate (e.g. BrdU incorporation, Ki67 or Phospho-histone H3)? Are the CF secreting ECM or other inflammatory cytokines? Are CF contractile? What is the impact on inflammation (TSP1 has been shown to induce inflammation in other heart disease models)? There are also a number of genetic markers of fibroblasts that might complement this study or allow the authors to isolate fibroblast populations by flow cytometry (Tcf21-MerCreMer, Postn-MerCreMer, Col1-CreERT2, Pdgfralpha-MerCreMer).

7) The authors also rely heavily on co-localization of SMA and vimentin and TSP1 and vimentin by Immunofluorescence. How is co-localization defined (the methods only state fluorescence intensity was quantified by image J)? How is a secreted factor (TSP1) co-localized with an intracellular CF marker? There seems to be considerable background expression of TSP1, perhaps in cardiomyocytes, and very little seems to be specific for CF.

8) The authors have not shown the mechanism whereby IH induces Stat3 dependent phosphorylation and activity, nor have they demonstrated Stat3 occupancy of a relevant binding site that regulates Tsp1 gene expression with IH. The authors have also not shown Stat3 nuclear accumulation in IH treated mice in vivo, either by immunostaining or nuclear-cytoplasmic fractionation and Western. Many of the observed results thus could be true-true and unrelated.

9) Figure 3D shows a very rapid induction of Tsp1 and SMA protein following IH (2 hrs). This is extremely rapid to see SMA induction even by exogenous TGF-β administration, which typically works on the order of days. Furthermore, the finding that TSP1 and SMA are both induced at 2 hrs of IH seems to argue against a linear IH-Stat3-Tsp1-SMA/myofibroblast model as proposed. It seems equally plausible that (If SMA is indeed induced within 2 hrs) that Stat3 or another mechanism acts independently of TSP1 on myofibroblast activation.

10) The authors treat mice with a Stat3 inhibitor and show it can block fibrosis and functional decline. Can this drug reverse damage already developed by IH?

11) Figure 2B refers to a significant induction of Tsp1 mRNA by IH alone, which is not true (mRNA is only induced by Ang II alone, or IH plus Ang II). This is surprising, considering the rapid and robust induction of TSP1 protein shown after only 2 hrs of IH in Figure 3D. What is the cause of this discrepancy? Perhaps looking at more acute timepoints would be useful in the mouse model.

https://doi.org/10.7554/eLife.49923.sa1

Author response

Essential revisions:

The paper is extensive and novel, and the data are generally clean with a clear message. However the reviewers have identified a number of significant issues that need to be addressed before publication in eLife. In particular, the characterisation of fibrosis using just SMA staining or gene expression in vitro and in vivo is insufficient (SMA is a poor marker of myofibroblasts), and the rapid induction of TSP1 and SMA in vitro within 2 hours begs the question of the validity of the mechanism of activation of fibrosis promoted. We would specifically like to see data on the mechanism of activation of STAT3 and via which JAKs if possible, stronger evidence for STAT3 activation in vivo, and determination that TSP1 is a direct target of STAT3. Strengthening the assessment of fibrosis and accompanying inflammation could be achieved by analysis of collagen and other ECM components directly, and inflammatory markers. Use of genetic markers such as Postn-Cre would be desirable but not essential. Also, many of the other questions and comments should be addressed, although we don't consider human data to be an essential revision.

We deeply appreciate the editor and reviewers for the comments and suggestions to improve the scientific merit of the manuscript. These comments are valuable for us and we have performed several experiments to address all the concerns. We have added more than 30 novel figure panels that represent new experiments or contain additional experiments. In the revised manuscript, we included more reparative myofibroblast markers (collagen I and periostin). We prolonged the treatment time of IH and looked at more acute timepoints for primary cardiac fibroblast (CF) activation in vitro. Upstream signaling of STAT3 was assessed and we found that JAK2 phosphorylation mediated IH-induced STAT3 activation. Evidence of IH-induced STAT3 activation was also observed in vivo. ChIP assay proved that TSP1 was a direct target of STAT3 in CFs. Assessment of fibrosis and accompanying inflammation were also strengthened with new experiments. Moreover, we determined the effect of AAV9-periostin promoter-shStat3 upon IH-induced fibrosis and further confirmed the vital role of fibroblast STAT3 in the process. Below please find our response to each of these comments.

1) The concept of Stat3-TSP1-TGFβ signaling pathway in cardiac fibrosis activation is interesting. However, the traditional signaling involved in ischemia/hypoxia mediated cellular signaling often leads to cellular injury and death. This study offers a potential opportunity to illustrate if Stat3 activation in cardiac fibroblasts is a specific and unique signaling event for IH vs. ischemia/hypoxia injury. It is also important to investigate the molecular mechanism involved in Stat3 activation under IH and Ang II stimulation. Which JAK molecule is involved? How was Stat3 selected to study among other possible signaling pathways?

We thank the reviewer for recognizing the novelty of our study and the valuable comments. To address the reviewer’s concern about the upstream of STAT3, we detected phosphorylation of JAK1-3 under IH. As shown in Figure 4A-B, we found that IH induced phosphorylation of JAK2 at Tyr1008 site time-dependently without altering the phosphorylation of JAK1 or JAK3. In addition, a JAK2 inhibitor, TG101209, significantly abolished IH-induced STAT3 activation (Figure 4C-D). These results indicated that JAK2 is involved in IH-induced STAT3 activation.

TSP1 has been reported to be transcriptionally regulated by STAT3 in astrocytes (Tyzack et al., 2014). Also, STAT3 has been found involved in other fibrosis diseases including skin fibrosis (Chakraborty et al., 2017) and liver fibrosis (Wang et al., 2018). These studies led us to hypothesize that STAT3 might mediate IH-induced TSP1 expression and cardiac fibrosis. We have cited these studies and added a brief introduction: “TSP1 was reported to be transcriptionally regulated by STAT3 in astrocytes, and emerging evidence has indicated that STAT3 is involved in fibrosis diseases. Here, we hypothesized that STAT3 might play a vital role in IH-induced TSP1 expression and cardiac fibrosis.”

2) Cellular specificity of the Stat3 signaling in cardiac fibrosis should be better demonstrated. Stat3 mediated signaling has a significant impact on cardiomyocyte hypertrophy and survival. Much of the in vivo characterization, including Stat3 measurement or inhibitor treatment, can potentially be contributed by myocytes rather than fibroblasts. Their interaction could also be an important contributor to the hypertrophic remodeling and contractile dysfunction. Therefore, authors should recognize the limitation of these data and their in vivo experimental approaches. They should carefully determine the myocyte morphology, apoptotic status and function in conjunction with fibrosis analysis. It is also important to recognize that the IH effect is global to all tissues. Overall, the lack of fibroblast specific manipulation of Stat3 or TSP1 in vivo appears to be a significant limitation of the current study.

We agree. Indeed, both STAT3 inhibitor treatment and IH challenge are global to all tissues, although we have demonstrated their roles in vitro. As the reviewer suggested, we assessed effects of IH on cardiomyocyte morphology and apoptotic status (Figure 1—figure supplement 1B-C) in the revised manuscript. Nuclear translocation of STAT3 induced by IH was also observed in CFs in vivo (Figure 4I). In addition, we performed new in vivo experiments with AAV9-periostin promoter-shStat3 to knockdown STAT3 in fibroblasts. In line with the STAT3 inhibitor experiments, STAT3 knockdown in CFs significantly attenuated IH and Ang II-induced cardiac fibrosis and improved cardiac function (Figure 6—figure supplement 1F-K). The new added experiments together with other results further demonstrated that CF STAT3 might mediate IH-induced cardiac fibrosis.

3) In characterization of the relevance of TSP1 induction among OSA patients, there was no data demonstrating potential spectrum of cardiac function among the OSA patients (their Echo parameters) which could be a major support for the functional relevance of the elevated plasma TSP1 in the cardiovascular complications among OSA patients.

We thank the reviewer for the valuable comments. It is indeed helpful to analyze whether the plasma TSP1 level is associated with cardiac function of OSA patients. Unfortunately, we did not find echocardiography records for the patients and healthy subjects. Since the plasma concentration of NT-proBNP is raised in cardiac impairment (Clin Endocrinol (Oxf). 1997 9;47(3):287-96.), we determined the NT-proBNP concentration in plasma of patients and analyzed the correlation of NT-proBNP and TSP1 concentrations, instead. As shown in Author response image 1, the two indexes did not show a significant correlation, which might be due to limited sample number. Another reason might be that the NT-proBNP level could be affected by multiple factors. Considering that the uncertain conclusion might mislead readers, we only present the result in the response letter.

Author response image 1
Correlation between NT-proBNP and TSP1 level.

(A) Plasma concentration of TSP1 and NT-proBNP in patients with obstructive sleep apnea (OSA), detected by ELISA and fluorescent immunochromatography. Correlation between NT-proBNP and TSP1 level. Pearson r = -0.2, P=0.3846.

4) The use of human serum samples is quite intriguing and hypothesis generating and unfortunately not very well developed. Clinical relevance and human applicability is always suspect in terms of mouse models so it is quite refreshing that it appears there is some signal of TSP1 in human serum – though this is not well fleshed out by further experiments. The paper, specifically Figures 1 and 2, are very reliant on finding a phenotype that reaches significance thresholds by overlaying with a model of hypertrophy using angiotensin II. The interaction between OSA and hypertension and heart failure is a very hot topic and I think that the authors here really missed a significant opportunity to make a clinical leap. They showed conclusively an additive interaction between IH and angiotensin. What would have really elevated this paper into clinical relevance is treating their mouse model somewhere with an ARB to see if they can mitigate either some of the signaling or mouse phenotype. That could have then been further investigated in their human samples, specifically the TSP1 levels stratified by whether the patient is on an ACEi or ARB. This could have been a real nidus for a clinical trial in terms of treating OSA induced hypertension with RAAS blockade.

We thank the reviewer for offering a professional clinical perspective. To investigate whether ACEi or ARB might have an effect on IH-induced TSP1 expression, we first performed in vitro experiments in primary CFs. As shown in Author response image 2, valsartan treatment had little effect on IH-induced TSP1 expression and STAT3 activation. This result might be explained in part by our new added results (Figure 4A-D). We found that JAK2 phosphorylation mediated IH-induced STAT3 activation. Hence, our in vitro results indicated that JAK2/STAT3 signaling rather than Ang II/AT1R signaling might play a predominant role in mediating IH-induced cardiac fibrosis. Although ACEi or ARB treatment in the mouse model might have beneficial effects and showed promise for clinical application, the global effects of RAAS blockade and the direct effect on blood pressure could not provide direct evidence to support our hypothesis. For the above reasons, we would like to focus on the potential targets found in the present study. In light of this, we hope that the reviewer will agree that in vivo analysis of the effect of ACEi/ARB is outside the scope of the current manuscript. Here, in the revised version, we have provided two potential therapeutic strategies for the IH-induced cardiac fibrosisin vivo. First, we used the STAT3 inhibitor as a pharmaceutical therapeutic strategy to rescue the IH-induced cardiac fibrosis. Second, STAT3 was knocked down in CFs by using AAV9-periostin promoter-shStat3. In line with the STAT3 inhibitor, STAT3 knockdown also showed impressive restorative effects in IH-induced cardiac fibrosis and cardiac dysfunction (Figure 6—figure supplement 1F-K).

Author response image 2
Effect of valsartan on IH-induced TSP1 expression.

(A) mPCFs were cultured under normoxia and IH for 24 hr with or without S3I-201 (S3I, 100 ng/mL) or valsartan (Val, 200 μM). The protein levels of collagen I (Col1), periostin, and α-SMA were detected by western blot analysis. (B) Quantification of Col1, periostin and α-SMA protein level in (A). Data are mean ± SEM, n=4 independent experiments, *P<0.05, 2-way ANOVA with Bonferroni post-test.

5) The authors use a published intermittent hypoxia model but do not show any indication that the hearts are hypoxic. Based on the mechanism proposed, it is important to distinguish between a direct impact of hypoxia on cardiac fibroblasts (CF) as implied from in vitro data, from myriad other mechanisms that may impact cardiac function including kidney dysfunction and inflammation, to name just two. Showing hypoxia with a hypoxia probe such as pimonidazole is important.

According to the reviewer’s suggestion, we performed the experiments with pimonidazole. Mice were exposed to IH for 7 days. Pimonidazole was injected at 60 mg/kg body weight. Mice were sacrificed 1 hr after pimonidazole injection. Then heart tissue was fixed and embedded into OCT for immunofluorescence staining. The pimonidazole-positive area was significantly increased under stimulation of IH. Data are shown in Figure 1—figure supplement 1A in the revised manuscript.

6) The authors have superficially evaluated the impact of IH and Ang II by evaluating only α-SMA expression as a readout of myofibroblast activity. The authors do not evaluate collagen or other ECM genes and do not evaluate the contribution of CF proliferation to the phenotype. Do CF proliferate (e.g. BrdU incorporation, Ki67 or Phospho-histone H3)? Are the CF secreting ECM or other inflammatory cytokines? Are CF contractile? What is the impact on inflammation (TSP1 has been shown to induce inflammation in other heart disease models)? There are also a number of genetic markers of fibroblasts that might complement this study or allow the authors to isolate fibroblast populations by flow cytometry (Tcf21-MerCreMer, Postn-MerCreMer, Col1-CreERT2, Pdgfralpha-MerCreMer).

We thank the reviewer for the valuable suggestions. Several new experiments were performed to address the concerns. First, we agree that α-SMA expression as a only readout of myofibroblast activity is insufficient. So, we observed the other two markers (collagen I and periostin) in most experiments for the revised manuscript (Figure 1, Figure 3, Figure 5 and Figure 6). Second, for the in vitro analysis, we observed the expression of collagen I and periostin in CFs (Figure 3I-K), CF proliferation (Figure 3C), CF contraction (Figure 3D) and expression of inflammatory cytokine (Figure 3I). Finally, following the suggestion of the reviewer, we sorted CF populations by flow cytometry and further confirmed the TSP1 upregulation in CFs of IH-treated mice (Figure 2G).

7) The authors also rely heavily on co-localization of SMA and vimentin and TSP1 and vimentin by Immunofluorescence. How is co-localization defined (the methods only state fluorescence intensity was quantified by image J)? How is a secreted factor (TSP1) co-localized with an intracellular CF marker? There seems to be considerable background expression of TSP1, perhaps in cardiomyocytes, and very little seems to be specific for CF.

We thank the reviewer for the comments. We quantified the colocalization of dual-color immunofluorescence images by using Image J with the colocalization plugin. Briefly, the plugin initially generates an 8-bit image with only the colocalized points, then it combines the three 8-bit images in an RGB image. Two points are considered colocalized if their respective intensities are strictly higher than the threshold of their channels. We detailed the quantification method in the subsection “Histology”.

We thank the reviewer for pointing out that TSP1 acts as a secretion protein and its localization with vimentin might not be convincing. In the revised manuscript, we replaced the statistical analysis of colocalization with quantification of TSP1 intensity (Figure 2E-F and Figure 6D-E). In addition, to address the reviewer’s concern about the source of TSP1, we also sorted fibroblast populations by flow cytometry and further confirmed the TSP1 upregulation in CFs of IH-treated mice (Figure 2G).

8) The authors have not shown the mechanism whereby IH induces Stat3 dependent phosphorylation and activity, nor have they demonstrated Stat3 occupancy of a relevant binding site that regulates Tsp1 gene expression with IH. The authors have also not shown Stat3 nuclear accumulation in IH treated mice in vivo, either by immunostaining or nuclear-cytoplasmic fractionation and Western. Many of the observed results thus could be true-true and unrelated.

We thank the reviewer for pointing out these important issues. To address the reviewer’s concerns about STAT3, we performed new experiments for the resubmitted manuscript. First, we detected the upstream factors of STAT3 and demonstrated that JAK2 but not JAK1 or JAK3 mediated IH-induced STAT3 phosphorylation. IH strongly induced phosphorylation of JAK2 at Tyr1008 but barely had an effect on phosphorylation of JAK 1 or JAK3 (Figure 4A-B). Additionally, a JAK2 inhibitor, TG101209, blocked the IH-increased levels of p-JAK2 and p-STAT3 in vitro (Figure 4C-D). These results are consistent with a previous report that JAK2-mediated STAT3 phosphorylation played a critical role in fibroblast activation in systemic sclerosis (Chakraborty et al., 2017.). Second, to investigate whether STAT3 could transcriptionally regulate the expression of TSP1 in CFs directly, we used ChIP assay and found that IH enhanced the enrichment of STAT3 at the TSP1 promoter in PCFs as compared with the control (Figure 5—figure supplement 1A). Finally, we also observed STAT3 subcellular localization in vivo. In line with our in vitro results, IH markedly induced STAT3 nuclear accumulation in CFs (Figure 4I). We believe that these new added results strongly strengthen our hypothesis.

9) Figure 3D shows a very rapid induction of Tsp1 and SMA protein following IH (2 hrs). This is extremely rapid to see SMA induction even by exogenous TGF-β administration, which typically works on the order of days. Furthermore, the finding that TSP1 and SMA are both induced at 2 hrs of IH seems to argue against a linear IH-Stat3-Tsp1-SMA/myofibroblast model as proposed. It seems equally plausible that (If SMA is indeed induced within 2 hrs) that Stat3 or another mechanism acts independently of TSP1 on myofibroblast activation.

We greatly appreciate the reviewer for asking a critical question. To investigate whether IH could really induce primary CF activation rapidly, we detected the expression of collagen I and periostin, which are more promising markers for myofibroblasts, in primary CFs after IH treatment for 2 hr. IH failed to alter their expression (Author response image 3). Therefore, although IH induced a transient change of α-SMA level, IH might not induce primary CF activation in 2 hr. As mentioned by the editor and reviewers, α-SMA might be a poor marker to indicate myofibroblasts. In the revised manuscript, we prolonged the stimulation time and included more reparative CF activation markers (collagen I and periostin). Collagen I and periostin expression was significantly increased from 24 to 48 hr (Figure 3I-K), which was consistent with the results observed in vivo.

Author response image 3
Effect of Intermittent hypoxia (IH) on the expression of collagan I and periostin.

(A) Expression of TSP1, collagen I (Col1) and periostin in mPCFs exposed to IH for the indicated time detected by western blot analysis. (B) Quantification of TSP1, Col1 and periostin in (A). Data are mean ± SEM, n=4 independent experiments, *P<0.05, 1-way ANOVA with Bonferroni post-test.

10) The authors treat mice with a Stat3 inhibitor and show it can block fibrosis and functional decline. Can this drug reverse damage already developed by IH?

The reviewer has pointed out an important question. To investigate whether the STAT3 inhibitor could reverse the damage already induced by IH, we treated mice with IH and Ang II for 4 weeks, then treated these mice with S3I-201 (5 mg/kg) for another 2 weeks. We detected the cardiac function index and fibrosis area. As shown in Figure 6—figure supplement 1A-E, the groups did not significantly differ in the indices, so STAT3 inhibition might not have beneficial effects on damage already induced. One of the possible reasons might be that although fibroblasts are the main effector cells in the pathogenesis of cardiac fibrosis, reversing fibrosis is an active process that may require co-operation of interstitial cell subsets, immune cells, vascular cells, and cardiomyocytes (J Am Coll Cardiol. 2019;73 (18): 2283-2285.).

11) Figure 2B refers to a significant induction of Tsp1 mRNA by IH alone, which is not true (mRNA is only induced by Ang II alone, or IH plus Ang II). This is surprising, considering the rapid and robust induction of TSP1 protein shown after only 2 hrs of IH in Figure 3D. What is the cause of this discrepancy? Perhaps looking at more acute timepoints would be useful in the mouse model.

Thanks for the suggestion. IH-induced TSP1 expression is a rapid process and we may have missed the peak time point to observe the change of TSP1 mRNA level. To test this possibility, we treated mice with IH and Ang II for 2 weeks and determined mRNA level of TSP1 in heart. As shown in Figure 2B, Thbs1 mRNA was significantly induced by IH alone, which was consistent with our in vitro results.

https://doi.org/10.7554/eLife.49923.sa2

Article and author information

Author details

  1. Qiankun Bao

    Tianjin key laboratory of Ionic-Molecular Function of Cardiovascular Disease, Department of Cardiology, Tianjin Institute of Cardiology, The Second Hospital of Tianjin Medical University, Tianjin, China
    Contribution
    Resources, Formal analysis, Funding acquisition, Validation, Investigation, Writing - original draft, Writing - review and editing
    For correspondence
    baoqiankun@tmu.edu.cn
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5221-2780
  2. Bangying Zhang

    Tianjin key laboratory of Ionic-Molecular Function of Cardiovascular Disease, Department of Cardiology, Tianjin Institute of Cardiology, The Second Hospital of Tianjin Medical University, Tianjin, China
    Contribution
    Formal analysis, Validation, Investigation
    Competing interests
    No competing interests declared
  3. Ya Suo

    Tianjin key laboratory of Ionic-Molecular Function of Cardiovascular Disease, Department of Cardiology, Tianjin Institute of Cardiology, The Second Hospital of Tianjin Medical University, Tianjin, China
    Contribution
    Formal analysis, Investigation
    Competing interests
    No competing interests declared
  4. Chen Liu

    Department of Clinical Laboratory, Peking University People's Hospital, Beijing, China
    Contribution
    Conceptualization, Resources, Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7120-5626
  5. Qian Yang

    Tianjin key laboratory of Ionic-Molecular Function of Cardiovascular Disease, Department of Cardiology, Tianjin Institute of Cardiology, The Second Hospital of Tianjin Medical University, Tianjin, China
    Contribution
    Resources, Validation, Investigation
    Competing interests
    No competing interests declared
  6. Kai Zhang

    Tianjin key laboratory of Ionic-Molecular Function of Cardiovascular Disease, Department of Cardiology, Tianjin Institute of Cardiology, The Second Hospital of Tianjin Medical University, Tianjin, China
    Contribution
    Formal analysis, Validation, Investigation
    Competing interests
    No competing interests declared
  7. Ming Yuan

    Tianjin key laboratory of Ionic-Molecular Function of Cardiovascular Disease, Department of Cardiology, Tianjin Institute of Cardiology, The Second Hospital of Tianjin Medical University, Tianjin, China
    Contribution
    Formal analysis, Validation, Investigation
    Competing interests
    No competing interests declared
  8. Meng Yuan

    Tianjin key laboratory of Ionic-Molecular Function of Cardiovascular Disease, Department of Cardiology, Tianjin Institute of Cardiology, The Second Hospital of Tianjin Medical University, Tianjin, China
    Contribution
    Resources, Formal analysis, Investigation, Writing - original draft
    Competing interests
    No competing interests declared
  9. Yue Zhang

    Tianjin key laboratory of Ionic-Molecular Function of Cardiovascular Disease, Department of Cardiology, Tianjin Institute of Cardiology, The Second Hospital of Tianjin Medical University, Tianjin, China
    Contribution
    Formal analysis, Funding acquisition, Writing - original draft
    Competing interests
    No competing interests declared
  10. Guangping Li

    Tianjin key laboratory of Ionic-Molecular Function of Cardiovascular Disease, Department of Cardiology, Tianjin Institute of Cardiology, The Second Hospital of Tianjin Medical University, Tianjin, China
    Contribution
    Conceptualization, Data curation, Supervision, Funding acquisition, Project administration, Writing - review and editing
    For correspondence
    tic_tjcardiol@126.com
    Competing interests
    No competing interests declared

Funding

National Natural Science Foundation of China (81800251)

  • Qiankun Bao

National Natural Science Foundation of China (81570304)

  • Guangping Li

National Natural Science Foundation of China (81800297)

  • Yue Zhang

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work was supported by grants from the National Natural Science Foundation of China (81800251, 81570304, 81800297).

Ethics

Human subjects: Ethical approval was obtained through the institutional ethical review board of Peking University People's Hospital (Permit Number: 2018PHB210-01). The study was conducted in accordance with the Declaration of Helsinki. Written informed consent was taken from all study participants.

Animal experimentation: Animal procedures were approved and conducted in accordance with the Experimental Animal Administration Committee of Tianjin Medical University (Permit Number: SYXK 2011-0006; SYXK 2016-0012).

Senior Editor

  1. Jonathan A Cooper, Fred Hutchinson Cancer Research Center, United States

Reviewing Editor

  1. Richard P Harvey, Victor Chang Cardiac Research Institute, Australia

Reviewer

  1. Barry Fine, Columbia University, United States

Publication history

  1. Received: July 4, 2019
  2. Accepted: January 10, 2020
  3. Accepted Manuscript published: January 14, 2020 (version 1)
  4. Version of Record published: January 30, 2020 (version 2)

Copyright

© 2020, Bao et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 493
    Page views
  • 85
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Human Biology and Medicine
    Naga Suresh Adapala et al.
    Research Article
    1. Human Biology and Medicine
    2. Neuroscience
    Iris Bachmutsky et al.
    Research Article Updated