Increased Connective Tissue Growth Factor Relative to Brain Natriuretic Peptide as a Determinant of Myocardial Fibrosis
Excessive fibrosis contributes to an increase in left ventricular stiffness. The goal of the present study was to investigate the role of connective tissue growth factor (CCN2/CTGF), a profibrotic cytokine of the CCN (Cyr61, CTGF, and Nov) family, and its functional interactions with brain natriuretic peptide (BNP), an antifibrotic peptide, in the development of myocardial fibrosis and diastolic heart failure. Histological examination on endomyocardial biopsy samples from patients without systolic dysfunction revealed that the abundance of CTGF-immunopositive cardiac myocytes was correlated with the excessive interstitial fibrosis and a clinical history of acute pulmonary congestion. In a rat pressure overload cardiac hypertrophy model, CTGF mRNA levels and BNP mRNA were increased in proportion to one another in the myocardium. Interestingly, relative abundance of mRNA for CTGF compared with BNP was positively correlated with diastolic dysfunction, myocardial fibrosis area, and procollagen type 1 mRNA expression. Investigation with conditioned medium and subsequent neutralization experiments using primary cultured cells demonstrated that CTGF secreted by cardiac myocytes induced collagen production in cardiac fibroblasts. Further, G protein-coupled receptor ligands induced expression of the CTGF and BNP genes in cardiac myocytes, whereas aldosterone and transforming growth factor-β preferentially induced expression of the CTGF gene. Finally, exogenous BNP prevented the production of CTGF in cardiac myocytes. These data suggest that a disproportionate increase in CTGF relative to BNP in cardiac myocytes plays a central role in the induction of excessive myocardial fibrosis and diastolic heart failure.
Epidemiological studies have established that 40% to 50% of patients with heart failure have normal or minimally impaired left ventricular (LV) ejection fraction, a clinical syndrome that is commonly referred to as diastolic heart failure (DHF). These patients typically have cardiac hypertrophy that is induced by long-standing hypertension or by primary hypertrophic cardiomyopathy, as well as increased passive LV stiffness.1 Among various molecular mechanisms that regulate LV stiffness,2 abnormalities in the transcriptional or posttranscriptional regulation of the collagen gene can result in the disproportionate accumulation of fibrous tissue and elevation of stiffness in the hypertrophied heart.2,3 Recent studies have shown that, in addition to mechanical load, autocrine, paracrine, and endocrine factors, such as angiotensin II, aldosterone (Aldo), endothelin-1 (ET1), natriuretic peptides, osteopontin, and transforming growth factor-β1 (TGF-β), play important roles in the development of myocardial hypertrophy and fibrosis.4,5 However, the precise molecular mechanisms that initiate and promote myocardial fibrosis and increases in ventricular stiffness remain largely unknown.
Connective tissue growth factor (CCN2/CTGF) belongs to the CCN (Cyr61, CTGF, and Nov) family of immediate early genes, which are highly conserved among species.6 This cysteine-rich secreted protein may contribute to progressive fibrosis and excessive scarring in various systemic and local fibrotic diseases.6 Further, CTGF expression is increased in the hypertrophied and failing myocardium of experimental animal models.7,8 CTGF is also an essential mediator for the biological actions of TGF-β6 and its downstream signal transduction elements.9 However, a recent in vitro study demonstrated that CTGF is 1 of the earliest growth factors transcriptionally induced by hypertrophic stimuli in cardiac myocytes (CMs).10
In this study, to confirm the involvement of CTGF in the myocardial fibrosis, we first investigated CTGF protein production in myocardial biopsy samples of patients with DHF. Secondly, by using the pressure overload rat model with a suprarenal abdominal aortic constriction (AC), which mimics a model of DHF,5 we determined and compared the temporal changes of CTGF, TGF-β, and an antifibrotic peptide, brain natriuretic peptide (BNP).11 Because the collagen accumulation level is reflected by the balance of profibrotic factors and antifibrotic factors,12 we investigated their functional interactions, especially between CTGF and BNP, in the development of myocardial fibrosis and DHF.
An expanded Methods section is available online at http:// hyper.ahajournals.org.
Forty-six consecutive patients with normal or minimally impaired LV ejection fraction (>40%), estimated by echocardiography, who underwent endomyocardial biopsy of the LV-free wall in Gunma University Hospital were enrolled in this study (Table S1). All of the patients were clinically stable when the biopsy was performed. Of these patients, 31 patients who had a previous history of overt heart failure within the preceding year in the absence of impaired systolic function as estimated by echocardiography were designated as the DHF group. Another 15 patients without a previous history of heart failure were designated as the nonfailing (NF) group. The clinical diagnosis and the exclusion criteria are described in the expanded Methods section.
AC was established with a 21G silver clip13 in male Wistar rats (Charles River, Japan) weighing 250 to 300 g. Cell culture, histochemical analysis and immunostaining, hemodynamic measurements in AC rats, RNA isolation and Northern blot analysis, Western blotting, and statistical analysis are described in the expanded Methods section online.
Elevated Levels of CTGF Protein in CM Correlates With Myocardial Interstitial Fibrosis in Patients With Preserved Ejection Factor
Clinical characteristics of the NF and DHF groups are summarized in Table S1. There were no significant differences in age, sex, clinical diagnosis, and frequency of complicated disease, except for atrial fibrillation, when comparing the 2 groups. Sixty-one percent of DHF patients had been given previous medication, including angiotensin-converting enzyme inhibitors and/or β-adrenoceptor blockers, because of their previous history of congestive heart failure. Pulmonary artery wedge pressure and LV end-diastolic pressure were not different when comparing the 2 groups. Furthermore, there were no significant differences in echocardiographic parameters when comparing the 2 groups except for left atrial dimension. Plasma BNP concentration was significantly elevated in the DHF group. Representative LV biopsy samples taken from a patient from the NF group with hypertension and a patient from the DHF group are illustrated in Figure 1. Biopsies from the NF patient showed mild hypertrophic myocytes but no interstitial fibrosis by Masson’s trichrome staining (Figure 1A). CTGF immunostaining of serial sections showed a small amount of CTGF protein in the myocytes (Figure 1B). By contrast, biopsies from the DHF patient showed interstitial fibrosis (Figure 1C) and an abundance of CTGF protein in CM (Figure 1D). Quantitative analysis revealed that myocardial fibrosis area (MFA) and CTGF-stained area were significantly elevated in DHF patients (Figure 1E and 1F). Interestingly, the CTGF-stained area correlated with MFA (r=0.638; P<0.001; Figure 1G).
CTGF and BNP Gene Expression Are Coordinately Induced Early in the Development of Cardiac Hypertrophy and Fibrosis
To investigate the role of CTGF for the development of cardiac fibrosis, we created a rat pressure overload cardiac hypertrophy model by constricting the abdominal aorta. In accordance with the increase of systolic blood pressure, on day 4 after AC operation and until day 28, LV weight/body weight ratio, a parameter of LV hypertrophy, significantly increased (Table S2). Furthermore, MFA was significantly increased on day 14 after AC.
Quantitative Northern blot analysis revealed that CTGF mRNA levels peaked on day 1, whereas TGF-β mRNA levels increased gradually and peaked on day 7 (Figure 2A). Furthermore, procollagen type 1α1 (COL1A1) mRNA levels were significantly increased on day 7 and continued to increase until day 28. Interestingly, the temporal course of changes in BNP mRNA was similar to that of CTGF mRNA, particularly from day 1 to day 7 (N=22; r=0.836; P<0.001; Figure 2B and 2D).
High CTGF/BNP Expression Ratio Is Associated With Myocardial Fibrosis and Ventricular Stiffness at a Later Stage of Cardiac Hypertrophy
Although a correlation between CTGF and BNP mRNA expression was observed during the entire experimental period (N=69; r=0.804; P<0.001; Figure not shown), the correlation was weaker between day 14 and day 28 (N=19; r=0.645; P<0.001; Figure 2C and 2E) when compared with the early stage of cardiac hypertrophy (day 1 to day 7; r=0.836; Figure 2B). As shown in Figure 2C, some rats expressed disproportionately abundant CTGF mRNA. Furthermore, those rats with high CTGF mRNA levels relative to BNP mRNA levels also showed marked upregulation of COL1A1 mRNA (Figure 2E, lanes 4 and 6). By contrast, rats with proportional increases in both CTGF and BNP mRNA levels showed only mild upregulation of COL1A1 mRNA (Figure 2E, lanes 3 and 5). Finally, rats with low CTGF mRNA levels relative to BNP mRNA levels showed low levels of COL1A1 mRNA (Figure 2E, lane 7).
Hemodynamic analysis was performed in rats on day 28. LV contractility indices, calculated using the pressure-volume relationship, were comparable when comparing sham-operated and AC rats (Table S3). Diastolic indexes, that is, time constant of relaxation (τ) and the slope of end-diastolic pressure-volume relationship (EDPVR slope), were significantly higher in the AC rats than in the sham-operated rats.
Representative hemodynamic data and histochemical staining of CTGF in a sham-operated and 2 AC rats with comparable or disproportionate mRNA levels for CTGF and BNP are illustrated in Figure 3A and Figure S1. A rat with high CTGF levels related to BNP mRNA levels showed a steeper slope of EDPVR (Figure 3A), a high E/A ratio in Doppler echocardiography (Figure S1A), severe interstitial fibrosis, and positive immunostaining against CTGF in CM (Figure S1B) when compared with a sham-operated rat and a rat with comparable mRNA levels of CTGF and BNP.
To further characterize hearts with high CTGF mRNA levels relative to BNP mRNA levels, AC rats were classified according to the upper or lower 50th percentile groups of the CTGF/BNP expression ratio on day 28. The mean ratio of the CTGF/BNP mRNA level was 1.2 (Figure 3B). AC rats with a higher CTGF/BNP mRNA ratio (n=7) showed elevated EDPVR slope, higher E/A ratio, and increased MFA (Figure 3B) relative to AC rats with the lower CTGF/BNP ratio (n=7). By contrast, LV relaxation (τ), contractility (ejection factor), or LV hypertrophy (LV weight/body weight ratio) was similar when comparing the 2 groups (Figure 3B). The protein content of sarcomeric α-actin was also similar when comparing the 2 groups (Figure 3B). Interestingly, the CTGF/BNP expression ratio correlated with EDPVR slope (r=0.720; P<0.001; Figure S2) and with COL1A1 mRNA expression (r=0.458; P<0.001; Figure S2). The ratio also significantly correlated with the E/A ratio, expression levels of procollagen type 3α1 (COL3A1), and MFA (Table S4). On the other hand, LV contractility indexes, τ, and mRNA expressions of TGFβ and sarcoplasmic reticulum Ca2+ ATPase (SERCA) 2a were not correlated with the ratio (Table S4). Finally, plasma concentration of Aldo was significantly elevated in AC rats with a higher CTGF/BNP ratio (Figure S3), whereas plasma TGF-β and ET-1 concentration was not significantly different when comparing the 2 groups.
CTGF Is Secreted From CM
The molecular basis of the production of CTGF in the heart and the functional interaction with other neurohumoral factors was investigated using rat neonatal primary CM and cardiac fibroblasts (CFBs). Immunofluorescent study with anti-CTGF antibody revealed production of CTGF in cultured CMs (Figure 4A) and CFBs (vimentin-positive cells; Figure S4). Administration of recombinant CTGF resulted in a dose-dependent increase in COL1A1 mRNA levels in cultured CFB (Figure 4B). Profibrotic stimulation with TGF-β, ET1, and Aldo resulted in increased CTGF production and release into the culture medium from the myocytes (Figure 4C). Furthermore, conditioned medium from these CMs enhanced COL1A1 mRNA levels in CFBs (Figure 4C), suggesting that CMs may regulate collagen production in CFBs. When the TGF-β–treated medium was preincubated with anti-CTGF antibody, COL1A1 mRNA induction was abolished in CFBs (Figure 4D). Pretreatment of the CM-cultured medium with both anti-CTGF and anti-TGF-β antibodies further suppressed the COL1A1 mRNA, suggesting that the induction of the COL1A1 gene by TGF-β and even the basal expression level of the COL1A1 gene in CFBs are mediated through TGF-β–dependent and CTGF-dependent pathways.
Common and Uncommon Stimuli Triggering CTGF and BNP Gene Transcription in CMs
To further characterize the correlation between CTGF and BNP mRNA induction in AC rats, the role of mechanohmoral and neurohumoral stimuli on CTGF and BNP induction was investigated. Cyclic stretch induced a rapid increase in CTGF and BNP mRNA levels (Figure 5A). Furthermore, G protein-coupled receptor ligands, such as ET1 (Figure 5B), norepinephrine, and angiotensin II (Figure S5A), increased CTGF and BNP levels in a dose-dependent manner. By contrast, Aldo and TGF-β stimulation resulted in increases in CTGF mRNA levels and decreases or no effect on BNP mRNA levels (Figure 5B). The differential effect of TGF-β and Aldo on CTGF and BNP mRNA levels was also confirmed by comparing the temporal induction pattern of these genes by TGF-β and Aldo with that induced by ET1 (Figure 5C).
BNP Suppresses CTGF Expression in CMs
Because BNP and TGF-β have opposing biological effects,14 the effect of BNP on CTGF expression was investigated. CTGF mRNA levels decreased 2 hours after administration of synthetic BNP (Figure S5B). Synthetic BNP-mediated inhibition of CTGF expression was completely blocked by the protein kinase G inhibitor KT5823, suggesting that the BNP-cGMP-protein kinase G pathway plays a critical role in regulating CTGF expression in CM (Figure S5C). Furthermore, the effect of BNP was also evident in the context of enhanced production of the CTGF protein in response to profibrotic stimuli, such as TGF-β, ET1, and Aldo (Figure 5D).
DHF, Fibrosis, and CTGF
In the present study, patients with DHF had greater amounts of interstitial fibrosis when compared with patients without a previous history of congestive heart failure. Excessive collagen deposition contributes to abnormal passive diastolic ventricular stiffness3 and leads to pulmonary edema.1 Importantly, MFA, the degree of the interstitial fibrosis, significantly correlated with the abundance of CTGF-positive CMs (Figure 1G). By contrast, neither MFA nor the percentage of CTGF-positive CMs was correlated with LV ejection fraction, an index of systolic function, in our study subjects (data not shown). Endomyocardial biopsy can merely disclose histological changes of a limited portion of whole heart, and this immunohistochemical analysis is not a quantitative measurement of CTGF protein. The amount of biopsy samples was not enough to isolate protein for Western blotting. However, multiple samplings from different portions of the same heart and the staining of serial section with normal IgG as a reference of nonspecific staining minimized sampling and technical variations. Our study suggests that excess fibrosis, through an increase of CTGF, significantly contributes to the development of DHF. Based on the staining pattern with CTGF antibody, strong staining was mainly observed in CMs rather than in the interstitium (Figure 1D), suggesting that CMs are largely responsible for the production of CTGF in the DHF heart.
CMs Produce CTGF
CTGF is overexpressed in numerous fibrotic diseases, and the degree of overexpression correlates with the severity of disease.6 Collagen is mainly produced by fibroblasts in many organs. However, various other cells, including fibroblasts, secrete humoral factors to initiate collagen production in fibroblasts.4,5 Although previous reports mainly focused on CFBs as CTGF-producing cells in the heart,15 the present study demonstrated that a significant amount of CTGF is produced by CMs in the hypertrophied rat heart and in the hearts of patients with DHF.
Interestingly, cultured CFB had higher basal levels of CTGF mRNA than CM (Figure S4A). However, unlike the significant induction in CM, CTGF mRNA levels in CFBs were only minimally affected by extrinsic stimuli (CFBs, Figure S4B; CMs, Figure 5B), which is consistent with observations by Kemp et al.10 The inducibility of CTGF in CMs, along with the fact that BNP, an antifibrotic factor, is also produced by CMs, raises the possibility that CTGF produced in CMs regulates collagen production in CFBs. Indeed, the present study demonstrated that CTGF was secreted into the cultured medium of CMs and that this conditioned medium induced increases in COL1A1 mRNA levels in CFBs (Figure 4C). In addition, neutralization of conditioned medium with CTGF antibody sufficiently blunted the fibrotic signal from CMs to CFBs (Figure 4D). These data suggest that there is molecular communication in a paracrine manner between CMs and CFBs and that this process regulates production of collagen.
CTGF/BNP Balance Regulates Cardiac Fibrosis
Because we could not obtain a well-working antibody for BNP immunostaining, we were unable to estimate the BNP protein level in myocardial biopsy samples. However, the percentage of CTGF-positive staining cells in biopsy samples correlated with the plasma BNP level (r=0.41; P<0.05; Figure not shown). A close correlation between CTGF and BNP mRNA induction was also seen in the pressure overload rat heart and in cultured CMs under cell stretch or stimulation with G protein-coupled receptor ligands. Although the precise mechanisms for this induction were not investigated in this study, preliminary studies demonstrated that the ET1-induced increase in CTGF and BNP mRNA levels was blocked by inhibitors of mitogen-activating protein kinases, protein kinase C, and protein kinase A (data not shown). These data suggest that coordinated expression of the CTGF and BNP genes may be mediated by these signaling pathways.
Most importantly, the CTGF/BNP ratio in CM significantly correlated with indices of fibrosis and diastolic function, such as the slope of EDPVR, E/A ratio, COL1A1 mRNA levels, and MFA (Figure 3 and Table S4). Furthermore, AC rats with comparable levels of CTGF mRNA and BNP mRNA expression showed mild production of CTGF protein and sparse fibrosis in the myocardium. Sarcomeric α-actin content was not different between the high CTGF/BNP ratio group and the comparable ratio group, which suggests that myocyte loss is not responsible for the change of the ratio (Figure 3B). SERCA2a is a principal protein responsible for the initiation of the diastolic phase through its ability to remove cytoplasmic Ca2+.2 However, SERCA2a mRNA level was not correlated with the CTGF/BNP ratio (Table S4). These data suggest that the CTGF/BNP ratio does not associate with LV diastolic function, which is related to Ca2+ removal from cytoplasm.
The identity of upstream factors responsible for the disproportionate expression of CTGF and BNP in CMs remains unclear. AC causes severe hypertension and increases in the levels of various neurohumoral factors, such as renin and angiotensin II.16 Interestingly, rats with a higher CTGF/BNP ratio had a higher plasma Aldo concentration and a tendency toward higher plasma TGF-β and ET1 concentrations than the rats with a lower CTGF/BNP ratio (Figure S3). In addition, in vitro study demonstrated that Aldo and TGF-β induced increases in CTGF mRNA but not in BNP mRNA in contrast to the response to cell stretch or G protein-coupled receptor ligands (Figure 5A through 5C). Therefore, at least in the present model, Aldo may be an upstream factor responsible for disproportionate CTGF expression.
In addition to the effect on body fluid homeostasis and blood pressure control, BNP can exert antihypertrophic and antifibrotic effects in the stressed myocardium.11 The present study demonstrated that BNP suppressed basal CTGF expression level in CMs via its effects on protein kinase G (Figure S4C). The effect of BNP on CTGF expression was also observed under various profibrotic stimuli, such as ET1, Aldo, and TGF-β (Figure 5D). Thus, the increase of CTGF and/or decrease of BNP in CMs may play a central role in the induction of excessive myocardial fibrosis and abnormal diastolic function (Figure 6).
To dissect the role of CTGF in the development of DHF, we used a rat pressure-overloaded model as a preserved systolic but impaired diastolic function model. Given that collagen accumulation is regulated by a balance of its synthesis and degradation, the pressure-overloaded model may be characterized as a “synthesis”-dominant model.12 On the other hand, myocardial infarction is a “accelerated synthesis and accelerated degradation” model with respect to collagen turnover.17 Myocardial infarction is another leading cause to provoke cardiac fibrosis. Therefore, our hypothesis should be also tested in the ischemic heart model, as well as the pressure-overloaded cardiac hypertrophy model.
CTGF is a secreted protein, and plasma CTGF concentration correlates with the severity of several systemic fibrotic disorders.18 Measurement of plasma CTGF concentrations is easier and less invasive than assessment of CTGF levels in biopsy samples. Furthermore, the present data suggest that plasma concentration of CTGF or the ratio of plasma concentration of CTGF:BNP may be a diagnostic marker for myocardial fibrosis. In addition, our data showing the inducibility of CTGF in CMs and the myocardial responsiveness to exogenously administered CTGF suggest that CTGF plays an active role in cardiac progressive fibrosis and, thus, becomes a good candidate molecule as a target of antifibrotic therapy.
The present study demonstrated the following: (1) production of CTGF from CMs is associated with the myocardial interstitial fibrosis and DHF; (2) increased CTGF expression relative to BNP expression triggers excessive cardiac fibrosis via BNP-mediated suppression of CTGF expression; and (3) Aldo and TGF-β induce a disproportionate induction of CTGF and BNP expression, whereas a mechanical stretch of CM and G protein-coupled receptor ligands induces proportionate CTGF and BNP expression. These data suggest that CTGF is a key molecule in the process of cardiac fibrosis and that it may serve as a diagnostic marker and therapeutic target for cardiac fibrosis and DHF.
We are grateful to Miki Yamazaki for her technical assistance with the cardiac myocytes culture and Yoshiko Nonaka for her excellent preparation of histological samples.
Sources of Funding
This work was supported in part by a Grant-in-Aid for Scientific Research (KAKENHI B-17390224 and S-15109010) from the Japan Society for the Promotion of Science.
- Received July 24, 2006.
- Revision received August 13, 2006.
- Accepted February 25, 2007.
Kass DA, Bronzwaer JG, Paulus WJ. What mechanisms underlie diastolic dysfunction in heart failure? Circ Res. 2004; 94: 1533–1542.
Manabe I, Shindo T, Nagai R. Gene expression in fibroblasts and fibrosis: involvement in cardiac hypertrophy. Circ Res. 2002; 91: 1103–1113.
Ohnishi H, Oka T, Kusachi S, Nakanishi T, Takeda K, Nakahama M, Doi M, Murakami T, Ninomiya Y, Takigawa M, Tsuji T. Increased expression of connective tissue growth factor in the infarct zone of experimentally induced myocardial infarction in rats. J Mol Cell Cardiol. 1998; 30: 2411–2422.
Kapoun AM, Liang F, O’Young G, Damm DL, Quon D, White RT, Munson K, Lam A, Schreiner GF, Protter AA. B-type natriuretic peptide exerts broad functional opposition to transforming growth factor-beta in primary human cardiac fibroblasts: fibrosis, myofibroblast conversion, proliferation, and inflammation. Circ Res. 2004; 94: 453–461.
Parker FB Jr, Streeten DH, Farrell B, Blackman MS, Sondheimer HM, Anderson GH Jr. Preoperative and postoperative renin levels in coarctation of the aorta. Circulation. 1982; 66: 513–514.
Jugdutt BI. Ventricular remodeling after infarction and the extracellular collagen matrix: when is enough enough? Circulation. 2003; 108: 1395–1403.
Dziadzio M, Usinger W, Leask A, Abraham D, Black CM, Denton C, Stratton R. N-terminal connective tissue growth factor is a marker of the fibrotic phenotype in scleroderma. Qjm. 2005; 98: 485–492.