1 Introduction

Several studies have shown that dysregulation in the degradation of the extracellular matrix (ECM) by matrix metalloproteinases (MMPs), specifically MMP-2 and MMP-9, is involved in the pathogenesis of a wide spectrum of cardiovascular disorders, including atherosclerosis, restenosis, congestive heart failure, myocardial infarction and cardiomyopathy [1, 11, 15, 21]. Diabetic cardiomyopathy is a multifactorial condition associated with a change of myocardial extracellular matrix (interstitial and perivascular fibrosis), interstitial inflammation, an activated cardiac renin-angiotensin-system, oxidative stress, impaired Ca2+-homeostasis and endothelial dysfunction, leading to impaired cardiac ventricular function [34, 37, 38, 44]. In diabetic cardiomyopathy, cardiac fibrosis [39] has recently been shown to be associated with a decrease in MMP-2 expression/activity [8, 43], whereas MMP-9 expression was found to be unchanged [8] or increased [43]. Glucose and angiotensin II, 2 factors associated with diabetic cardiomyopathy, reduce MMP activity in cultured fibroblasts [4].

The decrease in MMP-2 activity in the diabetic heart, which might contribute to the increased accumulation of collagen in the diabetic heart, triggered us to further investigate its regulation. Besides others, MMP-2 digests type I, II, and III collagen [2], important for left ventricular diastolic and systolic performance. Moreover, MMP-2 has recently been shown to be a direct mediator of ventricular remodelling and systolic dysfunction [6]. Similar to other members of the metalloproteinase family, MMP-2 is secreted in a latent form (pro-MMP-2) requiring proteolytic processing in order to become catalytically active [11]. The main activation of pro-MMP-2 takes place on the cell surface and is mediated by membrane type (MT)-MMPs, including MT1-MMP (MMP-14) [27], whereas the proteolytic activity of MMP-2 can be inhibited by binding of tissue inhibitor metalloproteinase (TIMP)-2 to the active catalytic domain of MMP-2. Moreover, the expression of MMP-2 is known to be increased by TGF-β1 signaling [16], of which Smad 7 is an intracellular antagonist [45].

However, the regulation of MMP-2 in different cardiac disorders is less well understood. We further characterized MMP-2 regulation under experimental diabetic conditions to investigate the hypothesis whether MMP-2 dysregulation may be among the mechanisms involved in the development of fibrosis in diabetic cardiomyopathy.

2 Materials and methods

2.1 Animals

Eight weeks old male Sprague Dawley (SD) rats (300–330 g, Charles River) were maintained on a 12 h light/dark cycle and fed with a standard chow ad libitum (n = 8/group). Diabetes mellitus (DM) was induced by a single injection of streptozotocin (STZ; 70 mg/kg; i.p.) prepared in 0.1 M sodium citrate buffer, pH 4.5 (Sigma, Munich, Germany) as described in detail previously [39]. The STZ-model using 70 mg/kg i.p. is a well established model to study diabetic cardiomyopathy [7, 37–39]. Insulin supplementation in animals injected with this dose of STZ leads to an improvement of left ventricle (LV) function, indicating the role of hyperglycemia in cardiac dysfunction in this model [28]. Only rats with blood glucose levels ≥300 mg/dl 5 days after STZ injection were used in the study. Age-matched non-diabetic SD rats were injected with sodium citrate buffer and used as controls. Rats were sacrificed 2 weeks (w) and 6 w after STZ-injection. Non-diabetic control rats were sacrificed 6 w after sodium citrate buffer injection. The investigation conforms with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85–23, revised 1996) and is approved by the Ethical Committee for the Use of Experimental Animals of the Charité, Berlin (Germany).

2.2 Hemodynamic characterization

The animals were anesthetized with sodium pentobarbital (60 mg/kg; ip), intubated and artificially ventilated. A 2F micro pressure Tip-catheter (Millar Instruments Inc., Houston, Texas, USA) was positioned in the LV via the right carotid artery as previously described [39]. Indices of systolic and diastolic cardiac performance were derived from LV pressure-data. LV systolic function was quantified by the end-systolic pressure (LVP) and the peak rate of rise in left ventricular pressure (dP/dt max). LV diastolic function was measured by end-diastolic pressure (LVEDP), peak dP/dt min and the time constant of isovolumic pressure relaxation (tau). Furthermore, the heart rate (HR) was determined by customized software.

2.3 Tissue preparation

After hemodynamic characterization, the hearts were excised, and circa 2 mm thick transverse sections of the hearts were rapidly frozen in liquid nitrogen and stored at -80°C for immunohistology and molecular biology purposes. A further section was paraffin embedded for Sirius Red staining.

2.4 Sirius Red staining

Total collagen content of the Sirius Red (Polyscience, Inc, Warrington, PA) stained sections was measured under circularly polarized light according to previously published methods [21] and quantified by digital image analysis (DIA) as described in detail elsewhere [39]. Perivascular fibrosis was not analyzed. All available fields (>30 fields) were measured, including the septum, the right and the left ventricle.

2.5 Immunohistology

Immunohistological staining was carried out with rabbit anti-collagen subtype I and III- and anti-pan-specific TGF-ß antibodies and quantified by DIA as described in detail elsewhere [39]. All available fields (>30 fields) were measured, including the septum, the right and the left ventricles. Data are represented as area fraction.

2.6 Real-time RT-PCR

Quantitative real-time reverse transcriptase (RT)-PCR (ABI PRISM® 7000 Sequence Detection System software version 1.0., Perkin Elmer) was performed to quantify LV rat collagen (Col) I, Col III, MMP-2 and MMP-9 cDNA levels, which were normalized relative towards the L32 housekeeping gene and multiplied by 10. Overview of the forward and reverse primers used is shown in Table 1. As DNA standards, the products of the conventional PCR for rat Col I, Col III, MMP-2, MMP-9 and L32 with the use of the primers designed for real-time PCR, were cloned into pGEM-T Easy vector (Promega).

Table 1 Primer sequences
Full size table

2.7 Gelatin zymography

Frozen adult rat myocardial tissue samples were homogenized in 5 x volumes of extraction buffer containing: 10 mmol/l cacodylic acid, pH 5.0, 0.15 mol/l NaCl, 1 µmol/l ZnCl2, 20 mmol/l CaCl2, 1.5 mmol/l NaN3, and 0.01% (v/v) Triton X-100 at 4°C overnight. Protein samples (50 µg) were mixed with loading buffer lacking reducing agents (10% w/v SDS, 4% w/v sucrose, 0.1% w/v bromophenol blue) and applied without boiling to 10% SDS/polyacrylamide gel containing 1 mg/ml of gelatin B and separated by gelelectrophoresis. Supernatant of HT 1080 (10 µl), human fibrosarcoma cells, served as positive control. Gelatinolytic activity was visualized as clear areas of lysis in the gel. Each sample lane was scanned and presented as percent of latent MMP-2 HT1080-standard. Quantification was performed by the size-fractionated banding pattern densitometrically.

2.8 Western Blot

LV samples were homogenized in lysis buffer containing proteinase inhibitors. An equal amount of protein (10–50 µg) was loaded into a 10% SDS-polyacrylamide gel. Smad 7 (Santa Cruz Biotechnology, Santa Cruz, CA, USA), MT1-MMP/MMP-14 (Sigma, Munich, Germany), TIMP-2 (IM 56L, Calbiochem, Bad Soden, Germany), β tubulin (Santa Cruz) and calsequestrin (PA1–913; Dianova, Hamburg, Germany) were detected with each specific antibody. β tubulin or calsequestrin were used to confirm equal protein loading. Data were expressed as the ratio of the band intensity of the protein of interest (Smad 7, MT1-MMP/MMP-14 or TIMP-2) versus β tubulin or calsequestrin and represented as the percentage of the non-diabetic control group SD-Co 6w. The blots were visualized with chemiluminescene (ECL) system (Amersham Bioscience, Buckinghamshire, UK). Quantitative analysis of the intensity of the bands was performed with Lab Works.

2.9 Statistical analysis

Statistical analysis was performed using JMP Statistical Discovery Software Version 4.05 (SAS Institute, Cary, NC, USA). Non-parametric tests were used for hemodynamic measurements, collagen and TGF-β data analysis, since normal distribution was excluded regarding all parameters. Quantitative and qualitative data were compared, by conducting the Wilcoxon-Kruskal-Wallis test on rank sums. The significant difference for multiple comparisons of all pairs was calculated by the Tukey-Kramer posthoc analysis. Paired and unpaired Student’s t tests were employed for zymography, Western Blot and real-time RT-PCR data analysis. Data are presented as mean ± SEM, except for collagen data which are presented as medians and interquartile ranges (25%–75%th percentile). Differences were considered significant at P ≤ 0.05 in all tests performed.

3 Results

3.1 Hemodynamic data

STZ-induced diabetes was associated with a time-dependent impairment of diastolic function, as indicated by increased dP/dt min and elongated Tau. LVEDP increased by 30%, 6 weeks after STZ-injection, but failed to reach significance. In addition, systolic function was decreased over time, illustrated by reduced LVP and dP/dt max (Table 2). HR was significantly reduced by SD-STZ 2w and SD-STZ 6w compared to SD-Co 6w, which is a characteristic of diabetes-associated dysfunctional cardiac autonomic neuropathy [40].

Table 2 Basal hemodynamic characteristics by micro-conductance pressure catheter measurements
Full size table

3.2 Effect of streptozotocin-induced diabetes mellitus on cardiac collagen and TGF-ß

Total collagen deposition increased moderately in the first 2w after STZ application, leading to a 2.1-fold (P < 0.05) higher total collagen content 6w post STZ-injection compared to non-diabetic controls (Fig. 1a, b). The increase in total collagen content 6w after STZ-application, was reflected in a 4.2-fold (P < 0.05) and 3.5-fold (P < 0.05) rise in collagen type I and type III-positive immunostainings, respectively (Fig. 1b). In contrast, mRNA expression of LV collagen type I and type III decreased time-dependently, leading to 8.8-fold (P < 0.005) and 9.3-fold (P < 0.0001) lower expression in SD-STZ 6w vs. SD-Co 6w LVs, respectively (Fig. 1c). TGF-ß positive staining was 54-fold (P < 0.05) increased in SD-STZ 6w compared to SD-Co 6w (Fig. 2).

Fig. 1
figure 1

Time-dependent effect of streptozotocin-induced diabetes mellitus on cardiac total collagen, type I and type III content and mRNA expression. a Cardiac total collagen deposition, expressed as area fraction (AF) %, in non-diabetic SD-Co 6w rats, in rats 2w (SD-STZ 2w) and 6w (SD-STZ 6w) after STZ injection. Data are represented as medians and interquartile ranges (25%–75%th percentile) (n = 6); * P < 0.05 vs. SD-Co 6w. b Representative pictures of total collagen deposition by Sirius Red staining (original magnification ×200) and of collagen type I and type III content visualized by immunostaining (original magnification ×200). Left panel baseline total collagen, collagen type I and III expression in a non-diabetic SD-Co 6w rat heart. Right panel total collagen, collagen type I and III abundance in a SD-STZ 6w rat heart. c LV collagen type I (black bar) and III (white bar) mRNA expression in non-diabetic SD-Co 6w, SD-STZ 2w and SD-STZ 6w rats, normalized to ribosomal L32, multiplied by 10 (n = 4–6). Data are represented as mean ± SEM; § P < 0.05 vs. SD-Co 6w and SD-STZ 6w, * P < 0.05 vs. SD-Co 6w and SD-STZ 2w

Full size image
Fig. 2
figure 2

Effect of streptozotocin-induced diabetes mellitus on cardiac TGF-ß expression. a Representative picture of TGF-ß staining (original magnification ×200). Left panel baseline TGF-ß expression in a non-diabetic SD-Co 6w rat heart. Right panel increased TGF-ß expression in a SD-STZ 6w rat heart. b Cardiac TGF-ß expression, depicted as area fraction (AF) %, in non-diabetic SD-Co 6w rats and in rats 6w after STZ injection (SD-STZ 6w). Data are represented as medians and interquartile ranges (25%–75%th percentile) (n = 6); * P < 0.05 vs. SD-Co 6w

Full size image

3.3 Effect of streptozotocin-induced diabetes mellitus on cardiac MMP-9 mRNA, MMP-2 mRNA expression and latent and active MMP-2 levels

LV MMP-9 mRNA tended to be higher in SD-STZ 6w versus SD-Co 6w (SD-Co 6w: 0.0017 ± 0.00051 vs. SD-STZ 6w: 0.0065 ± 0.0020; P = 0.06). LV MMP-2 mRNA expression in SD-STZ 2w and SD-STZ 6w was 2.7-fold (P < 0.005) lower compared to SD-Co 6w (Fig. 3a). In agreement, densitometric analysis of zymographic bands demonstrated that pro-MMP-2 at 72 kDa was 1.3-fold (P < 0.05) reduced in SD-STZ 6w compared to SD-Co 6w heart specimens (Fig. 3b). In addition, MMP-2 activity detected at 66 kDa in SD-Co 6w hearts (0.023 ± 0.0036), was undetectable in SD-STZ 6w hearts (Fig. 3b).

Fig. 3
figure 3

Effect of streptozotocin-induced diabetes mellitus on cardiac MMP-2 mRNA expression and pro-(latent) and active MMP-2 levels. a LV MMP-2 mRNA expression in non-diabetic SD-Co 6w, SD-STZ 2w and SD-STZ 6w rats, normalized to ribosomal L32, multiplied with 10 (n = 4–6). b Upper panel representative inverse picture of zymography of SD-Co 6w and SD-STZ 6w rat hearts, illustrating pro-(latent) and active MMP-2 bands. Lower panel bar graph representing the intensity of latent (black bar) and active (white bar) MMP-2 bands of SD-Co 6w and SD-STZ 6w rat hearts (n = 6). a, b Data are represented as mean ± SEM; * P < 0.05 vs. SD-Co 6w. B.D.: below detection limit

Full size image

3.4 Effect of streptozotocin-induced diabetes mellitus on cardiac Smad 7 protein levels

Smad 7 was found to be 1.3-fold (P < 0.05) higher expressed in SD-STZ 6w compared to SD-Co 6w (Fig. 4).

Fig. 4
figure 4

Effect of streptozotocin-induced diabetes mellitus on cardiac Smad 7 levels. Upper panel representative Western Blot of SD-Co 6w and SD-STZ 6w rat hearts, illustrating Smad 7 and β tubulin bands. Lower panel bar graphs of relative densitometric means, represented as the percentage of the non-diabetic control group SD-Co 6w of Smad 7 to β tubulin (n = 4). Data are represented as mean ± SEM; * P < 0.05 vs. SD-Co 6w

Full size image

3.5 Effect of streptozotocin-induced diabetes mellitus on cardiac MT1-MMP and TIMP-2 protein levels

Under non-diabetic conditions, the 45 kDa form of MT1-MMP was more extensively expressed as the 60 kDa form. Both forms were undetectable in SD-STZ 6w hearts (Fig. 5a), whereas TIMP-2 protein expression was 10-fold (P < 0.005) increased in SD-STZ 6w compared to SD-Co 6w hearts (Fig. 5b).

Fig. 5
figure 5

Effect of streptozotocin-induced diabetes mellitus on cardiac MT1-MMP and TIMP-2 levels. a Upper panel representative Western Blot of SD-Co 6w and SD-STZ 6w rat hearts, illustrating MT1-MMP and calsequestrin bands. Lower panel bar graphs of relative densitometric means, represented as the percentage of the non-diabetic control group SD-Co 6w of MT1-MMP towards calsequestrin (n = 6). b Upper panel Western Blot of SD-Co 6w and SD-STZ 6w rat hearts, illustrating TIMP-2 and calsequestrin bands. Lower panel bar graphs of relative densitometric means, represented as the percentage of the non-diabetic control group SD-Co 6w of MT1-MMP to calsequestrin. (n = 5). a, b Data are represented as mean ± SEM; * P < 0.005 vs. SD-Co 6w. B.D.: below detection limit

Full size image

4 Discussion

The salient finding of the present study is that dysregulation of ECM degradation contributes to the development of cardiac fibrosis in experimental Type I diabetic cardiomyopathy. Thereby, we show that decreased MMP-2 activity concomitant with increased Smad 7 protein levels, decreased MT1-MMP and increased TIMP-2 protein expression, are involved in this condition, which differs from mechanisms involved in dilated and ischemic heart disease.

Evidence from human [5, 31] and experimental models of diabetic cardiomyopathy [13, 25, 39] indicate cardiac fibrosis as pathological substrate of diabetic cardiomyopathy. Cardiac fibrosis leads to increased LV stiffness and decreased ventricular wall compliance, resulting in systolic and in particular diastolic dysfunction. Besides micro-ischemia and inflammation, induction of transforming growth factor (TGF)-β has been indicated to contribute to cardiac fibrosis, leading to the expectation that collagen type I and type III mRNA levels might be increased, at least in early phases of the development of cardiac fibrosis.

Cardiac total collagen deposition, collagen type I and type III content and TGF-ß protein expression were increased in STZ-induced Type I diabetic rats compared to non-diabetic controls. Moreover, the time-dependent increase in cardiac total collagen content was associated with an impairment of early diastolic and systolic function over-time. However, unexpectedly, LV collagen type I and type III mRNA expression declined time-dependently after STZ-injection. This suggests that the increase in collagen type I and type III abundance found in STZ-diabetic rats might lead to suppression of collagen type I and type III mRNA expression as a negative feedback mechanism. On the other hand it also suggests that this accumulation in collagen type I and type III might be due to post-transcriptional/post-translational changes including alterations in ECM regulation, which is mainly controlled by the MMP/TIMP system.

Among others, MMP-2 digests type I, II, and III collagen [2] and is secreted as an inactive pro-enzyme, which is mainly activated by a membrane-linked process involving MT1-MMP, whereas an excess of TIMP-2 inhibits MMP-2 activation by binding all MT1-MMP molecules [9]. In the present study, we demonstrated that MMP-2 mRNA expression declined already after 2w of STZ injection and that MMP-2 mRNA and pro-MMP-2 expression were reduced in STZ-diabetic rat hearts, 6w after STZ-injection compared to non-diabetic controls. Moreover, we could demonstrate that cardiac Smad 7 expression is increased under diabetic conditions. Smad proteins have recently been identified as important components of the TGF-β1 signaling pathway of which Smad 7 is an intracellular antagonist [18, 45]. The role of Smad 7 in MMP-2 regulation is still a matter of discussion: whereas Wang et al. [42] reported that ectopic over-expression of Smad 7 in primary cardiac myofibroblasts was associated with increased MMP-2 activity in the presence and absence of TGF-β1, others found that Smad 7 can suppress TGF-β1-induced MMP-2 production [16, 22], which is independent from its phosphorylation status [23]. The latter is in agreement with our findings under diabetic conditions. We suggest that the reduced MMP-2 mRNA expression and pro-MMP-2 expression found under diabetic conditions can be partly explained by increased Smad 7 protein expression, despite induced TGF-β1 (Fig. 6).

Fig. 6
figure 6

Hypothetical scheme illustrating the proposed cardiac signaling pathway contributing to cardiac fibrosis in streptozotocin-induced diabetes mellitus. The reduction in MMP-2 mRNA expression and latent MMP-2 protein levels under diabetic conditions can be explained by the increase of cardiac Smad 7 expression, which suppresses TGF-β1-induced MMP-2 production, whereas the suppression of MT1-MMP expression together with the increased TIMP-2 expression, leading to an excess of TIMP-2 in the presence of MMP-2 and MT1-MMP, are responsible for the decreased MMP-2 activation in experimental diabetic cardiomyopathy. These mechanisms leading to decreased MMP-2 activity contribute to cardiac fibrosis in streptozotocin-induced diabetes mellitus

Full size image

Besides a decrease in MMP-2 mRNA expression and latent MMP-2, also the active form of MMP-2 was significantly decreased under diabetic conditions. These results are comparable with observations in diabetes mellitus type 1 [12] and type 2 [17] experimental models of diabetic nephropathy and in cell culture [32, 35], illustrating that impaired matrix degradation is associated with reduced MMP-2 activity. However, this regulation seems to be specific for diabetic conditions since chronic pressure overload in MMP-2 knock out [19] mice is associated with less interstitial fibrosis, and MMP-2 transgenic [6] mice have been reported to be associated with fibrosis.

Moreover, we could demonstrate that MT1-MMP protein expression was reduced and TIMP-2 protein expression increased under diabetic conditions. This suggests that the decreased MMP-2 activity in experimental diabetic cardiomyopathy is due to (1) the reduction in MMP-2 mRNA expression and latent MMP-2, (2) the suppression of MT1-MMP expression and (3) the increase in TIMP-2 expression (Fig. 6). These findings are in line with in vitro studies on collagen lattices where glycation is associated with reduced MMP-2 activity and with decreased MT1-MMP expression [14, 24]. Moreover, we found that in non-diabetic rat hearts, especially the 45 kDa form of MT1-MMP was expressed. Since the activation of pro-MMP-2 is concomitant with an increase in the processing of MT1-MMP from the 60 kDa to the 45 kDa form [33], this finding underscores the activation of pro-MMP-2 under non-diabetic conditions. Interestingly, MT1-MMP also has collagenolytic activity on type I, II, and III collagen [20], implying that down-regulation of MT1-MMP leads to reduced proteolysis of collagen type I, II, III, directly via reduction of its own expression and indirectly via reduced MMP-2 activity.

In conclusion, we describe a mechanism contributing to cardiac fibrosis, which differs from mechanisms involved in dilated or ischemic cardiomyopathy. The latter are characterized by myocardial fibrosis associated with increased collagen type I and type III mRNA expression [21] and MMP activity [10, 26, 36], including MMP-2 [29, 30], and reduced Smad 7 expression [41]. In contrast, under STZ-diabetic conditions, in the absence of LV dilatation [3], cardiac fibrosis is associated with decreased collagen type I and type III mRNA expression, decreased MMP-2 activity and increased Smad 7 expression.

Although the role of other MMPs still has to be investigated, we suggest that the control of matrix regulation differs under certain pathological conditions, which may imply the need for differential therapeutic strategies to prevent the development of cardiac fibrosis under diabetic conditions.