Modulatory effect of interleukin-1α on expression of structural matrix proteins, MMPs and TIMPs in human cardiac myofibroblasts: role of p38 MAP kinase.

The proinflammatory cytokine interleukin-1 (IL-1) elicits catabolic effects on the myocardial extracellular matrix (ECM) early after myocardial infarction but there is little understanding of its direct effects on cardiac myofibroblasts (CMF), or the role of p38 mitogen-activated protein kinase (MAPK). We used a focused RT-PCR microarray to investigate the effects of IL-1α on expression of 41 ECM genes in CMF cultured from different patients, and explored regulation by p38 MAPK. IL-1α (10 ng/ml, 6h) had minimal effect on mRNA expression of structural ECM proteins, including collagens, laminins, fibronectin and vitronectin. However, it induced marked increases in expression of specific ECM proteases, including matrix metalloproteinases MMP-1 (collagenase-1), MMP-3 (stromelysin-1), MMP-9 (gelatinase-B) and MMP-10 (stromelysin-2). Conversely, IL-1α reduced mRNA and protein expression of ADAMTS1, a metalloproteinase that suppresses neovascularization. IL-1α increased expression of TIMP-1 slightly, but not TIMP-2. Data for MMP-1, MMP-2, MMP-3, MMP-9, MMP-10 and ADAMTS1 were confirmed by quantitative real-time RT-PCR. Tumor necrosis factor-alpha (TNFα), another important myocardial proinflammatory cytokine, did not alter expression of these metalloproteinases. IL-1α strongly activated the p38 MAPK pathway in human CMF. Pharmacological inhibitors of p38-α/β (SB203580) or p38-α/β/γ/δ (BIRB-0796) reduced MMP-3 and ADAMTS1 mRNA expression, but neither inhibitor affected MMP-9 levels. MMP-1 and MMP-10 expression were inhibited by BIRB-0796 but not SB203580, suggesting roles for p38-γ/δ. In summary, IL-1α induces a distinct pattern of ECM protein and protease expression in human CMF, in part regulated by distinct p38 MAPK subtypes, affirming the key role of IL-1α and CMF in post-infarction cardiac remodeling.

Introduction

Interleukin-1 (IL-1) is one of the initial stimuli that drive the acute inflammatory response after myocardial infarction (MI) (Guillen et al., 1995). Release of IL-1α from necrotic cells is a key trigger for inflammation in the liver (Chen et al., 2007), and it is plausible that this cytokine plays a similar role in the myocardium following myocyte necrosis. Cardiomyocytes express IL-1α (Westphal et al., 2007) and IL-1α protein levels are increased in the infarcted myocardium following myocyte necrosis (Timmers et al., 2008). Moreover, transgenic mice with cardiac-specific overexpression of IL-1α develop left ventricular hypertrophy (Nishikawa et al., 2006), and those ubiquitously over-expressing IL-1α die of heart failure (Isoda et al., 2001).

Cardiac fibroblasts are the most prevalent cell type in the heart and are key regulators of the myocardial extracellular matrix (ECM) (Porter and Turner, 2009). Activated cardiac fibroblasts, termed cardiac myofibroblasts (CMF), play an essential role in the early adaptive healing response that occurs following MI (Porter and Turner, 2009; van den Borne et al., 2010). In this acute phase, CMF secrete proteolytic matrix metalloproteinases (MMPs) that degrade the structural components of the ECM, permitting recruitment of inflammatory cells and CMF to the infarct areas and facilitating neovascularization. Subsequently, CMF synthesize new ECM protein (predominantly fibrillar collagens I and III) to form a scar, and actively contract the edges of the wound area. Thus, CMF play a key role in the early remodeling of the heart following MI and undergo transformation from an ECM-degrading phenotype to an ECM-synthesizing phenotype. Understanding the extracellular cues that drive these distinct processes at the level of the CMF is paramount for designing targeted therapeutic strategies that encourage infarct healing without promoting overt fibrosis.

There is strong evidence that the p38 mitogen-activated protein kinase (MAPK) pathway exacerbates myocardial injury following prolonged ischemia (Clark et al., 2007). Proinflammatory cytokines whose levels are elevated in the infarcted heart, such as IL-1 and tumor necrosis factor-alpha (TNFα), are potent stimuli for the p38 MAPK pathway and transgenic mice with cardiomyocyte-specific overexpression of IL-1α display increased myocardial p38 MAPK activity (Nishikawa et al., 2006). However, despite the key role of CMF in the myocardial remodeling process, very little is known about the role of the p38 MAPK pathway in modulating CMF function. There are four known p38 MAPK subtypes; p38-α (SAPK2a), p38-β (SAPK2b), and the more distantly related p38-γ (SAPK3) and p38-δ (SAPK4). Many studies ascribing roles to p38 MAPK have employed pyridinyl imidazole compounds (e.g. SB203580) as selective p38 MAPK inhibitors (Clark et al., 2007); although strictly speaking these agents inhibit only the α and β subtypes. In contrast, the diaryl urea compound BIRB-0796 inhibits all four p38 subtypes with similar efficacy (Kuma et al., 2005).

In this study we used a focused RT-PCR microarray to investigate the effects of IL-1α on expression of 41 ECM genes in human CMF, including structural ECM proteins, metalloproteinases and tissue inhibitors of metalloproteinases (TIMPs). We then determined the role of the p38 MAPK pathway in mediating IL-1α-induced changes in gene expression of key ECM proteases.

Results

Microarray analysis of IL-1-induced mRNA expression of ECM proteins and proteases

A SYBR Green-based real-time PCR array approach was used to investigate the effects of the proinflammatory cytokine IL-1α (10 ng/ml, 6 h) on mRNA expression of 41 ECM genes in cultured human CMF, including 23 structural proteins (collagens, laminins, fibronectin, vitronectin), 12 MMPs, 3 members of the ADAMTS (a disintegrin and metalloproteinase domain with thrombospondin motifs) family and 3 TIMPs. CMF from 3 different patients were employed in the array analysis to identify reproducible responses to IL-1α. Array data, expressed relative to mRNA levels of the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH), are presented in Table 1 (structural ECM proteins) and Table 2 (ECM proteases and TIMPs).

Table 1

Effects of IL-1 on mRNA levels of structural ECM proteins. Human CMF from 3 patients treated with 10 ng/ml IL-1α for 6 h before extracting RNA. Pooled RNA was reverse transcribed and analysed by real-time PCR array for 22 ECM proteins.

Gene	Protein	Basal (% GAPDH)	IL-1-stimulated (% GAPDH)	Fold change
COL1A1	Collagen I α1	181.1	142.2	0.79
COL4A2	Collagen IV α2	261.7	229.1	0.88
COL5A1	Collagen V α1	115.0	84.1	0.73
COL6A1	Collagen VI α1	556.3	567.8	1.02
COL6A2	Collagen VI α2	442.3	430.1	0.97
COL7A1	Collagen VII α1	1.2	6.1	5.08a
COL8A1	Collagen VIII α1	72.2	59.4	0.82
COL11A1	Collagen XI α1	0	0	–
COL12A1	Collagen XII α1	79.7	75.8	0.95
COL14A1	Collagen XIV α1	30.6	34.3	1.12
COL15A1	Collagen XV α1	2.1	0.8	0.38b
COL16A1	Collagen XVI α1	8.8	17.1	1.94
ECM1	Extracellular matrix protein 1	2.4	2.4	1.00
FN1	Fibronectin 1	5567.7	5474.1	0.98
HAS1	Hyaluronan synthase 1	0.2	1.5	7.50a
LAMA1	Laminin α1	0.4	0.1	0.25b
LAMA2	Laminin α2	17.0	13.3	0.78
LAMA3	Laminin α3	2.3	3.9	1.70
LAMB1	Laminin β1	92.0	49.6	0.54
LAMB3	Laminin β3	1.0	2.1	2.10a
LAMC1	Laminin γ1	79.8	102.8	1.23
VCAN	Versican	145.4	75.3	0.52
VTN	Vitronectin	5.7	5.5	0.96

Indicates marked increase (≥ 2 × control).

Indicates marked decrease (≤ 0.5 × control).

Table 2

Effects of IL-1α on mRNA levels of ECM proteases and TIMPs. Human CMF from 3 patients treated with 10 ng/ml IL-1α for 6 h before extracting RNA. Pooled RNA was reverse transcribed and analyzed by real-time PCR array for 12 MMPs and 3 TIMPs. Fold increases for non-detectable samples assume a detection limit of 0.003% GAPDH.

Gene	Protein pseudonym	Basal (% GAPDH)	IL-1-stimulated (% GAPDH)	Fold change
ADAMTS1	METH-1	260.6	92.2	0.4b
ADAMTS8	METH-2	0.7	0.7	1.0
ADAMTS13	vWF-CP	0.3	–	0.01b
MMP1	Collagenase 1	20.7	167.8	8.1a
MMP2	Gelatinase A	144.0	128.9	0.9
MMP3	Stromelysin 1	–	0.46	153.3a
MMP7	Matrilysin 1	–	–	–
MMP8	Collagenase 2	–	–	–
MMP9	Gelatinase B	–	0.62	206.7a
MMP10	Stromelysin 2	0.31	27.4	88.4a
MMP11	Stromelysin 3	0.73	0.38	0.5b
MMP13	Collagenase 3	–	–	–
MMP14	MT1-MMP	9.9	13.0	1.3
MMP15	MT2-MMP	–	–	–
MMP16	MT3-MMP	6.6	2.0	0.3b
TIMP1		2210.0	4403.4	2.0a
TIMP2		644.9	600.7	0.9
TIMP3		–	–	–

Indicates marked increase (≥ 2 × control).

Indicates marked decrease (≤ 0.5 × control).

Relatively high mRNA expression levels of multiple ECM proteins were observed in CMF under basal conditions, including collagen types I, IV, V, VI, VIII, XII and XIV, laminins α2, β1 and γ1, and fibronectin (Table 1). By far the most highly expressed transcript was that of fibronectin, with collagen type VI also being prevalent. Treatment of CMF with IL-1α (6 h) did not affect levels of these highly expressed mRNAs, but did appear to have modulatory effects on some of those with lower basal levels (Table 1). For example, IL-1α increased mRNA levels of hyaluronan synthase (7.5-fold), type VII collagen (5-fold) and laminin β3 (2-fold), and decreased mRNA levels of collagen type XV (60% reduction) and laminin α1 (75% reduction).

We next analyzed expression of a range of ECM proteases and TIMPs in human CMF (Table 2). CMF expressed relatively high levels of ADAMTS1, MMP-2, MMP-1, MMP-14 and MMP-16 under basal conditions. TIMP-1 and TIMP-2 mRNA were expressed at much higher levels than the MMPs, but TIMP-3 transcripts were not detected. Treatment of CMF with IL-1α for 6 h increased mRNA levels of MMP-9 (200-fold), MMP-3 (150-fold), MMP-10 (90-fold), MMP-1 (8-fold) and to a lesser extent TIMP-1 (2-fold). In contrast, ADAMTS1, MMP-11 and MMP-16 mRNA levels were reduced 50–70% following IL-1α treatment. Four MMPs were undetectable at the mRNA level in control and IL-1α-stimulated CMF, namely MMP-7, MMP-8, MMP-13 and MMP-15.

RT-PCR analysis of cytokine-induced metalloproteinase mRNA expression

Overall analysis of the microarray data set revealed that the major effects of IL-1α were on expression of proteases, rather than the structural ECM proteins (Tables 1 and 2). We therefore selected six protease genes for further study, four that were increased following IL-1α treatment (MMP-1, MMP-3, MMP-9, MMP-10), one that was reduced by exposure to IL-1α (ADAMTS1) and one that was expressed at high basal levels but unaltered by IL-1α treatment (MMP-2). CMF from 3 patients were treated with 10 ng/ml IL-1α for 2–6 h before extracting RNA and performing real-time RT-PCR with specific Taqman primer/probes sets for each protease (Fig. 1). These experiments confirmed the array results in terms of relative basal expression levels and the response to IL-1α (increased mRNA levels of MMP-1, -3, -9 and -10, decreased levels of ADAMTS1 and no change in MMP-2). The time course experiments also revealed interesting differences in the temporal profile of mRNA expression for each protease (Fig. 1). The mRNA profiles for MMP-1 and MMP-10 peaked within 5 h, whereas those of MMP-3 and MMP-9 were still rising 6 h after IL-1α treatment. As expected, MMP-2 levels remained constant after IL-1α treatment. ADAMTS1 mRNA levels were reduced by 55–65% over the whole 2–6 h time period.

Fig. 1

Temporal profile of metalloproteinase mRNA expression in response IL-1α. CMF from 4 patients were exposed to 10 ng/ml IL-1α for 2–6 h before extracting RNA and performing real-time RT-PCR with primers for MMP-1, MMP-2, MMP-3, MMP-9, MMP-10 and ADAMTS1. Data are expressed as percentage relative to GAPDH endogenous control. ANOVA: P < 0.001 (MMP-1, MMP-3, MMP-10, ADAMTS1), P < 0.01 (MMP-9) and P = 0.556 (MMP-2). Newman–Keuls post hoc test: ***P < 0.001, **P < 0.01, *P < 0.05 for effect of IL-1α (n = 4).

The finding that IL-1α could reduce ADAMTS1 mRNA expression was of particular novelty, since this is the first description in cardiac fibroblasts. We therefore performed Western blotting experiments to determine whether IL-1α could also modulate ADAMTS1 expression at the protein level. As shown in Fig. 2, IL-1α reduced cellular expression of the zymogen form of ADAMTS1 protein (~ 120 kDa) by 50% 4–8 h after IL-1α stimulation. Protein levels of ADAMTS1 returned to near basal levels 24 h after treatment (Fig. 2).

Fig. 2

Effect of IL-1α on ADAMTS1 protein expression. CMF from 3 patients were treated with 10 ng/ml IL-1α for 2–24 h. Whole cell homogenates were prepared and immunoblotted for ADAMTS1 expression. Blots were reprobed with β-actin antibody to confirm equal protein loading. Representative blots are shown with approximate positions of molecular weight markers (in kDa) to the right. Bar chart depicts pooled densitometry data from different experiments.

We next compared the relative potency of IL-1α with that of another important myocardial proinflammatory cytokine, TNFα, in inducing ECM protease expression in CMF (Fig. 3). TNFα treatment alone had little or no effect on MMP-1, MMP-3, MMP-9, MMP-10 or ADAMTS1 mRNA levels after 6 h, in marked contrast to the response to IL-1α. When both IL-1α and TNFα were added simultaneously, trends towards further increases in MMP-1, -3, -9 and -10 levels, and further decreases in ADAMTS1, were observed; however these were not statistically significant (Fig. 3). Neither IL-1α nor TNFα, alone or in combination, altered MMP-2 mRNA expression levels.

Fig. 3

Effect of IL-1α and TNFα on metalloproteinase mRNA expression. CMF from 4 patients were exposed to 10 ng/ml TNFα or 10 ng/ml IL-1α (alone or in combination) for 6 h before extracting RNA and performing real-time RT-PCR with primers for MMP-1, MMP-2, MMP-3, MMP-9, MMP-10 and ADAMTS1. Data are expressed as percentage relative to GAPDH endogenous control. Newman–Keuls post hoc test: ***P < 0.001, **P < 0.01, *P < 0.05 for effect of TNFα and/or IL-1α versus vehicle control (n = 4).

Role of p38 MAPK in mediating IL-1-induced changes in metalloproteinase gene expression

IL-1α stimulated the p38 MAPK pathway in CMF as evidenced by increased phosphorylation of p38 MAPK and the downstream substrate HSP27 (Fig. 4). The p38-α/β inhibitor SB203580 (10 μM) and the p38-α/β/γ/δ inhibitor BIRB-0796 (1 μM) effectively inhibited IL-1α-induced HSP27 phosphorylation, but did not inhibit p38 MAPK phosphorylation per se (Fig. 4); consistent with their mode of action as inhibitors of p38 activity (Clark et al., 2007).

Fig. 4

IL-1α-induced activation of the p38 MAPK pathway. Following a 1 h pre-treatment with vehicle (1% DMSO), 10 μM SB203580 or 1 μM BIRB-0796, CMF were stimulated without or with 10 ng/ml IL-1α (10 or 20 min) before preparing whole cell homogenates and immunoblotting with phospho-specific and total expression antibodies for p38 MAPK and HSP27. Blots are representative of n = 3.

When CMF were pretreated with SB203580, an 80% reduction in IL-1α-induced MMP-3 mRNA expression was observed (Fig. 5). Similar results were obtained with BIRB-0796 (Fig. 6). Both SB203580 and BIRB-0796 reduced ADAMTS1 mRNA either in the absence or presence of IL-1α treatment (Figs. 5 and 6). In contrast, neither inhibitor modulated MMP-2 or MMP-9 mRNA expression (Figs. 5 and 6). Interestingly, BIRB-0796 reduced IL-1α-induced MMP-1 and MMP-10 mRNA levels by 70–80%, but SB203580 was without effect (Figs. 5 and 6).

Fig. 5

Effect of p38-α/β MAPK inhibitor SB203580 on metalloproteinase mRNA expression. CMF from 5 different patients were pre-treated with vehicle (1% DMSO) or 10 μM SB203580 before stimulating without (control, C) or with 10 ng/ml IL-1α for 6 h and measuring mRNA levels of MMP-1, MMP-2, MMP-3, MMP-9, MMP-10 and ADAMTS1 by real-time RT-PCR. Data are expressed as percentage relative to GAPDH endogenous control. Newman–Keuls post hoc test: ***P < 0.001, **P < 0.01, *P < 0.05, NS not significant, for effect of IL-1α or inhibitor as indicated (n = 5).

Fig. 6

Effect of p38-α/β/γ/δ MAPK inhibitor BIRB-0796 on metalloproteinase mRNA expression. CMF from the same 5 patients as used for SB203580 experiments were pre-treated with vehicle (1% DMSO) or 1 μM BIRB-0796 before stimulating without (control, C) or with 10 ng/ml IL-1α for 6 h and measuring mRNA levels of metalloproteinases by real-time RT-PCR. Data are expressed as percentage relative to GAPDH endogenous control. Newman–Keuls post hoc test: ***P < 0.001, **P < 0.01, *P < 0.05, NS not significant, for effect of IL-1α or inhibitor as indicated (n = 5).

Discussion

In this study we used a focused RT-PCR microarray to quantify the effects of IL-1α on expression of 41 ECM genes in human CMF. The majority of effects were on expression of metalloproteinases rather than structural ECM proteins. IL-1α (but not TNFα) increased mRNA expression of MMPs 1, 3, 9 and 10, and reduced expression of ADAMTS1 at both mRNA and protein levels. We also demonstrated that p38-α/β was important for IL-1α-induced MMP-3 expression and basal expression of ADAMTS1. Data obtained with BIRB-0796 suggested a role for additional p38 MAPK subtypes (γ or δ) in mediating IL-1α-induced MMP-1 and MMP-10 expression.

IL-1 has been previously shown to reduce total collagen synthesis ([3H]-proline incorporation) in neonatal and adult rat cardiac fibroblasts (Siwik et al., 2000; Xiao et al., 2008). More detailed Northern blot examination revealed that IL-1β decreased mRNA levels of fibrillar type I and III collagens, but increased expression of non-fibrillar type IV collagen and fibronectin mRNA after 24 h (Siwik et al., 2000). We observed only small decreases (< 25%) in type I and type IV collagen expression in human CMF after 6 h IL-1α treatment. We did not study type III collagen as this was not included on the microarray. We selected the 6 h time point to identify genes regulated directly downstream of IL-1 receptor signaling, as opposed to those that may be induced in response to the plethora of autocrine factors secreted by the cells in response to IL-1α stimulation (Turner et al., 2009). It is plausible therefore that the long-term changes in collagen expression that have previously been reported in cardiac fibroblasts (Siwik et al., 2000; Xiao et al., 2008) are not due to IL-1α stimulation per se, but rather occur through an indirect mechanism. The IL-1α-induced changes in mRNA levels of hyaluronan synthase (increase), type VII collagen (increase) and type XV collagen (decrease) are in close agreement with studies using human dermal fibroblasts (Campo et al., 2006; Mauviel et al., 1994; Kivirikko et al., 1999).

Previous studies have shown that cardiac fibroblasts and/or CMF express several MMPs including MMP-1, 2, 3, 9, 13 and 14 (Porter and Turner, 2009). Although stimulatory effects of proinflammatory cytokines on MMP-3 and MMP-9 expression in cardiac (myo)fibroblasts have been reported (Porter and Turner, 2009), there are no prior reports that IL-1 or TNFα can induce MMP-1 (collagenase 1) or MMP-10 (stromelysin-2) expression in this cell type. Indeed, neither basal nor induced MMP-10 expression has been previously reported in CMF. We found no evidence of mRNA expression for MMP-7, 8, 13 or 15 in human CMF. However, MMP-13 has previously been reported to be expressed by rat cardiac fibroblasts and its activity and expression increased by IL-1 or reactive oxygen species (Siwik et al., 2000; Siwik et al., 2001). Both MMP-7 and MMP-8 expression is evident in mouse and human myocardium (Lindsey et al., 2005; Lindsey et al., 2006; van den Borne et al., 2009). After MI, MMP-7 expression is localized to cardiomyocytes and macrophages (Lindsey et al., 2006) and MMP-8 expression to infiltrating macrophages and neutrophils (van den Borne et al., 2009). Taking these studies together with our data, it seems unlikely that CMF actually express MMP-7 or MMP-8.

Our microarray data indicated that human CMF express high levels of mRNA encoding TIMP-1 and TIMP-2, but not TIMP-3. Secretion of TIMP-1 and -2 protein (but not TIMP-3 or -4) was similarly reported recently in cultured neonatal rat cardiac fibroblasts (Brown et al., 2007). In that study, IL-1 increased secreted TIMP-1 levels by 6-fold after 48 h, but did not affect TIMP-2 expression (Brown et al., 2007). Thus, our data using human cells are in general agreement, with 6 h IL-1α treatment inducing a modest 2-fold increase in TIMP-1 mRNA expression, but TIMP-2 mRNA levels remained unaffected. However, conflicting reports also exist. For example IL-1 and TNFα increased TIMP-1 mRNA expression, but decreased TIMP-1 protein secretion in neonatal rat cardiac non-myocytes (presumably mostly cardiac fibroblasts) (Li et al., 1999). In that study, cytokines did not modulate TIMP-2 (mRNA or protein), but reduced TIMP-3 mRNA and protein expression (Li et al., 1999). TIMP-4 was not detected in cardiac non-myocytes, although it was evident in cardiomyocytes (Li et al., 1999). The consensus of these studies is that cardiac fibroblasts express TIMP-1 and TIMP-2, but not TIMP-4, and TIMP-3 expression remains contentious. Although TIMP-1 expression is modulated by proinflammatory cytokines, TIMP-2 expression is not.

IL-1α treatment reduced ADAMTS1 mRNA and protein expression in CMF. ADAMTS1 is a secreted metalloproteinase that is a potent inhibitor of angiogenesis, due to its ability to bind vascular endothelial growth factor (VEGF), reduce VEGF receptor signaling and consequently reduce endothelial cell proliferation (Luque et al., 2003). IL-1α also stimulates VEGF-A mRNA expression in human CMF (NA Turner, unpublished data). The combined opposing effects of IL-1α on expression of ADAMTS1 (decrease) and VEGF (increase) may therefore be important for promoting neovascularization of the infarcted myocardium. The inhibitory effects of IL-1α on ADAMTS1 expression in CMF contrast with the reported transient increases in ADAMTS1 expression observed in the heart within 3–6 h of myocardial ischemia in rats in vivo (Nakamura et al., 2004). These differences may reflect the opposing effects of hypoxia and proinflammatory cytokines on ADAMTS1 expression. For example, hypoxia induces rapid increases in ADAMTS1 expression in endothelial cells, but not skin fibroblasts (Hatipoglu et al., 2009). In chondrosarcoma cells, hypoxia has no modulatory effect, whereas IL-1 reduces ADAMTS1 expression (Kalinski et al., 2007); findings in agreement with our results in CMF. In contrast, proinflammatory cytokines have been reported to increase ADAMTS1 expression in some other cell types (Ng et al., 2006; Bevitt et al., 2003). Hence transcriptional regulation of ADAMTS1 is highly dependent on the cell type studied, as well as on the stimulus applied.

Of the five MMP genes that we studied, MMP-3 was the only one modulated by SB202580 treatment. IL-1α-induced MMP-3 expression was similarly inhibited by the global p38 inhibitor, BIRB-0796. Thus, IL-1α-induced MMP-3 expression likely occurs via a p38α/β-dependent mechanism in human CMF. Whether this reflects increased gene transcription and/or increased mRNA stabilization remains to be determined, but previous studies on dermal fibroblasts have described p38-dependent stabilization of MMP-3 mRNA following cytokine stimulation (Reunanen et al., 2002). We are not aware of any previous studies that have investigated the effects of the global p38 inhibitor BIRB-0796 on MMP expression. Our findings that IL-1α-induced MMP-1 and MMP-10 mRNA expression was inhibited by BIRB-0796 but not SB203580, suggest a key role for p38-γ/δ. BIRB-0796 is a potent and highly specific inhibitor of all four p38 MAPK subtypes (and also JNK2), and has no inhibitory activity against more than 65 other protein kinases when used at 1 μM concentrations (Bain et al., 2007).

ADAMTS1 expression was significantly reduced by SB203580 and BIRB-0796 treatment, both in the absence and presence of IL-1α stimulation, suggesting that ADAMTS1 expression is positively regulated by p38 MAPK, but negatively regulated by IL-1α, presumably via a p38-independent mechanism. There are very few prior reports describing the signaling pathways that regulate ADAMTS1 expression. Studies in endothelial cells have uncovered roles for protein kinase C in mediating VEGF-induced ADAMTS1 expression (Xu et al., 2006), and the phosphatidylinositol 3-kinase and p38 MAPK pathways in the response to hypoxia (Hatipoglu et al., 2009). The promoter region of the ADAMTS1 gene contains several transcription factor binding sites including three Sp1/Sp3 sites (Doyle et al., 2004) that are potential downstream targets of p38 MAPK (D'Addario et al., 2006).

A strength of our study was the use of adult human CMF derived from multiple patients, rather than the more frequently used single strain cell lines (often neonatal) of rabbit and rodent cardiac fibroblasts, which can exhibit important differences compared with human cells (Porter and Turner, 2009; Agocha et al., 1997). The responses to IL-1α were reproducible across multiple patients, which contrasts with responses to TNFα in these cells which are more variable and appear to depend on the patient donor (Porter et al., 2004a). A limitation of our study was the focus on gene expression and the assumption that changes in mRNA levels are accompanied by changes in protein expression and/or activity. However, for ADAMTS-1 we did confirm that both mRNA and protein levels were similarly modulated by IL-1α. Our strategy of pooling RNA samples from three different patients for the array precluded statistical analysis. It is possible that smaller statistically significant changes were missed using this approach. It is also feasible that some of the larger changes in mRNA expression observed on the array may not have been evident in all three samples. However, all six of the genes that we pursued using real-time RT-PCR proved to be reproducibly modulated by IL-1α.

In summary, IL-1α treatment had minimal effect on expression of structural ECM proteins in human CMF, but selectively increased expression of several MMPs (1, 3, 9 and 10) and reduced mRNA and protein levels of ADAMTS1. Of these metalloproteinases, expression of MMP-3 and ADAMTS1 was regulated by p38-α/β, whereas MMP-1 and MMP-10 expression appeared to be mediated via p38-γ/δ subtypes. Thus, we have revealed important new insights into the expression profile of structural ECM proteins, ECM proteases and TIMPs in human CMF in response to IL-1α, a proinflammatory cytokine elevated in the heart after MI. Our findings may help to explain some of the key early catabolic processes that occur in the post-MI heart in man.

Experimental procedures

Materials

Recombinant human IL-1α and TNFα were purchased from Invitrogen (Paisley, UK). SB203580 was purchased from Calbiochem (Nottingham, UK) and BIRB-0796 was provided by the Division of Signal Transduction Therapy at Dundee University (Dundee, Scotland).

Cell culture

Right atrial appendage biopsies from patients undergoing elective coronary artery bypass surgery at the Leeds General Infirmary were obtained following local ethical committee approval and informed patient consent. Primary cultures of cardiac fibroblasts were harvested, characterized as myofibroblasts (α-smooth muscle actin- and vimentin-positive) and cultured as we have described previously (Porter et al., 2004b; Turner et al., 2003; Mughal et al., 2009). Experiments were performed on cells from passages 3–5.

RT-PCR microarray

CMF from 3 patients were treated with or without 10 ng/ml IL-1α for 6 h. We selected this concentration based on our previous report describing the optimal IL-1α concentration for inducing proinflammatory cytokine expression in these cells (Turner et al., 2009). Cellular RNA was extracted and pooled before reverse transcription (Turner et al., 2007). Expression levels of 41 structural ECM proteins, ECM proteases and TIMPs were assessed using a focused SYBR Green-based real-time PCR microarray (RT2 Profiler Human Extracellular Matrix and Adhesion Molecules, SABiosciences, www.sabiosciences.com). Expression data are presented as percentage of GAPDH expression using the formula 2−∆CT × 100. For undetectable transcripts, calculation of fold increase assumed the detection limit of the array was 15 cycles more than the housekeeping gene average (~ 0.003% GAPDH).

Quantitative RT-PCR

For expression studies, cells from 4 patients were treated with 10 ng/ml IL-1α or TNFα for appropriate times (2–6 h). For inhibitor studies, cells from 5 patients were pre-treated with vehicle (1% DMSO), 10 μM SB203580 or 1 μM BIRB-0796 for 1 h before addition of 10 ng/ml IL-1α for 6 h. After incubation, total RNA was extracted and cDNA prepared as described previously (Turner et al., 2007). Real-time RT-PCR was performed using an Applied Biosystems 7500 Real-Time PCR System and intron-spanning human MMP-1, MMP-2, MMP-3, MMP-9, MMP-10 and ADAMTS1 primers and Taqman probes (Applied Biosystems), as described in Supplementary Table 1. Data are expressed as percentage of GAPDH endogenous control levels (Hs99999905_m1 primers) using the formula 2−∆CT × 100, in which CT is the threshold cycle number.

Western blotting

Whole cell homogenates were prepared as described previously (Turner et al., 2001). ADAMTS1 protein expression was determined using a rabbit anti-human ADAMTS1 polyclonal antibody (A4851; Sigma, Poole, UK) raised against the propeptide region of the human ADAMTS1 zymogen (apparent molecular weight 110–120 kDa). Equal protein loading was confirmed with mouse anti-β-actin monoclonal antibody (ab8226; Abcam, Cambridge, UK). Phosphorylation of p38 MAPK(Thr180/Tyr182) and HSP27(Ser82) was determined by immunoblotting with rabbit phospho-specific polyclonal antibodies (#9211 and #2401; Cell Signaling Technology, Hitchin, UK). Membranes were re-probed with mouse monoclonal p38 MAPK and HSP27 expression antibodies (#9217 and #2402; Cell Signaling Technology) to confirm equal protein loading.

Statistical analysis

Results are mean ± SEM with n representing the number of experiments on cells from different patients. Data were analyzed as ratios using repeated measures one-way ANOVA and Newman–Keuls post hoc test (GraphPad Prism software, www.graphpad.com). P < 0.05 was considered statistically significant.

Disclosure statement

None.

Role of the funding source

The work was funded by a project grant from the British Heart Foundation. The funding source had no direct involvement in the study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the paper for publication.