|Home | About | Journals | Submit | Contact Us | Français|
Chronic β-adrenergic receptor (β-AR) overstimulation, a hallmark of heart failure, is associated with increased cardiac expression of matrix metalloproteinases (MMPs). MMP-1 has been shown to cleave and activate the protease-activated receptor 1 (PAR1) in non-cardiac cells. Here, we hypothesized that β-AR stimulation would result in MMP-dependent PAR1 transactivation in cardiac cells.
β-AR stimulation of neonatal rat ventricular myocytes (NRVMs) or cardiac fibroblasts (CFs) with isoproterenol (ISO) transduced with an alkaline phosphatase-tagged PAR1 elicited a significant increase in AP-PAR1 cleavage. This ISO-dependent cleavage was significantly reduced by the broad-spectrum MMP inhibitor GM6001. Importantly, specific MMP-13 inhibitors also decreased AP-PAR1 cleavage in ISO stimulated NRVMs, as well as in NRVMs stimulated with conditioned-medium from ISO-stimulated CFs. Moreover, we found that recombinant MMP-13 stimulation cleaved AP-PAR1 in NRVMs at DPRS42↓43FLLRN. This also led to the activation of ERK1/2 pathway through Gαq in NRVMs and via the Gαq/ErbBR pathways in CFs. MMP-13 elicited similar levels of ERK1/2 activation, but lower levels of inositol phosphates generation, in comparison to thrombin. Finally, we demonstrated that either PAR1 genetic ablation or pharmacological inhibition of MMP-13 prevented ISO-dependent cardiac dysfunction in mice.
In this study, we demonstrate that β-AR stimulation leads to MMP-13 transactivation of PAR1 in both cardiac fibroblasts and cardiomyocytes and this likely contributes to pathological activation of Gαq- and ErbB receptor-dependent pathways in the heart. We propose that this mechanism may underly the development of β-AR overstimulation-dependent cardiac dysfunction.
Seven transmembrane G-protein coupled (GPCR) receptors, such as adrenergic, angiotensin, endothelin and serotonin receptors, have been implicated in cardiac dysfunction and hypertrophy1. Among the GPCR family, the protease-activated receptors (PARs) are unique because they are activated by proteolytic cleavage of their extracellular N-terminal domain, unmasking a self-encoded, tethered ligand. The PAR family is composed of four members, PARs 1-4. PAR2 is activated by various proteases such as trypsin and tryptase, while PARs 3 and 4 are activated by thrombin 2. The high affinity tethered ligand for PAR1 is classically unmasked by the proteolytic activity of thrombin. Thrombin stimulation of PAR1 has been shown to trigger neonatal cardiomyocyte hypertrophy and cardiac fibroblast DNA synthesis 3, 4. Our group has recently demonstrated that while cardiac-restricted PAR1 overexpression led to heart failure 5, PAR1-deficient mice were less susceptible to myocardial injury, highlighted by their reduced left ventricular dilation and superior cardiac function compared to wild type (WT) mice after cardiac ischemia-reperfusion5. This finding in PAR1-deleted mice was confirmed by the cardioprotective effect of the selective PAR1 antagonist SCH79797 in a rat model of I/R 6. During ischemia, coagulation factors, such as thrombin, are known to leak through the damaged endothelial barrier, suggesting that in ischemic conditions, thrombin is initially mainly responsible of PAR1 activation in heart. Indeed, thrombin’s role in ischemic injury was demonstrated in a rabbit model of I/R where the thrombin inhibitor, hirudin, significantly reduced infarct size 7. Interestingly, thrombin is not the only protease capable of PAR1 cleavage and activation. In particular, matrix metalloproteinase 1 (MMP-1) has been shown to activate PAR1 in platelets, tumor cells and endothelial cells, promoting thrombosis, tumorigenesis or angiogenesis respectively 8–10. Though a homologue of MMP-1 has not been identified in mouse or rat, MMP-13 is considered to be the major interstitial collagenase in these species 11. MMP-13 was found to be expressed in the healthy adult mouse heart 12 but beyond its collagen-degrading function, its role in pathological cardiac remodeling remains unexplored. Interestingly, it has been described that pressure overload, β1-adrenergic receptor (β-AR) overstimulation or myocardial infarction elevates cardiac MMP-13 expression 13–15.
In this study, we determined whether activation of PAR1 in cardiac cells and the heart occurs in a non-ischemic pathological condition, i.e without implication of thrombin. We show that β-AR overstimulation leads to PAR1 cleavage and transactivation through MMP-13 in both cardiomyocytes and cardiac fibroblasts, and that this β-AR-elicited transactivation activity can be transmitted from fibroblasts to myocytes. We propose that the MMP-13/PAR1 axis plays a critical pathophysiological role in a non-ischemic model of murine heart failure.
Generation of PAR1 KO mice was previously described 5. Mini-osmotic pumps (1007D, Alzet Corporation), delivering filtered solutions of (−) l-isoproterenol bitartrate (ISO) or vehicle (0.002% ascorbic acid in saline), dose-adjusted to reflect 30mg/kg/day of (−) l-isoproterenol HCl (i.e. 43.8 mg/kg/day of (−) l-isoproterenol bitartrate), were implanted in male wildtype (WT) or male PAR1 KO mice (10–12 weeks old) under anesthesia (0.75% Isoflurane). WAY170523 (7.5 mg/kg) dissolved in a mixture of 5% DMSO, 40% PEG-400, and 55% saline, pH 7.4 or corresponding vehicle were injected intraperitoneal once daily for 7 days. All animal procedures were performed in accordance with the guidelines of the Department of Laboratory Animal Medicine and the University Committee on Animal Resources at the University of Rochester Medical Center.
Transthoracic 2D and M-mode echocardiography analysis was used to assess heart function in conscious mice with a VisualSonics Vevo 770 echocardiography machine equipped with a 30MHz probe (VisualSonics) (PAR1 null mice). Echocardiography on ISO +/− WAY170523 treated mice was performed using a VisualSonics Vevo 2100 equipped with an 18–38MHz probe (VisualSonics). Echocardiographic data from both systems were obtained and analyzed as we have recently described in detail 16, 17.
Cardiac fibroblasts (CFs) were obtained as described in the previous section. After 4 to 5 days in culture, cells were split using trypsin 0.05% and plated in 12-well plates. At 80–90% confluency, cardiac fibroblasts were serum-starved in DMEM and 1% pencillin/streptomycin for 24 hours and stimulated for 2, 4 or 8 hours with saline or ISO (5 μmol/L). Conditioned-media (CM) were collected. CM were incubated with the specific MMP-13 inhibitor WAY170523 (1 μmol/L), saline and propranolol (5 μmol/L) for 15 min and then applied on AP-PAR1 transduced NRVMs for 24 hours.
AP activity in the cell supernatant, which is proportional to the cleavage activity, was quantified using the Phospha-Light™ chemiluminescent reporter assay for secreted alkaline phosphatase kit (Tropix) as previously described 18.
MMP-13 cardiac activity was accessed using the SensoLyte 520 MMP-13 fluorometric Assay Kit (Anaspec). Briefly, mouse or human heart tissues were homogenized in assay buffer containing 0.1% (v/v) Triton-X 100, and then centrifuged for 15 min at 10,000xg at 4°C. 50 μL of the homogenate supernatant were incubated in a 96 well plate with the MMP-13 substrate for 30 min at 37°C. 50 μL of cardiac fibroblasts supernatant was similarly used to measure MMP-13 activity in cell culture medium. SensoLyte 520 MMP-13 fluorometric Assay Kit was also used to determine the IC50 value of WAY170523 for MMP-13. The fluorescence intensity was measured using a microplate reader.
Tissue was harvested from the LV free wall near the apex of 8 male patients (59.6 +/− 3.1 yrs of age) in end-stage (Stage D) HF at the time of LVAD placement (HeartMate or HeartMate II; Thoratec, Pleasanton, CA) as either destination therapy or bridge to transplant. Samples from four patients each with the diagnosis of ischemic or non-ischemic HF etiology were analyzed. Non-failing human heart tissue was obtained from the LV free wall of three male organ donor hearts rejected for transplant due to physical incompatibility. LV tissue obtained from surgery was immediately frozen in liquid nitrogen and stored at −140°C. All harvest and use of human tissue was performed in accordance with NIH and University of Rochester Medical Center institutional review board guidelines. Supplemental Methods and Materials can be found in Supplemental Data online.
The authors had full access to the data and take responsibility for its integrity. All authors have read and agree to the manuscript as written.
All results are expressed as mean ±SEM. Different groups were compared through one-way ANOVA followed by Newman-Keuls’s test. Comparison between two groups was assessed by t-test. All calculations were performed using the GraphPad Prism 5.0 program.
In the heart, chronic β-adrenergic receptor stimulation is known to induce MMP expression and activity 14. Therefore, we hypothesized that chronic β-AR stimulation with the agonist ISO could lead to MMP-dependent (and thrombin-independent) PAR1 cleavage in cardiac cells. To test this hypothesis, we first transduced NRVMs with the AP-PAR1 expressing adenovirus. Compared to vehicle-treated cells, isoproterenol (ISO, 10 μmol/L) stimulation led to a significant 3-fold increase in N-terminal PAR1 cleavage after 24h of stimulation (Figure 1A). This ISO-dependent cleavage was highly attenuated by the PAR1 specific blocking antibody H-111 (10 μg/mL) that recognizes the N-terminal domain of PAR1 (Supplemental Figure 1A). We excluded a potential involvement of thrombin in this process as hirudin (2.5U/mL), a specific inhibitor of thrombin, was unable to prevent the ISO-dependent PAR1 cleavage (Supplemental Figure 1A). To assess MMP involvement in ISO-induced PAR1 cleavage, we stimulated AP-PAR1 transduced NRVMs with ISO in the presence of the broad-spectrum MMP inhibitor (GM6001, 10 μmol/L) or the endogenous tissue inhibitor of MMP, TIMP-3 (10 ng/ml). Inhibition of general MMP activity with either GM6001 or TIMP-3 significantly reduced ISO-dependent PAR1 cleavage after 24h of stimulation (Figure 1A). A prior report suggested that among MMP-1/2/3/7/8/9, only MMP-1 (Collagenase 1) can cleave PAR1 in human cells 8. Rodents do not express MMP-1. However, MMP-13 is considered to act as the major interstitial collagenase in these species 11. Therefore, we sought to test our hypothesis that ISO-mediated PAR1 cleavage is mediated by MMP-13 in rodent cardiac cells. Indeed, inhibition of MMP-13 activity, using two different highly specific small molecule MMP-13 inhibitors (Figure 1B), lead to a significant reduction (approximately 50%) of ISO-induced PAR1 cleavage in NRVMs (Figure 1B). Importantly, unlike GM6001, neither of the two MMP-13 specific inhibitors modified the baseline cleavage observed in NRVMs (Supplemental Figure 1C).
Considering that CFs are known to be a major source of MMPs in the heart 19, we hypothesized that activated CFs could release MMPs capable of cleaving both fibroblast and myocyte PAR1 in an autocrine or paracrine manner, respectively. To investigate this possibility, we first stimulated CFs with vehicle or ISO (10 μmol/L) for 2h to 8h. We then collected the CM and applied it to AP-PAR1 transduced NRVMs. Importantly, we ruled out a direct effect of any residual ISO contained in the CM by adding the β-adrenergic receptor antagonist propranolol (5 μmol/L) before adding the CM to the NRVM culture. As shown in Figure 2A, CM obtained from non-stimulated (NS) CFs increased PAR1 cleavage in NRVMs. MMP-13 does not appear to be involved in this NS-CM effect, as the MMP-13 inhibitor WAY (1 μmol/L) had no significant effect on PAR1 cleavage under these conditions. Importantly, ISO-CM (2h, 4h and 8h) significantly induced PAR1 cleavage compared to NS-CM (≈7 fold increase with the 8h ISO-CM vs ≈3 fold increase for the NS-CM). This ISO-CM-dependent cleavage is mainly mediated by MMP-13 as demonstrated by the strong reduction of PAR1 cleavage in presence of WAY (Figure 2A). To confirm the involvement of MMP-13 in ISO-dependent cleavage of PAR1, we quantified MMP-13 activity in the CM of ISO stimulated CFs. ISO (10 μmol/L) stimulation significantly increased MMP-13 activity after 8 hours of stimulation compared to the non-stimulated condition (Figure 2B).
It has been previously shown that of several MMPs tested, human MMP-1 (collagenase 1) can cleave and activate PAR1 in human non-cardiac cells 8–10. Further, rodent MMP-13 is thought to play the role of MMP-1, which is not expressed in rodents 11. As suggested by our data obtained with ISO stimulation in NRVMs or CFs-CM experiments, we hypothesized that MMP-13 may directly cleave PAR1. To investigate this possibility, we transduced NRVMs with the alkaline phosphatase (AP)-tagged PAR1 adenovirus and applied different doses of recombinant MMP-13 protein to the cells. As shown in Figure 3A, MMP-13 induced a significant dose-dependent cleavage of PAR1 in NRVMs (> 2-fold increase with MMP-13 40 nmol/L compared to baseline). Several proteases are known to cleave the PAR1 N-terminal domain at different sites 20. To identify the exact MMP-13 cleavage site, we used a 26 amino acid peptide (TR26, PAR1 residues 36–61) corresponding to the PAR1 N-terminal domain, containing the thrombin cleavage site and flanking regions as previously described 9 (Figure 3B). When we incubated the TR26 peptide with thrombin, the expected cleavage peptide was obtained (residues 42–61) as determined by high-resolution mass spectrometry. However, when TR26 was incubated with MMP-13, we obtained a different peptide 43–61, demonstrating that MMP-13 cleaves the PAR1 N-terminal domain at a unique site DPRS42↓43FLLRN (Figure 3B). To functionally validate this new MMP-13 cleavage site, we transfected H9C2 cells, a myoblastic cell line originating from embryonic rat heart tissue, with WT AP-PAR1 or three different PAR1 mutants; an AP-R41A-PAR1 mutant which was described to be uncleavable by thrombin 21, an AP-P40N-PAR1 mutant previously reported to be insensitive to MMP-1 9 and an AP-F43R-PAR1 mutant where phenylalanine in position 43 was mutated to an arginine. As shown in Figure 3C, MMP-13 (40 nmol/L)-dependent PAR1 cleavage is totally absent in the AP-R41A-PAR1 or AP-F43R PAR1 transfected cells but was similar in WT AP-PAR1 and AP-P40N-PAR1 transfected cells. Interestingly, and in contrast to a prior report 9, we also found that thrombin (10 nmol/L)-dependent PAR1 cleavage is significantly reduced in AP-P40N-PAR1 transfected cells and to a smaller extent in AP-F43R PAR1 expressing cells (Figure 3C).
Although it remains somewhat unclear, some proteases, such as cathepsin G, can cleave PAR1 downstream of what is believed to be the functional SFLLRN tethered ligand. As a consequence, such proteolysis can inactivate PAR1 22. Unlike thrombin, we found that MMP-13 cleaves PAR1 at the DPRS42↓43FLLRN site. To investigate if this new MMP-13-generated tethered ligand could activate PAR1-dependent signaling pathways, we stimulated NRVMs or CFs, both of which express functional PAR1 4, with a low dose of MMP-13 (10 nmol/L) and assessed the activation of ERK1/2. In both cell types, MMP-13 stimulation produced a significant time-dependent increase of ERK1/2 phosphorylation (4–6 fold increase at maximal response) (Figure 4A–B). In the presence of the selective PAR-1 antagonist SCH79797 (1 μmol/L), the activation of ERK1/2 was significantly decreased in either NRVMs or CFs (Figure 4A–B). PAR1-dependent ERK1/2 activation by MMP-13 was confirmed by the significant reduction of ERK1/2 phosphorylation in presence of another PAR1 specific antagonist RWJ-58259 (5 μmol/L) or in the presence of the PAR1 blocking antibody S19 (15 μg/ml) in NRVMs or CFs, respectively (Figure 4C–D). Interestingly, the level and kinetics of MMP-13-dependent, PAR1-mediated activation of ERK1/2 was similar to that achieved by stimulation with thrombin at the same concentration (10 nmol/L and inhibited by SCH79797) in both CFs (Supplemental Figure 2) and NRVMs 4. Moreover, we found that the p38 MAP kinase was also activated by MMP-13 in a PAR1-dependent manner, as SCH79797 significantly reduced P38 phosphorylation in NRVMs (Supplemental Figure 3). To further demonstrate that MMP-13 activates PAR1, we measured specific PAR1 internalization in NRVMs or CFs using an ELISA method that quantifies PAR1 surface expression. Although not as strong as the thrombin (10 nmol/L)-dependent internalization, a significant decrease of PAR1 cell surface expression was measured in both cell types 2 hours after MMP-13 (40 nmol/L) stimulation (Figure 5A–B), demonstrating that MMP-13 promoted significant PAR1 internalization in cardiac cells.
Considering that PAR1 can be coupled to Gαq in heart 4, we investigated the capacity of MMP-13 to induce ERK1/2 signaling through the Gαq pathway. We transduced CFs with a GFP expressing control adenovirus or with a GqI-GFP expressing adenovirus to overexpress the GqI-GFP blocking peptide, a well-characterized competitor of endogenous Gαq proteins 23. As shown in Figure 6A, expression of the GqI-GFP peptide in CFs significantly reduced ERK1/2 phosphorylation after 5 and 10 min of MMP-13 stimulation (Figure 6A). However, late MMP-13-dependent ERK1/2 activation (20 to 30 min time points) was not modified in the presence of the GqI-GFP peptide compared to the control GFP. To further confirm that MMP-13 leads to Gαq pathway activation, we measured the generation of inositol phosphates in cardiac cells. A prior study demonstrated that thrombin can directly activate inositol phosphate release in cardiac tissue 24. We found that MMP-13 (40 nmol/L) stimulation elicited a statistically significant increase in IPs generation in NRVMs or CFs (+34% and +22% respectively), although is considerably lower than the levels of IPs generation with thrombin (10 nmol/L) (Figure 6B). Finally, we discovered that MMP-13 also activated ERK1/2 through transactivation of the ErbB receptor pathway in CFs, as the ErbB-1/4 tyrosine kinase inhibitor AG1478 (1 μmol/L) significantly attenuated ERK1/2 phosphorylation at 10 and 20 min of MMP-13 stimulation (Figure 6C). However, we found that MMP-13-dependent ErbB-1/4 transactivation was specific for CFs, as AG1478 failed to reduce ERK1/2 phosphorylation in cardiomyocytes (data not shown).
To test whether PAR1 stimulation could be involved in pathologic cardiac remodeling induced by chronic β-AR overstimulation, we used an acute pharmacological model of heart failure that consisted of ISO infusion via osmotic pumps. We first found that MMP-13 activity is increased in hearts of WT mice infused with ISO in a time-dependent manner, with a statistically significant increase at 7 days of infusion (+33% compared to baseline activity) (Figure 7A). After performing baseline echocardiography at day 0 (D0), we infused WT and PAR1 deficient mice (PAR1 KO) with ISO for 1 week. Importantly, PAR1 KO mice exhibit similar cardiac β-AR density as WT mice (Supplemental Figure 4A). Following 7 days of chronic ISO infusion, the cardiac hypertrophic response was not different between PAR1 KO and WT mice as shown by the similar increase in heart weight to body weight ratio (HW/BW) in both genotypes (Supplemental Figure 4B). However, in contrast to WT mice, cardiac function measured by echocardiography was fully conserved in PAR1 KO mice after 7 days (D7) of ISO infusion (Figure 7B). Moreover, left ventricular systolic diameter was significantly increased in WT mice but was not modified in PAR1 KO mice (Supplemental Figure 4C). To investigate a putative role of MMP-13 in ISO-dependent cardiac dysfunction, we infused WT mice with ISO for 7 days and concurrently delivered the specific MMP-13 inhibitor WAY170523 daily by i.p injection. Remarkably, we found that WAY170523 completely abolished ISO-dependent increase of the left ventricular systolic diameter (Supplemental Figure 4D) and preserved cardiac function in ISO-infused animals (Figure 7C) without modifying the hypertrophic response (Supplemental Figure 4E) similar to the PAR1 KO animals.
Finally, we found that MMP-13 protein expression, as well as MMP-13 activity, are significantly increased in the failing human heart to a similar extent in non-ischemic and ischemic failing tissues (Figure 7C–D).
We have recently shown that PAR1 is involved in pathological cardiac remodeling, including an important role in myocardial I/R 5. After ischemic cardiac injury, it has been proposed that prothrombin from the blood leaks into the myocardium through the damaged endothelial barrier, and the subsequent generation of thrombin leads to pathological PAR1 signaling in the heart 7. However, thrombin is not the only protease capable of activating PAR1, suggesting that cardiac PAR1 might be also activated in non-ischemic pathological conditions (i.e. possibly in the absence of thrombin). Our study provides novel data that PAR1 can be cleaved and transactivated by MMP-13 upon β-AR overstimulation, which may have pathologic consequences in cardiac remodeling in vivo. We show that (i) β-AR stimulation leads to PAR1 cleavage and transactivation, mainly through secretion of cardiac fibroblast-dependent MMP-13; (ii) MMP-13 cleaves PAR1 at the DPRS42↓43FLLRN site and activates PAR1 signaling in cardiac cells; and (iii) genetic ablation of PAR1 or pharmacological MMP-13 inhibition protects against adrenergic overstimulation-induced cardiac dysfunction which is associated with elevated MMP-13 expression and hyperactivity in both mouse and human heart.
The role of cardiac PAR1 activation by other proteases other than thrombin is poorly understood. Recently, MMP-1-dependent PAR1 activation was found to promote invasion and tumorigenesis in human cancer cells as well as to via activation of platelets induce arterial thrombosis8, 9. We have identified MMP-13 as a new protease for the PAR1 N-terminal domain. Classically, the PAR1 N-terminal domain is cleaved by thrombin at the DPR41↓42SFLLRN site, unmasking a SFLLRN tethered ligand, which binds and activates PAR1. We found that MMP-13 cleaves PAR1 upstream of the thrombin site at the following position DPRS42↓43FLLRN and observed that MMP-13 is unable to cleave the R41A or the F43R PAR1 mutants. Although the R41A mutation does not immediately flank the cleavage site, it suggests that a mutation of one amino acid downstream of the cleavage site would highly impact the MMP-13 cleavage efficiency. This concept is supported by the reduction of thrombin cleavage efficiency observed with the P40N mutant and to a smaller extent with the F43R PAR1 mutant. Based on our data, it appears that the canonical tethered ligand SFLLRN is not the only ligand capable of efficient PAR1 activation. For instance, it was recently shown that MMP-1, which does not cleave PAR1 at the thrombin site (although the exact cleavage site is still controversial 9, 25), can activate PAR1. Although not as efficient as a similar concentration of thrombin in terms of PAR1 cleavage, we found that 10 nmol/L of MMP-13 is sufficient to cleave PAR1 and activate PAR1-dependent signaling pathways (ERK, p38, IP3), demonstrating the capacity of the shorter FLLRN tethered ligand to activate endogenous PAR1 in cardiac cells.
MMP-13 is well described to play a major role in cartilage biology 26. In the healthy heart, MMP-13 is expressed at low levels and its role in pathological cardiac remodeling remains largely unknown, although its cardiac expression is increased in pathological conditions 27, 28. Here we show that β-AR stimulation of cardiac cells transactivates PAR1 mainly through MMP-13-mediated cleavage, demonstrating the existence of a new type of protease mediated cross-talk between a Gαs- and a Gαq-coupled receptor in heart. Indeed, based on our CFs conditioned-medium experiment, we propose that stimulation of β-AR on CFs can lead to an MMP-13-dependent autocrine activation of PAR1 but also to the activation of the PAR1 located at the surface of the cardiomyocytes through a paracrine release of MMP-13 by CFs. Confirming our data obtained at the protein level, it was recently described that β-AR activation leads to an increase of MMP-13 mRNA level in both adult mouse CFs and cardiomyocytes 29.
It has been suggested that transactivation of ErbBR after GPCRs stimulation could be a survival pathway in cardiomyocytes 30. However, MMP-13 transactivates ERK1/2 through the ErbBR pathway in cardiac fibroblasts but not in cardiomyocytes as previously described for thrombin PAR1-dependent activation of ERK1/2 4. This observation suggests that MMP-13 elicits the activation of a pathological PAR1/Gαq pathway 31 without activation of a pro-survival ErbBR-dependent signaling in cardiomyocytes. Conversely, GPCR-dependent ErbBR transactivation in cardiac fibroblasts is known to lead to pathological events such as proliferation 4, 32, inflammatory cytokine release 33 and extracellular matrix formation 34, which are associated with cardiac dysfunction.
Conditioned-medium from ISO-stimulated cardiac fibroblasts leads to increased MMP-13-mediated PAR1 cleavage compared to direct ISO stimulation of cardiomyocytes, suggesting that the main source of MMP-13 originates from cardiac fibroblasts. Although a MMP-13 selective inhibitor drastically attenuates ISO-induced PAR1 cleavage, the inhibition is not complete. As very recently demonstrated by Lee et al 35 using a mass spectrometry approach, and diverging from what was shown by Boire et al 8, MMP-3, MMP-8 and MMP-9 may also cleave PAR1 at the DPR41↓42SFLLRN site (whereas MMP-1 may cleave at the D39↓40PRSFLLRN site). Therefore, it is possible that the remaining ISO-dependent PAR1 cleavage observed in presence of the selective MMP-13 inhibitor is due to the activity of one or more of these three MMPs that are known to be expressed in the heart 12.
MMP-13 was previously described to induce ERK1/2 pathway in a mouse kidney-like cell line 36 but whether PAR1 triggers this intracellular pathway activation remains unknown. Importantly, the authors of that study experimentally ruled out the possibility that MMP-13 itself was able to cleave HB-EGF or TGF-α to subsequently induce ERK1/2 activation through EGF receptor transactivation. Here, we discovered that MMP-13 leads to ERK1/2 activation mainly through direct cleavage and subsequent activation of PAR1 in primary cardiac cells, as shown by the significant decrease of MMP-13-dependent ERK1/2 activation in presence of two different PAR1 specific antagonists, or in presence of a PAR1 N-terminal domain blocking antibody. Interestingly, while both thrombin and MMP-13 at the same concentration elicited similar levels of PAR1 mediate ERK1/2 activation, MMP-13 was not as robust as thrombin in the generation of IPs, suggesting a possible ligand-bias of downstream PAR1 signaling by MMP-13. These data likely suggest that the generation of the FLLRN ligand by MMP-13 mostly promotes the activation of the ERK1/2 pathway. As now clearly demonstrated for GPCRs, including the angiotensin 1 receptor (AT-1R) 37, biased ligands can selectively activate unique intracellular pathways by promoting a specific receptor conformation. Although a detailed study is required to substantiate this possibility for PAR1, we propose that MMP-13-dependent generation of the FLLRN ligand elicits a PAR1 conformational change that is divergent from that of thrombin, and that differentially activates downstream signaling pathways in terms of potency and specificity. In pathological conditions, βAR overactivation plays a critical role in the development of cardiac dysfunction 38. The underlying mechanism(s), however, remains elusive. We propose that transactivation of pathological PAR1/Gαq signaling through MMP-13, in both cardiomyocytes and cardiac fibroblasts, participates in the establishment of βAR overactivation-dependent cardiac dysfunction. Interestingly, PAR1 overactivation in human heart failure, measured by its level of phosphorylation, has been previously described in both ischemic and in non-ischemic failing human hearts 39, suggesting that PAR1 is chronically activated in pathological conditions. Here, we described that activation of PAR1 by MMP-13 is critical for βAR overactivation-induced cardiac dysfunction. Indeed, in contrast to WT mice, we found that mice with a genetic ablation of PAR1 or with a pharmacological inhibition of MMP-13 exhibit a compensated cardiac hypertrophy with no sign of cardiac dysfunction in a non-ischemic model of heart failure. Because we showed that MMP-13 expression and activity are increased in either ischemic or non-ischemic human failing tissue, we suggest that MMP-13 participates in pathological PAR1 signaling activation in failing human heart regardless of etiology. Consequently, it is likely that MMP-13 also elicits PAR1 activation in ischemic condition. We propose that in the early stage of an ischemic event, thrombin may be the main activator of PAR1 in the heart. However, we believe that MMP-13 mostly contributes to chronically activate deleterious PAR1-dependent pathways within the heart, leading to the maintenance and progression of pathologic cardiac remodeling. Notably, MMP-13 expression and activity remain elevated months to years post-ischemia (or in essentially any etiology of end-stage HF). Importantly, chronic heart failure of both ischemic and non-ischemic etiology is accompanied by chronically elevated circulating plasma catecholamines, and we have demonstrated that activation of PAR1 by MMP-13 is critical for βAR overactivation-induced cardiac dysfunction.
In summary, we demonstrate that β-AR overstimulation transactivates PAR1 through MMP-13 in both cardiac fibroblasts and cardiomyocytes, leading to the activation of the pathological Gαq pathway (Figure 8). We propose that this mechanism participates in the development of β-AR overactivation-dependent cardiac dysfunction. Since we found that PAR1 is involved in pathological cardiac remodeling in ischemic 5 but also now in non-ischemic conditions, we believe that specific PAR1 antagonists might represent a new therapeutic paradigm to prevent or treat cardiac dysfunction.
Chronic heart failure of both ischemic and non-ischemic etiology is accompanied by chronically elevated circulating plasma catecholamines. Chronic overstimulation of cardiac β-adrenergic receptors (β-AR) is known to be a key contributor to the establishment and maintenance of pathological cardiac remodeling. However, the mechanism by which β-AR overactivation contributes to heart failure remains poorly understood. Herein, we have discovered that cardiac protease activated receptor 1 (also known as the thrombin receptor) can be transactivated by β-AR overstimulation via the matrix metalloproteinase 13 (MMP-13, also known as collagenase 3). We propose that this mechanism plays a significant role in the establishment of cardiac dysfunction induced by chronic adrenergic overstimulation. This is of particular clinical interest, as PAR1 antagonists may therefore not only prevent PAR1 activation by thrombin (in case of an ischemic event) but could in addition antagonize aspects of adrenergic-dependent pathologic cardiac remodeling. Our work further suggests that MMP-13-mediated activation of PAR1 can also participate in the establishment of cardiac dysfunction. Therefore, we propose that the MMP-13/PAR1 axis represents an important novel target to treat heart failure. Drugs interfering with this axis are already available. Inhibitors of MMP-13 are currently under evaluation in animal models mostly as a cure for rheumatoid arthritis and osteoarthritis where MMP-13 has been shown to play a pathologic role. Moreover, antagonists of PAR1 are now undergoing clinical trials as novel antiplatelet agents. In conclusion, we suggest that MMP-13 inhibitors and PAR1 antagonists may hold therapeutic potential for the treatment of heart failure.
The authors wish to thank Dmitriy Migdalovich, Samantha N. Lomber and Heather Martin for their excellent technical contributions. The authors are also grateful to Dr Stephen Belmonte for critical reading of the manuscript.
Funding Sources: This work was funded in part by HL 084087 (NM and BCB) as well as HL 089885 and HL 091475 (BCB). FJ was supported in part by a fellowship of Fondation pour la Recherche Médicale.
Conflict of Interest Disclosures: NM is a consultant for Merck.