Matrix Metalloproteinases as Input and Output Signals for Post-Myocardial Infarction Remodeling

Abstract

Despite current optimal therapeutic regimens, approximately one in four patients diagnosed with myocardial infarction (MI) will go on to develop congestive heart failure, and heart failure has a high five-year mortality rate of 50%. Elucidating mechanisms whereby heart failure develops post-MI, therefore, is highly needed. Matrix metalloproteinases (MMPs) are key enzymes involved in post-MI remodeling of the left ventricle (LV). While MMPs process cytokine and extracellular matrix (ECM) substrates to regulate the inflammatory and fibrotic components of the wound healing response to MI, MMPs also serve as upstream signaling initiators with direct actions on cell signaling cascades. In this review, we summarize the current literature regarding MMP roles in post-MI LV remodeling. We also identify the current knowledge gaps and provide templates for experiments to fill these gaps. A more complete understanding of MMP roles, particularly with regards to upstream signaling roles, may provide new strategies to limit adverse LV remodeling.

Introduction

According to the European Cardiovascular Disease Statistics (www.ehnheart.org/cvd-statistics.html), each year cardiovascular disease (CVD) causes over 4 million deaths in Europe. In the United States, approximately 6 million Americans currently suffer from heart failure, and the high 5-year mortality rate of 50% results in an annual health care cost of >$34 billion.[1] In 70% of heart failure cases, myocardial infarction (MI) is the underlying etiology.[2-9] While immediate reperfusion of the occluded artery is an optimal therapeutic strategy for ST segment elevation MI, the number of MI patients that are not reperfused (due to late presentation or other exclusion criteria) accounts for approximately 250,000 new patients with permanently occluded arteries each year in the United States.[1] Therefore, adverse remodeling of the left ventricle (LV) following MI remains a significant cause of congestive heart failure.

Matrix metalloproteinases (MMPs) are a family of 25 proteolytic enzymes that regulate extracellular matrix (ECM) turnover and inflammatory signaling. While about half of the MMPs have been measured in the post-MI LV and several MMPs (e.g., MMP-2, -7, -9, -12, -14, and -28) have been studied in detail, there remains significant knowledge gaps.[10-19] To provide a framework for this review, we will use the four postulates for cardiac metalloproteinases actions (CarMA) previously developed to define the iterative process for proving MMP causality in post-MI LV remodeling.[20] The four postulates are: 1) the MMP increases post-MI; 2) the MMP stimulates cell signaling in vitro; 3) modulating MMP levels alters LV remodeling; and 4) MMP proteolytic products recapitulate at least partial components of MMP functions. Note that the four postulates match the order and substance of the Koch’s postulates to establish a cause and effect relationship between a microbe and a disease, rather than by how much evidence is currently available to support the postulate. Postulates 1 and 3 have the most literature available to date.

Below, we list the postulate, followed by the published evidence to support it, followed by the critical knowledge gaps that need to be addressed to provide mechanistic insight into MMP roles in the post-MI LV. We give examples of studies performed in mice with genetic deletions of MMP-7, MMP-9, MMP-14, or MMP-28 or inhibitors against MMP-12. We also discuss a few misconceptions that have arisen along the past 20 years of cardiac MMP research. We end with discussion of how the identified knowledge gaps can be filled, providing example experimental design templates including considerations for controls and potential limitations. While each of the four postulates has been shown for about half of the MMPs, there remains a lack of complete understanding for all MMPs. In addition, multiple MMPs have not been evaluated for even the first postulate. While we discuss each of the postulates to provide a comprehensive review, the second postulate (that MMPs stimulate cell signaling directly) provides an exciting new avenue of research that is only now coming to light.

CarMA Postulate 1: The MMP increases post-MI

This is the smoking gun postulate; the first evidence that must be shown is that the MMP is in a location and at a concentration high enough to have an effect. We and others have shown that MMP-2, -7, -8, -9, -12, and -14 increase post-MI.[12, 21-33] In particular, MMP-9 and MMP-12 have been shown to increase at both the gene and protein level, and both pro- and active protein forms are elevated post-MI.[12, 34] Both increase in the infarct at day 1 post-MI and remain elevated through days 5-7 post-MI.[12, 34]

In addition to measuring total amounts of mRNA or protein expression, it is also important to measure quality and location. With the exceptions of MMP-11, MMP-14, and MMP-28, most MMPs are secreted as pro-enzymes that need to be activated for substrate cleavage.[35, 36] Understanding if the MMP observed is in a pro, active, or inactivated/inhibited state provides knowledge useful for assessing function. Monitoring location and cell source are also important components. For example, total MMP-28 concentration actually decreases post-MI, due to the significant loss of cardiomyocytes that are the major pre-MI source of MMP-28 in the myocardium.[37] At the same time, the amount of MMP-28 contributed by the macrophage increases with the infiltration of macrophages into infarct. This postulate, therefore, is fulfilled for macrophage-derived MMP-28.

Originally, MMPs were classified based on the cell type where the MMP was first observed or the substrates first used to evaluate activity, and this led to confusion.[20] For example, MMP-1 was first named collagenase, but this MMP also processes tenascin and aggrecan into fragments.[20] MMP-8 was first termed neutrophil collagenase, whereas subsequent studies have shown macrophages to be robust expressors of this MMP and additional MMP-8 substrates include aggrecan, fibrinogen, CXCL5, and CXCL6.[38-40] MMP-12 was initially termed macrophage metalloelastase, whereas this MMP also processes osteopontin, and MMP-9 has greater enzyme affinity for elastin cleavage than MMP-12.[16, 41-43] In our own studies, we were surprised to find that post-MI LV neutrophils robustly express MMP-12 at day 1 post-MI.[12] Along with MMPs, the tissue inhibitors of metalloproteinases (TIMPs) have also been misunderstood. While TIMP-4 has been termed the cardiac-specifc TIMP due to its high expression in the myocardium, TIMP-4 is also expressed in the kidney, placenta, colon, and testes.[44]

While CarMA postulate 1 is fulfilled for a subset of MMPs, the current knowledge gap 1 is that each MMP has not been evaluated (e.g., MMP-15 and MMP-16) and not all MMPs (e.g., MMP-7 and MMP-28) have been mapped as extensively as others (e.g., MMP-2 and MMP-9).

CarMA Postulate 2: MMP stimulates cell signaling in vitro

This postulate dictates that MI-relevant cells stimulated with an MMP will display biological functions similar to what is observed during LV remodeling in vivo, if that MMP has a causal role. Basically, this postulate raises the possibility that MMPs can serve in direct signaling capacities, which separates this role from that of an enzyme. While there are examples to provide evidence for this postulate, this idea has not been examined in detail.

In vitro, MMP-9 directly activates macrophage polarization to an M1/M2 transition state.[45] MMP-9 stimulation increased the expression of the pro-inflammatory M1 gene Ccl5, but also decreased the expression of M1 markers Ccl3, interleukin (IL)-1β, and IL-6. Transforming growth factor (TGF)β1, an anti-inflammatory M2 marker, was also down-regulated after MMP-9 treatment. MMP-12 stimulates neutrophils in vitro to induce expression of the apoptosis markers, CD44, caspase 3, and caspase 8,[12] and MMP-9 also stimulates capase 3 expression in neutrophils.[46] CD44 regulates apoptosis by interacting with hyaluronic acid and is a critical mechanism in wound healing to clear inflammatory cells from injury sites.[47, 48] MMP-12 can also process CD44 to generate a 15 kDa fragment, indicating a feedback loop. CD44 cleavage prevents the clearance of the CD44 ligand hyaluronic acid, which is a stimulus for inflammation resolution during wound healing.[12] Combined, these results indicate that MMPs can be used as direct stimulating factors as well as output factors. This is an entirely new concept in MMP biology, and future studies evaluating how MMPs activate cell signaling (e.g., direct binding of receptors such as integrins or indirect effects through processing of substrates) are warranted.

In vivo, MMP-7 has been shown to have direct activity on myocardial electrical activity by cleaving connexin-43 in the C-terminal domain.[49] Infusion of recombinant MMP-7 into the jugular vein of mice induced heart block within 60 min of infusion, concomitant with reduced connexin-43 staining at the myocyte to myocyte borders. This study provides direct evidence that connexin-43 is an in vivo substrate of MMP-7 and that its processing results in a pathophysiological phenotype.

Knowledge Gap 2 is that the MMP signaling pathways that regulate cell function have not been mapped. There is a need to identify MMP signaling pathways that regulate post-MI relevant cell functions, including myocyte apoptosis; neutrophil apoptosis and degranulation; macrophage polarization and phagocytosis; and fibroblast proliferation, differentiation, and ECM expression. Included in this knowledge gap is the need to know which receptors are engaged by MMPs and whether the effects are actually directly occuring through receptor engagement and signaling or due to an indirect effect that has not been elucidated (e.g., substrate fragment binding to a receptor or shedding of an inhibitor in the signaling pathway). While this postulate has the largest unknown component, it also is one of the more exciting postulates due to its novelty.

CarMA Postulate 3: Modulation of an MMP alters LV remodeling

This postulate identifies tissue level functions controlled by MMP regulation of individual ECM components or of cell changes. Concentrations of a few specific MMPs (e.g., MMP-9) directly correlate to the extent of LV dysfunction post-MI. The assumption has been that an increase in an MMP is always detrimental and that MMPs should be inhibited, but this has not always been the case as we have recently seen for MMP-12 inhibition.[12] Inhibiting MMP-12 beginning at 3 hours post-MI exacerbates LV dilation and dysfunction, suggesting beneficial components of MMP-12 activity.[12] MMP-12 stimulates in vivo neutrophil apoptosis, as MMP-12 inhibition increased full-length caspase 3 by 58% and reduced cleaved caspase 3 by 50%.[12] CD18 (β2 integrin; ITGB2), a cell adhesion molecule that suppresses neutrophil apoptosis during endothelial transmigration, was elevated at day 7 post-MI with MMP-12 inhibition. This indicates reduced apoptosis and prolonged neutrophil accumulation when this MMP was inhibited.

The combined results from MMP-9 null mice and mice with transgenic overexpression of MMP-9 only in macrophages reveals that both situations paradoxically improve LV remodeling. [24, 50-53] These findings are due to MMP-9 exerting both negative and positive actions in the post-MI LV. MMP-9 deletion attenuates collagen and fibronectin cleavage to attenuate LV dilation; while MMP-9 overexpression promotes processing of angiogenic factors (e.g., plasminogen and collagen IV) to stimulate neovascularization.[34, 54]

MMP-28 deletion was also detrimental to LV remodeling. During aging, LV inflammation increases with age, and MMP-28 deletion further elevates inflammation and extracellular matrix responses, without altering macrophage numbers or collagen content.[55] MMP-28 deletion increased plasma macrophage inflammatory protein (MIP)-1α, MIP-1β and MMP-9 protein concentrations and elevated MIP-1a and MMP-9 gene and protein levels in the LV, indicating a higher overall inflammatory status when this MMP was genetically modified. After MI, MMP-28 deletion aggravated MI-induced LV dysfunction and rupture as a result of both a defective inflammatory response and reduced scar formation.[23] In both aging and MI, MMP-28 deletion suppressed M2 macrophage activation. This leads to Knowledge Gap 3: MMP signaling pathways that regulate the global LV response (structure and function) have not been identified. We need to understand how the compilation of different signals regulated by MMPs are coordinated to regulate LV remodeling.

CarMA Postulate 4: MMP proteolytic products regulate LV remodeling

According to this postulate, MMP cleavage products should directly serve as MMP substitutes in regulating particular aspects of the post-MI LV remodeling phenotype, such that adding back substrate in an MMP null background should recapitulate the MMP wild type phenotype or at least a component of the phenotype. While a number of MMP substrates have been identified and general consensus sequences have been compiled (Table 1), the complete MMP interactome has not been catalogued for any individual MMP. For MMP-2, the current substrate list is expansive and includes hundreds of proteins, while for MMP-28, the current substrate list only includes casein, Nogo-A (a myelin component), and neural cell adhesion molecule-1.[56, 57] Table 2 lists a sampling of MI-relevant substrates proteolytically processed by matrix metalloproteinases (MMPs).[20, 22, 46, 58-64] Below, we discuss a few interesting substrates for MMP-7, MMP-9, and MMP-12.

Table 1. MMP Substrate Consensus Sequences, based on analysis of known cleavage sites. Letters are abbreviations for amino acids. For details, see http://merops.sanger.ac.uk. – is any amino acid.

	1	2	3	4	Cleavage site	5	6	7	8	Number of known sites examined
MMP-2	-	P	-	-	↓	L	-	-	-	3417
MMP-3	-	PA	A	-	↓	L	-	-	-	180
MMP-7	-	PA	-	-	↓	L	-	-	-	196
MMP-8	G	PAS	-	G	↓	L	-	G	-	114
MMP-9	G	PA	-	G	↓	L	-	G	-	369
MMP- 12	G	PAG	AG	GA	↓	L	-	GA	-	218
MMP- 14	-	P	-	-	↓	L	-	-	-	132

Table 2.

Myocardial infarction (Ml)-relevant substrates proteolytically processed by matrix metalloproteinases (MMPs).[20, 22, 46, 58-64]

Substrate	MMPs
Aggrecan	-2,-9,-3, -8, -14
CD36	-9
CD44	-14
Citrate synthase	-9
Collagen	-1,-2,-8,-9,-13,-14
Complement C1q	-1, -2, -3, -9
Connexin 43	-7
Desmin	-9
Endothelin-1	-9
Fibroblast growth factor receptor 1	-2
Fibronectin	-2,-7,-9
Fibrinogen	-2, -3, -7, -9, -14
Galectin-3	-9, -2
Interleukin-1β	-9
Interleukin-8	-9, -1, -2, -3,
Insulin-like growth factor binding protein-3	-1, -2,-3
laminin	-9, -2, -3, -14
Latent TGF-β1, -β2, -β3	-2, -9
Monocyte chemotactic proteins-1, -2, -4	-1, -3
Monocyte chemotactic protein-3	-2, -14
Myosin Light Chain 1	-2
Osteoponin (cite Lis)	-2, -3, -7, -9
Pro-MMP-2, -9, -13	-9
Platelet factor 4	-9
serpina 1d	-9
Tenascin-C	-7,-9
Thrombospondin-1	-9
Troponin-I	-2
Tumor necrosis factor-α -	-1, -2, -3, -7, -9
VEGF	-9
Vitronectin	-2, -3,-9

In addition to connexin 43, MMP-7 substrates include denatured collagen and fibronectin.[49, 65] Several MMPs have the ability to activate other MMPs. For example, MMP-3, which can activate MMPs -1, -7, and -9 and is considered an early upstream MMP in the cascade.[66] Likewise, MMP-7 can activate in vitro MMPs -1, -2, -8, and -9, but not MMP-13.[67] In MMP-7 null LV, MMP-8 increases and MMP-13 decreases as a compensatory mechanism.[67] The MMP profile switch reveals that MMP-8 accumulation in the absence of MMP-7 downregulates MMP-13 in order to maintain a baseline collagenolytic function and preserve LV wall thickness and dimensions. This interplay between MMP-8 and MMP-13 indicates that these MMPs play reciprocal roles and demonstrates that when designing selective MMP inhibitors, we need to take into consideration potential compensation mechanisms that will regulate parallel MMP types. This also highlights that baseline MMP null phentoypes are seldom unchanged and measuring global phenotypes may not reveal subtle but important differences.

In addition to ECM molecules, MMP-9 cleaves a variety of inflammatory mediators to coordinate their in vivo functions. MMP-9-mediated proteolysis of cytokines and chemokines is one way by which MMP-9 influences leukocyte trafficking and creates positive or negative feedback loops. For example, the N-terminal cleavage of the CXC chemokine IL-8 by MMP-9 increases its chemotactic activity by 10-fold.[68] IL-8 promotes neutrophil chemotaxis which stimulates additional release of MMP-9 demonstrating a positive feedback loop. Similar to MMP-2, MMP-9 also processes a number of intracellular substrates, including citrate synthase.[59] In wild type mice, citrate synthase dramatically decreases by 1 day post-MI due to proteolytic processing by MMP-9, as MMP-9 deletion preserved citrate synthase concentrations in the mitochondria. This assigns a role for MMP-9 in impairing cardiac metabolism in the setting of ischemia.

MMP-12 cleaves immunoglobulins, which are key stimulators of neutrophil degranulation.[69-71] This indicates immunoglobulin processing may be a critical MMP-12 dependent trigger substrate to regulate neutrophil biology. MMP-12, therefore, regulates many aspects of neutrophil biology, including degranulation and apoptosis. While substrate identification is a straightforward process, making the connection between substrate processing and an effect on cell or tissue physiology in the setting of post-MI LV remodeling has only been comprehensively evaluated for MMP-9 generated collagen matricryptin.[61] A Collagen Iα1 matricryptin generated by MMP-9 cleavage at amino acids 1158 and 1159 (p1158/59) reduced LV dilation and facilitates LV remodeling post-MI by regulating scar formation through targeted ECM generation and stimulation of angiogenesis. From this information, we derive Knowledge Gap 4: MMP proteolytic products that regulate LV remodeling have not been catalogued. We need to understand the mechanisms whereby MMP proteolytic products regulate LV remodeling.

How to fill the knowledge gaps

Above, we summarized the current literature base and captured the critical knowledge gaps that remain in our understanding of how MMPs regulate LV remodeling post-MI. Here, we will provide example experimental design templates that would allow these gaps to be filled. We also discuss new technology that is required to help propel this research.

Knowledge gap 1

each MMP has not been evaluated (e.g., MMP-15 and MMP-16) and not all MMPs (e.g., MMP-7 and MMP-28) have been mapped as extensively as others (e.g., MMP-2 and MMP-9). Filling this knowledge gap is technically uncomplicated, as quantative PCR can be used to assess gene levels and there are many commercially available antibodies for all of the MMPs, which allows immunoblotting assessment. While there are no technical challenges, time and location of evaluation need to be considered in the experimental design. Assessing both the remote non-infarcted myocardium and the infarct region during a full time continuum with adequate sample size in each group to have sufficient sample size to detect differences makes these experiments complex. While evaluating at the mRNA level is quite feasible, running this number of samples by immunoblotting is difficult, particularly given issues with intra-blot comparisons. One strategy is to evaluate the time course with a pooled sample strategy, such that each time point is run as one sample, taking into account the caveat that a potential confounder to this strategy is the presence of any outliers in the pooled samples that could mask true effects. From there, one time point can be selected to compare against day 0 naïve or sham controls. What we need to accelerate the closure of this knowledge gap is an MMP protein array that allows quantitation of both pro and active MMP forms and includes the tissue inhibitors of metalloproteinsaes (TIMPs) in a more rapid and accurate way. While there are arrays available for a handful of MMPs, ideally an array that differentiates the pro and active forms of all MMPs and includes all TIMPs would aid in closing this knowledge gap.

Knowledge Gap 2

The MMP signaling pathways that regulate cell function have not been mapped. Table 3 provides an example experimental template for examining direct MMP effects on cell functions. We used MMP stimulation of neutrophil apoptosis and macrophage polarization as examples here.

Table 3.

Experiments to fill knowledge gap 2.

General design: isolate and culture cells of interest perform time and dose responses with MMP as stimulus examine cell response compare ex vivo and in vivo responses
Cell	Example responses (could be ↑ or ↓)	What to measure
All	Global gene expression Global protein expression ○ cellular or secreted	Genomics Proteomics
Neutrophil	Degranulation Apoptosis	Secreted markers Caspases CD44
Macrophage	Polarization Phagocytosis	Pro- and anti-inflammatory markers CD36; uptake of fluorescent bioparticles1
Fibroblast	Proliferation Migration Wound healing ECM production	BrdU uptake Transwell mobilization Scrape or heat injury2 ECM genes and proteins3

BrdU- 5-Bromo-2′-deoxyuridine; ECM- extracellular matrix

¹e.g., fluorescein-labeled Escherichia coli K-12 bioparticles (Molecular Probes)

²e.g., Electric Cell-substrate Impedance Sensing System (Applied Biophysics)

³MI relevant ECM includes collagen, fibronectin, laminin, and matricellular proteins

Neutrophils isolated from the blood or bone marrow and stimulated with an MMP in vitro can be compared to unstimulated cells as a negative control. The cells can be examined for marker expression or functional responses by flow cytometry, qRT-PCR, cytochemistry or proteomic evaluations. Conditioned media can also be examined by ELISA or proteomics, as long as the media does not contain a large concentration of serum (<0.1% is ideal). One positive control would be neutrophils isolated from the infarct region and immediately evaluated.

Macrophages isolated from the peritoneum or monocytes isolated from the blood or bone marrow and stimulated with an MMP in vitro can be compared to unstimulated cells as a negative control. The cells can be examined for M1 and M2 marker expression. An advantage of peritoneal macrophages is that enough cells can be isolated from one mouse to examine 6-8 different conditions, whereas blood monocytes from 6 mice need to be pooled to have the same numbers of isolated cells. The advantage of blood monocytes is that these cells will infiltrate into the LV post-MI, whereas peritoneal macrophages may not fully recapitulate the MI macrophage phenotype.[45, 72] Cells stimulated with interleukin (IL)-1β or lipopolysaccharide (LPS) + interferon (IFN)-γ serve as positive controls for M1 polarization, while IL-4 or IL-10 serve as positive controls for M2 polarization. Other positive controls are macrophages isolated from the infarct region or bone marrow derived macrophages, to compare back how well the in vitro setting recapitulates the in vivo situation. A co-stimulation or sequential stimulation experiment may be warranted, to assess cells under a condition that more closely reflects the in vivo post-MI environment.

Once it is established that the MMP activates (or inactivates) a cell function, the next step is to elucidate the mechanism whereby this occurs. The first consideration is whether this is a substrate mediated enzymatic event, such as MMP processing a ligand or receptor directly. If this is not the case, examining whether the MMP can directly bind to a receptor can be tested.

Target receptors can be assessed by proteomic methods, which provide unbiased appraisal. Using a solid phase extraction of glycopeptides (SPEG) with hydrazide chemistry protocol can be performed to identify glycopeptides in WT and MMP null infarct homogenates or cell culture conditioned media.[73, 74] Since most receptors are glycoproteins, this approach eliminates the potential issue of high levels of albumin in the conditioned media, since albumin is not a glycoprotein and will not be isolated or labeled by this approach. This approach is also useful for ECM-focused analysis (matridomics).[75-78]

Targeted assesssment of candidate receptors is also possible, and surface plasmon resonance can be used to test the in vitro binding kinetics. When testing viability of receptor targets, it is best to first use bioinformatics and simulations to rank candidate evaluation. Comparing homology of MMPs with known receptor stimulators will minimize the number of receptors that need to be tested. Binding simulations would also elucidate receptor candidates that could then be tested in vitro and in vivo.

Knowledge Gap 3

MMP signaling pathways that regulate the global LV response (structure and function) have not been identified. Filling this gap will identify tissue-specific actions, including effects on tissue mechanics.[79] To address this gap, permanent occlusion or reperfusion models are useful, and both have strengths and limitations. In a 2012 Scientific Statement on animal models of heart failure from the American Heart Association, reperfusion and permanent coronary artery ligation were deemed clinically relevant and significant models.[55] While MMP deletion or transgenic stable or conditional mouse models have been evaluated, a translational approach is to infuse or inhibit the MMP in a setting that recapitulates the clinic. A constant infusion protocol can be used to evaluate in vivo responses, and one protocol is to begin at 3 hours post-MI, which will capture the entirety of the inflammatory and wound healing responses. Starting too early post-MI (or pre-MI) changes the design in a way that limits clinical translation. To rule out the possibility that the MMP is regulating an early response (e.g., myocytes or neutrophils) to indirectly affect a later response (e.g., macrophages or fibroblasts), constant infusion beginning at day 2 post-MI can be used to offset this limitation.

Table 4 shows useful variables to assess the LV remodeling phenotype. LV geometry, strain analysis, and function can all be evaluated in vivo by ultrasound, magnetic resonance imaging, or hemodynamic evaluation. The LV (infarct and remote regions) can be frozen for biochemical analysis (gene or protein) or fixed for pathological examination of inflammation, ECM turnover, and scar formation. As a control for the MMP infusion, lungs and right ventricle can be weighed and myocyte cross sectional areas measured to rule out if effects are due to pulmonary issues and also assess heart hypertrophy and failure status. Consideration should also be given from the dose used; the dose should provide over-expression without being supra-pathological.

Table 4.

Measuring the LV remodeling phenotype

Phenotype	Variables measured
Infarct structure; LV function; hemodynamics; cardiac metabolism	LV geometry, strain analysis, and ejection fraction by echocardiography, magnetic resonance imaging, or hemodynamic assessment
Hypertrophy and heart failure	Organ weights at necropsy- heart and lungs; myocyte cross sectional areas
Infarct area	Viable and infarct myocardium areas by metabolic staining; hematoxylin and eosin stained histological sections
Scar formation	Picrosirius red stained histological sections
Inflammation and its resolution	Pro- and anti-inflammatory mediators (gene and protein)
Neutrophils	neutrophil extracellular traps; tissue destruction; degranulation markers
Macrophage polarization	M1 and M2 markers (gene and protein)
Fibroblast activation	α-smooth muscle actin levels; extracellular matrix production and turnover (MMP concentrations and ECM fragment formation)
Neovascularization	Griffonia (Bandeiraea) Simplicifolia Lectin 1 staining of histological sections; angiogenesis markers (e.g., VEGF)
Cell-cell interactions	Co-culture; secretome composition by mass spectrometry

Knowledge Gap 4

MMP proteolytic products that regulate LV remodeling have not been catalogued. Evaluating if substrate proteolyic products regulate the same signaling pathways will show direct cause and effect relationships. An experimental flow chart for how to answer this question is shown in Figure 1. We used this flow chart recently to examine the effects of a novel collagen fragment in post-MI LV remodeling.[61]

Figure 1.

Once a candidate MMP substrate is identified, cleavage can be confirmed using an in vitro assay consisting of the MMP and a recombinant of the candidate protein. Using mass spectrometry, the cleavage site(s) can be mapped. Once sites are known, comparing the literature for similar matricryptins will provide insights into putative functions. For example, fibronectin and laminin matricryptins have been studied extensively in the angiogenesis field.[83] To test for biological function, peptides 15-20 amino acids in length that are immediately upstream or downstream of the cleavage site can be commercially generated. If many cleavage sites exist, computational analysis can help to focus on the most likely candidates. Starting with in vitro cell-based assays is higher throughput and more cost effective and will eliminate candidates that have no activity on any cell type of interest (i.e., cardiomyocytes, leukocytes, fibroblasts, or endothelial cells). Using small peptides to test biological activities of naturally occurring fragments bypasses difficulties associated with larger peptide synthesis. A short peptide is easier to synthesize, purify, and concentrate, and a small size peptide limits response variability, which can occur with a longer peptide that may possess multiple activities. Using a small peptide increases the probability that an effect is a direct response to the sequence and not due to peptide structural properties.[92]

Future Directions and Conclusions

While this review focused on MMPs in the post-MI setting, a similar template can be applied for other relevant molecular components and as a broad application to other diseases that involve MMPs. For example, the A Disintegrin and Metalloproteinase (ADAM) family of enzymes are important for post-MI healing, and this family has been reviewed by others, including the Kassiri lab.[80, 81] Because a substantial amount of work remains to be done to establish cause and effect relationships between MMPs and LV remodeling components, we focused on MMP effects on single cell actions. Gene deletion or transgene strategies have been used to decrease or increase MMP expression, but these approaches occur in an artificial setting that is not directly translational. While it is relatively simple to overexpress MMPs systemically with recombinant proteins or peptides, tools available to limit MMP function are limited and suffer from low selectivity. Table 5 lists a sampling of known MMP inhibitors. Of note, issues with inhibitor selectivity and specificity continue; often inhibitors that are relatively specific at lower doses become broad spectrum inhibitors at higher doses.

Table 5.

Matrix metalloproteinase (MMP) inhibitors- note that very few are selective for specific MMPs.[88-91]

Inhibitor	MMPs inhibited
CGS27023A	MMP-2, MMP-3, MMP-9, MMP-12
matlystatin A	MMP-2, MMP-9
matlystatin B	MMP-2, MMP-9
ONO-4817	MMP-2, MMP-3, MMP-7, MMP-8, MMP-9, MMP-12
Ro 28-2653	MMP-2, MMP-9, MMP-14
SB-3CT	MMP-2, MMP-9
Doxycycline	MMP-2, MMP-9
Galardin	MMP-3, MMP-14
α1(I-III)Glyφ{PO2H-CH2}Leu THPI	MMP-9
RXP470.1	MMP-12
Imidapril	MMP-9
Captopril and Ramiprilate	MMP-2 and MMP-9
Lisinopril	MMP-9

A future direction is to incorporate intercellular communication to understand how one MMP can activate a cell response that feeds into activating another cell type. Computational models at the cellular scale, such as fibroblast models, will be useful in achieving this goal.[82] For example myocyte-neutrophil, neutrophil-macrophage, and macrophage-fibroblasts are all interconnected in ways that remain to be fully elucidated. The MatrixDB is an extracellular matrix interaction database that will continue to be useful as connections are made among ECM components.[83] In the review by Francis et.al., the Ripplinger lab discusses the interconnections among inflammation, fibrosis, and arrhythmias, highlighting the complexity of the system.[84] Understanding the interconnection between inflammation and fibrosis is an important future avenue of research.[85-87]

In conclusion, understanding the full complement of MMP roles will not only provide us with new therapeutic strategies to treat patients post-MI but will also provide novel diagnostic markers for the early detection of adverse LV remodeling as an early harbinger of ensuing heart failure.

Highlights.

• MMPs are key enzymes involved in post-MI remodeling of the left ventricle (LV).

• MMPs process extracellular matrix (ECM) and cytokine substrates.

• MMPs also serve as upstream initiators with direct actions on cell signaling.

• We review current MMP knowledge gaps in the cardiac remodeling field.