**The Association of Matrix Metalloproteinases with Chronic Kidney Disease and Peripheral Vascular Disease: A Light at the End of the Tunnel?**

**Michele Provenzano 1, Michele Andreucci 1, Carlo Garofalo 2, Teresa Faga 1, Ashour Michael 1, Nicola Ielapi 3,4,5, Ra**ff**aele Grande 6, Paolo Sapienza 6, Stefano de Franciscis 3,7, Pasquale Mastroroberto <sup>8</sup> and Ra**ff**aele Serra 3,7,\***


Received: 16 December 2019; Accepted: 14 January 2020; Published: 17 January 2020

**Abstract:** Chronic Kidney Disease (CKD) represents a risk factor for fatal and nonfatal cardiovascular (CV) events, including peripheral vascular disease (PVD). This occurs because CKD encompasses several factors that lead to poor prognoses, mainly due to a reduction of the estimated glomerular filtration rate (eGFR), the presence of proteinuria, and the uremic inflammatory milieu. The matrix metalloproteinases (MMPs) are a group of zinc-containing endopeptidases implicated in extracellular matrix (ECM) remodeling, a systemic process in tissue homeostasis. MMPs play an important role in cell differentiation, angiogenesis, inflammation, and vascular damage. Our aim was to review the published evidence regarding the association between MMPs, PVD, and CKD to find possible common pathophysiological mechanisms. MMPs favor ECM deposition through the glomeruli, and start the shedding of cellular junctions and epithelial-mesenchymal transition in the renal tubules. MMP-2 and -9 have also been associated with the presence of systemic vascular damage, since they exert a pro-inflammatory and proatherosclerotic actions. An imbalance of MMPs was found in the context of PVD, where MMPs are predictors of poor prognoses in patients who underwent lower extremity revascularization. MMP circulating levels are increased in both conditions, i.e., that of CKD and PVD. A possible pathogenic link between these conditions is represented by the enhanced production of transforming growth factor-β that worsens vascular calcifications and atherosclerosis and the development of proteinuria in patients with increased levels of MMPs. Proteinuria has been recognized as a marker of systemic vascular damage, and this may explain in part the increase in CV risk that is manifest in patients with CKD and PVD. In conclusion, MMPs can be considered a useful tool by which to stratify CV risk in patients with CKD and PVD. Further studies are needed to investigate the causal-relationships between MMPs, CKD, and PVD, and to optimize their prognostic and predictive (in response to treatments) roles.

**Keywords:** metalloproteinases; MMPs; TIMPs; CKD; peripheral vascular disease; biomarkers; proteinuria; eGFR; PAD.

#### **1. Introduction**

Chronic Kidney Disease (CKD) is defined as the presence of abnormalities in kidney function or structure for at least 3 months [1]. The Kidney Disease: Improving Global Outcomes (KDIGO) guidelines classify CKD according to the level of estimated glomerular filtration rate (eGFR), a marker of kidney function, and the amount of urine protein (proteinuria or albuminuria), which represents the principal marker of kidney damage and the primary cause of CKD [2]. The onset of CKD exerts a deleterious impact on individual health. Indeed, it has been demonstrated that either an eGFR reduction < 60 mL/min/1.73 m<sup>2</sup> or a small increase in proteinuria are associated with a poor prognosis, as shown by the increased rate of cardiovascular (CV) fatal and nonfatal events, all-cause mortality, and the progression of kidney disease resulting in the need for renal replacement therapies (kidney transplantation or dialysis) [3,4].

CV risk is severely increased in patients with CKD, and the impact of CV events in this population is crucial if one considers that the rate of CV events (such as myocardial infarction, stroke, arrhythmias, peripheral vascular disease, and chronic heart failure) over time is higher than the risk for kidney disease progression [5]. This strong association has been attributed to the coexistence of traditional and nontraditional CV risk factors in CKD patients, with the former being represented by hypertension, smoking, hypercholesterolemia, and metabolic abnormalities, and the latter by the two main prognostic measures of CKD, i.e., proteinuria and eGFR [6,7].

Among the wide spectrum of CV events, a relevant role is portrayed by peripheral vascular disease (PVD). It has been demonstrated that the presence of mild-to-moderate CKD increases the risk of peripheral artery disease, leg revascularization, leg amputation, and hospitalizations [8]. Either an eGFR reduction below 60 mL/min/1.73 m2 or slight increases of albuminuria (>30 mg/g) have been associated with a 1.5 to 4 times higher risk of peripheral artery disease (PAD). This evidence is strong, being derived from patients without PAD at basal visit, and reproducible, being confirmed in the general population, as well as in high-risk populations, regardless of the geographic area. Notably, the rate of PVD among patients with End-Stage Kidney Disease (ESKD) is higher than the incidence of acute myocardial infarction, stroke, and arrhythmias [9]. Hence, CKD patients warrant clinical surveillance and prompt the need for strategies to prevent the onset of PVD.

The magnitude of the association is so important that research has recently focused on discovering new biomarkers that are potentially useful in the clinical management of CKD patients at increased risk for PVD. Matrix Metalloproteinases (MMPs) are zinc-containing endopeptidases that are involved in regulating tissue development and homeostasis [10]. Although all MMPs are better acknowledged for their role in remodeling the extracellular matrix (ECM), they actually interact with both ECM and non-ECM substrates. Cell adhesion molecules and growth factors or their receptors represent these latter group of substrates. The wide range of interactions in which MMPs play an active role also explains why these endopeptidases participate in a number of functions such as cell differentiation, migration, apoptosis, and angiogenesis. On the other hand, MMPs have also been attributed to a profibrotic and pro-inflammatory role. MMP-2 and MMP-7 were increased in plasma and urine samples of CKD patients, and may affect direct damage to the kidney with the onset of albuminuria [11]. Moreover, imbalances in the expression and the levels of MMPs or their inhibitors have been linked to the structural changes that occur in the development of PAD, atherosclerotic plaque maturation, and arterial remodeling [12].

The purpose of this review is to examine the role of MMPs in increasing the risk of peripheral vascular disease by the specific aggravating condition of Chronic Kidney Disease.

#### **2. Materials and Methods**

The PubMed and ISI Web of science databases were searched for articles by using the following terms: 'chronic kidney disease", "chronic renal insufficiency", "metalloproteinases", "MMP" "atherosclerosis" "peripheral vascular disease". Titles and abstracts were screened by three authors (Michele Provenzano, Michele Andreucci, and Raffaele Serra) to identify potentially relevant studies. All potentially eligible studies were subsequently evaluated in detail by one reviewer and three authors (Michele Provenzano, Michele Andreucci, Carlo Garofalo, and Raffaele Serra) through consideration of the full text. Reference lists of retrieved articles were also searched for relevant publications. Clinical trial, meta-analyses, narrative review, and systematic reviews published in the last 10 years were included. Bibliographies of relevant articles and reviews were manually screened to identify additional studies. Studies were excluded if they were not in the English language, if they did not fit the research question, or if they had insufficient data.

#### **3. Results**

#### *3.1. Study Selection*

Initial database searches yielded 180 studies from PubMed and 445 from ISI Web of Science in the last 10 years. After the evaluation of the bibliographies of the relevant articles, we evaluated 67 eligible full text articles. The current evidence on MMP expression and kidney disease, the link between MMPs and CV risk specifically in CKD patients, and the association between MMPs and peripheral vascular disease are described below.

#### *3.2. Metalloproteinases and the Kidney*

MMPs are classified, according to their structure (or function) and the substrate selectivity, into six groups: Collagenases (MMP-1, MMP-8 and MMP-13), which cleave native collagen and with possible antifibrotic function; Gelatinases (MMP-2 and MMP-9), whose function is to cleave denatured collagens, type IV collagens in basement membranes and some chemokines; Stromelysins (MMP-3, MMP-10, MMP-11 and MMP-19), which degrade a number of substances such as fibronectin, laminin but are unable to cleave native collagen; Matrilysins (MMP-7 and MMP-26) act by degrading ECM components (laminin and entactin); Membrane-type MMPs (MMP-14, -15, -16, -17, -24, and -25), so called because they are anchored to the exterior of the cell membrane; and other MMPs that are tissue or cell-type specific [10,13]. The MMP activity is modulated by a series of four known enzymes called tissue inhibitors of metalloproteinases (TIMPs). TIMPs participate either in the activation or inhibition of MMP activity and, like MMPs, regulate several cellular functions such as cell proliferation, apoptosis, and angiogenesis [14]. It has been demonstrated that MMPs exert a role in the development of proteinuric kidney diseases in humans. Indeed, a number of studies depicted increased serum and urine levels of MMP-2, -8, and -9 in diabetic patients, with MMP-9 in particular being positively correlated with the degree of proteinuria in these patients [15–18]. It is remarkable that, in addition to MMP-9, urinary levels of neutrophil gelatinase-associated lipocalin (NGAL) have been found to increase in patients with diabetic nephropathy [19]. NGAL and MMP-9 are coexpressed, and their interaction prevents the degradation of MMP-9. It has, thus, been postulated that the increase in NGAL may prolong the action of MMP-9 as a trigger of kidney damage [20]. MMP-7, which is normally expressed in the proximal and distal convoluted tubules, as well as in the collecting duct, is found to be overexpressed in diabetic patients where it is also inversely correlated with the degree of kidney function [21]. Other than in diabetic patients, MMP expression is altered in many other glomerular diseases. Typical patterns of MMP-2 and MMP-9 are differentially expressed in patients with focal segmental glomerulosclerosis, minimal change disease, membranous nephropathy, and ANCA-associated vasculitis [22–26]. Regardless of the specific kidney disease involved, several mechanisms of damage have been put forward to explain the pathophysiologic effects of MMPs (Figure 1).

**Figure 1.** Expression of MMPs and TIMPs, and pathophysiological mechanisms of vascular and kidney damage [10,11,27–31]. EMT, epithelial-mesenchymal transition; VSMC, vascular smooth muscle cells; TGF-β, transforming growth factor-β.

MMPs intervene in all phases of renal fibrosis, from infiltration of mononuclear cells to cells proliferation and scarring. All these processes lead to a progressive decline of renal function in CKD patients. Henger and Colleagues performed a hierarchical clustering analysis to assess the differential gene expression in human kidney fibrosis. Interestingly, they observed that several MMP (MMP-3, -13, -14) genes were upregulated in different degrees of fibrosis [32]. With respect to inflammation, different MMPs play different, often opposing, actions. MMP-7 and MMP-9 expand inflammatory processes, particularly by their chemotactic effect on human dendritic cells [33,34]. Conversely, MMP-13 and MMP-14 have been hypothesized to act as anti-inflammatory mediators [35]. MMP-3 may also promote epithelial-to-mesenchymal transition, the conversion of tissue phenotype, from the epithelial to fibroblastic, thereby accelerating fibrosis [36]. Abnormalities in the accumulation/degradation of ECM due to imbalanced levels of MMPs and TIMPs have also been described in rat and humans [37–39]. Indeed, downregulation of MMP-1 and the overexpression of TIMP-1, MMP-2, MMP-7, and MMP-9 are associated with a profibrotic effect, as well as with a destructive effect on renal parenchyma [38,39]. Moreover, increased TIMP-1 plasma levels were predictors of incident CKD, regardless of other systemic and inflammatory biomarkers (C-Reactive Protein or Brain Natriuretic Peptide) and many clinical parameters (liver function, concomitant lipid-lowering, or antihypertensive medications) [40].

#### *3.3. Vascular E*ff*ects of Metalloproteinases in CKD Patients*

The structural remodeling of ECM, together with the profibrotic effect of MMPs, are detrimental for other organs apart from the kidney. Elevated blood concentrations of TIMP-1 have been associated with an increased risk of developing chronic heart failure (CHF) and, in patients already diagnosed with CHF, they were predictors of poor prognoses [41,42]. Moreover, the increase in MMP-9 and TIMP-1 conferred risk of all-cause mortality and incident CV disease in community studies [43,44]. Possible explanations for the relationship between MMPs and CV risk are varied. Whereas all MMPs are likely risk factors for atherosclerosis and cardiac dysfunction, a more specific mechanism of damage

has been postulated for MMP-9 and TIMP-1. Indeed, MMP-9 is involved in the intracellular cleavage of myosin filaments, a mechanism that leads to ventricular hypertrophy [45]. TIMP-1 has shown a direct relation with the left ventricular mass in the Framingham Heart Study participants [41]. The CKD condition is associated with an increased prevalence of CV morbidity and mortality. The United States Renal Data System (USRDS) showed that the frequency of each CVD, including myocardial infarction, coronary artery disease, and peripheral vascular disease was higher among patients with CKD compared with those without (Figure 2) [9].

**Figure 2.** Prevalence of cardiovascular diseases according to the presence (dark blue bars) or absence (turquoise bars) of Chronic Kidney Disease (CKD) in the United States, in the year 2015. AF, atrial fibrillation; AMI, acute myocardial infarction; CAD, coronary artery disease; CVA/TIA, cerebrovascular accident/transient ischemic attack; HF, heart failure; PAD, peripheral arterial disease.

The CV burden in these patients is significant if one considers that the rate of CV events is similar to that of reaching ESKD [46,47]. As a result of this epidemiological and public health evidence, great effort has been placed in finding early atherosclerosis biomarkers that predict CV events in CKD patients and serve as a possible target for new therapeutic agents [48]. MMPs were included in the set of investigated substances. Indeed, MMP-2 was positively associated with carotid Intima-Media Thickness (cIMT) and abdominal aortic calcification, suggesting that an association between this MMP and subclinical atherosclerosis is plausible [49–51]. Moreover, MMP-2 was also found to be higher in patients with a positive history of CV disease vs no history of CV disease [52,53]. MMP-9 was also strongly associated with cIMT, the development of carotid plaques, and systemic atherosclerosis [49,50]. MMP deregulation is intensified in patients with advanced CKD stages that are also associated with a reduction of their clearance. All these mechanisms enhance the inflammatory process that is chronically activated in CKD patients, due to oxidative stress, the uremic milieu, and the metabolic acidosis [54]. The combination of uremic status and imbalance in pro-inflammatory substances (such as MMPs) accelerates the atherosclerotic process, arterial stiffness, and vascular calcification, and impairs the vascular repair process as well [55]. Circulating MMP-2, -9, and -10 have been found to increase in CKD patients, and have been implicated in the vascular damage process. Moreover, MMP-2 and -9 are able to reduce the plaque stability in advanced CKD stages, thus rendering the plaque itself more prone to rupture [56,57].

#### *3.4. Metalloproteinases and Peripheral Vascular Disease*

MMPs play a role in the pathogenesis and prognosis of arterial and venous disease. With respect to the damage occurring in the arterial tissue, multiple studies have shown that MMPs are involved in one or more steps of atherogenesis and aneurysm development [58]. Evidence that confirms this hypothesis comes from both basic and clinical studies. During the progression of atherosclerotic plaques, a number of MMPs are produced, including TIMP-1, MMP-1, -2, -3, -9, and -14 [59]. An in vivo study on Fischer male rats showed that TIMP-1 directly regulates the smooth muscle cell migration [27]. A similar function has been recognized in human studies on MMP-14 and TIMP-1, with being both involved in the process of cellular migration to the plaque fibrous cap and plaque inflammation [28,29]. MMP-9 also contributes to the destruction of the fibrous cap itself in patients with increased CV risk [30]. Interestingly, increased concentrations of MMP-2, -3, and -9 have been found within the aneurysm human tissue, being mainly produced by the macrophages localized in the aneurysm wall [31]. These different enzymes are differentially expressed according to the aneurysm dimensions and severity [60,61]. MMP-2 is increased in small aneurysms (<5.5 cm), whereas MMP-9 is dominant in large aneurysms (5.5–7 cm). Moreover, it has also been shown that different localizations of MMP activity within the aneurysm wall modify the risk of rupture [60]. In clinical studies, abnormal circulating levels of MMP-1, -2, -8, -9, and TIMP-1 were found in patients with Peripheral Arterial Disease; their increase was attributed to the presence of ischemic tissue [12,62,63]. Circulating levels of MMP-1 and -8 were also found as predictors of poor prognosis, in terms of major amputation or death, in patients who underwent lower extremity bypass [64]. These multiple findings enhanced the importance of MMPs as biomarkers of arterial disease severity that also provide important prognostic information in clinical practice.

Regarding venous diseases, it has been observed that alterations in ECM remodeling are common in the case of varicose veins and chronic venous insufficiency (CVI). With regards to varicose veins, the expression of MMP-1, -2, -3, -7, and -9 is increased, particularly in the smooth muscular cell of the vein wall, both in human and mice models [65,66]. Moreover, an analysis of human saphenous vein showed that this expression is even higher in varicose veins with inflammation, as compared to those without inflammation [65]. Mechanisms underlying the association between MMPs and varicose vein physiopathology might involve the effect of MMPs on ECM degradation and the relaxation of the venous wall [66]. An upregulation of MMP-1, -2, -9, and -13, together with a downregulation of TIMP-1 and TIMP-2, have been described in patients with CVI [67,68]. The distribution of MMP varies based upon the stage (from CVI to active wound) and cells, suggesting that MMP-1 and TIMP-1 are needed in the re-epithelialization phase, while MMP-9 and -13 primarily participate in the remodeling of the collagenous matrix [69]. A summary of the principal pieces of evidence provided from this paper is depicted in Table 1.

**Table 1.** Summary of main paper results.


**Table 1.** *Cont*.


#### *3.5. Synthetic Metalloproteinases Inhibitors in Experimental and Clinical Research*

MMP activity is regulated at different levels, including either intracellular (mRNA expression and post-translational modification of MMP structure) or extracellular (stimulation or inhibition of their enzymatic activity from endogenous or exogenous substrates) process. The net activity of MMPs is a crucial step, since its up- or down- regulation could affect MMP activity, and ultimately lead to metabolic diseases, cancer, cardiovascular, and renal disorders [72]. For these reasons, new pharmacological agents that interfere with MMP activity have been developed and utilized as potential tools that could benefit a wide spectrum of patients. Synthetic MMPs inhibitors (MMPs-I) include broad-spectrum and specific MMPs-I. The vast majority of these compounds contain Zn2<sup>+</sup> in their structure, and have structured as Zn2<sup>+</sup> binding globulin (ZBG). Indeed, ZBGs inactivate MMPs by displacing the Zn2+- bound water in the MMPs, and favor the anchorage of the drug to the MMPs substrate binding-pocket [73]. ZBGs encompass hydroxamic acids Batimastat (BB-94), Marimastat (BB-2516), and Ilomastat (GM6001), that displays a broad-spectrum inhibition of MMPs. More selective ZBGs molecules have also been developed, and include hydrazides and sulfonylhydrazides that specifically inhibit MMP-1, -2, and -9. Hydroxamic ZBGs are effective, but they have poor oral bioavailability and, by inhibiting multiple MMPs, cause several musculoskeletal side effects [74]. Hence, heterocyclic bidentate chelators have been developed that have shown more biostability and lower toxicity in cells assays. Tetracyclines are antibiotic molecules that, by chelating Zn2<sup>+</sup> ion, are able to inhibit MMPs. Doxycycline inhibits MMP-2 and-9 [75]. Chemically-modified tetracyclines reach higher plasma levels for prolonged periods of time, require less frequent administration, are associated with lower rate of side effects when administered orally, and are thus preferred over conventional tetracyclines [76]. Apart from zinc-based compounds, several MMPs-I act by a noncompetitive, nonzinc-binding, mechanism of inhibition. They show high selectivity that minimizes the side effects, and thus, are considered very promising molecules [77]. MMPs-Is have already been used in pilot studies in mice and in human models of kidney damage [11]. MMPs-I BB-1101 has shown to reduce proteinuria in rats with anti-Thy1.1 nephritis, an experimental model of glomerular damage induced by antibody against Thy 1 gene [78]. A similar effect on proteinuria was found with the MMPs-I BB-94 in an experimental model of kidney allograft rejection in mice and the tetracycline antibiotic doxycycline in human patients with diabetic nephropathy already under renin-angiotensin-aldosterone inhibition [79,80]. Interestingly, doxycycline also reduced the aneurysm expansion in small randomized clinical trials enrolling patients with abdominal aortic aneurysm [81,82].

#### **4. Discussion**

The evidence for how the circulating levels or the expression of MMPs increase cardiovascular risk is well documented in both basic and clinical studies. However, to our knowledge, this represents the first review to encapsulate and describe the dual association between MMPs and CKD and MMPs and PVD.

Two large meta-analyses have separately summarized the strength of the relationship between MMP-2, TIMP-1, and subclinical atherosclerosis in CKD patients or between MMPs and vascular risk, regardless of the level of kidney function [58,83]. They concluded that imbalances in MMPs/TIMPS were implicated in the pathogenesis, clinical manifestations, and prognosis of arterial and venous diseases.

According to our results, we can also affirm that the CKD condition contributes to reinforcing the risk-pathways oriented from MMPs toward PVD. Numerous pieces of evidence support this: (1) Numerous MMP molecules are likely responsible for both CV (including PVD) and kidney damage and related clinical manifestations. Hence, such biomarkers warrant further investigation. The Gelatinases MMP-2 and MMP-9 are produced by both intrinsic glomerular and tubular cells [52,70]. It has been shown that the increased activity of MMP-2 and MMP-9 across the kidney tubules may lead to structural alterations in the tubular basement membrane, which starts epithelial-mesenchymal transition (EMT). EMT consists of the process in which a component of adherent junctions, the E-cadherin, and cell adhesion are mislaid, whereas epithelial cells acquire a mesenchymal phenotype by expressing and producing α-smooth muscle actin (αSMA) and matrix protein. All these mechanisms subsequently trigger tubular atrophy and fibrosis [71,84]. Moreover, Peiskerova and coworkers have demonstrated that levels of MMP-2 were significantly higher in CKD patients Stage III–V as compared to those with Stage I–II, and correlated with fibroblast growth factor 23 (FGF-23) and levels of serum phosphate, two important surrogates of oxidative stress and CV risk in these patients [52,71]. Similarly, gelatinases have also been implicated in the onset of structural changes associated with PAD. Circulating levels of MMP-2 and MMP-9 have been found to increase in patients with acute and chronic lower limb ischemia (i.e., intermittent claudication and critical ischemia), as well as in patients with hyperlipidemia when compared with healthy subjects [12,85,86]. The connecting link between CKD, gelatinases, and PAD is mainly represented by the increased atherosclerotic risk and arterial remodeling due to the imbalance of ECM enzymes. Indeed, MMP-2 and MMP-9 are capable of releasing latent transforming growth factor beta (TGF-β), which acts as a mediator in the cross-talk between endothelial cells and vascular smooth muscle cells [71]. Imbalances in MMPs and alterations in endothelial cells are a direct causative factor of abnormal extracellular matrices, vascular calcification with arterial stiffness, atherogenesis, and high pulse pressure [71,87]. TIMP-1 has been shown to exert a similar role. In CKD patients, TIMP-1 serum concentrations were significantly increased from Stage I–III to IV–V [88]. Moreover, TIMP-1 has been found to be a predictor of the new onset of CKD and heart failure, regardless of age, gender, and biomarkers of systemic inflammation (c-reactive protein and brain natriuretic peptide) [40]. TIMP-1 is also increased in patients with PAD, with ischemic muscle being a possible source for this enzyme [12]. Interestingly, an association between TIMP-1 levels and TGF-β was described in young patients with CKD, suggesting that this factor may be involved in the pathogenesis of inflammation and CV damage even at an early stage of the disease [89].

A further crucial role in the link between metalloproteinases, CKD, and PVD is indicated by NGAL. It has been shown that NGAL levels are increased in atherosclerosis and chronic inflammatory processes, and in particular, are overexpressed in atherosclerotic plaques which are vulnerable to rupture [90]. Moreover, NGAL circulating levels have been found to increase in patients with CKD, being directly related to diabetic status and inversely to the eGFR levels [91]. From a prognostic perspective, in both CKD and general populations, plasma NGAL was a significant predictor of CV fatal and nonfatal events, including PVD [92,93]. These findings were globally reinforced by the evidence that plasma and tissue NGAL expression contribute to the pathogenesis and severity of central and peripheral aneurismal disease [94]. In this context, it has been proposed that NGAL reflects the leukocyte-mediated inflammatory response, with higher values being associated with weakness of the vascular wall and impaired vascular damage [95]. (2) CKD is characterized by a persistent low-grade inflammation with the production of pro-inflammatory cytokines, and MMP activity has been implicated in these pathways. The uremic milieu may also induce oxidative stress and recurrences

in thrombotic events and infections, as well as impairment in arterial stiffening and calcification of both the intima and media of the arterial wall [55]. In the pathogenesis of all these processes, a crucial step is represented by ECM remodeling. MMPs are directly involved in ECM processes in almost all tissues. (3) In patients with already diagnosed CKD, serum levels of MMPs have been found to be significantly associated with the degree of proteinuria. This means that MMPs may be directly implicated in the development of glomerular or tubular damage typical of a wide spectrum of kidney diseases. The presence of proteinuria is a strong predictor of CV events (and PVD as well), and is recognized as a marker of endothelial systemic damage. Moreover, its prognostic role is even stronger than that attributed to the traditional CV risk factors in CKD patients (i.e., Framingham risk factors) [7,8,48]. Indeed, MMPs are detected and elevated in human kidneys with nephritic syndrome, which is a high-risk condition with severe proteinuria [51,96]. (4) It should not be excluded that MMPs play a predictive other than a prognostic role in patients with CKD or PVD. This means that the reduction of MMPs following a selective treatment could be associated with a better prognosis in these patients. Recently, several therapeutic agents that modulate the function of MMPs have been developed. The tetracycline antibiotic doxycycline and the nonselective inhibitors of MMPs, Batimastat and Marimastat, have been shown to cut vascular tissue remodeling by reducing MMP activity [97,98]. However, larger clinical studies are needed to generalize and translate these effects to clinical application. Moreover, the reduction in proteinuria that followed the use of doxycycline was associated with a reduction in the glomerular expression of MMP [58]. Also, the Renin-Angiotensin-Aldosterone system inhibitors (RAAS-I), that are currently considered the most effective renoprotective drugs against the risk of ESKD and CV events in CKD patients, have been shown to decrease the MMPs levels [99–101]. However, all these important pieces of evidence have led to the hypothesis that MMPs could be one of the causal links between CKD and CV risk, including PVD. (5) In the era of precision medicine, the need to ameliorate prognoses in patients with chronic diseases is currently increasing [102]. To this end, a number of clinical trials evaluating the beneficial effect of new drugs on cardiovascular (including peripheral vascular disease) and renal outcomes have been started [103]. Interventions varied between studies, with the effect of blood pressure lowering drugs, albuminuria lowering agents, diuretics, and endothelin receptor antagonist being tested. Results from these first studies increased the interest in a relatively new drug class, Sodium–glucose cotransporter 2 (SGLT-2) inhibitors. SGLT2-Is reduce the reabsorption of glucose from the proximal tubules of the kidney. The SGLT2-Is canagliflozin and empagliflozin reduced the risk of developing fatal and nonfatal CV events in previous trials, with this effect being associated with a reduction of albuminuria in the first 3 to 6 months of treatment [104,105]. However, an interesting recent analysis of human proximal tubular epithelial cells showed that empagliflozin is also able to reverse the renal suppression of Reversion Inducing Cysteine Rich Protein with Kazal Motifs (RECK), a membrane anchored endogenous MMP inhibitor whose expression is induced by hyperglycemia in animal and human models of diabetic kidney disease [106]. Interestingly, empagliflozin reduced the epithelial-to-mesenchymal transition, a mechanism that is strictly linked to MMPs activity and that anticipates kidney fibrosis, by directly reducing migration of tubular cells and expression of pro-inflammatory mediators (TRAF3IP2, NF-κB, p38MAPK, miR-21, and IL-1β, IL-6, TNF-α, and MCP-1). This means that SGLT2-Is may exert a protective effect on renal tubules that is independent from the albuminuria reduction. Thus, new large intervention studies evaluating the effect of drugs that interfere with MMPs activity are expected in the future to evaluate their impact on prognoses of albuminuric and nonalbuminuric CKD patients. This would be particularly important considering that nonalbuminuric CKD patients are growing in prevalence as referred patients in nephrology clinics [47,107]. In the context of intervention studies, MMPs may play a role as "predictive" biomarkers, since their assessment could identify a population of patients which is more likely to benefit from a specific drug. This point is crucial, since the development and optimization of new treatments that reduce CV risk in CKD patients may help to overcome the variability in response to the old standard treatments [108].

In addition to clinical trials perspective, further efforts are needed in the context of observational studies and risk stratification. Indeed, risk stratification of CKD patients may improve prevention strategies, acting to slow down CKD progression and reduce the high residual risk in CKD patients. In observational research, it may be useful to evaluate the potential role of MMPs as "prognostic" biomarkers used to identify the likelihood of a patient to develop a clinical outcome, regardless of treatment (i.e., under the standard of care). Such a measure may improve physicians' abilities to identify patients with poor prognoses [109]. One approach involves the discovery of novel biomarkers that may add proper prognostic information on top of already known risk factors [46,107,110–113]. Investigators from the Steno Diabetes Center, a prospective cohort that contributed to the comprehension of prognosis of diabetic patients with or without CKD, showed that MMP-1 and MMP-2 are associated with an increased risk of CV events and CV mortality, regardless of age, gender, eGFR, and albuminuria [114]. A similar effort in improving risk stratification is ongoing in the case of PAD management. It has indeed been demonstrated that patients suffering from PAD are exposed to a higher risk of CV events than those with coronary heart disease [115,116]. Hence, ameliorating prognoses of these patients is mandatory and urgent. A previous recent analysis showed that variations in the serum levels of MMPs before and after lower limb surgical revascularization in patients with critical limb ischemia are significantly associated with the subsequent outcome (major amputations or death). From our research, we can also assert that measuring the serum levels of MMPs could be even more useful in patients with CKD, PVD, or both conditions, as they potentiate the risk of progression for these conditions.

**Author Contributions:** Conceptualization, M.P., M.A., C.G., and R.S.; methodology, R.S.; validation, M.P., M.A., C.G., and R.S.; formal analysis, T.F., A.M., N.I., R.G., P.S. and S.d.F.; investigation, M.P., M.A., C.G., P.M. and R.S.; data curation, P.M., and P.S.; writing—original draft preparation, M.P., M.A., C.G., T.F., A.M., N.I., R.G., P.S., P.M., S.d.F., and R.S.; writing—review and editing, M.P., M.A., C.G., and R.S.; supervision, M.A., and R.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** The work was not founded.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Review* **Challenges in Matrix Metalloproteinases Inhibition**

#### **Helena Laronha 1,2, Inês Carpinteiro 1, Jaime Portugal 3, Ana Azul 1, Mário Polido 1, Krasimira T. Petrova 2, Madalena Salema-Oom 1,2 and Jorge Caldeira 1,2,\***


Received: 9 April 2020; Accepted: 30 April 2020; Published: 5 May 2020

**Abstract:** Matrix metalloproteinases are enzymes that degrade the extracellular matrix. They have different substrates but similar structural organization. Matrix metalloproteinases are involved in many physiological and pathological processes and there is a need to develop inhibitors for these enzymes in order to modulate the degradation of the extracellular matrix (ECM). There exist two classes of inhibitors: endogenous and synthetics. The development of synthetic inhibitors remains a great challenge due to the low selectivity and specificity, side effects in clinical trials, and instability. An extensive review of currently reported synthetic inhibitors and description of their properties is presented.

**Keywords:** matrix metalloproteinases; TIMP; synthetic inhibitors

#### **1. Introduction**

Matrix metalloproteinases (MMPs) are a protein family within the metzincin superfamily, comprising zinc-dependent endopeptidases with similar structural characteristics but with different substrate preferences. MMPs are produced and secreted from cells as inactive proenzymes depending, herein, on a structural alteration for activation [1–6]. In human tissues, there are 23 different types of MMPs expressed and they can be subdivided according to their substrate specificity, sequential similarity, and domain organization [1,2,4,7–17] (Table 1).

The most common structural features shared by MMPs are [1,2,4,5,7,8,10–14,16,18] (Figure 1) a pro-domain, a catalytic domain, a hemopexin-like domain, and a transmembrane domain for membrane type MMPs (MT-MMPs) although some MMPS do not have all the structural features represented in the figure. The pro-domain keeps MMP inactive by a cysteine switch, which interacts with the catalytic zinc making it impossible to connect the substrate. The catalytic domain has two zinc ions, three calcium ions, and three histidine residues, which are highly conserved [1–9,11–20]. In the terminal zone of the catalytic domain there is a region that forms the outer wall of the S1' pocket [1,14,17]. This pocket is the most variable region in MMPs and it is a determining factor for substrate specificity [1,2,6,7,11,17,18]. However, there are six pockets (P1, P2, P3, P1', P2', and P3') and the fragments of the substrates or inhibitors are named depending on the interaction with these pockets (R1, R2, R3, R1' or Ra, R2', and R3'). The linker is proline-rich, of variable length, allowing inter-domain flexibility and enzyme stability [4,8,12,13]. The hemopexin-like domain is necessary for collagen triple helix degradation and is important for substrate specificity [3,4,7,9,19].


**Table 1.** Matrix metalloproteinases (MMPs) classes.

**Figure 1.** Schematic representation of the general structure of MMP.

TheMMPs can process ECM proteins and glycoproteins, membrane receptors, cytokines, hormones, chemokines, adhesion molecules, and growth factors [1,3,4,6,7,9–11,13,14,20–26]. However, the presence and the activity of MMPs have been demonstrated to be intracellular [25,26]. For example, some studies show intracellular localization of MMP-2 in cardiac myocytes and colocalization of MMP-2 with troponin I in cardiac myofilaments [23]. The MMP-2 activity has also been detected in nuclear extracts from human heart and rat liver [23]. The MMPs are involved in many biologic processes, such as tissue repair and remodulation, cellular differentiation, embryogenesis, angiogenesis, cell mobility, morphogenesis, wound healing, inflammatory response, apoptosis, ovulation, and endometrial proliferation [1,2,4,6,8,10,11,13,16–18,20,27]. The deregulation of MMPs activity leads to the progression of various pathologies depending on which enzyme is involved [1,6,10,13–17,20,27]: cancer and metastasis, inflammatory processes, arthritis, ulcers, periodontal diseases, brain degenerative

diseases, liver cirrhosis, fibrotic lung diseases, otosclerosis, atherosclerosis, multiple sclerosis, dilated cardiomyopathy, aortic aneurysm, or varicose veins.

Although therapeutic strategies for specific inhibition of MMPs have been long researched, they are difficult to develop because these enzymes are involved in a myriad of pathways [2,5]. However, this inhibition can be done at the biomolecular expression and active enzyme terms [2,5,18]. The MMPs inhibitors can be divided into endogenous inhibitors, which can be specific or non-specific, and synthetic inhibitors [1,2,4,7,10,12–14,16,20,28,29] (Table 2).


**Table 2.** MMPs inhibitors classification.

#### **2. Specific Endogenous Inhibitor-Tissue Inhibitors of Metalloproteinases (TIMPs)**

Tissue inhibitors of metalloproteinases (TIMPs) are endogenous proteins responsible for the regulation of MMPs activity, but also of families such as the disintegrin metalloproteinases (ADAM and with thrombospondin motifs ADAMTS) and therefore for maintaining the physiological balance between ECM degradation and MMPs activity [1,2,8,9,18,30]. There are four TIMPs (TIMP-1, -2, -3, and -4) (Table 3), with 22–29 KDa and 41%–52% sequential similarity [2,4,12,13,16,20,31].


**Table 3.** Tissue inhibitors of metalloproteinases (TIMPs) classification.

TIMPs consist of a *N*- and *C*-terminal domain with 125 and 65 amino acids, respectively, each containing six conserved cysteine residues, which form three conserved disulphide bonds [2,4,7–9,12,31,32] (Figure 2a). The *N*-terminal domain is an independent unit, which can be inhibited by MMPs, in a 1:1 ratio [2,4,8–10,12,13,16,20]. This domain has two groups of four residues: Cys-Thr-Cys-Val and Glu-Ser-Val-Cys (Figure 2b), which are connected by disulphide bounds which are important for TIMP activity [7,12]. This is the main domain responsible for MMP inhibition through

its binding to the catalytic site in a substrate-like manner [31]. The several domains allow the TIMP and pro-gelatinases interactions [4].

**Figure 2.** (**a**) TIMP-1-catalytic domain of the MMP-3 complex. (**b**) TIMP-1-catalytic domain of the MMP-3 complex, where two conserved groups, Cys-Thr-Cys-Val and Glu-Ser-Val-Cys, are represented in yellow.

#### **3. Non-Specific Endogenous Inhibitors**

Non-specific endogenous inhibitors have been reported to inhibit MMPs (Table 4), however, the inhibition mechanism details have only been partially discovered [7,12].


**Table 4.** Non-specific endogenous inhibitors [4,7,12,13,33,34].

Human α2-macroglobulin is a glycoprotein with four identical subunits that act by entrapping MMP and the complex is cleared by endocytosis [2]. The α2-macroglobulin has been found in blood and tissue fluid [2,31]. The tissue factor pathway inhibitor (TFPI) is a serine proteinase inhibitor, which targets MMP-1 and -2, but this inhibition mode is still unknown [7,12]. The *C*-terminal proteinase enhancer protein and tissue factor pathway inhibitor have sequences with certain similarities to the *N*-terminal domain of TIMPs [31].

#### **4. Synthetic Inhibitors**

MMPs are molecular targets for the development of therapeutic and diagnostic agents [14]. The development of synthetic MMP inhibitors was initially based on the peptide sequence, recognized by proteases, with different chemical functionalities, capable of interacting potently with zinc ion [11–13,19]. The requirements for an effective inhibitor are [2,11,13,17,19,35]:


**Figure 3.** Examples of zinc binding groups (ZBGs). (**a**) Hydroxamate-based inhibitor; (**b**) thiolate-based inhibitor; (**c**) carboxylate-based inhibitor; (**d**) phosphorous-based inhibitor. The R group is the scaffold of inhibitor.

ZBGs have negative charges that prevent their penetration in the cell, restricting their activity to the extracellular space which reduces their cell toxicity [2]. Changes in the ZBG structure or in the point of attachment of the ZBG to the backbone of the MMP inhibitor can change its potency and selectivity [2]. When comparing what selectivity or potency of different ZBGs leaving the structure constant, Castelhano et al. arrived at the following list [36]: hydroxamic acid >> formylhydroxylamine > sulfhydryl > aminocarboxylates > carboxylate.

The selectivity of inhibitor is a primordial goal of MMPs' inhibitors (MMPis) design to increase efficacy and prevent side effects [2]. This selectivity is based on two molecular characteristics [11]: a chelating capacity for catalytic zinc and the presence of hydrophobic bridges of the active center for the S1' pocket. Numerous strategies have been suggested for creating selective MMP inhibitors [27] (Figure 4): endogenous-like inhibitors, exosite targeting inhibitors, a combination of exosite binding and metal chelating inhibitors, and function-blocking antibodies.

**Figure 4.** Types of MMPs' inhibitors: (**a**) endogenous-like inhibitors that chelate the catalytic zinc (II) ion; (**b**) exosite targeting inhibitors that alter the conformation of the enzyme; (**c**) a combination of exosite binding and metal chelating inhibitors; (**d**) antibodies inhibitors.

The synthetic inhibitors have had a great challenge in their development since first it is necessary to identify the enzymes that are involved in disease progression. Moreover, this goal has an additional difficulty as there are more than 50 human metalloproteinases (23 MMPs, 13 ADAM, and 19 ADAMTS) [2].

#### *4.1. Hydroxamate-Based Inhibitors*

The first generation of MMPis (1995–1999 [16]) were designed based on the knowledge of the triple helix collagen amino acid sequence (cleavage site) and the information derived from specific substrates [6,15,19,31,35,37]. These compounds contain a hydroxamic acid group as ZBG [5,6,15,18,27,28,31,37]. Hydroxamic acids (HA) were first described in 1986 [38,39]. They are easy to synthesize, are monoanionic compounds, bidentate chelating agents, and due the excellent zinc-chelating capability they are the more popular ZBG for MMPs [2,6,17,18,27–29]. The hydroxamic group established interactions with zinc ions, through two oxygens and two hydrogen bonds (NH and OH groups of HA with Ala and Glu, respectively), forming a distorted triangular bipyramid [2,5,18,19,28,29] (Figure 5a). Reich et al., in 1988 [40], used a hydroxamic acid compound, S*C*-44463 (Figure 5b), to block collagenase and prevent metastasis in a mouse model, which initiated the era of MMP inhibition therapeutics [40].

**Figure 5.** (**a**) Interaction between hydroxamate group and catalytic zinc (II) ion. The oxygen of the hydroxamate forms a strong hydrogen bond with the carboxylate oxygen of the catalytic Glu, while the NH of hydroxamate establishes another hydrogen bond with the carbonyl oxygen of Ala; (**b** S*C*-44463 inhibitor.

The structure–activity relationship (SAR) studies for a series of hydroxamic acids with a quaternary carbonyl group at R1 suggested that [13] (Figure 6):


**Figure 6.** General structure of hydroxamic acid. OH group: catalytic zinc binding group; Ra: α substituent; R1: P1' substituent group and this group is determinant to selectivity and activity; R2: P2' substituent and this substituent can be cyclized with Ra and R3; R3: P3' substituent.

Modification in Ra position (α substituent): a beneficial effect is conferred by lipophilic substituents capable of hydrogen bonding [19]. Analogues of Marimastat have been reported, where the α position is disubstituted by a hydroxyl group and a methyl group [19] (Figure 7a). The X-ray structures of the MMP-3-inhibitor complex showed that there were hydrogen bonds between the hydroxyl group and Ala165 [19]. By binding the positions Ra and R2 in a single chain, forming a cyclic inhibitor (Figure 7b), there was a substantial increase in water solubility [19,29,41,42]. This strategy led to the discovery of two inhibitors with similar potency to non-cyclized analogues, SE205 and SC903 [41] (Figure 7c,d).

**Figure 7.** (**a**) Analogue of Marimastat with the α position disubstituted; (**b**) analogue of Marimastat with Ra and R2 position connected; (**c**) SE205; (**d**) SC903.

The introduction of conformational restrictions through the addition of a three-membered ring between positions α and R1 (Figure 8a) has been reported by Martin et al. [43], and it resulted in reduced inhibition of MMP-9 [43]. However, the introduction of a six-membered ring between these same positions resulted in the inactivity of the compound [19] (Figure 8b).

**Figure 8.** Inhibitors with conformational restrictions between Ra and R1 positions. (**a**) Inhibitor with three-membered ring; (**b**) inhibitor with six-membered ring.

Exploring the depth of S1' pocket, bulky groups in Rα position confer selective inhibition for MMP-2, -8, and -9 [29]. An example is the presence of a biphenyl group, which showed higher inhibitory activity against MMP-9 [29].

Modification in R1 position: the incorporation of long groups in the R1 position can promote the selectivity of MMPis, since pocket S1' can undergo conformational changes to accommodate certain substituents [19].

Broadhurst et al. showed that an alkyl chain (C9) at the R1 position reduces the inhibition of MMP-1, but maintains the inhibitory activity against MMP-2, -3, and -9 [19] (Figure 9a). For matlystatin derivates in the C9 chain, R-94138 (Figure 9b) promotes the inhibition against MMP-9 [44]. The succinyl hydroxamates analogues in the C9 chain promote selectivity for MMP-2, however, a C10 chain (Figure 9c) results in MMP-1 inhibition, while a further increase of the chain to C16 (Figure 9d) leads to a loss of activity against MMP-1 [19].

**Figure 9.** Inhibitors with modification of R1 position. (**a**) Inhibitor with alkyl chain. This inhibitor has activity against MMP-2, -3, and -9, but the inhibition of MMP-1 is low; (**b**) R-94138, Matlystatin derivate. The inhibition of MMP-9 is 10 times higher than analogues with C8 or C10 chains; (**c**) succinyl hydroxamate analogue with C10 chain, which inhibits MMP-1; (**d**) succinyl hydroxamate analogue with C16 chain, which inhibits MMP-1.

Replacement of the R1-R2 bond of succinyl hydroxamates acid inhibitors by a sulfonamide bond (Figure 10a) results in substantial loss of inhibitory activity because the hydrogen bond (C=ONH; Figure 10b) with leucine is stronger than the new sulfonyl oxygen bond, due to the pyramidal nature of the sulfonamide [19].

**Figure 10.** (**a**) Succinyl hydroxamate acid with a sulphonamide bond. This compound presents low inhibitory activity because of the pyramidal nature of the sulphonamide group. (**b**) Succinyl hydroxamate acid with carbonyl bond.

Modifications in R2 position: modifications in the R2 position led to a modest effect in inhibitory activity, in vitro, and affects the pharmacokinetic properties [29]. Marimastat and Ro31-9790 (Figure 11) have a good oral activity because the bulky tert-butyl group assists the adjacent amide bond during absorption from an aqueous environment to the lipid environment of the cell membrane [45]. The beneficial combination of the tert-butyl group with the α-hydroxyl group increases the water solubility [19,29]. Babine and Bender suggest that the tert-butyl as R2 group leads to less Van der Waals interactions, comparing with other groups [46].

**Figure 11.** (**a**) Marimastat. The Ra position is substituted with a hydroxyl group (OH). (**b**) Ro31-9790. The Ra position has no substituents.

Ikeda et al. described compounds with phenyl R2 substituents (Figure 12) (KB-R7785), which are active orally, due to the beneficial effect of the R2 phenyl group on absorption, where the amide shielding and lipophilicity may assist in transepithelial resorption [47]. This inhibitor shows activity against MMP-1 in rats and its effectiveness in arthritis has been demonstrated [47].

**Figure 12.** KB-R7785 inhibitor.

Modifications in R3 position: the S3' pocket is an open area and several groups can be introduced at the R3 position [19]. The introduction of the benzhydryl group leads to compounds with selectivity to the MMPs-3 and -7 [19].

#### 4.1.1. Succinyl Hydroxamic Acid-Based Inhibitors

Succinyl hydroxamate derivates can be subdivided to peptide derivatives or non-peptide compounds [48]. The *N*-acetylcysteine has been reported to affect the tumoral invasion process and metastasis by MMP-2 and -9 inhibition [49]. The L-cysteine-2-phenylethylamide is an effective inhibitor, in which the phenyl group fills the S1' pocket of MMP-8 [50]. Foley et al. prepared several dipeptides derivatives containing cysteine (**R**CO-Cys-**AA**-NH2) and concluded [51]:


Batimastat (Table 5) was the first MMPi to enter in clinical trials for cancer as it inhibits MMP-1, -2, -7, and -9, but, due to its poor oral bioavailability, it was superseded by Marimastat [28,29,31] (Table 5), which has an alpha-hydroxyl group increasing the aqueous solubility [29]. Marimastat inhibits the activity of MMP-1, -2, -3, -7, -9, -12, and -13 [31]. However, Marimastat failed in clinical trials due to the absence of a therapeutic effect and the patients treated developed musculoskeletal

toxicity (MST) [6,31]. Batimastat, marimastat, and ilomastat are examples of succinyl hydroxamates, which have very analogous structure to that of collagen and inhibit MMPs by bidentate chelation of the Zn2<sup>+</sup> [2,6,29].

**Table 5.** Batimastat and Marimastat.

Several studies by Jonhson et al. [52] demonstrated that derivatives of succinyl hydroxamic acid (Figure 13a) are more potent for MMP-1 than the corresponding malonyl (Figure 13b) or glutaryl (Figure 13c) derivates.

**Figure 13.** (**a**) Derivates of succinyl hydroxamic acid; (**b**) malonyl acid; (**c**) glutaryl acid.

Marcq et al. [53] developed succinyl hydroxamates derivates selective for MMP-2, by modifications on Ilomastat structure (Figure 14a) to increase the overall hydrophobicity and, consequently, the selectivity [53]. This study resulted in a compound (Figure 14b) with an isobutylidene group of E geometry, which showed a 100-fold greater selectivity for MMP-2 over MMP-3, that is a 70-fold increase compared to Ilomastat [53].

**Figure 14.** (**a**) Ilomastat; (**b**) Ilomastat derivate with isobutylidene group.

Table 6 shows the IC50 and Ki values of some succinyl hydroxamic acid-based inhibitors [6,15–19,29,35,37,54,55].

#### 4.1.2. Sulfonamide Hydroxamic Acid-Based Inhibitors

In 1995, Novartis described the CGS-27023A (Figure 15a), a non-peptidic MMP-3 inhibitor, which has good oral availability but did not succeed in clinical trials [56]. The isopropyl group slows down the metabolization of the adjacent hydroxamic acid group and the 3-pyridyl substituent may aid partitioning into the hydrated negatively charged environment of the cartilage [56]. By analysis of the cocrystal structure of this inhibitor and MMP-12, it was possible to conclude that the binding mode between the hydroxamate moiety and the catalytic zinc ion was the same as the binding mode of hydroxamate-based inhibitors [6]. The interaction of CGS-27023A with MMP-3 was possible due to the p-methoxy phenyl substituent occupation of the S1' pocket and the pyridylmethyl and isobutyl groups occupation of the S2' and S1 pockets, respectively [29,56]. The modification of α to form the thioester derivate led to an increase of the inhibition of the deep pocket of the MMPs [29,56] (Figure 15b).

**Figure 15.** (**a**) CGS-27023A; (**b**) thioester derivate of CGS-27023A.

The NNGH (Figure 16a) (*N*-Isobutyl-*N*-(4-methoxyphenylsulfonyl)glycyl hydroxamic acid) was the starting point to many potent MMPis and is accommodated in the entry of the S1' pocket, but does not penetrate it [6]. Barta et al. described a series of arylhydroxamate sulphonamides, active against MMP-2 and -13 (Figure 16b) [57]. In this compound, the sulfonyl group formed a single hydrogen bond with Leu160 and the piperidine-*O*-phenyl moiety extends into the S1' pocket by Van der Waals interactions [57]. Noe et al. described a series of 3,3-dimethyl-5-hydroxy pipecolic hydroxamic acid, which possess potent inhibitory activity for MMP-13 [58]. In the first series of compounds, the 3-position of the piperidine ring was explored by the introduction of a polar functionality and it resulted in a compound with excellent activity on MMP-13 (Figure 16c), improved bioavailability, and lower metabolic clearance [58].

**6.**IC50andKivaluesofsuccinylhydroxamicacid-basedinhibitors.

120

122

**IC50**: MMP-1 = 4.6 μM; MMP-2 = 4 nM; MMP-3 = 42 nM;

MMP- 8 > 3.1 μM; MMP-9 > 2.128 μM; MMP-13 > 5.025 μM; MMP-14 > 5.290 μM; MMP-15 > 7.088 μM; MMP-16 > 5.554 μM

MMP-10 > 1.650 μM; MMP-13 > 5 μM;

MMP-14 = 163 nM; MMP-15 = 1.7 μM

> MMP-7 > 10 μM; MMP-9 = 120 nM

**IC50**: MMP-1 = 5.9 μM; MMP-9 = 560

**Table 6.** *Cont.*

**Ki**:

**Figure 16.** (**a**) NNGH; (**b**) arylhydroxamate sulphonamide compound; (**c**) 3-hydroxy-3-methylpipecolic hydroxamates.

The incorporation of a cyclic quaternary center α led to a strong inhibitory effect against MMP-1, -2, -3, -8, -9, -12, and -13 [29]. The RS-113,456 (Figure 17a) is an inhibitor with better oral bioavailability and metabolic stability compared to the hydroxamate derivates [35]. These two features were improved by shifting the cyclic group to the α-position of the hydroxamic acid (Figure 17b) [29].

**Figure 17.** (**a**) RS-113,456; (**b**) RS-130,830.

Table 7 shows the IC50 and Ki values of some sulfonamide hydroxamic acid-based inhibitors [6,15–19,29,35,37,54,55].

#### 4.1.3. Phosphamides Hydroxamic Acid-Based Inhibitors

The hydroxamic acids based on phosphamides are effective as MMPis due to the electronic environment of the phosphor atom [11]. The replacement of sulphonamide group by phosphinamide group leads to a potent inhibitor of MMP-3 (Figure 18), the collagenases and gelatinases [29]. The interactions between this inhibitor and the MMP-3 are realized by the phosphinamide phenyl group, that accommodates into the S1' pocket and by the phosphinamide oxygen, which establishes the hydrogen bonds with NH of Leu164 and Ala165 [29]. However, this group is susceptible to hydrolysis at low pH, limiting the inhibitory activity [29].

**Figure 18.** Inhibitor with phosphinamide group.

**Table 7.** *Cont.*

All hydroxamate-based inhibitors are very potent and they inhibit MMPs at low concentrations [18]. On the other hand, the hydroxamate acids have poor oral bioavailability, inhibit multiple MMPs, and cause side effects [2,17,27,28,35]. Additionally, this functional group may be metabolized via dehydroxylation or may be cleaved by endopeptidases and releasing hydroxylamine, can be hydrolyzed to carboxylic acids or reduced to *O*-glucuronyl or *O*-sulfate, leading to decreased effective inhibitor concentration and reducing its potency in vivo [18,27,28,35].

Table 8 shows the IC50 and Ki values of some phosphamides hydroxamic acid-based inhibitor [6,15–19,29,35,37,54,55].


*4.2. Non-Hydroxamate-Based Inhibitors*

The side effects caused by hydroxamate-based inhibitors, due to the lack of selectivity and in vivo lability, have been fostering the development of new compounds with alternative ZBGs [5,6,11,16,17,27–29]. The second generation of MMPis (1999–2003 [16]) was designed with a wide variety of peptidomimetic and non-peptidomimetic structures with higher selectivity and exploiting the deep S1' pocket present in some MMPs [6,15,16,18,59–61]. These compounds include carboxylic acids, sulfonylhydrazides, thiols, aminomethyl benzimidazole, phosphorous-based, nitrogen-based and heterocycles bidentate chelators, and can be monodentade, bidentade, and tridentade chelates [2,6,18,28,29].

The non-hydroxamate-based inhibitors open up a wide spectrum of affinities for the zinc ion from the catalytic site and new opportunities for targeting and inhibiting the active center [18,28]. They have weak Zn2<sup>+</sup> chelating ability and the rates of severe side effects, such as the musculoskeletal syndrome (MSS) decreased dramatically compared with the hydroxamate inhibitors [28].

#### 4.2.1. Thiolates-Based Inhibitors

The ability of the monodentate binding of thiols to zinc ion in proenzymes has served as inspiration for the design of several MMPis [5,29]. The potency of thiol inhibitors is intermediate between that of hydroxamate- and carboxylate- based inhibitors [29]. The first example of inhibitor thiol-based for MMP-1 is a bipeptidic analogue, where the incorporation of a thiol group as α substituent leads to improvement of activity (Figure 19a) [19]. Derivates with "linker" substituent between P1-P1' positions show a total loss of activity (Figure 19b) [19,62]. On the contrary incorporation of a methyl carboxylate group leads to a significant increase in activity (Figure 19c) [19]. The increased activity of these compounds may be a consequence of beneficial interactions between S1, the carbonyl ester, and the thiol group, participating in the bidentate coordination of the zinc [19].

**Figure 19.** (**a**) Thiol-based inhibitor with the thiol group as α-substituent. The stoichiometric is S when the thiol group is present. In its absence, the compound with R stoichiometric is more active than the S analogue; (**b**) thiol-based inhibitor with "linker" substituent; (**c**) thiol-based inhibitor with methyl carboxylate group.

Montana et al. have identified a series of inhibitors with mercaptoacyl, obtaining moderate inhibitors (Figure 20a) against a wide variety of enzymes with a deep pocket shown to be orally active in mouse models with arthritis [63]. The thiol and acyl carbonyl groups could cooperate in binding to the zinc of the active site [63]. Warshasky et al. have produced a variety of compounds in the Montana series, in which the amide nitrogen P2' is linked to the group P1' (Figure 20b) [19].

**Figure 20.** (**a**) Inhibitor with mercaptoacyl, where the thiol and acyl carbonyl groups could cooperate in binding to the zinc ion. (**b**) Variant of Montana compounds. n = 0, the compound does not have activity against MMPs-2, 3, and -12. n = 1, the compound has low activity against MMP-3.

The β-mercaptoacilamide represented in Figure 21 is active against the MMP-9 in vitro and exhibits oral activity in rats [19]. The 4-alcoxy substituent of cyclohexane group improved the activity against all MMPs [19]. Replacement of the 4-ethoxy substituent with 4-propyloxy leads to a significant reduction in MMP-1 activity and improves selectivity for MMP-3 [19]. The equivalent cyclopentyl compounds are inactive [19]. The mercaptoamide is unstable in solution hence, to overcome this issue, Campbell and Levin have prepared a series of mercaptoalcohols and mercaptoketones inhibitors [64]. The mercaptoalcohols have exhibited modest activity against MMP-1, -3, and -9, while the equivalent mercaptoketones could be optimized to active broad-spectrum inhibitors [64].

**Figure 21.** β-mercaptoacilamide inhibitor.

In 2005, Hurst et al. [65] reported a series of mercaptosulphides inhibitors that targeted MMP-1 [65]. The structure–activity relationship indicates that the five-membered ring increases the stability of the inhibitor compared to the linear structure, which can be quickly oxidized and lose its potency [65].

Table 9 shows the IC50 and Ki values of some thiolates-based inhibitors [6,15–19,29,35,37,54,55].

#### 4.2.2. Carboxylates-Based Inhibitors

The carboxylic inhibitors are synthetic precursors of the more popular hydroxamates yet they are weaker zinc (II) ligands than hydroxamates [17,27] and monodentate chelate [27]. Carboxylic acid is present in several MMPis that contain large lipophilic groups, such as biphenyls, since they fit in the S1' pocket [5,6]. These ZBGs are particularly appreciated for their high stability in vivo and their great positive effects on solubility, bioavailability, and selective properties [5,17]. The hydroxamate-based inhibitors are more potent in physiological conditions than carboxylate inhibitors, due to differences in acidity constants [29]. The carboxylate inhibitors bind more tightly to MMPs at low pH, while hydroxamate-based ones have a wider range of pH from 5 to 8 [29]. Fray et al. [66] compared the inhibition profiles of hydroxamates and carboxylic inhibitors (Figure 21a) and observed that the substitution of a carboxylate by a hydroxamate causes a 10-fold increase in potency of the inhibitor towards MMP-3 but decreases the selectivity against MMP-1, -2, -9, and -14 [66]. This effect is attributed to the fact that the strong zinc (II) affinity to the hydroxamic acid group is the main determinant of the binding energy, while in carboxylates this energy relies to a bigger extent on specific interactions with the specific pockets [66].

Hagmann et al. [67] described a series with *N*-carboxyalkyl group substituents, which presented inhibition for MMP-1, -2, and -3 [67] (Figure 21b). However, the substitution of the phenethyl group, in P1' position, for a linear alkyl chain removes the inhibitory activity for MMP-1 but it does not affect the activity for MMP-2 and -3 [67]. A similar effect was achieved by the 4-substitution of the phenyl ring of the phenethyl group with a small linear alkyl group [67] (Figure 22b). A similar range of P3 esters has been identified with "phthalamidobutyl" (Figure 22b), increasing activity against MMP-3 and further increasing selectivity [67].

**Figure 22.** (**a**) Fray et al. inhibitors. R = NH(OH), hydroxamate-based inhibitor. R = OH, carboxylate-based inhibitor. (**b**) Hagmann inhibitors. If X = H and Y = Me the compound presents inhibition to MMP-1, -2, and -3. When X = C4H9 and Y = Me, the inhibitor has a similar effect to the previous one. The inhibitor with X = H and Y = Phthbutyl (phthalamidibutyl) shows activity against MMP-3.

The interaction of the P1 biphenyl substituent with pocket S1 is an important factor contributing to the binding of the inhibitor [19]. The X-ray structure of the acyclic compound with MMP-3 revealed an important interaction between the phenyl terminal of the biphenyl group and the side chain of histidine (His224) [19]. The carboxylic acids derived from "D-valine" have a selective inhibition for MMP-2 and -3 [19]. The 4-substitution of the biphenyl ring helped to increase potency compared to the unsubstituted analogue and also helped to improve the pharmacokinetic properties [19].

With the aim of development inhibitors with high selectivity for a single MMP, Wyeth published, in 2005, a series of biphenyl compounds with carboxylates sulphonamides (Figure 23a). These compounds were tested for the treatment of osteoarthritis and indeed presented selectivity against MMP-13 [18]. Wyeth research developed a series of carboxylic acids-based inhibitors, which were potent and selective against MMP-13, with the carboxylate function connected to a benzofuran via a biphenyl sulphonamide spacer (Figure 23b) [16]. The presence of a bulky substituent in the benzofuran 4-position resulted in a compound 100-fold more selective for MMP-13 over MMP-2 [16].

**Figure 23.** (**a**) Inhibitor of Wyeth with carboxylate sulphonamide; (**b**) inhibitor of Wyeth with the carboxylate function connected to a bezofuran via biphenyl sulphonamide spacer.

Tanomastat (Bayer) inhibits MMP-2, -3, -9, and -13 but not MMP-1 [6,29]. This inhibitor participated in clinical trials, proving tolerable and no serious MSS, but the efficacy was negative in small-cell lung cancer because the median overall survival of patients treated did not increase [6,29].

Table 10 shows the IC50 and Ki values of some carboxylates-based inhibitors [6,15–19,29,35,37,54,55].

**10.**IC50andKivaluesofcarboxylates-basedinhibitors.

**Ki**: MMP-8

**Ki**

**Ki** (X **Ki** (X =

MM-3 = 8 nM

#### 4.2.3. Phosphorus-Based Inhibitors

The capacity of the phosphoric group to reproduce the gem-diol intermediate during peptide hydrolysis was explored with different structures to obtain potent MMPis [5]. The phosphorus-based of the peptide-analogous inhibitors can be phosphonates/phosphonic acids, phosphoramidates, phosphonamidates, and phosphinates/phosphinic peptides [27]. The phosphinic acid (PO(OH)-CH2) mimics the transition state obtained in substrate degradation, where each oxygen atom can coordinate both the catalytic zinc and the catalytic Glu [6,19,27]. The phosphinic acids are monodentate chelates [27]. In contrast to hydroxamate compounds, the phosphinic compounds interact with both the primed and unprimed side of the catalytic site [17,27,35] due to the placement of the ZBG in the middle of the scaffold and not at its *N*- or *C*-terminal, as in the cases of hydroxamate and carboxylate inhibitors [17]. Another advantage of phosphinic acids is the improved metabolic stability compared with hydroxamate acids [27].

The effectiveness of phosphoric acid inhibitors has been studied and it has been found that the three pockets unprimed are connected to obtain the maximum performance [19]. The S1-S2 pockets can be exploited using aromatic groups [19,68], that is why Reiter et al. prepared compounds with 4-benzyl (Figure 24) as a substituent to fill S2 pocket. They found that in the absence of this substituent or its replacement by small aliphatic or cyclohexyl methyl groups led to a loss of activity [19,68].

**Figure 24.** Compound prepared by Reiter et al., with a 4-benzyl substituent. The 4-benzyl group fills the S2 pocket and if the benzyl group was omitted or replaced by a small aliphatic or cyclohexyl methyl group, the activity is lost. The isobuthyl group fills the S1' pocket in a manner similar to other substrate-like inhibitors.

Matziari et al. [69] synthesized a series of phosphinic pseudopeptides bearing long P1' side chains, compounds that contain groups at the *ortho*-position of the phenyl ring and are selective for MMP-11 by the interaction of these groups with residues located at the entrance of the S1' cavity [69]. These results suggest that the development of compounds able to probe the entrance of the S1' cavity might represent an alternative strategy to gain selectivity [69].

Other phosphorus-based ZBGs are the carbamoyl phosphates, in which the two oxygens form a five membered ring with the zinc ion [18]. The negative charge of these inhibitors prevents their penetration into the cell and restrain them for extracellular space, contributing to low cytotoxicity [18]. Pochetti et al. [70] described a compound with high affinity to MMP-8 (Ki = 0.6 nM) but inhibits also MMP-2 (Ki = 5 nM) and MMP-3 (Ki = 40 nM) (Figure 25). The R enantiomer is more potent (1000 time more) than the S enantiomer (Ki = 0.7 μM) [70].

**Figure 25.** Pochetti's inhibitor.

The classical approach to synthesizing phosphinic compounds limits the full exploitation of this class of compounds for development of highly selective inhibitors of MMPS [35].

Table 11 shows the IC50 and Ki values of some phosphorus-based inhibitors [6,15–19,29,35,37,54,55].

#### 4.2.4. Nitrogen-Based Inhibitors

The nitrogen-based inhibitors have a binding preference to late transition metals and improved selectivity to zinc-dependent enzymes like MMPs [2]. The nitrogen-based inhibitors are studied by the Food and Drug Administration (FDA) and its metabolic availability and bioavailability are well described [2,18]. This ZBG type binds to Zn2<sup>+</sup> using the nitrogen atom and the carbonyl oxygen adjacent to nitrogen, which favors the formation of an enol because it is established by two hydrogen bonds [18].

The pyrimidine-2,4,6-trione inhibitors were published in 2001 by Hoffman-LaRoche. These compounds show relative specificity to gelatinases and potential usefulness as anticancer drugs [2]. The derivatization of position 5 of this compound promotes access to S1' and S2' pockets [18]. In development of osteoarthritis drugs, the pyrimidine-2,4,6-trione inhibitors have been optimized to inhibit MMP-13 [2].

#### 4.2.5. Heterocyclic Bidentate-Based Inhibitors

Heterocyclic bidentate ZBGs have better biostability and higher catalytic zinc ion binding capacity than hydroxamic acids, due to ligand rigidity [2]. Compared heterocycles bidentate and acetohydroxamic acid, the first are more potent to inhibit MMP-1, -2, and -3 and show low toxicity in cell viability assays [2].

Pyrones are biocompatible and they present good aqueous solubility [16]. The arylic portion was added to fit the MMP-3 hydrophobic S1' pocket, resulting in the compounds more potent than the corresponding hydroxamate-based inhibitors [16].

Table 12 shows the IC50 and Ki values of some heterocyclic bidentate-based inhibitors [6,15–19,29,35,37,54,55].

#### 4.2.6. Tetracyclines-Based Inhibitors

Tetracyclines are antibiotics that can chelate zinc and calcium ions and inhibit MMP activity [2,16,29]. Chemically modified tetracyclines (CMT) are preferred over conventional tetracyclines because they reach higher plasma levels for prolonged periods, consequently require less frequent administration, cause less gastrointestinal side effects, and have promising anti-proliferative and anti-metastatic activity [2,16,29]. The CMT binds to pro- or active MMPs, disrupt the native conformation of the protein, and leave the enzymes inactive [29]. In the search for new anticancer agents, the first series of CMT was obtained by removal of the dimethylamino group from the carbon-4 position, resulting in a compound without antimicrobial activity but with anticollagenolytic activity, in vitro and in vivo [16]. Preclinical studies demonstrated that CMT can inhibit gelatinases, stromelysins, collagenases, and MT-MMPs, by downregulating the expression of gelatinases, reducing the production of pro-enzymes and inhibiting the activation of pro-gelatinases and pro-collagenases [16,29].

Doxycycline (Figure 26a) is a semi-synthetic tetracycline that inhibits MMP-2, -9, -7, and -8 and is the only compound approved as an MMP inhibitor for the treatment of periodontitis [2,6]. The COL-3 (Figure 26b) showed specificity for MMP-2, -9, and -14, by decrease trypsinogen-2 and inducible nitric oxide (iNO) production, which are regulators of MMP activity [16]. Although COL-3 is currently being evaluated in clinical phase II trials, it showed poor solubility and stability [16].

#### 4.2.7. Mechanism-Based Inhibitors

The mechanism-based inhibitors coordinate with catalytic zinc ion, in a monodentate mode, allowing the nucleophilic attack by a conserved glutamic residue on the active site and forming a covalent bond [2,17]. This attack causes a conformational change in the catalytic site environment [17] preventing dissociation of the inhibitor and decreasing the rate of catalytic turnover and the amount of inhibitor needed to saturate the enzyme [2].

155

**Figure 26.** (**a**) Doxycycline; (**b**) COL-3.

Table 13 shows the IC50 values of some tetracyclines-based inhibitors [6,15–19,29,35,37,54,55,71].

In 2000, Mobashery et al. [72] were the first to report this novel type of MMPi that blocks gelatinases with a unique mechanistic mode [72]. The thiirane inhibitor showed a mechanism-based, slow-binding inhibition for MMP-2 and MMP-9 [72]. Bernardo et al. [73] also reported a slow-binding thiirane-containing inhibitor, (Figure 27), selective for MMP-2 and -9, where the sulfur group coordinates with the catalytic zinc ion, activates the thiirane group to interact with the active site glutamate, by nucleophilic attack causing a loss of activity [73]. These inhibitors are the first example of a suicide-inhibitor of MMPs [73].

**Figure 27.** Inhibitor developed by Bernardo el al. The sulfur group coordinates with the catalytic zinc ion and the activation of the thiirane group happened with interactions between the active site glutamate, by nucleophilic attack.

Thiirane-based ND-322 is a small molecule selective to MMP-2/MT1-MMP [2]. This inhibitor has been shown to reduce melanoma cell growth, migration, and invasion, and to delay metastatic dissemination [2].

SB-3CT is a selective inhibitor of MMP-2 and -9 [2]. The inhibition mechanism is similar to a "suicide inhibitor" in which a functional group is activated, leading to covalent modification of the active site [2]. SB-3CT also shows slow-binding kinetics with MMP-2, -3, and -9, contributing to slow dissociation of the MMP-inhibitor complex, but it is a reversible inhibitor which differentiates it from the truly irreversible suicide inhibitors [2]. O SB-3CT has potential benefits in brain damage caused by cerebral ischemia and has anti-cancer effects in T-cells lymphoma and prostate cancer models [2].

#### *4.3. Catalytic Domain (Non-Zinc Binding) Inhibitors*

The catalytic domain of MMPs contains other regions that can be exploited [17]. The first 3D-structure of the complex MMP-1 (catalytic domain)-synthetic inhibitor was reported in 1994 by Glaxo researchers [35]. Thereafter, other complexes have been studied and it was found that the S1' pockets have different depths among MMPS and this difference has been utilized in developing selective MMPis [28,35].

Stockman and Finel optimized two distinct series of MMP-3 inhibitors: PNU-141803 (amide, Figure 28a) and PNU-142372 (urea, Figure 28b) [19]. The connection between MMP-3 and PNU-142372 shows that the aromatic ring from the inhibitor extends to the S3 pocket (hydrophobic) and the thiadiazole sulfur group interacts with the catalytic zinc [19]. Moreover, the two nitrogen atoms form hydrogen bonds with Ala164 and Glu202 residues [19]. The alkylation of nitrogen atom or its replacement for carbon leads to the removal activity [19]. The replacement of a tyrosine for a serine within the S3 pocket (present in MMP-1) leads to the removal of inhibitory activity and explains the absence of activity against collagenases [19].

**Figure 28.** (**a**) PNU-141803; (**b**) PNU-142372.

Sanofi-Aventis developed a compound (Figure 29) for MMP-13 (IC50 = 6.6 μM), with very high selectivity [6]. This compound binds deeply to the S1' pocket and to a side pocket that has not been identified for other MMPs [6]. The pyridyl moiety is towards to the entrance of the S1' pocket, without interacting with the catalytic Zn(II) ion and the oxygen atoms neither from the amide (peptidic) bonds of the main chain (between Thr245 and Thr247) nor from hydroxyl group from the Thr247 side chain in the S1' pocket [6].

**Figure 29.** Sanofi-Aventis compound.

Many natural compounds have been shown to possess selective inhibition [28]. Wang et al. identified 19 potential MMPis from 4000 natural compounds isolated from medicinal plants [28]. The caffeates and flavonoids were found to be selective inhibitors against MMP-2 and -9, by occupying the S1' and S3 pockets [28].

The marine natural products are another pharmacological resource and include derivates from algae, sponges, and cartilages [28]. Some examples are Neovastat, Dieckol, and Ageladine A and they manifest anti-angiogenic, anti-proliferative, and anti-tumor effects [28].

Although the natural MMPis are more biocompatible and less toxic, they have disadvantages such as the effective dosages are in micromolar scale, which is thousands of times higher than synthetic inhibitors and are difficult to patent, making the pharmacological companies and investors reluctant to sponsor large-scale clinical trials [28].

#### *4.4. Allosteric and Exosite Inhibitors*

The catalytic zinc ion is common in all MMPs, therefore, if interactions of the substrates with this ion are minimized this would improve the inhibitor selectivity [2]. The hemopexin-like domain can move relatively to the catalytic domain and allosterically manipulate enzymatic activity by conformation deformation [28]. The allosteric drugs have a non-competitive inhibition mode [16,28], they bind and lock the MMP active site, forcing it to take less favorable conformation for substrate binding [2,16], avoiding off-target inhibition [28] and preventing the occurrence of side effects [28]. Exosite inhibitors are another alternative for selective MMPis since these inhibitors bind to alternative sites of MMPs [16,28].

Remacle et al. reported NSC405020, a small molecule that binds selectively to the hemopexin-like domain of MMP-14 [28]. This molecule inhibits the MMP-14 homodimerization and the interaction between the hemopexin-like domain and catalytic domain, preventing the type I collagen degradation [28].

Dufour et al. developed a peptide targeting the MMP-9 hemopexin-like domain, which blocks MMP-9 dimer formation and cell migration [28]. Scannevin et al. identified a highly selective compound (JNJ0966), which binds to the MMP-9 pro-peptide domain, inhibiting its activation, but not affecting the activity of MMP-1, -2, and -14 [28].

Xu et al. synthesized a peptide which inhibits the hydrolysis of type I and IV collagen by MMP-2, through binding to its collagen binding domain [28].

#### *4.5. Antibody-Based Inhibitors*

Antibodies are selective and have high affinity to MMP [27]. Clinical trials utilizing antibodies have provided evidence that selective MMP inhibitors do not induce MSS [27]. The therapeutic potential of anti-MMP antibodies has yet to be realized [27]. The antibodies may also undergo proteolysis, may be removed from circulation rapidly, and are costly [27].

Antibodies are large Y-sharped proteins which bind to an antigen via the fragment antigen-binding (Fab variable region) [28]. Monoclonal antibodies are highly specific, and they have affinity to MMPs [2,29]. The hemopexin domain can be a potential target for MMPs antibodies [2].

REGA-3G12 and REGA-2D9 are antibodies specific to MMP-9 [2,27,28] but not MMP-2 [28]. The inhibition mode of the REGA-3G12 involves the catalytic domain, the *N*-terminal region Trp116-Lys214 [28], and not the catalytic zinc ion or the fibronectin region [2]. The AB0041 and AB0046 are monoclonal anti-MMP-9 antibodies, which showed inhibition to tumor growth and metastasis in a model of colorectal carcinoma [27].

Andecaliximab (GS-5745), the humanized version of AD0041 [27], is a highly selective antibody and exerts allosteric control over tumor growth and metastasis in a colorectal carcinoma model (IC50 = 0.148 nM) [28]. This antibody is the only inhibitor that has undergone clinical trials [27,28] and it inhibits the pro-MMP-9 activation and inhibits non-competitively the MMP-9 activity [27].

DX-2400 is an antibody isolated from phage and it targets MMP-14 and the MMP-14-pro-MMP-2 complex, decreasing MMP activity [28]. DX-2400 inhibits the metastasis in a breast cancer xenograft mouse model [27]. However, the LAM-2/15 is the only selective inhibitor that inhibits MMP-14 catalytic activity, but not the pro-MMP-2 activation or MMP-14 dimerization [28]. The 9E8 is another antibody targeting MMP-14 which does not affect the catalytic activity of MMP-14 but inhibits the pro-MMP-2 activation [28].

Human scFv-Fc antibody E3 is bound to the catalytic domain of MMP-14 and inhibits type I collagen binding [27]. Human antibody Fab libraries were synthetized and the peptide G sequence (Phe-Ser-Ile-Ala-His-Glu) was incorporated resulting in Fab 1F8 antibody inhibitor, which inhibits the catalytic domain of MMP-14 [27].

Antibodies can also inhibit a specific activation of an MMP [27]. For example, the mAb 9E8 inhibits the MMP-14 activation of proMMP-2 but not the catalytic activities of MMP-14 [27]. The antibody LOOPAB also inhibits the MMP-14 activation of pro-MMP-2 but not collagenolysis activity of MMP-14 [27]. There are antibodies that reduce the MMP expression [71].

#### **5. Why Do MMP is Fail?**

MMI inhibitor side effects are predominantly related to off-target metal chelation [74]. The majority of MMPis used clinically are hydroxamic acid derivates with low selectivity [74] hence, they can inhibit other proteinases [28]. The most frequent side effects observed in MMPis clinical trials is the musculoskeletal syndrome (MSS) [17,28], which manifested as pain and immobility in the joints, arthralgias, contractures in the hands, and an overall reduced quality of life [14,15,74], leading to MMPis failing in the last phases of clinical trials [14]. Several studies indicate that the development of MSS is related to dose and time, with slightly different kinetic for the different MMPis, and the development of MSS is an indicator of successful MMPis [14,74]. The MMP inhibitors focused on chelating the catalytic

zinc ion have poor selectivity and resulted in MSS and gastrointestinal disorders [27]. However, the exact causes of MSS remains unknown [74], but can be related to a simultaneous inhibition of several MMPs [6,17,27].

Analysis of the expression of a target protein shows its presence at high levels when a disease is manifested or at low levels or absence in a healthy state [74]. However, these studies do not determine if a particular protein is directly associated with the disease process or if it is involved in ancillary event [74]. Studies of genetic manipulation in mouse as animal models determine the roles of MMPs in various pathological processes [74]. However, there are caveats in the use of animal models [74]:


#### **6. Conclusions**

Due to the side effects rising from the lack of selectivity and from the insufficient knowledge about the role of each MMP in the different pathological processes, none of the designed synthetic MMP inhibitors have yet passed the clinical trials and reached the market [6,27]. The poor performance of MMP inhibitors in clinical trials has globally been attributed to [27]:


In 1988, the first inhibitor was synthesized but after nearly 30 years, only one drug, Periostat®, doxycycline hydrate, had obtained approval from the FDA for the treatment of periodontal disease [6,16,17,27,28]. This inhibitor exhibited also therapeutic effects in treating aortic aneurysm, multiple sclerosis, as well as Type II diabetes [28].

**Funding:** This research was funded by A MOLECULAR VIEW OF DENTAL RESTORATION, grant number PTDC/SAU-BMA/122444/2010 and by MOLECULAR DESIGN FOR DENTAL RESTAURATION, grant number SAICT-POL/24288/2016. The work was supported by the Associate Laboratory for Green Chemistry-LAQV FCT/MCTES (UIDB/50006/2020), and Applied Molecular Biosciences Unit-UCIBIO FCT/MCTES (UID/Multi/04378/2019), which are financed by national funds.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **The Many Faces of Matrix Metalloproteinase-7 in Kidney Diseases**

#### **Zhao Liu 1, Roderick J. Tan <sup>2</sup> and Youhua Liu 1,3,\***


Received: 1 May 2020; Accepted: 22 June 2020; Published: 25 June 2020

**Abstract:** Matrix metalloproteinase-7 (MMP-7) is a secreted zinc-dependent endopeptidase that is implicated in regulating kidney homeostasis and diseases. MMP-7 is produced as an inactive zymogen, and proteolytic cleavage is required for its activation. MMP-7 is barely expressed in normal adult kidney but upregulated in acute kidney injury (AKI) and chronic kidney disease (CKD). The expression of MMP-7 is transcriptionally regulated by Wnt/β-catenin and other cues. As a secreted protein, MMP-7 is present and increased in the urine of patients, and its levels serve as a noninvasive biomarker for predicting AKI prognosis and monitoring CKD progression. Apart from degrading components of the extracellular matrix, MMP-7 also cleaves a wide range of substrates, such as E-cadherin, Fas ligand, and nephrin. As such, it plays an essential role in regulating many cellular processes, such as cell proliferation, apoptosis, epithelial-mesenchymal transition, and podocyte injury. The function of MMP-7 in kidney diseases is complex and context-dependent. It protects against AKI by priming tubular cells for survival and regeneration but promotes kidney fibrosis and CKD progression. MMP-7 also impairs podocyte integrity and induces proteinuria. In this review, we summarized recent advances in our understanding of the regulation, role, and mechanisms of MMP-7 in the pathogenesis of kidney diseases. We also discussed the potential of MMP-7 as a biomarker and therapeutic target in a clinical setting.

**Keywords:** matrix metalloproteinase-7; fibrosis; proteinuria; acute kidney injury; chronic kidney disease; apoptosis

#### **1. Introduction**

Matrix metalloproteinase-7 (MMP-7), also known as matrilysin-1, is one of the smallest secreted proteases of the MMP family, which consists of more than 20 structurally-related zinc-dependent endopeptidases with broad substrate specificity [1,2]. Collectively, MMPs degrade virtually all kinds of extracellular matrix (ECM) proteins and contribute to their turnover and remodeling. Furthermore, MMPs can also cleave many non-ECM substrates, making them a critical player in a wide variety of physiologic and pathologic processes, such as cell proliferation and apoptosis, endothelial cell function, inflammation, and tumor metastasis and invasion [3,4].

Extensive studies have shown that many MMPs are implicated in regulating kidney development and the pathogenesis of kidney diseases. Dysregulation of MMPs has been described in a wide variety of kidney disorders, including acute kidney injury (AKI), chronic kidney disease (CKD), diabetic kidney disease (DKD), glomerulonephritis, and inherited kidney diseases. Over the last two decades, due to the availability of MMP knockout models and specific pharmacological inhibitors, the role of MMPs, particularly MMP-2 and MMP-9, in regulating the development and progression of kidney diseases

have been extensively investigated. The detailed discussion of these MMPs in kidney pathology as well as in non-renal diseases is beyond the scope of the present paper, and the interested readers are referred to several comprehensive reviews on this topic [1,2,5]. Instead, the focus of this review is on MMP-7 in kidney diseases.

MMP-7 was first discovered in the rat uterus [6]. Structurally, MMP-7 comprises a pro-peptide domain and a catalytic domain. While it was formerly believed to cleave only ECM proteins, the evidence is emerging that MMP-7 also degrades a variety of non-ECM substrates, such as nephrin, E-cadherin, Fas ligand (FasL), pro-MMP-2, and pro-MMP-9 [7–10]. As such, MMP-7 is able to regulate a diverse array of biological processes, such as podocyte dysfunction, epithelial to mesenchymal transition (EMT), cell proliferation, and apoptosis [11,12]. It should be stressed that many actions of MMP-7 in the kidney are highly specific and unique. For instance, only MMP-7, but neither MMP-2 nor MMP-9, can cleave podocyte slit diaphragm protein nephrin and cause proteinuria [7]. In this context, it is conceivable that MMP-7 may be more important than other MMPs to the pathogenesis of kidney diseases.

In the past several years, significant progress has been made in our understanding of the biology of MMP-7 in kidney diseases. Numerous novel substrates of MMP-7 have been identified, enabling us to better comprehend the exact role of MMP-7 in the pathogenesis of kidney disorders. Mounting evidence suggests that urinary MMP-7 (uMMP-7) can be utilized as a noninvasive biomarker for predicting AKI prognosis and monitoring CKD progression in patients [13–15]. In this review, we discussed the expression, regulation, novel substrates, and mechanisms of MMP-7 in various kidney diseases.

#### **2. MMP-7 Structure, Activation, and Regulation**

The human *MMP-7* gene is located on chromosome 11 q22.3. The cDNA of *MMP-7* encodes a protein containing 267 amino acids. Structurally, MMP-7 only consists of a pro-peptide domain and a catalytic domain (Figure 1A), which separates it from most other MMPs that contain an additional hinge region and a hemopexin-like domain [1,2,16]. The crystal structure of MMP-7 containing the two domains aforementioned is shown in Figure 1B.

**Figure 1.** Structure of proMMP-7. (**A**) Full-length proMMP7 only consists of two domains: a pro-peptide domain (pro) and a catalytic domain (cat), which separates it from the prototype of MMPs. (**B**) The pro-peptide domain consists of three α-chains and connecting loops. The catalytic domain contains two zinc ions, two copper ions, and a ball-like structure consisting of three α-helices, five β-sheets, and multiple loops [16]. The image was prepared from Protein Data bank entries 2MZE (proMMP-7) using the PyMol (http://www.pymol.org). MMP, matrix metalloproteinase.

MMP-7 protein is produced and secreted as an inactive zymogen, which is maintained by a conserved cysteine residue that interacts with the zinc in the active site, rendering the protease inactive [16]. Disruption of this so-called cysteine switch is required for activation and can occur via proteolytic cleavage by many proteases, including trypsin, plasmin, or even other MMPs [17]. To generate a functional MMP-7 from the zymogen, the pro-peptide domain is proteolytically degraded in

a stepwise manner [18]. The latent form of MMP-7 is a 28 kDa protein. After removing an approximately 9 kDa sequence from the pro-peptide domain, the resultant 19 kDa peptide represents the active and functional endopeptidase. MMP-7 is also bound by two calcium ions, which plays an important role in stabilizing the secondary structure of the protein.

The activity of MMP-7 is regulated by a family of naturally occurring endogenous inhibitors known as tissue inhibitors of metalloproteinases (TIMPs). There are four known TIMPs; however, it remains elusive which TIMP has the greatest specificity for MMP-7. There are also potent inhibitors of MMP-7, such as MMP inhibitor II, that can reversibly block MMP-7 activity. However, the selectivity of these inhibitors for MMP-7 is uncertain, as they often inhibit, to a lesser degree, other MMPs as well. One of the challenges in the field is to develop potent and selective inhibitors that are specific for a given MMP.

The expression of MMP-7 is transcriptionally regulated by different cues, particularly the Wnt/β-catenin and transforming growth factor-β (TGF-β). The promoter of the human *MMP-7* gene contains a TATA box, an activator protein 1 (AP-1) site, and T cell factor (TCF)-binding elements. The AP-1 binding site is essential for mediating MMP-7 expression in response to growth factors, oncogenes, and phorbol ester, while the TCF-binding elements are responsible for mediating MMP-7 induction by Wnt/β-catenin. As TGF-β is known to activate β-catenin signaling [19], it remains elusive whether TGF-β controls MMP-7 expression directly or via β-catenin indirectly.

#### **3. MMP-7 Expression in the Kidney**

MMP-7 is commonly expressed in epithelial cells, including the liver, the ductal epithelium of exocrine glands in the skin, salivary glands, and pancreas, and the glandular epithelium of the intestine and reproductive organ and breast. Under normal physiologic conditions, adult kidney exhibits little MMP-7 expression [12,18,20]. Consistent with this notion, mice with global knockout of *MMP-7* are phenotypically normal, without any renal abnormality [12]. These data suggest that MMP-7 is dispensable for kidney structure and function in basal physiologic conditions.

The expression of MMP-7 is, however, induced in a wide variety of kidney diseases, including AKI, CKD, glomerular disease, inherited kidney disease, and renal cell carcinoma [11,21–24], suggesting that MMP-7 induction is a common feature of the kidney after various injuries. The expression and localization of MMP-7 protein in various kidney disorders are summarized in Table 1.

#### *3.1. Animal Models*

MMP-7 is markedly induced after AKI. In experimental animal models of AKI induced by ischemia-reperfusion injury (IRI), cisplatin, or folic acid, both mRNA and protein levels of MMP-7 are upregulated [11]. Following IRI, MMP-7 protein is mainly expressed and localized in the renal tubular epithelium, particularly in the S3 segment of proximal tubules, the epicenter of kidney injury in this model [11]. These results indicate a spatial correlation of MMP-7 induction with tubular injury and repair and regeneration. At this stage, the exact cues for triggering MMP-7 induction in AKI in vivo remain ambiguous, but it is most likely related to activation of Wnt/β-catenin signaling. In vitro studies show that human kidney proximal tubular cells (HKC-8) increase MMP-7 expression upon stimulation of Wnt/β-catenin, whereas inhibition of Wnt/β-catenin by small molecule inhibitor ICG-001 negates MMP-7 induction [21].

Induction of MMP-7 is a common finding in a wide variety of CKD characterized by renal fibrosis. MMP-7 expression is increased in renal tubular epithelia in the mouse model of unilateral ureteral obstruction (UUO) [21]. Similar results are obtained in adriamycin nephropathy, a model of focal and segmental glomerulosclerosis (FSGS). Immunohistochemical staining reveals that MMP-7 expression and Wnt/β-catenin activation are closely correlated in both UUO and adriamycin nephropathy [21]. This result is compatible with the findings in tumors that MMP-7 transcript overlaps with the accumulation of β-catenin protein [25,26]. Therefore, it is conceivable that MMP-7 expression in CKD is causatively linked to the activation of Wnt/β-catenin signaling.


**Table 1.** Expression of matrix metalloproteinase-7 (MMP-7) in kidney diseases.

<sup>1</sup> Acute kidney injury. <sup>2</sup> Unilateral ureteral obstruction. <sup>3</sup> Focal segmental glomerulosclerosis.

#### *3.2. Human Kidney Biopsies*

Consistent with animal studies, the induction of MMP-7 expression is also evident in the kidney from patients with various renal disorders. For instance, in the kidney biopsies of patients with an autosomal dominant polycystic kidney disease, MMP-7 staining is observed in tubular epithelial lining cysts and atrophic tubules [23]. In the specimens of patients with hydronephrosis, MMP-7 is detected in dilated and atrophic renal tubules [31].

Kidney tissues from patients with IgA nephropathy (IgAN) show positive MMP-7 staining in renal tubular epithelia, fluids in the tubular lumen, and glomerular podocytes [12,21]. The RNA sequencing data from IgAN patients also show a significant increase in the mRNA expression of MMP-7 in the kidneys [32]. MMP-7 protein is increased in both tubular epithelial cells and the tubulointerstitial compartment of human diabetic kidneys [12,33]. Similarly, renal biopsy specimen analysis reveals MMP-7 induction in lupus nephritis and chronic allograft nephropathy [27,29]. Induction of MMP-7 expression has also been observed in the kidneys of patients with FSGS [12,21].

#### *3.3. Mechanism of MMP-7 Regulation In Vivo*

As to the cues responsible for MMP-7 induction in vivo, many studies have pointed to Wnt/β-catenin signaling, which is activated in virtually every kind of nephropathy [21,34,35]. When Wnt ligands bind to their receptors, β-catenin is stabilized and subsequently translocates into the nucleus for binding to the TCF/lymphoid enhancer-binding factor to regulate the transcription of its target genes, including MMP-7 [34,36,37]. Bioinformatics analysis reveals the presence of putative TCF-binding sites in the promoter region of the *MMP-7* gene [21]. Chromatin immunoprecipitation confirms

that β-catenin activation promotes the binding of TCF to the *MMP-7* gene promoter, resulting in the expression of MMP-7 in kidney tubular epithelial cells [21].

There is also a correlation between MMP-7 and TGF-β in vivo, suggesting that TGF-β may play a role in mediating MMP-7 expression. Both TGF-β1 and MMP-7 are upregulated in streptozotocin-induced diabetic nephropathy rats. Sirtuin 1 (Sirt1) deacetylates Smad4 and inhibits the expression of MMP-7, indicating that the over-activation of TGF-β is related to the excessive acetylation of Smad4, which, in turn, causes MMP-7 induction [38,39]. Along these lines, resveratrol increases the expression of Sirt1, which inhibits MMP-7 and ultimately alleviates renal injury and fibrosis. Furthermore, TGF-β may indirectly augment MMP-7 by activating canonicalWnt signaling [19]. TGF-β also inhibits Dickkopf-1, an antagonist of Wnt, and potentiates the activity of β-catenin, thereby activating Wnt/β-catenin signaling [40,41]. Of interest, a study shows that MMP-7 can further induce TGF-β production via an MMP-7/Syndecan-1/TGF-β autocrine loop [42,43].

#### **4. MMP-7 As a Biomarker for Kidney Diseases**

Early identification and diagnosis are of importance in slowing the progression of kidney disease and preventing its complications [44]. Serum creatinine and blood urea nitrogen (BUN), two widely used markers for the diagnosis of kidney failure, increase only in the advanced stage of nephropathy. Consequently, kidney diseases are usually diagnosed at a later stage, and the implementation of therapeutic interventions is usually delayed. Therefore, there is an urgent need to develop novel biomarkers for early detection and prognostic assessment of kidney disorders [45]. MMP-7 is upregulated in various kidney diseases, and its protein is predominantly distributed in the apical region of tubular epithelial cells and is detected in the fluids present in the tubular lumen [12,21], suggesting that this protein could be secreted to the urine. Therefore, the level of uMMP-7 can be used as a potential non-invasive biomarker of kidney disease.

#### *4.1. uMMP-7 Predicts the Risk of AKI*

AKI is a relatively common disorder among hospitalized patients, occurring in more than 20% of all hospitalizations in large academic hospitals [46]. AKI causes approximately 2 million deaths each year worldwide. Current diagnostic criteria for AKI require changes in serum creatinine and urine output, which are not sensitive and represent delayed markers. Estimated glomerular filtration rate (eGFR) must decline by approximately 50% before any changes in serum creatinine can be detected [47,48]. Therefore, a sensitive and robust biomarker to predict the risk of AKI and its associated outcomes in patients is urgently needed. Over the past decades, there are tremendous efforts in the nephrology community to discover and validate potential biomarkers for AKI. Several biomarkers, such as kidney injury molecule-1 (Kim-1), neutrophil gelatinase-associated lipocalin (NGAL), and TIMP-2/insulin-like growth factor-binding protein-7 (IGFBP7), have been identified. Although some of them are applied to the clinic, they have various limitations.

There are two prospective cohort studies showing uMMP-7 as a valuable predictor of severe AKI after cardiac surgery [13,49]. In a prospective, multicenter, two-stage cohort study with 721 patients undergoing cardiac surgery, compared with the lowest quartile, a postoperative uMMP-7 level of 22.6 μg/g creatinine in children represents a >36-fold risk of severe AKI, whereas a postoperative uMMP-7 level of >15.2 μg/g creatinine in adults indicates a 17-fold risk of severe AKI [13]. In terms of predicting prognosis, higher uMMP-7 levels shortly after cardiac surgery are associated with an increased risk of acute dialysis or in-hospital deaths among children and adults, as well as longer duration of intensive care unit and hospital stays [13]. uMMP-7 outperforms other biomarkers, including urinary interleukin-18, NGAL, urinary angiotensinogen, albumin-to-creatinine ratio, and TIMP2/IGFBP7 [50–53], and the area under the receiver operating characteristic curve (AUC) of uMMP-7 is the largest for predicting severe AKI [13]. Of particular interest, uMMP-7 level peaks at 4 h after surgery, whereas the rise in serum creatinine occurs after 24 h in patients [13]. This is of clinical significance because it allows much earlier identification of the patients who have a high risk of

developing AKI than serum creatinine. Therefore, uMMP-7 could be a valuable and robust biomarker for predicting the AKI after cardiac surgery. More studies are needed to replicate the results in larger and more diverse populations and in different settings of AKI.

#### *4.2. uMMP-7 As a Biomarker of CKD Progression*

Kidney fibrosis is a common outcome of virtually all CKDs [54,55], which is characterized by excessive accumulation and deposition of ECM, leading to tissue scarring. Measuring the extent of renal fibrosis is essential for determining the prognosis of renal outcomes, monitoring CKD progression, and evaluating the therapeutic efficacy of new treatments [55]. Currently, renal fibrosis is assessed only via percutaneous renal biopsy [55,56]. It is necessary to find non-invasive surrogate biomarkers for evaluating the development and progression of kidney fibrosis. Because activation of Wnt/β-catenin is a common feature of fibrotic CKD [57–60], one would speculate that the activity of renal Wnt/β-catenin signaling parallels the severity of renal fibrosis. Because the uMMP-7 level reflects the activity of renal Wnt/β-catenin [21], it is then reasonable to measure uMMP-7 to estimate the extent of kidney fibrosis.

In patients with CKD, uMMP-7 levels are found to positively correlate with renal fibrosis scores and have an inverse association with the renal function [12]. Therefore, uMMP-7 levels may serve as a noninvasive biomarker for kidney fibrosis and a predictor for CKD outcomes, as well as monitor the dynamic of fibrosis progression. Consistent with the findings in the kidney, MMP-7 also affects liver and lung fibrosis after a chronic injury. MMP-7 is upregulated in biliary atresia-associated liver fibrosis, and its expression is considered the best strategy to distinguish between cirrhosis and pre-cirrhosis stages [61,62]. Elevated MMP-7 levels are also detected in the peripheral blood of patients with human idiopathic pulmonary fibrosis (IPF) and may be used as a biomarker for predicting disease progression and death [63–66]. Serum MMP-7 levels are also increased in IPF patients with severe obstructive sleep apnea [67]. It should be pointed out that uMMP-7 is much more robust than serum MMP-7 in predicting kidney injury, as it is mainly produced from the injured tubular epithelium and expressed apically in CKD. In this context, uMMP-7 is particularly suitable for assessing the severity of fibrosis in the kidney, compared to other organs.

The feasibility and validity of uMMP-7 as a biomarker to predict CKD progression are recently supported by a prospective cohort study for uMMP-7 to serve as a predictor for IgAN progression [15]. The course of IgAN is highly variable and heterogeneous. The clinical manifestations range from benign asymptomatic microscopic hematuria to severe hypertension and progressive CKD, and renal pathological appearances range from normal to different degrees of mesangial cell proliferation [68]. At present, the final diagnosis of IgAN mainly relies on renal biopsy [68]. On the basis of a histological assessment, variables including mesangial hypercellularity (M), endocapillary hypercellularity (E), segmental glomerulosclerosis (S) and tubular atrophy and interstitial fibrosis (T) give rise to MEST score and provide valuable information to prognostication of IgAN [69]. The lack of sensitive and specific surrogate indicators for long-term outcomes makes it challenging to improve treatment options [70]. A recent prospective observational cohort study shows that the level of uMMP-7 may serve as an independent and powerful predictor for IgAN progression, even for those patients who are still in the early stages of IgAN, as defined by an eGFR of <sup>≥</sup> 60 mL/min/1.73 m<sup>2</sup> [15]. High levels (<sup>&</sup>gt; 3.9 <sup>μ</sup>g/g of creatinine) of uMMP-7 increase the risk of IgAN progression by 2.7 times in an adjusted analysis [15]. Several earlier studies have suggested numerous possible biomarkers in serum or urine for predicting IgAN progressions, such as galactose-deficient IgA1, auto-antibodies against Gd-IgA1, fibroblast growth factor 23, angiotensinogen, epidermal growth factor, and Kim-1 [71–77]. Measurement of uMMP-7 level outperforms each of these biomarkers in risk prediction and improves the risk predictive power of a MEST score [15]. Some retrospective follow-up studies also show that an increase in serum MMP-7 levels is associated with a high risk of poor renal outcome and renal fibrosis [74]. In addition, uMMP-7 is also identified to serve as an independent predictor of tissue remodeling and renal interstitial fibrosis in children with CKD [78]. Another study shows that among people with type 2 diabetes and proteinuric DKD, uMMP-7 concentration is strongly associated with disease progression

and subsequent mortality [14]. Taken together, these findings indicate that uMMP-7 may hold promise as a noninvasive biomarker for kidney fibrosis and CKD progression.

#### **5. Roles of MMP-7 in Kidney Diseases**

Thanks to the availability of MMP-7 knockout mice, we now have a better understanding of the role of MMP-7 in the pathogenesis of kidney diseases. Although the specific action of MMP-7 in different nephropathy models has not been fully clarified, numerous studies have discovered discrete roles of MMP-7 in renal pathophysiology in different settings. Furthermore, the role of MMP-7 may also evolve with different disease types or the time course of kidney disorders. For instance, MMP-7 may be reparative in the early stages of AKI, whereas it may be detrimental as the disease progresses. Table 2 summarizes the roles of MMP-7 in various kidney diseases.


**Table 2.** Roles of MMP-7 in kidney diseases.

<sup>1</sup> Acute kidney injury. <sup>2</sup> Unilateral ureteral obstruction. <sup>3</sup> Chronic kidney disease.

#### *5.1. MMP-7 Protects Against AKI*

MMP-7 is induced specifically in renal tubular epithelium after AKI, a condition characterized by the death of tubular cells and the infiltration of inflammatory cells to the kidney [80–82]. In AKI patients after cardiac surgery, uMMP-7 levels increase rapidly and peak at 4 h, suggesting MMP-7 induction is an early event [13]. Using MMP-7-/- null mice, a recent study shows that MMP-7 has a renal protective effect against AKI as it alleviates injury by reducing the number of dead tubular cells, promoting the proliferation of tubular cells, and suppressing inflammation [11]. Compared with wild-type controls, MMP-7-/- mice display higher mortality, elevated serum creatinine, and more severe histologic lesions after IRI or cisplatin. These changes are accompanied by more prominent tubular cell death and interstitial inflammation in MMP-7-/- kidneys. In a rescue experiment, injection of exogenous MMP-7 protects against kidney injury in MMP-7-/- mice after IRI, confirming a renal protective action of MMP-7. Of note, MMP-7 may only play a protective role in the early stages of AKI because the long-term activation of MMP-7 leads to kidney fibrosis [12].

#### *5.2. MMP-7 Promotes Kidney Fibrosis and CKD Progression*

CKD is characterized by excessive ECM accumulation and interstitial fibroblast activation [54,83– 85]. Given its proteolytic potential to degrade ECM proteins, MMP-7 was originally thought to be

anti-fibrotic. Surprisingly, it is found that MMP-7 plays a critical role in the development of fibrotic CKD [12]. In mice, the genetic ablation of MMP-7 reduces the fibrotic lesions and ECM accumulation induced by UUO. Knockout of *MMP-7* also preserves E-cadherin protein and inhibits the de novo expression of vimentin in renal tubules of obstructed kidneys, suggesting that MMP-7 may promote EMT in vivo. Although the role of EMT in renal fibrosis remains controversial, partial EMT is an indispensable part of renal fibrosis [57,86]. Along these lines, it appears that MMP-7 plays an important role in the onset and progression of renal fibrosis by impairing the integrity of renal tubular epithelium and causing partial EMT [12]. Loss of E-cadherin mediated by MMP-7 leads to β-catenin liberation and nuclear translocation, which facilitates CKD progression [57,87]. Of note, MMP-7-mediated β-catenin liberation further induces MMP-7 expression, creating a vicious cycle [12]. Furthermore, TGF-β is known to induce MMP-7 expression, consistent with their role in promoting EMT and kidney fibrosis [38,39]. More studies are needed to confirm the detrimental role of MMP-7 in other models of CKD.

#### *5.3. MMP-7 Induces Podocyte Dysfunction and Proteinuria*

In glomerular diseases, MMP-7 is specifically induced in glomerular podocytes of the diseased kidney, implying its potential role in regulating podocyte injury and proteinuria. Indeed, a recent study shows that the infusion of MMP-7 protein or injection of *MMP-7* expression vector induces transient proteinuria in normal mice [7], suggesting that MMP-7 can trigger podocyte injury and impair the glomerular filtration barrier. Furthermore, MMP-7-/- mice are protected against proteinuria and glomerular injury induced by either angiotensin II or adriamycin [7]. Consistent with this finding, incubation of isolated glomeruli with MMP-7 ex vivo increases glomerular permeability and causes foot process effacement. These observations illustrate that MMP-7 impairs glomerular filtration and causes proteinuria in vivo.

CKD progression is characterized by increasingly widespread lesions in different compartments of the kidney parenchyma. There is ample evidence to suggest that the progression of the primary glomerular disease can trigger tubular injury and interstitial fibrosis, but whether tubular injury affects the function of the glomerulus remains ambiguous [88]. Recently, using conditional knockout mice with tubule-specific ablation of β-catenin, we discovered that tubule-derived MMP-7 is a pathological mediator of glomerular damage [7]. MMP-7 is secreted as a soluble protein from the tubules to the glomeruli and mediates the impairment of slit diaphragm integrity, leading to podocyte dysfunction and increased proteinuria [7]. This suggests that MMP-7 is the key mediator of tubular-to-glomerular crosstalk that promotes proteinuria and CKD progression.

#### **6. Mechanisms and Novel Targets of MMP-7 in Kidney Diseases**

The diverse actions of MMP-7 in kidney diseases are presumably mediated by its ability to cleave different substrates. As the primary role of activated MMP-7 is to break down ECM components, MMP-7 is well known to degrade macromolecules, including casein, type I, II, IV, and V gelatins, fibronectin, and proteoglycan [2,89]. Recent findings reveal, however, that MMP-7 is also capable of degrading a multitude of non-ECM substrates, such as FasL, E-cadherin, nephrin, and proMMP-2 and -9 [8–10,12]. These novel actions of MMP-7 play a crucial role in mediating its diverse functions in the pathogenesis of kidney disorders. Figure 2 illustrates the mechanisms of MMP-7 action in kidney disease.

**Figure 2.** The mechanisms of MMP-7 action in kidney disease. (**A**) In renal tubular epithelial cells, MMP-7 promotes cell proliferation and reduces cell death by degrading E-cadherin and FasL, respectively, and finally plays a role in protecting the kidney during acute kidney injury (AKI). (**B**) However, MMP-7 also degrades E-cadherin and activates proMMP-2 and -9, leading to renal fibrosis. (**C**) In podocytes, MMP-7-mediated degradation of nephrin impairs the integrity of the slit diaphragm, which subsequently causes an increase in proteinuria and eventually leads to renal fibrosis.

#### *6.1. FasL*

There are two signaling pathways leading to apoptosis: the intrinsic pathway and extrinsic pathway [90–92]. The FasL plays a central role in the death receptor-dependent extrinsic apoptosis pathway [93]. Dependent upon different pathological conditions and different cells, MMP-7 shows a bi-directional effect on the induction or degradation of FasL, which plays distinct roles in regulating cell survival/death in different settings. MMP-7 degrades FasL to decrease the apoptosis of renal tubular cells through FasL/Fas-associated death domain (FADD)/caspase-7 activation, which is one of the mechanisms underlying MMP-7 protection of kidney tubular cells against death in the early stage of AKI [11] (Figure 2). Consistently, MMP-7-induced cleavage of the membrane-bound FasL also plays a role in the process of pulmonary fibrosis [94].

Studies also show that MMP-7 affects the fate of renal interstitial fibroblasts in the opposite way. MMP-7 is shown to induce the expression of FasL in cultured fibroblasts, which promotes fibroblast apoptosis and facilitates its resolution after kidney repair following AKI [95]. This observation is supported by studies conducted on cancer cells, which also shows that MMP-7 induces the expression of FasL and subsequently activates the extrinsic apoptotic pathway [10,96]. Furthermore, MMP-7 enhances the staurosporine-mediated intrinsic apoptosis pathway [97,98], leading to fibroblast apoptosis [95]. In this regard, MMP-7-induced fibroblast apoptosis requires the synergistic action of both the intrinsic

and extrinsic pathways. At this stage, the mechanism by which MMP-7 induces FasL expression remains to be determined. It is likely that MMP-7 cleaves and activates an unidentified intermediate molecule, which, in turn, induces FasL expression.

#### *6.2. E-Cadherin*

E-cadherin, a classical member of the cadherin superfamily, is an epithelial marker that plays an important role in maintaining the integrity of tubular epithelial cells and cell–cell adhesion [99]. It is a calcium-dependent cell–cell adhesion receptor composed of five extracellular cadherin repeats, a transmembrane region, and a highly conserved cytoplasmic tail. MMP-7 is capable of degrading E-cadherin via ectodomain shedding. It has been shown that MMP-7 affects both CKD and AKI by cleaving E-cadherin [8,100]. Loss of E-cadherin impairs the integrity of tubular epithelial cells, which is recognized as the initial step of partial EMT, a process that is essential for tubular atrophy and kidney fibrosis [101,102]. Because E-cadherin and β-catenin associate with each other in a cellular adhesion complex [99,103], the cleavage of E-cadherin would result in the dissociation of the E-cadherin/β-catenin complex, leading to β-catenin liberation (Figure 2). In essence, MMP-7 can induce kidney fibrosis by activating β-catenin in the absence of Wnt ligands [12]. Notably, such β-catenin release caused by MMP-7-mediated degradation of E-cadherin is also found in mouse prostate cancer cells [104].

The liberation of β-catenin caused by MMP-7-mediated degradation of E-cadherin provides a rational explanation for why MMP-7 promotes renal fibrosis. The exact same pathway, however, is beneficial and plays a protective role against AKI (Figure 2). E-cadherin maintains tight cell–cell adhesion among tubular epithelia [105]. The ability of MMP-7 to degrade E-cadherin disrupts epithelial cell contact inhibition [106] and leads to cell proliferation [107–109]. E-cadherin degradation also leads to β-catenin liberation and downstream signaling. In vitro experiments have confirmed that proliferation-related proteins, such as proliferating cell nuclear antigen (PCNA) and c-Fos, are increased in isolated renal tubules cultured with mitogen-rich serum and MMP-7, compared with serum alone [11]. In short, by cleaving E-cadherin and releasing β-catenin, MMP-7 creates an environment conducive to cell proliferation and protects renal tubules against AKI [11]. However, loss of cell contact inhibition can also trigger EMT and lead to a profibrotic phenotype after the acute injury has resolved. Therefore, MMP-7 induction could be beneficial in the early phase of kidney injury but not in the late stage of chronic kidney disorders. The balance of beneficial versus detrimental effects may rely upon the time point at which MMP-7 activation occurs.

#### *6.3. Nephrin*

Nephrin is a key component of the slit diaphragm, which connects adjacent podocyte foot processes and plays a fundamental role in glomerular filtration [110]. We recently demonstrated that nephrin is a specific and direct substrate of MMP-7 [7] (Figure 2). MMP-7 is shown to degrade nephrin in cultured glomeruli, cultured cells, and cell-free systems, which is dependent on its proteolytic activity. Such action of MMP-7 on nephrin degradation is rapid, starting as early as 5 min after incubation. Furthermore, the action of MMP-7 on nephrin appears direct, cleaving purified recombinant nephrin protein in a cell-free system. It should be stressed that the action of MMP-7 is specific, as other MMPs, such as MMP-2 and MMP-9, are unable to degrade nephrin in the same conditions. These findings not only identify nephrin as a novel substrate of MMP-7 but also offer a mechanistic insight into the role of MMP-7 in the development of proteinuria and glomerular lesions [7]. Therefore, MMP-7-mediated cleavage of nephrin impairs slit diaphragm integrity to promote podocyte injury and proteinuria.

#### *6.4. Pro-MMP-2 and -9*

MMP-7 also promotes renal fibrosis by proteolytically activating MMP-2 and MMP-9 from their latent zymogen forms. In an in vitro experiment, it has been shown that MMP-7 activates pro-MMP-2 by propeptide removal [111]. Earlier studies in tumor cells also show that MMP-7 is a potential activator of pro-MMP-2 [112], through competitively binding to the TIMP-2 from MMP-2/TIMP-2 complexes, leading to the release of functional MMP-2 [113,114]. Because MMP-2 can convert pro-MMP-9 into MMP-9, it is conceivable that MMP-7 can activate MMP-9 by generating active MMP-2 in both direct and indirect ways [113–115]. At present, some studies show that MMP-2 and MMP-9 play an important role in the onset and progression of CKD via degrading type IV collagen, promoting partial EMT and mediating a complex interaction with various cytokines [116,117]. Therefore, MMP-7 could promote kidney fibrosis and CKD progression by activating MMP-2 and MMP-9.

#### **7. Conclusion and Perspectives**

Over the last several years, our understanding of MMP-7 in the pathogenesis of kidney disease has dramatically improved. MMP-7 is upregulated in AKI, CKD, and glomerular diseases and is predominantly localized in renal tubular epithelia. Recent findings on novel substrates of MMP-7, such as nephrin, have shed new light on its role and mechanisms of action in a wide variety of kidney diseases. By cleaving E-cadherin and FasL, MMP-7 protects the kidney from AKI by priming renal tubular cells for survival and proliferation. The same MMP-7-mediated degradation of E-cadherin in CKD setting, however, promotes tubular EMT and activates β-catenin in a Wnt-independent fashion, leading to kidney fibrosis. MMP-7 can directly degrade nephrin, resulting in podocyte dysfunction and proteinuria. Several clinical cohort studies suggest that uMMP-7 levels may be used as a noninvasive biomarker for predicting AKI prognosis and monitoring CKD progression in patients.

Although the present studies have provided invaluable insights into the biological functions of MMP-7 in kidney diseases, how to translate this knowledge into patient care is a daunting task. The revelation of MMP-7 upregulation provides hopes for the application of uMMP-7 as a noninvasive biomarker, and future clinical studies for validation in large and diverse populations of patients are warranted. Developing a therapy to target MMP-7 will rely upon highly selective inhibitors for MMP-7, which are not yet available. Future studies may focus on developing efficient strategies to inhibit MMP-7 expression in vivo for modulating the course of kidney disease. We hope that continuing this line of investigation will improve our understanding of the role of MMP-7 and its mechanism of action in the pathogenesis of various kidney disorders, and eventually translate into improved patient care.

**Funding:** This research was funded by the National Natural Science Foundation of China (grant 81521003 and 81920108007) and the National Institute of Health grant DK064005.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Biomolecules* Editorial Office E-mail: biomolecules@mdpi.com www.mdpi.com/journal/biomolecules

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18

www.mdpi.com