Recent insights into natural product inhibitors of matrix metalloproteinases

Members of the matrix metalloproteinase (MMP) family have biological functions that are central to human health and disease, and MMP inhibitors have been investigated for the treatment of cardiovascular disease, cancer and neurodegenerative disorders.

Abstract

Members of the matrix metalloproteinase (MMP) family have biological functions that are central to human health and disease, and MMP inhibitors have been investigated for the treatment of cardiovascular disease, cancer and neurodegenerative disorders. The outcomes of initial clinical trials with the first generation of MMP inhibitors proved disappointing. However, our growing understanding of the complexities of the MMP function in disease, and an increased understanding of MMP protein architecture and control of activity now provide new opportunities and avenues to develop MMP-focused therapies. Natural products that affect MMP activities have been of strong interest as templates for drug discovery, and for their use as chemical tools to help delineate the roles of MMPs that still remain to be defined. Herein, we highlight the most recent discoveries of structurally diverse natural product inhibitors to these proteases.

Introduction

Matrix metalloproteinases (MMPs) are a family of zinc-dependent endopeptidases involved in the proteolysis of a diverse series of protein targets, and whose activities lead to a myriad of biological functions that appear central to human health and disease. As such, MMPs have been the center of attention of research efforts for the last couple of decades. Many of the major diseases that afflict us have implicated pathological MMP functions as being key to the disease condition, and this includes disorders causing the highest rates of morbidity such as cardiovascular disease, cancer, inflammation, autoimmune disease and potentially several distinct neurodegenerative disease states1–12 (Fig. 1). Despite the considerable investment of research efforts and resources, the outcomes of initial clinical trials with the first generation of MMP inhibitors (MMPis) proved disappointing, with unwanted side effects and potential lack of efficacies being noted. Thus, new approaches to target MMPs are of great interest, and may involve structure-based drug discovery approaches, therapeutic antibodies (Abs) and potentially drugs developed from natural product precursors. Natural products that affect MMP activities have been of strong interest as templates for drug discovery, as they can provide novel and structurally diverse chemistries, as well as having uses as chemical tools to help further uncover MMPs functions. Thus, in this review, we focus on the most recent discoveries into natural products that function as inhibitors to these proteases, and we briefly highlight the current understanding of the biology of MMPs.

Fig. 1. Roles of MMPs in human health and disease.

Domain-based MMP structure

MMPs belong to the larger metzincin superfamily, they play a significant role in degradation of various protein targets, and they were initially identified by their roles within the extracellular matrix (ECM).13–16 MMPs have now been indicated to have diverse functions in remodeling of tissues and in embryonic development that includes having roles in the brain, as wells as in promoting the progression of various disease pathologies. MMP-1 was the first to be characterized, and it was noted to have proteolytic activity against collagen that results in ECM protein degradation.17 The MMP family has grown to 28 proteins in vertebrates, with 23 MMPs being present in humans and 15 of these are expressed in the vasculature.18,19 In addition to multiple ECM targets, MMPs have been defined as also being present within cells, occurring in the cytoskeletal intracellular matrix, and cellular compartments that include the mitochondria and nucleus, and they can proteolyze a broad selection of intracellular targets.20–23 The human family members are typically multi-domain proteins, and most share sequence homology to MMP-1 collagenase. MMPs are expressed as latent propeptides, containing a signal peptide in the N-terminus that is removed during secretory pathway steps, a N-terminal pro-domain that typically keeps the enzyme inactive, a metalloproteinase catalytic core domain containing a catalytic zinc ion, a hinge region and a C-terminal hemopexin (PEX) domain that has roles in localization, substrate specificity,24,25 and where in MMP-2 and MMP-9 this PEX domain has been shown to possess anti-apoptotic activity.26 MMP-2 and MMP-9 also contain three fibronectin type II repeats that aid in substrate recognition. MMP-7 and MMP-26 have the simplest domain structure in that they lack the hinge and C-terminal PEX domain. MMP-23 is the most distinct family member, where it also lacks these C-terminal components and instead contains an immunoglobulin-like domain after the catalytic core that has been implicated to have functions analogous to the PEX domain, and a C-terminal cysteine-rich domain that may have functions in altering ion channel activities.27

To keep the MMP in its inactive, proMMP zymogen state, a conserved cysteine switch motif ‘PRCGXPD’ in the pro-domain is used, in which the cysteine side chain of this motif chelates the catalytic core zinc ion. MMP activation occurs through the activity of other enzymes, or by free radicals, resulting in either removal or blocking of the zinc-chelating cysteine thiol group. The catalytic domain has a high degree of conservation among the MMPs, and it contains a conserved sequence motif of ‘HEXXHXXGXXH’, where the three histidines of this sequence bind to the active site Zn²⁺ ion (Fig. 2a). A conserved-glutamate residue also coordinates the active site metal ion, and is required for the catalytic mechanism of water-mediated nucleophilic attack on the substrate peptide bond.28,29 A conserved methionine is additionally present, and it occurs shortly after the conserved histidine-containing sequence motif. This methionine residue forms the hydrophobic base to the active site that is situated in a shallow cleft used to bind extended peptide substrate conformations (Fig. 2a). Sub-sites in this catalytic domain help determine substrate specificity, and this includes the S12 site that forms a hydrophobic pocket that varies in its depth among the MMPs (Fig. 2b). Exo-sites situated further from the active site also play a role in specificity, and these may include regions of the adjoining PEX domain. The differences in substrate specificities, together with overall domain organizations, have been used a means to classify the MMPs into four major groups, the gelatinases, matrilysins, furin-activated MMPs and the archetypal MMPs that consist of both stromelysins and collagenases. The furin-activated MMPs are classified into further subgroups, type-1 transmembrane MMPs, type-II transmembrane MMPs, GPI-anchored MMPs and the secreted MMPs. The secreted MMPs may still localize to the cell membrane, through interactions with CD44 or integrins.30–33

Fig. 2. Structural features of the catalytic domain of MMP-2. a) The crystal structure of MMP-2 catalytic domain, PDB code 1QIB.pdb, revealed that three histidines bind to an active site Zn²⁺ ion colored in dark grey. A conserved-glutamate residue is also present that is required for the catalytic mechanism of water-mediated nucleophilic attack on the substrate peptide bond. A conserved methionine residue forms the hydrophobic base of the active site. Protein surface is depicted in light grey. b) The size of the S12 site pocket of MMPs can help provide substrate specificity. MMP-2 has a deep, extended S12 pocket, as highlighted by the docking of anacardic acid, cyan backbone, into the protein structure that results in the extended aliphatic chain of this natural product binding into the S12 pocket. Major cavities of the MMP-2 catalytic domain structure, including S12, are highlighted in light grey.

Controls of MMP activity

The ECM consists of fibers, of which collagen is the main constituent, elastin provides required flexibility to tissues, and fibronectin binds cells to the ECM. MMP activity can target these three ECM components, in addition to nearly all other ECM structural components including laminin, vitronectin, aggrecan, enactin, tenascin and proteoglycans, along with cell surface receptors, cytokines and growth factors.34–41 In addition, intracellular MMP targets include as chaperones, cytoskeletal proteins, targets involved in protein biosynthesis and carbohydrate metabolism, signal transduction, targets involved in protein degradation through ubiquitination or lysosomal pathways, regulators of apoptosis, and regulators of transcription and translation (reviewed in ref. 20–23)). Thus, unrestrained MMP activities could be highly detrimental, and so activity is tightly regulated through expression of only small amounts of these enzymes, controlling their cellular localization and modulating their catalytic activity through the presence or removal of the pro-domain (Fig. 3). Further levels of control occur through modulating levels of transcription through interleukins, growth factor and TNF-α, by the anti-protease α-2 macroglobulin situated in the blood, and by the tissue inhibitors of metalloproteinases (TIMPs). The four TIMP family proteins function by binding directly to the MMP active site to inhibit catalytic activity, and their function is to provide a counter balance to the MMPs, to control overall net activity levels. TIMPs are two domain-containing proteins, and the N-terminal inhibitory domain binds in a one-to-one ratio to a MMP active site, and this domain alone can cause inhibition of MMP activity,42 but the C-terminal domain also interacts with MMPs, and its inclusion increases inhibition by approximately an order of magnitude.43 The four TIMPs have a broad and overlapping specificity, but the level of affinity for each MMP can vary considerably. The tightest interactions involved TIMP-2/MMP-2 and TIMP-1/MMP-9, and binding of these TIMPs may include interactions with the PEX domain of each of these MMPs.44 Interestingly, not all of the TIMP:MMP interactions are inhibitory, as for example TIMP-2 can interact with pro-MMP-2 and promote the activation of this pro-MMP by MT1-MMP.45 The activity of MMP-2 and MMP-9 can be further regulated by the reversion-inducing cysteine-rich protein with Kazal motifs (RECK), which is a GPI-anchored glycoprotein. RECK competitively inhibits the proteolytic activity of MMP-2 and MMP-9, and it effectively suppresses the extracellular release of proMMP-2 and proMMP-9.46

Fig. 3. Regulation of MMP activities. Intracellular MMP activation may occur by the action of convertases, including Furin, or at the cell surface. For example, MT1-MMP dimers can form trimeric complexes with TIMP and MMP-2, which results in cleavage of the MMP-2 propeptide, and thereby releasing activated MMP-2 into the extracellular milieu. MMPs can also be activated without removal of the inhibitory propeptide domain, by reactive oxygen species (ROS), reactive nitrogen species (ONOO^–) or by glutathiolation. The activity of MMPs can also modulated by activators, such as extracellular matrix metalloproteinase inducer (EMMPRIN) and inhibited by tissue inhibitor of metalloproteinases (TIMPs) and/or reversion-inducing-cysteine-rich protein with Kazal motifs (RECK). Growth factors, such as PDGF and TGF-β, can enhance the production of TIMPs. Cytokines or CD40 signaling can promote the exocytosis of Pro-MMPs stored in granules.

Certain serine proteases, along with other MMPs, can activate pro-MMPs, where for example MMP-3 cleaves pro-MMP-9 or pro-MMP2, to generate the active form of the MMP enzyme. Also, reactive oxygen species (ROS) and reactive nitrogen species (RNS) can control MMP expression and activation,47 and this includes a direct interaction with the Cys thiol group of the cysteine switch motif, resulting in catalytic activation. Hypochlorous acid (HOCl) can activate multiple pro-MMPs, including MMP-1, -7, -8 and -9 (ref. 48–52) and it was observed that low levels of hypochlorous acid activated MMP-7, but higher levels or extended exposure resulted in inactivation.53 RNS can exert a similar effect of the cysteine switch motif, and S-nitrosylation can activate MMP-1, -2, -8 and -9 in vitro, and such activation has been further observed in MMP-9 in vivo, in cerebral ischemia.54 The same set of MMPs have also been observed to be activated by S-glutathiolation in vitro. RNS can additionally inactivate the TIMPs, potentially magnifying the MMP activity. Interestingly, it was noted that RNS can activate MMPs that still contain the pro-domain, indicating that this form, normally referred to as inactive, could have catalytic activities under ROS or RNS oxidative stress or potentially even at physiological levels of these reactive species.

The membrane type (MT) MMPs all contain the furin cleavage motif, and are activated within the Golgi network by furin serine proteinase, or other members of the pro-protein convertase family.55–57 Secreted MMPs are also activated by serine proteases, such as plasmin, or by other MMPs, such as MT1-MMP that activates MMP-2. Definitions of the biological roles of the activated MMPs are still being defined, in part due to functional redundancy that occurs through at least partial overlapping substrate preferences. Transgenic animals lacking expression of individual MMPs only provide subtle phenotypes, with the exception of MT1-MMP. Null MT1-MMP knockout mice develop bone malformations, dwarfism and die before adulthood, revealing that MT1-MMP has critical roles in targeting its substrate, type 1 collagen, for turnover.58–60

Other mechanisms of control likely include phosphorylation and post-transcriptional regulation. Phosphorylation events may have roles in controlling MMP activity, where MMP-2 phosphorylation by protein kinase C reduces MMP-2 activity in vitro, along with dephosphorylation by alkaline phosphatase notably increasing the activity of this MMP.61 Also, MT1-MMP impaired phosphorylation of Tyr573 on its cytoplasmic domains results in inhibited tumor cell proliferation in vitro and tumor growth in vivo (ref. 62). Post-transcriptional regulation may also regulate MMP activity, as evidenced by a constitutively active 40 kDa MMP-11 intracellular isoform that lacks its secretory signal peptide and also its prodomain. Other MMPs have been suggest to have similar isoforms63 and this includes MMP-2 isoform that lacks its secretory signal peptide,61 as well as MT3-MMP being produced by alternative splicing to generate either a soluble or a transmembrane form.64

MMP functions in health and disease

Many cells secrete MMPs, including connective tissue, osteoblasts, leukocytes, fibroblasts and vascular smooth muscle (VSM) cells, promoting the biological processes of tissue remodeling during angiogenesis, embryogenesis, morphogenesis, antimicrobial activities and wound repair. However, MMP activities are also strongly implicated in a host of pathological conditions (Fig. 1), which includes inflammatory processes, autoimmune diseases that includes multiple sclerosis and systemic lupus erythematosus, osteoarthritis, periodontitis, fibrotic disorders, vascular diseases that may comprise of myocardial infarction, angiogenesis or aneurysm.1,3–12 MMPs have critical functions in cancer,2 being up-regulated in nearly all cancers, and with many of the MMPs being initially cloned as cancer-specific genes.65 Increased levels of MMPs were initially linked to metastasis, due to their activities facilitating the breakdown of the ECM. Cancer roles have now been extended to initiation processes, where MMPs may promote genome instability66,67 and to the progression of tumorigenesis,68,69 where degradation of the ECM by MMPs releases pro-tumorigenic molecules, in addition to aiding later steps of tumor metastasis and invasion, and angiogenesis.19 Mammary epithelial cells exposed to MMPs can cause an epithelial–mesenchymal transition, which makes the cells much more invasive,70–72 which likely relates to metastasis in a cancer setting. The pathological functions of MMPs further extend to neurodegenerative diseases and neuropsychiatric disorders.7,73,74 Thus, MMPs are highly attractive therapeutic targets to develop new therapies to treat these disease states. MT1-MMP, which is frequently up-regulated in tumors, promoting cellular invasion and metastasis is one such attractive MMP target. Studies on the regulation of MMPs have revealed that a potent means of activation of pro-MMP-2, is through the interaction of MT1-MMP together with TIMP-2, where TIMP-2 serves as the link between MT1-MMP and MMP-2.45 Extracellular matrix metalloproteinase inducer (EMMPRIN) can also play a significant role in inducing the secretion and activation of MMP-2, in addition to MMP-9, and thereby enhance the invasive ability of cells.

Efforts over the previous two decades to develop MMP inhibitors have proved unsuccessful in phase 2 and in phase 3 clinical trials. This setback in clinical trials was attributed largely due to poor bioavailability, lack of specificity, inadequate clinical efficacy75,76 and a potentially less appropriate study design,77 and previous lack of understanding of MMP functions on intracellular targets may have also played a role. However, such MMP inhibitors have proven useful as tools that have helped further define MMP functions at a cellular level. More recent support for a more appropriate use of MMP inhibitors in combating the early stages of tumorigenesis, including a study on mouse model using ‘SD-7300’ that is an MMP-2/-9/-13 inhibitor. In these studies, this compound markedly reduced the metastasis to the lungs of an aggressive mammary carcinoma, and also increased survival rates.78 Some MMPs are known to have anti-tumor activities, and thus non-selective inhibition of these MMPs would likely prove detrimental.79 However, the design of these initial clinical trials was likely not optimal for effective testing of MMP inhibitors, as the trials were carried out in patients with advanced metastasis, while MMPs are known to instead play a crucial role in the early stages of disease.50

The role of natural products in MMP regulation

Natural products have long been a vital source of bioactive molecules that allow researchers to interrogate biochemical pathways, and ultimately develop therapeutic interventions. In some cases, natural products themselves become approved drugs, while in others they serve as a starting point for synthetic chemists to develop new derivatives.80,81 Inspired by the four endogenous inhibitors (TIMPs) in humans, which include two proteins and two glycoproteins, many peptides and peptide mimics have been reported as synthetic MMP inhibitors.82,83 In addition, a wide variety of metabolites are known to directly inhibit MMPs or reduce MMP expression, from long-chain phenolic lipids like anacardic acids to polysaccharides and sulfated polysaccharides, as well as small molecule natural products from a range of compound classes (Fig. 4).84–86 One of the most prevalent groups is the flavonoids and polyphenols, including eckol and dieckol isolated in 2006 by Kim and coworkers.87 Complex terpenoids derived from both terrestrial and marine sources have been investigated, and the cancer cell migration inhibitor BU-4664L (diazepinomicin) advanced to stage II clinical trials in patients with glioblastoma multiform.88 The diversity of molecular structures has led to a rich starting point for ongoing investigations of natural product derivatives and MMP inhibition, especially targeting new cancer therapies. For more comprehensive coverage of the literature prior to 2017, we refer the reader to previous reviews on the topic.82–85,89,90 Herein, we highlight a few structurally unique examples together with more recently reported natural product MMP inhibitors.

Fig. 4. Natural products that inhibit or suppress expression of MMPs have remarkably diverse structures and derive from a variety of marine and terrestrial sources.

Sinulariolide is a marine diterpene of the cembranoid family that was isolated from the soft coral Sinularia flexibilis in 1975.91 Many sinulariolide derivatives are now known and display diverse biological activity.92 Sinulariolide has been shown to inhibit cancer cell migration and invasion in human hepatocellular carcinoma cells (HA22T) and human bladder cancer cells (TSGH-8301).93,94 Both sinulariolide and 11-epi-sinulariolide acetate result in reduced MMP-2 and MMP-9 expression and increased expression of tissue inhibitors TIMP-1 and TIMP-2, while also resulting in reduced phosphorylation of AKT and mTOR. The authors suggest the activity profile of the sinulariolides merits further evaluation of the cembranoids as agents to prevent tumor metastasis and invasion.

Lemnalol is a polycyclic ylangene-type sesquiterpenoid that was first isolated from the soft coral Lemnalia tenuis Verseveldt by Clardy and coworkers in 1982.95 Soon after, the anti-tumor activity of lemnalol was reported with DBA/MC fibrosarcoma cells in vitro and in a mouse model.96 It was recently identified as having beneficial activity in a gouty arthritis model, resulting in attenuated mast cell (MC) infiltration and degranulation.97 These effects were in part due to reduced expression of MMP-9 and the authors propose lemnalol may be a suitable therapeutic agent for preventing the bone destruction that accompanies gouty arthritis.

Aeroplysinin-1 is a bromotyrosine metabolite with reported antibiotic, anti-angiogenic and anti-tumor effects.98 Both enantiomers have been isolated from different species of sponges, however the (+)-enantiomer isolated from the yellow tube sponge Aplysina aerophoba has been more extensively studied. (+)-Aeroplysinin-1 inhibited the growth of BAE endothelial cells and resulted in decreased MMP expression.99 Specifically, it caused a decrease in the expression of MMP-2 and urokinase, while in other endothelial cell types it reduced levels of MMP-1, MMP-2 and interleukin 1 alpha (Il-1α), among others.98 The aqueous extracts of Aplysina aerophoba have also been demonstrated to reduce the activity and expression of MMP-2 and MMP-9 in rat astrocyte cultures, indicating that other, more polar compounds may contribute to the chemical defenses of this marine organism.100

The abyssinones are flavanone derivatives with diverse biological activities isolated from the East African medicinal plant Erythrina abyssinica.101 In 2010, Bergan, Scheidt and coworkers reported the asymmetric synthesis and testing of four individual abyssinones and the unnatural enantiomers.102 They demonstrated that these molecules selectively inhibit the growth of metastatic human prostate cancer cells at nontoxic concentrations and downregulate the expression of MMP-2, as measured by mRNA transcript levels. The asymmetric synthesis strategy could be useful for the production of unnatural analogs to further explore the structure–activity relationships of the abyssinone scaffold.

Due to the well-characterized structure of the MMP enzymes, including more than 15 experimental crystal structures of MMP-9-ligand complexes, there has been recent interest in using molecular docking to identify selective inhibitors in silico.103 In 2012, Li and coworkers performed a virtual screen of 4000 natural product derivatives and identified 60 candidates, primarily hydroxy-cinnamic acid esters and flavone derivatives.104In vitro testing of the candidates against a panel of MMP enzymes identified 19 active compounds, including a particularly potent inhibitor of MMP-9 (0.99 μM) and MMP-12, dimethyl lithospermate. This caffeic acid derivative was previously isolated from the Asian herbs Lindelofia stylosa and Dracocephalum forrestii and suppresses the migration of MDA-MB-231 tumor cells in a wound healing assay.105,106 In 2017, Zhang and coworkers performed a virtual screen of over 4400 natural product derivatives and identified four novel candidates for potential MMP-9 inhibitors.107 These included one glycosylated flavone and three hydroxy-cinnamic acid esters. Subsequent in vitro testing confirmed the activity of one candidate, the ferulic acid derivative cimicifugic acid B, which reportedly had an IC₅₀ of 13.4 μM against MMP-9. Interestingly, cimicifugic acid B had already been shown to inhibit HCT116 colon cancer cell growth.108 The confirmed hits by both groups bolster the use of receptor-ligand docking as a promising avenue to identify selective inhibitors, however follow-up using biochemical assays will remain an important validation step.

In 2013, Amanlou and coworkers investigated the widespread medicinal plant Onopordum acanthium, also known as Scotch thistle, and isolated a new caffeic acid derivative onopordia.109 Follow-up studies of plant extracts using a surface plasmon resonance (SPR)-based binding assay identified onopordia as being a strong binder of MMP-9.110 These results were supported by docking studies in the MMP-9 active site and an MMP-9 inhibition assay, providing an IC₅₀ of 1.39 μM.

Genipin is an iridoid natural product isolated from Gardenia jasminoides Ellis fruit that has long been used in oriental medicine for its anti-inflammatory effects.111 In 2012, genipin was shown to suppress the motility and invasiveness of human hepatocellular carcinoma cells HepG2 and MHCC97L at non-toxic doses.112 This activity was associated with up-regulation of TIMP-1, and genipin was later shown to inhibit MMP-1 and MMP-3 release from TNF-α-stimulated cells.113 Genipin also induces oxidative stress through inhibition of uncoupling protein-2 (UCP2), sensitizing multidrug resistant cancer cells to chemotherapeutic agents, such as doxorubicin and epirubicin.114 Genipin has also shown efficacy in several in vivo models, however many other molecular targets have been identified and the relative importance of MMP inhibition has not been established.111

UVB-induced inflammation has also been studied to identify the previously known hesperidin, a flavonoid glycoside with established anticancer activity.115 The anti-photoaging properties of Zanthoxylum rhetsa bark extract were traced to hesperidin, which inhibited the increase of pro-inflammatory cytokines and inhibited MMP-1, 3, and 9 in human dermal fibroblasts. Another molecule that has been reported to suppress MMP-1 in UVB- or tumor necrosis factor-α-stimulated dermal fibroblasts is the ginseng-derived compound K.116 This glycosylated terpenoid is one of the major constituents of Panax ginseng, a traditional medicine used in Asian countries for more than 2000 years. Compound K suppresses MMP-1 secretion by inhibition of extracellular signal-regulated kinase (ERK) activation, among other effects.116

In 2018, Lin and coworkers investigated natural products in human retinal pigment epithelial (ARPE-19) cells as a model for proliferative vitreoretinopathy (PVR), a disease that is the leading cause of failure in retinal detachment surgery.117 Kaempferol, a flavonoid natural product with established anticancer and anti-metastasis activity, was found to inhibit the activity and expression of MMP-2 in ARPE-19 via the ERK1/2 pathway. These studies showed that kaempferol inhibits cell invasion and migration by inhibiting MMP-2 function, supporting earlier reports of related flavonoid molecules such as quercetin and luteolin.118,119

Additional polyphenols and biflavonoids isolated from Selaginella tamariscina were investigated by Wang and coworkers.120 They reported micromolar in vitro inhibitory activity of several compounds in different cancer cell lines such as MGC-803, HepG2, A549 and T24. In particular, amentoflavone and robustaflavone showed moderate growth inhibition (IC₅₀ = 8.36–21.38 μM), with robustaflavone showing measurably stronger inhibition than its 4′ methyl ether. This cell-based activity was compared against inhibition of MMP and showed significant variation with structure (Table 1). Amentoflavone most strongly inhibited MMP-9 while robustaflavone was most active against MMP-2. Robustaflavone 4′ methyl ether showed very weak activity against MMP-2 but was the most active against MMP-3. Although the activites are modest, these structure–activity studies provide a starting point for docking studies and further SAR to access more potent analogs.

Table 1. Inhibitory activities of biflavonoids against MMP-2, MMP-3 and MMP-9 from Wang et al.

Natural Product	IC50 (μM)
	MMP-2	MMP-3	MMP-9
Amentoflavone	30.31 ± 0.17	>100	10.33 ± 0.66
Robustaflavone	10.22 ± 0.51	No inhib.	11.85 ± 0.49
Robustaflavone	>100	37.28 ± 0.57	26.08 ± 0.79
4′-Methyl ester

In 2019, Hsiao and coworkers reported on the mechanism of the retinoprotectant thiessenolactone C.121 This small lactone natural product is a promising drug lead for glaucoma and a potent inhibitor of MMP-9, which is activated in cases of elevated intraocular pressure (IOP). The fungal derivative thiessenolactone C was shown to exert ocular protective effects by suppressing neuroinflammation122 and this was the result of inhibition of MMP-9 expression and activation. Hsiao and coworkers suggest that the natural product could be further developed as a therapeutic agent for glaucoma and retinal ischemia diseases.

Synthetic studies have also been reported to investigate the MMP activity of natural product derivatives. Recently, two of our labs outlined a synthesis strategy for the anacardic acid (AA) family of phenolic lipids.123 We wondered whether a redox-relay Heck strategy could be used to join the aromatic head and non-polar tail sections in a modular fashion, with complete control of the alkene position and stereochemistry (Scheme 1). We identified a palladium catalyst system to couple aryl triflate 1 with commercially available unsaturated alcohols to give aldehydes such as 2 in >20 : 1 linear to branched ratio. A Wittig olefination followed by deprotection of the acetonide protecting group produced natural ginkgolic acid 3 and another diunsaturated AA derivative. Conversely, a Julia–Kocienski olefination was used to generate the unnatural E-derivative 4 for comparison. All molecules had very similar activities, with the natural Z-3 and unnatural E-4 inhibiting MMP-2 with IC₅₀s of 6.3 μM and 6.5 μM, respectively. With access to pure isoforms of AAs, ginkgolic acid and unnatural analogs, we demonstrated that the shape and rigidity of the tail had little impact on the inhibition of MMP-2.

Scheme 1. Synthesis of natural and unnatural anacardic acid analogs for isoform-specific bioactivity profiling.

The diversity of structures reported to inhibit the MMPs suggests multiple binding modes, and in some cases this has been confirmed with experimental co-crystal structures and docking studies. The three main categories are zinc-binding ligands, non-zinc-binding inhibitors and mechanism-based inhibitors that form covalent adducts with the protein.124 A large number of MMPis are known to bind zinc through the chelation of two heteroatoms such as the phenol and carboxylate of anacardic acid.125 Zhang's docking approach discussed above identified strong zinc-binding motifs in the flavone natural products as an important contributor to binding affinity.107 Nair and coworkers are among those who have identified potent inhibitors of MMP-2 and MMP-9 that do not bind zinc, such as the natural product derivative biacacetin.126 The conserved nature of the zinc active site among different isoforms may pose a challenge for the design of selective MMP inhibition. The initial MMP inhibitors often contained a hydroxamic acid group for chelation of this zinc,127,128 resulting in these inhibitors binding to the zinc ion in a non-selective fashion,129 which lead to musculoskeletal syndrome (MSS) in clinical trials. A significant contribution to MSS with these compounds was the off-target inhibition of MMP1 and ADAM17/TACE.130 Thus, zinc-binding compounds that have improved selectivities,124,131 in addition to compounds targeting non-zinc-binding motifs, are being pursued for the discovery of successful MMP inhibitors that can potentially reach the clinic. In this regard, structural information about specific binding modes will continue to provide new insight for on-going medicinal chemistry efforts, and in particular those based on natural product leads.

Outlook: MMPs back in the spotlight

Inhibition of MMPs through natural products could help provide a better understanding of the mechanistic attributes of the role and regulation of these versatile enzymes in complex biological processes. As described above, a large number of diverse bioactive natural products, including bioflavonoids, anthogenins, carotenoids and polyphenols, are known to act in a synergistic manner through multiple targets. The cumulative effect of these discoveries has led to a renewed and invigorated interest in formulating strategies for the application of natural products in therapeutic approaches to the control of cancer and other pathophysiological conditions, where MMPs are prime targets.90,132,133

A novel and important aspect of the utilization of natural products as a source of potential novel drugs is the concept of ‘Waste to Wealth’. The incentive here is to evaluate waste or underutilized material from natural sources and apply innovative phytochemical and analytical capabilities, resulting in significant value addition to abundantly available natural resources, which is attractive from a commercial, social, environmental and clinical perspective.134,135 One such material is anacardic acid (AA), which is a major component of cashew nut shell liquid (CNSL) that is an abundant waste product of the cashew industry, and is extracted from the outer shell of the cashew nut. Earlier studies from our laboratories revealed that AA is an effective inhibitor of both MMP-2 and of MMP-9 catalytic activity.125 Moreover, we also observed that AA decreased levels of MMP-2 mRNA, along with a significant inhibition of MMP-2 and MMP-9 protein expression and secretion.136 A dose dependent inhibition of EMMPRIN expression was also seen in the presence of AA, suggesting that this down-regulation of EMMPRIN by AA is, at least in part, responsible for the net reduction of MMP-2 and MMP-9 protein levels.136 The expression of the MMP inhibitor RECK protein was significantly induced by AA in a dose-dependent manner, also suggesting that this up-regulation of RECK may have a key role in inhibition of the gelatinases.136

Another pertinent example of effective utilization of natural product waste material is the isolation and characterization of oxyresveratrol (Fig. 4), a useful stilbenoid isolated from the coconut shell that is ordinarily discarded as waste. Studies with oxyresveratrol treatment of A375 human melanoma cells resulted in down-regulation of MMP-9 activity, along with a corresponding decrease in the expression of MMP-2, MMP-9, COX-2, VEGF and the EGF receptor. Molecular docking studies also demonstrated that oxyresveratrol likely binds in the catalytic site of both MMP-2 and MMP-9 to cause competitive inhibition.137 The available detailed structural information of various MMPs may aid in the quest for novel natural product-based MMP Inhibitors. For example, the shallow active site S2 pocket in MMP-1 limits the interaction of inhibitors with bulky groups that bind to this region, which is distinct from the deep pocket in the active site in MMP-2 and MMP-9. Leveraging such structural differences in inhibitors and the MMPs has already enabled the design of inhibitors that have enhanced interaction with certain MMPs over others,138 which is likely critical for success in the clinic.

Conclusions

The ability to develop selective MMP inhibitors is of critical importance, due to the significant contributions that MMPs play in major disease pathologies that include cardiovascular disease and cancer, the two leading causes of morbidity, in addition to neurodegenerative disorders and inflammatory diseases including arthritis. The failure of previous clinical trials with the broad-based and Zn²⁺-chelating MMP inhibitors highlights the need for MMPis with improved specificity and bioavailability, and also the need to take into consideration intracellular roles of MMPs. Recent approaches to generate improved inhibitors have included the use of Abs, biologics, peptides and small molecule based inhibitors.139 These potential therapies are being developed to either target the active site, to provide indirect blocking of the active site through mechanisms of allostery, to bind to and occlude other secondary binding sites within the larger protease that are required for function, or to prevent the activation of the proMMP molecule.139 Monoclonal antibodies (mAbs) and Ab fragments are being developed as therapeutic agents against the MMPs,25,140–146 and Abs that directly chelate the active site Zn may have an advantage by mimicking the recognition features of the TIMPs, while providing increased selectivity.147 Other sites within the catalytic domain may also be targeted, as highlighted by the development of REGA-3G12 mAb against MMP-9,148,149 and Andecaliximab that perturbs proMMP-9 activation and inhibits catalytic activity in a non-competitive manner.150,151 Additionally, recombinant human scFv Abs have been generated against the MMP-14 PEX domain.152 However, there are certain disadvantages of such Ab-based approaches, which includes their costs in the clinic, that they may be degraded and rapidly removed from circulation, and they may be unable to target intracellular MMP functions.

A biologics of significant interest is chlorotoxin (ClTx), which is a 36-amino acid peptide isolated from the venom of the deathstalker scorpion (Leiurus quinquestriatus).153 ClTx inhibits MMP-2 and chloride channel activity,154 and has both antiangiogenic and anti-invasion activities.154–157 CITx is of interest in the treatment of glioblastoma,156,158 as it can pass through the blood–brain barrier,157 and it also has a potential application in the development of cancer-imaging agents.157–161 Peptide approaches have included phosphinic peptides that function as transition state analogs.162 Phosphinate triple-helical peptides can additionally act as substrate (collagen) mimics and thereby can increase their MMP specificity through targeting of the S and S2 sites.163–165 Such triple helical peptides have been shown to inhibit MMP-2, MMP-9, and MMP-14,166 and have efficacy in mouse models of multiple sclerosis.167 The further development of small molecules targeting MMPs has lead to the generation of ND-322 and analogs that can selectively target MMP-2 and MMP-14 (ref. 168 and 169) and Roche 28-2653, an orally bioavailable pyrimidine-2,4,6-trione derivative that is selective for MMP-2, MMP-9 and MMP-14. Roche 28-2653 has anti-tumor and anti-angiogenic activites, and it is not associated with MSS.170 Small molecule compounds that target the PEX domain of MMPs is providing another area of interest for ongoing drug design efforts, where these compound could potentially prevent the dimerization and subsequent activation of MMPs. Several such compounds have been identified through either virtual or high-throughput screening approaches that inhibit MMP-9 function in cellular-based assays and where tested, in animal models of disease.171–175

Based on the recent experimental work described, computational studies will likely continue to play an important role in MMPi development.176–179 Docking studies using the known crystal structure as described above could be used to evaluate MMP specificity as part of pre-clinical development, and molecular dynamics should see increasing use to account for the flexibility of the MMP substrate-binding region. A challenge that is perhaps still to be addressed is that the same MMP may have different roles in promoting or inhibiting disease pathology. This is based upon a number of potential factors that includes extracellular or intracellular localization, the presence of distinct post-transcriptional isoforms, differences in post-translational modifications, and differences in activation states to due ROS or RNS mediated stresses. This can add increased complexity to the goal of therapeutic MMP inhibition, and where the cell permeability or lack thereof of small molecule inhibitors could play important roles. Thus, further definitions of intracellular MMP functions could likely be of significance to ongoing drug discovery efforts. Also, treatment timing may be of importance, where the ideal time point for delivery may be during the pre-metastatic stage of tumorigenesis.180

The discovery and characterization of diverse natural product-based inhibitors can potentially add great value to a field that clearly needs to develop new biomarkers and inhibitors. Such natural products may also provide novel approaches to develop MMP-focused therapies, where new chemistries and activities could ideally provide increased target selectivity. Advantageously, the discovery and characterization of new compounds can now be coupled with the increased understanding of MMP biology, improving the likelihood of success in translational research. Natural product-based selective inhibitors will also remain of great value as tools in characterizing areas of MMP cellular activity that need further clarification. This includes roles of MMPs in promoting genome instability or activities in neurodegenerative disease states, as well as helping to elucidate the dynamic spatial and temporal interplay of differing MMP activities that may occur during disease progression.

Conflicts of interest

There are no conflicts to declare.