mmp-9

MMP-9 (Gelatinase-B)

Matrix metalloproteinase-9 (MMP-9, gelatinase-B, 92K gelatinase) is a secreted zinc-dependent endopeptidase with the broadest substrate range among the collagenase/gelatinase subfamily. It is central to extracellular matrix (ECM) remodeling in aging skin, is a component of the SASP secretome in senescent fibroblasts, and is implicated in atherosclerotic plaque destabilization and blood-brain barrier disruption. MMP-9 is distinct from MMP-2 (72K gelatinase, gelatinase-A): these two enzymes share gelatin-binding domains but differ in regulation — MMP-9 is UV-inducible and NF-κB/AP-1-responsive, while MMP-2 is constitutively expressed and not induced by UVB irradiation or by tretinoin ¹.

Identity

• UniProt: P14780 (MMP9_HUMAN)
• NCBI Gene: 4318
• HGNC symbol: MMP9 (aliases: CLG4B)
• Ensembl: ENSG00000100985
• Mouse ortholog: Mmp9 (one-to-one ortholog)
• Protein length: 707 amino acids (pre-pro form); mature active form ~83 kDa after prodomain cleavage from the ~92 kDa proenzyme (by convention the “92K” designation refers to the proenzyme)
• EC: 3.4.24.35

Structure and unique features

MMP-9 shares the standard MMP domain architecture — signal peptide, propeptide (cysteine-switch), catalytic domain, hinge region, hemopexin-like C-domain — but is unique among MMPs in possessing three tandem fibronectin type II (FN-II) repeats inserted within the catalytic domain ². These gelatin-binding inserts enable tight docking to denatured collagen substrates (gelatin) and intact basement-membrane collagen IV, explaining MMP-9’s primary activity on these substrates.

Cofactors: 2 Zn²⁺ ions (one catalytic, one structural) and 3 Ca²⁺ ions per subunit. Post-translational modifications include N- and O-glycosylation and disulfide bonds ².

NGAL complex: MMP-9 forms a 1:1 covalent (disulfide-bonded) heterodimer with NGAL (neutrophil gelatinase-associated lipocalin; lipocalin-2, LCN2), yielding a ~125 kDa species detectable in plasma and in neutrophil-derived secretions ². This NGAL/MMP-9 complex is clinically exploited as a biomarker of neutrophil activation; NGAL protects MMP-9 from autocatalytic inactivation in the complex ². needs-replication — the proteolytic-protection mechanism for the NGAL complex is inferred from structural data; direct in-vivo stabilization kinetics in aged tissue are not established.

Key substrates

Substrate	Aging relevance
Denatured collagen (gelatin, type I/III fragments)	Cleaves collagen fragments generated by MMP-1; amplifies ECM degradation loop
Collagen IV (basement membrane)	Disrupts dermal-epidermal junction; contributes to rete ridge flattening with age
Collagen V, VII, XIV	Broader ECM remodeling
Elastin	Contributes to loss of skin elasticity; vascular wall weakening
Fibronectin	Disrupts cell-matrix adhesion signaling
KiSS1, ninjurin-1	Bioactive peptide processing (per UniProt annotation)

Activation: the cysteine-switch mechanism

Secreted as inactive proMMP-9 (~92 kDa). The cysteine-switch motif in the propeptide coordinates the catalytic Zn²⁺, blocking substrate access. Activation requires propeptide removal:

• MMP-3 (stromelysin-1) is the primary activating protease for proMMP-9 in inflammatory and wound-healing contexts — MMP-3 cleaves the MMP-9 propeptide to yield the active ~83 kDa species.
• Plasmin and trypsin can also activate proMMP-9 under inflammatory conditions.
• Note: MMP-2 activation proceeds via a distinct MT1-MMP/TIMP-2 cell-surface mechanism and is not regulated by MMP-3; the two gelatinases differ at this step as well as at the induction level.

Transcriptional regulation and aging-relevant induction signals

MMP-9 gene expression is driven by AP-1 (c-Fos/c-Jun) and NF-κB binding sites in its promoter — the same transcription factors that are activated by UV irradiation, ROS, pro-inflammatory cytokines, and SASP factors.

UV induction in human skin (in vivo): A single dose of UVB irradiation applied to adult human buttock skin induced MMP-9 (92K gelatinase) mRNA, protein, and enzymatic activity within 24 hours in a dose-dependent manner, with significant induction starting at 0.1 MED (~2–3 min sun exposure without visible reddening) and approaching maximal induction at 1 MED. Pretreatment with topical all-trans retinoic acid (tretinoin) reduced UVB-induced MMP-9 mRNA, protein, and activity by 50–80% via AP-1 transrepression — protein-protein interaction between retinoic acid receptors and Jun/Fos, without reducing UV-induced skin reddening ¹. MMP-2 (72K gelatinase) was not induced by UVB and was not altered by tretinoin in this experiment — a critical distinction frequently obscured in secondary literature.

Dimension	Status
Pathway conserved in humans?	yes
Phenotype (UV → MMP-9 induction) demonstrated in humans?	yes — direct human in vivo
Replicated in humans?	yes (Fisher 1996 is direct human biopsy data; n=9–10)

SASP context: Senescent fibroblasts produce MMP-9 as part of the SASP. Note on Coppé 2008: the SASP atlas used an antibody array of 120 selected proteins focused on cytokines, chemokines, and growth factors (e.g., IL-6, IL-8, GRO, GM-CSF, IGFBP family, MCP family, TIMP-1, TIMP-2) — MMP-9 was not among the proteins on this array and therefore was not directly measured or confirmed in this study ³. Evidence that MMP-9 is secreted by senescent fibroblasts derives from independent studies examining gelatinase activity in senescent cell conditioned media (see gap tag below). The SASP-derived MMP-9 is mechanistically plausible given the NF-κB/AP-1 promoter regulation and contributes to paracrine ECM degradation in aged tissue, amplifying the collagen-fragmentation cycle described for MMP-1 ⁴. needs-replication — direct quantification of MMP-9 protein in SASP secretome of senescent human fibroblasts (as distinct from gelatinase zymography or cell-line studies) is not established by Coppé 2008.

NF-κB constitutive activation in senescent cells (maintained by the IL-1α autocrine loop and cGAS-STING-derived signaling) drives persistent MMP-9 expression independent of fresh UV insult — this is the mechanism linking cellular senescence to ongoing matrix degradation in chronologically aged dermis, independent of photoaging.

Role in aging: skin

MMP-9 participates in the self-amplifying collagen fragmentation loop in aged dermis, primarily described for MMP-1 but with MMP-9 (and MMP-3) as secondary amplifiers. In the Fisher 2009 mechanistic framework ⁴:

• Elevated MMP-1 activity fragments dermal collagen I
• Collagen fragmentation reduces fibroblast cytoskeletal tension
• Reduced tension elevates mitochondrial ROS (~3-fold in vitro)
• ROS activates c-Jun/AP-1 + upregulates α2β1 integrin
• AP-1 drives further MMP-1 (and MMP-9, MMP-3) transcription
• Loop is self-sustaining

MMP-9 specifically degrades basement-membrane collagen IV (type IV collagen), contributing to flattening of the rete ridges — the dermal papillae interdigitations with the epidermis that provide mechanical interlocking between dermis and epidermis. Rete ridge flattening is a histological hallmark of aged skin and contributes to the fragility and easy bruising of elderly skin ². needs-replication — the specific contribution of MMP-9 vs other basement-membrane-active proteases (MT-MMPs, cathepsins) to rete ridge attenuation has not been isolated in in vivo human studies.

Role in aging: vascular

MMP-9 is expressed in atherosclerotic plaques, primarily by macrophages and foam cells in the fibrous cap region. Elevated MMP-9 degrades collagen IV and elastin in the plaque fibrous cap, reducing cap thickness and tensile strength. Cap thinning below ~65 μm is the histological threshold for “vulnerable plaque” (TCFA — thin-cap fibroatheroma) associated with rupture risk and acute coronary events ². no-mechanism — the causal contribution of MMP-9 specifically (versus MMP-2, MMP-8, or other MMPs in the same plaque microenvironment) to rupture events is not cleanly resolved by genetic or pharmacological evidence in humans.

Aging increases macrophage-derived MMP-9 secretion in plaques via multiple inputs: elevated circulating IL-6 and TNF-α (inflammaging), oxidized LDL-driven NF-κB activation, and telomere-driven DNA damage responses in plaque-resident macrophages. The net effect is accelerated plaque remodeling with age.

Role in aging: nervous system

MMP-9 contributes to blood-brain barrier (BBB) permeability by degrading collagen IV and laminin in the endothelial basement membrane. In aged mice, MMP-9 is elevated in the cerebral microvasculature and is induced by neuroinflammatory signalling via a ROS-CypA-MMP-9 axis ⁵. One surgical-stress study found that MMP-9 activation in aged mice mediates anesthesia/surgery-induced BBB disruption and postoperative cognitive dysfunction-like behavior, with MMP-9 inhibition (doxycycline treatment) reducing BBB leakage markers ⁶.

However, a Tarantini et al. 2021 genetic-depletion study produced a paradoxical worsening: MMP-9 null aged mice (20 months) showed higher ICH incidence than wild-type controls — 77% vs 35% (n=22 KO, n=17 WT; log-rank χ²=5.701, p=0.017) — following hypertension-induced intracerebral hemorrhage. The authors concluded that “MMP-9 deficiency does not ameliorate, rather increases, the incidence of hypertension-induced intracerebral hemorrhages in aged mice” ⁷. This is not a null result; MMP-9 appears to be protective in this ICH context, possibly because its loss triggers compensatory upregulation of other matrix metalloproteinases. contradictory-evidence — whether MMP-9 contributes to BBB aging in humans remains unresolved; the genetic-depletion (paradoxical worsening in ICH model) and pharmacological-inhibition (BBB-protection in anesthesia/surgery model) data give discordant signals in different cerebrovascular disease contexts.

Dimension	Status	Notes
Pathway conserved in humans?	yes	BBB MMP-9 biology shared across mammals
Phenotype conserved in humans?	partial	Elevated MMP-9 in human CSF and brain aging; causal role not established
Replicated in humans?	no	Genetic and intervention data are mouse-only

MMP-9 vs MMP-2: critical distinction

This distinction was explicitly confirmed in the Fisher 1996 primary source ¹ and must not be confused:

Feature	MMP-9 (92K, gelatinase-B)	MMP-2 (72K, gelatinase-A)
UniProt	P14780	P08253
UV (UVB) induction	Yes — dose-dependent from 0.1 MED	No — not induced by UVB
Tretinoin (t-RA) effect	Suppressed 50–80%	Not altered
Activation	MMP-3, plasmin, trypsin	MT1-MMP/TIMP-2 cell-surface mechanism
Promoter	AP-1 + NF-κB response elements	Lacks AP-1/NF-κB elements; constitutive
NGAL complex	Yes (disulfide-bonded heterodimer)	No
Primary expression context	Inflammatory cells, UV-stimulated keratinocytes/fibroblasts, SASP	Ubiquitous constitutive; dermal fibroblasts, smooth muscle

Druggability

Broad MMP inhibitors (marimastat, batimastat, ilomastat) achieved robust in vitro and in vivo activity but failed in multiple oncology Phase III trials due to dose-limiting musculoskeletal toxicity (MMP inhibitor arthralgia/fibromyalgia syndrome) caused by off-target inhibition of other MMPs (notably MMP-1, MMP-14) required for normal connective tissue homeostasis. This failure class foreclosed broad MMP inhibition as a therapeutic strategy.

Selective MMP-9 inhibitors represent the current investigational strategy:

• Andecaliximab (GS-5745) — a selective monoclonal antibody inhibitor of MMP-9 (allosteric mechanism, not active-site). Tested across multiple indications: (1) Phase 3 gastric/GEJ cancer (GAMMA-1 trial) failed to meet primary endpoint — median OS 12.5 vs 11.8 months (ADX vs placebo + mFOLFOX6; Shah MA et al., JCO 2021, PMID 33577358); (2) Phase 2/3 ulcerative colitis — failed to meet primary endpoint (histological remission at week 8; Panaccione R et al., J Crohns Colitis 2018, PMID 29767728); (3) Phase 1b rheumatoid arthritis — safety/PK study only, no efficacy endpoint (Genovese MC et al., Clin Ther 2018, PMID 29287749). No aging-indication trials have been initiated. long-term-unknown — no clinical-stage selective MMP-9 inhibitor data exist for an aging-relevant indication.
• Small-molecule selectivity remains challenging given the conserved active-site zinc-chelating pharmacophore across the MMP family.

Aging-context druggability-tier rationale: Tier 2 (high-quality probe exists in andecaliximab; clinical drug not FDA-approved; no aging-indication trial). If the aging-context criterion is applied strictly (no validated modulator for an aging indication), tier 3 would be defensible — tier 2 reflects that andecaliximab provides a selective Phase 3-tested probe.

Tretinoin (topical) reduces MMP-9 induction in UV-stressed human skin via AP-1 transrepression ¹, but tretinoin is not a direct MMP-9 inhibitor — it acts upstream at the transcriptional level. Its clinical efficacy in photoaging is established, but the MMP-9 pathway contribution (vs MMP-1 or TIMP induction effects) is not cleanly isolated.

Pathway and process linkages

• ap1-pathway — c-Jun/c-Fos transcriptional control of MMP-9 promoter; activated by UV and SASP-derived cytokines
• nf-kb — co-regulates MMP-9 transcription; central to SASP-driven MMP-9 in senescent cells
• sasp — MMP-9 is a SASP component in fibroblast and macrophage senescence; amplifies paracrine ECM degradation
• mmp-1 — primary collagenase in the skin aging fragmentation loop; MMP-9 acts as secondary amplifier on the basement membrane
• mmp-3 — MMP-3 (stromelysin-1) is the canonical activating protease for proMMP-9
• ecm-remodeling — MMP-9 is a central effector of ECM proteolysis in multiple tissues

Key interactors

• MMP-3 (stromelysin-1) — proMMP-9 activator
• NGAL / LCN2 — covalent complex partner from neutrophils; biomarker context
• TIMP-1 — primary endogenous inhibitor of active MMP-9 (TIMP-1 binds active MMP-9 with higher affinity than it inhibits most other MMPs; does not inhibit proMMP-9)
• Fibronectin, vitronectin — ECM scaffolds that regulate MMP-9 availability via non-covalent binding

Limitations and gaps

Gap	Tag	Notes
MMP-9 causal contribution to aging independent of MMP-1 not established in humans	needs-human-replication	Most human in vivo data are from Fisher 1996 showing co-induction with MMP-1; no study dissecting aging-specific MMP-9-only effects in aged human skin
SASP-driven MMP-9 in vivo quantification in aged tissue	needs-replication	Coppé 2008 is in-vitro; SASP-MMP-9 tissue levels in aged human dermis not reported separately from total gelatinase activity
Rete ridge attenuation attributed to MMP-9 specifically	no-mechanism	Mechanistic attribution is plausible from substrate profile; not demonstrated by MMP-9-specific genetic/inhibitor studies in human skin
BBB/neurological MMP-9 role discordant	contradictory-evidence	Tarantini 2021: MMP-9 KO mice showed paradoxically worse ICH incidence (77% vs 35%, p=0.017) — MMP-9 appears protective in ICH context; pharmacological inhibition reduces BBB leakage in anesthesia/surgery model; no human genetic data
Selective MMP-9 inhibition for aging indication	long-term-unknown	No clinical-stage trial; andecaliximab failed in oncology; aging-indication studies not initiated
GenAge entry absent	needs-canonical-id	MMP-9 has no GenAge (HAGR) entry; not catalogued as a longevity-associated gene in standard model-organism genetic screens

Footnotes

Footnotes

1. fisher-1996-photoaging-ap1-mmp · n=6–17 (varies per experiment; n=9–10 for MMP-9 endpoints) · in-vivo (human skin biopsy) · P<0.01 vs vehicle · model: adult Caucasian human buttock skin · UVB-induced MMP-9 (92K gelatinase) mRNA, protein, activity; tretinoin reduces 50–80%; MMP-2 not induced — text-stated findings verified against primary PDF ↩ ↩² ↩³ ↩⁴

2. UniProt P14780 (MMP9_HUMAN) · reviewed Swiss-Prot entry · accessed 2026-05-19 · source for domain architecture, cofactors, substrates, PTMs, and NGAL-complex annotation ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

3. doi:10.1371/journal.pbio.0060301 · Coppé JP, Patil CK, Rodier F, Sun Y, Muñoz DP, Goldstein J, Nelson PS, Desprez PY, Campisi J · PLoS Biol 2008 · n=multiple human fibroblast lines · in-vitro · antibody-array atlas · model: OIS and irradiation-induced senescent human fibroblasts · Note: antibody array covered 120 proteins (cytokines/chemokines/growth factors); MMP-9 was not among the measured proteins — cited here for SASP context; does NOT directly demonstrate MMP-9 in SASP secretome; PDF locally available ↩

4. fisher-2009-collagen-fragmentation-mmp · n=4–7 in-vivo; n=3–5 in-vitro · in-vivo + in-vitro · P<0.05 · model: aged human dermis biopsies + 3D collagen lattice cultures · collagen fragmentation loop; MMP-9 mentioned as secondary amplifier; note paper does NOT present independent quantitative MMP-9 data — primary data are for MMP-1 ↩ ↩²

5. doi:10.3389/fnagi.2022.1021129 · Liu LF, Hu Y, Liu YN, Shi DW, Liu C, Da X, Zhu SH, Zhu QY, Zhang JQ, Xu GH · Front Aging Neurosci 2022 · PMID: 36337710 · in-vivo (aged mice) · model: aged mice, anesthesia/surgery model · ROS-CypA-MMP-9 signaling axis driving delirium-like behavior via BBB impairment in aged mice ↩

6. doi:10.1016/j.intimp.2024.112290 · Hu Y, Hu XD, He ZQ, Liu Y, Gui YK, Zhu SH, Da X, Liu YN, Liu LX, Shen QY, Xu GH · Int Immunopharmacol 2024;135:112290 · PMID: 38796964 · in-vivo (aged mice) · model: aged mice, anesthesia/surgery · anesthesia/surgery activates MMP-9 → BBB disruption and neuroinflammation in aged but not young mice; supports age-dependent MMP-9 upregulation in BBB context ↩

7. doi:10.1007/s11357-021-00402-5 · Tarantini S, Yabluchanskiy A, Lindsey ML, Csiszar A, Ungvari Z · GeroScience 2021;43(5):2611–2619 · PMID: 34415518 · PMC8599521 · in-vivo (aged mice, MMP-9 KO) · model: 20-month-old MMP-9 null mice (n=22) vs WT controls (n=17), angiotensin II + L-NAME hypertension-induced ICH · Paradoxical worsening: MMP-9 KO had 77% ICH incidence vs 35% in WT (log-rank χ²=5.701, p=0.017) — MMP-9 deficiency increases ICH incidence; conclusion “MMP-9 deficiency does not ameliorate, rather increases, the incidence of hypertension-induced intracerebral hemorrhages in aged mice”; green OA via PMC8599521 ↩