🧬 Aging Wiki

Collagen Fragmentation Promotes Oxidative Stress and Elevates Matrix Metalloproteinase-1 in Fibroblasts in Aged Human Skin

14 min read

Collagen Fragmentation Promotes Oxidative Stress and Elevates MMP-1 in Fibroblasts in Aged Human Skin (Fisher et al. 2009)

TL;DR

Using aged human sun-protected buttock skin biopsies and three-dimensional collagen lattice cultures, Fisher et al. 2009 demonstrated that protein oxidation is twofold elevated and MMP-1 (collagenase-1) is elevated eightfold at the mRNA level in aged dermis (n=4–7, P<0.05). Collagen fragmentation — the consequence of elevated MMP-1 activity — itself amplifies MMP-1 expression through a mechano-oxidative feedback cycle: fragmented collagen reduces fibroblast cytoskeletal tension, elevates intracellular ROS (threefold in vitro), and activates c-Jun/AP-1 + α2β1 integrin signalling to drive further MMP-1 transcription. This constitutes a self-amplifying degradation loop that drives progressive dermal atrophy in aging skin. Antioxidant treatment with MitoQ₁₀ (mitoquinone mesylate, 1 nmol/L) reduces MMP-1 mRNA ~30% and protein ~40% in vitro (n=5, P<0.05), and partially restores procollagen I synthesis — providing proof-of-principle that the oxidative arm of the loop is pharmacologically tractable.

Central R38 correction note: YAP/TAZ is not described in this paper. The 2009 publication predates YAP/TAZ mechanotransduction in fibroblasts; the actual mechanosensing mechanism described is the c-Jun/AP-1 + α2β1 integrin axis. See dermal-fibroblasts, dermis, and skin-aging for confirmation of R38 propagation removing YAP/TAZ misattribution from downstream pages.

“~1% per year collagen loss” figure is NOT from this paper. That figure originates with Shuster 1975 (Br J Dermatol) and was reiterated by Varani 2006 1. Fisher 2009 does not quantify or cite an annualized collagen loss rate.

Background

Collagen I is the dominant structural protein of the dermis (~60–68% of dermal dry weight) 2. Loss of dermal collagen with aging is a clinically and histologically reproducible finding, but the mechanism driving progressive collagen depletion was incompletely characterised prior to this work.

Prior Fisher-group work had established that:

  • AP-1 transcription factor (c-Fos/c-Jun heterodimer) suppresses collagen I promoter activity and drives MMP-1 expression in UV-exposed skin
  • Retinoids (tretinoin) counter this by transrepressing AP-1 3
  • Age itself — independent of UV — elevates basal MMP-1 in sun-protected skin

Fisher 2009 asked: what is the upstream driver of elevated MMP-1 in chronologically (not UV-) aged skin? The novel hypothesis tested was that pre-existing collagen fragmentation, rather than external UV insult, is the primary amplifier of MMP-1 in intrinsic aging.

Methods

In-vivo cohort (skin biopsies)

  • Subjects: n=4 young (21–30 years) and n=4 aged (>80 years) sun-protected skin donors; biopsy site: buttock (photoprotected; biopsies obtained from sun-protected buttock skin per Methods, p.103)
  • Assays: MMP-1 mRNA and protein; c-Jun mRNA and protein; α2 and β1 integrin mRNA; protein oxidation (carbonyl content); collagen fibre architecture by SEM and H&E; in situ zymography for collagenase activity
  • Key quantitative results (in vivo):
    • Protein oxidation (carbonyls): twofold higher in aged vs young dermis; n=7, P<0.05 (Fig 1 — first result reported)
    • MMP-1 mRNA: eightfold higher in aged vs young dermis; n=4, P<0.02 (Fig 2A)
    • MMP-1 protein: twofold higher in aged vs young dermis; n=6, P<0.01 (Fig 2B)
    • c-Jun mRNA: 2.4-fold higher in aged dermis; n=6, P<0.0001 (Fig 3A)
    • c-Jun protein: fivefold higher in aged dermis; n=6, P<0.05 (Fig 3B)
    • α2 integrin mRNA: 2.9-fold higher in aged dermis; n=3–6, P<0.05 (Fig 4A)
    • β1 integrin mRNA: 9.8-fold higher in aged dermis; n=3–6, P<0.05 (Fig 4B)
  • MMP-3 and MMP-9: This paper does NOT present independent quantitative data for MMP-3 or MMP-9 in the in-vivo cohort. Prior Fisher-lab publications (e.g., Fisher 2001, ref 27 in this paper) reported MMP-8 elevation; MMP-3 and MMP-9 are mentioned as downstream processors of MMP-1 cleavage products in the discussion, but their elevation in aged dermis is not demonstrated by new experiments here.

Important caveats on in-vivo arm:

  • n=4 per group for the mRNA data is very small; statistical threshold P<0.02 is nominal and underpowered for effect-size inference
  • Age confounds: the study cannot separate pure chronological aging from cumulative UV history on the buttock (classified photoprotected, but lifetime UV exposure is never zero)
  • No cell-type dissection in the in-vivo biopsies — mRNA is from LCM-captured dermal fibroblasts (~200 cells per sample), not bulk dermis; this is an important technical note (LCM used, per Methods p.103)

In-vitro model — 3D collagen lattice cultures

Primary adult human dermal fibroblasts (DFs) were cultured in:

  • Intact collagen lattices — DFs embedded in cross-linked collagen I gel, providing physiological cytoskeletal tension
  • Fragmented collagen lattices — collagen pre-treated with bacterial collagenase to simulate the fragmented ECM of aged dermis, reducing fibroblast spreading and cytoskeletal tension

This is the mechanistic core of the paper: the in-vitro model isolates the structural change (collagen fragmentation) from all other aging variables (telomere attrition, epigenetic drift, UV history, circulating signals) to ask whether fragmentation alone is sufficient to drive the MMP-1 elevation seen in vivo.

Additional perturbations tested:

  • Antioxidant treatment (MitoQ₁₀ — mitoquinone mesylate 1 nmol/L, a mitochondria-targeted coenzyme Q10 analogue; MitoQ₅, MitoE₂, and NAC tested as comparators — MitoQ₁₀ was substantially more potent)
  • Forced reduction of cytoskeletal tension (cytochalasin B; actin disruption experiments described in supplemental figures to confirm the mechanical-tension pathway)
  • Blockade of α2β1 integrin (blocking antibody)
  • Overexpression of dominant-negative c-Jun to inhibit AP-1

Readouts:

  • Intracellular ROS levels (fluorescent probe assay)
  • MMP-1 mRNA (RT-PCR) and protein (ELISA/western)
  • Procollagen I mRNA and protein
  • c-Jun phosphorylation (AP-1 activation proxy)
  • α2β1 integrin expression (western / flow)

Key Findings

1. Protein oxidation and MMP-1 are elevated in aged human dermis (in vivo)

The paper’s first reported result is that protein oxidation (carbonyl content) is twofold higher in aged (>80 yr) vs young (21–30 yr) dermis, n=7, P<0.05 (Fig 1). This establishes that oxidative stress is elevated in aged dermis independent of exogenous manipulation.

MMP-1 mRNA is eightfold higher in aged vs young dermis, from sun-protected buttock skin; n=4 per group, P<0.02 (Fig 2A). MMP-1 protein is twofold higher in aged dermis by immunostaining quantification; n=6, P<0.01 (Fig 2B). The discordance between mRNA (8-fold) and protein (2-fold) is not explicitly discussed by the authors but likely reflects post-transcriptional regulation and/or rapid MMP-1 secretion reducing intracellular accumulation. This is a direct demonstration that MMP-1 elevation is an intrinsic aging feature, not solely UV-driven.

MMP-3 and MMP-9: This paper does NOT independently demonstrate elevated MMP-3 or MMP-9 in aged dermis. Those enzymes are mentioned in the discussion as secondary processors of MMP-1 cleavage products that could amplify the cycle, but no new MMP-3/MMP-9 quantitative data is presented here. Cite prior Fisher-lab publications for MMP-3/MMP-9 elevation claims.

2. Fragmented collagen raises intracellular ROS threefold (in vitro)

DFs cultured in MMP-1-fragmented vs intact collagen lattices show threefold higher intracellular oxidant levels (RedoxSensor Red CC-1 fluorescent probe, confocal microscopy); n=3, P<0.05 (Fig 6B). Protein carbonyl oxidation is correspondingly increased (Fig 6C, n=3, P<0.05). This demonstrates that the structural state of the ECM directly determines DF redox state, independent of any UV or senescence signal.

This is the mechanistic link between collagen damage (upstream) and oxidative stress (downstream) — a direction typically assumed to run the other way (ROS → ECM damage).

3. Reduced cytoskeletal tension is the proximal ROS driver

Cytochalasin B (actin polymerisation inhibitor), which reduces cytoskeletal tension without altering the collagen substrate, reproduces the ROS elevation seen in fragmented collagen cultures (supplemental figure data). This isolates reduced mechanical tension as the proximal driver of ROS elevation — confirming that the ECM structure → fibroblast shape → mitochondrial ROS pathway does not require chemical changes to collagen itself. Note: this actin-disruption experiment is in supplemental figures, not the main results.

4. ROS activates c-Jun/AP-1 and upregulates α2β1 integrin (in vitro)

In MMP-1-fragmented collagen lattice cultures:

  • c-Jun protein expression elevated twofold; phosphorylated (active) c-Jun DNA-binding elevated 3.5-fold (Fig 8A–B, n=3, P<0.05)
  • α2 integrin mRNA elevated 2.8-fold; β1 integrin mRNA elevated 1.7-fold (Fig 8C, n=3, P<0.03)

Both the AP-1 activation and α2β1 integrin upregulation are reduced by MitoQ₁₀ treatment (c-Jun expression reduced 33%; α2 integrin expression reduced 40% — reported as “data not shown” in the Discussion, p.109), confirming that ROS is the mediating signal between mechanical tension loss and the AP-1/integrin-driven transcriptional changes.

5. c-Jun/AP-1 + α2β1 integrin axis drives MMP-1 elevation

The mechanism of MMP-1 upregulation is established by two orthogonal perturbations:

  • Dominant-negative c-Jun overexpression reduces MMP-1 transcription in fragmented-collagen cultures, confirming AP-1 dependence
  • α2β1 integrin blocking antibody reduces MMP-1 mRNA in fragmented collagen cultures

Both arms of the mechano-oxidative signal converge on AP-1 to elevate MMP-1. This is the molecular mechanism: fragmented ECM → ↓cytoskeletal tension → ↑mitochondrial ROS → ↑c-Jun/AP-1 + ↑α2β1 integrin → ↑MMP-1 mRNA + protein.

YAP/TAZ is not described in this paper. The mechanosensing pathway described is the AP-1/integrin axis, not the Hippo/YAP/TAZ pathway. YAP/TAZ in dermal fibroblast mechano-regulation was characterised in subsequent work (post-2009); Fisher 2009 predates this literature. unsourced — YAP/TAZ in aged dermal fibroblasts requires a separate primary citation.

6. Fragmented collagen reduces procollagen I synthesis

In addition to elevating MMP-1, fragmented collagen lattice cultures show reduced procollagen I mRNA and protein — consistent with the clinical observation of reduced collagen synthesis capacity in aged DFs. This provides a second branch of the degradation loop: not only is collagen breakdown accelerated, but new collagen production is simultaneously suppressed.

7. MitoQ₁₀ reduces MMP-1 mRNA ~30% and protein ~40%; partially restores procollagen I

MitoQ₁₀ (mitoquinone mesylate, 1 nmol/L, 72 h) treatment of DFs in MMP-1-fragmented collagen lattices (Fig 10, n=5 experiments, P<0.05):

  • Reduces MMP-1 mRNA by approximately 30% (bar in Fig 10A at ~65–70% of CTRL; text p.108 states “MitoQ₁₀ reduced MMP-1 mRNA levels 30%”)
  • Reduces MMP-1 protein by approximately 40% (bar in Fig 10B at ~60% of CTRL; text p.108 states “MMP-1 protein 40%”)
  • c-Jun expression reduced ~33%; α2 integrin expression reduced ~40% (Discussion p.109, “data not shown”)
  • Partial restoration of procollagen I synthesis (stated in paper but quantitative data not shown in main figures)

MitoQ₅ and MitoE₂ (shorter-chain analogues with lower mitochondrial selectivity) were substantially less effective at reducing oxidant levels (Fig 9B), confirming the mitochondrial specificity of the antioxidant effect. NAC (N-acetyl cysteine) at 10 mmol/L was similarly ineffective at reducing oxidant levels (Fig 9B). needs-human-replication — MitoQ rescue in vivo in aged human skin has not been demonstrated.

Note: These are in-vitro (cell culture) findings; no in-vivo MitoQ skin-aging experiment was conducted in this paper.

The Self-Amplifying Fragmentation Loop

The unifying contribution of Fisher 2009 is the articulation of a self-sustaining degradation cycle:

Intact collagen → MMP-1 activity (basal) → collagen I fragmentation
                                                    ↓
                                    Reduced fibroblast stretch + cytoskeletal tension
                                                    ↓
                                       ↑ Mitochondrial ROS (~3-fold)
                                                    ↓
                                    ↑ c-Jun/AP-1 + ↑ α2β1 integrin
                                          ↙                ↘
                              ↑ MMP-1 mRNA/protein      ↓ Procollagen I synthesis
                                          ↓
                                 More collagen fragmentation → loop continues

Once initiated (e.g., by UV exposure, replicative senescence-associated MMP-1 SASP, or age-dependent endogenous ROS), the loop is self-perpetuating without further external input. MMP-3 and MMP-9 are mentioned in the discussion as potential amplifiers that could process MMP-1 cleavage products into smaller fragments, but they are not independently demonstrated in this paper and are not positioned as primary initiators or independent co-drivers of the feedback loop.

Mechanism: c-Jun/AP-1 + α2β1 Integrin Axis

The mechano-sensing mechanism is:

ComponentRole in the loop
α2β1 integrinCollagen-binding integrin; senses intact vs fragmented collagen conformation; upregulated by ROS on fragmented collagen; signals to activate AP-1
c-Jun / AP-1Transcription factor; activated by ROS and α2β1 integrin signalling; drives MMP-1 promoter; simultaneously suppresses collagen I/III promoters
Mitochondrial ROSMechanotransduction intermediate; rises when cytoskeletal tension falls; activates c-Jun and upregulates α2β1 integrin
MMP-1 (collagenase-1)Central effector; cleaves native fibrillar collagen I at a single Gly-Ile bond → generates 3/4 + 1/4 fragments

What this paper does NOT describe:

  • YAP/TAZ / Hippo pathway (post-2009 literature; hippo-yap-taz)
  • TGF-β/Smad3 axis (this is covered in a parallel stream of Fisher-lab work, notably Purohit 2016 on Smad3 decline in aged DFs)
  • NF-κB-mediated MMP-1 regulation (this is UV-specific; less prominent in intrinsic aging context)

Limitations

  • Very small in-vivo cohort. n=4 per group for MMP-1 mRNA; n=6–7 for protein/carbonyl data. P<0.02 for mRNA is nominal. Effect sizes should not be treated as precise; fold-changes could vary substantially in a larger cohort.
  • In-vitro model simplifications. The 3D collagen lattice is a useful surrogate for ECM stiffness/integrity but lacks: other ECM components (elastin, glycosaminoglycans, fibronectin), non-fibroblast cell types (keratinocytes, immune cells), and circulating aging signals (cytokines, growth factors). In-vitro n=3–5 throughout.
  • Single institution, single lab. Fisher lab at University of Michigan; no independent replication of the full mechanistic loop in a separate laboratory as of this paper.
  • No in-vivo antioxidant intervention data. MitoQ rescue is in-vitro only; no topical or systemic MitoQ skin study in aged human subjects appears in this paper.
  • MMP-3 / MMP-9 not quantified here. This paper does NOT present independent MMP-3/MMP-9 quantitative data; they are referenced in discussion only as potential amplifiers. Cite prior Fisher-lab publications for those claims.
  • No correction for UV history. Buttock skin is classified photoprotected but is not UV-naïve; the study cannot fully dissociate intrinsic and photoaging contributions at small n.

Significance and Impact

  • Citation count: 444 (DOI lookup, 2026-05-19); FWCI 17.97; citation percentile 100th — one of the highest-impact papers in dermal aging biology.
  • Established the ECM → fibroblast mechano-oxidative loop as a mechanistic framework that has become standard in skin aging discourse.
  • Provided the conceptual basis for antioxidant interventions in skin aging (topical retinoids, MitoQ, other antioxidants as MMP-1 suppressors).
  • Provides the mechanistic anchor for the collagen fragmentation section on dermal-fibroblasts and dermis.
  • Context for interpreting why tretinoin (topical retinoic acid) works: AP-1 transrepression by retinoic acid receptors breaks the loop at the c-Jun/AP-1 node 3.
  • Historically important for positioning MMP-1 as the primary effector in collagen fragmentation, with MMP-3 and MMP-9 as secondary amplifiers — a distinction with therapeutic implications (selective MMP-1 inhibition vs broad MMP inhibition).

Cross-references

  • dermal-fibroblasts — the collagen homeostasis / self-amplifying fragmentation loop section cites this paper as its mechanistic anchor; c-Jun/AP-1 + integrin mechanism confirmed per R38 verification
  • dermis — dermis.md verified against this paper per R38; confirmed MMP-1-centred framing; YAP/TAZ misattribution corrected
  • skin-aging — skin-aging.md confirmed per R38 verification; wrinkle / intrinsic aging mechanism section
  • loss-of-proteostasis — ECM proteostasis failure; collagen fragmentation as loss of extracellular protein homeostasis
  • cellular-senescence — senescent DF SASP drives MMP-1 secretion as the initiating event in many skin contexts
  • chronic-inflammation — SASP-derived MMP-1 and other matrix-degrading enzymes amplify local inflammation
  • Companion upstream study: 1 — Varani 2006 established baseline: aged fibroblasts produce less procollagen I intrinsically, separate from the fragmentation cycle
  • Companion mechanistic study: 4 — Brennan 2003 established MMP-1 as the dominant UV-induced collagenase in skin (pre-Fisher 2009)

Limitations and Gaps

GapTagNotes
n=4 per group for in-vivo MMP-1 mRNA; all quantitative in-vivo claims need independent larger-n replicationneeds-replicationP<0.02 is nominal in n=4; effect-size confidence interval is wide
MitoQ in-vitro only; no in-vivo skin rescue dataneeds-human-replicationMitoQ topical/systemic trials in aged skin not available at time of writing
~1% per year collagen loss figure is NOT in this paperunsourcedAttributed to Shuster 1975 and Varani 2006 — do not cite Fisher 2009 for this figure
MMP-3 / MMP-9 not quantified in this paper(correction)Paper does NOT present independent MMP-3/MMP-9 data; mentioned in discussion only as potential amplifiers. Do not cite Fisher 2009 for MMP-3/MMP-9 elevation claims
No UV-naïve buttock cohort; intrinsic/extrinsic aging dissociation incompleteno-mechanismPurely intrinsic contribution cannot be fully isolated at n=4
YAP/TAZ absent from this paper(correction note, not a gap)See hippo-yap-taz for YAP/TAZ in fibroblast mechanotransduction (requires separate citation)

Footnotes

Footnotes

  1. doi:10.2353/ajpath.2006.051302 · Varani J, Dame MK, Rittie L, Fligiel SEG, Kang S, Fisher GJ, Voorhees JJ · Am J Pathol 2006;168(6):1861–1868 · PMID:16723701 · observational + in-vitro · n=~9–10 donors per group (young/aged biopsies) + primary DF cultures · model: aged human dermis biopsies + 3D collagen cultures · demonstrates decreased procollagen I collagen synthesis in chronologically aged skin DFs; also provides the experimental basis for the “~1% per year collagen loss” claim often attributed to this paper (see also Shuster 1975) · archive status: bronze OA; download pending — not yet locally available; verify quantitative claims from OA URL before relying on specific figures ↩ ↩2

  2. doi:10.3390/ijms20092126 · Shin JW et al. · Int J Mol Sci 2019;20(9):2126 · review · model: human dermis histology · collagen I accounts for ~80–90% of total dermal collagen; total dermal collagen dry weight ~75% leading to collagen I representing ~60–68% of dermal dry weight; used as source for ECM composition figures on dermis page ↩

  3. Fisher GJ et al. · review/mechanistic · J Investig Dermatol Symp Proc or J Am Acad Dermatol (exact citation to verify) · AP-1 transrepression by retinoic acid receptors as the mechanism by which tretinoin counteracts MMP-1 elevation and skin aging unsourced — exact DOI not confirmed; cite as supporting context only pending DOI verification ↩ ↩2

  4. doi:10.1562/0031-8655(2003)078<0043:mmitmc>2.0.co;2 · Brennan M, Bhatti H, Nerusu KC, Bhagavathula N, Kang S, Fisher GJ, Varani J, Voorhees JJ · Photochem Photobiol 2003;78(1):43–48 · PMID:12929747 · in-vivo + in-vitro · model: UV-irradiated human skin organ culture · establishes MMP-1 as the major UV-induced collagenolytic enzyme (MMP-1 ~1.1 µg/ml in UV-treated vs <100 ng/ml control); companion context for Fisher 2009 which extends to intrinsic aging ↩

Graph View

Backlinks