Modulating IL-11-dependent matrix stiffness to delay ovarian aging

TL;DR

Wu et al. found that human ovarian cortical stiffness, hydroxyproline and collagen-I-rich matrix increased cross-sectionally with age and in chemotherapy-associated premature-ovarian-insufficiency, polycystic-ovary-syndrome and endometriosis. Primary-human-cell and mouse-culture experiments support a model in which TGF-β1 induces il-11, IL-11/IL11RA1 activates ERK1/2 in ovarian fibroblasts, and collagen deposition stiffens the matrix; experimentally stiff matrices impaired mouse follicle growth and granulosa-cell proliferation and steroidogenesis. Recombinant IL-11 worsened mouse ovarian stiffness and fertility, whereas germline Il11ra1 deletion protected aging, doxorubicin-POI and DHEA-PCOS models. Adult siIl11 lipid nanoparticles and local ovarian AAV-shIl11ra1 improved several stiffness and reproductive endpoints in aged rodents.

The human results are observational or ex vivo; every organ-level rescue was in rodents. Despite the paper’s use of “anti-IL-11 therapy,” no anti-IL-11 antibody was tested as an in-vivo ovarian intervention. The adult interventions tested progression during middle-age, not reversal of established fibrosis. gap/needs-human-replication gap/needs-replication

Design

Human tissue cohorts

All normal ovaries came from people undergoing oophorectomy for cervical or endometrial carcinoma. Samples were in the proliferative phase; donors had not received hormone therapy, radiotherapy or chemotherapy, and a pathologist excluded ovarian metastasis.

Cohort	Age	n	Source and definition
Reproductively young	18–28 y	30	Cancer-related oophorectomy; mean AMH 3.857 ± 1.647 ng/ml
Middle-aged	35–42 y	37	Same surgical context; mean AMH 1.809 ± 0.864 ng/ml
Older	47–52 y	40	Same surgical context; the cohort summary reports AMH <0.06 ng/ml, although Supplementary Table 1 contains several recorded values of 0.08–0.38 ng/ml
Age-matched control	30–40 y	20	Comparator in Fig. 1; its characteristics and whether it was independent of the 107 age-series donors were not reported in Methods or Supplementary Table 1
Chemotherapy-associated POI	30–40 y	16	Preoperative chemotherapy; FSH >25 IU/l
PCOS	30–40 y	10	Cancer-surgery patients with irregular cycles, polycystic ultrasound appearance and at least 12 small antral follicles in one ovary
Ovarian endometriosis	30–40 y	20	Marginal ovarian tissue from endometriotic-cyst dissection

AFM measured Young’s modulus in fresh-frozen, 30-µm cortex sections taken 50–80 µm below the surface; each location was indented three times and ten positions per tissue were averaged. Hydroxyproline, Masson’s trichrome, picrosirius red and immunohistochemistry assessed matrix/fibrosis. ^68Ga-FAPI-04 PET was shown only as representative young-versus-old images with no cohort size reported. TMT proteomics used n=4 biological replicates per age group; age-stratified collagen and TGF-β1 immunohistochemistry used n=9/group. These were cross-sectional associations, not longitudinal measurements of stiffening within the same person.

Cell and matrix experiments

• Low-passage (≤P3) primary human ovarian fibroblasts were stimulated for 24 h with TGF-β1 or recombinant human IL-11 (10 ng/ml). Each RNA-sequencing contrast used n=3/group; collagen, ACTA2, migration, phosphokinase and inhibitor assays tested the candidate pathway. Donor ages, number of independent donors and donor-to-replicate mapping were not reported.
• The in-vitro neutralizing anti-IL-11 antibody appeared only in these fibroblast experiments. Figure 2 and Extended Data Fig. 5 specify 2 µg/ml, whereas Methods give the antibody’s ND50 as 80 µg/ml in the presence of 10 ng/ml IL-11; Methods do not state a separate applied concentration, and the source workbook does not resolve the mismatch.
• Early secondary follicles isolated from 12–13-day-old C57BL/6 mice were cultured for 12 days in 0.5% (soft) or 2% (stiff) alginate. The legend reports growth n=80/group and steroid hormones n=15/group, but the source workbook labels the growth time series n=20/group and hormone timepoints n=10/group; the effective independent sample size cannot be reconciled.
• Primary mouse granulosa cells were cultured on collagen-coated 3-kPa or 30-kPa polyacrylamide substrates (cell area n=40/group; proliferation n=20/group; hormones n=15/group).

Rodent intervention matrix

Model	Perturbation	Route and timing	Principal group sizes and endpoints
Young C57BL/6 females, 8 wk	Recombinant mouse IL-11, 100 µg/kg	Methods and Reporting Summary: subcutaneous daily for 28 d; Fig. 3: twice weekly for 4 wk (unresolved)	Body weight n=13 control/16 IL-11; most ovarian endpoints n=5–8; estrous n=16/group; litter n=6/group
Physiologic aging	Germline C57BL/6J Il11ra1−/− vs WT littermates at 48 wk	Lifelong deletion	Umbrella n=25/group; most AFM, fibrosis, hormones, follicles, ovulation and IVF assays n=5–6; snRNA-seq n=3/group
Chemotherapy POI	WT or germline Il11ra1−/−; doxorubicin 10 mg/kg	Two intraperitoneal doses, once weekly, from 8 wk	Umbrella n=30/group; analyzed subsets generally n=5–6
PCOS-like injury	WT or germline Il11ra1−/−; DHEA 60 mg/kg	Subcutaneous daily for 28 d from 4 wk	Umbrella n=30/group; analyzed subsets generally n=6
Reproductive aging, mouse	siIl11 or control-siRNA lipid nanoparticles; 100 µl/injection	Tail-vein intravenous, twice weekly for 4 wk from 36 wk; euthanasia 7 d later	Umbrella n=30/group; AFM/follicles/ovulation n=6; pregnancy n=8; litter n=3 control/5 treated
Reproductive aging, rat	siIl11 or control-siRNA lipid nanoparticles; 100 µl/injection	Tail-vein intravenous, twice weekly for 4 wk from 40 wk	Umbrella n=30/group; most endpoints n=5–6; pregnancy n=10; litter n=5
Reproductive aging, mouse	AAV-shIl11ra1 vs control	Single ovarian-bursal microinjection at 48 wk; 4-wk follow-up	Umbrella n=25/group; reported endpoint n=6/group

The siRNA injection volume was specified but the siRNA concentration/mass dose and control-siRNA sequence were not. For the AAV experiment, vector serotype, promoter/construct, shRNA sequence, genome dose or titer, injection volume, vendor and control-vector identity were not reported anywhere in the article or supplements; its age, route, follow-up and n came only from the Extended Data Fig. 10 schematic/caption.

Key results

Human ovarian stiffness and ECM increase with age and ovarian pathology

AFM Young’s modulus increased stepwise from young to middle-aged to older human ovarian cortex. In the age-matched pathological series, POI, PCOS and endometriosis samples were each stiffer than controls. Hydroxyproline rose in parallel, and Masson/picrosirius-red staining showed more matrix. Polarized picrosirius red attributed most of the change to thick collagen-I-rich fibers; thin collagen-III-rich fibers were largely unchanged. The exploratory FAPI-PET images also showed higher uptake in older ovaries, but the authors explicitly called for a larger cohort.

Processed proteomics identified age-group differences enriched for ECM organization/extracellular-space Gene Ontology terms and TGF-β pathway annotations; individual COL1A1, COL1A2, COL3A1 and COL4A1 increases were demonstrated by immunohistochemistry (n=9/group), not by the differential-protein table. Multiplex staining showed COL1A1/TGF-β1/DCN coexpression, and COL1A1-high regions were mechanically stiffer than COL1A1-low regions (n=20 measured regions/group; donor independence was not reported). These data associate age and disease with fibrosis and stiffness but do not establish that stiffness caused the human endocrine or fertility phenotype.

Matrix stiffness directly impairs mouse follicle and granulosa-cell function in culture

Follicles in stiff alginate grew less over 12 days and secreted less estradiol and progesterone than follicles in soft alginate. Granulosa cells on 30-kPa substrates spread more, formed thicker actin stress fibers, proliferated less and produced less estradiol and progesterone than cells on 3-kPa substrates. These controlled mouse experiments support stiffness as a causal local constraint, but they did not test intact human follicles or reproduce the full ovarian niche.

TGF-β1–IL-11–ERK fibroblast mechanism

TGF-β1 (10 ng/ml, 24 h) strongly increased IL11 in primary human ovarian fibroblasts; CTGF/CCN2, bFGF and PDGF also increased IL-11 (the paper redundantly lists the synonymous CTGF and CCN2 names as separate treatments). Ovarian IL-11 rose with age in humans, mice and rats (n=6/age/species in expression assays), correlated positively with age and negatively with AMH in human tissue (n=40), and was elevated in the three pathological cohorts. The correlation test is internally inconsistent: Extended Data Fig. 5 says Pearson, whereas Methods specify Spearman.

Recombinant IL-11 induced ACTA2, collagen and fibroblast migration. In-vitro IL-11 neutralization attenuated TGF-β1-induced activation. IL-11 RNA sequencing showed induction of multiple collagen genes and ECM–receptor programs. Several pathway families were engaged, including NOTCH, PI3K–AKT, cGMP–PKG, MAPK and mTOR, but ERK1/2 inhibitors gave the strongest suppression of COL1A1. Phosphokinase arrays and immunoblotting showed ERK/p90RSK activation. The inhibitor-screen Methods and immunoblot/caption identify SCH772984 (5 nM), but Fig. 2m’s quantitative axes and its source-data sheet label U0126; because U0126 is absent from Methods, the inhibitor identity for that quantitative immunofluorescence panel is unresolved.

age / ovarian injury → TGF-β1 and other fibrotic inputs → IL-11
      → IL11RA1 → ERK1/2 → fibroblast activation + collagen secretion
      → matrix stiffening → impaired follicle growth and steroidogenesis

The complete arrow chain was assembled across human tissue, primary cells and rodents; it was not causally intervened on from end to end in humans.

Recombinant IL-11 is sufficient to worsen the mouse phenotype

Four weeks of recombinant IL-11 increased ovarian Young’s modulus, hydroxyproline, collagen, ACTA2 and pERK; lowered AMH and estradiol; raised FSH; and reduced litter size. Body weight was unchanged. The plotted and workbook values show fewer, not more, primary follicles (means 34.2 versus 19.7), antral follicles (26.3 versus 11.5) and total healthy follicles (218.2 versus 142.5) after IL-11. The Results prose says all three “increased,” which is an unresolved direction error; the numerical source data and the accompanying endocrine/fertility phenotype support the decrease direction.

Estrous-cycle regularity was reported as 70% versus 30% in control mice and 43.75% versus 56.25% after IL-11 (n=16/group). A 70% proportion is not attainable with n=16, and the reported Fisher–Freeman–Halton P=0.0003 is not compatible with the nearest integer counts, so this significance claim is not independently reproducible. The dosing schedule is also irreconcilable: Methods and the Reporting Summary specify daily subcutaneous dosing, while the Fig. 3 schematic says twice weekly.

Germline Il11ra1 loss protects aging and injury models

At 48 weeks, Il11ra1−/− ovaries had lower AFM stiffness, hydroxyproline, collagen, ACTA2 and pERK than WT. Secondary follicles increased, atretic follicles decreased and total healthy follicles showed a borderline trend (P=0.0503); AMH and estradiol were higher and FSH lower. Superovulation yielded more oocytes, but two-cell and blastocyst development rates did not differ. snRNA-seq retained 56,363 nuclei from n=3 animals/group and found fewer activated fibroblasts, with lower stromal ECM programs; cell-proportion inferences remain based on three biological replicates per genotype.

In doxorubicin-treated mice, knockout attenuated but did not normalize matrix stiffness, fibrosis, pERK and ovarian dysfunction. In DHEA-treated mice it attenuated stiffness/fibrosis, normalized elevated testosterone, improved estrous cycling and increased antral follicles and ovulated oocytes. In both models, fertilization/early-embryo development rates were not significantly changed. Because the deletion was germline and present before either injury, these are protection/prevention experiments, not post-injury therapeutic rescue.

Adult gene silencing improves middle-aged rodent endpoints

The siIl11 liposomes were approximately 116 nm (PDI 0.203) with >96% reported encapsulation. After four weeks, mouse ovarian stiffness was 35.7% lower than with control siRNA; hydroxyproline/collagen, ACTA2 and pERK fell; secondary, antral and total healthy follicles and ovulated oocytes increased; and atretic follicles decreased. Rat results converged: lower stiffness and collagen/ACTA2/pERK, more secondary/antral/total healthy follicles and corpora lutea, fewer atretic follicles, and larger litters.

Pregnancy proportions moved in a favorable direction but the reported significance is not credible under the stated Fisher’s exact test: mouse source data show 3/8 versus 6/8 pregnancies (figure P=0.0004; recalculated two-sided Fisher P≈0.315), and rat data show 3/10 versus 6/10 (figure P=3.34×10−5; recalculated P≈0.370). Mouse litter size increased among pregnancies, but the caption says n=3 versus 5 whereas the workbook contains three control and six treated litter values. These discrepancies preclude treating pregnancy success as a statistically established effect.

Local ovarian-bursal AAV-shIl11ra1 also lowered stiffness and collagen and increased primary, secondary, antral and total healthy follicles and litter size after four weeks. This supports an ovarian-local component, but the absent vector/dose/control details and lack of biodistribution or leakage testing prevent a definitive claim of ovary-restricted action. AAV gene silencing is not an antibody experiment and was not tested in people.

Null, negative and unresolved findings

• Recombinant IL-11 did not change body weight, most organ indices, or serum CK, ALT, LDH-L and creatinine; heart index decreased.
• In aged Il11ra1−/− mice, primordial, primary, antral and corpus-luteum counts were not significant, and total healthy follicles narrowly missed conventional significance (P=0.0503).
• Il11ra1 loss did not improve two-cell or blastocyst development rates in the aging experiment and did not change fertilization/early-development rates in the doxorubicin or DHEA models.
• Doxorubicin-associated ovarian stiffness remained higher than untreated controls even in knockout mice, indicating partial rather than complete protection.
• Collagen III was mostly unchanged in aging and pathological human ovaries; the matrix signal was dominated by collagen-I-rich fibers.
• No study arm tested an anti-IL-11 antibody in an intact ovary, animal or person. The antibody evidence was confined to primary-human-fibroblast culture.
• Pregnancy counts in the adult mouse and rat siRNA arms favored siIl11, but neither contrast is significant when recalculated with the stated two-sided Fisher’s exact test; the much smaller printed P values are irreconcilable with the source counts.
• Long-term persistence of reproductive/endocrine benefits, offspring outcomes, menopause timing and reversal of established ovarian fibrosis remain unknown. gap/long-term-unknown gap/no-mechanism

Safety and adverse observations

Short recombinant-IL-11 exposure left body weight and most serum organ-injury markers unchanged but decreased heart index; this was not investigated mechanistically. The siRNA and AAV arms did not report a dedicated systemic toxicology, biodistribution, immunogenicity or adverse-event package. Tail-vein nanoparticles were explicitly non-ovary-specific and could act in multiple organs. The local AAV experiment reduced that particular confound but introduced gene-therapy-specific safety and persistence questions.

The article’s statement that the siRNA/liposome approach is FDA-approved should not be read as approval of this siIl11 cargo and formulation; no clinical regulatory or safety dataset for the actual ovarian-aging intervention was presented.

Extrapolation to humans

Dimension	Status	Notes
Stiffness/ECM phenotype observed in humans?	yes, observational	AFM and collagen assays were performed in age-stratified and disease-associated surgical tissue; sampling was cross-sectional and cancer-surgery enriched.
IL-11 mechanism supported in human material?	partial	Human ovarian expression and primary fibroblast perturbations support TGF-β1→IL-11→ERK→collagen, but no intact human ovary was treated.
Organ-level rescue replicated across species?	rodent only	Genetic and adult silencing results converged in mice; siIl11 nanoparticles also improved rat endpoints.
Human fertility or endocrine benefit demonstrated?	no	No participant received IL-11 inhibition, and no pregnancy, live-birth or menopause endpoint was tested clinically.

The human cohorts establish association, not therapeutic efficacy. Surgical ascertainment, underlying cervical/endometrial cancer and narrow reproductive-phase inclusion limit population generalizability. A clinical translation path would require ovary-targeted exposure, reproductive and systemic safety, durable endocrine/fertility outcomes and evidence that established human fibrosis can be modified. gap/needs-human-replication

Limitations and conflicts

• Prevention versus reversal: the authors describe the 36-week mouse intervention as a model of rapid ovarian decline and explicitly state that it tested prevention of fibrosis progression. Older animals with established fibrosis are needed before claiming reversal.
• Developmental confounding: Il11ra1 deletion was germline in the aging, POI and PCOS arms. Prior reports associate complete Il11 or Il11ra1 loss with infertility; the authors distinguish adult partial silencing and report normal fertility in heterozygotes, but homozygous-developmental selection remains a major interpretive caveat.
• Systemic delivery and underreported reagents: tail-vein siIl11 nanoparticles were not ovary-targeted, so ovarian improvement could partly reflect extra-ovarian effects. The siRNA mass/concentration and control-siRNA sequence were omitted. AAV-bursal delivery supports a local component but did not reproduce the siRNA intervention; the article supplied no AAV Methods section, serotype, construct/promoter, shRNA sequence, titer/genome dose, injection volume, vendor, control-vector identity or biodistribution assessment.
• Small analyzed subsets: umbrella cohorts of 25–30/group often contributed only n=5–6 to individual endpoints; snRNA-seq used n=3 animals/group. No statistical method was used to predetermine sample size.
• Allocation and blinding: Methods broadly say animals were randomized, but the Reporting Summary specifies random assignment only for WT mice and rats; transgenic animals were grouped by genotype. Investigators were not blinded during allocation, experiments or outcome assessment.
• Proteomic reproducibility: the raw TMT mass-spectrometry files were irretrievably lost after physical hard-drive failure and had no complete off-site backup. Processed DEP lists and enrichment/source files remain, but raw-data reprocessing is impossible. DEP calls used nominal P≤0.05 and |fold change|≥1.2 without multiple-testing adjustment; pathway enrichment did use BH-FDR correction.
• Human observational design and reporting: no longitudinal human stiffness trajectory or human intervention was performed. Cancer-related surgery, chemotherapy selection in POI, and small PCOS and endometriosis cohorts create selection and confounding risks. The age-matched control group and FAPI-PET sample size were not characterized, and primary-fibroblast donor ages, donor count and donor-to-replicate mapping were not reported.
• Measurement/reporting inconsistencies: AFM was performed on frozen sections rather than living ovaries. The publisher files do not resolve the daily-versus-twice-weekly recombinant-IL-11 schedule, the Results-versus-source-data follicle direction, the 2-µg/ml figure-versus-80-µg/ml-ND50 Methods anti-IL-11 values, the SCH772984-versus-U0126 ERK-inhibitor identity, the Pearson-versus-Spearman correlation method, or the matrix-culture sample sizes. The adult-siRNA pregnancy P values and mouse litter n also disagree with the source workbook.
• Exclusions: Methods say no data were excluded from analyses, whereas the Reporting Summary says human samples not meeting diagnostic criteria were excluded; this appears to refer to eligibility screening, but the wording is not reconciled.
• Endpoint breadth: many histological, hormonal and reproductive outcomes were tested without a study-wide multiplicity framework. Serum sex steroids were measured by ELISA; the authors call for LC–MS/MS in longer-term studies.
• Conflicts: the authors declared no competing interests.
• Publication integrity: publisher, PubMed and PMC searches through 2026-08-09 found no correction, retraction, expression of concern or superseding primary study. Later commentary does not independently replicate the work.

gap/needs-replication gap/needs-human-replication gap/long-term-unknown gap/dose-response-unclear

Related pages

• ovary > Ovarian microenvironment aging and ovarian-stromal-fibroblasts
• il-11, il-11-receptor-alpha-1 and il-11-signaling
• tgf-beta, ras-mapk and extracellular-matrix-remodeling
• premature-ovarian-insufficiency, polycystic-ovary-syndrome and endometriosis
• il-11-gene-silencing
• anti-il-11-antibodies — target-class context only; no antibody was tested as an ovarian intervention in this study
• amargant-2020-ovarian-stiffness and ng-2021-il11-il11ra1-loss-of-function
• schafer-2017-il11-cardiovascular-fibrosis and widjaja-2024-il11-healthspan-lifespan

Cited by wiki pages

[Auto-populated by Obsidian backlinks panel]