Moderate exercise protects against osteoarthritis in mice through gut microbes

Moderate-intensity continuous treadmill exercise reduced cartilage damage, pain sensitivity and synovial inflammation in a mouse model of osteoarthritis, according to a new experimental study. The benefit was not reproduced by high-intensity continuous training and was largely abolished when the gut microbiota was depleted with antibiotics, indicating that the exercise effect depends at least partly on microbial changes. Mice receiving gut microbes transplanted from moderately trained donors also showed less osteoarthritis, while direct supplementation with Akkermansia muciniphila improved cartilage and bone measures. The findings link a specific microbial metabolite, lactate, to the receptor GPR81 and a signalling cascade that promotes cartilage breakdown.

The study used surgically induced osteoarthritis in male C57BL/6J mice through destabilization of the medial meniscus, or DMM. After two weeks of recovery, animals underwent either eight weeks of moderate exercise at 15 metres per minute for 60 minutes per session, six days a week, or high-intensity exercise at 25 metres per minute for 60 minutes per session. Moderate training increased the pain withdrawal threshold, reduced histological cartilage damage and synovitis, and lowered measures of abnormal subchondral bone volume and trabecular thickness. High-intensity training produced no significant improvement in these outcomes, although it also did not measurably worsen osteoarthritis in this experiment. The results therefore distinguish the benefits observed at the tested moderate intensity from those observed at the higher intensity rather than claiming that all vigorous exercise is harmful.

Microbial experiments provided the first mechanistic test. Investigators collected faeces from sedentary and moderately trained mice and transferred them into antibiotic-treated DMM recipients by oral gavage for eight weeks. Recipients of the MICT-derived microbiota had greater pain withdrawal thresholds, lower Osteoarthritis Research Society International cartilage-damage scores, less synovial inflammation and lower abnormal subchondral bone measures than recipients of saline or microbes from sedentary donors. They also showed more collagen type II alpha 1 and aggrecan, two markers associated with cartilage matrix production, and less matrix metalloproteinase 13 and collagen type X alpha 1, markers associated with matrix degradation and chondrocyte hypertrophy. The complementary depletion experiment produced a consistent result: antibiotics did not cause the exercise programme itself to worsen disease, but they erased much of MICT’s improvement in pain, cartilage damage, inflammation, bone structure and cartilage-marker expression.

Sequencing of 16S ribosomal RNA genes showed that MICT changed the composition of the caecal microbiota without significantly altering the measured within-group diversity indices. At the phylum level, MICT reduced the relative abundance of Bacteroidota and Desulfobacterota and increased Verrucomicrobiota. At the family level, Akkermansiaceae was the most responsive group to training in DMM mice. Correlations linked several DMM-enriched or exercise-responsive taxa with circulating lactate, lipopolysaccharide, interleukin-1β, collagen type II alpha 1 and matrix metalloproteinase 13, but correlation alone cannot establish which organisms caused which metabolic changes.

One of the clearest metabolic findings was that DMM increased circulating lactate, whereas MICT lowered it. Antibiotic treatment also reduced lactate in DMM animals, supporting a microbial contribution to systemic lactate levels, although the rise in inflammatory markers was more complicated: both MICT and antibiotics lowered serum interleukin-1β and lipopolysaccharide, and microbial depletion did not abolish these reductions. Blood was collected 48 hours after the final exercise session, when serum lactate was expected to have returned near baseline after acute exertion, limiting the contribution of short-lived exercise-induced lactate. The study therefore focused on persistent changes associated with the DMM environment and microbiota rather than on the immediate metabolic spike during a workout.

Akkermansia was then tested directly in antibiotic-treated DMM mice. Oral administration of 2 × 10⁸ colony-forming units per dose for eight weeks improved pain-related behaviour, cartilage histology, synovial inflammation and subchondral bone measures. It increased the cartilage markers collagen type II alpha 1 and aggrecan while reducing matrix metalloproteinase 13 and collagen type X alpha 1. These results place Akkermansia downstream of MICT in the proposed pathway, but they do not establish that the entire benefit of exercise is caused by this single genus. The experiments also do not show that the benefits of MICT in mice would be reproduced in people with osteoarthritis.

Cell experiments and injections into mouse knees identified lactate as a harmful signal in this setting. Lactate inhibited chondrogenic differentiation in a dose-dependent manner, reduced Col2a1 expression and increased Mmp13 expression in cells exposed to interleukin-1β. Weekly intra-articular injections of 10 mM lactate worsened cartilage damage and synovial inflammation after DMM surgery, reduced collagen type II alpha 1 and aggrecan, and increased matrix metalloproteinase 13 and collagen type X alpha 1. Lactate also increased Gpr81 expression in cartilage. These observations concern sustained or repeated lactate exposure in experimental systems and do not imply that lactate has the same effect in every tissue or physiological context.

Two genetic strategies supported GPR81 as a functional mediator. Global Gpr81 knockout preserved more cartilage matrix, reduced matrix metalloproteinase 13 and collagen type X alpha 1, and improved pain, cartilage damage and synovitis in DMM mice. The knockout also prevented lactate from suppressing chondrogenic differentiation and shifting Col2a1 and Mmp13 expression. More locally, an adeno-associated virus 2 vector carrying a Gpr81 short hairpin RNA was injected into the knee. The vector reduced GPR81 expression in articular cartilage and likewise improved pain, histology, inflammation, subchondral bone structure and the balance between anabolic and catabolic cartilage markers. These complementary models implicate GPR81 in cartilage metabolism, but the global knockout could also affect organs beyond the joint, while the local knockdown was tested only in mice.

Mechanistic assays placed GPR81 in a Gαi–cAMP–protein kinase A pathway linked to CREB and NF-κB. Blocking Gαi with pertussis toxin prevented lactate from worsening osteoarthritis in DMM mice and stopped lactate from increasing phosphorylation of the NF-κB p65 subunit. Raising cAMP with forskolin counteracted lactate-driven changes in chondrocyte metabolism, whereas inhibiting protein kinase A with H89 removed the metabolic advantages observed in Gpr81-deficient chondrocytes. Blocking CREB prevented the increased matrix production associated with Gpr81 deletion, while blocking NF-κB prevented lactate-induced catabolism. The proposed mechanism is that lactate activates GPR81, with Gαi signalling influencing CREB-dependent cartilage anabolism and NF-κB-dependent catabolism; the authors note that the pathway requires further transcriptomic and proteomic analysis.

Human samples provided clinical association rather than clinical proof of treatment. The researchers compared nine people aged 70–80 with primary knee osteoarthritis and 15 age-matched controls without joint symptoms or radiographic osteoarthritis, collecting faecal samples, synovial fluid and cartilage. People with osteoarthritis had less Akkermansia and more GPR81-expressing chondrocytes in damaged cartilage. Within the osteoarthritis group, faecal Akkermansia abundance correlated negatively with synovial-fluid lactate and cartilage GPR81 expression, while lactate correlated positively with GPR81 expression. The strongest reported correlation was between lactate and GPR81, with R² = 0.7817, compared with R² = 0.5697 for Akkermansia and GPR81 and R² = 0.2787 for Akkermansia and lactate. These small cross-sectional data are consistent with the animal mechanism but cannot establish temporal order, causation or whether exercise alters these measures in patients. Overall, the study identifies a testable microbiota–lactate–GPR81 mechanism for the cartilage protection observed with moderate training in mice, rather than demonstrating an effective human osteoarthritis therapy.

Subject of Research: Mechanistic mouse study of moderate exercise, gut microbiota, lactate and GPR81 in osteoarthritis.

Article Title: The inhibition of lactate/Gpr81 signalling driven by moderate-intensity continuous training protects against osteoarthritis

Article References: Sun, P., Guo, Y., Xue, A., Ni, J., Ren, Y., Sun, Y., Qian, X., Li, X., You, Y., Zhao, Y., Tang, L., Zhu, S., Li, Y., & He, L. (2026). The inhibition of lactate/Gpr81 signalling driven by moderate-intensity continuous training protects against osteoarthritis. Experimental & Molecular Medicine, 58(9), 2951-2971. https://doi.org/10.1038/s12276-026-01819-8

Image Credits: AI Generated

DOI: 10.1038/s12276-026-01819-8

Keywords: osteoarthritis, moderate exercise, gut microbiota, Akkermansia, lactate, GPR81, cartilage degeneration, mouse model, chondrocytes, microbiome, inhibition, signalling