Adipose-derived leptin and complement factor D mediate osteoarthritis severity and pain

Obesity is a risk factor for osteoarthritis (OA), and leptin is among the adipokines implicated in obesity-induced OA. However, the specific role of leptin in OA severity and pain is not known. Using lipodystrophic (LD) mice, we show fat-secreted factors are required for knee OA development, implicating a fat-cartilage crosstalk. Fat pad implantation or systemic leptin restoration in LD mice reintroduced structural OA and pain, whereas implantation of leptin-deficient fat pad did not change OA susceptibility. Isochronic parabiosis and spatial transcriptomics confirmed that fat-joint crosstalk likely occurred via soluble mediators. Global unsupervised multiomics of conditioned media from fat implants revealed that leptin exerts a regulatory effect on adipsin (or complement factor D), the activity of which modulates contrastive OA structural and pain phenotype. These findings suggest adipokines influence OA pathogenesis, providing conclusive evidence of a fat-joint crosstalk and implicating OA as a systemic disease of adipose tissue.

Methods
Animal Studies

All experimental procedures were approved by the Washington University School of Medicine Department of Comparative Medicine Institutional Animal Care and Use Committee (WUSTL IACUC 22-0306) and were conducted in accordance with ARRIVE guidelines. LD mice and WT (DTA/+) littermate controls were generated for these studies by crossing adiponectin-Cre (Jackson Labs: 028020) mice with homozygous lox-stop-lox-ROSA-diphtheria toxin A (Jackson Labs: 010527) mice on a C57BL/6 background. To breed at mendelian frequency, mice were bred and maintained at thermoneutrality throughout their lifespan (30 °C). Mixed genotype mice were group-housed such that n = 3 to 5 per cage.

Male LD mice received either a MEF transplant (MEF-rescue) from WT, leptin heterozygous (MEF-HET) or leptin homozygous knockout (MEF-KO) pups as previously described, which developed into adipose-like tissue between 3 and 5 weeks of age. The injection was delivered subcutaneously to the sternal aspect of a donor LD mouse (3–5 weeks old) under 2% isoflurane with a 27-gauge needle. Animal numbers used in each experiment can be found in the figure legend.

Exogenous Leptin Delivery

Osmotic pumps were surgically implanted to the dorsal aspect of mice at 15 weeks of age, one week prior to DMM. Leptin or saline was delivered to LD mice using Azlet 2006 (Cupertino, CA) at a release rate of 0.15 to deliver a total of 1mg of leptin over 6 weeks, and were replaced with new, sterile implants until 28 weeks of age. Leptin concentration was confirmed after implantation.

OA induction by DMM

At 16 weeks of age, surgical animals underwent destabilization of the medial meniscus surgery in their left knee joints, and the right limb served as a nonsurgical contralateral control. Behavioral and metabolic assays were conducted throughout the course of this study. Mice were sacrificed at 28 weeks of age to assess OA severity, bone microstructure, and serum and SF inflammatory profiles. Knee joints were fixed in 4% paraformaldehyde and decalcified in 10% formic acid  (Cal-Ex II) for 48-72 hrs. Histological assessment was performed by Modified Mankin Scoring, Synovitis Scoring, and Osteophyte scoring on formalin fixed paraffin embedded 5um sections as previously described. Bone marrow adipocytes were counted in ImageJ on histological sections.

Body Composition, Tissue Harvest and Storage

Mice were weighed weekly throughout the course of the study using a standard scale (g). Body fat was measured by DXA (Lunar Piximus) at 27 weeks of age to quantify body fat. Animals were euthanized at 28 weeks of age in accordance with the timelines in each figure. Body mass, liver mass, body fat, blood, L3-L5 dorsal root ganglia sensory neurons from Cfd-/- and WT mice and synovial fluid was harvested.

Serum and Synovial Fluid Profiling

All mice were fasted overnight prior to sacrifice. Serum was collected in a BD Microtainer STT serum separator tube and allowed to clot at room temperature. Synovial fluid was collected using an alginate pad that was digested in alginate lyase until a stop solution was applied. Both fluids were stored at −80 °C until analysis by Luminex multiplex 18-plex chemokine/cytokine array assay (Eve Technologies, Calgary). Enzyme-linked immunosorbent assay (ELISA) for serum leptin was conducted using a mouse/rat Leptin Quantikine ELISA Assay kit (MOB00; R&D Systems) at a dilution of 1:2 in LD mice, and all other groups were evaluated at the manufacturer’s recommended 1:20 dilution.

Bone Microstructure Analysis

Whole knee joints were scanned by microcomputed tomography (Bruker SkyScan1176) at a resolution of 18-μm isotropic voxel resolution according to previously reported methods. To reduce beam hardening, a 0.5-mm aluminum filter was used during scanning. Hydroxyapatite calibration phantoms were scanned to calibrate bone density. Scans were reconstructed to three-dimensional (3D) images using NRecon software, and CTAn software was used to segment subchondral and trabecular regions from the medial tibial plateau, lateral tibial plateau, medial femoral condyle, and lateral femoral condyle for analysis. The tibial epiphysis was identified using the subchondral plate and growth plate as references. The tibial metaphysis was defined as the 1-mm area directly below the growth plate. The main outcomes reported from microCT images are BMD, BV/TV (%), trabecular number, and trabecular thickness.

Insulin and glucose tolerance tests

Insulin- and glucose-tolerance tests were performed 4 weeks post-DMM surgery (20 weeks of age) after fasting mice for a minimum of 4 hours. Each test was conducted at least 1 week apart in all animals. For both tests, fasting glucose levels were measured by tail bleed at time point 0 to establish a baseline. For glucose-tolerance tests, animals were administered 1 mg/kg (10% dextrose: 1% volume/body mass) by intraperitoneal (IP) injection, and for insulin-tolerance tests, 0.75 U/kg body mass of insulin (Humulin R diluted to 75 mU/mL, 1% volume/body mass) was administered by IP injection. Serial blood glucose measurements were taken via tail vein at 20, 40, 60, and 120 min after injection with a glucose meter (Contour; Bayer).

Pain Assessments and Behavioral Testing

All animals were acclimatized to all equipment 1 d prior to the onset of testing. Two measures for pain were conducted at 27 weeks of age. To assess tactile allodynia in the DMM limb, an Electronic Von Frey assay was used as previously described(8). Hind paws of DMM limbs were stimulated three to five times. The intensity of the stimulus (grams) was recorded by the tester when the paw was withdrawn. To assess mechanical hyperalgesia, pressure-pain tests were conducted using a Small Animal Algometer (SMALGO) (Bioseb). Three to five trials of the surgical and nonsurgical limb were collected by applying a steadily increasing force to the lateral aspect of each limb until the limb was withdrawn. The average of three trials for each limb was reported, and maximum value of 450 g was employed to avoid tissue damage to the knee joint. 

Isochronic Parabiosis

Female LD and WT littermates were housed as a pair from weaning until 8-10 weeks of age, where we performed isochronic parabiosis. At 16 weeks, the LD mouse was challenged with DMM surgery unilaterally on their L outside limb. Mice were monitored for 12 weeks and euthanized at 28 weeks of age. Serum and hindlimbs were collected for further analysis.

Spatial Transcriptomics and Knee Joint Assessment 

Samples were fixed in 4% PFA for 24h from mice that were joined by isochronic parabiosis. A reduced decalcification time of 10h at 4C on an orbital shaker was used for knee joints. Joint tissues were processed according to previous protocols and cut on a Leica Microtome at 5uM. Serial sections were used for staining with hematoxylin & eosin, toluidine blue/oil red O, and an unstained slide onto Fisher superfrost plus slides or Dako Flex IHC adhesive slides, and prepared for spatial transcriptomics by 10x Visium using the standard CytAssist protocol. The concentration of each library was accurately determined through qPCR utilizing the KAPA library Quantification Kit according to the manufacturer’s protocol (KAPA Biosystems/Roche) to produce cluster counts appropriate for the Illumina NovaSeq6000 instrument.  Normalized libraries were sequenced on a NovaSeqX 10B or NovaSeq6000 S4 Flow Cell using the 151x10x10x151 sequencing recipe according to manufacturer protocol.  Read 1 was trimmed to the 10x Genomics recommendation of 28bp. A median sequencing depth of 50,000 reads/cell was targeted for each Gene Expression Library.  The data were then spatially analyzed using standard graph-based clustering methods and visualized using 10x Space Ranger and 10x Loupe browser version 7.

Metabolomics and Proteomics 

Fat was explanted, weighed, washed, filtered, and 100mg was prepared for bulk sequencing (n=3-5/group), while 250mg was cultured for 24h in 1% Dulbecco’s Modified Eagle’s Medium. The resulting conditioned media (CM) was collected and compared to CM derived from naïve WT fat of equal mass (250mg) (n=4-8/group). CM protein and metabolites were extracted and evaluated via liquid chromatography-mass spectrometry (LC-MS)(62). Significant metabolites and protein intensities were detected using clustering and false-discovery-rate corrections in MetaboAnalyst (p<0.05). Data were log transformed and compared using hierarchical cluster analysis (HCA), volcano plot analysis, principal component analysis (PCA) partial least-squares discriminant analysis (PLS-DA), and variable importance in projection (VIP) scores.