Inhibition of a signaling modality within the gp130 receptor enhances tissue regeneration and mitigates osteoarthritis

Abstract

Adult mammals are incapable of multitissue regeneration, and augmentation of this potential may shift current therapeutic paradigms. We found that a common co-receptor of interleukin 6 (IL-6) cytokines, glycoprotein 130 (gp130), serves as a major nexus integrating various context-specific signaling inputs to either promote regenerative outcomes or aggravate disease progression. Via genetic and pharmacological experiments in vitro and in vivo, we demonstrated that a signaling tyrosine 814 (Y814) within gp130 serves as a major cellular stress sensor. Mice with constitutively inactivated Y814 (F814) were resistant to surgically induced osteoarthritis as reflected by reduced loss of proteoglycans, reduced synovitis, and synovial fibrosis. The F814 mice also exhibited enhanced regenerative, not reparative, responses after wounding in the skin. In addition, pharmacological modulation of gp130 Y814 upstream of the SRC and MAPK circuit by a small molecule, R805, elicited a protective effect on tissues after injury. Topical administration of R805 on mouse skin wounds resulted in enhanced hair follicle neogenesis and dermal regeneration. Intra-articular administration of R805 to rats after medial meniscal tear and to canines after arthroscopic meniscal release markedly mitigated the appearance of osteoarthritis. Single-cell sequencing data demonstrated that genetic and pharmacological modulation of Y814 resulted in attenuation of inflammatory gene signature as visualized by the anti-inflammatory macrophage and nonpathological fibroblast subpopulations in the skin and joint tissue after injury. Together, our study characterized a molecular mechanism that, if manipulated, enhances the intrinsic regenerative capacity of tissues through suppression of a proinflammatory milieu and prevents pathological outcomes in injury and disease.

INTRODUCTION

Mammalian tissue regeneration involves compensatory new growth to completely restore original tissue architecture and function after damage (1). During prenatal development, mammals have an extraordinary ability to regenerate tissue; unfortunately, this capability declines with age, leading to fibrotic scarring and loss of original tissue integrity (2). Thus, it is of fundamental importance to decipher molecular mechanisms controlling degeneration and fibrosis versus regenerative responses after wounding.

Although different species and tissues may share some commonalities in their regenerative processes, the specific factors involved can vary markedly across the animal kingdom. For instance, unlike lower vertebrates, adult mammals have not been shown to regenerate complete limb structures (3). In higher-order mammals, the mechanism responsible for this progressive decline in regenerative abilities is partly due to cell-extrinsic factors, including changes in the milieu (1). During regeneration, inflammation often precedes the actual repair of the lesion in an attempt to reestablish homeostasis after an acute injury (3). Successful tissue regeneration is attributed to controlled inflammatory processes mediated by proinflammatory cytokines such as the interleukin-6 (IL-6) family of cytokines (3). The IL-6 family of cytokines [IL-6, IL-11, oncostatin M (OSM), leukemia inhibitory factor (LIF), and others], which signal through glycoprotein 130 (gp130), can promote both regeneration and pathogenesis upon injury depending on the perseverance of inflammatory signal (4–6). The regenerative capacity of IL-6 cytokines is seen in adult mammals in various organs and tissues (7–12). However, aberrant and prolonged activation of IL-6 cytokines is implicated in the pathogenesis of various fibrotic pathologies (13) and complex chronic diseases such as osteoarthritis (OA), where progressive inflammatory and degenerative changes in articular cartilage are accompanied by the excessive fibroplasia and collagen production in synovial tissue and subchondral bone (14). Human genetic data have also linked the IL-6 family of cytokines with OA (15, 16).

The mechanism underlying the bimodal ability of IL-6/g130 signaling to simultaneously enhance regeneration and promote inflammation and fibrosis is still unclear. We hypothesized that different functional outcomes downstream of gp130 may be mediated by specialized signaling residues of this receptor and that selective manipulation of these gp130 modules could promote regeneration in lieu of pathological inflammation and fibrosis. In the current study, we found that the tyrosine residue Y814 within gp130 serves as a major cellular stress sensor in response to IL-6–related cytokines and that inhibition of this signaling improves regenerative outcomes in multiple tissues.

RESULTS

Genetic knockout of a gp130 modality in vivo results in down-regulation of prodegenerative and profibrotic signaling cascades

Engagement of gp130 by IL-6 family members is required for activation of various downstream signaling cascades, including mitogen-activated protein kinase (MAPK)/extracellular signal–regulated kinases (ERKs), phosphoinositide 3-kinases (PI3Ks)/protein kinase B (AKT), Janus kinase (JAK)/signal transducer and activator of transcription 3 (STAT3), and SRC signaling (17, 18), but the role of each of these binding partners and their interactions with gp130 are context specific and not fully understood. Much work has focused on other pathways downstream of gp130, whereas studies of gp130-dependent SRC signaling in the context of the joint are still lacking. SRC was of an interest to us because previous studies reported SRC to be highly expressed in the synovia of patients with rheumatoid arthritis and constitutively activated in a rat model of collagen-induced arthritis (19). To elucidate the role of SRC signaling in extracellular matrix (ECM) degradation downstream of gp130, we first stimulated pig articular chondrocytes with OSM, a proinflammatory IL-6 cytokine member. We then used a pharmacological inhibitor to block SRC kinase by SU6656, which had a significant effect on OSM-induced transcription of matrix-degrading enzymes A disintegrin and metalloproteinase with thrombospondin motifs (ADAMTS4) (P = 0.003) and matrix metallopeptidase 13 (MMP13) (P < 0.0001) (fig. S1A). To determine which cytokine had the most substantial effect on SRC up-regulation, we stimulated pig articular chondrocytes with the IL-6 family of cytokines. OSM displayed the strongest up-regulation of phosphorylated (active) SRC (pSRC) (P < 0.0001) (Fig. 1A).

Fig. 1. Inhibition of gp130 Y814 signaling leads to down-regulation of prodegenerative and profibrotic pathways.

(A) Western blot data for pSRC (active) in pig articular chondrocytes stimulated with the IL-6 family of cytokines (10 ng/ml), including LIF, ciliary neurotrophic factor (CNTF), OSM, IL-11, hyper IL-6, and IL-6. Total SRC was used for normalization. Data represent means ± SD. n = 4 biological replicates per group. (B) Schematic of modified amino acids within the gp130 812 to 827 domain. Western blot of pSRC in Ba/F3 cells after transfection with the modified or wild-type (WT) gp130 plasmids stimulated with IL-6 (10 ng/ml) for 24 hours. Histone H3 was used for normalization. Horizontal lines with bars show the means ± SD. n = 4 biological replicates (C) Western blots for gp130 pY814 in WT and F814 mouse splenocytes treated with and without OSM, LIF, IL-11, or IL-6 (10 ng/ml) for 24 hours. Total gp130 was used for normalization. Graphs show means ± SD. n = 4 biological replicates. (D) Western blots for immunoprecipitated protein complex formation between gp130 and pSRC in WT and F814 mouse splenocytes stimulated with or without OSM (10 ng/ml) for 24 hours. Total gp130 was used for normalization. Graphs show means ± SD. n = 4 biological replicates. (E) Western blot pSRC in WT and F814 mouse splenocytes stimulated with or without OSM (10 ng/ml) for 24 hours. Total SRC was used for normalization. Graphs show the means ± SD. n = 4 biological replicates. Statistical analysis was performed using one-way ANOVA followed by the Tukey test to compare more than two groups. P values less than 0.05 were considered significant. (F) Distribution of differentially expressed genes in WT and F814 mouse synovial fibroblasts stimulated with or without OSM (10 ng/ml). Log ratio/mean average plots (log₂fold change versus log₂mean expression) for WT-OSM versus WT (left) and Y814-OSM versus Y814 (right) mouse synovial fibroblasts are shown. Each dot represents a gene. Red dots represent up-regulated genes (log₂fold change >1 and P < 0.05). Blue dots represent down-regulated genes (log₂fold change < −1 and P < 0.05). Genes that do not qualify any of these thresholds are indicated by gray dots.

Previously published data identified the acidic domain 812 to 827 of gp130 as the primary region of SRC activation (18). To determine the specific residue responsible for SRC activation, we modified amino acids through sequence substitutions using plasmids in vitro. Using a Ba/F3 cell line, which is completely deficient of gp130, we transfected the cells with wild-type (WT) gp130 or gp130 mutants and stimulated the cells with IL-6 to induce gp130 signaling. Substitution of tyrosine (Y) to phenylalanine (F) at residue Y814 (F814) resulted in the strongest reduction of IL-6–stimulated pSRC relative to the WT gp130, suggesting that Y814 might be responsible for regulation of SRC phosphorylation (Fig. 1B). To confirm this, we used the chondrogenic ATDC5 cell line where gp130 expression is minimal. Transfection of ATDC5 cells with the F814 plasmid reduced pSRC compared with the WT gp130 plasmid when stimulated with OSM (P < 0.0001) (fig. S1B). There was also a decreased expression of MMP13 (P < 0.0001) and ADAMTS4 (P = 0.002) and increased collagen type II (COL2A1) (P = 0.003) transcription in the F814-transfected cells (fig. S1C). These results suggested that the Y814 residue of gp130 could regulate SRC activation and cytokine-induced proinflammatory and profibrotic signaling in chondrocytes.

Because we have demonstrated that Y814 mutation may induce collagen biosynthesis (fig. S1C), we therefore wanted to assess Y814 phosphorylation in developing anabolic fetal mouse joints versus adult joints by generating a polyclonal antibody against phosphorylated (active) gp130 Y814 (pY814) (see Materials and Methods). As expected, pY814 activation was significantly lower in fetal chondrocytes compared with adult chondrocytes (P = 0.0005) (fig. S1D), suggesting that Y814 does not play a major role in rapidly proliferating and highly anabolic primary cartilage cells.

To understand the role of Y814 phosphorylation (activation) in vivo, we generated mutant gp130 Y814 mice (F814) using CRISPR-Cas9, and genetic sequencing was conducted to confirm the mutation. The mutant mice were viable and fertile and exhibited no apparent morphological differences in the musculoskeletal tissues or other organs (fig. S2, A and B). Analysis of the skin (fig. S2C) and immune cells from the spleen and bone marrow (figs. S3 to S7) also found no differences between WT and F814 mice. We also found no detectable pY814 in F814 mouse splenocytes endogenously or in response to 4 or 24 hours of OSM, IL-6, or IL-11 treatment, in contrast to WT cells (Fig. 1C and fig. S8). In addition, coimmunoprecipitation (Co-IP) experiments showed that pSRC did not interact with gp130 in F814-derived splenocytes in the presence or absence of OSM (Fig. 1D), and no increase in pSRC was detected after 4 or 24 hours of stimulation with OSM (Fig. 1E and fig. S9).

To determine differences in gp130 signaling, tumor necrosis factor (TNF), synovial fibroblasts isolated from WT and F814 mice were treated with or without OSM and subjected to RNA sequencing (RNA-seq). Normally, OSM up-regulates a variety of proinflammatory and profibrotic genes. However, RNA-seq results demonstrated that mutation of Y814 markedly reduced expression of these genes, including proteases [Mmps3, Mmps13, Adamts 4, Adamts5, and granzyme B (Gzmb)], cytokines and their receptors (Il-11, Il-6, Osm, Il-1r1/2, Osmr, and Il-17ra), and proinflammatory regulators, including nuclear factor NF-κB subunit 1 (NF-κB1) and prostaglandin endoperoxide synthase (Ptgs2)/prostaglandin endoperoxide synthase 2c (Cox2) in response to OSM (Fig. 1F). Ingenuity pathway analysis (IPA) indicated the role of macrophages, fibroblasts, and endothelial cells in rheumatoid arthritis (−log₁₀P value = 7.3 or P = 5.01187 ×10⁻⁸) (fig. S10), and other proinflammatory and fibrosis-related pathways were affected by this mutation. Log ratio/mean average plots for differential gene expression analysis confirmed that after treatment with OSM, WT mouse fibroblasts up-regulated a number of proinflammatory and profibrotic genes, whereas F814 cells had limited downstream gene expression (Fig. 1F).

STAT3 and Yes-associated protein (YAP) signaling pathways downstream of gp130 play a major role in regeneration, development, and diseases such as OA (20–22). In contrast to the decreased SRC and MAPK/ERK signaling in F814 cells in response to OSM, IL-6, or IL-11, WT and F814 mouse synovial fibroblasts did not show differences in YAP or STAT3 signaling when stimulated with the IL-6 cytokine LIF, a potent activator of STAT3 signaling (figs. S11 and S12) (23). In addition, treatment with LIF did not up-regulate gp130 pY814 in WT splenocytes (P = 0.059), suggesting that LIF does not induce Y814 signaling (Fig. 1C and fig. S8).

F814 mice demonstrate increased resistance to degenerative joint disease and enhanced skin regeneration after injury

To assess the effect of the Y814 mutation on ECM loss, we performed a neoepitope assay (23) using isolated femoral head explants cultured in the presence orabsence of OSM. The results confirmed a decrease in cartilage degeneration in femoral heads isolated from 4-week-old Y814 mice compared with WT as shown by low aggrecanase and collagenase activity (fig. S13). To address whether the Y814 mutation can ameliorate cartilage degeneration in joint disease, we used a destabilization of the medial meniscus murine model (24). F814 mice were partially resistant to surgically induced OA as reflected by reduced proteoglycan loss, synovitis, and synovial fibrosis 6 weeks after surgery (Fig. 2A and fig. S14).

Fig. 2. Gp130 Y814-deficient mice show improved responses to degenerative joint disease and wound healing.

(A) Histological staining and quantitative assessment of cartilage degradation in WT and F814 mouse knee joints 6 weeks after destabilization of the medial meniscus surgery after 3 months. Safranin O delineates proteoglycans (pink). The OARSI scoring system was used to quantify the extent of cartilage damage. Scale bars, 100 μm. Horizontal lines with bars show the means ± SD. n = 6 biological replicates. (B) WT and F814 mouse post-wound day (PWD) 21 wound sections after wound excision at 6 weeks. Representative images are shown for H&E, Picrosirius red, and COL1 staining of PWD 21 wound sections from WT and F814 mice. n = 5 biological replicates. Scale bars,100 μm. (C) Reanalysis of macrophage clusters from WT and F814 mouse skin wounds PWD 14 after wound excision at 6 weeks. Clusters are color coded in each analysis. Dot plots depict gene expression in each macrophage cluster. Dot sizes are proportional to the percentage of cells in each cluster expressing the indicated gene. The contribution of each sample to each cluster is shown as a stacked bar graph (WT, red; F814, blue). Cluster 6 (mainly WT cells) is highlighted by the red dotted line. (D) Reanalysis of fibroblast clusters from WT and F814 skin wounds PWD 14 after wound excision at 6 weeks. Clusters are color coded in each analysis. Dot plots depict gene expression in each fibroblast cluster. Dot sizes are proportional to the percentage of cells in each cluster expressing the indicated gene. The contribution of each sample to each cluster is shown as a stacked bar graph (WT, red; F814, blue). Statistical analysis was performed using two-tailed Student’s t test to compare two groups. P values less than 0.05 were considered significant.

Because proinflammatory signals have the potential to promote tissue regeneration, we wanted to assess this potential in F814 mice in vivo. We therefore used a skin excisional wound model where we accessed and compared the regenerative ability of the WT and F814 mouse skin using the wound-induced hair neogenesis (WIHN) assay (25). Histological hematoxylin and eosin (H&E) analysis showed that the WT wounds were thinner, and very few hair placodes were observed at post-wound day (PWD) 21 (Fig. 2B and fig. S15A). Staining for early hair placode marker P-cadherin and hair follicle bulge cell surface marker CD34 were also negative in the wound (fig. S15B). WT wounds also expressed profibrotic markers α-smooth muscle actin (α-SMA) and COL1 (Fig. 2B and fig. S15B). Picrosirius red staining of the dermis revealed collagen deposition and contraction, both indicative of fibrosis (Fig. 2B and fig. S15B). In contrast, the wound was thicker in F814 mice at PWD21, and several newly formed hair follicles were observed (Fig. 2B and fig. S15A). Alkaline phosphatase activity in mutant mice was also up-regulated relative to the WT (fig. S15A). α-SMA, CD34, and P-cadherin were expressed mostly within the hair placodes and follicles (Fig. 2B and fig. S15B). In addition, the intensity of Picrosirius red and COL1 staining in the wound dermis was lower in the F814 mouse relative to the WT mouse, and the wound was visually not as contracted, suggesting that there was less fibrosis (Fig. 2B and fig. S15A). On PWD 28, the F814 mice showed significantly (P < 0.0001) more regenerated hair follicles and longer hair fiber length (P < 0.0001) than WT mice (fig. S15A). Picrosirius red staining from periarticular subcutaneous adipose fibroblasts indicated that F814 cells were resistant to OSM-induced collagen (COL1/3) deposition compared with WT cells and may therefore contribute to the improved wound healing response (fig. S16).

To interrogate the cellular and molecular basis for the improved wound healing in F814 mice, we conducted single-cell RNA-seq (scRNA-seq) at PWD 14, which is an active phase for wound healing. Live Ter119⁻ cells (nonerythroid lineage) from two or three wounds from each genotype were sorted and processed using 10x Genomics technology. Uniform manifold approximation and projection followed by k-means clustering of the data delineated six different clusters, four of which expressed fibroblast genes (fig. S17). The fibroblast clusters were disproportionately composed of cells from the F814 genotype (fig. S17). Because macrophages have been implicated in wound healing and regenerative responses, we reanalyzed the two hematopoietic clusters after exclusion of CD3⁺ T cells and performed k-means clustering (Fig. 2C). This analysis yielded seven clusters that were annotated on the basis of their biomarkers (Fig. 2C). Of note, several clusters enriched for proinflammatory cytokines and proteins were disproportionately populated by cells derived from WT wounds (Fig. 2C). One of the macrophage clusters present in WT wounds was completely absent in F814 mouse wounds (cluster 6, red dashed rectangle; Fig. 2C), and this population strongly expressed proinflammatory factors, including Nos2, Ptgs2, Il1b, and Tnfa (26, 27). In contrast, a cluster expressing Ear2, Retnla, and Tgfb1, genes associated with anti-inflammatory, proregenerative macrophages (28), was largely composed of cells isolated from F814 wounds (cluster 2, blue; Fig. 2C). These data indicate that, in addition to fewer macrophages being present in F814 wounds, they express markers of a less inflammatory phenotype. Reclustering and k-means clustering of the fibroblasts from both wound types yielded two clusters (Wnt⁺ and Acta2⁺), with each populated primarily by one genotype (Fig. 2D). In the cluster dominated by cells from F814 wounds (Wnt⁺, dark blue), cells expressed genes associated with both papillary (Dpp4) and reticular (Dlk1) fibroblasts (29). This cluster was enriched for genes related to Wnt signaling, including Wnt2 and Sfrp2 (Fig. 2D), which are involved in regenerative fibroblast competency in WIHN (30). In contrast, the cluster that comprised mostly cells from WT wounds (light pink) was enriched for genes associated with myofibroblasts, including Acta2, Myl9, and Tagln (Fig. 2D) (31). These data suggest that fibroblasts present in F814 wounds have increased capacity to support regenerative wound resolution. Together, these results suggest that the enhanced healing observed in the F814 mouse may be due to reduced inflammatory macrophage polarization and fibroblast activation.

A SRC-targeting peptide disrupts physical interaction of gp130 and SRC and ameliorates bone matrix degeneration

We have shown thus far that gp130 Y814 can regulate SRC downstream signaling. Because SRC is known to associate with gp130, we hypothesized that gp130 and SRC physically interact on or near Y814 and that hindering this interaction could disrupt activated (phosphorylated) SRC signaling. We performed a peptide library screen with short, overlapping peptides surrounding the gp130 Y814 residue (4 to 15 amino acids) and tested for pSRC down-regulation. Pig articular chondrocytes were stimulated with or without OSM in the presence or absence of these peptides. Immunoprecipitation experiments showed that a 4–amino acid peptide containing a phosphorylated tyrosine residue (QQ[pY]F) was the most efficacious at physically hindering gp130-SRC protein-protein interaction in pig articular chondrocytes stimulated with OSM for 4 or 24 hours (Fig. 3A and fig. S18A). Docking for peptide-protein interaction interfaces in silico indicated that peptide QQpYF had a strong binding affinity to SRC as shown by the spatial proximity through protein folding in resolved three-dimensional structure (Fig. 3B), which limited SRC phosphorylation and downstream MAPK/ERK signaling (Fig. 3C and fig. S18B). In contrast, OSM-dependent AKT signaling and LIF-dependent STAT3 and YAP1 signaling were largely unaffected at either 4 or 24 hours after stimulation (Fig. 3C and fig. S18B). Cytotoxicity of peptide QQpYF was assessed by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay in pig articular chondrocytes, and no differences in viability were observed (fig. S19).

Fig. 3. QQpYF peptide prevents gp130-SRC signaling to ameliorate degenerative outcomes in human and porcine samples.

(A) Western blots for immunoprecipitated complex formation between gp130 and pSRC in pig articular chondrocytes stimulated with or without OSM (10 ng/ml) in presence or absence of control scrambled peptide (300 μg/ml) or peptide QQpYF (300 μg/ml) for 24 hours. Total gp130 was used for normalization. Graphs show the means ± SD. n = 4 biological replicates. (B) Computational analysis using GOLD software predicted a high-affinity binding of peptide QQpYF to the regulatory site of SRC (c-SRC). c-SRC as visualized by crystallography. The structure of the indicated c-SRC domains is shown in ribbon diagram representation (left) and with electrostatic potential (blue, positive charge; red, negative charge; white, neutral) mapped onto the molecular surface (right). Peptide QQpYF is shown in stick representation. (C) Western blots in pig articular chondrocytes stimulated with or without OSM (10 ng/ml) in presence or absence of control scrambled peptide (300 μg/ml) or peptide QQpYF (300 μg/ml) for 24 hours. Total respective proteins were used for normalization. Graphs show the means ± SD. n = 4 biological replicates. (D) qPCR analysis of human adult osteoarthritic articular chondrocytes treated with or without peptide QQpYF (300 μg/ml) for 48 hours. Graphs show the means ± SD. n = 3 biological replicates. (E) Pig knee cartilage explants were stimulated with or without OSM (10 ng/ml) and treated with or without peptide QQpYF at indicated doses for 72 hours, followed by a neoepitope assay and Western blot. Amounts of cleaved aggrecan (ACAN) and collagen II (COL2) neoepitopes in the supernatant were quantified with respect to the wet weight of the explant. Graphs show the means ± SD. n = 3 biological replicates. Statistical analysis was performed using one-way ANOVA followed by the Tukey test to compare more than two groups or two-tailed Student’s t test to compare two groups. P values less than 0.05 were considered significant.

To determine the effect of peptide QQpYF on matrix degeneration, we performed quantitative polymerase chain reaction (qPCR) to quantify gene expression of matrix-degrading enzymes in human adult OA articular chondrocytes. Transcription of ADAMTS4 (P = 0.001) and MMP13 (P = 0.0004) was markedly lower in cells treated with peptide QQpYF, whereas COL2A1 (P < 0.0001) and aggrecan (ACAN) (P = 0.074) mRNAs were increased (Fig. 3D). In addition, pig articular cartilage explants were stimulated with or without OSM and treated with or without peptide QQpYF at varying doses to assess ACAN and COL2 neoepitope generation. Significant decreases in aggrecanase (P = 0.003) and collagenase (P = 0.001) activity in QQpYF-treated cells suggested that the peptide had a protective effect against matrix degeneration (Fig. 3E).

The gp130-targeting small-molecule R805 promotes tissue regeneration after injury

We previously identified a small molecule capable of modulating gp130 receptor signaling, RCGD 423 (23). RCGD 423 prevents activation of MAPK/ERK and NF-κB pathways and demonstrates strong anti-inflammatory and antidegenerative outcomes (23). However, this molecule was also shown to activate STAT3 signaling and its downstream target, proto-oncogene MYC (23), which limits its clinical translation. Therefore, it was imperative to find an analog with similar anti-inflammatory properties that does not activate MYC signaling. Analysis of human primary OA chondrocytes has demonstrated greater pSTAT3 (P = 0.015) up-regulation compared with adult articular chondrocytes from healthy joints (fig. S20). We have previously shown that STAT3 signaling is highly up-regulated in rapidly growing, anabolic fetal chondrocytes (23), but this may not be necessary for therapeutic benefit in OA. We have developed an optimized analog of RCGD 423, termed R805 (Fig. 4A), and used it for further characterization because of its minimal effect on c-MYC in pig chondrocytes (fig. S21) or STAT3 signaling in human OA chondrocytes (fig. S22). Similar to RCGD 423 (23), R805 suppressed heterodimerization of gp130 with its α receptors OSMR, LIFR, and IL-11R in a dose-dependent fashion (figs. S23 and S24). R805 also inhibited proinflammatory signaling in human OA chondrocytes (fig. S22) and OSM-stimulated phosphorylation of gp130Y814 in pig chondrocytes (Fig. 4B and fig. S25A). We also found that R805 limited phosphorylation of gp130 Y759 (P < 0.0001), which is involved in Src homology-2 domain–containing protein tyrosine phosphatase-2 (SHP2)/MAPK activation, but did not have an effect on Y905 (P = 0.996), where the JAK/STAT3 complex is primarily recruited (Fig. 4B and fig. S25A). This suggests that, although the compound is not a selective inhibitor of the Y814 residue, it can nonetheless shift gp130 receptor activation of SRC and MAPK. To measure the effect of R805 on matrix destruction, we treated pig articular chondrocytes with R805 after stimulation with OSM. R805 treatment reduced MMP13 (P = 0.0004), ADAMTS4 (P = 0.0004), and ADAMTS5 (P = 0.067) gene expression (fig. S25B) as well as COL2 (P < 0.0001) and ACAN (P < 0.0001) neoepitopes in OSM-treated porcine knee explants (fig. S25C).

Fig. 4. The small-molecule R805 targets gp130 Y814 signaling to promote wound healing in mice.

(A) Chemical structure of R805. (B) Western blots for gp130 pY814, pY905, and pY759 in pig articular chondrocytes treated with or without OSM (10 ng/ml) and R805 (1 or 10 μM/ml) for 24 hours and normalized to total gp130. Graphs show the means ± SD. n = 4 biological replicates. (C) Mouse vehicle-treated (saline) or R805-treated (20 μM/ml) PWD 21 skin wound sections after wound excision at 6 weeks. Representative images are shown for alkaline phosphatase, H&E, and Picrosirius red staining. Magnification, ×10. Scale bars, 100 μm. n = 8 biological replicates. (D) Hair follicle (n = 3 biological replicates) and fiber length (n = 10 biological replicates) in R805-treated (10 μM or 20 μM/ml) mouse PWD 14 and untreated control wounds after wound excision at 6 weeks. Graphs show the means ± SD. (E) Reanalysis of macrophage clusters from skin wounds of vehicle-treated (saline) or R805-treated (20 μM/ml) mice PWD 14 after wound excision at 6 weeks. Clusters are color coded in each analysis. Dot plots depict gene expression in each macrophage cluster. Dot sizes are proportional to the percentage of cells in each cluster expressing the indicated gene. The contribution of each sample to each cluster is shown as a stacked bar graph (WT, red; R805, blue). Cluster 6 (mainly WT cells) is highlighted by the red dotted line. (F) Reanalysis of fibroblast from skin wounds of vehicle-treated (saline) or R805-treated (20 μM/ml) mice PWD 14 after wound excision at 6 weeks. Clusters are color coded in each analysis. Dot plots depict gene expression in each fibroblast cluster. Dot sizes are proportional to the percentage of cells in each cluster expressing the indicated gene. The contribution of each sample to each cluster is shown as a stacked bar graph (WT, red; R805, blue). Statistical analysis was performed using one-way ANOVA followed by the Tukey test to compare more than two groups. P values less than 0.05 were considered significant.

To compare the signaling changes after R805 treatment with the gp180 Y814F mutation, we assessed pSRC, MAPK (ERK 1/2), and NF-κB in response to OSM, IL-6, or IL-11, whereas pSTAT3 and YAP1 were evaluated in response to LIF. Although pSRC, MAPK (pERK 1/2), and pNF-κB (pNF-κBp65) were markedly suppressed by R805 after 4 and 24 hours of stimulation by OSM, IL-6, or IL-11 (figs. S26 and S27), the drug had minimal effect on pSTAT3 and YAP1 downstream of LIF (fig. S28) in pig articular chondrocytes. Furthermore, R805 was able to limit SRC (P = 0.046) binding to gp130 (fig. S29). These data suggest that R805 treatment, like the Y814F mutation, can modulate gp130 signaling to favor proregenerative signaling and limit activation of pathological signaling pathways.

To test whether R805 can foster tissue regeneration, we again used a skin excisional wound model. On PWD 21, superior hair follicle neogenesis and dermal regeneration were observed after topical administration of R805 relative to control (Fig. 4, C and D). Histological H&E observations showed that the vehicle-treated mouse wound was thinner, with fewer placodes (Fig. 4C). Picrosirius red, α-SMA, and COL1 staining in the dermis all revealed high collagen deposition and contraction, indicative of fibrosis (Fig. 4C and fig. S30). This was in contrast to R805-treated wounds, which were thicker and less contracted, with less collagen deposition (Fig. 4C and fig. S30). Higher expression of P-cadherin, CD34, and alkaline phosphatase activity was also observed in R805-treated wounds relative to the vehicle-treated wounds (fig. S30). Increased hair fiber length (P < 0.0001) and hair follicle number (P = 0.0005) (Fig. 4D) also suggested that R805 treatment led to improved wound healing in a concentration-dependent manner.

We next conducted scRNA-seq analysis on vehicle and R805-treated mouse skin wounds. Unsupervised k-means clustering defined a clear macrophage subset (gene expression for each cluster is depicted as dot plots) (fig. S31, purple; and table S1), and reclustering of this population identified 10 macrophage subsets (Fig. 4E). The macrophage cluster (yellow) contained more cells from R805-treated wounds and had higher expression of genes characteristic of proregenerative/anti-inflammatory macrophages, including Mrc1, Trem2, and Tgfb1 (Fig. 4E) (32). In contrast, macrophage cluster 6 (red) was populated primarily by cells from control wounds and had more proinflammatory gene expression (dashed rectangle), including Nos2, Pdgs2, Il1b, and Tnfa (Fig. 4E). Reclustering of fibroblasts yielded similar results to those observed with genetic manipulation of Y814, where there was a presence of more proregenerative fibroblasts (Wnt2⁺) in the drug treatment group (Fig. 4F), which supports the findings from our chondrocyte experiments.

Because both F814- and R805-treated mouse skin wounds had reduced Nos2 expression, we examined this mechanism in macrophages in more detail. In vitro experiments demonstrated that WT mouse macrophages isolated from bone marrow that were stimulated with IL-6 expressed significantly (P = 0.032) higher Nos2 mRNA relative to the F814 mutant cells (fig. S32A). Furthermore, M1-polarized proinflammatory macrophages treated with R805 also exhibited lower Nos2 gene expression (P = 0.017) compared with vehicle-treated cells (fig. S32B). Flow cytometry analysis of reactive oxygen species (ROS) production also demonstrated that the F814 mutation significantly decreased IL-6–induced ROS activity (P = 0.034) in bone marrow–derived macrophages (fig. S32C). Together, these data suggested that both genetic mutation in gp130 Y814 and pharmacological modulation by R805 can shift macrophages away from a proinflammatory phenotype; however, more studies are necessary to define the specific roles of each cell lineages in the wound healing process.

R805 improves symptoms of OA in small and large animal models

We then tested R805 in vivo in two commonly used models of posttraumatic OA. First, we used a rat medial meniscal tear model that has been useful for the assessment of anti-arthritic drugs and recently shown to have an IL-6 family–dependent pathology (33, 34). We found that intra-articular injection of R805 was protective in the joint through reduced Osteoarthritis Research Society International (OARSI) scores and preserved cartilage thickness (fig. S33, A and B). We then conducted a preliminary 90-day pharmacokinetic study in beagles to show that a 1-μg intra-articular injection maintained median inhibitory concentrations of R805 in the joint for 3 weeks (fig. S34A). On the basis of these data, we administered 10, 1, or 0.1 μg of R805 or saline by intra-articular injection at post-operative weeks 4 and 11 to beagles subjected to surgical arthroscopic meniscal release (fig. S34B).

Histological abnormalities due to loss of meniscal load sharing in saline and 0.1-μg dose groups included focal partial and full-thickness cartilage erosions in central nonmeniscus-covered portions of the tibial plateau and longer linear erosions in the femoral condyle. In adjacent cartilage, there was a loss of the superficial collagen layer and its chondrocytes, loss of proteoglycan staining, invasion of the calcified cartilage by chondroclasts and vasculature, and some thickening (sclerosis) of the subchondral plate. There was incremental improvement in the histological scores (P = 0.012) as the R805 dose increased; erosions were more shallow and smaller, and adjacent cartilage organization was preserved, as was proteoglycan staining (Fig. 5A and fig. S35C). Synovial membranes in the high- and middle-dose groups had mild intimal hypertrophy with occasional subintimal mononuclear cell infiltration, whereas low-dose and saline-treated dogs had a profound increase in subintimal vasculature, new connective tissue, and cellular infiltrate composed of lymphocytes and macrophages (Fig. 5B and fig. S34C). In the dogs, bone mineral density increased in the medial tibial plateau in response to meniscal deficiency, but dogs receiving a high dose of R805 (10 μg) did not develop the subchondral sclerosis in the contralateral knee that usually arises from compensatory load transfer (Fig. 5C). Changes in knee joint shape, which can be used as a biomarker of OA in people (35) and animals (36), were also limited in dogs receiving R805 (Fig. 5D). These data were consistent with pain scoring and gait analysis data that showed that the high-dose group had minimal pain (Fig. 5E) and lameness (Fig. 5F).

Fig. 5. R805 demonstrates disease-modifying effects in large animal models of OA.

(A) Representative images of Safranin O staining and OARSI scoring of canine joint sections after 18 weeks of R805 treatment in surgical OA model. Scale bars, 100 μm. Horizontal lines with bars show the means ± SEM. n = 5. (B) Synovial inflammation, determined by fibrillations and immune infiltration observed in H&E staining after R805 treatment in canine joint. Representative images are shown. Scale bars, 100 μm. Horizontal lines with bars show the means ± SEM. n = 5 or 6 biological replicates. (C) Assessment of canine bone mineral density (BMD) in the medial compartment of the operated stifle and contralateral, non-operated medial compartment by microcomputed tomography (CT). Horizontal lines with bars show the means ± SEM. n = 5 or 6 biological replicates. (D) Representative images of canine knee joint shape change from baseline as determined by serial CT imaging to visualize ectopic bone formation (red, left); quantification on right. Horizontal lines with bars show the means ± SEM. n = 5 or 6 biological replicates. (E) Canine Colorado pain scores were measured daily and totaled in R805 and vehicle-treated animals starting after the first intra-articular injection. Linear regression analysis of daily scores is presented with P values. n = 5 or 6 biological replicates. For (A) to (E), x axes indicate dose used for intra-articular injections. (F) Linear regression analysis of lameness scores of R805-treated (10 μg/ml) and vehicle-treated (saline) canines. n = 5 or 6 biological replicates. Statistical analysis was performed using one-way ANOVA followed by the Tukey test to compare more than two groups. P values less than 0.05 were considered significant.

To support the histological findings of reduced cartilage degeneration and synovial fibrosis, we conducted scRNA-seq on synoviocytes isolated from R805-treated and control beagles. Unsupervised clustering identified several macrophage clusters (fig. S35 and table S2); reclustering of these cells defined seven subsets (Fig. 6A). Cluster 3 (dashed rectangle) was almost entirely composed of control cells and was enriched for proinflammatory macrophage markers, including Nos2 and Ptgs2 (Fig. 6B), similar to the mouse scRNA-seq data. Analysis of synovial fibroblasts indicated that there were substantial differences in fibroblast populations based on drug treatment (Fig. 6C, dashed versus solid ovals). Gene ontology analysis revealed enrichment of categories related to inflammatory signaling and recruitment of inflammatory cells (Fig. 6C), In particular, IL36B, PTGS2, IL36G, SAA1, TLR4, CCL3, and IL6 were substantially up-regulated (table S3). Together, these data suggested that R805 treatment reduced inflammation after injury, thereby limiting cartilage degeneration and synovial fibrosis.

Fig. 6. R805 promotes an anti-inflammatory, antifibrotic milieu in canine joints after surgical induction of OA.

(A) Reanalysis of macrophage clusters from synovial joints of R805-treated (10 μg/ml) and vehicle-treated (saline) canines. Clusters are color coded in each analysis. The contribution of each sample to each cluster is shown as a stacked bar graph (WT, red; R805, blue). Cluster 3 (mainly WT cells) is highlighted by the red dotted line. (B) Dot plots depicting gene expression in each macrophage cluster from synovial joints of R805-treated and vehicle-treated (saline) canines. Color-coded clusters correspond to colors in (A). Dot sizes are proportional to the percentage of cells in each cluster expressing the indicated gene. (C) Reanalysis of fibroblast clusters from synovial joints of R805-treated and vehicle-treated (saline) canines. Cells are color coded by their sample of origin. The dashed oval indicates clusters of fibroblasts dominated by R805-treated cells, whereas the solid oval denotes a cluster mainly derived from control cells. Table shows gene ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways that are enriched (false discovery rate <0.05 and fold change >1.5) when comparing cells in the solid versus dashed ovals.

The flanking region adjacent to gp130 Y814 is inversely related to the regenerative potential of species

Although our data indicated that inhibition of the gp130 Y814 signaling promoted regenerative outcomes in several species, we wanted to assess whether the key gp130 modules are conserved across regenerative and nonregenerative vertebrates. We performed phylogenetic profiling of the amino acid sequences of gp130 Y759, Y814, and Y905 residues and their surrounding sequence. The analysis unexpectedly revealed that, although the codon encoding gp130 Y814 is highly similar across all mammalian species, the homology of the flanking regions adjacent to the DNA triplet encoding Y814 to human sequence is sharply declining with the phylogenetic tree and inversely related to the regenerative potential of the analyzed species (fig. S36). This marked difference was not observed in the sequences encoding gp130 residues Y759 and Y905 (fig. S36).

Because one of the primary biological functions of gp130 signaling is to regulate immune response and inflammation, we suspect that these changes may be associated with the evolutionary events to optimize immune system activation, especially in response to invading pathogens. Although alterations in Y814 signaling may have initially served as an evolutionary advantage for host defense, it now appears to be one of the factors responsible for the fibrotic response and limited regenerative capacity after tissue injury in mammals. However, more studies are necessary to fully understand the homeostatic and physiological roles of gp130 Y814 in mammals.

DISCUSSION

In this study, we have demonstrated that the Y814 residue within the gp130 intracellular domain serves as a major cellular stress sensor that regulates proinflammatory and profibrotic pathways along with SRC kinase recruitment. Because chronic inflammation can prevent normal wound healing, targeting Y814 may potentially improve the wound repair process.

Our previous studies have demonstrated that gp130-STAT3 signaling is highly up-regulated in proliferative, anabolic fetal chondrocytes (23) and that LIF is highly expressed in developing human joints (37). More recently, we found that LIFR-gp130-STAT3 signaling is also required for homeostatic maintenance of chondroprogenitors in mice and that genetic postnatal ablation of this triad results in premature growth plate fusion (38). Genetic overexpression of STAT3 in postnatal chondrocytes did not induce an OA phenotype (38). Together, these data challenged previous oversimplified views of gp130-STAT3 signaling as unfavorable in cartilage tissue; our data suggest that activation of this pathway in OA may initially represent a regenerative attempt, but prolonged and excessive activation of this mechanism is likely to be detrimental. This study demonstrates that genetic or pharmacological inhibition of the gp130-Y814 module designed to minimally interfere with the endogenous STAT3 signaling improves regenerative outcomes in multiple tissues. Although other groups have reported that Y814 is one of the four residues responsible for STAT3 recruitment (39), it may not be the primary docking site, which would explain why we observed this lack of STAT3 regulation in the F814 cells.

Here, we have shown that gp130 Y814 serves as a recruitment site for SRC kinase. However, it is still unknown whether Y814 solely binds to SRC or whether other SFK family members can interact at this site. Mutation in Y814 and SRC-targeting peptide QQpYF also down-regulated MAPK (ERK 1/2) signaling, which is a central pathway controlling cellular processes associated with fibrogenesis, including growth, proliferation, and survival (40, 41) as well as OA pathogenesis (42–44). It is unclear whether SRC and MAPK/ERK cross-talk on Y814, but such signaling is possible because SRC is known to activate ERK signaling in multiple pathologies (45, 46). In contrast, STAT3, YAP1, and AKT were minimally affected by Y814 mutation, indicating that this residue regulates specific downstream pathways.

A recent study demonstrated that activation of gp130 signaling can limit regeneration (47). Our data indicated that gp130 Y814 signaling may be a contributing factor (fig. S37). Our RNA-seq data from gp130 F814 mutant mouse skin showed reduced expression of multiple proinflammatory and profibrotic genes from macrophages and fibroblasts, respectively, which agreed with the improved responses that we observed in our wound repair and OA model.

In addition, we found that an optimized RCGD 423 analog, R805, prevented Y814 phosphorylation in response to IL-6 cytokines and reduced OA symptoms in both rat and canine models. Similar to the substitution of Y814 in mouse, scRNA-seq data suggested that R805 treatment reduced the pathological milieu in mouse skin wounds and diseased canine joints by promoting an anti-inflammatory macrophage phenotype.

We have shown that genetically and pharmacologically targeting Y814 signaling can stimulate intrinsic regenerative capacity by suppressing the local inflammatory milieu after injury. Our use of genetic mutation and small molecules in multiple complementary animal model systems highlights the importance of our findings. However, there are limitations in this study. Although we have shown that gp130 Y814 regulates SRC and MAPK signaling in wound repair and OA, more studies are necessary to fully dissect the effects of gp130 Y814 in each cell type involved in these diseases. It is also possible that additional pathways are affected by manipulation of Y814 phosphorylation. Moreover, although the peptide QQpYF was designed to mimic the gp130 motif surrounding Y814 and prevent SRC recruitment, it may also compete for SRC recruitment and activation by other receptors. Last, although we demonstrated that several inflammatory mediators were down-regulated by Y814F mutation in mouse, we did not functionally validate direct specific targets downstream of this mechanism. These limitations will need to be addressed in future studies.

In summary, we identified the gp130 Y814 residue as an important cellular stress sensor that can serve as a therapeutic target for drug development. Data from our small molecule studies with R805 demonstrate the potential for pharmacological manipulation of gp130 signaling to improve intrinsic regenerative capacity of tissues and resistance to degenerative changes, which could be used to treat a broad range of conditions.

MATERIALS AND METHODS

Study design

The goal of this study was to decipher the gp130-mediated molecular mechanism responsible for tissue degeneration and fibrosis versus regenerative responses after wounding and to demonstrate how selective manipulation of this mechanism can promote regeneration in lieu of pathological outcomes. Our initial in vitro studies were performed using specific mutated plasmids to demonstrate the importance of the gp130 Y814 residue in modulating proinflammatory signaling. To understand the role of Y814 in vivo, we generated a mutant gp130 Y814 mouse (F814) using CRISPR-Cas9, and genetic sequencing was conducted to confirm the mutation. The mouse was used for surgeries including destabilization of the medial meniscus surgery and skin wounding to demonstrate resistance to degenerative changes and superior regenerative potential, respectively. For mouse OARSI scoring, observers performing the analysis were blinded as to whether the samples were from treated or control animals. We further performed an innovative peptide library screen with short and overlapping peptides, mimicking the motif of gp130 around Y814 that was individually tested for pSRC down-regulation by hindering recruitment of pSRC to gp130 and, as a result, ameliorating degenerative outcomes. We also evaluated therapeutic effects of gp130 Y814 inhibitor R805 on mouse skin upon injury and in rat and dog models of OA. For the canine study, the study was double-blinded and randomized, and no exclusion/inclusion criteria were used. All results of the mechanism modulation were analyzed via conventional laboratory readouts, RNA-seq, and scRNA-seq. For all experiments, biological replicates (cells from independent specimens) were used to generate data. De-identified human cartilage samples were collected under University of California, Los Angeles International Review Board no. 10–001857. For experiments expected to yield large differences, standard practice of using three to five replicates was followed. Sample sizes, biological replicates, and statistical methods are provided in the corresponding figure legends. No data were excluded from analysis.

Statistical analysis

Numbers of repeats for each experiment are indicated in the figure legends. Pooled data are represented as means ± SD, unless otherwise indicated. Unless otherwise indicated, statistical analysis was performed using one-way analysis of variance (ANOVA) followed by the Tukey test to compare more than two groups or two-tailed Student’s t test to compare two groups. P values less than 0.05 were considered significant. For parametric tests, the data met the test assumptions for normality. Shapiro-Wilk test was used to determine normality.

Data and materials availability:

All data associated with this study are present in the paper or the Supplementary Materials. All RNA-seq data are deposited in GEO under accession number GSE168395. F814 mice and polyclonal antibodies for Y814 residue will be available to academic researchers via material transfer agreement by contacting the corresponding authors.