Synovial fibroblasts exacerbate inflammation and peritendinous adhesions in a tissue-on-chip model of tendon fibrosis

Introduction: : Developing clinically relevant models to study tendon fibrosis is of critical importance to elucidate disease mechanisms and develop targeted therapeutics. Tendon fibrosis is a debilitating condition that leads to chronic pain and loss of joint function. A key challenge during tendon repair is a chronic inflammatory response leading to the formation of restrictive adhesions. Adhesions are composed of disorganized matrix deposited around the tendon that severely limit joint motion and diminish quality of life. Infiltrating immune cells and extrinsic synovial fibroblasts can either support healing or contribute to this fibrotic pathology by migrating to the injury site, proliferating, and secreting excess collagen and fibronectin around the tendon [1,2]. Understanding the multicellular mechanisms that determine tendon fate during repair motivates the development of human-relevant in vitro models.

We previously engineered a human tendon-on-a-chip (hToC) that features a 3D tendon construct and a vascular channel separated by an ultrathin nanomembrane [3,4]. The hToC recapitulates immune cell and intrinsic tendon fibroblast interactions and demonstrates hallmarks of tendon fibrosis driven by TGF-β1, including myofibroblast differentiation and vascular activation. However, the model lacks the important contributions of synovial fibroblasts and does not form adhesions. Our objective here is to model synovial fibroblast crosstalk in the hToC to determine the role of extrinsic fibroblasts in promoting fibrotic adhesions. We hypothesize that synovial fibroblasts will enhance the secretion of inflammatory mediators and drive monocyte infiltration from the vascular channel. We expect this heightened inflammatory response to induce matrix deposition around the tendon construct, characteristic of adhesion formation.

Materials and

Methods: : Device Setup and Culture Timeline: We engineered modular microfluidic components that are assembled with peel-and-stick adhesives. This enables efficient and reproducible assembly and allows us to mature cell types separately before combining components (Fig. 1A). Patient-derived tendon fibroblasts and monocytes were encapsulated in a type I collagen hydrogel and cultured with macrophage colony-stimulating factor for 6 days in the bottom compartment. At Day -4, the porous membrane was coated with a type I collagen hydrogel embedded with primary fibroblast-like synoviocytes (+FLS) or without (-FLS). Human umbilical vein endothelial cells were cultured on the opposite side of the membrane in the vascular channel for 24 h. At Day 0, monocytes were added to the vascular channel, and serum-free media with or without 10 ng/ml TGF-β1 was added to the bottom compartment. Multicellular cultures (Fig. 1B) were assayed on Days 1 and 5, with n=4 replicates. Tendon Hydrogel Contraction: We imaged the tendon hydrogels every day from Day 0 to 5 and percent contraction was quantified in ImageJ. Immunofluorescence and Second Harmonic Generation (SHG) Imaging: Devices were fixed and labeled for α-SMA, a marker of myofibroblast differentiation, as well as matrix proteins collagen III and fibronectin. Devices were imaged with confocal microscopy or multiphoton microscopy to generate SHG images of collagen structure. Cytokine Secretion: Media from the vascular channel was assessed via Luminex for TGF-β1 and inflammatory cytokine levels. Monocyte Transmigration: Circulating monocytes were labeled with CellTracker Orange, and monocyte infiltration was quantified from confocal images at Days 1, 3, and 5.

Results, Conclusions, and Discussions:: We asked whether synovial fibroblasts exacerbate fibrotic hallmarks regardless of TGF-β1 addition. When comparing -TGF-β1 conditions, we found significant increases in contraction in +FLS devices starting at Day 3, with increasing differences by Day 5 (70.3±5.1% for -FLS, 52.5±1.3% for +FLS). However, in +TGF-β1 conditions, +FLS and -FLS devices exhibited the same bulk tissue contraction (Fig. 2A). Interestingly, we observed an extended matrix connecting the vascular barrier and tendon construct, but only in +FLS cultures (Fig. 2B). To further characterize these structures, we labeled devices with F-actin, collagen III, and fibronectin (Fig. 2C). We found that FLS produced fibronectin and collagen III rich networks that connect the synovial matrix to the tendon in a manner resembling nascent adhesions. While tendon fibroblasts migrated to the collagen beneath the vascular barrier in -FLS devices, we did not observe hallmarks of adhesions in this case.

When investigating the impact of FLS on the intrinsic tendon phenotype, we found a significant increase in %α-SMA+ cells in +FLS devices (Fig. 2D). There was also a more significant increase in %α-SMA+ cells in +FLS devices under -TGF-β1 conditions compared to +TGF-β1 conditions, suggesting that FLS have a greater impact on fibrosis without exogenous TGF-β1. SHG images revealed heightened signal intensity in +FLS cultures at Day 5, indicating higher collagen concentration and density (Fig. 2E). We also discovered a significant increase in monocyte infiltration in +FLS cultures, demonstrating that FLS drive a vascular inflammatory response (Fig. 2F). With respect to cytokine secretion, there were no differences in sampled TGF-β1 levels with the addition of FLS (Fig. 1G). However, in -TGF-β1 conditions, IL-6 and IL-8 were significantly elevated at Day 5, and FLS drove a significant increase in IL-6 compared to -FLS devices. Given the role of IL-6 in leukocyte recruitment and cell proliferation, IL-6 inhibition may attenuate adhesion formation.

Our in vitro platform is capable of representing not only fibrotic tendon tissue, but now also the restrictive peritendinous adhesions that drive poor patient outcomes. This model addresses a critical need for alternatives to animal models and could be leveraged to accelerate therapeutic strategies to mitigate tendon fibrosis.

Acknowledgements and/or References (Optional):: This research was supported by NIH UH3TR003281 and the National Science Foundation Graduate Research Fellowship. Schematics were created with BioRender.com.

References
[1] Zhang Q, Yang Y, Yildirimer L, Xu T, Zhao X. Advanced technology-driven therapeutic interventions for prevention of tendon adhesion: Design, intrinsic and extrinsic factor considerations. Acta Biomaterialia. 2021;124. doi: 10.1016/j.actbio.2021.01.027.

[2] Shen H, Yoneda S, Sakiyama-Elbert SE, Zhang Q, Thomopoulos S, Gelberman RH. Flexor Tendon Injury and Repair. The Influence of Synovial Environment on the Early Healing Response in a Canine Model. Journal of Bone and Joint Surgery. 2021;103(9):e36. doi: 10.2106/jbjs.20.01253.

[3] Ajalik R, Linares I, Alenchery R, Wright T, Miller BL, McGrath JL, Awad HA. Human Tendon-on-a-Chip for Modeling the Myofibroblast Microenvironment in Peritendinous Fibrosis. Advanced Healthcare Materials. 2024;14(2403116). doi: https://doi.org/10.1002/adhm.202403116

[4] Linares I, Chen K, Saffren A, Mansouri M, Abhyankar VV, Miller BL, Begolo S, Awad HA, McGrath JL. Fluid Flow Impacts Endothelial-Monocyte Interactions in a Model of Vascular Inflammatory Fibrosis. Scientific Reports. 2025;15(3227). doi: https://doi.org/10.1038/s41598-025-85987-z