TNF-α and IL-1β but not IL-18 counteract TGF-β1 mediated stimulation of collagen accumulation and increased extracellular matrix strength and stiffness Authors: Katerina St. Angelo, Arlen Brickman, Jeffrey R. Morgan Published: 2026 ABSTRACT Cytokine-mediated cross-talk between immune cells and fibroblasts is a driver of excessive ECM accumulation during fibrosis. In this study, we used a 3D in vitro model of a connective tissue to discern the roles of three pro-inflammatory cytokines; TNF-α, IL-18 and IL-1β, alone, and in combination with TGF-β1 to simulate the fibrotic environment. Ring-shaped tissues were formed by seeding human fibroblasts into circular molds of agarose, wherein the cells self-assembled, formed a 3D tissue and synthesized de novo a collagen-rich ECM. Cytokine treated tissues were analyzed at days 7 and 14 by histology and measured for thickness, collagen, DNA and strength and stiffness by tensile testing. Despite their pro-inflammatory nature, none of the cytokines increased collagen alone or in combination with TGF-β1. TNF-α and IL-1β reduced collagen, tissue strength and stiffness, and altered tissue morphology. When combined with TGF-β1, TNF-α and IL-1β counteracted TGF-β1-mediated increases in collagen, strength, and stiffness. In contrast, IL-18 had minimal effects alone or when combined with TGF-β1. These data suggest that IL-18 has no effect on fibroblast activation, whereas TNF-α and IL-1β may modulate TGF-β1’s effects. This 3D model of a human collagen-rich tissue can help define cytokine-mediated cross-talk between immune cells and fibroblasts.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-51200-y. Introduction Fibrosis, a pathological condition characterized by excessive extracellular matrix (ECM) deposition and tissue scarring, underlies a wide range of chronic diseases, including pulmonary fibrosis, liver cirrhosis, renal fibrosis, and cardiac fibrosis1–5. This maladaptive wound healing response results from prolonged or dysregulated tissue injury and repair mechanisms and can ultimately lead to organ dysfunction and failure1–5. Inflammation, cytokine signaling, and the resulting cross-talk between immune cells and fibroblasts are key drivers of fibrosis, acting as both initiators and perpetuators of fibroblast activation6–9. The cytokine milieu within fibrotic tissues is a complex interdependent network of pro-inflammatory and anti-inflammatory signals that dictates disease progression. However, it is ultimately the activated state of the fibroblast that is responsible for the synthesis and excessive accumulation of the collagen-rich ECM that underlies the pathology of fibrosis. Transforming growth factor-beta-1 (TGF-β1) is widely recognized as a key activator of the fibroblast and driver of the fibrotic process, but other pro-inflammatory cytokines are also present and implicated in fibrosis, including tumor necrosis factor-alpha (TNF-α), interleukin-18 (IL-18), and interleukin-1 beta (IL-1β)6,8–10. In many instances, immune cells and fibroblasts are secreting and responding to the same cytokines. Ultimately, this complex cross-talk is manifest in and interpreted by the fibroblast at the biochemical level when two or more signal transduction pathways converge within the cell. We’ve developed a 3D in vitro model of a collagen-rich connective tissue using human fibroblasts and have used the model to begin dissecting the role of different cytokines11–13. When seeded into a circular trough of agarose, human fibroblasts self-assemble and form a stable ring-shaped tissue that synthesizes and accumulates a collagen-rich ECM over four weeks without the aid of added scaffold. Addition of TGF-β1 stimulates collagen production and increases strength and stiffness of the tissues, properties that correlate with the in vivo pathology of fibrosis. In this study, we’ve used this model to understand the effects of TNF-α, IL-18, and IL-1β on untreated control tissues as well as tissues stimulated with TGF-β1 to simulate the fibrotic state. Unexpectedly, TNF-α and IL-1β altered tissue histology and reduced tissue size and collagen content, as well as the strength and stiffness of both control tissues and tissues stimulated with TGF-β1. In contrast, IL-18 had little effect on control tissues or tissues stimulated with TGF-β1. These data demonstrate the complexity of the cross-talk of pro-inflammatory cytokines, with IL-18 having a no effect and TNF-α and IL-1β having counter effects on the fibroblast, even when activated by TGF-β1. Results In vitro formation of 3D ring-shaped tissues We used a previously designed stainless steel molding system to form 5 mm diameter circular agarose troughs in 24-well plates (Fig. 1). After equilibration of the agarose with an optimized serum-free cell culture medium, cultured human fibroblasts were seeded into the molded agarose. Within 24 h, the cells aggregated and formed a 3D ring-shaped tissue that constricted around the central peg of agarose. The tissues, one per well, were visible and readily imaged using inverted microscopy while in the plate. To prepare for side-view imaging, the gel with tissue was removed from the plate, inserted into a specially designed chopping block, sliced on four sides, and each side of the tissue imaged using inverted microscopy (see Supplementary Fig. S1 online).Fig. 1Formation of ring-shaped tissues in vitro. Photograph of a ring-shaped tissue (5 mm inner diameter) in a molded agarose gel removed from a 24 well plate (a) and gel sliced to reveal the ring-shaped tissue (b). Conventional (c, e, g, i, k) and side-view (d, f, h, j, l) brightfield images of a day 14 control untreated tissue (c, d) and day 14 tissues treated with TNF-α (e, f), IL-18 (g, h), IL-1β (i, j) or TGF-β1 (k, l). Formation of ring-shaped tissues in vitro. Photograph of a ring-shaped tissue (5 mm inner diameter) in a molded agarose gel removed from a 24 well plate (a) and gel sliced to reveal the ring-shaped tissue (b). Conventional (c, e, g, i, k) and side-view (d, f, h, j, l) brightfield images of a day 14 control untreated tissue (c, d) and day 14 tissues treated with TNF-α (e, f), IL-18 (g, h), IL-1β (i, j) or TGF-β1 (k, l). TNF-α, IL-1β, and IL-18 alter the thickness of control tissues Twenty-four hours after cell seeding, rings were treated with either TNF-α (10 ng/ml), IL-18 (50 ng/ml), or IL-1β (50 ng/ml). Untreated rings served as the control. To quantify changes to tissue thickness, the dimensions of tissues treated with either TNF-α, IL-18, or IL-1β were measured and compared to untreated control rings (Fig. 2)(see Supplementary Table S1 online). TNF-α increased the thickness of day 7 tissues, but interestingly TNF-α decreased the thickness of day 14 tissues. IL-18 increased in thickness at days 7 and 14, whereas IL-1β treated rings decreased in thickness at days 7 and 14.Fig. 2TNF-α and IL-1β, but not IL-18, decrease the size of control tissues. Measurements of the x, y dimensions (a, b, d, e, g, h) and z dimensions (c, f, i) of tissues at day 7 (a, d, g) and day 14 (b, e, h). Control untreated tissues are compared to tissues treated with TNF-α, IL-18, or IL-1β (a-i). TNF-α and IL-1β, but not IL-18, decrease the size of control tissues. Measurements of the x, y dimensions (a, b, d, e, g, h) and z dimensions (c, f, i) of tissues at day 7 (a, d, g) and day 14 (b, e, h). Control untreated tissues are compared to tissues treated with TNF-α, IL-18, or IL-1β (a-i). The thickness of TGF-β1 stimulated tissues is decreased by TNF-α and IL-1β TGF-β1 is widely recognized as a major driver of fibrosis and previously, we’ve demonstrated that TGF-β1 increases size, total collagen, and tensile properties of ring tissues12. To determine if addition of a second cytokine could alter the TGF-β1 response, tissues were treated with TGF-β1 (10 ng/ml) plus either TNF-α, IL-18, or IL-1β. To determine cytokine influence on the TGF-β1-mediated increase in thickness, the dimensions of TGF-β1 treated tissues were compared to tissues treated with TGF-β1 plus TNF-α, TGF-β1 plus IL-18, or TGF-β1 plus IL-1β (Fig. 3)(see Supplementary Table S1 online). Acting alone, TNF-α increased the thickness of control tissues, but when combined with TGF-β1, TNF-α was not synergistic with the TGF-β1 mediated increase in thickness. The combination of TNF-α and TGF-β1 caused a significant decrease in thickness at day 7 as well as day 14 compared to tissues treated with TGF-β1 alone. The combination of IL-18 and TGF-β1 increased thickness at day 7 but not day 14. In contrast, tissues treated with a combination of IL-1β and TGF-β1 decreased in thickness at days 7 and 14.Fig. 3TNF-α and IL-1β, but not IL-18, decrease the size of TGF-β1 treated tissues. Measurements of the x, y dimensions (a, b, d, e, g, h) and z dimensions (c, f, i) of tissues at day 7 (a, d, g) and day 14 (b, e, h). Tissues treated with TGF-β1 are compared to tissues treated with TGF-β1 plus TNF-α, TGF-β1 plus IL-18, or TGF-β1 plus IL-1β (a–i). TNF-α and IL-1β, but not IL-18, decrease the size of TGF-β1 treated tissues. Measurements of the x, y dimensions (a, b, d, e, g, h) and z dimensions (c, f, i) of tissues at day 7 (a, d, g) and day 14 (b, e, h). Tissues treated with TGF-β1 are compared to tissues treated with TGF-β1 plus TNF-α, TGF-β1 plus IL-18, or TGF-β1 plus IL-1β (a–i). Differential effects of TNF-α, IL-1β, and IL-18 on total DNA of control and TGF-β1 stimulated tissues To determine if cytokine treatment altered total DNA, tissues harvested at days 7 and 14 were digested in papain and the amount of DNA measured using a fluorescent assay (Fig. 4) Total DNA was measured as an approximation of changes in cell number with the caveat that pyknotic nuclei are evident in histology. In prior studies, we’ve observed a decrease in total DNA of controls from day 7 to day 14. And in this study, we also see the same time dependent decrease for each of the treatments (Supplemental Table S2). When added to control rings, TNF-α had no effect on total DNA at days 7 or 14 but decreased total DNA of TGF-β1 stimulated tissues at days 7 and 14. In contrast, IL-18 increased total DNA of control tissues at days 7 and 14 but had no effect on TGF-β1 stimulated tissues. IL-1β had no effect on total DNA at day 7 of control or TGF-β1 stimulated tissues. However, at day 14, IL-1β increased total DNA of control tissues, but decreased total DNA of TGF-β1 stimulated tissues.Fig. 4Effects of TNF-α, IL-1β, and IL-18 on total DNA of control and TGF-β1 stimulated tissues. Measurements of the total DNA content of tissues at days 7 and 14 treated with TNF-α (a), IL-18 (b), IL-1β (c), TGF-β1 plus TNF-α (d), TGF-β1 plus IL-18 (e), or TGF-β1 plus IL-1β (f). Effects of TNF-α, IL-1β, and IL-18 on total DNA of control and TGF-β1 stimulated tissues. Measurements of the total DNA content of tissues at days 7 and 14 treated with TNF-α (a), IL-18 (b), IL-1β (c), TGF-β1 plus TNF-α (d), TGF-β1 plus IL-18 (e), or TGF-β1 plus IL-1β (f). TNF-α and IL-1β, but not IL-18, reduce total collagen of control and TGF-β1 stimulated tissues To determine if cytokine treatment altered total collagen, tissues harvested at days 7 and 14 were digested in papain and the amount of hydroxyproline measured using a colorimetric assay (Fig. 5)(see Supplementary Table S3 online). TNF-α decreased total collagen of control tissues at days 7 and 14. The decrease in collagen was even greater for TGF-β1 stimulated tissues. In contrast, IL-18 had no effect on total collagen of control or TGF-β1 stimulated tissues at days 7 or 14. IL-1β decreased total collagen of control tissues at days 7 and 14. Like TNF-α’s action, the IL-1β mediated decrease of total collagen was even greater when combined with TGF-β1 at days 7 and 14. We used TGF-β1 to mimic the fibrotic condition and when added to control rings, it consistently stimulated collagen production (see Supplementary Table S3 online).Fig. 5TNF-α and IL-1β, but not IL-18, reduce total collagen of control and TGF-β1 stimulated tissues. At day 7 and day 14 tensile tested tissues were frozen, solubilized by papain digestion and levels of hydroxyproline assayed to measure total collagen. Control untreated tissues are compared to tissues treated with TNF-α, IL-18, or IL-1β (a, b, c). Tissues treated with TGF-β1 are compared to tissues treated with TGF-β1 plus TNF-α, TGF-β1 plus IL-18, or TGF-β1 plus IL-1β (d, e, f). TNF-α and IL-1β, but not IL-18, reduce total collagen of control and TGF-β1 stimulated tissues. At day 7 and day 14 tensile tested tissues were frozen, solubilized by papain digestion and levels of hydroxyproline assayed to measure total collagen. Control untreated tissues are compared to tissues treated with TNF-α, IL-18, or IL-1β (a, b, c). Tissues treated with TGF-β1 are compared to tissues treated with TGF-β1 plus TNF-α, TGF-β1 plus IL-18, or TGF-β1 plus IL-1β (d, e, f). TNF-α and IL-1β, but not IL-18, reduce the strength and stiffness of control and TGF-β1 stimulated tissues To determine if cytokine treatment altered the biomechanics of the ECM, tissues at day 14 were removed from the gels and subjected to tensile testing to measure ultimate tensile strength (UTS, strength) and maximum tangent modulus (MTM, stiffness) (Fig. 6). All values were normalized to the cross-sectional area (CSA) of each tissue as calculated from their respective x, y and z measurements. TNF-α decreased the strength and stiffness of control and TGF-β1 stimulated tissues. In contrast, IL-18 had no effect on the strength or stiffness of control or TGF-β1 stimulated tissues. IL-1β had no effect on the strength or stiffness of control tissues but decreased the strength and stiffness of TGF-β1 stimulated tissues.Fig. 6TNF-α and IL-1β, but not IL-18, reduce strength and stiffness of control and TGF-β1 stimulated tissues. At day 14 (a–l), tissues were removed from the gel and tensile tested to measure to measure ultimate tensile strength (UTS, strength) (a, c, e, g, i, k) and maximum tangent modulus (MTM, stiffness) (b, d, f, h, j, l). Control untreated tissues are compared to tissues treated with TNF-α, IL-18, or IL-1β (a, b, c, d, e, f). Tissues treated with TGF-β1 are compared to tissues treated with TGF-β1 plus TNF-α, TGF-β1 plus IL-18, or TGF-β1 plus IL-1β (g, h, i, j, k, l). TNF-α and IL-1β, but not IL-18, reduce strength and stiffness of control and TGF-β1 stimulated tissues. At day 14 (a–l), tissues were removed from the gel and tensile tested to measure to measure ultimate tensile strength (UTS, strength) (a, c, e, g, i, k) and maximum tangent modulus (MTM, stiffness) (b, d, f, h, j, l). Control untreated tissues are compared to tissues treated with TNF-α, IL-18, or IL-1β (a, b, c, d, e, f). Tissues treated with TGF-β1 are compared to tissues treated with TGF-β1 plus TNF-α, TGF-β1 plus IL-18, or TGF-β1 plus IL-1β (g, h, i, j, k, l). TNF-α and IL-1β, but not IL-18, alter the histology of ring-shaped tissues At day 7 and day 14, tissues were formalin fixed and paraffin embedded, and sections were stained with hematoxylin and eosin (H&E) (Figs. 7, 8) or Masson’s Trichrome (see Supplementary Figures S2, S3 online). H&E and trichrome images of tissues were reviewed at day 7 and day 14 across multiple fields per slide, with attention to reproducible areas and consistent focus. Minor variation in stain intensity was present and did not change the overall impressions. Untreated controls showed the expected maturation from day 7 to day 14, thicker tissue with stronger trichrome blue and a shift toward parallel collagen bundles. Rings treated with TGFβ1 followed this trajectory, with collagen that is at least comparable to untreated controls and more clearly aligned by day 14. IL18 treated tissue appeared broadly similar to control at both time points, without a consistent change in collagen amount or architecture.Fig. 7TNF-α and IL-1β, but not IL-18, alter tissue histology. At day 7 (a, c, e, g) and day 14 (b, d, f, h), tissues were formalin fixed, paraffin embedded, and sections stained with hematoxylin and eosin (H&E). Shown are control untreated tissues (A, B) and tissues treated with TNF-α (c, d), IL-18 (e, f) or IL-1β (g, h). Scale bar = 100 um.Fig. 8TNF-α and IL-1β, but not IL-18, alter the histology of TGF-β1 stimulated tissues. At day 7 (a, c, e, g) and day 14 (b, d, f, h), tissues were formalin fixed, paraffin embedded, and sections stained with hematoxylin and eosin (H&E). Shown are tissues treated with TGF-β1 alone (a, b) and tissues treated with TGF-β1 plus TNF-α (c, d), TGF-β1 plus IL-18 (e, f), or TGF-β1 plus IL-1β (g, h). Scale bar = 100 um. TNF-α and IL-1β, but not IL-18, alter tissue histology. At day 7 (a, c, e, g) and day 14 (b, d, f, h), tissues were formalin fixed, paraffin embedded, and sections stained with hematoxylin and eosin (H&E). Shown are control untreated tissues (A, B) and tissues treated with TNF-α (c, d), IL-18 (e, f) or IL-1β (g, h). Scale bar = 100 um. TNF-α and IL-1β, but not IL-18, alter the histology of TGF-β1 stimulated tissues. At day 7 (a, c, e, g) and day 14 (b, d, f, h), tissues were formalin fixed, paraffin embedded, and sections stained with hematoxylin and eosin (H&E). Shown are tissues treated with TGF-β1 alone (a, b) and tissues treated with TGF-β1 plus TNF-α (c, d), TGF-β1 plus IL-18 (e, f), or TGF-β1 plus IL-1β (g, h). Scale bar = 100 um. Rings treated with TNFα or IL1β each produced a looser, less aligned matrix. The combinations TGFβ1 plus TNFα and TGFβ1 plus IL1β at day 14 showed the most pronounced effect, diminished trichrome intensity in many fields, loss of parallel fiber architecture with a whorled or interlacing pattern. H&E stains of these same tissues showed greater nuclear density with more shrunken dark nuclei consistent with increased pyknosis. Occasional small eosinophilic low nuclear zones were present in these groups, rare and focal. These morphologic trends align with the biochemical and biomechanical data that indicate reduced collagen content and impaired tensile properties in the TNFα and IL1β conditions, especially in the combinations, with the clearest separation at day 14. Discussion Cytokine-mediated cross-talk between immune cells and fibroblasts is well-established as a driver of the inflammatory state leading to excessive ECM accumulation during fibrosis6–8. While TGF-β1 is widely recognized as a key driver of fibrosis, other pro-inflammatory cytokines are known to signal fibroblasts and may modulate TGF-β1’s effects. However, a precise understanding of this cross-talk and how fibroblasts integrate the signals from multiple cytokines is lacking. In this study, we used a long-term 3D in vitro model of collagen-rich human connective tissue to investigate the effects of three pro-inflammatory cytokines: TNF-α, IL-1β, and IL-18, both alone and in combination with TGF-β1 to simulate a fibrotic environment. Despite their pro-inflammatory nature, none of the cytokines increased collagen accumulation when applied alone or in combination with TGF-β1 beyond the levels induced by TGF-β1 alone. Others have looked at ECM formation in 3D models by embedding fibroblasts in various scaffolds both natural and synthetic14–16. However, the material properties of the scaffold can confound biomechanical measurements of the ECM in these systems. Scaffold-free models have been used to measure the biomechanics of rings of smooth muscle cells, cartilage and mesenchymal stem cells17–20. The work in this study is focused on the effects of inflammatory cytokines on collagen production and the biomechanics of ECM produced by human fibroblasts. TNF-α, IL-1β, IL-18 and TGF-β1 are expressed by both fibroblasts and immune cells involved in fibrosis (e.g. macrophages). Fibroblasts and immune cells also respond to these cytokines, indicating that both autocrine and paracrine signaling occurs in the fibrotic environment. Fibroblast expression of these cytokines is context dependent and occurs during tissue injury, inflammation, immune cell signaling, or mechanical stress. This autocrine signaling is tightly regulated at multiple levels. TGF-β1 is secreted as an inactive pro-form bound to the ECM, requiring activation and release to be functional10,21. IL-1β and IL-18 are synthesized as inactive cytoplasmic pro-forms that require activation by an inflammasome prior to their release22–25. TNF-α is initially produced as a transmembrane protein and is released in a soluble form but only after proteolytic cleavage26–28. In this study, TNF-α and IL-1β reduced collagen, tissue strength and stiffness, and altered the morphology of the tissues. Likewise, TNF-α and IL-1β reduced collagen, tissue strength and stiffness, and altered the morphology of tissues stimulated with TGF-β1. Thus, when combined with TGF-β1, TNF-α and IL-1β counteracted or blunted the TGF-β1-mediated increases in collagen, strength, and stiffness. This suggests that TNF-α and IL-1β may modulate TGF-β1’s effects on fibroblasts in the fibrotic environment. In contrast, IL-18 had little to no measurable effect on collagen or ECM biomechanics of either control or TGF-β1 stimulated tissues. These data suggest that IL-18 may be more important as a paracrine signal to immune cells. Receptor-mediated cytokine signaling of fibroblasts occurs through well-characterized signal transduction pathways. Although TNF-α, IL-1β, and IL-18 bind distinctly different receptors, these receptors activate the same canonical pathways, namely NF-κB and MAPK22–24,26,27. TGF-β1 binds its own receptor, and its canonical pathway is the TGF-β/Smad pathway10,21,29. Our data suggest that the differential effects of TNF-α and IL-1β versus IL-18 are not the result of their common canonical signaling, but are likely context dependent and due to non-canonical signaling and yet to be defined interactions with an activated TGF-β/Smad pathway. This model may be useful for defining the complexity of these interacting pathways. Such studies would have to move beyond simply looking at collagen production and biomechanics and would need to examine RNA and protein markers indicative of fibroblast status as well as targeted inhibitors to dissect the signal transduction pathways. Previously, we’ve used this model to investigate the effects of IL-13, another pro-inflammatory cytokine implicated in fibrosis that signals via the JAK/STAT canonical pathway13. In that study, IL-13 alone had little effect, but when tissues were treated with a combination of IL-13 and TGF-β1 increases in collagen as well as tissue strength and stiffness were greater than when treated with TGF-β1 alone. These effects were sustained as long as 3 weeks. Unlike TGF-β1 and the cytokines tested in this study, IL-13 is a paracrine cytokine exclusive to immune cells (e.g., macrophages, eosinophils and Th2 cells). These data suggest that TGF-β1 activated fibroblasts may require a paracrine signal from immune cells to sustain increases in collagen, strength and stiffness. It would be interesting to test other pro-inflammatory paracrine cytokines such as IL-4, IL-17 and IFN-γ, both alone and in combination with TGF-β1. Conversely, it would be interesting to test the effects of anti-inflammatory paracrine cytokines such as IL-27. All are known to be expressed by immune cells in the fibrotic environment6,7,9. Most in vitro assays of cytokines have been performed on fibroblasts grown as 2D monolayers exposed to a cytokine for typically 24–48 hours30–32. In our model, we assess the effects of a cytokine on a 3D tissue for up to 4 weeks12. The data in this study suggests that fibroblast response to a cytokine may change with time. TNF-α and IL-1β effects on collagen production of control as well as TGF-β1 stimulated tissues were all evident within 7 days. Likewise, the effects of TGF-β1 alone were evident within 7 days and even earlier (data not shown). In contrast the effects of IL-13 on TGF-β1 activated tissues was evident only after 2–3 weeks of exposure13. These data suggest there may be a time dependency to cytokine signaling in the fibrotic environment with acute, more fast-acting signals (e.g. TGF-β1, TNF-α, IL-1β) and chronic signals (e.g., IL-13) that require prolonged exposure. The cross-talk of autocrine and paracrine cytokines that drive ECM accumulation in fibrosis is complex and regulated at multiple levels. The intention of this study is not to use single cytokines to represent the entire influence of immune cells in the complex crosstalk with fibroblasts. Our goal is to dissect which cytokines may be contributing to the fibrotic phenotype and or modulating the effect of TGF-β1. Using a human 3D model of a collagen-rich tissue, this study helps define the roles of TNF-α, IL-1β, and IL-18 on control tissues and tissues activated by TGF-β1. Stable for at least 4 weeks, this model can help define the effects of dose, exposure time as well as determine the non-canonical signaling of cytokine combinations that drive the dysregulated fibroblast to produce excessive ECM. The model can be used with fibroblasts from other organs and may also be useful for determining which if any cytokines are the causative agents that drive the formation of fibroblast subtypes found in fibrotic tissues in vivo that have been identified by single cell RNA sequencing33–38. Methods Culture of human fibroblasts Fibroblast tissue rings were made using juvenile normal human dermal fibroblasts (jNHDF), which were commercially sourced from PromoCell (Heidelberg, Germany). According to the company, cells were derived from tissues obtained from donors with informed consent. The company maintains donor privacy protections complying with strict ethical standards and European biomedical conventions. Cells are manufactured under ISO certification. We cultured the cells in Dulbecco’s Modified Eagle’s Medium (DMEM) with high glucose, phenol red, L-glutamine, and sodium pyruvate (Thermo Fisher Scientific, Waltham, MA; catalog #11,995,065), supplemented with 10% fetal bovine serum (FBS, Gemini Bio, West Sacramento, CA; catalog #100–500) and 1% penicillin/streptomycin (Thomas Scientific, Swedesboro, NJ; catalog #091,670,249). The cultures were maintained at 37 °C in a 10% CO2 incubator. Cells were propagated in T-175 flasks (Fisher Scientific, Hampton, NH; catalog #10-126-13) at a density of 0.5 × 106 cells per flask. Fibroblasts were used between the 4th and 8th passages and passaged using a standard trypsinization protocol. Cells were rinsed with phosphate-buffered saline (PBS, Thomas Scientific; catalog #SH30256.FS), followed by the addition of 0.05% trypsin (Thermo Fisher Scientific; catalog #SH3004201) in PBS for 5 min to detach the cells. Trypsin activity was neutralized with medium containing FBS. The resulting cell suspension was collected, centrifuged at 220 g for 5 min, resuspended in fresh medium, counted, and seeded into new flasks or agarose molds. The results in this study are based on neonatal human skin fibroblasts and results may vary with fibroblasts from other tissues and organs. In vitro formation of ring tissues Molds and molded agarose with a circular trough (5 mm in diameter) for forming ring tissues have been previously described. In brief, stainless-steel inserts with a circular feature and an aluminum base were fabricated through rapid prototyping (Protolabs, Maple Plain, MN). The aluminum base was designed to hold and position each stainless-steel insert into a single well of a 24-well plate (cat# 3527, Corning). Each well was filled with 1.5 mL of molten, sterile 2% (w/v) agarose (#BP160-500, Fisher Scientific, Hampton, NH) in PBS. A stainless-steel insert was added into each well, and the agarose was allowed to cool and gel for 15 min. After removing the inserts, each well contained molded agarose with a 0.75 mm wide cylindrical trough surrounding a 5 mm diameter agarose peg. Before adding the cells, the gels were equilibrated with serum-free DMEM containing 1% penicillin/streptomycin through several medium exchanges, at least 24 h before use. For ring tissue formation, an optimized culture medium (50:50) was used as previously described. This medium consisted of equal parts high glucose DMEM (#11,995,065), supplemented with 0.1 mM 2-phospho-l-ascorbic acid trisodium salt (#49,752, Sigma-Aldrich, St. Louis, MO), 50 μg/mL L-proline (#BP392-100, Fisher Scientific), and 1% penicillin/streptomycin, and advanced DMEM (#12,491,015, Thermo Fisher), supplemented with 4 mM GlutaMax (#35–050-061, Fisher Scientific) and 1% penicillin/streptomycin. Each agarose mold was seeded with 3 × 105 cells per ring. The medium was changed every 2–3 days. Prior to cytokine addition, rings were allowed to form for 24 h. TGF-β1(#100-21-50UG, ThermoFisher Scientific), TNF-α (#PHC3015, Fisher Scientific), IL-18 (#PHC0186, ThermoFisher Scientific), and IL-1β (200-01B-100UG, Thermo Fisher Scientific) were resuspended and aliquoted according to manufacturer’s protocols. TGF-β1 was reconstituted in 10 mM citric acid, pH 3.0 (Cell Signaling Technologies, #9871L) and diluted in 0.1% (w/v) final concentration of bovine serum albumin (BSA) in PBS. TNF-α was reconstituted in biology grade molecular water (#BP2819-1, Fisher Scientific). IL-18 was reconstituted in 1 M Tris HCL (ThermoFisher Scientific, #15,567,027). IL-1β was reconstituted in 0.1% (w/v) final concentration of bovine serum albumin (BSA) in PBS. Considering that each well has 1.5 ml of agarose and 1 ml of medium that could be removed, growth factors were added to 50:50 medium to achieve a final concentration of 10 ng/mL for TGF-β1, 10 ng/mL for TNF-α, 50 ng/mL for IL-18, and 50 ng/mL for IL-1β after equilibration with the agarose gel. These cytokine concentrations were chosen because they are at the high end of the dose response curve39–42. Medium was changed three times each week. Image acquisition and analysis To quantify time dependent and treatment dependent changes to ring size, brightfield imaging was performed at days 2, 4, 7, and 14 using a PCO.panda 4.2 USB3.1 sCMOS camera (Nikon, Tokyo, Japan) and a 2 × objective. Automated (x, y) thickness measurements were obtained through a custom General Analysis 3 (GA3) workflow within NIS-Elements AR (v6.10.01). This automated protocol utilized light-intensity thresholding to define ring boundaries, supplemented by manual inclusionary and exclusionary tools. To provide confidence in measurements of the ring thickness, the GA3 module generated 1000 equidistant measurements derived from 500 radial vectors originating from the sample’s geometric center. Raw data was compiled and standardized using a custom Python-based processing script. This automation consolidated individual sample measurements into a unified database, filtered for sampling outliers, and calculated the mean thickness and standard deviation for each ring. Following the final timepoint at day 14, side-view images (z) were acquired by removing the agarose gel with ring and placing it in into a custom-designed 3D printed chopping block (Supplemental Figure S1). The agarose gels were cut on four sides, laid on those sides and images taken of the rings. Four side view images were used to obtain 12 distinct thickness measurements across the z dimension of the ring. Cross-sectional area (CSA) and total volume were calculated using these measurements. Tensile testing of rings Tensile properties, maximum tangent modulus (MTM)(stiffness), and ultimate tensile strength (UTS)(strength) of tissues were measured as previously described12. Tensile values were normalized to the cross-sectional area (CSA) of each tissue as calculated from their x, y and z dimensions. Rings were carefully removed from the agarose molds, transferred onto the grips of a custom designed assembly. This assembly was submerged and attached to a 37 °C PBS bath mounted on an Instron 5943 (Norwood, MA) with a 5N (5mN resolution) load cell. After liberation from the agarose peg, rings were immediately placed onto grippers submerged in a saline bath. The assembly was secured on the tensile tester and the test commenced immediately. To expedite testing, the rings were not preconditioned prior to testing. The dominant failure mode was in one of the extended lengths of the ring. The upper grip was extended to an initial starting distance of 5 mm and then displaced at a rate of 0.1% initial length per second. Load (N) and extension (mm) were sampled at 20 Hz. Tests were terminated when the rings broke at a 40% drop in load. These testing parameters were chosen based on our prior work with rings12,13,43. Broken rings were frozen and stored at −80 °C for collagen and DNA measurements. Data were analyzed as previously described using a custom Python 3.6 code run on a Jupyter Notebook. The protocol and analysis was adapted from others43. The drag force of the grippers without a sample was removed and the tissue was considered to be in tension when the raw load was greater than the resolution of the load cell (5 mN). MTM and UTS were derived from the engineering stress (N/2A0) and engineering strain (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta L/L$$\end{document} gauge), where A0 is the starting cross-sectional area. Three to four technical replicates from three independent biological experiments were run for each condition for an n equal to ten. The mechanical properties of agarose are well documented, and values vary depending on concentration, molecular weight and the method of analysis44–46. The typical range for 2% agarose is ~ 20 to 200 kPa, thus the control rings measured in this study (~ 10 MPa) are significantly stiffer. Histology of ring tissues Tissues in the agarose mold were fixed in 10% buffered formalin and kept at 4 °C until embedding. In preparation for embedding, rings were kept on the peg and cut using a custom 3D printed chopping block. The squared gel with ring was moved to a 24-well plate and encapsulated with cooled 2% w\v agarose/PBS. After gelling, the agarose puck was transferred to a cassette and paraffin embedded. Embedded tissues were cut into 5–8 μm sections using a Leica RM2265 microtome (Leica Microsystems, Wetzlar, Germany). Slides were stained with H&E (Richard-Allan Scientific; Thermo Scientific) or Masson’s Trichrome (Electron Microscopy Sciences, Hatfield, PA) according to the manufacturer’s protocols. All slides were viewed and digitized using an Olympus VS200 Slide Scanner (Olympus, Tokyo, Japan) with associated VS200 ASW software followed by viewing in OlyVIA (Olympus, open access). Quantification of collagen and DNA of ring tissues To quantify total collagen, rings were digested in 125 ug/mL papain (#P4762, Sigma Aldrich) at pH 6.5 at 65 °C for 10 days and levels of hydroxyproline measured using a colorimetric assay. Digested tissues were diluted with an equal volume of 4 M NaOH (#S318-1, Fisher), incubated at 120 °C for an hour and neutralized with an equal volume of 4 M HCl (#SA56-1, Fisher). Chloramine T solution (#S318-1, Fisher) was added and samples incubated at room temperature for 20 min. Ehrlich’s reagent (#LC140802, LabChem) was added and samples incubated at 65 °C for 30 min. Samples were transferred to a 96-well plate and absorbance at 550 nm read using a UV plate reader (SpectaMax Plus 384, Molecular Devices, San Jose, CA). Values of collagen (ug/ring) were calculated from a standard curve (Type I rat tail collagen #5056, Advanced Biomatrix, Carlsbad, CA). Total DNA in the digests was measured using the Quant-iT PicoGreen dsDNA Assay Kit (#P11496, Invitrogen, Waltham, MA) per the manufacturer’s protocol. Digests were incubated with the picogreen solution at room temperature for 5 min, transferred to a 96-well plate and fluorescence (480 nm excitation, 520 nm emission) measured using a fluorescent plate reader (SpectraMax Gemini XS, Molecular Devices). Values of DNA (ng/mL) were derived from a standard curve (λ DNA #P11496, Invitrogen). Three to four technical replicates from three independent biological experiments were run for each condition for an n equal to ten. Measurements of DNA and collagen were normalized as per ring because the ring is the fundamental unit that did not change with time. Statistical analysis All data were graphically and statistically analyzed using GraphPad Prism: Version 11.0.1 (99)(https://www.graphpad.com/updates)(GraphPad Software, San Diego, CA). For comparisons between two unpaired groups, Welch’s t-test was applied to data following a Gaussian distribution, and the Mann–Whitney test was used for non-parametric data. Statistical significance is indicated by asterisks following the standard GraphPad settings (p < 0.05, p < 0.01, p < 0.001). Supplementary Information Supplementary Information. Supplementary Information.