Arthritis models for studying chikungunya virus joint inflammation

Abstract

Chikungunya virus (CHIKV) joint inflammation remains poorly understood, despite its clinical burden. This review surveys in vitro joint models which were originally developed for arthritis research and examine how they can be repurposed to study CHIKV joint pathology. We outline key considerations, limitations, and recommendations for adapting these models for infection studies, including facilitating infection susceptibility across joint models, tuning matrix composition, addressing cellular incompleteness through modeling relevant immune cells, improving culture longevity and reproducibility. Together, these models offer versatile and human-relevant approaches to replicate key features of joint inflammation in advancing the understanding of CHIKV-induced joint pathology.

A review highlights how in vitro joint models adapted from arthritis research can be leveraged to study chikungunya virus–driven joint inflammation, immune–stromal interactions, and matrix degradation underlying acute and chronic arthralgia.

Introduction

Chikungunya virus (CHIKV), an arthritogenic alphavirus transmitted by Aedes mosquitoes, has re-emerged as a significant global health concern. The World Health Organization reported a resurgence of CHIKV from 1 January to 30 September 2025, with 445,271 suspected/confirmed cases and 155 deaths across 40 countries; these cases spanned Africa, the Americas, South-East Asia, and the Western Pacific, and local transmission was reported in Europe¹. Beyond the acute febrile illness, which is typically self-limiting and resolves within one to two weeks, a substantial proportion of CHIKV-infected patients develop persistent joint inflammation that can last for years post-acute infection^2,3. This persistent joint pain, despite resolution of viremia, distinguishes CHIKV from other arboviral infections and underscores the importance of understanding its joint-specific pathology, particularly as the tissue-specific mechanisms driving persistence remain poorly defined.

Although CHIKV-induced joint inflammation shares certain immunological features with rheumatoid arthritis (RA), there are important distinctions between the two (Fig. 1). Both conditions involve CD4⁺ T cell activation and elevated pro-inflammatory cytokines, and transcriptomic analyses in mouse models have revealed overlapping gene signatures related to T cell activation, antigen presentation, and innate and adaptive immune responses^4–7. However, similarity in gene expression does not imply identical disease mechanisms. RA is driven by a breakdown in immune tolerance, leading to the production of autoantibodies such as anti-citrullinated protein antibodies (ACPAs) and rheumatoid factor (RF), and is sustained by chronic CD4⁺ T cell activation with extensive crosstalk involving macrophages, dendritic cells, B cells, and fibroblast-like synoviocytes (FLS)⁵. In contrast, CHIKV-induced joint inflammation is initiated by viral infection, involves CD4⁺ T cells and macrophage interactions but lacks consistent autoantibody involvement, and is shaped by a chemokine-dominant profile (including CXCL8, CXCL9, and CXCL10) associated with antiviral responses^8–11. While recent studies of CHIKV have primarily focused on systemic immune responses and macrophage-T cell interactions, it is likely that crosstalk between the resident cell types within the joint microenvironment, such as FLS, chondrocytes, and resident synovial macrophages, also contribute to CHIKV pathogenesis. These differences show that CHIKV-induced joint inflammation is a distinct pathophysiological process, and its mechanisms cannot be directly equated with those of autoimmune arthritides (Fig. 1).

Fig. 1. Differences between RA and CHIKV-induced joint inflammation.

The figure contrasts disease triggers, main disease mechanisms, the role of CD4⁺ T cells in pathogenesis, and factors contributing to disease persistence between RA and CHIKV-induced joint inflammation. Created in BioRender. Ng, L. (2026) https://BioRender.com/1cn0nmd.

Despite its obvious clinical relevance, the joint remains underexplored in CHIKV research. Previous in vivo studies and analyses of patient serum have advanced our understanding of systemic immune responses, but they offer only a partial view of the disease mechanisms. Complex interactions between immune cells, stromal cells, and extracellular matrix (ECM) components in the joint microenvironment may drive CHIKV-induced pathology and its potential persistence. Sophisticated joint-focused experimental systems are therefore required for dissecting the underlying tissue-specific processes, such as those established in the field of RA and osteoarthritis (OA). While OA is a progressive age-related degenerative disorder, CHIKV-associated joint inflammation is typically self-limiting, but can become chronic in a subset of patients. Despite these differences, OA-based models can provide a relevant platform to investigate inflammation-driven matrix degradation and chondrocyte dysfunction.

While several arthritogenic alphaviruses, including RRV, ONNV, and MAYV, can also cause joint inflammation and present with persistent arthralgia, the depth of clinical and mechanistic understanding varies across these infections¹². CHIKV has been extensively characterized, with a rich body of clinical and emerging tissue-level evidence. However, long-term and mechanistic data for other alphaviruses remain limited. The extent to which underlying pathogenic mechanisms are shared across these viruses remains poorly defined and reflect differences in available evidence rather than true biological divergence. To contextualize CHIKV within the broader group of arthritogenic alphaviruses, a comparative summary is provided (Table 1), highlighting differences in clinical presentation and depth of current mechanism. As such, while in vitro joint models are broadly applicable across arthritogenic alphaviruses, this review focuses on CHIKV as a representative system with a more established evidence base.

Table 1.

Comparison of clinical and mechanistic evidence across arthritogenic alphaviruses

Comparative features	CHIKV	RRV	MAYV	ONNV
Disease course	Acute polyarthritis across multiple clinical cohorts, with a significant subset of patients developing prolonged or chronic symptoms.	Acute polyarthritis though chronic persistence is not as common as in CHIKV.	Acute polyarthritis, with reports of chronic polyarthritis, potentially fatal outcomes with neuro-cardiac involvement	Acute polyarthritis, rarely progressing to chronic disease and typically non-deforming.
Depth of mechanistic understanding	Extensive (immune and emerging stromal/tissue-level studies)	Moderate (primarily immune-driven mechanisms)	Limited–moderate	Limited-moderate

Features are summarized based on available literature, with clinical course descriptions adapted in part from a comparative review of arthritogenic alphaviruses and represent general trends rather than direct comparative analyses¹².

CHIKV Chikungunya virus, RRV Ross River virus, MAYV Mayaro virus, ONNV O’nyong’nyong virus.

This review explores the ways in which experimental models originally developed for arthritis research can also be used to advance our understanding of CHIKV-induced joint pathology, drawing on tools such as two- and three-dimensional (2D and 3D) cell cultures, co-culture systems, and microfluidic joint-on-chip (JoC) platforms to interrogate viral-host interactions in a controlled, human-relevant context. This review was conducted as a narrative synthesis of the literature, with a focus on studies relevant to chikungunya virus-associated joint pathology and emerging joint models. Relevant articles were identified through searches of databases including PubMed, Web of Science, and Google Scholar using combinations of keywords such as “chikungunya virus”, “cartilage”, “synovial macrophages”, “fibroblastic-like synoviocytes”, and “joint models”. Rather than applying strict inclusion or exclusion criteria, studies were selected based on their relevance to the central themes of this review, with an emphasis on recent advances and mechanistic insights in CHIKV-associated joint pathology and joint modeling approaches.

CHIKV-induced joint pathology

To understand CHIKV joint pathology, it is important to consider the joint microenvironment, which provides the structural and biological context within which viral replication and immune activation occur. Firstly, the articular cartilage is an avascular and alymphatic tissue that lines the ends of bones within a joint and is composed primarily of chondrocytes embedded within an abundant ECM, allowing smooth articulation and absorbing load¹³. The surrounding synovial membrane lines the inner surface of the joint capsule and contains two main synoviocyte populations: type A macrophage-like synoviocytes, which function as phagocytic cells that clear debris and immune complexes, and type B FLS, which synthesize hyaluronan and lubricin to maintain joint lubrication and homeostasis¹⁴. Both synoviocytes and articular chondrocytes are immunologically active, capable of sensing pathogen- or damage-associated molecular patterns through pattern recognition receptors and responding via production of pro-inflammatory cytokines and chemokines¹⁵. The synovial membrane also produces synovial fluid, which fills the joint cavity and facilitates lubrication and nutrient and oxygen delivery to the avascular cartilage while serving as a reservoir for soluble mediators¹⁶. This tightly integrated joint microenvironment is highly susceptible to inflammatory disruption.

The immunopathology of CHIKV infection has been extensively attributed to systemic immune responses, with CD4⁺ T cells playing a central role in driving joint inflammation and tissue damage. In mouse models, early studies demonstrated that depletion of CD4⁺ T cells significantly reduced footpad swelling without affecting viral load, indicating an inflammatory rather than antiviral functionality during CHIKV infection⁶. Within the joint-footpad, activated CD64⁺MHCII⁺ macrophages have been identified and shown to engage in pathogenic interactions with CD4⁺ T cells, contributing to local immunopathology through the secretion of inflammatory mediators (Fig. 2A)⁸. Complementing these murine findings, recent human blood studies have demonstrated that CHIKV-specific CD4⁺ T cells persist long after acute infection, predominantly produce TNF-α, implicating them in the maintenance of post-viral inflammation¹⁷. Viral antigen was also detected in the synovial biopsies (specifically, in synovial macrophages) of a CHIKV patient 18 months after the initial clinic visit¹⁸. In addition, CHIKV was also detected in joint tissues of non-human primates for up to 3 months post infection, raising the possibility of viral persistence¹⁹.

Fig. 2. Mechanisms in the joint microenvironment during CHIKV infection.

This schematic illustrates the joint microenvironment during CHIKV infection, highlighting: A Crosstalk between CD4⁺ T cells and monocyte-derived macrophages; B HFLS-driven monocyte recruitment and differentiation toward osteoclast-like cells; C chondrocyte-derived mediators that promote cartilage catabolism; and D secretion of IL-6 and RANKL by osteoblasts with reduction of OPG, shifting the RANKL/OPG axis toward osteoclastogenesis and bone resorption. RANKL receptor activator of NF-κB ligand, OPG osteoprotegerin. Created in BioRender. Ng, L. (2026) https://BioRender.com/m5kx33b.

Several studies have examined the roles of resident cell populations in CHIKV-induced joint pathology. Type B FLS or synovial fibroblasts are one of the main components in the joint microenvironment. Primary human FLS were susceptible to CHIKV replication in vitro, producing high viral titers of 10⁵–10⁶ pfu/ml and displaying cytopathic effects within days of infection²⁰. Upon CHIKV infection, FLS predominantly secreted RANKL, IL-6, IL-8, and MCP-1, which recruited monocytes and drove their differentiation toward osteoclast-like cells (Fig. 2B). Notably, FLS were not major sources of TNF-α, IL-1β, or MMP-1, -2, or -13, which are known to induce arthritis. Instead, the resulting osteoclast-like cells contributed to bone resorption and secreted inflammatory cytokines such as TNF-α and IL-6, which exacerbated joint inflammation in vitro. This indicates that FLS are not direct mediators of joint destruction but instead act as initiators of an erosive cascade towards osteoclastogenesis. Single-cell transcriptomics have further implicated FLS as central effectors of CHIKV arthritis, highlighting their role in immune cell recruitment and RANKL-driven osteoclastogenesis²¹. Concordantly, proteomic profiling has demonstrated that CHIKV infection rapidly remodels the FLS proteome within 12 and 24 h post-infection, altering cellular trafficking and cytoskeletal, metabolic, and immune pathways in ways that may facilitate both viral replication and pro-inflammatory programming²². Finally, primary human FLS exhibited a miRNA-mediated dampening of innate signaling during CHIKV infection, with microarray profiling identifying approximately 26 host miRNAs predicted to target AKT3, TRIM26, SOCS7, and PELI1, consistent with suppression of the JAK-STAT, TLR, TNF, and NF-κB pathways that could favor viral persistence. Functionally, CHIKV upregulates miR-146a in human FLS, which downregulates TRAF6 and IRAK1/2, blunting NF-κB and thereby enhancing replication while modulating fibroblast inflammatory output²³. Collectively, these findings identify FLS as pivotal amplifiers linking stromal infection to immune recruitment and joint damage.

The role of type A macrophage-like synoviocytes (also called resident synovial macrophages) in CHIKV infection is less clear. Unlike the infiltrating monocyte-derived macrophages that dominate inflammatory settings, type A macrophages originate from embryonic precursors such as the yolk sac or fetal liver, and they are self-renewing and M2-like²⁴. These macrophages were characterized as a unique CX3CR1⁺ tissue-resident population and known to establish a protective barrier at the synovial interface, where they limit immune cell entry via tight junctions, actively resolve inflammation through neutrophil clearance, and sustain homeostatic balance within the synovial microenvironment^25,26. Their position at the synovial interface places these resident/barrier macrophages at a critical juncture between stromal integrity and viral pathogenesis, but whether they function as antigen-presenting sentinels or serve as potential reservoirs of persistence in the context of CHIKV infection is not yet known. Their roles remain largely unexplored due to the difficulty of isolating resident synovial macrophages, and most studies to date have relied on monocytic models such as THP-1 or U937 cell lines^27–29.

Very recently, chondrocytes were found to be permissive targets of CHIKV, whereby supporting viral replication and producing infectious virus³⁰. Using a NanoLuc-expressing CHIKV, robust viral replication within the articular cartilage of immunocompetent mice was observed, with signals localizing to chondrocyte-rich regions. Histopathology confirmed infected chondrocytes with focal cartilage erosion and periarticular inflammation in mice. In vitro primary human chondrocytes were also productively infected, released infectious progeny, and underwent apoptosis. Infection also induced expression of MMP-3 and MMP-9, enzymes linked to proteoglycan and gelatin degradation, thereby coupling viral cytopathology with catabolic matrix remodeling (Fig. 2C). A CHIKV-infected 3D chondrocyte model cultured from patient joint samples also showed productive viral replication. Quantitative proteomics revealed strong interferon antiviral responses and broad remodeling of ubiquitination, phosphorylation, and metabolism, changes that were not found in matched 2D monolayers³¹. Taken together, these data established human chondrocytes as active participants in antiviral signaling and demonstrated that matrix context shapes virus-host interactions. Although evidence for CHIKV infection of chondrocytes is still emerging, it is supported by broader observations of joint tissue tropism across other arthritogenic alphaviruses, including Ross River Virus (RRV) and Mayaro virus (MAYV)^32,33.

Meanwhile, osteoblasts have also been identified as stromal targets of CHIKV infection. Infection of primary human osteoblasts supports viral replication and elicits robust secretion of IL-6 and RANKL, accompanied by reduced osteoprotegerin (OPG), thereby shifting the RANKL/OPG axis toward osteoclastogenesis and bone resorption (Fig. 2-D)³⁴. The same cytokine imbalance was observed in the plasma of infected individuals, suggesting that the source of these mediators is osteoblasts³⁵. While definitive in vivo validation is still lacking for CHIKV, comparable work in RRV has shown infected osteoblasts to display the same IL-6 and RANKL/OPG profiles. In murine models, this translates to bone loss measurable by micro-CT, with IL-6 neutralization attenuating osteoclast recruitment and preserving bone volume³⁶. Importantly, this osteoblast program overlaps with observations from RA, where dysregulated RANKL/OPG signaling also fuels osteoclastogenesis and subchondral erosion^37,38. Thus, in alphavirus arthritis, osteoblast infection adds a bone remodeling axis that complements synoviocyte-driven immune recruitment and chondrocyte-mediated matrix degradation.

CHIKV-associated joint pathology is likely driven by multiple, non-mutually exclusive mechanisms. Among these, contributions from adaptive immune responses are relatively well supported. In contrast, several additional mechanisms have been proposed but remain poorly defined, including sustained activation and reprogramming of joint-resident stromal cells, persistence of replicating virus in tissue reservoirs, and retention of viral antigens in the absence of active replication. In this review, we adopt a framework in which persistent inflammation is primarily sustained by stromal–immune interactions following the acute phase, with viral or antigen persistence acting as potential initiating or amplifying factors. While evidence for long-term productive viral replication in joint tissues remains unclear, the presence of viral RNA or antigen and its role in local immune activation remains an area of ongoing investigation.

Together, these findings suggest that CHIKV-associated joint inflammation is not solely immune-driven but is also shaped by dynamic interactions between immune and stromal compartments. Elucidating how viral infection influences stromal cell behavior and how these interactions sustain inflammation will be critical for understanding disease progression. Leveraging in vitro joint models developed in the arthritis field, therefore, provides a powerful approach to dissect these processes and to investigate how early viral events transition into a self-sustaining, tissue-specific inflammatory response.

Complexity of in vitro joint models

Joint research has progressively advanced from 2D cultures to more complex 3D spheroids, co-cultures, and microfluidic platforms, each capturing increasing structural and functional complexity of the joint tissue. By analogy, advancing CHIKV-induced joint inflammation research would require a similar shift toward models that reflect the multicellular architecture, dynamic signaling, and biomechanical properties of the human joint.

2D cell culture models

FLS and chondrocyte monolayers, typically derived from primary cultures or tissue explants, have been established as standardized 2D cultures for dissecting joint pathology. For example, FLS were studied for their role in sustaining inflammation and tissue remodeling by stimulation with inflammatory cytokines such as IL-1β or TNF-α, which showed upregulation of adhesion molecules, including ICAM-1 and VCAM-1, and secretion of pro-inflammatory mediators such as IL-6 and CCL2 alongside matrix-degrading enzymes^39–41. Chondrocyte monolayers, despite their tendency to dedifferentiate in vitro, have likewise been extensively used to investigate cartilage responses⁴². Exposure to IL-1β or TNF-α induces expression of catabolic enzymes such as MMP-13 and ADAMTS-5 while suppressing anabolic markers, including COL2A1 and SOX9, consistent with the loss of cartilage matrix and identity observed in arthritic joints^43,44. These cell types have also been employed for interrogating inflammatory and catabolic signaling pathways, including NF-κB, MAPK, and JAK/STAT^45–47. Owing to their ease of use and reproducible and quantifiable readouts, these cultures remain widely used for mechanistic studies of joint pathology and for the evaluation of candidate therapeutics.

2D co-culture models are an initial step toward multicellular systems that recapitulate key cellular interactions in the joint. For example, an early 2D co-culture model of FLS and U937-derived macrophages was established to study cytokine crosstalk in RA⁴⁸. The study concluded that monocyte-synoviocyte contact is sufficient to drive IL-6-dominant inflammation, but additional immune or environmental factors are required to elicit full pro-inflammatory cytokine responses. More recently, an epigenetic mechanism linking macrophage-chondrocyte communication to osteoarthritis (OA) progression in 2D co-culture was uncovered: M1 macrophages released miR-350-3p-enriched ectosomes (microparticles that bud directly outward from the membrane) that, once internalized by chondrocytes, suppressed the histone methyltransferase NSD1, reducing H3K36 methylation and driving MMP-13 induction, cellular senescence, and matrix loss⁴⁹. This study thus identified a previously unrecognized macrophage-to-chondrocyte miRNA-epigenetic axis, revealing how inflammatory macrophages can rewire chondrocyte chromatin states to promote joint degeneration. Finally, a 2D macrophage-chondrocyte co-culture model was used to show that kinsenoside, a natural iridoid glycoside, inhibited NF-κB and MAPK signaling, shifting macrophages from an M1 to an M2 phenotype⁵⁰. This repolarization reduced IL-1β, IL-6, and TNF-α secretion, alleviated chondrocyte inflammation, and preserved matrix integrity, underscoring the potential of these models in OA research.

3D cell culture models

While 2D cultures have been significant in studying basic cellular and inflammatory responses, they fail to recapitulate the complex 3D architecture and dynamic cell-matrix interactions. This limitation is particularly evident for chondrocytes and FLS, in which their phenotype and function are strongly influenced by spatial organization and ECM cues. To address these limitations, 3D culture systems are now becoming central to arthritis research, where they help overcome the limitations of monolayers by restoring ECM deposition, stabilizing cell phenotype, and providing tissue-like architecture. A range of model types has been developed, including spheroids, pellets, and engineered scaffolds, each highlighting distinct aspects of joint pathophysiology.

Chondrocyte 3D spheroids, also called cartilage organoids, self-assemble to produce their own ECM and maintain a rounded morphology, enabling more physiologically relevant modeling of cartilage inflammation and catabolism than 2D cultures. Transcriptomic analyses show that these constructs recapitulate key OA signatures, including cytokine and protease cascade upregulation, together with chondrocyte clustering similar to that observed in vivo⁵¹. They are also responsive to pro-inflammatory stimuli such as LPS, IL-1β, and TNF-α, and they have been used to evaluate anti-catabolic compounds that preserve matrix integrity and suppress NF-κB-driven signaling (e.g., curcumin, resveratrol, and vitamin D)⁵². Building on this, cartilage organoids have also been engineered to model OA traits⁵³, alongside stem cell-derived spheroids showing regulated maturation pathways⁵⁴ and iPSC-derived systems offering renewable cartilage-like tissue for mechanistic and drug-screening studies⁵⁵. Collectively, these models provide tractable, human-relevant platforms for studying cartilage degeneration and testing therapeutic strategies.

Hydrogel-based 3D systems are widely used in modeling the synovial tissue environment, especially for FLS and its co-cultures, as they replicate a hydrated, viscoelastic architecture that resembles the native ECM of synovium. Articular chondrocytes can self-aggregate into pellets and produce their own ECM. Nevertheless, many studies still encapsulate chondrocytes in natural or synthetic hydrogels to precisely tune biochemical cues, stiffness, and diffusivity for disease modeling^56,57. As an example, FLS on its own self-organized in ECM micromass into a multilayer synovial-like lining that secreted lubricin and supported monocyte and macrophage co-compaction, with hyperplastic changes resembling RA also observed after cytokine stimulation⁵⁸. This was later extended with RA FLS organoids that showed higher baseline secretion of IL-6, IL-8, CCL2, and MMP-3 while retaining stromal markers when compared to 2D cultures⁵⁹.

Two-component 3D co-cultures

Two-component systems have provided valuable insights into intercellular crosstalk at the joint interface, particularly between synovium and cartilage. One model utilized human RA FLS and cartilage-like constructs formed from alginate-recovered bovine chondrocytes⁶⁰. Their findings showed that cartilage degradation was not driven solely by synoviocytes' secreted enzymes but required active chondrocytes' participation, possibly through production of cartilage-specific proteases, establishing mutual dependence and crosstalk at the synovium-cartilage interface. Building on this concept of intercellular interaction, a matrigel-free system pairing cartilage spheroids and RA synoviocytes quantified synovial invasion and reported ADAMTS5 overexpression in chondrocyte spheroids⁶¹. Cytokine stimulation increased synoviocyte invasion in this system, whereas the TNF inhibitor, etanercept, attenuated it, highlighting the therapeutic relevance of this platform. In contrast to these cartilage-focused models, another approach reconstituted a synovial lining by co-culturing RA FLS and CD14⁺ monocyte-derived macrophages to mimic the native synovial lining of the joint⁵⁹. In this co-culture, the bilayer self-organized within 7 days, showed hyperplasia with TNF-α, and shifted to fibrosis with TGF-β, capturing macrophage-FLS circuitry. This FLS-macrophage 3D system effectively recapitulated synovial structure and macrophage-driven inflammation seen in arthritis. Extending the complexity further, polycaprolactone nanofibrous microwells were used to co-culture FLS and human umbilical vein endothelial cells (HUVECs)⁶². The nanofibrous scaffold mimicked ECM structure, promoting stable spheroid formation, endothelial organization, and vascular marker expression. Upon TNF-α stimulation, the model recapitulated rheumatoid inflammation, while treatment with anti-RA drugs attenuated these responses, revealing donor-specific drug sensitivity. Finally, a PEGDA-RGD hydrogel co-culture of chondrocytes and activated macrophages demonstrated robust catabolic and inflammatory responses, with OA chondrocytes exhibiting exaggerated susceptibility compared to healthy chondrocytes. This emphasized the heightened susceptibility of diseased cartilage to macrophage-driven inflammation⁶³.

Collectively, these models share a focus on intercellular interactions, whether synovium-cartilage, synovium-macrophage, or synovium-endothelium, while differing in scaffold composition, cell sources, and arthritic context. Systems incorporating cartilage emphasize matrix degradation and protease activity, whereas macrophage-inclusive models capture inflammatory amplification and fibrosis. Endothelial co-cultures add vascular complexity and drug-response profiling, highlighting the value of integrated platforms that combine these elements for comprehensive joint modeling. However, given the phenotypic diversity of endothelial cells, the use of joint-derived endothelial cells should be prioritized.

Three-component 3D co-cultures

Several tri-culture systems have been developed to recapitulate complex cellular interactions of the synovial microenvironment. One approach employed a collagen-based scaffold to assemble RA FLS, HUVECs, and monocyte-derived macrophages into a multicellular 3D spheroid, which enabled analysis of inflammatory and angiogenic processes in a physiologically relevant 3D context⁶⁴. The authors found that VEGF and bFGF stimulation enhanced sprouting, while synovial fluid promoted macrophage retention, collectively recapitulating key features of the rheumatoid synovium. Building on this concept, a bacterial-cellulose hydrogel composed of RA FLS, HUVECs, and THP-1 M1 macrophages was introduced to mimic the inflammatory and vascularized synovial niche⁶⁵. The 3D environment improved cell viability and cytokine secretion, with M1 macrophages enhancing RA FLS proliferation, invasion, angiogenesis, and inflammation. Despite the use of THP-1 macrophages limiting physiological fidelity, the model effectively captures macrophage-stromal-endothelial crosstalk. A different strategy used a fibrin-GelMA hybrid hydrogel to tri-culture FLS, HUVECs, and monocyte-derived macrophages to investigate how macrophage phenotypes influence angiogenesis and fibrosis⁶⁶. The model successfully reproduced key aspects of synovial inflammation and vascular remodeling: M1 macrophages induced a surge of angiogenesis that peaked around day 21, while M2 macrophages sustained vessel growth and structural maturation over time. Bridging to cartilage components, another system combined synovial fibroblasts, U937-derived macrophages, and mesenchymal stem cell-derived chondrogenic pellets to mimic early OA⁶⁷. This system produced OA-relevant mediators (IL-8, MCP-1, MMP-1, and MMP-3) at physiological levels without external cytokine stimulation and induced glycosaminoglycan release from cartilage, reflecting the early catabolic-anabolic dynamics characteristic of OA.

Collectively, these studies highlight common themes such as the pivotal role of macrophages and endothelial cells in driving FLS-mediated inflammation and angiogenesis while differing in scaffold composition, macrophage source, and disease context. Models using primary macrophages offer greater physiological relevance, whereas those employing THP-1 cells trade fidelity for scalability. Importantly, cartilage-integrated systems broaden the scope to matrix degradation, underscoring the need for next-generation platforms that unify synovial and cartilage compartments for comprehensive joint modeling.

Microfluidics and chips

Microengineered joint-on-a chip (JoC) platforms put synovium-cartilage crosstalk under tight spatial and temporal control by using defined hydrogels or scaffolds, compartmentalized chambers connected by microchannels, and perfusion to maintain tissue gradients. Compared with 2D and 3D cultures, these chips maintain stable 3D architecture with tuneable matrix properties, enabling precise stimulus delivery and media washout. Individual tissues are kept in linked but physically separated chambers, allowing the isolation of paracrine signaling from direct cell-cell contact⁶⁸.

For example, a microfluidic synovium-on-a-chip platform incorporated patient-derived FLS organoids with integrated organic sensors, enabling real-time, label-free tracking of lining hyperplasia and matrix compaction under precise doses of TNF-α⁶⁹. Continuous perfusion preserved 3D architecture and allowed kinetic mapping of IL-6, IL-8, MMP-1, and MMP-3 secretion concurrent with organic photodiode scattering changes. The platform outperformed static 2D and 3D cultures by linking morphodynamics to secretomics with temporal resolution suited for mechanism and drug testing. Expanding this concept, a two-chamber microfluidic device combined matrigel-based FLS organoids and fibrin-encapsulated chondral organoids in interconnected but distinct compartments, enabling controlled paracrine exchange while maintaining tissue-specific 3D phenotypes⁷⁰. This configuration in co-culture increased chondral organoid size, roundness, and matrix organization, and lowered VEGF secretion levels compared to monoculture in the same system. Multiplex profiling confirmed bidirectional synovium-cartilage signaling, providing a reproducible bridge between micromass and fully integrated organoids for crosstalk-focused interventions. Moving towards osteochondral complexity, a human iPSC-derived osteochondral chip was engineered by differentiating iPSCs to mesenchymal progenitors, embedding them in photocrosslinked gelatine hydrogel, and culturing under dual perfusion to yield a biphasic cartilage-bone tissue in 28 days⁷¹. IL-1β was used to simulate OA by accelerating degeneration in the osteochondral context, implicating subchondral bone-driven catabolism. The anti-inflammatory COX-2 inhibitor, celecoxib, was also evaluated in systemic and intra-articular modes, which reduced expression of catabolic and inflammatory genes and partially rescued expression of COL2A1 and ACAN in the cartilage. Similarly, a compartmentalized JoC in which articular chondrocytes and FLS were cultured in spatially distinct, fluid-linked microchambers was engineered, preserving cartilage zonal architecture and a synovial-like lining while enabling precisely controlled gradients⁷². IL-1β and TNF-α gradients were found to drive cartilage catabolism in this system, whereas cartilage-derived factors attenuated synovial cytokine output. IL-1R and NF-κB blockade reduced catabolic signaling and partially restored ECM homeostasis. Finally, a personalized JoC platform using HA-PEGDA for cartilage and fibrin for synovium, separated by a synovial fluid-mimicking channel that maintained phenotypes and controlled diffusion⁷³. Exposure to OA patient synovial fluid induced elevated IL-1α, IL-6, IL-8, TNF-α, MMP-1, and MMP-13, and increased cellular senescence, while incorporating autologous mesenchymal stem cells into the synovial compartment demonstrated donor-specific therapeutic responses. This study established a patient-specific 3D co-culture platform for modeling synovial-cartilage crosstalk and screening of therapeutic interventions in a personalized OA context.

Together, these platforms introduced the capability to integrate perfusion, spatial compartmentalization, and real-time monitoring, enabling dynamic modeling of synovial-cartilage crosstalk and inflammatory signaling. The differences lie in their complexity, which ranges from single-tissue synovium chips to multi-compartment joint systems, and their application scope, from mechanistic studies to personalized drug screening. This represents a critical step toward physiologically relevant, customizable joint models that can be adapted for viral pathogenesis research and therapeutic development.

Joint models in alphavirus-induced joint inflammation

Despite advances in understanding CHIKV-associated joint pathology, several key mechanistic questions remain unresolved. A central gap lies in defining the primary drivers of tissue damage, specifically the relative contributions of direct virus infection in stromal cells versus immune-mediated inflammatory processes. Additionally, while monocyte and macrophage responses are well-characterized, the functional role of joint-resident stromal cells such as chondrocytes and FLS in sustaining inflammation and contributing to disease progression remains unclear. The mechanisms underlying persistent or disease relapse following viral clearance also remain poorly understood, including the potential roles of viral or antigen persistence and stromal reprogramming. Finally, the spatial and temporal dynamics of immune–stromal interactions within the joint microenvironment, including feedback loops that may sustain inflammation and matrix degradation, have not been fully resolved.

From a virological perspective, in vitro joint models provide a unique platform to dissect these unresolved mechanisms within a controlled, human-relevant environment. Unlike mouse models, which differ in immune architecture and joint anatomy, and patient serum studies, which reflect systemic responses but lack tissue specificity, these human-derived cellular models enable direct interrogation of joint-resident cell behavior under controlled conditions. This allows investigation of localized inflammatory responses, the contributions of individual cell types, cell-intrinsic pathways, and evaluation of therapeutic interventions in a context that more closely reflects human joint physiology.

Within this framework, different model systems address complementary virological questions. 2D joint models provide a foundational platform for studying virus entry, replication kinetics, and interferon responses in defined cell populations, offering high control and reproducibility. However, they lack matrix context and fail to capture tissue-level viral diffusion and cell–matrix interactions. In contrast, 3D models introduce the ECM as a key experimental variable, enabling investigation of alphavirus-induced chondrocyte catabolic activation, synovial fibroblast activation, and extracellular matrix (ECM) degradation in a tissue-relevant context. These systems remain limited by diffusion constraints and short culture duration, restricting their ability to model chronic and persistent CHIKV-associated joint inflammation. Co-culture systems introduce an additional layer of complexity by allowing interrogation of immune–stromal interactions that drive sustained inflammation and matrix remodeling in both 2D and 3D systems. This allows dissection of how interactions between cell types, such as macrophages, FLS, and chondrocytes, shape antiviral, immune, and cytokine signaling, catabolic responses, and tissue damage following CHIKV infection. However, controlling cellular composition, culture conditions, and reproducibility remains challenging. Meanwhile, JoC platforms provide the potential to study virus spread and longer-term inflammatory dynamics under perfused conditions with improved microenvironmental control, but their application to CHIKV remains in early stages and requires further validation. A comparative summary of in vitro joint model platforms is provided in Fig. 3, illustrating how each system can be applied to address specific aspects of CHIKV-induced joint pathology, including viral replication, immune–stromal interactions, and tissue-level responses, alongside their limitations and potential adaptations.

Fig. 3. Summary of in vitro joint models.

The figure illustrates differences across current in vitro joint models and describes their suitability in each aspect for modeling CHIKV infection. Created in BioRender. Ng, L. (2026) https://BioRender.com/xu57yuz.

Collectively, these models should be viewed as complementary tools that resolve distinct aspects of CHIKV-associated joint pathology, from viral replication and innate immune activation to multicellular crosstalk and tissue degradation. We therefore propose a flexible, context-dependent framework rather than a fixed checklist Box 1.

Given the complexity and model-specific nuances of these systems, their limitations and considerations for tuning toward CHIKV-relevant applications are discussed in greater detail in the following section.

Box 1 Recommended endpoints and design considerations for CHIKV studies in in vitro joint models.

• A.
  Core virological readouts

  • Viral RNA quantification
  • Infectious virus output
• B.
  Cell viability and infection-associated cytotoxicity

  • Cell viability assessment
  • Cell death pathways
  • Morphology assessment of cell/tissue integrity
• C.
  Antiviral and inflammatory responses

  • Interferon signaling
  • Pro-inflammatory mediators
• D.
  Matrix degradation and stromal dysfunction

  • Matrix degrading enzymes
  • Stromal identity markers
  • Imaging readouts of ECM responses
• E.
  Immune related readouts (if applicable)

  • Immune cell activation and polarization
  • Functional immune responses
  • Cytokine and chemokine signaling dynamics
  • Immune recruitment and infiltration

Perspectives, limitations, and applications of current joint models in CHIKV-associated joint inflammation

While the cellular and structural complexity of current joint models provides a promising structural and methodological foundation for studying viral joint inflammation, they remain unable to completely capture the full spectrum of CHIKV-induced joint biology and pathogenesis, as they were not originally designed to accommodate viral infection or persistence. These limitations span across biological and technical axes, including infection susceptibility, matrix tunability, lack of modeling relevant immune cells, culture longevity, and variable reproducibility, all of which collectively determine how faithfully the joint models capture the mechanisms and complexity of CHIKV-induced joint inflammation.

Importantly, these limitations should be viewed in the context of model design rather than as fundamental shortcomings. Rather than diminishing their utility, they highlight the need for context-specific adaptation and tuning. When appropriately applied and modified, these systems provide valuable and complementary platforms for dissecting distinct aspects of CHIKV-induced joint pathology. Perspectives and potential strategies to address these limitations are discussed below to further guide their application in a CHIKV-relevant context.

Infection susceptibility

Infection competence is the fundamental prerequisite for any joint model to be relevant to alphavirus research. Despite major advances in organoid and co-culture systems, the extent to which these platforms support viral entry, replication, and persistence remains insufficiently characterized in the joint context. Infection susceptibility is influenced by both structural and cellular factors, and the architectural configuration of the model dictates viral accessibility. In 2D monolayer cultures, cells grow as a homogeneous layer with equal exposure to the media and viral inoculum. This configuration facilitates uniform infection kinetics and synchronized host responses. In 3D constructs, proper cell-cell and cell-ECM interactions produce distinct microenvironmental niches that closely resemble in vivo tissue organization⁷⁴. However, these same interactions establish oxygen, nutrient, and signaling gradients that can limit molecular diffusion and viral accessibility. As 3D cultures increase in size with ECM density, the restricted transport of oxygen and metabolites can give rise to hypoxic or even necrotic cores; this alters local antiviral responses and infection dynamics, particularly in the context of chondrocytes or cartilage organoids⁷⁵. Nevertheless, CHIKV was found to infect chondrocyte spheroids with anti-CHIKV envelope protein detected in central regions³¹. Viral antigen accumulation was more pronounced along the spheroid periphery than in the core, underscoring the influence of diffusion gradients on infection dynamics. Despite these opportunities, direct infection studies in 3D or multicellular joint models remain exceedingly rare. Most platforms have been developed for inflammatory or regenerative studies, not for live viral infection. Even RA-derived synovial organoids, which are inherently immune-active through macrophage-fibroblast interactions, have not yet been adapted to assess viral replication and persistence or antiviral gene responses. Overall, the infection characteristics of 3D joint models remain poorly defined, requiring future systematic investigations of viral entry efficiency, receptor expression, interferon dynamics, and replication kinetics across different models⁷⁶.

Matrix tunability

The composition and density of hydrogels or scaffolds commonly used in constructing 3D joint models profoundly influence host–virus interactions. Matrix stiffness, charge, and porosity influence cytokine diffusion, viral mobility, and cell–cell communication, meaning that hydrogel design can influence both infection kinetics and viral sensing. Dense or highly crosslinked matrices can restrict viral diffusion and cell-to-cell spread, whereas softer matrices may enhance infection kinetics by promoting membrane contact and cytokine exchange^77–79. Beyond this, studies have shown that hydrogel stiffness and crosslinking density also modulate inflammatory tone and cytokine output in macrophages: stiffer matrices produce pro-inflammatory (M1-like) responses, whereas softer matrices promote anti-inflammatory (M2-like) profiles⁸⁰. Similarly, matrix softness enhances stromal secretion of TSG-6, which in turn dampens macrophage NF-κB activity through CD44-mediated signaling⁸¹. Together, these findings suggest a broader principle of hydrogel mechanics and biochemical composition actively tuning cellular immune responses. In the context of joint pathology, it is also important to note that chondrocytes are mechanosensitive cells and are known to transduce matrix cues through integrin and cytoskeletal pathways that converge on the NF-κB, MAPK, and JAK-STAT signaling pathways central to both inflammatory and antiviral responses⁸².

While hydrogels provide a valuable tool for reconstructing joint zonation and mechanical gradients, their physicochemical properties must be carefully calibrated to avoid introducing artificial diffusion barriers or unintended immune activation during infection studies. Importantly, this inherent tunability may also serve as a strength when applied with clear experimental intention. Defined synthetic hydrogels can enable systematic modulation of key parameters such as diffusivity, stiffness, and matrix composition to allow targeted investigation of viral spread or host responses. Furthermore, these systems offer the potential to recapitulate zonal stiffness gradients that closely resemble the native synovial-cartilage interface. Consistent with this, defined culture conditions are essential for successful 3D model development and should be maintained without modification once established to ensure reproducibility. However, most current formulations remain optimized for tissue repair and chondrogenic differentiation rather than infection modeling, and rigorous efforts to connect matrix mechanics with viral spread, cytokine signaling, and interferon feedback are still largely absent. This gap is particularly critical for CHIKV-induced joint inflammation research, where the extracellular environment directly shapes infection dynamics.

Modeling relevant immune components

The progression of CHIKV-induced joint inflammation is fundamentally immune-driven, and the inclusion of diverse immune and stromal populations is therefore critical for building physiologically relevant joint models. However, despite the multicellularity achieved in some co-culture platforms, most current systems still predominantly capture innate immune interactions between stromal and myeloid cells. During CHIKV infection, CCR2⁺ monocytes infiltrate the joint and differentiate locally into macrophages that regulate cytokine production, antiviral defense, and matrix remodeling⁸³. Concurrently, CD4⁺ Th1 cells engage in pathogenic crosstalk with CD64⁺ macrophages, mutually activating one another and amplifying joint inflammation through reciprocal signaling and recruitment of additional immune cells⁸. Integrating these adaptive immune components into co-culture or organoid systems will be critical for modeling both acute antiviral responses and chronic inflammatory persistence. Although no current 3D model has yet incorporated these adaptive immune populations under infectious conditions, a few emerging studies—so far reported only in abstract or pilot form—have begun co-culturing synovial fibroblasts with T cells in a 3D synovial organoid system, demonstrating the feasibility of extending existing organoid systems toward immune-complete and infection-competent joint constructs⁸⁴.

Next, most current arthritis models rely on surrogate macrophage systems—typically THP-1 or U937 cell lines or peripheral blood monocyte-derived macrophages—rather than true tissue-resident synovial macrophages, which understandably are difficult to obtain^64–67. While these substitutes provide tractable and reproducible platforms for studying immune and stromal crosstalk, they do not fully recapitulate the phenotypic heterogeneity, gene expression signature, or immune-suppressive homeostatic functions of resident macrophage subsets found in the synovial lining.

This is a critical distinction as resident barrier macrophages are self-renewing, are locally imprinted by stromal cues, and sustain immune quiescence and barrier integrity at steady state, whereas monocyte-derived or immortalized macrophages largely model the inflammatory influx²⁶. As a result, many 2D and 3D co-culture systems faithfully capture macrophage-driven inflammation but underrepresent resolution programs and lining-barrier functions that are central to chronic CHIKV joint inflammation. Closing this gap will require macrophage populations with resident-like phenotypes, generated either by direct isolation or iPSC-derived differentiation with tissue-specific imprinting⁸⁵.

Culture longevity and reproducibility

Most multicellular joint models are validated over week-scale windows, with their long-term culture constrained by matrix remodeling; mass transport limits of oxygen, nutrients, and waste; and media incompatibilities between different cell types^86,87. In the context of CHIKV-induced joint inflammation, existing in vitro platforms are well-suited to acute-phase readouts but remain ill-equipped for persistent, chronic disease modeling. Clinically, CHIKV-induced joint inflammation can persist for weeks to years, with viral RNA detected in synovial macrophages even after systemic clearance¹⁸. Therefore, joint models that aim to capture the post-acute persistent joint inflammation should support prolonged culture (e.g., from weeks to months) under perfusion with controlled oxygenation, matrices that accommodate compaction, and compartment-specific media that preserve native phenotypes while permitting controlled crosstalk. Without a long-lasting multicellular culture system, it is not possible to recapitulate the physiologically relevant crosstalk, matrix reorganization, and immune remodeling that sustain chronic joint inflammation.

Despite standardized protocols, primary cell lines may exhibit considerable inter-donor heterogeneity in ECM production, cytokine baselines, and innate immune tone, which may be shaped by donor age, ethnicity, and sex^88,89. While this heterogeneity poses challenges for reproducibility, it may also reflect clinically relevant diversity and enhance the translational value of in vitro models. In parallel, variability in hydrogel and scaffolds—particularly lot-to-lot differences in natural matrices like Matrigel and unreported differences in gel stiffness, crosslinking, and residual growth factors—are well-recognized drivers of organoid variability⁹⁰. When biological variability combines with scaffold variation, the resulting spread in cytokine readouts may mask subtle infection-induced effects. Identifying and adopting a suitable non-animal origin matrix may help to mitigate batch-to-batch variation and improve reproducibility across models.

Finally, tuneable JoC platforms represent a key emerging approach towards improving long-term culture and reproducibility, which are critical for modeling the chronic and persistent nature of CHIKV-associated joint inflammation. By implementing controlled perfusion, oxygen regulation, defined matrices, and compartment-specific media, these systems can better sustain tissue viability and microenvironment stability, as highlighted in musculoskeletal roadmaps⁶⁸. Further practical improvements may include culturing under physioxia, the use of synthetic or hybrid hydrogels alongside lot-qualified natural matrices, stratifying donors, and inclusion of reference constructs. Importantly, the application of microfluidic JoC platforms in CHIKV studies may also be constrained by biosafety requirements. While CHIKV is commonly handled under BSL-2, containment requirements may vary across institutions and regulatory frameworks, and higher biosafety settings can further complicate the integration of microfluidic systems due to challenges in containment, leakage risk, and waste handling. While these technologies are still emerging, they are advancing the field to enable physiologically relevant and sustained CHIKV disease modeling.

Future directions

CHIKV-induced joint inflammation lies at the interface of virology, immunology, and joint cell biology. Previous work in arthritis research has established that stromal and immune compartments coordinate matrix turnover and inflammatory tone, presenting the joint as a dynamic immunological organ rather than a passive target. While these models have been extensively used in arthritis, they remain underutilised in the context of viral infections. However, they now offer an opportunity to dissect how alphavirus infection reprograms the joint microenvironment and shifts the balance between immune-driven and stromal-driven injury. In this context, an ideal joint model would combine multicellularity, immune competence, and longevity with preservation of native phenotypes and crosstalk under controlled perfusion and matrices, thereby facilitating exploration of the transition from acute infection to chronic inflammatory pathology while also providing human-relevant systems for evaluating antiviral and anti-inflammatory therapies.

Recent advances highlight JoC and multi-organ-on-chip platforms as particularly promising directions for achieving these goals^91–93. By incorporating controlled perfusion, spatial compartmentalization, and dynamic microenvironmental regulation, these systems enable modeling of viral spread, immune cell recruitment, and sustained inflammatory signaling in ways not achievable in static cultures. Integration with multi-organ platforms further provides the potential to capture systemic–tissue interactions, such as immune cell trafficking and cytokine exchange between circulation and joint compartments, which may be critical for understanding CHIKV-associated disease progression.

Looking forward, the incorporation of emerging technologies such as artificial intelligence and advanced imaging pipelines could further enhance these platforms by enabling automated data acquisition, real-time monitoring of infection dynamics, and high-dimensional analysis of cellular responses. Such interdisciplinary integration has the potential to transform in vitro joint models from descriptive systems into predictive platforms for studying CHIKV pathogenesis and therapeutic responses. Together, these advances position joint-on-chip and multi-organ systems as key next-generation tools for bridging mechanistic insight and translational application in viral arthritis research.

Conclusion

In conclusion, in vitro joint models provide complementary platforms to investigate CHIKV-associated joint pathology at the level of tissue-resident cells and multicellular interactions. While each system presents inherent limitations, their combined use enables mechanistic insights into viral infection dynamics, inflammation, and matrix degradation that are difficult to resolve in vivo. Continued refinement of these models will be essential for advancing CHIKV research and therapeutic development.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

L.F.P.N. conceived the initial perspective and overarching direction of the manuscript and contributed to scientific discussions, critical editing, and manuscript refinement. I.J.H.S. developed the conceptual framework, performed the literature review and synthesis, prepared the figures, and wrote the original draft of the manuscript. L.L.Y.C. contributed to scientific discussions, conceptual refinement, and critical editing of the manuscript together with I.J.H.S. L.R. contributed to manuscript refinement and critical revision. All authors reviewed and approved the final manuscript.

Peer review

Peer review information

Communications Biology thanks J Kennedy Amaral, Hugh Watson, and Cheng Wang for their contribution to the peer review of this work. Primary handling editors: Joao Valente. A peer review file is available.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s42003-026-10559-4.