This study demonstrates the therapeutic efficacy of EVs as a treatment in the AIA model of inflammatory arthritis. All EV treatments prompted amelioration of clinical symptoms of AIA with increased effect seen in reduction of joint swelling when treated with EV-2%O2 or EV-Pro-Inflam. In line with this, histopathological examination of the joints showed improved histological features following EV treatments compared to controls. Mechanistically, EVs acted remarkably different from their cells of origin (MSCs) when co-cultured with CD4 + T cells from healthy spleens. EVs reduced Th17 polarization without affecting T cell expansion and proliferation, while MSCs increased T cell expansion and Th17 polarization, and lessened T cell proliferation. Ex vivo, CD4 + T cells isolated from spleens of arthritic mice treated with EVs showed significantly reduced Th17 polarization that rebalanced the Treg:Th17 ratio. Together, this suggests the reduction in Th17 cells, that led to restoration of the Treg:Th17 ratio which is typically unbalanced in inflammatory arthritis, as the main therapeutic mode of action of MSC-derived EVs.
To further dissect the therapeutic mechanism of action of MSC-derived EVs, circulating levels of TNF-α, which is a key driver of pathogenesis in RA and therapeutic target in biological treatments, and IL-10, which is a master regulator of anti-inflammatory immune responses, were measured in mice with AIA at day 3. While IL-10 was not detected in our assay, similarly low levels of TNFa were detected in untreated and EV-treated mice. This suggests that TNFa blockade and IL-10 modulation are not the mechanisms by which MSC-derived EVs improve AIA and therefore EV treatment may represent an alternative option for patients who do not respond to biological interventions such as anti-TNF therapy. Nevertheless, a wider timecourse of serum TNF-α measurement (e.g. 14, 21 and 28 days) remains to be investigated.
Our previous study demonstrated reductions in Th17 cells following CM-MSC treatment compared to control and MSC treatments. It also showed an increase in IL-10 expression and proportions of IL-4 expressing Th2 cells as well as cellular expression levels of IL-4 following MSC treatment(3). We also demonstrated an improved Treg:Th17 effector cell ratio following CM-MSC treatment compared to MSC treatments or controls(3). With these results in mind, the present study indicates that MSCs, not their secretome, were driving increased IL-10 expression and were responsible for the reduction in Th1 and increase in Th2 cells seen in our previous study. Remarkably, EVs, a component of the CM-MSC, are capable of significantly reducing Th17 cell numbers. This represents a significant finding in the search for immunomodulatory therapeutics for treating autoimmune disorders where an imbalance in T cell polarisation (Th1:Th2 or Treg:Th17) is integral in disease pathology. Moreover, EV treatment in this study resulted in a highly improved (2.40-fold over controls) Treg:Th17 ratio compared to our previous results using whole CM-MSC treatment (2.13-fold over control) or MSC treatment (1.47-fold over control) (3).
The normal ratios of Treg:Th17 cells can be calculated from previously published studies by looking at proportions of either of these cell types in experimental investigations. The proportion of regulatory T cells in a healthy human is commonly reported in the range of 2.3–4.9 ± 1.1%(45–50), and the proportion of Th17 cells is reported at 0.50–2.13%(49–53) in peripheral blood, with typical murine Treg proportions reported at 10.2–12.1% in spleens(45, 47, 48, 54, 55) and Th17 cells at 1.17% in spleens(47, 56) respectively.
Therefore, using published research as above, a typical ratio of Treg:Th17 cells in healthy humans can be estimated to range from 1.1:1 to 9.8:1 and in healthy mice from 6.0:1 to 7.1:1.
In human RA sufferers the ratio of Treg:Th17 becomes imbalanced with peripheral blood Treg cells ranging from 1.0 to 16.9%(41, 53, 57) and Th17 cells ranging from 0.91 to 0.96%(42, 51) in peripheral blood, increasing to 2.20–9.09% in synovial fluid(42, 53). An analysis of published data from a range of studies shows that RA sufferers experience a drop in FOXP3 + Tregs from 5.94 ± 1.40% in healthy subjects to 4.44 ± 0.96% in patients, representing a 0.77 ± 0.07-fold decrease (p = 0.021, n = 17)(58). These data can be analysed to demonstrate a Treg:Th17 ratio ranging from 1.0:1 to 18.6:1 in peripheral blood and from 0.1:1 to 7.7:1 in synovial fluid of RA sufferers.
The proportions of Treg and Th17 cells in RA sufferers has been directly linked to the severity of disease, and restoring the Treg:Th17 balance has the potential to promote homeostasis and positive clinical outcomes by minimising inflammatory responses in a range of autoimmune disorders. Here, we demonstrate that untreated (control) spleens of mice with AIA display a Treg:Th17 ratio of 5:1 that is increased to 12:1 upon EV and EV-depleted CM-MSC treatments. This result reinforces the use of CM-MSC as a therapeutic for RA, with EVs being a convenient and chemically definable option for delivering its therapeutic action. Furthermore, the data in this study and our previous research suggest that CM-MSC might be a more effective therapeutic than MSCs for targeting inflammatory immune disorders such as RA, whilst MSCs might be more suitable for the treatment of inflammatory disorders, including Crohn’s Disease, Lyme’s arthritis, Multiple Sclerosis, Systemic Lupus Erythematosus, where tissue regeneration and the Th1/Th2 balance are an issue(59). The concomitant rise in IL-4 expressing Th2 cells and IL-10 in medium of MSC treatments but not CM-MSC seen previously would therefore be attributed to cellular interactions rather than secretome components.
Th17 and Treg differentiation are antagonistic responses to TGF-β expression, with low levels of TGF-β prompting synergy with IL6 and IL21 that in turn increases the expression of IL-23 receptor, resulting in upregulation of Th17 cell production. Conversely, highly expressed TGF-β prompts naïve helper T cells to express FOXP3 and retinoic acid receptor-related orphan receptor-γt (RORγt), and to repress the expression of IL-23 receptor, thus tilting the balance towards the generation of Treg cells where RORγt function is inhibited due to TGF-β induced FOXP3(60). Addition of IL-6, IL-21 and IL-23 restores RORγt function and associated Th17 differentiation(60–62). In this regard, our results demonstrate inhibition of IL17a production (or repressed IL-23 receptor expression) through EV treatment, inhibiting IL17a production whilst leaving the induction of FOXP3 expressing Treg cells unaffected.
Here, EV treatments did not induce IL-10 production, indicating that MSC-derived EV do not mediate anti-inflammatory effects via IL-10 as those seen in our previous study where MSCs and CM-MSC induced IL-10 release upon co-culture with T cells(3). IL-10 functions to inhibit monocyte-derived cell production of immunomodulatory cytokines such as IFN-γ and IL-4. This increases the likelihood that MSCs will convey a greater impact on Th1 and Th2 expressing cells than their secretome alone(7). Additionally, IL-10 expression prevents dendritic cell trafficking to lymph nodes, which presents a hypothesis for MSC action in reduced Th1 cell recruitment; however, here we demonstrate that lymph nodes show reductions only in IL17a expressing Th17 cells when treated with EVs, in accordance with the lack of IL-10 expression in medium previously shown (3).
In vivo, dendritic cell EVs which are high in expression of intercellular adhesion molecule 1 (ICAM-1) show high affinity binding to the surface of activated CD4 + T cells via the integrin leukocyte function-associated antigen 1 (LFA-1) without the need for T cell receptor specificity(26). This recruitment enables activated T cells to present acquired MHC class II peptide complexes from DCs and this can influence the activation and proliferation of the wider activated T cell population(63). MSCs are constitutively low in ICAM-1 expression(64, 65) however these cells have been shown to upregulate expression of ICAM-1 when co-cultured with activated CD4 + T cells in response to the presence of increased pro-inflammatory cytokine expression, leading to increased immunosuppressive properties of MSCs(64, 66). It follows then that priming MSCs with pro-inflammatory cytokines will enhance T cell adhesion to secreted EVs through increased binding efficiency and this has been shown to increase treatment potency(67, 68). It has been suggested that the immunomodulatory properties of EVs are contingent on pro-inflammatory priming(21, 69). However, the present work demonstrates that alternative priming methodologies, namely hypoxia, convey similarly enhanced immunosuppressive capabilities to secreted EVs.
Most studies utilise pro-inflammatory IFN-γ(70, 71) and/or TNF-α(72, 73) to facilitate priming, although IL-1β(74) and IL-17a(75) have also been applied. Regardless of cytokine selection, the broad outcomes of exposing cells in culture to a pro-inflammatory environment is to initiate strategies that would re-establish homeostatic control in vivo, such as anti-inflammatory effects prompted through alteration of the EV cargo(76) and signalling to reduce recruitment of inflammatory mediators, expression of pro-inflammatory cytokines, T cell polarisation and proliferation(71, 72, 77), and restoration of homeostatic ratios of leukocytes and T cells(73, 78). A primary mechanism involved here is IFN-γ-mediated upregulation of indoleamine 2,3-dioxygenase (IDO)(70, 72, 73).
Our study demonstrates that pro-inflammatory priming of MSCs does indeed increase the efficacy of EVs in treating inflammatory arthritis as observed in these studies. Our results suggest that the in vivo response to pro-inflammatory primed EV treatment in the AIA model of inflammatory arthritis is not primarily through inhibition of T cell proliferation, but through suppression of CD4 + Th17 effector polarisation.
However, we also show that hypoxic priming of MSCs results in the production of EVs that alleviate joint swelling as effectively as EVs secreted by MSCs primed with pro-inflammatory cytokines. This reduced swelling by EV-2%O2 treatment did not translate to significant improvement in the fine histological structures assessed in our arthritis index. Conversely, whilst both EV-2%O2 and EV-Pro-Inflam treatments inhibited polarisation of T cells towards pro-inflammatory Th17, EV-Pro-Inflam did not reduce the expression levels of IL-17a in Th17 cells as seen in EV-2%O2 treatments. We hypothesise that whilst the effectiveness of these treatment methodologies is similar in potency, the underlying mechanisms of action differ. Interestingly, MSCs under hypoxic culture have been shown to promote anti-inflammatory M2 macrophage polarization through a mechanism also dependent upon ICAM-1 adhesion(79), although hypoxic MSCs exhibit reduced reactive oxygen species (ROS) levels and increased resistance to ROS stress and upregulate secretion of growth factors and cytokines such as TGF-β, IL-8, IL-10 and PGE2, which are also implicated in macrophage polarisation and MSC immunomodulatory capacity(36, 79, 80).
Hypoxia inducible factor 1α (HIF-1α) is a primary sensor of hypoxia, and hypoxic conditions are physiologically common in regions of inflammation(81). Reduced oxygen environments are physiologically appropriate for MSCs resident in the bone marrow, where the available oxygen tensions are equivalent to 1–9% dependent upon distance from the vasculature, with the MSC niche likely to be resident in regions experiencing the lower end of this scale(36, 82). This makes EV-2%O2 a physiologically relevant treatment option derived from healthy conditions. We see primarily Th17 immunomodulation through EV treatment. The underlying mechanism by which EV-2%O2 treatment affects immunomodulation may be different to the mechanism(s) responsible for EV-Pro-Inflam-mediated effects, yet both hypoxic and pro-inflammatory cytokine priming systems may converge in affecting STAT3 activation. As described, RORγt is a master regulator of Th17 differentiation(60); however, STAT3 induces RORγt expression and is activated by IL-6, IL-21 and IL-23, which are key cytokines in promoting Th17 differentiation(60, 83). Moreover, STAT3 activation has been shown to be mandatory for the development of Th17 cells and Th17 related autoimmunity, such as seen in RA, and an absence of STAT3 activation inhibits Th17 formation and promotes Th1 differentiation through its absence, skewing pro-inflammatory immune responses from Th17 to Th1(83). STAT3 has been suggested as a therapeutic target in the treatment of autoimmune inflammatory disorders such as RA(81). In normoxic conditions the HIF-1 gene is inhibited by proteasome mediated degradation of HIF-1α and hydroxylation of HIF-1 by Factor Inhibiting HIF-1 protein (FIH)(84, 85). Hypoxia inhibits the function of key hydroxylases, von-Hippel-Lindau (VHL)-ubiquitin ligase complexes that mark HIF-1α protein for degradation in normoxia, and FIH. This in turn promotes increased HIF-1α protein in the cell due to reduced catabolism in degradative processes(84). Hif-1α dimerization with HIF-1β in the nucleus leads to transcriptional upregulation of genes in response to the hypoxia cues, primarily to drive metabolism from oxidative phosphorylation and oxygen-dependent ATP production towards glycolytic energy metabolism and decreased mitochondrial oxygen consumption. This reduces cellular dependency on oxygen, alongside activity that will restore oxygen supply to an affected region, such as angiogenesis achieved through regulation of target genes such as vascular endothelial growth factor (VEGF) and cathepsin D (CTSD)(81, 84, 85).
Priming of MSCs with hypoxia activates HIF-1α and increases cell signalling to induce these shifts in tissue energy metabolism and angiogenic responses to deal with the compromised inflammatory environment. It therefore follows that treatment with hypoxically induced changes to signalling molecules, such as EVs, conveys these coping mechanisms to the environment into which treatments are induced, promoting tissue stabilisation and homeostasis into the diseased joint cavity. These effects, in combination with enhanced immunomodulatory capacity due to increased TGF-β and IL-10 secretion and reduction in ROS due to hypoxic culture, led us to hypothesise that this mechanism of action may be more beneficial to cells. Prior exposure to pro-inflammatory cytokines prompt crisis responses in MSCs as evidenced through the induced cell death following continued exposure to pro-inflammatory cocktails we witnessed in our cell cultures if cytokine exposure exceeded 48 hours (data not shown).
Hypoxic culture of MSCs has been shown to aid maintenance of stemness and pluripotency in cells whilst very low oxygen tensions can assist in maintaining quiescence in resident stem cells, which may be beneficial in clinical therapies dependent upon cell secretome rather than tissue regeneration(36). In this study, we observed a trend for increased protein content in hypoxic EV preparations compared to pro-inflammatory preconditioned cells, suggestive of a more potent EV treatment utilising this method of cell priming, which more closely mimic the physiological microenvironment. Previous studies have also demonstrated benefits of hypoxic treatment of MSCs and cell therapies for tissue repair in regenerative medicine treatments for arthritic disorders(86–93).
We have previously demonstrated CM-MSC reduced cartilage degradation by aggrecanase activity through ADAMTS and MMPs cleavage3. RA disease manifests with increased circulating citrullinated epitopes of degraded proteoglycans, including aggrecan, in > 60% of sufferers. This has implications in the production of pro-inflammatory cytokines directly linked to increased Th17 polarisation(41, 42, 58, 59). Circulating MHC-II complexes with cartilage epitope fragments may be implicated in the autoimmunity developed in RA through polarisation and activation of T cells, primarily Th17 effector cells(63). MMP activation occurs in response to pro-inflammatory signalling(63), and EVs sourced from stromal cells have been shown to carry both matrix metalloproteinases (MMPs), including MMP1, MMP2, MMP3, MMP7, MMP9 and MMP10, and ADAM9, ADAM10 and ADAMTS12(94, 95), and also tissue inhibitors of metalloproteinases (TIMPS) such as TIMP1, TIMP2 and TIMP3(94). Our previous study provided evidence of reduced activity of catabolic enzymes following CM-MSC treatment even beyond the effects seen with MSC infusion(3), thus the cargo of EVs may be the source of this mechanism of action in CM-MSC tests. EVs may function as a source of inhibitors of aggrecanases or equally introduce enzymes involved in epitope formation. As such, it would present an optimisation strategy to select EVs that carry a cargo high in TIMPs and anti-inflammatory cytokines, and low in aggrecanases. This presents a potential avenue for future investigation.
One common limitation in studies examining EVs is the presence of contaminating proteins present in EVs isolation through ultracentrifugation(24). Given the knowledge that both hypoxic and pro-inflammatory priming of MSCs can lead to increased expression of secreted proteins, it is important to ascertain whether the effects seen here could be the result of EVs contamination with proteins that also act to promote immunosuppression. In this study we utilised serum-free culture during EVs isolation to eliminate protein or EV contaminations from serum, and we included a PBS wash step during 100,000G ultracentrifugation to reduce residual protein. When measured, we saw no detectable IL-10 in T cell co-cultures suggesting an absence of IL-10 contamination in EVs preparations.
In this study, our primary aim was to evaluate the efficacy of EV treatments in vivo during AIA. We show that MSC priming leads to the release of EVs that if administered to mice with acute inflammatory arthritis significantly ameliorate disease pathogenesis, mainly by inhibiting Th17 polarization. Future studies will define the composition and sub-vesicular localisation of proteins in EV cargos.