Targeting Tregs in Juvenile Idiopathic Arthritis and Juvenile Dermatomyositis—Insights From Other Diseases

Regulatory T cells (Tregs) are believed to be dysfunctional in autoimmunity. Juvenile idiopathic arthritis (JIA) and juvenile dermatomyositis (JDM) result from a loss of normal immune regulation in specific tissues such as joints or muscle and skin, respectively. Here, we discuss recent findings in regard to Treg biology in oligo-/polyarticular JIA and JDM, as well as what we can learn about Treg-related disease mechanism, treatment and biomarkers in JIA/JDM from studies of other diseases. We explore the potential use of Treg immunoregulatory markers and gene signatures as biomarkers for disease course and/or treatment success. Further, we discuss how Tregs are affected by several treatment strategies already employed in the therapy of JIA and JDM and by alternative immunotherapies such as anti-cytokine or co-receptor targeting. Finally, we review recent successes in using Tregs as a treatment target with low-dose IL-2 or cellular immunotherapy. Thus, this mini review will highlight our current understanding and identify open questions in regard to Treg biology, and how recent findings may advance biomarkers and new therapies for JIA and JDM.


INTRODUCTION
CD4 + FOXP3 + regulatory T cells (Tregs) are a subset of CD4 + T helper cells present in lymphoid and non-lymphoid tissues, and are crucial for mediating tolerance to self, preventing allergies and controlling immune reactions after infections (1). They develop in the thymus or are induced in the periphery and exhibit contact-dependent and -independent mechanisms of action (1). Inactivating mutations in FOXP3 lead to multi-organ autoimmune disease [immune dysregulation, polyendocrinopathy, enteropathy, X-linked syndrome (IPEX)], highlighting the importance of Tregs (2). Importantly, it is becoming clear that the local microenvironment affects the phenotype and function of tissue-localized Tregs, which also have additional roles in repair and regeneration (3).
Treg-tissue interaction might be particular important in autoimmunity with tissue-specific presentation, such as juvenile idiopathic arthritis (JIA) and juvenile dermatomyositis (JDM). While JIA is the most common inflammatory rheumatic disease in children, JDM is rare. JIA is characterized by persistent arthritis and subtype-dependent symptoms [reviewed in (4)]. Here, we focus on polyarticular and oligoarticular JIA, which present without involvement of systemic organs or skin. JDM is characterized by inflammation of muscles and skin, resulting in muscle weakness and rashes [reviewed in (5)]. Interestingly, for both conditions researchers may take advantage of clinical sample collection from the site of inflammation: synovial fluid (SF) drained during therapeutic joint injection (JIA) and biopsies (mostly muscle, JDM). While some patients respond to therapy, others do not and studying the underlying differences may lead to better understanding and treatments.
Here, we discuss recent advances in the understanding of Treg biology in oligo-/ polyarticular JIA and JDM, and what we can learn about Treg-related disease mechanisms, treatments and biomarkers from other diseases.

ALTERED TREGS IN JIA AND JDM
The phenotype of CD4 + FOXP3 + Tregs in JIA has been considerably characterized in the past (6) with the molecular roles of FOXP3 in JIA reviewed by Copland and Bending in this special collection (7). It is now clear that the Treg TCR (T cell receptor) repertoire is highly restricted in JIA, both at the site of inflammation (8-11) and in circulation (10, 12). Interestingly, in blood only Tregs but not conventional CD4 + non-Treg cells (Tconv) are more clonal (10, 12). Some suggest that the TCR repertoires of Tregs from SF and peripheral blood (PB) significantly overlap (8), while others only found a very small overlap (9, 11). These differences might be explained by different sequencing depth and analysis strategies and/or by different Treg subsets studied: total (11) or effector Tregs defined by HLA-DR (8) or CD161 expression (9). Further, one study found that SF Tregs, but not Tconv, share specificity at an amino acid sequence level among different patients (10), suggesting disease-associated Treg clones might foster JIA.
Besides a restricted TCR repertoire, Tregs from the JIA inflammatory sites show unstable FOXP3 and CD25 (13), altered homing markers (9), cytokine production (6, 9), deficiency in specific chemokine production (14), and low responsiveness to IL-2 (13)-indicating impaired Treg function in JIA. Nevertheless, many reports found that JIA SF and PB Tregs are fully demethylated (8, 13), thus committed to the Treg-lineage, and suppressive in vitro (6, 8, 9, 13, 15). Hence, JIA Tregs are likely functioning inappropriately or insufficiently in the context of the inflammatory microenvironment. Interestingly, adding SF to in vitro cultures can both increase/stabilize Treg FOXP3 expression (11, 16) and in situ induce effector T cells to be resistant to Treg-mediated suppression ex vivo (17, 18). Thus, more research is needed to decipher the effects of the inflammatory microenvironment on Treg function.
In comparison, we know little about the contribution of Tregs to JDM pathogenesis. Similar to JIA, the Treg repertoire is restricted with a lack of diversity (12). FOXP3 + Tregs were found to be enriched in JDM muscle compared to muscle tissue from patients with Duchenne muscular dystrophy (19). Since the latter is already enriched in Tregs compared to normal muscle (20), this suggests a hyper-enrichment in JDM in response to autoimmune inflammation. PB Tregs of active JDM also appear less suppressive in vitro with decreased expression of CTLA4 (19). Adult DM/ polymyositis muscle biopsies are also enriched with Tregs (21). Interestingly, both Treg and effector T cell numbers decreased post immunosuppressive therapy in adult myositis, suggesting that Treg enrichment is a response to inflammation. However, juvenile and adult DM have different clinical presentation (22) and JDM PB express more Th17-type and FOXP3 transcripts (23). JDM and other myopathies are characterized by a type 1 IFN signature (24-26) and interferons may be a potential therapeutic target (27), but their effects on Tregs remain to be investigated.
Tregs are crucial in resolving muscle injury in animal studies (28) and Treg-deficient mice develop more severe myopathies in response to antigen, while adoptive Treg transfer prevents inflammation (29,30). Thorough immune-profiling recently revealed pan-tissue and tissue-specific signatures and enhancers of murine Tregs (31). The muscle Treg signature was highly enriched in cell cycle genes, showed a dynamic response to injury and was more similar to circulating Treg signatures than to other tissue Tregs (31), indicating that muscle Tregs might acutely infiltrate muscle and are not necessarily long-term resident cells. While myopathy is a defining characteristic of JDM, skin inflammation and rash are other symptoms (5). Skinresident Tregs are crucial for immune homeostasis (3) and have been characterized in health and various disease settings (32). However, studies on JDM-affected skin are lacking, and more work is needed to characterize JDM skin-resident Tregs.

TREGS AS A BIOMARKER?
JIA and JDM can exhibit an unpredictable disease course. While mounting evidence indicates that an early aggressive treatment is best for severe disease (4, 27, 33,34), the disease course is unpredictable at presentation. Additionally, due to potential short-and long-term side effects children should not be exposed to unnecessary medication. Unfortunately, once a patient appears to be in clinical remission (on or off medications), disease may flare without any notice or obvious trigger ( Figure 1A). Indeed, among JIA patients who are in clinical remission, 30-50% experience flares (35,36).
Hence, reliable biomarkers need to predict (i) the future disease course, (ii) treatment response, and (iii) the safety for medication withdrawal during clinical remission ( Figure 1B).
Inflammation markers in the serum can indicate disease activity and potentially treatment response in JIA [reviewed in (36)]. In JDM, histology of biopsies and myositis-specific autoantibodies can indicate future disease severity or complications [reviewed in (27)].
Only a few putative biomarkers probe the immunoregulatory balance in autoimmune arthritis and myopathies. The frequency of inflammation-associated Tregs (HLA-DR + ) in PB was proposed as a biomarker for disease activity in arthritis (8). TCR sequence overlap of these PB HLA-DR + Tregs with SF Tregs in JIA, and an increase of HLA-DR + Tregs in active rheumatoid arthritis (RA) were found. Low expression of the immunoregulatory receptor CD39 has been suggested as an FIGURE 1 | Current and desired disease progression models. (A) Current "Trial-and-error" model: Upon diagnosis the first line of treatment is started, which may lead to remission, or partial remission. Often, a second, third or fourth treatment strategy needs to be implemented when the previous treatments are not effective. Choice of treatments is guided by previous experience, e.g., upon presentation non-steroidal anti-inflammatory drugs with glucocorticoids are used, after a couple of month many patients are switched to methotrexate as first disease modifying therapy, often followed by anti-TNF-α agents with/without methotrexate (4). This disease progression and subsequent staggering of therapy can result in irreversible damage and long-term therapy. (B) Desired personalized/biomarker-driven model: Biomarkers could be used for prediction and aid decisions at the following stages: disease trajectory, selection of treatment, early response to therapy, remission/minimal disease activity and the risk of flare upon withdrawal of therapy, ongoing monitoring of immune activity and risk of flares. Biomarkers might thus contribute to more efficient therapy, pre-empt flares, and minimize short-and long-term effects of flares and reduce long-term damage. Lines represent models of disease progression, with line thickness representing frequency estimates.
indicator of methotrexate resistance in RA (37). Also, response to the TNF-α blocker adalimumab could be predicted by a Treg increase in PBMCs from RA patients cultured with adalimumab prior to treatment (38). Finally, the soluble form of the high affinity IL-2 receptor α chain (CD25), crucial for Treg phenotype and function, might be a biomarker for adult myositis disease activity (39).
In the recent past, gene signatures have been defined as multi-parameter biomarkers. Thus, far, efforts to define JIA immune-based gene biomarkers have focused on whole genome expression profiling (40)(41)(42) and epigenomic signatures (43,44). JIA displays an altered immune signature, which changes during remission, but does not return to a state comparable to healthy controls (41,42). Myositis is characterized by type 1 IFN signatures (27). While interesting and highlighting potential disease mechanisms, whole-genome/exome expression profiling is not feasible for routine clinical practice due to cost, logistics and data interpretation. We have recently developed a Treg gene signature associated with Treg competency using the clinicallyapplicable multiplex platform nanoString (45). NanoString is fast and fewer than 10,000 lysed cells are sufficient without the need to purify RNA. Although the proportion of Tregs that express FOXP3 was similar between type 1 diabetes (T1D) and controls, there was a significant change in their Treg signature (45). Future work will elucidate whether the Treg gene signature may also be used as a biomarker in JIA and other autoimmune conditions.
In summary, some progress has been made, but more biomarkers are needed for biological disease activity, prognosis, treatment success, and risk of flares. Further, a consensus of criteria to describe active/inactive disease is needed to better estimate the currently widely variable incidence of clinically inactive JIA disease (46). For JDM, a comprehensive set of criteria to assess disease activity and damage has been proposed (47).

TREGS AS THERAPEUTIC TARGET/TOOL?
Convincing evidence demonstrates that functioning Tregs are crucial to prevent autoimmunity and our understanding of how different immunotherapies affect Tregs has improved.

(Unforeseen) Treg Effects of Immuno-Therapy
High levels of TNF-α in the inflamed JIA joint (32, 48) offer a clear rationale for anti-TNF therapy in JIA with marked success (4, 49). Anti-TNF therapy has also been used in refractory JDM (50), but with mixed evidence for its effectiveness (27, 51-53).
Whether drugs such as ustekinumab and tocilizumab act on Tregs directly or through changing the microenvironment is unclear. Human Tregs can express the receptors for IL-6 (86), IL-12 (87), and IL-23 (88), but evidence for direct drug action on Tregs is lacking.

Treg (-Targeted) Therapy
Adoptive transfer of Tregs has been shown to be safe and possibly effective at reducing inflammation, inducing transplant tolerance, preventing graft-vs.-host disease (GVHD) and treating autoimmunity [reviewed in (101)].
Important considerations for Treg-therapy currently under investigation are the source of therapeutic cells, antigenspecificity and possibly tailoring homing characteristics for improved activity. Isolating and expanding sufficient numbers of Tregs from patients awaiting transplantation, under immunosuppression or with autoimmune disease is feasible and can restore their function (101-103), although achieving clinically relevant Treg numbers from pediatric JIA and JDM patients might prove challenging. Third-party Tregs from umbilical cord blood have been found safe and possibly effective as GVHD prophylaxis in adults (104,105) and pediatric thymus-routinely removed during pediatric cardiac surgerymight be a plentiful source for highly functional therapeutic Tregs (106,107). Antigen-specific Tregs are more effective than polyclonal Tregs for therapy and with recent successes of chimeric antigen receptor (CAR) T effector therapies for cancer, there has been a surge to adapt this technology to generate CAR-Tregs [reviewed in (108)]. While generation of antigen-specific Tregs recognizing allogeneic HLA-molecules is relatively straightforward in transplantation, generation of CAR-Tregs for autoimmunity without known antigen (i.e., JIA) might be difficult. Still, CAR-Tregs reacting with antigen found at the site of inflammation (i.e., JIA joints or JDM muscle) could activate Tregs locally. Alternatively, Tregs could be conditioned in vitro to home to specific sites (107) or Tregs could be injected locally, as shown with intra-dermal injection of Tregs to inhibit murine allograft skin inflammation (109).
However, increasing Treg numbers alone might not be sufficient to overcome the highly inflammatory environment and effector cell resistance. Thus, to achieve sustained remission combination-therapy might be necessary to reduce the inflammatory milieu paralleled with boosting Tregs to maintain a renewed tolerance.

CONCLUDING REMARKS
Taken together, it is clear that Tregs present challenges and opportunities in JIA and JDM research and clinical management (Figure 2). Their phenotype and function are clearly altered in JIA and JDM, targeting them might improve disease outcome and Tregs could be used as biomarkers to gage the state and progress of disease.
The role of the microenvironment on Treg function and phenotype in JIA-and JDM-affected tissues remains to be explored further. Researchers should take advantage of biopsies taken for clinical diagnosis (JDM) and SF aspirated during therapeutic joint injections (JIA). Novel techniques, such as single cell sequencing, multidimensional mass/flow cytometry and microscopy, will aid using clinical samples to their full potential (125)(126)(127). Additionally, co-culture with SF or muscle-derived cells could highlight how the microenvironment affects Tregs. Since JDM in particular is a rare disease, collaborations between groups are crucial to increase sample size for fundamental research, biomarkerfinding and -validation studies and controlled treatment trials. This could be achieved by consortiums similar to juvenile diabetes research foundation (JDRF) biomarker working group for T1D (114) and the immune tolerance network trials (128).
While there is progress toward unified measures of disease activity (46,47), these will need to be tested and verified, followed by development of feasible, reliable and cost-effective biomarkers to predict disease activity, risk of flare and ideal treatment strategies. The ultimate goal, aided by biomarkers, is to go from a trial-anderror treatment approach toward a more efficient and personalized medicine approach with more patients achieving drug-free remission without major long-term disabilities (Figure 1).
Success of various agents affecting the immunoregulatory balance in other diseases point to potential uses in JIA and JDM. Any (new) therapy will need to be considered in regards to both effector cells AND Tregs, since some therapies might have unexpected effects on Tregs. Thus, it is important to continuously build our understanding of how various agents affect the immunoregulatory balance.
In conclusion, important recent advances might lead to valid future contributions to the widened arsenal of treatment options available to restore the immunoregulatory balance in a heterogeneous disease spectrum.

AUTHOR CONTRIBUTIONS
AP conceived the review. RH and AP reviewed the literature and co-wrote the manuscript.