T cells in myositis

T cells of both the CD4 and CD8 lineage are commonly found in affected tissues of patients with idiopathic inflammatory myopathies, but understanding the contribution of these cells to immunopathogenesis remains challenging. Given recent advances in identifying more myositis-associated autoantibodies and their putative targets, we suggest that studies on autoreactive T cells targeting those autoantigens are one way forward. Another (so far, more frequently used) approach comes from studies on effector T cells in the context of myositis. This review summarizes recent advances and current hypotheses in both of these contexts.


Background
Idiopathic infl ammatory myopathies (IIMs), also known as myositis, are a group of rheumatic disorders clinically characterized by muscle weakness, leading to disability, decreased quality of life, and a reduced life expectancy. Although this is a relatively rare disease, our understanding of risk factors and the underlying immunopathogenesis has increased substantially in recent years. Myositis shares many features with rheumatoid arthritis and systemic lupus erythematosus, namely as diff erent examples of disabling chronic infl ammatory syndromes, which can be reevaluated in the light of distinct genetic and environmental contributions [4]. Common traits between these rheumatic disorders include a major histocompatibility complex (MHC) class II association, infl am matory cell infi ltrates in aff ected tissues, and the presence of predictive or disease activity-associated autoantibodies (or both). Taken together, these observations point to a central role for adaptive immune reactions in disease manifestation.
Th e spectrum of infl ammatory myopathies is getting broader, and the classifi cation criteria for the IIMs, designed by Tony Amato on behalf of the Muscle Study Group, proposed the following categories: (i) inclusion body myositis, (ii) polymyositis (PM), (iii) dermato myositis (DM), (iv) non-specifi c myositis, and (v) immunemediated necrotizing myopathy [5]. Some IIMs share common histopathological features of leukocyte infi l tration, preferentially T cells and macrophages in skeletal muscle tissue, whereas others display no or spare perivascular and perimysial infi ltrates. Novel studies of this latter group which are based on detailed immunopathology suggest that the predominant abnormal histological feature is instead membrane attack complex (MAC) deposition on the sarcolemma in both nonnecrotic and necrotic muscle fi bers [5,6]. Many patients have manifestations besides in the muscles, such as in the lungs (mostly PM), skin (DM), and sometimes in the joints. Additionally, some patients display more than one rheumatic diagnosis, and systemic sclerosis is the most common connective tissue disease associated with IIM [7].
Both CD4 + and CD8 + T cells have been described to be present and active in patients with myositis. Th e presence of cytotoxic CD8 + T cells has been attributed to virus or intracellular bacterial infections, which would generate potent eff ector cells. CD8 + T cells are often subdivided on the basis of their diff erentiation level, fi rstly into naïve and activated/memory T cells; the latter subset can be further subdivided in three groups (central memory T cells (T CM ), eff ector memory T cells (T EM ) and T EMRA ) on the basis of their surface expression of diff erent lymph node homing markers [8]. A summary of candidate infectious agents associated with myositis was published recently [9].
Th e presence of CD4 + T cells could also be associated with infectious agents, but in the context of myositis, it is more likely that these cells develop as a consequence of an autoimmune reaction [1]. Owing to how immune responses are orchestrated by CD4 cell-derived cytokines, CD4 + T cells are traditionally regarded as helper cells. Indeed, the most common way of subdividing CD4 T cells is based on secretion of specifi c cytokines, together with activity of so-called master transcription factors. In this fashion, CD4 T cells can be subdivided Abstract T cells of both the CD4 and CD8 lineage are commonly found in aff ected tissues of patients with idiopathic infl ammatory myopathies, but understanding the contribution of these cells to immunopathogenesis remains challenging. Given recent advances in identifying more myositis-associated autoantibodies and their putative targets, we suggest that studies on autoreactive T cells targeting those autoantigens are one way forward. Another (so far, more frequently used) approach comes from studies on eff ector T cells in the context of myositis. This review summarizes recent advances and current hypotheses in both of these contexts.
into diff erent T helper subsets such as Th 1, Th 2, Th 9, Th 17, and Th 22 and regulatory T (Treg) cells [10]. However, in recent years, it has become clear that CD4 + T cells can also diff erentiate into cytotoxic eff ector cells reminiscent of CD8 cells and natural killer (NK) cells [11]. Such cells have been named CD4 + CD28null T cells and fall outside of the classic T helper subsets. Th ey represent terminally diff er en tiated cells, which in addition to being potent inter feron-gamma (IFNγ) and tumor necrosis factor (TNF) producers have acquired many NK-related receptors and cytotoxic capacity by expressing both perforin and gran zymes [12].
It is well established that T cells can be found at all of the diff erent sites of disease manifestations in patients with myositis. But the importance of the presence of these cells is still a matter of debate, as is their antigen specifi city. In recent years, an increasing number of myositis-associated and myositis-specifi c autoantibodies have been identifi ed. Th e targets of those autoantibodies might represent diff erent candidate autoantigens [1]. Clearly, such data implicate antigens that could also be studied with regard to T-cell function. Below, we will discuss T cells in the three major aff ected tissues in patients with myositis rather than in subgroups according to the classic disease sub-entities PM, DM, and IBM.

Muscle-infi ltrating T cells
Th e main manifestations, shared by all three subsets of myositis, are proximal muscle weakness and muscle fatigue. Some patients have persistent cellular infi ltrates, which are associated with sustained muscle weakness.

How do the fi rst T cells migrate to muscle?
Th ere have been ample eff orts in dissecting how and which T cells (and other infl ammatory cells) migrate into muscle. Here, chemokines govern the migration of leukocytes to sites of infl ammation (Figure 1), and several studies have addressed this issue and demonstrated expres sion of the α-chemokines CXCL9 and CXCL10 and the β-chemokines CCL2, CCL3, CCL4, CCL19, and CCL21 in IIM muscle. Th e chemokines can be produced by infi ltrating infl ammatory cells but potentially also by muscle fi bers themselves. Th e reason why muscle fi bers would express chemokines could be infection, trauma, and genetic predisposition.
Although many studies have investigated mRNA from muscle biopsies, making it diffi cult to elucidate the cellular source of the chemokines, there are data that IIM muscle fi bers themselves can produce chemokines. In this context, CCL2 expression has been demonstrated by immunohisto chemical staining of muscle biopsies [13,14] and in myoblast cell cultures in which co-stimulation with IL1β leads to elevated CCL2 mRNA levels [15]. CCL2 is a chemoattractant for CCR2-and CCR4-express ing cells, including monocytes, memory T cells, and dendritic cells. Additional chemokines are interesting in this context, such as CCL3 and CCL4, which are chemo attractants for macrophages and T cells, and CCL3 is a potent regulator of Th 1-committed T cells. Indeed, a signifi cant upregulation of CCL3 had been demonstrated in IBM myofi bers. Both CCL3 and CCL4 are present in muscleinfi ltrating mononuclear cells in DM, PM, and IBM [16].
Similarly to the traffi c via high endothelial venules into lymph nodes, CCR7-CCL19 interaction has been suggest ed to contribute to amplifying/sustaining T-cell traffi c to sites of infl ammation. Also, muscle fi bers positive for this chemokine and infi ltrating lymphocytes positive for the receptor have been demonstrated [17]. Further dissection for extranodal lymphoid microstructures has been performed, and indeed such structures can be found [18]. Th e existance of such lymph node structures indicates that lymphocyte activation and diff erentiation could take place within the muscle, and there is support from studies of B cells, plasma cells, and immunoglobulin sequences that plasma cell diff erentiation can take place in this location [19].
A common observation in biopsies of aff ected muscle from patients with IIM is the focal distribution of the infl ammatory infi ltrates. Th e reasons behind this have not yet been delineated, but it is tempting to speculate that migration of the fi rst cells into muscle is a rare event. Once a few infl ammatory cells have entered the muscle, a feedback loop is started because of chemokine production by the infl ammatory cells.

Which T cells are found in aff ected muscles?
A long-standing dogma has been that CD8 cells are more common in PM and IBM but that CD4 cells are more pronounced in DM. T-cell receptor (TCR) profi ling by performing CDR3 spectratyping of the TCR Vβ chains for CD8 + T cells in PM and IBM demonstrated that CD8 T cells (potentially autoreactive) are clonally expanded and can persist for years [20][21][22]. A strong bias of Vβ expression in the IBM muscle as compared with the corresponding blood further suggests that the T cells are clonally expanded in situ, or are specifi cally recruited to the muscle, and may be driven by muscle-specifi c autoantigens [20]. However, within the Vβ subfamilies, there was a high variability in clonal restriction between patients for both PM [23,24] and IBM [25,26], possibly suggesting the presence of several local autoantigens and epitope spreading. For CD4 + T cells, in contrast, CDR3 spectratyping has revealed much more polyclonal patterns [21]. Th is, together with the observation that many muscle-infi ltrating cells express perforin or granzyme B or both, could be seen as evidence for an exclusively cytotoxic CD8 T cell-mediated immuno patho logy. How ever, CD4 T cells have been less studied, but it has been demonstrated that CD4 + T cells can be a major part of the muscle infi ltrate [12], so more studies are needed to fully evaluate this. Moreover, today we know that CD4 + CD28null T cells predominate the infi ltrate in aff ected muscle and those cells express cytotoxic eff ector substances [12,27], allowing us to reevaluate older fi ndings in a new context. FOXP3 + Treg cells are critical cells in maintaining immune homeostasis and preventing autoimmune and chronic infl ammatory disease. Th erefore, it is appealing to suspect that patients with IIM may have a numerical defi cit in Treg cells. Th is was, however, not the case [28], and similar data exist for other rheumatic diseases. An alternative hypothesis is that the Treg cells are functionally defi cient or that the infl ammatory milieu does not allow Treg cell suppression as indicated in rheumatoid arthritis [29]. During investigations of biopsies before and after glucocorticoid therapy, both the overall T-cell count and the Treg cell count were reduced after therapy whereas persistent CD4 + CD28null T cells could be observed [30].

How could T cells perpetuate local disease?
Cytokine staining of muscle biopsies has demonstrated only modest levels of T cell-derived cytokines. However, since many cytokines function directly on neighboring cells, this could be due in part to rapid consumption. Another interesting eff ector function is the perforin/ granzyme B axis and its eff ect on muscle fi bers (Figure 2). We recently started to address whether CD28null T cells could directly interact with muscle fi bers by an autologous co-culture system, and we have preliminary data in support of this [31]. A granzyme attack normally would be expected to induce apoptosis in the aff ected cell, but one needs to remember that muscle fi bers are multinucleated cells and, as such, may not undergo classic apoptosis [32]. Instead, one could think that the common observation of regenerating fi bers could be a result of such an insult. Such regenerating fi bers also express high levels of Jo-1, one of the candidate auto antigens [33]. In the context of CD28null T cells, an additional concern is their longevity, and this subset is the reason why persistent infi ltrates are sometimes seen even after aggressive therapy [30].

Skin-infi ltrating T cells
DM is the IIM subtype that aff ects both skin and muscle. Historically, this disease has been looked upon as more CD4-driven than PM and IBM. Granzyme B-expressing T cells have been found in other rheumatic diseases with skin involvement (such as systemic lupus erythematosus) but were found to be low in DM [34]. Also, the degree of FOXP3 + cells has been reported to be low [35,36]. In contrast, type I IFN appears strong in aff ected skin of patients with DM [37] and could lead to accumulation of CXCR3 + lymphocytes. Indeed, enrichment of CXCR3 + cells has been reported in DM skin [38]. Another interesting T-cell population, CXCR5 + T helper cells, has been studied in peripheral blood of patients with juvenile DM and found to display Th 2 and Th 17 activities and to be associated with disease activity [39].

Lung involvement in myositis
Interstitial lung disease (ILD) is a heterogeneous group of non-infectious lung disorders characterized by infl am matory cell infi ltration and interstitial fi brosis. It remains one of the greatest contributors to morbidity and mortality in myositis [40] and is associated with a poor quality of life for patients with myositis. Myositis-asso ciated ILD is closely linked to the appearance of auto antibodies raised against anti-histidyl tRNA synthetase (Jo1) and the socalled anti-synthetase syndrome [41]. In anti-Jo-1 + patients, there are a few reports on T lympho cytes with specifi city toward this particular autoantigen (reviewed in [42]). Additionally, T cells from bronchoalveolar lavage fl uid from patients with myositis-asso ciated ILD showed a strong bias toward Vbeta3 T-cell receptor expression as compared with healthy controls, suggesting a specifi c role for T cells in the development of ILD [43,44]. Since ILD often precedes myositis symptoms, it is suggested that the autoimmune reaction might start in the lungs (Figure 3). Indeed, it has been demonstrated that the Jo1 antigen is highly expressed in the lung compared with other organs [33]. Interestingly, a proteolytically sensitive conformation of Jo-1 is found in the lungs and leads to cleavage by granzyme B [45]. Granzyme B generates unique fragments of this autoantigen. It is suggested that these fragments are taken up by immature dendritic cells, which get activated, mature into professional antigen-presenting cells, and stimulate CD4 + T cells, initiating downstream immune cascades. Both CD8 + T cells and CD4 + CD28null T cells may play a role in cytotoxicity in both the muscle and the lungs. Th ey could contribute not only to destruction of target tissues in myositis but also to initiation of autoimmunity through cleavage of Jo1 mediated by granzyme B.
Th e trigger for the initiation of the autoimmune response has not yet been clarifi ed. On the one hand, it is hypothesized that the infl ammatory cascade might be initiated by an infection with an as-yet-unidentifi ed virus (for example, Coxsackie, infl uenza, HIV, hepatitis C virus, and cytomegalovirus) [42]. Th ose viruses may enter the respiratory tract, where they may lead not only to cell death and an associated release of Jo-1 antigen into the extracellular space but also to a modifi cation of this enzyme. Another possibility is genetic predisposition to reduced apoptotic clearance or increased apoptosis in conjunction with environmental stimuli such as smoking [46]. Th e release of Jo1 and its special conformations into the extracellular milieu are believed to trigger not only activation of T cells by antigen-presenting cells but also migration of cells expressing CCR5, including den dritic cells and T cells [47], especially Th 1 cells that were shown to express predominantly CCR5 [48]. Th is may explain an infi ltration of mostly Th 1 cells in the lung of patients with myositis-associated ILD [49]. In addition, it was shown that there is a disease-specifi c association between Jo-1, ILD, and serum levels of CXCL9 and CXCL10 [50], two IFN-γ inducible chemokines attracting CXCR3-expressing cells, including NK and Th 1 cells, further enhancing the infi ltration of pro-infl ammatory Th 1 cells and emphasizing a role for these cells in disease pathogenesis.

T cell-independent autoantibodies?
IgG autoantibodies are generally believed to be a consequence of T-cell help but could potentially arise in a T cell-independent fashion via the cytokines BAFF Epidemiologic studies suggest cigarette smoking as a key environmental risk factor. Smoking, or other irritants, could initiate Jo-1 cleavage or modifi cation. Granzyme B-cleaved Jo-1 will attract mononuclear cells, which can process and present the Jo-1 autoantigen to the immune system. Similarly, Jo-1 is overexpressed in aff ected muscle and hence a similar pathway as described for the lung could also perpetuate adaptive immune reactions in the muscle. DC, dendritic cell; NK, natural killer; Th1, T helper 1.
(B-cell activating factor) and IL-21 [51]. Moreover, high BAFF levels have been found in the circulation of anti-Jo-1 + patients and an even higher concentration of BAFF is associated with myositis-associated ILD [52]. Th is leads to a higher appearance of B cells and plasma cells that possibly could locally produce autoantibodies. Th e autoantibodies are believed to build immune complexes with Jo-1 fragments and nucleic acid released from dying cells, which may activate plasmacytoid dendritic cells (pDCs). Th ese cells are the major source of type I IFN, and IFN-α can be detected in muscle, skin, and peripheral blood of anti-Jo-1 + myositis patients where pDCs are highly enriched [53][54][55]. IFN-α released by pDCs in turn upregulates expression of BAFF, creating a positive feedback loop and ultimately leading to a break of tolerance. In addition, IFN-α upregulates expression of intercellular adhesion molecule (ICAM) on lung epi thelial cells, enhancing recruitment of even more infl am matory cells and further amplifying the infl ammatory cascade, and furthermore may contribute toward CD28 downregulation and thereby the appearance of the CD28null phenotype [56].
Since the association between autoimmune IIM and ILD was initially described, 35 years ago, a great deal of knowledge has been added, especially regarding diagnosis and therapy of myositis-associated ILD. Th e initiating trigger and detailed pathogenesis of this disease remain to be elucidated. Having a closer look at how myositis-associated ILD is initiated and what major key players are involved at what time point not only will improve our understanding of disease mechanisms but also may reveal therapeutic possibilities.

How to continue dissecting T cells in myositis?
Our understanding of T-cell function and regulation is continually growing. Transferring this knowledge to clinical settings can allow the identifi cation of new biomarkers. But to truly understand the contribution of T cells to myositis, we will need focused studies in which patient material is fi rst stratifi ed for autoantibodies and HLA type and in which autoreactive T cells are character ized in detail.
A third, indirect, way of increasing our understanding of T cells in this disorder involves studies before and after diff erent treatment regimes. Assessing T-cell eff ector functionality at baseline and after a given time of therapy can also reveal T-cell involvement. Ultimately, we may be able to predict which patients with myositis have a disease infl uenced by T cells and which do not and thereby pave the way for individual ized treatment strategies.