Autoreactive T Cells from Patients with Myasthenia Gravis Are Characterized by Elevated IL-17, IFN-γ, and GM-CSF and Diminished IL-10 Production

Myasthenia gravis (MG) is a chronic autoimmune disorder of neuromuscular transmission (1). Patients present with characteristic weakness and fatigability, particularly of the skeletal muscles (2, 3). Immunopathology in the most common subtype of the disease is directly related to the presence of acetylcholine receptor (AChR) autoantibodies (4). The AChR is a pentameric transmembrane glycoprotein ion channel, composed of five (α₂βδε) subunits (5). Autoantibodies specific for each subunit can be found in MG patients (6), although the majority recognize the α subunit (7). These AChR-targeting autoantibodies, primarily of the IgG₁ and to a lesser extent the IgG₃ subclass (8), affect the disease by inactivating the AChR at the neuromuscular junction (1) primarily through internalization and localized complement-mediated tissue damage (9). Both active and passive transfer of AChR Abs from humans to animal models affect the disease, demonstrating the direct role these molecules play in its pathology (4, 10–12). Although their production has been well delineated at a descriptive level, the details and features of the underlying cellular immunobiology of MG require further understanding. Specifically, the contribution of T cells to the mechanisms of autoantibody production remain to be more clearly defined.

Autoantibody-producing B cells in MG include evidence of class switching and somatic hypermutation, indicating that they are products of affinity maturation (13, 14), which suggests that Ag-specific CD4⁺ T cells provide B cell help during this process. Although they have been investigated less thoroughly than B cells and autoantibodies in MG, the studies of MG-related T cells have collectively defined several important characteristics. Circulating T cells that recognize the human AChR (15) are present in patients with MG. These autoreactive T cells exhibit an inflammatory response to AChR subunits by proliferating and inducing the production of the Th1 cytokine IFN-γ (16–19). T cell recognition of the α subunit is most common; however, autoreactive MG T cells reflect the pattern of B cell specificity toward the AChR as epitopes derived from each subunit can affect T cell proliferation (16–19) and induce production of IFN-γ (20). The AChR epitopes recognized by MG T cells can vary among patients; however, a majority of MG patients recognize a common set of epitopes. These “universal” epitopes are most often found on the AChR α subunit while recognition of regions within the other subunits are reported, albeit less frequently (21). Contemporary studies of T cells in MG have identified a defective regulatory T cell (Treg) population (22, 23), but studies specifically investigating other potential pathogenic contributors such as Th17 cells have not been reported.

Autoreactive CD4⁺ T cells are associated with the pathogenesis of autoimmune disorders. Both Th1 and Th17 cells play critical roles in experimental autoimmune models and have become linked to multiple autoimmune diseases through their induction of proinflammatory mediators and recruitment of immune cells to sites of inflammation (24–26). Th17 cells can function as B cell helpers through induction of robust proliferative responses, triggering Ab production along with class switch recombination (27). Th17 cells have been implicated in the pathology of autoimmune diseases mediated by B cells and the pathogenic autoantibodies they produce, such as neuromyelitis optica (28). GM-CSF–producing T cells display a distinct transcriptional profile and represent a new Th subset that contributes to autoimmune pathology (29–31). A requirement for inducing an inflammatory autoimmune demyelinating disease in mammals is the activation of Th1/Th17 autoreactive T cells that secrete pathogenic IL-17, GM-CSF, and IFN-γ (30–34), illustrating the critical role of Th17 cells in the development of autoimmunity. Conversely, T cells producing both IL-17 and IL-10 are protective and function in suppressing inflammatory responses (35, 36).

The recent success in the treatment of MG with biologics such as anti-CD20 (37, 38) and the shift toward the use of similar highly specific immune-targeting treatments for autoimmune disease has highlighted gaps in our knowledge concerning the cellular immunobiology contributing to the autoimmune dysregulation. A deeper understanding of these mechanisms will provide a refined picture of MG immunopathology and inform us how novel biologics may target these processes, so that the effectiveness and durability of such treatments can be anticipated. To this end, we sought to investigate the contribution of T cells to MG pathology with particular attention given to the Th17 compartment. We further focused on AChR-reactive T cells, which are exceptionally infrequent among the circulating population. Molecular profiles of AChR-reactive T cells were established to identify key functional differences of the CD4⁺ T cell compartment in MG patients and healthy subjects. The technical challenges of interrogating the function of rare human autoreactive T cells were addressed by using recent advances in the generation of T cell libraries (39).

This study was approved by the Human Research Protection Program at Yale School of Medicine. Peripheral blood was obtained from healthy individuals and patients with MG after acquiring informed consent. Patients were recruited from the Yale Myasthenia Gravis Clinic. We obtained specimens from 11 patients with MG (age range 33–80 y [at the time of entry into the study], female/male ratio 5:6) (Table I). Ten of the 11 patients had detectable anti-AChR Abs by RIA testing at the time of entry into the study (Athena Diagnostics, Worcester, MA). All patients had clinical and electrodiagnostic features consistent with MG. Collected clinical data included demographics, duration of disease, immunosuppressive medications, AChR autoantibody titer, thymectomy status, and thymus pathology, Myasthenia Gravis Foundation of America Clinical Classification (Table I). Healthy control (HC) subjects had no history of autoimmune disease or malignancies and no acute or chronic infections. These controls were matched to the patients as closely as possible with regard to age (age range 24–55 y), sex (female-to-male ratio 5:6), and HLA genotype (8/11 were DR3 or DR5).

Genomic DNA was isolated from whole blood derived from each subject using the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany). Because the AChR peptides used in the study are associated with HLA-DR3 and HLA-DR5 (40), HLA-DR3– and HLA-DR5–specific genotyping (Table I) was carried out as previously described (41, 42). Amplification of β−actin served as a positive control for the PCR.

The following mAbs were purchased from BD Biosciences (San Jose, CA): anti-CD45RA (Hl100), anti-CD45RO (UCHL1), anti-CD25 (M-A251), anti-CD196 (CCR6, G034E3), anti-CD3 (UCHT1), anti-CD28 (28.2), anti–IFN-γ (B27), and anti–IL-17 (BL168). The complete media, used in all experiments, included RPMI 1640 medium (Life Technologies, Carlsbad, CA) supplemented with 2 mM l-glutamine, 5 mM HEPES, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 50 U/ml penicillin, 50 U/ml streptomycin (all from Lonza, Walkersville, MD), and 5% human serum (Immune Tolerance Network, San Francisco, CA). Recombinant human IL-2 was obtained through the AIDS Research and Reference Reagent Program (Division of AIDS, National Institute of Allergy and Infectious Diseases, National Institutes of Health).

PBMCs were isolated by standard Ficoll–Paque PLUS (GE Healthcare, Piscataway, NJ) gradient centrifugation and were used after liquid nitrogen-based cryopreservation in 90% human AB serum (GemCell, West Sacramento, CA) plus 10% DMSO (Sigma-Aldrich, St. Louis, MO). Monocytes, which served as APCs for T cell library assays, were preselected using a human CD14⁺ positive isolation kit, followed by CD4⁺ T cell negative selection using a human CD4⁺ T cell enrichment kit (STEMCELL Technologies, Vancouver, BC). Live cells were identified with a viability dye (Molecular Probes, Eugene, OR) and then gated before identification of naive (CD45RA⁺CD45RO⁻CD25⁻), CCR6⁻ memory (CD45RA⁻CD45RO⁺CD25⁻CCR6⁻), and CCR6⁺ memory (CD45RA⁻CD45RO⁺CD25⁻CCR6⁺) CD4⁺ T cells, which were sorted on a FACSAria (BD Biosciences) to a purity >98% confirmed by postsort analysis. The frequencies of the different T cell subsets, relative to the total CD4⁺ T cell population, were determined directly from the respective sorting gates.

Amplified T cell library assays were carried out as previously described (39). Naive, CCR6⁻, and CCR6⁺ memory CD4⁺ T were cultured in 96-well, round-bottom plates (Costar, Cambridge, MA) at 2 × 10³ cells/well in complete RPMI 1640 medium, and stimulated by 1 μg/ml PHA (Roche, Nutley, NJ) and 40 U/ml IL-2 in the presence of irradiated (45 Gy) allogeneic feeder cells (2.5 × 10⁴/well). Then IL-2 was added on days 4, 7, and 10. The culture was split into two 96-well plates. The stimulation mixture was washed out after 2 wk of stimulation and expansion. Library screening was performed by culturing ∼10⁶ T cells/well with autologous monocytes (∼10⁵), which were either unpulsed or pulsed for 3 h with 10 μg/ml AChR peptides (AChR-P1, AChRα_195–212, DTPYLDITYHFVMQRLPL; AChR-P2, AChRα_257–269, LLVIVELIPSTSS). Staphylococcus aureus (Calbiochem, San Diego, CA) served as a positive control because this Ag mixture has been previously reported to induce proliferation and cytokine production by CD4⁺ T cells (43). On day 5, [³H]thymidine (Perkin Elmer, Waltham, MA) was added and the mixture was further incubated for 16 h, after which proliferation was measured by [³H]thymidine incorporation on a scintillation beta-counter (Perkin Elmer). Culture supernatants were taken on day 7 and measured for cytokine production by ELISA.

Culture supernatants from amplified T cell libraries assay were tested for cytokine content. ELISA measurement of cytokines was performed with purified coating and biotinylated detection Abs: anti–IFN-γ (clone 2G1) and biotin-labeled anti-human IFN-γ mAb (Thermo Scientific); anti–IL-10 (clone JES3-19F1) and biotin-labeled anti-human/viral IL-10 (BD Biosciences). Production of IL-17 and GM-CSF were measured with a DuoSet ELISA kit (R&D Systems, Minneapolis, MN). The absorbance was measured, and concentrations were determined with a microplate reader and reader-associated software (Bio-Rad, Hercules, CA).

Amplified T cell subsets were labeled with CFSE (Life Technologies, Carlsbad, CA) and then restimulated with monocytes and AChR peptides. After 5 d, live CFSE^low CD4⁺ T cells were single-cell sorted into 96-well plates using a FACSAria cell sorter (BD Biosciences). Clones thus obtained were expanded for 28 d by stimulation with irradiated allogeneic peripheral blood mononuclear cells in the presence of soluble anti-CD3 (1 μg/ml), anti-CD28 (1 μg/ml), and IL-2 (10 U/ml). The single-cell clones were stimulated by AChR peptides to confirm their specificity. Thymidine incorporation and intracellular staining were then used to measure proliferation and cytokine production, respectively.

Single-cell clones were restimulated with irradiated autologous monocytes that were pulsed with AChRα_195–212 and AChRα_257–269 for 7 d. GolgiStop (BD Biosciences, Franklin Lakes, NJ) was added to the culture 4 h before harvesting. Cells were then stained with live/dead fixable marker (Molecular Probes). After fixation and permeabilization with intracellular staining buffer (eBioscience), cells were stained with anti–IFN-γ (clone B27) and anti–IL-17 (clone BL168) (Biolegend, San Diego, CA). Results were acquired on a Fortessa flow cytometer (BD Biosciences) and analyzed on FlowJo software (Tree Star, Ashland, OR).

Samples were quantified for the generation of dot plots and statistical analysis using the T cell library assay as previously described (44). For the development of the heat map, the following approach was applied: samples were quantified using the T cell library assay as described earlier. Each set of wells was ascribed to a given patient sample cell type, and stimulation was first summarized by the mean of the assay readout for cpm, IFN-γ, IL-17, GM-CSF, and IL-10. For each of the 10 replicates, where one replicate consists of an HC sample and an MG patient, Z-scores were calculated for each assay readout across all of the cell types and stimulations. For the consolidated heat map, these data were then further collapsed by calculating the mean of these Z-scores across the 10 replicates.

To study autoreactive CD4⁺ T cells and their potential role in MG pathogenesis, we adapted a novel T cell library assay (Fig. 1) and carried it out as previously described (39). Live naive, CCR6⁻ memory, and CCR6⁺ memory CD4⁺ T cells were sorted (Fig. 2A) into subsets, seeded at 2000 cells/well. The difference in the frequencies of these T cell subpopulations between the MG and HC subjects (Table I) was unremarkable (Fig. 2B) and was consistent with the same observation in patients with multiple sclerosis (MS) and HC subjects (44). To allow for polyclonal expansion, we excluded CD25^high Tregs from these distinct populations. Sorted subsets were amplified by PHA and IL-2, and subsequently stimulated by autologous monocytes and AChR peptides or control Ags.

Each of the 3120 unique expanded T cell libraries was interrogated for Ag reactivity. Cell viability from patients and control subjects was comparable before and after (day 5) Ag stimulation (Supplemental Fig. 1). The proliferative response of Ag-reactive T cells to AChR peptides was measured by [³H]thymidine incorporation; a representative MG subject (MG-02) and a paired HC subject are shown in Fig. 3. Libraries generated from naive populations (Fig. 3A) of HC subjects stimulated by AChR peptides were not different from those measured in the absence of Ag. Increased proliferation in a fraction of the stimulated naive T cell libraries derived from MG subjects was observed but did not reach statistical significance (Supplemental Table I). Libraries generated from the CCR6⁻ memory populations generally exhibited minimal functional responses to AChR Ags (Fig. 3B). A considerable fraction of the libraries generated from CCR6⁺ memory populations originating from MG patients displayed a distinctly increased (Supplemental Table I) and significantly different (p = 0.0029) proliferative response to AChR-derived peptides relative to those from healthy subjects (Fig. 3C).

We next focused on a phenotypic analysis of the AChR-reactive T cells in 10 MG patients (MG-1 through MG-10), each paired with an HC subject (Table I). To that end, an ELISA was carried out to measure specific cytokine production from each library. Given that Th1 and Th17 cells play a critical role in experimental models of autoimmunity (30–33), we measured their respective hallmark proinflammatory cytokines, IFN-γ and IL-17, in each population. Negligible levels of these cytokines were detected in the naive and CCR6⁻ memory CD4⁺ T cell compartments (Fig. 4A, 4B), and values from the HC and MG-derived libraries were not significantly different (Supplemental Table I). Interestingly, this lack of cytokine expression was observed in naive libraries that displayed a minor proliferative response to the Ags (Fig. 3A). Libraries derived from CCR6⁺ memory CD4⁺ T cells of MG patients produced levels of IFN-γ (p = 0.0061) and IL-17 that were significantly elevated (p < 0.0001) compared with that from HC subjects when stimulated by AChR-derived peptides (Fig. 4C, Supplemental Table I). The positive control Ag induced IFN-γ and IL-17 production in both the MG and the HC libraries (Fig. 4C).

Distinct protective and pathogenic Th17 cell subsets were recently identified in both autoimmune models and humans (30–33, 43). T cells that produce both IL-17 and GM-CSF represent a pathogenic phenotype and may contribute to autoimmunity. Conversely, T cells producing IL-17 and IL-10 are protective and maintain immune homeostasis (30, 31, 33, 34, 43). To further characterize the functionality of AChR-reactive T cells, we measured GM-CSF and IL-10 production in 7 of the 10 patients for whom additional PBMCs were available (MG-01 and MG-05 through MG-10). Minimal levels of GM-CSF production were detected in the libraries derived from either the naive or CCR6⁻ cells (Fig. 4A, 4B), but increased production was observed in the libraries stimulated by the S. aureus in both the MG and HC libraries derived from CCR6⁺ memory cells (Fig. 4C, Supplemental Table I). Marked GM-CSF production (p < 0.0001) was measured in the CCR6⁺ memory MG-derived libraries stimulated by AChR peptides, which was much lower in the negative control and libraries derived from the HC subjects (Fig. 4C, Supplemental Table I). Marginal production of IL-10 was detected in the naive or CCR6⁻ memory cell libraries derived from the MG subjects (Fig. 4A, 4B). Similar results were observed with the same libraries derived from the HC subjects with the exception of some IL-10 production by the unstimulated naive cells, the S. aureus–stimulated CCR6⁻ memory cells, and both the naive cells and the CCR6⁻ memory cells stimulated by the AChR peptides (Fig. 4A, 4B), all of which were statistically different from that produced by the same MG-derived libraries (Supplemental Table I). Marked production of IL-10 was measured in the CCR6⁺ memory HC-derived libraries stimulated by AChR peptides, in contrast with the negligible IL-10 production in the MG libraries stimulated (p < 0.0001) by the same peptides (Fig. 4C, Supplemental Table I). Interestingly, S. aureus stimulation induced less IL-10 production by the CCR6⁺ memory libraries derived from MG patients than that produced by the same libraries from the HC subjects (p = 0.0002).

To capture a unified portrait of the T cell proliferation and cytokine programs, we analyzed data libraries from the 10 MG and HC pairs and then presented them in a heat map (Fig. 5, Supplemental Fig. 2). The naive and CCR6⁻ memory compartments displayed similar proliferation and cytokine production levels in both the MG patients and the HC subjects. However, the CCR6⁺ memory CD4⁺ T cell compartment revealed conspicuous differences between the MG patients and HC subjects. Higher frequencies of AChR-reactive T cells were present in the MG subjects, and these cells secreted the proinflammatory cytokines IFN-γ, IL-17, and GM-CSF. Moreover, the HC subjects produced the anti-inflammatory cytokine IL-10 in response to both the AChR peptides and the S. aureus, whereas the same T cell subset from the MG subjects did not produce detectable levels of IL-10.

The distinct functional responses evident in the CCR6⁺ T cell libraries could arise from oligoclonal populations, meaning that the cytokine production profiles we measured may have been derived from contributions made by a number of cells, rather than coproduction by the same cells. We therefore assessed the specificity and functional programs of individual AChR-reactive T cells isolated from these libraries. Expanded libraries from an MG patient (MG-11) and a healthy subject were prelabeled with CFSE and restimulated by AChR peptides. Representative oligoclonal libraries were selected to confirm that this subject responded similarly to the AChR peptides as observed in the 10 MG and HC pairs examined earlier. Both comparable proliferation and cytokine production were observed in response to the AChR peptides (Supplemental Fig. 3).

Clones were then generated from the remaining oligoclonal libraries by sorting CFSE^low CCR6⁺ memory CD4⁺ T cells as single cells followed by expansion. Ag specificity of each single-cell–derived clone was tested. Among the 15 clones generated from the MG patient, 2 showed specific Ag-stimulated proliferation. None of the 14 clones generated from the HC subject were AChR specific (Fig. 6A). Evaluation of the cytokine profile from the AChR-specific single-cell clones revealed coproduction of the proinflammatory cytokines, IFN-γ and IL-17 (Fig. 6B), whereas minimal IL-10 secretion was observed (data not shown).

Given that MG represents an archetype of autoantibody-mediated autoimmune disease, the majority of investigations have been directed toward understanding B cells and the autoantibodies they produce. Not unexpectedly, MG T cells are less well studied than their counterparts in the humoral compartment. However, T cells appear to be required for MG pathology as transfer of MG CD4⁺ cells to animal models results in the development of myasthenic weakness (45). Furthermore, genetic studies have identified MHC types among MG risk alleles. More recent genome-wide association studies have identified CTLA4 and TNFRSF11A associations (46): the former transmits an inhibitory signal to T cells, and the latter is an important regulator of the interaction between T cells and Ag-presenting dendritic cells. A large collection of investigations (15–21) has focused on the identification of epitopes on the AChR subunits and has implicated a Th1-driven response because proliferation is accompanied by IFN-γ production. Although these earlier studies of MG demonstrated the presence of AChR autoreactive T cells, more recent studies have shifted focus toward other T cell subsets such as Tregs. Their role in experimental autoimmune MG, the AChR immunization-based model for the disease, indicates that they can potently suppress disease severity, preserve muscle AChR content, attenuate muscular weakness, and reduce circulating levels of AChR autoantibodies (47–49). Defects in the ability of CD4⁺, CD25^high FOXP3⁺ Tregs to suppress responder cells have been demonstrated for a number of human autoimmune diseases including MS (50), type 1 diabetes (51), and psoriasis (52). Tregs derived directly from the thymus of patients with MG are present at normal frequencies, but their ability to suppress is defective (53), and T responder cells also derived from MG thymus resist Treg mediated-suppression (23). In the periphery, Treg-mediated suppression of T responder cells is impaired in MG patients but can be restored using Tregs isolated from HC subjects (22).

Further studies seeking to understand the role of T cells in MG have focused on additional T cell subsets. Broad cytokine measurements in serum and cultures of PBMCs have been used to implicate both the activation and the pathogenic role of particular T cell subsets. For example, measurements of cytokine levels, such as IL-17, in serum or cell culture media of stimulated T cells provide some indication of Th17 cell involvement in MG (54), but it is difficult to assign production of cytokines to a particular cell type with such approaches. Identification of skewed T cell frequencies, such as Tregs or Th17 cells, has implied that immune dysregulation is governed by these particular T cell subsets (55). However, these broad approaches provide little data to support direct implication of particular T cell subsets, and there is no information that can be linked to Ag-specific T cells.

Large numbers of cells are required to deeply analyze self-reactive T cells. One strategy enabling enrichment and analysis of these rare T cells involves the use of tetramers composed of immunodominant epitopes loaded into fluorescent MHC class II molecules. Accordingly, tetramers have emerged as an important tool for characterization of the specificity and phenotype of CD4⁺ T cell immune responses (56, 57). However, a number of limitations, such as patient preselection based on HLA genotype, prevent their use in some settings (58). To circumvent the limitations inherent to class II tetramers, we applied an innovative, unbiased T cell library assay to identify and characterize AChR-reactive T cells in MG patients. With this approach, we were able to define the cytokine expression pattern of these extraordinarily rare cells in a multiplexed manner. We found increased frequencies of AChR-reactive T cells in MG patients compared with HC subjects. These autoreactive T cells resided in the CCR6⁺ memory CD4⁺ T cell compartment and secreted high levels of proinflammatory cytokines when stimulated by peptides derived from the target of the pathogenic autoimmune response in MG. We noted that the AChR peptides we chose are associated with the HLA-DR3 or DR5 haplotype. Many of our MG patients and control subjects (also 8/11) carried these haplotypes, thereby demonstrating that the absence of reactivity in our control cohort was not due to their HLA type. In addition, three MG subjects were not DR3 or DR5, but still responded to the AChR-derived peptides, which was consistent with previous observations (40) indicating an association with the HLA-DR3 or DR5 haplotypes, but not restriction. Collectively, these data indicate that the autoantigen-reactive T cells in MG patients are of the proinflammatory, Th1/Th17 phenotype and are likely to play a critical role in MG pathogenesis. Data from a model of MG support such a role for these cells as Ag-specific T cells derived from the experimental autoimmune MG model produce IL-17. Moreover, the loss of B cell tolerance leading to the production of AChR autoantibody and consequent disease in the model is dependent on these IL-17–producing cells (59).

Proliferation and cytokine production were recorded in a subset of the amplified libraries stimulated by either the AChR peptides or S. aureus. However, a number of subsets did not include T cells specific for these Ags, although the complete library represents a multifold expansion of the existing T cell repertoire. This demonstrates that each well of the complete library is a unique representation of the circulating repertoire and that the approach demonstrates the rarity of these autoreactive cells in the circulation. Although a considerable number of AChR-derived peptide Ags capable of stimulating MG-derived T cells have been described, we chose to focus on the two that have consistently demonstrated activity in a large portion of patients with MG (40, 60). Given that epitopes from each of the AChR subunits can stimulate MG-derived T cells, we reason that the relative frequency of autoreactive T cells in MG patients, although small, is greater than what we estimated in this study with two epitopes from the AChR α-chain. This limitation of our study, and others that have focused on particular AChR peptide Ags, is not easily overcome. Ideally, whole-protein Ag could be used so that naturally processed peptides could be produced in vitro. Although we recognize that naturally processed Ag is preferred over synthetic epitopes, this cannot be easily achieved with this particular Ag because of its complexity. The AChR is a transmembrane protein composed of five separate subunits; thus, it has proved difficult to acquire in a highly purified form either through recombinant means or from postmortem tissue, which has prevented its use as an Ag in T cell assays.

The cytokine profile of the clones from the MG patients showed that responding cells from the expanded library were Ag specific and potentially pathogenic in that we infer they coexpress IFN-γ, IL17, and GM-CSF, but not IL-10. Interestingly, S. aureus stimulation did not induce equivalent IL-10 production by the CCR6⁺ memory libraries derived from MG patients to that which was clearly measurable in the same libraries from the HC subjects. This may indicate an intrinsic defect in the production of the immunosuppressing IL-10 cytokine in MG patients. A diminished ability to produce IL-10 is characteristic of a number of autoimmune diseases (36, 61), and evidence for such defects in MG have recently been reported (22).

The pathogenic role of self-reactive T cells has been well characterized in a number of mouse models of human autoimmune disease, such as experimental autoimmune encephalomyelitis, which models MS. Myelin-reactive T cells, also located in the CCR6⁺CD4⁺ T cell population in MS patients, exhibited a distinct transcriptome that is homologous with encephalitogenic CD4⁺ T cells isolated from mice with experimental autoimmune encephalomyelitis and further revealed a transcriptional profile that distinguishes myelin-reactive T cells in MS from those present in healthy subjects (44). The distinct cytokine program that characterized these autoreactive T cells (elevated IL-17, IFN-γ, and GM-CSF expression and diminished IL-10 production) was equivalent to what we report in this article. Given that a growing number of convergent mechanisms exist between autoimmune diseases that are clinically distinct, it is interesting to consider this shared immunopathology between MS and MG. Although MS and MG are respectively regarded as T cell– and B cell–mediated diseases, similar T cell abnormalities appear to contribute to both at a fundamental level.

Our cohort of MG patients included some heterogeneity reflected in their clinical features. Our study subjects primarily included those who were immunotherapy naive (6/11). Others experienced different therapeutic approaches: four received prednisone and three underwent a thymectomy. Those who had their thymus gland removed continued to require treatment, indicating that the surgical procedure was not completely effective in eliminating the autoreactive mechanisms that affect disease. Their autoreactive T cell program, measured in this study, was similar to that of the remaining MG subjects who did not undergo the procedure. Although it is widely accepted that the thymus plays a role in MG pathology (62), our data and that from many other studies imply that autoimmunity may extend beyond this tissue compartment, perhaps after the disease initiates.

We further found that proinflammatory, AChR-specific T cells were present in a single patient for whom AChR autoantibody was not detectable. We tested a second serum sample from this patient, collected at the time of publication, for the presence of either AChR or the muscle-specific tyrosine kinase autoantibodies using both the standard RIA and cell-based assays, which can improve detection of such autoantibodies over the RIA (63). This patient had not seroconverted for either AChR or the muscle-specific tyrosine kinase autoantibodies, but the clinical presentation of ocular MG was consistent with that commonly observed for the AChR disease subtype. That we found autoreactive T cells in the absence of autoantibodies is consistent with other studies, which indicated that seronegative MG patients harbor T cells that respond to AChR (64) and have thymic abnormalities that are often similar to Ab-positive patients (65). Our cohort also included both early- and late-onset MG. Again, our data regarding the autoreactive T cell populations were the same in both groups, which is consistent with the overlapping clinical features of early-onset MG and late-onset MG, and that there are few well-established major differences in the two subtypes (66). We recognize that MG is a complex and heterogeneous disease. Accordingly, we acknowledge that a limitation of our study was that the MG cohort was not strictly homogeneous in terms of clinical demographics and treatment. Indeed, a considerably larger study would be required to evaluate the contribution of Th17 cells to disease subsets and response to treatment modalities. We leave open the possibility that appreciable differences in both the frequency and cytokine program of AChR-specific T cells may associate with MG clinical subtypes or clinical response to therapy. Overall, our collective data suggest that, irrespective of AChR autoantibody status or disease subset, many MG subjects harbor autoreactive, proinflammatory Th17 cells, indicating that these cells may be a fundamental feature of the disease and consistent with a continuous clinical spectrum of a single condition.

Given that B cell Ag-driven affinity maturation relies on the help of T cells, it is expected that AChR-reactive T cells contribute to the activation and maturation of AChR Ab-producing B cells. This component of the autoimmune response in MG is of particular importance when considering the durability of emerging MG treatment strategies. Of particular interest, because of their recent increase in use, are the treatments that target B cells, such as rituximab (37, 38). Indeed, encouraging results with B cell depletion in MG have materialized and a phase 2 trial is now under way (http://www.clinicaltrials.gov, NCT02110706). It still remains to be seen how patients who respond to this treatment will progress once the B cell compartment has reestablished. It should be considered that the reemerging B cell repertoire may be influenced by these autoreactive T cells, leading to induction of new pathogenic Abs by the reconstituted B cell compartment with disease recurrence and clinical relapse. There are numerous broadly acting immunotherapies for MG, such as steroids, azathioprine, and cyclophosphamide, but no interventions that specifically target T cells have been widely applied. Combination therapy that selectively targets more than one component of the autoimmune mechanism may be required to induce a state of tolerance and provide sustained clinical remission in patients with MG. Continued efforts to more thoroughly understand the cellular immunobiology underlying MG pathology will support rationales for such targeted therapeutic approaches. Such studies that are intimately connected with longitudinal clinical trials of immune-modulating biologics are of particular importance toward reaching this goal. Development of patient-tailored treatments and precision medicine depends on this much-needed insight; accordingly, integration of immune-based mechanistic studies with all future MG clinical trials should be carefully considered.

We thank our colleagues at Yale School of Medicine, Drs. William Housley and Aditya Kumar for critically reading the manuscript and Drs. Margarita Dominguez-Villar and David A. Hafler for valuable consultation and discussions.

The authors have no financial conflicts of interest.