Myasthenia Gravis (MG) is an autoimmune disease characterized by muscle weakness and fatigability, which, in most cases, is the result of autoantibody production against the acetylcholine receptor (AChR) at the neuromuscular junction (1, 2). Although the disease is strongly associated with autoantibody production, the AChR-specific CD4⁺ T cells have a central regulatory role and, indeed, also patients lacking anti-AChR antibodies host peripheral blood CD4⁺ T cells that react to AChR-peptides (3, 4). Importantly, MG patients have reduced levels of regulatory T cells (Tregs) (5–8) and restoring the levels of Tregs in the experimental autoimmune myasthenia gravis (EAMG) model suppressed disease (9). Hence, reinstating a functional Treg population could be a curative therapeutic intervention in MG (10). Both Th1 and Th17 cells have been implicated in driving the autoimmune attack, but the precise contribution of the respective subset remains unclear (11–13). Evidence in support of an involvement of Th17 cells in the EAMG model has been documented in several recent studies, while earlier studies, for example, in Tbet^−/− mice demonstrated reduced susceptibility to EAMG as a consequence of fewer autoreactive Th1 cells (14).

Current treatments for autoimmune diseases are not curative, but rather are directed against the immunopathology and symptoms (15, 16). Also in MG, the first line of treatment is symptomatic, counteracting the muscle fatigability using acetylcholinesterase inhibitors, while some patients need additional medication with steroids or other forms of immunosuppressants, such as azathioprine, mycophenolate, or methotrexate. In the more severe cases, monoclonal antibodies such as anti-CD20 (rituximab) or, for short-term benefit, intravenous immunoglobulin or even plasmapheresis/immunoadsorption are being used (17). The medication has to be taken life-long with an increased risk of acquiring infectious diseases or even cancer, secondarily to severe side effects (15). Therefore, there is growing interest in developing novel therapies able to suppress autoaggressive T cells and reinstate tolerance in the immune system (18).

One such approach is to use adoptive cell transfer of ex vivo expanded autologous CD4⁺ Treg cells to inhibit disease development as has been successfully demonstrated in the EAMG model (9). Other investigators have focused on identifying immunodominant epitopes in the AChR to raise Tregs by immunization and in this way protect against disease development (19). Thus, restoring a functional Treg population by immunization with immunodominant epitopes from the AChR appears to be an attractive therapeutic approach for curbing MG disease. A major challenge, though, has been to translate the very promising findings in the rodent EAMG models into effective immunization protocols for treatment of MG patients (20, 21). This could partly be explained by the lack of effective formulations for tolerance-induction in humans as both disease-relevant peptides and proteins have already been identified but clinical testing still awaits to be done (18, 22–25). Hence, we need better ways of formulating our candidate epitopes to secure a strong induction of tolerance also in the clinic.

We have developed CTA1R9K-X-DD, which is a targeted immunomodulating fusion protein that can carry different disease-relevant peptides and which is an effective tolerogenic vector for the suppression of autoaggressive CD4 T cells (26). The fusion protein is an inactivated mutant of the CTA1 subunit of cholera toxin, which in its native form exerts strong ADP-ribosylating effects (26). The DD is a dimer of a fragment of Staphylococcus aureus proteinA, which targets classical dendritic cells (27). Using the collagen-induced arthritis (CIA) mouse model, we could demonstrate that a collagen peptide (aa 259–274) inserted into the fusion protein, CTA1R7K-COL_259–274-DD, and given as an intranasal (i.n.) therapy or orally in the form of edible plants effectively protected against CIA (26, 27). Following treatment with CTA1R7K-COL_259–274-DD, we found suppression of specific antibody levels in serum, reduced effector Th1 and Th17 CD4 T cell responses to peptide concomitant with an increased production of IL-10, while IL-6 and MMP3 levels were strongly reduced (26, 27). Although we observed increased numbers of circulating Foxp3⁺ Tregs, the increased IL-10 production emanated from regulatory Foxp3⁻ CD4 T cells, i.e., Tr1-like cells (28).

In the present study, we have extended our work to the EAMG model. We hypothesized that the mechanism of action in the EAMG model could be the induction of Tr1 cells and the subsequent reinstatement of tolerance to the AChR. Therefore, a disease-relevant peptide, the 146–162 amino acid peptide from AChR, was expressed as a fusion protein, CTA1R9K-AChR_146–162-DD (hereinafter referred as mCTA1–T146). This fusion protein, thus, carried a dominant epitope from the AChR for treatment of EAMG in C57Bl/6 mice (29, 30). We report here the successful use of this fusion protein in both acute and chronic stages of EAMG disease. Treated mice developed significantly less symptoms and exhibited less tissue destruction and lower serum anti-AChR antibody titers. Lymph node cells in treated mice demonstrated upregulated gene expression of TGFβ, IL10, IL27, and Foxp3, a suppressive gene signature, concomitant with suppressed Th1 and Th17 CD4⁺ T cell development (31, 32).

The present study identifies a novel approach to treating MG disease by reinstating tolerance against the AChR using the targeted fusion protein mCTA1–T146. We provide compelling evidence in the mouse EAMG model that i.n. treatment with mCTA1–T146 is effective at an early as well as at a later chronic stage of the disease with significantly less tissue destruction, i.e., higher muscle AChR content, and lower serum anti-AChR antibody titers than in untreated mice. This has important implications for the feasibility of developing a clinical treatment for MG patients. Moreover, the effect on lowering the clinical score was associated with an immune suppressive gene signature involving upregulation of genes encoding for IL10, IL27, TGFβ, and FoxP3 in the draining lymph nodes and spleen. These genes have been found strongly linked to a protective effect on autoimmune reactions (35). Most importantly the production of AChR-specific IFNγ and IL17, key cytokines representing Th1 and Th17 functions, respectively, were significantly suppressed in mCTA1–T146-treated mice. In this way, the present findings corroborate and extend our previous documentation of a Tr1-mediated suppressive effect of the fusion protein on autoaggressive CD4⁺ T cells in the CIA model (26).

The pioneering experiments showing that mucosal administration of native TAChR in rats induces tolerance dates back to the early 1990s when nasal or oral delivery was shown to suppress EAMG disease (29, 36, 37). Since then, several studies have been published that convincingly demonstrate that mucosal delivery of AChR or its recombinant extracellular α1 domain effectively suppresses induction of EAMG in both the rat and mouse EAMG models (38). Notably, it was observed that in ongoing EAMG it was significantly more difficult to suppress the disease and it usually, but not always, required much larger doses of antigen (19, 39). Also peptides have been used for tolerization and it has been well documented that most often large doses (50–100 µg or more) are required for suppressing induction of EAMG disease (29, 40, 41). It is in this context, we believe our targeted fusion protein, mCTA1–T146, adds to the therapeutic arsenal of AChR-specific tolerance-inducing strategies.

We found that repeated nasal treatments with a low dose of mCTA1–T146 effectively ameliorated EAMG disease resulting in less reduction in muscle AChR content and lower levels of serum anti-AChR antibodies. Because the peptide T146–162 is incorporated into the fusion protein, it has to be taken up and be processed by antigen-presenting cells (APC) to induce tolerance (27). This is contrary to the peptide approach where uptake by APC and processing are not involved and, indeed, the term apitope has been coined to identify an antigen-processing independent epitope (42–44). Therefore, the mechanisms of action can potentially be different between a peptide approach and that of the fusion protein. In particular, lower molar doses would be expected to be required for a tolerance-inducing effect by the fusion protein, which is also supported in our study when comparing the fusion protein with an equimolar dose of peptide that failed to induce tolerance. However, in the experimental autoimmune encephalomyelitis (EAE) model, apitope-induced tolerance was also mediated by IL-10 producing Foxp3-negative Tr1 cells, similar to what we think our mCTA1–T146 fusion protein does in the EAMG model (45). Nevertheless, we believe the fusion protein approach may have certain merits as it is possible to attach targeting elements to the peptides. In the mCTA1–T146 fusion protein, we have explored the DD-moiety for its DC targeting ability. Given i.n. the migratory DCs of the CD103⁺ phenotype are the primary targets for inducing tolerance (46, 47). Indeed, in preliminary experiments we could also show that our fusion protein accumulated in CD103⁺ migratory DC following i.n. administration (C. Hansson, Univ. of Gothenburg, Sweden, personal communication).

The critical disease-preventing element in the fusion protein is the AChR-specific peptide T146–162. Hence, the immune system in general is, therefore, unperturbed and functional for protection against, e.g., pathogens. Moreover while the fusion protein is carrying the T146–162 peptide, rather than the whole extracellular α1 of AChR, i.e., aa 1–210, the peptide approach largely avoids an immunogenic and potentially pathological reactivity to B cell epitopes in the protein (3). Hence, this also limits the risk of disease exacerbation following treatment should a enhancing rather than a suppressive anti-AChR-specific antibody response be elicited, including potential epitopes due to renaturation of conformation-dependent epitopes (3, 39, 48, 49). This is also the reason why investigators have avoided the main immunogenic region (i.e., aa 66–76) loop when developing α1 region based tolerogenic treatments and why cytoplasmatic domains rather than the extracellular α1 region of the AChR are being used for tolerance induction (3, 50). However, it appears that rather large doses are required for tolerance induction using this approach, at least in the rat EAMG model (48).

According to the guidelines for the mouse EAMG model, the quality of the TAChR preparation is important to get a high take of disease (21). In this study, we achieved 95% incidence of disease. Although, our preparation could have contained low levels of contaminating other synaptic proteins (see Materials and Methods), we specifically tolerized against the T146 peptide of the AChR, and, hence, we suppressed CD4 T cell reactivity against the AChR with positive clinical manifestations. However, additional incorporated peptides, representing multiple synaptic proteins, may have improved suppression even further. It has been debated whether a single or few peptides from a given protein will suffice in the clinic to reinstate tolerance (44). One argument against the use of a single peptide has been that since MG patients usually recognize multiple epitopes in AChR, it will be difficult to reinstate tolerance in multiple CD4 T cell clones with diverse epitope recognition (3). Given that many AChR-epitopes are dominant and probably do not host B cell epitopes, it would be feasible to combine several of these epitopes in a fusion protein (49, 51). This would also be a mechanism to broaden the scope of peptide-binding to surface molecules encoded by the HLA-D-locus in a genetically diverse population, although the 146–162 amino acid peptide has proven to bind to quite a large and diverse range of HLA-DQ subset of haplotypes prevalent in Caucasians as well as in Afro-Americans and Asians (52).

Some researchers have investigated the concept of a therapeutic intervention by reinstating Tregs in MG patients (9). Whereas the technology appears clinically cumbersome, costly and possibly associated with significant risk it may not be a feasible future therapeutic approach. The idea that polyclonal autologous inducible Tregs (iTregs) expanded ex vivo and adoptively transferred to the patient suffers from the inconsistency that also other immune responses will be suppressed in these patients, potentially placing these patients at risk of acquiring infectious diseases or developing cancer. An alternative strategy to expanding Tregs is to transfer autologous immature or tolerogenic DCs prior to or subsequent to inducing EAMG (53, 54). In this way, Tregs will expand in vivo and suppress disease. These ex vivo generated DC could be treated with TGFβ, IL10, or substances that promote a tolerogenic phenotype like rapamycin or statin (11, 55, 56). However, even though this strategy seemed promising in the experimental EAMG model its clinical application appears unrealistic. By contrast, the fusion protein would achieve the induction of Tregs in vivo and in this way circumvent ex vivo expansion and handling of Tregs.

Our mCTA1–T146 fusion protein acts through suppressing the autoaggressive Th1 and Th17 cells directed against epitopes in the AChR. As aforementioned the mechanism of action is through induction of regulatory CD4⁺ T cells, of which both Foxp3⁺ Tregs and Tr1 cells appeared to be induced in the present study. Although we found upregulated Foxp3-gene expression in the present study, we have previously observed clear evidence of an IL-10 dependent mechanism associated with Foxp3⁻ CD4 T cells in the CIA model (26). Recently, we have further strengthened a Tr1-mediated mechanism by showing that CD49b⁺LAG3⁺ CD4 T cells were induced in the draining lymph node after i.n. immunizations with our tolerogen for treatment of the EAE mouse model (C. Hansson, Univ. of Gothenburg, Sweden, unpublished observation) (57). Moreover, the same suppressive gene signature with upregulated gene expression for IL10, IL27, TGFβ, and Foxp3 was observed also in this latter model (45, 58–61). The upregulated Foxp3-gene expression could, however, indicate that also iTregs are induced and participate in the suppression of AChR autoreactive CD4⁺ T cells in treated mice. Perhaps, induction of both Tr1 and iTregs is the more likely scenario to explain the effective suppressive function of mCTA1–T146 in the EAMG model. Contrary to the acute CIA model the EAMG model has a chronic ongoing phase of disease (21, 33). It is promising to note that our tolerogen was equally effective in the late chronic phase of EAMG as in the early phase of disease development. We have previously demonstrated that oral administration of the Tα146–162 synthetic peptide in milligram doses to mice reduced the T-cell proliferative response to TAChR or Tα146–162 peptide and peptide doses given i.n. in 100 µg could achieve a similar effect (29). Hence, the mCTA1–T146 fusion protein is significantly more effective and requires much lower and fewer doses (a total dose of 50 µg of mCTA1–T146) to substantially reduce ongoing EAMG than peptide alone or derivatives of AChR. Ongoing studies aim to define the minimal protocol for significant improvement of EAMG disease in the mouse model using mCTA1–T146. The present study clearly demonstrates the effectiveness of the targeted mCTA1–T146 fusion protein and helps identify a new promising strategy in tolerance-inducing treatments of autoimmune diseases in general, and of MG-disease, in particular.

This study was carried out in accordance with the recommendations of the Institute Ethical Board and Italian Ministry of Health (1064/2015-PR) and were performed in respect to the Italian Principle of Laboratory Animal Care (DDL 116/92 and DLgs 26/2014), and in accordance to European Communities Council Directive 86/609/EEC and 2010/63/UE.

AC and ER performed the animal experiments. AC, SS, KS, and CL-F analyzed the data and interpreted the results. AC and CL-F prepared the figures. NL and FB designed the experiments. AC, NL, and FB wrote the manuscript.

NL has intellectual property related to MG treatment. NL is affiliated with the Toleranzia AB company that develops MG-specific treatments based on the fusion protein described in the present study. Disclosures are managed in compliance with the policies of University of Gothenburg, Sweden. AC, SS, KS, CL-F, ER, and FB have nothing to disclose.