Innate and adaptive immune responses accompany the degenerative process in the fatal neurodegenerative disease amyotrophic lateral sclerosis (ALS) (1). The complex dynamics that parallel disease onset and progression include not only cells of the central nervous system (CNS), but also peripheral immune cells: infiltration of peripheral immune cells into the brain and spinal cord has been well-documented in ALS patients and different animal models of the disease (2–5). However, whether they contribute to human disease, and how, is not completely understood. T cells have gained particular attention in ALS, as they represent a significant proportion of the peripheral cells that infiltrate the CNS of ALS patients, their phenotype is altered in ALS, and they bear high therapeutic potential. CD4⁺ T cells have been shown to be protective in ALS, but their abundance and regulatory capacities are diminished in ALS patients (6); the number and functionality of regulatory T cells (T_reg) correlates negatively with ALS progression and suggests a novel avenue for therapeutic interventions (7, 8). In addition, cytotoxic CD8⁺ T cells were recently also implicated in ALS pathogenesis by the demonstration that they can selectively kill motor neurons (9). In that study, the interaction of CD8⁺ T cells with motoneurons relied on the MHC-class I molecules’ association with self-antigens and their interaction with T-cell receptors. However, the underlying autoantigens remain unknown, which might be critical for further understanding of the involvement of T cells and their different subsets in ALS. In line with this, more recent studies found a clonal expansion of cytotoxic CD8⁺ T cells to an unknown antigen in familial ALS with mutation in the SETX gene (10) and an increase in activated (HLA-DR⁺) CD8⁺ T cells in the blood and CSF of patients with sporadic ALS (11). Therefore, it has been suggested that a common ALS-associated antigen like TAR DNA-binding protein 43 (TDP-43) or SOD1 might trigger T-cell mediated autoimmune processes (12). Interestingly, in a similar manner it was recently shown that T cells in Parkinson’s Disease (PD) selectively recognize autoantigens generated from alpha-synuclein, a protein which plays a major role in the etiology of the disease (13). Several proteins linked to ALS share the pathological features of alpha-synuclein in PD, including generation of toxic oligomers, high molecular-weight aggregation, aberrant dynamics of proteostasis and subcellular mislocalization. Among these proteins, TDP-43 stands out by being the histopathological hallmark of the disease, as well as by its linkage to genetic forms of ALS (14–18). In motoneurons, TDP-43 pathology results in mislocalization, aggregation and aberrant processing of the protein (19). We therefore hypothesized that TDP-43 might be a so-far unidentified autoantigen contributing to motoneuron demise in ALS. To test this hypothesis, we investigated the reactivity of T cells from healthy controls and ALS patients to ten TDP-43-derived peptides.

To obtain a comprehensive picture of the circulating T cells in ALS patients, we performed flow cytometry with peripheral blood mononuclear cells (PBMCs) from 24 ALS patients and 25 age-matched healthy controls ( Supplementary Table 1 ). To determine the net frequencies of T cells in PBMCs, we selected CD3⁺ cells and quantified cell numbers as percent from the total singlet PBMCs ( Figure 1A ). In line with previous observations (30–32), we found a net decrease of the relative numbers of CD3⁺ cells, which was driven by a net decrease of CD4⁺ cells, while net CD8⁺, double-negative (DN) and double-positive (DP) T cells were unchanged in ALS ( Figure 1B ). We did not observe statistically significant differences in the relative numbers of CD4⁺, CD8⁺, DN and DP cells when quantified as percent of all CD3⁺ cells ( Supplementary Figure 1 ).

Both CD4⁺ and CD8⁺ T cells can be activated upon recognition of their cognate antigen. To investigate whether there is an increased abundance of antigen-activated T cells in ALS, we analyzed T-cell activation by flow cytometry for HLA-DR expression, a well-established marker of T-cell activation, before and after expansion with IL-2. We chose HLA-DR, as it specifically represents the late stage of T-cell activation, i.e. it allows to quantify how many of the circulating T cells have already come to an initial contact with their cognate antigen. In addition, both CD4⁺ and CD8⁺ T cells express HLA-DR after activation. While T-cell activation was comparable on baseline between healthy controls and ALS patients, we found a significant increase of HLA-DR-expressing T cells in ALS, affecting both CD4⁺ and CD8⁺ cells, upon culturing and expansion of T cell for 14 days in vitro with IL-2 (20 ng/mL) ( Figures 2A, B , n=9/9; Supplementary Table 2 ). Furthermore, T-cell activation was observed in all three main T cell memory types: central memory, effector memory and TEMRA, while effector memory T cells showed the strongest in-vivo expansion, as expected ( Figure 2C , Supplementary Figures 2 , 3 ). This change suggests an increased presence of antigen-experienced T cells that express IL2R after antigen-mediated activation in ALS.

We then hypothesized that the TDP-43 protein, which is prone to misfolding and aggregation and at the same time represents the major neuropathological marker in almost all ALS patients, acts as a T-cell activating antigen in ALS. To test this hypothesis, we designed a set of 10 peptides derived from the wild-type human TDP-43 (UniProt:Q13148) ( Figure 3B , Supplementary Table 3 ). Most peptide sequences were chosen to be flanked by at least one predicted or known protease cleavage site, thus imitating the natural processing of TDP-43. Five of the peptides had a length of 9-11 amino acids, which is required to fit in the peptide cleft of the MHC class I complex, while the other five peptides have a length of 15 amino acids, fulfilling the minimal length for MHC class II-mediated antigen presentation. MHC-I and MHC-II peptides were distributed evenly across the sequence of TDP-43 and were designed for high binding affinity (IC50<50nM) to a wide range of HLA alleles. All main domains of TDP-43 were covered by the peptides, including the ALS mutational hotspot in the C-terminal glycine-rich domain ( Figure 3B ). First, we excluded that TDP-43 peptides result in unspecific activation of PBMCs, as they did not activate isolated peripheral blood monocytes at three different concentrations ( Supplementary Figure 4 ). We then used the peptides to stimulate PBMCs from ALS patients and healthy controls in culture (n=33/32, mean disease duration 49 ± 38 months, mean ALSFRS-R 32 ± 11, Supplementary Table 4 ) and quantified the T-cell response by ELISPOT for IL-5 and IFN-γ ( Figure 3A , Supplementary Table 5 ). Stimulation of PBMCs with the potent mitogen PHA (positive control) induced a strong, antigen-unspecific activation of PBMCs, with a slight, non-significant decrease in activation of PBMCs from ALS patients ( Figure 3 , Supplementary Table 5 ). As expected, PBMCs from ALS patients and healthy controls responded weakly to peptides from the autoantigens actin or myelin oligodendrocyte glycoprotein (MOG, data available in 24/27 cases), with no significant difference between healthy controls and ALS patients. The response to TDP-43 peptides was comparable or lower to the response to MOG and actin in magnitude; however, we found significantly lower responses by IL-5-producing T cells in ALS. The percentage of responders (persons with a response higher than the spontaneous SFC-activity in the DMSO vehicle control) was comparable to that for actin and less in ALS for both IL-5 (>25%) and IFN-γ producing cells (>30%) (Fisher’s exact: 0.43 (IL-5), 0.46 (IFN- γ)). The median response in the responder groups was significantly lower in ALS in IL-5 producing, but not in IFN-γ producing cells ( Supplementary Figure 5 ). Taken together, this data suggests a possibly lower CD4⁺-mediated T-cell response to TDP-43 in ALS. We then analyzed in detail which TDP-43 peptides elicit the response of PBMCs ( Figure 3B , Supplementary Tables 6 , 7 ). We observed that all TDP-43 peptides contributed to the activation of PBMCs, with a slight tendency for a stronger reaction to the peptides closer to the C-terminus. In addition, the weak antigenic response was observed both to peptides designed for MHC-I and for MHC-II-restricted activation. In general, we observed more often responses by IL-5 producing T cells than by IFN-γ producing cells, but the responses by IFN-γ producing cells were stronger in amplitude. The IL-5 response correlated positively with the IFN-γ response per individual and peptide, which can be attributed to individual differences in T-cell activity or to differences in the binding affinity of the peptides to the different HLA alleles expressed by the different persons. The response to most peptides was decreased in ALS, both in IL-5 producing and in IFN-γ producing cells. Moreover, we did not find any difference between the abundance of naïve and of antigen-experienced T cells reactive to TDP-43 ( Supplementary Figure 6 ). As a complementary measure of autoimmunity, we also quantified the abundance of naturally-occurring autoantibodies against full-length wild-type TDP-43 monomers and oligomers ( Figure 4 ) in blood plasma. We found low, but measurable levels of TPD-43-specific autoantibodies, but no difference between healthy controls and ALS patients. Levels of TDP-43 autoantibodies were significantly higher than the negative control in this assay (****p<0.0001, luciferase alone) and higher than the levels of naturally occurring antibodies to alpha-synuclein ( Supplementary Figure 7 ); however, they were much lower than the levels of antibodies to alpha-synuclein in one person with known high blood concentrations of alpha-synuclein antibodies ( Supplementary Figure 7 ), as measured with a similar assay.

In the present study, we found elevated levels of activated T cells in ALS blood after in vitro expansion with IL-2. By assaying HLA-DR expression as activation marker, we captured the frequency of circulating T cells, which have already come in contact with their cognate antigen. HLA-DR is a late T-cell activation marker, which is up-regulated after other common activation markers (CD25, CD69), which could not be assessed in this study. Similar to some previous reports (30–32), we found a significant decrease of total CD3⁺ and net CD4⁺ cells in ALS. This decrease in CD3⁺, driven by a decrease in CD4⁺ cells, might reflect an increased infiltration of T cells into the ALS CNS, as reported previously (33). Apart from this, we found comparable composition of T cells in healthy controls and ALS patients. The CD4/CD8 ratio was normal in both groups and numbers of CD8⁺ cells, DN, DP, naïve and different types of memory T cells, as well as baseline HLA-DR⁺ cells, were comparable between both groups. Furthermore, we demonstrate that TDP-43 peptides induced a weak T-cell response, comparable to the control self-antigens actin and MOG. However, in contrast to PHA, actin and MOG, we found a statistically significant decrease in the activation of T cells from ALS patients by TDP-43. This suggests that although TDP-43 is a weak self-antigen, it still might be associated as such with ALS. One limitation of this study is the use of IL-2 for T-cell expansion, which has been shown to influence specifically regulatory T-cell function (34, 35). Thus, our data might be heavily influenced by the preferential expansion of regulatory T-cells in culture.

TDP-43 is a nucleo-cytoplasmic protein, which suggests the presentation of TDP-43 epitopes by the MHC-class I complexes. In addition, we have previously shown that TDP-43 oligomers can be released from cells into the extracellular space (27) and thus presentation by MHC-class II complexes should also be possible. In our experimental setup, cross-presentation of extracellular antigens by immature dendritic cells and possibly monocytes allows for the activation of both MHC-class I restricted and MHC-class II-restricted responses to the exogenously applied peptides. Indeed, we found both MHC-class I and MHC-class II-restricted weak antigen-specific responses, which implies that both cytotoxic and cytoprotective immune responses could be evoked by TDP-43. The presence of an MHC-class I-restricted T-cell response suggests cytotoxic effects on cells presenting TDP-43 antigens. Motor neurons express MHC-class I complexes in physiological as well as pathological conditions, like ALS (36). MHC-I also has a role in the stabilization of synapses and their recovery from injuries (37). The expression of MHC-class I is upregulated upon exposure to IFN-γ, also produced by activated cytotoxic T cells and activated microglia (38). In line with this, Coque et al. demonstrated recently that cytotoxic T cells can selectively induce the death of motoneurons in the SOD1^G93A mouse model (9). The presence of MHC-class II mediated responses, on the other hand, allows for cytoprotective immune responses. Interestingly, we found less activation of T cells by TDP-43 in ALS patients, suggesting a possible failure of cytoprotective T-cell mechanisms. Indeed, regulatory T cells are decreased in ALS and negatively correlate with disease progression (6). Thus, reduced T-cell responses to ALS-related antigens might be detrimental. In addition, the presence of antigen-reactive T cells is also an important requirement for the development of strong, successful humoral responses to antigens. Although T-cell-independent activation of antigen-specific B cells is possible, most antigens will require the simultaneous presence of an antigen-specific T-cell and B-cell for the development of strong humoral response. It is therefore conceivable that such antibody-mediated immune responses could halt the cell-to-cell transmission of TDP-43 pathology that we have previously demonstrated (27). In fact, a similar therapeutic strategy has already entered the clinical phase for alpha-synuclein in PD (39). Interestingly, we did not find a significant difference in the levels of TDP-43-specific auto-antibodies between ALS patients and controls. This is in disagreement with some previous studies, which found either elevation (40, 41) or a decrease (42) in naturally-occurring TDP-43 auto-antibodies in ALS. The discrepancy most likely stems from differences in methodology, as we employed a novel assay which allows the binding of the auto-antibodies to TDP-43 in liquid phase and includes monomeric, oligomeric and likely post-translationally modified TDP-43. Unfortunately, it was not possible in this study to correlate the levels of auto-antibodies to TDP-43 to the T-cell activation in the same patient, so this should be considered in future studies to elucidate better the relationship between T-cell activation by TDP-43 and the humoral response to it.

Importantly, the antigen-specific response to TDP-43 was very low across all conditions: the disease/control state, MHC-class I/II activation type, detection method, assayed cell types. Thus, our data does not directly provide evidence for the abovementioned processes in ALS. Further work is necessary to either identify other autoantigens in ALS or to demonstrate stronger TDP-43 responses in disease-relevant circumstances, like a strong activation of the innate immunity. The activation of innate immune cells is a prerequisite for a strong activation of the adaptive immune response. The TDP-43 peptides used in this study did not induce a significant activation of monocytes, supporting the specificity of the observed T-cell response (e.g. no contamination with endotoxin), but also likely reducing the magnitude of T-cell activation that would be seen if innate immune cells were also activated by TDP-43. Presentation by non-activated dendritic cells is an important mechanism for the induction of tolerance to an antigen, while presentation by activated dendritic cells will induce a strong T-cell response. Due to the ubiquity and importance of TDP-43 for the cell, tolerance to intracellular TDP-43 antigens is probably critical and reflected by the weak T-cell response and widespread reactivity we found also in healthy controls (~ 40% of donors), similarly to MOG and actin. Thus, we consider antigen presentation by PBMCs, as usual in the mixed-culture ELISPOT assay, sufficient for this model. However, future studies should consider the application of mature DCs as APCs to investigate the antigenic response to TDP-43 in full scale. Moreover, we speculate that the T-cell response to TDP-43 in vivo will be much stronger than what we observed in culture, if the source of TDP-43 antigens are pathologic forms of TDP-43 (oligomers, aggregates) that could also activate innate immune cells. Indeed, it is well-established that aggregated proteins may act as alarmins and induce a strong activation in APCs (43). In line with this, we have recently shown that not only aggregates, but also oligomers and mutant forms of alpha-synuclein can strongly enhance the activation of innate immune cells (44). The mounting of an adaptive immune response to abnormal TDP-43 in the CNS will require the presentation of TDP-43 antigens derived from CNS neurons by professional antigen-presenting cells (APC). Peripheral APCs will need to first collect the antigen from the CNS, a concept that was traditionally considered not possible unless there is some leakage from the blood-brain-barrier. However, recent studies have demonstrated that presentation of antigen from the CNS is possible in the glymphatic system and that peripheral lymphocytes may indeed circulate through the adult CNS. Alternatively, microglia and astrocytes may act as APCs in the CNS (45) and prime T cell for an immune response. In ALS, activated microglia are found throughout the affected CNS areas and express HLA-DR (2, 46, 47). Furthermore, studies have consistently reported infiltration of T cells into the CNS of ALS patients and an association with areas of degeneration (48, 49).

In summary, we show that human T cells bear the potential to recognize peptides derived from TDP-43. Taken together from our data and the current literature, the response is comparably weak, seen in a significant proportion of probands, may have role in tolerance, and could contribute to the neurodegenerative processes in ALS in vivo.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethics Committee of Ulm University. The patients/participants provided their written informed consent to participate in this study.

SR and BL performed the experiments, VG and MM analyzed the data, KK and SW collected samples, AL and JW provided intellectual input, VG, SR and KD wrote the manuscript. All authors reviewed and approved the manuscript.