Multiple sclerosis (MS) is the most prevalent chronic inflammatory demyelinating disease of the central nervous system (CNS) (1). The etiology of MS remains unclear (2). Nevertheless, it is generally accepted that a complex interaction between genetic predisposition and environmental factors results in the disruption of immunological self-tolerance to myelin antigens of the CNS (3). Established environmental risk factors for MS include insufficient sun exposure (4), infection with Epstein-Barr virus (EBV) (5) and dietary intake (6). In particular, several studies have comprehensively elucidated how nutrition can act as a possible co-factor influencing the inflammatory cascade in MS patients either directly, by effecting various cellular elements (7) or indirectly, by interacting with the gut microbiota (8).

For example, the beneficial effect of a diet comprised of fish (9), short chain fatty acids (10), fruits and vegetables (11) on reducing the severity of disease progression in either MS or experimental autoimmune encephalomyelitis (EAE), a commonly used animal model of MS, has been suggested by different reports. Furthermore, a significant epidemiological association between consumption of bovine milk and milk products and the prevalence of MS has been established by other researchers (12–14).

A mechanistic link between the potential exacerbation of autoimmune responses in MS patients and the consumption of milk has been suggested by our group and others (15–17). Stefferl et al. previously reported sequence homologies between myelin oligodendrocyte glycoprotein (MOG) and butyrophilin (BTN), a milk fat globule membrane protein (15). We extended this concept of molecular mimicry to dairy proteins as possible etiological factors in the immunopathogenesis of MS by providing evidence of antibody cross-reactivity between another milk protein, casein, and myelin-associated glycoprotein (MAG) (17).

Despite these findings, among others, that highlight the negative effect of dairy intake on the aggravation of MS symptoms (6, 7, 18), milk and milk products have been an integral part of human diet for centuries. Benefits of milk consumption range from adequate intake of calcium and bone mineralization (19, 20) to nourishment with essential amino acids, vitamins, minerals and fatty acids (21, 22).

With most of these studies (23, 24) focusing on cattle-based milk, here we used enzyme-linked immunosorbent assays (ELISAs) as a method to screen for an immunoglobulin(Ig)G response to different types of animal-sourced and plant-based milk in MS patients and healthy controls. Furthermore, we also correlated the IgG titers between specific bovine milk antigens and their potential cross-reactive myelin antigens in MS patients.

Milk proteins, specifically bovine milk antigens, have been shown to share peptide homology with different neuronal and non-neuronal antigens (15–17, 29–32). Therefore, loss of oral tolerance to these ingested milk antigens (33, 34) can lead to aggravation of ongoing autoimmunity as a result of molecular mimicry, as in the case of MS. Indeed, the immune repertoires are greatly heterogenous between individual patients (35) and the possible consequences of this tolerance breakdown only apply to those who already harbor milk-antigen reactive T cells and B cells. Accordingly, the purpose of this study was to analyze the overall prevalence of antibodies against different sources of milk and milk antigens in a cohort of MS patients and correlate these findings to the presence of antibodies against CNS tissue lysate and specific recombinant antigens in the same patients.

To this end, as the first approach, we screened for IgG reactivity to different sources of milk in a cohort of 35 MS patients and 20 healthy donors. Our data indicate that IgG reactivity in MS patients in comparison to healthy donors was highest against cow milk followed by goat milk, whereas the antibody titers to sheep milk were comparable between the two groups. Loss of oral tolerance to specific milk antigens initiated by disturbance in antigen uptake and presentation in genetically susceptible individuals (34) can be an explanation why MS patients harbor higher IgG titers to animal milk than healthy donors. A potential factor could be the disruption of the intestinal epithelial barrier resulting in increased permeability of potentially pathogenic antigens into the body (36). Leaky gut syndrome has been reported to play a critical role in several autoimmune diseases like MS resulting in dysbiosis and alterations in the composition of the gut microbiome (37).

Interestingly, however, our healthy control cohort displayed a higher IgG response to A2 milk (P = 0.008) compared to the MS patients. As mentioned earlier, A2 refers to one of the two variants of bovine milk β-casein (26) with reports suggesting that consumption of milk containing only A2 β-casein is associated with fewer gastrointestinal symptoms than conventional milk in those with milk allergies (38, 39). Furthermore, elevated levels of β-casein antibodies in patients with type 1 diabetes (40) have been attributed to the A1 variant of β-casein as the antigenic trigger (41). We observed a similar trend in our cohort where, on the one hand, MS patients harbored significantly higher IgG titers to β-casein ( Figure 2 ) in comparison to the healthy controls. Of note, the β-casein antigen used in this study does not specifically contain either only the A1 or A2 variant of β-casein ( Table 2 ). On the other hand, the overall antibody reactivity to A2 (cow) milk was significantly lower in the MS cohort. Taken together ( Figures 1 , 2 ), therefore, one may speculate that in MS patients the overall reactivity to bovine milk that was observed is, at least partially, due to an elevated IgG response to the A1 β-casein. This would also explain the low OD values to the A2 milk which does not contain the allergen and antigen A1 β-casein. Future studies focusing on adsorption assays using A1 vs. A2 β-casein would validate this observation.

Another finding of this study was that both MS patients and healthy donors showed similar patterns and levels of IgG reactivity to the different types plant-based milk ( Figure 1B ), with reactivity to almond milk being the lowest and to cashew and coconut milk, on average, the highest. Indeed, the main sources of plant proteins that are consumed as an alternative to animal proteins, like nuts, wheat and soy, are considered to be important allergenic triggers (42). On the one hand, healthy subjects have been reported to possess antibodies to coconut, almond and soy-based milk substitutes (43). On the other hand, an increased antibody (IgG) titer to certain plant antigens, like those in soybean and corn, has also been reported in a subset of patients with autoimmune neuroinflammation (44). Furthermore, this reactivity has been attributed to the expression of aquaporins (AQPs) in plants that are known to facilitate the transport of water and sustain plant life (44). Plant AQPs have been shown to share homology with human aquaporins (45, 46), in particular AQP4, which is a known autoantigen in the neuroimmune disorder neuromyelitis optica (NMO) (47).

Vaishnav et al. reported that serum from NMO patients cross-reacted with a sequence found in plant aquaporins which shared homology with the human AQP4 epitope (48, 49). They also demonstrated that this cross-reactivity was higher in NMO patients compared to non-NMO controls (48). Whether the plant-based milk alternatives that were used in this study (i.e., coconut, oat, cashew, almond and hazelnut) also express the human AQP4-cross-reactive pathogenic epitope mentioned by Vaishnav et al. remains to be clarified (48). Furthermore, one can speculate that because MS is not an autoimmune astrocytopathy (50), our patient cohort did not have any significantly higher levels of antibodies to the plant-based milk types in comparison to the healthy donors. Finally, one important point to consider is the low concentration of potentially pathogenic antigen(s) present in the plant-based alternatives. The percentage of plant protein and therefore the relevant antigen(s) present in the different sources of commercially available plant-based milk are < 10% which might not be sufficient to provoke a cross-reactive antibody response in patients with an underlying autoimmunity.

When segregating our data according to gender, we observed no significant differences in IgG titers to the different sources of milk (animal- or plant-based) between male and female MS patients. A similar result was observed also for the healthy donors, except for the reactivity to sheep milk where female healthy controls had a significantly higher titer (P = 0.009) in comparison to their male counterpart. However, this finding can also be explained by the relatively low N numbers in both the male and female cohorts.

In this study, we investigated the IgG titers against the three most abundantly present bovine milk antigens (51) and observed significantly elevated IgG titers to β-casein (P = 0.0049) in the MS cohort. When the MS patients with IgG to β-casein were further categorized into those with an OD value > 2.0 vs. < 2.0, a significance between the two groups in terms of their corresponding EDSS score was observed ( Supplementary Figure S2 ), indicating that patients with high titers to β-casein also suffered from a greater disability. Nevertheless, in this study, the dairy intake of the recruited MS patients or the corresponding healthy controls has not been documented. Careful monitoring of milk intake and its effect on the IgG titers to the individual milk antigens needs to be performed in a longitudinal manner to confirm this preliminary data.

We also correlated the overall IgG response against cow milk and the individual cow milk antigens in the MS cohort. The correlation matrix showed the highest correlation between the presence of antibodies to α-lactalbumin and cow milk (r = 0.47) followed by antibodies to β-lactoglobulin and cow milk (r = 0.46). Why only a moderate linear relationship was observed between a response to casein antigens and cow milk in this study can perhaps be explained by the nature of the different sources of antigens used. For example, the individual bovine caseins (i.e., α-casein, β-casein, κ-casein) that were used in this study were purified antigens. While the source of cow milk was store-bought and therefore the coating antigen did not only comprise purified bovine milk proteins but rather a heterogenous mixture of other unrelated proteins. Therefore, our correlation data needs to be interpreted with caution, also acknowledging that “correlation does not imply causation”.

Our data also highlighted that the presence of antibodies to human brain lysate in MS patients strongly correlated with the IgG titer to bovine β-lactoglobulin (r = 0.77). On the one hand, there are reports suggesting that β-lactoglobulin shares sequence and structure homology with a number of different antigens expressed in human brain tissues, like glutamate decarboxylase (GAD)-65 (52, 53) and β2-microglobulin (54, 55). Furthermore, both GAD-65 and β2-microglobulin are also involved in different brain diseases like Stiff-Person’s syndrome (56) and glioblastoma (57). On the other hand, β-lactoglobulin is restricted to bovine milk and this protein is not expressed in humans (58), making it a highly allergenic and pathogenic antigen (59), with particular relevance to different human diseases. To what extent the high correlation we detected in our MS cohort can be attributed to antibody cross-reactivity between bovine β-lactoglobulin and GAD-65 and β2-microglobulin remains unknown. Future studies should focus on validating the nature of this antibody cross-reactivity as well as exploring the possibility that there may be other bovine β-lactoglobulin cross-reacting antigens expressed in the brain tissue.

We were able to confirm the simultaneous presence of IgG to bovine casein and MAG in MS patients, where MAG is the cross-reactive antigen for bovine casein (17). Indeed the correlation was moderate but this could be, once again, due to the nature of the antigens used. While the casein antigens were purified by chromatography from bovine casein with a purity between ≅ 70% (for α- and κ-casein) and ≅ 98% (for β-casein), recombinant MAG was used ( Table 2 ). Of note, the significant correlation between the occurance of IgG to bovine β-lactoglobulin, β- and κ-casein vs. human brain tissue lysate in the MS cohort was not observed in the healthy control group ( Supplementary Figure S3 ). Similarly, there was also no correlation between the presence of antibodies to MAG and the individual casein antigens (especially, β- and κ-casein) in the healthy cohort ( Supplementary Figure S4 ).

Finally, the IgG titers to neither BTN1A1 nor MOG were significantly higher in the MS patient cohort in comparison to the healthy controls, which is in line with the literature (16, 60). Indeed, the fact that we did not detect significantly higher titers of anti-MOG antibodies in the MS cohort could be a technical issue since whether these antibodies recognize linear or conformational epitopes of MOG remains under debate. Nevertheless, we observed a very high correlation between BTN1A1 and its cross-reactive antigen MOG ( Figure 4 ), in which case both were recombinantly expressed proteins, with r = 0.98 and r = 0.64 in the MS and healthy cohorts, respectively. This was not surprising given that studies have previously extensively demonstrated that MHC class II-restricted CD4⁺ T cell responses to BTN and MOG are mutually cross-reactive and that BTN can be processed and presented in vivo to simulate a cross-reactive T cell response to MOG in a rat model of MS (15). Furthermore, it has also been reported that BTN and MOG cross-react in a subset of MS patients (16). Therefore, the high correlation between IgG to MOG and BTN1A1 in our patient cohort was a successful extension of the data already available since the early 2000s.

While the encephalitogenic properties of BTN have already been discussed (16), cross-reactivity between MAG and casein resulting in CNS demyelination in a mouse model has previously been demonstrated by our group (17). Future studies should focus on further elucidating pathomechanistic processes of potential molecular mimicry, e.g., by passively transfering antibodies against different milk antigens into mice in the context of a compromised blood-brain barrier to determine whether the transfer will exacerbate CNS inflammation and demyelination. This could be done in combination with adsorption experiments to validate the specificity of the antibodies. In addition, further efforts could be made to identify additional self-antigenic targets of milk-specific antibodies.

Taken together, we hypothesize that consumption of animal-based milk antigens that share sequence or structural homology with human tissue-specific proteins can result in mimicry-induced misfires of the immune system in susceptible individuals (61–64). However, one should not assume that consumption of any plant-based milk alternatives by these patients is automatically a better choice. Therefore, on the one hand our result calls for a thorough testing for the determination of antibody titers against an array of different mammalian and plant-based milks/milk substitutes on an individual basis. On the other hand, removal of such antigens, for example, a dairy-free diet in individuals with a genetic predisposition can be used in the development of personalized medicine.

One of the caveats of this observational study is that the effect of different immunomodulatory therapies and the amount of daily dairy and dairy product consumption on the IgG response to milk and CNS antigens in the MS cohort have not been taken into account. Furthermore, the antibody response has been measured only for a single time point. Therefore, future studies should validate the findings presented in this manuscript using a larger and stratified cohort of MS patients whose IgG titers against the different dietary antigens can be measured in a longitudinal manner, also including paired cerebrospinal fluid (CSF) and serum samples. This will provide a deeper insight into the possible sequestration of milk antigen-specific antibodies into the CSF/CNS compartment. Additionally, the impact of immunomodulatory or B cell depleting therapies on the IgG response to milk antigens needs to be studied in a systemic manner. In the current study, the MS cohort was neither under any special diet nor did the patients suffer from any known milk allergies. Questionnaires provided to the MS patients to document the amount of their daily consumption of animal-sourced milk or vegan milk alternatives will provide crucial information on how diet may have an impact on the immune response to different milk antigens. This will be a necessary step for creating a customized diet plan tailored according to every patient’s need.

Finally, we have previously demonstrated that antibodies to bovine milk casein cross-react with MAG (17). While the current study confirms the correlation between anti-casein and anti-MAG IgG in the MS cohort, we do not have any further knowledge regarding the precise biochemical structure of the cross-reacting epitopes. Whether antibodies developed against cow milk casein(s) target the glycans of MAG, or bind to the protein itself remains to be clarified. For example, in the case of the rare peripheral polyneuropathy, anti-MAG IgM gammopathy, antibodies are specifically directed against the carbohydrate HNK-1 independent of the protein backbone (65). If this would be the case for the anti-casein antibodies where they recognize the glycans rather than the protein, enzymatic deglycosylation of caseins (or other milk-related cross-reactive antigens) before consumption may be an important point to consider.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by the Ethics Committee of the FAU Erlangen-Nürnberg, Germany (file 185_18B). The patients/participants provided their written informed consent to participate in this study.

RC and SK designed research and wrote the manuscript. RC performed the experiments. TH recruited the patients and was responsible for their clinical monitoring. All authors contributed to the article and approved the submitted version.