SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Subjects and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Appendix
  8. References
  9. Supporting Information

Objectives

Pediatric-onset multiple sclerosis offers a unique window into early targets and mechanisms of immune dysregulation. It is unknown whether heightened T-cell reactivities documented in adult patients, to both target-organ and environmental antigens, emerge in parallel or develop as early or late events. Our objectives were to determine the presence, pattern, and specificity of abnormal T-cell reactivities to such antigens in the earliest stages of the multiple sclerosis process.

Methods

Peripheral T-cell proliferative responses to self-, dietary, and control antigens were blindly evaluated in a large cohort of well-characterized children (n = 172) with central nervous system (CNS) inflammatory demyelination (n = 63), recent-onset type 1 (insulin-dependent) diabetes mellitus (T1D; n = 41), nonautoimmune neurological conditions (n = 39), and healthy children (n = 29).

Results

Children with inflammatory demyelination, CNS injury, and T1D exhibited heightened T-cell reactivities to self-antigens, and these responses were not strictly limited to the disease target organs. Children with autoimmune disease and CNS injury also exhibited abnormal T-cell responses against multiple cow-milk proteins. Responses to specific milk epitopes distinguished T1D from inflammatory demyelination and other neurological diseases.

Interpretation

Abnormal T-cell reactivities to self- and environmental antigens manifest in the earliest clinical stages of inflammatory demyelination and T1D. The pattern of heightened T-cell reactivities implicates both shared and distinct mechanisms of immune dysregulation in the different autoimmune diseases. Abnormal T-cell responses in children with tissue injury challenge the prevailing view that CNS autoreactive cells inherently mediate the disease in early multiple sclerosis. Ann Neurol 2007

The cause of multiple sclerosis (MS) involves a complex interplay between environmental factors and genetic predisposition, associated with loss of self-tolerance and heightened reactivity to central nervous system (CNS) antigens. This process is thought to be mediated in part by some form of in vivo priming or abnormal sensitization of peripheral T cells reactive to CNS targets including myelin epitopes (reviewed in Hafler and colleagues1 and Bar-Or2).Whether these CNS autoreactive T cells are essential in mediating particular stages of disease (eg, initiation vs propagation), or whether they represent heightened immune reactivity generated in response to CNS tissue injury remains unknown. The specific epitopes and the environmental triggers that may lead to abnormal sensitization of myelin-reactive T cells, during either initiation or propagation of disease, have not been established. A variety of environmental triggers has been proposed, including common infections,3, 4 vitamin D deficiency,5 and dietary exposures.6 The possible contribution of dietary exposures to early events in MS, and in other autoimmune diseases such as type 1 (insulin-dependent) diabetes mellitus (T1D), has been debated over the years6–13 and continues to be of considerable interest14–19 because of the broad economic and health policy implications.

A major impediment to elucidating earliest events and putative environmental triggers in human autoimmune diseases, in general, relates to the considerable lag time between biological disease onset and clinical diagnosis. Childhood-onset clinically isolated syndromes (CIS) and MS provide a window into early disease mechanisms less susceptible to confounding by irrelevant exposures.

Here, we compare T-cell reactivities with target-organ and environmental dietary antigens in childhood-onset CIS/MS and in pediatric control cohorts, including children with T1D who represent an autoimmune disease that presents at a similar age and is thought to share a pathophysiology that involves a susceptible host, environmental triggers, and abnormal sensitization of T cells targeting self- (pancreatic) antigens. Our findings point to the presence of a relatively nonspecific abnormality of T-cell regulation in the early stages of both autoimmune diseases and challenge our view of the relation between CNS injury and CNS autoreactivity.

Subjects and Methods

  1. Top of page
  2. Abstract
  3. Subjects and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Appendix
  8. References
  9. Supporting Information

Subjects

We studied 7 pediatric cohorts (Table 1): (1) CIS (n = 37); (2) clinically definite MS (n = 26); (3) T1D (n = 41, <6 months after onset); (4) long-standing CNS insult (stroke, craniotomies, or surgery for epilepsy; n = 8); (5) epilepsy with no surgery (n = 14); (6) other, likely nonautoimmune neurological diseases (OND; including headache, developmental delay; n = 17); and (7) healthy children (n = 29). Informed consent/assent were obtained for all participants under approved protocols.

Table 1. Clinical and Demographic Features of All Subjects
  1. MS = multiple sclerosis; CIS = clinically isolated syndrome; CNS = central nervous system; OND = other neurological diesease.

inline image

None of the CIS children had experienced a second, MS-defining event at time of sampling. All MS children fulfilled the established McDonald20 or Poser21 criteria for clinically definite MS, having experienced two or more separate and well-defined clinical episodes of demyelination (n = 22) or magnetic resonance imaging evidence of new lesions (n = 4) evolving more than 30 days after an initial demyelinating event and involving two or more separate areas of the CNS. Four children with MS experienced an initial demyelinating event characterized by polyfocal neurological deficits, encephalopathy, and magnetic resonance imaging evidence of involvement of white matter and deep gray nuclei that was clinically indistinguishable from acute disseminated encephalomyelitis.22 All subjects subsequently experienced additional, nonacute disseminated encephalomyelitis–like demyelinating events, leading to a diagnosis of MS.

Recruitment and blood draws occurred an average of 0.49 year after presentation in the CIS cohort and an average of 4.6 years after the initial demyelinating episode for the MS cohort. Although infectious exposures associated with the initial acute presentation were therefore less likely to influence the measured T-cell responses, extensive infectious workup was performed at the time of initial presentation in all children with fever and/or symptoms/signs of possible infection (Table 2). All studies were negative with the exception of a single child whose serum (but not cerebrospinal fluid) tested positive for mycoplasma. T1D diagnosis was based on the American Diabetes Association criteria. The diabetic cohort was recruited as part of a long-term, prospective collaboration between the Children's Hospital of Pittsburgh and the Toronto Hospital For Sick Children (SickKids).19 All other participants were enrolled at the SickKids, including healthy children through orthopedic and dental clinics, with interviews and blood samples obtained on the day of elective outpatient surgery for minor procedures. Children with CNS insult, epilepsy, or OND were recruited and interviewed in neurology clinics, and blood samples were obtained at the time of requisite blood work. None of the MS or CIS participants had other immune conditions including asthma, atopic dermatitis, allergies, or psoriasis. Asthma was present in four of the OND control subjects and in one healthy control subject.

Table 2. Clinical Features of the Multiple Sclerosis and Clinically Isolated Syndrome Cohorts
  1. CIS = clinically isolated syndrome; MS = multiple sclerosis; ON = optic neuritis; TM = transverse myelitis; mono = monosymptomatic; poly = polysymptomatic; ADEM = acute disseminated encephalomyelitis; RRMS = relapsing-remitting MS; NA = not applicable; EDSS = expanded disability status scale37; IVIg = intravenous immunoglobulin.

inline image

Sample Processing and T-Cell Assays

Peripheral blood mononuclear cells (PBMCs) were isolated by standard Ficoll–Hypaque gradient centrifugation within 24 hours of blood draw. Sample processing was standardized, and blinded T-cell proliferation assays were conducted in a single laboratory using a protocol recently validated by the Immune Tolerance Network (http://www.immunetolerance.org)23 as employed in the multicenter TrialNet and Trial to Reduce Insulin-Dependent Diabetes in the Genetically at Risk (TRIGR) studies of diabetes.19, 24, 25 Test antigens (see Supplemental Table) included positive and negative controls, putative target tissue (myelin, pancreatic,24 glial) self-antigens, and selected dietary antigens. These were applied to dry, flat-bottom, 96-well plates in 50μl volumes containing 0.01 to 30μg antigen. PBMCs were washed, seeded into plates (105/well), and cultured for 5 days before measuring proliferation using 16 to 18 hours of3H-thymidine incorporation. Average counts of replicate wells were normalized to generate stimulation indices (SIs = counts per minute with test antigen/counts per minute with media alone).24 SIs of less than 1.5 were arbitrarily defined as nonreactive (negative). As a quality control for sample integrity, PBMCs from all 172 children demonstrated the expected T-cell proliferative response to a mitogen (phytohemagglutinin).

Results

  1. Top of page
  2. Abstract
  3. Subjects and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Appendix
  8. References
  9. Supporting Information

Clinical Demographics

Children with MS and CNS insult were slightly older, whereas children with CIS were similar in age as the healthy cohort and children with T1D, epilepsy, or OND (see Table 1). A family history of MS in first-, second-, or third-degree relatives was noted for 11% of the MS, 16% of the CIS, 7% of the epilepsy, 12% of the OND, 24% of the healthy cohorts, and none of the children with CNS insults. Family history data were not available from the T1D cohort. The prevalence of MS family histories in the healthy cohort was significantly greater than expected (p < 0.05), possibly reflecting greater interest in study participation given a family member with MS.

Table 2 details presenting clinical features of the children with CIS and MS (at the time of their first and second clinical events). Also shown are the total number of MS attacks, Expanded Disability Status Scale scores, results of infectious workup, and immunomodulatory or immunosuppressive therapies as determined on the day of blood sampling. Although we found no statistically significant difference between the overall T-cell responses of treated versus untreated children with CIS or MS, the number of antigens to which abnormal T-cell responses were seen tended to be lesser in the treated patient group (data not shown), and we cannot conclude “no treatment effect.” Such a treatment effect would serve to underestimate differences between the CIS/MS and control T-cell responses. Spinal fluid analysis for oligoclonal bands was available in a subset of patients. Oligoclonal bands were present in 12 of 17 (71%) children with MS, and in 6 of 11 (55%) of children with CIS.

T-Cell Reactivity to Target-Tissue Antigens

Figure 1 displays a dot plot of PBMC proliferation (depicted as SI) in response to individual test antigens expressed by myelin, glia, and pancreatic islet tissue for subjects in the CIS, MS, T1D, CNS insult, epilepsy, OND, and healthy control cohorts. The proportions of autoimmune and control cohorts exhibiting reactivities to the individual myelin and pancreatic antigens are depicted in Figure 2. Fewer than 10% of healthy children exhibited T-cell reactivities (defined as SI > 1.5 in our assay) to any one of the individual myelin antigens (whole myelin basic protein [MBP], exon-2 of MBP, or myelin oligodendrocyte glycoprotein [MOG]). Abnormally heightened reactivities to any individual myelin antigen were most notable in the children with CNS insult, diabetes, and CIS. Heightened T-cell responses were more apparent in the CIS cohort compared with MS, possibly because children with CIS were usually studied at time of acute relapse, whereas children with established MS typically provided blood at a scheduled visit, which may not be temporally related to acute disease relapse. The greater proportion of children treated with immune modulators (see Table 2) in the MS (27%) versus CIS (0%) cohorts may also have contributed.

thumbnail image

Figure 1. T-cell reactivities in autoimmune and control cohorts to individual target-tissue antigens. T-cell proliferative responses against myelin, pancreatic, and glial antigens are plotted for each child in the clinically isolated syndrome (CIS), multiple sclerosis (MS), type 1 (insulin-dependent) diabetes (T1D), central nervous system (CNS) insult, epilepsy, other neurological diseases (OND), and healthy cohorts. Note that only 17 of 41 of the T1D cohort were tested for T-cell reactivity to myelin oligodendrocyte glycoprotein (MOG). Stimulation index (SI) units are plotted on the y-axis. EX = exon; GAD = glutamic acid decarboxylase; GFAP = glial fibrillary acidic protein; MBP = myelin basic protein.

Download figure to PowerPoint

thumbnail image

Figure 2. Proportion of autoimmune and control cohorts reactive to individual myelin, pancreatic, and control antigens. The proportion (%) of children in the clinically isolated syndrome (CIS), multiple sclerosis (MS), type 1 (insulin-dependent) diabetes (T1D), central nervous system (CNS) insult, epilepsy, other neurological diseases (OND), and healthy cohorts exhibiting T-cell reactivity (defined as a proliferation stimulation index > 1.5) to myelin, pancreatic, and control antigens. EX = exon; GAD = glutamic acid decarboxylase; GFAP = glial fibrillary acidic protein; MBP = myelin basic protein; MOG = myelin oligodendrocyte glycoprotein; PHA = phytohemagglutinin.

Download figure to PowerPoint

When considering responses to one or more of the MBP antigens (either whole MBP, exon-2, or both), abnormal reactivities relative to healthy controls were notable in increased proportions of children with CNS insult (75% of the group; p < 0.0001) and children with diabetes (73%; p < 0.0001), followed by CIS (51%; p < 0.0001), as well as MS (46%; p < 0.0001), OND (29%), and epilepsy (20%). Of those children with detectable T-cell reactivities to whole MBP, more than 85% responded also to exon-2 of MBP, an antigen expressed only in immature myelin or during remyelination.26 Exon-2 has also been previously shown to be an immunodominant MBP antigen in adult MS.24 Because of a limited supply of MOG, only 62 participants (12 CIS, 12 MS, 17 T1D, 5 OND, 3 epilepsy, 8 CNS insult, and 5 healthy control subjects) were tested for responses to this antigen. Compared with healthy control children and children with T1D, who exhibited no abnormal T-cell proliferation to MOG, increased reactivities were observed in children with CIS and some of the children with MS, but also in CNS insult, epilepsy, and OND. We also assessed autoreactivities to glial fibrillary acidic protein (GFAP) and S100β, antigens expressed by both CNS astrocytes and peri-islet cell glia in the pancreas.25 Although less than 11% of healthy subjects, and approximately 14% of children with epilepsy and 18% of children with OND, demonstrated reactivities to either GFAP or S100β, we observed heightened reactivities to these self-antigens in children with CIS (GFAP = 22%; S-100 = 24%), MS (GFAP = 39%; S-100 = 35%), CNS insult (GFAP = 50%; S-100 = 50%), and T1D (GFAP = 63%; S-100 = 59%) (all p < 0.01 relative to healthy control subjects). We further examined T-cell reactivities to the pancreatic antigens human-proinsulin, glutamic acid decarboxylase 65kD (GAD65), and one of its major T1D-associated epitopes, the pGAD555-567,27 as well as to ICA69 and its dominant “Tep69” T-cell epitope. Approximately 90% of the T1D cohort responded to each of Tep69, GAD, and GAD555, compared with less than 13% of children in any of the other cohorts (all p values 0.001). Consistent with prior studies, 88% of children with T1D responded abnormally to proinsulin.19 T-cell responses to proinsulin were detected in 30% of children with CIS, 19% of MS patients, 75% of children with CNS insult, 14% of the epilepsy group, and 29% of children with OND compared with 14% of healthy control subjects. Our results in pediatric-onset MS contrast with prior reports in adults with MS, where T-cell responses to proinsulin (in the same assay) were seen in more than 60% of patients. These adult MS patients also had abnormal T-cell reactivities to other pancreatic antigens,24 suggesting that the heightened autoreactivity increases and spreads with longer disease duration.

T-Cell Reactivities to Dietary Antigens

PBMC proliferation in response to dietary milk antigens is depicted for the CIS, MS, T1D, CNS insult, epilepsy, OND, and healthy control cohorts in Figures 3 and 4. Up to 22% of healthy children exhibited T-cell reactivities to milk but not egg antigens. Responses to multiple milk antigens were seen in children with CIS (62%), MS (50%), CNS insult (75%), and T1D (93%) (p < 0.001 vs healthy controls), and to a lesser extent, in children with epilepsy (21%) or OND (41%). The overall pattern of reactivities to milk antigens detected in children with MS resembled the pattern previously seen in T1D and adult MS patients.16 Although these results initially suggest that abnormally heightened T-cell reactivity to multiple milk antigens is not a unique feature of either CIS/MS or T1D, we also found remarkable disease specificity at the level of particular epitopes. Proliferative responses to bovine serum albumin (BSA)-150 (ABBOS) were common in the T1D patients (93%), yet rare in children with CIS (8%), and largely undetectable in the MS and control cohorts. In contrast, 65% of CIS patients, 54% of the MS cohort, 75% of children with CNS insult, 21% of the epilepsy group, and 41% of the OND cohort targeted the BSA-193 epitope, which is encephalitogenic in rodents, whereas none of the T1D patients responded to this milk peptide.

thumbnail image

Figure 3. T-cell reactivities to dietary antigens. T-cell proliferative responses against specific milk antigens are plotted for each child in the different cohorts. Stimulation index (SI) units are plotted on the y-axis. BLG = beta lactoglobulin; BSA = bovine serum albumin; CIS = clinically isolated syndrome; MS = multiple sclerosis.

Download figure to PowerPoint

thumbnail image

Figure 4. Proportion of autoimmune and control cohorts reactive to milk antigens. The proportion (%) of children in the different cohorts exhibiting T-cell proliferative responses against the various milk antigens are contrasted. Note the high proportion of type 1 (insulin-dependent) diabetes (T1D) patients responding to the milk protein, Abbos (a protein rarely recognized by T cells in clinically isolated syndrome [CIS], multiple sclerosis [MS], central nervous system [CNS] insult, epilepsy, other neurological diseases [OND] patients, or healthy control subjects) in contrast with the responses to bovine serum albumin (BSA)-193. Cntl = control; PHA = phytohemagglutinin.

Download figure to PowerPoint

Reactivities to Milk Antigens as a Marker of T-Cell Autoreactivity

Although reactivity to one or more milk antigens was common in children with autoimmune disease, the presence of such reactivity did not appear to reflect a general loss of tolerance to dietary proteins, as only 2 of the 172 participants demonstrated T-cell proliferative response to the egg protein, ovalbumin (p < 0.0001, not shown). Of the 102 children with evidence of some reactivity to milk, 93 (92%) also exhibited responses to one or more myelin, glial, or islet-cell self-antigens. In contrast, of the 70 children with no evidence of reactivity to milk antigens, only 2 (<2%) recognized even a single self-antigen (p < 0.0001, Fisher's exact test) (Fig 5).

thumbnail image

Figure 5. T-cell reactivities to tissue antigens in children with negative reactivity to milk antigens. Lack of responses to dietary milk antigens is associated with lack of responses to tissue antigens. Dot-plot depiction of the virtual absence of T-cell reactivities detected against self-antigens in children from the different cohorts who do not exhibit T-cell reactivities against milk antigens. BSA = bovine serum albumin; CIS = clinically isolated syndrome; CNS = central nervous system; GAD = glutamic acid decarboxylase; GFAP = glial fibrillary acidic protein; MBP = myelin basic protein; MOG = myelin oligodendrocyte glycoprotein; MS = multiple sclerosis; OND = other neurological diseases; SI = stimulation index; T1D = type 1 (insulin-dependent) diabetes.

Download figure to PowerPoint

T-Cell Proliferative Responses as a Function of Age

An intriguing observation was that the reactivity profile of children to the different antigens tended to be influenced by age and, furthermore, was quite different across cohorts (Fig 6). With few exceptions, healthy children exhibited a lower level (SI < 2), and a more restricted range, of proliferative responses to the panel of antigens, particularly at early ages. Only 8% of the younger healthy children (age, 3–14 years) exhibited any antigen reactivity; older healthy children tended to have a broader range of reactivities to the same panel of antigens, though, even then, their reactivities were of relatively low amplitude (SI < 3). There were no obvious differences when comparing healthy children with or without family history of autoimmune disease. In contrast with the healthy control subjects, the majority of CIS children (including many of the youngest) exhibited a broad range of proliferative responses to the test antigen panel, and often with SIs well greater than 3. Responses in diabetic children were similar to the CIS cohort, with a broad range of proliferative responses seen even in the youngest children. Although reactivities in children with MS appeared to be intermediate to those seen for CIS patients and healthy control subjects, the small number of younger children in the MS group limits interpretation. Thus, it appears that a shared feature of both childhood-onset T1D and CIS is the presence of an abnormally broad range of T-cell reactivities to multiple dietary and self-antigens, and that this is seen in even the youngest patients studied.

thumbnail image

Figure 6. T-cell reactivities as a function of age. Dot plot of T-cell responses to a panel of tissue and dietary antigens, depicted in the clinically isolated syndrome (CIS), multiple sclerosis (MS), type 1 (insulin-dependent) diabetes (T1D), and healthy control cohorts, as a function of age. Dots in each vertical column represent the range of proliferative responses (stimulation index [SI] units) to the antigens, exhibited in a single child of a particular age.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Subjects and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Appendix
  8. References
  9. Supporting Information

T-cell–mediated responses are viewed as important contributors to the causative and active inflammatory injury of target tissues in autoimmune diseases such as MS and T1D. Loss of tolerance to target-specific autoantigens, enhanced capacity of activated T cells to migrate into the target, and aberrant T-cell functional profiles have all been considered relevant to early disease events and to the triggering of subsequent relapses and disease propagation. However, the identity of early disease targets is not established, and neither the host nor the environmental factors contributing to abnormal priming or sensitization of autoreactive T cells have been elucidated. Indeed, whether the CNS autoreactive T cells are essential to the different phases of the MS disease process, or whether they represent a heightened immune reactivity generated in response to CNS tissue injury remains unknown. Recently, Correale and Tenembaum28 reported similar myelin-directed T-cell reactivities in children and in adults with MS, raising the possibility that particular myelin epitopes are among the early disease targets. However, the presence of circulating autoreactive T cells, including those directed against myelin epitopes, is recognized as a normal feature of the immune repertoire. An important question is whether abnormal reactivities can be established when comparing patients with appropriate control subjects. Our parallel evaluation of children with CIS/MS and T1D, as well as other carefully assembled control cohorts, enabled us to generate novel insights into both shared and distinct mechanisms of human autoimmune diseases, and the relation between CNS injury and CNS autoreactivity.

The focus on childhood-onset disease is particularly relevant when assessing T-cell reactivities to early target antigens, because in adults with long-standing biological disease, abnormal self-reactivity may well develop secondary to the chronic injury. In the context of CNS inflammatory demyelination, we provide the first demonstration of abnormal T-cell reactivities against myelin antigens, detectable even in very young children with CIS and MS compared with age-matched control subjects. However, although these findings support the contention that aberrant T-cell activation is an early feature of the human inflammatory demyelinating disease spectrum, similarly abnormal reactivities to myelin antigens were seen in some neurological control subjects, and particularly those with the shared feature of chronic CNS insult. The heightened autoreactive T-cell responses would also appear to be accentuated as a function of age or disease duration, given both our current results, as well as the considerably higher autoreactivities reported using the same assay in prior studies of adults with MS.24

Chronic target organ injury is also a feature of autoimmune diabetes, in which pancreatic damage is not restricted to islet cells and is thought to extend to pancreatic neural elements, including glia and myelinated fibers.25 This would be consistent with our observation of abnormal T-cell reactivities to MBP and glial antigens in children with T1D. It is noteworthy that MOG is thought to be less broadly expressed in the periphery compared with MBP,29 which may explain our finding that heightened reactivities to MOG are not seen in the diabetic children, whereas they are observed in children with CNS inflammation or insult. Our findings thus raise the question whether myelin autoreactive T cells in MS/CIS cohorts are generated exclusively as a consequence of CNS tissue injury. If true, this would challenge the prevailing view that CNS autoreactive cells inherently mediate (initiate and/or propagate) the disease in CIS/MS. In this context, it has been suggested that inherent abnormalities in CNS tissue of MS patients, such as aberrant myelin structure, may result in abnormal immune responses.30, 31 Such a model would consider that heightened T-cell reactivities to CNS antigens, particularly early in the disease, may reflect the attempt of the immune system to normally respond to injured tissue.

It should be kept in mind that such heightened CNS autoreactivity may not be entirely “adaptive” and could subsequently contribute to propagating further CNS injury. Indeed, our findings do not exclude a pathophysiological role for all CNS autoreactive T cells in either pediatric or adult CIS/MS. Although heightened T-cell proliferation to myelin antigens is not specific to our pediatric CIS/MS cohort, it is possible that there are differences in the qualitative aspects of these responses, compared with the reactivities seen in CNS insult. For example, there may be more proinflammatory (Th1, ThIL-17) responses in CIS/MS cohorts, compared with the CNS insult cohort whose T-cell responses may not play an ongoing role in propagating CNS injury. Our recent observations in animal systems emphasizes that interplay between autoreactive T-cell pools and target tissues can be critical in separating normal autoimmunity from tissue-destructive autoimmune disease.32 It remains possible that additional target autoantigens, such as immature myelin antigens, are relevant in pediatric CIS/MS but not in adults. It is also possible that autoreactive T cells play different roles in different stages of an autoimmune disease process. For example, in animal models of diabetes, the initial pool of autoreactive T cells directed at pancreatic antigens is not capable of transferring disease. In our pediatric CIS/MS cohort, we might have expected a particularly aggressive profile of CNS myelin autoreactive T cells if these are indeed directly involved in the initiation and/or mediation of early CNS tissue injury.

Several common themes emerged as T-cell reactivities were compared between children with CIS/MS and T1D. The observation that children with diabetes, and to some extent also CIS and MS, exhibited heightened reactivities to multiple self-antigens derived from different tissues (CNS myelin, glia, pancreas) extends similar findings reported in adult populations24 and suggests that abnormalities that are not target-organ specific may not merely reflect chronic immune dysregulation. This is further supported by the striking observation that, within the pediatric CIS/MS and T1D cohorts (and quite distinct from healthy control subjects), even the very youngest patients exhibit an abnormally broad range of reactivities to multiple self- and dietary antigens. Together, these results may imply a common mechanism of the autoimmune disease process in both CIS/MS and T1D, with the early emergence of a tissue nonspecific abnormality in the regulation of T-cell responses. That abnormal reactivities to multiple pancreatic antigens appear more prevalent in adults with MS24 again suggests that such autoreactivity may increase and/or spread with longer MS disease duration.

Studies of T-cell proliferation in animal models may shed further light on the significance of our findings in human populations. For example, in the diabetes-prone nonobese diabetic (NOD) mouse model, circulating autoreactive T cells to both islet cell antigens and MBP have been demonstrated.16 The islet antigen-reactive T cells arise early in prediabetes, long before the development of overt T1D. Despite the concurrent presence of circulating myelin-reactive T cells, spontaneous CNS inflammatory demyelination does not occur. Inflammatory CNS disease (AENOD) is, however, readily induced after pharmacological breach of the blood–brain barrier. Thus, CNS autoreactivity can arise spontaneously in animal models, but some additional manipulation is still required to produce target-organ disease. In the NOD mouse, the “natural” target appears to be the islet, rather than the CNS. However, the propensity for these animals to contract EAE and diabetes16 underscores the potential for certain shared mechanisms across autoimmune diseases; indeed, in the NOD mouse, several autoimmune risk loci (idd4, eae7, orch3, strep) may map to the same mutant gene, TRPV1, on chromosome 11.32

Although our pediatric studies would support an early abnormality in the generation and regulation of autoreactive T-cell pools that may be shared in CIS/MS and T1D, the near-reciprocal pattern of reactivities to the BSA-193 and BSA-150 (ABBOS) dietary cow-milk epitopes that we observe in our cohorts invokes the presence of early processes that are quite specific for CIS/MS and diabetes, respectively. Abnormal sensitization to BSA has been previously reported in both adults with MS and patients with T1D.24 Similar to what has been described in adult MS, we identify in our pediatric CIS and MS cohorts significantly enhanced reactivities to the BSA epitope BSA-193, whereas such reactivity is absent in our T1D cohort. This is of interest because the BSA-193 peptide has been shown to be capable of inducing inflammatory demyelination in the commonly used SJL animal model of MS/acute disseminated encephalomyelitis.14, 16 The BSA-193 peptide has some structural similarities to MBP exon-2, and there were only seven MS or CIS patients where the presence or absence of BSA-193 and exon-2 responses was discordant. In contrast with the findings with BSA-193, we find that children with CIS and MS exhibit essentially no reactivity to the ABBOS epitope of BSA (extending a similar observation in adults with MS24), although increased reactivity to ABBOS is seen in most of our pediatric T1D cohort. ABBOS, which has strong antigenic homology with the Tep69 epitope of the T1D-associated autoantigen ICA69,19 is known to bind with high affinity to human and mouse T1D-risk–associated DQ/IA alleles; in diabetes-prone NOD mice, ABBOS can be diabetes protective and tolerogenic depending on dose and administration protocols.16

One interpretation of our observations is that antigenic mimicry between particular dietary-milk and self-antigens may represent an early disease mechanism shared by patients with CIS/MS and with T1D. This would be consistent with the early and prominent reciprocal abnormalities in T-cell responses to the BSA-193 and ABBOS dietary-milk epitopes observed in our pediatric CIS/MS and T1D cohorts. The strong association observed in our study between the presence or absence of increased reactivity to dietary antigens and the corresponding presence or absence of increased reactivity to self-tissue antigens is also intriguing.

However, an alternate interpretation of our findings is that induction of autoimmune disease and subsequent tissue injury may promote T-cell populations that can also cross-recognize environmental antigens. This would be most directly supported by our overall dataset, particularly the observations that heightened reactivities to dietary-milk antigens are also prominently seen in the CNS insult controls (see Fig 4), arguing against a simple “molecular mimicry” model. Indeed, in the diabetes animal model, although molecular mimicry was demonstrated, it was ruled out as the likely disease mechanism for the ABBOS-Tep69 epitope pair, using ICA69 knock-out mice.16 It is also quite possible that our observations are influenced by differences in ethnicity and/or distinct major histocompatibility complex backgrounds between cohorts, though the sample size does not provide sufficient power for such analyses. In this context, a particular major histocompatibility complex background may predispose to the independent generation of enhanced reactivities to dietary antigens on one hand and target-organ antigens on the other hand. Such heightened reactivities could thus represent an association, rather than a causal link.

Our study highlights the challenges involved in efforts to elucidate the relation between dietary antigens and autoimmunity. In both human T1D and its rodent models, heightened cow-milk reactivities have been reported19, 33, 34 and linked to observations that delayed weaning to foreign protein diets may protect T1D-prone hosts from disease.12, 34 This hypothesis currently is being tested in a global diabetes prevention trial, TRIGR, that compares weaning to standard versus severely hydrolyzed, nonantigenic infant formula.15 Although similar observations are unavailable for MS, high milk consumption was proposed as a possible MS risk factor years ago.7, 35 The prospective follow-up, from birth, of immune reactivity to milk (and relevant self-antigens) in the TRIGR trial may shed new light on the stepwise appearance of abnormal T-cell reactivity in hosts that were either exposed or not exposed to intact foreign dietary-milk proteins at weaning. Accurate assessment of early-life exposure to dietary cow milk is complex, particularly in Western societies in which infants are rarely breast-fed exclusively. In our pediatric cohorts, review of early-life nutrition showed no difference between the proportion of children exposed to infant formula compared with those who were breast-fed exclusively, but the power of this analysis is limited.

Collectively, we have shown that children with CIS, MS, T1D, and CNS injury harbor circulating T cells that exhibit heightened reactivities to a range of self-antigens, and that these abnormal autoantigen responses are not restricted to the target tissue of the disease. Our findings may point to a relatively nonspecific abnormality of T-cell regulation as an early feature of both CIS/MS and T1D, and highlight the relation between tissue injury and autoimmunity. The coexistence of a relatively nonspecific dysregulation of T-cell responses, combined with additional host and environmental factors, may then determine the emergence and possibly also the target-organ selectivity of autoimmune disease. A more complete examination of such a model will require studies to assess the functional properties of the abnormal T-cell autoreactivities and whether they are attributable to the memory versus naive T-cell populations, the study of additional environmental exposures including immune-pathogen interactions (eg, to common viral antigens such Epstein–Barr virus in MS36), and a deeper understanding of genetic and target-organ susceptibilities.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Subjects and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Appendix
  8. References
  9. Supporting Information

This study was funded by the Wadsworth Foundation (B.B., A.B.-O., L.B.K.), Canadian Institutes for Health Research (CIHR) (H.-M.D.), the NIAIDS (RO1 DK024021-27, D.J.B.), and a Multiple Sclerosis Society of Canada (MSSC) Don Paty Career Scientist Award (A.B.-O.).

We thank Dr Judd, Dr Tompson, Dr Barlow, Dr Narayanan, Dr Feldman, L. Medeiros, S. Parker, A. Dhillon, and the anesthesia staff at the Hospital for Sick Children for their assistance in recruitment of control participants. We also gratefully acknowledge the administrative support of J. Hamilton, M. Tantses, and L. MacMillan.

Appendix

  1. Top of page
  2. Abstract
  3. Subjects and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Appendix
  8. References
  9. Supporting Information

Additional Members of the Wadsworth Pediatric Multiple Sclerosis Study Group include Silvia Tenembaum, Hospital de Pediatria, Buenos Aires, Argentina; Anita Belman, SUNY, Stony Brook, NY; Alexei Boiko and Olga Bykova, Russian State Medical University, Moscow, Russia; Jayne Ness, Children's Hospital of Alabama, Birmingham, AL; Jean Mah and Cristina Stoian, University of Calgary, Calgary, Alberta, Canada; Marcelo Kremenchutzky, London Health Sciences, London, Ontario, Canada; Mary Rensel, Cleveland Clinic Foundation, Cleveland, OH; Jin Hahn, Stanford University, Stanford, CA; Bianca Weinstock-Guttman and Ann Yeh, SUNY, Buffalo, NY; Kevin Farrell, University of British Columbia, Vancouver, British Columbia, Canada; Mark Freedman, University of Ottawa, Ottawa, Ontario, Canada; Emmanuelle Waubant, University of California at San Francisco, San Francisco, CA; Martino Ruggieri, University of Catania, Catania, Italy; Matti Iivanainen, University of Helsinki, Helsinki, Finland; Virender Bhan, Dalhousie University, Halifax, Nova Scotia, Canada; Marie-Emmanuelle Dilenge, Montreal Children's Hospital, Montreal, Quebec, Canada.

References

  1. Top of page
  2. Abstract
  3. Subjects and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Appendix
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Subjects and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Appendix
  8. References
  9. Supporting Information

This article includes supplementary materials available via the internet at http://www.interscience.wiley.com/jpages/0364-5134/suppmat

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.