Abnormal effector and regulatory T cell subsets in paediatric-onset multiple sclerosis

Abstract

Elucidation of distinct T-cell subsets involved in multiple sclerosis immune-pathophysiology continues to be of considerable interest since an ultimate goal is to more selectively target the aberrant immune response operating in individual patients. While abnormalities of both effector (Teff) and regulatory (Treg) T cells have been reported in patients with multiple sclerosis, prior studies have mostly assessed average abnormalities in either limb of the immune response, rather than both at the same time, which limits the ability to evaluate the balance between effectors and regulators operating in the same patient. Assessing both phenotypic and functional responses of Teffs and Tregs has also proven important. In studies of adults with multiple sclerosis, in whom biological disease onset likely started many years prior to the immune assessments, an added challenge for any reported abnormality is whether the abnormality indeed contributes to the disease (and hence of interest to target therapeutically) or merely develops consequent to inflammatory injury (in which case efforts to develop targeted therapies are unlikely to be beneficial). Paediatric-onset multiple sclerosis, though rare, offers a unique window into early disease mechanisms. Here, we carried out a comprehensive integrated study, simultaneously assessing phenotype and functional responses of both effector and regulatory T cells in the same children with multiple sclerosis, monophasic inflammatory CNS disorders, and healthy controls, recruited as part of the multicentre prospective Canadian Pediatric Demyelinating Disease Study (CPDDS). Stringent standard operating procedures were developed and uniformly applied to procure, process and subsequently analyse peripheral blood cells using rigorously applied multi-parametric flow cytometry panels and miniaturized functional assays validated for use with cryopreserved cells. We found abnormally increased frequencies and exaggerated pro-inflammatory responses of CD8⁺CD161^highTCR-Vα7.2⁺ MAIT T cells and CD4⁺CCR2⁺CCR5⁺ Teffs in paediatric-onset multiple sclerosis, compared to both control groups. CD4⁺CD25^hiCD127^lowFOXP3⁺ Tregs of children with multiple sclerosis exhibited deficient suppressive capacity, including diminished capacity to suppress disease-implicated Teffs. In turn, the implicated Teffs of multiple sclerosis patients were relatively resistant to suppression by normal Tregs. An abnormal Teff/Treg ratio at the individual child level best distinguished multiple sclerosis children from controls. We implicate abnormalities in both frequencies and functional responses of distinct pro-inflammatory CD4 and CD8 T cell subsets, as well as Treg function, in paediatric-onset multiple sclerosis, and suggest that mechanisms contributing to early multiple sclerosis development differ across individuals, reflecting an excess abnormality in either Teff or Treg limbs of the T cell response, or a combination of lesser abnormalities in both limbs.

See Hohlfeld (doi:10.1093/brain/awz008) for a scientific commentary on this article.

Introduction

Elucidation of immune cell subsets and their responses that contribute to multiple sclerosis immune-pathophysiology continues to be of major interest. Cellular immune responses relevant to multiple sclerosis inflammatory disease activity almost certainly reflect interactions between different immune cell types (such as T cells, B cells and myeloid cells) and their pro-inflammatory and anti-inflammatory subsets (Li et al., 2015a). Among these, effector and regulatory T cells (Teffs and Tregs, respectively) continue to be regarded as important participants in mediating CNS injury in multiple sclerosis. While abnormalities in both Teffs (CD4⁺ and CD8⁺) and Tregs have been implicated in multiple sclerosis (Viglietta et al., 2004; Mahad et al., 2006; Venken et al., 2008; Annibali et al., 2011; Sato et al., 2012; Abrahamsson et al., 2013; Balint et al., 2013; Darlington et al., 2013; Vargas-Lowy et al., 2013), these findings have not always been replicated and it has been difficult to conclusively define disease-relevant immune cell subsets (Kleinewietfeld and Hafler, 2013; Carbajal et al., 2015; Grant et al., 2015). In part, such difficulties reflect inherent limitations of conducting these studies in adults, in whom biological disease onset likely started many years before biological sampling. This has made it challenging to distinguish abnormalities that reflect pathogenic roles of implicated immune cells from abnormalities that develop secondary to ongoing injury or chronic immune dysregulation in persons with long-standing disease. In contrast, a potentially narrower window between biological- and clinical-disease onset in paediatric multiple sclerosis offers an opportunity to elucidate early and putatively disease-relevant immune cell subsets and their abnormal response profiles.

Prior studies in adult- and paediatric-onset multiple sclerosis have generally focused on either effector or regulatory T cell measures, but not both. This includes two studies comparing paediatric multiple sclerosis to healthy paediatric controls, one reporting average abnormalities in Teffs (Vargas-Lowy et al., 2013) and the other implicating the Treg limb (Balint et al., 2013) of the immune response in multiple sclerosis children. An integrated examination of both Teff and Treg subsets in the same cohort would address key unanswered questions, including whether individual patients preferentially harbour abnormalities in effector or regulatory limbs of their immune response (or whether both abnormalities tend to co-exist within the same individual patients), and to what extent failure to contain effector responses relates to functional defects in the regulatory capacity of the Treg limb, versus resistance of the Teff limb to be regulated [as suggested in adults (Schneider et al., 2013)]. The importance of assessing not only phenotypes but also functional response capacities of disease-implicated cells has been emphasized in recent work in adult multiple sclerosis (Cao et al., 2015). We showed previously that, even in paediatric-onset disease, distinguishing abnormal immune responses that actually contribute to injury from those that emerge as part of the immune response to injury is not straightforward (Banwell et al., 2008). Comparing immune profiles of children with multiple sclerosis soon after their sentinel clinical attack not only with healthy control children, but also with children who experienced a monophasic inflammatory demyelinating episode of the CNS (i.e. non-multiple sclerosis CNS inflammatory disease), permits evaluation of early immune abnormalities of multiple sclerosis and, in particular, features that are more likely to be specific to multiple sclerosis rather than non-specific features of immune response to inflammatory CNS injury.

While studies in paediatric multiple sclerosis are attractive as noted above, such efforts have been constrained both by the small volumes of blood available and by the rarity of the disorder which often requires multi-site recruitment (with inherent challenges of ensuring consistency in sample processing and assay implementation). Here, we developed standardized methodologies for high quality multicentre sample procurement and handling, assay miniaturization and centralized implementation, and applied these to prospectively-followed children recruited at time of incident acute acquired demyelinating syndrome (ADS). This enabled us to prospectively define for the first time phenotypic and functional abnormalities in both Teff and Treg subsets that distinguish children confirmed to have multiple sclerosis from children with non-multiple sclerosis monophasic inflammatory CNS disorders (referred to as monoADS), and healthy controls subjects.

Materials and methods

Participants

We studied children recruited at time of presentation with incident acute ADS and followed prospectively as part of the Canadian Pediatric Demyelinating Disease Study (CPDDS) to ascertain those with a diagnosis of multiple sclerosis versus those remaining with monophasic disease (monoADS). Inclusion/exclusion criteria and clinical characteristics of the overall CPDDS have been previously described (Banwell et al., 2011) The Supplementary material details the criteria used to select representative participants for the current study and Table 1 identifies demographics of children with ADS ascertained as having multiple sclerosis, children with ADS remaining monophasic (monoADS), and healthy controls, including age at time of incident ADS presentation, age at time of blood sampling, and total follow-up duration from ADS onset. All protocols and informed consent/assent were approved by the institutional ethics review boards.

Standard operating procedures for CPDDS biological sample procurement, processing and sample analysis

Since reliable study of functional immune response profiles in children with a relatively rare condition is predicated on the ability to interrogate high-quality viable cell samples using limited volumes of blood obtained in the context of multicentre recruitment, our Experimental Therapeutics Program (ETP) developed and implemented standard operating procedures (SOPs) for all steps of venous blood sample procurement, handling, shipping, central lab processing (Ficoll density gradient centrifugation to peripheral blood mononuclear cells; PBMC), cryopreservation (liquid nitrogen vapour phase), storage and subsequent thawing of PBMC (Supplementary material, Appendices 1 and 2). Recruiting sites were all trained and regularly monitored, and the SOPs were consistently implemented throughout the CPDDS, with the identical approach also used for samples obtained from the healthy control children. The multi-parametric flow cytometry panels and functional assays (described below), together able to characterize both the phenotype and function of immune-cell subsets of interest, were also standardized, miniaturized and validated for use with cryopreserved cells obtained using the study SOPs.

Identifying and functionally-phenotyping T cell subsets of interest

Comprehensive literature review (Supplementary material andSupplementary Table 1) was carried out to identify effector (CD4⁺ and CD8⁺) T cell subsets putatively implicated in multiple sclerosis as well as other human autoimmune diseases. Multi-parametric flow-cytometry panels (Supplementary Tables 2 and 3) designed to characterize the T cell subsets within PBMC were developed and validated for use with cryopreserved samples (Supplementary material). PBMC were rapidly thawed and rested for 2 h. In each experiment, phenotypic and functional analyses of PBMC were analysed in batch using all samples passing predefined quality control (Supplementary material) from a balanced number of children with multiple sclerosis, monoADS and healthy controls. Rested cells were either left unstimulated or activated for 4 h with PMA/ionomycin and Golgi stop, then stained for T cell subset surface markers, and for intracellular cytokines, using appropriate isotype-controls (Supplementary Table 3). Samples were acquired on LSR-Fortessa (BD Biosciences), and analysed using FlowJo (version 10.7, Tree Star, Ashland, OR, USA). Supplementary Fig. 1 depicts the basic strategy for CD4⁺ and CD8⁺ T cell subset gating. HLA-DRB1 genotyping was carried out using allele specific PCR amplification as previously described (Disanto et al., 2011).

Cross-over Treg suppression assays and Teff resistance assays

Treg suppression assays were first developed using freshly isolated adult PBMC (Nyirenda et al., 2015), then miniaturized, adapted and validated for use with paediatric cryopreserved PBMC, and applied in a standardized way to balanced batches of samples from children with multiple sclerosis, monoADS and healthy controls (Supplementary material). Briefly, thawed PBMC were sorted into CD4⁺CD25^hiCD127^low Tregs; CD4⁺CD25⁻CD127⁺ (conventional responder T cells, Tresps); and CD4⁺CCR2⁺CCR5⁺ Teffs, with confirmed subset purities >95% (Supplementary Fig. 5). Treg suppression assays were performed in a ‘cross-over’ design to compare the ability of Tregs sorted from the different cohorts to suppress the same Tresps (conventional or Teffs), as well to assess the ability of responder cells sorted from the different cohorts to resist suppression by the same Tregs. CD4⁺CD25^hiCD127^low Tregs were titrated into co-culture with fixed numbers of Tresps (1.2 × 10³ stimulated with plate-bound anti-CD3 and anti-CD28, achieving Treg/Tresp ratios of 0:1, 1:8, 1:4 and 1:2 in 96-well U-bottom plates). After 5 days at 37°C, 16-h tritiated-thymidine incorporation was measured and cell proliferation was expressed as counts per minute (cpm). Per cent suppressive capacity of Tregs was determined as: (1 − cpm of Treg:Tresp co-culture/cpm of Tresp alone) × 100.

Statistical analysis

Between-group comparisons of average frequencies of surface-molecule expressing immune cell subsets and cytokine-expressing or FOXP3-expressing immune cell subsets were analysed by using two-tailed non-parametric Mann-Whitney U-test for independent groups and P-values are based on mono and multiple sclerosis z-scores derived from the healthy control group [using mean and standard deviation (SD)]. For Treg suppression assays, the mean ± SEM counts per minute (cpm) of triplicate cultures was calculated for each co-culture condition, and data were analysed using two-tailed paired comparisons with the Wilcoxon signed rank test for paired samples (monoADS and multiple sclerosis). All statistical analyses were performed using SPSS software (IBM SPSS Statistics for Windows, Version 24.0. Armonk, NY: IBM Corp). P-values < 0.05 were considered statistically significant.

Data availability

The data that support the findings of this study are available from the corresponding author, upon reasonable request.

Results

Study participants

Based on the planned selection criteria (see ‘Materials and methods’ section and Supplementary material), 71 children were studied (Table 1), including 23 with confirmed multiple sclerosis (female:male = 14:7; age at ADS: 12.8 ± 2.4 years; age at sampling: 14.1 ± 2.3 years), 26 with monophasic disease (female:male = 15:11; age at ADS: 9.6 ± 3.2 years; age at sampling: 12.3 ± 2.9 years) for whom biological samples were available for comparison, and 22 healthy control children without a personal or family history of multiple sclerosis or autoimmune disease (female:male = 11:11; age at sampling: 14.4 ± 2.1 years).

Abnormal frequencies and responses of CD4⁺CCR2⁺CCR5⁺ and CD8⁺ MAIT T cells in paediatric multiple sclerosis

Overall frequencies of CD3⁺ or of CD4⁺ and CD8⁺ T cell subsets (Supplementary Fig. 1) were no different between children with multiple sclerosis, monoADS or healthy controls, nor were there differences in the frequencies of activation-induced IFNγ or IL-17 expressing T cells within the total CD4 or CD8 T cell subsets across cohorts (Supplementary Figure 2). However, children with multiple sclerosis harboured an average of 6.6% CD8⁺ CD161^hiTCR-Vα7.2⁺ cells within their circulating CD8 T cell pool (a phenotype commonly associated with mucosal-associated invariant chain T cells; MAIT cells), compared to both healthy control (2.8%) and children with monoADS (2.6%) (Fig. 1A and B). We confirmed that these CD8⁺ CD161^hiTCR-Vα7.2⁺ T cells were indeed MAIT cells by co-staining them with MR1 tetramer (MR1–5-OP-RU⁺) (Supplementary Fig. 3). PBMC activation revealed abnormally increased induction of both IFNγ and IL-17 expression by the MAIT cells of children with multiple sclerosis (25.8% and 5.4%, respectively) compared to both healthy controls (4.3% and 2.2%) and monoADS control groups (3.8% and 2.8%) (Fig. 1C–E). Children with multiple sclerosis also selectively harboured increased frequencies of CCR2⁺CCR5⁺ CD4⁺ T cells (Fig. 2A and B) averaging 6.5% of circulating CD4+ T cells, compared to both healthy control (3.5%) and children with monoADS (also 3.5%). These cells in the children with multiple sclerosis also exhibited abnormally increased activation-induced Th1 and Th17 pro-inflammatory cytokine responses (7.95% and 8.9%, respectively), compared to both healthy control children (4.5% and 3.2%) and children with monoADS (5.4% and 5.1%) (Fig. 2C–E).

Reduced FOXP3 expression and suppressive capacity of Tregs in children with multiple sclerosis

While overall frequencies of CD4⁺CD25^hiCD127^low T cells (an estimate of regulatory T cells; Tregs) did not differ across groups (average frequencies for multiple sclerosis: 3.5%; healthy control: 4.1%; monoADS: 3.4%) (Fig. 3A and B), their average expression of FOXP3 (Fig. 3C) was slightly lower in children with multiple sclerosis compared to both monoADS and healthy control children. Subset analysis indicated that the frequency of memory (CD45RO⁺FOXP3⁺) Tregs was increased, whereas frequencies of naïve (CD45RO⁻FOXP3⁺) Treg and RTE (recent thymic immigrant) (CD31⁺CD45RO⁻FOXP3⁺) Treg subsets, as well as Tregs expressing CD39, were mildly decreased in children with multiple sclerosis compared to both control cohorts (Supplementary Fig. 4). Since FOXP3 expression has been correlated with Treg suppressive capacity, we compared the relative capacities of CD4⁺CD25^hiCD127^low Treg sorted from the different groups (Supplementary Fig. 5 for gating strategy) to suppress proliferation of conventional responder (CD4⁺CD25⁻CD127⁺) T cells, sorted from healthy control subjects (healthy control Tresps). While Tregs of monoADS and healthy control children exhibited similar capacities to suppress proliferation of the same normal Tresps, Tregs of multiple sclerosis children exhibited deficient suppressive capacity (Fig. 3D).

Paediatric multiple sclerosis Tregs exhibit reduced capacity to suppress disease-implicated Teffs that are resistant to suppression

We next assessed the capacity of Tregs from multiple sclerosis children to suppress the very effector cells implicated as abnormal in these children. We focused on the CD4⁺CCR2⁺CCR5⁺ Teff subset (whose activation is more readily induced in short-term cultures, as compared to MAIT cells). Compared to Tregs of both monoADS and healthy control children, Tregs of multiple sclerosis children exhibited deficient capacity to suppress the CD4⁺CCR2⁺CCR5⁺ Teffs (Fig. 3E). Our cross-over design enabled us to further examine the relative susceptibility of responder cells to Treg suppression. Both conventional (CD4⁺CD25⁻CD127⁺) responder T cells, as well as the disease-implicated effector (CD4⁺CCR2⁺CCR5⁺) T cells, isolated from children with multiple sclerosis were relatively resistant to regulation compared to the responder T cell subsets of healthy children (Fig. 3F).

An abnormal Teff:Treg balance distinguishes multiple sclerosis children from controls

Since resistance to Treg suppression has recently been reported in CD4⁺ Th17 cells from patients with rheumatoid arthritis that co-express CD161⁺ (Basdeo et al., 2015), we assessed co-expression of CCR2, CCR5 and CD161 on CD4⁺ T cells. First, we found that multiple sclerosis children harbour abnormally increased frequencies of circulating CD4 T cells expressing CD161 (average 10.1%) compared to both healthy control (6.3%) and monoADS controls (6.1%) (Fig. 4A and B).

Compared to the frequencies of CCR2⁺CCR5⁺ T cells within total CD4⁺ T cells (Figs 2A, B and 4D), CCR2⁺CCR5⁺ T cells were enriched within the CD4⁺CD161⁺ cell population, and this enrichment was substantially exaggerated in the children with multiple sclerosis (Fig. 4C and D). Further supporting a pathogenic role of this particular Teff subset in multiple sclerosis, was the observation that the balance between these CD4⁺CD161⁺CCR2⁺CCR5⁺ T cells and the Tregs (assessed using the simple ratio between the Teff/Treg frequencies in individual children), was highly abnormal in the multiple sclerosis children, better distinguishing them from both healthy and monoADS children (Fig. 4E) than any single abnormality noted above.

Serial assessment of Teff:Treg balance in children with multiple sclerosis and monoADS controls

We considered whether the Teff/Treg abnormalities noted in individual children reflected episodic abnormalities (potentially related to disease activity) or a characteristic of the individual that is relatively stable over time. To address it, we serially studied the Teff and Treg frequencies as well as Teff/Treg ratios in 10 children (six with multiple sclerosis and four with monoADS) at three time points (separated over at least 2 years). To avoid the potential confounding effects of immune therapies we focused on children for whom we had collected serial untreated samples (on no disease-modifying therapy and at least 60 days following steroid treatment, for all time points). We first considered the trajectories of the CD4⁺CD161⁺CCR2⁺CCR5⁺ Teff frequencies and CD4⁺CD25^hiCD127^lo Treg frequencies in all children captured at the three time points (TP1–3 in Fig. 5A and B). While fluctuations were observed, overall, there appeared to be particular individuals with consistently lower frequencies, and other individuals with overall higher frequencies of both Teffs and Tregs, over time. We then calculated the Teff/Treg ratios, shown separately for the four children with monoADS (Fig. 5C) and the six children with multiple sclerosis (Fig. 5D). The Teff/Treg ratios of the monoADS children tended to be low, and relatively stable over time. In contrast, there appeared to be greater variability across the children with multiple sclerosis; several of the multiple sclerosis children exhibited greater magnitude fluctuations though, overall, particular individuals again tended to have lower ratios and others, higher ratios, over time.

Discussion

While abnormalities of Teffs and Tregs have been reported in patients with multiple sclerosis, prior studies have typically assessed either limb of the immune response. We hypothesized that mixed contributions of abnormalities in effectors and regulators may exist across individuals with multiple sclerosis. Testing this would require both limbs of the immune response to be assessed at the same time. To better understand how such an imbalance between disease-implicated effectors and regulators may manifest in patients with multiple sclerosis, we chose to focus on paediatric-onset multiple sclerosis since these individuals, on average, would be expected to be closer to biological disease onset, and any imbalances between effectors and regulators may more likely reflect their actual involvement in the disease process rather than manifestation as epi-phenomena or consequences of chronic disease and immune response to injury. We also considered that past difficulties duplicating individual study findings (such as consistent disease-implication of particular immune cell subsets) may in part reflect methodological differences between studies, including the limited number of studies that assess both phenotype and function of subsets of interest and the lack of inclusion of controls with other (non-multiple sclerosis) inflammatory CNS disease.

We report abnormal frequencies and pro-inflammatory functions of distinct subsets of CD4 and CD8 T cells in children with multiple sclerosis compared to both healthy and other CNS inflammatory disease (monoADS) controls. In findings that further support prior reports in adults with multiple sclerosis, we identify abnormalities in the function of paediatric multiple sclerosis Tregs, as well as decreased susceptibility to suppression of their disease-implicated pro-inflammatory T cells. Based on our integrated analysis of both effector and Treg populations, we suggest that multiple sclerosis can manifest when the balance between the effector and regulatory limbs of the immune response in individual children is disrupted, whether the abnormality is predominantly due to exaggerated pro-inflammatory responses of their distinct effectors, insufficient regulation by their regulatory cells, or lesser combinations of both (Fig. 6).

Recruitment of children following the sentinel clinical presentation of their disease, enabled us to evaluate immunological features of paediatric multiple sclerosis without the confound of immunomodulatory therapy exposure. We note that selection of samples from patients who remained untreated over the first years of disease could bias our sample to individuals who opted not to be treated, including those with milder disease. However, such a bias would be expected to diminish rather than enhance differences between disease-relevant immune cell subsets assessed in children with multiple sclerosis compared to controls. To overcome operational challenges inherent to the study of rare patient populations, and to minimize the impact of pre-analytical variables that can severely hamper sample quality and interpretation of results in such studies, we developed and validated rigorous methodologies for all phases of sample processing and analysis (Supplementary material, Appendices 1 and 2). Consistent implementation of these protocols which we validated for use with the small-volume samples, multisite procurement, and cryopreserved specimens, enabled us to comprehensively characterize distinct immune cell subsets, both phenotypically and functionally, as well as evaluate how distinct Teff and Treg subsets perform within a given patient and across patients and controls.

We implicate MR1-tetramer-positive CD8⁺CD161^high TCR-Vα7.2⁺ (MAIT) cells for the first time in paediatric multiple sclerosis, reporting not only abnormally increased circulating frequencies, but also a greater functional propensity to respond with expression of IL-17 and IFNγ compared to these cells in both children with monoADS and healthy controls. MAIT cells are best known for their antimicrobial activity (Le Bourhis et al., 2010; Kurioka et al., 2015) and are described as tissue-targeting cells that secrete IL-17 (Dusseaux et al., 2011). While MAIT cells have been implicated as pathogenic in several autoimmune diseases (Supplementary Table 1), including in adult multiple sclerosis where they have been described within brain lesions, there remains controversy regarding the roles of these cells (Illes et al., 2004; Annibali et al., 2011; Dusseaux et al., 2011; Miyazaki et al., 2011; Abrahamsson et al., 2013; Willing et al., 2014, 2018; Held et al., 2015; Salou et al., 2016). On one hand, a number of these studies have reported increased IL-17⁺ (presumably proinflammatory) cells in patients with multiple sclerosis, and the substantially decreased relapsing multiple sclerosis disease activity seen following bone marrow transplant in multiple sclerosis has been associated with decreased IL-17 responses of both CD4 and CD8 T cells and diminished frequencies of circulating MAIT cells in the treated patients. (Abrahamsson et al., 2013; Darlington et al., 2013). On the other hand, MAIT cell enrichment or adoptive transfer of MAIT cells significantly reduced the incidence and severity of experimental autoimmune encephalomyelitis (EAE) implicating a regulatory function for these cells in the animal model (Croxford et al., 2006), and MAIT cells were shown to suppress IFNγ production from human T cells (Miyazaki et al., 2011). While our findings point to a pro-inflammatory role of MAIT cells in paediatric multiple sclerosis, future work is required to more fully characterize the functional response profiles of these cells early in the multiple sclerosis process. It is noteworthy that decreased (rather than increased) frequencies of MAIT cells have been described in the circulation of patients with several autoimmune diseases compared to controls, for example in type 1 diabetes (Gulden et al., 2017; Rouxel et al., 2017) and Crohn’s disease (Chiba et al., 2018). Such observations have advanced the plausible explanation that the cells of interest are decreased in the blood since they have trafficked into the target organ. In multiple sclerosis, multiple studies over the years have identified increased (rather than decreased) frequencies and/or inflammatory response propensities of other disease-implicated T cell subsets in the circulation (such as Th1/Th17 cells; Kebir et al., 2009). Thus, how to interpret an increase or decrease in counts or frequencies of a circulating cell subset when identified in patients compared to controls is likely to be context dependent and may remain speculative until a directed intervention can more definitively ascribe in vivo function. As recently reviewed by Chiba et al. (2018) there have been variable reports on frequencies of MAIT cells in the circulation of adult patients with multiple sclerosis compared to controls, with some reporting decreased frequencies, no changes in frequencies, as well as increased frequencies or increased frequencies at time of relapse. It is intriguing to speculate that during early (e.g. paediatric onset) disease, a particular cell subset may initially be found at increased frequencies in the circulation (as the subset expands in the periphery and begins to traffic to the target organ), and later in disease decreases in the circulation (as the cells have largely trafficked into the target organ). A recent laser microdissection study of T cells from multiple sclerosis lesions, followed by single cell PCR characterization of paired TCRα and TCRβ chains, revealed clonally expanded populations expressing the canonical Vα7.2(+) MAIT cell chain, pointing to an antigen-driven process with further evidence that the same MAIT cell clones may persist for years in the circulation of adult patients with multiple sclerosis (Held et al., 2015). Elucidating the antigenic specificities of MAIT cells in paediatric multiple sclerosis in an effort to enumerate and functionally study these cells, and investigating their potential relation to the microbiome (Tremlett et al., 2016), will be of considerable interest in better defining their relevant function(s) in multiple sclerosis. Ultimately, selective targeting of MAIT cells in patients may be the only strategy to definitively elucidate whether they play an injurious or regulatory role in the human disease (Rahimpour et al., 2015).

T cells expressing CCR2 [which binds several chemokines including CCL2 (MCP-1)] or CCR5 [which binds CCL3 (MIP-1a) and CCL5 (RANTES) amongst others] have individually been implicated in autoimmune diseases including colitis and multiple sclerosis (Supplementary Table 1). In multiple sclerosis, earlier immunohistological studies identified both CCR2 and CCR5 expressing cells in perivascular lesions of multiple sclerosis brain with co-localization of their ligands around the vessels (Balashov et al., 1999; Simpson et al., 2000). Our findings in paediatric-onset multiple sclerosis support a pro-inflammatory role for the subset of CD4⁺ T cells that co-express both CCR2 and CCR5, a chemokine receptor profile particularly suited for cell homing to the CNS (Balashov et al., 1999; Fife et al., 2000; Simpson et al., 2000; Kivisakk et al., 2002; dos Santos et al., 2005; Mahad et al., 2006; Sato et al., 2012). Unlike CD8⁺ MAIT cells that have been more variably attributed either pro-inflammatory or anti-inflammatory functions, CD4⁺CCR2⁺CCR5⁺ T cells have been consistently implicated as pro-inflammatory, including prior findings from adult patients with multiple sclerosis, in whom CD4⁺CCR2⁺CCR5⁺ T cell frequencies were found increased in the CSF, as well as in the blood at time of relapse (Sato et al., 2012). Functionally, circulating CCR5⁺ T cells isolated from PBMC of multiple sclerosis patients were previously shown to secrete high levels of IFNγ (Balashov et al., 1999) and the interaction of CCR2 with its ligand MCP-1 has been shown to play an important role in effector CD4⁺ T cells migration across the blood brain barrier (Mahad et al., 2006). By assessing functional responses of both effector and regulatory T cells in our paediatric population, we provide support for a pro-inflammatory role of CD4⁺CCR2⁺CCR5⁺ T cells in paediatric-onset multiple sclerosis, not only with the demonstration that they tend to circulate at higher frequencies and exhibit abnormally increased IL-17 and IFNγ responses but also with the novel observation that this T cell subset is relatively resistant to regulation in the children with multiple sclerosis. Future studies with technologies such as single cell RNASeq using 10× genomics applied to CSF cells are warranted to assess whether the Teff subsets implicated here in the periphery of children with multiple sclerosis are also implicated in the CNS of these children.

Both reduced functional capacity of Tregs and resistance of effector cells to suppression by Tregs have been proposed to play a role in the incomplete control of pro-inflammatory effector cells in multiple sclerosis (Goodman et al., 2011; Wehrens et al., 2011; Schneider et al., 2013; Trinschek et al., 2013; Nyirenda et al., 2015). The cross-over design of our functional assays extends prior work (Balint et al., 2013) by demonstrating that defective Treg function in multiple sclerosis children is not a general feature of paediatric CNS inflammatory demyelinating conditions. It demonstrates further that, among the overall responder T cell population, resistance to suppression is a feature of the very CD4⁺CCR2⁺CCR5⁺ effector T cell subset we implicate here as abnormally pro-inflammatory in the multiple sclerosis children. The molecular mechanisms underlying Teff resistance to suppression are not fully elucidated and may include decreased expression by the Teff of receptors that bind inhibitory Treg cytokines (such as IL-10 or TGFβ or limiting Treg function through downregulation of Foxp3 within the Treg or local secretion of cytokines including IL-6, IL-17 or TNFα by other cells (Grant et al., 2015). While the present study focuses on T cells given their strong implication in multiple sclerosis, it is well-recognized that the T cell contribution to multiple sclerosis likely involves important interactions with other cells (Jelcic et al., 2018). In this regard, the remarkable success of selective B cell depletion using anti-CD20 therapies in multiple sclerosis patients (Hauser et al., 2017) together with the elucidation of non-antibody dependent functions of B cells (including their ability to shape pro-inflammatory T cell responses through secretion of TNFα, IL-6 and GM-CSF (Bar-Or et al., 2010; Barr et al., 2012; Li et al., 2015b) and reviewed in (Li et al., 2015a) implicate important B cell:T cell interactions in the human disease. Of interest in future studies will be to examine such interactions as well as both antibody-dependent and antibody-independent functions of B cells and their relation to disease activity in patients with paediatric- and adult-onset multiple sclerosis.

In addition to abnormally increased frequencies and exaggerated pro-inflammatory cytokine responses, another common feature of the two Teff subsets that we implicate in children with multiple sclerosis is their co-expression of CD161 (KLRB1; killer cell lectin-like receptor subfamily B member). While functions of CD161 are not fully elucidated, its expression has been associated with resistance to suppression by Treg in patients with rheumatoid arthritis (Basdeo et al., 2015). Of note, we found that CCR2⁺CCR5⁺ T cells are enriched within CD4⁺ T cells expressing CD161, and to a much greater extent in children with multiple sclerosis compared to controls. Moreover, children with multiple sclerosis exhibited a highly abnormal ratio between their circulating CD4⁺CD161⁺CCR2⁺CCR5⁺ Teffs and Tregs. Together our results point to the CD4⁺CD161⁺CCR2⁺CCR5⁺ subset of T cells as the most abnormally implicated subset of Teffs in children with multiple sclerosis, exhibiting increased frequencies, exaggerated pro-inflammatory responses and potential resistance to Treg suppression (that may be due to both active molecular mechanisms of resistance, as well as numerical abundance relative to Tregs).

In a small subset of paediatric patients, we explored the behaviour of the balance between Teffs and Tregs (assessed as the Teff/Treg ratios in individual children) over time, using serial samples collected over at least a 2-year period. Specifically, we wished to assess whether the Teff/Treg abnormalities we noted in individual children with multiple sclerosis reflected episodic abnormalities (that could potentially reflect the state of disease activity) or a characteristic of the particular individual that is relatively stable over time. These samples were selected from children who were off disease modifying treatment, to avoid potential confounds of immune therapy. This could introduce a bias towards children with relatively little disease activity over time, although many of these samples were obtained at a time that most children with multiple sclerosis remained untreated with disease-modifying therapies unless they had particularly active disease. Because we also avoided recent steroid use, these samples were not collected around the time of acute clinical relapses (though this would not preclude subclinical disease activity), hence our assessment is not ideally suited for defining whether multiple sclerosis patients experience major fluctuations in their Teff/Treg ratios in relation to acute disease activity. Rather, our findings are in keeping with the view that, overall, greater variability exists across individuals, while the ratio within an individual tends to be a relatively stable characteristic of that individual, over time. This finding is consistent with results from Kivisäkk et al. (2003) who reported previously that expression of CCR2 and CCR5 by CD4⁺ T cells is stable over time in the same person, though varies across individuals.

The samples we selected to compare children with multiple sclerosis and monoADS were well balanced with respect to HLA-DRB1 genotype (with ∼65% homozygous-negative, 25% heterozygous and less than 10% homozygous-positive in both cohorts). Human leucocyte antigen (HLA) may be particularly relevant in the context of antigen-specific MHC-restricted responses, while our study compared overall T cell subset frequencies and responses to polyclonal activation. The balanced HLA-DRB1 genotype between the patients with multiple sclerosis and monoADS in our cohort suggests that this genotype is not the driving force for the abnormal Teff/Treg balance seen in children with multiple sclerosis. One could speculate that HLA influences multiple sclerosis disease susceptibility through mechanisms that are different or complementary to those that modulate the state of disease in patients with established disease. This also raises the interesting possibility that the differences in T cell subset responses that we report here between multiple sclerosis and monoADS cohorts may be shaped by other genetic/environmental/epigenetic features. Future studies with larger numbers of paediatric multiple sclerosis patients and controls could examine the relationship between the genome-wide association study-implicated genetic variants as well as other suspected disease modulators (e.g. vitamin D levels, EBV status, adiposity and smoking exposure) on the Teff and Treg measures in these individuals, and potentially relate these to clinical and MRI measures of disease activity.

Collectively, our findings indicate that the ultimate capacity to regulate disease-relevant effector T cell responses in children with multiple sclerosis may reflect a combination of deficient suppressive capacity by the regulatory compartment, as well as relative resistance to suppression by particular pathogenic effector subsets. Future studies will establish whether measurement of the particular Teff/Treg ratio (rather than each measure in isolation) will better serve to distinguish multiple sclerosis patients from controls and, if validated, emerge as a useful measure of disease state and response to particular therapies. An abnormal balance between CD4⁺ pro-inflammatory Teffs co-expressing CD161, CCR2 and CCR5 may not only be a distinguishing feature of individuals with early multiple sclerosis, but may also emerge as a useful measure of disease state and possibly response to therapy, warranting future study.

Abbreviations

Abbreviations
 
• CPDDS

  Canadian Pediatric Demyelinating Disease Study

 
• IFNγ

  interferon gamma

 
• MAIT

  mucosal-associated invariant chain T cells

 
• (mono)ADS

  (monophasic) acquired demyelinating syndrome

 
• PBMC

  peripheral blood mononucelar cell

 
• TCR

  T-cell receptor

 
• Teff

  T effector cell

 
• Treg

  T regulatory cell

 
• Tresp

  responder T cell

Acknowledgements

We thank Sandeep Vanamala, Julien Sirois, Ada Villalobos, Boli Fan and Chahrazed Belabani, members of the Experimental Therapeutics Program, for their superb and consistent handling of samples, critical for success of the assays. The authors gratefully acknowledge the site investigators of the Canadian Paediatric Demyelinating Disease Network: Drs Katherine Wambera, Mary B. Connolly, Jerome Yager, Jean K Mah, Fran Booth, Guillaume Sebire, David Callen, Brandon Meaney, Marie-Emmanuelle Dilenge, Anne Lortie, Daniela Pohl, Asif Doja, Sunita Venkateswaran, Simon Levin, E. Athen MacDonald, David Meek, Ellen Wood, Noel Lowry, David Buckley, Conrad Yim, Mark Awuku, Pamela Cooper, Francois Grand’Maison, J. Burke Biard, and Virender Bhan. The authors further wish to acknowledge the invaluable assistance of the coordinator(s) at each site, as well as Ms. Rozie Arnaoutelis and Ms. Danusha Nandamalvan. The authors also thank the participating children and their families; this study would not have been possible without their cooperation and commitment. The MR1 tetramer technology was developed jointly by Dr. James McCluskey, Dr. Jamie Rossjohn, and Dr. David Fairlie, and the material was produced by the NIH Tetramer Core Facility with permission to distribute provided by the University of Melbourne.

Funding

The MS Scientific Research Foundation of Canada (A.B-O., D.L.A., B.B., R.A.M., A.D.S. and E.A.Y.); I.M. was supported by the Canadian Institutes of Health Research (CIHR) Vanier Graduate Scholarship, the CIHR Training Grant in Neuroinflammation, and the Neuroscience Graduate Group, University of Pennsylvania, M.N. was supported by an MSSOC Research Fellowship, R.L. was supported by the Banque Nationale Fellowship; J.O’M. was supported by the Waugh Family Multiple Sclerosis Society of Canada Doctoral Studentship award; A.R. was supported in part by a Studentship of the CIHR Training Grant in Neuroinflammation. B.B. is funded by the National MS Society, NIH, and the Canadian Multiple Sclerosis Scientific Research Foundation.

Competing interests

I.M., M.N., R.L., J.O’M., A.Re., A.Ro., C.M., T.N., D.S. and L.C. have nothing to disclose. D.L., D.L.A., B.G., A.Y., R.A.M. and A.B.O. have no disclosures relevant to this manuscript. B.B. has served as a central MRI reviewer and speaker for Novaritis, and as a non-remunerated advisor on clinical trial design to Novartis, Sanofi, Teva Neuroscience and Biogen.

References

Author notes