Circulating Follicular Regulatory T Cells Are Defective in Multiple Sclerosis

Given the prompt emergence of somatically mutated B cells during B cell maturation in the germinal center (GC) response, it seems evident that regulatory pathways need to be engaged to secure the maintenance of self-tolerance (1, 2). In this GC process, follicular Th cells (T_FH) are believed to select B cells based on their capacity to bind and present the specific Ag (3). This survival-of-the-fittest mechanism requires a limited amount of T_FH cells to ensure competition and thereby elimination of undesirable (autoreactive) B cells. The recently identified follicular regulatory T cells (T_FR) propose themselves as ideal candidates for regulating the normal GC response and preventing emergence of autoreactive B cells (4–6). T_FR cells comprise 5–25% of the GC T cells in mice and originate from Foxp3⁺ thymic-derived regulatory T cell (Treg) precursors (4). Like T_FH cells, T_FR differentiation depends on CD28, B cell lymphoma 6 (BCL-6), signaling lymphocytic activation molecule–associated protein (SAP), and B cells (4). Phenotypic analysis of T_FR overlaps with the surface profile of T_FH cells (CD4⁺CXCR5⁺PD-1⁺ICOS⁺), and both cell types also express BCL-6. In contrast with T_FH, T_FR do not express IL-20, IL-4, or CD40L. Besides sharing surface markers with T_FH cells, T_FR cells also express regulatory markers such as Foxp3, CD25, CTLA-4, glucocorticoid-induced TNFR-related protein, and IL-10 (4–6). The transcription factor NFAT2 has been identified as crucial for T_FR development in mice, because it is essential for the induction of CXCR5 in these cells (7). CXCR5 directs the migration of T_FR into the GC where they suppress the magnitude of the GC response by regulating the number of T_FH, B cells, and Abs. Taken together, T_FR are a distinct population of Tregs located primarily in the GC, showing overlap with both Tregs and T_FH.

Because regulation of (auto)antibody production primarily takes place in the GC, it is probable that an aberrant GC response can contribute to autoimmunity. In the absence of T_FR, overwhelming outgrowth of nonspecific B cells leads to lower amounts of Ag-specific B cells (8). Furthermore, a lack of CXCR5⁺ Tregs in mice results in increased GC activity, including affinity maturation and differentiation toward plasma cells. In addition, PD-1 was found to be crucial for the homeostasis of T_FR, because PD-1–deficient mice show increased numbers of T_FR in lymph nodes. Also, the PD-1 pathway seems to impair T_FR function, considering PD-1 deficiency results in an increased T_FR suppressive capacity (9). NFAT2-deficient T_FR fail to control lupus-like disease manifestations in mice, supporting the proposition that T_FR play a role in maintaining self-tolerance (7).

In this study, we sought to identify the involvement of T_FR in human disease by using multiple sclerosis (MS) as a model for an autoimmune disease (AID) with Treg disturbances (10) and elevated autoantibody levels (11, 12). Peripheral blood (PB) is the most accessible source to analyze immune responses in healthy control subjects (HCs) and patients with AID. Therefore, we first determined whether blood T_FR are a good representation to investigate T_FR responses ongoing in secondary lymph nodes. To do so, a pairwise comparison was made between T_FR derived from blood and tonsils of HCs. Next, circulating T_FR were monitored after influenza vaccination to define GC-induced changes in this subset. Finally, circulating T_FR alterations were investigated in MS patients and compared with HCs to evaluate their involvement in AID.

Ethics approvals were obtained from each institute’s human ethics committee. Tonsils and blood were obtained from adult patients without AIDs who were undergoing routine tonsillectomies at Hospital East-Limburg (Genk, Belgium; for detailed information, see Supplemental Table I). Adult healthy volunteers were recruited for the vaccination of inactivated influenza vaccine Influvac S 2013/2014 (ABBOTT BIOLOGICALS B.V., Brussels, Belgium; for detailed information, see Supplemental Table I). MS patients were recruited from the Rehabilitation and MS-Center (Overpelt, Belgium). Detailed clinical characteristics are listed in Supplemental Table I.

PB was collected in heparin-coated tubes (Venosafe plastic tubes; Terumo Europe N.V., Leuven, Belgium). Tonsils were cut into small pieces and singles cells were obtained using a cell strainer (EASYstrainer 70 μM; Greiner Bio-One BVBA/SPRL, Wemmel, Belgium). After collection of the plasma, density centrifugation was used to isolate the PBMCs (Lympholyte; Cedarlane Laboratories, SanBio B.V., Uden, the Netherlands). In line with recent publications, specific flow-cytometric markers were used to identify circulating T_FR and T_FH in human blood (13, 14). For flow-cytometric analysis of the T cell subsets, the following Abs were used: CD4 allophycocyanin, CD4 PE-CF594, CD4 FITC, CD45RO PE-CF594, CXCR5 Alexa Fluor 488, CXCR3 PE-CF594, CCR6-PerCP-Cy5.5, BCL-6 PE-CF594, Foxp3 PE-CF594 (all from BD Biosciences, Erembodegem, Belgium), CD25 PerCP-Cy5.5, CD127 PE, PD-1 PE-Cy7, CD45RA allophycocyanin-H7, CD31 allophycocyanin, CCR7 PE, CD62L allophycocyanin-Cy7, SAP-PE, Helios allophycocyanin (all from eBioscience, San Diego, CA); and CD25 allophycocyanin-Cy7 and CD27 allophycocyanin (from Biolegend, ImTec Diagnostics N.V., Antwerp, Belgium). B cell analysis was performed with CD19 PerCP-Cy5.5, IgD allophycocyanin-Cy7, and CD27 PE-Cy7 (all from BD Biosciences). Appropriate isotype controls were used to establish the proper gating strategies (all from BD Biosciences). All flow-cytometric analyzes were performed on a FACSAriaII flow cytometer and analyzed with FACSDiva software (BD Biosciences). For Foxp3 intranuclear staining, the eBioscience kit was used; for other intracellular stainings, the BD Cytofix/Cytoperm kit (BD) was used according to the manufacturers’ guidelines.

CD4⁺ T cells were purified using CD4 negative selection (STEMCELL Technologies SARL, Grenoble, France). CD25 positive selection (STEMCELL Technologies SARL) was used to obtain a CD25⁺-enriched population and a CD25⁻ population. CD4⁺CD25⁺CD127⁻ responder T cells (Tresp) and CD4⁺CD25⁺CD127⁻CXCR5⁺PD⁻1⁺ T_FH were sorted from the CD25⁻ population, CD4⁺CD25^hiCD127^lo Tregs and CD4⁺CD25^hiCD127^loCXCR5⁺PD-1⁺ T_FR were sorted from the CD25⁺ population using the following Abs: CD4 FITC (BD), CD25 PerCP-Cy5.5 (eBioscience), CD127 PE (eBioscience), CXCR5 Alexa Fluor 647 (BD), and PD-1 PE-Cy7 (eBioscience) using a FACSAria II (BD). Purity of the isolated cells was confirmed. Flow-cytometric analysis was performed using FACSDiva software (BD Biosciences) and FlowJo V10.

A 96-well round-bottom plate (Nunc, Roskilde, Denmark) was coated for 2 h on 37°C with 0.01 μg/ml anti-CD3 (HIT3; BD) and washed with PBS. CD4⁺CD25⁻CD127⁺ Tresp (labeled with 4 μM CFSE; Invitrogen) were cultured at 1 × 10⁴ cells/well with 1 × 10⁵ irradiated autologous PBMCs (feeder cells) in the presence or absence of the same number of Tregs or T_FR in duplicate. The isolated Treg population thus also includes the CXCR5⁺PD-1⁺ T_FR population. Cell cultures were also stimulated with soluble anti-CD28 (BD) for 4 d. The following controls were used: 1) a nonlabeled stimulated control to serve as reference for setting the Treg gate, 2) a labeled nonstimulated control to serve as reference for setting the nonproliferated gate, and 3) a labeled stimulated control with double amount of responder cells to exclude possible nutrient deprivation effects. Cocultures were analyzed on a FACSAriaII on day 4. The suppressive capacity (percentage) of Tregs toward Tresp in coculture was calculated relative to the maximal proliferation of the Tresp alone: [100 − (% proliferation Tresp alone/% proliferation Tresp + Treg)]*100.

CD4⁺CD25⁺ cells (purified as described earlier) were sorted to obtain CD4⁺CD25^hiCD127^lo Tregs and CD4⁺CD25⁻CD127⁺CXCR5⁺PD-1⁺ T_FR, and CD4⁺CD25⁻ cells were sorted to obtain CD4⁺CD25⁺CD127⁻CXCR5⁺PD-1⁺ T_FH cells using FACSAria II (BD). Purity of the obtained cells was confirmed. Purified cells were pelleted and frozen at −80°C. Next, the proportions of cells with a demethylated FOXP3 intron 1 allele were quantified by quantitative PCR on bisulfite-treated genomic DNA, as described previously (15).

Plasma samples from HCs were obtained before and 21 d after vaccination, for Ab titration against the 2013–2014 influenza H1N1pdm vaccine strain A/California/7/2009 (A/California/7/2009 NYMC X-179A, Cat. No. FR-1184; Influenza Reagent Resources, Centers for Disease Control and Prevention) and the 2013–2014 influenza H3N2 vaccine strain A/Texas/50/2012 (A/Texas/50/2012 X-223, Cat. No. FR-1185; Influenza Reagent Resources, Centers for Disease Control and Prevention), using a hemagglutination inhibition (HAI) assay. In brief, 1:2 serial dilutions of inactivated human plasma samples were preincubated with a standardized amount of virus before the addition of 1% RBC (glutaraldehyde-stabilized freshly prepared guinea pig RBCs, Cat. No. 88R-P001; Bio-connect). After incubation, HAI titers are recorded. Controls samples were included in all analyses. Samples were tested in duplicate, and assays were independently repeated. The titer analyzed was the geometric mean of these test results. Patients who showed nonspecific agglutination of RBCs were excluded.

Statistical analyses were done using SAS 9.3, SAS Jump, and GraphPad Prism 6. Graphics were made using GraphPad Prism 6. Data sets were checked for normality, and for effect of age and sex. Analyses of the vaccination study was done using a mixed model (multiple measurements SAS 9.3). Analysis of multiple groups was done using ANOVA, nonparametric testing (Kruskal–Wallis), or linear correlations using SAS Jump. A Mann–Whitney U test was used for nonparametric unpaired data. Wilcoxon matched-pair test was used for nonparametric paired data. Tests were considered significant when p < 0.05 (two-sided tests).

All human blood samples and tonsils were obtained with ethical approval of each institute’s human ethics committee, the Medical Ethical Commission of University Hospital Leuven and Hospital East-Limburg, respectively. Written, informed consent was obtained from all study subjects.

The presence and origin of genuine follicular subsets in the circulation remains controversial (2, 16). This led us to question whether circulating T_FR are phenotypically bona-fide T_FR or rather represent a distinct population.

First, circulating T_FR were phenotypically characterized in detail comparing blood and tonsils of HC (for detailed information, see Supplemental Table I). Distinct subpopulations of Treg (CD4⁺CD25⁺CD127⁻) and conventional T cells (CD4⁺CD25⁻CD127⁺) were found to express the follicular markers CXCR5 and PD-1 in blood and in tonsils (Fig. 1A). Based on that, CD4⁺CD25⁺CD127⁻CXCR5⁺PD-1⁺ and CD4⁺CD25⁻CD127⁺CXCR5⁺PD-1⁺ were defined as (circulating) counterparts of, respectively, T_FR and T_FH. In tonsils, T_FR and T_FH comprise a much larger population compared with the blood (Fig. 1A, 1B).

We next analyzed whether circulating follicular cells are phenotypically similar to those derived from tonsils using both follicular (ICOS, SAP, and BCL-6) and regulatory (Foxp3, Helios, and CD31) markers (Fig. 1C–E). Circulating T_FR and T_FH did not express ICOS, whereas tonsil-derived cell subsets did. SAP, essential for T–B cell interaction, was significantly less expressed on circulating compared with tonsil-derived T_FH, whereas no significant differences were found for T_FR. A significantly higher proportion of T_FR from tonsil express BCL-6 compared with circulating T_FR, whereas T_FH did not significantly differ in percentage of BCL-6⁺ cells. In contrast, the mean fluorescent intensity (MFI) of BCL-6 is significantly decreased on both blood T_FR and T_FH compared with their tonsil-derived counterparts. In addition, the expression levels of essential follicular markers (CXCR5, PD-1, BCL-6, and ICOS) are shown in Fig. 1D, again highlighting the phenotypical difference in follicular expression on blood T_FH and T_FR. The regulatory markers Foxp3 and Helios were equally expressed by T_FR in tonsils and blood. No expression of regulatory markers was seen in T_FH from any source. CD31 is a key molecule for the regulation of T cell homeostasis, effector function, and trafficking (17–19). We found an increased expression of CD31 on the surface of circulating follicular cells, whereas the tonsil-derived counterparts did not express this marker. Taken together, these data show that human circulating T_FR are not bona-fide T_FR because they lack high expression of CXCR5, PD-1, BCL-6, and ICOS, all essential follicular marker proteins.

Lastly, we determined the differentiation stage and effector phenotype of human circulating T_FR. As a negative and positive control, we compared these cells with naive T cells and memory T cells, respectively, based on their CD45RO expression. Tonsil-derived T_FR and T_FH cells have a CD45RO⁺ and CD45RO⁻ cell population unlike the circulating follicular cell subsets, which are all CD45RO⁺ indicating a memory phenotype (Fig. 2A). To further characterize this memory phenotype, we used CCR7 and CD62L to distinguish effector memory T cells (T_EM) (Fig. 2B) from central memory T cells (T_CM) (Fig. 2C). Although all tonsil-derived follicular T cells have a T_EM phenotype (CCR7⁻CD62L⁻), only half of the circulating T_FR have a T_EM and a minority is T_CM. We next examined the effector phenotype based on the expression of chemokine receptors CXCR3 and CCR6. On T_FH (CD4⁺CXCR5⁺), CXCR3 expression is reported to represent a Th1 phenotype, whereas CCR6 indicates a Th17 phenotype (20). Combining both markers gives a more elaborate view on the effector phenotype (Th2; CXCR3⁻CCR6⁻, Th17; CXCR3⁻CCR6⁺, and Th1; CXCR3⁺CCR6⁻). We found that circulating T_FR have a significantly higher percentage of CXCR3⁺CCR6⁻ cells (Th1-like phenotype) compared with tonsil-derived T_FR (p = 0.03; Fig. 2D). A trend toward an increase in the percentage of CXCR3⁻CCR6⁺ cells (Th17-like phenotype) was found (Fig. 2E), whereas the percentage of CXCR3⁻CCR6⁻ cells (Th2-like phenotype) was significantly decreased (p = 0.0079; Fig. 2F). An overview of the gating strategy of the flow-cytometric markers can be found in Supplemental Fig. 1.

Together, these data suggest that human circulating T_FR are phenotypically distinct from their counterparts in the secondary lymphoid organs (tonsils were used as a model) because they express lower levels of follicular markers, are a memory population, and have a proinflammatory effector phenotype.

To investigate whether follicular cell activity is measurable in the blood, we next assessed the effect of vaccination on circulating follicular and B cell subsets. PB from HCs (n = 24; for detailed information, see Supplemental Table I) was taken before (D0), and 1 d (D1), 1 wk (D7), and 3 wk (D21) after influenza vaccination. The percentage of circulating T_FR cells significantly increased after 24 h (p < 0.001) and 7 d (p < 0.001) after vaccination, and returned back to baseline after 3 wk (Fig. 3A). Memory T_FR (CD45RO⁺CD45RA⁻ T_FR) significantly increased (D1 and D7, p < 0.0001) after vaccination (Fig. 3B), whereas the percentage of naive T_FR (CD45RO⁻CD45RA⁺) decreased significantly after 1 and 7 d (p = 0.01, data not shown). In line with other groups (21, 22), we confirmed a significant increase in circulating T_FH after vaccination (all time points, p < 0.001; Fig. 3C). Similar to memory T_FR, memory T_FH increased significantly after influenza vaccination (D1 and D7: p < 0.001, D21: p < 0.05; Fig. 3D). CXCR5 and PD-1 levels of circulating T_FR and T_FH decreased after vaccination (D1 and D7, p < 0.001) and returned to baseline after 3 wk (Fig. 3E–H).

Next to circulating T cells, we also evaluated various B cells subtypes after vaccination. Similar to a previous report (21), we show that percentage of plasmablasts (CD19⁺CD27 ⁺CD138⁺) increased after vaccination, although it did not reach statistical significance (D7, p = 0.07; Fig. 3I). Although no difference in the percentage of class-switched B cells (CD19⁺IgD⁻CD27⁺) was found (data not shown), the ratio of non–class-switched B cells (CD19⁺IgD⁺CD27⁺) over class-switched B cells was significantly increased after 21 d compared with D1 (p < 0.001) and D7 (p = 0.03) (Fig. 3J).

To investigate whether the changes in the circulating T_FR compartment are linked with the production of protective Ab responses, we performed HAI assays on plasma samples collected at baseline and D21 using two vaccine virus strains (A/California/7/2009 [H₁N_1pdm] and A/Texas/50/2012 [H₃N₂]). For both strains, the percentage of circulating T_FR at D7 was positively correlated with postvaccination (D21) geometric mean titers (GMTs; p = 0.05 and p = 0.03; Fig. 3K, 3L). Moreover, the percentage of plasmablasts significantly correlated with postvaccination H1N1pdm GMTs titers (p = 0.02; Fig. 3M), but not with postvaccination H3N2 GMTs (Fig. 3N).

In summary, we show that seasonal influenza vaccination in HCs leads to a significant increase in circulating (memory) T_FR and T_FH, which significantly correlate with titers of anti-vaccine Abs. We therefore conclude that circulating T_FR are a relevant source to measure T_FR activity and regulation of Ab responses in response to GC reactions.

To assess the functionality of circulating T_FR, we made a pairwise comparison of the suppressive capacity of sorted blood- and tonsil-derived T_FR and total Tregs of HCs (n = 7, Supplemental Table I) using an in vitro coculture suppression assay. In brief, CFSE-labeled Tresp were cocultured in a 1:1 ratio with T_FR, total Tregs, or no Tregs, and this in the absence or presence of CD3/CD28 stimulation for 4 d (23–25). Fig. 4A shows a representative example with the various conditions that were incorporated in the assay. Blood-derived T_FR and Tregs were equally suppressive as their tonsil-derived counterparts, indicative of an equal functional activity (Fig. 4B). In addition, the methylation status of the FOXP3 gene was assessed as previously described (15). No significant difference in demethylated FOXP3 was found between T_FR from blood compared with tonsils, indicating equal differentiation into Treg lineage (Fig. 4C). Sorted total Tregs and T_FH were used as a positive and negative control, respectively. Together, these results show that circulating T_FR are equally suppressive and have the same level of demethylated FOXP3 as T_FR of secondary lymphoid organs, governing them as a good alternative for functional analyses in the context of (auto)immunity.

Various groups, including ours, have reported an impairment in the Treg compartment of MS patients (10, 26–28). Moreover, elevated levels of autoantibodies have been reported in MS (29, 30). In light of these findings, we investigated whether the T_FR compartment known to regulate humoral immunity is disturbed in MS. We show a decreased frequency of circulating T_FR in MS patients (n = 172; for more information, see Supplemental Table I) compared with HCs (n = 107; for more information, see Supplemental Table I; p < 0.0001; Fig. 5A). Moreover, the T_FH/T_FR ratio was significantly increased in blood of MS patients (p < 0.0001; Fig. 5B), indicating a relative imbalance between these interacting T cell populations. Age and sex had no effect on the frequency of circulating T_FR in MS patients. We also looked into the effect of treatment but did not observe any effect (Supplemental Fig. 2A). Moreover, in a 1-y follow-up study of patients treated with fingolimod, we did not detect an effect of treatment on percentage of blood T_FR (Supplemental Fig. 2B). In addition, no significant association could be found between T_FR frequencies and Expanded Disability Status Scale, MS disease type, and disease duration (Fig. 5C–E).

Next, we assessed the T_FR effector phenotypes in MS. T_FR with a Th17-like phenotype are increased in MS patients (n = 14 for MS patients and n = 16 for HC, p = 0.0033; Fig. 5F; for more information, see Supplemental Table I), whereas other effector phenotypes did not differ. In contrast with T_FR, no alterations were found in the T_FH compartment of MS (Supplemental Fig. 3).

Taken together, our results show that the frequency of circulating T_FR is significantly decreased in patients with MS, and that the Th17 effector subpopulation of MS-derived T_FR is increased.

The functionality of MS-derived circulating T_FR was tested using the in vitro coculture suppression assay described earlier (Fig. 6A). MS-derived T_FR displayed a strongly impaired suppression compared with HCs (HC: n = 15, MS: n = 12, p ≤ 0.0001; for more information, see Supplemental Table I; Fig. 6B). We further confirmed that conventional Tregs from these MS patients were also significantly impaired (p < 0.0001; Fig. 6B). When correlating the suppressive function of both Tregs and T_FR from the same donor, a significant positive association was found for both MS patients and HCs (Fig. 6C). For one MS patient (relapsing-remitting MS, no treatment), the suppressive function was assessed at two time points 1 mo apart. The suppressive capacity of Tregs and T_FR (at a 1:1 ratio) at both time points was in the same range (data not shown), confirming the reproducibility and stability of the assay, as well as the constant nature of Treg/T_FR function.

In conclusion, we found that both Tregs and T_FR isolated from MS patients had a reduced capacity to suppress the proliferation of Tresp showing a functional impairment of circulating T_FR in MS patients.

In this study, we provide an insight into the phenotype and function of circulating T_FR in humans. Although circulating T_FR represent a population that is phenotypically distinct from their tonsil-derived counterparts, they are an eligible source to measure GC responses ongoing in secondary lymphoid organs. We showed a decreased frequency and impaired functionality of circulating T_FR in MS patients, indicating their involvement in breakdown of self-tolerance in human AIDs. These conclusions led us to two main questions: What is the origin and fate of these cells? And what explains the impairment of the circulating T_FR population in MS?

Circulating T_FR are not considered bona-fide T_FR because they lack high expression of essential follicular markers, such as BCL-6 and ICOS. These results are consistent with findings in circulating T_FH and likely indicate that follicular markers upregulate only after homing to the GC (16). This notion is further supported by the decreased expression of CXCR5 and PD-1 on circulating T_FR after influenza vaccination, which could reflect homing of CXCR5^hiPD-1^hi T_FR to the GC. In mice, T_FR were shown to originate from thymic-derived Tregs (4). In line with these findings, we showed that human blood T_FR express Helios and demethylated FOXP3 to a similar extent as tonsil-derived T_FR and are fully functional. Furthermore, we characterized these cells as central memory with a higher expression of both CCR7 and CD62L compared with tonsil-derived T_FR, allowing recirculation to the lymph nodes. He at al. (16) showed that circulating T_FH also have a higher expression of these markers. In addition, we found a significant increase in the percentage of memory T_FR after seasonal influenza vaccination. It is therefore possible that they originate from GC T_FR, migrate to the circulation after a GC response, and become a central effector memory population that is long-lived and has the capacity to recirculate. Another theory proposed by He at al. (16) suggested that circulating T_FH cells are a population of cells that leave the GC response in its early phase, before developing in mature T_FH cells, governing a “precursor memory” population. We showed that although the expression of Foxp3 and Helios was the same in T_FR, blood-derived follicular cells are CD31⁺ whereas the tonsil-derived follicular cells are negative. It has been reported that CD31 expression regulates T cell activation and can thus prevent hyperactivation (17, 19). In addition, recent evidence showed that the loss of CD31 after activation leads to a stable interaction of T cells with B cells (18). This could indicate that circulating T_FR are indeed a central memory population that is quiescent by nature and will lose CD31 expression upon activation. In line with this, a recent mouse study (31) reported that circulating T_FR are a long-lived memory population that homes to GC after reactivation. A human study showed that follicular T cell populations do not need an ongoing GC response for their maintenance because treatment with rituximab, known to eliminate GC B cells, had no effect on the follicular T cell compartment (32). Together all these data indicate that circulating T_FR are a distinct effector memory population that persists for a long time and is able to recirculate to the lymph nodes when needed.

MS, an AID of the CNS, was used as a model to investigate the role of circulating T_FR in autoimmunity. A functional impairment of Tregs in MS was shown by various groups (26–28, 33, 34). We confirmed an impairment of Tregs in MS patients and found that T_FR from the same patients are also defective in their capacity to suppress Tresp. Also, a decreased percentage of blood T_FR in MS patients was found compared with HCs. Based on the chemokine markers suggested by Morita et al. (20), we show that the amount of Th17-like T_FR is increased in MS patients. A more proinflammatory phenotype could explain the decreased suppressive function of circulating T_FR. Furthermore, this impairment could originate from a defect in CTLA-4 signaling because CTLA-4 is essential in T_FR function (35). An alternative explanation for the reduced frequency and suppressive activity in the circulating T_FR compartment could be that the most potent T_FR homed to the lymph nodes to suppress the ongoing GC response. Sage et al. (31) showed that circulating T_FR in mice require dendritic cells for their development and cytokine production. Circulating dendritic cells in MS were shown to have a decreased regulatory character and could in this way contribute to an impairment of functional circulating T_FR (36).

In conclusion, we believe that follicular populations in the blood form a source of responsive memory cells that quickly reacts when encountering an Ag again. Impairment of circulating T_FR and T_FH could contribute to the pathogenesis of various AIDs, including MS, highlighting their importance in conserving normal immunity.

We thank Igna Rutten, Kim Ulenaers, Anne Bogaers, Dr. Van Woensel (Hasselt University and Ubilim), and Anita Knevels (MS research coordination, MS rehabilitation center) for helping with sample collection. We also thank the medical staff from Hospital East-Limburg for the collection of blood and tonsil samples.

The authors have no financial conflicts of interest.