Research ArticleAUTOIMMUNITY

Human blood Tfr cells are indicators of ongoing humoral activity not fully licensed with suppressive function

+ See all authors and affiliations

Science Immunology  11 Aug 2017:
Vol. 2, Issue 14, eaan1487
DOI: 10.1126/sciimmunol.aan1487

Suppressing Sjögren syndrome

T follicular regulatory (Tfr) cells regulate antibody production in the germinal center, yet individuals with the autoimmune disease Sjögren syndrome have increased numbers of circulating Tfr cells compared with healthy individuals. Fonseca et al. compared blood Tfr cells with tissue Tfr cells and found that blood Tfr cells were phenotypically distinct from their tissue counterparts. Moreover, blood Tfr cells did not preferentially suppress humoral responses and had a naïve-like phenotype. These cells were not thymically derived but were generated during germinal center responses, exiting the tissue to enter the blood. These data explain why increased number of blood Tfr cells does not correlate with increased suppression potential and suggest that, instead, increased numbers of blood Tfr cells indicate ongoing humoral activity.

Abstract

Germinal center (GC) responses are controlled by T follicular helper (Tfh) and T follicular regulatory (Tfr) cells and are crucial for the generation of high-affinity antibodies. Although the biology of human circulating and tissue Tfh cells has been established, the relationship between blood and tissue Tfr cells defined as CXCR5+Foxp3+ T cells remains elusive. We found that blood Tfr cells are increased in Sjögren syndrome, an autoimmune disease with ongoing GC reactions, especially in patients with high autoantibody titers, as well as in healthy individuals upon influenza vaccination. Although blood Tfr cells correlated with humoral responses, they lack full B cell–suppressive capacity, despite being able to suppress T cell proliferation. Blood Tfr cells have a naïve-like phenotype, although they are absent from human thymus or cord blood. We found that these cells were generated in peripheral lymphoid tissues before T-B interaction, as they are maintained in B cell–deficient patients. Therefore, blood CXCR5+Foxp3+ T cells in human pathology indicate ongoing humoral activity but are not fully competent circulating Tfr cells.

INTRODUCTION

Germinal center (GC) responses are crucial for the generation of high-affinity antibodies during T-dependent immune responses. Within the GC resides a specialized subset of CD4+ T cells—the T follicular helper (Tfh) cells—which are essential for GC development and function (1, 2). It is now clear that Tfh cells play a central role in productive vaccine responses, whereas defects in their formation or function can contribute to immunodeficiency or autoimmunity (3, 4). More recently, the discovery of T follicular regulatory (Tfr) cells, a subset of suppressive regulatory T (Treg) cells that participate in the GC, added an additional layer of complexity in the biology of GC responses (58).

Tfr cells, generally defined by Bcl-6+CXCR5+PD-1+ICOS+Foxp3+, are a distinct subset of thymic Foxp3+ Treg cells present in lymphoid tissues. Like the Tfh cell differentiation pathway, Tfr cell commitment requires both dendritic cell and B cell interactions, as well as CD28, SAP (SLAM-associated protein), ICOS (inducible costimulator), and PD-1 (programmed cell death–1) signaling (6, 9, 10). A tight balance between expression of transcription factors Bcl-6 and Blimp-1 regulates the differentiation of Tfr cells (6). Tfr cells have specialized functions in controlling the magnitude of GC responses and in limiting the outgrowth of non–antigen-specific B cell clones (5, 6). However, the precise mechanisms of Tfr cell suppression remain elusive, although cytotoxic T lymphocyte–associated antigen 4 (CTLA-4) and regulation of metabolic pathways seem to play a key role (1113).

Although Tfh and Tfr cells are characterized by their location in lymphoid tissues, an increasing number of studies have described putative circulating counterparts of these cells in peripheral blood. This is particularly relevant for studying the biology of these cells in humans, because access to secondary lymphoid tissues can be limiting. Human blood CXCR5+ T cells have been established as memory Tfh–like cells based on their ability to recapitulate bona fide Tfh cell functions: Human blood CXCR5+ T cells can promote plasmablast differentiation, activation-induced cytidine deaminase expression, and class switch recombination by naïve B cells. However, they are phenotypically distinct from tissue Tfh cells and do not express the transcriptional repressor Bcl-6 (1416). Furthermore, an immunization leading to GC and antibody responses correlates with an increase in the frequency of circulating ICOS+ Tfh cells, suggesting that they indicate ongoing Tfh cell responses in secondary lymphoid tissues (14, 1619). Human circulating Tfh cells comprise a heterogeneous population concerning their phenotype and the quality of help they provide to B cells (14, 17). In mice, CXCR5+Foxp3+ Tfr-like cells were found in peripheral blood after immunization and were shown to represent a circulating counterpart of tissue Tfr cells (9, 10).

Although CXCR5-expressing Treg cells and GC Foxp3-expressing T cells have been found in humans (7, 20, 21), so far, no studies have addressed the biological importance of these putative circulating Tfr-like cells in humans. Human tonsil CD25+CD69 T cells have been shown to directly suppress B cell responses, but the relationship of these putative Treg cells to Bcl-6+CXCR5+PD-1+ICOS+Foxp3+ Tfr cells is unclear (22, 23). Peripheral blood CXCR5+ Treg cells are being studied as circulating Tfr cells in many different human diseases, despite the fact that the biological relevance of these cells is unclear (2428). Additionally, Foxp3 up-regulation by nonregulatory human T cells and transient CXCR5 expression by T cells undergoing activation challenge the assumption that peripheral blood CXCR5+ Treg cells are bona fide circulating Tfr cells (29, 30).

Here, we found that human blood Tfr cells, defined as CXCR5+Foxp3+ T cells, are generated in peripheral lymphoid tissues as humoral immune responses are established. However, in contrast to tissue Tfr cells and conventional CXCR5 Treg cells, circulating Tfr cells have a naïve-like phenotype. Our data demonstrate that blood Tfr cells are generated following the initial steps that lead to GC responses being distinct from tissue Tfr cells.

RESULTS

Blood Tfr cells indicate ongoing GC responses

To address the impact of Tfr/Tfh ratio in human autoimmunity, we studied Sjögren syndrome (SS), a systemic autoimmune disease characterized by the lymphocytic infiltration of salivary and lachrymal glands, with the formation of ectopic lymphoid structures demonstrating the pathogenic involvement of T-B cell interactions (31, 32). We studied a cohort of 25 patients with recently diagnosed SS according to American-European Consensus Group (AECG) criteria (33) under no immunosuppressive treatment other than prednisolone (less than 7.5 mg/day or equivalent) or hydroxychloroquine (table S1). Unexpectedly, we found a notable increased frequency of circulating Tfr cells in SS as compared with age-matched healthy donors (HDs) (Fig. 1A and fig. S1, A and B). Whereas SS patients showed fewer Treg and Tfh cells per milliliter of blood, in agreement with their lymphopenic state (fig. S1C), the absolute number of CXCR5+ Tfr cells was not different when compared with HDs (fig. S1B), in line with the increased frequency of blood CXCR5+ Tfr cells. The increased frequency of the CXCR5+ Treg cell subset was specifically increased, providing an explanation for the high Treg cell frequency observed in SS patients (Fig. 1A). SS patients showed a substantial increase in the Tfr/Tfh ratio compared with HDs (Fig. 1B) (8, 10). Furthermore, among SS patients, the increased Tfr/Tfh ratio is associated with patients with serum autoantibodies (Fig. 1C). On the contrary, we found no correlation between high Tfr/Tfh ratio with C-reactive protein and disease activity score [EULAR Sjögren’s syndrome disease activity index (ESSDAI)] (Fig. 1, D and E).

Fig. 1 Blood Tfr cells are indicators of ongoing humoral activity.

(A) Frequency of total Treg cells, Tfr cells, and CXCR5 Treg cells in peripheral blood of SS patients and age-matched HDs (n = 25; unpaired Student’s t test with Welch’s correction for variance). Representative plots (left) and pooled data (right) are shown. (B) Blood Tfr/Tfh ratio in SS patients and HDs (n = 25; unpaired Student’s t test with Welch’s correction for variance). (C) Blood Tfr/Tfh ratio in SS patients with and without serum autoantibodies (AAb) (anti-SSA/Ro52, anti-SSB/Ro60, and anti-SSB/La) (n = 25; unpaired Student’s t test). (D) Variation of blood Tfr/Tfh ratio according to C-reactive protein (CRP). Analysis by linear regression was used. (E) Variation of blood Tfr/Tfh ratio according to disease severity score (ESSDAI). Analysis by linear regression was used. Bars represent SEM (A and C) or minimum and maximum values (in box and whisker graphs) (C).

Blood and tissue Tfr cells present different follicular and regulatory markers

To test whether CXCR5+Foxp3+ Tfr cells in human peripheral blood are circulating counterparts of tissue Tfr cells, we studied peripheral blood from a cohort of 42 healthy volunteers between 22 and 92 years (mean age, 46.76 ± 18.14 years; 30 females and 12 males). We found that CXCR5 was expressed by 18.57 ± 6.55% of total Treg cells (defined as CD4+CD25+Foxp3+ T cells) (Fig. 2A). The frequency and number of CXCR5+Foxp3+ T cells did not change with aging (Fig. 2B and fig. S2A).

Fig. 2 Blood Tfr cells show expression of follicular and regulatory markers.

(A) CXCR5+ Tfr cells constitute 18.57 ± 6.55% of Treg cells (left) and 0.93 ± 0.56% of total CD4+ T cells (middle), representing 9985 ± 9043 cells per milliliter of blood (right) (n = 42; adult HDs). (B) Variation of blood Tfr cell frequency (left) and absolute number per milliliters of blood (right) according to age (age range, 22 to 92 years) (n = 42; linear regression). (C) Expression of Foxp3, CD25, CD69, CTLA-4, CXCR5, ICOS, PD-1, Bcl-6, and CD57 by Tfh cells (blue), CXCR5 Treg cells (black), and Tfr cells (red) in children blood (top) and tonsils (bottom). Naïve CD4+ T cells were used as control (gray). Representative plots from six healthy children are shown. CXCR5+ subsets in tonsils were defined as CXCR5+ICOS+ cells (fig. S2B). Tconv, conventional T cells. (D) Immunofluorescence microscopy of formalin-fixed, paraffin-embedded human tonsils stained for DAPI (blue), CXCR5 (yellow), CD4 (red), and Foxp3 (green). Top, middle, and bottom outlined areas indicate top, middle, and bottom enlarged areas on the right, respectively. Data are representative of tonsil sections from five healthy children. (E) Blood Tfr/Tfh ratio in adult blood, children blood (tonsil donors), and tissues (tonsils). Black and red dots represent blood and tonsil results, respectively (n = 42 for adults and n = 6 for children; Student’s t test). Error bars represent SEM.

Because CXCR5 is used to identify human circulating Tfh cells, we compared the phenotype of circulating CXCR5+Foxp3+ Tfr cells with that of circulating Tfh cells and CXCR5 conventional Treg cells. Peripheral blood CXCR5+Foxp3+ T cells share characteristics with both circulating Tfh cells and CXCR5 Treg cells (Fig. 2C). Taking advantage of routine tonsillectomies performed because of tonsil hypertrophy in otherwise healthy children, we compared the cell phenotype of paired blood and tissue samples from the same child (Fig. 2C and fig. S2B). We found that circulating Tfh cells were phenotypically distinct from their tissue counterparts, in line with previous reports, especially regarding their PD-1, ICOS, and Bcl-6 expression (Fig. 2C) (14, 15). In a similar way, circulating CXCR5+Foxp3+ T cells were also ICOSPD-1Bcl-6CD57 and consequently distinct from tonsil Tfr cells (Fig. 2C). In addition, we confirmed that these cell populations displayed a similar phenotype in adults (fig. S2, C and D). Our results are consistent with murine studies showing that blood and tissue Tfr cells are phenotypically distinct (9). ICOS was not differentially expressed by Treg and Tfr cells in tonsils (Fig. 2C). Bcl-6 expression was not detected in any population by real-time polymerase chain reaction (PCR) (fig. S1F), consistent with previous reports showing that blood Tfh cells do not express Bcl-6 (1518, 34). We also confirmed that tissue CXCR5+Foxp3+ T cells are localized within GCs, therefore corresponding to Tfr cells (Fig. 2D and fig. S2G). Curiously, we observed different Tfr/Tfh ratios in the blood and tonsils (Fig. 2E).

CXCR5+Foxp3+ Tfr cells are a distinct subset of suppressive Foxp3+ T cells

It has been described that CXCR5 expression can transiently occur upon human T cell activation (30, 35, 36). Moreover, human T cells can also transiently express Foxp3 upon in vitro T cell receptor (TCR) stimulation in a transforming growth factor–β (TGF-β)–dependent manner (29, 37).

To address whether ex vivo CXCR5+Foxp3+ Tfr cells were bona fide regulatory cells, we sorted that cell population, as well as CXCR5 conventional Treg cells (fig. S3, A and B), and cultured them with CellTrace Violet (CTV)–labeled conventional T cells. Proliferation of responder cells was analyzed after 5 days of soluble αCD3 stimulation (Fig. 3A). Blood CXCR5+Foxp3+ Tfr cells reduced conventional T cell proliferation (Fig. 3, B and C), definitely demonstrating their regulatory function.

Fig. 3 Blood Tfr cells are a distinct subset of suppressive Treg cells.

(A) Schematic representation of in vitro suppression assay. FACS-sorted 25 × 103 CXCR5CD25CD127+CD4+ Tconv cells were cocultured with 25 × 103 CXCR5CD25+CD127CD4+ Treg cells or CXCR5+CD25+CD127CD4+ Tfr cells under stimulation by anti-CD3 (1 μg/ml) in the presence of 105 irradiated (25 Gy) allo-PBMCs. After 5 days, responder cells were analyzed for CTV dilution by flow cytometry. Sorting strategy is described in fig. S2 (A and B). (B) Proliferation of Tconv cells without Treg cells or in the presence of either CXCR5 Treg or Tfr cells. Representative plots (left) and pooled data (right) (n = 3, each with technical triplicates; one-way ANOVA with posttest Turkey’s multiple comparisons) are shown. ns, not significant. (C) Suppression curve of CXCR5 Treg and Tfr cells in different ratios using the same conditions described in (A) and (B) (n = 1, with technical triplicates; two-way ANOVA). (D) Stability of Foxp3 expression by sorted CXCR5 Treg cells and CXCR5+ Tfr cells after 5 days of in vitro culture under αCD3/αCD28 (1 μl per well) stimulation. Percentage (left) and cell number (right) (n = 5, each with technical triplicates; Student’s t test) are shown. (E) Relative expression of Foxp3 and CXCR5 by sorted Tconv cells, Tfh cells, CXCR5 Treg cells, and Tfr cells from blood by real-time reverse transcription PCR. Gene expression was normalized to housekeeping genes (B2M, G6PD, and ACTB) (n = 2, each with technical duplicates; Student’s t test). (F) Expression of Foxp3, CD25, CTLA-4, and CXCR5 by sorted CXCR5 Treg and Tfr cells at baseline [day 0 (d0)] and after 5 days of in vitro culture under αCD3/αCD28 (1 μl per well) stimulation. Data are representative histograms of three independent experiments, each one with technical triplicates. Error bars indicate SEM.

Stability of Foxp3 expression is required for the suppressive function of Treg cells (38). To determine whether blood Tfr cells have stable Foxp3 expression, we stimulated sorted Tfr cells and CXCR5 Treg cells with anti-CD3/CD28 microbeads for 5 days in the absence of exogenous interleukin-2 (IL-2). In the absence of IL-2, Treg cells do not survive well in culture. Under these conditions, both CXCR5 Treg cells and Tfr cells retain a similar frequency of Foxp3+ cells, albeit lower than in the beginning of the culture (Fig. 3D). The frequency of recovered live Foxp3-expressing cells was slightly higher for sorted Tfr cells as compared with CXCR5 conventional Treg cells. Next, we analyzed the relative expression of Foxp3 and CXCR5 in sorted conventional T cells, Tfh cells, Tfr cells, and CXCR5 Treg cells from human blood by real-time PCR. Although Foxp3 protein expression was lower in Tfr cells than in CXCR5 Treg cells (Fig. 2C and fig. S2D), Foxp3 gene expression was similar between the two subsets (Fig. 3E). In addition, circulating Tfh and Tfr cells also showed comparable CXCR5 gene expression (Fig. 3E).

To investigate whether activation of blood Tfr cells triggers up-regulation of Foxp3, CD25, and CTLA-4, a phenomenon known to be associated with increased Treg cell–suppressive function (38), we analyzed the phenotype of sorted CXCR5+ and CXCR5 Treg cells after 5 days of culture in the presence of αCD3/CD28 microbeads. We found an up-regulation of Foxp3 and CD25 by Tfr cells, whereas CTLA-4 was increased in both populations (Fig. 3F and fig. S3C). The levels of expression of these markers by blood Tfr cells after activation resembled those from tissue Tfr cells (compare with Fig. 2C). CXCR5 up-regulation was not detected in sorted CXCR5 Treg cells, showing that CXCR5+ Tfr cells are a distinct subset of human blood Treg cells.

Blood Tfr cells do not preferentially suppress humoral responses

To address the function of blood Tfr cells, we first investigated whether this population could directly suppress Tfh cells. Using a similar in vitro assay used to prove the regulatory capacity of blood Tfr cells, but with sorted Tfh cells as responders, we found that blood Tfr cells strongly suppressed Tfh cell proliferation, but without a specific advantage when compared with CXCR5 Treg cells (Fig. 4A).

Fig. 4 Blood Tfr cells do not show specialized humoral regulatory capacity.

(A) Proliferation of CXCR5+CD25CD127+CD4+ Tfh cells without Treg cells or in the presence of either CXCR5 Treg or Tfr cells after 5 days of in vitro culture as described in Fig. 3A. Representative plots (left) and pooled data (right) (n = 3, each with technical triplicates; one-way ANOVA with posttest Turkey’s multiple comparison) are shown. (B) Schematic representation of suppression coculture assay. FACS-sorted 25 × 103 CXCR5+CD25CD127+CD4+ Tfh cells (or CXCR5CD25CD127+CD4+ Tconv cells) were cocultured for 5 days with 25 × 103 CXCR5+CD25+CD127CD4+ Treg cells (or CXCR5CD25+CD127CD4+ Treg cells) under stimulation by SEB (1 μg/ml) and in the presence of 30 × 103 CD27IgD+CD19+ naïve B cells. (C) Up-regulation of CD38 and down-regulation of IgD by naïve B cells (top) and proliferation of Tfh cells by CTV dilution (bottom) without Treg cells or in the presence of either CXCR5 Treg or Tfr cells. Representative plots (left) and pooled data (right) (n = 5, each with technical triplicates; one-way ANOVA with posttest Turkey’s multiple comparisons) are shown. (D) Suppression curve of CXCR5 Treg and Tfr cells in different ratios using the same conditions described in (B) and (C) (n = 1, with technical triplicates; two-way ANOVA). (E) ELISA determination of IgA, IgM, and total IgG in supernatants after 10 days of in vitro coculture performed as described in (D) but using SEB (1 μg/ml) + SEA (10 g/ml) + SEE (10 ng/ml) + TSST-1 (10 ng/ml) as superantigen stimulation (n = 3, each with technical triplicates; one-way ANOVA with posttest Turkey’s multiple comparisons). (F) In vitro migration of 75 × 103 sorted Tconv cells, Tfh cells, CXCR5 Treg cells, and Tfr cells toward a CXCL13 gradient (2 μ/ml) expressed by chemotaxis index (n = 3, each with technical triplicates; one-way ANOVA with posttest Turkey’s multiple comparisons). Error bars indicate SEM. SA, superantigens.

Next, to directly assess the impact of blood Tfr cells on B cell activation, we used in vitro T-B cocultures in the presence of staphylococcal enterotoxin B (SEB) superantigen (Fig. 4C). After 5 days of culture, B cells up-regulated CD38 and down-regulated immunoglobulin D (IgD) only in the presence of Tfh cells (Fig. 4C). Both CXCR5 and CXCR5+ Treg cells impaired the generation of CD38+IgD GC-like B cells. Consistent with our results from suppression assays with Tfh cells (Fig. 4A), Tfh cell proliferation was similarly inhibited by CXCR5 and CXCR5+ Treg cells (Fig. 4, C and D, and fig. S4A). As expected, Tfh cells showed better proliferation responses in coculture with B cells.

To further address the function of blood Tfr cells on humoral responses, we analyzed class switch recombination by naïve B cells 10 days after superantigen stimulation. We found that blood Tfr cells, although able to reduce activation of naïve B cells and proliferation of Tfh cells as shown before, did not limit class switch recombination by B cells because no impact on IgA or IgG production was observed (Fig. 4E). On the contrary, CXCR5 Treg cells efficiently suppressed humoral responses (Fig. 4E).

CXCR5/CXCL13-dependent migration to GC is critical for suppression of humoral responses by Tfr cells (5, 7), and plasma CXCL13 levels have been correlated to ongoing GC responses in humans (39). To prove that blood Tfr cells were capable of migrating toward a CXCL13 gradient, we conducted in vitro chemotaxis assays with sorted populations from human peripheral blood. We found that, although the CXCR5 mean fluorescence intensity (MFI) of peripheral Tfh and Tfr cells was slightly different (Fig. 2C and fig. S2D), both populations shared their ability to migrate toward a CXCL13 gradient, showing functional capacity of blood Tfr cells to enter CXCL13-enriched tissues (Fig. 4F).

Blood Tfr cells have a distinctive naïve-like phenotype

To explain the unexpected observation that blood Tfr cells do not suppress antibody production, we hypothesized that this population could represent thymus-derived precursors of Tfr cells not yet fully committed to regulate humoral responses. We found that blood Tfr cells were predominantly CD45ROFoxp3lo resting Treg cells (Fig. 5A), expressing high levels of CD45RA, CCR7, CD62L, and CD27 and low levels of human leukocyte antigen–DR (HLA-DR), reminiscent of a naïve phenotype (Fig. 5B). Virtually, all blood Tfr cells were quiescent Ki-67 nonproliferating cells when analyzed ex vivo (Fig. 5C). Moreover, circulating Tfr cells were virtually devoid of CD45RO+CCR7 effector memory cells, in notable contrast to CXCR5 Treg cells, a phenotype more similar to circulating Tfh cells (Fig. 5D). Although the vast majority of blood Tfh cells were CD45RO+CCR7+ central memory cells, consistent with previous reports (1417), a substantial proportion of Tfr cells were CD45ROCCR7+ naïve cells (Fig. 5D). Furthermore, the few CD45RO Tfh cells did not express high levels of CD45RA, indicating that those cells were not really naïve, in contrast to Tfr cells (fig. S5A). Therefore, blood Tfr cells constitute a pool of naïve resting cells.

Fig. 5 Blood Tfr cells are immature but are not committed in the thymus.

(A) Backgate of CXCR5 and CXCR5+ Treg cells according to CD45RO and Foxp3 expression. (B) Expression of CD45RO, CD45RA, CCR7, CD62L, HLA-DR, and CD27 by Tfr cells (red) and CXCR5 Treg cells (black) in the blood. (C) Expression of Ki-67 by CXCR5+ Treg cells and CXCR5 Treg cells in the blood (n = 22; Student’s t test). (D) CD45RO+CCR7 effector memory, CD45RO+CCR7+ central memory, and CD45ROCCR7+ naïve subsets of Tfr cells and CXCR5 Treg cells in adult blood. Representative plots (left) and pooled data (right) (n = 22; Student’s t test). Tfh cells are represented in blue, CXCR5 Treg cells in black, and Tfr cells in red. (E) Variation of CD45ROCCR7+ naïve Tfr cell and CXCR5 Treg cell frequency in blood according to age (n = 22; linear regression). (F) Expression of CXCR5 by Foxp3+CD4+ thymocytes (n = 4). (G) Expression of CXCR5 by cord blood Foxp3+CD4+ T cells (n = 3). (H) Expression of CD45RO, CCR7, and ICOS by cord blood Treg cells (n = 3). Bars represent SEM.

To test whether blood Tfr cells were thymus-derived precursors of tissue Tfr cells, we analyzed the frequency of these cells according to age. Contrary to thymus-derived naïve Treg cells, CD45ROCCR7+ naïve Tfr cells did not decrease with increasing age (Fig. 5E). In addition, the expression of CD31, a marker used to identify recent thymic emigrants in human blood (4042), was not specifically enriched in the population (fig. S5, B and C). Although these observations suggest that blood Tfr cells are not a thymic population, this was not conclusive. Therefore, we directly examined CXCR5-expressing T cells in the human thymus and neonatal cord blood. There was not a population of CXCR5+ Treg cells detected in any of those tissues (Fig. 5, F and G, and fig. S5D). Although CXCR5-expressing Treg cells were not found in cord blood, some Foxp3+ cells expressed CD45RO, suggesting that additional activation signals not present before birth are required to shape a CXCR5 phenotype in circulating Treg cells. Consistent with our previous data, ICOS+ Treg cells were detected in cord blood, indicating that ICOS cannot be used as a specific follicular marker in circulating human Treg cells (Fig. 5H).

Blood Tfr cells emerge from lymphoid organs before B cell interaction

Having demonstrated that circulating Tfr cells did not egress from the thymus, we investigated whether Tfr cells recirculate from secondary lymphoid tissues before being fully committed to tissue Tfr cells. We compared both CXCR5+ and CXCR5 Treg cell subsets from children paired blood and tissue (tonsils) concerning their effector memory, central memory, and naïve composition. We found that CD45RO+CCR7 effector Tfr cells were present in lymphoid tissues but not in the blood, suggesting that effector Tfr cells are selectively retained in tissues, similarly to effector Tfh cells (Fig. 6A). Therefore, it is unlikely that blood CD45RO Tfr cells derive from the fully mature tissue Tfr cells that express CD45RO, as the few CD45RA reexpressing end-stage memory CD4+ T cells do not become CD45RO (fig. S6A) (43, 44).

Fig. 6 Blood Tfr cells are lymphoid tissue–derived Tfr precursors.

(A) CD45RO+CCR7 effector memory, CD45RO+CCR7+ central memory, and CD45ROCCR7+ naïve subsets of Tfr cells and CXCR5 Treg cells in children blood (top) and tissues (bottom). Representative plots (left) and pooled data (right) (n = 6; Student’s t test) are shown. Tfh cells are represented in blue, CXCR5 Treg cells in black, and Tfr cells in red. CXCR5+ subsets in tonsils were defined as CXCR5+ICOS+ cells (fig. S2B). (B) Blood Tfh and Tfr cells from X-linked agammaglobulinemia (BTK-deficient) patients compared with sex- and age-matched HDs. Representative plots (left) and pooled data (right) (n = 5; Student’s t test) are shown. (C) Model of CXCR5+ Tfh and Treg cell generation and recirculation in humans upon antigen stimulation. Tfh cells are shown in red, and Tfr cells in blue. DC, dendritic cell. (D) Frequency of peripheral blood Tfr cells on the day of influenza vaccination (d0) and 7 days later in healthy volunteers. Schematic representation and representative plots (left) and pooled data (right) (n = 32; Student’s t test) are shown. Bars represent SEM.

Our data suggest that blood Tfr cells are generated in secondary lymphoid tissue before full differentiation toward mature Tfr cells. It has been known that full differentiation of follicular T cells requires a two-step process, with an initial activation mediated by dendritic cells and a subsequent B cell interaction in the T-B border. We investigated whether blood Tfr cells, given their immature phenotype, could be generated before the B cell interactions required for acquisition of terminal differentiation. To investigate this issue, we analyzed peripheral blood from X-linked agammaglobulinemia [Bruton’s tyrosine kinase (BTK)–deficient] patients, with a complete absence of CD19+ cells. We observed a notable decrease in blood Tfh cells in those patients, in line with previous reports (Fig. 6B) (45). However, frequency of blood Tfr cells was not decreased in B cell–deficient patients (Fig. 6B). These observations are conclusive in establishing that blood Tfr cells enter the circulation before B cell contact, whereas most of the blood Tfh cells require B cell interactions. To investigate whether CD45RO+ and CD45RO blood Tfr cells could discriminate between Tfr cells recirculating before and after B cell interaction, we analyzed these two populations in peripheral blood of patients with B cell deficiency, as well as in SS patients. We found no differences in CD45RO+ or CD45RO Tfr cells in these two diseases (fig. S5E), although CD45RO up-regulation occurs irrespective of B cell interaction on Tfr cells.

We therefore hypothesized that blood Tfr cells are generated in secondary lymphoid tissue and enter the circulation before full differentiation toward tissue Tfr cells (Fig. 6C). To test our model in vivo, we analyzed samples from healthy adults undergoing influenza vaccination. Previous studies have shown a positive correlation between circulating Tfh cell subsets and antibody responses after influenza vaccination in healthy adults (15, 18, 19). We therefore analyzed the impact of vaccination in circulating Tfr cells. Consistent with our hypothesis, we found that circulating Tfr cells increased on day 7 after influenza vaccination (Fig. 6D). This observation is in line with our prediction that, during ongoing GC responses, Tfr cells are generated and some exit from the tissue to the peripheral blood.

DISCUSSION

Our comprehensive evaluation of human Tfr supports a model in which blood Tfr cells are generated following the initial steps that lead to GC responses in secondary lymphoid tissues, exiting the tissue before interactions with B cells that are required for complete differentiation toward tissue-resident Tfr cells. Although some studies have quantified blood CXCR5+ Treg cells as circulating Tfr cells in different diseases, the human biology of CXCR5+ Treg cells remains elusive (24, 25, 27, 46, 47). Moreover, most of the literature studies define blood Tfh cells as cells that contain both Tfh and CXCR5+ Tfr cells, whereas many other studies identify Treg cells as a mixture of bona fide conventional Treg cells and CXCR5+ Tfr cells. Hence, results may be confounded by combining effector and regulatory cell populations. As an example, our cohort of SS patients shows an increase in the frequency of Foxp3+ Treg cells compared with the control population. However, only CXCR5+ Tfr cells, and not conventional CXCR5 Treg cells, are increased in those patients. As a consequence, the apparent increase of Treg cells in the blood of SS patients is an increase of CXCR5+ Tfr cells that reflect the ongoing humoral activity. It was the search for an explanation for this apparent counterintuitive observation that led us to establish the ontogeny and function of human circulating Tfr cells.

We found that Tfr cells in tonsils have follicular and regulatory markers and were found within GCs, whereas blood Tfr cells do not express ICOS, PD-1, or Bcl-6, apparently diverging these cells from follicular imprinting. Previous studies have described low ICOS and PD-1 expression and no Bcl-6 expression in human blood Tfh cells (14). In mice, blood Tfr cells have also lower expression of ICOS (9). It was also reported that murine circulating Tfr cells can bypass the B cell zone and do not gain full activation as part of a memory programmed state (9). In line with these studies, the absence of ICOS, PD-1, and Bcl-6 from human blood CXCR5+ Tfr cells does not exclude their follicular ontogeny.

Our results show key differences between mice and humans regarding the function of blood CXCR5+ Tfr cells: Although murine blood Tfr cells appear to be specialized in suppressing antibody production (despite their lower suppressive capacity when compared with tissue Tfr cells) (9, 10, 12, 13), human blood Tfr cells do not have the ability to fully suppress humoral responses.

Nevertheless, we found that blood Tfr cells specifically migrated toward CXCL13 gradient, suggesting that these cells have the capacity to reach the follicles. CXCR5 conventional Treg cells did not up-regulate CXCR5 upon in vitro activation, further confirming CXCR5-expressing Tfr cells as a distinctive subset.

We also found that blood Tfr cells have a prominent naïve phenotype. However, they are absent from the thymus and cord blood (where activated Treg cells can already be found). These observations provide compelling evidence that activation signals generated in peripheral lymphoid organs are required to shape a CXCR5+ phenotype on human Foxp3+ T cells. Conversely, tissue Tfr cells are almost all CD45RO+ antigen-experienced effector cells. Together, these observations led us to hypothesize that blood Tfr cells leave lymphoid tissues as immature cells before B cell interaction in T-B border and full differentiation into Tfr cells. This view was supported by the presence of blood Tfr cells in peripheral blood of patients lacking B cells due to genetic defects. This finding provides an explanation for the incomplete suppressive function of blood Tfr cells.

An important limitation of our study is the difficulty to isolate tissue Tfr cells for functional assays because CD25 and CD127 are not reliable to identify tonsil Foxp3-expressing cells. However, the phenotype between blood and tissue Tfr cells is remarkably different, in particular, with respect to maturation markers.

Our results from vaccination and SS patient cohorts show that blood Tfr cells are indicative of ongoing humoral activity. In SS patients, where ongoing GC reactions promote the production of autoantibodies (31, 32), blood Tfr cells were substantially increased (directly contributing to an increased Tfr/Tfh ratio). Although we expected to find a decrease in this putative humoral suppressive cell population in autoimmune conditions, our results suggest that blood Tfr cells indicate ongoing humoral activity and are not a measurement of suppressive potential. This is in line with recently published reports showing an increase in blood CXCR5+ Treg cells in other autoimmune conditions and infectious diseases (27, 28, 47). Therefore, studies regarding blood CXCR5+ Treg cells in different human settings should be carefully interpreted.

Given that circulating Tfr cells have an immature phenotype, it is not unexpected that blood Tfr cells are not fully endowed with suppressive function, because the suppressive capacity of conventional Treg cells has been ascribed predominantly to those cells with a more mature phenotype. Although the TCR repertoire of Tfr cells is different from Tfh and probably skewed toward autoantigens (48), it is possible that circulating Tfr cells represent a pool of cells ready to be recruited into subsequent GC responses as they retain the ability to migrate toward CXCL13.

In conclusion, our data support a model in which CXCR5+Bcl-6 T cells egress from secondary lymphoid tissues during antigen-driven immune responses. Whereas the frequency of blood Tfh cells is reduced in the absence of B cells, Tfr cells do not require interactions with B cells. Thus, the acquisition of a CXCR5+Foxp3+ phenotype in the tissues precedes access to the follicle, where the cells acquire a fully mature phenotype. As a consequence, circulating Tfr cells represent lymphoid tissue–derived Tfr precursors not yet endowed with full B cell and humoral regulatory function.

MATERIALS AND METHODS

Study design

Sample sizes were estimated on the basis of previous studies and according to each cohort (see the next section). No outliers were excluded. The number of biological and technical replicates is stated in the figure legends. Human samples from different conditions were used (see the next section) with appropriate age-matched controls. This experimental study was performed unblinded.

Human samples

Fresh peripheral blood samples were collected from patients referred to the Rheumatology Department of Hospital de Santa Maria, Centro Hospitalar Lisboa Norte, for salivary gland biopsy due to clinical suspicion of SS. Blood samples were collected on the day of salivary gland biopsy. All patients with exclusion criteria for SS (33) or treated with biologic drugs, disease-modifying antirheumatic drugs, or prednisolone (more than 7.5 mg/day) were excluded. Patients diagnosed with an infectious disease in the previous month were also excluded, as well as those who received any vaccine in the same period of time. Patients were diagnosed as having SS if they met the AECG diagnosis criteria (n = 25) (33). Routine C-reactive protein plasma levels (mg/dl) closest to blood collection were used. Age-matched healthy volunteers (from the cohort described below) were used for statistical comparison. Fresh peripheral blood samples were collected from healthy adult volunteers (n = 42). Fresh buffy coats (blood collection in less than 24 hours) were used for in vitro suppression and coculture assays. Tonsils and peripheral blood samples were collected from healthy children submitted to tonsillectomy due to tonsil hypertrophy (n = 6). Children with any clinical condition, under any drug treatment, or submitted to tonsillectomy due to chronic tonsillitis were excluded. Umbilical cord blood samples were collected from healthy pregnant women during delivery (n = 3). Thymus tissue was collected from children submitted to cardiac surgery due to congenital heart disease who were otherwise healthy (n = 4). Blood samples were also collected from X-linked agammaglobulinemia (BTK-deficient) patients during routine blood tests (n = 5). All blood samples were collected in EDTA-coated tubes. These studies were approved by the Lisbon Academic Medical Center Ethics Committee (reference no. 505/14). Informed consent was obtained from all adult volunteers, parents, or legal guardians.

For vaccination studies, we used healthy adult volunteers (n = 32) recruited from the Cambridge BioResource as part of the vaccination study during the 2014–2015 winter season. Participants were excluded if they have had a previous adverse reaction to any vaccination, have a known allergy to any components of the vaccine, were taking immunomodulating medication, and are pregnant or breastfeeding. Participants were administered the inactivated influenza vaccine (split virion) BP vaccine (Sanofi Pasteur) by intramuscular injection in the right deltoid. Blood samples were collected in EDTA-coated tubes on the day of vaccination (before administration of the vaccine) and 7 days after vaccination. The influenza vaccination study protocol was approved by the Health Research Authority, National Research Ethics Service Committee South Central, Hampshire A, UK (REC reference: 14/SC/1077).

Cell isolation and flow cytometry

Peripheral blood mononuclear cells (PBMCs) were isolated from blood samples by Ficoll gradient medium (Histopaque-1077, Sigma-Aldrich) using SepMate tubes (STEMCELL Technologies). Lymphocytes from tonsils and thymocytes were also isolated by Ficoll gradient medium after mechanical disruption. Before cell sorting, PBMCs from buffy coats were enriched for CD4+ T cells using MojoSort Human CD4 T Cell Isolation Kit (BioLegend). The CD4+ fraction was used for cell sorting of CD4+ T cell subsets. The CD4 fraction was used for cell sorting of naïve B cells (see fig. S1A for sorting strategy). For flow cytometry, cells were stained with anti–Bcl-6 (K112-91, BD Biosciences), anti-CCR7 (#150503, R&D Systems), anti-CD127 (eBioRDR5, eBioscience), anti-CD19 (HIB19, BioLegend), anti-CD25 (BC96, eBioscience), anti-CD27 (LG.7F9, eBioscience), anti-CD3 (OKT3, eBioscience), anti-CD31 (WM-59, eBioscience), anti-CD38 (HB-7, BioLegend), anti-CD4 (OKT4, BioLegend), anti-CD45RA (HI100, eBioscience), anti-CD45RO (UCHL1, BioLegend), anti-CD57 (HNK-1, BioLegend), anti-CD62L (DREG-56, BioLegend), anti-CD69 (FN30, BioLegend), anti-CD8 (RPA-T8, eBioscience), anti–CTLA-A (L3D10, BioLegend), anti-CXCR5 (J252D4, BioLegend), anti-Foxp3 (PCH101, eBioscience), anti–HLA-DR (G46-6, BD Biosciences), anti-ICOS (C398.4A, BioLegend), anti-IgD (IA6-2, BioLegend), anti–Ki-67 (Ki-67, BioLegend), and anti–PD-1 (EH12.2H7, BioLegend). For Bcl-6, CTLA-4, Foxp3, and Ki-67 intracellular staining, Foxp3 Fix/Perm Kit (eBioscience) was used according to the manufacturer’s instructions. For cell viability staining, Live/Dead Fixable Aqua Dead Cell Stain Kit (Life Technologies) was used. CellTrace Violet Cell Proliferation Kit (Life Technologies) was used for cell proliferation assessment. Cell sorting was performed in Aria IIu and Aria III instruments (BD Biosciences). Flow cytometry analysis was performed in an LSRFortessa instrument (BD Biosciences) and further analyzed with FlowJo v10 software (Tree Star).

Cell culture and functional assays

For in vitro suppression assays, 25 × 103 CXCR5CD25CD127+CD4+ conventional T cells or 25 × 103 CXCR5+CD25CD127+CD4+ Tfh cells were plated with CXCR5CD25+CD127CD4+ Treg cells or CXCR5+CD25+CD127CD4+ Treg cells in a 1:1 ratio. Cells were cultured with anti-CD3 (1 μg/ml) (OKT3, eBioscience) in the presence of 105 irradiated (25 Gy) allo-PBMCs. After 5 days, cells were harvested and responder cells were analyzed for CTV dilution by flow cytometry. For TCR stimulation assays, 25 × 103 CXCR5CD25+CD127CD4+ Treg cells and CXCR5+CD25+CD127CD4+ Treg cells were plated with anti-CD3/anti-CD28 MACSiBead particles (1 μl per well) (T Cell Activation Kit, Miltenyi Biotec). For coculture in vitro suppression assays, 25 × 103 CXCR5+CD25CD127+CD4+ Tfh cells were plated with CXCR5CD25+CD127CD4+ Treg cells or CXCR5+CD25+CD127CD4+ Treg cells in a 1:1 ratio in the presence of 30 × 103 CD27IgD+CD19+ naïve B cells. Cells were cultured with SEB (1 μg/ml) (Sigma-Aldrich). After 5 days, responder Tfh cells were analyzed for CTV dilution, B cells for CD38 up-regulation, and Treg cells for follicular and activation markers. For immunoglobulin measurement, 25 × 103 CXCR5+CD25CD127+CD4+ Tfh cells were plated with CXCR5CD25+CD127CD4+ Treg cells or CXCR5+CD25+CD127CD4+ Treg cells in a 1:1 ratio in the presence of 30 × 103 CD27IgD+CD19+ naïve B cells. Cells were cultured with SEB (1 μg/ml) (Sigma-Aldrich) + staphylococcal enterotoxin A (SEA) (10 ng/ml) (Toxin Technology) + staphylococcal enterotoxin E (SEE) (10 ng/ml) (Toxin Technology) + toxic shock syndrome toxin–1 (TSST-1) (10 ng/ml) (Toxin Technology). After 10 days, supernatants were collected and immunoglobulin concentration was determined by enzyme-linked immunosorbent assay (ELISA). Cultures were performed in U-shaped 96-well plates in RPMI 1640 (Life Technologies) supplemented with 10% heat-inactivated fetal bovine serum (Life Technologies), 1% Hepes (Sigma-Aldrich), 1% sodium pyruvate (Life Technologies), 1% penicillin-streptomycin (Life Technologies), and 0.05% gentamicin (Life Technologies) in 37°C, 5% CO2 incubator conditions.

Enzyme-linked immunosorbent assay

IgA, IgM, and total IgG concentrations were determined in supernatants from T-B coculture (as described above) by ELISA using Human ELISA Ready Set Go Kit according to the manufacturer’s instructions (eBioscience).

Migration assays

For in vitro chemotaxis assays, 75 × 103 CXCR5CD25CD127+CD4+ conventional T cells, CXCR5+CD25CD127+CD4+ Tfh cells, CXCR5CD25+CD127CD4+ Treg cells, and CXCR5+CD25+CD127CD4+ Treg cells were loaded on top wells of HTS Transwell 96-well permeable supports (5-μm pore size) (Corning). Plain RPMI 1640 (Life Technologies) or medium supplemented with CXCL13 (0.2 μg/ml) (PeproTech) was added to the bottom wells of the plate. After 4 hours of incubation (37°C, 5% CO2), filters were removed and cells that migrated to the lower chamber were counted in an LSRFortessa instrument (BD Biosciences) and further analyzed with FlowJo v10 software (Tree Star). Chemotaxis index was calculated as the ratio of cells migrating toward CXCL13 and cells randomly migrating.

Real-time reverse transcription PCR

Total RNA was extracted and reverse-transcribed from FACS (fluorescence-activated cell sorting)–sorted CXCR5CD25CD127+CD4+ conventional T cells, CXCR5+CD25CD127+CD4+ Tfh cells, CXCR5CD25+CD127CD4+ Treg cells, and CXCR5+CD25+CD127CD4+ Treg cells using RNeasy Micro Kit (Qiagen) according to the manufacturer’s instructions. Complementary DNA was generated using SuperScript III reverse transcriptase (Life Technologies) according to the manufacturer’s instructions. Real-time PCR was set up with Power SYBR Green PCR Master Mix (Applied Biosystems) and performed on ViiA 7 Real-Time PCR System (Applied Biosystems) according to the manufacturer’s instructions. The expression of each gene was normalized to housekeeping genes (B2M, ACTB, or G6PD) and calculated by change-in-threshold method (ΔCT) using QuantStudio Real-Time PCR software v1.1 (Applied Biosystems). The following primers (Invitrogen) were used: Foxp3, 5′-GCAAATGGTGTCTGCAAGTG-3′ (forward) and 5′-GCCCTTCTCATCCAGAAGAT-3′ (reverse); CXCR5, 5′-CTGGAAATGGACCTCGAGAA-3′ (forward) and 5′-GCAGGGCAGAGATGATTTTC-3′ (reverse); Bcl-6, 5′-TTCCGCTACAAGGGCAAC-3′ (forward) and 5′-CGAGTGTGGGTTTTCAGGTT-3′ (reverse); B2M, 5′-TATGCCTGCCGTGTGAACCAT-3′ (forward) and 5′-CGGCATCTTCAAACCTCCATG-3′ (reverse); ACTB, 5′-CTCTTCCAGCCTTCCTTCCT-3′ (forward) and 5′-AGCACTGTGTTGGCGTACAG-3′ (reverse); G6PD, 5′-CCAAGCCCATCCCCTATATT-3′ (forward) and 5′-CCACTTGTAGGTGCCCTCAT-3′ (reverse).

Immunofluorescence microscopy

After paraffin removal and antigen retrieval by heat (HIER pH 9, Leica Biosystems), 3-μm sections of formalin-fixed, paraffin-embedded human tonsil were stained with anti-human CXCR5–Alexa Fluor 488 (J252D4, BioLegend), anti-human CD4 (SP35, Cell Marque), and anti-human Foxp3 (PCH101, eBioscience) primary antibodies. Alexa Fluor 488 (anti-mouse), Alexa Fluor 546 (anti-rabbit), and Alexa Fluor 488 (anti-rat) were used as secondary antibodies. 4′,6-Diamidino-2-phenylindole (DAPI) was used as nuclei counterstaining. Images were acquired with ZEN 2012 software on a Zeiss LSM 710 confocal point-scanning microscope (Carl Zeiss) using a dry Plan-Apochromat 20× objective (×200 magnification) and with a numerical aperture of 0.80. Images were further analyzed using ImageJ Fiji software.

Statistical analysis

Unpaired, paired Student’s t test, one-way analysis of variance (ANOVA) with posttest Turkey’s multiple comparisons, and two-way ANOVA with posttest Bonferroni’s multiple comparison were used as described. Linear regression analysis was also conducted for some data. Results are presented as means ± SD. P values of less than 0.05 were considered statistically significant. GraphPad Prism v5 software was used for statistical analysis.

SUPPLEMENTARY MATERIALS

immunology.sciencemag.org/cgi/content/full/2/14/eaan1487/DC1

Table S1. Clinical characteristics of 25 patients with primary (pSS) and secondary (sSS) Sjögren syndrome.

Fig. S1. Gating strategy for Tfh cells, CXCR5 Treg cells, and Tfr cells in human blood.

Fig. S2. Variation of blood Tfh cells and total Treg cells according to age.

Fig. S3. Sorting strategy for human blood naïve B cells, Tfh cells, CXCR5 Treg cells, and Tfr cells.

Fig. S4. Representative plots of CXCR5 Treg and Tfr cell suppression curves for coculture assay.

Fig. S5. Expression of CD45RO, CD45RA, and CD31 by human blood Tfh and Tfr cells.

REFERENCES AND NOTES

Acknowledgments: We thank A. C. H. Pinto for the recruitment of healthy adult volunteers and for making the diagram. We thank the Instituto Português do Sangue e da Transplantação, the Centro Hospitalar Lisboa Ocidental–Hospital de Santa Cruz, and the Centro Hospitalar Lisboa Norte–Hospital de Santa Maria (Rheumatology, Otorhinolaryngology, and Obstetrics departments) for human samples and collaborations. We thank the Cambridge BioResource staff for help with volunteer recruitment. We thank members of the Cambridge BioResource Scientific Advisory Board and Management Committee for support of our study and the National Institute for Health Research Cambridge Biomedical Research Centre for funding. We acknowledge the participation of all volunteers. We also thank J. Faro for opinions and statistical review. Funding: This study was funded by HMSP-ICT/0034/2013, FAPESP/19906/2014, PTDC/IMI-IMU/7038/2014 research grants, and LISBOA-01-0145-FEDER-007391, projeto cofinanciado pelo FEDER através POR Lisboa 2020–Programa Operacional Regional de Lisboa, do PORTUGAL 2020, e pela Fundação para a Ciência e a Tecnologia. The vaccination study was funded by the European Research Council Starting Grant TWILIGHT (to M.A.L.). W.P. was funded by a Newton International Fellowship from the Royal Society. M.A.L. was funded by the Bioscience and Biotechnology Research Council. Author contributions: V.R.F. designed research, performed experiments, analyzed data, and wrote the paper. A.A.-D. designed research, performed experiments and statistical analysis, and reviewed the paper. A.R.M., F.R., and A.R.P. performed experiments. W.P. performed the vaccination studies. V.C.R. and S.L.d.S. selected SS and BTK-deficient patients, respectively. J.E.F., A.E.S., and M.A.L. designed research and reviewed the paper. L.G. designed research and wrote the paper. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article