In the attempt to understand the origin of autoantibody (AAb) production in patients with and at risk for type 1 diabetes (T1D), multiple studies have analyzed and reported alterations in T follicular helper (Tfh) cells in presymptomatic AAb+ subjects and patients with T1D. Yet, whether the regulatory counterpart of Tfh cells, represented by T follicular regulatory (Tfr) cells, is similarly altered is still unclear. To address this question, we performed analyses in peripheral blood, spleen, and pancreatic lymph nodes (PLN) of organ donor subjects with T1D. Blood analyses were also performed in living AAb− and AAb+ subjects. While negligible differences in the frequency and phenotype of blood Tfr cells were observed among T1D, AAb−, and AAb+ adult subjects, the frequency of Tfr cells was significantly reduced in spleen and PLN of T1D as compared with nondiabetic control subjects. Furthermore, adoptive transfer of Tfr cells delayed disease development in a mouse model of T1D, a finding that could indicate that Tfr cells play an important role in peripheral tolerance and regulation of autoreactive Tfh cells. Together, our findings provide evidence of Tfr cell alterations within disease-relevant tissues in patients with T1D, suggesting a role for Tfr cells in defective humoral tolerance and disease pathogenesis.
Type 1 diabetes (T1D) is an autoimmune disease characterized by the destruction of pancreatic β-cells (1). Abnormalities in T cell activation, function, and differentiation drive development of autoimmunity in patients and mouse models of T1D (2). In humans, detection of multiple islet-specific autoantibodies (AAb) and impaired glucose tolerance divide progression to T1D into two progressive stages leading to insulin dependence (3). Asymptomatic individuals positive for two or more diabetes-related AAb are classified as patients with stage 1 T1D, while those who have additionally developed impaired glucose tolerance are classified with stage 2 T1D.
T follicular helper (Tfh) cells are a subset of CD4+ T cells, highly specialized in promoting efficient antibody production from B cells in germinal centers (GCs) of the spleen and secondary lymphoid organs (SLOs) (4–7). Tfh cells are characterized by the expression of chemokine (C-X-C motif) receptor type 5 (CXCR5) (8,9), production of IL-21 (10), and high levels of activation markers inducible T cell costimulator (ICOS) and programmed cell death protein 1 (PD-1) (11,12). Higher Tfh cell activity is linked to production of AAb in T1D (13,14). T follicular regulatory (Tfr) cells also express CXCR5, PD-1, and ICOS but are a subset of FOXP3+ T regulatory (Treg) cells that inhibit AAb production (15–17).
Tfh cells transiting in peripheral blood can be segregated into four functionally distinct populations based on expression of chemokine receptors, CXCR3 and CCR6 (17–19). Activated (PD-1hiICOS+) blood Tfh cells transiently expand following vaccination and infection and were found increased in patients with T1D and other autoimmune diseases (13,20–22). Blood Tfr cells also increase after vaccination or infection and are elevated in the blood of patients with autoimmunity, but their role in progression to T1D is unknown (23–25). Recently, Fonseca et al. (26) found an increased number of blood Tfr cells in patients with Sjögren syndrome that lacked suppressive potential in vitro, and suggested that an increased number of blood Tfr cells indicates ongoing humoral activity. The same group showed that GC and circulating Tfr cells arise independently via a bifurcated trajectory from FOXP3+ precursor Treg cells, suggesting that blood Tfr cells are not good indicators of the SLO-residing Tfr cells (27).
The development of high-affinity class-switched islet-specific AAb in GCs may be controlled by Tfh and Tfr cells (14,28). In peripheral blood, Tfh cell–associated genes such as BCL6, IL21, CXCR5, PDCD1, CXCL13, and ICOS were upregulated in memory CD4+ T cells from adults with long-standing T1D (22). Consistent with this observation, other studies have shown the frequency of circulating activated Tfh cells to be elevated in adult T1D patients (29,30). Activated Tfh cells were also increased in the peripheral blood of children with stage 1 or stage 2 T1D, with the relative frequency of Tfh cells increasing in stage 2 T1D (13). The expansion of activated blood Tfh cells in patients with T1D may be due to migration of cells from SLO, as a recent histological study showed a reduction in the number of GCs in the pancreatic lymph node (PLN) in T1D patients (31). Previously, we reported an elevated frequency of Tfr cells in peripheral blood of children with new-onset T1D as compared with healthy children (25). However, much still is unknown about the frequency, phenotype, and function of Tfr cells in T1D and, particularly, in the tissues where these cells function and reside.
We hypothesized that there would be an alteration of Tfr cell number and immunosuppressive function in SLO of patients with T1D, leading to the generation of AAb. Given the limited number of studies assessing the contribution of Tfr in islet autoimmunity, we analyzed Tfh and Tfr cells in peripheral blood, spleen, and PLN of patients with T1D and control organ donors without diabetes, as well as from peripheral blood of living T1D patients and their relatives with risk of developing T1D (from 18 to 66 years of age). While Tfr cells were unaltered in blood, we found them reduced in spleen and PLN of patients with T1D compared with nondiabetic control subjects. In mice adoptive transfer of BDC2.5 Tfh cells alone into NOD.SCID animals induced T1D, while cotransfer of Tfr cells delayed disease development. These results suggest that a reduction in Tfr cells in SLO of humans with T1D might contribute to the generation of islet-specific AAb due to a reduced regulation in Tfh–B cell interactions.
Research Design and Methods
Subjects and Data Collection
Human blood and tissue collection was conducted under protocols approved by the San Raffaele Hospital Ethics Committee, and a University of Florida institutional review board approved exempt human subjects’ protocol. Peripheral blood was collected from adults in groups, shown in Supplementary Tables 1 and 2. Spleen and PLN were derived from brain-dead donors without diabetes (control subjects) of the San Raffaele Hospital islet transplantation program and from T1D donors from the Network for Pancreatic Organ donors with Diabetes (nPOD). A summary and detailed list of organ donors used in this study are provided in Supplementary Tables 3–5. Patients with T1D were positive for at least one islet-specific AAb (GAD65 AAb, insulin AAb [IAA], islet cell antibodies [ICA], zinc transporter 8 AAb, and islet antigen 2 [IA-2]) (Supplementary Table 2). Relatives of patients with T1D enrolled in the Type 1 Diabetes TrialNet Pathway to Prevention trial (TN01) were screened for islet-specific AAb and considered at risk for developing T1D. Subjects were tested for five islet-specific AAb (GAD65 AAb, IAA, ICA, zinc transporter 8 AAb, IA-2) and were classified as follows: AAb− subjects; positive for one AAb (1 AAb+); stage 1 T1D, subjects with two or more islet AAb (≥2 AAb+) and normal glucose tolerance test results; and stage 2 T1D, subjects with ≥2 AAb+ and impaired glucose results of the glucose tolerance test (32).
Whole blood, peripheral blood mononuclear cells (PBMCs), and single cell suspensions from spleen and PLN were prepared as previously described (33). Immune populations were stained with a combination of antibodies specific for human CD45 (REA747), CD4 (SK3), CD3 (SK7), CXCR5 (RF8B2), CD25 (2A3), CD19 (4G7), CD8 (BW135/80), CD14 (TUK4), CXCR3 (IC6), CCR6 (G034E3), CD45RA (REA562), PD-1 (eBioJ105), and ICOS (ISA-3) for 20 min at room temperature (Supplementary Table 6). Erythrocytes in stained whole blood with EDTA were lysed with ammonium-chloride-potassium and washed, and cells were fixed in PBS with 0.4% PFA. PBMCs were isolated from heparinized venous blood by density gradient centrifugation with Lymphoprep (STEMCELL Technologies) according to the manufacturer’s instructions. Stained PBMCs, spleen, and PLN were fixed-permeabilized with a Foxp3 fixation kit (eBioscience) and stained for FOXP3 (259D) according to the manufacturer’s instructions. Cells were acquired on a FACSCanto II (BD Biosciences) and analyzed with FlowJo software (v10; BD Biosciences). Detailed staining panels can be found in Supplementary Table 6. Whole blood was used to analyze Tfh cell subsets (CXCR3+CCR6− Tfh1, CXCR3−CCR6− Tfh2, CXCR3−CCR6+ Tfh17, and CXCR3+CCR6+ Tfh1/17 cells and PD-1+CXCR3− Tfh cells). PBMCs were used to analyze Tfh, Tfr, and CXCR5− Treg cells.
PLN from subjects with T1D and control subjects (Supplementary Table 5) were fixed with formalin, embedded in paraffin, cut to 5-μm sections, mounted on slides, and dried for 24 h. Slides were permeabilized with confocal buffer for 1 h, washed with PBS, and blocked with mouse and/or goat serum for 1 h and then incubated with primary antibodies overnight at 4°C (Supplementary Table 7). Slides were washed three times for 15 min in PBS and stained with secondary antibodies for 2 h (Supplementary Table 7). Tissues were stained with a nuclear marker (JoPro) at a 1:15,000 dilution for 20 min and then mounted with Fluoromont-G mounting media (Thermofisher). A Nikon (A1) confocal microscope using NIS-Elements AR software was used to capture z-stacked images (512 × 512 pixels) of PLN. The “live un-mixing” function in NIS-Elements AR software was used for compensation of the emitted fluorescence, with the emission spectrum from single-stained tissues used to separate the fluorescence into corresponding channels. Further quantitative imagining analysis was performed with histocytometry as previously described on three serial tissue sections of PLN (34).
CXCL13 Plasma Levels Evaluation
Whole peripheral blood (EDTA) was centrifuged at 1,000 rpm for 15 min, and plasma was collected. Plasma was further centrifuged at 13,000 rpm for 10 min to remove any debris. CXCL13 was evaluated in plasma by ELISA assay (Human CXCL13/BLC/BCA-1 Quantikine ELISA Kit; R&D Systems) according to the manufacturer’s instructions.
NOD-SCID and BDC2.5 mice were maintained in a specific pathogen-free barrier in the animal facility at San Raffaele Scientific Institute. Mouse splenocytes were sorted on a BD FACSAria II sort flow cytometer, washed, and resuspended in sterile PBS. CD4+ Tfh (CXCR5+CD25−) cells (2.5 × 105) from BDC2.5 mice, with or without 2.5 × 105 BDC2.5 CD4+CXCR5+CD25+ Tfr cells, as indicated were injected intravenously (i.v.) into NOD-SCID recipients. B cells from normoglycemic NOD mice were isolated with CD19 B Cell Isolation Kit (Miltenyi), and 3 × 106 cells were coinjected i.v. Mice were monitored for diabetes with measurement of blood glucose twice per week. Diabetes was defined as two consecutive readings >11.1 mmol/L. All animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee of San Raffaele Scientific Institute.
Pancreata were collected either when mice turned diabetic or 30 days after adoptive transfer, when control, BDC2.5 conventional (CXCR5−CD25−) T (Tconv) cell–treated mice first turned diabetic. After being placed in 4% formalin, pancreata were embedded in paraffin and stained with hematoxylin-eosin (H-E) as previously described (35).
Quantitative data are expressed as mean and categorical data expressed as n (%). Normality of distribution was assessed with the Shapiro-Wilks test. Nonparametric tests were used for analysis. Comparisons between two groups were performed with Mann Whitney U test. Comparisons among more than two groups were performed with Kruskal-Wallis test with Dunn posttest for multiple comparisons. Long-rank Mantel-Cox test was used for multiple comparisons of murine data. Spearman correlation coefficient was computed to study the strength of correlation between quantitative variables. Statistical analyses were performed with GraphPad Prism software, version 7. A P value <0.05 was considered significant.
Data and Resource Availability
The data sets generated during or analyzed during the current study are available from the corresponding author upon reasonable request.
Results and Discussion
Blood Tfh and Tfr Cells in AAb+ Adult Subjects With and Without Clinical T1D
The role of islet-specific AAb that are usually detected months to several years before the clinical onset of T1D in humans remains unknown. Tfh cells control Ig production from B cells. Alterations in Tfh cell function and/or their regulation by Tfr cells were shown to associate with the presence of islet-specific AAb in humans with presymptomatic and diagnosed T1D (22,36,37). Here, we determined the frequency of Tfh and Tfr cells in living subjects with T1D, their first- and second-degree relatives, and control subjects recruited through TrialNet Milan and from San Raffaele Hospital (donors are described in Supplementary Tables 1 and 2). The frequency of circulating Tfh (CXCR5+FOXP3−) and Tfr (CXCR5+FOXP3+) and conventional CXCR5− Treg (CXCR5−FOXP3+) cells was determined in AAb− (n = 14), AAb+ (n = 26), and T1D (n = 34) subjects after gating on total CD4+CD3+ (CD19−CD8−CD14−) cells, as we previously described (25). We and others previously showed that CXCR5+ CD4 T cells are either CD45RA− or CD45RAint, suggesting they are memory or antigen experienced (38,39). Hence, no CD45RA− gating step was included in the analysis. While the frequency of circulating Tfr cells was similar in all groups, the frequency of Tfh cells was lower in individuals with T1D compared with age-matched AAb+ subjects (mean ± SD 19.93 ± 7.92% vs. 26.24 ± 6.38%, respectively) (Fig. 1A and B). Years with T1D were not significantly associated with the distribution of Tfh, Tfr, and CXCR5− Treg cells (Supplementary Fig. 1A). Altered Tfh–to–Tfr cell ratios have been associated with an increased prevalence of AAb in autoimmune BXD2 mice (40). In analyzing the Tfh/Tfr in our cohort, we found no differences among the groups (Fig. 1B).
Although the frequency of Tfr cells was similar between AAb−, AAb+, and T1D subjects, we sought to evaluate potential differences in the levels of the remaining (non-Tfr) Treg cells. The frequencies of conventional CXCR5−FOXP3+ Treg were similar between AAb+ subjects and patients with T1D but higher in AAb+ subjects (7.51 ± 2.77%) as compared with AAb− control subjects (5.06 ± 2.20%), possibly as a consequence of active autoimmunity in AAb+ individuals in advance of metabolic dysfunction (Fig. 1C). We further divided the AAb+ subjects into stage 1 and stage 2 T1D (32) and assessed the distribution of circulating Tfr and Tfh cells according to this classification, but no differences were seen even in comparisons with AAb− and T1D groups (Supplementary Fig. 1B). Specific genetic alleles of the HLA are strongly associated with increased susceptibility to T1D (41). Class II HLA haplotypes that confer increased risk to develop T1D are HLA-DRB1*03 (DR3) and HLA-DRB1*04 (DR4), with subjects heterozygous for DR3/DR4 having the highest risk (42). We analyzed the proportion of Tfh, Tfr, CXCR5− Treg, and total Treg cells in our cohort of AAb+ TrialNet subjects without diabetes who had HLA-DRX/DRX (i.e., neither HLA-DR3 nor HLA-DR4, n = 6), DR3/DRX (n = 9), DR4/DRX (n = 4), or DR3/DR4 (n = 4). We found a trend of higher Tfr cell frequency in HLA-DR3/DR4 AAb+ subjects compared with HLA-DRX/DRX subjects (Supplementary Fig. 1C).
Elevated CXCL13 plasma levels, a marker of GC activity, have been shown to correlate with the relative frequency of circulating Tfh cells (43). Consistent with the lower circulating frequency of Tfh cells in our subjects with T1D, we found lower plasma CXCL13 levels in individuals with T1D compared with AAb− and at-risk AAb+ subjects (24.85 ± 18.22 pg/mL T1D vs. 57.94 ± 40.19 pg/mL AAb− vs. 46.45 ± 32.66 pg/mL AAb+ (Fig. 1D). Plasma CXCL13 levels associated positively with the frequency of Tfr cells but not with that of the Tfh or CXCR5− Treg cells or the Tfh–to–Tfr cell ratio (Supplementary Fig. 1D).
Circulating Tfh cells are composed of phenotypically and functionally distinct subsets based on the expression of the CXCR3 and CCR6 chemokine receptors, with each Tfh subset displaying distinct capacities to help B cells (18,19). Here, we determined the frequencies of Tfh1, Tfh2, Tfh17, and CXCR3+CCR6+ Tfh1/17 cells in the same cohort of adult subjects (using whole blood). We found that Tfh1/17 cell frequency was higher while Tfh2 cell frequency was lower in patients with T1D in comparison with AAb+ subjects (Supplementary Fig. 2A and B). We also examined the frequency of PD-1+CXCR3− Tfh cells, which were found to strongly correlate with the production of neutralizing antibodies in HIV patients (12). Although AAb+ subjects had a slightly elevated frequency of PD-1+CXCR3− Tfh cells, there were no significant differences among any of the groups (Supplementary Fig. 2C and D). No differences in Tfh1, 2, or 17 or PD-1+CXCR3− Tfh cells were observed when AAb+ subjects were divided according to their disease stage or HLA type (data not included).
Next, we assessed the functional competence of Tfh cells from the blood of living control (n = 19) and T1D (n = 9) subjects in coculture experiments with naïve or memory B cells stimulated with Staphylococcal enterotoxin B (SEB) (Fig. 1E). No differences in plasmablast (CD20−CD38+) differentiation or IgG or IgM production were seen from either naïve or memory responder B cells, showing that T1D and control-derived Tfh cells were similarly potent at providing B-cell help (Fig. 1E and F). As expected, memory B cells produced higher levels of IgG and IgM compared with naïve B cells cultured with Tfh cells (Fig. 1F).
The pathological role of islet-specific AAb detected months to several years before the clinical onset of T1D in humans is unknown. Tfh cells control Ig production from B cells. Therefore, alterations in Tfh cell function and/or their regulation by Tfr cells likely contribute to the presence of islet-specific AAb in humans with presymptomatic and diagnosed T1D. Our data confirmed alterations in blood Tfh cells of presymptomatic AAb+ individuals reported previously (44), showing that peripheral blood of adults with islet-specific AAb had a higher frequency of Tfh cells, even compared with T1D patients. One report showed that circulating Tfh cells were also increased in children with stage 2 but not stage 1 presymptomatic T1D (13).
Contrary to a previous report showing increased circulating Tfr in T1D patients (pooled adults and children) (36), no significant alterations in circulating Tfr cells were observed in our cohort of adults reported here. Viisanen et al. (45), on the other hand, found an increase in circulating Treg cells but not Tfr cells in children with newly diagnosed T1D. We previously described an expansion of circulating Tfr cells expressing reduced levels of PD-1 in children with T1D (13,46). These results suggest that T1D immunopathogenesis might be different in children compared with adults, as has been previously speculated (47–49). Differences between our study and those previously reported (13,36) may be due to different gating strategies, cohorts, age distribution, and duration of T1D.
Human Tfr Cells Are Highly Enriched in PLN Compared With Spleen and Blood in Control Donors
Tfh and Tfr cells are most commonly identified in SLOs, including tonsils, spleen, and lymph nodes, where they interact with B cells during the GC response (50). Although more easily accessible, peripheral blood Tfh and particularly Tfr cells may not accurately represent their counterparts in SLOs (51). Thus, we determined the frequency of Tfh (CXCR5+FOXP3−), Tfr (CXCR5+FOXP3+), and CXCR5− Treg (CXCR5−FOXP3+) cells in PLN and spleen of 15 control organ donors without diabetes (aged 39–64 years) as compared with PBMCs from 26 living healthy control (HC) subjects (aged 18–46 years). Spleen and PLN samples were derived from the same donors (Supplementary Table 3) (all samples derived from our hospital). With the caveat that blood samples were derived from living donors while spleen and PLN were from deceased donors, we found an increased frequency of Tfr cells in PLN as compared with spleen and blood (mean ± SD 11.86 ± 4.28% PLN vs. 2.02 ± 1.26% spleen vs. 1.66 ± 1.14% PBMC [Fig. 2A and B]), suggesting that the PLN could be an important immune checkpoint site. We further found that the frequencies of CXCR5− Treg cells in the PLN, spleen, and blood were similar (Fig. 2C). The Tfh-to-Tfr ratio was significantly lower in PLN (2.46 ± 1.3 PLN vs. 16.19 ± 12.06 spleen vs. 24.8 ± 23.01 PBMC). We also found that the expression of PD-1 on Tfr, Tfh, and CXCR5− Treg was highest in spleen followed by PLN and lowest in PBMC (Fig. 2D and E).
Similarly to Kumar et al. (27), we found a predominant presence of Tfr cells, but not Tfh, and increased PD-1 expression by Tfr cells in the spleen and PLN of nondiabetic control subjects in comparison with blood of individuals without diabetes. Our observations suggest that Tfr cells in spleen and PLN possibly play an important role in regulating self-tolerance and inflammation. To date, no studies related to T1D have addressed Tfh and Tfr cells in human tissues other than the blood. Next, we analyzed Tfh and Tfr cells in spleen and PLN of patients with T1D.
Tfr Cells Are Decreased in Spleen and PLN of Patients With T1D
Immunophenotyping and histological studies of SLOs, including spleen and PLN, from patients with T1D are very limited (52). Willcox et al. (31) showed decreased primary B cell follicle frequency and fewer follicular DC networks in PLN from patients with recent-onset T1D but did not explore what role Tfr cells may have played. To gain further insights into the distribution of Tfh, Tfr, and CXCR5− Treg cells in these tissues, we performed flow cytometry and histological studies in spleen and PLN from cadaveric donors without diabetes (control) (n = 15 for flow cytometry, n = 8 for histological studies) and cadaveric donors with T1D (n = 9 for flow cytometry, n = 8 for histological studies) (Supplementary Tables 4 and 5). In comparisons with control subjects, Tfr cells of T1D subjects were reduced in spleen (mean ± SD 2.02 ± 1.26% control vs. 0.63 ± 0.52% T1D) and PLN (11.86 ± 4.28% control vs. 6.02 ± 7.32% T1D) (Fig. 3A and B), while Tfh cell frequency was similar between T1D and control subjects (Fig. 3A and B). CXCR5− Treg cells were also reduced in PLN and spleen of subjects with T1D compared with control subjects, but the difference was not significant in the flow cytometry study (Fig. 3A and B). A caveat in this analysis is the use of two different cohorts, where the T1D cohort was derived exclusively from nPOD, with collection and processing likely different at several sites, partly explaining the great variability in the results. Another caveat is the age difference between the T1D and control cohorts, with the control subjects being much older overall (Supplementary Table 4). Hence, we undertook a histological analysis of PLN and analyzed two cohorts (T1D vs. non-T1D) that were more similar in age (Supplementary Table 5). Of note, we observed that samples of the youngest T1D (7.37 ± 8.18%) and oldest control (12.56 ± 3.99%) subjects do not represent extremes in phenotype. Furthermore, samples of older T1D (1.75 ± 1.56%) and younger control (15.4 ± 0.9%) subjects are not similar. The Tfr frequency in older T1D subjects (average age 48 years) is 1.75 ± 1.56%, much lower than that seen in younger control subjects (average age 44.5 years), 15.4 ± 0.9%, in PLN. In spleen, the oldest T1D subjects (0.5 ± 0.31%) have Tfr frequency similar to that of the youngest control subjects (0.54 ± 0.19%).
The relative decrease in Tfr cells in patients with T1D was confirmed by immunofluorescence in the PLN tissue (Fig. 3C). In this study, B cell follicles (CD20+ areas) and T cell zones (CD20−CD4+ areas) in human PLN were analyzed. Since CXCR5 was not used in the immunofluorescence staining, FOXP3+CD4+ T cells in follicles were considered as Tfr cells and FOXP3−PD-1+ T cells as Tfh cells. Similarly, FOXP3+ and FOXP3− CD4 T cells were considered conventional Treg and T helper cells in the T-cell zone, respectively. Arrows in Fig. 3C (top left) show Treg cells (CD4+FOXP3+) localized in B cell follicles of a control tissue. We found that the number of FOXP3+ cells was lower in the follicles of PLN from T1D donors compared with control subjects, while there were no differences in the number of CD4+FOXP3-PD-1+ cells (Fig. 3D). We also calculated the number of CD4+FOXP3+ Treg cells in the T-cell zone (Fig. 3C). Interestingly, the number of Treg cells in the T cell zone was also lower in T1D donors (Fig. 3E). Thus, our results show that the distribution of Tfr (CD4+FOXP3+) cells is reduced within the B cell follicles of PLN from T1D donors, which could impact the architectural structure of B cell follicles and cause abnormal generation of islet-specific AAbs. Of note, different studies have reported that depletion of Treg in mouse models causes an increase in antigen-specific Tfh cells in LN and circulating AAb, including anti-dsDNA (15,16,53), suggesting that Tfr cells might be involved in islet-specific AAb insurgence and T1D progression.
Tfr Cells Can Delay Disease Onset in a Mouse Model of T1D
Next, to address the role of Tfr cells in T1D development, we tested their ability to regulate T1D in an adoptive transfer mouse model of the disease (54). (See Supplementary Fig. 3A for sorting gating strategy.) First, we examined whether Tfh-like (CXCR5+CD25−) cells isolated from the spleens of BDC2.5 transgenic mice were able to induce T1D in NOD-SCID mice, similarly to conventional (CXCR5−CD25−) T (Tconv) cells isolated from the same donors. (See Supplementary Fig. 3B and C for frequencies of Tfh- and Tfr-like cells in the spleen of BDC2.5 mice.) BDC2.5 Tfh cells were as potent as BDC2.5 Tconv cells at inducing T1D in NOD-SCID mice (Fig. 4A). Interestingly, cotransfer of Tfr-like cells isolated from BDC2.5 mice substantially delayed but did not protect against T1D onset in NOD-SCID mice (Fig. 4A). In contrast to conventional BDC2.5 Treg cells (Fig. 4A), Tfr-like cells were unable to offer permanent protection from the disease (Fig. 4A). Analysis of transferred cells showed presence of FOXP3+ Treg cells in mice injected with CXCR5− and CXCR5+CD25+ BDC2.5 Treg cells (Supplementary Fig. 3D and E), suggesting that Tfr-like cells possibly downregulated CXCR5 but maintained FOXP3 expression under these experimental conditions. Histological analysis of the pancreata 30 days after adoptive transfer revealed destructive insulitis in mice treated with Tconv or Tfh alone, while Treg of Tfr cotransfer provided protection (Fig. 4B).
The functional differences between murine Tfr and Treg cells might have been due to their FOXP3 content, stability, mechanism of function, trafficking, or the pathogenicity/Treg resistance of the effectors. Xu et al. (36) performed a similar experiment transferring Tfr cells sorted from prediabetic NOD mice into NOD-SCID mice and similarly found that Tfr cells could prevent T1D. In a different mouse model, Kenefeck et al. (22) transferred DO11 T cells enriched in CXCR5 expression into RIP-mOVA–expressing mice, thereby demonstrating that Tfh cells can transfer diabetes and cause pancreas infiltration. Similarly, our results indicate that Tfh-like cells sorted from BDC2.5 mice have diabetogenic function and, importantly, Tfr cells can delay it. These results further suggest that Tfr cells are needed to maintain peripheral tolerance by regulating diabetogenic Tfh and B cells and that their absence might potentiate the onset of T1D. Thus, Tfr cell adoptive cell therapy should be explored as a potential treatment to prevent T1D.
In conclusion, our findings support the hypothesis that the reduction of tissue-resident Tfr cells might be at the root of a failed peripheral tolerance, leading to the activation of islet-specific Tfh cells and the production of AAb in human T1D. Further studies of tissue-resident Tfr cells in AAb+ subjects are needed to reveal their possible role in AAb production and disease pathogenesis. Therefore, our data provide significant evidence for a regulatory role for Tfr cells in preventing progression to T1D and provide support for exploring Tfr cells as potential therapeutic modalities to treat T1D.
This article contains supplementary material online at https://doi.org/10.2337/figshare.16712971.
R.D.F. is currently affiliated with Istituto Tumori IRCCS “Giovanni Paolo II,” Bari, Italy.
Acknowledgments. The authors thank Bechara Mfarrej for help with FACS staining and all nurses from the Pediatric Department at San Raffaele Hospital for help with blood collection. The authors thank Amanda Posgai Simmons (Department of Pathology, Immunology, and Laboratory Medicine, College of Medicine, University of Florida, Gainesville, FL) for help with editorial assistance.
Funding. Samples from at-risk subjects were obtained through a TrialNet ancillary study to TN01 funded by National Institutes of Health grants (U01 DK061010, U01 DK061034, U01 DK061042, U01 DK061058, U01 DK085465, U01 DK085453, U01 DK085461, U01 DK085463, U01 DK085466, U01 DK085499, U01 DK085504, U01 DK085505, U01 DK085509, U01 DK103180, U01-DK103153, U01-DK085476, U01-DK103266, and P01 AI42288 and to TMB) and The Leona M. and Harry B. Helmsley Charitable Trust and JDRF International. This work was funded by Young Researchers Award from the Italian Ministry of Health to G.F. (GR-2011-02348732). This research was performed with the support of nPOD (RRID SCR_014641), a collaborative T1D research project sponsored by JDRF (nPOD: 5-SRA-2018-557-Q-R) and The Leona M. and Harry B. Helmsley Charitable Trust (grant no. 2018PG-T1D053). A list of organ procurement organizations partnering with nPOD to provide research resources is available from https://www.jdrfnpod.org/for-partners/npod-partners/.
The content and views expressed are the responsibility of the authors and do not necessarily reflect the official view of nPOD.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. A.V. performed experiments, analyzed data, and wrote the manuscript T.J., J.G., K.S., R.D.F., G.G., E.I., A.R.S., H.R.S., M.F., G.M., A.S., and F.R. performed experiments. P.G., EBi., A.L., A.C., R.N., R.M., EBo., and L.P. collected samples and contributed to discussion. M.P.C., A.A., and M.B. discussed data. N.D., T.B., and C.P. discussed data and reviewed and edited the manuscript. G.F. performed experiments, analyzed data, and wrote the manuscript. G.F. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.