CD4⁺CD45RA⁻FoxP3^high activated regulatory T cells are functionally impaired and related to residual insulin-secreting capacity in patients with type 1 diabetes

Abstract

Accumulating lines of evidence have suggested that regulatory T cells (T_regs) play a central role in T cell-mediated immune response and the development of type 1A and fulminant type 1 diabetes. CD4⁺forkhead box protein 3 (FoxP3)⁺ T cells are composed of three phenotypically and functionally distinct subpopulations; CD45RA⁺FoxP3^low resting T_regs (r-T_regs), CD45RA⁻FoxP3^high activated T_regs (a-T_regs) and CD45RA⁻FoxP3^low non-suppressive T cells (non-T_regs). We aimed to clarify the frequency of these three subpopulations in CD4⁺FoxP3⁺ T cells and the function of a-T_regs with reference to subtypes of type 1 diabetes. We examined 20 patients with type 1A diabetes, 15 patients with fulminant type 1 diabetes, 20 patients with type 2 diabetes and 30 healthy control subjects. A flow cytometric analysis in the peripheral blood was performed for the frequency analysis. The suppressive function of a-T_regs was assessed by their ability to suppress the proliferation of responder cells in a 1/2:1 co-culture. A flow cytometric analysis in the peripheral blood demonstrated that the frequency of a-T_regs was significantly higher in type 1A diabetes, but not in fulminant type 1 diabetes, than the controls. Further, the proportion of a-T_regs among CD4⁺FoxP3⁺ T cells was significantly higher in patients with type 1A diabetes with detectable C-peptide but not in patients with type 1A diabetes without it and with fulminant type 1 diabetes. A proliferation suppression assay showed that a-T_regs were functionally impaired both in fulminant type 1 diabetes and in type 1A diabetes. In conclusion, a-T_regs were functionally impaired, related to residual insulin-secreting capacity and may be associated with the development of type 1 diabetes.

Introduction

Type 1 diabetes (T1D) is characterized by insulin deficiency as a result of the destruction of beta cells in the pancreas. According to the classification made by the American Diabetes Association and the World Health Organization, type 1 diabetes is divided further into two categories: autoimmune type 1 (type 1A) diabetes and idiopathic (type 1B) diabetes 1,2. Type 1A diabetes is now believed to result from destruction of beta cells by autoreactive immunocytes. Insulin is the most likely primary antigen in rodent models and humans 3,4. Several islet-related autoantibodies, such as islet cell antibodies (ICA), anti-glutamic acid decarboxylase (GAD) antibodies, insulin autoantibodies (IAA), anti-insulinoma-associated antigen 2 (IA-2) antibodies and anti-zinc transporter 8 (ZnT8) antibodies have been recognized as markers 1,2,5. However, these antibodies are not cytotoxic, and mononuclear cells of T cells and macrophages that are infiltrating to islets have been thought to damage beta cells. Those islets-reactive T cells are also detected in the peripheral blood of patients with type 1A diabetes 6,7. Therefore, type 1A diabetes is now labelled as T cell-mediated autoimmune disease to beta cell antigens 8, and T cells play a central role in the development of this disorder.

At present, little is known about type 1B diabetes. However, various studies have established the presence of a clinical entity of fulminant type 1 diabetes as a distinct subtype within type 1 diabetes, at least in Asian countries, where this subtype accounts for about 20% of acute-onset ketosis-prone type 1 diabetes. Fulminant type 1 diabetes is characterized by almost complete insulin deficiency resulting from the destruction of pancreatic beta cells 9,10. The clinical characteristics of this disease are: (1) remarkably abrupt onset; (2); very short duration of diabetic symptoms; (3) acidosis at the time of diagnosis; (4) negative findings in general for islet-related autoantibodies; (5) virtually no C-peptide secretion at the time of diagnosis; and (6) elevated serum pancreatic enzyme levels 9–11. Massive cellular infiltration of T cells and macrophages has been detected in both islets and exocrine pancreas just after disease onset 12. Increased T cell responses against pancreatic beta cell antigens, as detected by an enzyme-linked immunospot (ELISPOT) assay, have been proposed in this subtype as well as in type 1A diabetes 13. There is low expression of cytotoxic T lymphocyte antigen-4 (CTLA-4) in helper T cells, which is correlated significantly with higher proliferation of those cells in patients with fulminant type 1 diabetes 14. These findings suggest that the development of fulminant type 1 diabetes is associated with the destruction of pancreatic beta cells through increased T cell activation due to decreased CTLA-4 expression of helper T cells.

Accumulating lines of evidence have indicated that regulatory T cells (T_regs) play a central role in suppressing the T cell-mediated immune response. Therefore, their dysfunction could cause T cell-mediated autoimmune disease. Several markers for T_regs, such as human transcription factor forkhead box P3 (FoxP3), CTLA-4, CD25^high and CD127^low, have been clarified. Of these molecules, CTLA-4 could be functionally most essential for T_regs to suppress the immune response 15–17. Indeed, low expression of CTLA-4 in helper T cells and soluble splice variants of CTLA-4 (sCTLA-4) in sera from patients with fulminant type 1 diabetes has been observed in a previous study 14,18. Nevertheless, functional defects of T_reg, defined as CD4⁺CD25^high T cells, could not be found in patients with fulminant type 1 diabetes or in type 1A diabetes 14. Defects in the T_reg function have been found in some studies of patients with type 1 diabetes 19–23, but not in other studies 24,25. These studies have defined T_reg as CD4⁺CD25^high, or at least CD4⁺FoxP3⁺ T cells.

CD4⁺FoxP3⁺ T cells are composed of three phenotypically and functionally distinct subpopulations: CD45RA⁺FoxP3^low resting T_regs (r-T_regs), CD45RA⁻FoxP3^high activated T_regs (a-T_regs), both of which were suppressive in vitro, and interleukin (IL)-17-secreting CD45RA⁻FoxP3^low non-suppressive T cells (non-T_regs) 26,27. a-T_regs express intracellular CTLA-4 and C-C chemokine receptor type 4 (CCR4) to the highest degree among the three subpopulations of CD4⁺FoxP3⁺ T_regs (a-T_reg, r-T_reg and non-T_reg), whereas r-T_regs hardly express the former 26,27. a-T_regs are anergic and suppress the proliferation of responder cells and other FoxP3⁺ T_regs (r-T_regs and non-T_regs), whereas r-T_regs are not anergic and are able to proliferate themselves upon T cell receptor (TCR) stimulation 26–29. Moreover, r-T_regs are able to convert to a-T_regs after and/or during TCR stimulation and suppress the proliferation of responder cells and themselves in a negative feedback manner 26,27. Non-T_regs, which can produce IL-2, interferon (IFN)-γ and IL-17, are not able to suppress responder cells 26–29. These facts indicate clearly that CD4⁺FoxP3⁺ T_regs whose surface marker is CD4⁺CD25^high are not all suppressive and anergic.

The current study sought to clarify the precise function of T_regs in patients with type 1 diabetes by assessing the proportion of the three subpopulations in CD4⁺FoxP3⁺ T cells, and also the suppressive function of CD4⁺CD45RA⁻FoxP3^high a-T_regs.

Methods

Human subjects

Peripheral blood samples for the frequency analysis were obtained from 20 patients with type 1A diabetes, 15 patients with fulminant type 1 diabetes, 20 patients with type 2 diabetes and 30 age-matched healthy control subjects (Table 1). Functional analysis was performed in some of those subjects (selected randomly). All patients with type 1A diabetes were GAD antibody- and/or IA-2 antibody-positive. Patients with fulminant type 1 diabetes were diagnosed according to the inclusion criteria proposed by the Committee of the Japan Diabetes Society (JDS) 11. The value for HbA1c (%) is estimated as a National Glycohemoglobin Standardization Program (NGSP) equivalent value (%), calculated by the formula HbA1c (%) = HbA1c (JDS; %) + 0·4%, considering the relational expression of HbA1c (JDS; %) measured by the previous Japanese standard substance and measurement methods and HbA1c (NGSP) 30.

Table 1.

Patients demographics

	n	Gender (M/F)	Age (years)	GAD/IA-2 antibody positive (%)	HbA1c (%)	Duration (years)
Type 1A diabetes	20	5/15	48·6 ± 15·6	100/50 (n = 8)	8·15 ± 1·55	11·5 (1·3–38·0)
Fulminant type 1 diabetes	15	8/7	49·7 ± 16·0	7/0	7·84 ± 0·86	3·7 (0·0–9·6)
Type 2 diabetes	20	8/12	49·6 ± 9·6	0/n.d.	8·07 ± 1·43	n.d.
Healthy control subjects	30	10/20	49·5 ± 11·8	n.d./n.d.	n.d.

M/F: male/female; n.d.: not determined; GAD: glutamic acid decarboxylase.

This study was approved by the ethics committee of Osaka Medical College. Written informed consent was obtained from all patients.

Flow cytometric analysis

Fresh peripheral blood mononuclear cells (PBMCs) were obtained from whole blood by density gradient centrifugation (Lymphoprep; Axis-Shield PoC AS, Oslo, Norway) and subjected immediately to cellular staining. The cells were stained with the following antibodies: fluorescein isothiocyanate (FITC), anti-hCD45RA (clone HI100), phycoerythrin (PE), anti-hCD25 (M-A251), V450 anti-hCD25 (M-A251), V450 anti-hCCR4 (1G1), peridinin chlorophyll protein (PerCP), anti-CD4 (SK3), PE anti-hCTLA4 (BNI3; BD Bioscience, San Jose, CA, USA), allophycocyanin (APC) and anti-hFoxP3 (clone 236A/E7; eBioscience, San Diego, CA, USA). The following isotype control antibodies were used: FITC mouse immunoglobulin (Ig)G2b, κ (X27–35), PerCP mouse IgG1 (X40), PE mouse IgG1, κ (MOPC-21), PE mouse IgG2a, κ (G155-178), V450 mouse IgG1, κ (X40; BD Bioscience) and APC mouse IgG1, κ (MOCP-21; eBioscience). The cells were stained for 30 min (4°C) in the dark and washed twice with cold phosphate-buffered saline (PBS). Surface-stained cells then underwent methods for intracellular CTLA-4 or FoxP3 staining using an anti-human FoxP3 staining kit (eBioscience), according to the manufacturer's recommendations. Stained cells were then subjected to flow cytometric analysis using a BD FACSAria™ Cell Sorter (BD Bioscience). BD FACSDiva Software was used for analysis of the cytometric data. We defined the target cell populations based on approximately 99·5% of isotype controls. At least 30 000 events were acquired from each sample. We performed the analysis in a blinded manner, with no knowledge regarding the disease status of the sample.

Definition of activated T_reg, resting T_reg and non-suppressive T cells

As in the previous study 25, we defined activated T_reg (a-T_reg) as CD4⁺CD45RA⁻FoxP3^high or CD4⁺CD45RA⁻CD25⁺⁺⁺ T cells, resting T_reg (r-T_reg) as CD4⁺CD45RA⁺FoxP3^low or CD4⁺CD45RA⁺CD25⁺⁺ T cells and non-suppressive T cells (non-T_reg) as CD4⁺CD45RA⁻FoxP3^low or CD4⁺CD45RA⁻CD25⁺⁺ T cells (Fig. 1c,g).

Fig. 1.

Flow cytometric analysis of regulatory T cells (T_regs) from fresh peripheral blood. Representative plots showed one healthy control sample gated on lymphocytes and CD4⁺ T cells (a) showing isotype control (b) or CD45RA and intracellular staining for forkhead box protein 3 (FoxP3) (c) or FoxP3 (d) or CD25 and FoxP3 (e) or CD25 (f) or CD45RA and CD25 (g). Histograms showed the expression of cytotoxic T lymphocyte antigen-4 (CTLA-4) (h) and C-C chemokine receptor type 4 (CCR4) (i) by each fraction defined in (c); bottom; a-T_reg, r-T_reg and non-T_reg) from healthy control subjects.

T cell isolation

PBMCs were obtained from whole blood by density gradient centrifugation and were then stained immediately with FITC anti-hCD45RA, PE anti-hCD25 and PerCP anti-CD4. a-T_regs and r-T_regs were sorted and CD4⁺CD45RA⁻CD25⁻ fractions were sorted as responder cells with a BD FACSAria™ Cell Sorter (Fig. 1g). T cell-depleted accessory cells were isolated by negative selection from PBMCs using anti-CD3 microbeads (BD Bioscience), according to the manufacturer's recommendations; these cells were irradiated at 3000 rad.

Cell culture

T cells and accessory cells were cultured in RPMI-1640 medium (Gibco, Carlsbad, CA, UJSA) supplemented with L-glutamine, 25 mmol/l HEPES, 50 μg/ml penicillin/streptomycin (Gibco) and 5% human type antibody serum (Gemini Bio-Products, West Sacramento, CA, USA) in U-bottomed 96-well plates (Becton Dickinson, Franklin Lakes, NJ, USA).

Cell stimulation and suppression assay

Cell stimulation and suppression assays were performed by culturing sorted responders (1 × 10⁴/well) with a-T_regs at various ratios (1:0, 1:1/2, 1:1/4, 1:1/8 and 0:1/2), and with r-T_regs (1:1) in the presence of 5 × 10⁴ irradiated accessory cells. r-T_regs (1 × 10⁴/well) were also co-cultured with a-T_regs at a 1:1/2 ratio to evaluate that a-T_regs were able to suppress the proliferation of r-T_regs in a negative feedback manner. These co-cultures were stimulated using a combination of 5 μg/ml soluble anti-CD3 (clone HIT3a) and 5 μg/ml soluble anti-CD28 (clone CD28·2; BD Bioscience). All T cell culture conditions were tested in triplicate. The cells were cultured for 4 days, 100 μl of supernatant was removed from each well and 100 μl of fresh medium that contained 1 μCi of [³H]-thymidine was added for the final 16 h of culture to assess proliferation. The percentage suppression by a-T_regs was calculated as 100 − [mean counts per minute (cpm) of co-cultures − a-T_regs alone/mean cpm of responders alone] × 100 at a ratio of 1:1/2 (responder cells : a-T_regs).

Cytokine and sCTLA-4 determination

The IL-10 level was measured in the supernatants from suppression assay co-cultures (responder cells : a-T_regs = 1:1/2) using the Quantikine ELISA Kit (R&D Systems, Inc., Minneapolis, MN, USA), according to the manufacturer's instructions. The level of sCTLA-4 was measured using the Platinum ELISA Kit (eBioscience), according to the manufacturer's instructions.

Statistical analysis

The two-tailed unpaired Student's t-test and Mann–Whitney U-test were employed. Relationships between the proportion of a-T_regs among CD4⁺FoxP3⁺ T cells. P < 0·05 was considered to be significant.

Results

Frequency of a-T_regs, r-T_regs and non-T_regs

CD4⁺ T cells co-expressing CD45RA⁻ FoxP3^high, CD45RA⁺ FoxP3^low and CD45RA⁻FoxP3^low were counted to determine the frequency of three phenotypically different subpopulations (a-T_regs, r-T_regs and non-T_regs; Fig. 1a–c). a-T_regs expressed intracellular CTLA-4 and CCR4 to the highest degree among a-T_regs, r-T_regs and non-T_regs in healthy control subjects (Fig. 1h,i). The frequency of a-T_regs among CD4⁺ T cells was significantly higher in patients with type 1A diabetes (2·25 ± 0·68%, n = 20 versus 1·28 ± 0·48% in healthy control subjects, n = 30, P < 0·0001; versus 1·43 ± 0·59% in patients with fulminant type 1 diabetes, n = 15, P = 0·0008; versus 1·47 ± 0·51% in patients with type 2 diabetes, n = 20, P = 0·0002; Fig. 2a), whereas the frequency of r-T_regs and non-T_regs among CD4⁺ T cells did not differ among patients with type 1A diabetes (1·15 ± 0·73%, 2·52 ± 1·03%), patients with fulminant type 1 diabetes (1·81 ± 1·00%, 2·61 ± 0·83%), patients with type 2 diabetes (1·36 ± 0·64%, 2·94 ± 1·83%) and healthy control subjects (1·54 ± 1·00%, 2·32 ± 0·90%; Fig. 2b,c).

Fig. 2.

Variation in CD4⁺forkhead box protein 3 (FoxP3)⁺ T cell subpopulations and frequency of CD4⁺FoxP3⁺, CD4⁺CD25⁺FoxP3⁺ and CD4⁺CD25⁺ T cells. Graph shows the frequency of CD4⁺CD45RA⁻FoxP3^high activated regulatory T cells (T_regs) (a-T_regs; (a), CD4⁺CD45RA⁺FoxP3^low T_regs (r-T_regs; (b), CD4⁺CD45RA⁻FoxP3^low non-regulatory T cells (non-T_regs; (c), CD4⁺FoxP3⁺ T cells (d), CD4⁺CD25⁺FoxP3⁺ T cells (e) and CD4⁺CD25⁺ T cells (f) from healthy control subjects (HC; n = 30), patients with type 1A diabetes (T1AD; n = 20), fulminant type 1 diabetes (FT1D; n = 15) and type 2 diabetes (T2D; n = 20).

Frequency of CD4⁺ FoxP3⁺, CD4⁺CD25⁺ FoxP3⁺ T cells and CD4⁺CD25⁺ T cells

The frequency of FoxP3⁺ T cells and CD25⁺ FoxP3⁺ T cells among CD4⁺ T cells did not differ among patients with type 1A diabetes (5·79 ± 1·78%, 5·54 ± 1·59%, n = 20), patients with fulminant type 1 diabetes (5·75 ± 1·81%, 5·45 ± 1·73%, n = 15), patients with type 2 diabetes (5·82 ± 2·40%, 4·94 ± 1·78%, n = 20) and healthy control subjects (5·13 ± 1·66%, 5·36 ± 1·54%, n = 30; Figs 1d,e, 2d,e). The frequency of CD4⁺CD25⁺ T cells also did not differ among patients with type 1A diabetes (27·0 ± 10·9%), patients with fulminant type 1 diabetes (23·6 ± 8·3%), patients with type 2 diabetes (27·7 ± 11·1%) and healthy control subjects (24·9 ± 6·5%; Figs 1f, 2f).

Proportion of a-T_regs, r-T_regs and non-T_regs among CD4⁺ FoxP3⁺ T cells

Although the frequency of a-T_regs among CD4⁺ T cells was significantly higher in patients with type 1A diabetes in comparison to the other three groups (versus healthy control subjects, fulminant type 1 diabetes and type 2 diabetes), the total number of CD4⁺FoxP3⁺ T cells did not differ. We further attempted to determine the proportion of three subpopulations (CD45RA⁻ FoxP3^high a-T_regs, CD45RA⁺FoxP3^low r-T_regs and CD45RA⁻ FoxP3^low non-T_regs) among the CD4⁺FoxP3⁺ T cells. The proportion of a-T_regs among CD4⁺FoxP3⁺ T cells was significantly higher in patients with type 1A diabetes (38·9 ± 8·9% versus 25·9 ± 8·2% in healthy control subjects; P < 0·0001 versus 25·3 ± 8·5% in patients with fulminant type 1 diabetes; P < 0·0001 versus 26·9 ± 8·3% in patients with type 2 diabetes; P < 0·0001), whereas that of r-T_regs among CD4⁺FoxP3⁺ T cells was significantly lower in patients with type 1A diabetes (19·2 ± 9·5% versus 28·9 ± 11·9% in healthy control subjects, P = 0·0039; versus 29·9 ± 10·1% in patients with fulminant type 1 diabetes, P = 0·0030; Fig. 3a). No significant differences in the proportion of non-T_regs among CD4⁺FoxP3⁺ T cells were observed among the four groups (healthy control subjects versus type 1A diabetes versus fulminant type 1 diabetes versus type 2 diabetes; Fig. 3a).

Fig. 3.

Proportion of activated regulatory T cells (a-T_regs) T_regs, resting T_regs (r-T_regs) and non-T_regs among CD4⁺FoxP3⁺ T cells. (a) Pie charts showed the proportion of a-T_regs, r-T_regs and non-T_regs among CD4⁺forkhead box protein 3 (FoxP3)⁺ T cells from healthy controls (HC) (n = 30), patients with type 1A diabetes (T1AD) (n = 20), patients with fulminant type 1 diabetes (FT1D) (n = 15) and patients with T2D (n = 20). (b) The graph shows the proportion of a-T_regs among CD4⁺FoxP3⁺ T cells when we divided 20 patients with T1AD into two groups (serum C-peptide level < 0·01 ng/dl; T1AD without residual insulin-secreting capacity, n = 8 and ≥0·01 ng/dl; T1AD with residual insulin-secreting capacity, n = 12). *P < 0·05; **P < 0·001; ***P < 0·0001. Bars represent the mean ± standard error.

Higher proportion of a-T_regs among CD4⁺FoxP3⁺ T cells in patients with type 1A diabetes with residual insulin-secreting capacity

Twenty patients with type 1A diabetes in this study were divided into two groups (serum C-peptide level < 0·01 ng/dl, type 1A diabetes without residual insulin-secreting capacity, n = 8 and ≥0·01 ng/dl, type 1A diabetes with residual insulin-secreting capacity, n = 12). The proportion of a-T_regs among CD4⁺FoxP3⁺ T cells was significantly higher in patients with type 1A diabetes with residual insulin-secreting capacity (43·3 ± 5·0%, n = 12 versus 32·3 ± 9·7% in type 1A diabetes without residual insulin-secreting capacity, n = 8, P < 0·01; versus 25·9 ± 8·2%, n = 30 in healthy control subjects, P < 0·0001; versus 25·3 ± 8·5%, n = 20 in patients with fulminant type 1 diabetes, P < 0. 0001; versus 26·9 ± 8·3%, n = 20 in patients with type 2 diabetes, P < 0·0001) but not in patients with type 1A diabetes without residual insulin-secreting capacity (Fig. 3b).

An evaluation of the correlation between the proportion of a-T_regs among CD4⁺ FoxP3⁺ T cells and the patient demographics (age, gender, disease duration, HbA1c level, GAD or IA-2 antibody level) in 20 patients with type 1A diabetes revealed that there was no significant correlation with the age, gender, disease duration, HbA1c level, GAD or IA-2 antibody level.

Suppressive function of a-T_regs

Responder cells were co-cultured with a-T_regs at various ratios (1:0, 1:1/2, 1:1/4, 1:1/8 and 0:1/2), to assess the function of a-T_regs among the three groups (healthy control subjects, type 1A diabetes and fulminant type 1 diabetes). The responder cells isolated from the three groups proliferated well, whereas a-T_regs isolated from the same subject groups were relatively anergic (Fig. 4a,d). Among three groups, responder cell proliferation in type 1A diabetes tended to decrease, but not significantly (Fig. 4a). Responder cell proliferation was suppressed in a dose-dependent manner by a-T_regs in each group (healthy control subjects, type 1A diabetes and fulminant type 1 diabetes; Fig. 4b). There were significant defects in the suppression of target responder cell proliferation by a-T_regs (responder cells : a-T_regs = 1:1/2) both from patients with fulminant type 1 diabetes (median: 70·7%, range: 22·0–97·6%, n = 10 versus 88·1%, 57·9–98·1%, n = 10 in healthy control subjects, P = 0·0233) and type 1A diabetes (median: 63·9%, range: −0·7–77·8%, n = 10 versus healthy control subjects, P = 0·0019; Fig. 4c). There was no correlation between suppressive function and C-peptide level either within any of the groups or in total (data not shown). r-T_regs proliferated well by themselves and suppressed the proliferation of responder cells when they were co-cultured with r-T_regs at a 1:1 ratio (Fig. 4d) in healthy control subjects (n = 8). Further, a-T_regs suppressed the proliferation of r-T_regs in healthy control subjects (n = 3) when r-T_regs were co-cultured with a-T_regs at a 1:1/2 ratio (data not shown).

Fig. 4.

Suppressive function of activated regulatory T cells (a-T_regs) and resting T_regs (r-T_regs). (a) The graph shows the proliferation ([³H]-thymidine incorporation) in separated cell populations and co-culture (responder cell : a-T_reg = 1:0, 1:1/2 and 0:1/2) from healthy controls (HC) (open bars; n = 10), patients with type 1A diabetes (T1AD) (grey bars; n = 10) and patients with fulminant type 1 diabetes (FT1D) (closed bars; n = 10). Bars represent the mean ± standard error (s.e.). (b) Representative percentage suppression of responder cells (1 × 10⁴) by a-T_regs from HC, patients with T1AD and patients with FT1D in the presence of various ratios of a-T_regs (responder cell : a-T_reg = 1:1/2, 1:1/4, 1:1/8). Bars represent the mean ± s.e. (c) Suppressive function of a-T_regs co-cultured with responder cells at a 1:1/2 ratio. Date plotted represent the percentage suppression of proliferation by a-T_regs from HC (open circles; n = 10), patients with T1AD (grey circles; n = 10) and patients with FT1D (closed circles; n = 10). Bars represent the median. (d) Proliferation in separated cell populations and co-culture (responder cell : a-T_reg = 1:0, 1:1/2, 1:1/4, 1:1/8 and 0:1/2; open boxes) and the ability to proliferate and suppress the proliferation of responder cells in r-T_regs from HC (responder cell : r-T_reg = 1:0, 1:1 and 0:1; grey boxes; n = 8). The arrow means ability to suppress the proliferation of responder cells in r-T_regs. Bars represent the mean ± s.e. Graph shows the level of interleukin (IL)-10 (e) and soluble, cytotoxic T lymphocyte antigen 4 (sCTLA-4) (f) in supernatants from suppression assay co-cultures (responder cell : a-T_reg = 1:1/2) from HC (open circles), patients with T1AD (grey circles) and patients with FT1D (closed circles). Bars represent the median.

The level of IL-10 and sCTLA-4 in co-cultures

Measuring the levels of IL-10 and sCTLA-4 in supernatants from suppression assay co-cultures (responder cells: a-T_regs = 1:1/2) revealed that the level of IL-10 did not differ among three groups (healthy control subjects, type 1A diabetes and fulminant type 1 diabetes; Fig. 4e), but the level of sCTLA-4 was increased significantly in patients with type 1A diabetes (median: 0·525 ng/ml, range: 0·067–1·721 ng/ml, n = 9 versus 0·076 ng/ml, 0·000–0·300 ng/ml, n = 8 in healthy control subjects, P = 0·0014, versus 0·018 ng/ml, 0·000–0·340 ng/ml, n = 9 in fulminant type 1 diabetes, P = 0·0009; Fig. 4f).

Discussion

FoxP3⁺ regulatory T cells could be divided into three different subpopulations, a-T_regs, r-T_regs and non-T_regs 26–29. Among them, a-T_regs play a most important role as a suppressor for immune reactions. The present study showed that the frequency of a-T_regs in the peripheral blood was higher in type 1A diabetes but not in fulminant type 1 diabetes than that in healthy control subjects and type 2 diabetes. This study also revealed that a-T_regs from patients with fulminant type 1 diabetes and type 1A diabetes had a lower ability to suppress the responder cells in comparison to healthy control subjects. The concomitant presence of lower ability of a-T_regs and of the higher prevalence of a-T_regs in the peripheral blood could be explained by a-T_regs being increased in the peripheral blood to compensate for their decreased ability of suppression in patients with type 1A diabetes.

This study evaluated the suppressive capacity of T_regs by using neither total CD4⁺CD25^high nor CD4⁺CD25⁺CD127^low T cells but CD4⁺CD45RA⁻CD25⁺⁺⁺, a-T_regs, and found that a-T_regs are functionally deficient both in fulminant type 1 diabetes and in type 1A diabetes. Isolating a-T_regs without contamination of r-T_regs and non-T_regs enabled a more precise evaluation of T_reg function than before. CD4⁺CD25^high, CD4⁺CD25⁺FoxP3⁺ and CD4⁺CD25⁺CD127^low T cells have been used as T_regs in the previous studies 19–25, but these cells could contain r-T_regs and non-T_regs at various ratios. The contamination of r-T_regs and non-T_regs in isolated CD4⁺CD25^high and CD4⁺CD25⁺CD127^low T cells may well cause discrepancies in the results of T_reg function 19,20–25. Indeed, isolated a-T_regs were relatively anergic and expressed intracellular CTLA-4 and CCR4 to the highest degree among a-T_regs, r-T_regs and non-T_regs, whereas r-T_regs proliferated well by themselves and hardly expressed CTLA-4. In this study, some of these r-T_regs converted to a-T_regs 26,27 and suppressed the proliferation of responder cells. These results indicate that, at present, a combination of CD25 and CD45RA is the best marker for purifying human FoxP3⁺ T_regs as a-T_regs and r-T_regs.

The mechanism of suppressive function by a-T_regs has not been clarified fully, but the current study showed no differences in the level of IL-10 production among three groups, indicating that the functional defects in a-T_regs is not due to the defect in IL-10. T_regs appear to suppress the effector T cells (T_effs) directly, independent of IL-10 31. In most models, direct cell–cell interaction between T_regs and T_effs was also found necessary to suppress immune responses 32. Conversely, the level of sCTLA-4 on supernatants from suppressor assay in patients with type 1A diabetes was increased significantly in comparison to that of fulminant type 1 diabetes and healthy control subjects. This increase in the level of sCTLA-4 might be a result of a compensation mechanism for the functional defects of a-T_regs in type 1A diabetes. sCTLA-4 expression induces suppression directly in the T_reg function, and thereby modulates disease risk in a mouse model 33. The suppressive function in patients with type 1A diabetes and fulminant type 1 diabetes had no correlation with their C-peptide level, indicating that the impaired function in a-T_regs is independent of residual insulin-secreting capacity. Further studies, such as the IL-2 signalling pathway in a-T_regs and the resistance of responder cells for a-T_reg suppression, might help to clarify the impaired function in a-T_regs.

The frequency of a-T_regs in the peripheral blood was higher in type 1A diabetes, but not in fulminant type 1 diabetes, than that in healthy control subjects and type 2 diabetes. The proportion of a-T_regs among CD4⁺FoxP3⁺ T cells was higher in patients with type 1A diabetes with the residual insulin-secreting capacity, but not in those without it. The existence of the residual insulin-secreting capacity implies existence of the remaining β cells. Insulin in β cells could be a key autoantigen in the development of type 1A diabetes 3,4, thus indicating that the immune reaction to destroy β cells would be stronger in patients with a residual insulin-secreting capacity than in those without it 34,35. Consequently, the frequency of a-T_regs might increase to compensate for the impaired function of a-T_regs in type 1A diabetes to protect β cells. In other words, r-T_regs would be converted into a-T_regs based on the residual β cell antigens, such as insulin or GAD65. As a result, the proportion of a-T_regs among CD4⁺FoxP3⁺ T cells was significantly higher whereas, conversely, that of r-T_regs tended to be low in patients with type 1A diabetes. The increased frequency of a-T_regs in type 1A diabetes is not explained by changes in glucose metabolism, because that in type 2 diabetes was not associated with changes in a-T_regs.

Despite functional defects in a-T_regs, there was no accelerated conversion from r-T_regs to a-T_regs in patients with fulminant type 1 diabetes, which was different from type 1A diabetes with the remaining insulin-secreting capacity. Further, the level of sCTLA-4 was not higher in patients with fulminant type 1 diabetes. These differences between fulminant type 1 diabetes and type 1A diabetes might be due to the fact that β cells are almost completely destroyed in fulminant type 1 diabetes, and therefore the target antigen does not exist, nor do the compensation mechanisms such as an increase in the frequency of a-T_regs or in sCTLA-4 levels. Low CTLA-4 expression could be involved, at least in part, in the mechanism of functional defects in a-T_regs in patients with fulminant type 1 diabetes 14.

There are some limitations in this study. First, the methylation status of the FoxP3, i.e. the T_reg-specific demethylated region (TSDR), was not analysed, while the marker profiles used are established within the literature 26–29. Secondly, [³H]-thymidine was used to measure cell proliferation. This method does not allow for the discrimination of cell proliferation of T_regs, responder T populations or analysis of cell death during culture. Thirdly, we did not match the human leucocyte antigen (HLA) types between the different groups. This might have affected the frequency and function of a-T_regs.

In conclusion, CD4⁺CD45RA⁻FoxP3^high a-T_regs were functionally impaired both in fulminant type 1 and type 1A diabetes, related to residual insulin-secreting capacity, and may be associated with the development of type 1 diabetes. These findings could provide new insight into the dynamics of CD4⁺FoxP3⁺ T cells and their role in the pathogenesis of type 1 diabetes.

Disclosure

The authors declare that there are no conflicts of interest.