Increased expression of regulatory T cell-associated markers in recent-onset diabetic children

CD4CD25 T cells are thought to be crucial for the maintenance of immunological tolerance to self antigens. In this study, we investigated the frequencies of these cells in the early stage of type 1 diabetes, as well as in a setting of possible pre-diabetic autoimmunity. Hence, the expression of FOXP3, CTLA-4, and CD27 in CD4 CD25 T cells was analyzed using flow cytometry in 14 patients with recent onset type 1 diabetes, in 9 at-risk individuals, and 9 healthy individuals with no known risk for type 1 diabetes. Our results show there were no differences in the frequency of CD4CD25 cells between groups. However, compared to controls, recentonset type 1 diabetic patients had higher expression of FOXP3, CTLA-4, and CD27 in CD4 CD25 cells from peripheral blood. The median fluorescence intensity of FOXP3 was significantly higher in CD4CD25 cells from patients with type 1 diabetes than from controls. Furthermore, a positive correlation between the frequency of FOXP3 cells and the median fluorescence intensity of FOXP3 was observed among patients with type 1 diabetes. These data suggest that the frequency of CD4CD25 FOXP3 T cells in the periphery is not decreased but rather increased at onset of type 1 diabetes. Thus, functional deficiencies rather than reduced numbers of CD4CD25 cells could contribute to the development of type 1 diabetes.


INTRODUCTION
There is little doubt today that a regulatory subset of T cells necessary for peripheral tolerance exists, and that absence of these cells causes autoimmunity in a variety of experimental settings [1].Naturally occurring regulatory T cells (Treg) are CD4 + T cells which constitutively express the Interleukin (IL)-2 receptor α chain (CD25), the transcription factor FOXP3 and Cytotoxic T lymphocyte associated antigen 4 (CTLA-4, CD152), and are normally produced in the thymus [2].Expression of the transcription factor FOXP3 is vital to the development and function of Treg [3][4] and has therefore been used to delineate regulatory T cell populations.However, activation of human T cells induces transient expression of FOXP3 in non-regulatory T cells without conferring a regulatory phenotype to the affected cells [5][6][7][8][9].Some report this phenomenon to be a consequence of T cell receptor (TCR) stimulation [7], while others have postulated that TCR stimulation does not produce FOXP3 expression at either gene or protein level [10].Tran et al recently found that high levels of FOXP3 could be induced in CD4 + CD25 -T cells by TCR stimulation, only in the presence of transforming growth factor (TGF)-β.FOXP3 expression induced in this way was maintained for weeks in the presence of IL-2 [9].CTLA-4 is a costimulatory molecule with potent suppressive function constitutively expressed on Treg but also on activated effector T cells [11,12].In addition to its negative co-stimulatory effect, CTLA-4 up-regulates Indolamine 2,3-dioxygenase in dendritic cells [13], resulting in catabolism of tryptophan to kynurine, which has potent local immunosuppressive effects [14].Expression of the transmembrane costimulatory receptor CD27 was used to define Treg in inflamed synovia in conjunction with CD25 [15].Other evidence suggests that Treg expressing CD27 are more suppressive than CD27-negative counterparts [16].
Defects in the function of Tregs have been hypothesized to be involved in the pathogenesis of numerous autoimmune diseases, including type 1 diabetes [17].In mice, islet antigen specific FOXP3 transduced T cells were able to suppress recent onset type 1 diabetes [18].However, studies of human Tregs in type 1 diabetes have produced contradictory results.Kukreja et al. reported reduced numbers of CD4 + CD25 hi cells [19], while Lindley et al. and Brusko et al. could find no change in CD4 + CD25 hi frequency between healthy and diabetic individuals [20,21].A meta-analysis comparing these findings suggested that the lack of consensus between studies is a consequence of differently matched diabetic and control groups.The authors advocated further studies of phenotypical markers associated with Treg, including FOXP3 [22].
A conundrum of type 1 diabetes is the lack of good indicators before disease onset, since it is of interest to study individuals at risk of developing the disease.Thus, we are interested in describing T cell populations in at-risk individuals, with possible autoimmune activity prior to diagnosis of type 1 diabetes.Most people progressing to type 1 diabetes produce autoantibodies to one or more islet autoantigens, most commonly against insulin, glutamic acid decarboxylase (GAD), and the tyrosine phosphatase like protein IA-2 [23].Therefore, a group of healthy children with autoantibodies was selected as a risk population in this study.To our knowledge, no previous studies have examined the expression of FOXP3 and CTLA-4 in autoantibody-positive children, nor in children with recent onset T1D with as closely age-matched healthy controls as in the present study.This is important as the immune system changes with age from childhood, through puberty until adulthood, and as the autoimmune process leading to type 1 diabetes is more rapid and aggressive in children than in higher age groups.
We hypothesized that patients with type 1 diabetes as well as healthy individuals expressing high levels of autoantibodies against islet antigens would have a decreased proportion of Treg compared to healthy individuals without autoantibodies.Therefore we analyzed the expression of proteins related to Treg function to determine their frequency in type 1 diabetes patients, healthy subjects and healthy subjects at risk of type 1 diabetes.Here, we report an increased expression of FOXP3 and CTLA-4 on CD4 + CD25 + cells in patients with type 1 diabetes and subjects at risk of type 1 diabetes.

Study Population
All the participants and their parents received written information on the study, and consent was obtained according the Declaration of Helsinki.Via proxy consent was obtained for participating children, and presumed consent was obtained from parents upon completing and submitting questionnaries on entering the study.
Venous blood samples were collected from 14 patients with recent onset type 1 diabetes with a median age of 10 years (range 3 -17 years, SEM 1,123) and a diabetes duration of three months.Nine 8-year-old healthy children with autoantibodies against Insulin, GAD65, or IA-2 in the 95 th percentile or higher at 5 and/or 2.5 years of age were included as a risk population.Some, but not all, at-risk children exhibited an antibody response to more than one autoantigen (Table 1).And some, but not all, were positive for autoantibodies on several occasions a few years apart prior to sampling.Nine healthy children, 8 years old, with no known type 1 diabetes-associated HLA-genotypes, allergy, or autoimmune disease were included as reference.All at-risk and control subjects were participating in the ABIS study (All Babies in Southeast Sweden).
Samples were acquired on a four-color BD FAC-SCalibur flow cytometer.The cytometer was calibrated daily using BD Calibrite 3 beads, with added APC beads (BD Biosciences).Compensation was set manually and gates were set subjectively (Figures 1(a)-(c)).Analysis was performed using Cellquest Pro software (BD Biosciences).All analyses were performed in a blinded manner, the evaluator did not know the identity of the sample.CD4 + CD25 hi cells were defined by first gating on small lymphocytes by forward and side scatter, and then on CD4 and high CD25 expression [24].The CD25 hi gate was adjusted to contain CD4 + cells that expressed higher levels of CD25 than the discrete population of CD4 - cells [25].This gate contained approximately 2% of small lymphocytes, and 2% -6% of CD4 + cells.Approximately 5 × 10 5 small lymphocytes were collected from each tube, while approximately 10 4 cells were acquired from tubes with unstained cells, cells stained with isotype controls, and tubes used to set compensation.

Statistical Analysis
Some, but not all, of our material was normally distributed according to the D'Agostino & Pearson omnibus normality test.We decided to use non-parametric tests based partly on this and because of the small size of our population.Analysis of variance between groups was performed for each parameter using the Kruskal-Wallis test, followed by Dunn's multiple comparison test.Selected groups were compared using the Mann-Whitney U-test; two-tailed P-values were obtained throughout.Statistical testing was carried out with GraphPad Prism 5.01 and SPSS 14.0, both for Windows.

RESULTS
Expression of Treg associated markers in CD4 + CD25 hi T cells is more frequent among at-risk and recent onset type 1 diabetic children than among controls.
To determine the frequency of cells with a Treg phenotype, we compared the percentages of CD4 + CD25 hi cells expressing FOXP3, CTLA-4, and CD27 determined by FACS.Lymphocytes were thus gated based on forward and side scatter (Figure 1(a)), followed by CD4 and CD25 expression (Figure 1(b)).Quadrant lines demarcate the isotype controls.Gated cells were analyzed for expression of FOXP3, CTLA-4, and CD27 (Figure 1(c)).Results are expressed as percentages of cells positively stained for each molecule.For clarity, no-risk individuals will be termed controls.
Approximately 2% small lymphocytes and 2% -6% CD4 + lymphocytes were gated as CD4 + CD25 hi .The groups were not significantly different in their frequencies of CD4 + CD25 hi cells (Figure 1(a)).The percentage of FOXP3-expressing CD4 + CD25 hi cells was significantly higher in diabetic children compared to controls (p = 0.0061, Figure 1(b)), and tended to be higher in diabetic children compared to at-risk children (p = 0.0518, Figure 1(b)).CD4 + CD25 hi cells from at-risk and diabetic children more frequently expressed CTLA-4 than cells from control children (p = 0.0078, p = 0.0006, Figure 1(c)).CD27 was also more frequently expressed in CD4 + CD25 hi cells among diabetic children compared to controls (p = 0.0092, Figure 1(d)).In addition, we analyzed the frequencies of FOXP3 + CTLA-4 + CD4 + CD25 hi  cells.In concordance with separately determined FOXP3 and CTLA-4 levels, co-expression on single cells was higher among children with type 1 diabetes than controls (p = 0.0018, Figure 1(e)).Single cells expressing both FOXP3 and CD27 were common among CD4 + CD25 hi cells regardless of group, as were CTLA-4 + CD27 + cells.As the majority of CD4 + CD25 hi cells expressed CD27, the frequencies of FOX P3 + CD27 + cells are essentially the same as those of FOXP3 single positive cells.Hence, both FOXP3 + CD27 + and CTLA-4 + CD27 + co-expression was significantly higher in CD4 + CD25 hi cells from diabetic chil-61 dren compared to controls (p = 0.0021, Figure 1(f), p = 0.0006, Figure 1(g)).Recent onset diabetic children had a higher percentage of FOXP3 + CD27 + coexpressing CD 4 + CD25 hi cells than at-risk children (p = 0.0033), whereas at-risk children more frequently co-expressed CTLA-4 + CD27 + than no-risk controls (p = 0.0315).While the at-risk population did not always differ significantly from the control and diabetic groups, a clear trend is discernable from the graphical presentation of our findings, where the at-risk group has higher expression of most analyzed markers than controls, and lower expression than recent onset type 1 diabetic patients.Even though the most distinct variations in FOXP3 and CTL A-4 expression were present within the CD4 + CD25 hi subset, the groups remain significantly different when analyzing CD4 + cells in general (data not shown).
The FOXP3 Median Fluorescence Intensity (MFI) of CD4 + CD25 hi FOXP3 + cells is higher among recent onset Type 1 diabetic children than among controls.The median fluorescence intensity for FOXP3 PE in CD4 + CD25 hi FOXP3 + cells was higher in diabetic children than in controls (p = 0.0061, Figure 2(a)).Intriguingly, CD4 + CD25 hi FOXP3 + cells had higher FOXP3 MFI than CD4 + cells among children with type 1 diabetes (p = 0.0007, Figure 2(b)).The same pattern was evident in children with risk for type 1 diabetes (p < 0.0001).In contrast, no difference in the FOXP3 MFI was detected between CD4 + cells and CD4 + CD25 hi- FOXP3 + cells in the control group.Furthermore, there is a distinct correlation between the frequency of CD4 + CD25 hi FOXP3 + cells and the MFI of FOXP3 among CD4 + CD25 hi FOX P3 + cells (Figure 2(c)).This correlation is found exclusively in the diabetic population when the groups are examined independently.CD27 cannot define CD4 + CD25 hi cells in peripheral blood.
To explore whether CD27 could define CD4 + CD25 hi cells in peripheral blood, the expression of this receptor was analyzed.CD27 was commonly expressed on more than 95% of CD4 + cells.Further, CD27 expression did not differ between the CD4 + CD25 hi and the CD4 + population, regardless of the autoimmune state of the individual, defined as diagnosed type 1 diabetes or the pres-ence of autoantibodies.

DISCUSSION
In this study we found that the percentages of CD4 + CD25 hi cells expressing FOXP3 and CTLA-4 was higher in the peripheral blood of diabetic and at-risk children compared to healthy individuals.It has been argued that the age of the control population relative to the studied population may affect the outcome of such comparisons [21].In the present study, both at-risk and control subjects were 8-year-old children, and the recent-onset type 1 diabetic patients were only slightly older (median 10 years).Thus, the differences between at risk and control groups cannot be explained by differences in age.
In agreement with most of the studies including adults [20,21,26], we could not detect a difference in the frequency of CD4 + CD25 hi cells.One previous publication on the subject has reported reduced frequencies of CD4 + CD25 + cells in children with type 1 diabetes, but the control population was considerably older than the diabetic population [19].Brusko et al. found no changes in the frequency of CD4 + CD25 + FOXP3 regulatory cells in type 1 diabetics [27].However, the samples from first degree relatives and healthy controls were from individuals considerably older than the diabetic children.Lawson et al. has also reported no difference in the percentages of CD4+CD25 hi cells co-expressing FOXP3, but in contrast to our study the patients had long-standing diabetes, and both patients and controls were adult subjects [28].
It has been shown recently that the effector cells of diabetic subjects are resistant to regulation via Treg, and that this resistance is intrinsic to the effector population [29].Thus, the increased level of FOXP3 expression we detect might be due to a resistance to Treg-mediated suppression in effector T cells.Marwaha et al. recently demonstrated that recent-onset type 1 diabetes patients have a higher percentage of non-suppressive CD45RA-CD25 int FOXP3 low cells that secrete IL-17 [30].Since CD 45RA was not included as a marker in the present study, it cannot be excluded that the increased percentage of FOXP3-expressing cells presented here may represent a population of non-suppressive cells.
Our study also showed higher frequencies of CD4 + CD25 hi cells expressing intracellular CTLA-4 among recent-onset type 1 diabetes children.This is in agreement with a previous result from a study in type 1 diabetic adults [20].In addition we observed that children with risk for type 1 diabetes also had increased percentages of CD4 + CD25 hi expressing CTLA-4.CTLA-4 is constitutively expressed by Treg and has been linked to Treg function in vitro [12,31,32].However, sharply contrasting results indicate that CTLA-4 blockade does not alter the ability of Treg to suppress proliferation of responder T cells [24].FOXP3 + Treg capable of in vitro suppression are present in CTLA-4 deficient mice, which further questions the role of CTLA-4 in the mechanism of Treg suppression [33].Our results indicate that it is unlikely that a lack of CTLA-4 is a causative factor in type 1 diabetes development in children.
CD27 has previously been suggested to define Treg in combination with CD25 [15].It has also been reported that CD27 expression correlates with FOXP3 expression in peripheral blood of patients with relapsing-remitting multiple sclerosis [34].In the present study, expression of CD27 did not vary noticeably between CD4 + and CD4 + CD25 hi cells.It has been shown that 80% of CD4 + CD25 int cells from synovial fluid expressed CD27, arguing against the use of CD27 as a marker to define Treg, since it is unlikely that as many as 80% of CD4 + CD25 int cells are Treg [35].Thus, our results indicate that CD27 is not suitable to define Treg in peripheral blood from children with type 1 diabetes.The median fluorescence intensity of FOXP3 has been shown to be higher in Treg than in effector T cells [8].Thus, increased FOXP3 expression in children with type 1 diabetes might represent Treg and not be due to activation of effector T cells.Both diabetic and at-risk children exhibited higher FOXP3 MFI in CD4 + CD25 hi cells compared to CD4 + , whereas controls did not.This could represent a higher fraction of CD4 + CD25 hi cells with a regulatory phenotype in the at-risk and diabetic groups compared to controls.Finally, we found a correlation between the frequency of FOXP3 + cells and FOXP3 MFI among patients with type 1 diabetes but not in at-risk or healthy children.This is in agreement with a previous study where the authors observed a correlation between FOXP3 + cell frequency and FOXP3 MFI in patients with multiple sclerosis [34].They further demonstrated that the suppressive capacity of CD4 + CD 25 hi FOXP3 + cells correlates with the MFI of FOXP3 in-vitro.Thus, it cannot be excluded that an abundant population of functional Treg exists in the peripheral blood of children with type 1 diabetes.
In conclusion, the result of our study provide evidence of an altered frequency of cells with a regulatory T cell phenotype in peripheral blood of children with type 1 diabetes and in children with risk for developing the disease.It will be important to clarify whether the increased frequency of CD4CD25hiFOXP3 cells is a failed attempt at controlling autoimmunity.
shows the frequency of CD25 high cells among CD4 + cells.The percentages of CD4 + CD25 hi cells expressing FOXP3, CTLA-4, and CD27 are depicted in (b), (c), and (d), respectively.The percentage of CD4 + CD25 hi cells coexpressing both FOXP3 and CTLA-4 is given in (e).(f) and (g) show percentages of CD4 + CD25 hi cells that are CD27 + FOXP3 + and CD27 + CTLA-4 + , respectively.Ns = not significant.* = p < 0.05, ** = p < 0.01, *** = p < 0.001.Significance was determined by Kruskal-Wallis test followed by Dunn's post test.A clear trend of increasing Treg-associated molecules with increasing autoimmune activity is discernable throughout.The manual setting of the lymphocyte gate is shown in H, with 5% of all events shown in the plot.In I, the manually set CD4 + CD25 hi gate is depicted.Thirty-three percent of lymphocyte-gated events are shown, and the gate is set according to where the expression of CD25 in the CD4-negative population becomes scarce.J shows a typical plot of CD27 FITC (x-axis) and FOXP3 PE (y-axis); all collected events in the CD4 + CD25 hi gate are shown.All plots show permeabilized cells.Isotype staining is compared to FOXP3 stained cells in K.

Figure 1 .
Figure 1.Differential expression of FOXP3 and CTLA-4 among healthy, at risk and recent onset diabetic children.
illustrates the FOXP3 MFI of CD4 + CD25 hi cells, whereas (b) shows FOXP3 MFI between the CD4 + and CD4 + CD25 hi populations in each group.(c) shows the correlation of CD4 + CD25 hi FOXP3 + cell frequency and FOXP3 MFI in CD4 + CD25 hi cells in a mixed population and among healthy, at risk and diabetic children separately.r and p values were calculated using Spearman's correlation test.

Table 1 .
Presence of autoantibodies in at-risk children.