High Frequency of Vα24⁺ Vβ11⁺ T-Cells Observed in Type 1 Diabetes

Peripheral blood samples were obtained from 164 Japanese diabetic patients and 67 healthy subjects (39 men, 28 women; aged 40.5 ± 14.3 years) with informed consent for use in this study. The diabetic patients were classified into three categories: classic type 1 diabetes (n = 78; 41 men, 37 women; aged 40.6 ± 17.1 years; disease duration 4.0 ± 4.5 years; HbA_1c 8.4 ± 1.9%), LADA (27) (n = 26; 19 men, 7 women; aged 51.3 ± 12.1 years; disease duration 6.7 ± 5.6 years; HbA_1c 8.4 ± 1.9%), and type 2 diabetes (n = 60; 33 men, 27 women; aged 61.3 ± 10.1 years; disease duration 9.8 ± 6.7 years; HbA_1c 7.1 ± 1.4%). The diagnosis of classic type 1 diabetes was based on the criteria of the American Diabetes Association for type 1 diabetes, with pancreatic β-cell destruction as the primary cause of the disorder and a tendency toward ketoacidosis (28,29), irrespective of islet-associated autoantibody positivity. Briefly, type 1 diabetic patients were diagnosed on the basis of classic presentation of acute abrupt onset of symptoms (polyuria, polydipsia, body weight loss, and so on) with diabetic ketosis (or ketoacidosis) due to both sustained hyperglycemia and a defect of residual insulin secretion; in fact, serum C-peptide levels were 0.35 ± 0.07 ng/ml in classic type 1 diabetes near the time of diagnosis in this study. Moreover, we measured glutamic acid decarboxylase antibody (GADA) and/or insulinoma-associated protein-2 antibody (IA2A) as islet-associated autoantibodies, as previously described (30), and if either of the antibodies was detected, the patient was classified as islet-associated autoantibody positive. Therefore, classic type 1 diabetes consists of both islet-associated autoantibody-positive patients (Ab⁺; n = 46) and -negative patients (Ab⁻; n = 32). LADA patients were defined as adult diabetic patients who were positive for GADA and/or IA2A and who did not require insulin therapy for at least 12 months after the diagnosis of diabetes but who ultimately progressed to insulin dependence (27).

Heparinized peripheral blood (100 μl) was placed in 5-ml polystyrene round-bottom tubes (Becton Dickinson, Franklin Lakes, NJ). Each sample was incubated with 3 μl human phycoerythrin-cyanine 5 (PC5)-conjugated CD3 antibody (UCHT1, mouse IgG1; Immunotech, Marseille, France), 4 μl human fluorescein isothiocyanate (FITC)-conjugated Vα24 antibody (C15, mouse IgG1; Immunotech), and 4 μl human phycoerythrin (PE)-conjugated Vβ11 antibody (C21, mouse IgG2a; Immunotech) simultaneously on ice for 20 min. To exclude the possibility of nonspecific staining in this system, mouse IgG1-FITC (679.1Mc7; Immunotech) and mouse IgG2a-PE (U7.27; Immunotech) were used as isotype controls. After lysing erythrocytes, the samples were vortexed gently and incubated at room temperature for another 10 min. After centrifuging the tube at 450g for 5 min, the supernatant was removed, and the cells were washed with 0.1% BSA-PBS. The prepared cells were analyzed using a three-color flow cytometer, Epics Altra (Coulter). For enumeration of Vα24⁺ Vβ11⁺ CD3⁺ triple-positive cells, at least 100,000 events were analyzed. After the lymphocytes were gated by forward scatter and side scatter, at least 25,000 counts of CD3⁺ cells, which were identified by gating cells stained with human CD3 antibody-PC5, were analyzed. Then, we counted the number of Vα24⁺ Vβ11⁺ CD3⁺ triple-positive cells. The frequency of Vα24⁺ Vβ11⁺ CD3⁺ triple-positive cells was shown as a percentage of CD3⁺ cells analyzed.

Results are presented as means ± SE. Differences in Vα24⁺ Vβ11⁺ T-cell frequency were analyzed using the Bonferroni/Dunn test or Mann-whitney U test for nonparametric unpaired observations. Differences in patient age were analyzed using the Kruskal-Wallis test for nonparametric unpaired observations. The correlation between Vα24⁺ Vβ11⁺ T-cell frequency and disease duration (or age) was analyzed by Spearman’s rank-order correlation test. Multiple regression analysis was used to evaluate the association among Vα24⁺ Vβ11⁺ T-cell frequency, disease duration, and patient age.

NKT cells are now considered to be phenotypically diverse and complicated (5,31); therefore, it is difficult to accurately define the phenotype of NKT cells. Recently, some investigators reported that human NKT cells express a restricted TCR repertoire including an invariant Vα24 paired with Vβ11 (5,18,31–33), and they clonally expanded both in vivo (34) and in vitro in the presence of α-GalCer (35–37). Moreover, Poulton et al. (32) recently demonstrated that the combination of Vα24 antibody and Vβ11 antibody is an excellent surrogate NKT cell marker, comparable to CD1d tetramers loaded with α-GalCer. Thus, in this study, we measured the frequency of peripheral Vα24⁺ Vβ11⁺ CD3⁺ triple-positive cells as a representative of human NKT cells. The frequency was expressed as percentage of CD3⁺ cells analyzed.

Unlike the findings of several recent studies (21,22,25), it is intriguing that the frequency of Vα24⁺ Vβ11⁺ T-cells in the peripheral blood of Japanese type 1 diabetic patients overall (including classic type 1 diabetes and LADA, 0.112 ± 0.012%) was significantly higher than that in patients with type 2 diabetes (0.041 ± 0.006%; P < 0.0001) and healthy subjects (0.065 ± 0.009%; P = 0.0019). We then divided type 1 diabetic patients into the following two groups: classic type 1 diabetes (both Ab⁺ and Ab⁻) and LADA. There was a significantly higher frequency of Vα24⁺ Vβ11⁺ T-cells in classic type 1 diabetes (0.127 ± 0.015%) than in LADA patients (0.066 ± 0.010%; P = 0.0043), type 2 diabetic patients (P < 0.0001), and healthy subjects (P < 0.0001). However, there were no significant differences among LADA patients, type 2 diabetic patients, and healthy subjects. A difference in the frequency of CD3⁺ cells among the groups was not observed (data not shown).

Moreover, we divided classic type 1 diabetic patients into the following two groups: islet-associated autoantibody (GADA and/or IA2A) positive (Ab⁺; n = 46) and negative (Ab⁻; n = 32). As shown in Fig. 1, a higher frequency of Vα24⁺ Vβ11⁺ T-cells was similarly observed in the Ab⁺ (0.116 ± 0.012%) and Ab⁻ (0.142 ± 0.033%) groups compared with LADA patients (P = 0.0294 and P = 0.0021, respectively), type 2 diabetic patients (P < 0.0001 and P < 0.0001, respectively), and healthy subjects (P = 0.0046 and P = 0.0001, respectively). However, there was no significant difference between Ab⁺ and Ab⁻ classic type 1 diabetic patients. Thus a higher frequency of peripheral Vα24⁺ Vβ11⁺ T-cells was observed in classic type 1 diabetes, irrespective of the positivity of islet-associated autoantibodies. To assess the reproducibility of our assay, the measurement of Vα24⁺ Vβ11⁺ T-cell frequency was repeated at an interval of 3–6 months in the same subjects and then was compared with the first result. As previously reported by Poulton et al. (32), there was no significant difference in the Vα24⁺ Vβ11⁺ T-cell frequency at such an interval, irrespective of metabolic controls (blood glucose level and glycohemoglobin) (data not shown).

It is generally considered that higher disease (insulitis) activity is expected in type 1 diabetes with shorter disease duration (38), because a severe β-cell mass reduction is suspected in cases of established (long-standing) type 1 diabetes. To assess whether the frequency of Vα24⁺ Vβ11⁺ T-cells in peripheral blood reflects disease activity in type 1 diabetes, we investigated the correlation between Vα24⁺ Vβ11⁺ T-cell frequency and disease duration. Classic type 1 diabetic patients were divided into the following four groups: recent-onset (disease duration <3 years) Ab⁺ (n = 26), recent-onset Ab⁻ (n = 19), established (disease duration ≥3 years) Ab⁺ (n = 20), and established Ab⁻ (n = 13). We previously reported that a significant difference of serum interferon-inducible protein-10 (IP-10) level, which presumably reflects the disease (insulitis) activity in type 1 diabetes, was observed between disease duration <3 years and ≥3 years in type 1 diabetes (38). As a result, recent-onset Ab⁺ patients showed a significantly higher Vα24⁺ Vβ11⁺ T-cell frequency than established Ab⁺ patients (0.143 ± 0.017% vs. 0.081 ± 0.013%; P = 0.0075), and recent-onset Ab⁻ patients also showed a significantly higher Vα24⁺ Vβ11⁺ T-cell frequency than established Ab⁻ patients (0.191 ± 0.050% vs. 0.072 ± 0.023%; P = 0.0474). Moreover, an inverse correlation between Vα24⁺ Vβ11⁺ T-cell frequency and disease duration was observed in Ab⁺ (ρ = −0.455; P = 0.0023) and Ab⁻ (ρ = −0.432; P = 0.0162) classic type 1 diabetes (Fig. 2), but not in LADA or type 2 diabetes.

When we combined the data of Vα24⁺ Vβ11⁺ T-cell frequency in Ab⁺ and Ab⁻ classic type 1 diabetic patients and similarly analyzed them, recent-onset (Ab⁺ and Ab⁻ combined) classic type 1 diabetic patients showed a significantly higher Vα24⁺ Vβ11⁺ T-cell frequency than those with established (Ab⁺ and Ab⁻ combined) classic type 1 diabetes (0.165 ± 0.022% vs. 0.069 ± 0.011%; P < 0.0001), LADA patients (P < 0.0001), type 2 diabetic patients (P < 0.0001), and healthy subjects (P < 0.0001). Similarly, an inverse correlation between Vα24⁺ Vβ11⁺ T-cell frequency and disease duration was observed in (Ab⁺ and Ab⁻ combined) classic type 1 diabetes (ρ = −0.423; P = 0.0002). Thus, irrespective of the positivity of islet-associated autoantibodies, a higher frequency of Vα24⁺ Vβ11⁺ T-cells was observed in recent-onset classic type 1 diabetes.

To assess whether Vα24⁺ Vβ11⁺ T-cell frequency was affected by patient age, the frequency was also compared with age. There was a significant difference in age among the four groups (classic type 1 diabetes, LADA, type 2 diabetes, and healthy subjects; P < 0.0001). Although there was a tendency for an inverse correlation between Vα24⁺ Vβ11⁺ T-cell frequency and patient age in Ab⁺ and Ab⁻ classic type 1 diabetes (ρ = −0.294, P = 0.0108), multiple linear regression analysis identified disease duration, rather than age, as a significant independent parameter determining the frequency (Table 1). Moreover, although there was a tendency for an inverse correlation between frequency and age in healthy subjects (ρ = −0.286), the correlation was very weak and was considered statistically meaningless. In another report (18), there was no correlation between Vα24⁺ Vβ11⁺ T-cell frequency and age. Accordingly, we believe that we can ignore the effect of age on Vα24⁺ Vβ11⁺ T-cell frequency in the present study. These results suggest that Vα24⁺ Vβ11⁺ T-cell frequency may be specifically associated with disease (insulitis) activity in classic type 1 diabetes.

It has been reported that a numerical and functional deficit of NKT cells (including Vα24⁺ Vβ11⁺ T-cells in humans) may be associated with the pathogenesis of type 1 diabetes (25,26). On the other hand, we demonstrated here the striking finding that peripheral Vα24⁺ Vβ11⁺ T-cells, some of which are considered to be human NKT cells, were significantly increased in Japanese type 1 diabetes, especially in recent-onset classic type 1 diabetes irrespective of islet-associated autoantibody positivity. It seems that our results conflict with the previous reports by Wilson et al. (25) and Kukreja et al. (26). Regarding the report by Wilson et al., a difference in onset age (child or adult) and/or disease duration, neither of which was described in their report (25), could account for the discrepancy between their report and ours, although we cannot exclude the possibility of a difference in genetic background and the definition of NKT cell markers between the two reports. Moreover, they focused on selected subjects, i.e., twin/triplet cases, and NKT cell frequencies in “usual” type 1 diabetic patients and the general population were not compared in their study.

Similarly, regarding the report by Kukreja et al. (26), there is the possibility that a difference in onset age and/or genetic background (Caucasian versus Japanese) could account for the discrepancy between their report and ours. Indeed, the newly diagnosed type 1 diabetic patients reported by Kukreja et al. were very young (9.4 ± 2.16 years old) compared with our newly diagnosed type 1 diabetic patients (35.2 ± 13.1 years old). Therefore, at least age should be matched; analysis of children should be done to address this question. Moreover, further assessment of the Vα24⁺ Vβ11⁺ T-cell population using other useful NKT cell-associated markers and cytokine profile of Vα24⁺ Vβ11⁺ T-cells are necessary to determine the reason for the discrepancies between their reports and ours.

Although we did not evaluate the cytokine profile of Vα24⁺ Vβ11⁺ T-cells in the present study, we speculate that increased NKT cells (probably including Vα24⁺ Vβ11⁺ T-cells), which would have an extreme Th1 phenotype in diabetic subjects as shown by Wilson et al. (25), may produce IFN-γ upon activation and stimulate diabetogenic lymphocytes to promote the destruction of pancreatic β-cells. On the other hand, considering that NKT cells were originally recognized as regulatory cells, we can also speculate that these cells may be secondarily increased to regulate the progression of recent-onset classic type 1 diabetes. Although measurement of Vα24⁺ Vβ11⁺ T-cell frequency is believed to be useful as a marker of cellular immunity in type 1 diabetes, further studies are necessary to confirm the significance of increased Vα24⁺ Vβ11⁺ T-cells in the development of type 1 diabetes.

Recently, we reported a correlation between levels of serum IP-10 (a chemokine that is considered to cause migration of activated Th1-type T-cells to the local lesion) and disease (insulitis) activity in type 1 diabetes (38). However, there has been no general consensus on a useful marker to assess disease activity of type 1 diabetes thus far. We showed here a high frequency of Vα24⁺ Vβ11⁺ T-cells in recent-onset classic type 1 diabetes, and the Vα24⁺ Vβ11⁺ T-cell frequency was inversely correlated with disease duration. These observations raise the possibility that the increased Vα24⁺ Vβ11⁺ T-cells may reflect active disease (or insulitis) status in classic type 1 diabetes. To confirm this possibility, it is necessary to assess the relationship between the severity of insulitis in the human pancreas and Vα24⁺ Vβ11⁺ T-cell frequency in the peripheral blood of type 1 diabetic patients in a future study. Interestingly, we found no direct correlation between Vα24⁺ Vβ11⁺ T-cell frequency and serum IP-10 level (data not shown), suggesting that the two factors can be used as independent markers of cellular immunity in type 1 diabetes.

Although we attempted to assess the relationship between Vα24⁺ Vβ11⁺ T-cell frequency and serum C-peptide, there was no correlation between the two (data not shown). Considering that type 1 diabetes progresses as an inflammatory β-cell dysfunction without actual β-cell destruction until late in the disease process (1), a low level of C-peptide does not necessarily reflect β-cell destruction (or disease [insulitis] activity), and at times may reflect β-cell dysfunction. Therefore, although no correlation between Vα24⁺ Vβ11⁺ T-cell frequency and C-peptide was observed in the present study, this negative finding does not refute our reports on the relationship between Vα24⁺ Vβ11⁺ T-cell frequency and disease (insulitis) activity, and rather Vα24⁺ Vβ11⁺ T-cell frequency may be a useful disease activity marker independent of C-peptide value.

Because there is no report of an increase in Vα24⁺ Vβ11⁺ T-cells in other autoimmune diseases in humans thus far, we would like to emphasize that a high Vα24⁺ Vβ11⁺ T-cell (partially including human NKT cell) frequency in peripheral blood is a unique finding in recent-onset classic type 1 diabetes and is not seen in any other autoimmune diseases (18). We observed no increase in the frequency of this subset in patients with other autoimmune diseases such as rheumatoid arthritis or Grave’s disease (data not shown). These findings will contribute greatly to understanding the roles of Vα24⁺ Vβ11⁺ T-cells in autoimmune diseases and to diagnosing and evaluating disease activity in type 1 diabetes.