CD8+ regulatory T cells are critical in prevention of autoimmune-mediated diabetes.

Type 1 diabetes (T1D) is an autoimmune disease in which insulin-producing pancreatic β-cells are destroyed. Intestinal helminths can cause asymptomatic chronic and immunosuppressive infections and suppress disease in rodent models of T1D. However, the underlying regulatory mechanisms for this protection are unclear. Here, we report that CD8+ regulatory T (Treg) cells prevent the onset of streptozotocin -induced diabetes by a rodent intestinal nematode. Trehalose derived from nematodes affects the intestinal microbiota and increases the abundance of Ruminococcus spp., resulting in the induction of CD8+ Treg cells. Furthermore, trehalose has therapeutic effects on both streptozotocin-induced diabetes and in the NOD mouse model of T1D. In addition, compared with healthy volunteers, patients with T1D have fewer CD8+ Treg cells, and the abundance of intestinal Ruminococcus positively correlates with the number of CD8+ Treg cells in humans.

week-old mice were analysed as described in Fig. 1d, and the frequency is shown. Mice were treated with STZ, and blood glucose concentrations (b) and plasma insulin (c) were measured as described in Fig. 1a, b. (d) Sixty-week-old mice depleted of CD8Tregs were used for T1D induction, and blood glucose concentrations were monitored. Values represent the mean ± S.D. of 10 mice. Asterisks denote statistical significance at p < 0.05 calculated by the two-sided unpaired Student's t-test(a), the two-way ANOVA (b, d) Tukey post-hoc analysis (c). NS indicates non-significant. All experiments were repeated four times with similar results.

. In situ uppression of IFN-g-producing T cells in CD8 + Treg cells-dependent manner during Hp infection. (a)
Pancreatic mononuclear cells isolated from the indicated mice were analysed for production of IFN-g after stimulation with a plate-bound anti-CD3 antibody by flow cytometry. (b)Numbers of total recovered cells, and the frequency and number of IFN-g-producing CD4 + and CD8 + T cells are calculated. Values represent the mean ± SD from 5 mice. Asterisks denote statistical significance at p < 0.05 calculated by the Tukey post-hoc analysis. All experiments were repeated at least three times with similar results.    Intestinal contents of infected and control mice (a), HES antigens, DMEM (c), and 50 ng trehalose (e) were analysed using the GC/MS full-scan mode (m/z 45 -600) without methoximation. Representative mass chromatograms (m/z 361) showed that a compound equivalent to the peak was increased in mice infected with Hp and in HES antigens. A mass chromatogram (m/z 361) showed that the retention times were very similar among a, c, and e, and the mass spectrum of the peak corresponding to the infected sample in a (b) and to HES antigens in c (d) had high similarity to that of trehalose (f), suggesting that these peaks were derived from trehalose. Intestinal contents of infected and control mice (a), and 500 ng maltose (c) were analysed using the GC/MS full-scan mode (m/z 45 -600) with methoximation. Representative mass chromatograms (m/z 361and 480) of an infected sample indicated that at least two compounds were present. It was likely that the peak at 55.100 min was derived from trehalose ( Supplementary Fig. 5). However, the peak at 55.150 in the control sample was dominant in the control sample and not in the infected sample. (b) Mass spectrum of the peak corresponding to the control sample in a had high similarity with that of methoximated maltose (d). Mass chromatogram (m/z 361) showed that the retention times were very similar among a and c, suggesting that the peak at 55.150 min in (a) was derived from maltose. (d) Mass spectrum of the peak in (c) had a distinctive fragment ion of m/z 480, which was not present in the mass spectrum of trehalose. (e) Standard mass spectrum of methoximated maltose collected in the spectrum library of the GCMS solution based on NIST11. (f) Maltose was analysed using the GC/MS full-scan mode with methoximation. The fragment ion m/z 480 specific for maltose was used for the quantification. 2-IPM (0.5 µg) was used for the internal standard. All experiments were repeated two times with similar results. (a) As indicated in the spectrum of pure trehalose crystals (black line in the right panel), this sugar had a unique vibration band at 992 cm -1 (arrow), which reflected an a,a-1,1 bond between two glucose molecules. FITR microscopic observations using this peak were performed to visualize the location of trehalose in L3 larvae. The resulting FTIR image is shown in the centre panel together with the corresponding bright field image (left panel). The image size is approximately 200 µm wide and 500 µm long. The peak intensity at 992 cm -1 was expressed along with the colour scale from the darkest red to the lightest violet. The concentrated trehalose regions were located along the surface of the worm body. The FTIR spectra of R1 and R2 regions in the FTIR image are shown in the right panel. The spectrum of the R1 region exhibited a distinct peak at 992 cm -1 , which confirmed the existence of trehalose. (b) Trehalose concentrations in the preservative solution in the absence (DW) or presence of 200 L3 larvae/ml were measured using a trehalose assay kit. Values represent the mean ± SD from triplicate cultures. Asterisks denote statistical significance at p < 0.01 calculated by the two-tailed Mann-Whitney test. FTIR observation was performed once and treahalose concentrations were measured three times with similar results.   Fig. 2l was evaluated as described in Fig. 3f. Values represent the mean ± SD. Asterisks denote statistical significance at p < 0.05 calculated by the Tukey post-hoc analysis. Experiments using aged mice were repeated at least three times with similar results and those using NOD mice were performed once.   Fig. 4 were also analysed for CD4Tregs. The gated CD4 + cells were plotted onto CD25 and Foxp3 (left panels). The numbers indicate the percentages of CD25 + Foxp3 + CD4Tregs among the gated CD4 + cells. The frequency of CD4Tregs is presented as described in Figure 4b (right panels). Values represent the mean ± SD. NS indicates non-significant using the two-sided unpaired Student's t-test. All experiments using human samples were performed once.
Supplementary Figure 11 Gating strategies for Fig. 1d, f, and Fig. 3a, j are shown.      Metabolites in the intestinal contents (5 controls and 4 infected) were analyzed using GC/MS MRM mode. Forty-eight metabolites were identified manually using GC/MS solution software version 4.41 (Shimadzu). Each area of metabolite was normalized by the area of 2-IPM and the weight of intestinal content.
Intra-assay variability was examined by peak areas of 2-IPM in all the measurements. Its %CV value was 8.9%, which showed the measurement was stable.
The p values were calculated with the two-tailed unpaired Student's t-test. Metabolites were sorted according to the p values. Among the 48 metabolites, only the trehalose level was significantly high in the infected group with the significance level of 0.05 when it was adjusted with Bonferroni's method (p < 0.001).
When false discovery rate (FDR) was considered using Benjamini and Hochberg (BH) method, the aspargine level, in addition to trehalose, was also significantly high in the infected group with FDR 0.1, which was shown as asterisks.
Intra-assay variability was also examined using quality control (QC) samples. The QC sample that was prepared by mixing all the target samples (n = 9) was processed in the same way as the target samples. Four independent measurements were done for the QC samples. Concerning the QC measurements, the %CV values of peak areas that were normalized by the areas of 2-IPM, indicated that the measurement was stable enough for most of the metabolites.
All the retention times for the identified metabolites were close to the set values.
Supplementary Intra-assay variability was also examined using QC samples. The QC sample that was prepared by mixing HES and DMEM samples was processed in the same way as the target samples. Four independent measurements were done for the QC samples. Concerning the QC measurements, the %CV values of peak areas that were normalized by the areas of 2-IPM, indicated that the measurement was stable enough for most of the metabolites. All the retention times for the identified metabolites were close to the set values.
Supplementary Table 3. Summary of clinical characteristics of the T1D participants and healthy volunteers.