NOD mice, NY8.3 transgenic mice (12, 13), BDC2.5 TCR transgenic mice on the NOD genetic background (2), G9Cα-/- TCR transgenic mice (11), and NOD.Scid mice were bred in-house at Cardiff University. The mice were housed in specific pathogen-free facilities with a 12-hour light-dark cycle and provided ad libitum access to food and water. All experimental procedures involving animals were approved by the Cardiff University Animal Welfare and Ethics Review Committee and conducted under UK Home Office License approval, in accordance with the United Kingdom Animals (Scientific Procedures) Act, 1986, and associated guidelines.

For in vitro transcription of mRNA, we used DNA templates encoding peptide/β₂m/CD3-ζ, incorporating a modified InsB_15-23 (LYLVCGERV) peptide designated INS construct or IGRP_206-214 (VYLKTNVFL), designated IGRP construct for generation of INS-eTregs and IGRP-eTregs respectively, shown schematically in Figures 1A, B . These use endogenous MHC-I heavy chains and the engineered constructs, which comprise the chimeric polypeptide chains expressed with endogenous MHC class I (8). Human β₂m (hβ₂m) was used in this study, as it associates efficiently with mouse MHC-I heavy chains and can be used for detecting expression of the resulting MHC-I complexes (8). The pGEMT vector (Promega, Madison, WI) was used for direct cloning of PCR products. Template DNA for in vitro transcription of mRNA was cloned into the pGEM4Z/GFP/A64 vector (8) and prepared as previously described (8). mRNA constructs were transcribed in vitro using the T7 mScript Standard mRNA Production System (Cell-Script, USA) to generate 5’-capped mRNA. Cells were washed and resuspended in Opti-MEM medium (Gibco) containing the appropriate amount of in vitro-transcribed mRNA (10-20 μg). Electroporation was performed using BTX Harvard Apparatus ECM830 (1 millisecond, 300 V) into the prepared BDC2.5 CD4⁺ Treg cells (see below) to generate engineered Treg, hereafter designated eTreg. eTreg cells were only used for experiments if > 50% of the cells expressed the constructs, and this was assessed by performing flow cytometry before any in vitro or in vivo study, as described below in the flow cytometry section.

Spleens from NOD BDC2.5 non-diabetic mice (6-10 weeks old) were homogenized, and erythrocytes were removed. CD4⁺ T cells were isolated using magnetic negative selection (MACS;Miltenyi Biotec, UK). The isolated T cells were stimulated for 96 hours with CellXVivo™ Mouse Treg Cell Differentiation Kit (R&D, CDK007) in RPMI Medium at 37°C, 5% CO₂. This process yielded a highly enriched population of FoxP3⁺CD25⁺ Treg cells. Differentiation of BDC2.5 CD4⁺ T Cells into Treg cells was confirmed by FoxP3 and CD25 expression, measured by flow cytometry ( Figure 2 ). All the Tregs used in this study were BDC2.5 CD4⁺ Treg.

For extracellular staining, cells were incubated with TruStain (Biolegend, UK) for 10 minutes at 4°C to block Fc receptors. Multi-parameter flow cytometry was carried out using the antibodies against the surface markers indicated as follows: CD8-FITC (53-6.7), CD4-AF700 (RM4-5), CD69-BV711 (H1.2F3), LAP-PerCP-Cy5.5 (TW7-16B4), PD1-BV605 (29F.1A12), Lag3-PE-Dazzle594 (C9B7W), CD304-PerCP-Cy5.5 (3E12), CD39-PE-Cy7 (Duha59), TIGIT-PE (1G9), CD95-FITC (SA367H8), CCR7-BV650 (4B12), (all from BioLegend), and FR4-BV605 (12A5), CXCR1-BV786 (5A12), TNF-α−PE-CF594 (3/23), CD49b-BV786 (HMα2) (all from BD Bioscience), and dead cells were excluded from the analysis using Live/Dead Viability Dye (Thermo Fisher).

For intracellular cytokine analysis, cells were treated with phorbol 12-myristate-13-acetate (PMA) (50 ng/ml), ionomycin (500 ng/ml), and monensin (3 μg/ml) (all from Sigma-Aldrich) for 3 hours. Fc receptors were blocked using TruStain, and after extracellular staining, cells were fixed using fixation/permeabilization according to the manufacturer’s instructions. Antibodies against intracellular markers, including FoxP3-APC (FJK-16s, eBioscience), CTLA4-PE (UC10-4B9, Thermo Fisher), IL-2 (JES6-5H4, BD Biosciences) IL10-APC (JES5-16E3, BD Biosciences), and IFNγ-PE-Cy7 (XMG1.2, BioLegend), or appropriate isotype controls, were used for staining.

For assessment of expression of hβ₂m, transfected cells were first incubated in Fixation/Permeabilization buffer (eBioscience, 00-5523-00), followed by staining with anti-hβ₂m FITC (clone 2M2, Thermo Fisher) together with antibodies against surface and intracellular markers. The other markers stained for included anti-FoxP3 Alexa Fluor 647 (clone 3G3, eBioscience). This protocol allowed for the assessment of genetic construct expression in the transfected eTregs.

Flow cytometry was performed on LSRFortessa (FACS Diva software), and analysis was performed using Flowjo software 10.8.1 (Treestar).

BDC2.5 T regs cells were isolated, activated, and transfected with INS-construct or IGRP-construct mRNA and labelled with PKH lipophilic dye (Sigma Aldrich). As a control, BDC2.5 Treg cells were subjected to the electroporation procedure without added mRNA (mock-transfected) and labelled in the same way. The construct-transfected or mock-transfected eTregs (designated MOCK-eTreg) cells were adoptively transferred into 5-6-week-old NOD mice by intravenous injection into the tail vein. After 24, 48 hours and 5 days, mesenteric, and pancreatic lymph nodes as well as spleens were taken from the mice for measurement of the in vivo expression of the genetic constructs by staining for hβ₂m in the INS-eTregs or IGRP-eTregs. In addition, other markers stained for included anti-FoxP3 Alexa Fluor 647 (clone 3G3, eBioscience), and cells were gated on PKH 26-positive cells. This protocol allowed for the assessment of genetic construct expression in the transfected eTregs.

The BDC2.5 CD4⁺ Treg cells were transfected with mRNA encoding INS or IGRP constructs to generate eTregs. CD8⁺ T cells were isolated from the spleens of 5- to 6-week-old G9Cα-/- or NY8.3 mice and separated by negative selection (MACS; Miltenyi Biotec). The purified CD8⁺ T cells were then activated and stimulation using plate-bound anti-CD3 (0.5 µg/ml) (clone 2C11, Bio X Cell) and anti-CD28 (5 µg/ml) (clone 37.51, Bio X Cell) antibodies in 96-well plates. Activated mRNA-transfected eTreg cells were incubated with the activated, purified INS- or IGRP-CD8⁺ T cells respectively for 72 hours at different cell ratios of Treg: CD8⁺ T cell ratio. Death was assessed in the live/dead gate on flow cytometric plots.

Naive INS-CD8⁺ T cells were isolated and co-cultured with INS-eTreg at a 1:1 ratio. After 24 hours, the INS-eTregs were purified using a negative selection kit for CD4⁺ T cells (Miltenyi Biotec), removing the CD8⁺ T cells. Polyclonal CD4⁺ T cells were purified from splenocytes of NOD mice using negative selection (Miltenyi Biotec) and labelled with CellTrace Violet (CTV) (Thermo Fisher). These labelled polyclonal splenic CD4⁺ T cells were then incubated with the INS-eTreg and stimulated by anti-CD3/CD28 Dynabeads according to the manufacturer’s instructions (Thermo Fisher) at different ratios of INS-eTreg to polyclonal CD4⁺ T cells, maintaining the number of polyclonal CD4⁺ T cells constant (15x10³ cells), in 96-well flat-bottomed plates. The cells were cultured in RPMI complete medium at 37°C, 5% CO₂ for 72 hours, and the inhibitory effect of the INS-eTreg on proliferation of the CTV-labelled polyclonal CD4⁺ T cells was investigated by assessing the dilution of CTV using flow cytometry.

INS-CD8⁺ T cells were isolated, labelled with Carboxyfluorescein succinimidyl ester (CFSE) (Thermo Fisher) and co-cultured with INS-eTreg, stimulated by anti-CD3/CD28 Dynabeads, according to the manufacturer’s instructions (Thermo Fisher), for 72 hours at different cell ratios. Proliferation of the INS-CD8⁺ T cells was assessed by dilution of CFSE using flow cytometry.

BDC2.5 Treg cells were generated and transfected with the either INS or IGRP constructs. These INS-eTreg cells (10⁷) or IGRP-eTreg cells (10⁷) were then co-transferred with 10⁷ INS-CD8⁺ or IGRP-CD8⁺ T cells respectively into 5-6-week-old female NOD.scid mouse recipients by intravenous injection into the tail vein. In this experiment, as irrelevant peptide transfected controls, we also used cotransfer of INS-CD8⁺ T cells with IGRP-eTreg and IGRP-CD8⁺ T cells with INS-eTreg. Mice were monitored daily for glycosuria using Diastix strips (Bayer Diastix). Diabetes was confirmed by blood glucose levels >13.9 mmol/l following two consecutive positive readings. eTreg were only used for the experiments if >50% of eTreg expressed the construct, as assessed by staining for hβ2m.

BDC2.5 Treg cells were generated and transfected with either INS or IGRP constructs. Ten million INS-eTregs were transferred into female NOD mice at 5-6 weeks of age, and a further ten million INS-eTreg cells transferred one week later (n=16). A separate set of female NOD mice of the same age were transferred with MOCK-eTregs as controls at the same times as the INS-eTregs (n=8). Similarly, in a separate cohort of mice, ten million IGRP-eTregs were transferred into NOD mice at 7-8 weeks of age, and a further ten million IGRP-eTreg cells transferred one week later (n=16). A control set of mice of the same age were transferred with MOCK-eTregs as controls at the same times as the IGRP-eTregs (n=8). Furthermore, we also have included data from untreated NOD mice from our NOD colony at the same time as the experimental quality controls for diabetes development, as we maintain a check on overall diabetes incidence routinely, but these were not part of the experiment. The mice were monitored weekly for glycosuria using Diastix strips (Bayer Diastix) until 35 weeks of age or until diabetes developed, confirmed by blood glucose levels >13.9 mmol/l, following two consecutive positive readings. eTreg were only used for the experiments if >50% of eTreg expressed the construct as measured by staining for hβ2m, as described above.

The pancreas was fixed in paraformaldehyde-lysine-periodate buffer overnight, infused with 10% sucrose in phosphate buffer, followed by 20% sucrose, as described previously (6). The pancreas was then embedded in OCT and snap-frozen for immunohistochemistry. Frozen sections, 10 μm thick, were stained with rat-anti-mouse CD4-biotin (clone RM4-5, Thermo Fisher), CD8-biotin (clone 53.6.7, Thermo Fisher), B220-biotin (clone RA3-6B2, Thermo Fisher), and anti-insulin-biotin (ICBTACLS) antibodies and detected with streptavidin-alkaline phosphatase and a Vector Red AP substrate kit (Vector Laboratories, Peterborough, UK), followed by counterstaining with hematoxylin. Insulitis was assessed in 4-5 mice per group, and 100-150 islets were scored. Scoring for insulitis is shown in the figure legend.

Unpaired Student’s t-tests were used to compare data between groups, or 2-way ANOVA when multiple group comparisons were done. The Chi-square test was used for the analysis of insulitis. Log-rank analysis was carried out for the adoptive transfer experiments. A p-value of less than 0.05 was considered statistically significant. All statistical analysis was carried out using Graphpad Prism v 10.1.1.

BDC2.5 TCR transgenic CD4⁺ T cells express the BDC2.5 TCR which recognizes an insulin-chromogranin A hybrid peptide, expressed in the pancreatic islets of NOD mice (14). These cells very effectively target the islets through recognition of this endogenous peptide, but the NOD BDC2.5 TCR transgenic mice do not themselves develop spontaneous diabetes, as endogenous Treg cells prevent this from occurring (15). Thus, we chose to use the BDC2.5 TCR transgenic CD4⁺ T cells, in order to engineer Tregs in vitro to endow these cells with the ability to target the pancreatic lymph nodes and islets by virtue of their TCR. The differentiation process was carried out over 4-5 days (illustrated in Figure 2A ), yielding a population of Tregs with a high expression of FoxP3 and CD25 ( Figure 2B ) and other markers including the activation markers CD69, FR4, CD95, CCR7 and IL-2 ( Figure 2C ). We also stained for other markers previously demonstrated on Treg including CD39, PD-1, LAP, and IL-10 ( Figure 2C ); the expression of other regulatory markers including LAG3, TIGIT, CTLA-4 and CD304 was very low (data not shown).

To redirect the BDC2.5 TCR transgenic CD4⁺ Treg such that they would interact specifically with antigen-specific pathogenic CD8⁺ T cells, the BDC2.5 TCR transgenic CD4⁺ Treg were electroporated with the INS construct or the IGRP construct. Hereafter, the cells will be designated as INS-eTreg or IGRP-eTreg. Expression of these mRNA constructs was assessed, in vitro, 4 and 24 hours after transfection ( Figure 3 ), by staining with anti-hβ₂m. A mean of 67.5% (range 57.8-81%) of the INS-eTreg cells and 80.5% (range 71.6-88%) IGRP-eTreg expressed the constructs at 4 hours, decreasing to a mean of 53.3% INS-eTreg cells and 55.5% IGRP-eTreg cells at 24 hours ( Figure 3 ). At 72 hours a mean of 14.1% INS-eTreg cells and 28.3% IGRP-eTreg cells expressed the constructs (data not shown).

To investigate the expression of the constructs in vivo, we labelled the transfected eTreg that were positive for FoxP3, with the lipophilic dye PKH and 4 hours post labelling, we transferred the PHK-labelled eTreg, intravenously into NOD mice ( Figure 4A ). We tracked the presence of eTreg cells in the mesenteric lymph nodes (MLN), pancreatic lymph nodes (PLN) and spleen, 24 hours after transfer. The transfected mRNA constructs could still be clearly visualized up to 24 hours after adoptive transfer, as shown by expression of hβ₂m in the PKH+ T cells ( Figure 4B ). Next, the cells were further tracked and mice were examined for presence of the Tregs, and the expression of the transfected constructs investigated at 5 and 8 days after transfer. Whereas the CD4⁺ PKH-labelled e-Treg cells, expressing FoxP3, could be detected in the lymph nodes at 5 and 8 days after transfer, the cells no longer expressed the constructs on their cell surface (data not shown).

To investigate the regulatory function of the eTregs, these cells were cultured for 72 hours, in a Treg suppression assay with CTV-labelled polyclonal CD4⁺ T cell responders, which were purified from NOD splenocytes and stimulated with anti-CD3 and anti-CD28 Dynabeads ( Figure 5A ). With increasing proportions of eTregs to responder CTV-labelled polyclonal CD4⁺ T cells, the responder polyclonal CD4⁺ T cell proliferation decreased, as demonstrated by decrease in CTV dilution ( Figure 5B ). This assay tested general Treg function, and there was no difference overall in the general ability of eTregs to suppress CD4⁺ T cells, whether they were mock-transfected or transfected with the MHC-peptide complex.

To further test the regulatory function of the transfected eTregs, after encounter with their antigen-specific target T cells, the INS-eTregs were cultured with INS-CD8⁺ T cells as shown in the experimental schematic in Figure 5A . The eTreg were placed in culture with insulin peptide-specific INS-CD8⁺ T cells at different ratios of CD8⁺ T cell:eTreg, ranging from 1:1 to 20:1 to specifically prime the eTreg. After 24 hours, the eTregs were then separated from the priming CD8⁺ T cells by depleting the CD8⁺ T cells. As CD8⁺ T cells have cytotoxic function, we assessed the eTreg viability after the culture with INS-CD8⁺ T cells. We found that the incubation of eTreg with naive antigen-specific CD8⁺ T cells did not significantly affect the viability of the eTreg at different ratios of Treg: CD8 T cells ( Supplementary Figure S1 ). Moreover, the eTreg were then used in a CD4⁺ Treg suppression assay culture for a further 72 hours, with anti-CD3- and anti-CD28-stimulated polyclonal CD4⁺ T cells, and we observed that the eTregs maintained, and enhanced their suppressive function toward polyclonal CD4⁺ T cells, after interaction with the CD8⁺ T cells, as shown for 2 different ratios of CD8:eTreg pre-culture ( Figures 5C, D ). Although mock-transfected eTreg increased suppressive activity, possibly related to the inflammatory cytokines produced from the CD8⁺ T cell incubation, it was clear that the antigen-specific eTreg suppressive function was more enhanced, and this effect was greater with the increased CD8⁺ T cell:eTreg ratio of 20:1 compared to 1:1. Thus, the eTregs maintained/enhanced their regulatory function after specific interaction with the antigen-specific CD8⁺ T cells.

We have shown that the eTregs maintain polyclonal suppressive function after transfection with the specific MHC-peptide constructs, and next we investigated the specific effects of the eTregs on the antigen-specific CD8⁺ T cells. To do this, we first evaluated whether the antigen-specific eTreg induced death of the antigen-specific CD8⁺ T cells, and observed more loss of the antigen-specific CD8⁺ T cells after culture with the antigen-specific eTregs, compared with the cells incubated with MOCK-eTregs ( Supplementary Figure S2 ). We then labeled the antigen-specific CD8⁺ T cells with CFSE and tested the CD8⁺ T cell proliferative responses by examining the dilution of the dye after incubation with the eTreg. We found that at a low ratio of CD4⁺ eTregs to CD8⁺ T cells, there was approximately 20% suppression, but with increased CD4⁺ eTregs to CD8⁺ T cell ratio, the CD8⁺ T cell proliferation was further reduced by 95% ( Figure 6 ). Very little suppression of the antigen-specific CD8⁺ T cells (less than 10%) was seen in the presence of the eTregs transfected with an irrelevant peptide (IrrPept-eTreg).

Having demonstrated that the eTregs suppressed the proliferation of naïve antigen-specific CD8⁺ T cells, we assessed the phenotype of the CD8⁺ T cells following culture with the antigen specific-eTregs and compared with either IrrPept-eTregs transfected with irrelevant peptide ( Figures 7A–I ) or MOCK-eTregs ( Figures 7J–M ). We found that there was upregulation of markers of exhaustion (PD1, FR4), as well as regulatory proteins (CD49b, CTLA-4, CD39, LAG3, CD304) activation markers (CD69), migration proteins (CXCR1, CCR7), and cytokines (TNFα, interferon-γ), although LAP was not increased.

We had shown that eTregs suppressed both polyclonal CD4⁺ T cells and antigen-specific CD8⁺ T cells in vitro, and next the important question was whether eTregs suppressed the diabetogenic T cells in vivo. Thus, we tested the suppressive function of eTreg cells on pathogenic INS-CD8⁺ T cells and IGRP-CD8⁺ T cells in adoptive co-transfer into NOD.Scid mice. The experimental scheme is shown in Figure 8A . Activated splenic INS-CD8⁺ T cells, isolated from G9Cα^-/- mice, or activated IGRP-CD8⁺ T cells, isolated from NY8.3 mice were transferred into NOD.Scid mice. When the activated antigen-specific CD8⁺ T cells were adoptively co-transferred with either INS-eTregs together with INS-CD8⁺ T cells (1:1) or IGRP-eTregs together with IGRP- CD8⁺ T cells (1:1) diabetes development was significantly reduced, compared with 100% diabetes occurring in the absence of the eTregs ( Figure 8B ) (p<0.001). When the pathogenic CD8⁺ T cells were co-transferred with control eTregs that were transfected with construct linked with an irrelevant peptide, all the mice development diabetes, and no protection was seen. Comparing the histology of the non-diabetic mice co-transferred with the pathogenic CD8⁺ T cells, the overall level of infiltration of CD8⁺ T cells was less severe, with lower levels of both infiltrating CD8⁺ T cells as well as CD4⁺ T cells ( Figure 8C ).

Having shown that the eTregs were able to protect against adoptively-transferred CD8⁺ T cell-induced diabetes, we then tested eTreg effect on the incidence of spontaneous diabetes in NOD mice. We infused INS-eTreg cells into NOD mice at 5-6 weeks of age and repeated the infusion one week later again at 6-7 weeks of age. In a separate group of mice, the IGRP_-eTregs were adoptively transferred into 7-8 week-old female NOD mice and this infusion was repeated one week later in 8-9 week old mice. These times were chosen related to the previously published reports that insulin-reactive CD8⁺ T cells appeared early (10, 11) whereas IGRP-CD8⁺ T cells occurred later (10). These mice were tested weekly for glycosuria and diabetes confirmed by blood glucose reading and the mice were terminated, either when they became diabetic, or at 35 weeks at the end of the experiments as shown in Figure 9 .

We examined the insulitis in the pancreas samples of eTreg recipients, when the recipient mice were terminated (either when they developed diabetes or when non-diabetic at the end of the experiment at 35 weeks). We found reduced severity of insulitis in the protected mice ( Figure 10 ), with less infiltration of CD8⁺ T cells, CD4⁺ T cells as well as B cells compared with those that developed diabetes. In contrast, the majority of those control recipients receiving the MOCK-eTreg cells became diabetic.