Animals

NOD mice have been inbred at the KU Leuven animal facility since 1989, maintained under semi-barrier conditions and housed according to protocols approved by the KU Leuven Animal Care and Use Committee. This study was approved by the local ethics committee (Leuven, Belgium; project number 151-2020). Mice were screened three times a week for glucosuria using Diastix (Ascensia Diabetes Care, Machelen, Belgium). Diabetes was confirmed when glucosuria was detected, followed by non-fasting blood glucose concentrations exceeding 11.1 mmol/l on two consecutive days (AccuCheck, Roche Diagnostics International, Basel, Switzerland).

Treatments and follow-up

Female NOD mice with recent-onset diabetes were randomly assigned to one of four treatment groups: untreated, verapamil+IgG, rabbit-anti-mouse ATG (mATG) or verapamil+mATG, with 15–26 mice per group. mATG (PharmAbs, Leuven, Belgium) or the corresponding IgG isotype control (BioXcell, West Lebanon, NH) was administered intravenously at 250 μg per day on days 0 and 3 after disease onset (total dose of 500 µg). Verapamil (Thermo Fisher Scientific, Waltham, MA) was dissolved in sterile drinking water and administered ad libitum at a concentration of 1 mg/ml throughout the 56 day follow-up period. Daily verapamil dosing ranged from 4 to 44 mg per day, based on an average daily water consumption of 6.6 ± 1.4 ml (n=10) for normoglycaemic and 39.9 ± 13.5 ml (n=11) for diabetic NOD mice (data not shown). Mice were monitored three times per week for body weight and blood glucose concentrations, and follow-up continued for 56 days after therapy start. Disease reversal was defined by the absence of glucosuria and normalisation of blood glucose concentrations (<11.1 mmol/l). Mice were withdrawn from the study before the 56 day endpoint if their blood glucose concentrations exceeded 33.3 mmol/l in two consecutive measurements. The experimental design is illustrated in electronic supplementary material (ESM) Fig. 1.

Complete blood counts and plasma cytokine and C-peptide analyses

EDTA-anticoagulated peripheral blood was collected via submandibular bleeding on days 0, 3 and 14 after therapy start. Whole blood samples were analysed using a Vet abc Plus+ (Scil Animal Care Company, Viernheim, Germany) to determine the percentages (%) and absolute numbers (103/mm3) of white blood cells (WBCs), lymphocytes, monocytes and granulocytes. Whole blood samples were centrifuged at 4°C for 10 min at 2000 g, and plasma was collected to detect the presence of the following cytokines: IFN-γ, IL-2, IL-10, IL-15 and IL-21 (cat. no. K15069L-1, Meso Scale Diagnostics, Rockville, MD), and random C-peptide concentrations by ELISA (Merck Millipore, Burlington, MA).

Pancreatic insulin content

Pancreases were harvested on days 0 and 14 after therapy start for insulin content determination. Pancreases were homogenised in acidic ethanol (91% ethanol, 9% 1 mol/l H3PO4) at 4°C overnight and then sonicated (30 s). Insulin content in the supernatant was determined by ELISA (cat. no. 10-1247-10, Mercodia, Uppsala, Sweden) and normalised to the weight of the pancreas.

Multi-parameter flow cytometry

EDTA-anticoagulated peripheral blood was collected via submandibular bleeding on days 0, 3 and 14 after therapy start. Pancreatic draining lymph nodes (PLNs) and pancreases were harvested on days 0 and 14. The pancreas was mechanically disrupted using surgical blades, followed by enzymatic digestion with collagenase-D and DNAse I at 37°C, and purified using 40% Percoll. PLNs were crushed through a 70 µm cell strainer. Single-cell preparations were stained with the following antibodies: CD3 (RRID: AB_1107000, clone 145-2C11, cat. no. 45-0031-82, eBioscience, Thermo Fisher Scientific, Waltham, MA, 1:100 or RRID: AB394595, clone 145-2C11, cat. no. 553062, BD Biosciences, Erembodegem, Belgium, 1:100), CD4 (RRID: 1645235, clone GK1.5, cat. no. 560181, BD Biosciences, 1:160), CD8 (RRID: 1272198, clone 53-6.7, cat. no. 48-0081-82, eBioscience, Thermo Fisher Scientific, 1:300), CD25 (AB_469608, clone PC61.5, cat. no. 25-0251-82, eBioscience, Thermo Fisher Scientific, 1:625), CD44 (AB_465044, clone IM7, cat. no. 11-0441-81, eBioscience, Thermo Fisher Scientific, 1:400), CD62L (AB_469409, clone MEL-14, cat. no. 17-0621-81, eBioscience, Thermo Fisher Scientific, 1:400) and FOXP3 (AB_465935, clone FJK-16s, cat. no. 12-5773-80, eBioscience, Thermo Fisher Scientific, 1:100). Zombie Aqua (BioLegend, San Diego, CA) was used as viability marker. FOXP3 detection was performed using a staining buffer set (cat. no. 00-5523-00, eBioscience, Thermo Fisher Scientific), following the manufacturer’s instructions. Cell acquisition was performed on a BD Canto II AIG flow cytometer (BD Biosciences) and analysed using FlowJo software v.10.10 (FlowJo, Ashland, OR). All analyses were conducted on a fixable viability dye-negative singlet population. The gating strategy is shown in ESM Fig. 2.

Insulitis grading

Formalin-fixed paraffin-embedded (FFPE) pancreases collected on day 14 after therapy start were sectioned, deparaffinised and rehydrated. Tissue sections were stained with Mayer’s Haematoxylin Solution (1 g/l) followed by 0.2% eosin. H&E-stained pancreas sections were examined using a BA310 LED trinocular microscope equipped with a Moticam S3 (MoticEurope, Barcelona, Spain). Islets were graded by an independent investigator (P-JM). At least 25 islets per pancreatic sample were assessed for immune infiltration, and the degree of insulitis was scored as follows: 0, no infiltration; 1, peri-insulitis; 2, lymphocyte infiltration in <50% of the islet area; and 3, lymphocyte infiltration in >50% of the islet area or complete islet destruction.

Pancreas immunohistochemistry

First, 5 µm sections from FFPE pancreas tissues collected on days 0 and 14 after therapy start were mounted on Superfrost Plus slides (VWR International, Radnor, PA) and dried overnight at 40°C. The slides were baked for 60 min at 60°C, deparaffinised in xylene (2 × 5 min), rehydrated in 95% ethanol (2 × 2 min) and rinsed for 2 min under running distilled water. For immunostaining, non-specific binding sites were blocked for 30 min with Blocker Casein in PBS (cat. no. 37528, Thermo Fisher Scientific) diluted 1:4 in PBS + 0.1% Tween-20 (PBST). Sections were incubated overnight at 4°C with primary antibody for insulin (guinea pig anti-insulin, Diabetes Research Center [DRC], Vrije Universiteit Brussel [VUB], Brussels, Belgium) diluted in BOND Primary Antibody Diluent (cat. no. AR9352, Leica Biosystems, Nussloch, Germany). After incubation, slides were rinsed 3 × 20 min in PBST and subsequently incubated for 30 min with biotinylated secondary antibody (RRID: AB_2340451, biotin-SP-labelled donkey anti-guinea pig, Jackson ImmunoResearch), diluted 1:1,000 in BOND Primary Antibody Diluent. Slides were rinsed 3 × 10 min in PBST and subsequently incubated for 30 min with Vectastain Elite ABC-AP Standard (Vector AK-5000, Vector Laboratories, Burlingame, CA). Slides were rinsed 3 × 10 min in PBST and subsequently incubated for 10 min with ImmPACT Vector Red (Vector SK-5105, Vector Laboratories). Slides were rinsed for 2 min under running distilled water. Finally, slides were dried for 60 min at 60°C and mounted with Pertex mounting medium (VWRKAM-0801, VWR International). Brightfield image acquisition for beta cell volume analysis was done with a Leica Aperio GT450 (Leica Biosystems). Image analysis was conducted with Fiji image processing software v.1 (NIH, Bethesda, MD).

Pancreas RNA in situ hybridisation

RNA in situ hybridisation (RNA-scope), immunostaining and imaging were performed by the Visual and Spatial Tissue Analysis (VSTA) core facility at VUB (Brussels, Belgium) (https://vsta.research.vub.be) using the RNA-scope Multiplex Fluorescent v2 Assay kit (cat no. 323100, Advanced Cell Diagnostics [ACD], Newark, CA) according to the manufacturer’s instructions. The probe used was Mus musculus Txnip (cat no. 457221, ACD). For immunostaining after in situ hybridisation, sections were blocked for 30 min with Blocker Casein in PBS (cat no. 37528, Thermo Fisher Scientific) diluted 1:4 in PBST. Primary antibodies diluted in BOND Primary Antibody Diluent (Leica Biosystems) were guinea pig anti-insulin (DRC, VUB, 1:5,000) and rabbit anti-glucagon (RRID: AB_10561971, clone EP3070, cat no. ab92517, Abcam, 1:5,000), incubated overnight at 4°C, and detected using anti-guinea pig AlexaFluor 488-conjugated IgG (RRID: AB_2337438, Jackson ImmunoResearch, 1:500) and anti-rabbit DyLight 755-conjugated IgG (RRID: AB_2556615, Thermo Fisher Scientific, 1:200), respectively. Nuclei were stained with Hoechst 33258 (Sigma-Aldrich, St Louis, MO, B2261) added at 2 µg/ml to the secondary antibodies. Fluorescence image acquisition was done with a Zeiss Axio Scan.Z1 (Zeiss, Jena, Germany) at ×20 magnification.

Blinding/masking

Experimenters were not blinded throughout the study.

Statistical analysis

Statistical comparisons were performed using GraphPad Prism v.9.3.0 (GraphPad Software, La Jolla, CA). The respective number of replicates (n) is indicated in the figures or figure captions. Power analyses were used to predetermine appropriate sample sizes for animal experiments (power 0.8, α 0.05). Kaplan–Meier life table analysis with logrank Mantel–Cox regression was applied to compare the survival curves between experimental groups. p values were determined by an appropriate statistical test such as unpaired two-tailed Student’s t test or Mann–Whitney U test if the data did not assume Gaussian distribution, as indicated in the figure legends. Differences between paired data were determined by a mixed-effects model test with post hoc Tukey’s honestly significant difference test. Outliers were determined by the Grubbs test (alpha 0.05). p values ≤0.05 were considered significant (*p≤0.05, **p≤0.01, ***p≤0.001).