Rats were derived from the Worcester DP-BB and diabetes-resistant (DR) BB strain, but were maintained and bred at our institutional Central Animal Facility under specific-pathogen-free and viral-antibody-free conditions . The animals received humane care in compliance with the principles of laboratory animal care (NIH publication no. 85-23; revised 1985) and the Dutch law on experimental animal care. The university Ethical Board for Animal Studies approved all animal experiments reported in this study.
Dietary intervention protocol
Two dietary intervention experiments were performed.
In the first, experiment 1, DP-BB rats were given the HC diet (n = 18) or a standard diet (n = 16) from weaning onwards and monitored for the development of autoimmune diabetes. At 65 days of age, just before DP-BB rats start to develop autoimmune diabetes, a lactulose–mannitol (LA/MA) assay as described by Meddings et al.  was performed to establish intestinal permeability in the small intestine. After subjection to the LA/MA test, the rats were monitored until 140 days of age. Animals were weighed three times per week. In case of weight loss or suspicion of diabetes, blood glucose was measured in tail vein blood using a glucose sensor (Reflolux S; Boehringer, Mannheim, Germany). When blood glucose exceeded 15 mmol/l (non-fasting), rats were considered diabetic and killed.
In the second intervention, experiment 2, another group of DP-BB rats was fed the HC diet (n = 37) or the standard diet (n = 33) from weaning onwards. At fixed time points (25, 35, 45, 55 and 65 days of age) rats were killed. Ileum tissue was collected to investigate the effect of the HC diet over time on the intestinal barrier function by measuring mRNA expression of tight junction proteins and the ileal transepithelial electrical resistance (TEER). The TEER was measured on ileal tissue obtained at 65 days of age. In addition, ileal mRNA expression of the cytokines Ifn-γ (also known as Ifng), Tnf-α (also known as Tnf), Il-10 (also known as Il10) and Tgf-ß (also known as Tgfb1) was measured. DR-BB rats of 21 to 50 days of age (n = 5) or 51 to 70 days of age (n = 25) were used as control rats.
The standard plant-based diet used was a standard laboratory rodent diet (RMH-B diet; Arie Blok, Woerden, the Netherlands). The HC diet (TD99482; Harlan-Teklad Custom Research, Madison, WI, USA) is a modification of the AIN-93G diet and contained 200 g/kg HC (as source of amino acids), 3 g/kg l-cysteine, 509.8 g/kg corn starch, 120 g/kg sucrose, 70 g/kg soy-bean oil, 50 g/kg cellulose, 35 g/kg mineral mix, 10 g/kg vitamin mix, 2 g/kg choline bitartrate and 0.20 g/kg butylated hydroxyanisole antioxidant.
LA/MA test for measuring intestinal permeability in vivo
The LA/MA test is a non-invasive technique to measure intestinal permeability in vivo. The sugar alcohol mannitol, which is a small molecule, permeates the intestinal mucosa via a transcellular pathway through the water-filled pores of the cell membrane, whereas the larger disaccharide molecule lactulose uses a paracellular route through the junctional complexes between adjacent enterocytes. An increased uptake of lactulose (associated with a high lactulose:mannitol ratio) indicates a damaged barrier. Therefore, a reduction of the intestinal permeability will lead to a lower lactulose uptake and a decreased lactulose:mannitol ratio in the urine. This test allowed us to investigate whether the rats with high intestinal permeability in the prediabetic phase would indeed develop autoimmune diabetes.
A LA/MA assay as described by Meddings et al.  was performed to establish intestinal permeability in the small intestine. Briefly, a stock solution was made containing 4 g mannitol and 6 g lactulose per 100 ml distilled water. Enough solution was made to give each rat 2 ml of the probe.
Rats were placed in stainless steel metabolism cages with wire bottoms to separate faeces from urine. Rats were denied access to water for 3 h, at which point they were allowed free access to water for the remainder of the experiment. Urine was collected for a total of 24 h, after which rats were returned to their normal cages. Urine volumes were measured and the urine composition was analysed by HPLC as described previously .
Final data were reported as a ratio of fractional excretions (lactulose:mannitol). Fractional excretion is defined as the fraction of the gavaged dose recovered in the urine sample.
Snapwell assay for measuring TEER
An indicator of intestinal permeability, TEER was measured on ileum samples in vitro by snapwell assay as described by Watts et al. , with some minor modifications. A small sample (length ~50 mm) was taken from the ileum. During the time between killing of the animal and mounting in the snapwells (Corning, Schiphol-rijk, the Netherlands), the samples were kept in DMEM (with 4.5 g/l glucose) (Gibco, Breda, the Netherlands) at 4°C. Before mounting, the intestinal pieces were cleaned of faeces by gently pulling the intestine over a 1 ml pipette and then cut open longitudinally. The intestine was cleaned further by gently moving it, with the mucosal layer facing downwards, in a Petri-dish with approximately 1 ml medium. A piece was cut out, placed on a filter with the mucosal layer facing upwards, sandwiched between two disks and put in the snapwell insert. This insert was placed in a pre-warmed six-well plate containing 2 ml DMEM in each well. On top of the insert 450 μl DMEM was added. The tissue was mounted within 15 min after excision and the plates were incubated at 37°C. The TEER measurements were done using a Millipore Millicell-ERS meter (Millipore, Amsterdam, the Netherlands) at 30, 60, 90, 120 and 180 min after excision. The TEER shown in this paper was measured at 60 min after tissue excision.
Quantitative PCR of tight junction proteins and cytokines
RNA was isolated from ileal tissue and expression of Myo9b, Cldn1, Cldn2 and Ocln mRNA (encoding the tight-junction-related proteins myosin IXb, claudin-1, claudin-2 and occludin) and Ifn-γ, Il-10, Tgf-ß and Tnf-α mRNA (encoding cytokines) was analysed. To isolate RNA from the intestine, frozen tissue (±1 cm, stored at −80°C) was homogenised in 1 ml of TRI reagent (Sigma-Aldrich, Zwijndrecht, the Netherlands) and mRNA was isolated using the TRI reagent mRNA isolation method. The concentration of the isolated mRNA was determined using a nanodrop (ND-1000; Isogen, Maarsen, the Netherlands). Measurement was done at the 230 nm absorption spectrum for RNA. Isolated mRNA (5 μg) was converted to cDNA using a kit (SuperScript II Reverse Transcriptase kit; Invitrogen Life Technologies, Breda, the Netherlands). To measure differences in expression of genes for tight junction-related proteins, transcript levels of Myo9b, Cldn1, Cldn2, Ocln and the housekeeping gene hypoxanthine phosphoribosyl-transferase (Hprt) were quantified using real-time PCR (for primer sequences see Electronic supplementary material [ESM] Table 1). Real-time PCR analysis was performed using iQ SYBR Green Supermix (Bio-Rad Laboratories, Veenendaal, the Netherlands) according to the manufacturer’s instructions; detection was by iCycler iQ Real-Time PCR Detection System (Bio-Rad), using the following programme: 3 min 95°C, 40 cycles of 30 s at 95°C and of 30 s at 60°C, and of 10 s at 58°C, then 80 times an increase in temperature of 0.5°C to create a melting curve. Results were expressed as ratio of target gene, Hprt, according to a mathematical method described by Pfaffl et al. .
Serum zonulin levels were analysed by the UM Center for Celiac Research at the University of Maryland Medical Center. The ELISA was performed as described by Sapone et al. .
Difference in survival was calculated by the logrank test for Kaplan–Meier survival curves. Differences in zonulin levels and lactulose:mannitol ratio were calculated using the Mann–Whitney U test. Differences in TEER and expression of tight junction proteins and cytokine levels were calculated by Kruskal–Wallis test followed by the Mann–Whitney U test to identify the difference between the groups. Correlations were tested for significance using the Spearman correlation method. A p value of < 0.05 was considered significant.