Salmonella typhimurium infection halts development of type 1 diabetes in NOD mice



Infectious disease has been proposed as an environmental modifier of autoimmunity in both human populations and the NOD mouse. We found that infection of NOD mice with attenuated, but not killed, Salmonella typhimurium can reduce the incidence of type 1 diabetes (T1D), even if infection occurs after the development of a peri-islet pancreatic infiltrate. Functional diabetogenic effector T cells are still present, as demonstrated by the initiation of diabetes in NOD-scid recipients of transferred splenocytes. High levels of IFN-γ are secreted by splenocytes of infected mice, but there is no evidence of involvement of IL-10 in the protective effect of the infection. Finally, prolonged changes in cell subsets are observed in infected mice involving invariant Vα14Jα281 NKT and dendritic cells. These data reinforce the idea that prevention of T1D in the NOD mouse cannot be reduced to the simple Th1/Th2 paradigm and that different infections may involve different protective mechanisms.


α-Galactosyl β-ceramide

iNKT cell:

Invariant NKT cell


Type 1 diabetes

1 Introduction

Salmonellae are Gram-negative bacteria belonging to the family of Enterobacteriaceae. Salmonella can infect a large number of animal species and humans. During systemic infections, Salmonellae reside within macrophages, granulocytes and dendritic cells (DC) 1, 2. In the early phases of the infection, intracellular bacterial growth is inhibited by the production of reactive oxygen and nitrogen intermediates, by the concerted action of a number of cytokines (TNF-α, IFN-γ, IL-12, IL-18, and IL-15) as well as by the recruitment and activation of inflammatory phagocytes in the tissues 3. Later, Salmonella infections induce the activation of CD4+ TCR-αβ T cells resulting in Th1-type immunological responses and memory and antibody responses with a predominant IgG2a profile. CD8+ T cells with cytolytic activity (CTL) against Salmonella-infected target cells also appear during infection 3.

The NOD mouse provides a good model of type 1 diabetes (T1D). Around 6 weeks of age the female mice develop an infiltration in the pancreas that is initially peri-islet, but progresses to intra-islet when the mice are 10–12 weeks old with beta cell destruction arising through a Th1-mediated autoreactive response. Between 60% and 80% of female NOD develop clinical signs of the disease when the mice reach the age of 30 weeks. As in humans, both genetic 4 and environmental factors contribute to the development of autoimmune diabetes in the NOD mouse. Viral, bacterial and helminth infections can prevent the development of autoimmunity in these mice 57. Tolerogenic protocols using islet antigens which prevent diabetes onset in NOD mice are often most effective if started before the infiltration into the pancreas has occurred, showing that timing of intervention is crucial 8, 9. Complete Freund's adjuvant (CFA) prevents NOD mice from developing diabetes and complete protection can be achieved by administering the CFA between 4 and 10 weeks of age 10. We recently showed that soluble antigens from Schistosoma mansoni parasite were effective in preventing diabetes in the NOD mouse, but only when given between 4 and 8 weeks of age 11. NOD mice are deficient in a population of T cells expressing NK cell receptors and the invariant Vα14Jα281 TCR (iNKT cells) and expansion of this cell type prevents diabetes onset in NOD mice 1215. We have previously argued that one possible mechanism of diabetes protection mediated by S. mansoni soluble antigens might be the activation and increase of iNKT cells 11. In the present study we found that after Salmonella typhimurium infection iNKT cells cannot be detected in the spleen or in the liver of NOD mice by flow cytometric analysis using tetramers comprising CD1d and α-galactosyl β-ceramide (α-GalCer). In contrast to our studies with S. mansoni, we could also find no evidence for a potential involvement of IL-10 in S. typhimurium-mediated diabetes prevention. The simple explanation that a Th1/Th2 switch can prevent the development of diabetes in the NOD mouse is not sufficient to explain diabetes prevention by Salmonella infection. Alternative explanations might include CD4+ T cell expansion and effects of secreted cytokines on the innate immune system.

2 Results

2.1 Live, but not killed, S. typhimurium infection reduces the incidence of T1D in NOD mice

Infection of NOD mice with S. typhimurium significantly reduces the incidence of T1D in female NOD mice (Fig. 1a). Table 1 shows protection can be achieved by infection with S. typhimurium infection at any time between 4 and 12 weeks of age.

Figure 1.

Infection with S. typhimurium protects against the development of T1D. (A) Cumulative prevalence of diabetes following infection with S. typhimurium at 8 weeks of age (filled squares) and in uninfected controls (open circles). Each group comprised 16 mice. (B) Histological scoring at 25 weeks of age of pancreatic islets from 6 control (open bars) and 10 mice infected with S. typhimurium (filled bars) at 6 weeks of age (19 weeks post infection). Data are the median number of islets within each category expressed as a percentage of the median total for each group.

Table 1. S. typhimurium infection protects from development of T1Da)
Age at infection (weeks)Prevalence of diabetes (%) at 33 weeks
NOD controlS. typhimurium-infected NODHeat-killed S. typhimurium-injected NOD
  1. a) Groups of mice were injected with S. typhimurium i.v., heat-killed bacteria or left as controls and monitored for the onset of clinical diabetes. p values are from Fisher's exact test.

48/10 (80%)3/10 (30%)p=0.0709/10 (90%)p=NS
614/16 (88%)2/16 (13%)p<0.0017/10 (70%)p=NS
812/16 (75%)0/16 (0%)p<0.0017/10 (70%)p=NS
1212/16 (75%)2/16 (13%)p=0.00110/12 (83%)p=NS

Histological examination of the pancreas was carried out to determine whether Salmonella infection had influenced pancreatic infiltration. Fig. 1b shows that at 25 weeks of age the pancreases of non-diabetic control NOD mice contain a higher percentage of intra-islet infiltrates compared to the pancreases of age-matched S. typhimurium-infected NOD mice. The percentages of islets that were either non-infiltrated or had a peri-islet infiltrate were correspondingly higher in S. typhimurium-infected mice than in non-diabetic control mice.

The cellular composition of the pancreatic infiltrates was determined by FACS analysis of mononuclear cells isolated from the pancreas of infected and non-infected mice 5 weeks after the infection. Characterization of the infiltrates showed the presence of CD4+ T cells, CD8+ T cells, B cells, NK cells, macrophages and DC (Fig. 2A–F). Comparison of the relative composition of the pancreatic infiltrates in the pancreas of control and infected mice showed a significant difference only in the percentages of CD8+ T cells and the percentage of CD4+ T cells expressing CD62L (Fig. 2G): The percentages of CD8+ T cells were significantly increased (Fig. 2B), while the percentage of CD4+ T cells which expressed CD62L (Fig. 2G) was reduced in the infiltrates in the pancreas from Salmonella-infected mice. Furthermore, amongst the infiltrating T cells in the pancreas of Salmonella-infected mice, there was an increase in the percentage of CD4+ T cells capable of producing IFN-γ (Fig. 2H).

Figure 2.

(A–H) FACS analysis of pancreatic infiltrating lymphocytes after S. typhimurium infection. Mice were infected with S. typhimurium at 8 weeks of age and lymphocytes extracted from the pancreas 5 weeks later. Cells were analyzed for expression of CD4 (A), CD3 and CD8 (B), CD19 (C), DX5 (D), CD11b+ CD11c- (E), and CD11c+ (F). CD4+ cells were also analyzed for CD62L expression (G) and for production of IFN-γ (H). p values compared to uninfected littermate controls are from a Mann Whitney U-test.

To investigate whether live bacteria were required for diabetes prevention, NOD mice at different ages were injected with heat killed S. typhimurium. Table 1 shows that the killed organisms were not able to prevent T1D in NOD mice.

2.2 Splenic lymphocytes from S. typhimurium-infected NOD mice re-stimulated in vitro secrete large amounts of IFN-γ and TNF-α and low amounts of IL-10

Spleens from control and infected mice were taken at different time points during the normal progression towards diabetes in control mice and non-specifically stimulated ex vivo (Fig. 3A–H). As expected, cells taken at an early time point after infection were biased towards the secretion of IFN-γ with little IL-10 production. Elevated production of pro-inflammatory cytokines was sustained for a long period of time after the infection. Nineteen weeks after infection, splenocytes from infected mice still produced larger amounts of IFN-γ than those from non-diabetic control mice (Fig. 3G). IL-4 and IL-10 levels were lower in the supernatants of splenocyte cultures from infected mice at all time points.

Figure 3.

(A–H) Sustained biasing to Th1 cytokine responses after S. typhimurium infection. Splenocytes from uninfected controls (open bars) and NOD mice infected with S. typhimurium at 6 weeks of age (shaded bars) were taken at various time points after infection (A, B 2 weeks; C, D 4 weeks; E, F 8 weeks; G, H 19 weeks) and cultured either in medium alone or restimulated with anti-CD3 or Con A as indicated. Culture supernatants were taken for cytokine determination at 24 h for IL-4 and at 48 h for IL-10 and IFN-γ. Results are means ± SD for data generated from means of six sets of triplicates set up from individual mice per group; p values are for a two-tailed Student's t-test.

The specific response to Salmonella antigen in vitro was assessed by measuring IFN-γ production by purified CD4+ T cells taken from the spleens of mice 4 weeks after infection and re-stimulated in vitro with C5 Salmonella antigen (C5Ag). As can be seen in Fig. 4, CD4+ T cells secrete large amounts of IFN-γ in response to C5Ag. This production of IFN-γ is not seen with purified CD4+ T cells from the spleens of control mice. We have also observed this specific IFN-γ secretion in response to C5Ag by CD4+ T cells isolated from the pancreatic lymph nodes (LN) or from the spleens of mice 22 weeks after infection (data not shown).

Figure 4.

Mice were infected at 6 weeks of age and killed 4 weeks after infection (at 12 weeks of age, when the control mice were not yet diabetic). Purified splenic CD4+ T cells from six control (▴) and six infected mice (○) were cultured for 72 h with or without different concentrations of C5Ag (1 or 5 μg/ml). IFN-γ secretion was measured from culture supernatants by CBA assay. Values from individual mice are expressed in pg/ml; comparison of C1 and C5 data sets by two-tailed Mann-Whitney U test gives p=0.002 in both instances.

These data suggest that a component of IFN-γ secretion, in Salmonella-infected mice, is specific for the bacteria and might be due to long-lived memory T cells.

2.3 Splenic lymphocytes from S. typhimurium-infected non-diabetic NOD mice transfer T1D to NOD-scid recipient mice

To evaluate whether autoreactive effector cells were still present after Salmonella infection, splenocytes from individual diabetic control, non-diabetic control and non-diabetic infected mice were transferred into NOD-scid recipient mice. Splenocytes taken from mice (at 25 or at 35 weeks of age) that had been infected with Salmonella at 6 weeks of age were able to transfer diabetes to the NOD-scid recipients with a time course and final prevalence that was identical to that seen in the recipients of splenocytes from non-infected donors (Table 2).

Table 2. Autoreactive T cells are still present after S. typhimurium-induced protectiona)
Experimental groupPrevalence of diabetes(%)
  1. a) Groups of NOD-scid mice received splenocytes from 25- or 35-week-old control diabetic or non-diabetic NOD mice, or mice which had been infected with S. typhimurium at 6 weeks of age, and monitored for the onset of clinical diabetes.

NOD-scid control recipients(diabetic donors, killed at 25 weeks)6/6(100%)
NOD-scid control recipients(non-diabetic donors, killed at 25 weeks)6/6(100%)
NOD-scidS. typhimurium recipients(non-diabetic infected donors, killed at 25 weeks)8/10(80%)
NOD-scid control recipients(non-diabetic donors, killed at 35 weeks)12/13(84.6%)
NOD-scidS. typhimurium recipients(non-diabetic infected donors, killed at 35 weeks)12/13(84.6%)

2.4 α-GalCer positive iNKT cells are not detectable in the spleen or in the liver of S. typhimurium infected NOD mice

iNKT cell activation and proliferation have been proposed as a possible mechanism in a variety of protocols for diabetes prevention in the NOD mouse, including protection through pathogen infection 11, 16. We have enumerated iNKT cells by FACS-staining of spleen, liver and pancreatic LN from S. typhimurium infected and control mice using CD1d tetramers loaded with α-GalCer. Table 3 shows that for at least 14 weeks after infection, surface expression of the invariant Vα14Jα281 TCR, characteristically expressed by iNKT cells, was undetectable on cells from spleen and liver by FACS analysis using tetramers, since values did not differ significantly from unloaded CD1d tetramer controls.

Table 3. Sustained reductions in iNKT numbers in tissues of S. typhimurium infected NOD micea)
Group% Positive for α-GalCer CD1d-tetramer staining
SpleenLiverPancreatic LN
  1. a) α-GalCer CD1d tetramers were used either alone or together with anti-CD3 to enumerate iNKT cells in cell suspensions prepared individually from the liver, spleen and pancreatic LN of control NOD mice or mice infected with S. typhimurium at 6 weeks of age. Values shown are for individual mice; median values and p values compared to controls from Mann-Whitney U test are given in parentheses.

NOD control(mice 7 weeks old)0.98, 1.00, 1.14, 1.25, 1.38, 1.40(1.20)5.7, 6.21, 6.36, 6.58, 8.09, 10.91(6.47)
NOD S. typhimurium(mice 7 weeks old, 1 week after infection)0.21, 0.25, 0.33, 0.43, 0.45, 0.45(0.38; p=0.004)1.15, 2.03, 2.09, 2.25, 2.34, 2.81(2.17; p=0.004)
NOD control(mice 8 weeks old)0.92, 0.96, 1.03, 1.08, 1.23(1.03)4.23, 6.36, 6.57, 7.52, 9.50,11.20(7.05)
NOD S. typhimurium(mice 8 weeks old, 2 weeks after infection)0.02, 0.02, 0.02, 0.08, 0.2, 0.24(0.05; p=0.006)1.40, 1.69, 2.05, 2.50, 2.70, 2.90(2.28; p=0.004)
NOD control(mice 12 weeks old)0.71, 0.74, 0.77, 0.78, 0.82, 0.86(0.78)7.18, 7.30, 8.61, 9.03, 9.22, 11.14(8.82)0.75, 0.80, 0.83, 0.87, 1.03, 1.14(0.85)
NOD S. typhimurium(mice 12 weeks old, 6 weeks after infection)0.11, 0.17, 0.22, 0.23, 0.24, 0.25(0.22; p=0.004)0.45, 0.88, 0.98, 1.01, 1.27, 1.51(1.00; p=0.004)0.5, 0.52, 0.53, 0.65, 0.70, 0.76(0.59; p=0.007)
NOD control(mice 20 weeks old)1.17, 1.26, 1.37, 1.37, 1.61, 1.63(1.37) 3.18, 3.37, 4.45, 4.91, 5.75, 6.51(4.68)0.11, 0.61, 0.69, 0.74, 0.82, 0.91(0.72)
NOD S. typhimurium(mice 20 weeks old, 14 weeks after infection)0.37, 0.43, 0.50, 0.51, 0.75(0.50; p=0.006)0.76, 1.17, 1.24, 2.02, 2.28, 2.32(1.63; p=0.004)0.28, 0.31, 0.51, 0.80, 1.01(0.51; p=NS)

2.5 Sustained DC subset changes associated with S. typhimurium infection

To determine whether S. typhimurium infection might be altering diabetes susceptibility through long-lasting effects on antigen-presenting cell (APC) behavior, we examined DC subset composition by flow cytometry, using spleens and pancreatic LN from previously infected animals and their littermate controls. Using this approach we detected a decrease in the relative numbers of CD8α+ CD11c+ cells in the spleen that was still significant even 20 weeks after the infection (Table 4).

Table 4. Sustained reduction in CD8α+:CD8α- DC ratio after S. typhimurium infectiona)
Spleen samples% of CD11c+ Positive for CD8α staining
  1. a) Splenocytes from individual control NOD mice and mice infected with S. typhimurium at 6 weeks of age were gated on CD11c+ and the percentage staining positive for CD8α determined. Values shown are for individual mice; median values and p values compared to controls from Mann-Whitney U test) are given in parentheses.

NOD control(mice 13 weeks old)36.16, 38.96, 39.29, 40.09, 42.10, 43.91(39.69)
NOD S. typhimurium(mice 13 weeks old, 7 weeks after infection)22.85, 26.73, 29.40, 34.72, 35.08, 37.35(32.06; p=0.008)
NOD control(mice 26 weeks old)30.34, 34.46, 34.71, 34.72, 35.84, 39.10(34.72)
NOD S. typhimurium(mice 26 weeks old, 20 weeks after infection)25.38, 27.33, 29.88, 30.19, 33.28, 36.21(30.04; p=0.007)
NOD control(mice 32 weeks old)30.86, 31.78, 41.06(31.78)

NOD S. typhimurium

(mice 32 weeks old, 24 weeks after infection)

23.47, 26.94, 28.41, 31.11, 31.70, 33.92(29.76; p=NS)

2.6 Lymphocytes from BDC 2.5 transgenic NOD mice can traffic and proliferate in the pancreatic LN of Salmonella-infected NOD mice

To address whether the changes observed in DC subsets were somehow responsible for diabetes prevention through changes in T cell trafficking or autoantigen presentation in pancreatic LN, we used an adoptive transfer of autoreactive transgenic T cells. T cells from the transgenic BDC2.5 NOD mouse express a TCR with known islet antigen reactivity. Labeling with the cytoplasmic dye 5,6-carboxyfluorescein diacetate succinimidyl ester (CFSE) enables the trafficking and proliferation of transferred cells to be followed. Splenic lymphocytes from transgenic BDC2.5 NOD mice were labeled with CFSE and injected intravenously into control or S. typhimurium-infected mice (three mice per group). At 72 h after injections the cells were recovered from the pancreatic LN, the mesenteric LN and the spleen and proliferation analyzed by flow cytometry. Autoantigen-specific BDC2.5 T cells, identified by expression of the transgene encoded Vβ4 TCR, were detectable in all tissues at 2, 4, 8 and 12 weeks after infection. As expected, proliferation was only seen in the pancreatic LN (Fig. 5). No differences in trafficking or proliferation were observed between control and infected mice at any time point (Fig. 5 and data not shown).

Figure 5.

(A–D) BDC2.5 NOD T cells can home and proliferate in pancreatic LN of control and infected mice. CFSE-labeled BDC2.5 NOD T cells were transferred i.v. into control or S. typhimurium-infected mice. Tissues were harvested at 72 h and cell homing and proliferation assayed. Representative histograms show CFSE staining gated on Vβ4+ T cells from mice at 2 weeks after infection. Percentage of cells that have proliferated are shown (MLN: mesenteric LN; PLN: pancreatic LN).

3 Discussion

In this study we have shown that Salmonella infection can prevent the onset of T1D in NOD mice. Our data show that, as with exposure to CFA 10, diabetes prevention is achieved even when pancreatic infiltration is already established. These observations contrast with our findings with helminth infections 5 or helminth extracts 11, where diabetes could only be prevented if NOD mice were exposed to helminth antigens at around 4 weeks of age. The fact that only live Salmonella infection, and not just bacterial antigens, is able to prevent diabetes suggests that an active interaction with the immune system or the more sustained presence of the pathogen is required to modify the autoimmune outcome. To understand if the protection from diabetes achieved by infecting NOD mice with Salmonella had expanded a stable, transferable population of T regulatory cells we performed adoptive transfer experiments. We found that either at 25 or 35 weeks after infection, splenocytes from Salmonella-infected non-diabetic NOD mice could still transfer diabetes to recipient NOD-scid mice. These data suggest that the anti-diabetogenic effect of the S. typhimurium infection is not due to an expansion of T regulatory cells and that potentially diabetogenic effector cells are not deleted in Salmonella-infected mice. The protection from diabetes achieved by infecting NOD mice with Salmonella strongly contrasts with our previous work on helminth infection, where comparable adoptive transfer experiments demonstrated a lack of active diabetogenic effectors in splenocytes from mice exposed to S. mansoni antigens 11.

As expected, several weeks post infection, the cytokine profile from infected mice was still biased towards the secretion of IFN-γ, and at least a significant component of this can be attributed to a Salmonella-specific memory response. An increase in IFN-γ-secreting CD4+ T cells after infection with live S. typhimurium has been reported previously and shown to be pathogen specific 2. This same study additionally showed that the increase in CD4+ T cell IFN-γ production was significantly greater in mice given the live rather than heat-killed infection, offering an interesting parallel to our finding that protection from diabetes is dependent upon infection with live bacteria, and further reinforcing the potential significance of IFN-γ in this protection. Other investigators have also described an increase in IFN-γ production following S. typhimurium infection 17, but have suggested that this might not be solely pathogen specific, raising the possibility of activation of T cell clones with alternative specificities.

The role of IFN-γ in the development of T1D is a controversial issue. Both inhibition and administration of this cytokine can prevent diabetes in the NOD mouse and in the BB rat model 1821. On the other hand, and more importantly, IFN-γ-deficient mice still develop diabetes and fail to be protected from diabetes by either CFA and BCG, suggesting a key role for IFN-γ in the action of these agents 22. This is of particular interest given our finding of IFN-γ-producing cells within the pancreatic LN and the pancreatic infiltrates. However, it remains unclear how IFN-γ may play a protective role. A number of issues remain to be addressed, including the antigen specificity of the IFN-γ-producing T cells we observe within the pancreas, and whether the effects of IFN-γ are directly on autoreactive T cells or are mediated via APC, as discussed below.

To explore whether changes in the innate immune system, caused by Salmonella infection, can be responsible for the anti-diabetogenic effects of the infection, we decided to investigate the fate of iNKT cells and DC in infected and control NOD mice. iNKT cells are the earliest T cell subset to secrete cytokines in response to infection and to modulate the adaptive immune system 23. Furthermore, these cells have been shown to be numerically deficient in NOD mice and their expansion by exogenous administration of α-GalCer has been shown to prevent diabetes onset 16. This population is also expanded following administration of helminth antigens 11. Of relevance to the studies reported here, Kukreja et al. 13 have shown that administration of BCG and CFA raises iNKT-TCR levels as detected by RT-PCR for expression of invariant Vα14Jα281 TCR mRNA in lymphocytes of NOD mice protected from diabetes.

The observation we report here of decreased iNKT numbers is in marked contrast to these previous reports and our own findings following helminth infection of NOD mice, where increased iNKT cell numbers were observed using this same technique. We are continuing to investigate the fate of iNKT cells after S. typhimurium infection, to differentiate between possible outcomes including internalization of the Vα14Jα281 TCR, cell migration or apoptosis, as previously suggested 24.

DC have been shown to take up live S. typhimurium and prime T cell responses in vivo2. Furthermore, immunohistochemical studies on the spleens of mice during an acute infection revealed a redistribution of DC subsets. In our experiments Salmonella-infected mice show a sustained reduction in CD8α+ CD11c+ DC. It has recently been suggested that CD8α expression is correlated with DC maturation, representing the last stage of a common differentiation pathway 25. Up-regulation of CD8α expression by previously CD8α DC resident in the splenic marginal zones may parallel the up-regulation of other markers, including CD205, observed upon migration of these DC into T cell areas of the white pulp in response to stimuli such as bacterial lipopolysaccharide 26. Thus, it was important to establish whether a reduction in the CD8α+:CD8α DC ratio was a general feature observed in all NOD mice protected from diabetes, effectively as an outcome rather than an cause of protection. We therefore compared pre-diabetic female NOD mice with age-matched male NOD and female NOD-E mice, the latter being protected from the development of diabetes by the expression of H-2E 27. In these mice we failed to detect differences in the CD8α+:CD8α DC ratio (data not shown), indicating a degree of specificity to our observations in S. typhimurium-infected animals. These findings suggest that the S. typhimurium infection has somehow altered DC developmental pathways and hence antigen-presenting behavior either directly through effects on precursor populations, or through effects mediated via other cell types. Our finding that α-GalCer CD1d+ cells are not present in the spleens or livers of previously infected animals suggests one possible pathway, since CD1d-restricted iNKT cells have been shown to be regulators of DC subset development and impact upon CD8α expression 16.

Islet antigen is still presented in the pancreatic LN of Salmonella-infected, diabetes-protected NOD mice, as shown by the proliferation of transferred CFSE-labeled BDC2.5 TCR transgenic T cells. The outcome of such a proliferative event remains to be clarified.

Apart from the possibility that a lack of iNKT cells contributes to DC subset changes, it may also be the case that high levels of IFN-γ produced in response to infection have a direct effect on DC phenotype and function. In this regard we note experiments showing that DC exposed to IFN-γ are able to transfer diabetes protection 28 and that IFN-γ-conditioned CD8α+ DC inhibit T cell responses through effects on tryptophan metabolism 29. Intriguingly, a deficiency in the response to IFN-γ has been shown in the CD8α+ DC population of young female NOD mice, and has been correlated with the predisposition for T1D in this strain 30. Since Salmonella infection protects NOD mice from diabetes development at early as well as late time points, this suggests that other mechanisms in addition to the actions of IFN-γ on DC must account for the protective effects seen in young female mice.

Thus, an alternative model of immunomodulation emerges involving in part high levels of IFN-γ secreted during infection with potential effects on antigen presentation by DC and up-regulation of tolerogenic pathways.

In conclusion, our data presented here offer evidence for the action of an infection to modify the outcome of a spontaneous autoimmune process, even at a stage when infiltration of the target tissue by cells of the immune system is well established and when other immunomodulatory interventions have been ineffective. By further study of this model and comparison with autoimmune outcomes following other infections, we would hope to be able to better understand the mechanisms through which the environment may modify spontaneous autoimmune disease on a susceptible genetic background. Ultimately, such studies may help to shed light on potential common mechanisms and suggest suitable targets for therapeutic intervention.

4 Materials and methods

4.1 Animals

NOD, NOD-scid and BDC2.5 NOD mice were obtained from breeding colonies established in the Pathology Department, University of Cambridge animal facilities. During the experiments all the mice were maintained in the same standard conditions with free access to food and water.

4.2 Bacteria

S. typhimurium SL3261 is an aroA attenuated live vaccine strain with an i.v. LD50 for BALB/c mice of 107 cfu 31. Bacteria were grown at 37°C as stationary overnight cultures in Luria Bertani (LB) broth (DIFCO). The inoculum was diluted in PBS and mice were injected with 5×105 cfu S. typhimurium SL3261 in a lateral tail vein. The inoculum was checked by pour plating on LB agar.

4.3 Histological examination

Pancreases were processed for wax histology. Five-micrometer sections were taken at eight levels (200 μm apart) and stained with hematoxylin and eosin. Total islets per section were counted and the degree of cellular infiltration was scored. Scoring was as follows: non-infiltrated islet, peri-islet infiltrate (i.e. up to 20% of the islet is infiltrated) and intra islet infiltrate (21–100 % of the islet is infiltrated).

4.4 Extraction of pancreatic infiltrating lymphocytes

Each pancreas was harvested individually and torn into pieces in cold PBS containing 5% FCS, 56 mM glucose (Sigma) and Complete Mini Protease inhibitors (Roche). The tissues were washed twice in cold PBS, prior to incubation in 2 ml pre-warmed PBS containing 15% FCS and Liberase CI (Boehringer Mannheim). After digesting, the tissues were washed and cell suspensions prepared by forcing through a cell strainer. Suspensions were twice left to settle and the supernatants decanted to remove stromal debris, before the cells were washed and used for FACS analysis or intracellular cytokine staining as below.

4.5 Adoptive transfer of T1D into NOD-scid mice

Splenocytes (2×107/mouse) from Salmonella-infected or non-infected NOD female mice were injected i.v. into 6-week-old NOD-scid female mice. Recipient NOD-scid mice were tested weekly for the development of diabetes by measurements of urine glucose.

4.6 Splenocyte culture and cytokine assays

Single-cell suspensions were prepared and adjusted to 5×105/well (for non-depleted splenocytes) in 96-well round-bottom plates (Falcon) in 200 μl RPMI 1640 (Gibco) supplemented with 5% FCS (Harlan), 2 mM L-glutamine (Gibco), 100 mg/ml Streptomycin (Gibco), 100 U/ml Penicillin (Gibco). Anti-CD3 (10 μg/ml, clone 145–2C11, BD PharMingen) or Con A (2μg/ml, Sigma-Aldrich) were added in vitro and supernatants harvested at appropriate time points for cytokine measurements. Concentrations of IFN-γ (48 h), IL-10 (48 h) and IL-4 (24 h) were quantified in supernatants, using a sandwich ELISA, according to the manufacturer's instructions (R&D Systems).

CD4+ T cells were positively selected from spleens and pancreatic LN from control and Salmonella-infected mice by sorting using PE-conjugated anti-CD4 (BD PharMingen) and a MoFlo cell sorter (DakoCytomation). The purified CD4+ T cells were cultured (1×105/well) in vitro in with irradiated syngeneic APC (CD4+ T cell and APC at a 1:1 ratio) with or without C5Ag (1 or 5 μg/ml). The supernatants were harvested at 72 h and IFN-γ secretion was detected by mouse Th1/Th2 cytokine cytometric bead array assay (CBA, BD Bioscience) according to the manufacturer's instructions.

4.7 Purification and isolation of C5Ag

Overnight, stationary culture of S. typhimurium C5 in LB broth was pelleted, washed once in PBS containing 5 mM EDTA, and once more in PBS. The suspension was sonicated on ice, cellular debris were removed by centrifugation at 13,000×g. The supernatant was sterile filtered through a 0.22−μm filter (Sartorius) and stored at –70°C. The protein concentration was determined using the bicinchroninic acid Kit (Pierce Biochemicals) following the instructions of the manufacturer. Alkali-treated antigen was prepared by addition of NaOH up to 0.25 M, incubating at 37°C for 3 h before the mixture was neutralized with HCl and sterile filtered through a 0.22-μm filter.

4.8 Intracellular cytokine staining

Cells were incubated in medium containing 50 ng/ml PMA (Sigma) and 500 ng/ml ionomycin (Sigma) for 4 h, with the addition of 5 μl/ml Brefeldin A (Sigma) after 2 h. Cells were then washed and blocked in 2.4G2 before staining with anti-CD3 FITC and anti-CD4 PerCP. After washing, cells were incubated for 15 min at room temperature in fixation medium (Caltag) before staining with anti-IFN-γ-PE (XMG1.2), anti-IL-10-PE (JES5–16E3) or appropriate isotype control (all antibodies from BD PharMingen) in permeabilization buffer (Caltag). Washed cells were fixed overnight in 1% formaldehyde prior to analysis using a FACScan flow cytometer (Becton Dickinson).

4.9 Flow cytometric analysis

Single-cell suspensions were prepared from liver, spleen and pancreatic LN harvested from NOD mice killed at different time points after infection and from littermate controls. Nonspecific binding was blocked by incubation with antibody clone 2.4G2 (from tissue culture supernatant) and neutravidin (Molecular Probes). Cells were washed and resuspended in staining buffer (PBS containing 2% BSA and 0.05% NaN3). As an additional step for DC subset analysis, spleens and LN were first cut into small fragments and digested with frequent mixing for 25 min at 37°C in medium containing Liberase CI. DC-T cell complexes were then disrupted by the addition of EDTA (0.1 M, pH 7.3). Cells were stained with appropriate combinations of FITC, PE or PerCP conjugates (all from BD PharMingen) of: anti-CD3 (145–2C11), anti-CD4 (RM4–5), anti-CD8α (53–6.7), anti-CD11c (HL3) and appropriate isotype control antibodies. α-GalCer-loaded CD1d-PE tetramers 32 were also used along with CD1d-PE tetramers left unloaded as a negative staining control. Cells were then washed and analyzed as above.

4.10 CFSE labeling and proliferation assays

Single-cell suspensions of splenic lymphocytes from BDC2.5 NOD transgenic mice were prepared and red blood cells lysed in ammonium chloride buffer. Washed cells were resuspended at 5×107/ml in PBS with 5 μM CFSE (Molecular Probes) and incubated at 37°C for 30 min. Cells were then blocked in PBS containing 5% FCS. After washing, cells were resuspended and 2×107 injected into a lateral tail vein of each mouse. After 72 h, mice were killed and spleens, pancreatic and mesenteric LN harvested. Single-cells suspensions were prepared, stained for Vβ4 PE (clone KT4, BD PharMingen) and analyzed as described above.

4.11 Statistics

Appropriate statistical tests were performed on all data as described in the figure and table legends.


This work was supported by Diabetes UK and the Wellcome Trust. The authors are grateful to Dr. J. Phillips and Dr. R. Allen for helpful discussion. We are also thankful to Dr. A. Stewart for critical reading of the manuscript and Mr. Nigel Miller and Mr. B. Potter for technical assistance.


  1. 1


  2. 2


  3. 3


  4. 4


  5. 5