In contrast, none of the LPS-treated males developed diabetes (Fig. S1). Initiation of the treatment in NOD females at 12 weeks LDK378 of age, when mononuclear infiltration of Langerhans islets is readily detectable ([48] and not shown), also prevented progression to diabetes (Fig. 1B). However, administration of LPS after positive scoring for diabetes did not revert disease (data not shown). We next tested shorter LPS treatments. A single LPS injection into 7.5-week-old NOD females delayed diabetes onset by an average of 7 weeks but was not sufficient to significantly decrease diabetes incidence (Fig. 1C). Finally,

administration of LPS in 4-week-old female mice for 1 month resulted in 15 weeks delay in diabetes progression as compared with age-matched PBS-injected controls (Fig. 1D). We conclude that LPS is a potent inhibitor of diabetes occurrence in NOD mice.

The finding that continuous exposure to LPS protects HIF inhibitor NOD mice from diabetes, even after extensive infiltration of the pancreatic islets, suggests that LPS prevents insulitis progression. Our evidence that interruption of LPS treatment systematically leads to reactivation of disease, and hence diabetes establishment, supports the notion that the LPS effect is transient and it is exerted by maintaining diabetogenic T cells at check. Thereafter, to perform the cellular and functional analysis of LPS-protected NOD females, we chose the robust and long-lasting weekly regimen initiated in 6- to 8-week-old mice (Fig. 1A). It is still not known why few NOD females do not spontaneously progress to diabetes while they all reach Megestrol Acetate the stage of insulitis. Yet, it is well established that female NOD mice raised in germ-free conditions all develop disease [49]. Therefore, it was conceivable that LPS treatment would mimic an environmental factor of bacterial

origin present, although limited, in our SPF conditions. This reasoning prompted us to compare the two types of disease-free animals, namely LPS-treated and spontaneously protected, in what concerns sub-clinical signs of autoimmunity (Fig. 2A, B). To this aim it was necessary to focus our analysis on rather old animals (5–6 months of age), to increase the odds that the untreated normoglycemic controls were indeed spontaneously protected animals. In a first step, we evaluated whether the protective regimen affected directly the degree of islet infiltration. As expected, the majority of the islets in diabetic females presented severe infiltration; moreover, islet destruction was evident as indicated by a low number of detectable pancreatic islets (data not shown). Strikingly, LPS-treated mice were indistinguishable from age-matched healthy controls, as the majority of islets were devoid of infiltrates (60% and 66%, respectively), while the remaining islets displayed various degrees of infiltration, from mild to severe.