Following RRV rotavirus infection as an infant, it is shown here that virus replication in the intestine and MLN is greater in diabetes-prone NOD mice than diabetes-resistant mouse strains. Furthermore, the extent of diarrhoea varies between mouse strains and is related to the degree of RRV replication in the intestine, MLN and thymus. Higher gastroenteritis rates also correlate with increased numbers of CD8+ IEL showing prolonged elevations in PD-L1 expression. We have demonstrated for the first time that RRV spreads to the mouse thymus and can alter T cell development. Although thymocyte production is reduced at 3 weeks after RRV infection in NOD mice, thymic Treg numbers are maintained. Furthermore, at 17 weeks of age in NOD mice given RRV as infants, PLN T cells show decreased proliferative ability in response to a strong stimulus. It is proposed that the maintenance of thymic Treg numbers at the time of reduced T cell production leads to this decreased T cell proliferative response in the PLN, which could contribute to the delayed diabetes onset in NOD mice following RRV infection as an infant.
IEL also accumulate during murine viral infection with reovirus 
, and intestinal T cell numbers are increased in rotavirus-infected calves 
. RRV infection induces a rapid expansion of CD8αβ TCRαβ IEL in infant NOD mice () but not in adult NOD (Fig. S1
) or C57BL/6 mice 
. RRV infection had no effect on CD8+
TCRγδ IEL, which increased as a proportion of total T cells from 19±4% on day 14 post infection to 47±2% on day 21 and 59±1% on day 35 in both mock- and RRV-infected NOD mice (C. Zufferey and B. S. Coulson, unpublished data). These results confirm previous findings that the TCRγδ IEL frequency in mice reaches ~50% by 20 days of age, and are consistent with the lack of effect of reovirus infection on TCRγδ IEL frequency compared to control mice 
. The low level of RRV replication in infant C57BL/6 mice may explain their unaltered numbers of CD8αβ TCRαβ IEL and rapid decrease in PD-L1 expression intestinally. The extensive RRV replication by day 1 post infection in NOD and BALB/c mice results in higher numbers of CD8αβ TCRαβ IEL to fight infection. Endothelial MHC class I levels are higher in NOD than BALB/c mice, influencing T cell proliferation, adhesion and transmigration of CD8+
T cells in NOD mice 
. The elevated CD8αβ TCRαβ T cell numbers in the IEL compartment of NOD over BALB/c mice () could result from increased RRV antigen presentation on MHC class I.
The increased frequency of CD8αβ TCRαβ IEL and their extended PD-L1 expression is evident in BALB/c and NOD but not C57BL/6 mice, and might relate to the degree of virus replication. The increased PD-L1 expression at 2 days post infection suggests a role for PD-L1 in an enhanced immune response for clearance of rotavirus infection. PD-L1 interaction with PD-1 plays an important role in preventing autoimmune disease by inducing T cell anergy and tolerance 
. However, in pathogen-specific T cell responses, PD-L1 is involved in T cell activation, proliferation and cytokine production 
. In line with this latter PD-L1 role, our results show that PD-L1 expression correlates with the extent of intestinal rotavirus replication in all mouse strains. PD-L1 expression on CD8+
T cells in other tissues may prove to be a useful marker of rotavirus infection.
Our detection of RRV in thymic macrophages extends the earlier identification of macrophages as the likely main RRV source in the infant NOD mouse pancreas, and supports the proposal that macrophages disseminate rotavirus infection 
. Altered thymic T cell development following infection by any virus that delays diabetes has not been reported to date. However, in an adult SJL/J mouse model of virus-induced diabetes, coxsackievirus B4 infection of the thymus, pancreas and other organs was associated with an increased DN thymocyte frequency but unaltered overall thymocyte numbers at 21 days post infection 
. In fetal thymic organ cultures from CD-1 mice, coxsackievirus B4 increased the frequency of DN, CD4 SP and CD8 SP cells and decreased DP cells 
. Thymic RRV infection in NOD mice also was associated with increased proportions of CD8 SP and DN, and decreased DP cells, although CD4 SP cell proportions were unaltered. However, total thymocyte numbers were decreased due to reductions in CD4 SP, DP and DN cells, in contrast to the coxsackievirus B4 studies 
. Thus, reduced overall thymyocyte numbers, combined with maintenance of CD4 SP cells, might contribute to the RRV-mediated diabetes delay. However, the effect of RRV on thymocytes is mouse strain-dependent, with most subsets and total cell numbers decreased in C57BL/6 mice, but no effect in BALB/c mice. The differing thymic effects of coxsackievirus and rotavirus in diabetes mouse models may relate to their contrasting effects on diabetes and the mouse strain utilized.
NOD mice show a normal ability to generate and maintain Treg in the thymus 
. However, diabetes develops as Treg become functionally defective in the periphery with age, and self-reactive Teff increasingly develop resistance to Treg-mediated suppression 
. We found that RRV-infected NOD mice maintain thymic Treg production at day 21 post infection when CD4+
Teff and other thymocyte subsets are reduced. At this time, the number and proliferative function of T cells in the PLN and MLN of mock- and RRV-infected NOD mice are similar ( and data not shown), as expected once infection is cleared and homeostasis has been restored. PLN T cells from NOD mice aged 17 weeks, which had been RRV-infected as infants, are less proliferative to a strong T cell stimulus than cells from age-matched, mock-infected mice. It is proposed that RRV infection of infant NOD mice may prolong Treg immunosuppression, which could contribute to their delayed diabetes onset. A lowering of Teff resistance to Treg-mediated regulation 
during RRV infection also may be a factor in this reduced T cell proliferation that occurs following a strong stimulus.
The wide range of immunological defects in NOD mice includes resistance to thymic deletion and increased production of autoimmune T cells 
. For example, DP thymocytes from NOD mice have a lower activation threshold than those of C57BL/6 and BALB/c mice 
. Although we have associated Treg maintenance and decreased T cell proliferation after a strong T cell stimulus with the delayed diabetes after RRV infection, it is also possible that rotavirus infection of the NOD thymus induces stronger activation of the normally weakly activated self-T cells, leading to their deletion and contributing to the delayed diabetes onset. Further studies now are warranted to address the effects of rotavirus infection on thymic T cell activation, and the role of immunosuppression by Treg in the delayed diabetes mediated by RRV.