Whereas numerous studies have investigated the role of commensal bacteria in mucosal homeostasis, it is easy to forget that the gut microbiome also contains fungi and viruses. Previous work has associated the dysbiosis of commensal viruses with inflammatory bowel diseases, but the underlying mechanism was not known. Reporting in Nature Immunology, Liu et al. describe a role for commensal viruses in maintaining intraepithelial lymphocytes (IELs) through the retinoic acid inducible gene I (RIG-I)-dependent production of IL-15.

Credit: Vicky Summersby/Springer Nature Limited

These authors used an antiviral cocktail (AVC) to deplete commensal viruses in the gut of C57BL/6J wild-type mice. After 6 weeks of treatment, starting at 2 weeks after birth, the number of faecal virus-like particles (from both DNA and RNA viruses) was reduced markedly. AVC-treated mice had decreased numbers of IELs in the small intestine and colon, in particular CD8αα+TCRαβ+ and CD8αβ+TCRαβ+ IELs. There were no differences between AVC-treated and control mice in terms of lymphocyte numbers in other organs or percentages of the major immune cell subpopulations. AVC treatment further reduced the number of IELs in antibiotic-treated mice, which suggests that the effect involves eukaryotic viruses and is not owing only to bacteriophage depletion.

Further studies showed that virus recognition for IEL maintenance depends on RIG-I, but not other virus-sensing receptors. Ddx58–/– mice (which lack RIG-I) and mice deficient for mitochondrial antiviral signalling protein (MAVS; the adaptor protein for RIG-I) had a similar phenotype to AVC-treated mice, and AVC treatment of Ddx58–/– mice had no further impact on IEL numbers. Intraperitoneal supplementation with a RIG-I ligand rescued the IEL loss in AVC-treated Ddx58+/+ mice but not in Ddx58–/– mice. Therefore, commensal viruses support IEL numbers through RIG-I–MAVS signalling. Mice with defective RIG-I–MAVS signalling had normal numbers of thymic precursors for IELs but lower levels of IEL proliferation and higher levels of IEL apoptosis. Thus, RIG-I–MAVS signalling regulates IEL homeostasis rather than development. Although AVC treatment and Ddx58 knockout can also result in bacterial dysbiosis, the authors used co-housing experiments to rule this out as a cause of IEL loss.

A series of adoptive bone-marrow transfer experiments showed that IEL-extrinsic RIG-I signalling in haematopoietic cells is required for IEL homeostasis. Conditional knockout of Ddx58 in CD11c+ dendritic cells (DCs) and macrophages resulted in decreased numbers of CD8αα+TCRαβ+ and CD8αβ+TCRαβ+ IELs, which indicates a crucial role for RIG-I signalling in antigen-presenting cells (APCs). RIG-I signalling had no effect on the development of intestinal macrophages and DCs. Rather, APCs from Ddx58–/– mice and AVC-treated mice expressed much lower levels of IL-15 than wild-type and control-treated APCs. Furthermore, viral vector-mediated gene delivery of Il15 to Ddx58–/– mice and AVC-treated mice restored their numbers of IELs. Together, the results indicate that APC-derived IL-15 maintains IELs in response to commensal virus-induced RIG-I–MAVS signalling. Furthermore, the transcription factor IRF1 was shown to have a crucial role in IL-15 induction mediated through this pathway.

IELs have well-described roles in homeostasis and host defence of the gut mucosa. In keeping with this, both AVC-treated and Mavs–/– mice were more susceptible than control-treated and wild-type mice to dextran sulfate sodium (DSS)-induced colitis; this susceptibility could be reduced by Il15 gene delivery to restore IEL numbers.

In summary, this study indicates that commensal viruses have similar, but independent, physiological roles to commensal bacteria in maintaining intestinal homeostasis and thus should not be overlooked in studies of intestinal health and disease.