Commentary: Endogenous glucocorticoids control host resistance to viral infection through  tissue-specific regulation of PD-1 expression on NK cells

The key purpose of the immune system is to protect the host from infectious diseases. However, it has long been known that the induced host response in many cases represents a double-edged sword, and proper regulation of the magnitude and the temporal and spatial context of this response is critical to avoid immune-mediated bystander tissue damage. A classical model of such damage is intracerebral infection of immunocompetent mice with the noncytolytic arenavirus lymphocytic choriomeningitis virus. Although T-cell-deficient mice suffer little or no disease following infection by any route, immunocompetent mice succumb to T-cell-mediated lethal meningitis.^1,2 Peripheral infection of such immunocompetent mice is associated with limited disease, and the same type of effector T cells that cause lethal disease are also central in controlling viral infection in organs.^3,4 Through the staggered adoptive transfer of effector T cells into virus-infected, immunosuppressed hosts, it has clearly been demonstrated that the timing of the immune response relative to maximal viral load in the CNS is the decisive factor.⁴ An early potent response such as that observed in previously immunized mice will protect, whereas a similar response initiated when the viral load is at its peak causes an immunological conflict that kills the host. Although it is a somewhat contrived model of immunopathology, this infection has nevertheless served as the prototype of virus-induced immunopathology for many years. However, as our knowledge about infection immunology has increased, it has become very  clear that host immunity is often associated with some form of collateral damage. The notorious cytokine storm associated with certain types of flu virus infection is one such situation.^5,6 Following up on this realization, much effort is currently being devoted to attempting to understand how the immune response is regulated to minimize immune-mediated tissue damage but maintain efficient host defense.

Glucocorticoids are well known for their capacity to restore homeostasis by reducing and counteracting many types of inflammation.⁷ In their recent paper on how endogenous glucocorticoids may modulate the NK cell response to murine cytomegalovirus (MCMV) in a tissue-specific manner, Quantrini et al.⁸ have added a new dimension to our understanding. It has long been known that adrenalectomy causes mice to become more susceptible to MCMV-induced disease and that resistance is restored by corticosterone replacement.⁹ However, until now, the underlying mechanism remained unclear. In their study, Quantrini et al. used mice with a conditional deletion of cytosolic glucocorticoid receptors in innate immune cells expressing the activating NK cell receptor NCR1 (NKp46). Using this approach, the authors showed that endogenous glucocorticoids produced as a result of viral infection reduced splenic inflammation in wild-type mice compared with mice with the target GR defect. A comparison of the functionality of activated splenic NK cells in wild-type and GR-deficient mice revealed that the ability of wild-type NK cells to produce IFN-γ was reduced relative to that of matched GR-deficient cells. In contrast, the expression of granzyme B was similarly upregulated in both cases. RNA-based next-generation sequencing revealed downregulation of Pdcd1 in NK cells from GR-deficient mice relative to corresponding cells from infected wild-type mice; following that lead, the authors demonstrated that the inhibitory receptor PD-1 was upregulated on splenic NK cells from wild-type mice but not on the same cells from infected GR-deficient mice. Interestingly, PD-1 expression was not upregulated on liver NK cells, a finding consistent with the fact that MCMV-induced liver inflammation was similar regardless of genotype. Regarding the reason why liver and spleen NK cells behaved differently, the authors pointed to a significant difference in the cytokine environment in the infected liver vs the spleen. Confirming the essential role of the GR-PD-1 axis as a regulator of MCMV-induced IFN-γ production, the authors demonstrated that injection of PD-1 blocking antibody reduced IFN-γ production in the spleen and MCMV-induced mortality in wild-type mice, whereas neutralizing anti-IFN-γ “rescued” GR-deficient mice.

Thus, via GR-dependent regulation of PD-1 expression on splenic NK cells, endogenous glucocorticoids modulate virus-induced IFN-γ production in the spleen and prevent severe immunopathology in this organ. Notably, this regulatory pathway operated in an organ-specific manner and did not impair virus control. The latter observation is consistent with other data showing that NK cells in the spleen exert their antiviral functions in a perforin-dependent manner.¹⁰ In contrast, in the liver, NK cells primarily operate via IFN-γ production.^10,11 Based on this organ-specific mode of NK operation, the GR-mediated anti-inflammatory effect on NK cells avoids negatively impacting virus control in either organ.

These results are extremely interesting and uncover a new way in which host regulatory mechanisms may manage to modulate the immune response to minimize collateral damage without losing significant antimicrobial activity. In the future, it will be interesting to see how generally this and other forms of specific GR-mediated immune regulation may be observed in the context of infectious diseases. In addition, with respect to tumor immunology, a clearer understanding of the interplay between physiological stress, glucocorticoids, and immune cell activity is pertinent and could open new avenues for further improving our possibilities to exploit the immune system in the control of cancer.

Competing interests

The author declares no competing interests