The significance of CD8
+ T cells in polyomavirus control is emphasized by studies demonstrating a strong association between an early JCPyV-specific CD8
+ T cell response and a favorable clinical outcome in PML patients [
275,
276]. In clinical studies exploring pembrolizumab, an inhibitor of the immunosuppressive protein programmed cell death protein 1 (PD-1), as a treatment for PML, an increased JCPyV-specific CD8
+ T cell response was linked with lower viral loads and, ultimately, clinical improvement [
3]. CD8
+ T cells have been shown to colocalize with JCPyV-infected glial cells, further validating the importance of CD8
+ T cells in controlling JCPyV [
277,
278]. The necessity of CD8
+ T cells to viral control in an immunocompromised setting is demonstrated
in vivo in STAT1 null mice, which are immunocompromised and exhibit high viral loads in the brain when infected with MuPyV. When depleted of CD8
+ T cells, these mice display severe weight loss and increased mortality by two weeks post-infection compared to WT animals [
185].
CD8
+ resident memory T (T
RM) cells constitute a subset of CD8
+ T cells and are critical responders to viral reinfection in non-lymphoid tissues, such as the brain. T
RM cells are semi-permanent fixtures of non-lymphoid tissues and are responsible for rapidly countering repeated antigen encounters [
279,
280]. The expression of chemokine receptor CXCR6 is associated with the transition of CD8
+ T cells from circulating effectors to resident memory [
281,
282]. Previous studies have shown CXCR6 and its chemokine ligand CXCL16 to be important for T
RM formation in the lung [
283], skin [
284,
285] and liver [
286]. In the brain, T
RM cells require CXCL16-CXCR6 signaling after resolution of West Nile virus infection for maintenance, as loss of CXCR6 leads to reduced numbers of CD8
+ cells [
287]. This chemokine signaling was also required for T
RM formation in an Alzheimer’s disease mouse model, and CXCR6 null mice had reduced CD8
+ T cells with increased cognitive decline [
288]. The dominant source of CXCL16 in the brain is myeloid cells including microglia [
287,
288]. With MuPyV infection, we have observed CXCR6
+ CD8
+ T cells in the kidney and brains of infected mice, as well as expression of CXCL16 by brain microglia (M Lauver, S Spencer and A Lukacher, unpublished observations). In CXCR6 deficient mice, CD8
+ T cells show reduced expression of T
RM marker CD69 [
289] after infection, suggesting a requirement of CXCR6 for the establishment of T
RM cells. Mice without CXCR6 signaling infected with MuPyV showed reduced expression of PD-1 on CD8
+ T cells in the brain (S Spencer and A Lukacher, unpublished observations).
PD-1, a classic inhibitory immune receptor, is an intrinsic attribute of virus-specific CD8
+ brain-resident T
RM (bT
RM) cells established during MuPyV infection [
111]. Notably, PD-1 expression is upregulated on JCPyV-specific CD8
+ cells in PML patients [
289]. Increased PD-1 and PD-L1 expression have also been observed in autopsied PML lesions [
3]. MuPyV-specific CD8
+ bT
RM cells highly express PD-1, a characteristic not shared by MuPyV-specific memory CD8
+ cells in the spleen, supporting PD-1 as a brain-intrinsic trait of CD8
+ T
RM cells [
111]. Disrupting PD-1 signaling influences bT
RM cell differentiation. MuPyV-infected mice lacking programmed cell death ligand 1 (PD-L1), one of the two canonical ligands for PD-1, present a higher frequency of virus-specific CD103
+ CD8
+ cells compared to wild-type mice [
111]. CD103 is an alpha integrin considered to be a broad marker of T
RM cells [
290]. Wild-type mice treated with a blocking PD-L1 antibody also exhibit this increase in CD103
+ cell frequency, an observation recapitulated by PD-1
fl/fl Rosa-Cre
ERT2 mice, in which PD-1 can be conditionally deleted with tamoxifen administration. Upon PD-1 deletion after MuPyV infection, these mice demonstrate an increase in CD103
+ cell frequency compared to vehicle-treated animals [
291]; A Butic and AE Lukacher, unpublished observations]. PD-1 has also been shown to constrain the effector functions of MuPyV-specific CD8
+ bT
RM cells
ex vivo. Bone marrow-derived dendritic cells (BMDCs) from wild-type and PD-L1
-/- mice were used to stimulate brain CD8
+ cells isolated from MuPyV-infected mice. PD-L1 null BMDCs incited a lower frequency of cells that were positive for IFN-γ, a classically inflammatory and antiviral cytokine, compared to wild-type BMDCs [
291]. Inflammatory gene expression in wild-type and PD-L1 deficient mice indicate that inflammatory genes were upregulated in PD-L1
-/- mice during acute infection [
291]. Similarly, brain sections from acutely infected, tamoxifen-treated PD-1
fl/fl Rosa-Cre
ERT2 mice exhibit increased glial fibrillary acidic protein (GFAP) and ionized calcium binding adaptor molecule 1 (Iba1) immunofluorescent staining compared to vehicle-treated mice [A Butic and AE Lukacher, unpublished observations]. GFAP and Iba1 are markers of reactive gliosis, an aspect of neuroinflammation. More specifically, GFAP is upregulated in activated astrocytes, while Iba1 is increased in microglia [
292,
293]. Unexpectedly, inflammatory gene expression is reduced in PD-L1 null mice during persistent infection [
291]. Collectively, these findings portray PD-1 as a dynamic regulator of MuPyV-induced neuroinflammation and a potentially important factor in MuPyV-specific bT
RM cell differentiation. Exploring the factors that contribute to bT
RM cell differentiation and also function and maintenance is critical to designing therapies for PML as well as improving patient stratification for therapies such as pembrolizumab. Although pembrolizumab has shown clinical effectiveness in some PML patients, its effects are still inconsistent, with some patients showing no response [
3,
294].