Background
Infection of the central nervous system (CNS) presents unique challenges to effective pathogen control, as brain infection may rapidly progress causing substantial damage or even death. Neuroimmune responses are critical for antiviral defense, but extensive damage to this generally non-regenerating tissue must be avoided [
1]. It is well established that different immune mechanisms are very specifically tailored to control infections in particular organs. Recent studies have demonstrated that after clearance of many acute viral infections, CD8
+ T lymphocytes generate a population of long-lived, non-recirculating tissue-resident memory cells (T
RM) in non-lymphoid tissue; and it is becoming increasingly clear that these T
RM cells play critical roles in controlling re-encountered infection and accelerating the process of pathogen clearance [
2‐
5].
The CNS can be a target of acute viral infection, as well as a reservoir of latent and persistent virus. During acute viral infection, most pathogens are rapidly cleared through the generation of a large number of short-lived effector T cells (SLEC). Simultaneously, the T cell response is triggered to generate a subset identified as memory precursor effector cells (MPEC). These MPEC begin to develop into a tissue-resident memory (T
RM) phenotype shortly after infection. Recent work by several groups provides evidence that there is a clear distinction between terminal effector and memory cells based on heterogeneity in expression of killer cell lectin-like receptor G1 (KLRG1) [
6‐
8]. We have recently characterized brain-infiltrating T cells which persist within the tissue after acute murine cytomegalovirus (MCMV) infection. We showed that infiltrating CD8
+ T cell populations shift from SLEC to clear infection to MPEC that protect against re-challenge. The shift of prominent SLEC populations to MPEC populations is concomitant with transition from acute through chronic phases of infection. In addition, these cells were found to selectively express the integrin CD103, a marker of brain T
RM (bT
RM) cells and persist long-term within the CNS [
9].
Resolution of adaptive immune responses and generation of immunological memory is an essential process to confer long-term protective immunity particularly in immune-privileged tissue-like brain. Inflammation within different anatomical sites of brain dramatically increases the infiltration and migration of lymphocytes and effector molecules. We understand much about the infiltrating T cell mediated immune response and the penetration of T cells within the infected brain parenchyma [
10]. However, better understanding of the association between inflammation and the establishment of T
RM will inform us about the protective effects of neuroimmune responses to re-infection or viral reactivation.
T
RM cells are characterized by their non-recirculating, resident nature in tissues. It is well reported that T
RM cells often express α
Eβ
7. α
E, otherwise known as CD103, has been identified as a marker of particular types of T
RM cells. High expression of CD103 and CD69 is a common feature of resident memory cells observed in epithelial tissue [
11,
12]. Whereas, effector and resident memory cells in circulation appear to lack expression of both CD103 and CD69 [
13,
14]. It has been shown that CD69 expression is required for the optimal formation of T
RM following herpes simplex virus (HSV) infection in tissues such as the skin and dorsal root ganglia [
2,
15]. In addition, experiments using the skin, lung, and gut show differential expression of CCR7, as well as CXCR3, which define the migration properties of T cells [
16‐
18]. However, further insight into factors responsible for development of T
RM is required. Given the importance of the formation of brain (bT
RM) cells, there is surprisingly little known about how glial cells contribute to their formation.
The programmed death receptor-1 (PD-1): programmed death ligand-1 (PD-L1) pathway is central in controlling interactions between host defense and invading pathogens. Accumulating evidence suggests that during neuroinflammation, PD-L1 expression is increased on microglial cells, as well as astrocytes [
19]. These findings suggest that resident glial cells limit CNS pathology through suppression of proinflammatory cytokine production from brain-infiltrating T cells via activation of the PD-1: PD-L1 pathway [
20]. PD-L1 expression on glial cells has also been shown to limit immune-mediated tissue damage in models of multiple sclerosis, as well as during acute viral encephalitis [
19,
21,
22].
We have previously investigated the role of PD-1: PD-L1 signaling in regulating immunopathology through functional inhibition of effector CD8
+ T cells within the post-encephalitic brain following MCMV infection [
19]. In the present study, we investigated the involvement of PD-1: PD-L1 signaling in the retention of CD8
+-gated CD103
+CD69
+ T cells and the development of bT
RM. Using our murine model of MCMV infection, we performed phenotypic analysis of CD8
+ lymphocytes residing within the chronically infected brain to characterize bT
RM. We also compared their development in wild-type (WT) animals to that in PD1- and PD-L1 knockout mice.
Discussion
The most significant finding presented in this study is that the PD-1: PD-L1 pathway contributes to development of bT
RM cells within the MCMV-infected brains. Upon resolution of acute viral infection, the greatly expanded effector CD8
+ T cell population rapidly contracts, leaving behind a small number of cells that survive to form long-lived memory cells [
31,
35]. Some of these memory T lymphocytes persist long term in non-lymphoid tissues as T
RM cells, which defend against re-infection [
3,
14,
36]. We and others have previously shown that effector CD8
+ T cell populations exhibit heterogeneity in expression of KLRG1 during activation and expansion [
2,
3,
7,
9]. Through study of both acute and long-term CNS viral infection using WT, PD-L1 KO, and PD-1 KO animals, we report here that brain-infiltrating CD8
+ T cells display distinct phenotypes of SLEC and MPEC populations from acute to chronic infection. In accordance with other studies where it has been reported that CD127 and KLRG1 are inversely expressed on SLEC and MPEC, our results show that during acute MCMV infection, KLRG1
+ CD127
− (i.e., the SLEC population) cells dominate. In contrast, later time points correlate with development of KLRG1
− CD127
+ cells in WT animals [
9,
37,
38]. Importantly, CD127 expression was significantly reduced in PD-L1 KO and PD-1 KO animals. Taken together, these data demonstrate that the PD-1: PD-L1 pathway within the CNS promotes development of a bT
RM cell population following viral infection.
Studies of HIV-1 infection have reported expansion of CD8
+CD127
− effector-like T cells as a consequence of heightened immune responses [
39]. Experiments using acute LCMV and Listeria infections in mice have demonstrated emergence of CD127-expressing CD8
+ T cells that arise during the effector phase and acquire phenotypical and functional properties of memory T cells [
37,
40]. The down-regulation of CD127 during these chronic viral infections has been attributed to ongoing repetitive TCR stimulation, whereas elevated expression of CD127 on HCV-, HBV-, and RSV-specific memory CD8
+ T cells has been explained by a lack of persisting antigen [
38,
41]. Thus, the frequency of CD127 expression on bT
RM cells in WT animals despite persistence of the latent viral genome may suggest an absence of ongoing TCR triggering within the MCMV-infected brain. In contrast, significantly reduced expression of CD127 on bT
RM cells indicates prolonged, effector-like T cell responses in PD-1 KO and PD-L1 KO animals.
Phenotypic signatures indicative of bT
RM, consisting of CD103, CD69, and CD127 expression, were observed at higher levels among WT animals than among PD-L1 KO and PD-1 KO mice during chronic infection. Similar to findings reported for other non-lymphoid organs, as well as from the brain with vesicular stomatitis virus [
42,
43], we found 87.3 ± 5.6% of the CD8
+ T cells persisting within the MCMV-infected brain express CD69. In contrast to brain infection with LCMV, which showed that CD103 was expressed only on a portion of bT
RM [
44], we observed that the vast majority of CD8
+ T cells co-expressed CD103 and CD69 in WT mice during long-term infection. Expression kinetic studies show early induction of CD69 on brain-infiltrating T cells, as shown by Mutnal et al. [
28], and CD69
+CD103
− cells appear to show effector function early after brain infection. It has been previously reported that expression of CD69 is required for efficient effector T cell retention in the skin and subsequent formation of T
RM cells [
11,
15,
45,
46]. This is because CD69 expression by T
RM cells downregulates cell surface expression of S1P1, thereby blocking T cell movement out of tissues supporting their stationary state [
3,
47,
48]. In this study, PD-1 KO and PD-L1 KO animals show a dominating population of CD69
+CD103
− at 30 dpi, a time point at which these mice have significantly fewer co-expressing CD69
+ CD103
+ cells when compared to WT. Accumulating evidence also indicates a contribution of cytokines like TGFβ, IL-15, IL-7, and IFN-α/β in the induction CD103 and CD69 [
2,
49‐
51].
Development of T
RM cells in a particular tissue clearly involves various factors such as T cell migration, entry into the tissue, retention, and survival. These factors are likely regulated or induced by locally derived signals. Therefore, effector T cell populations during acute infection and the retention of T
RM within non-lymphoid tissue under specific environment conditions, such as the infected brain, are critical to understand. The chemokines CXCL9 and CXCL10 have been shown to facilitate entry of T cells into epithelium during infection of mucosal surfaces with HSV-2 [
32]. Similarly, CXCR3, the receptor for CXCL9 and CXC10, is required for the appropriate localization of effector T cells and for subsequent formation of T
RM [
16]. Expression of CXCR3 on circulating T cells or its chemokine ligands, CXCL9 and CXCL10, in tumor tissues has been reported to be associated with elevated intratumoral T cell infiltration in melanoma and colorectal cancer patients [
52‐
54]. Interestingly, previous studies from our laboratory have shown that microglial cells produce high levels of CXCL9 and CXCL10 in response to MCMV brain infection [
33]. Additionally, reports using a skin model suggest that CCR7 is responsible for exit of T cells out of the tissue, whereas CCR7
− T cells remained in the skin as T
RM cells [
3,
11]. Likewise, results presented here show negligible expression of CCR7 at 7 and 30 dpi in all groups of animals, which are in line with other studies [
3,
17]. Furthermore, differential expression of CXCR3
+ on CD103
+CD8
+ T cells was observed among WT, PD-1 KO, and PD-L1 KO animals. A significantly higher level expression of CXCR3 on CD103
+CD8
+ T cells from the brains of WT mice in comparison to PD-1 KO and PD-L1 KO mice at 30 dpi was observed. Moreover, the PD-1: PD-L1 pathway has been reported to negatively regulate chemokine expression in various contexts. For example, increased expression of the chemokine CXCL9 and its receptor is associated with blocking of PD-L1 in dry eye disease mice [
55]. Here, differential expression profiles of CXCR3 on bT
RM between WT, PD-1 KO, and PD-L1 KO mice were observed. This finding reveals a role for PD-1: PD-L1 in regulating the expression of CXCR3, which in-turn may regulate the retention of bT
RM following MCMV infection.
The PD-1: PD-L1 pathway is well known to limit immune-mediated tissue damage caused by over-reactive T cells, particularly in immune-privileged sites like the brain. Previous reports which show upregulation of PD-L1 in inflamed brain suggest a role for this pathway in regulating T cell activation, as well as controlling immunopathological damage [
19,
56]. Additionally, PD-1 was first regarded as an inhibitory marker and found to be upregulated on exhausted T cells, as defined by reduced ability to proliferate and produce cytokines [
57]. It has previously been proposed that increased expression of inhibitory markers, such as PD-1 and CTLA-4, on brain T
RM cells may serve as a means to prevent this population from unintentional activation and unnecessarily self-attack [
31]. Similarly, our flow cytometry analysis reveals upregulation of PD-1 on CD103
+CD8
+ T cells, along with negligible expression on CD103
−CD8
+ cells in WT animals. Interestingly, inverse expression of PD-1 on the CD103
+ and CD103
− population was observed among the PD-L1 KO animals. It appears that expression of PD-1 by bT
RM cells is not only a mechanism by which the immune system exerts brakes on unnecessary T cell stimulation and proliferation, but it also may itself promote longevity. Furthermore, the decreased expression of PD-1 on T
RM cells in PD-L1 KO animals indicates a dysregulation of bT
RM cells in the absence of the PD-1: PD-L1 pathway.
Evaluation of Bcl-2 expression in memory cells during chronic infection showed significant higher levels of this pro-survival factor in CD103
+ and CD69
+ CD8 T cells among WT when compared to PD-L1 KO animals. However, an evaluation of the proliferation potential of memory cells using Ki67 staining revealed no major difference among the two groups of animals. These data indicate increased survival of memory cells without any change in proliferation with an intact PD-1: PD-L1 pathway. Our finding was similar to Grayson et al. who reported surviving memory cells contain higher levels of Bcl-2 than naïve cells. These elevated levels of Bcl-2 may lead to diminished death phase after secondary infection, resulting in a net increase in memory cells [
58].
Data indicate that T
RM cells residing in a variety of tissues accelerate and improve clearance of pathogens upon re-challenge. However, the driving mechanisms still remain the subject of intense investigation [
31,
59]. It has recently been reported that T
RM cells respond to viral reactivation by the production of inflammatory cytokines, such as IFN-γ, along with immune cells being rapidly recruited from the circulation [
60]. The function of T
RM cells in the brain may largely depend upon rapid IFN-γ production in combination with release of cytotoxic granules and perforin because IFN-γ-deficient bT
RM fails to provide sufficient non-cytolytic antiviral function [
44]. The positioning of bT
RM within the brain parenchyma could be critical to rapidly eliminate infected cells in response to reinfection or reactivation of latent CNS infections.