Tissue resident memory T cells and viral immunity

Abstract

Tissue resident memory T cells (T_RM) constitute a recently identified T cell lineage that is responsible for frontline defense against viral infections. In contrast to central and effector memory T cells, which constitutively recirculate between tissues and blood, T_RM reside permanently within tissues. As the main surveyors of non-lymphoid tissues, T_RM are positioned to rapidly respond upon reinfection at barrier sites. During a viral reinfection, T_RM trigger the local tissue environment to activate and recruit immune cells and establish an antiviral state. Consistent with this function, there is empirical evidence that T_RM accelerate control in the event of reinfection or possible reactivation of latent infections in solid organs and barrier tissues. Here we review recent literature highlighting the protective functions of T_RM in multiple viral challenge models and contextualize the implications of these findings for vaccine development.

Introduction

Tissue resident memory T cells (T_RM) constitute a recently identified T cell lineage that is responsible for frontline defense against viral infections. Prior to encountering cognate antigen, naïve T cells restrict trafficking to blood and secondary lymphoid organs (SLO) such as the spleen and lymph nodes (LN). Therefore, to prime a T cell response against a virus, antigens from viral infections occurring in peripheral tissues must reach SLOs, either directly or through migratory antigen presenting cells. Once activated, naïve T cells proliferate extensively to become numerically relevant and then must migrate to sites of infection. If the infection is cleared, an expanded memory cell population remains that can be found not only in blood and SLOs (like naïve cells) but also most tissues throughout the rest of the body. Thus, memory T cells are not only more abundant, but are also anatomically more broadly distributed than their naïve counterparts, enabling quicker and more effective protection against secondary infections that are initiated outside SLOs (e.g. by those pathogens encountered at barrier tissues such as the mucosae and skin).

Pioneering work parsed memory T cells isolated from human blood into two lineages based on the expression of homing molecules [1]. Memory T cells that expressed SLO homing molecules, such as CCR7 and CD62L, were termed central memory T cells (T_CM). It was proposed that this population was specialized to recirculate among the T cell zones of SLOs, and in the event of viral antigen detection, proliferate vigorously and give rise to a second wave of effector T cell differentiation and migration. The second subset lacked SLO homing molecules, but expressed higher levels of integrins and chemokine receptors that were suggestive of nonlymphoid tissue (NLT) migration. These cells were termed effector memory T cells (T_EM) because they also expressed heightened levels of effector molecules, and they were proposed to be principal surveyors of NLTs. A common interpretation was that T_EM constitutively recirculated, meaning they routinely entered NLTs using blood as a conduit, then egressed from NLTs via the lymphatics on their way back to blood (Fig 1) [2–4].

Figure 1. Patterns of memory T cell migration.

Memory T cells are parsed into subsets that have different migration patterns. The prevailing model is described here. Central memory T cell (T_CM) migration is similar to that of naïve T cells and emphasizes immunosurveillance of secondary lymphoid organs (SLOs, such as lymph nodes, intestinal Peyer’s patches, and the white pulp of spleen). Specifically, T_CM migrate from blood into SLOs, exit via lymphatic vessels, then rejoin the blood supply to being this recirculation pattern anew. Effector memory T cells (T_EM) recirculate through blood, nonlymphoid tissue (NLT), then back into the blood via the lymph (while transiently passing through SLOs). Tissue resident memory cells (T_RM) are parked within tissues and do not recirculate. Studies indicating that T_RM may dominate immunosurveillance of NLTs have exposed gaps in our understanding of bona fide T cell recirculation through these compartments.

This model has undergone significant revision. Evidence now suggests the existence of a third major T cell population that resides in tissues without recirculating through blood or lymph. These have been referred to as resident memory T cells (T_RM), and they constitute a transcriptionally and phenotypically unique T cell lineage. T_RM often express CD69 and sometimes express CD103 [5,6]. It should be noted that CD69 is transiently upregulated on T cells of any subset following restimulation [7], and CD103 may be expressed by regulatory T cells [8], and at low levels by recirculating naïve T cells. While imperfect, these markers are frequently used to identify T_RM when stringent migration studies are not feasible (reviewed in [6]). Transcription factors that regulate T_RM include Hobit, Blimp-1 KLF2, Tbet and Eomes, although many other factors may be involved in T_RM ontogeny and differentiation [9–11]. While uninfected specific pathogen free laboratory mice harbor few T_RM, mouse models of infection demonstrate that T_RM populate most, if not all, tissues [12,13]. Analyses of human cadavers confirm an abundance of T_RM phenotype cells in numerous tissues [14,15], and they may comprise the largest memory T cell lineage in free-living organisms with physiological infectious experience. Indeed, as a direct indication of T cell residency in humans, Clark et. al. demonstrated that Alemtuzumab, an anti-CD52 antibody, depleted circulating T cells in patients undergoing treatment for cutaneous T cell lymphoma. However, skin T cells were protected from depletion, supporting the hypothesis that they did not recirculate through blood and thus were resident [16]. Here we discuss the function and regulation of virus-specific T_RM and consider the implications for vaccination against viral pathogens.

T_RM provide rapid protection against viral reinfection

The control of secondary infection events by T_RM has been robustly demonstrated in numerous animal infection models. Here, we review key evidence of T_RM-mediated viral control to illustrate the breadth of relevant infectious contexts and anatomic locations surveyed by T_RM (Fig 2). Later in the review, we will discuss mechanisms of T_RM-mediated protection.

Figure 2. Broad evidence that T_RM accelerate viral control upon rechallenge.

This figure summarizes several mouse studies demonstrating that T_RM can protect a wide array of tissues against many viral challenges.

Many studies utilizing mouse models of cutaneous and intravaginal infection with herpes simplex virus (HSV) have demonstrated protective capabilities of T_RM at these barrier sites. For example, intravaginal vaccination with an attenuated strain of HSV generated CD4+ T_RM in the female reproductive tract (FRT) capable of providing enhanced protection against a subsequent HSV challenge [17]. In addition to vaginal protection, virus-specific T_RM can also mediate cutaneous protection against HSV challenge. Cutaneous infection with HSV resulted in virus-specific CD8+ T_RM in the skin that were predominantly localized to the specific site of challenge and were absent at the opposite flank [18]. Importantly, immunized skin was protected against local HSV challenge compared to contralateral skin [18]. Experimental approaches for inducing site-specific T_RM include the generation of focal skin inflammation, which recruits recently activated CD8+ T cells that subsequently differentiate into T_RM upon migration into the epidermis. HSV-specific skin T_RM generated in this manner enhanced protection from local HSV challenge, as mice with circulating memory T cells alone supported high viral titers and skin lesions whereas in the presence of skin T_RM, there was no detectable infectious virus by day 6 and mice did not exhibit overt disease pathology [19]. Remarkably, in a skin smallpox reinfection model, circulating memory T cells took 26 days to completely eliminate virus compared to only 6 days when T_RM were present [20].

One postulated function of T_RM is to protect against local reactivation of latent viruses [21]. Studies using the herpesvirus murine cytomegalovirus (MCMV), which establishes latency in the salivary gland (SG), resulted in salivary gland T_RM that contribute to local viral control [22]. HSV establishes latency in the sensory ganglia and interestingly, human and mouse studies have found HSV-specific CD8+ T cells that reside in sensory ganglia and localize to latently infected neurons [23–25]. Moreover, these CD8+ T_RM can respond to and restrict HSV reactivation from latency in mice [25–27]. Consistent with this, human studies demonstrated that HSV-specific CD8+ T cells persisted in the genital skin and mucosa of patients who suffered from HSV lesions eight weeks prior. Notably, these CD8+ T cells were found to be positioned at the dermal-epidermal junction, likely poised to intercept reactivated HSV exiting axon terminals [28,29].

Many infections, including vesicular stomatitis virus (VSV), polyomavirus, and lymphocytic choriomeningitis virus (LCMV), result in T_RM positioned within the CNS; a compartment where successful pathogen control can come at the price of immunopathology [30–32]. For instance LCMV infection can result in T cell-induced cerebral meningitis in humans [33]. Likewise, circulating LCMV-specific memory T cells may trigger fatal immunopathology upon intracranial reinfection in mice [32,34]. Interestingly, however, LCMV-specific brain T_RM which were recently shown to be capable of protecting against reinfection without inducing immunopathology [32].

The lung presents an interesting case as the ontogeny and maintenance of lung T_RM remain not well defined. Nevertheless effective heterosubtypic protection against flu appears to be dependent upon pulmonary T_RM [35–37], and these results likely extend to other lung infections, including the mouse parainfluenza virus, Sendai [38,39]. Mechanistic studies have identified a role for flu-specific CD4+ T cells in the regulation of the induction of CD8+ T_RM within the pulmonary mucosae [40] and for direct control of reinfection [41]. Human lungs are abundantly populated with T cells bearing a T_RM phenotype [42] and particularly elegant work demonstrated that RSV infection of humans generated more frequent RSV-specific CD8+ T cells in the lung airways than in the blood. Importantly, higher airway CD8+ T_RM correlated with reduced disease severity [43].

Mechanisms of T_RM protection

The eponymous function of cytotoxic CD8+ T lymphocytes is to kill infected target cells. Certain populations of T_RM, including those within the intestinal mucosa, constitutively express highly elevated levels of granzyme B [44], consistent with the hypothesis that T_RM may be poised for rapid lytic activity at frontline sites of pathogen exposure. While killing is difficult to assess in nonvascular compartments of nonlymphiod tissues in vivo, T_RM cytotoxicity has been demonstrated directly ex vivo in chromium release assays [44] (Fig 3).

Figure 3. CD8+ T_RM functions.

Some CD8+ T_RM express high basal levels of granzyme B suggesting reactivated T_RM may be constitutively poised to kill infected cells, although this has not been formally tested in vivo. Upon recognition of cognate antigen, T_RM also secrete pro-inflammatory cytokines, namely IFNγ, IL-2 and TNFα, which act to upregulate chemokines such as CXCL9 and CXCL10, as well the adhesion molecule VCAM-1 on endothelial cells. This facilitates recruitment of recirculating subsets of memory CD8+ T cells and B cells into the tissue. Interferon stimulated genes (ISGs) are also upregulated in cells surrounding reactivated T_RM which may increase their resistance to viral infection. Furthermore, recognition of viral antigen by T_RM also precipitates upregulation of granzyme B by natural killer cells (NK) and bystander memory CD8+ T cells and maturation of local dendritic cells (DCs). In summary, T_RM trigger a cascade of immunostimulatory and antiviral responses at sites of viral re-exposure.

One variable that likely regulates the efficiency of T_RM mediated immunosurveillance, target acquisition, and killing is cellular motility. Interstitial migration is likely to be tissue dependent, for instance, being significantly slower in skin epidermis compared to dermis [45,46]. Moreover, context matters. Indeed, during acute vaccinia virus or HSV infection within skin, evidence was provided that cytotoxic T cells migrate preferentially towards infected cells [47,48]. If generalizable, this important concept dramatically alters models of the efficiency of T cell immunosurveillance in tissues.

T_RM function, however, likely extends far beyond the mere capacity to kill infected host cells. Upon reactivation, T_RM orchestrate numerous changes within hours of cognate antigen recognition in the tissue, upregulating pro-inflammatory cytokines which work to activate and recruit immune cells, and establish an antiviral state [49–51] (Fig 3). LCMV-specific T_RM in the female reproductive tract (FRT) produced the cytokines IFNγ, IL-2 and TNFα upon recognition of cognate peptide [50]. This resulted in upregulation of adhesion molecules and chemokines, which promoted the recruitment of circulating memory T cells and B cells into the tissue; a process which was dependent on T_RM-derived IFNγ. Recruited memory T cells also exhibited signs of bystander activation as there was increased granzyme B expression [50]. Similar studies in the skin demonstrated that activated HSV-specific skin T_RM produced IFNγ resulting in interferon stimulated gene (ISG) expression in surrounding cells [49]. This is consistent with other infection models demonstrating IFNγ-dependent T_RM protection [32,38].

In addition to adaptive immune cells, T_RM activation may also activate leukocytes of the innate immune system. Granzyme B upregulation in NK cells was seen in the FRT after LCMV-specific T_RM reactivation and this was through an IL-2rβ-dependent mechanism. Moreover, dendritic cell (DC) maturation occurred largely in a TNFα–dependent manner, marked by expression of CCR7 and upregulation of CD80 and CD86, molecules important for co-stimulation during T cell activation [50].

Notably, reactivation of T_RM only requires cognate peptide, and the immune response that is generated is capable of transiently protecting against antigenically unrelated pathogens, highlighting the potent capabilities of T_RM to establish an antiviral state in tissues [49,50]. In summary, T_RM protective functions likely rely on multiple mechanisms, including cytolyitc activity and cytokine mediated alarm functions.

Regulation of T_RM establishment

Most activated T cells downregulate lymph node homing molecules and are transiently endowed with the capacity to enter nonlymphoid tissues. At this early point during the immune response, T cells entering tissues initially appear phenotypically similar to their counterparts in blood and SLOs. After migration into nonlymphoid tissues, however, KLRG1-T_RM precursors begin to take on characteristic markers of residence, supporting the hypothesis that the tissue environment promotes T_RM differentiation [10]. A well-studied example of this is the production of TGFβ in certain tissues, which promotes CD103 expression. CD103 (αE) pairs with beta7, which together bind to E-cadherin and promote retention of T_RM within certain epithelial compartments, although this requirement remains incompletely characterized for many organs [10,52–54]. Other factors also play a role in T_RM establishment. For example, evidence indicates that down-regulation of KLF2 and S1P1 coupled with the transcription factors Hobit, Blimp-1, and surface expression of CD69 also enforces residence [9–11]. Additionally, IL-15 expression can regulate T_RM survival in certain tissues, but not all, highlighting that the regulation of T_RM may differ depending on the anatomic compartment [10,55].

Antigen-independent inflammation can also promote T_RM establishment within tissues. For example, HSV gB-specific T cells primed by flu-gB (which does not infect the skin) home to the skin and become T_RM when a non-antigen specific inflammatory agent (DNFB) was topically applied [19]. Consistent with this, vaginal application of chemokine alone during the effector phase of a primary immune response can promote T_RM seeding in the FRT [56]. Inflammation, however, is not absolutely necessary for T_RM establishment, as naïve CD8+ T cells transferred into a lymphopoenic host resulted in seeding of CD8+ T cells into many NLTs which subsequently adopted phenotypic markers of residence [52].

Antigen is not required for T_RM establishment in many contexts. However, some studies reveal a striking reliance on local antigen for optimal T_RM development. This was demonstrated in a study that utilized two recombinant vaccinia viruses expressing distinct antigens. After transferring naïve cells specific for either antigen, researchers infected the skin of both ears with either virus and demonstrated that T_RM preferentially developed within skin where cognate antigen was expressed [57]. Defining the local and systemic parameters that regulate T_RM development and distribution remain a priority for the field and it is apparent that the rules will not be the same for all tissues and infectious contexts.

Implications for vaccines

It is clear that T_RM are capable of providing robust and rapid immune protection at critical barrier sites frequently exposed to pathogens. A critical question is whether the establishment of T_RM through vaccination can increase protective efficacy against certain pathogens. Addressing this issue poses logistical hurdles, extending even to mere immunogenicity measurements, as T_RM cannot be evaluated in blood (ergo tissues must be biopsied). Moreover, T_RM must be extracted from tissues before assessment with conventional techniques for assessing antigen specificity (e.g. ELISPOT and flow cytometric assays), and the isolation efficiency of this T cell subset may be very poor [12]. However, these barriers are not impossible to overcome. A recent study explored the mechanism of action of a newly developed malaria vaccine, PfSPZ, which has exhibited very promising efficacy in human clinical trials. Perplexingly, patients that were protected exhibited low levels of T cells (and vaccine-specific antibody) within blood. Closer examination (aided by nonhuman primate studies) revealed that the frequency of T_RM within the liver (the relevant site for T cell-mediated protection) was 100-fold higher [58]. This study may provide valuable lessons for vaccine development as well as a cautionary tale against limiting immunogenicity studies to blood.

In summary, T_RM are positioned at frontline tissues and are poised for rapid detection of reinfection events and mobilization of antiviral mechanisms of host protection. T_RM defense strategies likely extend beyond canonical cytolytic function and include cytokine-mediated activation of other components of innate and adaptive immunity. Importantly T_RM remain hidden from view when analyses are limited to peripheral blood, which creates logistical challenges. However, protection studies, both in animal models and limited human studies, suggest that advancement of T_RM inducing vaccine modalities holds promise for the future.

Highlights.

• Tissue resident memory T cells (T_RM) are the main surveyors of non-lymphoid tissues.

• T_RM activate innate and adaptive immunity and establish an antiviral state within tissues.

• T_RM provide protection against numerous viral pathogens in multiple tissues.

• Vaccines aimed at promoting T_RM hold promise for the future.