Cellular interactions in resident memory T cell establishment and function

Abstract

Tissue resident memory T cells (T_RM) are enriched in non-lymphoid tissues and represent a formidable barrier against invading pathogens and tumors. T_RM are armed with deployment ready effector molecules which combined with their frontline location allows them to be early organizing centers of our immune defense. Despite their autonomous nature, T_RM rely on careful collaboration with other immune and non-immune cells located within the barrier organ to exert their superior protective role. Here, we highlight recent studies focusing on cellular interactions that regulate T_RM establishment and function. A deeper understanding of these processes is instrumental in designing new means to target T_RM for desirable outcomes in infectious diseases, cancers and autoimmunity.

Introduction

The realization that a large fraction of memory T cells in our body are stably contained within non-lymphoid tissues (NLTs) and are not routinely circulating, has brought a remarkable shift in our understanding of immunological memory. Resident memory T cells (T_RM) are now considered the numerically dominant memory T cells in most peripheral organs [1,2]. The numerical abundance of T_RM in barrier tissue is coupled with their rapidity of effector functions upon reencounter with antigen. In a study that sampled herpes simplex virus-2 (HSV-2) reactivation every 6hr, it was shown that the presence of local T_RM rapidly limited viral replication within 24hrs of initial viral recrudescence [3]. While T_RM’s protective role was initially described in the setting of viral infections, recent studies have broadened their antipathogenic capabilities to cover numerous bacterial and fungal diseases as well as cancers. Spatially, an individual resident cell patrols a limited territory, but considering all T_RM together, they comprise a highly effective network that can be rapidly mobilized against an incoming pathogen [4,5]. As such, density of T_RM is directly linked to their protective ability. Our understanding of T_RM’s functional capabilities is undergoing tremendous expansion. It was recently demonstrated that CD8 T_RM are a key player in establishing cancer-immune equilibrium through their expression of effector molecule TNF in a mouse model of melanoma [6]. This is suggestive of a highly effective and durable immunosurveillance activity that may span decades.

Immunosurveillance activities of T_RM in non-barrier organs has also received significant attention recently. The liver of individuals infected with Hepatitis B virus has been shown to harbor a unique population of IL-2high CD8 T_RM that could exert partial control of the virus [7]. Similarly, in a mouse model of malaria, liver T_RM were instrumental in patrolling the sinusoids and providing protection against sporozoite replication [8]. Additionally, T_RM in brain have been shown to control fatal viral encephalitis in multiple contexts [9,10]. While these potent effector functions are responsible for accelerated clearance of pathogens, they also indicate T_RM’s potential for harm. T_RM are likely central to the pathogenesis of numerous chronic inflammatory diseases and recent work by Steinbach et al, suggests presence of brain T_RM early in life can indeed potentiate the development of autoimmune lesions in the central nervous system [11].

While the potent anti-pathogenic as well as pathologic roles of T_RM are well established, the cellular interactions that underpin T_RM fate decisions, differentiation and functions are less well understood. Here we review the current state of the field emphasizing the cellular circuitry that modulate these processes.

T_RM differentiation and maintenance

As part of the anti-pathogen effector response, T_RM precursors gain entrance to non-lymphoid tissues (NLTs) to eliminate infection. Although these T_RM precursors are phenotypically similar to KLRG1^lo memory precursor effector cells that give rise to long-lived circulating memory T cells, recent studies have identified important transcriptional differences [12]. Even at this early stage in memory differentiation, many T_RM precursors in NLTs already adopt transcriptional and epigenetic features of mature T_RM. By combining lineage tracing and single cell transcriptomic analysis approaches, recent studies have identified the presence of committed T_RM precursors within the effector CD8 T cell pool with increased level of T_RM- associated genes in the blood or target NLTs [13,14].

Early T_RM programming

While the specification of the early T_RM fate is a product of the integrating a diverse array of T cell-intrinsic and extrinsic signals, most research has focused so far on investigating T cell-specific transcriptional regulators that impact T_RM formation (comprehensively reviewed in [15,16]). Identifying external cues and their cellular sources that may impact early T_RM fate decisions will be crucial to modulate their abundance and functionality. A recent study by Mani et al, showed that migratory dendritic cells can precondition naïve CD8 T cells towards an epidermal T_RM fate through TGF-β [17]. Such homeostatic programming even before T cells are activated, suggests a significant heterogeneity in the T_RM-differentiation potential of the pre-immune T cell repertoire. Studies in mouse models of influenza and vaccinia virus infection suggest a critical role of DNGR-1+ dendritic cells (DCs) in priming T_RM precursors [18]. These cross-priming DCs produce cytokines IL-12 and IL-15 which along with molecule CD24 are required for optimal priming of T_RM precursors. IL-6 produced by fibroblastic reticular cells was shown to enhance T cell survival and promote T_RM differentiation [19]. Similarly, monocytes derived IL-10 was important in lung T_RM differentiation [20]. As more tools become available to temporally manipulate the priming environment of T cells, the identity of additional molecules and cellular interactions that underpin acquisition of T_RM fate will be revealed.

T_RM differentiation and maintenance in non-lymphoid tissues

Once in the target tissue, CD8 T_RM precursors undergo further differentiation under the influence of the local environment. Cells in the NLTs can influence entry of T_RM precursors, their accurate positioning and ultimately their immunosurveillance activities during steady state as well as after secondary infection.

a). CD4 T cells:

CD4 T cell help is important for functional long-lived memory CD8 T cells but is dispensable for effector CD8 T cell generation. In contrast, T_RM differentiation requires CD4 T cell help in several viral infection models. Upon HSV-2 infection effector CD8 T cell migration to the female genital mucosa required CD4 T cell derived IFN-γ [21]. However, this migratory defect was not observed in systemic and intracranial lymphocytic choriomeningitis infection, vaccinia virus skin infection and human papilloma virus vector immunization [9,22-24]. This suggests a highly pathogen specific role of CD4 T cells in creating a T_RM permissive environment. In the context of influenza infection of respiratory mucosa, CD4 T cell deficiency was shown to impair formation of CD103+ CD8 T_RM [25]. Although there was no migrational defect, the CD8 T_RM were localized away from airways and had poor protective ability against heterosubtypic infection. A recent report suggests that T-bet expressing type-1 regulatory T cells improve TGF-β bioavailability and promote CD8 T_RM establishment through expression of integrin β8 by the Treg cells [26]. In the context of persistent murine polyoma virus infection, IL-21 from CD4 follicular helper-like T cells has been shown to be important for brain CD8 T_RM differentiation [27]. A similar mechanism has also been proposed for acute influenza virus infection [28]. Importantly this help was not only needed for CD8 T_RM positioning but also their recall function.

b). Immune cell aggregates:

T_RM in the female reproductive mucosa and skin have been shown to be present in “cluster” with other T cells and professional antigen-presenting cells (APCs) (Fig. 1a) [29,30]. These memory lymphocyte clusters (MLCs) are distinct from tertiary lymphoid organs (TLOs) but thought to play a similar critical role in initiating early responses against reinfections. The chemokine CCL5 derived from the APCs or CD8 T cells is important for maintenance of CD4 T_RM within these microanatomical niches. CCL5 expression in turn is maintained via a positive feedback loop by low level IFN-γ produced by T_RM cells present within the MLCs. Similarly in influenza infection, CD4 T_RMs are maintained within TLOs termed inducible bronchus-associated lymphoid tissues where they promote survival and function of CD8 T_RM as well as resident B cells (Fig. 1b) [28,31].

Figure 1. Cells in diverse anatomic niches supporting T_RM maintenance.

(A) T_RM in skin. CD8 T_RM in the skin epidermis require TGF-β for its appropriate differentiation and long-term maintenance. The source of this TGF-β is the CD8 T_RM itself but for the activation of the latent TGF-β, T_RM rely on integrin αvβ6 and αvβ8 found on surrounding keratinocytes. Both CD4 T_RM and CD8 T_RM also require IL-15 and IL-7 produced by the cells in hair follicle for their maintenance. CD4 T_RM in the skin dermis are found in aggregates with CD11b+ macrophages and CD8 T_RM. Similar memory lymphocyte clusters are found in the female reproductive tract (FRT) as well. But unlike FRT, where these clusters are maintained by CCL-5 produced by macrophages, in skin, the CD8 T_RM are the source of CCL-5 and influence long-term maintenance of the CD4 T_RM in this unique memory niche. (B) T_RM in lung. CD4 helper T_RM colocalize with resident B cells and CD8 T_RM within the inducible bronchus-associated lymphoid tissues (iBALT) found in the lung. The CD4 T_RM rely on autocrine IL-2 for survival as well as IL-7 and IL-15. They in turn provide IL-21 which influence positioning, survival and function of CD8 T_RM. CD4 T_RM also potentiate function of locally resident B cells in the tissue and enhance their survival through co-stimulatory interactions mediated by CD40L. HEV, high endothelial venule.

c). Myeloid cells:

Differentiation and long-term persistence of CD8 T_RM are also regulated by other hematopoietic cells. Inflammatory monocytes recruited to the lamina propria of small intestine during Yersinia pseudotuberculosis infection were shown to be important for CD103neg CD8 T_RM differentiation [32]. Lung alveolar macrophages can negatively regulate CD8 T_RM abundance whereas pulmonary monocytes promote T_RM differentiation after influenza infection [33,34]. Macrophages in the salivary gland were shown to be critical to the routine immunosurveillance of the organ at steady state by CD8 T_RM [35].

d). Non-hematopoietic cells:

Stromal cells also provide critical survival factors required for T_RM maintenance. In the skin epidermis, keratinocytes regulate T_RM differentiation and maintenance [36]. Keratinocytes in the skin and epithelial cells in the small intestinal mucosa express TGF-β activating integrins α_vβ₆ and/or α_vβ₈. Ablation of these integrins interfered with long-term maintenance of T_RM in skin epidermis and small intestine epithelium. Interestingly a recent study showed that the source of TGF- β are the CD8 T_RM themselves not keratinocytes (Fig.1a) [37]. The reliance of epidermal T_RM on TGF- β is continuous, as established T_RM are depleted by blocking this signaling axis [36]. Competition for active TGF-β has been proposed to be a mechanism to selectively enrich antigen-specific T_RM abundance in skin epidermis. Besides TGF-β, the production of cytokines IL-7 and IL-15 by cells in the hair follicle is also important for the localization and maintenance of both CD4 and CD8 T_RM [38,39]. However, these cytokine requirements appear to be organ-specific as a number of tissues can support T_RM in the absence of IL-15 [40]. Identification of tissue-specific survival signals and their cellular sources will be key towards devising new therapeutic means to alter T_RM density in NLTs.

T_RM reactivation

After antigen reencounter, T_RM elaborate an array of effector functions to rapidly eliminate invading pathogens. While the initial response follows the classic T cell receptor activation, many of the downstream anti-pathogenic responses involve collaborative interactions with other cells present locally or in central circulation.

Sensing and alarm function

Upon reactivation both CD8 and CD4 T_RM rapidly release copious amounts of effector cytokines and chemokines including IFN-γ, TNF-α, CCL-3 and CCL-4 [29,41-43]. The speed of this response often rivals that of innate immune cells and is possible because of the transcriptionally poised state of T_RM ready to translate these effector molecules [39,44,45]. Besides inflammatory cytokines, most mucosal T_RM also carry preformed cytotoxic granules including granzyme-B and perforin. While the ex vivo cytolytic activity of T_RM is well established, evidence of in vivo cytotoxicity has only been shown in a few studies [9,46]. It is worth noting that a non-cytolytic purging of infection has the added advantage of maintaining the integrity of the vital organs and allows for a rapid return to homeostasis.

Besides their direct anti-pathogenic roles, T_RM-elicited cytokines and chemokines help to broadcast the alarm signal to cells in the surrounding tissue. T_RM induced IFN-γ upregulates VCAM-1 on adjacent endothelial cells as well as chemokines CXCL-9 and CXCL-10 that allow for fast recruitment of T and B lymphocytes from the periphery (Fig.2)[47]. These recruited memory B cells were the chief source of virus-specific antibodies in the reproductive mucosa in an HSV-2 infection model [48]. Interestingly in a recent simian-human immunodeficiency virus vaccine study in macaques, the presence of T_RM in the genital tissue reduced the threshold of neutralizing antibody required for protection [49]. Whether this protection was mediated through actions of T_RM-recruited antibodies is not known, but this finding underscores the potential synergy between local T_RM and antibody responses.

Figure 2. Immune interactions after T_RM reactivation.

Restimulated T_RM accelerate pathogen clearance through a range of direct and indirect actions. (1) CD8 T_RM mediate the killing of infected cells via elaboration of cytotoxic granules perforin and granzyme-B. (2) Cytokines and chemokines secreted by the T_RM, influence activation of local innate and adaptive immune cells as well as promote recruitment of B cells and T cells from the peripheral circulation. These recruited cells further aid in pathogen elimination through secretion of effector molecules including antibodies. (3) In situ proliferation of T_RM generate new effectors to fight infections. (4) Some of these T_RM progeny give rise to local secondary T_RM after elimination of the infection. But a small group of cells can egress out of the tissue and redifferentiate to become circulating T_CM or T_EM.

T_RM induced TNF-α and IL-2 are critical for maturation of dendritic cells and induction of granzyme-B among natural killer as well as bystander CD8 T cells respectively [47]. Activation of other immune cells in the affected organ combined with recruitment of cells from circulation contributes to the rapid control of the pathogen. In the case of cutaneous leishmaniasis, recruited inflammatory monocytes were a principal mechanism of parasite control through their production of reactive oxygen species and nitric oxide [50]. In a similar but pathologic context, resensitization of Th2 T_RM in allergic airway disease led to activation of eosinophils in the airway and mucus metaplasia [51].

CD8 T_RM display significant promiscuity in their ability to be activated by various types of APCs. Interestingly the quality of the T_RM’s functional response can be modulated by the identity of APCs. In an influenza recall infection, it was recently shown that antigen-presentation by hematopoietic cells reduced the expression of T_RM induced cytokines and chemokines [52]. By contrast non-hematopoietic cell mediated antigen presentation resulted in restraining of type-1 interferon stimulated gene (ISG) program and upregulation of cell cycle genes. This suggests that by focusing antigen-presentation to different compartments, one can tune T_RM functionality.

In situ proliferation and developmental plasticity

In addition to their immediate effector functions, restimulated T_RM undergo in situ proliferation and provide a new wave of activated T cells to rapidly contain pathogen spread [53,54]. After clearance of the pathogen, a number of these cells further differentiate into long-lived secondary T_RM and increase T_RM density in the local tissue. However, recent work suggests some T_RM progeny do not necessarily adopt the T_RM fate and rather exit from the NLT, enter systemic circulation and resemble central or effector memory T cells [55-58]. These so called ex-T_RM cells retain transcriptional and epigenetic marks of their prior T_RM-self and display migrational bias to their tissue of origin. Similar evidence of migrational plasticity has also been described in a small population of blood borne human CD4 T cells bearing canonical epithelial T_RM marker, CD103 [59]. Importantly these cells show significant overlap in their transcriptome and TCR clonotype with bonafide T_RM and maintain the ability to migrate and establish residence in human skin xenografts. Such migrational flexibility are likely to be critical in exerting T_RM associated functions (both good and bad) in distant sites that are not seeded during primary immune responses. However, it is unknown how T_RM progenies decide to become T_RM again or escape the resident fate. Single cell RNA-seq studies have suggested significant heterogeneity among TRM populations. T_RM cells expressing higher levels of CD28 and the transcription factor Id3 exhibit enhanced capacity to produce circulating memory T cells [14,60]. Such heterogeneity is well established during primary T cell response. Whether similar cellular program control the heterogenous fate of T_RM progenies is an ongoing area of study and their identification will greatly improve our understanding of T_RM biology.

Restraining T_RM functions

Positioning of highly weaponized T cells in vital organs also raises issues of their aberrant activation leading to immunopathology. Accordingly, many T_RM express molecules that can attenuate activation including the regulatory cytokine IL-10 and a whole range of co-inhibitory molecules. Most human and mouse T_RM express PD-1, Tim-3, CTLA-4 and CD101 in the absence of persistent antigen [45,61]. Most often, the presence of these markers does not limit the functional responsiveness of T_RM suggesting that the receptors are not actively engaged but nonetheless available for regulation. An example of such regulatory control is seen in influenza infection where PD1 expression by a population of lung T_RM was shown to limit the latter’s inflammatory actions that might contribute to fibrosis [62]. Similarly, ex vivo characterization of human pancreatic tissue revealed that the effector functions of pancreatic CD8 T_RMs are attenuated by resident macrophages through PD-1/PD-L1 pathways [63]. Interestingly during murine polyomavirus infection, interaction between PD-L1 expressing myeloid cells and PD-1+ CD8 T_RM show examples of temporal regulation. The signaling is important to limit neuroinflammation during acute infection but is needed to maintain a certain level of inflammation to control virus during persistent infection [64].

Conclusions

Research over the last decade has positioned T_RM as central in our anti-pathogen defense arsenal. However, T_RM do not operate in isolation and must adapt to the physical, metabolic, and environmental constraints of their location. They make strange bedfellows with neighboring keratinocytes, pneumocytes, neurons, and other resident immune cells, all within the confines of a dense extracellular matrix. They are transcriptionally distinct from circulating memory T cells and based on their tissue location, display considerable heterogeneity in their maintenance requirements and functions. This is reflective of their tissue-specific adaptation that builds upon complex local interactions with numerous immune and non-immune cells. We have just begun to unravel the crosstalk that occurs between T_RM and other cell types using classical immunology approaches. High throughput next-generation technologies could significantly propel these discoveries and build a comprehensive map of cellular interactions underlying every aspect of T_RM biology.

Highlights.

• Resident memory T cell (T_RM) differentiation and function is dependent on a number of extracellular interactions

• Through secretion of cytokines, chemokines and other immune mediators, these cells can influence generation of T_RM precursors, migration of T_RM to target tissue and their long-term residence

• Upon T_RM reactivation after pathogen encounter, T_RM engages the local antipathogenic cells to rapidly curtail pathogen spread