Decoding Tissue-Residency: Programming and Potential of Frontline Memory T Cells

Abstract

Memory T-cell responses are partitioned between the blood, secondary lymphoid organs, and nonlymphoid tissues. Tissue-resident memory T (Trm) cells are a population of immune cells that remain permanently in tissues without recirculating in blood. These nonrecirculating cells serve as a principal node in the anamnestic response to invading pathogens and developing malignancies. Here, we contemplate how T-cell tissue residency is defined and shapes protective immunity in the steady state and in the context of disease. We review the properties and heterogeneity of Trm cells, highlight the critical roles these cells play in maintaining tissue homeostasis and eliciting immune pathology, and explore how they might be exploited to treat disease.

The cardinal feature of an effective memory T-cell response is the ability to rapidly locate and contain the spread of disease. Early analyses of peripheral blood and lymph uncovered two major subsets of memory T cells: central memory T (Tcm) cells that typically traffic between secondary lymphoid organs (SLOs) and the blood, and effector memory T (Tem) cells that were found to circulate continuously through peripheral nonlymphoid tissues (NLTs) (Sallusto et al. 1999, 2004). However, these circulating memory T (Tcirc) cells can be restricted in their ability survey and quickly access barrier sites where the majority of pathogens gain entry to the body (Masopust et al. 2010; Wakim et al. 2010; Mackay et al. 2012a). We now know that a large proportion of memory T cells are actually anchored in peripheral NLTs, comprising a dedicated layer of tissue memory that is maintained independently of the Tcirc cell pool. These tissue-resident memory T (Trm) cells form a first line of defense against invading pathogens and developing malignancies and are critical to control tissue homeostasis. In this review, we will examine how T-cell residency is defined and shapes tissue memory in the context of infection, cancer, and autoimmunity. We will discuss how memory T-cell persistence is regulated within tissues, our evolving understanding of heterogeneity within the Trm cell pool, and the consequences of tissue residency for spatiotemporal immune protection.

DEFINING TISSUE RESIDENCY

In the context of T-cell memory, “tissue residency” refers to the permanent maintenance of cells in a given organ without recirculation through blood. Conclusive evidence for the existence of CD8⁺ Trm cells stemmed from experiments in mice where donor-derived CD8⁺ memory T cells were shown to be retained in grafted skin or gut tissue following transplantation (Gebhardt et al. 2009; Masopust et al. 2010). Several additional methods to demonstrate T-cell tissue residency have since been used, each with their own strengths and weaknesses (for review, see Masopust and Soerens 2019). Parabiosis—in which the circulatory systems of two animals are surgically fused—has been instrumental in revealing that the vast majority of CD4⁺ and CD8⁺ T cells situated in peripheral tissues are nonrecirculating, as they do not equilibrate between conjoined partners (Steinert et al. 2015; Beura et al. 2019). In addition, CD8⁺ Trm cells have been shown to persist in various tissues following antibody-mediated depletion of Tcirc cells (Gebhardt et al. 2009; Schenkel et al. 2013; Mackay et al. 2015b; Park et al. 2016) and are retained in peripheral tissues following treatment with the sphingosine-1-phosphate (S1P) receptor antagonist FTY720, which prevents T-cell egress from SLOs (Masopust et al. 2010). Using these methods, bona fide CD8⁺ and CD4⁺ Trm cells have now been identified in almost every mouse organ including the brain (Wakim et al. 2010), skin (Gebhardt et al. 2009; Jiang et al. 2012), intestines (Masopust et al. 2010; Beura et al. 2019), salivary glands (Hofmann and Pircher 2011), lung (Teijaro et al. 2011; Takamura et al. 2016), liver, kidney, pancreas (Steinert et al. 2015), female reproductive tract (FRT) (Mackay et al. 2012a; Schenkel et al. 2013; Iijima and Iwasaki 2014), adipose tissue (Han et al. 2017), thymus (Hofmann et al. 2013), spleen, and lymph nodes (LNs) (Schenkel et al. 2014b). Unlike blood-tropic T cells, CD8⁺ Trm cells in most tissues do not stain with intravascularly administered antibodies, indicating that they are situated within the tissue parenchyma (Anderson et al. 2012). Although intravascular labeling is useful to illustrate the spatial location of T cells within a single snapshot of time (Anderson et al. 2014), the use of this technique alone is insufficient to denote permanent residency.

The phenomenon of T-cell tissue residency has proven more difficult to examine in humans. Early work demonstrated that non-recirculating CD8⁺ and CD4⁺ T cells were retained in human skin following antibody-mediated depletion of blood Tcirc cells, where they improved resistance to infections (Clark et al. 2012; Watanabe et al. 2015). More recently, analyses of donated skin (Lian et al. 2014), gut (Zuber et al. 2016; Bartolomé-Casado et al. 2019), lung (Snyder et al. 2019), and kidneys (de Leur et al. 2019) after transplantation have revealed that donor-derived CD8⁺ and CD4⁺ Trm cells are maintained long term in grafted tissues without being detectable in blood. Collectively, these findings strongly support that non-recirculating Trm cells akin to those found in mice are also abundant in humans.

CD8⁺ Trm cells were first formally identified in the skin epithelium and intestinal intraepithelial layer of mice, where they were shown to differ from Tcirc cells by elevated expression of the surface markers CD69 and CD103 (αE integrin) (Gebhardt et al. 2009; Masopust et al. 2010). However, several studies have revealed that bona fide CD69⁺ Trm cells in nonepithelial tissue compartments such as the gut lamina propria (Bergsbaken and Bevan 2015) and visceral tissues including the liver (Fernandez-Ruiz et al. 2016), kidney, FRT (Steinert et al. 2015), and adipose tissue (Han et al. 2017) frequently lack expression of CD103. Populations of CD69⁺CD103⁺ and CD69⁺CD103^– T cells are also ample across human NLTs and SLOs (Sathaliyawala et al. 2013) where they accumulate over the life span (Thome et al. 2014). Importantly, donor-derived CD8⁺ T cells retained in transplanted human tissues similarly express CD69, and most also express CD103 (Bartolomé-Casado et al. 2019; Snyder et al. 2019). Given that CD69 is stably expressed by the majority of Trm cells in both NLTs and SLOs (Schenkel et al. 2014a; Steinert et al. 2015), CD69 positivity is now frequently used to infer tissue residency in both mice and humans irrespective of CD103 expression. However, CD69 can be rapidly and transiently induced on recently activated T cells and Tcirc cells following antigen recognition or exposure to certain inflammatory cytokines (Beura et al. 2015; Cibrián and Sánchez-Madrid 2017; Duhen et al. 2018). As such, the veracity of CD69 as marker to identify Trm cells may be limited in some tissues or circumstances, especially in those where sustained local antigen presentation or ongoing inflammation occur (Fig. 1). For example, whereas CD8⁺CD69⁺ T cells in SLOs are resident and relatively rare in specific pathogen-free (SPF) mice (Schenkel et al. 2014b), their frequency increases following colonization with “wild” microbiota and they partially equilibrate during parabiosis (Beura et al. 2018b). CD69⁺CD103^– CD8⁺ T cells have also been shown to recirculate in the skin (Mackay et al. 2015b), thymus (Park et al. 2016), kidney, and liver (Steinert et al. 2015) of SPF mice in some settings. These studies stress that expression of CD69 alone can be insufficient to demarcate T-cell residency with certitude. Profiling additional molecules that are typically up-regulated by CD8⁺ Trm cells including the chemokine receptor CXCR6 (Fernandez-Ruiz et al. 2016; Wong et al. 2016) or collagen-binding integrin CD49a (Gebhardt et al. 2009; Mackay et al. 2013) may increase confidence in attempts to distinguish Trm cells from their circulating counterparts. However, a considerable fraction of Trm cells in the pancreas, FRT, and salivary glands may evade detection altogether via phenotyping, as they express neither CD69 nor CD103 (Steinert et al. 2015).

Figure 1.

Regulation of T-cell residency formation and persistence in peripheral tissues. Trm cell formation is controlled at multiple stages of T-cell homeostasis, activation, and differentiation. Naive CD8⁺ T cells can be preconditioned for T-cell residency potential by interacting with dendritic cells (DCs) that activate transforming growth factor β (TGF-β) in lymph nodes (LNs) prior to T-cell priming. Activated T cells migrate to the periphery where Trm differentiation occurs in situ following exposure to tissue-derived cytokines that can include TGF-β and interleukin 15 (IL-15). Full commitment to the T-cell residency fate is often associated with the up-regulation of molecules including CD69 and CD103 and down-regulation of factors that promote T-cell egress from tissues such as Klf2 and S1pr1. In addition, Trm cells acquire a unique transcriptional profile that may include up-regulation of residency-promoting factors such as Znf683, Runx3, and Bhlhe40. In the context of chronic infections or cancer where persistent antigen may be present, residency-associated molecules might also be dynamically regulated in recently activated or effector T cells that could eventually return to the circulation.

CD8⁺ Trm cells across diverse tissues share a unifying transcriptional signature that distinguishes them from Tcirc cells following the resolution of infection and that is also enriched in resident innate immune cells (Wakim et al. 2012; Mackay et al. 2013, 2016). This transcriptional profile includes induction of genes that enhance T-cell retention including Cd69, Itgae (encoding CD103), and Cdh1 (encoding E-cadherin) and down-regulation of genes that promote tissue egress (e.g., Klf2, S1pr1) or acquisition of alternate memory T-cell fates (e.g., Tcf7, Eomes) (Mackay et al. 2013, 2015b, 2016; Skon et al. 2013). Comparing gene expression by putative tissue-resident cells to this canonical Trm cell genetic profile may reveal transcriptional similarities that can be used to imply tissue residency. However, many of the genes comprising this Trm-associated gene signature including Cd69, Itgae, S1pr1, and Klf2 can also be dynamically altered in Tcirc cells following antigen stimulation, exposure to cytokines, or in response to stress (Cyster and Schwab 2012; Duhen et al. 2018). Use of transcriptional profiling to identify Trm cells might therefore be confounded in the context of chronic infection or cancer, where gene expression by recently activated Tcirc cells or dysfunctional “exhausted” T cells could overlap with bona fide Trm cells. Moreover, although gene expression by human Trm cells generally mirrors that seen in mouse Trm cells (Kumar et al. 2017; Snyder et al. 2019), there may be differences between species. For example, whereas the critical residency-promoting gene Znf683 (encoding Hobit, Homologue of Blimp-1 in T cells) is elevated in Trm cells across multiple mouse tissues compared to Tcirc cells (Mackay et al. 2016), expression of Hobit by human Trm and Tcirc cells appears to be more variable (Kumar et al. 2017; Pallett et al. 2017; Stelma et al. 2017; Zhang et al. 2018; Snyder et al. 2019).

In sum, we still lack definitive markers that dependably and comprehensively denote tissue-resident cells in many settings. Coexpression of CD69 and CD103 can reliably identify bona fide CD8⁺ Trm cells under homeostatic conditions (Gebhardt et al. 2009; Masopust et al. 2010; Wakim et al. 2010), but these markers fail to capture all resident T-cell populations across tissues. Phenotypic hallmarks of Trm cells may be particularly intangible during chronic disease when ongoing exposure to antigen or inflammation is likely. Delineating combinations of markers that can reliably distinguish Trm cells from Tcirc cells across various organs and disease settings is a long-standing goal of the field and would greatly advance the study of these cells in humans, which typically rely on phenotypic and transcriptional analyses for Trm cell identification.

IMPRINTING AND REGULATION OF TISSUE-RESIDENT MEMORY T CELLS

The generation of Trm cells in peripheral tissues is a multistep process involving naive T-cell activation in SLOs, migration of effector cells to the site of challenge, in situ differentiation, and long-term retention (Fig. 1). Although complete commitment to the Trm cell fate occurs after effector T cells enter peripheral tissues (Mackay et al. 2013), initial priming or preconditioning events might also impact the potential for Trm cell development. Recent work suggests that migratory dendritic cells (DCs) facilitate transforming growth factor β (TGF-β) imprinting of naive CD8⁺ T cells in LNs, which predisposes these cells toward effective Trm cell differentiation following activation (Mani et al. 2019). Early signals integrated during T-cell activation, including inflammatory cues, direct differentiation into KLRG1^loCD127^hi memory precursor (MP) and KLRG1^hiCD127^lo terminal effector (TE) cells (Chang et al. 2014). Studies in mice indicate that Trm cells primarily develop from KLRG1^lo MP-like effector cells (Mackay et al. 2013; Sheridan et al. 2014) or from MP cells that temporarily express KLRG1 but later down-regulate its expression (Herndler-Brandstetter et al. 2018). T-cell receptor (TCR) lineage tracking of recipient-derived T cells following organ transplantation suggests this may also be the case in humans (Bartolomé-Casado et al. 2019). Furthermore, CD8⁺ T cells activated by lower affinity antigens might be more likely to give rise to lung Trm cells than high-affinity clones in mice (Fiege et al. 2019), although the reverse may be true during persistent viral infection of the brain and kidney (Frost et al. 2015). Recent reports indicate that CD8⁺ Trm cells might depend more stringently on cross-presenting Batf3-dependent DCs for their activation than Tcirc cells (Iborra et al. 2016). However, Trm and Tcm cells can both be generated from naive T-cell progenitors bearing the same TCR (Gaide et al. 2015) and the requirement for Batf3⁺ DC priming in Trm cell formation is not absolute (Iborra et al. 2016).

T-cell migration to peripheral tissues is guided by homing and chemokine receptors that can be imprinted during activation and confer tissue-specific tropism (Agace 2006). For example, expression of the a4β7 integrin and chemokine receptor CCR9 can direct T-cell migration to the gut (Mora et al. 2003), whereas E- and P-selectin ligands, CCR4, CCR8, and CCR10, can promote migration toward the skin (Schaerli et al. 2004; Malik et al. 2017; Zaid et al. 2017). However, inflammatory cytokines induced in tissues following infection or injury might dampen the impact of tissue-directed imprinting and promote more promiscuous T-cell migration patterns. The inflammatory cytokines CXCL9 and CXCL10 are rapidly induced in diverse tissues during infection, drawing CXCR3-expressing T cells toward afflicted sites, promoting their epithelial positioning, and subsequent Trm cell differentiation (Shin and Iwasaki 2012; Mackay et al. 2013; Laidlaw et al. 2014; Bergsbaken and Bevan 2015; Fernandez-Ruiz et al. 2016; Gilchuk et al. 2016). E- and P-selectins can also be widely expressed on the vascular endothelium following infection (Hobbs and Nolz 2017) and CXCR6 can enhance Trm cell formation in diverse anatomical sites including the skin and liver (Tse et al. 2014; Zaid et al. 2017) and may promote Trm cell positioning in the lung airways (Takamura et al. 2019; Wein et al. 2019). Furthermore, strong systemic infections (Masopust et al. 2004; Casey et al. 2012), repeated reactivation of Tcirc cells via boosting (Jiang et al. 2012; Davies et al. 2017; Park et al. 2018b), and lymphopenia-driven homeostatic proliferation (Casey et al. 2012) can promote widespread dissemination of Trm cells throughout the body. Migratory receptors are often quickly down-regulated following T-cell activation, restricting peripheral migration, and thus Trm cell formation to a relatively short window of time after antigen encounter (Masopust et al. 2004, 2010; Wakim et al. 2010; Mackay et al. 2012a; Zhang and Bevan 2013). However, during chronic or persistent infection, the expression of homing receptors on Tcirc cells might be sustained for prolonged periods, promoting ongoing peripheral recruitment (Zhang and Bevan 2013; Beura et al. 2015) and potentially continuous Trm cell generation (Smith et al. 2015).

Following tissue entry, Trm cell differentiation can occur in an antigen-independent fashion (Gebhardt et al. 2009; Casey et al. 2012; Mackay et al. 2012a; Holz et al. 2018). Exceptions to this include in the nervous system (Wakim et al. 2010; Casey et al. 2012; Mackay et al. 2012a) and lung (Lee et al. 2011; Takamura et al. 2016), where in situ antigen recognition appears critical for Trm cell formation. Although noncompulsory, the presence of local antigen within the skin, liver, or intestines can boost the number of Trm cells formed after infection or vaccination (Gebhardt et al. 2009; Fernandez-Ruiz et al. 2016; Khan et al. 2016; Muschaweckh et al. 2016; Davies et al. 2017; Holz et al. 2018). Recent experiments in which major histocompatibility complex I (MHCI) expression was ablated following establishment of the Trm cell pool have revealed that ongoing MHCI:TCR interactions are dispensable for the long-term maintenance of CD8⁺ Trm cells in the skin (Lauron et al. 2019) and lung (Wang et al. 2019). In fact, prolonged antigen presentation may actually oppose Trm cell differentiation under certain circumstances by inhibiting CD103 up-regulation (Zhang and Bevan 2013; Tomov et al. 2017; Wang et al. 2019) or Znf683 expression (Kragten et al. 2018; Behr et al. 2019), which may have consequences for Trm cell generation in chronic infection or cancer.

Full commitment to the Trm cell fate requires induction or suppression of key molecules and transcription factors (TFs) that regulate T-cell retention and survival. CD69 can be induced upon antigen recognition or exposure to tissue-derived cytokines including type I interferon (IFN), TGF-β, and interleukin 33 (IL-33) (Casey et al. 2012; Skon et al. 2013; Mackay et al. 2015b) and can promote early local retention of Trm cell precursors (Mackay et al. 2015a; Takamura et al. 2016) through competitive inhibition of S1PR1, which facilitates emigration (Mackay et al. 2015a). Whereas CD69 expression is critical for Trm cell formation in the skin (Mackay et al. 2015a), lung (Lee et al. 2011), thymus (Park et al. 2016), and kidney (Walsh et al. 2019), it may not be essential in the intestines, salivary gland, or liver (Walsh et al. 2019). Nevertheless, suppression of tissue-egress mediators such as KLF2, S1PR1, and CCR7 represents an initial key step in Trm cell differentiation (Mackay et al. 2013; Skon et al. 2013), promoting accumulation of effector T cells in tissues where they can be further imprinted for long-term residency by local signals. TGF-β drives up-regulation of CD103 on Trm cells (El-Asady et al. 2005; Mackay et al. 2013; Zhang and Bevan 2013; Wakim et al. 2015), which can bind E-cadherin on surrounding cells (Cepek et al. 1994; Cresswell et al. 2001) and promote either accumulation (Sheridan et al. 2014) or long-term retention (Wakim et al. 2010; Lee et al. 2011; Casey et al. 2012; Mackay et al. 2013). Trm cells themselves also express E-cadherin, which has been shown to support their maintenance in the salivary glands (Hofmann and Pircher 2011). In addition, very late antigen-1 (VLA-1; CD49a) and LFA-1 have been shown to enhance Trm cell retention in the lungs (Ray et al. 2004), skin (Bromley et al. 2020), and liver (McNamara et al. 2017), respectively.

Multiple TFs cooperate to orchestrate Trm cell formation and function. Down-regulation of the T-box TFs T-bet and Eomes is necessary for Trm cell differentiation in the skin and lungs of mice (Laidlaw et al. 2014; Mackay et al. 2015b), but low levels of T-bet are maintained to ensure sustained IL-15Rα and CXCR3 expression (Mackay et al. 2015b). In addition, T-bet induces the TF Hobit in differentiating Trm cells, which collaborates with Blimp-1 to inhibit expression of tissue-egress genes including Klf2, S1pr1, and Ccr7 as well as the Tcm-associated TF Tcf7 (encoding T-cell factor 1 [TCF-1]) (Mackay et al. 2016). Loss of both Hobit and Blimp-1 leads to a reduction of both Trm cells and resident innate cells in multiple organs including the liver, skin, lung, and gut (Mackay et al. 2016; Behr et al. 2019). Similarly, the TF Runx3 is up-regulated during Trm cell development and promotes Trm cell survival and cytotoxicity in both tissues and tumors (Milner et al. 2017). Similarly, the TF Bhlhe40 has been shown to support Trm cell persistence and functionality in the lungs and melanoma (Li et al. 2019). Additional TFs that might promote Trm cell survival and metabolic homeostasis include the aryl hydrocarbon receptor (Ahr) (Zaid et al. 2014), Nr4a1 (Boddupalli et al. 2016b), and Notch (Hombrink et al. 2016).

Differentiation and maintenance of Trm cells is regulated by tissue-derived cytokines and requires metabolic adaptations to life in barrier tissues. Tissue-derived TGF-β is a key signal for Trm cell commitment. TGF-β can induce CD103 expression (Pauls et al. 2001) and KLRG1 down-regulation (Schwartzkopff et al. 2015) in vitro, and deficiencies in TGF-β signaling have been shown to result in defects in Trm cell generation in the skin (Mackay et al. 2013), intestines (Casey et al. 2012; Zhang and Bevan 2013), kidney (Ma et al. 2017), lungs (Lee et al. 2011), and salivary glands (Thom et al. 2015), which correlate with reduced CD103 expression (Casey et al. 2012; Mackay et al. 2013; Zhang and Bevan 2013; Bergsbaken and Bevan 2015). However, the role of TGF-β in Trm cell development extends beyond CD103 up-regulation, as TGF-βRII-deficient cells can display reduced Trm cell formation even when CD103 is dispensable (Sheridan et al. 2014). In line with this, TGF-β also regulates expression of CD69, CD49a (Zhang and Bevan 2013), Klf2, and S1pr1 (Skon et al. 2013) and drives the down-regulation of Eomes and T-bet following tissue entry (Mackay et al. 2015b). TGF-β is produced in a latent form and must be activated by αvβ6-expressing epithelial cells to license Trm cell differentiation (Mohammed et al. 2016). Blocking its activation after establishment of the memory T-cell pool has demonstrated that TGF-β is also required for long-term maintenance of gut and skin Trm cells (Mohammed et al. 2016). In addition, the cytokine IL-15 is required for Trm cell development and survival in several tissues (Mackay et al. 2013; Schenkel et al. 2016), where it can promote expression of residency-associated TFs including T-bet (Mackay et al. 2015b), Hobit (Mackay et al. 2016), and Runx3 (Milner et al. 2017), and modulate expression of Klf2 (Skon et al. 2013) and CD69 (Pallett et al. 2017). Other cytokines including IL-12, tumor necrosis factor (TNF), and IL-7 can also influence Trm cell development and function in some settings (Adachi et al. 2015; Bergsbaken et al. 2017). Trm cell formation and maintenance may also require metabolic adjustments; differentiating Trm cells gradually up-regulate certain isoforms of fatty acid–binding proteins determined by their tissue location to facilitate exogenous lipid uptake, whereas Tcirc cells do not (Pan et al. 2017; Frizzell et al. 2020). In addition, Trm cells in many tissues express the purinergic receptor P2RX7, which may serve as a means to sense microenvironmental stressors and disruptions to tissue homeostasis (Borges da Silva et al. 2018, 2020; Stark et al. 2018).

HETEROGENEITY OF TISSUE-RESIDENT MEMORY T CELLS

Although Trm cells across tissues share many unifying features, they can also exhibit key developmental and phenotypic differences. Whereas CD8⁺ Trm cells in the skin (Mackay et al. 2013, 2015b), liver (Herndler-Brandstetter et al. 2018; Holz et al. 2018), salivary gland, and kidney (Schenkel et al. 2016) require IL-15 signaling for differentiation and survival, CD8⁺ Trm cells in the gut, pancreas, FRT (Schenkel et al. 2016), and SLOs (Schenkel et al. 2014b) are IL-15 independent. These requirements may also be malleable within a single tissue in the context of chronic infection; salivary gland CD8⁺ Trm cells require IL-15 for formation following acute lymphocytic choriomeningitis virus (LCMV) but not during chronic murine cytomegalovirus (MCMV) infection (Smith et al. 2015; Schenkel et al. 2016). Moreover, whereas Trm cells across NLTs and SLOs share a degree of transcriptional overlap, they can also display tissue-specific gene-expression patterns (Mackay et al. 2013, 2016; Beura et al. 2018b). For instance, LN Trm cells share some transcriptional similarities with NLT Trm cells but others with Tcm cells (Beura et al. 2018b). Lung Trm cells may be particularly unique; unlike Trm cells in most other NLTs, they require antigen for their formation (Wakim et al. 2015; Takamura et al. 2016; McMaster et al. 2018), develop independently of Hobit (Behr et al. 2019), and experience a high rate of cell turnover (Wu et al. 2014; Slütter et al. 2017). Some studies indicate that lung Trm cell populations may be sustained by continual recruitment of Tcirc cells (Ely et al. 2006; Slütter et al. 2017), whereas Trm cells in other tissues are maintained independently of Tcirc cell input (Gebhardt et al. 2009; Masopust et al. 2010). These observations emphasize that Trm cells residing in different tissues can be functionally distinct and display tissue-specific requirements for local antigen, TFs, and cytokines.

Heterogeneity between Trm cells also exists at the CD4⁺ and CD8⁺ subset level. Parabiosis indicates that the majority of CD4⁺ T cells are resident in peripheral tissues such as the lung, gut, FRT, and liver (Teijaro et al. 2011; Kumar et al. 2017; Beura et al. 2019). Although mouse and human CD4⁺ and CD8⁺ Trm cells display similarities in their transcriptional signatures (Kumar et al. 2017; Strutt et al. 2018; Beura et al. 2019), they may also differ with respect to their expression of certain genes and molecules. For example, CD4⁺ Trm cells are less likely to localize to epithelial tissue compartments and tend to express lower levels of CD103 and Ly6C than CD8⁺ Trm cells (Gebhardt et al. 2011; Watanabe et al. 2015; Collins et al. 2016; Strutt et al. 2018; Beura et al. 2019). CD4⁺ T cells may be less prone to long-term residency than CD8⁺ T cells in epithelial barrier tissues including the skin and salivary glands following infection (Collins et al. 2016; Beura et al. 2019), and recent work suggests that human CD4⁺CD103⁺ Trm-like cells might continuously migrate between skin and blood despite expressing the core Trm cell–associated gene signature (Klicznik et al. 2019). CD4⁺ Trm cells may depend more stringently on IL-7 than CD8⁺ Trm cells with possible tissue- or context-specific requirements for IL-15 and IL-2 and a lower dependency on TGF-β (Adachi et al. 2015; Hondowicz et al. 2016; Romagnoli et al. 2017; Strutt et al. 2018). CD4⁺ Trm cells can also display diverse effector profiles depending on tissue location and stimulation history; in addition to resident Th1 cells (Beura et al. 2019), resident Treg (Sanchez Rodriguez et al. 2014), Th17 (Park et al. 2018a; Amezcua Vesely et al. 2019), and Th2 (Hondowicz et al. 2016; Turner et al. 2018; Bošnjak et al. 2019) subsets have been identified.

Trm cells and tissue-resident innate cells residing in the same organ share transcriptional similarities, implying a tissue-specific overlay of resident lymphocyte imprinting independent of cell type (Mackay et al. 2016). Despite this, even CD8⁺ Trm cells residing in the same tissue can be heterogeneous with respect to development and functionality. CD69⁺CD103⁺ and CD69⁺ CD103^– Trm cells found in either the lamina propria or brain can exhibit distinct proliferative and cytotoxic potential and may display differential dependencies on cytokines including TGF-β, type I IFN, and IL-12 (Wakim et al. 2012; Bergsbaken et al. 2017; Bartolomé-Casado et al. 2019). In addition, CD69⁺CD8⁺ T cells in the lung parenchyma display enhanced cytokine and GzmB production and an altered surface phenotype compared to those residing in the airways (Wein et al. 2019). Furthermore, within the CD69⁺CD103⁺ subset, skin CD8⁺ Trm cells can display diverse cytokine profiles, whereby CD49a⁺ Trm cells producing IFN-γ display enhanced cytotoxicity and those lacking CD49a expression secrete IL-17 (Cheuk et al. 2017). In mice, both IL-17- and IFN-γ-producing skin Trm cells can develop from naive CD8⁺ T-cell precursors with identical TCRs and antigen specificity in response to the same commensal microbes (Naik et al. 2015; Linehan et al. 2018; Harrison et al. 2019). Together, these findings highlight disparities between Trm cells both within and among tissues and imply that infection- and tissue-independent factors can also influence Trm cell identity. Understanding interorgan and interpopulation variation between CD4⁺ and CD8⁺ Trm cells may allow for site-specific modulation of these cells for immunotherapy.

TISSUE MEMORY IN IMMUNE PROTECTION AND PATHOLOGY

Trm cells are poised for immediate detection and control of microbes that breach barrier surfaces (Mackay et al. 2012a; Park et al. 2018b), contrasting with Tcirc cells that may take days to be recruited to the site of infection. CD8⁺ Trm cells have been shown to enhance protection against a broad range of viral, bacterial, and parasitic pathogens in mice including herpes simplex virus (HSV) or vaccinia virus in the skin (Gebhardt et al. 2009; Jiang et al. 2012; Mackay et al. 2012b), LCMV infection of the salivary glands or brain (Hofmann and Pircher 2011; Steinbach et al. 2016), listeria infection of the gut (Sheridan et al. 2014), influenza infection of the lung (Ray et al. 2004; Wu et al. 2014; McMaster et al. 2015; Wakim et al. 2015; Pizzolla et al. 2017), and malaria infection of the liver (Fernandez-Ruiz et al. 2016). In some of these settings, Trm cells can efficiently protect against peripheral infections in the absence of Tcirc cells (Hofmann and Pircher 2011; Mackay et al. 2015b; Zens et al. 2016; Park et al. 2018b). CD4⁺ Trm cells have also been found to confer antimicrobial protection during influenza or streptococcal infections in the lung (Teijaro et al. 2011; Smith et al. 2018), Leishmania infection of the skin (Glennie et al. 2015), and HSV infection of the FRT (Iijima and Iwasaki 2014). In humans, CD8⁺ Trm cells cluster around nerve endings of neurons latently infected with HSV-2 where they are associated with reduced bouts of viral reactivation (Zhu et al. 2007, 2013; Schiffer et al. 2010; Peng et al. 2012) and accumulate at sites of viral persistence during Epstein–Barr virus (EBV) (Woodberry et al. 2005; Woon et al. 2016) and hepatitis (Pallett et al. 2017; Stelma et al. 2017) infection, suggesting that Trm cells in humans serve a similar protective function to those found in mice.

More recently, a critical role for Trm cells in cancer protection has emerged. CD8⁺ Trm-like tumor-infiltrating lymphocyte (TIL) expressing CD103 are abundant in human melanoma (Salerno et al. 2014; Boddupalli et al. 2016a; Edwards et al. 2018), breast (Wang et al. 2016; Savas et al. 2018), ovarian (Webb et al. 2010, 2014; Bösmüller et al. 2016), colorectal (Quinn et al. 2003), brain (Masson et al. 2007), lung (Djenidi et al. 2015; Ganesan et al. 2017), bladder (Wang et al. 2015; Hartana et al. 2018), endometrial (Workel et al. 2016), oropharyngeal (Solomon et al. 2019), and cervical (Komdeur et al. 2017) cancers where they often correlate with improved patient prognoses. Studies in mice have affirmed an anticancer role for Trm cells by revealing that they can protect against challenge with melanoma or head and neck cancer cells bearing their cognate antigen (Enamorado et al. 2017; Malik et al. 2017; Nizard et al. 2017; Gálvez-Cancino et al. 2018; Park et al. 2019). Trm-like TIL often exhibit additional transcriptional features of tissue residency including S1pr1 and Klf2 down-regulation and elevated Znf638 expression (Djenidi et al. 2015; Ganesan et al. 2017; Guo et al. 2018; Savas et al. 2018; Zhang et al. 2018; Clarke et al. 2019), although the extent to which these changes always reflect long-term residency is unclear. Heterogeneous populations of Trm-like TIL that differ with respect to proliferative capacity and effector function can coexist in human lung and breast tumors (Bengsch et al. 2018; Savas et al. 2018; Clarke et al. 2019). At least some of this diversity may result from the presence of bystander Trm cells that reside in transformed tissue but do not actually recognize cancer cells, and recent work indicates that coexpression of PD-1 and CD39 might distinguish tumor-reactive clones from Trm cells of unrelated specificities (Duhen et al. 2018; Simoni et al. 2018). Given that most cancers arise from epithelial tissue layers (Ferlay et al. 2015), Trm cells are likely vital during the early stages of carcinogenesis, as Tcirc cells might have difficulty entering these peripheral tissue compartments in the absence of overt inflammation. However, bona fide tissue-resident cells may become less critical in progressively growing tumors that can recruit their own vascular supply, spread beyond the epithelium, and provide a constant source of antigen. In line with this, Tcirc cells may be sufficient to protect against subcutaneously transplanted tumors in mice (Enamorado et al. 2017), although the extent to which this protection depends on their conversion into de novo Trm cells in the tumor microenvironment remains unclear.

In addition to providing protection against infection and cancer a role for Trm cells in orchestrating autoimmunity is becoming apparent. IFN-γ-producing CD8⁺CD103⁺ skin Trm cells have been shown to mediate or promote vitiligo in both mice (Malik et al. 2017; Richmond et al. 2018, 2019) and humans (Cheuk et al. 2017), whereas IL-17-producing Trm cells have been implicated in the pathogenesis of psoriasis (Boyman et al. 2004; Cheuk et al. 2017). However, commensal-specific IL-17⁺ Trm cells can also promote tissue homeostasis in other settings by accelerating wound healing after skin injury (Linehan et al. 2018), pointing to context-dependent functions of Trm cells that share functional profiles. Further evidence indicates that Trm cells might promote autoimmune pathology in multiple sclerosis (Sasaki et al. 2014; Steinbach et al. 2019), diabetes (Kuric et al. 2017), and inflammatory colitis (Zundler et al. 2019). Trm cells can also induce allergic reactions in the settings of asthma (Hondowicz et al. 2016; Bošnjak et al. 2019) and inflammatory skin hypersensitivity (Adachi et al. 2015; Gamradt et al. 2019). The permanent retention of these cells in afflicted tissues may explain relapsing and remitting bouts of disease at the same site over time.

Trm cells constantly survey their surroundings for potential threats (Gebhardt et al. 2011; Ariotti et al. 2012; Zaid et al. 2014; Beura et al. 2018a) with the degree of immune protection they confer being linked to their density within the tissue (Schiffer et al. 2010; Park et al. 2018b). Whereas Trm cells in the skin and intestines display a slow and restricted mode of migration (Zaid et al. 2014; Thompson et al. 2019b), those in the FRT and liver exhibit more mobile behavior (Fernandez-Ruiz et al. 2016; Beura et al. 2018a; Holz et al. 2018). CD8⁺ Trm cells from both mice and humans can constitutively express cytotoxic molecules such as GzmB (Masopust et al. 2006; Casey et al. 2012; Mackay et al. 2012b; Ganesan et al. 2017) and transcripts for proinflammatory cytokines including IFN-γ (Wakim et al. 2012; Hombrink et al. 2016), although this is not always the case (McMaster et al. 2015; Pizzolla et al. 2017). Following antigen recognition, both CD4⁺ and CD8⁺ Trm cells rapidly up-regulate an arsenal of effector proteins including GzmB, IFN-γ, TNF, and IL-2 (Teijaro et al. 2011; Schenkel et al. 2014a; McMaster et al. 2015; Watanabe et al. 2015; Beura et al. 2019). Together, these molecules may enable Trm cells to limit the spread of infection or cancer by killing or inhibiting proliferation of infected or abnormal cells or may contribute to the induction of autoimmune pathology (Fig. 2). Direct Trm cell-mediated killing of target cells has been demonstrated in vitro (Djenidi et al. 2015), but whether this process also occurs in vivo remains unclear. CD8⁺ Trm cell-derived TNF and IFN-γ are critical for antiviral protection in the lung (McMaster et al. 2015), brain (Steinbach et al. 2016), and FRT (Schenkel et al. 2014a) and allow Trm cells to halt melanoma progression without completely eliminating cancer cells (Park et al. 2019). Similarly, Trm cells can suppress reactivation of latent HSV infections without mediating substantial tissue damage (Mackay et al. 2012b; Zhu et al. 2013). Trm cells in the skin (Collins et al. 2016; Lauron et al. 2019), brain (Wakim et al. 2010), FRT (Iijima and Iwasaki 2014), and gut (Bergsbaken and Bevan 2015; Bergsbaken et al. 2017) can be seen clustering with subsets of antigen-presenting cells including DCs that may cross-present antigen in situ to activate Trm cells in the absence of direct contact with infected cells (Wakim et al. 2010; Shin et al. 2016; Lauron et al. 2019). Importantly, cytokines such as IFN-γ produced by activated CD8⁺ and CD4⁺ Trm cells can recruit and activate downstream innate and adaptive effector cells including DCs, Tcirc cells, B cells, and natural killer (NK) cells to amplify local immunity and inducing a tissue-wide state of pathogen alert that helps to contain viral spread in an antigen-independent manner (Schenkel et al. 2013, 2014a; Ariotti et al. 2014; Beura et al. 2019). In addition, Trm cell activation can promote DC maturation and subsequent antigen spreading during tumor challenge (Menares et al. 2019). The “sense-and-alarm” function of Trm cells may explain why the protection they provide or pathology they induce is often amplified in the presence of Tcirc cells (Enamorado et al. 2017; Nizard et al. 2017; Park et al. 2019; Richmond et al. 2019).

Figure 2.

Functions of Trm cells in infection, cancer, and autoimmune disorders. Upon restimulation, Trm cells can rapidly produce a variety of effector molecules with diverse immune functions, including interferon γ (IFN-γ), GzmB, tumor necrosis factor (TNF), interleukin (IL)-2, and IL-17. Many of the cytokines produced by Trm cells can recruit and activate downstream immune cells found either locally in the surrounding tissue or in the circulation, including B cells, natural killer (NK) cells, and circulating T cells. This “sense-and-alarm” function of Trm cells may combine with their capacity for direct effector function (e.g., via cell killing or direct functions of cytokines on abnormal cells), allowing them to provide superior protection against infection and cancer by comparison with circulating T cells or endowing them with the capacity to orchestrate autoimmunity.

Following restimulation CD8⁺ Trm cells in the skin and FRT can proliferate in situ to maintain or expand the Trm cell pool (Beura et al. 2018a; Park et al. 2018b). Recently recruited Tcirc cells may also divide to give rise to de novo Trm cells at the site of rechallenge (Enamorado et al. 2017; Beura et al. 2018a; Park et al. 2018b; Osborn et al. 2019). The extent to which Trm cells and their progeny maintain functional and anatomical plasticity following reactivation is incompletely understood. Trm cells isolated from the gut and liver and transferred to the circulation of new hosts may retain the capacity to give rise to Tcirc cells and disseminate throughout the body when restimulated (Masopust et al. 2006; Behr et al. 2020; Fonseca et al. 2020). Skin or gut Trm cells reactivated in situ might also be capable of relocating to draining LNs (Beura et al. 2018b) and giving rise to secondary resident or effector memory T cells (Behr et al. 2020), respectively. Restimulated Trm cells that remain in their tissue of origin might also adapt their cytokine profile in response to environmental cues. Following tissue injury or exposure to the inflammatory cytokine IL-18 and antigen recognition, IL-17⁺ CD8⁺ skin Trm cells can acquire the ability to produce Th2-associated cytokines including IL-5 and IL-13 (Harrison et al. 2019). In addition, a proportion of CD4⁺ lung Trm cells that produce IL-17 during differentiation can instead acquire the ability to produce IFN-γ following restimulation (Menares et al. 2019). Whether Trm cell properties can also be altered in the absence of antigenic restimulation remains unclear, but these observations imply that Trm cells may maintain a degree of plasticity in the face of contextual fluidity. Understanding Trm cell fate following reactivation will be of particular importance in active or evolving immune environments including cancer or persistent infection.

EXPLOITING TISSUE-RESIDENCY FOR IMMUNOTHERAPY

The critical role of Trm cells in protecting against or inducing disease implies they could be exploited to improve immunotherapies. First, their regional localization allows them to rapidly curtail pathogen replication at its origin, making Trm cells a highly attractive target for vaccine design (Muruganandah et al. 2018). Immunization protocols designed to embed Trm cells at sites of infection can provide superior prophylactic protection against multiple pathogens when compared to strategies that only induce Tcirc cells, including HSV (Mackay et al. 2012a; Shin and Iwasaki 2012; Çuburu et al. 2015), influenza (Wakim et al. 2015; Zens et al. 2016; Pizzolla et al. 2017), and malaria (Fernandez-Ruiz et al. 2016; Holz et al. 2018) in mice, as well as HIV in primates (Petitdemange et al. 2019). Moreover, cancer vaccines that generate greater frequencies of Trm or Trm-like cells enhance tumor protection in both mice and humans (Murray et al. 2016; Nizard et al. 2017; Gálvez-Cancino et al. 2018). Immunizing at mucosal surfaces may enhance the generation of Trm cells and therefore the protective efficacy of T-cell-based vaccines against both infection (Liu et al. 2010; Jiang et al. 2012) and cancer (Sandoval et al. 2013), but intravenous vaccination can also generate large numbers of peripherally localized Trm cells in some settings (Thompson et al. 2019a). Repeated immunizations may lessen requirements for mucosal priming by encouraging widespread dissemination or boosted accumulation of protective Trm cells in tissues (Jiang et al. 2012; Davies et al. 2017; Park et al. 2018b; Van Braeckel-Budimir et al. 2018).

Antibody-mediated blockade of inhibitory checkpoint (IC) molecules has proven effective for the treatment of various solid cancers but not all patients respond favorably to therapy (Wei et al. 2018). The presence of high numbers of Trm-like TIL correlates with positive responses to immune checkpoint blockade (ICB) in melanoma (Edwards et al. 2018; Savas et al. 2018), and Trm cells can constitutively express various IC molecules including PD-1, CTLA-4, TIM-3, and LAG-3 (Mackay et al. 2013; Ganesan et al. 2017; Park et al. 2018b). Expression of IC molecules may be elevated on T cells following chronic antigen stimulation (Boldison et al. 2014; Gamradt et al. 2019; Wang et al. 2019) and in the tumor microenvironment (Djenidi et al. 2015; Boddupalli et al. 2016a; Workel et al. 2016; Ganesan et al. 2017; Clarke et al. 2019), but the impact these molecules have on Trm cell functionality is incompletely understood. Recent work suggests IC molecules can restrain Trm cell effector functions during persistent antigen recognition in the skin (Gamradt et al. 2019) or lung (Wang et al. 2019), possibly as a means to prevent unwanted autoimmune pathology. Blocking interactions between IC molecules and their ligands may therefore enhance Trm cell effector functions both in vitro (Djenidi et al. 2015; Shwetank et al. 2017) and in vivo (Gamradt et al. 2019; Wang et al. 2019). However, skin Trm cells retain the ability to efficiently protect against viral challenge despite expressing various IC molecules (Park et al. 2018b) and Trm-like TIL often express high levels of effector and IC molecules simultaneously (Djenidi et al. 2015; Savas et al. 2018; Clarke et al. 2019). Studies in mice indicate that only some CD8⁺ T-cell populations are efficiently targeted by ICB; in particular, proliferation of a “stem-like” TCF-1^hi TIL subset appears critical for therapeutic success (Kurtulus et al. 2019; Siddiqui et al. 2019). Trm-like TIL may become more abundant following ICB (Edwards et al. 2018; Clarke et al. 2019), but whether this reflects direct proliferation of the Trm cell pool or expansion of less differentiated “stem-like” TIL that give rise to de novo Trm cells is unknown. Separate approaches through which Trm cell responses might be boosted for cancer treatment include adoptive cell therapy (ACT) and chimeric antigen receptor (CAR) T-cell therapies. In line with this, forced expression of the residency-promoting TFs Runx3 or Bhlhe40 in tumor-specific effector T cells can promote TIL accumulation and melanoma protection in mouse models of ACT (Milner et al. 2017; Li et al. 2019).

On the other hand, devising ways to dislodge Trm cells from tissues may prove beneficial for the treatment of localized autoimmune and allergic disorders. Trm cells in most tissues appear numerically and temporally stable and are not easily displaced by newly formed Trm cells (Holz et al. 2018; Park et al. 2018b). IL-15 blockade was recently shown to deplete skin Trm cells and alleviate disease in a mouse model of vitiligo (Richmond et al. 2018), suggesting that approaches that remove unwanted Trm cells might ameliorate autoimmunity. Ideally, identifying markers that are specifically expressed by Trm cells within afflicted tissues or expressed by subsets of pathological Trm cells may permit selective ablation of pathogenic Trm cells while preserving those that are essential for antipathogen and anticancer protection.

CONCLUDING REMARKS

The critical role for Trm cells in coordinating antipathogen and cancer immunity is now well established. Targeting these cells during vaccination or cancer immunotherapy is therefore an attractive prospect. However, many questions surrounding Trm cell responses remain unanswered and decoding these may unlock novel strategies to boost tissue memory to improve disease treatments. In particular, elucidating molecules and genes that can reliably identify Trm cells across diverse tissues and in active disease settings including cancer or chronic infection will be critical. Recent work has revealed that microbial exposure dramatically alters both the distribution and phenotype of memory T cells in mice (Beura et al. 2016), emphasizing the importance of studying Trm cell biology in humans in addition to laboratory animals. Uncovering more reliable markers for the identification of CD4⁺ and CD8⁺ Trm cells will be imperative to advance the study of these cells in patients where analyses are typically limited to phenotypic or transcriptional profiling.

Trm cells are frequently positioned in barrier tissues that can be coinhabited by commensal microbes and that are subjected to frequent tissue damage and low nutrient availability. Our understanding of how such microenvironmental factors influence the protective capacity and longevity of Trm cells remains incomplete. How are extrinsic cues such as nutrient availability and stress sensed and perceived by Trm cells and how might these regulate their functions, survival, and plasticity? How do Trm cells interact and communicate with surrounding commensals and immune, stromal, and neural cells that comprise the tissue networks in which they reside? Resolving how Trm cell responses are fine-tuned within tissues and tumors may provide insight into the genesis and regulation of both protective immune responses and detrimental autoimmune or inflammatory diseases.

The diversity of Trm cells in different organs and even within a single given tissue is becoming increasingly clear, but the extent of this heterogeneity is incompletely understood. Can we specifically target Trm cells in a particular given organ to better treat tissue-tropic infections, cancers, or autoimmune diseases? Are some Trm cells more functional than others, and if so can we identify those that are the most protective? Can Trm cells become “exhausted” or dysfunctional following exposure to persistent infection or cancer? Or do chronically stimulated Trm cells de-differentiate and give rise to alternate effector and memory T-cell subsets? Are the relative roles of Trm and Tcirc cell subsets different during acute infection compared to chronic disease? Deciphering differences in the requirements and functionality of Trm cells across different tissues and understanding how Trm cell generation and persistence might be differentially regulated during acute versus chronic infection or in the tumor microenvironment will be critical to manipulate these cells for therapeutic purposes.