Tissue-specific Control of Tissue Resident Memory T Cells

Abstract

Tissue-resident memory T (T_RM) cells have emerged to be a major component of T cell biology. Recent investigations have greatly advanced our understanding of T_RMs. Common features have been discovered to distinguish memory T cells residing in various mucosal and non-mucosal tissues from their circulating counterparts. Given that most organs and tissues contain unique microenvironment, local signal-induced tissue-specific features are tightly associated with the differentiation, homeostasis and protective functions of T_RMs. We will discuss the recent advances in T_RM field with a special emphasis on the interaction between local signals and T_RM cells in the context of individual tissue environment.

Introduction

T lymphocytes or T cells are the central component of adaptive immunity. To prepare for the vast majority of potential antigenic encounter, T cells harbor a large repertoire of different T Cell Receptors (TCRs) with diverse reactivity. For each given TCR specificity, only a small number of T cells are present in human and un-manipulated mice due to the limit of total T cells that an individual can host.¹ Therefore, to effectively patrol the most parts of a body for potential pathogen invasion or other antigenic challenge, circulation and migration is an essential feature tightly associated with T cell function.

Under steady state, naïve T cells circulate through secondary lymphoid organs, blood and lymphatic vessels.² Upon antigen stimulation, naïve T cells differentiate into effector T cells with newly equipped effector functions that actively eliminate antigenic sources.³ At later stages of an immune response, most effector T cells die via apoptosis and a small number of survived cells further differentiate into memory T cells, which carry the unique feature of adaptive immunity—immunological memory.⁴ Although long-lasting debates have been focused on the identity of memory precursors and the differentiation path of memory T cells,⁵ recent works have provided solid epigenetic evidence that virus-specific memory T cells transit through an effector stage, not directly derive from naïve cells in both mouse and human.^6,7

Early studies in human peripheral blood have identified two distinct memory T cell populations based on their unique migratory patterns, namely central memory T cells (T_CM) and effector memory T cells (T_EM).⁸ T_CMs carry lymph node homing receptors CCR7 and CD62L, and share a similar circulation path as naïve T cells. In contrast, T_EMs lack CCR7 and CD62L, and prefer non-lymphoid peripheral tissues during circulation. In addition to divergent migration patterns, different proliferative potential and effector functions have been attributed to T_CM and T_EM. Similar circulating memory T cell subsets have been confirmed in mouse and other animal models.⁹

Partition of memory T cells into T_CM and T_EM provides a convenient model to investigate the migration and function of memory T cells. However, in contrast to the generally accepted notion that T_EM cells patrol non-lymphoid tissues under steady state, recent studies have discovered that during local inflammation, T_CMs, but not T_EMs or long-lived effector T cells, migrate into inflamed tissues due to their superior capacity to induce O-glycosylation and generate P/E-selectin ligands, which facilitate the trans-endothelial extravasation of T cells.^10,11 Comparing with T_EM, T_CM cells generally express higher level of chemokine receptor CXCR3, which also enhances the migration of T_CM into inflamed peripheral tissues.¹² Thus, the migration pattern of memory T cells is dynamically controlled by inflammatory signals independent of antigenic stimulation.

A population of non-circulating tissue-resident memory T cells (T_RM) has been identified in almost all non-lymphoid tissues in both human and mouse.^13–19 It has been estimated that the number of T_RM cells exceeds the number of T cells in all lymphoid tissues and entire blood volume combined in both adult human and immunized mice. Therefore, as a newly discovered major T cell population, T_RM is a focus of extensive and active investigations.

Based on the results from decades of research on mucosal T cells, it is quickly realized that mucosal lymphocyte surface marker integrin αEβ7 (CD103) marks mucosa-associated T_RM cells.^20,21 Although with various specificity and accuracy, CD103, together with CD69 has been widely accepted as the common markers to identify T_RM cells in mucosal and some non-mucosal tissues in both mouse and human. Transforming growth factor-β (TGF-β) is a pleiotropic cytokines that control various aspects of T cell biology including thymocyte development, naïve T cell homeostasis and effector/memory T cell differentiation.^22,23 For more than two decades, it has been known that TGF-β enhances the expression of CD103 on activated T cells during in vitro culture.²⁴ Later, it has been validated in different in vivo models that TGF-β signaling is tightly linked with T_RM biology, which will be discussed in details in the following sections.

In current review, we will focus on the recent advances in T_RM biology and will specifically address the following topics: I) Tissue-specific features of T_RM cells; II) Transcriptional control of T_RM cells and III) CD4⁺ T_RM cells. As a rapidly expanding field, exciting overlap between T_RM cells and tumor infiltrating T cells has been observed. Due to the scope of current review, infection-induced T_RM cells will be the major topic with a special emphasis on the relationship between TGF-β signaling and T_RM biology. As CD4⁺ T_RM cells will be discussed in the last section, CD8⁺ T_RM cells will be the main focus in the first two sections.

I. Tissue specific features of T_RM cells

T_RM cells are broadly distributed in both mucosal and non-mucosal tissues outside lymphoid compartment. In addition, a small number of CD8⁺ T_RM cells reside in secondary lymphoid organs isolated from infected mice and Peyer’s patches from naïve mice.^25,26 Interestingly, a substantially increased population of memory T cells bearing T_RM markers is present in the secondary lymphoid organs isolated from adult human^27,28 presumably due to prolonged history of antigen exposure.²⁹ A recent report has demonstrated a direct link between non-lymphoid tissue resident CD8⁺ T cells and T_RM cells in the draining LNs.³⁰ Secondary lymphoid organ T_RMs are largely differentiated from non-lymphoid tissue T_RMs during re-infection. However, the biological importance of secondary lymphoid organ-resident CD8⁺ T_RM is not entirely clear. In this section, we will focus our discussion on the recent discoveries of T_RM cells in a collection of non-lymphoid organs.

A. Skin

Skin harbors a large number of antigen-specific CD8⁺ T_RM cells following various infections in both mouse and human.^31,32 As one of the pioneer focuses of T_RM research, skin T_RM population is relatively well characterized. During the early phase of skin infection, effector CD8⁺ T cells with a Killer Cell Lectin Like Recrptor G1⁻ (KLRG1^-) phenotype (i.e., the common precursors for memory T cells) migrate to the skin via a P/E-selectin ligand- and CXCR6-dependent manner.^33–35 Skin T_RMs isolated from both mouse and human share common TCR sequences with circulating memory T cells in the lymph nodes from the same individual, suggesting that common precursor effector T cells give rise to both skin-resident and circulating memory T cells.³⁶

Using skin Vaccinia virus (VACV) infection model, it has been demonstrated that DNGR-1⁺ dendritic cell (DC)-mediated cross-Priming is specifically required for the formation of skin T_RM, but not for that of circulating memory T cells.³⁷ CD8α⁺ DCs in mouse lymphoid organs and CD103⁺ DCs in non-lymphoid organs express chemokine receptor Xcr1 and C-type lectin DNGR-1 (encoded by Clec9a). This subset of DCs develop in a transcription factor Batf3-dependent manner and exhibit superior capacity to cross-Present exogenous antigen to CD8⁺ T cells.^38,39 Interestingly, it is in the draining lymph node (LN) during the very early phase of naïve CD8⁺ T cell priming that DNGR1⁺ DCs deliver critical signals to instruct CD8⁺ T cells to differentiate into skin T_RM at later stages.³⁷ Cross-Priming DCs extend the retention of activated CD8⁺ T cells in draining LNs via repressing transcription factor Kruppel Like Factor 2 (K1f2) and its target Sphingosine-1-Phosphate Receptor 1 (S1pr1). Defects in cross-Priming DCs result in early egress of CD8⁺ effector T cells from the draining LNs and enhanced accumulation of KLRG1⁺ effector CD8⁺ T cells in the skin at early stages following infection. Further, DNGR-1⁺ DCs provide IL-12, IL-15 and CD24 signals, all of which are required for optimal formation of skin T_RM cells.³⁷ A separated line of research has established Xcr1⁺ cross-Priming DCs as an essential player to convey CD4-help signals during CD8⁺ T cell priming in the LNs.^40,41 Even though CD4-help is not required for the initial recruitment of CD8⁺ effector T cell to the skin,³³ the role of CD4-help in the formation and long-term maintenance of skin T_RM population remains to be determined. Indeed, CD4⁺ T cell depletion leads to enhanced CD8⁺ T cell recruitment to the skin,³³ phenocopying the accelerated lymph node egress and increased skin CD8⁺ T cell accumulation in cross-Priming DC deficient animals at the early phases of an infection.³⁷ Together, it is likely that through cross-Priming DCs, early CD4-help is required for the formation of skin T_RM cells.⁴² However, the molecular programs linking CD4 helped effector CD8⁺ T cells in the LNs with later formed skin T_RMs remain to be elucidated, although a recent work started to dissect the connections.⁴³

After arrival at the skin, CD8⁺ T cells up-regulate CD69 and CD103 in a progressive order.³⁴ CD69 promotes the early retention of CD8⁺ T cells in the skin before the expression of Klf2 and S1pr1 are efficiently suppressed.⁴⁴ Mechanistically, CD69 inhibits the function of S1pr1 and blocks the egress of T cells.⁴⁵ Even though the down-regulation of Klf2 and S1pr1 is a common signature of T_RM,⁴⁶ the rapid induction of CD69 helps to retain T_RM precursors when there is residual activity of S1pr1 at early stages of T_RM differentiation. In the absence of CD69, skin T_RM population is greatly reduced. However, CD69 deficient T cells can differentiate into CD103⁺ T_RMs in the skin, similar as the situation in lung T_RM cells.⁴⁷ These results demonstrate that CD69 per se is not required for the subsequent differentiation of T_RM cells. The induction of CD69 in skin T_RM cells is independent of TGF-β and type I interferon (IFN). Local antigen is not required for skin T_RM formation.⁴⁸ However, local antigen greatly promotes CD69 induction and skin T_RM differentiation.^49–51 Using VACV skin infection model, it has been demonstrated that the early recruitment of activated CD8⁺ T cells to the skin is cognate antigen-independent. After arrival, skin T cells compete for antigen-Presenting cells for cognate antigen recognition, which leads to the induction of CD69. Notably, local TCR signal only provides differentiation, but not proliferation signals to T_RM precursors.^50,51

TGF-β is required for the induction of CD103 and long-term maintenance of skin T_RM cells.³⁴ The expression of CD103 reduces the mobility of skin T_RM cells as demonstrated by multi-photon microscopy.³⁵ Integrin αvβ6 and αvβ8 expressed by keratinocytes cooperate to activate latent TGF-β and are essential for the maintenance of skin T_RM population.⁵² Interestingly, the activity of αvβ6 and αvβ8 is continuous required even after the establishment of skin T_RM population. This observation suggests that the unique T_RM transcription program is not permanently fixed. Instead, constant environmental cues (e.g., TGF-β) are essential to maintain the identity of skin T_RM cells at least under the circumstance of systemic viral infection. Most CD8⁺ skin T_RM cells reside in the epidermis layer. Hair follicle derived IL-7 and IL-15 and a special metabolic program involving exogenous lipid uptake are required for the long-term survival of skin T_RM cells.^53,54 Further, as a skin homing chemokine receptor, CCR10 is required for the long-term homeostasis of both circulating and skin-resident memory CD8⁺ T cells during skin infection.³⁵

A small number of CD8⁺ T cells can seed distal non-infected regions of the skin and provide critical local protection, suggesting that a low number of skin T_RMs are sufficient to provide effective protection.^33,49 Re-encounter of cognate antigen induces rapid activation of skin T_RM cells. Activated T_RMs stimulate both innate and adaptive immune components of the skin and recruit circulating memory T cells in an IFN-γ-dependent manner.⁵⁵ Skin T_RMs provide sufficient protection in adult human as demonstrated by the lack of infections in alemtuzumab-treated cutaneous T cell lymphoma patients whose circulating T cells are depleted while skin T_RM cells are spared.⁵⁶ During the recall response, skin T_RM cells undergo in situ expansion and contraction, and do not rejoin circulating effector/memory T cell pool. Thus, skin T_RM population is a relatively stable local immune component during the subsequent challenges.⁵⁷

Most previously mentioned skin T_RM cells are generated in response to a specific pathogen introduced by intradermal injection or scarification, which causes both local infection and skin damage. Without causing skin injury, topical application of certain strains of skin commensal bacteria leads to a typical antigen-specific CD8⁺ T cell response including expansion, contraction and long-term maintenance of a memory T cell population in the skin carrying a common T_RM phenotype (i.e., CD69⁺CD103⁺).⁵⁸ In addition to IFN-γ producing cells, this commensal bacteria-induced skin T_RM population contains a significant subset of IL-17 producing cells. Cross-priming DCs are required for the formation of these IL-17⁺ CD8⁺ T_RM cells. In addition to the divergent effector cytokines, regular pathogen-induced T_RMs are different from commensal-specific T_RMs in three major aspects: 1) Pathogen infection-induced skin T_RM cells are largely restricted to the injured site while commensal-specific T_RMs are scattered; 2) Pathogen infection-induced T_RMs directly respond to infected epidermal cells to produce IFN-γ while IL-17 production from commensal-specific T_RM cells requires CD11b⁺ local DCs;58–60 and 3) the unique population of IL-17 producing CD8⁺ skin T_RM cells differentiate from non-classical MHC-Ib-restricted CD8⁺ T cells and promote tissue repair.⁵⁹

As a common T_RM signature, skin T_RM cells exhibit a T-bet^loEomes^neg phenotype.⁶⁰ In T-bet deficient cells and therefore complete lack of T-box transcription factors, skin T_RM cells up-regulate transcription factor RORγt and become IL-17 producing cells.⁶⁰ Similar IL-17-Producing CD8⁺ T cells have been observed in T-bet/Eomes double deficient T cells in lymphoid organs after systemic viral infection.⁶¹ Commensal bacteria-induced IL-17⁺ skin T_RM cells carry minimal amount of T-bet while maintain a high level of RORγt, suggesting that upon the suppression of T-box transcription factors, RORγt-mediated type 17 effector program may be an important default path of CD8⁺ T cell differentiation. Importantly, IL-17 producing CD8⁺ T cells are present in both human and non-human primates.⁵⁹ In human skin, the presence or absence of CD49a expression can divide T_RMs into IFN-γ or IL-17-Producing cells and associated with type 1 or type 17 effector T cell-related disease settings.⁶² However, the molecular and cellular control of the type 17 effector program in CD8⁺ T cells or T_RM cells remains unclear.

Together, skin T_RM cells are differentiated from common memory T cell precursors in the circulation. Local antigen is not required, but significantly promotes the induction of CD69 and differentiation of skin T_RMs. Local signals including TGF-β, IL-7 and IL-15 controls the formation and homeostasis of skin T_RMs. Pathogen-induced and commensal-specific T_RMs exhibit distinct features.

B. Lung

Lung T_RM cells exhibit both common T_RM features and lung-specific properties. Most lung parenchyma and interstitium-resident CD8⁺ memory T cells carry either CD69⁺CD103⁺ or CD69⁺CD103⁻ phenotype, similar as T_RMs isolated from most other mucosal sites. Lung CD8⁺ T_RM cells are essential for the local protection against influenza viral infection in mouse.^63–66 Further, it has been recently confirmed that influenza-specific T_RM cells isolated from human lungs mount a robust proliferative response with superior effector functions.⁶⁷

The differentiation of CD103⁺ lung T_RM cells is TGF-β-dependent^64,68 and requires CD4-help and cross-Priming signals from DNGR-1⁺ DCs during the initial priming phase.^37,42 4–1BB signal to T cells is critical for the formation of lung T_RM cells in a competitive setting.⁶⁹ In addition, Notch signaling and Notch inducing transcription factor EGR2 are up-regulated in CD103⁺ lung T_RM cells and essential for the formation and maintenance of lung T_RM cells.^70,71 The survival of CD103⁺, but not CD103⁻ lung T_RM requires IL-15 signaling.⁶⁰ Interestingly, Notch may promote the maintenance of lung T_RM cells via an IL-15-independent and metabolism-related mechanism.⁷⁰

In contrast to antigen-independent and local inflammation-driven T_RM differentiation in the vagina and salivary gland, local cognate antigen is essential for lung T_RM differentiation,^47,64 consistent with the findings that T cells with different TCR specificity elicit distinct T_RM forming potential during polyclonal response against influenza virus infection in mice.^67,72,73 Further, TCR signal can induce the expression of anti-viral protein IFITM3 (Interferon Induced Transmembrane Protein 3) in lung T_RM cells. IFITM3 protects lung T_RM cells from direct viral infection-induced cell death.⁷⁴

Distinct from long lasting protection provided by T_RM cells residing in other mucosal sites, lung T_RM cells wane over time due to enhanced apoptosis of CD103⁺ T_RM in lung microenvironment.^63,75 The maintenance of CD8⁺ memory compartment in the lung requires continuous recruitment of circulating memory T cells.^76,77 In contrast to lung T_RM differentiation during the acute phase of influenza virus infection, at the memory phase, previously infected lung is permissive to de novo T_RM differentiation from circulating T_EM cells in a cognate antigen-independent, but IL-33 and TNF-dependent manner.⁷⁵ The waning of lung T_RM cells is caused by the increase of circulating T_CM and decrease of T_EM, and therefore the decline of continuous T_RM induction.

Not completely exclusive from the above explanation, recent publications have identified a specific niche in injured lungs that supports T_RM differentiation and maintenance. These lung T_RM niches are the tissue repair-associated regions and co-localize with the production of T_RM promoting factors TGF-β and IL-15.^47,72 Notably, the regeneration of damaged airway epithelium is also TGF-β-dependent,⁷⁸ providing an example of complex functions of TGF-β signaling in both local immunity and tissue homeostasis. The gradual decline of lung T_RM cells is caused by the completion of injured lung regeneration and the shrinking of lung T_RM niches. These observations are also consistent with the lack of lung T_RM formation in most systemic infection models that do not induce significant lung injury.

Thus, gradual changes of both circulating memory T cells and lung microenvironment may be together responsible for the decline of lung T_RM cells over time. However, there is one piece of observation needed to be reconciled. Considering that T_EM cells continuously migrate and differentiate into lung T_RM cell after influenza viral infection,⁷⁵ it seems counterintuitive that seven weeks of parabiosis does not lead to significant de novo lung T_RM formation in a similar animal model.⁴⁷ One possible explanation may be that parabiosis surgery itself causes unexpected inflammation and tissue damage. Systemic and local inflammatory signals introduced by surgical procedures may alter the migration of circulating memory T cells as TCM cells are sensitive to inflammation-induced O-glycosylation and migration.¹¹ It is well documented that surgery has immediate impacts on the immune system of human patients.⁷⁹ Therefore, even as the golden standard in T_RM research, the results from parabiosis experiments should be carefully interpreted along with the experiments involving less invasive procedures.

CD69 promotes the early migration and retention, but not the differentiation or long-term maintenance of lung T_RM cells.⁴⁷ Similar as skin T_RMs, early expression of CD69 inhibits the residual activity of K1f2/S1pr1 pathway. In addition, CD69 may directly facilitate effector CD8⁺ T cell migration into inflamed lung via interacting with its ligands myosin light chain 9, 12a and 12b.⁸⁰

Using VACV immunization and infection models, it has been demonstrated that intranasal, but not systemic intra-Peritoneal infection induces the differentiation of two populations of lung CD8⁺ T_RM cells, i.e., a major population of CXCR3^lo interstitium T_RM and a minor population of CXCR3^hi airway T_RM. CXCR3^lo T_RM cells provide critical protection.⁸¹ An independent investigation also confirms that CXCR3^hi and CXCR3^lo lung CD8⁺ T cells represent different differentiation stages in response to local inflammation (e.g., IL-12 and IL-15) and occupy distinct niches in the lung. Further, cooperative action from both CXCR3^hi and CXCR3^lo lung T_RMs is required for the protection against lethal respiratory VACV challenge.12 Considering that CXCR3^hi airway-resident CD8⁺ T cells are established protectors against respiratory infections,^82,83 lung T_RM cells may not represent a homogenous population of cells. Instead, different subsets or differentiation stages of T_RM cells may occupy different niches and cooperate to achieve maximal protection.

In terms of effector functions for lung T_RM, in addition to IFN-γ production, which is a common effector cytokine produced by T_RMs isolated from various tissues, tissue-specific production of IL-22 is labeled as a unique feature for lung T_RM cells upon cognate antigen re-stimulation.⁸⁴

Together, during the effector phase of a respiratory infection, lung T_RM cells are formed in a cognate antigen-dependent and CD69-dependent manner. During the memory phase, de novo T_RM formation and maintenance may be mediated by an antigen- and CD69-independent mechanism. Local signals, such as TGF-β, IL-15, IL-33 and TNF promote lung T_RM differentiation and homeostasis. Different microenvironment inside the lung supports various subsets of lung T_RM cells.

In addition to lung T_RM, a series of elegant investigations have established upper respiratory tract as a key site to support local T_RM against respiratory viral infection.^73,85,86 Virus-specific CD69⁺ and CD69⁺CD103⁺ CD8⁺ T cells carrying common T_RM signature genes can be readily isolated from upper respiratory tract including nasal tissue and nasal-associated lymphoid tissues in mouse and tonsils in human. Distinct from TGF-β- and cognate antigen-dependent induction of lung T_RM cells, both TGF-β and cognate antigen recognition are dispensable for upper respiratory tract T_RM cells. Further, in contrast to the gradual decline of lung T_RM cells over time, upper respiratory tract T_RM cells persist at a steady level and are sufficient to provide protective immunity.⁷³ Thus, upper and lower respiratory tract associated T_RM cells provide us a perfect example that different local environment supports T_RM differentiation and maintenance through distinct mechanisms.

C. Intestine

Intestine contains one of the largest mucosal surfaces in the body. The complete overview of intestinal T cell components is beyond the scope of current review. In this section, we will limit our discussion to CD8αβ⁺TCRαβ⁺ memory T cells residing in the intraepithelial lymphocyte (IEL) and lamina propria (LP) compartments of the intestines.

Both local and systemic infection leads to the generation of gut CD8⁺ T_RM cells. Local infection is often more effective in inducing gut T_RMs. Generally, most T_RM cells carry a CD69⁺CD103⁺ surface phenotype in the IEL compartment while LP T_RM cells contain both CD69⁺CD103⁺ and CD69⁺CD103⁻ subsets. CD103⁺ T_RM cells are evenly distributed. CD103⁻ T_RM cells are clustered around infected loci in the LP of both small and large intestines via a CXCR3-dependent mechanism and critical for local immunity.⁸⁷ In contrast to most other mucosal and non-mucosal tissues, local antigen is not required for the differentiation of gut T_RM cells. Indeed, persistent local antigen may inhibit gut T_RM cell formation revealed by delayed induction of CD103.⁸⁸

During the early phase of oral infection, intestinal CD8⁺ T cells congregate around infected cells and receive inflammatory signals (e.g., IL-12). IL-12 prevents the induction of CD103. However, in the absence of IL-12 signaling, although the initial induction of CD103 is accelerated, the long-term survival of both CD103⁻ and CD103⁺ gut T_RM cells is significantly impaired.⁸⁹ These results provide an elegant example of the complicated impacts of local inflammation on the differentiation and maintenance of T_RM cells.

During chronic viral infection, TGF-β inhibits the expression of gut-homing receptor integrin α4β7 on effector CD8⁺ T cells isolated from secondary lymphoid organs and therefore dampens the migration of effector CD8⁺ T cells to the intestine.⁹⁰ Interestingly, low dose rapamycin treatment during the effector phase of an immune response inhibits the expression of gut homing receptors and greatly reduces the formation of gut T_RM population.⁹¹ Considering that TGF-β inhibits the serine and threonine kinase mammalian target of rapamycin (mTOR) in NK cells,⁹² the possible crosstalk between TGF-β signaling and mTOR pathway in gut T_RM cell biology warrants future investigation.

In the intestinal tissues, TGF-β signaling is required for the induction of CD103, but dispensable for CD69 expression.⁸⁷ TGF-β is essential for the differentiation of gut T_RM cells during both local and systemic infections as TGF-β unresponsive T_RMs (both CD103⁺ and CD103⁻) are dramatically reduced in both IEL and LP compartments of the intestines at the memory phase of an immune response.^90,93 However, in contrast to the generally accepted notion that CD103 helps gut T_RM cell retention by interacting with epithelial derived E-cadherin, CD103 deficient T cells only exhibit a two-fold reduction in the initial establishment, but not in the long-term maintenance of gut T_RM cells in the IEL compartment.⁹³ CD103 is not involved in the homeostasis of gut T_RMs in the LP compartment.^88,93 These results demonstrate that TGF-β mediates essential functions via CD103-independent mechanisms in gut T_RM cells.

Through abrogating the function of latent TGF-β activating integrin αvβ6 in the gut, it has been recorded that continuous TGF-β signaling is required for the maintenance of gut T_RM cells in the IEL, but not LP compartment.⁵² However, underlying mechanisms explaining the difference between IEL and LP compartments are not addressed. Different TGF-β dependency of T_RM subsets or additional molecules mediating the activation of local TGF-β in the LP may be the possible explanations. The factors that mediate the long-term survival of gut T_RM cells are not clear. In contrast to lung and skin T_RM cells, IEL and LP gut T_RM cells are maintained in an IL-15-independent manner.⁹⁴

In addition to the common effector molecules associated with memory CD8⁺ T cells, gut T_RM cells have been demonstrated to produce both type I and type III IFNs to activate the innate antiviral status of gut epithelium.⁹⁵ However, whether these properties are gut-specific or generally associated with T_RMs isolated from other tissues awaits future clarification.

Together, intestinal T_RM is formed during both local and systemic infection in a cognate antigen-independent manner. CD103⁺ and CD103⁻ T_RM cells exhibit different location and function. TGF-β, but not IL-15 is required for the initial differentiation and long-term maintenance of gut T_RM cells. The capacity to respond to (e.g., IL-12R) or being recruited to (e.g., CXCR3) local inflammatory loci controls the formation and homeostasis of gut T_RM cells.

D. Female reproductive tract

Female reproductive tract (FRT) represents another well-studied mucosal tissue in T_RM field. Similar to skin, CD8⁺ T_RMs are highly enriched in the epithelial layer of FRT.³² Remarkably, different segments of FRT exhibit distinct immunological properties that impact CD8⁺ T cell priming and T_RM formation. The unique local environment of lower FRT restricts the immediate production of type I and type III IFN following vaginal viral infection, which in turn results in defective DC maturation and delayed CD8⁺ T cell priming.⁹⁶ Interestingly, the dampened innate immune response is restricted to lower FRT while upper FRT mounts a relatively normal response. The mechanisms underlying this striking difference between lower and upper FRT remain unknown. In addition to variable immune components, different epithelial structure and the restricted association of microbiome with lower FRT may be the potential contributing factors.⁹⁷

After priming, the migration of activated CD8⁺ T cells to vaginal mucosa is tightly regulated by local immune environment. Following vaginal infection, CD4⁺ T cell-derived IFN-γ activates FRT epithelium to produce CXCL9/10 and enhance the migration of antigen-specific effector CD8⁺ T cells in a CXCR3-dependent fashion.⁹⁸ Dysbiosis-induced IL-33 production leads to greatly enhanced ILC2 (type 2 innate lymphoid cell) mediated accumulation of eosinophils and defective recruitment of both CD4⁺ and CD8⁺ T cells after vaginal herpes virus infection,⁹⁹ suggesting a potential crosstalk between lower FRT associated microbiome and T_RM formation. Different from current paradigm that DCs carry local antigens to the draining LNs to prime antigen-specific naïve T cells, lower FRT mucosa is able to support naïve CD8⁺ T cell priming and proliferation in situ without the involvement of secondary lymphoid organs.¹⁰⁰ The significance of mucosa initiated naïve T cell priming remains to be validated in different infection settings. Further, whether different priming sites (i.e., vaginal mucosa versus draining LNs) impact the formation of T_RM cells is left to be demonstrated.

Local antigen recognition is not required for FRT T_RM differentiation.⁸⁸ Non-specific local inflammation or exogenous chemokines are sufficient to attract circulating CD8⁺ T cells and allow newly recruited cells to further differentiate into long-lasting CD69⁺CD103⁺ FRT-resident memory T cells.^48,101 IL-15 is not required for the homeostasis of FRT T_RM population.⁹⁴ Upon antigenic recall, T_RMs quickly produce IFN-γ and function as an alarming system to activate both local innate and adaptive immune components and recruit circulating memory T cells.^102,103 CD301b⁺ LP DCs are required to activate vaginal T_RMs upon vaginal herpes virus re-challenge.¹⁰⁴ In contrast, dorsal root ganglia resident T_RMs are re-activated by recruited monocyte-derived DCs¹⁰⁵ and skin CD8⁺ T_RMs are reactivated by almost any directly infected epidermal cells carrying cognate antigens.¹⁰⁶ Recent results have demonstrated that FRT T_RM cells undergo expansion and differentiation in situ during re-challenge. This T_RM-autonomous response dominants the local CD8 recall response. Thus, in addition to a sentinel system, mucosal T_RM function as a robust self-sufficient defense system and can function independent of circulating T cells.^104,107

Together, the formation and maintenance of FRT T_RM are independent of local antigen and IL-15. Different regions of FRT harbor distinct immune environment that impacts CD8⁺ T cell response. Local DCs are required for the recall response of FRT T_RMs. The involvement of TGF-β signaling in FRT T_RMs remains to be determined.

E. Non-mucosal tissues

Following systemic infection, kidney supports the differentiation and maintenance of a significant population of both CD69⁺ and CD69⁻ T_RM cells.¹⁰⁸ Similar as other non-mucosal T_RM cells, most kidney T_RM cells do not express CD103. TGF-β is required for the optimal differentiation of kidney T_RMs via facilitating effector CD8⁺ T cell extravasation. Mechanistically, TGF-β signaling promotes the expression of CXCR3 and E/P-selectin ligands on effector CD8⁺ T cells. Both CXCR3 and E/P-selectin ligands participate in the transendothelium migration of CD8⁺ effector T cells in the kidney.¹⁰⁹ The potential functions of local antigens in kidney T_RMs have not been determined. One study has found that kidney T_RMs are enriched for T cells with high-affinity TCRs during chronic viral infection,¹¹⁰ suggesting that local antigen may facilitate kidney T_RM induction. Common T_RM transcriptional program is active in kidney T_RM cells as deficiency in transcription factors Blimp-1, Hobit or Runx3 leads to impaired maintenance of kidney T_RM cells.^111,112 Similar to skin, lung and salivary gland T_RM cells, the long-term maintenance of kidney T_RM cells is IL-15-dependent.^60,94 However, the protective function of kidney T_RM cells remains to be demonstrated. During polyomavirus BK reactivation following kidney transplant in human patients, the presence of CD69⁺ kidney T_RM cells is associated with diminished effector functions and poor virus control while CD69⁻ kidney CD8⁺ T cells are associated with better clinical outcomes.¹¹³ Together, TGF-β promotes the formation of kidney T_RMs. IL-15 is required for the long-term survival of kidney T_RMs while the protective function of kidney T_RMs remains to be determined.

Even though considered as a non-mucosal tissue, salivary gland supports the differentiation and maintenance of a significant population of CD69⁺CD103⁺ intraepithelial CD8⁺ T_RM cells. Similar as intestinal mucosal, but distinct from most other tissues which have been examined, the differentiation of salivary gland T_RM does not require cognate antigen recognition.^114–116 Local inflammation does not affect T_RM differentiation in salivary gland as least during murine cytomegalovirus infection. Integrin α4β1 is required for the accumulation of CD8⁺ T cells in the salivary gland mediated via the interaction with endothelial VCAM-1 (Vascular Cell Adhesion Molecule-1).^115,117 Often used as an epithelium marker, E-cadherin is highly expressed by salivary gland T_RM and promotes CD8⁺ T cell accumulation presumably via homotypic interactions between E-cadherin.¹¹⁶ TGF-β signaling is required for the induction of CD103 and long-term maintenance of salivary gland T_RMs. Similar as the situation for intestinal T_RM cells, CD103 itself is only involved in the initial establishment, but not long-term maintenance of salivary gland T_RMs. Thus, CD103-independent but TGF-β-dependent mechanisms may be essential for the maintenance of T_RM cells. The initial induction of CD69 is TGF-β- and type I IFN-independent and may involve the signals from IL-33 and TNF.114,118 IL-15 is not required for the initial differentiation,¹¹⁴ but essential for the long-term survival of salivary gland T_RMs.⁹⁴ In summary, initial CD69 induction on salivary gland T_RMs is independent of local antigen and TGF-β. The long-term maintenance of T_RMs requires both TGF-β and IL-15. Adhesion molecules including integrin α4β1, CD103 and E-cadherin promote the accumulation of salivary gland T_RMs.

Home to a large collection of diverse immune cell types, liver has been proposed as a lymphoid organ and functions as an essential battle field against various liver-targeting pathogens, such as malaria, hepatitis B and hepatitis C virus.¹¹⁹ Murine liver T_RM cells are identified as CD69⁺CXCR6⁺CXCR3⁺CD11a⁺CD103⁻.^120,121 In contrast to most T_RM cells that are located outside the vasculature, the vast majority of liver T_RM cells reside inside the blood vessels and display active crawling behavior to patrol hepatic sinusoids.^{108,120–122} This unique feature excludes the usage of intravascular labeling technique in liver T_RM research. Even constantly exposed to blood circulation, liver T_RM are not travelling along the bloodstream and considered as bona fide liver-resident cells as demonstrated by parabiosis experiments. Their liver residency is dependent on integrin LFA-1 (Lymphocyte Function-associated Antigen-1) and chemokine receptor CXCR3.^120,121 Local antigen is required for liver T_RM formation. At the transcription level, both Blimp-1 and Hobit are required for the maintenance of liver T_RM in mouse.¹¹¹ However, liver T_RM isolated from human hepatitis B virus infected patients display a Blimp-1^hiHobit^lo phenotype.¹²³ In contrast to the situation that most mouse liver T_RMs are CD103⁻, a distinct CD69⁺CD103⁺ T_RM population is present in human liver. TGF-β together with IL-15 may mediate liver T_RM formation in human. However, the function of TGF-β in mouse liver T_RM has not been determined. Regarding the effector functions, both human and mouse liver T_RMs are associated with enhanced local protection. Interestingly, in response to TLR4 or TLR9 signals, inflammatory monocytes forms cocoon-like cell aggregates in mouse liver to support local proliferation of CD8⁺ T cells. These cellular structures may represent a key site for liver T_RM function.¹²⁴ Human liver T_RMs display an IFN-γ^hi IL-2^hi GranzymeB^lo phenotype¹²³ while mouse liver T_RMs are GranzymeB^hi IFN-γ^{hi 120} and produce colony-stimulating factor-2.⁸⁴ In addition to the species difference, various infection settings may also contribute to the phenotypic and functional distinctions in liver T_RM populations. Together, liver T_RMs are closely associated with blood vasculature. Local antigen recognition is required for the induction of liver T_RMs. LFA-1 and CXCR3 promote liver T_RM formation. The requirement of TGF-β signaling remains to be determined. Prominent distinctions have been identified between mouse and human T_RMs in the liver.

Following local infection, CD8⁺ effector T cells migrate to the brain and differentiate into both CD69⁺CD103⁻ and CD69⁺CD103⁺ T_RM cells.^125,126 Local antigen presentation is required for the differentiation of brain T_RM cells, consistent with the findings that during persistent brain infection, the TCR affinity of brain T_RM cells gradually increases.¹¹⁰ TGF-β signaling likely promotes the induction of CD103⁺ brain T_RM cells as depletion of Foxp3⁺ regulatory T cells (Tregs) and Treg-derived TGF-β impairs the formation of brain T_RM population.¹²⁷ Further, similar as kidney T_RM cells, TGF-β promotes the trans-endothelial migration of CD8⁺ effector T cells into the brain.¹⁰⁹ Locally produced survival cytokines IL-7 and/or IL-15 may promote brain T_RM homeostasis as a sizable population of brain T_RMs contains phosphorylated STAT5 and undergoes homeostasis proliferation in vivo.¹²⁸ Interestingly, pSTAT5⁺ and proliferating brain T_RM cells are enriched around the brain surface comparing with brain parenchyma. However, the cellular source and location of IL-7 and IL-15 have not been determined. Brain T_RM cells up-regulate the expression of inhibitory receptors PD-1 and CTLA-4.¹²⁶ PD-1/PD-L1 interaction is required to limit the accumulation of PD-1^hiCD103⁻ CD8⁺ T cells and promote brain T_RM population.¹²⁹ In contrast to lung and FRT T_RMs, the differentiation and maintenance of brain T_RM cells are independent of CD4-help.^126,128 Cognate antigen re-challenge activates brain T_RMs to recruit circulating memory T cells. However, in the absence of circulating memory T cells, activated brain T_RM cells proliferate in situ and provide sufficient immune protection.¹²⁸ Thus, in addition to functioning as a component of local alarming system, brain T_RMs can function as an organ-autonomous defense system. Together, brain T_RM induction depends on local antigen encounter and is independent of CD4-help. TGF-β promotes T_RM formation and IL7/15 may provide the survival signals for T_RMs in the brain.

In response to oral infections, mesenteric white adipose tissue supports efficient CD69⁺CD103⁻ T_RM differentiation. Surprisingly, white adipose tissue contains more T_RM cells and provides better protection than intestinal LP against intestinal infections.¹³⁰ White adipose tissue T_RM cells carry the receptors for survival cytokines IL-15 and IL-7. However, the local signals which drive the differentiation and maintenance of while adipose tissue T_RM remains unknown. Considering the unique metabolic requirement of T_RM cells, i.e., the uptake of exogenous lipid,⁵⁴ white adipose tissue may provide an ideal environmental niche for T_RMs.

With proper infection settings, almost all non-lymphoid organs support the differentiation of T_RM cells. Different microenvironment and local signals dictate the phenotype and behavior of T_RM cells isolated from various tissues. The tissue specific features of CD8⁺ T_RM isolated from various non-lymphoid organs are summarized in Table 1. Be aware that when T_RM is studied in different tissues, it is often involved distinct infection models. Therefore, in addition to tissue-specific local environment, infection-specific properties may also impact T_RM cells.

Table 1.

Tissue-specific features of CD8⁺ T_RMs.

	Local Ag	TGF-β	IL-15	Xcr-1+DC	CD4-help	Unique Effector
Skin	N.R., but promote	Required*	Required	Required	N.R. ?	IL-17 from a subset
Lung	Required	Required*	Required	Required	Required	IL-22
URT	N.R.	N.R.	Unknown	Unknown	Unknown	Unknown
SI	N.R./Suppress	Required*	N.R.	Unknown	Unknown	Type I/III IFN
FRT	N.R.	Unknown	N.R.	Unknown	Required	Unknown
Kidney	Unknown	Promote	Required	Unknown	Unknown	Unknown
SG	N.R.	Required*	Required	Unknown	Unknown	Unknown
Liver	Required	Maybe	Maybe	Unknown	Unknown	CSF-2
Brain	Required	Maybe	Maybe	Unknown	N.R.	Unknown
WAT	Unknown	Unknown	Maybe	Unknown	Unknown	Unknown

URT, Upper respiratory tract; SI, Small intestine; FRT, Female reproductive tract; SG, Salivary gland; WAT, White adipose tissue.

Ag, Antigen; N.R., not required.

^*, TGF-β is required for CD103⁺ TRM, but not for CD69⁺CD103⁻ TRM.

^?, The involvement of CD4-help in skin T_RM formation may require future investigation.

II. Transcriptional regulation of T_RM cells

Transcription factors (TFs) control the development of multiple immune cell types through activating and/or repressing genes that are critical to cell identity.¹³¹ Comparing with effector and memory T cells in lymphoid tissues and circulation, T_RM cells in non-lymphoid tissues are a unique and distinct memory T cell population that displays a specific TF expression pattern. T_RM cell fate is determined by the integrated activity of multiple TFs, which contributes to optimal survival and function within their local environment.¹³² Functional illumination of TFs-modulated T_RM formation will facilitate future manipulation of these TFs to foster T_RM accumulation, which ultimately yield desirable and effective protective memory T cells in tissues. In this section, TFs with a well-established role in T_RM formation are discussed in details below.

A. Krüppel-Like Factor 2 (KLF2)

Krüppel-like factors (KLFs) are a family of zinc-finger TFs including 15 mammalian family members, in which Klf2 is one of the core transcriptional regulators that affect T cell trafficking.^133,134 The reduction of Klf2 is required to establish tissue-residency of various immune cells, including mouse and human CD8⁺ T_RM,^46,86,111 CD4⁺ T_RM,135 NK and NKT cells,^111,136 and CD8αα⁺ TCRαβ T cells and TCRγδ T cell in the IEL compartment of the gut.¹³⁷ As a possible exception to the universal down-regulation of Klf2 in tissue-resident lymphocytes, a significant population of conventional TCRαβ T cells (both CD4⁺ and CD8β⁺) in the IEL compartment of the large intestine, but not in that of the small intestine maintain a high level of Klf2 expression.¹³⁷ The biological significance of this unique expression pattern of Klf2 in large intestine IEL remains unclear. Klf2 controls the expression of receptors required for emigration and peripheral trafficking, including S1pr1, CD62L, CCR7 and β7 integrin.¹³³ Once entry into peripheral non-lymphoid tissues, local cytokines such as TGF-β, IL-33 and TNF cooperate to extinguish the expression of both Klf2 and its target S1pr1, which potentiates the retention of T_RM cells in the tissue.^46,75,88 Mechanistically, PI3K/Akt pathway is activated by cytokine signals to inhibit the expression of transcription factor Foxo1 and therefore enforce the down-regulation of K1f2 46 Various combinations of pro-inflammatory cytokines can suppress the expression of Klf2 in activated CD8⁺ T cells in vitro, including type I IFN, IL-12 and IL-18.^46,88 However, type I IFN does not significantly alter T_RM differentiation⁴⁴ and IL-12 inhibits early differentiation and promotes long-term maintenance of gut T_RM.⁸⁹ Thus, the function of different cytokines in T_RM cells will require further clarification in different tissues under various inflammatory conditions in vivo. TCR signal is not involved in the down-regulation of Klf2 during acute viral infections. In summary, as a key regulator of T cell trafficking, local cytokine-mediated repression of Klf2 is essential to establish tissue residency of most T_RMs.

B. T-bet and Eomes

T-bet (encoded by Tbx21) and Eomesodermin (encoded by Eomes), as two members of T-box binding TFs, are essential regulators for the differentiation and function in distinct immune cells including CD4⁺ T, CD8⁺ T, NKT, NK, innate lymphocytes and B cells.¹³⁸ In CD8⁺ T cells, Tbx21 expression is highest in short-lived effector cells, whereas Eomes expression is increased in long-lived memory cells.¹³⁹ Temporal and spatial down-regulation of both Tbx21 and Eomes represents a pivotal step in the lodging and maturation of skin T_RM cells, in which Eomes is virtually extinguished, either before or after CD8⁺ T cells enter the epithelium and prior to the acquisition of CD103^.60 However, Tbx21 deficient CD103⁺CD8⁺ T_RM cells ultimately vanish over time, because complete loss of Eomes during the final maturation of CD103⁺CD8⁺ T_RM cells renders them dependent on low level of T-bet for persistent survival. This phenomenon is supported by the notion that at least one T-box TF, in particular T-bet, is necessary to maintain the expression of cytokine receptor subunit CD122 (IL-2/IL-15R β chain), which delivers a survival signal to certain memory T cell populations.^60,140,141 However, the requirement of IL-15 in the long-term survival of T_RMs is tissue type-dependent.⁹⁴ Whether IL-15-independent T_RMs requires residual expression of CD122 and T-bet remains to be clarified. In addition, complete lack of T-box TFs in CD8⁺ T cells may lead to the activation of RORγt-mediated type 17 effector program. The transcriptional regulation of type 17 CD8⁺ effectors is not entirely understood.

Smad3 is required for TGF-β mediated CD103 (encoded by Itgae) expression. Both Smad3 and T-bet directly bind to the first intron of Itgae locus, suggesting the potential mechanisms by which T-bet might repress Itgae transcription. T-bet may directly compete with Smad3 for DNA binding, interact with Smad3 to prevent its transcription, or recruit other transcriptional repressors to the Itgae locus.⁴² Further, the well-orchestrated down-regulation of both T-bet and Eomes strengthens TGF-β signaling pathway that reciprocally inhibits the expression of T-bet and Eomes, indicating a feed-forward loop forms to optimize CD103⁺CD8⁺ T_RM cell formation.⁶⁰ Interestingly, previous reports have documented that enhanced T-bet expression and defective T_RM formation are often associated in various scenarios. CD8⁺ T cell priming in the absence of CD4 help,⁴² deficiency in cross-Priming DCs³⁷ or in infant animals and human¹⁴² all lead to increased T-bet expression and defective T_RM formation. The common factors controlling T-bet expression in above-mentioned settings remain unknown. In addition to TGF-β and IL-15, other signals that control the expression of T-bet and Eomes during T_RM differentiation remain to be discovered.

TGF-β signaling pathway is composed of a complicated network of molecular interactions.¹⁴³ Briefly, upon ligand binding, TGF-β receptor complex phosphorylates Smad2 and Smad3. Phosphorylated Smad2/3 associate with Smad4 and translocate to the nucleus. In addition to TGF-β, other members of TGF-β superfamily can also activate Smad (e.g., Activins and Bone Morphogenetic Proteins). For instance, Smad4-mediated Myc expression is essential for T cell homeostasis and function via a TGF-β-independent and presumably other TGF-β superfamily member(s)-dependent fashion.¹⁴⁴ Except for Smad4, other factors, such as Tripartite Motif Containing 33 (encoded by Trim33) can cooperate with phosphorylated Smad2/3 to initiate transcription.145 Further, TGF-β also activates non-Smad pathways including MAP kinase pathways, Rho-like GTPase and PI3K/Akt pathways.¹⁴⁶ Accumulating evidence suggest that TGF-β does not control T_RM differentiation via the canonical Smad4-dependent pathway. Indeed, Smad4 deficient and TGF-β unresponsive CD8⁺ T cells exhibit opposite phenotypes during T_RM differentiation. Smad4 deficient CD8⁺ T cells exhibit dramatically enhanced differentiation of CD103⁺T_RM cells comparing with TGF-β unresponsive CD8⁺ T cells that fail to up-regulate CD103 under similar settings.68 Similarly, TGF-β and Smad4 antagonize each other during Th17 differentiation in CD4⁺ T cells. TGF-β signaling reverses Smad4-mediated suppression of RORγt via a SKI-dependent mechanism.¹⁴⁷ Whether similar mechanisms are underlying TGF-β-induced T_RM differentiation remains to be demonstrated. In addition, TGF-β regulates several T cell-related target genes (e.g., Eomes and RORγt) in a Smad2/3-independent manner.^148,149 TGF-β represses the expression of Klf2 via the non-Smad PI3K/Akt pathway.46 However, direct genetic evidence to support a TGF-β-dependent and Smad-independent mechanism in T_RM differentiation remains to be established. Considering the importance of TGF-β in Th17 CD4⁺ T cell differentiation, it is interesting to determine the role of TGF-β in type 17 CD8⁺ T_RM cells and the interconnected regulation between TGF-β, T-bet, Eomes and RORγt. Together, considering the facts that Smad3 directly binds to Itgae locus, TGF-β may control the differentiation and homeostasis of T_RMs via both Smad2/3-dependent and Smad-independent pathways.

C. Blimp1 and Hobit

Homolog of B lymphocyte-induced maturation protein (Blimp1) in T cells (Hobit, encoded by Zfp683 or Znf683) and Blimp1 (encoded by Prdm 1) play a universal role in tissue residency of distinct immune cells.¹¹¹ In CD8⁺ T cells, Blimp1 is increased in effector cells and important for efficient effector function and terminal differentiation, whereas Hobit shows relatively low expression in effector and circulating memory CD8⁺ T cells.¹⁵⁰ In contrast, Hobit expression is specifically up-regulated in T_RM cells including CD8⁺, ILC1 and NKT cells from different anatomical sites. Transcriptional analysis indicates that both Blimp1 and Hobit deletion in CD8⁺ T cells re-activates genes associated with tissue egress including Ccr7, S1pr1 and Klf2, which in turn enhances the ability of T cells to exit from peripheral tissues and abolishes tissue residency. Deficiency in either Blimp1 or Hobit leads to partial reduction of CD8⁺ T_RM cells, while simultaneously abolishing both Blimp1 and Hobit near-completely inhibits the formation of CD8⁺ T_RM cells in diverse organs including the skin, liver, gut and kidney, suggesting the synergistic function of both TFs is required to establish tissue residency.¹¹¹

In contrast to the situation in mouse T_RM, the function of Hobit in human is less clear. The unique induction of Hobit in T_RM cells is only observed in mouse. Human T_EM cells express a significant level of Hobit. Therefore, even though most other T_RM core signature genes are highly conserved between mouse and human, the function of Hobit in human T_RMs may require additional evidence.^{70,123,151–153}

D. Runx3

Runx protein family (Runx1, 2 or 3) has a unique DNA-binding α subunit, which forms a complex with cofactor CBFβ (Core-binding factor subunit β) that stabilizes the Runx-DNA interaction.¹⁵⁴ As DNA-binding TFs, Runx proteins control thymocyte differentiation and determine the fate of CD4⁺ and CD8⁺ T cell lineages. Specifically, Runx3 represses TF Th-Pok (encoded by Zbtb7b), therefore inhibits CD4 lineage potential and contributes to the development of CD8⁺ single-Positive thymocytes.^155,156 Further, Runx3 induces the expression of CD103 in CD8⁺ single-Positive thymocytes. Synergizing with T-bet and Eomes, Runx3 is required to maintain the cytotoxicity program of activated CD8⁺ T cells via transcriptional regulation of key effectors including IFN-γ, perforin and granzyme B.^157,158 In addition, Runx3 deploys epigenetic marks (i.e., H3K27me3) to guard the fate of effector CD8⁺ T cells and prevent the expression of follicular helper T cell-related genes.¹⁵⁹ As to be discussed in the next section, a small population of CD4⁺ T cells acquire the expression of Runx3 and differentiate into CD8αα⁺ CD4⁺ T_RM cells in the gut IEL compartment.^160,161 More recently, using computational and pooled in vivo RNAi screens, Runx3 is reported to be a critical regulator in the establishment of T_RM cell populations in both non-barrier tissues (salivary gland and kidney) and barrier tissues (IEL, skin and lung parenchyma) even though the expression of Runx3 is not specifically induced in T_RM cells. Runx3 supports the expression of tissue-residency genes and represses genes associated with egress and recirculation. In addition, Runx3 is also a driver for both human and mouse CD8⁺ tumor-infiltrating lymphocytes (TILs) that exhibit characteristics of T_RM cells.¹¹²

E. Other TFs and Perspectives

Nur77 (encoded by Nr4a1), together with Nurr1 (encoded by Nr4a2) and NOR-1 (encoded by Nr4a3), constitute the NR4A subfamily of orphan nuclear receptors in the steroid thyroid receptor family.¹⁶² In the thymus, Nur77 controls CD8⁺ T cell development by suppressing the expression of Runx3.¹⁶³ In the periphery, as an immediate early response gene downstream of TCR signaling, Nur77 regulates CD8⁺ T cell expansion and effector function through transcriptional repression of Irf4. Lack of Nur77 leads to enhanced CD8⁺ T cell expansion, especially in KLRG-1⁺ terminally differentiated effector cells.¹⁶⁴ T cells from Nr4a1^−/− mice display reduced capacity to generate T_RM cells, suggesting the potential role of Nur77 in the generation and/or tissue residency of T_RM cells. Interestingly, at memory phase of influenza infection, lung and liver T_RM cells exhibit a 2 to 4 fold reduction while gut IEL T_RMs display a 90 fold decrease in the absence of Nur77 expression. The mechanisms leading to this dramatic tissue-specific requirement of Nur77 remain unclear.¹⁶⁵ Notably, under different infection settings, both lung and liver T_RMs, but not gut T_RMs require local antigen recognition. The connection between local TCR signal and the requirement of Nur77 in T_RMs remains to be visited in the future.

The expression of Notch and its down-streaming TF recombination signal binding protein for immunoglobulin kappa J (encoded by Rbpj) is enriched in human and mouse lung CD103⁺ T_RM cells. Simultaneous disruption of both Notch1 and Notch2 in T cells results in a two-fold reduction in CD103⁺ T_RM cells in the lung after influenza virus infection. Activation of Notch signaling pathway promotes the persistence of CD103⁺ T_RM cells via controlling metabolic programs.70 The involvement of Notch signals in T_RM cells isolated from other tissues is largely unknown.

Further, the targeting genes of hypoxia-inducible factor-1α (HIF-1α) are significantly enriched in human lung T_RM cells.70 Interestingly, oxygen-sensing prolyl-hydroxylase (PHD) proteins degrade HIF-1α via their enzymatic activity. Disruption of PHD proteins in T cells results in elevated HIF-1α expression and enhanced both CD4⁺ and CD8⁺ T cell response specifically in the lung, but not other tissues.¹⁶⁶

In addition, arylhydrocarbon receptor (AhR) is required for the maintenance of skin T_RM167 and γδ T cells in the epithelial surface and CD8αα⁺ αβ T cells in the IEL compartments of the gut.^168,169

TFs act as a link between signals from extrinsic microenvironment and intrinsic regulation of cellular response. Fluctuations of environmental cues, including cytokines, chemokines, pathogen insult and the persistence of microbiome can modulate the expression of disparate TFs. The cooperation among these TFs in turn instructs T cell differentiation and/or homeostasis. Considering the dramatically variable microenvironment inside different tissues, the knowledge of tissue-specific control of TFs will be essential to understand T_RM biology. After leaving the circulation, tissue-specific transcriptional reprogram represents a key step for T_RMs to adapt to the new environment and remains largely ill defined. Further, studies of TF cofactors or epigenetic regulators in T_RM formation are just in their infancy. These studies will facilitate our understanding of how TFs from the same family, such as T-bet and Eomes, Blimp1 and Hobit, perfo_RM both segregated and cooperative functions at the molecular level. In addition, these investigations will help to address the question that how numerous TFs function in a temporal-and spatial-dependent manner.

III. CD4⁺ T_RM cells

The vast majority of recent T_RM studies have been focused on CD8⁺ T cells. CD4⁺ T_RM cells represent a critical adaptive component of local immunity.¹⁷⁰ We will use the last section to summarize the recent findings about CD4⁺ T_RM cells in various tissues. We will not include Foxp3⁺ regulatory T cells in our discussion as recent reviews have covered the related findings.^171,172

A. Skin

Early research in mouse has demonstrated that after skin herpes viral infection, CD8⁺ T cells fo_RM a distinct population of T_RM cells in the epidermis while memory CD4⁺ T cells are largely located in the dermis and continue to recirculate.³² As unique structure components of the skin, hair follicles produce survival cytokines IL-7 and IL-15, which are essential for the maintenance of skin T cells.⁵³ As a consequence, skin CD4⁺ T cells are often clustered around hair follicles. Widely used T_RM markers CD69 and CD103 are generally believed to contribute to T_RM cell retention. Interestingly, even though a significant population of skin CD4⁺ T cells carry both CD69 and CD103, they reach equilibration with the circulation at steady state, which further questions the function of CD103 in T_RM cell biology. Local inflammation promotes CD8⁺ T cell-and CD11b⁺ myeloid cell-mediated recruitment and retention of skin CD4⁺ T cells.¹⁷³ Therefore, in the absence of local inflammation, skin CD4⁺ T cells are a component of circulating memory cells even with typical T_RM markers.

Different infection models can induce the formation of bona fide skin-resident CD4⁺ memory T cells. Skin infection of Candida albicans in C57BL/6 mice results in acute infection cleared in less than two weeks. Interestingly, C. albicans infection induces a distinct population of IL-17 producing and largely sessile CD69⁺CD4⁺ T_RM cells in the superficial layer of the dermis providing C. albicans-specific protection.¹⁷⁴ These IL-17⁺CD4⁺ T_RM cells are often co-localized with CD11c⁺ dendritic cells months after the clearance of the infection, suggesting a role for residual antigen or inflammatory cues for the retention of skin CD4⁺ T_RM population. Consistent with previous findings that many dermis CD4⁺ T cells are rapidly exchanging with the blood, a substantial subset of circulating CD4⁺ T cells is present in the deeper layer of the dermis. These mobile skin CD4⁺ T cells express low levels of CD69 and do not produce IL-17. These results suggest that different local environment within the dermis can support different subsets of memory CD4⁺ T cells, including both circulating and tissue-resident cells with distinct effector functions.

Another recent example of skin CD4⁺ T_RM comes from Leishmania major (L. major) infection model in C57BL/6 mice, which leads to prolonged skin lesion that lasts for 12 weeks. After the clearance of the infection, skin CD4⁺ T_RM cells can be identified at both infected and non-infected skin,¹⁷⁵ similar as CD8⁺ T_RM cell spreading to non-infected skin after repeated skin infections.³³ In contrast to skin CD4⁺ T cells residing in naïve mice,¹⁷³ L. major-specific CD4⁺ T_RM cells are not exchanging with circulating cells as demonstrated by grafting infected skin into naïve animals. Consistent with the dermal location of skin CD4⁺ T cells,³² L. major-specific CD4⁺ T_RM cells are sensitive to antibody-mediated depletion. Functionally, during high dose re-challenge, CD4⁺ T_RM -mediated and IFN-γ-dependent recruitment of circulating memory T cells are required for long-term protection.¹⁷⁵ However, during low dose re-challenge, CD4⁺ T_RM and T_RM-recruited inflammatory monocytes are sufficient to provide immediate protection without the contribution from circulating T cells.¹⁷⁶

Human skin harbors a large number of CD69⁺CD4⁺ T cells in the dermis, which is identified as the major T_RM population resistant to antibody-mediated depletion. Further, both CD103⁺CD4⁺ and CD103⁺CD8⁺ T cells are enriched in the epidermis of human skin.³¹ Keratinocyte derived TGF-β is likely involved in the induction of CD103⁺CD4⁺ T_RM cells in human. However, genetic evidence to support a role of TGF-β in skin CD4⁺ T_RMs in mouse is missing.

Strictly speaking, bona fide skin CD4⁺ T_RM cells are only formed under certain infection circumstances in mouse. Local inflammatory signals and cognate antigens may regulate the dynamic behavior of skin CD4⁺ T cells. CD103 is not a reliable marker for skin CD4⁺ T_RM in mouse and the definitive evidence supporting a role of TGF-β in skin CD4⁺ T_RM cell differentiation is lacking.

B. Lung

In both mouse and human lungs, distinct populations of CD4⁺ T cells carrying typical T_RM markers can be identified.^{27,28,67,152,177,178} Comparing with CD8⁺ lung T_RM, CD4⁺ T_RM cells usually carry less CD103 or express CD11a instead of CD103.^170,178,179 Genome wide transcriptional analysis reveals that CD4⁺ lung T_RM cells resemble CD8⁺ lung T_RM cells.^151,152,179 Similar transcription programs including the down-regulation of T-bet and Eomes, and the up-regulation of Blimp-1 and Notch signaling, direct the local differentiation of lung CD4⁺ T_RM cells.¹⁵² Interestingly, human and mouse infant T cells (including both CD4⁺ and CD8⁺ T cells) express enhanced levels of T-bet and exhibit defective lung T_RM cell formation, further suggesting that down-regulation of T-bet is a conserved common mechanism underlying both CD4⁺ and CD8⁺ T_RM differentiation.¹⁴²

Local signals that induce lung CD4⁺ T_RM cells are not entirely known. Similar as CD4⁺ T_RM clusters in the skin and vagina, clusters of lung CD4⁺ T_RM have been identified after influenza virus infection,¹⁷⁰ suggesting a common mechanism underlying CD4⁺ T_RM differentiation and/or maintenance. In contrast to lung CD8⁺ T_RM cells, one report has suggested that mouse lung CD4⁺ T_RM is TGF-β-independent, consistent with a CD103⁻ phenotype. Distinct from CD103⁺CD8⁺ T_RM, IL-15 is required during the early differentiation, but not the long-term maintenance of CD4⁺ lung T_RM cells in mouse.¹⁷⁹ Following both acute viral infection (a Th1 response) and brief allergy exposure-induced lung inflammation (a Th2 response), a clear population of lung-resident antigen-specific CD4⁺ T cells that is separated from bloodstream forms in an IL-2-dependent manner and play essential functions in local immunity.^180,181 In response to prolonged allergen exposure, a similar CD69⁺ and Th2-biased CD4⁺ T_RM population persists in the lung parenchyma.¹⁸² Even in the absence of circulating T cells, these CD4⁺ T_RMs are sufficient to mount a robust recall response. Similar autologous recall response has also been observed for helminth-induced lung CD4⁺ T_RMs after T cell migration has been blocked.¹⁸³ Together, similar as lung CD8⁺ T_RM cells, lung CD4⁺ T_RM cells are formed under various infectious and inflammatory settings. However, the cellular and molecular mechanisms underlying their differentiation and long-term maintenance remain to be demonstrated.

C. Other Non-lymphoid Tissues

Murine cytomegalovirus infection induces a distinct population of CD69⁺CD4⁺ T_RM cells in the salivary gland. In contrast to antigen-independent differentiation of CD8⁺ T_RM cells in the same tissue and same infection model, CD4⁺ T_RM formation requires local antigen in the salivary gland.¹¹⁴ The role of TGF-β in salivary gland CD4⁺ T_RM remains undetermined.

In both naïve and mucosal infected mice, CD69⁺CD103⁻CD4⁺ T_RM cells are identified in the white adipose tissue and provide potent protective response.¹³⁰ White adipose tissue CD4⁺ T_RM cells are isolated from circulation as demonstrated by parabiosis experiments. Further, they carry mucosal homing integrin α4β7.¹³⁰ Considering that TGF-β inhibits integrin α4β7 expression on effector CD8⁺ T cells,⁹⁰ it will be interesting to test the involvement of TGF-β signaling in white adipose tissue CD4⁺ T_RM.

Further, genital mucosa supports CD4⁺ T_RM cell differentiation after herpes simplex virus infection. Similar as skin and lung CD4⁺ T_RM cells, genital CD4 T_RM cells fo_RM clusters.¹³⁵ In contrast to the more mobile behavior of skin CD4⁺ T_RM cells isolated from mouse dermis, vaginal CD4⁺ T_RM cells are locally restricted and isolated from the circulation. Remarkably, even mucosal vaccination at a remote site (i.e., intra nasal priming) induces protective CD4⁺ T_RM formation at vaginal mucosa in a Chlamydia infection model. Similar as CD8⁺ T_RM cells, vaginal CD4⁺ T_RM differentiation is limited to the early stage of effector phase when a large number of activated T cells exit secondary lymphoid organs and migrate to peripheral mucosal sites.¹⁸⁴ The relationship between TGF-β and the differentiation of CD4⁺ T_RM in FRT remains unknown.

Comparing with CD8⁺ T_RMs, CD4⁺ T_RMs are a minor cell population within the IEL compartment of the small intestines in naïve specific pathogen free mice.²¹ Most of these unique CD4⁺ T_RM cells carry surface expression of CD 8αα. Both conventional effector and regulatory CD4⁺ T cells can differentiate into IEL CD8αα+ CD4⁺ T_RM cells in response to specific microbiota stimulation.¹⁸⁵ In addition, gut enriched local signals, such as TGF-β and retinoic acid are essential for the differentiation of CD8αα⁺CD4⁺ T_RM cells in the IEL compartment. The down-regulation of lineage specific TF Th-POK (conventional CD4⁺) or Foxp3 (Treg) and the up-regulation of T_RM and CD8⁺ related TF Runx3 are crucial for their differentiation.^160,161,186

Distinct from microbiota-induced CD8αα⁺CD4⁺ T_RM cells, listeria oral infection leads to the differentiation of CD4⁺ gut T_RM cells without the surface expression of CD 8aa. These CD4⁺ T_RM cells carry CD69 and reside in both IEL and LP compartments.¹⁸⁷ A small percentage of IEL CD4⁺ T_RM cells also express CD103. These gut CD4⁺ T_RMs provide essential Th1 response-mediated protection against re-infection and maintained in an IL-15-independent manner, similar as gut CD8⁺ T_RM cells. Intestine-restricted helminth infection induces a large population of Th2 CD4⁺ memory T cells in both gut LP compartment and peritoneal cavity. However, whether these protective Th2 memory CD4⁺ T cells are bona fide T_RMs remains elusive.¹⁸⁸

Together, consistent with the complexity and plasticity of effector CD4⁺ T cell differentiation programs, different local signals drive the differentiation of distinct CD4⁺ T_RM cells at the intestinal mucosal surface.

D. Secondary Lymphoid Tissues

A series of recent discoveries have demonstrated that a significant population (up to 50%) of CD44^hi effector/memory CD4⁺ T cells reside in secondary lymphoid organs (i.e., spleen, LNs and Peyer’s patches) without continuous recirculation for a prolonged period of time.^26,189–191 Similar as T_RM isolated from non-lymphoid organs, these secondary lymphoid organ CD4⁺ T_RM express higher levels of CD69 and lower levels of S1pr1 than their circulating counterparts. In contrast to the widely accepted notion that secondary lymphoid organs are mainly occupied by circulating T cells, antigen-specific CD4⁺ memory T cells residing in the draining lymph nodes are largely sessile after immunization.¹⁹¹ Further, similar CD4⁺ T_RM cells are present in the secondary lymphoid organs of naïve specific pathogen free mice presumably due to prolonged TCR stimulation by self-antigens and microbiome-derived antigens. The population of secondary lymphoid organ CD4⁺ T_RM expands with age.²⁶ Interestingly, a significant population of secondary lymphoid organ CD4⁺ T cells express T_RM marker CD69 in adult human.¹⁵¹ These secondary lymphoid organ T_RMs may represent a conserved phenomenon in CD4⁺ T cell biology although the functional importance and molecular control of these CD4⁺ T_RMs remain to be determined.

In summary, with unique tissue-specific features, CD4⁺ T_RM cells are present in a variety of lymphoid and non-lymphoid tissues, and play a non-redundant function in local immunity. Considering the complexity of effector CD4⁺ T cell lineages, the differentiation and maintenance of CD4⁺ T_RM cells may be controlled in a tissue-specific and inflammation-specific manner while preserve a common gene signature that restricting the recirculation. TGF-β controls the differentiation of almost every individual lineage of effector CD4⁺ T cells under certain conditions. However, the contribution of TGF-β signal to CD4⁺ T_RM cells remains largely unknown presumably due to the difficulties to dissect the roles of TGF-β in tissue residency from those in effector CD4⁺ lineage specification.

Conclusions

T_RM cells represent a major memory T cell population without continuous recirculation. Recent advances have established that common transcriptional and metabolic programs distinguish T_RM cells from circulating T cells and are closely associated with the behavior and function of T_RM cells. In various infection settings, it has been demonstrated that T_RM cells can function both as an alarming system and a self-sufficient defense system without significant contribution from circulating T cells, which further emphasize the importance of inducing desired T_RM populations in future vaccine design.

In addition to the common differentiation programs, to establish residency in different tissues, T_RM cells are required to adapt tissue-specific programs to accommodate unique local environmental cues. Recent investigations have accumulated evidence to support the paradigm that local signals, including cognate antigens, TGF-β, survival and inflammatory cytokines may impact T_RM cells in a tissue-specific and infection-specific manner. The mechanisms underlying tissue-specific T_RM regulation remain largely unknown. Even within a given tissue, the microenvironment is not homogenous. Thus, it is conceivable that the heterogeneity of T_RM population in a given tissue may be tightly linked with their functions. Together, studies on the intercellular and intracellular programs that are induced by local environmental signals may provide important information to deepen our understanding of T_RM biology and guide the development of TRM-focused future vaccine strategies.

Glossary

T_CM

Central memory T cell

DC

Dendritic cell

T_EM

Effector memory T cell

FRT

Female reproductive tract

IEL

Intraepithelial lymphocyte

LP

Lamina propria

LN

Lymph node

TCR

T cell receptor

T_RM

Tissue-resident memory T cell

TF

Transcription factor

TGF-β

Transforming growth factor-β

VACV

Vaccinia virus