Tissue-resident lymphocytes: from adaptive to innate immunity

Abstract

Efficient immune responses against invading pathogens often involve coordination between cells from both the innate and adaptive immune systems. For multiple decades, it has been believed that CD8⁺ memory T cells and natural killer (NK) cells constantly and uniformly recirculate. Only recently was the existence of noncirculating memory T and NK cells that remain resident in the peripheral tissues, termed tissue-resident memory T (T_RM) cells and tissue-resident NK (trNK) cells, observed in various organs owing to improved techniques. T_RM cells populate a wide range of peripheral organs, including the skin, sensory ganglia, gut, lungs, brain, salivary glands, female reproductive tract, and others. Recent findings have demonstrated the existence of T_RM in the secondary lymphoid organs (SLOs) as well, leading to revision of the classic theory that they exist only in peripheral organs. trNK cells have been identified in the uterus, skin, kidney, adipose tissue, and salivary glands. These tissue-resident lymphocytes do not recirculate in the blood or lymphatic system and often adopt a unique phenotype that is distinct from those of circulating immune cells. In this review, we will discuss the recent findings on the tissue residency of both innate and adaptive lymphocytes, with a particular focus on CD8⁺ memory T cells, and describe some advances regarding unconventional T cells (invariant NKT cells, mucosal-associated invariant T cells (MAIT), and γδ T cells) and the emerging family of trNK cells. Specifically, we will focus on the phenotypes and functions of these subsets and discuss their implications in anti-viral and anti-tumor immunity.

Introduction

The bridge between the innate and adaptive immune systems is critical for efficient immune responses against invading pathogens, and proper coordination and balance between various innate and adaptive cell types are necessary in the process. In the classic view, naive T cells are stimulated by their cognate antigens presented by antigen-presenting cells (APCs) and undergo clonal expansion and differentiate into effector T cells that migrate to the site of infection.¹ After elimination of the infection, a minority of these effector T cells remain alive and differentiate into memory T cells, which can be found in the blood, secondary lymphoid organs (SLOs), and tissues throughout the rest of the body.² Natural killer (NK) cells, on the other hand, are widely distributed throughout the body. These NK cells are now termed conventional NK (cNK) cells; they originate in the bone marrow and migrate through the circulation to different tissues.³ For many decades, it has been believed that CD8⁺ memory T cells and NK cells constantly and uniformly recirculate throughout the body.

In contrast to the classic view of lymphocyte recirculation, studies in the past decade have led to the characterization of lymphocyte populations permanently residing in nonlymphoid organs. These populations include tissue-resident memory T (T_RM) cells, unconventional T cells (invariant NKT (iNKT) cells, mucosal-associated invariant T (MAIT) cells, and γδ T cells), and the emerging family of tissue-resident NK (trNK) cells. Although these cells are different, they share a number of features pertaining to their tissue-resident functions.⁴ The existence of these populations has been proved in various settings due to improvements in technologies. The use of parabiosis, in which the circulatory systems of two animals are conjoined for an extended period of time, is the most direct way to demonstrate tissue residency. Recirculating cells will migrate between conjoined animals and establish equilibrium between the two, while tissue-resident cells will not, demonstrating a lack of recirculation.⁵ Other techniques used to prove tissue residency include tissue transplantation and the blockade of cell recirculation.

T_RM cells have been found in the skin, sensory ganglia, gut, lungs, brain, salivary glands, female reproductive tract, and other sites, while trNK cells have been identified in the uterus, skin, kidney, adipose tissue and salivary glands. These tissue-resident lymphocytes do not recirculate in the blood or lymphatic system and often adopt unique phenotypes and functions distinct from those of circulating lymphocytes.⁶ In this review, we will discuss the recent findings on the tissue residency of both innate and adaptive lymphocytes, with a particular focus on T_RM cells, and provide some description of advances regarding unconventional T cells (iNKT cells, MAIT cells, and γδ T cells) and the emerging family of trNK cells. As most findings on T_RM cells have been made regarding CD8⁺ T_RM cells, we use the term “T_RM cells” to refer to CD8⁺ T_RM cells throughout the review.

Tissue-resident memory CD8⁺ T cells

Naive T cells migrate between blood and SLOs such as the spleen, lymph nodes (LNs), and mucosal-immune system (MIS) to search for their cognate antigens. They activate and proliferate extensively once they encounter cognate antigens presented by APCs. These activated cells differentiate into effector T cells capable of cytokine production and cytolytic activity, and they then migrate to the site of infection.¹ When the infection is successfully eliminated, the majority of effector T cells die through apoptosis during the contraction phase, and the remaining cells differentiate into memory T cells, which can be found in the blood, SLOs, and tissues throughout the rest of the body.² These memory T cells rapidly upregulate their effector functions and respond quickly upon rechallenge, thereby protecting against secondary infections outside of SLOs due to their broad distribution and high clonal frequency.⁷

Memory T cells play an important role in adaptive immunity. Researchers have defined the existence of two subsets of memory T cells based on their homing molecules and recirculation patterns.⁸ One of these subsets is the central memory T cells (T_CM), which express the SLO homing molecules CCR7 and CD62L, primarily recirculate between SLOs and blood and rapidly proliferate upon antigen restimulation.^9,10 The other subset is the effector memory T cells (T_EM), which, unlike T_CM, lack expression of SLO homing molecules and instead express higher levels of integrins and chemokine receptors that indicate the potential to migrate between nonlymphoid tissues.⁹ T_EM express high levels of effector molecules and provide immediate effector functions when monitoring inflamed peripheral sites.

Due to technical shortcomings, the classic theory that divides memory T cells into two subsets has been assumed correct for a very long time, and it has always been believed that memory T cells constantly and uniformly recirculate between peripheral tissues and blood. However, significant revisions of this long-standing concept have been made thanks to improved techniques such as parabiosis, tissue transplantation, and blockade of T cell recirculation. The existence of noncirculating memory T cells that remain resident in the peripheral tissues, termed T_RM cells, was defined in the skin and sensory ganglia after acute infection with herpes simplex virus (HSV), in the small intestine epithelium after lymphocytic choriomeningitis virus (LCMV) infection, and in the skin after vaccinia virus (VacV) infection.^11–13 The work of these two laboratories has transformed the classic knowledge of memory T cell subsets in tissues and led to the discovery of T_RM cells in many different organs.¹⁴ The T_RM subset has become the cutting-edge population and the spotlight of research on T cells.

Signature phenotypes of CD8⁺ T_RM cells in various tissues

T_RM cells are both phenotypically and transcriptionally unique. They populate a wide range of peripheral organs, including the skin, sensory ganglia, gut, lungs, brain, salivary glands, female reproductive tract, and other tissues. Although migration is the main determinant of T_RM cells, they are also identified by elevated expression of CD69 and the integrin CD103.⁷ Coexpression of CD69 and CD103 is observed on the majority of T_RM cells, however, some fall into other phenotypic categories.¹⁴ For example, although CD103⁺ T_RM cells are generally found in the small intestine and gut,^15–20 oral infection with Yersinia pseudotuberculosis results in the development of both CD103⁺ and CD103⁻ T_RM cells in the lamia propria.^21,22 Brain T_RM cells express CD103 in models of intranasal (i.n.) infection with vesicular stomatitis virus (VSV),²³ intracranial infection with Theiler’s murine encephalomyelitis virus (TMEV),²⁴ and toxoplasma infection,²⁵ but lack CD103 expression when elicited by LCMV infection. Both CD103⁺ and CD103⁻ T_RM cells can be found in the kidney.^17,26

Previous findings have suggested a lack of CD103 expression in several organs such as the spleen,^16,17,23,27 LNs^16,23,27, and liver.^28–30 However, contrary recent findings have identified T_RM cells within SLOs and suggested that restimulation of nonlymphoid memory CD8⁺ T cells within the skin or mucosa can give rise to bona fide T_RM cells specifically within draining LNs.³¹ In addition, populations of CD69⁺CD103⁺CXCR6⁺ T_RM cells expressing low levels of sphingosine-1-phosphate receptor 1 (S1PR1) and KLF2, and CD69⁺CD103⁺CXCR6⁺CXCR3⁺ T_RM cells with a T-bet^loEomes^loBlimp-1^hiHobit^lo phenotype have been identified in the liver.^32,33 CD103⁺ T_RM cells are often present in the sensory ganglia,¹¹ skin,^{11,13,15,16,34} lungs,^15,35–37 salivary gland,^16,17,38 reproductive tract,^17,39 thymus⁴⁰ and pancreas,^17,41 and a minimal number of CD103⁺ T_RM cells can also be found in the heart.¹⁷

Of note, intravascular staining has enabled the discovery that 95% of CD69⁺CD103⁺ T_RM cells isolated from the mouse lung via standard methods are confined to the pulmonary vasculature instead of the lung tissue, suggesting an overestimate of the T_RM population within the lung.²⁸ On the other hand, parabiosis and quantitative immunofluorescence microscopy studies have revealed that conventional isolation methods and identification of T_RM cells by phenotypic markers underestimate the size of the T_RM population.⁴² In addition, different phenotypic markers and subsets have been identified in different organs. For example, CCR8 has been proposed as a marker for long-lived memory T cells in human skin, and CCR8⁺ T cells bear all the hallmarks of T_RM cells, including surface expression of both CD103 and CD69.⁴³ CD49a has recently been identified as a marker to differentiate CD8⁺ T_RM cells into CD49a⁺ T_RM cells that produce IFN-γ and CD49a⁻ T_RM cells that produce IL-17 in human skin epithelia.⁴⁴ CD8⁺CD28⁻ T_RM cells have been identified in the skin lesions of patients in the early stages of systemic sclerosis.⁴⁵ CD69⁺S1PR1⁻ T_RM cells were found in the nasal polyps of patients with chronic rhinosinusitis.⁴⁶ CD103⁺CD39⁺ tumor-infiltrating CD8⁺ T cells have been found for six different malignancies including melanoma, lung cancer, head and neck squamous cell carcinoma (HNSCC), ovarian cancer, and rectal cancer,⁴⁷ and high numbers of perforin-expressing T_RM cells are found in patients with urothelial urinary bladder cancer (UBC).⁴⁸

These findings indicate that (1) there may be substantial phenotypic heterogeneity among T_RM cell subsets; (2) the tissue microenvironment and infection route can contribute to the regulation of phenotype and function of these cells; (3) phenotypic markers of T_RM cells may differ between humans and mice; and (4) limitations in experimental techniques may confound our knowledge of T_RM populations; cytometric identification based on phenotypic markers sometimes cannot represent a whole population. Therefore, identification of T_RM cells based solely on the coexpression of CD69 and CD103 may not reliably identify all T_RM cells. Distinct phenotypic markers should be used in studying T_RM cells of different tissues, and lack of recirculation and using imaging techniques should be helpful methods in defining the T_RM population.

Differentiation and maintenance of CD8⁺ T_RM cells

A large number of effector T cells enter the site of infection to clear pathogens during the acute phase of an immune response. Killer cell lectin-like receptor G1 (KLRG1) is induced on antigen-stimulated effector CD8⁺ T cells and has been considered as a marker of senescence. KLRG1⁺ cells receiving intermediate amounts of activating and inflammatory signals downregulate KLRG1 during the contraction phase in a Bach2-dependent manner,⁴⁹ and the KLRG1⁻ effector population contains T_RM precursor cells. KLRG1⁻ T_RM precursor cells stop migrating immediately after pathogen clearance and begin to differentiate into T_RM cells with phenotypic markers of residence, often as a result of the differential effects of stimulation by transcription factors, cytokines, chemokines and cognate antigen.^12,15,16,50

The majority of T_RM cells express CD69 and CD103. The integrin CD103 binds to E-cadherin, and this interaction promotes the retention of T_RM cells within certain epithelial tissues.³⁴ Deficiency in CD103 expression results in impaired tissue retention of T_RM cells in a variety of peripheral tissues.^{15,17,18,23,51,52} However, the necessity of CD103 for T cell residence in many organs remains unclear.^15,17,51,53 Notably, CD103⁻ T_RM cells do not rely on CD103 for their retention in specific tissues. CD69 is also important for T_RM differentiation; it blocks the functional activity of S1PR1 by interfering with its cell surface expression and therefore resists S1P-mediated signals that enable tissue egress.^16,38,54 Other molecules also participate in the differentiation and accumulation of T_RM cells. For example, lung antigen-specific CD8⁺ T cells express both the very late antigen-1 (VLA-1) and CD103 after respiratory mucosal immunization⁵⁵; the tumor necrosis factor (TNF) superfamily molecule LIGHT can promote the generation of lung-resident CD8⁺ T cells after an acute respiratory virus infection⁵⁶; and the immune checkpoint ligand B7-H1 is essential for the maintenance and accumulation of virus-specific T_RM cells in the central nervous system (CNS): Long-term maintenance of T_RM cells is diminished in B7-H1 knockout mice.²⁴

Transcriptional factors involved in the regulation of T_RM differentiation include Hobit, Blimp-1, KLF2, Eomes and T-bet.^15,57,58 In addition, the transcription factor Runx3 was recently identified as a key regulator of T_RM cell differentiation that supports the expression of tissue-residency genes and suppresses genes associated with tissue egress and recirculation.⁵⁹

The cytokine TGF-β promotes the expression of CD103 in barrier tissues,^17,60 and a deficiency in TGF-βR on CD8⁺ T cells consequently leads to the failure of T cells to differentiate into T_RM cells.^34,38,61 Interestingly, TGF-β induces liver-adapted T_RM cells and controls the formation of CD69⁺CD103⁻ T_RM cells in the kidney.^26,33 Additionally, IL-15 drives the differentiation of T_RM in some but not all tissues; the absence of TGF-β and IL-15 signaling results in impaired differentiation of CD103⁺ T_RM cells and the loss of this population from the skin over time.^15,58,62,63 Sequential signaling by IL-15 followed by TGF-β also induces liver-adapted T_RM cells.³³ New findings have also suggested the importance of pro-inflammatory cytokines, such as IFN-γ and IL-12, on the differentiation of T_RM populations; impaired IFN-γ or IL-12 signaling results in defective differentiation of CD103⁻CD69⁺ T_RM cells and reduces T_RM persistence in the intestine.⁶⁴

In addition to cytokines, chemokines are also important in the trafficking and localization of T_RM cells. For example, the chemokine receptors CXCR6 and CCR10 are required for optimal formation of a T_RM population in the skin; lack of either CXCL10 or CXCR3 compromises the mobilization of T_RM cells within latently infected trigeminal ganglia and results in the absence of signals required for differentiation^15,65,66; CXCR3 is also critical for T cell accumulation in uninfected salivary glands,⁶⁷ and CXCL17 is required for the mobilization of T_RM cells in the vaginal mucosa (VM).⁶⁸

Differentiation of recirculating T_EM cells depends on prolonged cognate antigen stimulation. However, antigen stimulation is not always required in the context of T_RM cells, and whether it is required for T_RM cell differentiation depends on the tissue of residence. For example, antigen recognition is required for brain T_RM cells to differentiate²³; is critical for T_RM formation in the skin after infections with VacV-expressing model antigens⁶⁹; and is essential for T_RM development in the lungs and airways after infection with respiratory syncytial virus (RSV).⁷⁰ On the contrary, long-lived intraepithelial CD103⁺ T_RM cells can be generated in the absence of antigen recognition in the skin and mucosa in HSV and vaginal challenge models¹⁶; prolonged cognate antigen stimulation is dispensable for intestinal T_RM ontogeny¹⁷; and noninflammatory vaccination enables the establishment of T_RM cells in the female genital tract.⁶¹ Therefore, differentiation and maintenance of T_RM cells differ from classic models of those processes for T_EM cells, and the local microenvironment drives the phenotypic diversity of T_RM cells.

Interestingly, although T_RM cells have not been identified in SLOs in past studies, a new study using parabiosis of “dirty” mice has demonstrated that restimulation of nonlymphoid memory CD8⁺ T cells within the skin or mucosa results in the accumulation of T_RM cells within draining LNs, indicating that nonlymphoid cells can give rise to SLO T_RM cells.³¹ Secondary skin T_RM cells can also be formed from pre-existing T_RM cells, as well as from precursors recruited from the circulation, without displacement of the pre-existing T_RM cell pool.⁷¹ Reactivation of mucosal T_RM cells in the reproductive tracts triggers the recruitment of recirculating memory T cells that differentiate independently of antigen stimulation and contribute substantially to the boosted secondary T_RM cell population.⁷²

The length of T_RM persistence varies between different tissues. For example, CD103⁺CD69⁺ T_RM cells in nasal tissues are relatively stable for at least three months after a total respiratory tract (TRT) infection, while the number significantly declines in the lung.⁷³ In contrast, skin T_RM cells displace epidermal niches and can stay there for up to a year after HSV infection.⁷⁴ Therefore, the persistence of T_RM cells in different microenvironments is dependent on multiple tissue-specific factors.⁷⁵ In addition, survival signals are essential for T_RM cells to repopulate and maintain their niche. For example, T_RM cells require IL-15 to support their maintenance after HSV-1 infection,¹⁵ while they rely on the aryl hydrocarbon receptor (AhR) for long-term persistence in the epidermis.⁷⁴ Loss of these signals results in the loss of T_RM cells. On the other hand, a recent study by Takamura et al. has revealed that the progressive loss of temporarily created “spaces” (RAMDs) over time may be the reason for the shorter lifespan of lung T_RM cells.⁷⁶

In conclusion, the differentiation and maintenance of T_RM cells are tissue-specific and depend on a variety of factors including but not limited to: (1) certain molecules that are essential for the differentiation of T_RM cells; (2) transcription factors that regulate differentiation of T_RM cells; (3) tissue-specific environmental cues such as cytokines, chemokines, antigen stimulation, etc. that are involved in the differentiation of T_RM cells; (4) nonlymphoid cells that contribute to the accumulation of T_RM cells; and (5) pre-existing T_RM cells or recirculating memory T cells that give rise to secondary T_RM cells (Fig. 1). The fates of T_RM cells (whether they become exhausted, migrate to other regions, transform into other cells (lose phenotypic markers, etc.), or are reactivated) and the reasons they undergo a specific fate merit further research.

Fig. 1.

Differentiation, maintenance, and function of T_RM cells. Naïve T cells are activated and transform into KLRG1⁻T_RM precursor cells. KLRG1⁻ T_RM precursor cells stop migrating immediately after pathogen clearance and begin to differentiate into T_RM cells with phenotypic markers of residence, which often results from the differential effects of transcription factors, cytokines, chemokines, and cognate antigen stimulation. Recirculating memory T cells and pre-existing T_RM cells give rise to secondary T_RM cells upon restimulation. Activated T_RM cells are involved in a variety of effector functions, including cytolytic activity, the secretion of proinflammatory cytokines such as IFN-γ and TNF-α, and the recruitment of other adaptive and innate immune cells

CD8⁺ T_RM cells in anti-viral immunity

T_RM cells, as a subset of memory T cells that reside permanently in the peripheral tissues, are believed to provide stronger protective immune responses than circulating memory T cells do.⁷ T_RM cells provide frontline defense against invading pathogens due to their massive presence in the barrier tissues. The function and role of T_RM cells in the control of viral infections have been demonstrated in numerous animal infection models and human patients. Here, we review the most recent evidence of T_RM-mediated viral control in different tissues.

Due to their tissue-specific locations, T_RM cells provide local protection immediately upon antigen re-encounter and promote rapid immune responses.⁷⁷ This property of T_RM cells has been demonstrated in various animal infection models and even in human patients.^{11,38,67,68,78–81} An increased number of CXCR8⁺ T_RM cells has been detected in the VM of mice after intravaginal HSV-1 infection, and these cells contribute to the clearance of viral infection.⁶⁸ Defective accumulation of T_RM cells in the VM results in more virus replication and a reduced number of functional CD8⁺ T cells in the local tissue.⁶⁸ Significant increases in both the number and function of HSV-specific CXCR3⁺ T_RM cells have been detected in the trigeminal ganglia of mice following UV-B-induced HSV-1 reactivation, which protects the host from recurrent HSV infection, and a lack of T_RM is again associated with recurrent ocular HSV infection.⁶⁵ In addition, accumulation of T_RM cells in the skin provides enhanced control against viral infection with HSV-1.^11,62,82 T_RM cells generated by the “prime and pull” protocol may reduce the spread of infectious HSV-2 into the sensory neurons and prevent development of clinical disease.⁶¹

Accumulation of T_RM cells has been established in the brain of newborn mice after intraperitoneal mouse cytomegalovirus (MCMV) infection, which provides protection against primary MCMV infection and reduce brain pathology.⁸³ Depletion of these cells results in virus reactivation and enhanced inflammation in the brain.⁸³ Models using LCMV have revealed the protective role of brain-resident T_RM cells in restricting viral infection in the CNS.⁸⁴ T_RM cells are generated and maintained in the CNS tissues after intracranial infection with TMEV, and a lack of this cell population results in compromised control of heterologous virus rechallenge.²⁴ In addition, T_RM cells accumulate within the brain after chronic Toxoplasma gondii infection and contribute to parasite control within the CNS.²⁵

In the respiratory tract, T_RM cells may confer protective immunity against viruses such as RSV,^80,85 Sendai virus,⁸⁶ and influenza.^36,73,87,88 High levels of lung-resident T_RM cells are induced by administration of the mucosal Sendai virus-engineered recombinant anti-TB vaccine (SeV85AB), which leads to a rapid and strong recall response against Mtb challenge infection.⁸⁹ Generation of T_RM cells by intranasal vaccination of RSV antigen-expressing MCMV results in earlier T cell responses and viral clearance after RSV challenge.⁹⁰

Human immunodeficiency virus (HIV)-infected women display a high frequency of CD103⁻CD8⁺ T_RM cells residing close to the epithelial basal membrane; accumulation of this subset is associated with HIV viral load.⁹¹ HIV-specific T_RM cells are established in the VM and are capable of initiating a tissue-wide immune response after combined intranasal and intravaginal mucosal immunization with recombinant influenza-HIV vectors.⁸⁸

Upon activation of T_RM cells, they can rapidly secrete a number of proinflammatory cytokines, such as IFN-γ, TNF-α and IL-12 (Fig. 1).^81,86,92,93 T_RM cells generated by intranasal administration of RSV antigen-expressing MCMV secrete IFN-γ after RSV challenge.⁹⁰ In human skin, CD8⁺CD49a⁺ T_RM cells produce IFN-γ, whereas CD8⁺CD49a⁻ T_RM cells produce IL-17, which promotes local inflammation in the skin.⁴⁴ CD103⁺ T_RM cells produce both IFN-γ and TNF-α after chronic Toxoplasma gondii infection.²⁵ In addition, allergen-induced epidermal accumulation of IL-17A-producing and IFN-γ-producing skin-resident T_RM cells correlates with the magnitude of the challenge response.⁹⁴

These mediators may drive the local activation and recruitment of both innate and adaptive immune cells (Fig. 1). For example, secretion of IFN-γ by T_RM cells in the female mouse reproductive tract leads to the recruitment of immune cells including T cells, B cells and NK cells, while secretion of TNF-α by T_RM cells leads to the maturation of APCs.⁹³ Alternatively, activated T_RM cells in the mouse skin can alter the local tissue environment by secreting cytokines, which leads to the induction of an IFN-γ-dependent antiviral program.⁹² The activation of HIV-specific CD8⁺ T_RM cells results in the recruitment of both adaptive and innate immune cells in the VM.⁸⁸ On the other hand, T_RM cells can lyse infected target cells directly (Fig. 1). For example, CD8⁺CD49a⁺ T_RM cells in human skin may induce the expression of perforin and granzyme B upon stimulation with IL-15, which in turn promotes a strong cytotoxic response.⁴⁴ In general, T_RM cells play a protective role in viral immunity, in which they provide stronger and faster immune responses against invading pathogens. Defective accumulation of T_RM cells in local tissues often results in impaired immune responses, elevated viral load and recurrent infection. T_RM cells confer their effector functions through the secretion of proinflammatory cytokines, the recruitment of other immune cells, and the enhancement of cytotoxic responses against target cells (Fig. 1).

CD8⁺ T_RM cells in anti-tumor immunity

CD8⁺ T cells are extremely important in protection against tumor cells. However, in order for CD8⁺ T cells to fight against tumor cells, they must first migrate into the local tumor microenvironment to respond to tumor antigens. Although the role of T_RM cells in anti-viral immune responses has been extensively studied over the last decade, the role of T_RM cells in anti-tumor immune responses has yet to be fully discerned. Recent findings have revealed an accumulation of CD103⁺ T_RM cells in several human solid tumors including ovarian,^95–97 breast,⁹⁸ lung,^99–102 liver,¹⁰³ and urinary bladder⁴⁸ cancers. The abundance of these T_RM cells has been associated with prolonged survival and better prognosis in patients with pulmonary squamous cell carcinoma (pSCC),¹⁰¹ hepatocellular carcinoma (HCC),¹⁰³ triple-negative breast cancer (TNBC),¹⁰⁴ recurrent laryngeal squamous cell carcinoma (LSCC),¹⁰⁵ or HNSCC.⁴⁷ In addition, a high number of T_RM cells infiltrating the tumors is associated with lower tumor stage in patients with UBC.⁴⁸ In contrast, the accumulation of T_RM cells in melanoma tumors has not been associated with prolonged survival or a better prognosis, suggesting that T_RM cells are associated with a better prognosis in some cancers but not in others^106,107; T_RM cells may have different local anti-tumor responses in different malignancies.

The functions of CD103⁺ T_RM cells remain to be explored. Several studies have proposed a protective role of these T_RM cells in the tumor microenvironment. CD103⁺CD39⁺ CD8⁺ tumor-infiltrating lymphocytes (TILs) from HNSCC patients kill autologous tumor cells in an MHC class I-dependent manner.⁴⁷ Binding of the adhesion-associated protein paxillin (Pxn) to the subunit tail of CD103 expressed on tumor-specific cytotoxic T lymphocyte (CTL) clones may alter both adhesion and spreading of freshly isolated CD8⁺CD103⁺ lung TILs and CD103⁺ tumor-specific CTL clones and severely compromise the functions of these CTL clones against autologous tumor cells.¹⁰⁰ Skin T_RM cells generated as a result of autoimmune vitiligo produce IFN-γ and are critical for protection against melanoma rechallenge.⁵² T_RM cells generated by OVA-encoding VacV may also provide effective anti-tumor immunity against OVA-expressing melanoma.¹⁰⁸ In addition, CTLs from CD103^hi tumors in lung cancer display features of enhanced cytotoxicity,⁹⁹ and CD103⁺CD8⁺ TILs from nonsmall-cell lung carcinoma (NSCLC) patients display increased activation-induced cell death and specific cytolytic activity toward autologous tumor cells.¹⁰²

One of the major functions of T_RM cells is the recruitment of other immune cells from the circulation, the cooperation of T_RM and recirculating T_CM cells has been identified in a mouse model of vitiligo in which they work together to maintain disease.¹⁰⁹ Thus, it is reasonable that T_RM cells may confer protective functions against tumor cells through the recruitment of circulating tumor-specific T cells. Of note, circulating CD8⁺ T cells and CD8⁺ T_RM cells cooperate in anti-tumor immunity, and circulating CD8⁺ T cells may adopt a T_RM phenotype within the tumor and reside in the local tissue after tumor elimination.¹⁰⁸

Two recent studies have pointed out that local proliferation of pre-existing T_RM cells gives rise to secondary T_RM cells upon rechallenge, and these newly recruited antigen-specific or bystander T_RM cells are generated without replacement of the pre-existing T_RM cells.^71,72 A study of breast cancer also demonstrated that newly arrived T_RM cells in the tumor are functional, while T_RM cells established previously are dysfunctional; the persistence of these tumor-active T_RM cells in the tumor site does not depend on antigen stimulation but is sustained by tumor-associated macrophage (TAM)-derived IL-15.¹¹⁰ These findings suggest that irrelevant cells may be present in the tumor and can compete with functional newly arrived T_RM cells for cytokine resources.

On the other hand, immune checkpoint molecules are particularly enriched within T cells with phenotypic and genomic features of T_RM cells in tumors.¹⁰⁷ Some CD103⁺CD8⁺ TILs isolated from NSCLC patients have transcriptomic and phenotypic signatures of T_RM cells and frequently express PD-1 and Tim-3.¹⁰² CD103⁺ TILs within the tumor epithelium in ovarian cancer also express PD-1, LAG-3 and Tim-3.⁹⁷ PD-1 is additionally overexpressed in patients with urinary bladder cancer, HBV-related liver cancer, head and neck cancer, or breast cancer.^{47,48,103,104} These findings suggest that the T_RM subset of TILs may adopt an exhausted phenotype and may be a major target of immune checkpoint blockade. Furthermore, TILs secrete less IL-2, IFN-γ and TNF-α compared with circulating counterparts.¹⁰⁶ A study by Gabriely et al. has suggested that tumor-associated CD103⁺ CD8 T cells have regulatory properties, demonstrated by increased expression of CTLA-4 and IL-10, and they also have protumorigenic properties: adoptive transfer of CD103⁺ CD8 T cells promotes tumor growth.¹¹¹ Therefore, the role of T_RM cells in anti-tumor immunity is still debated, as they are protective in certain cancers but not in others. Further knowledge on the generation of appropriate T_RM cells to act against tumor cells is essential for improved tumor immunotherapy.

Tissue-resident unconventional T cells

In contrast to the “conventional” CD8⁺ T_RM cells, “unconventional” or “innate-like” T cells expressing T cell receptors (TCRs) with limited diversity, such as αβTCR-expressing iNKT cells, MAIT cells, and γδ T cells, have also been shown to be tissue-resident.⁴ These lymphocytes often recognize nonclassical and nonpolymorphic major histocompatibility complex (MHC)-like molecules, or MHC-unrelated presenting molecules.⁴

iNKT cells are lipid-sensing innate T cells expressing a semi-invariant αβTCR that only recognizes glycolipid antigens presented by the MHC class I-like molecule CD1d.^112,113 These cells often express the transcription factor promyelocytic leukemia zinc finger (PLZF) and play a role in tumor surveillance and control of certain viral and bacterial infections.^5,114 Three subsets of iNKT cells, termed NKT1, NKT2 and NKT17, have been described. These subsets express distinct transcriptional factors and corresponding cytokines and have been shown to localize to different tissues.^115–121 The long-lived thymus-resident population of mature NKT cells is capable of rapid and prolonged production of IFN-γ and IL-4.¹²² Liver-resident NKT cells are retained locally through constitutive LFA-1-intercellular adhesion molecule (ICAM)-1 interaction induced by PLZF.¹²³ However, iNKT cells accumulating in adipose tissue lack PLZF but express the transcriptional factor E4BP4,¹²⁴ they produce IL-2 and IL-10 and control the homeostasis of Treg cells and macrophages in this tissue.¹²⁴ Of note, iNKT cells also express CD69, a hallmark of T_RM cells.⁵

MAIT cells are very abundant in humans and exhibit innate-like functions similar to those described for iNKT cells. Like iNKT cells, they also express a semi-invariant αβTCR that only recognizes bacterial metabolites derived from the synthesis of vitamin B presented by the MHC-related protein 1 (MR1).^4,125,126 In addition, MAIT cells also respond quite sensitively to non-TCR signals, such as inflammatory cytokines including IL-7, IL-12, IL-15, IL-18, and IFN-α/β.¹²⁷ Interestingly, MAIT cells confer a robust IFN-γ and granzyme B response to inflammatory signals but have limited responsiveness when stimulated directly by their TCR.¹²⁸ MAIT cells are often marked by high expression of the C-type lectin CD161^129,130 and expression of the transcription factors PLZF and RORγt.^131,132 They are especially enriched in mucosal tissues, including the lung, liver and intestinal tract¹³³ and are also abundant in peripheral blood where they coexpress CD161 and CD26.¹³⁴ MAIT cells exhibit tissue homing properties and produce inflammatory cytokines.¹³⁵ Human MAIT cells display a chemokine receptor expression pattern (CCR9^intCCR7⁻CCR5^hiCXCR6^hiCCR6^hi) that indicates preferential homing to the tissues, such as the intestine and liver. They produce IFN-γ and granzyme B as well as high levels of IL-17 after phorbol myristate acetate (PMA) and ionomycin stimulation.¹³⁶ IL-12 and IL-18 activate liver-resident MAIT cells to produce a substantial amount of IFN-γ.¹³⁷ MAIT cells have been found within primary and metastatic tumors. However, whether they play an aggravating role in malignancies or contribute to anti-cancer immunity is still unclear.¹²⁷

T cells expressing the γδTCR represent another innate-like T cell subset that recognizes conserved nonpeptide antigens.¹³⁸ γδ T cells are greatly enriched in mucosal and epithelial sites, such as the skin, respiratory, digestive and reproductive tracts. They also comprise a small proportion (1–5%) of the circulating lymphocytes in the peripheral blood. γδ T cells derive from thymic precursors and migrate into tissues early during development where they then persist as tissue-resident cells.^139,140 These cells frequently express tissue-specific TCRs, which are invariant or closely related, resulting in distinguished roles for γδ T cells in different tissues.^140,141 Similar to MAIT cells, IFN-γ and IL-17 are produced by different γδ T cell subsets.^142,143 Dermal γδ T cells express the tissue-resident T cell markers CD69 and CD103 and additionally bear skin homing receptors and produce IL-17 and IL-22.¹⁴⁴ CD69⁺CD103⁺ tissue-resident γδ T cells are expanded in the lungs of mice reinfected with B. pertussis and produce significantly more IL-17 than γδ T cells from infected unprimed mice.¹⁴⁵ Resident memory γδ T cells in the LNs also produce IL-17A.¹⁴⁶ γδ T cells contribute to the protective immunity against pathogens, tumor surveillance, and regulation of the innate and adaptive immune responses.¹⁴⁷ They are involved in various diseases, including viral and microbial infections, autoimmune diseases and cancer.^140,148

The emerging family of Tissue-Resident NK cells

Tissue residency is a hallmark of the innate lymphoid cell (ILC) family. ILCs can be subdivided into three groups based on differential expression of phenotypic markers and transcription factors and production of different cytokines. Group 1 ILC (ILC1) cells, including trNK cells, are directed by the transcription factor T-bet and produce IFN-γ; group 2 ILC (ILC2) cells are directed by the transcription factor GATA-3 and produce IL-15 and IL-13; group 3 ILC (ILC3) cells are directed by the transcription factor RORγt and produce IL-17. Interestingly, trNK cells have been distinguished from the cNK cells and are considered the “innate counterparts” of T_RM cells.^5,149,150

The new era of trNK cell biology began with the discovery of CD49a⁺ liver-resident NK cells in a contact hypersensitivity (CHS) model through a parabiosis study.^3,151,152 Two distinct subsets of murine NK cells were identified in this study: CD49a⁻DX5⁺ cNK cells that circulate in the blood, and CD49a⁺DX5⁻ trNK cells that remain resident in the liver.^151,153 CD49a⁺ trNK cells reside in the liver sinusoidal blood, they possess memory potential and confer hapten-specific CHS responses upon hapten challenge.¹⁵¹ In addition to CD49a, liver-resident NK cells express higher levels of CXCR6, CXCR3, CD69 and the TNF-related apoptosis-inducing ligand (TRAIL),¹⁵⁴ and they strictly require the transcription factors T-bet, Hobit and PLZF for their development but are independent of Eomes.¹⁵⁴ A recent study also demonstrated the requirement for AhR in the maintenance of liver-resident CD49a⁺TRAIL⁺CXCR6⁺DX5⁻ NK cells and their hapten memory function.¹⁵⁵ Hepatic CD49a⁺ NK cells are induced by culturing cells with IL-2, IL-12, IL-15, IL-18, or the cytokine cocktail (IL-2/IL-12/IL-15/IL-18) and may produce high quantities of IFN-γ and TNF-α.¹⁵⁶ These cells degranulate less efficiently than cNK cells, however, due to their higher expression of TRAIL, they are capable of inducing cell death in TRAIL-sensitive target cells.³

Findings on liver trNK cells provide new insights into the discovery of CD49a⁺DX5⁻ NK cells in the uterus, skin, kidney, lung, and adipose tissues.^157–160 trNK cells in the skin are T-bet-dependent and lack Eomes expression.¹⁵⁷ However, unlike liver and skin trNK cells, uterus and kidney trNK cells are T-bet-independent.^158,160,161 Eomes⁺CD49a⁺ NK cells are most abundant during early pregnancy, while Eomes⁻CD49a⁺ NK cells dominate before puberty.¹⁶² Interestingly, a recent study using an immune-competent NK cell-specific reporter mouse showed that although both cNK and trNK cells accumulate in the mouse uterus, only trNK cells proliferate, and these proliferating trNK cells are the source of uterus NK cells during endometrial decidualization.¹⁶³ trNK cells in salivary glands are positive for both CD49a and DX5¹⁴⁹ and are present in normal numbers in T-bet-, Eomes-, and NFIL3-deficient mice. However, lack of TGF-β signaling significantly decreases salivary gland NK cell numbers.^149,164 Unlike liver NK cells, lung NK cells control viral proliferation after primary influenza virus infection but do not protect mice against secondary influenza virus infection,¹⁶⁵ suggesting the absence of a memory phenotype and function in lung NK cells. An in vitro cytokine-based feeder-free system has been developed as an approach to generate CD49a⁺Eomes^−/+ NK cells using IL-15 and IL-4; IL-15 is essential for the development and maintenance of CD49a⁺ NK cells, while IL-4 induces the expression of Eomes and converts Eomes⁻CD49b⁻ NK cells into CD49a⁺Eomes⁺ NK cells.¹⁶⁶

Inspired by these findings, several groups have attempted to define trNK cells in humans. A T-bet⁺Eomes⁻CD49a⁺ NK cell subset with a phenotype homologous to that of CD49a⁺ trNK cells in the murine liver has been identified in the human liver but not in afferent or efferent blood of the liver.¹⁶⁷ This trNK cell subset is CD56^bright and expresses killer cell Ig-like receptor (KIR), NKG2C, and low levels of CD16, CD57 and perforin.¹⁶⁷ This subset expresses high levels of inflammatory cytokines such as IFN-γ, TNF and GM-CSF and degranulates poorly upon stimulation.¹⁶⁷ Although trNK cells are believed to arise from precursors distinct from cNK cells and lack Eomes expression,^168,169 a recent study has identified an Eomes^hi population of NK cells in the human liver that is completely absent in the blood.¹⁷⁰ Eomes^lo NK cells circulate freely, whereas Eomes^hi NK cells are unable to leave the liver.¹⁷⁰ This population accounts for more than 50% of human liver NK cells and largely overlaps with CD56^brightCXCR6⁺ NK cells. However, these cells do not overlap with CD49a⁺ liver-resident NK cells.¹⁷¹ Thus, two nonoverlapping NK cell populations have been identified in the human liver: CD49a⁺ NK cells and Eomes^hi (largely CD56^brightCXCR6⁺) NK cells (Table 1). It was found that immature CD16⁻ NK cells can differentiate into both hepatic-specific CD49a⁺ and CXCR6⁺ NK cells.¹⁷² Interestingly, CyTOF analysis has revealed CD49e as a discriminating marker in human hepatic NK cells and suggested that CD49e⁻ NK cells are the human liver-resident NK cells instead of CD49a⁺ NK cells.¹⁷³ Unlike murine liver-resident NK cells, both CD49e⁻ and CXCR6⁺ NK cell populations in humans express Eomes rather than T-bet (Table 1). Furthermore, a CD49a⁺Eomes⁺ subset of NK cells has been identified at the maternal-fetal interface in both humans and mice.¹⁷⁴ A decrease in this NK cell subset impairs fetal development and results in fetal growth restriction.¹⁷⁴

Table 1.

Unique features of human liver-resident NK (lrNK) cells

	CD49a+ lrNK cell	Eomeshi lrNK cell	CD49e- lrNK cell
% of total intrahepatic NK	~0–13%	~50–60%	~60%
Phenotypic marker	CD56brightCD16−	CD56brightCD16−	CD56brightCD16−
CD49a	+	−	ND
CD49e	−	−	−
CD69	+	++	++
CD103	+	−	ND
CD127	−	−	ND
CXCR6	−	+	+
NKG2A	−	+	++
NKG2C	++	−	ND
NKG2D	+	+	+
KIRs	+	+	ND
NKp30	+	ND	ND
NKp44	−	+	+
NKp46	+	++	ND
CD226	+	+	ND
Transcription factor
Eomes	+	++	++
T-bet	++	+	+
Cytokine secretion
IFN-γ	+	++	+++
TNF-α	+++	+	+
GM-CSF	+	+	ND
Cytolytic molecule
CD107a	+	+	+
Perforin	+	+	ND
Granzyme A	+	ND	ND
Granzyme B	++	+	ND
Granzyme K	ND	++	ND
References	a,b	b–d	e

− not expressed, + low expression, ++ intermediate expression, +++ high expression, ND not determined

^aMarquardt et al.¹⁶⁷

^bHydes et al.¹⁵⁶

^cStegmann et al.¹⁸³

^dCuff et al.¹⁷⁰

^eAw Yeang et al.¹⁷³

Although CD103 is one of the signature markers of T_RM cells, there is little evidence regarding expression of CD103 on trNK cells. Human intrahepatic CD3⁻CD49a⁺CD56⁺ NK cells only express low levels of CD103,¹⁶⁷ while neither Eomes^hi nor CD49e⁻ NK cells express CD103 (Table 1). In addition, human Eomes⁺CD49a⁺ decidual NK (dNK) cells,¹⁷⁵ CD56⁺ NK cells in nasal lavage,¹⁷⁶ and CD56^brightCD49a⁺ NK cells in the lung¹⁷⁷ express CD103.

trNK cells perform functions that are significantly different than those of cNK cells. Liver-resident trNK cells have a more potent and faster response to haptens and viruses through cytokine production or clonal-like expansion.^151,178,179 Furthermore, they also contribute to immunosurveillance; the absence of CD49a⁺CD103⁺ NK cells results in accelerated tumor growth.^180,181 In addition, a greater proportion of lung-resident CD56^brightCD49a⁺ NK cells express surface CD107a compared with CD56^brightCD49a⁻ NK cells in lung explants infected ex vivo with Influenza A Virus; the former population provides early and important control of viral infection¹⁷⁷ and plays a dominant role in controlling metastatic tumor growth in the lung¹⁸².

Concluding remarks

The existence of lymphocytes remaining resident in the peripheral tissues has challenged the long-standing concepts of memory T cells and NK cells, and the newly defined tissue-resident subsets have been the subject of numerous cutting-edge studies.

T_RM cells are located in a wide range of peripheral organs including the skin, sensory ganglia, gut, lungs, brain, salivary glands, female reproductive tract, and other sites. However, a recent study by Beura et al. demonstrated the existence of T_RM cells in the draining LN as well,³¹ which further enhances our knowledge of T_RM cells. Although coexpression of CD69 and CD103 is observed on the majority of T_RM cells, CD103⁻ T_RM cells also exist, and other phenotypic markers such as CCR8, CXCR3, CXCR6, and CD49a have been used to identify and classify T_RM cells. CD49a has recently been identified as a marker to differentiate T_RM cells into CD49a⁺ T_RM cells, which produce IFN-γ, and CD49a⁻ T_RM cells, which produce IL-17, in human skin epithelia,⁴⁴ suggesting the presence of different subsets within T_RM cells. CD49a is also a typical marker of liver-resident NK cells in mice. CD49a⁻DX5⁺ NK cells are cNK cells that circulate in the blood, while CD49a⁺DX5⁻ NK cells are liver-resident NK cells that exhibit very different transcriptional and phenotypic features than those of the cNK cells.¹⁵¹ CD49a⁺DX5⁻ NK cells have also been found in the uterus, skin, kidney, and adipose tissue. Unlike in the murine liver, where only one population of liver-resident NK cells has been identified, two nonoverlapping NK cell populations have been identified in the human liver: CD49a⁺ NK cells and Eomes^hi (largely CD56^brightCXCR6⁺) NK cells.¹⁶⁷ A recent finding has also suggested that CD49e⁻ NK cells may be human liver-resident NK cells,¹⁷³ implying a new possibility in the identification of human liver-resident NK cells. Although tissue-resident lymphocytes are identified through the expression of the surface receptors discussed above, it is still not clear whether additional phenotypic markers and subsets are involved. The tissue microenvironment and infection route can both contribute to the regulation of phenotype and function of these cells, and phenotypic markers may differ between humans and mice.

Of note, studies done using intravascular staining have been suggested to result in an overestimate of T_RM populations within lung tissues,²⁸ while parabiosis and quantitative immunofluorescence microscopy studies have revealed that conventional isolation methods and identification through phenotypic markers underestimate the size of T_RM population.⁴² These findings suggest that limitations in experimental techniques can hamper our knowledge; cytometric identification through phenotypic markers sometimes cannot represent an entire population. Lack of recirculation and histology and/or other imaging techniques will fully reveal tissue-resident cell population and their functions.

The differentiation and maintenance of tissue-resident populations are crucial to long-term host survival. It is still unclear which populations of cells give rise to the T_RM and trNK subsets. T_RM cells may be derived from KLRG1⁻ T_RM precursor cells, from nonlymphoid or recirculating memory CD8⁺ T cells, or from pre-existing T_RM cells. The factors that determine the fate of T_RM cells in different tissues are not well defined. Additionally, it remains questionable whether a common developmental pathway exists. Of note, the possibility of peripheral tissue-derived precursor cells that give rise to T_RM cells should not be ruled out, as such precursors have been identified for trNK cells.¹⁵¹

T_RM cells provide stronger anti-viral immune responses to a variety of viruses upon rechallenge compared to circulating memory T cells, while trNK cells provide more potent and faster responses to haptens, viruses, or viral-like particles. T_RM cells are now considered to be promising mediators of long-lived peripheral immunity to be elicited by future vaccines; however, effective responses in vaccine settings still merit further research. The role of T_RM cells in anti-tumor immune responses is yet to be fully discerned; their accumulation has been confirmed in several human solid tumors and has been associated with prolonged survival and a favorable prognosis. However, some of the T_RM cells in the tumor microenvironment tend to adopt an exhausted phenotype, suggesting that the role of T_RM cells in anti-tumor immunity is still unclear; they are protective in certain cancers but not in others. trNK cells also contribute to immunosurveillance, the absence of which results in accelerated tumor growth. Few have suggested using T_RM cells as the targets in checkpoint immunotherapy due to their higher expression of co-inhibitory receptors. However, targeting T_RM cells in this setting should be approached carefully, as they sometimes protect the host against tumor cells. Generally, a deeper and more comprehensive understanding of the immune responses at sites of infection is necessary for tissue-resident cell-based immunotherapies to develop. The underlying mechanisms and molecular regulators involved in the activation, persistence and effector functions of these tissue-resident cells could be used in order to enhance the immune responses in local tissues.

Conflict of Interest

No potential conflicts of interest were disclosed.