Introduction

Resident memory T cells (T_RM) are a recently described subset of memory T cells that persist long-term in peripheral tissues. T_RM undergo a distinct differentiation program that discriminates them from circulating T cells; this program likely evolved to populate epithelial barrier tissues - skin, gut, lung and reproductive tracts - with highly protective T cells specific for the pathogens most commonly encountered through each tissue (1, 2). In this way, the immune system distributes memory T cells to the tissue sites where they will likely to be needed in the future. However, dysregulation of T_RM can give contribute to human autoimmune and inflammatory diseases. T_RM differentiate and accumulate in tissues after pathogen infection but evidence suggests they also develop after sensitization to otherwise harmless environmental or self antigens. Pathogenic autoreactive T_RM induce the fixed, recurrent skin lesions of psoriasis and T_RM specific for environmental allergens likely underlie development and worsening of allergic asthma and contact dermatitis. Moreover, the T_RM differentiation program is likely also engaged when T cells are activated in non-barrier tissues such as the kidney, brain and joints. As a result, pathogenic T_RM likely contribute to chronic inflammatory diseases in non-barrier tissues as well. This review will discuss both the basic biology and human diseases associated with T_RM, exploring how these cells support healthy immune function and contribute to human inflammatory disease.

The discovery of resident memory T cells

In prior decades, it was thought that tissue tropic T cells remained in the circulation until recruited to sites of inflammation and that non-inflamed peripheral tissues contained very few T cells. T cells recruited to tissues during infections were thought to either exit tissues or undergo apoptosis following clearance of the infection. However, CD8⁺ T cells that remained long term in mouse lung following influenza virus infection were observed as early as 2001 (3, 4) and antigen-specific CD8⁺ T cells were found to migrate to non-lymphoid tissues and remain as long-lived memory cells following infections with Listeria and vesicular stomatitis virus (5). In 2004, it was observed that skin grafts of normal appearing, non-lesional skin from psoriatic patients gave rise to active psoriatic skin lesions after transplantation onto immunodeficient mice, demonstrating that a population of resident pathogenic T cells existed in the non-lesional skin of patients with psoriasis (6, 7). Subsequently, it was discovered that the healthy skin surface of an adult human being contained nearly 20 billion T cells, approximately twice as many as are present in the entire blood volume (8). Human skin T cells were all CD45RO⁺ memory T cells, co-expressed the skin homing addressins CLA and CCR4, had potent effector functions and a diverse T cell repertoire (8–11). Studies designed to determine the location of skin tropic memory T cells found that 98% were located in human skin under non-inflamed conditions and only 2% were in the circulation. (8) These findings challenged the idea that T cells must be recruited from peripheral blood during infectious challenges and suggested that at least some subsets of tissue tropic T cells spend the majority of their time in peripheral tissues. Subsequent studies demonstrated large numbers of antigenically diverse T_RM in human lung, gastrointestinal tract, peritoneum, reproductive tract and bone marrow (12–19).

The Mission of Resident Memory T cells: To Persist, Protect & stay Put

The large numbers of memory T cells in human epithelial barrier tissues is consistent with the idea that some T cells may reside long-term in peripheral tissues but a series of elegant mouse models were required to clarify how these cells are generated, distributed and maintained. Mice kept in clean, pathogen-free barrier facilities had few skin T_RM, likely because these cells are generated by the infections barrier facilities are designed to prevent. However, experimentally administered skin infections with HSV and vaccinia virus both generated populations of CD8⁺ T cells that remained in skin long after resolution of the acute infection and provided rapid viral clearance after skin reinfection (20–24). Dendritic cells (DC) were capable of presenting antigen to local populations of HSV-specific T_RM, leading to a local recall immune response entirely within the skin (25). T_RM generated by prior vaccinia skin infection were more protective than circulating T cells and rapidly cleared virus from the skin after reinfection in the complete absence of both antibodies and circulating T cells (21, 23, 24). Moreover, local skin infection with vaccinia virus led to seeding of the entire cutaneous surface with long lived, highly protective T_RM, although the highest concentration of these cells occurred at the site of infection (21) (Figure 1). Subsequent re-infections at different skin locations led to progressive increases in the number of protective T_RM throughout the skin; in this way local skin infections led to a global protection of the entire skin by long lived, highly protective T_RM (21). Interestingly, preferential deposition of T_RM at inflammatory sites following HSV infection was found to be antigen independent (22). T_RM progenitor cells generated by recent skin infection migrated in greatest numbers into areas of inflamed skin, regardless of whether the inflammation was caused by local HSV infection or by irritation of the skin by the contact allergen 2,4-dinitrofluorobenzene (DNFB) (22).

The best possible outcome of vaccination is the generation of highly protective T_RM in the epithelial barrier tissue most likely to be infected. Skin scarification with live vaccinia virus, associated with local keratinocyte infection, generated long-lived CD8⁺ T cell-mediated immunity that was 100,000 times more effective than subcutaneous, intradermal and intramuscular vaccination in protecting against viral reinfection of the skin (24). This protection was mediated entirely by skin T_RM and did not require antibodies or recruitment of circulating T cells from blood (21, 24). These studies call into question the current vaccination strategy of injecting vaccines into muscle. Muscle lacks a resident population of antigen presenting cells and intramuscular vaccination is unlikely to effectively focus immune responses to the relevant peripheral tissues.

A surprising additional benefit of local skin infection with vaccinia was the generation of a smaller but functionally important population of lung T_RM that could, in the absence of antibody and circulating T cells, partially protect mice from an otherwise lethal pulmonary challenge with vaccinia virus (23, 24). This spillover of immune protection occurred as a result of early release from skin draining lymph nodes of a subset of T cells proliferating in response to antigen following skin infection. These released T cells traveled to distant lymph nodes draining other peripheral tissues where they continued to proliferate in the absence of antigen and developed tissue tropism for these distinct, noncutaneous tissues (23) (Figure 2, left panel). In other words, local infection of one barrier tissue led to at least some immune T_RM-mediated protection of other epithelial barrier tissues. A similar mechanism clearly exists in humans, where smallpox vaccination, delivered by skin scarification, has historically provided effective protection against smallpox, which is primarily transmitted as a respiratory infection (Figure 2, right panel).

The immune protection provided by T_RM is more than skin deep; infection models in mouse gut, lung and reproductive tract confirm that T_RM persist, protect and do not recirculate. Gut infection with lymphocytic choriomeningitis virus, listeria, and other pathogens generated long-lived intraepithelial T cells with potent protective capacity (26). CD4⁺ T_RM in the lung generated by influenza infection persisted, did not recirculate (14) and provided potent in situ protection against an otherwise lethal influenza challenge (27, 28). Both intranasal and intraperitoneal infection with influenza induced viral specific effector memory T cells but only nasal infection generated T cells that mediated protection against an otherwise lethal intranasal influenza challenge (29). In the genitourinary tract, intravaginal infection with HSV induced the accumulation of highly protective T_RM (30). Moreover, topical intravaginal chemokine application “pulled” circulating effector T cells generated by subcutaneous HSV infection into the vaginal mucosa where they differentiated into protective T_RM. Likewise, intravaginal inflammation induced by topical nonoxynol-9 application pulled circulating HSV-specific effector cells into the vaginal mucosa where they differentiated into protective T_RM (22).

T_RM generation in peripheral tissues: A distinct differentiation program

Studies in both mice and humans have demonstrated that T_RM from peripheral tissues express CD69 and a subset also express CD103 (8, 16, 20, 21, 28, 31–33). In mice, epithelial resident CD8⁺CD103⁺ T have been most fully characterized although CD4⁺ T_RM have also been observed and are in fact the dominant population of T_RM in human skin (34, 35). After HSV infection in mice, sessile CD8⁺ T_RM took up residence in the epidermis whereas CD4⁺ T_RM remained in the dermis and were locally more mobile (34). Interestingly, CD8⁺ T_RM that took up residence in the epidermis after HSV infection in mice progressively displaced existing populations of dendritic epithelial T cells (DETC) from their epidermal niches (36). In contrast to CD8⁺ T_RM, which take up residence in epidermis only after skin infection, γδ DETC are present in mouse epidermis prior to infections, although they are not observed in humans. γδ DETC were fixed in place within the epidermis whereas CD8⁺ T_RM moved laterally between keratinocytes, frequently interacting with Langerhans cells, but remained within a limited territory of migration (36). Lastly, decreased expression of the KLF2 transcription factor leading to sphingosine 1 phosphate receptor 1 (S1P1) down-regulation was also necessary for retention of CD8⁺ T_RM in skin (37). CD69 suppresses the activity of S1P1, suggesting CD69 expression by T_RM may assist in retaining these cells in peripheral tissues (38).

Gene profiling of CD103⁺ CD8⁺ T_RM isolated from mouse skin, lung, gut and brain has demonstrated that T_RM are the product of a distinct differentiation pathway that renders T_RM extracted from different peripheral tissues more similar to each other than they are to circulating T cells (1, 2). The T_RM differentiation program is engaged only after migration of recently activated T cells into peripheral tissues such as the skin (1) (Figure 3). The exact precursor cell type for T_RM remains unidentified but in mice they are derived from the KLRG1⁻ T cell population (1). Localization of HSV specific CD8⁺ T_RM in the epidermis was dependent on a chemokine mediated process that could be inhibited by pertussis toxin. Maintenance of epidermal T_RM cells required IL-15 and CD103 expression on these cells was induced in a TGF-β dependent manner after entry into the epidermis (1). A second study confirmed that TGF-β was important for establishing epithelial T_RM residence and demonstrated additional requirements for IL-33 and TNFα (37).

Peripheral tissues also contain populations of recirculating T cells and one common feature of these T cells is their expression of CCR7. CCR7 expression is required for the exit of T cells from peripheral tissues via the afferent lymphatics (39–41). Studies in Kaede mice demonstrated that skin contained both non-recirculating CD4⁺CCR7⁻ T_RM and recirculating CD4⁺ CD69⁻CCR7⁺L-selectin^lo T cells that exited skin in a CCR7 dependent mechanism (40). Human skin contains a recirculating population of L-selectin⁺CCR7⁺ central memory T cells that co-express the skin addressins CLA and CCR4. (35)

Resident Memory T Cells in Humans

Mouse model systems have provided invaluable insights into many biologic processes. However, there are significant differences in the immune systems of humans and mice and these differences remain incompletely characterized. Some mouse models, such as the C3H model for alopecia areata, closely resemble the human disease but others, including proposed mouse models of psoriasis, only partially recapitulate the human disease (42–44). It is therefore critical that observations made in mouse models be confirmed in humans.

Evidence has accumulated that the three critical characteristics of mouse T_RM - that these cells persist, protect and do not recirculate - also hold true in humans. The presence of 20 billion T cells in human skin and the observation that T cell numbers in skin remain constant even in patients in their 90s is circumstantial evidence that these cells persist long-term (8, 45). However, the persistence and protective nature of human T_RM has now has been elegantly and directly demonstrated in studies of patients with genital HSV (46, 47). HSV specific CD8αα T cells were found to persist at the dermo-epidermal junction long term after resolution of acute HSV outbreaks (47). During subsequent episodes of subclinical viral reactivation, asymptomatic viral shedding was noted accompanied by HSV protein production by infected keratinocytes. T_RM were observed to cluster around the infected keratinocytes and produce perforin. Suppression of viral reactivation was observed when T_RM were present in high density whereas low numbers of HSV specific T_RM were associated with full viral reactivation and progression to clinically apparent lesions (47). These elegant studies, utilizing multicolor immunofluorescence and high throughput TCR sequencing, are illustrative of the elegant work that can now be done to characterize T_RM in human tissues.

Additional evidence that T_RM in the skin persist, protect and do not recirculate comes from studies of patients with leukemic cutaneous T-cell lymphoma (L-CTCL). Alemtuzumab, an anti-CD52 antibody, used in the treatment of CLL and CTCL, depletes via ADCC requiring accessory cells that are most frequent in the circulation (35). Low dose alemtuzumab treatment depleted all circulating T cells and purged the skin of both malignant and benign L-selectin⁺/CCR7⁺ central memory T cells (T_CM), a skin tropic subpopulation of highly migratory T cells (35, 48). Despite complete depletion of all circulating and recirculating T cells, low dose alemtuzumab was not associated in this patient population with an increased risk of infection. Biopsy of the skin demonstrated that large numbers of T_RM with marked effector functions and diverse antigen specificities remained in the skin of treated patients. These studies demonstrated that human skin is colonized by a diverse, highly protective and non-recirculating population of T_RM. In addition, these studies demonstrated that a subset of L-selectin⁺/CCR7⁺ T_CM, a subset of T cells thought to primarily recirculate between the blood and lymph nodes, also express skin homing addressins in humans and can enter and recirculate through the skin (35).

Large numbers of diverse CD69⁺ T_RM with marked effector functions and enriched for influenza specificity have also been detected in histologically noninflamed human lung (14, 19). CD8⁺ T cells in human lung expressing the T_RM marker CD103 were found to be the subset that were specific for influenza, compared to non-resident T cell populations from the same patients (14, 49). More recent studies in other tissues have demonstrated the presence of T_RM in human stomach, colon, small intestine and cervix (12, 16, 18). CD8⁺ T cells were excluded from high-grade cervical dysplastic lesions (18) but patients vaccinated against oncogenic papilloma virus proteins had robust T cell infiltration of high-grade cervical intraepithelial neoplasias (50).

Findings in both human and mice suggest that the T_RM populations in each epithelial barrier tissue may be enriched for the pathogens commonly encountered through those sites. Although the antigen specificity of T_RM in human peripheral tissues has been difficult to interrogate directly, a recent study evaluated the antigen specificity of circulating effector T cells tropic for the skin (CLA⁺), gut (α4β7 integrin⁺) and other sites (including lung) (51). T cells tropic for skin were enriched for specificity against skin associated pathogens and commensals (S. epidermidis, C. albicans, HSV-1), gut-tropic T cells were enriched for specificity against gut associated pathogens (S. Typhi, Rotavirus) and remaining T cells (including lung tropic T cells) were enriched for pathogens encountered through the lung (Influenza A and Myocplasma).

T_RM in human disease

The T_RM genetic differentiation program evolved to progressively populate epithelial barrier tissues with nonrecirculating memory T cells with potent effector functions that are specific for the pathogens encountered through those sites. Under conditions of health, these cells provide extremely rapid on-site immune protection against known pathogens. However, when T cells specific for allergens or auto-antigens enter peripheral tissues and differentiate into T_RM or when T_RM themselves become malignant, T_RM cease to be the solution and instead become part of the problem (Figures 3–5).

The clinical characteristics of T_RM mediated human diseases reflect T_RM biology. Because T_RM do not recirculate, inflammatory lesions caused by T_RM tend to be well demarcated, with distinct margins and an abrupt cutoff with normal tissue (Figure 4). T_RM mediated lesions tend to persist long-term in a particular location and, although they may appear to resolve with anti-inflammatory therapy, will often recur in the same location and grow to the previously observed size after therapy is discontinued if the inciting agent is still present. Because T_RM are already on site and can respond to antigen presented by local tissue resident dendritic cells (25), T_RM mediated tissue inflammation is extremely rapid and can occur within hours of antigen exposure. In general, onset of inflammation within 24 to 48 hours after antigen exposure is suspicious for T_RM involvement. Lastly, because of the tendency of T_RM to progressively accumulate in tissues with repeated antigen exposures, T_RM mediated inflammatory diseases tend to worsen over time, with an increasingly rapid onset of inflammation and increasing inflammatory severity with each exposure.

Two prototypic T_RM skin diseases: Mycosis fungoides and psoriasis

Conclusions and perspectives

In the years to come, it will be critical to determine how long T_RM remain in peripheral tissues, how often they turn over, what environmental niches within the peripheral tissue support their continued long-term survival and what if anything can be done to mobilize these cells from peripheral tissues if their presence is detrimental. It will be critical to identify which human diseases have a T_RM mediated component and to evaluate the therapies for their impacts on T_RM. Many treatment modalities in current clinical use, including total body irradiation, skin brachytherapy, phototherapy and a variety of systemic chemotherapy drugs are completely uncharacterized with respect to their effects on T_RM. For example, alemtuzumab, a medication previously thought to deplete all T and B cells in the body, depletes only circulating and recirculating T cells but leaves at least skin T_RM intact (35). A better understanding of how and where these drugs act on memory T cells could lead to more selective therapies that preferentially target the pathogenic T cells while sparing nonpathogenic T cell subsets.

The discovery and characterization of T_RM has proved once again that nature and biology are smarter than we are. A system that distributes memory T cells specific for the pathogens likely to be encountered through those sites directly to the tissues at risk is a remarkably elegant way to focus potentially dangerous effector T cells only to the places they are most likely to be needed. Work remains to be done to identify which vaccination strategies generate the greatest number of T_RM in the target tissues and similar approaches are needed to induce the formation of T_RM a tumor sites, where they can locally protect against recurrence. However, the strengths of T_RM become a problem when these cells are maladaptive and cause inappropriate inflammation. The long lived, non-recirculating nature of the cells combined with their potent effector functions results in inflammatory diseases that are remarkably intransigent and tend to recur in the same anatomic locations. Further studies of the basic biology of these cells may identify an Achilles’ heel in T_RM biology that will allow their depletion and local control within tissues.