An Overview of Tissue-Resident Memory T Cells in the Intestine: From Physiological Functions to Pathological Mechanisms

The human intestine contains a complex network of innate and adaptive immune cells that provide protective immunity. The dysfunction of this network may cause various chronic diseases. A large number of T cells in the human intestine have been identified as tissue-resident memory T cells (TRM). TRM are present in the peripheral tissues, and they do not recirculate through the blood. It is known that TRM provide rapid immune responses at the frontline of pathogen invasion. Recent evidence also suggests that these cells play a role in tumor surveillance and the pathogenesis of autoimmune diseases. In this review, we discuss the general features of intestinal TRM together with their role in intestinal infection, colorectal cancer (CRC), and inflammatory bowel disease (IBD).


INTRODUCTION
The intestinal mucosa is predominantly exposed to environmental antigens, such as food, innocuous microbes, and enteric pathogens. A complex network of innate and adaptive immune cells in the intestine mediates the protective immune responses against harmful pathogens and immune unresponsiveness to benign antigens. Among these immune cells, memory T cells provide effective and efficient immune responses to antigens previously encountered (1).
Memory T cells were primarily divided into two subsets: central memory T cells (T CM ) and effector memory T cells (T EM ) (2). T CM express high levels of secondary lymphoid organ homing receptors CCR7 and CD62L and recirculate between the blood and secondary lymphoid organs (3). On the other hand, T EM express integrins and chemokine receptors and circulate through the non-lymphoid tissues (4). With advances in technology, another lineage of memory T cells called tissue resident-memory T cells (T RM ) (5) has been identified. Unlike T CM and T EM , T RM are permanently retained in the peripheral tissues and do not recirculate via the bloodstream. Apart from the location and migration patterns, they mediate local immune response to reencountered pathogens (5). In fact, most memory T cells in the intestinal mucosa possess a resident nature with distinct phenotypes and transcriptional profiles (6)(7)(8), implicating the significance of T RM in intestinal immunity.
Several studies focused on the intestinal immune system have explored the potential roles of T RM in pathogen defense (9), cancer immunosurveillance (10), and immunopathies (11). Understanding how T RM function in intestinal immunity is vital for developing novel therapies and vaccines targeting T RM . Thus, this review aims to summarize the features of intestinal T RM and their role in intestinal health and disease.

T RM IN THE INTESTINE
Intestinal T RM are phenotypically different from T EM . CD69 and CD103 are the two key cell surface markers identified for T RM (12,13). Lectin CD69, a marker for early T cell activation, is reexpressed in the intestinal T RM , and it can be used to distinguish T RM from their circulating counterparts (7). CD69 can promote T cell retention by downregulating the expression of the sphingosine-1-phosphate receptor 1 (S1PR1) protein and interacting with the transmembrane domains of the S1PR1 to suppress its function, subsequently preventing the T cells from sensing the sphingosine-1-phosphate (S1P) gradient and exiting from the gut (14)(15)(16)(17). However, according to a recent report, the elevated CD69 expression may only serve as a passive marker rather than an essential functional regulator in driving T RM cell formation in the gut (18). Another marker for T RM , CD103 (aE integrin), is expressed by most CD8 + cells and few CD4 + cells in the intestine (7,12,13,19,20). The aEb7 integrin binds to Ecadherin expressed on the intestinal epithelial cells, promoting T RM retention in the epithelium (21)(22)(23). However, the absence of CD103 does not rule out long-term maintenance of T cells in the intestine. Evidence showed that a small population of CD8 + CD69 + CD103 -T cells displayed persistence in the human intestine transplantation model for more than one year (12,24). The immune repertoire of CD8 + CD69 + CD103 -T cells showed low clonal overlap with the CD8 + CD103 + subset and was rather similar to T cells from peripheral blood in terms of phenotype and function (12,24). Therefore, it can be speculated that the CD8 + CD103 -T RM subset represents an intermediate state between recently recruited T cells and CD8 + CD103 + T RM . Interestingly, CD8 + CD103 -T RM express high levels of b2-integrin, but whether it plays a role in CD8 + CD103 -T RM differentiation remains unclear (24). Furthermore, a fraction of CD8 + CD103 -T RM express KLRG1, another ligation of E-cadherin (12,24,25). The comparison of CD103 + T RM and CD103 -T RM is summarized in Table 1. For CD4 + T RM and CD8 + T RM , the phenotypical and functional differences between their CD103 + and CD103subsets are similar (24). In addition to CD69 and CD103, other molecules have been described as potential phenotypical markers of intestinal T RM , including CD49a (integrin a1 chain) (8,12,13), CD101 (8) and CD161 (13,24,28).
T RM have a distinct pattern of transcription factor expression that coordinates T RM development and survival. Hobit and Blimp-1 have been identified as two key transcription factors expressed in mice T RM (29). Hobit and Blimp-1 work in synergy to repress genes required for tissue egress by binding to the S1pr1, Tcf7, and Ccr7 loci (29). Additionally, Blimp-1 alone downregulates the expression of KLF2 and subsequently S1PR1 (29). Runx3 plays a primary role in CD8 + T RM differentiation by enhancing the expression of core residency signature and repressing the expression of signature genes of T CM (30,31). Downregulation of T-box transcription factor Eomes and T-bet is also necessary for early-stage T RM differentiation (32,33). Recent studies have found that effector-like and memory-like CD8 + T RM subsets can be distinguished by differential expression of two transcriptional factors Blimp-1 and Id3. It was shown that Blimp-1 hi Id3 lo CD8 + T RM displayed an effector gene signature and dominated the early phase of infections, whereas Blimp-1 lo Id3 hi CD8 + T RM exhibited enhanced memory potential and accumulated at the late stage of infections (34).
The differentiation and maintenance of intestinal T RM are highly regulated by the local microenvironment of the gut (26,35). Upon antigen encounter, naïve CD8 T cells become activated and differentiate into short-lived effector cells (SLEC) and memory precursor cells (MPEC) (36). Early adoptive transfer experiments in mice identified the MPEC with low KLRG1 expression as the precursors of T RM (21). However, a fate-mapping study on KLRG1 reporter mice revealed that more than half of the T RM population in the small intestine of mice originate from the effector T cell population previously expressing KLRG1 (37). In addition, CD103 -T RM within the intraepithelial (IE) preferentially developed from T cells that transiently expressed KLRG1 rather than those that never express KLRG1 (37). The expression of KLRG1 may be downregulated by the cytokine transforming growth factor b (TGF-b) and T cell receptor (TCR) triggering in the intestine (38). TGF-b plays a vital role in the development of the intestinal T RM (39). TGF-b selectively accelerates the apoptosis of SLEC in the intestine, thus contributing to the rapid MPEC phenotype formation (26). TGF-b also downregulates KLF2 expression and subsequently downregulates the expression of S1PR1 expression (40). Additionally, TGF-b upregulates the expression of CD103 directly through Smad3 and indirectly through counteracting suppression of CD103 expression mediated by Eomes, T-bet, and TCF1 (26,32,41). IL-12 and TNF-b produced by inflammatory monocytes can suppress the TGF-b-induced CD103 expression, leading to the formation of CD103 -T RM cells in the lamina propria (LP) with different transcriptional profiles (27). When exposed to extracellular ATP, the purinergic receptor P2RX7 can promote TGF-b receptor expression in CD8 + T RM precursor, suggesting the importance of P2RX7 in T RM generation (42). Recently, Peng et al. found that local ICOS signaling also promotes the generation of intestinal T RM by activating the PI3K pathway and downregulating KLF2 expression (43). While the maintenance of T CM and T EM depends on the presence of IL-15, IL-15 is not necessary for T RM retention (44).

T RM AND PROTECTION AGAINST PATHOGEN
Numerous studies have revealed that the intestinal CD4 + and CD8 + T RM can be generated by intravenous or oral infections and provide enhanced regional immunity (19,20,26). In mice, CD4 + and CD8 + T RM have been shown to provide strong protection against oral infection with Listeria monocytogenes (19,26). In humans, attenuated oral typhoid vaccine can elicit activated CD4 + and CD8 + T RM response (45,46). Here, we review the two possible mechanisms by which the intestinal T RM protect against pathogens rapidly and efficiently upon reinfection ( Figure 1).
First, T RM upregulate the production of proinflammatory cytokines, serving as an alarm for both innate and adaptive immune responses. Previous experiments have demonstrated the robust cytokine polyfunctionality of certain pathogen-specific CD4 + and CD8 + T RM that produce high levels of interferon-g (IFN-g), IL-2, and tumor necrosis factor-a (TNF-a) (20,47). The functions  of T RM primarily rely on the cytokines they secrete. Upon cognate peptide rechallenge, the released IFN-g induces the expression of various antiviral and antibacterial genes in the surrounding cells to limit the initial pathogen invasion (48,49). Furthermore, T RMinduced IFN-g is required for the rapid recruitment of the circulating B and memory T cells via vascular cell adhesion molecule-1 (VCAM-1) upregulation on the endothelial cells (50). IL-2 stimulates granzyme B expression in the natural killer cells and the bystander memory T cells. The proinflammatory cytokine TNFa facilitates the maturation of dendritic cells by upregulating CCR7 and other co-stimulatory molecules (49). Comparing subsets of intestinal T RM , LP CD8 + CD103 + T RM are more polyfunctional in cytokine production than LP CD8 + CD103 -T RM and IE CD8 + CD103 + T RM (12,25). For LP CD4 + T RM , both CD103 + and CD103subsets are very potent producers of cytokines IFN-g, IL-2, and TNF-a (13). Second, some T RM exhibit direct effector function. A fraction of intestinal T RM can directly lyse the infected cells by expressing high levels of cytotoxic granules such as granzyme B at an early stage (23,51). This effector-like phenotype of the intestinal T RM mainly depends on the cytokine milieu of the intestinal microenvironment rather than the persistent antigen stimulation (23). Interestingly, it was found that the expression of cytotoxic granules is quite similar between CD4 + and CD8 + T RM (19). CD8 + CD103 -T RM , especially those still expressing KLRG1, exhibit high cytotoxic and proliferative potentials (12,25). As mentioned above, they may represent T RM populations recently recruited and maintained in the intestinal mucosa. Previous studies using two-photon laser scanning microscopy have shown restricted motility of the intestinal T RM (52). They may stay at the sites of previous infections to surveil potential recurring pathogens. The actual contribution of the effector functions in immunologic protection mediated by the intestinal T RM needs further evaluation.
As shown in the mouse infection model, intestinal T RM undergo proliferation in situ in response to antigen rechallenging ( Figure 1) (19,53). Two recent studies have demonstrated that reactivated T CM maintain the potential to form CD103 -T RM and few CD103 + T RM , but the efficiency of reactivated T CM is extremely lower than naïve T cells (53,54). Under steady-state conditions, the intestinal T RM populations are maintained mainly by longevity rather than local proliferation (12).
Recent evidence suggested that intestinal T RM are not terminally differentiated as previously thought (Figure 1) (54). After pathogen rechallenge, intestinal T RM are able to egress from the intestine and differentiate into circulating memory T cells. And some "ex-T RM " can migrate to the draining lymph nodes and give rise to the local T RM (54)(55)(56). Such a process may be initiated by the antigen-driven downregulation of Hobit expression (56). In addition, "ex-T RM " that re-joined the circulating pool are advantaged to migrate back and re-differentiated into intestinal T RM (54). A T RM differentiation program is epigenetically maintained for those cells despite their developmental plasticity (54,55). The differentiation potential of T RM enables them to shape systemic T cell responses after reinfection (56).
In general, the intestinal T RM can exert protective functions locally by expressing proinflammatory cytokine and cytolytic granules. Based on these findings, developing vaccines that induce T RM to combat recurrent infections may be a promising strategy to enhance immunological protection.

T RM AND COLORECTAL CANCER
CD8 + T cells contribute to the immunosurveillance against cancer, and their protective effects are highly correlated to their ability to enter and survive in the immunosuppressive microenvironment of the tumor compartments (10). Therefore, T RM present within the tumor may be critical for controlling tumor growth. Recent reports have recognized the immunosurveillance function and prognostic significance of T RM in CRC.
Over a decade ago, large numbers of CD103 + CD8 + tumorinfiltrating lymphocytes (TIL) were observed in the microsatellite instability (MSI) sporadic CRC (57). Further investigations suggested that these T RM -like TIL may be divided into two subsets based on the expression of CD39 (58). CD39, an ectonucleotidase that hydrolyzes ATP and ADP, marks CD8 T cells for chronic antigen stimulation. The expression level of CD39 in the TIL varies significantly in the CRC patients and is remarkably high in patients with high MSI (59). CD39 + CD103 + TIL are enriched for tumor antigenspecific T cells which may contribute to the control of tumor growth. In comparison, CD39 -CD103 + TIL recognize cancerunrelated epitopes and are named "bystander" T cells (59,60). When activated in vitro, the CD8 + CD39 + CD103 + TIL are highly capable of killing the tumor cells in an MHCdependent manner (59). One possible explanation is that the interaction between the CD103 on the TIL and E-cadherin on the tumor cells induces the polarization and exocytosis of cytotoxic granules at the immune synapse. As a result, the effector function of the tumor-resident memory T cells is enhanced (61). Consistent with that, a co-culture system of intestinal tumor organoids and T cells showed the significance of CD103/E-cadherin signals for antitumor immune response (62). However, whether intestinal T RM are able to suppress tumors in vivo by target cell killing has not been validated. In addition to direct cell killing, intestinal T RM may suppress tumor cell growth by secreting cytokine TNF-a and IFN-g (63,64). Current studies also confirmed that infiltration of CD103 + CD8 + TIL is an independent predictor of survival for patients with CRC (65). CD39 + CD103 + TIL promote the gene transcription of typical markers related to exhaustion and immunomodulation (PD-1, CTLA-4, and TIM-3) (59), which implies a possible immune escape mechanism of tumor cells. Such T RM -like TIL are an appealing target of immune checkpoint inhibitor (ICI) therapy due to their expression of immunomodulatory markers and proliferative potential (58,59,66). In fact, CD103 expression highly correlates with clinical response to anti-PD-L1 immunotherapy in patients with lung and bladder cancer (67). A recent study also identified IFN-g-producing CD8 + T RM as a potential therapeutic target of ICI-colitis (68). It was found that most activated T cells in ICI-colitis co-express CD69 and CD103 (68). Moreover, the proportion of activated CD8 + T RM in intestinal mucosa is highly correlated with clinical and endoscopic scores of ICI-colitis (68).
Taken together, T RM contribute to the tumor immunosurveillance in the intestinal mucosa and act as a prominent target of ICI therapy and ICI-colitis. A more comprehensive understanding of the identification, regulation, and function of T RM -like TIL is required for improving current cancer treatment.

T RM AND INFLAMMATORY BOWEL DISEASE
IBD, including Crohn's disease (CD) and ulcerative colitis (UC), is a chronic inflammatory disease of the intestinal tract characterized by local relapsing flares (69). The pathogenesis of IBD involves abnormal immune responses against intestinal microbiome (70). Some properties of T RM suggest that they may be the key player in the onset and relapse of IBD. T RM possess the capacity to secrete pro-inflammatory cytokine and recruit other leukocytes. Also, T RM in the intestine mucosa are confined to certain regions with limited ability to migrate, which may explain the localized pattern of flares seen in IBD (52).
Recent evidence suggests that CD4 + T RM may serve as a driver for IBD (71). Zundler et al. reported that knockout of Hobit and Blimp-1, the two core transcriptional factors of T RM , prevented the development of colitis in several experimental mouse models (71). Furthermore, in the T cell transfer colitis model, the mouse with Hobit-Blimp-1 double knockout CD4 + T cells showed impaired secretion of pro-inflammatory cytokines (IFN-g, IL-13, and IL-17A) and recruitment of granulocytes and macrophages (71). In humans, the same authors observed that CD4 + CD69 + T cells in the LP of IBD patients produce markedly more pro-inflammatory cytokines (IFN-g, IL-13, IL-17A, and TNF) when compared to CD4 + CD69 -T cells. High CD4 + CD69 + CD103 + T RM proportion in LP was prospectively associated with earlier relapse in both UC and CD patients (71). Consistently, a study by Lamb et al. showed that the high expression of CD103 in CD4 + T cells is associated with elevated pro-inflammatory cytokine production and lowered regulatory markers expression in UC patients (72). Furthermore, CD4 + CD103 + T RM are enriched for Th17/Th1 lineage cells, which co-express IL-17A and IFN-g (72). Bishu et al. investigated colon samples of CD patients and found that CD4 + T RM in CD patients expresses higher levels of IFN-g and IL-17A relative to controls (73). Particularly, CD4 + T RM are identified as the major mucosal TNF-a-producing T cell subsets in the CD patients (73). Despite the pathogenic role of CD4 + T RM, it remains controversial whether CD4 + CD103 + T RM are increased in the gut of IBD patients (71,74).
The heterogeneity and functional profile of CD8 + T RM in IBD were also investigated. Bottois et al. observed two distinct subsets of CD8 + T RM in the ileum of CD patients, defined by the mutually exclusive expression of CD103 and KLRG1 (25). CD8 + CD103 + T RM were decreased in inflamed mucosa from IBD patients when compared to non-inflamed mucosa and controls, while CD8 + KLRG1 + T RM showed a significant increase (25). CD8 + CD103 + T RM in CD patients expressed higher levels of IL22, IL26, and CCL20. These three Th17-related cytokines are involved in tissue homeostasis and innate immune response (25). Similarly, single-cell RNA-sequencing of CD8 + T cells from the colon of UC patients showed that the expression of IL26 was enriched in the CD8 + CD103 + population (75). In addition, CD8 + CD103 + T RM express high levels of CD39 and CD73, two key functional markers of regulatory T cells (76). CD39 and CD73 can hydrolyze extracellular ATP into adenosine, which is a potent immunoregulator (77). Roosenboom et al. observed that CD8 + CD103 + T RM were significantly decreased in patients with active IBD compared to patients with endoscopic remission and healthy controls (74). These findings suggest that CD8 + CD103 + T RM may contribute to tissue homeostasis and immunoregulation. By contrast, CD8 + KLRG + T RM have higher proliferative and cytotoxic potential (12,25).
The pathogenic role of T RM makes them a promising target for IBD treatment. For instance, etrolizumab is a humanized monoclonal antibody that targets the b7 subunits of aEb7 and a4b7 integrins. In a completed phase III trial in UC, etrolizumab met its primary endpoint in two induction studies but not in any maintenance studies (78)(79)(80)(81). Phase III trials assessing etrolizumab in CD are still ongoing with positive results in an exploratory induction cohort (82). The clinical efficacy of Etrolizumab may be partially explained by impairing the retention of CD103 + T RM . In fact, a phase II trial in UC showed that the response to etrolizumab is associated with the number of CD103 + cell in the colon samples (83). However, CD103 is also expressed on a subset of intestinal dendritic cells (84), so it remains to be investigated whether T RM are the real target of etrolizumab. Another pharmacological that might target T RM is S1PR modulators. Recently, the S1PR modulator ozanimod has been approved for UC (85). By triggering the internalization and degradation of the S1P receptors, S1PR modulators inhibit lymphocyte egress from lymphoid organs and may impair local generation of intestinal T RM (86).
Collectively, a growing body of evidence implicates that T RM participate in the pathogenesis of IBD. Therefore, they might be the potential target of some currently developed drugs, especially etrolizumab and S1PR modulators. It is not surprising that T RM have also been identified as key mediators in other intestinal immunopathologies, such as celiac disease (87) and acute graftversus-host disease (88). Further exploration is needed to better understand the role T RM play in different intestinal immunopathologies and to develop therapies that specifically target pathogenic T RM .

CONCLUSIONS
A large number of CD8+ and CD4+ T RM are permanently located in the human intestinal LP and IE (12,13). After developing within the gut microenvironment, CD8+ and CD4+ T RM differ from their circulating counterpart both at the phenotypical level and transcriptional level (29). In recent years, studies in mouse models and human biopsies have demonstrated a crucial role for intestine T R M in the protection against enteric pathogens, immunosurveillance of intestinal malignancy, and probably inducing intestinal chronic inflammation (26,59,71). Advancing our knowledge of the properties and functions of intestinal T RM may provide insights into developing efficacious vaccines and therapies against intestinal diseases.

AUTHOR CONTRIBUTIONS
YL and YZ collected the papers and data, analyzed the conclusions, and drafted the manuscript; JS presented the idea of this paper, supported the funding, analyzed the conclusions, drafted and revised the manuscript. All authors contributed to the article and approved the submitted version.