It is considered that the surface markers contribute to the identification and maintenance of hepatic T_RM cells. Similar to other tissue-specific T_RM cells, hepatic T_RM cells downregulate the expression of tissue egression markers, like soingosine-1-phosphate 1 (S1PR1), and the homing receptors such as CD62L and CCR7 (43, 44). Furthermore, hepatic T_RM cells usually express some adhesion molecule and chemokine receptors, including CD69 (44), CD103 (17, 45), CD49a (36), CXCR3 (17, 23) and CXCR6 (46, 47), which are involved in their localization and maintenance in the hepatic sinusoids and portal veins.

The lectin CD69 is constitutive expressed on the majority of liver T_RM subsets. Upon exposure to antigens or pro-inflammatory mediators, the expression of CD69 is strongly upregulated on activated CD8⁺ T cells within peripheral tissues as a result of the downregulation of Krüppel-like factor 2(KLF2) (44, 48, 49). Meanwhile, as an antagonist of S1P1, CD69 complexs with S1P1 on the cell surface and leads to its internalization and degradation (50). Besides, CD69 also contributes to the retention status of hepatic T_RM cells by downregulating sphingosine 1 phosphate receptor (S1PR1)-mediated tissue egress (44). Therefore, it is likely to that its primary role is to restrict the egress of T_RM cells from the liver to the blood and lymphatic vessels.

CD103 is an α-chain of the integrin αEβ7. It is upregulated in activated peripheral CD8⁺ T lymphocytes upon exposure to TGFβ (51). CD103 is a receptor for E-cadherin, an adherens junction protein interlocking epithelial cells (52). Interestingly, E-cadherin is widely expressed by hepatocytes and cholangiocytes (36, 53, 54).The interaction of E-cadherin and CD103 expressing on the liver-infiltrating lymphocytes may be involved in positioning, adhesion and retention of hepatic T_RM cells (36). Furthermore, CD103 may define two different functional subsets of T_RM cells in human liver. The CD69⁺CD103⁺ subpopulations are antigen-specific autoreactive cytotoxic T cells in human liver, exhibiting more potent effector function than CD69⁺CD103^- counterparts (45, 55, 56). Interestingly, there are differences between mouse and human liver T_RM cells regarding CD103 expression. Interestingly, it appears that another liver-specific homing marker, lymphocyte function associated antigen 1 (LFA-1), rather than CD103, may be responsible for the retention of hepatic T_RM cells in mice (16, 57).

CD49a, another adhesion molecule of T_RM cells, is the α1 component of the integrin α1β1. CD49a pairs with integrin β1 to form the heterodimer VLA-1 which bind to collagen IV. This interaction is believed to be critical for retention of the resident population at the epithelium (58). In general, CD49a is upregulated following T cell activation and can be found on circulating T cells (59). Expression of CD49a contributes to protect cells from undergoing apoptosis (60). Importantly, blockade of CD49a with antibodies as well as genetic deletion of CD49a results in a diminution of T_RM cells (59, 61). However, CD49a was not essential for the recruitment of CD8 T cells to the lung in mice, but for their persistence as memory cells (59). Therefore, CD49a may promote the survival, retention or proliferation of T_RM cells. Moreover, CD49a may define different functional subsets of T_RM cells. In the skin, CD49a expressing CD8⁺ T_RM cells produce large amounts of IFN-γ, perforin and granzyme B, while CD49a negative counterparts prefer to produce IL17 (62). However, the effector function bias based on CD49a expression of liver T_RM cells have not been comprehensively interrogated.

Chemokines and chemokine receptors have been extensively used to describe the correct localization, residence and effector function of immune cells within lymphoid organs and non-lymphoid tissues (63). Despite their expressions on T_RM cells of different tissues have great heterogeneity, it is reported that the maintenance and effector function of T_RM require constant chemokine stimulation (64–66). Chemokine receptors CXCR3 and CXCR6 have been extensively reported to be constitutively expressed on the surface of intrahepatic T_RM cells (16, 17, 23, 46, 67). CXCR3 is a vital homing marker that may contributes to the retention of liver CD8⁺ T_RM cells. It binds to multiple chemokines, such as CXCL9, CXCL10 and CXCL11, which are predominantly secreted by monocytes, liver sinusoidal endothelial cells and fibroblasts (17). On the other hand, CXCR6 also plays an important role in the maintenance of liver T_RM cells (46, 68). CD8⁺ T cells lacking CXCR6 migrate to the liver normally after immunization, whereas perform a marked decrease capacity to form hepatic CD8⁺ T_RM cells and severely impairs their effector functions against infection in the liver (46). In addition, CXCR6 also contributes to the maintenance of liver T_RM cells via binding to CXCL16 secreted by liver sinusoidal endothelial cells (46, 68). These studies suggest that CXCR6 is essential for retention rather than recruitment of CD8⁺T cells to the liver. Additionally, deficiency of CXCR6 results in decreased survival of hepatic NKT cells patrolling the liver sinusoids, affecting hepatic intravascular immune surveillance (68).

Besides surface markers, multiple transcription factors are involved in the regulation of the distinct features of liver T_RM cells.

The network of transcription factors underlies the unique features of T_RM cells, including liver T_RM cells ( Figure 1 ). These transcription factors include B lymphocyte-induced maturation protein 1 (BLIMP1; also known as PRDM1), homologue of BLIMP1 in T cells (HOBIT; also known as ZFP683), runt-related transcription factor 3 (RUNX3), Notch, Peroxisome proliferator-activated receptor-γ (PPAR-γ), BHlHe40, TBX21(T-bet), aryl hydrocarbon receptor (AhR), Eomesodermin (EOMES), and NR4A family of orphan nuclear receptors (NR4As). The combined action of these transcription factors contributes to the residency status of liver T_RM cells (64, 69).

HOBIT is specifically up-regulated in T_RM cells and, together with related Blimp1, mediates the development of T_RM cells in lymphoid organs and non-lymphoid tissues (70). The co-expression of HOBIT and BLIMP1 instructs the downregulation of CCR7, transcription factor 7 (TCF7), KLF2, and S1PR1 in T_RM cells (71). CCR7 is the receptor for chemokine ligand 19 (CCL19) and chemokine ligand 21 (CCL21) that responsible for cell migration to secondary lymphoid tissues (72).Meanwhile, TCF7, KLF2 and S1PR1 are involved in the tissue egression of lymphocytes (71). Interestingly, KLF2 regulates the expression of S1PR1 in lymphocytes of tissues, which directs them returning to circulation (44). Consequently, the Hobit-Blimp1 transcriptional module retains T_RM cells within tissues through silencing the genes related to recirculation in addition to suppressing the markers related to egression. Furthermore, a murine study demonstrated that the transcriptional repressor Capicua (CIC) controls the development of liver T_RM cells. Mechanistically, they found that CIC could regulate the expression of HOBIT by inhibiting the ETS variant transcription factor 5 (ETV5) (73). RUNX3 and Notch are essential for the maintenance of T_RM cells by repressing the expression of genes involved in the formation of circulating memory T cells and inducing the expression of retention molecules, including CD103 (74). The collaboration of HOBIT, BLIMP1 and RUNX3 also drives immediate effector function in T_RM cells by inducing and sustaining granzyme B production (75–77). Notch, predominantly expressed in newly developed T_RM cells, not only regulates expression of IFN-γ upon restimulation but also contributes to the mitochondrial fatty acid β-oxidation (FAO) in T_RM cells (74, 75). Importantly, exogenous free fatty acids uptake and their FAO are required for the survival and effector function of T_RM cells (78). Meanwhile, PPAR-γ facilitate the uptake of free fatty acids by upregulating fatty acid binding proteins 1 and 4 (FABP1 and FABP4) in T_RM cells (78, 79). Bhlhe40, a stress-responsive protein, promotes the survival and function of T_RM cells under stress conditions by sustaining mitochondrial fitness (80).

T-bet is crucial to sustain the expression of the IL15 receptor β subunit (IL15Rβ) and therefore enable the long-term lineage stability of T_RM cells, albeit at low levels (81). Activation of aryl hydrocarbon receptor (AhR) may be associated to the maintenance of liver T_RM cells by increasing the expression of CD69 (82, 83).Recent studies reveal that EOMES directly inhibited expression of IFN-γ in vitro, while EOMES deletion in T cells led to substantially increased frequency and percentage of T_RM precursor in the liver (84, 85). Therefore, the downregulation of EOMES in T_RM cells is required to not only their formation, but also their effector function.

Additionally, the NR4As are composed of NR4A1 (Nur77), NR4A2 (Nurr1), and NR4A3 (Nor1). During the memory phase of influenza infection, Nur77 deficiency in CD8⁺ T cells reduces the frequency of CD8⁺ T_RM cells in the liver without any effect on lung or bone marrow CD8⁺ T_RM cells and other memory CD8⁺ T cells such as T_CM and T_EM (86), indicating a specific role of Nur77 on liver T_RM cell differentiation. In addition, the expression of the transcription factors involving in T_RM differentiation (BLIMP1 and T-bet) is decreased, while the expression of EOMES is increased in absence of Nor1 in CD8⁺ T cells (87). Interestingly, NR4As are particularly enriched in the highly functional CD28⁺ subset of CD8⁺ T_RM cells. Importantly, deficiency of Nurr1 specifically reduces the percentage of these CD28⁺ T_RM subsets (88). To conclude, NR4As are important regulators involved in the differentiation of CD8⁺ T_RM cells. However, not all NR4As are comprehensively interrogated at the specific differentiation steps of CD8⁺T_RM cells. Therefore, figure out which signals promote the expression of NR4As in addition the role of NR4As in CD8⁺ T_RM cell differentiation await further investigation.

Although these transcription factors described above have been shown to be critical for T_RM cells, it is difficult to determine which are the specific key regulators of T_RM differentiation and maintenance, as they are also expressed in other CD8⁺ effector or memory subsets. Therefore, the differentiation and maintenance of T_RM may be regulated by the cooperation of multiple transcription factors.

Similar to other tissue T_RM cells, liver T_RM cells also have timely, potent and durable effector functions. When pathogens enter the liver, T_RM cells can take advantage of tissue residency to generate a rapid and effective protective immune response by secreting multiple chemokines and cytokines in a deployment-ready mode (75). The cytotoxic cytokines enable them to directly eliminate infected or malignant cells as well as control invading pathogens, while chemokines and pro-inflammatory cytokines recruit and activate other immune cells, thereby remodeling the local liver microenvironment for more potent effector functions. Furthermore, liver CD8⁺ T_RM cells express high levels of Ki-67 and TCF1, showing their proliferative and self-renewal potential (89). Actually, T_RM cells can persist in the liver for years and exert durable protective effect (17). In addition, T_RM cells may help to significantly promote the repopulation of locally resident and circulating memory T cells after infection, suggesting their role in establishing secondary memory T cells to prevent future reinfection with the same pathogen (90, 91). Accordingly, T_RM cells have been used to develop vaccines that generate stronger and longer-lasting immune responses than conventional vaccines (15, 28, 29). Meanwhile, CD8⁺ T_RM cells are able to attract hepatic stellate cells (HSCs) in a CCR5-dependent manner and predispose activated HSCs to FasL-Fas-mediated apoptosis, thereby promoting liver fibrosis regression (39). However, every coin has two sides, as do liver T_RM cells. Once T_RM cells are interfered by cognate antigens and damage hepatocytes and cholangiocytes, it may lead to the occurrence of AILD. Meanwhile, auto-aggressive liver CXCR6⁺CD8⁺ T_RM cells cause hepatic immune pathology in NASH in an MHC-class-I-independent manner (47). Therefore, clarifying the biological characteristics and development of liver T_RM cells so as to accurately manipulate liver T_RM cells can enhance the effector functions of T_RM cells and avoid weaknesses.

Hepatoviral infection is mainly caused by the hepatitis B (HBV) and hepatitis C (HCV) viruses and the course can be acute or chronic. Chronic infection with hepatotropic virus can cause liver damage, cirrhosis, liver failure, development of HCC, and even liver transplantation. It has been demonstrated that hepatic T_RM cells play a major antiviral immune response during chronic hepatic virus infections.

Pallett, J et al. were the first to report the virus-specific liver CD8⁺ T cells in chronic HBV infection, in which approximately 90% of them have a T_RM cell-like phenotype (CD69⁺CD103⁺ or CD69⁺CD103⁻) (17). CD8⁺ T_RM cells can persist in the liver for several years after primary infection and expand in patients with HBV. Importantly, virus-specific CD8⁺ T_RM cells could still be detected in spontaneously recovered HBV patients, with effector functions equivalent to those from chronic HBV-infected patients (18), suggesting the long-term viral control of hepatic CD8⁺ T_RM cells. Virus-specific CD8⁺ T_RM are very efficient in their function. During HBV viral infection, PD-L1 expression is upregulated in hepatic sinusoidal endothelial cells and hepatocytes (118). PD-L1 on intrahepatic cells can interact with PD1 on T_RM cells, thereby dampening pro-inflammatory T_RM cell responses (19). Nevertheless, even though T_RM cells express high levels of the PD1, they readily produce IFN-γ, TNF-α, perforin, and IL2 upon stimulation (17). IFN-γ and TNF-α mediated control of HBV replication, while perforin may contribute to the directly elimination of infected hepatocytes (20, 21). Furthermore, IL2 production is most strikingly enhanced within CD69⁺CD103⁺ T_RM cells, which contributes to overcome PD-L1-mediated inhibition and exhaustion, stressing their ability for survival and maintenance (21, 119). Additionally, CD8⁺ T_RM cells are enriched in HBV patients who achieved viral control, and their abundance is inversely correlated with HBV viral load, stressing that the virus-specific liver T_RM cells can control viral replication and contribute to the functional cure for HBV patients (17, 22). Therefore, liver T_RM cell expansion may be a potential therapeutic target for chronic HBV infection.

Additionally, a portion of HBV patients are co-infected with hepatitis D virus (HDV), which often indicates a poor prognosis. As the smallest known human virus, HDV has perfectly adapted to escape recognition by CD8⁺ T cells restricted by common human leukocyte antigen (HLA) class I alleles (120). A recent study suggested that antigen-nonspecific activation of hepatic CD8⁺ T_RM cells may be involved in intrahepatic inflammation and disease progression in HDV infection (121).

CD8⁺ T_RM cells also play an essential role in long-term antiviral response in chronic HCV infection (23–25). In the chimpanzee model of HCV reinfection, depletion of CD8⁺ T cells resulted in prolonged the virus persistence and prevented effective viral clearance, while recovery of CD8⁺ T cells lead to virus eradication (26). Meanwhile, a large number of CD69⁺CD8⁺ T cells were detected in the liver of animals recovered after HCV infection, but not in the peripheral blood. These subsets may be hepatic T_RM cells, which are required for protection from persistent HCV Infection (26). Consistently, liver CD8⁺ T_RM cells are highly increased in chronic HCV patients and possess a specific activation and cytolytic potential and are important in controlling chronic HCV infection (27).

Besides hepatotropic virus infection, liver CD8⁺ T_RM cells contribute to the effective clearance of Lymphocytic choriomeningitis virus (LCMV) as well. In the murine model of LCMV infection, virus-specific T_RM cells in the liver could be influenced by other liver-resident immune cells. For example, deficiency of liver-resident natural killer (LrNK) cells increased both the frequency and antiviral activity of hepatic T_RM cells via the interaction of PD1 and PD-L1. Consistently, transfer of LrNK cells into LrNK-cell-deficient mice as well as PD-L1 inhibition restrain hepatic T_RM cell function, resulting in impaired viral clearance (122). Furthermore, during LCMV infection, other liver-resident T cells, such as γδ T cells, also expand and promote viral clearance by producing IFN-γ and TNF-α (123).

Current studies suggest that hepatic T_RM cells may be involved in the clearance of viral infection, protect patients from persistent viral infection, and improve disease prognosis. However, the role of TRM cells in different viral infections in the liver remains to be further elucidated.

Besides viral infections, several studies have investigated the role of liver T_RM cells in parasitic infections, including malaria and leishmaniasis.

Malaria is an insect-borne infectious disease caused by the infection of Plasmodium through the bite of Anopheles mosquitoes or the transfusion of the blood of a person carrying Plasmodium (124). Plasmodium has a complex life cycle, including three stages in the liver, blood and mosquito. During infection of malaria, Plasmodium promotes the development of antigens-specific T_RM cells (16, 125–127). These T_RM cells could mediate protective immune responses through killing infected cells by producing pro-inflammatory cytokines, such as IFN-γ and TNF-α (16, 30). Additionally, T_RM cell depletion abrogated an efficient immune response to a murine model of Plasmodium infection (31).Due to the protective immune response of T_RM cells against malaria, vaccination strategies that maximize intrahepatic Plasmodium-specific T_RM development have emerged (16, 28, 29, 32–34, 127). An example is the Plasmodium ribosomal protein vaccine (15). One of the antigens for this vaccine is PbRPL6_120-127, a highly conserved H2-K^b-restricted epitope from the 60S ribosomal protein L6, expressed throughout the parasite life cycle, across Plasmodium species (15). It may be an optimal antigen for endogenous liver T_RM development and protection against malaria. A single dose of this vaccine could provide effective and prolonged sterilizing immunity against high dose sporozoite challenges (15). Indeed, people living in malaria-endemic areas do not acquire effective protection against reinfection from malaria (128), while attenuated Plasmodium falciparum sporozoite (SPZ) vaccine is highly protective against controlled human malaria infection 3 weeks after immunization (129), suggesting multiple, complex factors are likely responsible for the lack of development of sterilizing immunity to malaria through natural infection. Furthermore, the protection and long-term efficacy of existing vaccines are not satisfactory. Accordingly, to improve the T_RM-based vaccination against malaria in human, further investigation of the mechanisms that mediate Plasmodium-specific T_RM generation and function, assessment of the feasibility of currently known antigens, as well as identification of novel target epitopes are required.

Recently, the role of T_RM cells in Leishmaniasis was studied as well. Leishmaniasis is a zoonotic disease caused by Leishmania, which can cause cutaneous and visceral kala-azar in humans (130). There are various types of Leishmania in which Leishmania infantum (L. infantum) primarily infects the liver (131–133). During chronic L. infantum infection, liver T_RM cells are generated and play a protective role. Importantly, induction by the Leishmania proteins LirCyP1 and LirSOD promotes the expansion of hepatic T_RM cells, which could be a promising strategy for prophylactic or therapeutic vaccine formulations (131).

Taken together, hepatic T_RM cells are critical in parasitic infections, and the T_RM-based vaccination strategies could hold remarkable promise in providing long-term protection.

AILD is a group of liver inflammatory damage diseases mediated by abnormal autoimmunity, including autoimmune hepatitis (AIH), primary biliary cholangitis (PBC), primary sclerosing cholangitis (PSC), IgG4-related sclerosing cholangitis (IgG4-SC), etc. AIH is an inflammatory liver disease dominated by T cell-mediated hepatocyte injury. Antigen-specific CD8⁺ T_RM cells have been reported to characterize the liver tissue of subjects with indeterminate pediatric acute liver failure (PALF) and may serve as a novel biomarker for PALF due to AIH (37, 38). Recently, our group demonstrated that CD69⁺CD103⁺CD8⁺ T_RM cells play an important role in the pathogenesis of AIH, and histological remission is accompanied by decreased hepatic CD8⁺ T_RM cells in AIH patients (36). In addition, hepatic CD8⁺ T_RM cells from AIH patients expressed a higher level of PD-1, CXCR3 and granzyme B than those of healthy controls. Consistently, in AIH liver, both expression of IL15 and TGFβ, cytokines that induce T_RM cells in vitro, were elevated, suggesting that the immunological microenvironment facilitates hepatic CD8⁺ T_RM cells development and residency (36). Intriguingly, E-cadherin, the natural ligand of CD103, is widely expressed in hepatocytes of AIH patients, and located closely to CD8⁺ T_RM cells, which may contribute to the residency of CD8⁺ T_RM cells in the liver. Furthermore, E-cadherin is also widely expressed in cholangiocytes (53, 54), suggesting that CD103⁺ T_RM cells may be involved in pathology of bile duct injury in cholestatic liver diseases, such as PBC and PSC. Interestingly, a recent study on biliary immune atlas revealed the presence of CD8⁺ T_RM cells in areas of biliary inflammation in PSC patients (134).

Nonalcoholic fatty liver disease (NAFLD) is considered a hepatic manifestation of metabolic syndrome, hypertension and type 2 diabetes. Several studies have demonstrated that liver-resident T cells and the proinflammatory immune response they elicit are involved in NAFLD disease progression (135–138). Generally, liver-resident γδT cells induce chronic liver inflammation by producing proinflammatory cytokines such as IL17A, IFN-γ, and TNF-α, contributing to the pathogenic immune response to NAFLD (123, 137, 139). Furthermore, systemic inflammation in obese patients is associated with increased T_RM cells in the liver and may be further involved in NAFLD disease progression. Importantly, activated T_RM cells are significantly increased in the liver and visceral fat of obese patients. These activated T_RM cells produce multiple pro-inflammatory cytokines, such as IL1β, IL2, IL12, and IL15 (140), further contributing to the generation of T_RM cells in addition to the overall pro-inflammatory phenotype in obese patients.

Interestingly, a recent study revealed that CD69⁺CD103^-CD8⁺ T_RM cell may perform a protective role in resolving liver fibrosis of nonalcoholic steatohepatitis (NASH) (39). They demonstrated that the reduction of these CD8⁺ T_RM cells significantly delayed fibrosis resolution via influencing predisposed HSCs apoptosis, while adoptive transfer of these cells protected mice from fibrosis progression in a CCR5-dependent manner (39). Therefore, the paradoxical roles of T_RM cells in NAFLD and their specific mechanisms remain to be further investigated.

HCC accounts for the majority of primary liver cancers and is currently one of the leading causes of cancer-related deaths worldwide. The development of HCC is a complex multistep process caused by multiple risk factors, whereas the function of tumor-infiltrating T cells is important for moderating antitumor immunity in HCC development and determining the clinical fate of HCC patients (40). There are strong evidences that CD103⁺ T_RM cells are enriched in HCC patients and associated with better prognosis (19, 41, 42).

In murine model of HCC, hepatic T_RM cells were significantly expanded, and their frequencies decreased during HCC progression (141). Meanwhile, hepatic T_RM cells in HCC have an exhausted phenotype, manifested by expression of PD1, LAG3, and TIM3 (40). Given that PD1 expression in T_RM cells in HCC is associated with poor disease outcome (142), immunotherapy targeting checkpoint inhibition has been applied to HCC (143, 144). During immunotherapy for HCC, PD1^high T_RM cells are the most sensitive cells to anti-PD-1 therapy to overcome tumor growth and progression (145). Additionally, other markers of exhaustion and inhibition, such as TIM3 and CTLA4, and pro-inflammatory cytokines, such as IFN-γ and TNF-α, can also be simultaneously expressed on T_RM cells in HCC patients (142), suggesting that hepatic T_RM cells may be involved in direct killing of tumor cells. Overall, hepatic T_RM cells might play an extremely important role in both HCC development and anti-tumor therapy.

Liver transplantation is the treatment of last option for end-stage liver disease of various causes and severe acute liver failure. It has been reported that donor-derived T_RM cells are detectable in the liver allografts and that their abundancy could be correlated with organ survival and reduced rejection (146–148). Specifically, long-term persistence of lung donor-derived T_RM cell is associated with reduced incidence of clinical events that precipitate allograft injury, including primary graft dysfunction (PGD) and acute cellular rejection (ACR) (149). However, the association of liver donor-derived T_RM cells with the incidence of clinical events remains to be further elucidated (150). In liver allograft tissues, approximately 2-6% of CD8⁺ T cells had a donor-derived T_RM phenotype at 11 years post-transplantation (18), well demonstrating the longevity of human liver T_RM cells. Additionally, donor-derived T_RM cells from an HBV-infected liver allograft could migrate to draining lymph nodes with down-regulation of some T_RM-specific markers. However, they were not detectable in blood vessels (18). Interestingly, the same study demonstrated that a lower quantity of recipient-derived virus-specific T cells with a T_RM-like phenotype were detected in the liver and blood (18), further revealing the extrahepatic origin of T_RM cells in the liver. Nevertheless, CMV-specific T_RM cells in human liver allografts did not acquire a T_RM phenotype in the liver, possibly due to the lack of relevant antigens in the liver.