Egress of resident memory T cells from tissue with neoadjuvant immunotherapy: implications for systemic anti-tumor immunity

Abstract

Introduction:

Resident memory T (T_RM) cells are embedded in peripheral tissue and capable of acting as sentinels that can respond quickly to repeat pathogen exposure as part of an endogenous anti-microbial immune response. Recent evidence suggests that chronic antigen exposure and other microenvironment cues may promote the development of T_RM cells within solid tumors as well, and that this T_RM phenotype can sequester tumor-specific T cells into tumors and out of circulation resulting in limited systemic antitumor immunity. Here, we perform a review of the published English literature and describe tissue-specific mediators of T_RM cell differentiation in states of infection and malignancy with special focus on the role of TGF-β and how targeting TGF-β signaling could be used as a therapeutical approach to promote tumor systemic immunity.

Discussion:

The presence of T_RM cells with antigen specificity to neoepitopes in tumors associates with positive clinical prognosis and greater responsiveness to immunotherapy. Recent evidence indicates that solid tumors may act as reservoirs for tumor specific T_RM cells and limit their circulation – possibly resulting in impaired systemic antitumor immunity. T_RM cells utilize specific mechanisms to egress from peripheral tissues into circulation and other peripheral sites, and emerging evidence indicates that immunotherapeutic approaches may initiate these processes and increase systemic antitumor immunity.

Conclusions:

Reversing tumor sequestration of tumor-specific T cells prior to surgical removal or radiation of tumor may increase systemic antitumor immunity. This finding may underlie the improved recurrence free survival observed with neoadjuvant immunotherapy in clinical trials.

Introduction

One of the hallmarks of adaptive immunity is the capacity for memory, which enables rapid clearance of pathogens upon re-exposure to the initial antigen¹. T cells that remain after being licensed by the primary antigen exposure exist as central, effector, and resident memory T cell subsets (T_CM, T_EM, and T_RM, respectively) and occupy different anatomic spaces where they fulfill distinct roles in protective immunity^2,3. T_RM cells are classically described as non-recirculating memory T cells that remain positioned at common sites of re-exposure, including barrier tissues such as the skin⁴ and mucosa⁵. In these barrier tissues, abundant CD4 and CD8 T_RM clonotypes that target a variety of antigens can be found^6,7. These tissue-retained T cells are distinguished by expression of the integrins CD69^8,9 and CD103^10,11. CD69 is a membrane-bound type II C-lectin receptor that acts primarily via S1PR1 (Sphingosine-1-Phosphate Receptor 1) to promote tissue residency^8,9. S1PR1 receptor signaling overrides retention mediated by G alpha i-coupled receptors. CD69, whose expression is rapidly induced on the surface of T lymphocytes following T cell activation, binds the S1PR1 receptor to induce receptor activation and internalization without lipid ligands, thus promoting retention^12,13. Integrin, alpha E (ITGAE), also known as CD103, is highly expressed at mucosal sites and binding to E-cadherin promotes retention in peripheral tissues¹⁴. Other markers such as CD49a^15,16 and CD44^17,18 have also been extensively explored to distinguish T_RM cells from other T cell subsets, both in mice and humans. CD44 is a type I transmembrane glycoprotein that has long been used to differentiate memory and effector T cells from their naïve counterparts¹⁹. CD49d/α4-integrin is selectively expressed in mucosal endothelium and mediates both cellular rolling and firm adhesion by binding MCAM-1, thus functioning as a T cell retention receptor in mucosal lymphoid tissues^20,21. Local tissue environments supply a diverse array of cytokines^22,23 that mediate expression of distinct chemokine receptors such as CCR8 in the skin²⁴ and CXCR6 in respiratory epithelium²⁵ that contribute to maintenance of a tissue residence phenotype in T cells. T_RM cells undergo a distinct differentiation program that discriminates them from circulating T_CM cells and T_EM cells^4,26,27 even when these subsets share a common progenitor, as evidenced by T cell receptor (TCR) repertoire overlap in distinct tissue compartments^28,29.

This tissue resident program develops to populate barrier tissues such as skin and mucosa with highly protective T cells specific against the pathogens most commonly present in those environments. Importantly, similar T_RM cell phenotypes have also been observed in solid tumors and found to be enriched in T cell clonotypes specific for tumor antigens³⁰. Therefore, a better understanding of the cues that drive the development of a T_RM cell phenotype amongst tumor-infiltrating T cells and the implications of such programs on the compartmentalization of anti-tumor immunity is needed. For example, if development of a T_RM cell phenotype occurs upon chronic TCR stimulation and cytokine cues in tumors such that the tumor acts as a reservoir for a significant proportion of the tumor specific T cells in a patient, definitive surgical removal or radiation ablation treatment may remove most of the patient’s antitumor immunity. Here we discuss the environmental factors and transcriptional programming steps that drive the differentiation of naïve T cells to a T_RM cell phenotype, with a specific focus on CD8⁺ T cells and the role of the multi-functional cytokine transforming growth factor-beta (TGF-β), summarized in Figure 1. We additionally discuss existing evidence suggesting that tumor specific T cell tumor residency can be therapeutically manipulated, and the clinical implications of these observations.

Figure 1: Mechanisms of development and maintenance of T_RM cells.

Tissue-specific dendritic cells (DCs) migrate to the draining lymph nodes (dLN) where they present antigens to naïve T cells. Once primed, these T cells have the capacity to become CD8⁺ T_RM precursors and migrate to the injured tissue, following a chemotactic gradient. In the tissue, exposure to transforming growth factor beta (TGF-β) initiates the resident memory program, namely by driving the expression of the integrin alpha E (ITGAE/CD103) that binds to E-cadherin expressed by the tissue, enforcing retention. TGF-β is also important in driving the general transcriptional profile of T_RM cells, which requires the expression of Hobit, Blimp-1, Runx3 and Notch and downregulation of Tcf-1, Eomes, T-bet and Klf2. T_RM cells are often characterized by the expression of immune checkpoint receptors associated with T cell exhaustion such as Lag-3, TIM-3, CTLA-4, and PD-1.

Review Methods

Literature was identified by searching PubMed for “resident memory T cell” in combination with terms including differentiation, immunotherapy, cancer, and recirculation, searching publication dates from March 1993 to March 2023. Both pre-clinical and clinical reports were included. The relative contribution of reach report was described within the context of the larger field along with additional questions that it raised. One reviewer independently screened each report and synthesized findings, which were verified by all authors.

Discussion

TGF-β and other mediators of T_RM differentiation

The tissue resident memory programming may be initiated in draining lymph nodes (dLN) when dendritic cells (DCs) present naïve T cells with antigens from the initial site of exposure³¹. Before migrating out of the dLN and entering circulation, activated T cells undergo extensive proliferation in the days following antigenic exposure³². Yoon and colleagues reported that activated CD8⁺ T cells undergo up to 5–6 divisions within 2–4 days post antigen encounter³². More recently, Kurd et al. performed a time course experiment in LCMV-infected mice and, using single-cell RNA-sequencing (scRNA-seq) to track individual T cell clonotypes over time and location, were able to identify rapid induction of a T_RM signature in T cells that traffic into peripheral tissues³³. There is some evidence, although, that CD8⁺ T cells can be preconditioned in the dLN for T_RM differentiation upon interaction with integrins on migratory DCs³⁴. Another study tracing lineage patterns of adoptively transferred barcode-labeled ovalbumin-specific transgenic CD8⁺ T (OT-I) cells in immunized mice found a subset of antigen-specific CD8⁺ effector T cells that began expressing a T_RM transcriptional signature while still in the dLN³⁵. This demonstrates the potential for CD8⁺ T cells to commit to a memory and possibly tissue resident fate prior to peripheral tissue entry. Acquisition of a T_RM cell phenotype in a subset of CD8⁺ T cells upon entry into peripheral tissues is well established, but whether commitment to T_RM cell fate necessarily begins in the dLN priming phase requires further investigation.

The T_RM program in peripheral tissues is promoted by engagement with chemokine receptors that drive expression of adhesion molecules like integrins^36–39. After initial chemotaxis to tissues mediated at least by the chemokines CXCL9 and CXCL10⁴⁰, T cell tissue entry begins when signals such as IL-1⁴¹, TNFα^41–43, LPS⁴⁴, and IL-4⁴² promote expression of the cell adhesion molecules E-selectin and P-selectin on endothelial cells at site of active inflammation that facilitate rolling and firm adhesion within vessels. Following T cell extravasation through a process that requires at least CD44⁴⁵, T cells within tissues are exposed to TGF-β secreted by fibroblasts, epithelial cells, and leukocytes^46,47. TGF-β promotes tissue residency by upregulating adhesion molecules including the integrins CD103^14,48,49, αEβ7, α1, CD69⁴⁸, αvβ6, αvβ8⁵⁰, and CD49a¹⁶. Interestingly, IL-12, which has also been reported to induce CD49a in vitro¹⁶, negatively affects the expression of CD103^51,52. This might help explain why IL-12, together with IFN-β, in highly proinflammatory microenvironments promotes the differentiation of a subset of T_RM cells that are phenotypically CD103⁻CD69⁺²³. CD103⁻ CD8 T_RM cells develop in the intestinal LP where they cluster in areas of infection together with other inflammatory immune cells such as CX3CR1⁺ macrophages and dendritic cells²². Of note, CD103⁻ T_RM cells seem to display distinct functional capabilities and tissue specificities when compared with their CD103⁺ counterparts. CD103⁻ T_RM cells display a transcriptional profile similar to circulating T cells and display higher migratory potential, making them the primary responders in sites of secondary infection⁵³. Consequently, CD103⁻ T_RM cells exhibit more plasticity, being capable of modifying their phenotype following migration, whereas CD103⁺ T_RM cells may be more resistant to transdifferentiation⁵⁴.

Recent work has also highlighted the importance of TGF-β signaling in maintaining a subset of stem-like CD8⁺ T cells that is crucial for long-term T cell response in tissues^55–57. These cells, termed precursor of exhausted T cells (T_PEX), have a high proliferative capacity in lymphoid tissues where its progeny is committed to an exhausted T cell fate. Although they share functional similarities with CD8⁺ memory T cells formed during acute infection, chronic antigen exposure enforces a distinct transcriptional and epigenetic program that characterizes this population as TCF-1⁺PD-1⁺CXCR5⁺Tim-3^lo58–61. In a model of chronic LCMV infection, parabiosed mice were used to track virus-specific CD8⁺ T cells, leading to the observation that stem-like T_PEX CD8⁺ cells proliferated in LNs to give rise to terminally differentiated progeny that formed a resident memory population. Unlike in the acute setting, there was minimal trafficking of stem-like cells between mice, which further supports the concept of acquired residency⁶². Using a mouse model of melanoma, Li et al demonstrated that TGF-β signaling was the main driver of differentiation of tumor-specific stem-like CD8⁺ T cells into T_RM phenotype in the tumor dLN. Furthermore, the authors were able to show that stem-like CD8⁺ T cells also differentiate into T_RM-like cells in dLN of human head and neck squamous cell carcinoma (HNSCC) patients⁵⁶. These TCF1⁺ tumor specific CD8⁺ T cells seem to maintain antitumor responses by giving rise to effector cells that take up residency in tumors^63,64. They expand in the dLN and traffic to the tumor⁶³ or proliferate in regions of the tumor that are enriched with antigen presenting cells (APCs)⁶⁴, processes that involve TGF-β-mediated upregulation of CD103. Clearly TGF-β plays a crucial role in the development of T_RM phenotype once activated T cells enter peripheral tissues, such as skin and mucosa.

T_RM transcriptional program

The transcriptional signature of T_RM cells was initially established through adoptive transfer of gBT-I cells, which express a transgenic TCR specific for herpes simplex virus (HSV), into HSV-immunized mice. Microarray analysis of CD103⁺ gBT-I cells isolated from skin epithelium revealed that T_RM cells develop from KLGR1⁻ precursors that upregulate CXCR3²⁷. KLRG1 (Killer Lectin-like Receptor G1) is a co-inhibitory receptor that has been used to identify antigen-experienced T cells and to distinguish memory precursor cells from effector T cells⁶⁵. During effector differentiation, CD8 T cells gain expression of KLRG1. KLRG1 is then downregulated as cells further differentiate into memory T cell lineages, including T_RM cells, a process mediated by the transcriptional repressor Bach2. In the gBT-I model, after entering the epidermis, KLGR1⁻ precursors are signaled by TGF-β and IL-15 to enhance retention and survival of T_RM cells by promoting CD103 and Bcl-2 expression, respectably²⁷. A transcriptional analysis comparing circulating T_EM and T_CM cells to different T_RM subsets revealed that T_RM cells selectively upregulate the expression of the aryl hydrocarbon receptor (Ahr) which acts as a promoter of T_RM cell differentiation and function⁶⁶, the G protein signaling genes RGS1 and RGS2 that modulate T cell trafficking⁶⁷, and the receptor of the immunoglobulin (Ig) superfamily CD244 (SLAMF4; 2B4), known to increase tissue infiltration of not only CD8⁺ T cells but also dendritic cells (DCs) and myeloid-derived suppressor cells (MDSCs)⁶⁸. Inversely, the authors found that the transcriptional signature was characterized by the downregulation of S1PR1 and Fam65b⁶⁹, an inhibitor of the small GTPase RHOA⁷⁰. Interestingly, expression of both S1PR1⁷¹ and Fam65b⁶⁹ has been shown to be negatively affected by TGF-β signaling. As mentioned, at the protein level, surface expression of S1PR1 is disabled upon forming a complex with CD69^13,72. In HSV-immunized mice, adoptively transferred gBT-I cells with deficient CD69 expression (CD69^−/−) failed to maintain tissue residency to the same extent as wildtype cells, but upon treatment with the S1P receptor agonist, FTY720, accumulation of CD69^−/− cells in infected skin was restored¹³. Transcriptionally, CD8 T_RM cells in the parenchyma of non-lymphoid tissues (NLTs) have also been shown to lack expression of the transcription factor KLF2 and its target S1pr1⁷³. These two different mechanisms highlight the importance of S1PR1 downregulation for tissue retention and maintenance of long-term immunity in peripheral tissues.

Although the transcriptomic signature of T_RM cells is heterogenous across tissues⁷⁴, several transcription factors have been identified as master regulators of T_RM cell fate. The T-box transcription factors Eomesodermin (Eomes) and T-bet, for instance, have been described as key regulators of T_RM cell differentiation^75,76. Eomes seems to play tissue-specific roles, having been reported to repress T_RM cell formation in the skin, liver, and kidney^75,77, while supporting maintenance of established T_RM cells in the small intestine, partially by inducing the antiapoptotic molecule Bcl-2⁷⁸. T-bet, whose expression is proportional to the strength of the TCR signaling, has been widely described to promote T_E/T_EM formation⁷⁹ and suppress T_RM cell differentiation⁷⁵. Like T-bet, the transcriptional repressor Blimp1 is also upregulated by inflammation and/or TCR triggering. However, contrary to T-bet, Blimp1 works to maintain the T_RM phenotype while also being able to promote the differentiation of T_E/T_EM cells^80–82. T-bet also mediates the expression of the transcription factor Hobit, which is jointly upregulated with Blimp1 to promote T_RM cell differentiation^80,82,83. Hobit has been found to be upregulated in a subset of LCMV-specific CD8⁺ T cells located within peripheral tissues that were identified to be T_RM cell precursors, since depletion of these cells substantially decreased T_RM cell development⁷⁷. Hobit, as Blimp1, was shown to bind and regulate the expression of Tcf7, S1pr1, Klf2 and Ccr7, and loss of these transcription factors might enable T_RM cells to leave the tissue and re-enter circulation⁷⁷. Hobit and Blimp1 are at the core of transcriptional program of tissue residency, mostly working together but with some exceptions, as is the case for the differentiation of CD8+ T_RM cell in the lungs which has been reported to exclusively dependent on Blimp1⁸⁴. KLF2 is a member of the Krüppel-like transcription factor family of proteins that directly controls the expression of CD62L and S1P₁⁸⁵. KLF2 is downregulated during T_RM cell development, leading to decreased expression of S1P₁ and facilitating the retention of these cells in peripheral tissue^11,71,86. More recently, the transcription factor Runx3 has also been shown mediate T_RM cell differentiation and survival^87,88. In a murine melanoma model, adoptively transferred tumor specific CD8⁺ T cells deficient in Runx3 were unable to maintain tumor residency resulting in uncontrolled tumor growth⁸⁸. Interestingly, Runx3 enforces tissue residency of CD8⁺ T cells through the promotion of chromatin accessibility at TGF-β regulated genes⁸⁸. Another regulator of the T_RM cell formation that acts on the TGF-β-driven residency program is the SKI proto-oncogene. SKI negatively regulates TGF-β signaling by directly interacting with Smads and repressing the transcription of TGF-β responsive genes such as CD103⁸⁹.

In addition to these highly conserved pathways, tissue-specific transcriptional regulators of T_RM cells have been identified by comparing gene expression patterns of tissue isolated from different compartments of immunized mice. In the small intestinal epithelium, unique regulators include Nr4a2, Junb, Fosl2³³ and Hic1⁷⁶. In the respiratory epithelium, signaling through the TNF receptor family members CD137 (4–1BB)⁹⁰ and GITR⁹¹ is required for the generation of CD4⁺ and CD8⁺ T_RM cells.

T_RM cells can, therefore, be identified using certain core transcriptional signatures and several cell surface protein markers. However, some circulating T cells also express genes associated with tissue-residency at levels comparable to T cells infiltrating NLTs, including canonical T_RM markers such as CXCR6, CD69 and ITGA1^11,92,93. A key feature that distinguishes T_RM cells in tissue from T_RM-like cells in circulation may be their functional status. Recently, Noé and colleagues investigated liver-infiltrating T_RM cells following malaria prime-target vaccination (PTV) and compared them to CD69⁺ T_RM-like circulating cells and found that both subsets shared expression of conventional-residency markers but differed metabolically and functionally⁹⁴. Specifically, T_RM-like cells in circulation upregulated the zinc transporter SLC39A7 and were metabolically active, while tissue infiltrating T_RM cells were metabolically quiescent but ready for rapid effector function upon activation. As such, although T cells expressing markers associated with tissue residency can be found both in tissue and in circulation, these cell population may be distinguished from each other functionally.

Recirculation of T_RM cells

Although T_RM cells were initially portrayed as T lymphocytes permanently embedded in nonlymphoid peripheral tissue like skin and mucosa, under certain conditions T_RM cells can leave the tissue of residency and recirculate^95,96. Often these circulating T_RM cells are described as ex-T_RM cells and express some EM phenotype markers such as KLRG1 and CX3CR1⁹⁷. In murine models, local restimulation of T_RM cells results in the egress of cells from tissues of residency^97–99. Using a (VSV)-expressing ovalbumin (VSVova) infection model and resorting to skin-grafts, Fonseca et al. demonstrated that upon reactivation of T_RM cells within the previously immunized mice, OT-I cells started accumulating in the dLN and circulating OT-I cells with T_CM and T_EM signatures were observed in distant LNs 2–3 weeks later⁹⁸. This data indicates that T_RM cells are capable of not only tissue egress, but also transdifferentiation into other memory phenotypes in circulation, which is suggestive of developmental plasticity. Behr and colleagues also observed recirculation of tissue-resident OT-I cells that developed after oral infection with Listeria monocytogenes-expressing OVA (Lm-OVA). Hobit expression, which is necessary to enact the T_RM program in T cells within this model, was downregulated following secondary antigen recognition⁹⁷. Consequently, reduced or abrogated Hobit expression following T_RM cell reactivation might be crucial for T_RM cell tissue egress and differentiation into other T cell phenotypes in circulating cells.

Another example of systemic immunity arising from egressed T_RM cells comes from the work of Klicznik et al¹⁰⁰. Using transcriptional profiling of rare cell populations by RNA sequencing (RNA-seq) the authors were able to identify a population of skin T_RM CD4⁺ T cells that recirculate through blood and thoracic duct lymphatics in the steady state. These CD4⁺ T cells express the cutaneous lymphocyte antigen (CLA), characteristic of skin–resident memory T cells, together with CD103, and their exit from tissue was associated with downregulation of CD69. Interestingly, these circulating CD4⁺CLA⁺CD103⁺ T_RM cells were able to reseed distant skin sites, regaining CD69 expression upon reentering the tissue¹⁰⁰. The Klicznik work did not definitively identify the mechanism of egress of T_RM cells from the skin, but a later study proposed that regular tissue-damaging conditions result in the activation of T_RM cells and subsequent emigration. Specifically, this study observed an increased percentage of circulating T_RM cells of skin origin in the blood of patients with graft-versus-host disease (GVHD) and cutaneous infections, implicating local inflammation in re-activation and tissue egress of TRM cells¹⁰¹.

Tissue residency in malignancy

In addition to their role in pathogen defense, cells phenotypically similar to T_RM have been identified within epithelial malignancies (carcinomas) where they constitute an important part of the antitumor immune response. The gene expression signature of subset of CD4+ and CD8+ T cells infiltrating carcinomas parallels that of T_RM cells present in healthy barrier tissue like skin and mucosa³⁰, with these tumor infiltrating lymphocytes (TILs) expressing CD103 and CD69^102,103. However, chronic antigen persistence in the tumor results in a fundamentally different microenvironment compared to the setting of resolution of infection and inflammation where classic T_RMs have been described. Chronic TCR signaling and other features such as TGF-β exposure in tumors drive a transcriptional program in T cells that shares hallmark features of T_RM generated after infection¹⁰⁴. To date, these CD103+ TILs have mostly been described simply as T_RM cells, but recent evidence suggests that they may more accurately be referred to a ‘T_RM-like’ cells^105–107.

Steele and colleagues demonstrated that TCR affinity may play an important role in the accumulation of CD8⁺ T cells in tumors¹⁰⁸. In their study, interaction with high-affinity antigens, but not low-affinity, led to the downregulation CXCR4 and upregulation of the CXCL12 decoy receptor, ACKR3. CXCR4 is a chemokine receptor that binds to CXCL12, also known as SDF-1, which is produced by Bone Marrow (BM) stroma, thus functioning as a homing signal to the BM¹⁰⁹. CXCR4 expression can therefore prevent infiltration of T cells into tumors because the surrounding stroma expresses CXCL12.¹¹⁰ In the YUMMER1.7 mouse melanoma model, reduction in the sensitivity to CXCL12 promoted T cell tumor infiltration and improved tumor control in the context of PD-L1 immunotherapy. Objectively, entry of T cells into tumors seems to be a critical step in the development of an effective T cell response.

Accumulation of T_RM cells in tumors is associated with increased survival^30,103,111 and positive response to immunotherapies, including anti-PD-1 and anti-CTLA-4 immune checkpoint blockade (ICB)^112,113. Immunotherapy strategies such as ICB may reverse an exhausted (and dysfunctional) state in T cells caused by chronic antigen exposure, characterized by the expression of high levels of the inhibitory receptors PD-1, Lag-3, TIM-3, CTLA4, and Tigit^114–117. Indeed, gene expression programs indicating exhaustion and tissue residency appear to be closely associated in tumor specific T cells. Although it is clear that the development of exhaustion and tissue residency gene expression programs occurs downstream of TCR signaling¹¹⁸, additional understanding of the complex similar or dissimilar signals from the tumor microenvironment that drive either program are needed and may help inform the development of novel therapeutics beyond immune checkpoint blockade that modulate one or both programs.

Examples of the importance of T_RM-like cells in response to ICB exist across several cancer types. In a non-small cell lung cancer (NSCLC) patient cohort, Djenidi et al. demonstrated that CD8⁺ tumor-infiltrating lymphocytes (TIL) upregulate genes encoding adhesion molecules including CD69, CD103, RGS1, ICOS1 and downregulate egress mediators S1P₁ and ITGB2. Besides markers of residency, these TILs also acquired the expression of PD-1 and Tim-3³⁰, consistent with the idea of persistent antigen exposure and T cell exhaustion in the tumor microenvironment (TME). Furthermore, the authors observed that the presence of CD103⁺ TILs associated with increased survival of early-stage NSCLC patients and that neutralization of PD-1–PD-L1 interactions improved the specific cytotoxic activity of CD8⁺CD103⁺ T cells against autologous tumor cells³⁰. In early stage primary triple negative breast cancer, enrichment in CD8⁺CD103⁺ T_RM cells has also been associated with improved overall survival after standard chemotherapy¹¹¹. In their study, Savas and colleagues only reported on the contribution of infiltrating T cells to clinical outcomes, but more recently Egelston et al. suggested that this effect may be mediated by enhanced infiltration into the tumor parenchyma rather than the stroma by a subset of CD8⁺CD103⁺ TILs relative to their CD103⁻ counterparts¹¹⁹. Hence, spatial localization of CD8⁺CD103⁺ TILs within cancer islands is an important determinant of patient survival.

In patients with hepatocellular carcinoma, pre-treatment biopsies with a higher proportion of CD8⁺ T_RM cells to exhausted CD8⁺ T (T_EX) cells demonstrated enhanced responsiveness to anti-PD-1 therapy¹¹³. Similarly, in patients with advanced stage melanoma, pre-treatment enrichment of CD8⁺ T_RM cells in the tumor was associated with responsiveness to immune-checkpoint blockade (ICB) and recurrence free survival (RFS). Multiplex immunohistochemistry showed that CD8⁺ T_RM cells localized in closer proximity to melanoma cells compared to other CD8⁺ T cells, supporting the idea that spatial localization matters for cytolytic effects¹¹².

Quantity, however, is not the only determinant of improved prognosis. Quality, particularly diversity of T cell specificity, is also a crucial aspect of T_RM cells association with favorable clinical outcomes. T cell responses depend on antigen recognition and the type of tumor antigens presented by cancer cells may shape the TCR repertoire observed. Notably, evaluation of the TCR repertoire of TILs has been a major focus of predictive work in cancer prognostics^120–122. TCR repertoire analysis of CD8⁺ T cells within the mucosa of patients with gastric cancer revealed a positive association between clonotype diversity and survival¹²³. In patients with pancreatic cancer, repertoire diversity has been linked to enhanced survival following neoadjuvant therapy and surgical resection¹²⁴. The degree of clonal infiltration is not a one-size-fits-all phenomenon, with tumor control being shaped by the functional attributes of tumor-specific TILs. In ovarian cancer, Tsuji et al. found that favorable prognosis was associated with strong monoclonal TCR repertoires¹²⁵. This might be due to the presence of a significant number of bystander TILs that recognize non-cancer antigens¹⁰³. In patients with metastatic melanoma, TCRs with neoantigen specificity are more likely to be identified amongst clonotypes present in higher frequencies¹²⁶. Investigation of the TCR repertoire of CD8⁺ T cells that harbor a T_RM phenotype has thus facilitated the identification of TCR clones that recognize tumor-associated (TAAs) and tumor-specific (TSAs) antigens.

Building systemic immunity against cancer through egress of T_RM cells

As in the setting of infection, recirculated T_RM-like cells may have the potential to confer enhanced protection against distant disease or locoregional disease relapse. Given this potential, mechanisms that regulate T cell trafficking are of interest since enhanced systemic antitumor immunity could be achieved by egress of T_RM cells from the primary tumor or dLN prior to surgical removal or radiation ablation. Initially, responsiveness to ICB was related to early expansion of circulating CD8⁺ T_EM with diverse repertoire clonality^127–130. Recent studies suggest that ICB can also induce the egress of reinvigorated, exhausted, T_RM-like cells. In a study by Luoma and colleagues, tumor-infiltrating CD8⁺ T cells that expressed elevated tissue-resident memory and cytotoxicity signatures clonally expanded during neoadjuvant anti-PD-1 or anti-PD-1/CTLA-4 immunotherapy. Treatment also resulted in the identification of increased frequencies of the same tumor-infiltrating CD8⁺ T cell clonotypes in the blood of the patients after treatment compared to before, demonstrating that ICB therapy may drive egress of tissue-infiltrating CD8⁺ T cells and enhance systemic immunity⁹⁵. More recently, in a neoadjuvant immunotherapy clinical study using dual PD-L1 blockade and TGF-β neutralization, Sievers et al. demonstrated expansion of antigen-specific exhausted and proliferating T cells that harbored markers of residency including CD103⁹⁶. Dual PD-L1 blockade and TGF-β neutralization reduced intratumoral expression of CD103, a TGF-β-induced integrin whose downregulation was validated in vitro to be linked to TGF-β neutralization. This treatment established a pool of neoepitope-specific CD8⁺ T cells that were undetectable in the blood prior to treatment. Whether tumor specific T cells identified at greater frequency in circulation after treatment with ICB egress form the primary tumor, tumor dLN or both, requires further study. Concomitantly, whether TGF-β neutralization is required in addition to ICB to facilitate efficient egress of T_RM cells from peripheral tissues into circulation or whether this can be achieved with ICB alone also requires further study. Figure 2 illustrates the concept of using immunotherapy to induce recirculation of T_RM-like cells in patients with cancer prior to surgical removal.

Figure 2: Neoadjuvant therapy as a strategy to induce recirculation of T_RM-like cells and systemic immunity against cancer.

Through expression of tissue resident markers such as CD103, exhausted, dysfunctional CD8⁺ T cells reside in tumor tissue. PD-1 targeting immune checkpoint blockade with or without TGF- β blockade enhance egress of CD8⁺ T cells out of the tumor and into circulation prior to surgical removal of the tumor. This treatment result may increase the systemic anti-tumor immunity of the patient and improve recurrence free survival.

These findings have important clinical implications because CD8⁺ T_RM cells, at least in the context of prior infection, have the potential to sustain a long-term systemic antitumor immunity^131,132. Some early evidence exists that this may also be true for T_RM cells that develop in the contact of malignancy. In a patient with dormant metastatic melanoma, CD8⁺CD103⁺ T cells were enriched in micrometastases near melanoma cells, providing control of cancer growth¹³¹. In a patient with vaginal melanoma refractory to ICB, tumor specific CD8⁺ T_RM cells were found only in metastatic lesions that shared TCRs with T_EM in the primary tumor. Although the isolated CD8⁺ T_RM cells showed strong functional responsiveness to melanoma antigens in vitro, loss of major histocompatibility complex (MHC) class I expression blunted tumor responsiveness in vivo¹³³. In a cohort of patients with metastatic melanoma who had a durable response to immunotherapy, scRNA-seq and paired scTCR-seq were performed across tissues at time points that spanned up to 9 years. Interestingly, tumor-reactive clonotypes were found not only as T_RM clones in the tumor and distant skin, but also as T_EM cells in circulation¹³² . This supports a growing body of literature indicating that T_RM cells expand locally in response to ICB, but then traffic through the blood to confer systemic antitumor immunity^95,134. How ICB alters tissue retention of other T_RM cells in non-malignant tissue (intestines, for example) remains to be determined.

Another route of protection against metastasis may be driven by precursor cells that differentiate into T_RM cells following recirculation from the tumor. Using a murine model of triple negative breast cancer (TNBC), Christian et al. performed TCR-β repertoire sequencing on CD8⁺ T cells isolated from the tumor and benign distant tissues and found that T_RM clonotypes in the distant tissue that overlapped with those in the tumor shared developmental ontogeny. The distant mucosal T_RM cells originated from a subset of CXCR6⁻ intratumoral effector T cells (T_eff)/ T_EM¹³⁵. CXCR6 promotes tissue residency through interaction with its unique ligand CXCL16 expressed by epithelial cells¹³⁶, and immune cells such as dendritic cells (DCs)¹³⁷. Disrupting CXCR6-mediated retention through intratumoral injection of a CXCL16 neutralizing antibody resulted in the reduction of the metastatic disease burden in distal tissues¹³⁵. This suggests that a subset of tumor infiltrated T_eff/T_EM cells retains potential to egress from tumors and fully differentiate into T_RM cells at distant sites, thereby conferring protection from metastasis.

In a study using a mouse model of melanoma-associated vitiligo, Molodtsov and colleagues¹³⁸ demonstrated that the presence of CD8⁺ T_RM cells in regional lymph nodes confers immunity to metastatic melanoma. Through scRNA sequencing, the authors found that some of the most expanded clonotypes in vitiligo-affected skin also occurred in the LNs where they were overwhelmingly maintained as T_RM cells. In turn, these CD8⁺ T_RM cells afforded long-lived protection against melanoma seeding in LNs. They further expanded these findings to humans, finding CD8⁺CD69⁺ T cells within human melanoma-infiltrated sentinel lymph nodes (SLNs) that exhibited T_RM features¹³⁸. This suggests that the potential to generate T_RM cells in the SLNs of melanoma patients could confer protection against metastasis.

The clinical implications of T_RM cells in cancer

Multiple independent reports now indicate that tumor specific T cells harboring a T_RM phenotype are sequestered in solid tumors and likely dLN and are present at very low frequencies or undetectable in circulation in patients with cancer^95,96. If this reservoir of tumor specific T cells in the tumor and dLN is removed with surgery or ablated with radiation treatment, a significant proportion of the patient’s antitumor immunity could be lost. Multiple reports also indicate that peripheral blood frequencies of tumor specific T cells that display a T_RM phenotype in the tumor can be substantially increased with ICB-based immunotherapy prior to definitive treatment^95,96. Additionally, multiple lines of pre-clinical and clinical evidence suggest that enhancement of systemic anti-tumor immunity confers protection against disease relapse or metastasis^{131–133,135}. Together, these data indicate that for patients with newly diagnosed cancer, tissue egress and re-circulation of tumor specific T_RM cells may underlie the substantially improved recurrence free survival observed in patients that receive neoadjuvant ICB. Proceeding with definitive surgical resection or radiation treatment without first activating and inducing tissue egress of tumor specific T_RM cells into circulation may remove a substantial proportion of the patient’s antitumor immunity and reduce the chance of long-term relapse free survival. In addition to surgical stress¹³⁹ and surgery associated NK cell dysfunction¹⁴⁰, reduced systemic antitumor immunity after surgical removal of a tumor harboring tumor specific T_RM cells may also result from so called ‘surgery-induced immunosuppression’, which is consistent with observations of outgrowth of metastatic deposits following removal of a primary tumor across many cancer types¹⁴¹.

The rationale for aiming to therapeutically induce T_RM cell egress from primary tumors into circulation may be clinical context dependent. For example, inducing tissue egress of tumor specific T_RM cells from tumors and dLN into circulation may be desirable in patients with newly diagnosed cancer to maintain systemic anti-tumor immunity following definitive surgical resection or radiation ablation, but may be undesirable in a patient with relapsed disease where the goal is to maintain the greatest possible density of tumor specific T cells within disease deposits. Accordingly, use of therapeutic interventions that aim to induce tissue egress of T_RM cells into circulation may have the strongest scientific rationale in the neoadjuvant or induction setting.

Conclusions

Recognition of compartmentalization of tumor specific T cells into tumors through induction of a T_RM-like phenotype may have significant implications for how we treat cancer. Yet, mature clinical studies definitively linking improved recurrence free survival to tissue egress of functional T_RM cells that can detect and eliminate residual or circulating cancer cells following ICB in neoadjuvant or induction studies in patients with newly diagnosed cancer are still needed. Investigation into the additional contribution of therapeutically inhibiting TGF-β or other environmental cues that promote T_RM cell differentiation to tissue egress and enhanced systemic antitumor immunity is specifically important because of the roles that TGF-β signaling also plays in tumorigenesis. Early in tumor progression, TGF-β inhibits epithelial cell proliferation, mainly through cell cycle arrest and induction of apoptosis. In late-stage cancers, accumulation of genomic alterations within the epithelial compartment often results in the loss of TGF-β-associated anti-proliferative signaling^142,143. As a result, the tumor promoting and immunosuppressive properties of TGF-β dominate in established malignancies, making it an attractive therapeutic target to both directly inhibit tumor cell proliferation and survival as well as to enhance anti-tumor immunity. In this review, we mainly explored the role of TGF-β in the differentiation of tumor-resident CD8 T cells and how targeting TGF-β signaling could lead to the egress of tumor-reactive T cells from the tumor into circulation and the resulting enhanced systemic immunity. Considering this mechanism, TGF-β inhibition is rational in the neoadjuvant setting prior to surgical removal of the tumor. The decision to inhibit TGF-β in other setting such as relapsed or metastatic disease must consider other effects of TGF-β blockade on the tumor cell and immune compartments. What is clear, however, is that failing to ‘use the tumor as a vaccine’ and promote systemic antitumor immunity through therapeutic reversal of tumor sequestration of tumor specific T cells prior to definitive treatment is failing to offer patients with newly diagnosed cancers a better chance at long term recurrence free survival.

Highlights.

• Tissue resident T cells (T_RM) express adhesion molecules and embed within tissues

• T_RM promotion in tumors leads to sequestration of T cells out of circulation

• Definitive surgery or tumor ablation may remove most tumor-specific T cells

• Neoadjuvant immunotherapy promotes egress of T_RM from tumors into circulation

• This effect may underlie the clinical benefit observed with neoadjuvant immunotherapy