Emerging Concepts of Tissue-Resident Memory T Cells in Transplantation

Abstract

In this review, we summarize and discuss recent advances in understanding the characteristics of tissue-resident memory T cells (TRMs) in the context of solid organ transplantation (SOT). We first introduce the traditionally understood noncirculating features of TRMs and the key phenotypic markers that define this population, then provide a detailed discussion of emerging concepts on the re-circulation and plasticity of TRM in mice and humans. We comment on the potential heterogeneity of transient, temporary resident and permanent resident T cells and potential interchangeable phenotypes between TRM and effector T cells in nonlymphoid tissues. We review the literature on the distribution of TRM in human nonlymphoid organs and association of clinical outcomes in different types of SOT, including intestine, lung, liver, kidney and heart. We focus on both tissue-specific and organ-shared features of donor- and recipient-derived TRMs after transplantation whenever applicable. Studies with comprehensive sample collection, including longitudinal and cross-sectional controls, and applied advanced techniques such as multicolor flow cytometry to distinguish donor and recipient TRMs, bulk and/or single-cell T cell receptor sequencing to track clonotypes and define transcriptome profiles, and functional readouts to define alloreactivity and pro-/anti-inflammatory activities are emphasized. We also discuss important findings on the tissue-resident features of regulatory αβ T cells and unconventional γδ T cells after transplantation. Understanding of TRM in SOT is a rapidly growing field that urges future studies to address unresolved questions regarding their heterogeneity, plasticity, longevity, alloreactivity and roles in rejection and tolerance.

Summary

SOT provides an opportunity to study the dynamics of human tissue-resident T cells, given the abundance of T cells at multiple organ sites and the ability to track donor- and recipient-derived cells based on specific HLA-markers. Many questions regarding bidirectional alloimmune responses between graft and host and the anatomical and environmental diversities of each organ type remain to be addressed. While animal studies provide an opportunity to more precisely manipulate designated factors to investigate relevant mechanisms, not all of the findings can appropriately translate to humans. Important questions include the heterogeneity, plasticity, alloreactivity and persistence of human TRMs in solid organs. Emerging concepts (Figures 1, 2) include the developmental plasticity of TRMs with interchangeable Teff phenotypes and their recirculating features to secondary lymphoid tissues and peripheral blood; the relocation of ex-TRMs to original NLTs and re-differentiation in situ; and the association of rejection and dynamic turnover of intragraft T cells, not only αβ conventional T cells, but also Tregs and γδ T cells. Multiomics provide promising platforms to address the above questions by identifying the phenotype, clonotype, alloreactivity and functional gene profiles of tissue TRMs and their spatial interaction with other cell types and their milieu.

1. Overview of tissue‐resident memory T cells (TRMs) in solid organ transplantation (SOT)

Tissue-resident T cells in nonlymphoid tissues (NLTs) include conventional CD4 and CD8 αβ T cells, regulatory T cells (Tregs), innate lymphoid cells, and several types of unconventional T cells, such as γδ T cells, invariant NKT cells and mucosal-associated invariant T cells.^1,2 In this review, we focus on T cells with memory features in such tissues, which are termed TRMs. TRMs differ from their circulating counterparts in phenotype, transcriptional regulation, survival requirements, and function.^3,4 TRMs provide a frontline defense against reinfections with pathogens at body surfaces. However, their role in SOT is largely unexplored.^5,6 The success of SOT is limited by rejection and risks of infection and cancer, reflecting challenges with immunosuppression.^7,8 Given that TRMs are largely excluded from the circulation⁹ and have lower reliance on costimulation,^10,11 they may be shielded from the effects of immunosuppressive drugs,^12,13 thereby protecting the organ against infection or promoting tissue homeostasis.¹⁴ TRMs carried within the allograft and graft-infiltrating recipient T cells that gradually acquire TRM phenotypes may contribute to graft-versus-host disease (GVHD) and transplant rejection, respectively.^5,6,15

Studies from our group and others have demonstrated the presence of both donor- and recipient-derived TRMs in human small intestine,^16–19 lung,^20,21 liver,^22,23 and kidney²⁴ allografts. Remarkably, donor graft-derived tissue lymphocytes can remain within their tissue of origin for months to years after transplantation. In organs highly enriched for TRMs, such as intestines and lungs, the dynamics of donor T cell replacement by the recipient in the graft mucosa are closely associated with clinical outcomes, where slower replacement associates with less rejection and better graft survival.^16,17,20 By integrating T cell clonotypes, mixed lymphocyte reaction (MLR)-determined alloreactive clonotype analysis^25,26 and single-cell RNA (scRNA) profiling in human intestinal transplantation (ITx), our previous^16,17 and ongoing studies^27–29 have highlighted the role of bidirectional alloresponses in TRM-enriched grafts in determining clinical outcomes (Section 4.1). Moreover, scRNA-seq studies^27–29 provide preliminary evidence for interconversion between TRMs and effector T cells (Teffs) among intragraft T cells after ITx, consistent with the evolving concepts of heterogeneity and recirculating features of TRMs (Section 3). TRMs adapt to local environments that vary in cytokines, metabolites, cell interactions, and matrix proteins. In Section 4, we discuss studies of human TRMs located in gut, lung, liver, kidney and heart and their contribution to clinical outcomes after transplantation. The balance between donor- and recipient-derived T cells in the allograft may not be limited to conventional CD4 and CD8 αβ T cells, but can be extended to Tregs (Section 5) and unconventional γδ T cells (Section 6). This review highlights recent advances and emerging concepts around TRMs in transplant medicine and urges further studies to gain deeper understanding of the impact of TRM on transplant outcomes and develop therapeutic interventions.

2. Definition of TRMs: traditionally nonrecirculating features and key phenotypic markers

Historically, the defining feature of TRMs in both animals and humans is their commitment to peripheral tissue sites and lack of recirculation.^2,3,9 Strategies such as parabiosis surgery,^30–32 transplantation,^33–35 in vivo intravascular antibody staining,^36–38 in situ labeling,^39,40 T cell depletion,⁴¹ and blockade of lymphocyte trafficking^32,35,36 have been used in animal studies to assess migration patterns of TRMs. In humans, persistence of donor TRM after different types of SOT, including small intestine,^16–19 lung,^20,21 liver,^22,23 and kidney,²⁴ suggests a similar propensity for TRMs to be retained within tissues. Tissue residency of TRMs is regulated by the induction of a series of retention signals and the repression of tissue egress pathways, consistent with their low migratory and proliferative potential.^3,42 Therefore, TRMs are transcriptionally, phenotypically, and functionally distinct from recirculating central memory (TCM) and effector memory (TEM) T cells.^3,43

TRMs lack expression of several transcription factors (TFs) and receptors associated with lymph node (LN) homing and recirculation, such as KLF2, KLF3, L-selectin, S1PR1 and CCR7.⁴² TRMs express surface markers that include C-type lectin CD69 and integrins CD103 (αE) and CD49a (α1). CD69 prevents surface expression of S1PR1, preventing tissue egress.⁴⁴ CD103 binds to the epithelial cell marker E-cadherin,⁴⁵ thereby favoring the retention of TRMs in tissues enriched with epithelial cells, such as intestines, lungs and skin. These NLTs with TGF-β-rich environments drive the expression and maintenance of CD103 on TRMs.⁴⁶ Co-expression of CD69 and CD103 is more frequently seen in CD8 compared to CD4 TRMs.^16,20 CD49a, the α subunit of VLA-1, is expressed on skin, lung and intestinal TRMs, likely promoting tissue retention via binding to collagen and laminin.⁴⁷ Expression of CD49a has been associated with cytotoxic function of CD8 TRMs in human skin.⁴⁸ Cytotoxic features of TRMs are reflected by their high expression of GZMB, perforin, IFNγ and TNFα.⁴⁹ Phenotypes of TRMs are controlled by their TF profiles that generally include Runx3, Notch, Blimp1, Hobit, BATF and AHR, although this appears to be subset (CD4 vs CD8) and species (mouse vs human) dependent.⁴⁹ Despite these common characteristics, identification of TRM is complicated by the fact a single set of phenotypic markers does not appear to be exclusive to this subset. Recent advances in multi-omics technologies will potentially overcome this limitation by measuring a list of TRM signature genes/proteins in combination with clonal tracking⁵⁰ and even evaluating the environmental milieu of TRM residence in a particular tissue through spatial immune profiling.^51,52

Representative flow cytometry gating of TRMs (CD69⁺CD103^+/−) in human NLTs in steady state^42,53 and transplantation settings^16,17 and schematic representative lists of TRM signature genes⁴² have been previously presented by us and others. The densities of TRMs in normal human NLTs have been summarized in a review,⁹ reflected by the percentages of CD69⁺CD103^+/− T cells among total T cells: human skin (70–90%), lungs (60–80%), intestine (80–95%) and liver (60–80%) are highly enriched for TRM. Donor age is also a contributing factor for the composition of TRMs in human NLTs, as younger donors (0–2 years old) have significantly lower proportions of CD8 TRMs in mucosal sites compared with young adults (15–25 years old).⁵³

3. Emerging re-circulating features of TRM in mice and humans

Recent identification of recirculating features of TRMs in mice and humans challenges the previous paradigm that TRMs retain long-term residency in NLTs without participating in systemic recall responses.⁵⁴ In fact, TRMs exhibit a significant level of developmental plasticity, being capable of tissue egress and re-entry into the circulation in both steady state and inflammatory conditions. Changes in gene expression, such as downregulation of tissue-resident markers (CD69 or CD103) and transient upregulation of exit signals (CD62L), allow “ex-TRMs’ to exit from tissues and re-differentiate to TCM and TEM in the circulating T cell pool.^55,56 Interestingly, circulating ex-TRMs retain a propensity to return to their tissue of origin and even populate other NLT sites. In this section, we summarize and discuss the evolving re-circulating features of TRM in murine and human studies (Figure 1).

Figure 1. Emerging re-circulating features of TRM in mice and humans.

Recent studies demonstrated that TRMs exhibit a significant level of developmental plasticity, being capable of tissue egress and re-entry into the circulation and SLOs in steady state and/or inflammatory conditions. Some TRMs in NLTs may enter into a transitioning stage (T-TRM) with Teff phenotypes before they egress the tissue to become circulating ex-TRMs or TRMs in SLOs. T-TRMs and ex-TRMs undergo changes in gene expression, such as downregulation of TRM markers (CD69 or CD103) and transient upregulation of exit signals (CD62L), compared to TRMs in NLTs. TRMs in circulation and SLOs can trans-differentiate to TCM and TEM. Circulating ex-TRMs retain a propensity to return to their tissue of origin and even populate distal NLT sites. aLym: afferent lymph. eLym: efferent lymph.

3.1. Murine studies of re-circulating TRMs

Epithelial barrier tissues contain a mixture of resident and recirculating T cells in mice. In herpes simplex virus (HSV)-infected skin, Gebhardt et al⁵⁷ identified 2 distinct HSV-specific memory subsets: a CD8 T cell population sequestered in the infected epidermis and a dynamic CD4 T cell population that trafficked rapidly through the dermis. Kaede transgenic mouse skin⁵⁸ carrying photo-convertible protein upon exposure to violet light and parabiotic pairs between CD45.1 or CD45.2 congenic mice were used to track the fate of cutaneous CD4 T cells in secondary lymphoid organs (SLOs) and circulation.^59,60 Bromley et al⁵⁹ demonstrated that a subset of CD4 memory T cells exits from the skin and reenters draining LNs, circulation, distal LNs, and sites of nonspecific cutaneous inflammation. These migrating CD4 T cells expressed a transitional phenotype (CD69⁻CD103^+/−CCR7^+/intCD62L^intESL⁺). Collins et al⁶⁰ demonstrated that a vast majority of skin CD4 T cells equilibrate with the circulation rather than lodge in the tissue at steady state. Almost half of skin-infiltrated CD4 T cells in parabiosis experiments expressed CD69 and CD103, similar to their host counterparts. Photo-converted Kaede CD4⁺ T cells migrating from the skin to the draining LNs partially expressed CD103 but not CD69, indicating a modulated phenotype of translocating CD4 TRMs. Using the lymphocytic choriomeningitis virus (LCMV) infection model, Masopust et al⁶¹ found that CD4 TRMs share overlapping transcriptional signatures and location-specific features with CD8 TRMs, including high CD69 and GZMB expression in the small intestine. A population of bona fide CD4 TRMs specific to LCMV infection was identified in SLO that share transcriptional characteristics with CD4 TRMs from NLTs. CD69⁺ CD8 TRMs were detected in the red pulp of spleen and medullary area of LNs.⁶² Utilizing OT-I-Kaede immune chimeras with Vesicular Stomatitis Virus (VSV)-ovalbumin infection, local reactivation in skin and female reproductive tract was shown to induce migration of antigen-specific CD8 TRMs from NLTs to the draining LNs.⁶³

TRMs in NLTs can also give rise to circulating effector and memory T cells and further relocate to the local environment upon reactivation.^55,64,65 Restimulated CD8 TRMs in murine intestines undergo retrograde migration to rejoin the circulating pool and exhibit developmental plasticity to differentiate into TCM, TEM and TRM.⁵⁵ Ex-TRMs downregulated CD69 and CD103, upregulated CD62L and maintained CCR9 expression after infection. They also maintained a heritable capacity to relocate to their tissue of origin during recall responses and re-differentiated into local TRMs, leading to an “outside-in” differentiation model.⁵⁵ To investigate TRM progeny in secondary responses, Behr et al⁶⁴ developed a lineage tracer mouse model exploiting the TRM-defining TF Hobit. Reinfection with Listeria monocytogenes-expressing ovalbumin (Lm-OVA) induced local expansion of OT-I TRMs, accumulation of secondary TRMs in draining LNs and a sizeable fraction of circulating secondary memory T cells that developed downstream of TRMs. These secondary TRM responses were substantially impaired by specific ablation of primary local TRMs. OT-1 TRMs reactivated by Lm-OVA lost some TRM markers and their retention profiles (Hobit, CD69, RGS1) and upregulated genes related to egress (S1PR1, KLF2). These ex-Hobit⁺ secondary memory T cells (ex-TRMs) largely consisted of TEM cells coexpressing KLRG1 and CX3CR1. The same group⁶⁵ also performed adoptive transfer and LCMV reinfection models to assess secondary responses of TCM and TEM at mucosal sites. Both TCM and TEM appeared compromised in their ability to form CD103⁺ TRMs in the gut. However, activated intestinal TRM, but not liver TRM, efficiently reformed CD103⁺ TRMs.

3.2. Human studies of re-circulating TRMs

Studies by Klicznik et al⁶⁶ challenged the concept of strict tissue compartmentalization of CD4 TRMs in humans. The authors identified a population of circulating CD4 T cells in blood and thoracic duct lymph of healthy individuals with phenotypic, transcriptional, and clonal signatures that suggested that they were ex-TRMs originating from human skin. Using explant cultures from human skin and mass cytometric profiling of circulating CD4 T cells from healthy subjects, a fraction of human circulating CD4 T cells was shown to downregulate CD69, but still express CD103 and cutaneous lymphocyte antigen (CLA), a glycan promoting skin entry. A cluster of these circulating CD4 T cells expressed the skin-tropic chemokine receptors CCR4, CCR6 and CCR10. Clonal analysis demonstrated a greater overlap between CD4⁺CLA⁺CD103⁺ T cells in the blood and skin than other matched subsets. By generating human engineered skin on immunodeficient NSG mice followed by xenografting human skin from healthy donors onto the same NSG mice, the authors confirmed that human cutaneous CD4 TRMs can reenter the circulation and relocate to secondary human skin sites and reassume a TRM phenotype. Recirculating CD4 TRMs (CLA⁺CD103⁺) represent a rare population in blood of healthy humans, which, on average, accounts for <2% of circulating CD4⁺CLA⁺CD45RA⁻ memory T cells and <0.2% of total CD4⁺ T cells. Estimated number of CD4⁺ CLA⁺CD103⁺ cells in the blood is between 2×10⁶ and 2×10⁷, which is approximately 250-fold lower than in the skin.⁶⁶

The concept of recirculating ex-TRMs was also supported by findings in human disease settings.^67–72 Diani et al⁶⁷ found that CCR6⁺ or CXCR3⁺ CD8 memory T cells co-expressing CD69 are increased in the blood of psoriatic arthritis (PsA) patients, which was associated with increased systemic inflammation. In vitro transwell migration assays demonstrated preferential migration of CD8 TEMs, with a higher percentage of CXCR3⁺ cells and a lower percentage of CCR6⁺ cells, towards synovial fluid of PsA patients. In fact, an accumulation of CXCR3⁺ CD8 T cells was observed in synovial fluid of PsA patients. A previous study from the same group⁶⁸ correlated circulating CCR4⁺ CD8 memory T cells co-expressing CD103 with both systemic inflammation and disease severity in psoriasis patients. These data support the hypothesis that recruitment of specific chemokine receptor-bearing CD8 T cells to inflamed joints is an important downstream event in systemic inflammation, and that a fraction of such cells may constitute recirculating ex-TRMs.

In human celiac disease (CeD), an intestinal autoimmune disease driven by dietary gluten and gluten-specific CD4 T cell responses, Han et al⁶⁹ identified a large increase in circulating CD38⁺, αE (CD103)/β7 integrin-expressing CD8⁺ αβ and γδ T cells after gluten challenge. These T cells had a restricted TCR repertoire. Single-cell analysis⁷⁰ of γδ and CD8⁺ αβ TCR sequences from both blood and gut of CeD patients before and during gluten challenge revealed extensive clonotype sharing across tissue and time, even prior to gluten challenge. More expanded clonotypes and clonal sharing between blood and gut were seen in subjects with a challenge-induced surge. However, γδ and CD8⁺ αβ TCR repertoires between individual patients were rather diverse, suggesting they may not be specific for the gluten antigen. These may be NKG2D⁺ T cells that exhibit TCR-independent cytolytic activity against epithelial cells expressing stress signals, as described in the intestinal intraepithelium of CeD patients.⁷³ Gluten-specific CD4 TCR repertoires exhibit predominant public features, as 10% of TCRα, TCRβ, or paired TCRαβ amino acid sequences of total 1813 TCRs generated from 17 CeD patients were observed in 2 or more patients, and are shared between blood and gut tissue in CeD patients over decades.⁷¹ It is possible that recirculating ex-TRMs from both CD4 and CD8 compartments contribute to these phenomena, consistent with the “outside-in” model proposed in mice.⁵⁵

In patients undergoing allogeneic hematopoietic stem cell transplantation who developed skin and gastrointestinal GVHD, Strobl et al. demonstrated that there was a population of circulating recipient skin-derived T cells with a TRM phenotype (cTRMs: CD103⁺CLA⁺CD69⁻/lowCD45RO⁺) that can produce Th2 and Th17 cytokines. Single cell RNA sequencing showed a trend toward increased TRM gene expression among these recipient-derived cTRMs and they were demonstrated to be able to exert damage to keratinocytes in skin and home to distant tissue sites during GVHD, including the gastrointestinal tract, as they expressed the gut-homing marker integrin α4β7.⁷²

3.3. Transient, temporary resident and permanent resident T cells and potential interchangeable phenotypes between TRM and Teff in NLTs

Long-term residency of TRMs in NLTs is overlayed with elements of migration and developmental flexibility, although it is not clear whether these are features of all TRMs or only a subset of them. Systemic distribution of TRMs originating from different NLTs might contribute to broad protection against pathogens that escape local defense.⁵⁶ It has been proposed that T cells within peripheral tissues may consist of cells at 3 different stages, namely transient, temporary resident, and permanent resident, characterized by rapid, slow and no recirculation between tissue and blood/lymphatics, respectively.⁷⁴ Given that circulating ex-TRMs undergo changes in gene expression compared to TRMs in NLTs, it is possible that such modulation might occur within tissues before translocation to the circulation. Some TRMs in NLTs may enter into a transitioning stage (T-TRM)⁷⁵ with Teff phenotypes before they egress the tissue to become circulating ex-TRMs (Figure 1). Our ongoing scRNA-seq studies^27–29 provide preliminary evidence that clonally-defined alloreactive and nonalloreactive T cells in intestinal allografts can distribute in different clusters that cover both TRM and Teff phenotypes, supporting the above notion (Figure 2). Epigenetic analysis could help to further describe the identity and plasticity of organ-specific TRMs and their crosstalk with the local environment. The presence of recirculating TRMs might provide diagnostic biomarkers and targets for development of novel therapies for systemic inflammation, autoimmune diseases and allograft rejection.

Figure 2. Role of donor- and recipient-derived tissue-resident memory T cell (TRM)-mediated bidirectional alloresponses after human intestinal transplantation (ITx), as determined by integration of T-cell clonotypes, mixed lymphocyte reaction (MLR)-determined alloreactivity, and scRNA profiling.

Major local and systemic immunological events described in our previous and ongoing studies are summarized. (I) Intestinal donor T cells are dominated by a TRM phenotype (CD69⁺CD103^+/−CD28^low). Donor GvH-reactive T cells expand within the graft (intestine/liver/LNs), likely acquiring Teff phenotypes (CD69^low/−CD103^low/−CD28^+/high) in response to recipient APCs that enter the graft early and control recipient HvG-reactive T cells locally. Expanded GvH-reactive Teff then migrate into the recipient circulation and BM, where they attack host hematopoietic cells and counteract HvG responses (bold black arrows with red “X” to show inhibition effect). This LGVHR (II) makes hematopoietic “space” for donor cell engraftment early post-Tx and (III) allows the survival and expansion of donor HSPCs from the graft that enter the circulation, BM, and thymus, resulting in de novo donor T cell generation and promoting persistent multilineage chimerism and potentially promoting immune tolerance. Intestinal HSPCs undergo replacement by the recipient from a circulating pool. (IV) Circulating recipient T cells (CD69⁻CD103⁻CD28^+/high) infiltrating donor intestinal allografts are primed by donor APCs, likely in gut-associated lymphoid tissues (GALTs), and become HvG-reactive T cells with Teff phenotypes that repopulate the graft mucosa early post-Tx. (V) These recipient Teff cells gradually acquire a TRM phenotype during quiescence and seed the entire GI tract, including the residual recipient intestine. They can regain features of circulating Teff during late rejection (eg: upregulation of CD28 and NKG2D). Some of these recipient Teff cells may enter into lymphoid follicles in graft mucosa and become follicular helper T cells (Tfh). (VI) Persistent alloreactive recipient T cells as TRM and/or Tfh may pose a constant risk of recurring rejection or become tolerized (tolerant HvG). De novo-generated HvG-reactive T cells post-Tx and donor- and recipient-derived Tregs may also influence the bidirectional alloresponses after ITx.

4. Distribution of TRM in human NLTs and association with clinical outcomes in different types of SOT

TRMs residing in different organs must accommodate to unique local environments, given that each anatomic site differs in cytokines, nutrients, and composition of epithelial and connective tissues. Murine studies have indicated that tissue topography, which considers tissue spatial structure and cell-cell interaction in a particular microenvironment, such as epidermis (epithelial tissue), exocrine glands (epithelial-connective tissue) and lymphoid organs (connective tissue), may influence CD8 TRM retention and surveillance strategies.⁷⁶ TRMs rely on chemokine- and integrin-related mechanisms to be retained in epithelial barrier sites.⁷⁷ However, TRMs in exocrine glands display autonomous motility that is supported by tissue macrophages and independent of chemoattractants and adhesive molecules.⁷⁸

In considering the role of TRM and graft-versus-host (GvH) alloreactivity in transplant outcomes, it should be remembered that intestine, lung and liver are rich lymphoid organs, whereas kidney and heart are not, which may affect the balance of bidirectional alloresponses after each type of SOT (Figure 2). Blood contamination should also be considered in T cell phenotype and clonotype analysis of vascular organs like liver and kidney. In this section, we discuss human TRMs in different types of SOT and associated graft outcomes.

4.1. TRMs in ITx

As the only long-term option for patients who suffer intestinal failure, ITx is complicated by high rejection rates and consequences of high levels of immunosuppression such as infection, renal dysfunction and de novo malignancy.^79,80 Our previous studies showed that peripheral blood macrochimerism, defined as the presence of ≥4% of donor T cells, developed frequently after multivisceral transplantation, usually without causing GVHD. Blood macrochimerism is associated with significantly reduced graft rejection or donor-specific antibody development and slower replacement of donor T cells in the graft by the recipient.^16,17,81,82 These observations link local and systemic immunological events. A faster rate of recipient T cell predominance over donor T cells in the graft mucosa correlated with early rejection, which was associated with a preponderance of host-versus-graft (HvG) T cell clones.^16,17 Donor T cells persisted in the mucosa for several years in patients lacking rejection, only gradually being replaced by recipient T cells^16–19. While intragraft donor T cells were dominated by a TRM phenotype (CD69⁺CD103^+/−CD28^low), recipient Teffs, including HvG-reactive T cells, infiltrating the intestinal mucosa slowly acquired a TRM phenotype during quiescence and regained features of circulating Teff during late rejections (eg: upregulation of CD28 and NKG2D).¹⁶ The persistence of alloreactive recipient HvG T cells as TRM may pose a constant risk of rejection, perhaps contributing to high intestinal allograft rejection rates (Figure 2).

Studies of intestinal TRMs (CD69⁺CD161⁺CD103^+/−) after human ITx have also been performed by Jahnsen and colleagues^18,19 in patients without rejection. Donor TRMs from duodenal grafts (transplanted with pancreas) were shown to persist for at least 52 weeks. In normal donors,^18,19 lamina propria CD8 TRMs demonstrated a polyfunctional profile (IFN-γ⁺IL-2⁺TNF-α⁺) and were potently cytotoxic following stimulation.¹⁹ Similarly, the vast majority of lamina propria CD4 TRMs were polyfunctional Th1 cells (IFN-γ⁺IL-2⁺TNF-α⁺), and a fraction produced GZMB and perforin after activation.¹⁸

Given that rejection episodes are closely associated with accelerated replacement kinetics of graft T cells after ITx,¹⁶ and the potential involvement of recirculating ex-TRMs with re-differentiation plasticity, our group is actively pursuing multiomic studies to integrate T cell clonotypes, alloreactivity and gene expression profiles.^{16,17,27–29} Clonal and phenotypic tracking of donor and recipient T cells in serial intestinal allograft biopsies, peripheral blood and bone marrow (BM) post-Tx provide a deeper understanding of their tissue origin, migration pattern and phenotypic maturation. We also use a unique platform¹⁷ that integrates bulk TCRβ-seq and scRNA-seq that combines 5’ transcriptional analysis with TCRαβ-seq. T cells are further annotated as CD4 or CD8 alloreactive or nonalloreactive or as nonmappable by interrogation of the sequence set defined from pre-Tx MLRs,^16,25 allowing functional characterization of known alloreactive T cells within the allograft tissue.

We recently demonstrated that donor GvH-reactive T cells expand within the intestinal allograft in response to recipient antigen-presenting cells (APCs) that enter the graft early. These GvH-reactive T cells appear to control recipient HvG-reactive T cells locally, then migrate into the recipient circulation and BM, where they attack host hematopoietic cells and counteract HvG responses¹⁷ (Figure 2). This lymphohematopoietic graft-versus-host response (LGVHR) usually occurs without causing GVHD. Single-cell transcriptional profiling of BM-infiltrating donor T cells reveals dominant clusters of cytotoxic Teffs with GvH allorecognition identifiable by TCR sequences. This LGVHR makes hematopoietic “space” for donor cell engraftment early post-Tx and allows the survival and expansion of donor hematopoietic stem and progenitor cells from the graft that enter the circulation, BM, and thymus, resulting in de novo donor T cell generation and promoting persistent multilineage chimerism and potentially promoting immune tolerance.^81,82 Individual GvH clones were detected in either the ileal mucosa or PBMCs before detection in recipient BM,¹⁷ consistent with an origin in the intestinal allograft, and consistent with our hypothesis that circulating and BM-infiltrating GvH-reactive T cells originated as microbe-reactive TRMs, carried in the intestinal allograft, that cross-react on recipient alloantigens.²⁸ We are currently utilizing the above multiomic platform to test the hypothesis that alloreactive HvG T cells infiltrating the intestinal allograft join the TRM pool and may become Teffs that either participate in recurring rejection and/or are tolerized²⁷ (Figure 2).

In a cohort of patients who developed GVHD after ITx, Weiner et al. described a putative circulating TRM population (CD69⁺CD62L⁻HLA⁻DR⁺CD57⁺PD-1⁺) that they believed to have been derived from the donor intestinal graft and was associated with GVHD.⁸³ Interestingly, the percentages of CD69⁺ TRMs not only increased in the blood of patients with GVHD compared to those without, but also in parallel increased in the grafted and native intestines of GVHD patients. These were demonstrated to be donor-derived using multiplex immunostaining in the cases involving HLA-A2 host-donor mismatch. These observations support the scenario whereby TRMs with transitioning phenotypes (T-TRMs) translocate from donor intestines to recipient blood and further home to native NLTs to contribute to GVHD pathophysiology.

4.2. TRMs in lung transplantation (LuTx)

Like the intestine, human lung is also an immunological barrier organ that is highly enriched for TRMs. Therefore, bidirectional alloresponses may also contribute to graft outcomes after LuTx. Blood chimerism after LuTx^20,84,85 is detected, but is not as robust or persistent as that observed in ITx.^17,81,82 Blood chimerism was detected in 3 of twenty LuTx recipients and disappeared by 2 months after transplant.²⁰ However, donor T cells were found to persist in bronchoalveolar lavage (BAL) samples at high frequencies even after 1 year, at levels up to 55% and 85% for CD4 and CD8 T cells, respectively.²⁰ These donor-derived BAL T cells expressed high levels of TRM markers including CD69, CD103 and CD49a. Consistent with observations in ITx patients,^16,17, persistence of donor TRMs after human LuTx was associated with better clinical outcomes, as reflected by less acute cellular rejection and primary graft dysfunction.²⁰ Earlier studies reported a much more rapid replacement of donor-derived lung TRM by recipient T cells within 40 days post-Tx when assessed by serial transbronchial biopsies.⁸⁶ Again, low numbers of donor lymphocytes in the allografted lung seemed to correlate with a worse clinical course. Taken together, monitoring TRM dynamics in lung allograft and BAL samples may be clinically informative.

Not surprisingly, acquisition of TRM phenotypes within graft-repopulating recipient T cells after LuTx^6,20,87 also mirrored the phenomenon observed after ITx.^16,18,19 A longitudinal study of LuTx showed that 3 to 9 months after transplant, the numbers of CD4 and CD8 T cells in the lung can reach normal healthy donor levels. However, the number of CD3 and CD8 (but not CD4) T cells in post-Tx lung allografts continued to increase over time, regardless of the development of chronic rejection,⁸⁷ suggesting that graft-associated CD8 T cells participate in both homeostasis and chronic rejection. CMV⁻ LuTx recipients who received CMV⁺ allografts demonstrated an influx of de novo CMV-specific CD8 T cells into the airways and allografts. These cells were maintained within the transplanted lung at higher frequencies than within PBMCs and were functionally and phenotypically distinct from circulating CMV-specific CD8 T cells.⁸⁸ A longitudinal study demonstrated low frequencies of TRM markers among recipient BAL T cells at early times (2 to 4 weeks) after LuTx (20 to 40% CD69⁺ CD4⁺ and CD8⁺ T cells); however, by 6 months post-Tx, >50% of recipient BAL T cells were CD69⁺, with many recipient CD8⁺ T cells co-expressing CD103. By 3 to 6 months post-Tx, recipient BAL T cells expressed CD69 and CD103 at frequencies similar to those observed in control BAL fluid. Recipient-derived T cells in the lung BAL maintained multifunctional profiles (IFN-γ⁺ IL-17⁺ IL-2⁺GZMB⁺) associated with mucosal memory T cells. scRNA-seq revealed both non-TRM (putative circulating TEM lacking CD69 and CD103 expression) and TRM-like (CD69⁺CD103^+/−ITGA1⁺CXCR6⁺RUNX3⁺) subpopulations among recipient BAL T cells.²⁰ As in ITx recipients,¹⁶ the gradual acquisition of TRM markers by recipient-derived T cells infiltrating lung allografts may reflect the conversion to this phenotype of HvG T cells that pose a constant risk of rejection or the repopulation of donor lung tissue by circulating TEM counterparts of TRMs that acquire TRM phenotypes or by recirculating ex-TRMs.

4.3. TRMs in liver transplantation (LiTx)

The liver is immunologically unique. On 1 hand, it is exposed to a variety of microbes from the systemic circulation or through the portal vein from the gut. On the other hand, it preferentially induces immune tolerance,⁸⁹ in the context of both MHC-mismatched LiTx (in rodents) and liver infection. However, intrahepatic immune responses can be robustly induced under certain circumstances such as viral or autoimmune hepatitis. Unlike in gut and lung, CD69⁺ CD8 TRMs in human liver constitute a large proportion (>90%) of CD103⁻ cells and demonstrate low cytotoxicity.^90,91 CD103⁺ liver CD8 TRMs express liver-homing and retention markers CXCR6 and CXCR3, and robustly produce IL-2 and IFNγ upon antigen stimulation.⁸⁴ Kim et al⁹² demonstrated that liver CD69⁺CD103⁻ CD8 T cells have a terminally differentiated TRM phenotype, and their effector functions depend on hypoxia-inducible factor (HIF)-2α, suggesting that they are predominantly located in hypoxic regions. Furthermore, activation of liver CD69⁺CD103⁻ CD8 T cells with HIF-2α upregulation is observed during acute (hepatitis A) and chronic (cirrhosis) liver pathology. Swadling et al⁹³ found that an increased rate of basal autophagy is a hallmark of human liver CD8 TRMs. Enhanced autophagy in CD8 TRMs can be imprinted by IL-15 or primary hepatic stellate cells and adapts liver CD8 TRMs to combat mitochondrial depolarization and acquire tissue residence. These findings highlight the importance of tissue-specific adaptations of TRMs.

In the LiTx setting, a small population of donor CD4 and CD8 T cells with TRM phenotypes (CD69⁺CD103^+/−CXCR3^hi) was detectable in liver grafts even more than a decade after an HLA-mismatched transplant.²³ Recipient CD4 and CD8 T cells are also persistently observed.^22,23 These graft-repopulating T cells may have a TRM phenotype, although with a less definitive residency program, such as lower levels of CXCR3 compared to donor-derived CD8 TRMs.²³ Despite the requirement of unique environmental conditions for liver TRMs, the persistence of intrahepatic donor TRM and the gradual acquisition of TRM phenotypes by graft-infiltrating recipient T cells after LiTx are reminiscent of findings after human ITx and LuTx, underlining the common features of donor- and recipient-derived TRMs in allograft organs. The authors further showed that TRMs lacking CXCR6 expression were detectable in the local draining LNs but did not egress into the hepatic vasculature. Whether these TRMs migrate from the liver graft or represent an independent population developed in situ will need further investigation, for example using TCR clonal tracking.²³ While the association of rejection and the dynamic replacement of intragraft donor T cells by the recipient was not explored, studies of two antiviral responses (HBV, CMV) revealed that donor-derived virus-specific CD8 TRMs persist long-term post-Tx and may be supplemented by recipient responses.²³

4.4. TRMs in Kidney transplantation (KTx)

Human kidneys contain small numbers of lymphocytes compared to intestine, lungs and liver. The composition and profile of immune cell subsets in human kidneys is largely unknown. Park et al⁹⁴ identified a predominant CD3⁺ T cell (47%±12%) population and a low proportion of CD14⁺ or CD68⁺ myeloid cells (<10%) in healthy human kidney sections. Kidney T cells included 44% CD4 and 56% CD8 subsets. An average of close to 50% of T cells displayed a TRM phenotype (CD69⁺CCR7⁻CD45RA⁻), while the rest had a TEM phenotype (CD69⁻CCR7⁻CD45RA⁻). It should be borne in mind that non-TRM populations in the healthy kidney may include circulating T cells present in the rich vasculature of the organ. Among kidney TRMs, CD103⁻CD49a^+/− cells were predominant in CD4 cells and CD103⁻CD49a^+/− and CD103⁺CD49a⁺ subsets were predominant in CD8 cells.

Drachenberg and colleagues revealed a correlation of CD8⁺CD103⁺ cytolytic T cells (CTLs) that are CD62L⁻CD11a^hiperforin⁺ with clinical renal allograft rejection by analyses of transplant nephrectomy specimens. These CD103⁺ CD8 CTLs comprised 40–50% of the graft-infiltrating lymphocyte population during late acute rejection and were also present in biopsies with signs of chronic rejection.^95,96 CD103⁺ CD8 CTLs were biased towards an intratubular localization, while a CD103⁻ subset of graft-infiltrating CD8 T cells that also exhibited a CTL phenotype was restricted to the graft interstitium.⁹⁶

De Leur et al tracked the turnover dynamics of donor intragraft T cell replacement by the recipient in transplant nephrectomies.²⁴ High proportions (1.7–17.4%) of donor-derived CD4 and CD8 T cells were only observed in early rejecting allografts removed within the first month post-Tx. Grafts that failed greater than 5 months post-Tx mainly contained recipient-derived CD8 TRMs (CD103^+/−CCR7⁻CD45RO⁺) that produced IFNγ, TNFα and GZMB.

4.5. TRMs in heart transplantation (HTx)

Both protective⁹⁷ and detrimental^98,99 roles of donor T lymphocytes carried in cardiac allografts have been reported in animal models. Meanwhile, studies on the lymphocyte compartment of human hearts are very limited. Hu et al¹⁰⁰ created a single-cell atlas of human nondiseased cardiac arteries obtained from HTx patients. T cells were the fourth largest cell population in coronary arteries, at 14.9%. CD4 T cells represented a higher percentage than CD8 T cells in coronary arteries and TRM markers CD69 and CD44 were highly expressed while CD103 and CD49a were poorly expressed. While the authors suggest that these may be specific features of vascular TRM, it is uncertain that these cells are truly TRMs rather than activated T cells. CD8 T cells in human cardiac arteries also highly expressed cytotoxic markers including granzyme family members and perforin, but had low expression of proinflammatory cytokines TNF and IFNγ. Whether these cells play a specific role in the outcome of clinical HTx will require further investigation. Graft-infiltrating T cells after HTx have been associated with rejection in both animals¹⁰¹ and humans.¹⁰² However, a lack of longitudinal phenotypic and functional studies makes comparison with other types of human transplants impossible at this time.

5. Tregs with TRM features in transplantation

The presence of Tregs in a variety of NLTs has been documented in both mice and humans, including intestinal mucosa, lung, liver, skin, kidney, adipose tissue and skeletal muscle, where they maintain host homeostasis and self-tolerance.^103–107 Emerging data from mouse studies shows that tissue Tregs exhibit unique phenotypic and transcriptional signatures that are controlled by epigenetic reprogramming.^{105,106,108,109} Tissue Tregs express TRM surface markers such as CD69, CD103 and CCR4 and express the TRM TF Blimp1.^110–112 Integrated accessible-chromatin and scRNA-seq¹⁰⁹ indicated that adaptation of Tregs to visceral adipose tissue, skeletal muscle and colon reflected a combination of tissue-shared and tissue-specific modulations. Tissue adaptation of human Tregs is still largely undefined. Transcriptional analysis of human skin Tregs demonstrated a TRM phenotype similar to skin-tropic (CLA⁺) helper CD4 T cells and CD103⁻ CD8 T cells, but these Tregs were distinct from blood-derived CLA⁺ T cells.¹¹³ A comprehensive study to characterize the transcriptome of human mucosal tissue (lung and colon) Tregs from the normal area of cancer resections and their peripheral blood counterparts¹⁰⁶ identified TNIP3 as a shared Treg-specific gene that is involved in the regulation of NF-κB signaling. The most prominent genes differentiating lung Treg from gut or blood-derived Treg were Wnt singling genes, suggesting potential crosstalk between lung Tregs with nonmtor immune tissue-specific cells and a role for lung Tregs in epithelial repair and regeneration.

The role of tissue Tregs in mediating tolerance after human SOT was investigated in a limited number of studies. Foxp3⁺ Tregs in renal allografts with subclinical rejection are associated with significantly better graft function 2 and 3 years post-Tx.¹¹⁴ Urinary Foxp3 mRNA is diagnostic of T cell-mediated rejection (TCMR), and can also predict TCMR reversibility after KTx.¹¹⁵ Direct evidence that tissue Tregs participate in regulating KTx tolerance was reported by our group^116,117 in the setting of HLA-haploidentical combined bone marrow and kidney transplants (CKBMT). Enrichment of Tregs measured by Foxp3 expression were found in long-term protocol biopsies of renal allografts from tolerant patients after CKBMT.¹¹⁷ By applying high throughput TCRβ-seq, more than 200 Treg clones identified in sorted circulating Treg populations, were detected in each kidney biopsy from 3 CKBMT subjects.¹¹⁶ Using TCRβ-seq to identify the donor-specific Treg repertoire, we demonstrated expansion of circulating donor-specific Treg clones in tolerant subjects, but not in a patient who failed tolerance, at 6 months post-Tx, implicating donor-specific Tregs in initiating tolerance induction.¹¹⁸ A study in human LuTx patients showed that most Tregs (CD4⁺CD25⁺CD127^loFoxp3⁺) in BAL were recipient-derived, even in samples with significant T cell chimerism, suggesting rapid replenishment of Tregs from the circulation after transplant.²⁰ Our ongoing single-cell profiling^27–29 of T cells in intestinal allografts has identified a small fraction of tissue Tregs (Foxp3⁺) with TRM features (RGS1⁺CXCR6⁺CCR6⁺) among both donor- and recipient-derived T cell populations 600–1800 days post-Tx, suggesting long-term residency of donor-derived tissue Tregs and the potential acquisition of TRM features of graft-infiltrating recipient Tregs. Further investigations are needed to understand tissue adaptation of Tregs after transplantation.

6. γδ TRM in transplantation

γδ T cells have both innate and adaptive properties and are implicated in immune surveillance and modulation.¹¹⁹ γδ TCRs can recognize structurally diverse and biologically unrelated antigens mainly through MHC-independent mechanisms,^119,120 with only minor, if any, alloreactivity reported in earlier in vitro studies.^121,122 Therefore, γδ T cells have garnered interest as mediators of graft-versus-leukemia effects without GVHD in the setting of allogeneic hematopoietic stem cell transplantation.¹²³ Accumulating evidence indicates that innate- and adaptive-like features of human γδ T cells may be driven by differential γδ TCR repertoires, generally defined as Vγ9⁺δ2⁺ and non-Vγ9δ2, respectively.¹²⁴ Immune repertoires can be shaped by tissue compartmentalization, age and history of antigen exposure.^124–127 Although γδ T cells only account for <10% of T cells in human peripheral blood, with a dominant semi-invariant TCR Vγ9Vδ2, they are often enriched in human solid organs and barrier sites and have heterogenous TCRs, such as Vγ2/3/4/5/8 and Vδ1/3/5. These can constitute between 10–100% of T cells in gut, lung, liver and skin.^124,128 The non-Vγ9δ2 repertoire appears to be shaped by TCR-dependent selection events including CMV infection and cancer.¹²⁵ This association of V-gene usage with tissue distribution and functional development of human γδ TCR is reminiscent of their mouse counterparts.^129,130

The role of γδ T cells in SOT outcomes remains unclear. γδ T cells might contribute to both allograft acceptance and rejection, and could impact infection and post‐Tx malignancy.¹²⁰ Earlier functional studies were limited to in vitro systems of human peripheral blood-derived γδ T cells,^131,132 which are dominated by Vδ2 clonotypes. γδ T cells had been shown to exhibit either direct veto-type suppression of alloreactions,¹³¹ or indirect stimulation of alloreactive αβ T cell proliferation by inducing maturation of autologous dendritic cells and B cells into functional APCs.¹³² A recent study¹³³ demonstrated that the γδ T cell pool in liver explants collected from patients who underwent LiTx for end-stage liver disease included a TRM compartment that is CD69⁺CXCR3⁺CXCR6⁺CD45RA^lo and enriched for “private” Vδ1 clonotypes. These liver-resident Vδ1 T cells were found to be polyfunctional and responded to both TCR and cytokine stimuli in vitro. Although a higher ratio of Vδ1/Vδ2 in the peripheral blood of liver allograft recipients has been shown to correlate with stable graft function¹³⁴ and operational tolerance,¹³⁵ there are few studies investigating the association of rejection with the turnover dynamics and clonal reconstitution of intragraft γδ T cells after human LiTx.

Our recent published¹⁷ and ongoing studies¹³⁶ have provided further insights into the role of γδ T cells in modulating 2-way alloresponses locally and systemically after human ITx. Single-cell profiling of BM-infiltrating donor γδ T cells revealed a dominant “public” Vδ2⁺ clonotype with cytotoxic Teff phenotypes similar to their CD8 αβ counterparts. Graft-repopulating recipient γδ T cells show an activated Teff phenotype early post-Tx and gradually develop into TRMs with a dominant “private” Vδ1⁺ clonotype, likely participating in graft defense and regulating graft rejection. There are still many gaps to explore in this area.

Abbreviations

APCs

antigen-presenting cells

BAL

bronchoalveolar lavage

BM

bone marrow

CeD

celiac disease

CKBMT

combined bone marrow and kidney transplants

CLA

cutaneous lymphocyte antigen

CTLs

cytolytic T cells

GALTs

gut-associated lymphoid tissues

GvH

graft-versus-host

GVHD

graft-versus-host disease

HIF

hypoxia-inducible factor

HSPCs

hematopoietic stem and progenitor cells

HSV

herpes simplex virus

HTx

heart transplantation

HvG

host-versus-graft

ITx

intestinal transplantation

KTx

Kidney transplantation

LCMV

lymphocytic choriomeningitis virus

LGVHR

lymphohematopoietic graft-versus-host responses

LiTx

liver transplantation

Lm-OVA

Listeria monocytogenes-expressing ovalbumin

LNs

lymph nodes

LuTx

lung transplantation

MLR

mixed lymphocyte reaction

NLTs

non-lymphoid tissues

post-Tx

post-transplant

pre-Tx

pretransplant

PsA

psoriatic arthritis

scRNA-seq

single cell RNA sequencing

SLOs

secondary lymphoid organs

SOT

solid organ transplantation

TCM

central memory T cell

TCMR

T cell-mediated rejection

Teffs

effector T cells

TEM

effector memory T cell

Tfh

follicular helper T cell

Tregs

regulatory T cells

TRM

tissue-resident memory T cell

T-TRM

transitioning TRM

Tx

transplantation