Memory T-cell trafficking: new directions for busy commuters

Abstract

The immune system is unique in representing a network of interacting cells of enormous complexity and yet being based on single cells travelling around the body. The development of effective and regulated immunity relies upon co-ordinated migration of each cellular component, which is regulated by diverse signals provided by the tissue. Co-ordinated migration is particularly relevant to the recirculation of primed T cells, which, while performing continuous immune surveillance, need to promptly localize to antigenic sites, reside for a time sufficient to carry out their effector function and then efficiently leave the tissue to avoid bystander damage. Recent advances that have helped to clarify a number of key molecular mechanisms underlying the complexity and efficiency of memory T-cell trafficking, including antigen-dependent T-cell trafficking, the regulation of T-cell motility by costimulatory molecules, T-cell migration out of target tissue and fugetaxis, are reviewed in this article.

Introduction

Fifty years ago, J. Gowans¹ discovered that lymphocytes possess the unique property of recirculating continuously between the blood, lymphoid tissues and lymph. Extravasation of most leucocytes is unidirectional and mediated by cell-specific but non-tissue-selective inflammatory stimuli. In contrast, T cells are capable of recirculating from the blood to lymphoid and non-lymphoid tissue and back via lymphatic vessels. Like all leucocytes, T cells undergo a number of co-ordinated adhesive interactions with the endothelium, assisted by the integrin-activating function of chemokine receptors, which allow their migration out of the blood stream (reviewed by Marelli-Berg et al.²).

The sequential operation of adhesion and chemokine receptors during migration from blood to tissue has led to the proposal of the multi-step model of transmigration,³ which now appears in every textbook.

Co-ordinated migration of naïve and memory T cells is the key to effective immunity. While naïve T cells predominantly recirculate through secondary lymphoid tissue until they encounter antigen, primed T cells efficiently localize to antigen-rich lymphoid and non-lymphoid tissue. In order to carry out efficient immune surveillance, effector/memory T cells are able to mount fast and effective responses upon re-challenge. These responses are targeted to the affected tissues by both inflammatory signals and the specific homing phenotype acquired by the T cells during activation and differentiation. While a large number of molecular mediators and interactions guiding T-cell extravasation to both lymphoid and non-lymphoid tissue following priming have been identified, relatively little is known about the molecular mechanisms regulating the targeted delivery of memory T cells to antigen-rich sites, their retention in these sites, their subsequent egression from them, and their trafficking patterns afterwards. We here summarize recent key observations addressing these issues (Fig. 1).

Figure 1.

Migratory events in memory T cell trafficking. (a) After generation in the thymus, naïve T cells (TNaive) emigrate to the blood circulation and migrate into lymph nodes (LN) by utilizing the lymphoid tissue homing receptors L-selectin and CCR7 (1). Here, T cells become activated (TM) by antigen-presenting dendritic cells (DC) and acquire homing receptors (2). CD28 signals promote migration of primed T cells to non-lymphoid target tissue. CTLA-4-mediated signals inhibit the migration of primed specific T cells to antigenic sites, and their interactions with antigen presenting cells (APCs). Successfully activated T cells leave LNs via sphingosin receptor-dependent mechanisms (3). (b) Primed T cells migrate to non-lymphoid tissue (4, 5). In conjunction with homing receptors, T-cell receptor (TCR) engagement by endothelium and tissue resident APCs promotes specific T cell recruitment and retention. Memory T cells leave target tissue by CCR7-mediated mechanisms and/or fugetaxis (6). Some memory T cells can return to lymphoid tissue (7). (c) At the end of an immune response, memory T cells can either continue recirculating, or return in niches (8): memory T cells can reside in the spleen in chronic antigen persistence, or they can locate to the bone marrow following acute but transient antigen challenge. MHC, major histocompatibility complex; HEV, high endothelial venules; T_CM, central memory T cells; T_EFF, effector T cells.

Memory T-cell recirculation

Unlike naïve T lymphocytes, which constitutively traffic through lymphoid tissue, memory T cells are more diverse with respect to their migratory properties. Antigen-experienced T cells can be subdivided into central memory (T_CM), effector memory (T_EM) and effector (T_EFF) cell subsets based on distinct migratory and functional characteristics,^4,5 although the real situation is more fuzzy. T_CM cells retain expression of the lymph node (LN) homing receptors L-selectin and chemokine (C-C motif) receptor 7 (CCR7), and, like naïve T cells, are well represented in all secondary lymphoid organs.⁶ T_CM cells can also localize to peripheral tissues and sites of inflammation.^4,7

In contrast, T_EFF and T_EM cell subsets are defined as CCR7-negative, and most of them are also L-selectin^−/low.^4,7 T_EM cells are long-lived [interleukin-7 receptor-positive (IL-7R⁺)], while T_EFF cells are mainly short-lived recently activated T cells. Both T_EFF and T_EM cells largely lack the ability to enter peripheral lymph nodes (PLNs) in the steady state and they home preferentially to non-lymphoid tissues. However, they can migrate into reactive lymph nodes to modulate the immune response in a chemokine (C-X-C motif) receptor 3 (CXCR3)- or P-selectin-dependent fashion.^8,9

The capacity of distinct subpopulations of T or B cells to travel back selectively into compartments of initial antigen contact has long been recognized and is referred to as ‘homing’.¹ In primed T cells, topographical memory is endowed by the stable expression of homing and chemokine receptors that promote their interactions with ligands expressed by the endothelium of specific organs, such as the skin and the gut.⁷

Memory T cells with tropism for the skin are characterized by the expression of the carbohydrate epitope cutaneous lymphocyte antigen (CLA),¹⁰ and the chemokine receptors CCR4¹¹ and/or CCR10.¹² CLA mediates the tethering and rolling of T cells through interaction with its endothelial counter-receptor, E-selectin, which is constitutively expressed on skin post-capillary venules. The ligands for CCR4 and CCR10, which are, respectively, chemokine (C-C motif) ligand 17 (CCL17) thymus and activation-regulated chemokine (TARC) and CCL27 cutaneous T cell-attracting chemokine (CTACK), have been found on inflamed and non-inflamed skin endothelium.^11,13 CCL17 (TARC) was shown to selectively induce integrin-dependent adhesion to intercellular adhesion molecule 1 (ICAM-1) of skin-derived memory T cells under static conditions and under physiological flow,¹¹ while CCL27 (CTACK) was found to be preferentially produced by epidermal keratinocytes, and its chemotactic effect on T cells was demonstrated in in vitro assays.¹³

Constitutive memory T-cell trafficking into the lamina propria of the small intestine requires the interaction of the integrin α₄β₇ and the chemokine receptor CCR9 on the lymphocyte surface¹⁴ with mucosal vascular addressin cell adhesion molecule 1 (MAdCAM-1) and CCL25 thymus-expressed chemokine (TECK) on endothelial cells of gut lamina propria venules, respectively.¹⁵ T cells lacking β₇-integrin chain expression are severely impaired in their ability to localize to the intestinal mucosa¹⁶ and CCL25 blockade or genetic ablation of CCR9 significantly reduces antigen-specific CD8⁺ T-cell migration to the small intestine.¹⁷

Additional adhesion molecules, such as vascular adhesion protein-1 (VAP-1¹⁸) and CD44,¹⁹ may contribute to a significant diversity of potential address codes, but selectins, α₄-integrins, β₂-integrins, and chemokine receptors and their respective ligands appear to be the workhorses of the system with differential but broadly overlapping functions at the various destinations of lymphocyte trafficking.

Acquisition and plasticity of topographical memory

The paradigm of organ-specific homing is based on the assumption that T-cell priming within a specific tissue environment, such as cutaneous and mesenteric lymph nodes (MLNs), leads to an imprinting of the expression of specific homing receptors.^17,20,21

Recent studies have shown that tissue-derived dendritic cells (DCs) are key mediators of the induction of T-cell tissue-specific homing potential.^22–24 CD8⁺ T-cell stimulation by endogenous DCs isolated from MLNs and Peyer’s patches was shown to up-regulate lymphocyte expression of the α₄β₇ -integrin and CCR9, which are associated with homing to the gut.^22–24 Conversely, skin-derived DCs were shown to induce E- and P-selectin ligands that are associated with homing to the skin.^24,25 The capacity of DCs to instruct T-cell homing properties is related to their ability to produce active metabolites from tissue-derived factors. Gut-derived DCs produce retinoic acid, which leads to imprinting of the gut-homing phenotype and suppression of the skin-homing phenotype on T cells.²⁶ Similarly, the active form of vitamin D3, 1,25(OH)(2)D(3), which is produced by skin DCs, induces T-cell expression of the skin-selective chemokine receptor CCR10, while inhibiting the expression of gut-homing receptors α₄β₇ integrin and CCR9.²⁷ Interestingly, recent data also suggest that the DCs are not the starting point but are instructed by local stromal cells.^28,29

Albeit the induction of a specific homing phenotype in primed T cells has been occasionally referred to as ‘imprinting’,²³ recent data have rather challenged the concept of permanent imprinting and favour the assumption of flexibility in the expression of homing receptors.²⁵ Hypothetically, organ-specific homing could also be explained by continuing selection or re-induction of a given receptor upon recirculation through selected tissues providing antigen-exposure and organ-specific co-signals.³⁰

Efforts to demonstrate the stability of differentially expressed homing receptors in vivo have been made only recently. The expression of ligands for E/P-selectins that serve as homing receptors for inflamed skin has been shown to persist for at least several weeks in vivo only on a subfraction of T cells. However, upon repeated stimulation under ligand-inducing conditions (presence of IL-12), the stable fraction was increased, and ex vivo isolated selectin-ligand-positive effector/memory cells turned out to be almost completely stable.³¹ This shows that imprinting of a stable homing phenotype appears possible, but requires repeated stimulation under permissive conditions, similar to findings for the imprinting of a cytokine memory in T cells.³²

The above-mentioned studies on the mucosal homing receptor α₄β₇ in CD8⁺ T cells suggested that expression of this receptor is not permanent after initial induction.²⁵ In CD4⁺ T cells, repeated stimulations in the presence of retinoic acid were found to result in a largely persistent expression of α₄β₇, and, again, ex vivo isolated α₄β₇-high memory CD4⁺ cells remained positive for weeks after adoptive transfer (B. Szilagyi and A. Hamann, unpublished).

In contrast, stable expression of the chemokine receptor CCR9, which is also induced on CD8⁺ cells by retinoic acid and considered to contribute to mucosal homing, was not observed (Mora et al.²³ and B. Szilagyi and A. Hamann, unpublished).

These data suggest that homing and chemokine receptors display individual and variable degrees of imprinting and that prolonged exposure to instructive signals appears to be crucial for the establishment of a topographical memory.

However, it has to be considered that, even in the absence of stable expression, memory might exist as a type of ‘commitment’. This has been demonstrated for T effector cells; Polarized T helper type 1 (Th1) or Th2 cells are predestined to secrete the respective effector cytokines, but require re-stimulation to do so. A committed cell needs only stimulation, for example by the T-cell receptor (TCR), rather than the full range of instructive signals, to re-acquire the specific phenotype. Further investigations are required to determine whether a similar type of memory also exists in the case of homing receptors, as some data suggest.³¹

Antigen-driven T-cell trafficking

Unlike other leucocytes, memory T cells must locate to their cognate antigen (Ag) in non-lymphoid tissue to exert their function. It is a longstanding question to what extent the accumulation of specific T lymphocytes within the parenchymal tissue is directly influenced by antigen recognition. In early studies it was reported that antigen-reactive T and B cells become concentrated within antigen-rich tissue, which can even lead to the complete disappearance of the reactive cells from the circulation. Evidence for antigen-specific trapping has been presented for lymphoid tissue,^33,34 for the liver^35,36 and for peripheral tissue.³⁷

The retention of specific T cells in antigen-rich tissue has also been demonstrated in models of autoimmunity, such as experimental autoimmune encephalomyelitis (EAE)^38–40 and diabetes,⁴¹ and in infection.⁴²

In principle, several mechanisms could lead to an accumulation of antigen-specific T cells at antigen-bearing sites: (i) the trapping of antigen-reactive cells, for example upon TCR-triggered activation of integrin adhesion or effects on motility^43–45; (ii) local proliferation of antigen-specific cells; or (iii) a direct effect of antigen recognition on the recruitment of T cells.

While trapping or local expansion may be operative during primary T-cell responses, it is tempting to speculate that these mechanisms per se would be insufficient to explain the efficacy and speed of specific T-cell accumulation in target tissue, particularly in recall responses.

Antigen presentation by the endothelium has been repeatedly reported, both in vitro and in vivo,^46–48 raising the possibility that this event may directly contribute to the recruitment of specific T cells. In support of this hypothesis, cognate recognition of B7-deficient human and murine endothelial cells was shown to enhance T-cell trans-endothelial migration without inducing unresponsiveness in vitro.^49,50 Indirect evidence that similar mechanisms may exist to sustain the recruitment of specific T cells in vivo was first provided by the observation that major histocompatibility complex (MHC) class II molecule expression by microvascular endothelium in the central nervous system precedes, and is required for, the formation of T-cell infiltrates in an EAE model in guinea pigs.⁵¹ Similarly, homing of insulin-specific CD8⁺ T cells to the islets of Langerhans during the onset of autoimmune diabetes in non-obese diabetic (NOD) mice in vivo was impaired in interferon (IFN)-γ-deficient NOD mice.⁴⁷ In particular, T-cell diapedesis was significantly diminished. This effect was reversible by treatment of the animals with recombinant IFN-γ.

Further in vivo studies provided direct evidence that antigen presentation by the endothelium contributes to the development and specificity of T-cell infiltrates. Islet-specific homing by insulin-specific H2-K^d-restricted CD8⁺ T cells was abrogated in mice lacking MHC class I expression, and in mice displaying impaired insulin peptide presentation by the local endothelium as a result of deficient insulin secretion, suggesting that endothelial cells can cross-present tissue antigens.⁵²

In addition, up-regulation of H2 molecules by local vessels led to peritoneal recruitment of HY (male)-specific H2-D^b-restricted CD8⁺ T cells in male but not female mice.⁴⁸ Consistent with previous studies,^47,51 intravital microscopy revealed that antigen presentation by the endothelium selectively enhanced T-cell diapedesis into the tissue, without affecting rolling and adhesion.

Direct cross-talk between the TCR, chemokine receptors and flow has recently been shown to be essential for antigen-induced T-cell migration.^17,52–55 The zeta-associated protein 70 (ZAP-70), a key element in TCR signalling, is required for CXCR4 signal transduction in human T cells.⁵⁶ CXCL12 (the ligand for CXCR4) stimulates the physical association of CXCR4 and the TCR and utilizes the ZAP-70 binding immunoreceptor tyrosine-based activation motifs (ITAMs) of the TCR for signal transduction.⁵⁷

Other studies, however, have found no influence of antigen on the entry of lymphocytes into a given tissue.^58,59 In a transgenic delayed-type hypersensitivity (DTH) model, there was enhanced recruitment of both antigen-non-specific and antigen-specific effector T cells into antigenic cutaneous tissue but no selective antigen-specific T cells trapping was found.⁶⁰ However, the specific T cells that arrived at the site started to proliferate locally after a few days, resulting in a cellular infiltrate that was strongly enriched for cognate T cells (C. Doebis and A. Hamann, unpublished).

The relative contribution of TCR-induced and non-antigen-specific signals to memory T-cell recruitment is likely to be determined by the severity of the inflammatory process. It is plausible that TCR-mediated control of primed T-cell localization to target sites may be essential to ensure efficient, rapid memory responses in the presence of limited inflammatory signals, for example at the early stages of a recall response. For example, insulin-specific H2-K^d-restricted T cells are efficiently recruited to pancreatic islets of various H2-K^d-positive mouse strains that are free of pre-existent inflammation.⁵² By contrast, very severe inflammation or a pre-existing large antigen-specific T-cell repertoire (for example during direct alloresponses⁵⁹) may override the requirement for selective antigen-dependent T-cell recruitment.

Experimentally, however, it is often difficult to discriminate direct effects of antigenic stimulation on recruitment processes from indirect effects where a few antigen-specific T cells (‘pioneer cells’⁶⁰) are required to boost (non-specific) recruitment of T cells into the tissue.

Regulation of T-cell migration by costimulatory molecules

Costimulatory signals (such as those mediated by CD28) delivered to T cells, in conjunction with TCR engagement, are required to sustain T-cell division, differentiation and survival.^61–63 Negative costimulators [such as cytotoxic T-lymphocyte antigen 4 (CTLA-4)] counteract these effects, thus promoting homeostatic mechanisms and preventing autoimmunity.

These costimulators have been shown to regulate adhesion molecules and intracellular mediators of cytoskeletal rearrangement in vitro.^64–70

In vivo, CD28-mediated signals promote the localization of T cells to target tissue following priming. A prominent feature of CD28-deficient immune responses is the inefficient localization of primed T cells to non-lymphoid antigenic sites.^61,71,72

We recently reported that intact CD28 signalling is required for primed T cells to leave lymphoid tissue and migrate to antigenic sites following priming.⁷³ TCR-transgenic T cells carrying a mutation in the cytoplasmic tail of CD28 (CD28^Y170F) that abrogates phosphatidylinositol-3-kinase (PI3K) recruitment, without leading to defects in clonal expansion,⁷⁴ failed to localize to target tissue following priming.

The mechanism by which CD28 promotes migration of primed T cells to target tissue is unclear. CD28 does not appear to directly mediate adhesion,⁷⁵ but may favour primed T-cell migration to non-lymphoid tissue by inducing integrin-mediated adhesion.⁷³ The long-term effect of CD28-mediated signals on T-cell migration⁷³ suggests that additional mechanisms, such as transcriptional regulation of chemokine receptor expression,⁷⁶ are likely to be involved.

Despite sharing these adhesion-inducing and pro-migratory properties in vitro,⁷⁷ CTLA-4-mediated signals lead to effects antagonistic to those induced by CD28 on T-cell migration in vivo. CTLA-4 ligation reduced conjugate formation with cognate DCs and their retention in lymph nodes in response to antigen, suggesting that CTLA-4 engagement may limit the expansion of specific T cells by reducing their cumulative interactions with cognate DCs. In addition, tissue infiltration by a murine HY-specific H2-K^k-restricted T-cell clone was abrogated by CTLA-4 ligation,⁷³ suggesting that CTLA-4 engagement can antagonize recruitment of primed T cells to target tissue mediated by antigen-induced signals.

A number of costimulatory molecules other than CD28 and CTLA-4 have been implicated in the regulation of memory T-cell migration. For example, OX40/OX40 ligand interactions are required for T-cell migration to germinal centres,⁷⁶ and the inducible costimulatory molecule (ICOS) regulates human memory T-cell migration through tumour necrosis factor (TNF)-α-treated endothelial cells.⁷⁸

Egression of memory T cells from target tissue

Most studies on the location of effector/memory T cells in non-lymphoid tissues have focused on entry (homing) or proliferation and survival as determining factors of lymphocyte content in a given tissue. Recent findings have shown that exit from the tissue is an active process controlled by chemotactic mechanisms. The chemokine receptor CCR7 was shown to be required for T-cell exit from inflamed peripheral tissue.^79,80 Another chemotactic agent, sphingosin-1-phosphate (S1P), and its receptors are required for the exit from lymph nodes, a finding emerging from studies with the drug FTY 720, which displays immunosuppressive effects. Both CCR7 and S1P receptors are modulated in the course of T-cell activation, and thereby might cause the transient retention of recently activated T cells in the lymph node.⁸¹ When CCR7 is knocked out, the number of T cells retained in an inflamed tissue doubles, confirming its importance for continuous circulation.⁸²

For technical reasons, quantification of exit rates for specific subsets of cells and specific tissues is more difficult. However, a variety of data are available from early studies in which cannulation of the thoracic duct or even single lymph nodes was applied, which provided clear evidence not only that naïve cells entering a lymph node via the high endothelium pass through the tissue within half a day and exit it, but also that large numbers of effector/memory cells attracted to an inflamed tissue, or generated by local proliferation, exit the tissue via the efferent lymph.⁸³ It is conceivable that the process of emigration also underlies a variety of regulatory influences; T-cell activation upon antigen encounter within the tissue may be one factor, but an influence of inflammation-generated mediators such as prostaglandins has also been described.⁸⁴

Chemorepulsion and fugetaxis

The directional movement towards a chemical compound plays a major role in the recruitment as well as the egress of T cells from the site of an immune response. Leucocytes are able to integrate signals from multiple chemoattractants in their migration.⁸⁵ In fact, cells migrating away from a local chemoattractant source actually chemotax towards distant attractants. The ability to navigate through chemoattractant arrays may be sufficient to explain entry and egress of T cells during an immune response. However, recent evidence supports the existence of both chemoattractants and chemorepellents that guide the directed movement of leucocytes into and out of tissues. Chemorepulsion is defined as the migration away from peak concentrations of a chemokine and was initially studied in the context of axonal guidance, where the same molecule may act as a chemoattractive or chemorepulsive cue depending on the receptor expressed on the cell surface.^86,87

The term ‘fugetaxis’ is considered a synonym for chemorepulsion, but the former has been introduced to define an agent that can either attract or repel cells, depending on its concentration, while using the same receptor.⁸⁸

Chemotaxis and chemorepulsion in the context of T-cell trafficking have been studied in the process of thymic emigration. Egress of mature thymocytes from the medulla to the periphery has been shown to be orchestrated by chemoattraction exerted by S1P and a simultaneous fugetactic function of CXCL12, which induces cells to leave the thymus.^81,89

A bimodal effect of chemokines on memory T-cell trafficking has also been demonstrated in cancer. Certain growing tumours initially generate the chemokine CXCL12 at a level that induces T-cell chemoattraction, but ultimately establish an immune-privileged site through the chemorepellent effect of high levels of CXCL12 on tumour-specific T cells. In this setting, T-cell chemorepulsion impairs cytotoxic T lymphocyte-mediated lysis of tumour cells, which requires that the effector makes direct contact with the target cell.⁹⁰

Fugetaxis and chemorepulsion may coexist in situations where the concentration of the chemokine drives cells from chemotaxis to fugetaxis, but dual receptor engagement may take place. In fact, it has been shown that the chemokine CXCL12 mediates a concentration-dependent chemorepulsive effect on diabetogenic T cells by altering firm adhesion. As this effect is G-protein-coupled receptor dependent but is only partially reversed by CXCR4 blockade, it has been suggested that alternative downstream CXCL12 signalling pathways mediated by protein coupled receptor 1 (RDC1)/CXCR7 may trigger chemorepulsion.⁹¹

Memory T-cell anchorage

Memory plasma cells reside on CXCL12-expressing stromal cells of bone marrow and rest there for a long periods.^92–94 Until recently, evidence demonstrating the existence of survival niches for memory CD4 T cells has been elusive.^95,96 In immune reactions characterized by long-term antigen persistence (virus or adjuvants), memory-phenotype CD4 T cells are found in the spleen and lymph nodes for long periods.^97,98 In contrast, following immunization in the presence of soluble adjuvants (lipopolysaccharide or monophosphoryl lipid A), memory CD4 T cells in the spleen or lymph nodes substantially decrease in number 1 week after immunization.^99,100 These T cells have been shown to locate to the bone marrow and rest on IL-7-expressing stromal cells of the bone marrow.⁹⁹ The relocation of antigen-experienced CD4 T cells to the bone marrow is dependent on integrin α2β1, a collagen receptor. Inhibition of integrin α2β1 on primed CD4 T cells results in defective relocation of antigen-specific CD4 T cells to the bone marrow and reduced B-cell help (e.g. reduced affinity maturation). It is still unknown how the memory T cells migrate to their survival niches in the bone marrow, although they express CCR2 and CXCR6.⁹⁹ The bone marrow is presumably the most best tissue for long-term localization of CD4 T cells primed by blood-borne antigen. In addition, the gut and skin have also been reported to provide anchorage niches for memory CD4 T cells.¹⁰¹

Conclusions

Every decade of immunological research appears to reveal novel functional subsets of T cells. How this expanding universe of specialists becomes co-ordinated and appropriately targeted to the hot-spots of immunoreactivity would have remained a mystery if, at the same time, our knowledge of the mechanisms of cell trafficking had not greatly improved. Co-operating adhesion molecules and chemokine receptors equip the migrating cells with an almost unlimited combinatorial diversity which allows them to recognize the signatures defining tissues and compartments, to distinguish different inflammatory processes depending on the kind of triggers, site of inflammation, or involved cell populations and so on. The recent key advances discussed in this review are summarized in Fig. 1. Monitoring of the migration of T-cell subsets associated with immune-mediated diseases may prove to be essential in allowing us to understand pathogenic mechanisms, to design prognostic and therapeutic tools and to predict therapeutic responses.¹⁰² If these goals are to be achieved, we must address the many unanswered questions highlighted in this review.

Disclosures

The authors have no financial conflict of interest.