Lymphocyte homing to allografts.

Abstract

A complex system of lymphocyte trafficking has evolved to provide immune surveillance and recognition of foreign antigens (1). When the regulation of this process is disrupted leukocyte recruitment continues inappropriately resulting in pathological inflammation (2, 3). During graft rejection activated lymphocytes are recruited to and retained in the graft (4) and therefore the factors that regulate lymphocyte recruitment are crucial in determining the outcome of allorecognition. In this article we review the current understanding of how lymphocytes recognize and bind to endothelial cells and how these interactions regulate T cell recruitment into tissue. We discuss the implications of leukocyte recruitment in the context of allograft rejection. LYMPHOCYTE RECIRCULATION UNDER PHYSIOLOGICAL CONDITIONS Lymphocyte recruitment from the circulation into tissue is dependent on the ability of lymphocytes to bind to molecules on endothelial cells that promote adhesion and transendothelial migration into tissue. A multi-step model of leukocyte adhesion to vascular endothelium has been described and is broadly applicable, although the details of the signals involved will differ depending on the setting. In the generally accepted model (1, 2, 5) tethering or rolling receptors expressed on endothelial cells, capture free flowing leukocytes. These receptors may be members of either the selectin family of adhesion proteins (6) or the immunoglobulin superfamily (7–9). Once captured, the leukocyte can receive activating messages presented by endothelial cells in the form of chemotactic cytokines or “chemokines,” which bind to specific G protein coupled receptors on the leukocyte surface (10–12). Under some circumstances cell surface receptors, including the L-selectin ligand glycosylation dependent cell adhesion molecule 1 (GLYCAM-1), CD31 and CD73, can also trigger rapid integrin activation (13, 14). Occupancy of these receptors triggers a cascade of intracellular signals that result in presentation of high affinity integrin receptors on the leukocyte surface. These activated integrins then bind competently to their immunoglobulin family receptors expressed on the endothelium to promote arrest and firm adhesion of the leukocyte to the vessel wall (15–17). In the presence of the appropriate migratory signals the leukocyte will then migrate across the endothelium into tissue where it follows a hierarchy of chemotactic signals towards the focus of inflammation (see Fig. 1) (18). Figure 1: Adhesion of lymphocytes to endothelial cells under conditions of flow. Free-flowing lymphocytes in the circulation are captured by tethering receptors (usually carbohydrate dependent selectins) expressed on endothelial cells that induced the cell to roll on the vessel wall bringing it into contact with chemokines immobilized in the endothelial glycocalyx. Chemokine activate specific, G-protein-linked receptors on the lymphocyte that results in a conformational activation of lymphocyte integrins to a high affinity state, permitting firm adhesion to endothelial-expressed immunoglobulin adhesion molecules. Chemokine recognition also results in cytoskeletal reorganization within the adherent lymphocyte, which facilitates migration across the endothelial monolayer and into the tissue. Once within the tissue the leukocyte follows a hierarchy of chemotactic gradients toward the site of inflammation. The shaded triangles represent the relative contribution of the different classes of adhesion molecule to each step of the cascade (adapted from Adams and Shaw, Lancet 1994 and Picker and Butcher, Science 1996).HOMING PATTERNS OF NAÏVE AND MEMORY/EFFECTOR LYMPHOCYTES Adhesion of lymphocytes to endothelial cells within different tissues appears to follow this paradigm, under both physiological and pathological conditions. However, subsets of lymphocytes display different receptors that will alter their propensity to be recruited to different sites. This is best illustrated by the marked differences between naïve and memory T lymphocytes (Fig. 2). Figure 2: Adhesion molecules involved in lymphocyte endothelial interactions. Details are given in the text. PNAd, Peripheral node addressin a complex that contains CD34, podocalyxin, GlyCAM-1, and Sgp 200. Adapted from Salmi, Adams, and Jalkanen, Am J Physiology 1998.Naïve lymphocyte migration. These cells migrate almost exclusively between the circulation and secondary lymphoid tissues whereas memory cells are largely excluded from lymph nodes and instead migrate into tissue, returning to the circulation via lymphatics (19). The recruitment of naïve T cells to secondary lymphoid tissue is regulated by their ability to recognise specific molecules on lymph node high endothelial venules (HEV) (20). These molecules include endothelial adhesion molecules, such as the peripheral node addressin (PNAd) which binds to L-selectin on naive T cells and the chemokine SLC (secondary lymphoid tissue chemokine) that binds to a receptor, CCR7, preferentially expressed on naive T cells (1, 21–24). Antigen presenting cells (APCs) in lymph nodes have the capability to activate naïve T cells and these interactions are also regulated by chemokines. For example, another CCR7 ligand called, EBV-induced molecule-1 ligand chemokine (ELC) , is secreted by dendritic cells (DCs) in the T cell compartment of the lymph node and acts to bring T cells and activated DCs (which also express CCR7) together thereby promoting optimal T-APC interactions (22, 25, 26). Memory/effector T cell migration. After activation in secondary lymphoid tissue, naïve T cells differentiate into activated effector cells and long-lived memory T cells. These cells exhibit different migratory pathways dictated by changes in their cell surface expression of adhesion receptors (27). For example, activated T cells lose expression of L-selectin and CCR7 (which prevents them from binding efficiently to HEV in lymph nodes) but up-regulate molecules such as integrins and the chemokine receptors CCR2, CXCR3, and CCR5 that promote adhesion to activated endothelium in inflamed tissue (28–31). If the lymphocyte is activated in tissue it down-regulates the expression of chemokine receptors such as CCR2 and CCR5 while activating integrin-mediated adhesion to cellular and extracellular matrix ligands resulting in immobilization of the lymphocyte at the site of antigen exposure (32, 33). Thus in the main, memory T cells respond to tissue derived inflammatory signals but not to the physiological signals that drive recruitment to lymph node. However, recent studies suggest that a subset of Th1 memory T cells maintain CCR7 expression allowing them to recirculate through lymph nodes but excluding them from B cell areas and preventing them from providing B cell help (34, 35). Differences in adhesion molecules determine the migration patterns of Th1 and Th2 cells. An even more subtle regulation of adhesion molecule expression has been reported on Th1 and Th2 lymphocytes. Because these functional subsets are delineated by distinct patterns of cytokine secretion differences in Th1 versus Th2 cell recruitment could effect the outcome of alloactivation. Th1 cells secrete interleukin- (IL) 2, interferon-γ (IFN-γ) and tumor necrosis factor (TNFβ) whereas Th2 cells secrete IL-4, IL-5, IL-10, tumor growth factorβ (TGFβ), and IL-13 and have been associated with suppression of allograft rejection in some models (36). There are marked differences in the expression of chemokine receptors and adhesion molecules on Th1 and Th2 cells reflecting their different requirements for recruitment to tissue (37). Although both Th1 and Th2 lymphocyte subsets express the P-selectin ligand PSGL-1, only Th1 cells bind both E- and P-selectin (38), a property determined by their ability to glycosylate PSGL-1. The fucosyl transferases (39, 40) responsible for this glycosylation show differential expression between Th1 and Th2 cells (41) and determine whether the cell will bind P-selectin at sites of inflammation (42). The expression of these enzymes is influenced by the local cytokine milieu. For example, IL-4 down-regulates α3 fucosyltransferases in Th2 cells thereby altering their ability to home to inflamed sites. However, although the differences between Th1 and Th2 cells are clear-cut in vitro, current evidence suggests that the expression of chemokine receptors and the ability of T cells to bind selectins may be more strongly associated with tissue specificity rather than in determining Th1 versus Th2 responses in vivo (43, 44). TISSUE SPECIFIC EXPRESSION OF ADHESION RECEPTORS AND CHEMOKINES In addition to acquiring molecules which promote their recruitment to inflamed tissue there is evidence that memory cells display tissue-specific homing receptors that allow them to recognize endothelium in the tissue draining into the lymph node where they were originally activated (1, 18, 27). Hence, a T cell activated in an axillary lymph node will subsequently recirculate preferentially to the skin, whereas one activated in a mesenteric node will be programed to home to the gut (19). This tissue tropism is facilitated by organ-specific expression of endothelial molecules; particularly those involved in lymphocyte capture. For example, recruitment of lymphocytes to the gut is mediated by an endothelial molecule mucosal addressin cell adhesion molecule-1 (MAdCAM-1) which is largely restricted to mucosal vessels and its ligand α4β7 integrin, which is preferentially expressed on T cells that display gut tropism (45). In contrast memory T cells that infiltrate the skin express a unique skin-homing receptor called the cutaneous lymphocyte-associated antigen (CLA) which results from fucosyltransferase VII-mediated glycosylation of the cell-surface receptor P-selectin glycoprotein ligand-1 (PSGL-1) (46). Although PSGL-1 is expressed constitutively on most human T cells, CLA is found on memory T cells in the skin but not on T cells infiltrating other inflammatory sites (47, 48). Expression of CLA by T cells facilitates their homing to inflamed skin by promoting adhesion to E-selectin on dermal vessels (49). Tissue-specific endothelial adhesion molecules have also been proposed for the liver (50–52) and lung (53–55). There is now evidence that chemokine receptor expression also determines where lymphocytes are recruited. Thus, skin homing T cells express high levels of CCR4, the receptor for TARC, a chemokine that is expressed at inflammatory sites in the skin (56). In contrast gut and liver homing T cells express very little CCR4 but instead high levels of CCR5 the receptor for the chemokines regulated with activation normal T cell expressed and secreted (RANTES) and macrophage inflammatory protein- (MIP) 1α and β (30) and gut-derived cells express CCR9 (18, 57). Two factors, therefore, determine where a lymphocyte is recruited; the adhesion molecules and chemokine receptors expressed on the lymphocyte (which will be determined by the microenvironment in which the cell was activated) (58) and the combination of chemokines and adhesion counter-receptors expressed on the endothelium in the target tissue. REGULATION OF ADHESION MOLECULE AND CHEMOKINE EXPRESSION BY ENDOTHELIUM Under basal conditions the expression of endothelial adhesion molecules in a particular tissue will be determined by signals from the microenvironment. For example lymphotoxin (TNFβ) is essential for the development of mucosal lymphoid tissue and the expression of MAdCAM-1 (59). The nature of many of these signals is currently poorly understood but will be crucial to understanding tissue specificity of homing (60). Adhesion molecules expressed by activated endothelium. The signals responsible for activating endothelium in the presence of inflammation are understood. Molecules such as intercellular adhesion molecule-2 (ICAM-2) and CD31 are constitutively expressed on endothelial cells and change little with inflammation. Other molecules such as ICAM-1 are normally expressed at low levels but increase after activation, whereas others such as E-selectin and vascular cell adhesion molecule-1 (VCAM-1) only appear after activation (61). The local presence of proinflammatory cytokines such as IL-1, TNFα, or IFNγ, causes increased expression of many endothelial adhesion receptors, thereby extending the range of leukocyte subsets that can be recruited (62). The precise phenotype of activated endothelium depends on the nature of the local inflammatory response because some cytokines have differential effects on adhesion molecules. For example, a combination of IL-4 and TNFα selectively up-regulates VCAM-1 on endothelial cells in culture (63) (64), and IFNγ can inhibit activation-induced selectin expression (65). The time course of endothelial adhesion molecule expression differs. For example P-selectin is stored as intact protein in endothelial Weibel-Palade bodies and can be rapidly mobilised to the cell surface in response to agonists such as hydrogen peroxide and histamine (66). In contrast, E-selectin, in common with most endothelial adhesion receptors, requires de novo protein synthesis and takes several hours to be expressed (61). The leukocytes recruited to tissue can themselves amplify endothelial activation by secreting cytokines and via poorly understood adhesion-dependent mechanisms. For example, engagement of endothelial ICAM-1 by lymphocyte leukocyte function associated molecule-1 (LFA-1) results in increased secretion of chemokines and expression of adhesion molecules such as VCAM-1 that promote transendothelial migration (67–69). In addition endothelial cells express CD40, a member of the TNF receptor superfamily that can interact with its ligand, CD154 expressed on activated lymphocytes. Expression of CD40 increases with cytokine treatment of endothelial cells in vitro and engagement of CD40 with trimeric CD154 increases expression of the adhesion molecules, E-selectin, VCAM-1, and ICAM-1. CD40 can be detected immunocytochemically on endothelium in rejecting human allografts (70) and may thus acts as a signaling receptor to promote lymphocyte infiltration to the graft (71). Thus as the local inflammatory response develops the phenotype of the endothelium will change over time accounting for the sequential recruitment of different subsets of leukocytes. Endothelial secretion and presentation of chemokines. The nature of an inflammatory stimulus will also determine the local secretion and presentation of chemokines by the endothelium. Their expression can be rapidly induced in most cell types following stimulation by a variety of agents including bacterial products, viruses, cytokines and activation of cell surface receptors (11, 72). Activated T cells express a range of chemokines at both the mRNA and protein levels and non-hematopoeitic cells including endothelial and epithelial cells are potent sources of many chemokines. The chemokines secreted in response to particular stimuli show differences between cell types. For instance epithelial cells secrete large amounts of the CXC chemokines ENA-78 and IL-8 in response to LPS whereas they fail to respond to IL-10 or IFNγ (73). However at sites of chronic inflammation local IFN-γ secretion causes expression of the CXCR3 ligands IP-10, MIG and I-Tac by endothelium thereby promoting the recruitment of effector lymphocytes and monocytes (30, 74). Infiltrating monocytes and activated lymphocytes are a major source of the chemokines that will determine the subsequent composition and duration of the inflammatory response. For example, CD8+ CTL, which are antigen specific for myelin proteolipid protein peptides, a putative antigen in multiple sclerosis, secrete the chemokines MIP-1α, MIP-1β, IL-16, and IP-10 that act to recruit CD4+ T cells of the same TCR specificity. Thus CD8+ cytotoxic T cells can promote and maintain inflammatory responses by recruiting specific CD4 subsets (75). If they are to trigger adhesion and migration effectively at the endothelial surface chemokines must be retained at the vessel wall to permit interaction with circulating leukocytes. This is achieved by immobilizing chemokines on proteoglycans in the endothelial glycocalyx (16, 76) or by their binding to nonsignaling receptors such as the Duffy antigen on red cells (DARC) (77) and D6 receptors (78). Chemokines show differential binding to proteoglycans and because proteoglycans vary from tissue to tissue and with activation, this provides a mechanism by which tissues can selectively express a particular cohort of chemokines (79). The sophistication of chemokine presentation by endothelium is further enhanced because endothelial cells can both present chemokines secreted by underlying cells by a process of transcytosis (62) and also capture (80) chemokines produced “upstream” in the circulation (81, 82). HOW DO ADHESION MOLECULES AND CHEMOKINES ALTER THE OUTCOME OF ALLOGRAFT TRANSPLANTATION? Alterations in adhesion molecule and chemokine expression by endothelial cells and of adhesion molecules and chemokine receptors by leukocytes determine when and where leukocytes adhere to the vessel wall and penetrate tissue. It thus follows that such mechanisms will regulate the recruitment of leukocytes to allografts and determine the outcome of allorecognition. Vascular endothelial cells within transplanted organs can be activated to express a wide range of adhesion molecules and chemokines. The sequence of endothelial activation in allograft tissue follows that seen in other inflammatory settings and coincides with T cell infiltration into the graft (83). Chemokines are up-regulated early on several graft structures, including the epithelium and endothelium (84, 85) and local secretion is then amplified by infiltrating lymphocytes (86). There is now good experimental and clinical evidence that endothelial activation occurs as a consequence of damage to the graft during the process of preservation and surgical manipulation and that this promotes the recruitment of leukocytes during reperfusion injury (87, 88). Early graft damage also activates the alloresponse, probably by providing “natural adjuvants” for lymphocyte activation as well as activating endothelium thereby enhancing the subsequent recruitment of alloreactive lymphocytes during rejection (89, 90). Experimental and clinical studies have demonstrated that adhesion molecules, including VCAM-1, ICAM-1, and E-selectin and chemokines such as IL-8 and MIP-1β are up-regulated in graft tissue during preservation. Furthermore, this early activation has been associated with subsequent rejection (85, 91–93) and may explain the link between reperfusion injury and graft rejection (92). The role of selectins in lymphocyte trafficking to allografts. Endothelial E- and P-selectin are particularly important in regulating neutrophil recruitment during reperfusion injury. This is confirmed by the ability of selectin ligand mimetics to reduce neutrophil accumulation and the severity of reperfusion injury (94). Selectins are also important in lymphocyte recruitment. However, it is likely that selectins will play different roles in different tissues or organs. Thus Th1 responses in the skin are associated with enhanced lymphocyte binding to endothelial selectins whereas T cells at Th1 driven sites of inflammation in the gut do not express E- or P-selectin receptors (43). This might explain why lymphocytes in rejecting liver allografts do not bind E-selectin (48). In contrast there is in vivo evidence that P-selectin is important for lymphocyte recruitment to lung allografts. When a sialyl Lewis X analogue (SLX), was used to block P-selectin function in a rat model of lung transplantation there was a reduction in the severity of graft rejection (95). Selectins expressed by other cell types such as platelets and lymphocytes may also be of significance during allograft rejection. Platelet adhesion to graft endothelium occurs very early (96) and the high levels of P-selectin expressed by such activated, adherent platelets can serve as a focus to capture circulating lymphocytes (97). There is also evidence for involvement of lymphocyte expressed L-selectin in lymphocyte recruitment to rejecting allografts (98). Lymphocytes fail to migrate into skin grafts in L-selectin-deficient mice and here the severity of rejection is reduced compared with wild- type animals (99). Functional L-selectin ligands have been detected on the endothelium of vascularised allografts during rejection in animals and in human transplants (100, 101). Toppila and colleagues (102) reported increased expression of L-selectin ligands on the endothelium of human heart transplants, which correlated directly with lymphocyte infiltration and the severity of graft rejection. Furthermore, this study revealed that the expression of L-selectin ligands was an early event in rejection associated with the initiation as well as the maintenance of lymphocyte infiltration (101). The ability to inhibit selectin-mediated adhesion with glycans/oligosaccharides makes them attractive new targets for immunotherapy in transplantation (103). The role of other adhesion molecules in lymphocyte trafficking to allografts. In some circumstances molecules other than selectins may mediate the capture phase of lymphocyte recruitment. Two examples are the gut addressin MAdCAM-1, which can promote lymphocyte rolling via interactions with either L-selectin or the integrin α4β7 and vascular adhesion protein-1 (VAP-1), which mediates sialic acid dependent capture of lymphocytes to lymph node high endothelial venules and liver sinusoids (50). Both these molecules display preferential expression in particular tissues (51, 104, 105) but whether they maintain this tissue specificity in the context of allograft transplantation is not known. Given the role of MAdCAM-1 in lymphocyte recruitment to the gut it would seem likely that it will be involved in lymphocyte trafficking to intestinal transplants and preliminary animal studies support this (106). CD44 is another adhesion molecule that can function both as a costimulatory molecule and to support lymphocyte adhesion to endothelium. Blockade of CD44 led to reduced lymphocyte infiltration in a rat skin transplant through different mechanisms depending on which isoform of CD44 was inhibited. Antibodies against the standard isoform blocked homing of activated lymphocytes into the graft, whereas antibodies against the CD44 variant 6 inhibited clonal expansion of donor-specific T cells (107). The role of integrins in mediating lymphocyte adhesion to graft endothelium. There is evidence from clinical and animal studies for the involvement of both β1 and β2 integrins in mediating lymphocyte adhesion to graft endothelium. The endothelial expression of VCAM-1 and ICAM-1 increases during rejection and both molecules can support lymphocyte adhesion as demonstrated by binding studies (108–110). In several models of vascularized transplants treatment with blocking antibodies to either LFA-1/ICAM-1 or VLA-4/VCAM-1 reduced lymphocyte infiltration into the graft and prolonged graft survival (108, 111–114). The relative efficacy of blocking each pathway varies between models (115) with maximal effects usually seen when both pathways are blocked (116, 117). Thus, although it is accepted that both β1 and β2 integrin dependent mechanisms can promote lymphocyte endothelial adhesion the precise role of each pathway will differ between tissues and in response to different inflammatory stimuli. Furthermore, the interpretation of studies in which LFA-1/ICAM-1 and VLA-4/VCAM-1 are blocked in vivo is complicated because in addition to supporting lymphocyte recruitment both pathways are also involved in alloactivation (118). Because the early animal studies showed promise, anti-adhesion molecule therapy has been tried in several clinical studies. As early as 1991 a murine monoclonal antibody directed against the alpha chain (CD11a) of the human LFA1 molecule was used to treat histologically documented episodes of acute renal rejection. Although the mAb was well tolerated in most patients it was ineffective in controlling rejection (119). In 1993 a phase I clinical trial of anti-ICAM-1 therapy in high-risk renal transplantation recipients reported decreased rejection and good survival in patients who achieved adequate anti-ICAM-1 serum levels (120). More recently the European Anti-ICAM-1 Renal Transplant Study (EARTS), reported the results of cadaveric renal recipients given either anti-ICAM-1 (enlimomab) or a placebo for 6 days together with conventional triple immunosuppression. There was no significant differences in the incidence of rejection, delayed graft function, patient or graft survival (121). The role of chemokines in lymphocyte trafficking to allografts. The requirement for integrin activation to convert rolling adhesion into static adhesion during lymphocyte recruitment suggests that chemokines expressed by graft endothelium will be crucial regulators of lymphocyte entry. Evidence from animal and human studies shows increased expression of several chemokines that act on lymphocytes in allograft rejection. Human studies remain relatively limited but have documented increased expression of RANTES, MIP-1α, MIP-1β, and IP-10 during graft rejection (84, 85, 122). Animal studies allow the dynamics and function of chemokine expression to be studied in more detail (123, 124). Using a skin allograft model Kondo et al. (125) demonstrated that the expression of RANTES, MIP-1α, MIP-1β, and IP-10 in the graft was increased in presensitized mice and during second set rejection and that this expression was dependent on the presence of T cells. Recent studies have shown the functional importance of chemokines in allograft rejection. The use of met-RANTES, a nonfunctional analogue of RANTES, leads to reduced lymphocyte recruitment and graft damage as a consequence of inefficient lymphocyte capture by graft endothelium (126). More recently suppressed acute and chronic rejection of cardiac allografts has been reported in mice that lack the RANTES MIP-1α receptor CCR1 (127). The use of antibodies that recognise specific chemokine receptors allows lymphocyte subsets capable of responding to specific chemokines to be detected. The RANTES, MIP-1α, MIP-1β receptor CCR5 has been shown to be increased on infiltrating lymphocytes in both renal and liver allograft rejection (128), strengthening the role for CCR5 ligands in allograft rejection. Another chemokine receptor which appears to be critical for effector cell entry into inflamed tissue is CXCR3 (30, 129). The ligands for CXCR3, ITAC, IP-10, and Mig are inducible by interferon and preliminary evidence suggests that they are up-regulated in graft rejection in association with infiltration by CXCR3 expressed T cells (130). In the future it will be important to determine which chemokines and chemokine receptors are involved in determining the outcome of allorecognition. For instance if graft acceptance is promoted by a switch in local chemokine production to support the recruitment of Th2 over Th1 type cells, manipulation of chemokines or chemokine receptor function may allow allogeneic responses to be modulated toward the recruitment of lymphocytes that promote graft acceptance. The development of animal models with deficient chemokines and chemokine receptors will determine the specific roles of particular receptor/ligands more precisely (127) and will be reviewed in a subsequent article. DOES ALLOANTIGEN RECOGNITION ALTER LYMPHOCYTE RECRUITMENT TO GRAFT TISSUE? Although mechanisms that regulate lymphocyte endothelial adhesion are predominantly antigen-independent, recent work suggests that cognate recognition of alloantigens on endothelial cells by T cells can effect not only the state of lymphocyte activation but also their ability to undergo transendothelial migration (131). Because endothelial cells in vascularised allografts express MHC molecules they have the potential to activate interacting T cells (132) via the TCR, a powerful stimulus signal for the induction of the high affinity integrin conformation (33), resulting in lymphocyte arrest. A recent study has provided more evidence for an antigen-dependent mechanism in lymphocyte recruitment to allografts. Marrelli-Berg and colleagues (133) demonstrated that presentation of antigens by MHC molecules on endothelium to T cell clones, enhanced the rate of transendothelial migration suggesting that antigen presentation by endothelium can augment the recruitment of antigen-specific T cells into tissues (133). CONCLUSIONS The last 10 years have seen major advances in understanding the molecular regulation of lymphocyte endothelial interactions. It is likely that the endothelium within vascularized allografts will behave like any other acute inflammatory tissue and express molecules that promote lymphocyte recruitment during graft rejection. Understanding the nature and function of these molecules has important implication for therapy. The endothelium in the graft may be modified by genetic manipulation to prevent expression of molecules such as ICAM-1 and VCAM-1 which appear to be critical for lymphocyte entry (134). The future development of drugs or biological agents that inhibit adhesion molecule function may add to the armamentarium of immunosuppressive therapy. One problem with the latter approaches is that targeting widely expressed adhesion molecules will also inhibit lymphocyte recirculation to host tissues. If specific molecules regulate lymphocyte recruitment to particular tissues the inhibition of these molecules might deliver tissue-specific immunosuppression and leave generalized lymphocyte recirculation intact. Potentially the most exciting approach is to modulate the nature of the lymphocyte subsets recruited to the graft so that harmful cells are excluded and beneficial subsets preferentially recruited. For example, the increasing understanding of the signals, particularly chemokines, that control the recruitment of lymphocyte subsets to tissue raises the possibility of therapeutic strategies in which the recruitment of protective immunosuppressive cells is favoured. However, in vivo attempts to inhibit lymphocyte endothelial interactions have been disappointing, probably reflecting the complexity of the molecular interactions involved and the large numbers of chemokines produced by the graft. Furthermore, studies using animals deficient in chemokine receptors have shown surprising results suggesting that predicting the outcome of inhibiting particular chemokine/chemokine receptor interactions will be difficult (135).