Extracellular matrix proteins in organ transplantation.

Abstract

Traffic of leukocytes from the bloodstream to specific sites of inflammation is regulated by signals that stimulate cellular adhesion to vascular endothelium or extracellular matrix (ECM) proteins (1, 2). To migrate to the extravascular areas, circulating leukocytes have to acquire strong adhesion interactions to the vessel wall while resisting to continuous sheer forces (3). The emerging relevance of leukocyte-ECM interactions to cell adhesion, migration, and positioning within specific tissue microenvironments emphasizes the role ECM proteins may play in the functioning of the immune system in both physiological and pathological conditions (1, 4). This review will focus on some of the best characterized ECM proteins, fibronectin (FN) in particular, one of the most studied ECM components in organ transplantation. The text is divided into three integral parts, which include a general overview of ECM proteins and their cellular receptors, a narrative of ECM proteins in organ transplantation, and a discussion of ECM-leukocyte interactions as potential targets for therapeutic intervention. ECM PROTEINS AND THEIR CELLULAR RECEPTORS FN is a large dimeric glycoprotein composed of a series of independently folding modular domains known as FN repeats I, II, III (5). These repeating units contain regions or domains that interact with a number of other matrix molecules as well as cells, and mediate functions such as cell migration, matrix assembly, differentiation, and proliferation (6, 7). Structural diversity in FN arises by regulated alternative splicing of a single gene transcript in three segments, termed EIIIA, EIIIB, and V in rats, and ED-A, ED-B, and IIICS, respectively, in humans (8, 9) (Fig. 1). The EIIIA (or ED-A) and EIIIB (or ED-B) domains are either included or excluded as intact type III homology repeats. The V (or IIICS) may be excluded, partially included, or fully included. The latter form contains the connecting segment-1 (CS1) region known to mediate leukocyte adhesion (10). The FN variants that include the EIIIA and EIIIB segments are prominent in FN produced during embryogenesis, and their expression in the adult tissues is minimal unless under certain pathological conditions (11, 12). For example, adult liver synthesizes the plasma form of FN, and this form excludes the EIIIA and EIIIB domains. However, the EIIIA and EIIIB segments are markedly increased during cutaneous wound healing (13, 14), neointimal hyperplasia (15, 16), and glomerular nephritis (17, 18). Alternative splicing of the FN gene allows for the generation of multiple isoforms, with 20 isoforms possible in humans (19). The role of FN in lymphocyte adhesion, migration, and activation has been extensively illustrated (20–24). Figure 1: FN structure and variants. A diagrammatic representation of FN showing the three regions of alternative splicing designated EIIIA, EIIIB, and V in rats. FNs are mainly composed of three types of homologous structural units called I (rectangles), II (triangles), and III (ovals), which define several domains interacting with different molecules. Note that the cell binding domain RGD that interacts with the α5β1 integrin and the cell adhesive segment CS1 that interacts with the α4β1 integrin are included in FN-III(8–11) and in V120, respectively.Tenascin (TN), another ECM oligomeric protein functionally associated with FN, consists of an amino- terminal cystein-rich region followed by a series of EGF-like domains, a series of FN type III repeats, and a fibrinogen-like region at the carboxyl terminal (25, 26). The findings that TN is present normally and exclusively in the adult human thymus and thymus-dependent areas of lymph nodes (27) and abnormally in the matrix of tumors (28) are attracting a growing interest to this ECM component. TN is considered a major ECM glycoprotein that can interfere with the activity of FN by inhibiting cell adhesion and spreading (27). TN together with thrombospondin and SPARC, form a family of ECM proteins with counter-adhesive properties that are able to provoke the loss of focal adhesions in well spread endothelial cells (29). TN inhibits both monocyte and T cell adhesion to FN (30–33). Moreover, soluble TN inhibits proliferation of human T cells and high level induction of interleukin (IL)-2R in response to anti-CD3 antibody (Ab) co-immobilized with FN (31). TN may also support the tethering and rolling of peripheral blood lymphocytes under flow conditions, with several more binding sites, compared with the rolling on E selectin (34). Laminin (LN), one of the first ECM protein detected during embryogenesis, is found predominantly in basement membranes (35, 36). LN is composed of three subunits, designated α, β, and γ, where the association of different isoforms of those subunits into heterotrimers gives rise to the seven known LNs (37). The expression of specific LN isoforms is tissue specific, and it may be altered in disease, suggesting that the interactions of leukocytes with different LN isoforms can elicit differences in leukocyte functions (38). Indeed, LN has been found to play a role in T cell migration (39, 40), and T cell differentiation (41). Cellular adhesion to ECM proteins is mediated primarily by a superfamily of heterodimeric molecules, the integrins (42). The leukocytes have been shown to interact with sequences within the type III repeats of FN primarily through two different receptors of the β1 integrin family, the α4β1 and α5β1 (43, 44) (Table [tbc]). Thus, α4β1 integrin, interacts with the CS1, which is located within the V region (44). In addition to the CS1 binding domain, activated α4β1 integrins are able to recognize the KLDAPT (45) and RGD domains on FN. For example, adhesion of B lymphocytes to FN is exclusively mediated by the interaction between α4β1 and the CS1 domain (46). However, antigen activation can induce α4β1 integrin expression on B cells to recognize the RGD domain (47). The flexibility of the α4β1 integrins in recognizing other domains in the FN molecule supports the important role this ECM protein may play in an array of lymphocyte biological functions. The other main receptor for FN is the α5β1 integrin that interacts through the RGD sequence, and it is known to be present on a subpopulation of resting CD45RA dim memory T cells (48). Although, the RGD loop is the critical recognition site for the α5β1 integrin, it is thought that in humans the PHSRN sequence located in FN-III9 is required for high affinity binding (49). Other integrins have also been identified as alternative receptors for FN, such as α4β7 (50), α3β1 (6), α8β1 (51), αvβ1 (52), αIIβ3 (53), αvβ3 (54), αvβ6 (55), and α?β8 (56) (Table). Adhesion to TN has been attributed to several receptors, such as α2β1, α9β1, and FN, although the mediators of leukocyte adhesion to this protein need to be identified (34). Recently, it has been suggested that α9β1 is critical for neutrophil migration on both vascular cell adhesion molecule-1 (VCAM-1) and TN (55). Finally, α6β1 integrin represents the principal leukocyte LN receptor on macrophages, neutrophils, and T cells (57–59). The adhesive capacity of integrins expressed on circulating leukocytes is highly regulated. For example, adhesion of T cells to FN has been shown to be regulated by the engagement of antigen-specific T-cell reactivity (TCR)/CD3 complex, as well as CD2, CD7, and CD28 (60–62). The fact that TCR engagement up-regulates adhesion of T lymphocytes to FN highlights the importance of this protein in host immune responses. Integrin-dependent biding of lymphocytes to FN is also regulated by a number of chemokines, including the C-C chemokines, such as RANTES, monocyte chemotactic protein 1 (MCP-1) and macrophage inhibitory protein 1 (MIP-1) (63–65). Modifications in ECM expression may also have a regulatory role in the integrin-mediated leukocyte adhesion. Indeed, there is good evidence that ECM protein binding can modulate integrin expression (66, 67).Table 1: Fibronectin and its integrin receptorsECM PROTEINS AND ORGAN TRANSPLANTATION There is a growing number of reports describing changes in ECM composition associated with reperfusion injury or acute and chronic rejection in human organ transplantation (68–71). Because much of mechanistic studies on the immune modulation by ECM proteins derive from experimental setting, this review emphasizes work done primarily in animal transplantation models. In our own immuno-histological and in situ hybridization studies of rat cardiac allografts, a markedly increased expression of FN and LN, mostly vascular, in the early posttransplantation period preceded cellular infiltration (72, 73). This initial up-regulation of vascular FN and LN is a common step in both allotransplantation and isotransplantation and may reflect a response to injury or ischemia occurring during the interval of about 45 min of cardiac engraftment (73). The initiation of vascular FN and LN synthesis so early may represent an important signal that triggers lymphocyte recruitment at the graft site. Indeed, administration of anti-LN Ab to T-cell deficient (B) rat recipients of cardiac allografts significantly decreased the accumulation of adoptively transferred lymphocytes selectively in the peripheral lymph nodes and in the cardiac grafts (39). At later posttransplantation periods, simultaneous detection of FN and LN in cardiac allografts by laser scanning confocal microscopic analysis, revealed a preferential accumulation of FN in the interstitial and perivascular areas where infiltrating mononuclear cells (MNC) localize. Sensitized lymphocytes adoptively transferred to test recipients also localized in FN-rich areas of both cardiac transplants and lymph nodes (72). Furthermore, treatment of rats with a neutralizing anti-TNF-α serum significantly prolonged cardiac allograft survival, downregulated local production of FN and reduced intragraft MNC infiltration (74). This supports a key role of FN as an in vivo adhesive factor for lymphocytes to home to specific tissue microenvironments, including an organ transplant (75). Our combined immunohistological and in situ hybridization analyses have demonstrated that the prime sources of FN in rejecting cardiac allografts are macrophages in the myocardium and smooth muscle and endothelial cells in the vessels (73, 74). It remains to be determined how different cell types regulate FN expression in vivo. Although transforming growth factor (TGF)-β1 is known to increase FN expression by fibroblasts, its role in modulating FN expression in other cell types remains unclear (76, 77). Our data suggests a potential role for TNF-α in modulating macrophage appearance, and potentially FN expression, during cardiac allograft rejection (74). Other data also indicate that, in coronary arteriopathy after transplantation, endothelial and smooth muscle cells produced increased FN under the regulation of IL-1β and TNF-α (78, 79). Moreover, soluble TNF-α receptor reduces the expression of FN and leukocyte infiltration within areas of intimal thickening (80). In cardiac allografts, FN exists in multiple isoforms with a distinct temporal and spatial expression patterns (73). Newly synthesized FN in cardiac grafts includes EIIIA+, EIIIB+, and CS1 variants that are generated by alternative splicing of FN premRNA (Fig. 1). Local synthesis of FN by cells in the transplanted organ is not likely the only FN present. The extravasation of plasma FN along with other plasma proteins, such as fibrinogen, seems to be a ubiquitous feature of all inflammatory sites examined (81). Extravasated plasma proteins constitute a provisional matrix that is a prominent feature of early healing wounds and cutaneous delayed hypersensitivity reactions (82, 83). An important consequence of FN alternative splicing that occurs during inflammation is the gradual replacement of plasma FN, which largely lacks the EIIIA and EIIIB segments, with locally produced FNs that contain these segments. Very little is known about the function of the EIIIA and EIIIB domains of FN, although, by comparing the patterns of EIIIA and EIIIB expression in cardiac allografts and isografts, our data support the hypothesis that these FN variants are involved in lymphocyte adhesion, migration, or differentiation in cardiac allografts. Both EIIIA and EIIIB domains are highly expressed in the myocardium of cardiac allografts associated with increased numbers of infiltrating T cells and appearance of infarcts. In contrast, the FN splicing variants are not detected in the myocardium of cardiac isografts, neither are a similar increase in T cells, nor are infarcts. Because the EIIIA segment mediates adhesion of fibroblasts and promotes the differentiation of lipocytes in the liver to become myofibroblasts (84), one may envision similar processes in lymphocyte biological functions in organ allografts. For example, CD8+ T-lymphocyte-mediated lysis of target cells requires the formation of tight conjugates between T cytotoxic cells and the targeted cell. Whereas the cellular receptors for EIIIA and EIIIB domains are not yet fully characterized, the CS1 domain recognized by the α4β1 integrin is known to mediate lymphocyte adhesion (10, 43). The CS1 subunit of FN is the first alternatively spliced domain found to be increased in cardiac allografts, it is up-regulated as early as 3 hr after transplantation, and it is preferentially expressed after day 4 (73). The presence of specific FN variants may provide a spatial address at which appropriate co-stimulatory signals could initiate or augment critical T cell signaling events. Supporting this assumption, are our recent findings in rat recipients rendered tolerant to MHC-incompatible cardiac grafts by exogenous immunosuppressive therapy. In long-term tolerant recipients, FN is up-regulated in the vessels in the absence of other vascular adhesion molecules, such as VCAM-1 and intercellular adhesion molecule-1 (ICAM-1), highlighting the putative role of FN in the recruitment of leukocytes at the graft site (85). Moreover, despite the high number of MNC infiltrating well-functioning grafts, intramyocardial infiltrating macrophages failed to express FN. This contrasted with long-term control recipients undergoing chronic rejection, in which FN expression was readily detectable in the myocardium (85). The observation that MNC do accumulate in the myocardium in the absence of cellular FN in tolerant hosts, indicates another role for this ECM protein that is unrelated to cell migration or tissue positioning. Furthermore, FN may regulate IL-2 and IFN-γ expression, which indicates an active role of this ECM protein in cellular activation (Coito et al., April, 1999, unpublished data). Our results are consistent with the concept that FN exerts synergistic effects on T cell activation by acting as a co-stimulator for both CD4+ and CD8+ T cells through TCR (23, 86) and cytokine release. For example, the density of immobilized CD3 or TCR monoclonal antibody (mAb) required to induce degranulation and tyrosine phosphorylation of cellular proteins by CD8+ T cells is about 10 times lower in the presence of FN (23). Several studies have also shown that adhesion to FN activates tyrosine phosphorylation of several T cell proteins (23, 87, 88). The production of IL-2, IFN-γ, and TNF are also stimulated in vitro by interactions between CD4+ cells and FN (89–92). Moreover, it has been reported that the binding of CD4+ cells to ECM proteins, most likely through conformational changes resulting in better presentation to their receptor, may in turn enhance cytokine activity (93–95). For instance, TGF-β, a cytokine with complex yet still ill-defined roles in organ transplantation, is secreted by most cells in an inactive form (96), and its active or functional form is highly dependent on the in vivo association to thrombospondin (93). In addition to regulating cell migration and differentiation, there is evidence that ECM may also function as a leukocyte survival factor in cardiac allografts. Apoptosis, also called programmed cell death, has been reported in organ transplants undergoing chronic rejection (97, 98). Moreover, recent studies have indicated a role of apoptosis mediated by activated MNC expressing FasL in the survival of renal allografts (99), and in the vascular pathologic lesions of chronic rejection (100). In parallel, it has also been illustrated that anchorage-dependent cells that are prevented from attaching to ECM proteins may undergo apoptosis (101, 102). Indeed, the FN-α5β1 interactions support survival of cell lines expressing α5β1 or TGF-β1 costimulated CD8 cells (103–105). Taken together, these findings support the idea that ECM proteins are active participants in the immune cascade leading to graft rejection. MNC-ECM INTERACTIONS: POTENTIAL TARGETS FOR THERAPEUTIC INTERVENTION IN TRANSPLANT RECIPIENTS The immunopathological role of MNC-ECM adhesive interactions has been documented in both autoimmune disease and transplantation animal models. Multiple-organ MNC infiltration occurs in TGF-β1 knockout mice, followed by cachexia, and death (106). In addition, cells from TGF-β1-deficient mice exhibit increased adhesion to FN matrices in vitro. Such an increased adhesion in culture has been inhibited after incubation of cells with synthetic FN peptides that interact with integrins (RGD/CS1) and/or cell surface proteoglycans (C/H-I/II/III/V). It is interesting that daily systemic injections of bioactive FN peptide preparations virtually blocked leukocyte tissue sequestration, and the development of autoimmune-like lesions, and retarded the lethal wasting syndrome characteristic for this model. The efficiency of FN peptides has been also documented in the rat model of erosive polyarthritis (107). Not only were FN peptides inhibitory to acute and chronic synovial pathology, but were also found to prevent and reverse local leukocyte recruitment, and the evolution of arthritis. Figure 2 illustrates T cell interactions in the context of FN, which provide the basis for possible therapeutic intervention. Figure 2: Model of T cell interactions in the context of FN. Cellular interactions with FN are mediated primarily by the α4β1 and α5β1 integrins. The interactions of T lymphocytes with FN may have distinct roles during allograft rejection: (1) T cell interactions with FN expressed by endothelial cells may facilitate the up-regulation of other vascular adhesion molecules, such as VCAM-1, and the recruitment of leukocytes at the graft site; (2) T cell interactions with intra-myocardial FN expressed mainly through macrophages may act as a co-stimulatory signal on T cell activation and cytokine release. Arrows indicate FN stained by immunofluorescence (white) in a rejecting rat cardiac allograft.Our own data are consistent with the model in which in vivo interactions between the α4β1 integrin receptor and the cell-associated CS1 motif of FN are critical in the acute allograft rejection cascade (108)). Indeed, treatment of rat recipients of cardiac allografts with a 7-day course of bioactive CS1 peptides accomplished the following: (1) abrogation of acute rejection, and doubled the transplant survival time, (2) diminishment of vascular expression of adhesion molecules such as VCAM-1 and ICAM-1, (3) reduction of intragraft infiltration by CD4+ and CD8+ cells, and (4) decrease of allo-Ag activation at the graft site, as evidenced by decreased infiltration by CD25+ cells, and diminished expression of T helper cell 1 and T helper cell 2 type cytokines. These immunosuppressive effects could be reversed and acute rejection recreated after adjunctive treatment with recombinant IL-2, suggesting that CS1 peptides may induce a temporary state of cytokine-responsive T cell anergy, in vivo. Most of the adhesion functions mediated by α4β1 integrin are attributed to interactions with its two known ligands, the CS1 motif in the FN molecule, and VCAM-1 expressed on endothelial cells (Table and Figure 2). Although it is possible that CS1 peptide may interfere with α4β1 - VCAM-1 interactions, α4β1 has been shown to have distinct ligand/binding sites for VCAM-1 and FN (109), suggesting that the CS1 peptide that binds α4β1 integrin does not likely inhibit the binding by VCAM-1. Moreover, although VCAM-1 and CS1-FN may share spatially overlapping biding sites on α4β1, the concentration of FN peptides that interfere with the α4β1 - VCAM-1 binding are several times higher than those required for the α4β1 - CS1 blockade (110). Hence, it seems quite unlikely that the effects of relatively low dose CS1 peptide regimen on the kinetics of, and local cell recruitment associated with, acute graft rejection, could be attributed to the blockade of α4β1 - VCAM-1 interactions. Supporting this notion are findings from mouse cardiac transplant models of persistent T cell infiltration in allografts, despite anti-VCAM-1 mAb therapy, and transient T cell infiltrate in isografts, which are devoid of VCAM-1 expression (111). These strongly suggest a role for VCAM-1-independent collateral system(s) that mediate leukocyte sequestration during the course of acute graft-deteriorating rejection. It has been shown that the blockade of α4β1-dependent leukocyte interactions by treatment with anti-α4β1 mAb (TA-2) acted synergistically with cyclosporine to prolong the survival of rat small bowel allografts (112). Moreover, mAb-facilitated combined blockade of α4β1 and VCAM-1 prolonged murine cardiac allograft survival and induced tolerance in some cases (113). In pancreatic islets, the blockade of α4β1 with TA-2 resulted only in a slight prolongation of the graft survival time (114). However, in a more recent study, administration of CS1 peptides to mice recipients of islet allografts not only abrogated graft rejection, but also prevented graft infiltration, suggesting a role for α4β1 - FN interactions in leukocyte homing to the graft site (115). In contrast to the blockade of α4β1-FN interactions by CS1 peptides, the protective effects afforded by targeting the α4β1 integrin by mAb alone were rather short-lived and insignificant on allo-Ag-driven mononuclear and endothelial cell activation, leading to graft rejection in transplant recipients. We have also tested the efficacy of CS1 peptides in sensitized rat recipients in which therapy with rapamycin (RPM) abrogates accelerated rejection at 24 hr but does not prevent the ultimate cardiac allograft loss at 40–50 days (116). These long-term grafts show progressive development of arteriosclerosis, a hallmark of chronic rejection (117). Indeed, an adjunctive course of CS1 peptides in the early posttransplantation phase resulted in the disappearance of the inflammatory and smooth muscle cells from the arterial intima of long-term recipients. Unlike medium or large arteries in rats undergoing monotherapy, those following adjunctive CS1 peptides had normal internal elastic lamina, and were free of neointimal thickening, confirming that CS1 peptides successfully prevented the development of arterial chronic injury. Moreover, treatment with CS1 peptides reduced gene transcript and product levels for T cell- and macrophage-derived cytokines and chemoattractants, which are known to contribute to the development of progressive chronic-type allograft failure. A similar finding was observed in a cholesterol-fed rabbit model in which treatment with CS1 peptides specifically blocked vascular changes, and markedly diminished accelerated coronary arteriopathy (118). Therapy with FN peptides effectively depressed intragraft transendothelial infiltration by macrophages and T cells, documenting the functional role of FN in the trafficking of inflammatory cells in progression of cardiac arteriopathy. However, unlike in our rat study, CS1 peptides did not affect the grade of cardiac rejection in the rabbit model, as judged by the extensive myocardial cell infiltration, myocyte necrosis, hemorrhage, and fibrosis. As integrin-FN interactions are critical for T cell activation, adhesion, and local retention to perpetuate chronic inflammatory responses, antagonism of cellular activation, and recruitment by FN peptides provides an important mechanism for modulating the multi-step adhesion process, and attenuating aberrant inflammatory responses (119). Given the efficacy of FN peptides to block neointimal thickening in coronary vessels, this novel therapy may interfere with the migration of smooth muscle cells from the media into the intima, with resultant reduction of neointimal hyperplasia, consistent with the expression of α4β1 integrin by smooth muscle cells (118). The CS1-mediated prevention of chronic rejection may result from a far more reaching effects on long-term in vivo interactions between MNC, endothelial vascular lining, and the ECM scaffolding. Treatment with RPM abrogates rejection in presensitized hosts by depressing humoral and T cell cytotoxic and proliferative responses (120, 121). At this point, host-vs.-graft interactions are limited to the cellular contact with ECM and the endothelial lining. Adjunctive administration of CS1 peptides during this early posttransplantation phase may impair the engagement of T cells with FN as part of the transplantation structural framework. At the same time, CS1 peptides may modify intragraft vasculature by reducing its adhesiveness for host leukocytes. During the ensuing maintenance phase, the alterations in intragraft MNC homing and positioning are mimicked by the prevailing macrophage response. The perpetual cycle of late macrophage-associated pro-inflammatory mediators, smooth muscle proliferation, neointimal formation, and subsequent vasculature occlusion fails to be initiated. Other findings supporting the ability of cell populations to discriminate and choose microenvironments to home, as originally defined by “ecotaxis” (122), was also observed in mice cardiac recipients infected with Trypanosoma cruzi. In this model, syngeneic mouse hearts of chronic chagasic recipients are rejected through a CD4+ T cell-dependent mechanism (123), although treatment with an antibody against the LN receptor abrogates the ability of autoreactive CD4+ cells to migrate into the graft and then trigger rejection (124). Moreover, the blockade of CD44 decreased homing of activated lymphocytes and prolonged allograft survival in models of skin (125) and kidney (126) transplantation. CD44 is a cell surface receptor that binds to ECM components such as hyaluronate (HA) and osteopontin (OPN). HA, a major carbohydrate component of the extracellular matrix, can mediate activated T cell extravasation into sites of inflammation by promoting the primary adhesion (rolling) to vascular endothelium under conditions of physiologic shear stress (127, 128). However, OPN, a secreted acidic glycoprotein, is a potent monocyte chemoattractant (129). Indeed, highly up-regulated OPN levels may be readily detected in cardiac allografts in parallel with increased intra-graft macrophage infiltration (Coito, June, 1995, unpublished results). Collectively, these data advance the hypothesis that local synthesis of ECM components is an ongoing feature of, and adhesive ECM-MNC associations are critical for the development of, acute and chronic rejection in transplant recipients. Further elucidation of ECM-MNC interactions may ultimately offer potential novel sites for intervention in the control of transplant rejection, and may lead to the development of refined strategies based upon new concepts of host immunomodulation.