Monocyte Adhesion to Xenogeneic Endothelium during Laminar Flow Is Dependent on -Gal-Mediated Monocyte Activation1
http://www.100md.com
免疫学杂志 2005年第12期
Abstract
Monocytes are the predominant inflammatory cell recruited to xenografts and participate in delayed xenograft rejection. In contrast to allogeneic leukocytes that require up-regulation of endothelial adhesion molecules to adhere and emigrate into effector tissues, we demonstrate that human monocytes adhere rapidly to unstimulated xenogeneic endothelial cells. The major xenoantigen galactose(1,3)galactose(1,4)GlcNAc-R (-gal) is abundantly expressed on xenogeneic endothelium. We have identified a putative receptor for -gal on human monocytes that is a member of the C-type family of lectin receptors. Monocyte arrest under physiological flow conditions is regulated by -gal, because cleavage or blockade results in a dramatic reduction in monocyte adhesion. Recruitment of human monocytes to unactivated xenogeneic endothelial cells requires both 4 and 2 integrins on the monocyte; binding of -gal to monocytes results in rapid activation of 2, but not 4, integrins. Thus, activation of monocyte 2 integrins by -gal expressed on xenogeneic endothelium provides a mechanism that may explain the dramatic accumulation of monocytes in vivo.
Introduction
Xenotransplantation, when ultimately successful, will alleviate the current severe organ shortage. At present, the rejection of xenografts by both the innate and adaptive immune systems is vigorous and poorly controlled by tolerable immunosuppressive protocols. Delayed xenograft rejection (DXR)3 is an intermediate form of graft rejection characterized by overwhelming monocyte and NK cell infiltration, Ab deposition along the endothelium, endothelial cell activation, thrombosis, and eventually graft loss (1, 2). Many components of the innate and acquired immune systems, including xenoreactive Abs, complement, coagulation factors, monocytes, and NK cells, can act independently or in concert to effect xenograft damage, but the mechanisms that initiate DXR are poorly understood (3, 4, 5, 6, 7, 8). Since DXR represents the most important immunological hurdle preventing routine use of xenografts for the treatment of end-organ failure, additional studies are necessary to elucidate the mechanisms underlying DXR.
Monocyte accumulation within discordant xenografts is striking. Monocyte infiltration occurs early after xenografting in both rodent- and pig-to-primate combinations, although the kinetics and severity of monocyte accumulation are more variable in pig-to-primate than in rodent models (2, 4, 5, 9, 10, 11, 12). Several lines of evidence support the role of monocytes and NK cells as effectors of DXR: depletion of monocytes or NK cells in xenograft recipients delays the onset of DXR, adoptive transfer of activated monocytes from rejected xenografts accelerates rejection in newly transplanted animals, and NK cells directly recognize and lyse xenogeneic endothelial cells through Ab-independent cytotoxic pathways (13, 14, 15, 16, 17). Furthermore, monocytes recruited into the xenograft elaborate multiple proinflammatory cytokines and chemokines, such as TNF-, IFN-, and MCP-1, that amplify the inflammatory response (4, 9, 12, 15). Direct monocyte or NK cell contact also activates xenogeneic endothelial cells to up-regulate adhesion molecule, cytokine, and tissue factor expression (13, 18, 19, 20). Recent work from our laboratory has elucidated some of the mechanisms underlying monocyte-induced endothelial cell activation (21). Few studies have addressed the trafficking of monocytes to xenogeneic endothelium. Leukocyte emigration into inflamed tissues involves intimate contact with vascular endothelium and multiple adhesive and signaling events and follows a multistep paradigm of rolling, arrest, firm adhesion, and diapedesis before migration into inflamed tissues (22). Human leukocytes are also capable of adhering to both quiescent and cytokine-activated xenogeneic endothelium (21). Robinson et al. (23) demonstrated that human neutrophil adhesion to TNF-stimulated porcine endothelium is orchestrated through known adhesion molecule receptor-ligand pairing. Study of human monocyte adhesion to porcine endothelium indicates that VCAM-CD49d (4 integrin) interaction supports monocyte adhesion to TNF-stimulated porcine endothelium, albeit less efficiently than monocyte adhesion to HUVEC (24). However, these authors could not explain the increased frequency of human monocyte adhesion to unstimulated porcine endothelium compared with human endothelium.
Previous studies suggest that the major xenoantigen galactose(1,3)galactose(1,4)GlcNAc-R (-gal), a terminal carbohydrate expressed on xenogeneic endothelium, participates in monocyte adhesion. For instance, transfection of HUVEC with an 1,2-fucosyltransferase reduced surface expression of -gal and decreased monocyte adhesion by 50% (25). These authors did not definitively elucidate whether -gal functions as an adhesion molecule, modulates the affinity of existing adhesion molecules, or leads to cellular activation by acting as a ligand for some undefined signaling receptor.
Our objective was to directly compare adhesion of human monocytes to unstimulated porcine (PAEC) and human (HAEC) aortic endothelial cells under conditions of physiological flow. Specifically, we examined the contributions of the endothelial carbohydrate -gal in regulating monocyte adhesion to xenogeneic endothelial cells. As expected, naive allogeneic human monocytes do not adhere to quiescent unstimulated human endothelial cells, yet human monocytes rapidly adhere to unactivated porcine endothelial cells under conditions of laminar flow. We found that monocytes possess a putative receptor for -gal that activates 2 integrin binding and promotes the arrest of monocytes on porcine endothelial cells.
Materials and Methods
VCAM-1/Fc, ICAM-1/Fc, ICAM-2/Fc, and -gal-BSA were immobilized on plastic according to a protocol described previously with the following modifications (28). Goat anti-human IgG (Fc-specific) F(ab')2 was passively adsorbed onto the center of a 35-mm polystyrene tissue culture dish (Corning) by incubating a 10-μl drop for 60 min in a humidified atmosphere (100 μg/ml; 24°C). Dishes were washed with PBS, followed by passive adsorption of -gal-BSA or BSA alone at the specified concentration (60 min; 24°C). Excess -gal-BSA was removed by washing with PBS, and nonspecific binding sites were blocked with 5% FBS (60 min; 24°C). The anti-Fc-coated area was then incubated with a saturating concentration of VCAM-1/Fc, ICAM-1/Fc, or ICAM-2/Fc (10 ml-drop; total concentration, 20 μg/ml; 60 min; 24°C), which corresponds to a coating density of 1200 molecules/μm2. In some experiments the density of immobilized VCAM-1/Fc was modulated by adjusting the relative molarity with nonimmune human IgG. For plates containing -gal-BSA or BSA alone, -gal-BSA or BSA was immobilized in the center of plastic dishes at 1, 10, 50, or 100 μg/ml (60 min; 24°C), followed by washing and blockade of nonspecific binding sites with 5% FBS. The efficacy of -gal-BSA adsorption was confirmed by ELISA; triplicate wells of a 96-well plate coated with -gal-BSA were labeled with BS-IB4-biotin (60 min; 24°C), followed by streptavidin-HRP (60 min; 24°C) and 3,3',5,5'-tetramethylbenzidine (Rockland) for 15 min, and the OD was read on an ELISA plate reader at 450 nm.
Parallel plate flow chamber assays
All adhesion assays were performed using a parallel plate flow chamber (Glycotech) at a shear stress of 0.8 or 1.0 dyne/cm2 unless otherwise specified. Confluent monolayers of HAEC and PAEC, or dishes coated with VCAM-1, ICAM-1, ICAM-2, or -gal-BSA (or a combination) were mounted in the flow chamber and placed on the heated stage (37°C) of a Diaphot 300 inverted phase-contrast microscope (Nikon). Experiments were videotaped with a Sony DXC-151A color video camera and Sony SVT-S3100 time-lapse video cassette recorder for later offline analysis. Purified monocytes (0.5 x 106) were resuspended in 500 μl of HBSS (with 1 mM Ca2+/Mg2+) supplemented with 10 mM HEPES and 0.5% FBS (assay buffer) and injected into the flow chamber. Adherent monocytes were counted in six consecutive high power fields after 4 min of flow. Monocytes were defined as firmly adherent if they remained stationary for >4 s. It was unusual for adherent monocytes to subsequently release.
The strength of monocyte adhesion to -gal coimmobilized with an integrin ligand was tested in a cell detachment assay by injecting 300 μl (5 x 105 monocytes/ml) into an inverted flow chamber of plastic dishes coated with BSA or -gal-BSA plus ICAM-1, ICAM-2, or VCAM-1. Monocytes were allowed to settle for 5 min, after which increasing levels of shear stress were applied in 30-s increments by pulling the assay buffer across the flow chamber with a programmable syringe pump (Genie; Kent Scientific). Assay buffer was maintained at 37°C in a heated water bath, and the stage of the microscope was heated with an infrared heat lamp coupled to a temperature controller. Adherent cells were counted at each level of shear stress.
Flow cytometric analysis
The capacity for human monocytes to bind directly to -gal was determined by incubating PBMCs with increasing concentrations of either soluble -gal-FITC (0070-PAA-FP) or control glycoconjugate-FITC (0000-PAA-FP) for 30 min at 4°C. Before incubation with glycoconjugates, nonspecific binding sites were blocked with 500 μg/ml 0000-PAA for 20 min at 4°C. In some experiments, PBMCs were also pretreated with either 20 μg/ml isotype or blocking mAbs to monocyte 4, 1, and 2 integrins and L-selectin (10 μg/ml Dreg 200). Specific binding of soluble -gal to monocytes was determined by coincubating PBMCs with the lectin BS-IB4, which specifically recognizes -gal residues. The correct gating for monocytes was previously determined with specific Abs for CD14. The mean fluorescence intensity (MFI) was measured for each population.
Exposure of the mAb24 epitope is a reporter of the 2 integrin high affinity state. Monocytes were incubated with soluble -gal-BSA or BSA alone for 1–30 min at 37°C. At the end of each incubation period, ice-cold PBS was added, and the monocytes were washed once, followed by addition of 10 μg/ml mAb24 or an isotype control for 20 min at 4°C. After washing, monocytes were labeled with a secondary PE-conjugated F(ab')2 for 20 min at 4°C, washed, and analyzed by FACS.
Statistical analysis
Statistical analysis was performed with SAS software (SAS version 8.1; SAS Institute). Continuous variables are expressed as the mean ± SEM. Comparisons of continuous variables between groups were performed with unpaired Student’s t tests or ANOVA. Orthogonal contrasts between control and treatment groups were made to determine significance where appropriate. A two-tailed value of p < 0.05 was considered statistically significant.
Results
Human monocytes adhere to unstimulated porcine but not human endothelium
Adhesion of human monocytes is much greater to PAEC than to HAEC under physiological flow conditions. Unstimulated human monocytes bind more avidly to unstimulated PAEC than to HAEC (33.7 ± 5.6 vs 2.5 ± 0.6 monocytes/field, respectively; p = 0.009; Fig. 1A). Monocyte accumulation begins immediately after injection into the flow chamber and continues in a linear fashion for the 4-min period of observation. Most monocytes arrested from fast-rolling, rather than slow-rolling, monocytes (see supplemental online video).4 Monocytes continued to accumulate on PAEC for 30 min, and very little adhesion was observed for HAEC even after 30 min (data not shown). In contrast, monocyte adhesion to TNF-stimulated endothelial cells did not differ between species (69.4 ± 5.0 vs 65.5 ± 6.4 for PAEC vs HAEC, respectively; p = NS; Fig. 1A). Adhesion of human monocytes to PAEC is dependent on the level of shear stress. We perfused unstimulated human monocytes over PAEC at four different levels of shear stress for 4 min: 0.5, 1.0, 2.0, and 5.0 dyne/cm2. Firmly adherent monocytes per field were counted at the end of 4 min and averaged over six fields. Monocyte adhesion was significantly lower for increasing levels of shear, particularly at shear levels of 2.0 and 5.0 vs 0.5 dyne/cm2 (p < 0.05; Fig. 1B).
To ensure that increased monocyte adhesion to xenogeneic endothelial cells was not due to basal differences in endothelial adhesion molecule expression, we examined the basal expression of various adhesion molecules on unstimulated endothelial cells. As expected, quiescent human endothelial cells expressed ICAM-1 (not shown). Attempts to determine resting ICAM-1 levels on porcine endothelial cells were unsuccessful due to lack of Abs reacting or cross-reacting with porcine cells. Otherwise, human and porcine endothelial cells did not express significant levels of VCAM-1, E-selectin, or P-selectin, yet significantly increased the expression of VCAM-1, E-selectin, and ICAM-1 (on HAEC) after TNF- stimulation (Fig. 2). In contrast, we detected increased basal levels of L-selectin ligand on unactivated PAEC compared with HAEC (Fig. 2).
Monocyte 4 and 2 integrins, and not L-selectin, are necessary for adhesion to xenogeneic endothelium
Selectin binding to an appropriate ligand initiates tethering and rolling of monocytes on endothelial cells; rolling contributes to firm arrest by allowing sufficient time for both integrin activation and bond formation between leukocyte integrins and their counter-receptors (29, 30). Because quiescent PAEC expressed higher levels of L-selectin ligand compared with HAEC, we examined the contribution of L-selectin on monocyte adhesion to unstimulated PAEC. We also focused on monocyte integrins as potent mediators of monocyte firm adhesion. As a first step, we evaluated the requirements for extracellular calcium on monocyte adhesion to porcine endothelial cells, because both monocyte L-selectin and monocyte integrins contain binding sites for divalent cations that are necessary for their adhesive function. The absence of extracellular calcium from the perfusion buffer completely inhibited monocyte arrest on PAEC under laminar flow compared with control calcium-containing buffer (1.7 monocytes/field expressed as percentage of control arrest events; Fig. 3A; p = 0.04). To further characterize the role of monocyte integrins and L-selectin on adhesion, we blocked specific integrins and L-selectin with function-blocking mAbs. Blockade of 4 and 2 integrins resulted in a significant reduction in firm adhesion under laminar flow compared with an isotype-matched control Ab (6 ± 2.7 and 12 ± 2.8 vs 21 ± 4.4 monocytes/field for anti-4, anti-2, and isotype, respectively; p = 0.01; Fig. 3B). A combination of blocking Abs against 4 and 2 completely inhibited adhesion (0.64 ± 0.16 vs 21 ± 4.4 monocytes/field; p < 0.001; Fig. 3B), whereas blockade of monocyte L-selectin did not inhibit adhesion compared with isotype control (23 ± 5.5 vs 21 ± 4.4 monocytes/field; p = NS; Fig. 3B).
Human monocyte adhesion to xenogeneic endothelium is dependent on the major xenoantigen -gal
Porcine endothelial cells express high levels of -gal on their surface, which may function as a xenospecific adhesion molecule. Because basal adhesion molecule levels (except L-selectin ligand) were similar between human and porcine endothelial cells, yet L-selectin does not mediate increased arrest of monocytes on PAEC, we addressed the contribution of endothelial -gal to monocyte arrest on porcine endothelium by specifically cleaving terminal galactose residues with an exogalactosidase. Binding of the -gal-specific BS-IB4 was used to monitor the efficacy of -gal cleavage. Galactosidase treatment of porcine monolayers effectively cleaved the terminal Gal(1,3)Gal residues from the endothelium, as evidenced by a marked reduction of BS-IB4-FITC fluorescence compared with control monolayers (MFI, 120 vs 35 for control vs -galactosidase, respectively; Fig. 4A, right axis). Under these conditions, adhesion of monocytes to porcine endothelial cells was reduced by 54% vs control, from 35.8 ± 4.4 to 16.5 ± 4.0 monocytes/field (p = 0.035; Fig. 4A). In contrast, treatment of PAEC with a control galactosidase, which does not cleave -gal residues, resulted in a 26% reduction in monocyte adhesion vs control, from 35.8 ± 4.4 to 26.6 ± 8.0 monocytes/field, respectively (p = NS). We also blocked terminal -gal residues to confirm its function in mediating monocyte arrest on PAEC. Pretreatment of PAEC with BS-IB4 resulted in a 57% reduction in monocyte arrest vs control monolayers (10 ± 3.3 vs 24 ± 3.3 monocytes/field; p = 0.03; Fig. 4B). Treatment of monolayers with a control lectin that does not bind -gal, Ulex europeaus agglutinin 1 (UEA-1), had no effect. In contrast, up-regulation of classical adhesion molecules with TNF- before blockade with BS-IB4 minimized the effect of -gal on monocyte arrest. Monocyte arrest after BS-IB4 blockade on TNF--pretreated monolayers was reduced from 67 ± 11 and 72 ± 17 to 55 ± 13 monocytes/field for control and UEA-1 vs BS-IB4, respectively, but this reduction was not statistically significant (Fig. 4B; p = NS).
-Gal on xenogeneic endothelium is a putative ligand for human monocytes
Since xenogeneic endothelial cell -gal clearly contributes to monocyte arrest under conditions of laminar flow, we hypothesized that human monocytes could directly bind to -gal, possibly through a direct -gal-integrin interaction. We first determined whether monocytes could bind to -gal by incubating PBMCs with a soluble form of -gal conjugated to FITC. At a saturating concentration, all the monocytes bound to -gal (MFI, 180.1 vs 9.7, for -gal vs control; Fig. 5A). In contrast, most lymphocytes did not bind to -gal (MFI, 12.3 vs 1.3, for -gal vs control), with the exception of a small population that could have represented NK cells. The binding of -gal to human monocytes was dose dependent and saturatable (Fig. 5B). Saturation of the putative -gal receptor on monocytes was achieved at a dose slightly >500 μg/ml -gal. We also demonstrated -gal binding to human monocytes using -gal-BSA and directly biotinylated -gal (data not shown).
As additional evidence that the binding of -gal to monocytes was specific, we inhibited binding with the lectin BS-IB4. Incubation of monocytes with soluble -gal in the presence of BS-IB4 reduced monocyte binding of -gal by 90% of control-FITC levels (MFI, 77 ± 21, 18 ± 5.8, and 11 ± 1, for -gal-FITC, -gal-FITC plus BS-IB4, and control-FITC, respectively; p < 0.005 for -gal-FITC vs -gal-FITC plus BS-IB4; Fig. 5C). Thus, the putative -gal receptor on human monocytes demonstrates specificity and saturability for its ligand, characteristics consistent with classical receptor-ligand binding.
If monocyte integrins directly bound to -gal, then presumably this binding would also require extracellular calcium. Incubation of soluble -gal-FITC with monocytes in the absence of calcium led to a dramatic reduction of monocyte-bound -gal (MFI, 77 ± 21 vs 17 ± 6.2, for -gal-FITC vs -gal-FITC plus no calcium; p < 0.005; Fig. 5C). To definitively address the possibility that monocyte adhesion molecules might be receptors for -gal, we blocked monocyte 1, 2, and 4 integrins as well as L-selectin with function-blocking Abs. As demonstrated in Fig. 5D, monocytes bound to saturating levels of soluble -gal equally well in all groups, indicating that monocytes do not bind directly to -gal through these adhesion molecules (p = NS for no Ab and isotype control vs specific integrin- and L-selectin-blocking Abs). Because L-selectin is a C-type lectin, we tested two separate cell lines stably transfected with human L-selectin, K562 wild-type and U937, for their ability to bind soluble -gal. K562 parent cells do not express endogenous L-selectin and failed to bind either control or soluble -gal (Fig. 6A). In addition, the expression of L-selectin on the cell surface did not confer -gal binding to K562 wild-type or U937 cells, confirming that L-selectin is not the putative -gal lectin (Fig. 6). Taken together, these data suggest that monocytes possess an alternate receptor specific for -gal; the requirement of extracellular calcium for ligand binding indicates that the putative receptor may be a member of the C-type lectin family of carbohydrate receptors.
Immobilized -gal does not directly mediate monocyte rolling or firm arrest
Since human monocytes bind directly to -gal, we hypothesized that -gal may function as an adhesion molecule for monocytes. -gal-BSA or BSA alone was adsorbed onto tissue culture dishes, with saturation of all binding sites achieved at a dose of 100 μg/ml; the efficacy of -gal binding to plastic tissue culture plates was measured by ELISA (Fig. 7A). Monocytes did not roll (data not shown) or arrest on plates coated with a saturating amount of -gal (100 μg/ml) or IgG (20 μg/ml), whereas significant numbers of monocytes arrested on plates coated with a saturating level of VCAM-1 (1200 molecules/μm2; Fig. 7B; p < 0.0001 for -gal and IgG vs VCAM-1). We also modified the coating by first adsorbing an anti-Fc F(ab')2, followed by an anti-BSA Ab, and, lastly, -gal-BSA or BSA alone in an attempt to orient -gal to flowing monocytes. Under these conditions, the number of monocytes that firmly arrested to increasing doses of immobilized -gal or control BSA was similar after 4 min of laminar flow, suggesting that -gal does not function directly as an adhesion molecule (Fig. 7C; p = NS).
Human monocyte adhesion to VCAM-1 is not augmented by -gal
Monocyte adhesion to porcine endothelial cells is partly dependent on 4 integrin; therefore, we also examined whether -gal could modulate the affinity for ligands of 4 such as VCAM-1. As expected, monocyte adhesion was dependent on the coating density of VCAM-1, but coimmobilized -gal did not result in greater monocyte arrest than VCAM-1 alone (Fig. 8A; p = NS). Similarly, -gal coimmobilized with VCAM-1 did not result in greater strength of adhesion compared with VCAM-1 and BSA, because the number of adherent monocytes remaining after graded increases in shear stress were similar between conditions (Fig. 8B; p = NS). We also used flow cytometry to detect rapid and transient increases in 4 integrin affinity in response to soluble ligand binding, as previously described (31). Using this method, known chemoattractants, such as fMLP, induce high affinity 4 integrin ligand binding, but soluble -gal at 10 and 100 μg/ml failed to do so (data not shown).
-Gal induces high affinity 2 integrin ligand binding on human monocytes and subsequent arrest on ICAM-1 and ICAM-2
Although -gal does not directly mediate adhesion of monocytes to PAEC, we hypothesized that -gal may increase the affinity or avidity of monocyte 2 integrin for its ligand, ICAM-1. Indeed, monocyte adhesion after graded increases in shear stress was greater on -gal-BSA coimmobilized with ICAM-1 than to control BSA and ICAM-1 (Fig. 9A), suggesting that -gal binding to monocytes resulted in either 2 integrin up-regulation or increased affinity of 2 for ICAM-1. 2 integrins on monocytes include L2 (LFA-1) and M2 (Mac-1), and activation of either LFA-1 or Mac-1 could mediate increased strength of adhesion to ICAM-1. In contrast, only LFA-1 binds to ICAM-2 (32); therefore, we tested the strength of monocyte adhesion to -gal coimmobilized with ICAM-2 to determine whether LFA-1 mediated increased monocyte adhesion on ICAM-1. Monocytes adherent to -gal-BSA plus ICAM-2 resisted detachment at shear stresses of 4 and 10 dyne/cm2, more so than ICAM-2 plus BSA, suggesting that -gal activated LFA-1 (Fig. 9B).
To address the possibility that -gal binding to monocytes results in increased affinity of 2 integrin for ICAM-1 and ICAM-2, we coincubated monocytes with -gal-BSA or BSA alone and measured the exposure of the mAb24 epitope, a marker of the high affinity state for 2 integrins. Soluble -gal binding to monocytes resulted in activation of 2 integrins to a higher affinity state after only 1 min of coincubation and peaked after 5 min, compared with the BSA control (Fig. 9C). Incubation of human monocytes with a different -gal glycoconjugate, linked to PAA, also resulted in up-regulation of 2 integrins to a higher affinity state, similar to that observed after treatment with fMLP (data not shown).
Discussion
In the present study we demonstrate the following observations: monocytes arrest and firmly adhere on an order of magnitude greater (13-fold increase) to unstimulated xenogeneic than to allogeneic endothelial cells under physiologic flow conditions; monocytes possess a putative C-type lectin for -gal, the major xenoantigen, and -gal induces high affinity ligand binding of monocyte 2 integrins leading to firm adhesion to porcine endothelial cells. These findings propose a xenospecific adhesion mechanism that may explain the dramatic accumulation of monocytes observed early after xenografting.
The firm arrest of human monocytes on porcine aortic endothelial cells after 4 min of laminar flow was dramatic. This observation confirms a recent finding by our group that demonstrated that human monocytes adhered to unstimulated porcine endothelial cells as early as 15 s after flow was initiated. In both studies, predominantly fast-rolling, rather than slow-rolling, monocytes arrested on porcine endothelial monolayers, whereas very few monocytes rolled or arrested on human endothelial monolayers. In contrast, the number of monocytes that firmly arrested on TNF--activated PAEC or HAEC was not different, suggesting that human monocytes adhere to xeno- or alloendothelium through similar adhesion molecule-ligand pairings when endothelial adhesion molecule and chemokine expression is maximal.
Since very few monocytes adhered to unstimulated human endothelial cells, we hypothesized that the substantially greater binding of monocytes to porcine endothelial cells under flow conditions was due to differential basal adhesion molecule expression, terminal xenospecific endothelial glycosylation, monocyte activation, or a combination of these factors. To rule out basal differences in adhesion molecule expression as the reason for increased monocyte adhesion to PAEC, we compared adhesion molecule expression on unstimulated endothelial cells. Flow cytometric analysis of HAEC and PAEC confirmed an equally low basal level of adhesion molecule expression, with the exception of ICAM-1 and L-selectin ligands. Although ICAM-1 is normally constitutively expressed on quiescent vascular endothelium, we could only detect its presence on unactivated HAEC (33). We were unable to obtain Abs that cross-react between human and porcine ICAM; therefore, we cannot rule out the possibility that basal expression of ICAM-1 was higher on PAEC. One notable difference between HAEC and PAEC was increased basal expression of L-selectin ligand on PAEC. Giuffre et al. (34) also found that unactivated bovine aortic endothelial cells express high levels of L-selectin ligand, which did not increase after TNF- stimulation.
L-selectin binding to its ligand primarily mediates monocyte tethering and rolling along the endothelium (35, 36). Because unactivated PAEC, and not HAEC, express L-selectin ligands, we tested the contribution of L-selectin on the greater adhesion of monocytes to PAEC. The binding domain of L-selectin is contained within a terminal calcium-dependent (C-type) lectin (37); therefore, the complete inhibition of monocyte adhesion that we observed in calcium-free conditions suggested that L-selectin could potentially mediate monocyte tethering to PAEC (Fig. 3A). Surprisingly, blocking mAbs against L-selectin had no effect on either monocyte rolling or firm adhesion under flow (Fig. 3B), suggesting that monocytes tethered to PAEC via an alternate adhesion molecule. This is consistent with our observation that monocyte arrest was not preceded by slow rolling, which is usually characteristic of L-selectin-mediated events (see supplemental online video).
The presence of divalent cations, such as calcium and magnesium, is also critical for adhesive function of monocyte integrins; therefore, we also inhibited 4 and 2 integrins with blocking Abs. These integrins mediate monocyte firm arrest, and 4 may also support the capture and rolling of monocytes (38). Indeed, both 4 and 2 integrin were required for monocyte firm adhesion to PAEC, and 4 probably also contributed to monocyte capture, because L-selectin inhibition had no effect on the number of firmly adherent monocytes (Fig. 3B). The virtual absence of VCAM-1 on unstimulated PAEC suggests an alternate ligand for 4.
There are several mechanisms by which -gal might increase the adhesiveness of xenogeneic endothelium to human monocytes. -gal is a terminal pentasaccharide expressed on the endothelium of most mammals and New World monkeys, but it is absent from the endothelium of humans and Old World monkeys (39). Our earlier studies demonstrated that human monocyte adhesion was greater to PAEC than to HAEC, and cleavage of -gal from the surface of PAEC led to reduced monocyte adhesion under static conditions.5 Other strategies aimed at minimizing -gal expression also lowered monocyte adhesion to PAEC. For example, overexpression of an (1,2)-fucosyltransferase in porcine endothelial cells led to reduced -gal expression and subsequently lower monocyte adhesion and activation (25). However, these results must be interpreted with caution, because overexpression of a fucosyltransferase reduces terminal sialylation as well as -gal expression.
To precisely determine the contribution of -gal to adhesion, we studied monocyte adhesion to xenogeneic endothelial cells under laminar flow. Under low shear flow conditions, -gal mediated the arrest of human monocytes to unstimulated PAEC, because either cleavage or blockade of -gal led to a dramatic reduction in adhesion (Fig. 4). However, -gal immobilized to tissue culture dishes failed to stimulate monocyte rolling or firm arrest, even if -gal was presented efficiently above the plate using Abs (Fig. 7, B and C). The Gal(1, 3)Gal(1,4)GlcNAc-R structure was sufficient for soluble -gal binding to monocytes; however, additional glycoprotein domains may be necessary for -gal to function as an adhesion molecule in vitro or in vivo, as is the case for the dual chemokine and adhesion molecule, fractalkine. The soluble chemokine domain of fractalkine binds to its receptor, CXC3R1; however, the chemokine domain immobilized to plates does not mediate leukocyte adhesion unless it is presented on the mucin domain (40). Therefore, it is possible that -gal could function as an adhesion molecule if it were presented in a more physiological manner.
Monocytes possess a repertoire of cell surface adhesion molecules and receptors that orchestrate the rolling, activation, and firm arrest on vascular endothelial cells (22). In addition to classical adhesion molecules, monocytes express lectins that bind to specific carbohydrate sequences on bacteria and glycosylated Ags (41, 42). Preformed xenoreactive Abs target the major xenoantigen, -gal, because of a common terminal carbohydrate sequence with bacterial -gal, and similar cross-reactivity may occur between monocyte lectins and endothelial -gal (39). Indeed, we found that unstimulated human monocytes specifically bound to soluble -gal. Most lymphocytes did not bind to -gal, and the -gal binding lectin (BS-IB4) completely inhibited -gal binding to monocytes. Extracellular calcium was also required for -gal binding to monocytes, suggesting that the putative receptor is a C-type lectin. Furthermore, the failure of two separate leukocyte cell lines that were stably transfected with human L-selectin to bind soluble -gal provides proof that the putative -gal-binding, C-type lectin is not L-selectin (Fig. 6).
Since immobilized -gal did not directly mediate monocyte arrest on coated plates, we hypothesized that -gal binding to its putative receptor could modulate the affinity of 4 and 2 integrins for their ligands. Chemokines and chemoattractants are the prototypical integrin activators, but carbohydrate binding to cell surface lectins can induce a host of intracellular signaling pathways (43, 44, 45). -Gal coimmobilized with increasing densities of VCAM-1 did not augment monocyte firm arrest, nor did it increase the strength of monocyte adhesion to VCAM-1 (Fig. 8) Furthermore, soluble -gal did not induce high affinity ligand binding of monocyte 4 for the binding sequence of VCAM-1 (data not shown). In contrast, -gal increased the strength of adhesion for both ICAM-1 and ICAM-2, as evidenced by an increased resistance to detachment at varying levels of shear stress, suggesting that -gal induced high affinity 2 integrin binding (Fig. 9, A and B). This was confirmed by the ability of soluble -gal to induce the mAb24 epitope, which is a reporter of the high affinity state for 2 integrins (Fig. 9C) (46). We have previously demonstrated that coculture of human monocytes with PAEC for 30 min resulted in up-regulation of 2 surface levels in an -gal-dependent manner.5 Thus, -gal may regulate both 2 integrin avidity and affinity, but affinity regulation probably accounts for the rapid arrest of flowing monocytes on PAEC.
Understanding the initial cascade of activating sequences culminating in firm monocyte adhesion to xenogeneic endothelium is crucial to assure that the damage to xenografts inflicted by host innate immune cells is minimized. Numerous in vivo and in vitro studies have established host monocytes and NK cells as primary mediators of delayed xenograft rejection (9, 12, 13, 14, 15, 16, 17, 47, 48). Once firmly adherent to xenograft endothelium, host monocytes or NK cells fulfill their effector functions, including elaboration of proinflammatory and prothrombotic mediators, direct endothelial cytolysis, and induction of endothelial adhesion molecule expression. The aim of this study was to understand monocyte-activating events that resulted in firm adhesion, so that eventually this knowledge can be translated into strategies preventing xenograft monocyte infiltration. Hopefully, transgenic strategies to alter xenograft endothelial glycosylation will allow prolonged survival of xenografts by reducing the insult from both innate and acquired immune responses. Homozygous knockout pigs for the 1,3-galactosyltransferase gene responsible for decorating the endothelium with -gal are now available, so these goals may soon be a clinical reality (48). Nevertheless, understanding novel lectin-dependent adhesion pathways may prove useful in this context, because -gal knockout animals may unmask additional glycosylated epitopes recognized by monocyte lectins. Evidence that monocytes accumulate in atherosclerotic plaques displaying advanced glycation end products implies that monocytes could potentially recognize altered glycosylation patterns in human diseases, such as atherosclerosis or allograft vasculopathy, via lectin-dependent mechanisms (49, 50).
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by the Canadian Institutes of Health Research and the Cystic Fibrosis Foundation. M.D.P. was supported by the Thoracic Surgery Foundation for Research and Education.
2 Address correspondence and reprint requests to Dr. Thomas K. Waddell, Toronto General Hospital, EN 10-233, 200 Elizabeth Street, Toronto, Ontario, Canada M5G 2C4. E-mail address: tom.waddell{at}uhn.on.ca
3 Abbreviations used in this paper: DXR, delayed xenograft rejection; BS-IB4, Bandeiraea simplicifolia isolectin B4; -gal, galactose(1,3)galactose(1,4) GlcNAc-R; HAEC, human aortic endothelial cell; MFI, mean fluorescence intensity; PAA, poly[N-(2-hydroxyethyl)acrylamide]; PAEC, porcine aortic endothelial cell; UEA-1, Ulex europeaus agglutinin 1.
4 The online version of this article contains supplemental material.
5 E. Vlasova, M. D. Peterson, and T. K. Waddell. Submitted for publication.
Received for publication November 19, 2004. Accepted for publication March 25, 2005.
References
Lin, S. S., B. C. Weidner, G. W. Byrne, L. E. Diamond, J. H. Lawson, C. W. Hoopes, L. J. Daniels, C. W. Daggett, W. Parker, R. C. Harland, et al 1998. The role of antibodies in acute vascular rejection of pig-to-baboon cardiac transplants. J. Clin. Invest. 101: 1745-1756.
Leventhal, J. R., A. J. Matas, L. H. Sun, S. Reif, R. M. Bolman, III, A. P. Dalmasso, J. L. Platt. 1993. The immunopathology of cardiac xenograft rejection in the guinea pig-to-rat model. Transplantation 56: 1-8.
]
Palmetshofer, A., U. Galili, A. P. Dalmasso, S. C. Robson, F. H. Bach. 1998. -Galactosyl epitope-mediated activation of porcine aortic endothelial cells: type II activation. Transplantation 65: 971-978.
Blakely, M. L., W. J. van der Werf, M. C. Berndt, A. P. Dalmasso, F. H. Bach, W. W. Hancock. 1994. Activation of intragraft endothelial and mononuclear cells during discordant xenograft rejection. Transplantation 58: 1059-1066.
Itescu, S., P. Kwiatkowski, J. H. Artrip, S. F. Wang, J. Ankersmit, O. P. Minanov, R. E. Michler. 1998. Role of natural killer cells, macrophages, and accessory molecule interactions in the rejection of pig-to-primate xenografts beyond the hyperacute period. Hum. Immunol. 59: 275-286.(Mark D. Peterson*,, Rongy)
Monocytes are the predominant inflammatory cell recruited to xenografts and participate in delayed xenograft rejection. In contrast to allogeneic leukocytes that require up-regulation of endothelial adhesion molecules to adhere and emigrate into effector tissues, we demonstrate that human monocytes adhere rapidly to unstimulated xenogeneic endothelial cells. The major xenoantigen galactose(1,3)galactose(1,4)GlcNAc-R (-gal) is abundantly expressed on xenogeneic endothelium. We have identified a putative receptor for -gal on human monocytes that is a member of the C-type family of lectin receptors. Monocyte arrest under physiological flow conditions is regulated by -gal, because cleavage or blockade results in a dramatic reduction in monocyte adhesion. Recruitment of human monocytes to unactivated xenogeneic endothelial cells requires both 4 and 2 integrins on the monocyte; binding of -gal to monocytes results in rapid activation of 2, but not 4, integrins. Thus, activation of monocyte 2 integrins by -gal expressed on xenogeneic endothelium provides a mechanism that may explain the dramatic accumulation of monocytes in vivo.
Introduction
Xenotransplantation, when ultimately successful, will alleviate the current severe organ shortage. At present, the rejection of xenografts by both the innate and adaptive immune systems is vigorous and poorly controlled by tolerable immunosuppressive protocols. Delayed xenograft rejection (DXR)3 is an intermediate form of graft rejection characterized by overwhelming monocyte and NK cell infiltration, Ab deposition along the endothelium, endothelial cell activation, thrombosis, and eventually graft loss (1, 2). Many components of the innate and acquired immune systems, including xenoreactive Abs, complement, coagulation factors, monocytes, and NK cells, can act independently or in concert to effect xenograft damage, but the mechanisms that initiate DXR are poorly understood (3, 4, 5, 6, 7, 8). Since DXR represents the most important immunological hurdle preventing routine use of xenografts for the treatment of end-organ failure, additional studies are necessary to elucidate the mechanisms underlying DXR.
Monocyte accumulation within discordant xenografts is striking. Monocyte infiltration occurs early after xenografting in both rodent- and pig-to-primate combinations, although the kinetics and severity of monocyte accumulation are more variable in pig-to-primate than in rodent models (2, 4, 5, 9, 10, 11, 12). Several lines of evidence support the role of monocytes and NK cells as effectors of DXR: depletion of monocytes or NK cells in xenograft recipients delays the onset of DXR, adoptive transfer of activated monocytes from rejected xenografts accelerates rejection in newly transplanted animals, and NK cells directly recognize and lyse xenogeneic endothelial cells through Ab-independent cytotoxic pathways (13, 14, 15, 16, 17). Furthermore, monocytes recruited into the xenograft elaborate multiple proinflammatory cytokines and chemokines, such as TNF-, IFN-, and MCP-1, that amplify the inflammatory response (4, 9, 12, 15). Direct monocyte or NK cell contact also activates xenogeneic endothelial cells to up-regulate adhesion molecule, cytokine, and tissue factor expression (13, 18, 19, 20). Recent work from our laboratory has elucidated some of the mechanisms underlying monocyte-induced endothelial cell activation (21). Few studies have addressed the trafficking of monocytes to xenogeneic endothelium. Leukocyte emigration into inflamed tissues involves intimate contact with vascular endothelium and multiple adhesive and signaling events and follows a multistep paradigm of rolling, arrest, firm adhesion, and diapedesis before migration into inflamed tissues (22). Human leukocytes are also capable of adhering to both quiescent and cytokine-activated xenogeneic endothelium (21). Robinson et al. (23) demonstrated that human neutrophil adhesion to TNF-stimulated porcine endothelium is orchestrated through known adhesion molecule receptor-ligand pairing. Study of human monocyte adhesion to porcine endothelium indicates that VCAM-CD49d (4 integrin) interaction supports monocyte adhesion to TNF-stimulated porcine endothelium, albeit less efficiently than monocyte adhesion to HUVEC (24). However, these authors could not explain the increased frequency of human monocyte adhesion to unstimulated porcine endothelium compared with human endothelium.
Previous studies suggest that the major xenoantigen galactose(1,3)galactose(1,4)GlcNAc-R (-gal), a terminal carbohydrate expressed on xenogeneic endothelium, participates in monocyte adhesion. For instance, transfection of HUVEC with an 1,2-fucosyltransferase reduced surface expression of -gal and decreased monocyte adhesion by 50% (25). These authors did not definitively elucidate whether -gal functions as an adhesion molecule, modulates the affinity of existing adhesion molecules, or leads to cellular activation by acting as a ligand for some undefined signaling receptor.
Our objective was to directly compare adhesion of human monocytes to unstimulated porcine (PAEC) and human (HAEC) aortic endothelial cells under conditions of physiological flow. Specifically, we examined the contributions of the endothelial carbohydrate -gal in regulating monocyte adhesion to xenogeneic endothelial cells. As expected, naive allogeneic human monocytes do not adhere to quiescent unstimulated human endothelial cells, yet human monocytes rapidly adhere to unactivated porcine endothelial cells under conditions of laminar flow. We found that monocytes possess a putative receptor for -gal that activates 2 integrin binding and promotes the arrest of monocytes on porcine endothelial cells.
Materials and Methods
VCAM-1/Fc, ICAM-1/Fc, ICAM-2/Fc, and -gal-BSA were immobilized on plastic according to a protocol described previously with the following modifications (28). Goat anti-human IgG (Fc-specific) F(ab')2 was passively adsorbed onto the center of a 35-mm polystyrene tissue culture dish (Corning) by incubating a 10-μl drop for 60 min in a humidified atmosphere (100 μg/ml; 24°C). Dishes were washed with PBS, followed by passive adsorption of -gal-BSA or BSA alone at the specified concentration (60 min; 24°C). Excess -gal-BSA was removed by washing with PBS, and nonspecific binding sites were blocked with 5% FBS (60 min; 24°C). The anti-Fc-coated area was then incubated with a saturating concentration of VCAM-1/Fc, ICAM-1/Fc, or ICAM-2/Fc (10 ml-drop; total concentration, 20 μg/ml; 60 min; 24°C), which corresponds to a coating density of 1200 molecules/μm2. In some experiments the density of immobilized VCAM-1/Fc was modulated by adjusting the relative molarity with nonimmune human IgG. For plates containing -gal-BSA or BSA alone, -gal-BSA or BSA was immobilized in the center of plastic dishes at 1, 10, 50, or 100 μg/ml (60 min; 24°C), followed by washing and blockade of nonspecific binding sites with 5% FBS. The efficacy of -gal-BSA adsorption was confirmed by ELISA; triplicate wells of a 96-well plate coated with -gal-BSA were labeled with BS-IB4-biotin (60 min; 24°C), followed by streptavidin-HRP (60 min; 24°C) and 3,3',5,5'-tetramethylbenzidine (Rockland) for 15 min, and the OD was read on an ELISA plate reader at 450 nm.
Parallel plate flow chamber assays
All adhesion assays were performed using a parallel plate flow chamber (Glycotech) at a shear stress of 0.8 or 1.0 dyne/cm2 unless otherwise specified. Confluent monolayers of HAEC and PAEC, or dishes coated with VCAM-1, ICAM-1, ICAM-2, or -gal-BSA (or a combination) were mounted in the flow chamber and placed on the heated stage (37°C) of a Diaphot 300 inverted phase-contrast microscope (Nikon). Experiments were videotaped with a Sony DXC-151A color video camera and Sony SVT-S3100 time-lapse video cassette recorder for later offline analysis. Purified monocytes (0.5 x 106) were resuspended in 500 μl of HBSS (with 1 mM Ca2+/Mg2+) supplemented with 10 mM HEPES and 0.5% FBS (assay buffer) and injected into the flow chamber. Adherent monocytes were counted in six consecutive high power fields after 4 min of flow. Monocytes were defined as firmly adherent if they remained stationary for >4 s. It was unusual for adherent monocytes to subsequently release.
The strength of monocyte adhesion to -gal coimmobilized with an integrin ligand was tested in a cell detachment assay by injecting 300 μl (5 x 105 monocytes/ml) into an inverted flow chamber of plastic dishes coated with BSA or -gal-BSA plus ICAM-1, ICAM-2, or VCAM-1. Monocytes were allowed to settle for 5 min, after which increasing levels of shear stress were applied in 30-s increments by pulling the assay buffer across the flow chamber with a programmable syringe pump (Genie; Kent Scientific). Assay buffer was maintained at 37°C in a heated water bath, and the stage of the microscope was heated with an infrared heat lamp coupled to a temperature controller. Adherent cells were counted at each level of shear stress.
Flow cytometric analysis
The capacity for human monocytes to bind directly to -gal was determined by incubating PBMCs with increasing concentrations of either soluble -gal-FITC (0070-PAA-FP) or control glycoconjugate-FITC (0000-PAA-FP) for 30 min at 4°C. Before incubation with glycoconjugates, nonspecific binding sites were blocked with 500 μg/ml 0000-PAA for 20 min at 4°C. In some experiments, PBMCs were also pretreated with either 20 μg/ml isotype or blocking mAbs to monocyte 4, 1, and 2 integrins and L-selectin (10 μg/ml Dreg 200). Specific binding of soluble -gal to monocytes was determined by coincubating PBMCs with the lectin BS-IB4, which specifically recognizes -gal residues. The correct gating for monocytes was previously determined with specific Abs for CD14. The mean fluorescence intensity (MFI) was measured for each population.
Exposure of the mAb24 epitope is a reporter of the 2 integrin high affinity state. Monocytes were incubated with soluble -gal-BSA or BSA alone for 1–30 min at 37°C. At the end of each incubation period, ice-cold PBS was added, and the monocytes were washed once, followed by addition of 10 μg/ml mAb24 or an isotype control for 20 min at 4°C. After washing, monocytes were labeled with a secondary PE-conjugated F(ab')2 for 20 min at 4°C, washed, and analyzed by FACS.
Statistical analysis
Statistical analysis was performed with SAS software (SAS version 8.1; SAS Institute). Continuous variables are expressed as the mean ± SEM. Comparisons of continuous variables between groups were performed with unpaired Student’s t tests or ANOVA. Orthogonal contrasts between control and treatment groups were made to determine significance where appropriate. A two-tailed value of p < 0.05 was considered statistically significant.
Results
Human monocytes adhere to unstimulated porcine but not human endothelium
Adhesion of human monocytes is much greater to PAEC than to HAEC under physiological flow conditions. Unstimulated human monocytes bind more avidly to unstimulated PAEC than to HAEC (33.7 ± 5.6 vs 2.5 ± 0.6 monocytes/field, respectively; p = 0.009; Fig. 1A). Monocyte accumulation begins immediately after injection into the flow chamber and continues in a linear fashion for the 4-min period of observation. Most monocytes arrested from fast-rolling, rather than slow-rolling, monocytes (see supplemental online video).4 Monocytes continued to accumulate on PAEC for 30 min, and very little adhesion was observed for HAEC even after 30 min (data not shown). In contrast, monocyte adhesion to TNF-stimulated endothelial cells did not differ between species (69.4 ± 5.0 vs 65.5 ± 6.4 for PAEC vs HAEC, respectively; p = NS; Fig. 1A). Adhesion of human monocytes to PAEC is dependent on the level of shear stress. We perfused unstimulated human monocytes over PAEC at four different levels of shear stress for 4 min: 0.5, 1.0, 2.0, and 5.0 dyne/cm2. Firmly adherent monocytes per field were counted at the end of 4 min and averaged over six fields. Monocyte adhesion was significantly lower for increasing levels of shear, particularly at shear levels of 2.0 and 5.0 vs 0.5 dyne/cm2 (p < 0.05; Fig. 1B).
To ensure that increased monocyte adhesion to xenogeneic endothelial cells was not due to basal differences in endothelial adhesion molecule expression, we examined the basal expression of various adhesion molecules on unstimulated endothelial cells. As expected, quiescent human endothelial cells expressed ICAM-1 (not shown). Attempts to determine resting ICAM-1 levels on porcine endothelial cells were unsuccessful due to lack of Abs reacting or cross-reacting with porcine cells. Otherwise, human and porcine endothelial cells did not express significant levels of VCAM-1, E-selectin, or P-selectin, yet significantly increased the expression of VCAM-1, E-selectin, and ICAM-1 (on HAEC) after TNF- stimulation (Fig. 2). In contrast, we detected increased basal levels of L-selectin ligand on unactivated PAEC compared with HAEC (Fig. 2).
Monocyte 4 and 2 integrins, and not L-selectin, are necessary for adhesion to xenogeneic endothelium
Selectin binding to an appropriate ligand initiates tethering and rolling of monocytes on endothelial cells; rolling contributes to firm arrest by allowing sufficient time for both integrin activation and bond formation between leukocyte integrins and their counter-receptors (29, 30). Because quiescent PAEC expressed higher levels of L-selectin ligand compared with HAEC, we examined the contribution of L-selectin on monocyte adhesion to unstimulated PAEC. We also focused on monocyte integrins as potent mediators of monocyte firm adhesion. As a first step, we evaluated the requirements for extracellular calcium on monocyte adhesion to porcine endothelial cells, because both monocyte L-selectin and monocyte integrins contain binding sites for divalent cations that are necessary for their adhesive function. The absence of extracellular calcium from the perfusion buffer completely inhibited monocyte arrest on PAEC under laminar flow compared with control calcium-containing buffer (1.7 monocytes/field expressed as percentage of control arrest events; Fig. 3A; p = 0.04). To further characterize the role of monocyte integrins and L-selectin on adhesion, we blocked specific integrins and L-selectin with function-blocking mAbs. Blockade of 4 and 2 integrins resulted in a significant reduction in firm adhesion under laminar flow compared with an isotype-matched control Ab (6 ± 2.7 and 12 ± 2.8 vs 21 ± 4.4 monocytes/field for anti-4, anti-2, and isotype, respectively; p = 0.01; Fig. 3B). A combination of blocking Abs against 4 and 2 completely inhibited adhesion (0.64 ± 0.16 vs 21 ± 4.4 monocytes/field; p < 0.001; Fig. 3B), whereas blockade of monocyte L-selectin did not inhibit adhesion compared with isotype control (23 ± 5.5 vs 21 ± 4.4 monocytes/field; p = NS; Fig. 3B).
Human monocyte adhesion to xenogeneic endothelium is dependent on the major xenoantigen -gal
Porcine endothelial cells express high levels of -gal on their surface, which may function as a xenospecific adhesion molecule. Because basal adhesion molecule levels (except L-selectin ligand) were similar between human and porcine endothelial cells, yet L-selectin does not mediate increased arrest of monocytes on PAEC, we addressed the contribution of endothelial -gal to monocyte arrest on porcine endothelium by specifically cleaving terminal galactose residues with an exogalactosidase. Binding of the -gal-specific BS-IB4 was used to monitor the efficacy of -gal cleavage. Galactosidase treatment of porcine monolayers effectively cleaved the terminal Gal(1,3)Gal residues from the endothelium, as evidenced by a marked reduction of BS-IB4-FITC fluorescence compared with control monolayers (MFI, 120 vs 35 for control vs -galactosidase, respectively; Fig. 4A, right axis). Under these conditions, adhesion of monocytes to porcine endothelial cells was reduced by 54% vs control, from 35.8 ± 4.4 to 16.5 ± 4.0 monocytes/field (p = 0.035; Fig. 4A). In contrast, treatment of PAEC with a control galactosidase, which does not cleave -gal residues, resulted in a 26% reduction in monocyte adhesion vs control, from 35.8 ± 4.4 to 26.6 ± 8.0 monocytes/field, respectively (p = NS). We also blocked terminal -gal residues to confirm its function in mediating monocyte arrest on PAEC. Pretreatment of PAEC with BS-IB4 resulted in a 57% reduction in monocyte arrest vs control monolayers (10 ± 3.3 vs 24 ± 3.3 monocytes/field; p = 0.03; Fig. 4B). Treatment of monolayers with a control lectin that does not bind -gal, Ulex europeaus agglutinin 1 (UEA-1), had no effect. In contrast, up-regulation of classical adhesion molecules with TNF- before blockade with BS-IB4 minimized the effect of -gal on monocyte arrest. Monocyte arrest after BS-IB4 blockade on TNF--pretreated monolayers was reduced from 67 ± 11 and 72 ± 17 to 55 ± 13 monocytes/field for control and UEA-1 vs BS-IB4, respectively, but this reduction was not statistically significant (Fig. 4B; p = NS).
-Gal on xenogeneic endothelium is a putative ligand for human monocytes
Since xenogeneic endothelial cell -gal clearly contributes to monocyte arrest under conditions of laminar flow, we hypothesized that human monocytes could directly bind to -gal, possibly through a direct -gal-integrin interaction. We first determined whether monocytes could bind to -gal by incubating PBMCs with a soluble form of -gal conjugated to FITC. At a saturating concentration, all the monocytes bound to -gal (MFI, 180.1 vs 9.7, for -gal vs control; Fig. 5A). In contrast, most lymphocytes did not bind to -gal (MFI, 12.3 vs 1.3, for -gal vs control), with the exception of a small population that could have represented NK cells. The binding of -gal to human monocytes was dose dependent and saturatable (Fig. 5B). Saturation of the putative -gal receptor on monocytes was achieved at a dose slightly >500 μg/ml -gal. We also demonstrated -gal binding to human monocytes using -gal-BSA and directly biotinylated -gal (data not shown).
As additional evidence that the binding of -gal to monocytes was specific, we inhibited binding with the lectin BS-IB4. Incubation of monocytes with soluble -gal in the presence of BS-IB4 reduced monocyte binding of -gal by 90% of control-FITC levels (MFI, 77 ± 21, 18 ± 5.8, and 11 ± 1, for -gal-FITC, -gal-FITC plus BS-IB4, and control-FITC, respectively; p < 0.005 for -gal-FITC vs -gal-FITC plus BS-IB4; Fig. 5C). Thus, the putative -gal receptor on human monocytes demonstrates specificity and saturability for its ligand, characteristics consistent with classical receptor-ligand binding.
If monocyte integrins directly bound to -gal, then presumably this binding would also require extracellular calcium. Incubation of soluble -gal-FITC with monocytes in the absence of calcium led to a dramatic reduction of monocyte-bound -gal (MFI, 77 ± 21 vs 17 ± 6.2, for -gal-FITC vs -gal-FITC plus no calcium; p < 0.005; Fig. 5C). To definitively address the possibility that monocyte adhesion molecules might be receptors for -gal, we blocked monocyte 1, 2, and 4 integrins as well as L-selectin with function-blocking Abs. As demonstrated in Fig. 5D, monocytes bound to saturating levels of soluble -gal equally well in all groups, indicating that monocytes do not bind directly to -gal through these adhesion molecules (p = NS for no Ab and isotype control vs specific integrin- and L-selectin-blocking Abs). Because L-selectin is a C-type lectin, we tested two separate cell lines stably transfected with human L-selectin, K562 wild-type and U937, for their ability to bind soluble -gal. K562 parent cells do not express endogenous L-selectin and failed to bind either control or soluble -gal (Fig. 6A). In addition, the expression of L-selectin on the cell surface did not confer -gal binding to K562 wild-type or U937 cells, confirming that L-selectin is not the putative -gal lectin (Fig. 6). Taken together, these data suggest that monocytes possess an alternate receptor specific for -gal; the requirement of extracellular calcium for ligand binding indicates that the putative receptor may be a member of the C-type lectin family of carbohydrate receptors.
Immobilized -gal does not directly mediate monocyte rolling or firm arrest
Since human monocytes bind directly to -gal, we hypothesized that -gal may function as an adhesion molecule for monocytes. -gal-BSA or BSA alone was adsorbed onto tissue culture dishes, with saturation of all binding sites achieved at a dose of 100 μg/ml; the efficacy of -gal binding to plastic tissue culture plates was measured by ELISA (Fig. 7A). Monocytes did not roll (data not shown) or arrest on plates coated with a saturating amount of -gal (100 μg/ml) or IgG (20 μg/ml), whereas significant numbers of monocytes arrested on plates coated with a saturating level of VCAM-1 (1200 molecules/μm2; Fig. 7B; p < 0.0001 for -gal and IgG vs VCAM-1). We also modified the coating by first adsorbing an anti-Fc F(ab')2, followed by an anti-BSA Ab, and, lastly, -gal-BSA or BSA alone in an attempt to orient -gal to flowing monocytes. Under these conditions, the number of monocytes that firmly arrested to increasing doses of immobilized -gal or control BSA was similar after 4 min of laminar flow, suggesting that -gal does not function directly as an adhesion molecule (Fig. 7C; p = NS).
Human monocyte adhesion to VCAM-1 is not augmented by -gal
Monocyte adhesion to porcine endothelial cells is partly dependent on 4 integrin; therefore, we also examined whether -gal could modulate the affinity for ligands of 4 such as VCAM-1. As expected, monocyte adhesion was dependent on the coating density of VCAM-1, but coimmobilized -gal did not result in greater monocyte arrest than VCAM-1 alone (Fig. 8A; p = NS). Similarly, -gal coimmobilized with VCAM-1 did not result in greater strength of adhesion compared with VCAM-1 and BSA, because the number of adherent monocytes remaining after graded increases in shear stress were similar between conditions (Fig. 8B; p = NS). We also used flow cytometry to detect rapid and transient increases in 4 integrin affinity in response to soluble ligand binding, as previously described (31). Using this method, known chemoattractants, such as fMLP, induce high affinity 4 integrin ligand binding, but soluble -gal at 10 and 100 μg/ml failed to do so (data not shown).
-Gal induces high affinity 2 integrin ligand binding on human monocytes and subsequent arrest on ICAM-1 and ICAM-2
Although -gal does not directly mediate adhesion of monocytes to PAEC, we hypothesized that -gal may increase the affinity or avidity of monocyte 2 integrin for its ligand, ICAM-1. Indeed, monocyte adhesion after graded increases in shear stress was greater on -gal-BSA coimmobilized with ICAM-1 than to control BSA and ICAM-1 (Fig. 9A), suggesting that -gal binding to monocytes resulted in either 2 integrin up-regulation or increased affinity of 2 for ICAM-1. 2 integrins on monocytes include L2 (LFA-1) and M2 (Mac-1), and activation of either LFA-1 or Mac-1 could mediate increased strength of adhesion to ICAM-1. In contrast, only LFA-1 binds to ICAM-2 (32); therefore, we tested the strength of monocyte adhesion to -gal coimmobilized with ICAM-2 to determine whether LFA-1 mediated increased monocyte adhesion on ICAM-1. Monocytes adherent to -gal-BSA plus ICAM-2 resisted detachment at shear stresses of 4 and 10 dyne/cm2, more so than ICAM-2 plus BSA, suggesting that -gal activated LFA-1 (Fig. 9B).
To address the possibility that -gal binding to monocytes results in increased affinity of 2 integrin for ICAM-1 and ICAM-2, we coincubated monocytes with -gal-BSA or BSA alone and measured the exposure of the mAb24 epitope, a marker of the high affinity state for 2 integrins. Soluble -gal binding to monocytes resulted in activation of 2 integrins to a higher affinity state after only 1 min of coincubation and peaked after 5 min, compared with the BSA control (Fig. 9C). Incubation of human monocytes with a different -gal glycoconjugate, linked to PAA, also resulted in up-regulation of 2 integrins to a higher affinity state, similar to that observed after treatment with fMLP (data not shown).
Discussion
In the present study we demonstrate the following observations: monocytes arrest and firmly adhere on an order of magnitude greater (13-fold increase) to unstimulated xenogeneic than to allogeneic endothelial cells under physiologic flow conditions; monocytes possess a putative C-type lectin for -gal, the major xenoantigen, and -gal induces high affinity ligand binding of monocyte 2 integrins leading to firm adhesion to porcine endothelial cells. These findings propose a xenospecific adhesion mechanism that may explain the dramatic accumulation of monocytes observed early after xenografting.
The firm arrest of human monocytes on porcine aortic endothelial cells after 4 min of laminar flow was dramatic. This observation confirms a recent finding by our group that demonstrated that human monocytes adhered to unstimulated porcine endothelial cells as early as 15 s after flow was initiated. In both studies, predominantly fast-rolling, rather than slow-rolling, monocytes arrested on porcine endothelial monolayers, whereas very few monocytes rolled or arrested on human endothelial monolayers. In contrast, the number of monocytes that firmly arrested on TNF--activated PAEC or HAEC was not different, suggesting that human monocytes adhere to xeno- or alloendothelium through similar adhesion molecule-ligand pairings when endothelial adhesion molecule and chemokine expression is maximal.
Since very few monocytes adhered to unstimulated human endothelial cells, we hypothesized that the substantially greater binding of monocytes to porcine endothelial cells under flow conditions was due to differential basal adhesion molecule expression, terminal xenospecific endothelial glycosylation, monocyte activation, or a combination of these factors. To rule out basal differences in adhesion molecule expression as the reason for increased monocyte adhesion to PAEC, we compared adhesion molecule expression on unstimulated endothelial cells. Flow cytometric analysis of HAEC and PAEC confirmed an equally low basal level of adhesion molecule expression, with the exception of ICAM-1 and L-selectin ligands. Although ICAM-1 is normally constitutively expressed on quiescent vascular endothelium, we could only detect its presence on unactivated HAEC (33). We were unable to obtain Abs that cross-react between human and porcine ICAM; therefore, we cannot rule out the possibility that basal expression of ICAM-1 was higher on PAEC. One notable difference between HAEC and PAEC was increased basal expression of L-selectin ligand on PAEC. Giuffre et al. (34) also found that unactivated bovine aortic endothelial cells express high levels of L-selectin ligand, which did not increase after TNF- stimulation.
L-selectin binding to its ligand primarily mediates monocyte tethering and rolling along the endothelium (35, 36). Because unactivated PAEC, and not HAEC, express L-selectin ligands, we tested the contribution of L-selectin on the greater adhesion of monocytes to PAEC. The binding domain of L-selectin is contained within a terminal calcium-dependent (C-type) lectin (37); therefore, the complete inhibition of monocyte adhesion that we observed in calcium-free conditions suggested that L-selectin could potentially mediate monocyte tethering to PAEC (Fig. 3A). Surprisingly, blocking mAbs against L-selectin had no effect on either monocyte rolling or firm adhesion under flow (Fig. 3B), suggesting that monocytes tethered to PAEC via an alternate adhesion molecule. This is consistent with our observation that monocyte arrest was not preceded by slow rolling, which is usually characteristic of L-selectin-mediated events (see supplemental online video).
The presence of divalent cations, such as calcium and magnesium, is also critical for adhesive function of monocyte integrins; therefore, we also inhibited 4 and 2 integrins with blocking Abs. These integrins mediate monocyte firm arrest, and 4 may also support the capture and rolling of monocytes (38). Indeed, both 4 and 2 integrin were required for monocyte firm adhesion to PAEC, and 4 probably also contributed to monocyte capture, because L-selectin inhibition had no effect on the number of firmly adherent monocytes (Fig. 3B). The virtual absence of VCAM-1 on unstimulated PAEC suggests an alternate ligand for 4.
There are several mechanisms by which -gal might increase the adhesiveness of xenogeneic endothelium to human monocytes. -gal is a terminal pentasaccharide expressed on the endothelium of most mammals and New World monkeys, but it is absent from the endothelium of humans and Old World monkeys (39). Our earlier studies demonstrated that human monocyte adhesion was greater to PAEC than to HAEC, and cleavage of -gal from the surface of PAEC led to reduced monocyte adhesion under static conditions.5 Other strategies aimed at minimizing -gal expression also lowered monocyte adhesion to PAEC. For example, overexpression of an (1,2)-fucosyltransferase in porcine endothelial cells led to reduced -gal expression and subsequently lower monocyte adhesion and activation (25). However, these results must be interpreted with caution, because overexpression of a fucosyltransferase reduces terminal sialylation as well as -gal expression.
To precisely determine the contribution of -gal to adhesion, we studied monocyte adhesion to xenogeneic endothelial cells under laminar flow. Under low shear flow conditions, -gal mediated the arrest of human monocytes to unstimulated PAEC, because either cleavage or blockade of -gal led to a dramatic reduction in adhesion (Fig. 4). However, -gal immobilized to tissue culture dishes failed to stimulate monocyte rolling or firm arrest, even if -gal was presented efficiently above the plate using Abs (Fig. 7, B and C). The Gal(1, 3)Gal(1,4)GlcNAc-R structure was sufficient for soluble -gal binding to monocytes; however, additional glycoprotein domains may be necessary for -gal to function as an adhesion molecule in vitro or in vivo, as is the case for the dual chemokine and adhesion molecule, fractalkine. The soluble chemokine domain of fractalkine binds to its receptor, CXC3R1; however, the chemokine domain immobilized to plates does not mediate leukocyte adhesion unless it is presented on the mucin domain (40). Therefore, it is possible that -gal could function as an adhesion molecule if it were presented in a more physiological manner.
Monocytes possess a repertoire of cell surface adhesion molecules and receptors that orchestrate the rolling, activation, and firm arrest on vascular endothelial cells (22). In addition to classical adhesion molecules, monocytes express lectins that bind to specific carbohydrate sequences on bacteria and glycosylated Ags (41, 42). Preformed xenoreactive Abs target the major xenoantigen, -gal, because of a common terminal carbohydrate sequence with bacterial -gal, and similar cross-reactivity may occur between monocyte lectins and endothelial -gal (39). Indeed, we found that unstimulated human monocytes specifically bound to soluble -gal. Most lymphocytes did not bind to -gal, and the -gal binding lectin (BS-IB4) completely inhibited -gal binding to monocytes. Extracellular calcium was also required for -gal binding to monocytes, suggesting that the putative receptor is a C-type lectin. Furthermore, the failure of two separate leukocyte cell lines that were stably transfected with human L-selectin to bind soluble -gal provides proof that the putative -gal-binding, C-type lectin is not L-selectin (Fig. 6).
Since immobilized -gal did not directly mediate monocyte arrest on coated plates, we hypothesized that -gal binding to its putative receptor could modulate the affinity of 4 and 2 integrins for their ligands. Chemokines and chemoattractants are the prototypical integrin activators, but carbohydrate binding to cell surface lectins can induce a host of intracellular signaling pathways (43, 44, 45). -Gal coimmobilized with increasing densities of VCAM-1 did not augment monocyte firm arrest, nor did it increase the strength of monocyte adhesion to VCAM-1 (Fig. 8) Furthermore, soluble -gal did not induce high affinity ligand binding of monocyte 4 for the binding sequence of VCAM-1 (data not shown). In contrast, -gal increased the strength of adhesion for both ICAM-1 and ICAM-2, as evidenced by an increased resistance to detachment at varying levels of shear stress, suggesting that -gal induced high affinity 2 integrin binding (Fig. 9, A and B). This was confirmed by the ability of soluble -gal to induce the mAb24 epitope, which is a reporter of the high affinity state for 2 integrins (Fig. 9C) (46). We have previously demonstrated that coculture of human monocytes with PAEC for 30 min resulted in up-regulation of 2 surface levels in an -gal-dependent manner.5 Thus, -gal may regulate both 2 integrin avidity and affinity, but affinity regulation probably accounts for the rapid arrest of flowing monocytes on PAEC.
Understanding the initial cascade of activating sequences culminating in firm monocyte adhesion to xenogeneic endothelium is crucial to assure that the damage to xenografts inflicted by host innate immune cells is minimized. Numerous in vivo and in vitro studies have established host monocytes and NK cells as primary mediators of delayed xenograft rejection (9, 12, 13, 14, 15, 16, 17, 47, 48). Once firmly adherent to xenograft endothelium, host monocytes or NK cells fulfill their effector functions, including elaboration of proinflammatory and prothrombotic mediators, direct endothelial cytolysis, and induction of endothelial adhesion molecule expression. The aim of this study was to understand monocyte-activating events that resulted in firm adhesion, so that eventually this knowledge can be translated into strategies preventing xenograft monocyte infiltration. Hopefully, transgenic strategies to alter xenograft endothelial glycosylation will allow prolonged survival of xenografts by reducing the insult from both innate and acquired immune responses. Homozygous knockout pigs for the 1,3-galactosyltransferase gene responsible for decorating the endothelium with -gal are now available, so these goals may soon be a clinical reality (48). Nevertheless, understanding novel lectin-dependent adhesion pathways may prove useful in this context, because -gal knockout animals may unmask additional glycosylated epitopes recognized by monocyte lectins. Evidence that monocytes accumulate in atherosclerotic plaques displaying advanced glycation end products implies that monocytes could potentially recognize altered glycosylation patterns in human diseases, such as atherosclerosis or allograft vasculopathy, via lectin-dependent mechanisms (49, 50).
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by the Canadian Institutes of Health Research and the Cystic Fibrosis Foundation. M.D.P. was supported by the Thoracic Surgery Foundation for Research and Education.
2 Address correspondence and reprint requests to Dr. Thomas K. Waddell, Toronto General Hospital, EN 10-233, 200 Elizabeth Street, Toronto, Ontario, Canada M5G 2C4. E-mail address: tom.waddell{at}uhn.on.ca
3 Abbreviations used in this paper: DXR, delayed xenograft rejection; BS-IB4, Bandeiraea simplicifolia isolectin B4; -gal, galactose(1,3)galactose(1,4) GlcNAc-R; HAEC, human aortic endothelial cell; MFI, mean fluorescence intensity; PAA, poly[N-(2-hydroxyethyl)acrylamide]; PAEC, porcine aortic endothelial cell; UEA-1, Ulex europeaus agglutinin 1.
4 The online version of this article contains supplemental material.
5 E. Vlasova, M. D. Peterson, and T. K. Waddell. Submitted for publication.
Received for publication November 19, 2004. Accepted for publication March 25, 2005.
References
Lin, S. S., B. C. Weidner, G. W. Byrne, L. E. Diamond, J. H. Lawson, C. W. Hoopes, L. J. Daniels, C. W. Daggett, W. Parker, R. C. Harland, et al 1998. The role of antibodies in acute vascular rejection of pig-to-baboon cardiac transplants. J. Clin. Invest. 101: 1745-1756.
Leventhal, J. R., A. J. Matas, L. H. Sun, S. Reif, R. M. Bolman, III, A. P. Dalmasso, J. L. Platt. 1993. The immunopathology of cardiac xenograft rejection in the guinea pig-to-rat model. Transplantation 56: 1-8.
]
Palmetshofer, A., U. Galili, A. P. Dalmasso, S. C. Robson, F. H. Bach. 1998. -Galactosyl epitope-mediated activation of porcine aortic endothelial cells: type II activation. Transplantation 65: 971-978.
Blakely, M. L., W. J. van der Werf, M. C. Berndt, A. P. Dalmasso, F. H. Bach, W. W. Hancock. 1994. Activation of intragraft endothelial and mononuclear cells during discordant xenograft rejection. Transplantation 58: 1059-1066.
Itescu, S., P. Kwiatkowski, J. H. Artrip, S. F. Wang, J. Ankersmit, O. P. Minanov, R. E. Michler. 1998. Role of natural killer cells, macrophages, and accessory molecule interactions in the rejection of pig-to-primate xenografts beyond the hyperacute period. Hum. Immunol. 59: 275-286.(Mark D. Peterson*,, Rongy)