Glycoprotein V-deficient platelets have undiminished thrombin responsiveness and Do not exhibit a Bernard-Soulier phenotype.

Adhesion of platelets to extracellular matrix via von Willebrand factor (vWF) and activation of platelets by thrombin are critical steps in hemostasis. Glycoprotein (GP) V is a component of the GPIb-V-IX complex, the platelet receptor for vWF. GPV is also cleaved by thrombin. Deficiency of GPIb or GPIX results in Bernard-Soulier syndrome (BSS), a bleeding disorder in which platelets are giant and have multiple functional defects. Whether GPV-deficiency might also cause BSS is unknown as are the roles of GPV in platelet-vWF interaction and thrombin signaling. We report that GPV-deficient mice developed normally, had no evidence of spontaneous bleeding, and had tail bleeding times that were not prolonged compared with wild-type mice. GPV-deficient platelets were normal in size and structure as assessed by flow cytometry and electron microscopy. GPV-deficient and wild-type platelets were indistinguishable in botrocetin-mediated platelet agglutination and in their ability to adhere to mouse vWF A1 domain. Platelet aggregation and ATP secretion in response to low and high concentrations of thrombin were not decreased in GPV-deficient platelets compared with wild-type. Our results show that (1) GPV is not necessary for GPIb expression and function in platelets and that GPV deficiency is not likely to be a cause of human BSS and (2) GPV is not necessary for robust thrombin signaling. Whether redundancy accounts for the lack of phenotype of GPV-deficiency or whether GPV serves subtle or as yet unprobed functions in platelets or other cells remains to be determined.

Distinct granule populations in human neutrophils and lysosomal organelles identified by immuno-electron microscopy.

In this paper, we illustrate the fine structural localization of distinct marker proteins in the organelles of human neutrophils and outline our preferred methods for processing ultrathin cryosections for use with immunoelectron microscopy. Previous work has determined the subcellular localization of certain marker proteins within intact polymorphonuclear neutrophilic leukocytes (PMN) and PMN fractions. These are as follows: myeloperoxidase (MPO) for azurophilic granules, lactoferrin for specific/secondary granules, gelatinase for gelatinase/tertiary granules, albumin for the secretory vesicles, and HLA class I and L-selectin for the plasma membrane. In addition to analyzing the heterogeneity of the PMN granule populations, new information on the lysosomal system of this cell is reviewed and extended by the localization of the lysosome-associated membrane proteins (LAMPs) and the cation-independent mannose 6-phosphate receptor (CI-M6PR). LAMPs were absent in all identified granule populations, but were found in the membranes of vesicles, multivesicular bodies (MVB), and multilaminar compartments (MLC). We show here that MVB contain CI-M6PR whereas MLC do not. Furthermore, since MLC contain LAMPs but not the receptor, they probably correspond to the late endosome. By current criteria, the true lysosomes of the resting PMN are MVB and MLC. Finally, although azurophil granules contain acid hydrolases their membranes do not contain LAMPs and they cannot be classified as lysosomes, but rather are more similar to regulated secretory granules.

Cysteine protease inhibitors alter Golgi complex ultrastructure and function in Trypanosoma cruzi.

Cruzain, the major cysteine protease of the protozoan parasite Trypanosoma cruzi, is a target of rational drug design for chemotherapy of Chagas' disease. The precise biological role of cruzain in the parasite life cycle and the mechanism involved in the trypanocidal effect of cysteine protease inhibitors are still unclear. Here we report biological and ultrastructural alterations caused by cysteine protease inhibitors in T. cruzi epimastigotes. Cruzain, a glycoprotein that transits the Golgi-endosomal pathway, localized to pre-lysosomes/lysosomes in the posterior end of untreated epimastigotes by fluorescent microscopy utilizing either a biotinylated cysteine protease inhibitor to tag the active site, or a specific anti-cruzain antibody. Radiolabeled or biotinylated cysteine protease inhibitors bound exclusively to cruzain in intact epimastigotes confirming that cruzain is accessible to, and is targeted by the inhibitors. Treatment of T. cruzi epimastigotes with specific cysteine protease inhibitors arrested growth, altered the intracellular localization of cruzain, and induced major alterations in the Golgi complex. Following treatment, cruzain accumulated in peripheral dilations of Golgi cisternae. There was a concomitant 70% reduction in gold-labeled cruzain transported to lysosomes. Cisternae abnormalities in the Golgi compartment were followed by distention of ER and nuclear membranes. Brefeldin A increased the number and size of cisternae in epimastigotes. Pre-treatment of epimastigotes with cysteine protease inhibitors followed by exposure to brefeldin A induced a more rapid appearance of the cysteine protease inhibitor-induced Golgi alterations. Our results suggest that cysteine protease inhibitors prevent the normal autocatalytic processing and trafficking of cruzain within the Golgi apparatus. Accumulation of cruzain may decrease mobility of Golgi membranes and result in peripheral distention of cisternae. These major alterations of the Golgi complex parallel the death of T. cruzi epimastigotes.

Azurophilic granules of human neutrophilic leukocytes are deficient in lysosome-associated membrane proteins but retain the mannose 6-phosphate recognition marker.

During granulocyte differentiation in the bone marrow (BM), neutrophilic leukocyte precursors synthesize large amounts of lysosomal enzymes. These enzymes are sequestered into azurophilic storage granules until used days later for digestion of phagocytized microorganisms after leukocyte emigration to inflamed tissues. This azurophil granule population has previously been defined as a primary lysosome, i.e., a membrane-bound organelle containing acid hydrolases that have not entered into a digestive event. In this study, azurophil granules were purified and shown to contain large amounts of mannose 6-phosphate-containing glycoproteins (Man 6-P GP) but little lysosome-associated membrane proteins (LAMP). In addition, the fine structural localization of Man 6-P GP and LAMP was investigated at various stages of maturation in human BM and blood. Man 6-P GP were present within the azurophilic granules at all stages of maturation and in typical multivesicular bodies (MVB) as well as in multilaminar compartments (MLC), identified by their content of concentric arrays of internal membranes. LAMP was absent in all identified granule populations, but was consistently found in the membranes of vesicles, MVB, and MLC. The latter compartment has not been previously described in this cell type. In conclusion, the azurophilic granules, which contain an abundance of lysosomal enzymes and Man 6-P GP, lack the LAMP glycoproteins. By current criteria, they therefore cannot be classified as lysosomes, but rather may have the functional characteristics of a regulated secretory granule. Rather, the true lysosomes of the resting neutrophil are probably the MVB and MLC. Finally, the typical "dense bodies" or mature lysosomes described in other cells are not present in resting neutrophils.

The human antibacterial cathelicidin, hCAP-18, is synthesized in myelocytes and metamyelocytes and localized to specific granules in neutrophils.

hCAP-18 is the only human member of the antibacterial and endotoxin-binding family of proteins known as cathelicidins. The antibacterial and endotoxin binding domains reside in the C-terminal 37 amino acids of the protein (LL-37) and this is believed to be unleashed from the neutralizing N-terminus by proteases from peroxidase positive granules. In human neutrophils, peroxidase positive and peroxidase negative granules can be subdivided into granule subsets that differ in protein content and ability to be exocytosed. To determine the localization of hCAP-18, we performed high-resolution immuno-electron microscopy and subcellular fractionation on Percoll density gradients. Biosynthesis of hCAP-18 was investigated in isolated human bone marrow cells. hCAP-18 was found to colocalize and comobilize with lactoferrin, but not with gelatinase in subcellular fractions. This was confirmed by electron microscopy. hCAP-18 is synthesized at the same stage of myeloid cell maturation as lactoferrin, and is efficiently targeted to granules. Like the peroxidase negative granule's matrix metalloproteinases, collagenase and gelatinase, hCAP-18 is also stored in unprocessed form. hCAP-18 is a major protein of specific granules where it is present in equimolar ratio with lactoferrin.

Leukocyte activation induces surface redistribution of P-selectin glycoprotein ligand-1.

Binding of P-selectin on activated endothelium to P-selectin glycoprotein ligand-1 (PSGL-1) on neutrophils mediates the initial tethering and rolling of neutrophils on the vessel wall at inflammatory sites. Upon activation of rolling cells by locally expressed signaling molecules, integrin-dependent adhesion mechanisms are engaged and transendothelial migration proceeds. P-selectin binding sites are uniformly distributed on the surface of quiescent neutrophils, but are redistributed to the uropod of activated neutrophils. It is unclear whether this activation-induced change in the surface topography of P-selectin binding sites is due to surface redistribution of PSGL-1, shedding of PSGL-1 from the lamellapod, and/or movement of PSGL-1 from an intracellular compartment to the uropod of the polarized cell. With the use of immunogold electron microscopy we previously demonstrated that PSGL-1 was localized to the tips of microvilli on neutrophils. Here we document a similar localization for PSGL-1 on eosinophils, basophils, monocytes, and lymphocytes. On quiescent neutrophils, approximately 80% of the PSGL-1 label was on tips of microvilli, which are randomly distributed around the cell circumference. On activated, polarized neutrophils, the PSGL-1 label was restricted to a segment of approximately 42% of the cell circumference even though total labeling decreased by only approximately 26%. Latex microbeads coated with anti-PSGL-1 mAb bound preferentially to the uropod of activated neutrophils. Subcellular fractionation and immunogold analysis of frozen thin sections of neutrophils failed to detect PSGL-1 in any intracellular compartment. Taken together, these data indicate that the activation-induced change in the surface topography of PSGL-1 is due to surface redistribution of PSGL-1. This process may facilitate transendothelial migration by disrupting bonds between P-selectin and PSGL-1 at the leading edge of migrating cells.

Thrombin receptors on human platelets. Initial localization and subsequent redistribution during platelet activation.

Platelet responses to thrombin are at least partly mediated by a G-protein-coupled receptor whose NH2 terminus is a substrate for thrombin. In the present studies we have examined the location of thrombin receptors in resting platelets and followed their redistribution during platelet activation. The results reveal several new aspects of thrombin receptor biology. 1) On resting platelets, approximately two-thirds of the receptors were located in the plasma membrane. The remainder were present in the membranes of the surface connecting system. 2) When platelets were activated by ADP or a thromboxane analog, thrombin receptors that were initially in the surface connecting system were exposed on the platelet surface, increasing the number of detectable receptors by 40% and presumably making them available for subsequent activation by thrombin. 3) Platelet activation by thrombin rapidly abolished the binding of the antibodies whose epitopes are sensitive to receptor cleavage and left the platelets in a state refractory to both thrombin and the agonist peptide, SFLLRN. This was accompanied by a 60% decrease in the binding of receptor antibodies directed COOH-terminal to the cleavage site irrespective of whether the receptors were activated proteolytically by thrombin or nonproteolytically by SFLLRN. 4) The loss of antibody binding sites caused by thrombin was due in part to receptor internalization and in part to the shedding of thrombin receptors into membrane microparticles, especially under conditions in which aggregation was allowed to occur. However, at least 40% of the cleaved receptors remained on the platelet surface. 5) Lacking the ability to synthesize new receptors and lacking an intracellular reserve of preformed receptors comparable to that found in endothelial cells, platelets were unable to repopulate their surface with intact receptors following exposure to thrombin. This difference underlies the ability of endothelial cells to recover responsiveness to thrombin rapidly while platelets do not, despite the presence on both of the same receptor for thrombin.

A differential role for cell shape in neutrophil tethering and rolling on endothelial selectins under flow.

We investigated the role of neutrophil microvilli in interactions with E-selectin and P-selectin in hydrodynamic shear flow by disruption with cytochalasin B, hypotonic swelling, and chilling. Cytochalasin B only marginally reduced microvilli numbers (from 30 +/- 6 to 16 +/- 6 per cell perimeter, p < 0.005) as shown by electron microscopy, completely disrupted tethering in shear flow to E-selectin and P-selectin, increased the strength of rolling adhesions on E-selectin and P-selectin, and increased cell deformability in shear flow with a likely increase in the area of cell:substrate contact. Hypoosmotic swelling markedly reduced microvilli number (to 6 +/- 5 per perimeter, p < 0.005), almost completely inhibited tethering on E- and P-selectin, and increased the strength of rolling adhesions on P-selectin but not on E-selectin. Chilling almost completely abolished microvilli (to 3 +/- 3 per perimeter, p < 0.005), but pseudopod-like structures were present, and had little effect on tethering in flow. Immunogold labeling of L-selectin, which is normally clustered on tips of microvilli, showed that in the absence of microvilli it remained in small clusters. Our studies show that alterations in cell morphology and viscoelasticity can have opposing effects on tethering and rolling, showing that they are independently regulatable. Furthermore, our results suggest that the association of molecules that mediate rolling with microvilli tips may be important not just to enhance presentation, but for other functions such as to promote resistance to extraction from the membrane or cooperative interactions among clustered receptors.

Quantitation of L-selectin distribution on human leukocyte microvilli by immunogold labeling and electron microscopy.

L-Selectin is a leukocyte cell adhesion receptor that contributes to neutrophil (PMN) rolling on activated endothelium at sites of inflammation and mediates lymphocyte attachment to high endothelial venules in peripheral lymph nodes. Localization of this receptor to the tips of PMN and lymphocyte microvilli has been demonstrated. However, its distribution on these cells has not been quantified, and its localization on other leukocytes and the morphometry of microvilli on different leukocyte subpopulations have not been previously examined. In this study, PMN and mononuclear leukocytes were isolated from anticoagulated blood by dextran sedimentation and density centrifugation, fixed in 2% paraformaldehyde and 0.05% glutaraldehyde, immunogold-labeled for L-selectin, and embedded in Epon resin. The distribution of L-selectin was determined by counting gold particles on the plasma membrane of sectioned cells, and the surface microstructure of these cells was surveyed on two-dimensional transmission electron micrographs. On average, 78% of PMN, 72% of monocyte, and 71% of lymphocyte L-selectin was observed on the microvilli, with more variance on lymphocytes than the other cell types. Typical PMN and monocyte sections had 26 microvilli, whereas typical lymphocyte sections had 23. Quantitation of the distribution of L-selectin and leukocyte surface topology offers a foundation from which to study the requirement of microvilli or microvillus-localized L-selectin for leukocyte tethering and rolling in model systems that mimic microvascular environments.

Neutrophil rolling, arrest, and transmigration across activated, surface-adherent platelets via sequential action of P-selectin and the beta 2-integrin CD11b/CD18.

Platelets bound to thrombogenic surfaces have been shown to support activation-dependent firm adhesion of neutrophils in flow following selectin-mediated tethering and rolling. The specific receptor(s) responsible for mediating adhesion-strengthening interactions between neutrophils and platelets has not previously been identified. Furthermore, the ability of adherent platelets to support the migration of bound neutrophils has not been tested. We studied neutrophil interactions with activated, surface-adherent platelets as a model for leukocyte binding in vascular shear flow and emigration at thrombogenic sites. Our results demonstrate that the beta 2-integrin Mac-1 (CD11b/CD18) is required for both firm attachment to and transmigration of neutrophils across surface-adherent platelets. In flow assays, neutrophils from patients with leukocyte adhesion deficiency-1 (LAD-I), which lack beta 2-integrin receptors, formed P-selectin-mediated rolling interactions, but were unable to develop firm adhesion to activated platelets, in contrast to healthy neutrophils, which developed firm adhesion within 5 to 30 seconds after initiation of rolling. Furthermore, the adhesion-strengthening interaction observed for healthy neutrophils could be specifically inhibited by monoclonal antibodies (mAbs) to Mac-1, but not to lymphocyte function-associated antigen-1 (LFA-1; CD11a/CD18) or intercellular adhesion molecule-2 (ICAM-2; CD102). Further evidence for a beta 2-integrin-dependent neutrophil/platelet interaction is demonstrated by the complete inhibition of interleukin (IL)-8-induced neutrophil transmigration across platelets bound to fibronectin-coated polycarbonate filters by mAbs to Mac-1. Thus, Mac-1 is required for firm adhesion of neutrophils to activated, adherent platelets and may play an important role in promoting neutrophil accumulation on and migration across platelets deposited at sites of vascular injury.

Subcellular distribution of docking/fusion proteins in neutrophils, secretory cells with multiple exocytic compartments.

Neutrophils contain at least four distinct types of secretory organelles, which undergo exocytosis during infection and inflammation. The signaling pathways leading to secretion of individual granules and their kinetics of exocytosis vary greatly, causing temporal and regional differences in docking and fusion with the plasma membrane. As a step toward understanding the processes underlying differential granular secretion in neutrophils, we assessed the presence and distribution of a number of proteins reported to be involved in vesicular docking and/or fusion in other systems. Specific Abs were used for immunoblotting of cells fractionated by density gradients and free-flow electrophoresis, and for localization by confocal immunofluorescence and electron microscopy. Syntaxin 1, VAMP (vesicle-associated membrane protein)-1, synaptosome-associated protein-25 (SNAP-25), synaptophysin, and cellubrevin were not detectable in human neutrophils. In contrast, syntaxin 4, VAMP-2, and the 39-kDa isoform of secretory carrier membrane protein (SCAMP) were present. SCAMP was found mainly in secondary and tertiary granules and in a fraction containing secretory vesicles, but was virtually absent from the primary (lysosomal) granules. This profile is consistent with the proposed "post-Golgi" distribution of SCAMP. VAMP-2 was largely absent from primary and secondary granules, but concentrated in tertiary granules and secretory vesicles. This pattern of distribution parallels the increasing sensitivity of these exocytic compartments to intracellular free calcium. Accordingly, ionomycin induced translocation of VAMP-2 toward the plasma membrane. Syntaxin 4 was found almost exclusively in the plasma membrane, and it accumulated in lamellipodia of migrating cells. This regional accumulation may contribute to localized secretion into the phagosomal lumen.

Factor V is complexed with multimerin in resting platelet lysates and colocalizes with multimerin in platelet alpha-granules.

Factor V stored in platelets is an important source of factor Va for the prothrombinase complex. Investigations of potential platelet factor Va-binding proteins, using factor Va light chain affinity chromatography, identified a disulfide-linked multimeric protein with a reduced mobility of 155 kDa in the column eluate. Immunodepletion and immunoblotting indicated that this protein was multimerin. Multimerin specifically bound factors V and Va and the isolated factor Va light chain, but not the heavy chain of factor Va. Factor V stored in platelets, but not plasma factor V, was found to be complexed with multimerin. Multimerin immunodepletion of resting platelet lysates was associated with the removal of factor V and the loss of factor V coagulant activity. Immunoelectron microscopic studies colocalized factor V with multimerin in the alpha-granules of resting platelets. With thrombin-induced platelet activation, we observed dissociation of factor Va-multimerin complexes, multimerin-independent membrane binding of factor Va, and prothrombinase activity that was not inhibitable by multimerin antibodies. This study indicates that platelet factor V is stored as a complex with multimerin and suggests a possible role for multimerin as a carrier protein for factor V stored in platelets.

P-selectin glycoprotein ligand-1 mediates rolling of human neutrophils on P-selectin.

Neutrophils roll on P-selectin expressed by activated platelets or endothelial cells under the shear stresses in the microcirculation. P-selectin glycoprotein ligand-1 (PSGL-1) is a high affinity ligand for P-selectin on myeloid cells. However, it has not been demonstrated that PSGL-1 contributes to the rolling of neutrophils on P-selectin. We developed two IgG mAbs, PL1 and PL2, that appear to recognize protein-dependent epitopes on human PSGL-1. The mAbs bound to PSGL-1 on all leukocytes as well as on heterologous cells transfected with PSGL-1 cDNA. PL1, but not PL2, blocked binding of 125-I-PSGL-1 to immobilized P-selectin, binding of fluid-phase P-selectin to myeloid and lymphoid leukocytes, adhesion of neutrophils to immobilized P-selectin under static conditions, and rolling of neutrophils on P-selectin-expressing CHO cells under a range of shear stresses. PSGL-1 was localized to microvilli on neutrophils, a topography that may facilitate its adhesive function. These data indicate that (a) PSGL-1 accounts for the high affinity binding sites for P-selectin on leukocytes, and (b) PSGL-1 must interact with P-selectin in order for neutrophils to roll on P-selectin at physiological shear stresses.

A functional integrin ligand on the surface of platelets: intercellular adhesion molecule-2.

Activated platelets express P-selectin and release leukocyte chemoattractants; however, they have not been known to express integrin ligands important in the stabilization of leukocyte interactions with the vasculature. We now demonstrate the presence of intercellular adhesion molecular-2 (ICAM-2) (CD102), and lack of expression of other beta 2-integrin ligands, ICAM-1 (CD54) and ICAM-3 (CD50), on the surface of resting and stimulated platelets. ICAM-2 isolated from platelets migrates as a band of 59,000 M(r) in reducing sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Staining of bone marrow aspirates with anti-ICAM-2 mAb demonstrates strong reactivity to megakaryocytes. Using frozen thin sections and immunogold labeling, the antigen was shown to be present on the plasma membrane and surface-connected canalicular system of resting platelets. The average number of ICAM-2 molecules per platelet is 3,000 +/- 230 and does not change after activation. In adhesion assays, resting and stimulated platelets were capable of binding through ICAM-2 to purified leukocyte function-associated antigen-1. Activation of T lymphocytes with PMA stimulated binding to platelets that was Mg2+ dependent and could be specifically inhibited by mAbs to either ICAM-2 or leukocyte function-associated antigen-1. ICAM-2 is the only known beta 2-integrin ligand present on platelets, suggesting that it may play an important role in leukocyte-platelet interactions in inflammation and thrombosis.

Changes in subcellular localization and surface expression of L-selectin, alkaline phosphatase, and Mac-1 in human neutrophils during stimulation with inflammatory mediators.

The localization of the adhesion protein L-selectin in human neutrophils was determined by subcellular fractionation and immunoelectron microscopy and compared with the localization of Mac-1 (alpha m beta 2) and alkaline phosphatase, the marker for secretory vesicles. L-selectin was found to be localized exclusively on the plasma membrane of unstimulated cells and also of stimulated cells, although markedly diminished. This was in contrast to Mac-1, which was also localized in secretory vesicles and in specific/gelatinase granules as shown previously [Sengeløv, H., et al. J. Clin. Invest. (1993) 92, 1467-1476]. Stimulation of neutrophils with inflammatory mediators such as tumor necrosis factor (TNF), platelet-activating factor (PAF), or f-Met-Leu-Phe (fMLP), induced parallel up-regulation of the surface membrane content of alkaline phosphatase and Mac-1 and down-regulation of L-selectin, as evidenced by flow cytometry. Preimbedding immunoelectron microscopy confirmed that L-selectin was present mainly on tips of microvilli in unstimulated cells and showed that alkaline phosphatase and Mac-1 were randomly distributed on the surface membrane of fMLP-stimulated cells. These studies indicate that the transition of neutrophils from L-selectin-presenting cells to Mac-1-presenting cells induced by inflammatory mediators is mediated by incorporation of secretory vesicle membrane, rich in Mac-1 and devoid of L-selectin, into the plasma membrane.

Granulophysin is located in the membrane of azurophilic granules in human neutrophils and mobilizes to the plasma membrane following cell stimulation.

Granulophysin, a protein described in platelet dense granule membranes, has been shown to be similar or identical to CD63, a lysosomal membrane protein. We have previously shown granulophysin to be present in neutrophils using immunofluorescence. We now localize granulophysin to the neutrophil azurophilic granules by fine structural immunocytochemistry. Granulophysin expression on the surface membrane of the neutrophil is increased following stimulation of the cells, demonstrated by flow cytometry and fine structural immunocytochemistry. A similar pattern is shown for an anti-CD63 antibody. Incubation of activated neutrophils with D545, a monoclonal antibody to granulophysin, blocks subsequent binding of anti-CD63 antibodies to the cell surface, and anti-CD63 antibodies prevent subsequent binding of D545 as assessed by flow cytometry and immunoblotting. Our results support the homology of CD63 and granulophysin previously demonstrated in platelets and confirm CD63 as an activation marker in neutrophils and the first azurophilic granule membrane marker of neutrophils.

Neutrophil tethering to and rolling on E-selectin are separable by requirement for L-selectin.

Neutrophil tethering and rolling in shear flow are mediated by selectins and have been thought to be two indistinguishable manifestations of a single molecular interaction between selectin and ligand. However, we report that under physiologic flow conditions, tethering to E-selectin requires a ligand distinct from the one that supports neutrophil rolling. Tethering under shear to E-selectin requires a carbohydrate ligand that is closely associated with the lectin domain of L-selectin on the neutrophil surface, as enzymatic removal of L-selectin, chemotactic factor-induced shedding of L-selectin, and L-selectin MAbs effectively block tethering. In contrast, this ligand is dispensable for the ability to roll on E-selectin, since rolling adhesions formed after static incubations were not affected by the presence or absence of L-selectin. Thus, E-selectin interactions with ligands on neutrophils persist after L-selectin shedding. These findings add an additional step for regulation of leukocyte localization in inflammatory sites.