Transendothelial Migration of Neutrophils CD99 Is a Key Mediator

Home , CD99

CD99 Is a Key Mediator of the Transendothelial Migration of Neutrophils Olivia Lou, Pilar Alcaide, Francis W. Luscinskas and William A. Muller This information is current as of September 30, 2021. J Immunol 2007; 178:1136-1143; ; doi: 10.4049/jimmunol.178.2.1136 http://www.jimmunol.org/content/178/2/1136 Downloaded from

References This article cites 46 articles, 22 of which you can access for free at: http://www.jimmunol.org/content/178/2/1136.full#ref-list-1

Why The JI? Submit online. http://www.jimmunol.org/

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average by guest on September 30, 2021

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2007 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

CD99 Is a Key Mediator of the Transendothelial Migration of Neutrophils1

Olivia Lou,*† Pilar Alcaide,‡ Francis W. Luscinskas,‡ and William A. Muller2*†

Transendothelial migration of leukocytes is a critical event for inflammation, but the molecular regulation of this event is only beginning to be understood. PECAM (CD31) is a major mediator of monocyte and neutrophil transmigration, and CD99 was recently defined as a second mediator of the transmigration of monocytes. Expression of CD99 on the surface of circulating polymorphonuclear cells (PMN) is low compared with expression of CD99 on monocytes or expression of PECAM on PMN. We demonstrate here that, despite low expression of CD99, Fab of Abs against CD99 blocked over 80% of human neutrophils from transmigrating across HUVEC monolayers in an in vitro model of inflammation. Blocking CD99 on either the neutrophil or endothelial cell side resulted in a quantitatively equivalent block, suggesting a homophilic interaction between CD99 on the neutrophil and CD99 on the endothelial cell. Blocking CD99 and PECAM together resulted in additive effects, suggesting the two molecules work at distinct steps. Confocal microscopy confirmed that CD99-blocked neutrophils lodged in endothelial cell junc- Downloaded from tions at locations distal to PECAM-blocked neutrophils. The CD99-blocked PMN exhibited dynamic lateral movement within endothelial cell junctions, indicating that only the diapedesis step was blocked by interference with CD99. Anti-CD99 mAb also blocked PMN transmigration in a second in vitro model that incorporated shear stress. Taken together, the evidence demonstrates that PECAM and CD99 regulate distinct, sequential steps in the transendothelial migration of neutrophils during inflammation. The Journal of Immunology, 2007, 178: 1136–1143. http://www.jimmunol.org/ n exuberant neutrophil response is a hallmark of acute during the subsequent transmigration step (11). The leukocyte is then inflammation. To enter tissues, neutrophils (polymor- capable of transendothelial migration (TEM, transmigration, diapede- A phonuclear cells (PMN))3 must traverse the endothelial sis), the movement of the leukocyte in amoeboid fashion across the barrier, usually in a paracellular manner across endothelial cell interendothelial cell space and into the extracellular matrix (12, 13). (EC) borders (1, 2). Exit of neutrophils and other leukocytes from Transmigration is largely an irreversible process for neutrophils and vascular flow is a multistep process that begins with rolling of the commits these cells to enter the inflammatory site. It is therefore an leukocyte along the EC wall via interactions between selectins attractive step for anti-inflammatory therapeutic intervention. (CD62E, L, and P) and their sialylated receptors (3–5). The cap- PECAM-1 (PECAM and CD31), a 130-kDa member of the Ig by guest on September 30, 2021 tured leukocytes interact with chemokines or other agents pre- gene superfamily, is a well-characterized molecular mediator of sented on the apical surface of the endothelium, which activate diapedesis (14–17). Anti-PECAM blocking reagents have been ␤ ␤ leukocyte 1 and 2 integrins (LFA-1, Mac-1, and VLA-4) to bind demonstrated to block the transmigration of monocytes (14), neu- tightly to their ligands (ICAM-1, ICAM-2, and VCAM-1) (6–9). trophils (14, 18–20), NK cells (21), and eosinophils (22). A recent This causes arrest of leukocytes on the endothelial surface. The study using an in vitro model of inflammation described the in- leukocyte moves from the site of firm adhesion to the nearest junc- volvement of CD99, a heavily glycosylated 32-kDa transmem- tion in a newly identified step termed locomotion, which is medi- brane protein, as a second molecule in the transmigration of mono- ␤ ated by the 2 integrins and their counterreceptors ICAM-1 and cytes across EC borders (23). This study demonstrated that CD99 ICAM-2 (10). Indeed, LFA-1 and ICAM-1 remain tightly apposed controlled a process subsequent to PECAM-dependent interaction, suggesting that diapedesis itself can be a multistep process. This was also the first characterization of CD99 as an EC marker. *Department of Pathology and Laboratory Medicine and †Program in Immunology and Microbial Pathogenesis, Weill Medical College of Cornell University, New York, Considering the dominant role of neutrophils in the acute in- NY 10021; and ‡Center for Excellence in Vascular Biology, Department of Pathol- flammatory response (24), we set out to determine whether CD99 ogy, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA has a role in neutrophil diapedesis. Whereas PECAM is highly 02115 expressed by both neutrophils and monocytes, CD99 is expressed Received for publication July 11, 2006. Accepted for publication October 24, 2006. at low density on the surface of neutrophils compared with robust 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 expression on monocytes. Therefore, it was not apparent whether with 18 U.S.C. Section 1734 solely to indicate this fact. CD99 would play a role in neutrophil transmigration. Using an in 1 This work was supported by National Institutes of Health Grants HL064774 and vitro model of inflammation that allows simultaneous dissection of HL046849 (to W.A.M.) and HL53993 and HL36028 (to F.W.L.), a Fulbright-Spanish molecular interactions during diapedesis (25), we report here that Ministry of Education and Science (to P.A.), and a predoctoral fellowship from the Cancer Research Institute (to O.L.). blocking Abs against CD99 inhibit over 80% of neutrophil TEM, demonstrating that CD99 is a critical molecule in neutrophil trans- 2 Address correspondence and reprint requests to Dr. William A. Muller, Department of Pathology and Laboratory Medicine, Weill Medical College of Cornell Uni- migration across inflamed endothelium. versity, 1300 York Avenue C-312, New York, NY 10021. E-mail address: [email protected] 3 Abbreviations used in this paper: PMN, polymorphonuclear cell; EC, endothelial Materials and Methods cell; hpf, high-powered field; JAM, junctional adhesion molecule; M199, medium Human subjects 199; TEM, transendothelial migration. All human subject protocols were approved by the Institutional Review Copyright © 2007 by The American Association of Immunologists, Inc. 0022-1767/07/$2.00 Board. www.jimmunol.org The Journal of Immunology 1137

Antibodies Flow cytometry Monoclonal mouse-anti-human Abs hec2 (anti-CD99), hec7 (anti- All leukocyte staining and washing steps were conducted in PBS without PECAM), and hec1 (anti-VE-cadherin) were produced from hybridomas divalent cations plus 1% heat-inactivated FBS and analyzed with a FACS- generated in the lab as described previously (26). The anti-CD99 mAb Calibur (BD Biosciences) using CellQuest software. YG32 was purchased from DiNonA, and 12E7 and O662 were a gift from Dr. A. Bernard (Institut National de la Santeét de la Recherche Me´dicale Production of stable transfectants expressing huCD99-Fc Unite´ 576, Nice, France). Anti-CD11b (OKM1) was purchased from chimera American Type Culture Collection. Polyclonal Ab 177 was raised in rab- The oligonucleotides 5Ј-CCGGAATTCCGGCCGTGCCCAGCACCTGA bits against the extracellular domain of PECAM (Covance) and column Ј Ј Ј purified on protein A-Sepharose (Amersham Biosciences). Anti-junctional AC-3 and 5 -CGCGGATCCCTATTTACCCGGAGAC-3 were used to amplify the Fc region of human IgG1 from the DC67 plasmid (gift from adhesion molecule (JAM)-A mAb (clone 1H2A9) was generated as de- Ј scribed previously (27). Fab and F(abЈ) were cut from hec2 IgG1 using Genzyme, Framingham, MA). The oligonucleotides 5 -CCGCTCGA 2 GCGGGCCACCATGGCCCGCG-3Ј and 5Ј-CCGGAATTCCGGGGCGT immobilized ficin (Pierce) according to the manufacturer’s protocol and Ј purified using protein G column chromatography. Abs for immunofluores- CGGCCTCTTCCCCT-3 were used to amplify the extracellular region of cence were coupled to Alexa-488 and Alexa-546 according to the manu- human CD99 from a construct encoding human CD99 in pcDNA/neo (23). Human IgG1 Fc and the human CD99 extracellular region were sequen- facturer’s protocol (Molecular Probes/Invitrogen Life Technologies). Pu- tially inserted into the pcDNA3.1(Ϫ) vector (Invitrogen Life Technolo- rity of whole Ig or Ab fragments was confirmed by SDS-PAGE and gies), and the resulting construct was transformed into TOP10 chemically Coomassie staining, and endotoxin levels were confirmed to be at trace competent Escherichia coli (Invitrogen Life Technologies), as per the man- levels (Limulus assay; BioWhittaker). ufacturer’s protocol. Clones were screened by restriction enzyme digest and verified by DNA sequencing (Genewiz). CD99-Fc-pcDNA3.1 was electroporated into a CHO cell line, ldlD (29), a gift from Dr. M. Krieger HUVEC culture (Whitehead Institute, Massachusetts Institute of Technology, Cambridge, Downloaded from HUVEC were isolated by standard methods (26, 28) and cultured on fi- MA). Clones of stable transfectants were screened by ELISA using an bronectin-coated tissue culture dishes in medium 199 (M199; Invitrogen alkaline phosphatase-coupled Ab against the Fc portion of human Ig. Life Technologies) supplemented with 20% heat-inactivated normal hu- CD99-Fc chimera protein was harvested from the supernatant of spent man serum (from healthy volunteer donors) and penicillin and streptomy- cultures of selected clones, and purified by protein A-Sepharose column cin (Mediatech). HUVEC at passage 2 were cultured on hydrated collagen chromatography (Amersham Biosciences). ␮ type I (Vitrogen from Cohesiontech) gels set in 96-well plates with 50 l Immunofluorescence staining and confocal microscopy of collagen in each well (26). For the neutrophil transmigration assay, http://www.jimmunol.org/ ␤ HUVEC were stimulated with IL-1 (50 pg/ml) in 5% CO2 at 37°C for the The transmigration assay was conducted as described in the presence of final 4–6 h before transmigration. hec2 or 177 to block transmigration or hec1 as a control. The endothelial monolayers along with adherent and transmigrated PMN were fixed in 2% paraformaldehyde for 15 min. The collagen gels with the fixed EC and Leukocyte isolation from peripheral blood and neutrophil TEM PMN were then incubated with hec1-Alexa 488 (anti-VE-cadherin) to vi- assay sualize EC and anti-CD11b-Alexa 546 to visualize neutrophils. After ex- tensive washing, the collagen gels were examined using a Zeiss LSM510 Leukocytes were prepared from heparinized peripheral blood by density confocal microscope to acquire a stack of confocal images in representative sedimentation in a discontinuous gradient of Ficoll (Amersham Bio- areas of the monolayer. X-Z orthogonal reconstructions were performed

sciences) and Histopaque-1119 (Sigma-Aldrich). Care was taken to mini- using Metamorph software. by guest on September 30, 2021 mize physical disruption or temperature changes to the neutrophil preparation to avoid activation by experimental manipulation. All steps were Live imaging of PMN TEM under static conditions conducted at room temperature until the transmigration step, when the Neutrophils were added to HUVEC cultured to confluency atop collagen cocultures were placed in a 37°C tissue culture incubator. Low numbers of gels set into 30-mm tissue culture dishes with glass coverslip bottoms (10). remaining RBC, which do not interfere with the transmigration assay, were Interactions between neutrophils and EC were visualized by time-lapse not lysed to avoid osmotic shock to the PMN. The transmigration assay photography using a Zeiss Axiovert 200M widefield microscope. Differ- was performed as previously described (25), with slight modifications to ential interference contrast images were captured at ϫ40 magnification study neutrophils. PMN were resuspended in M199 plus 20% autologous every 20 s for 20 min by a charge-coupled device camera connected to the 5 serum at a concentration of 5 ϫ 10 cells/ml for transmigration. microscope and controlled by Metamorph software (Universal Imaging). EC monolayers activated for 4 h with IL-1␤ before the experiment were The images were converted into movie files using Metamorph or Quick- washed in warm M199, then PMN were added and allowed to transmigrate time (Apple Computers). for 20 min at 37°C. In certain experiments, some EC monolayers were left unstimulated or were activated with TNF-␣ (25 ng/ml) for4hascontrols. Live imaging of PMN TEM under fluid shear stress Abs at a concentration of 20 ␮g/ml were incubated with the PMN at room temperature for 15 min before TEM and allowed to remain for the duration Neutrophil-endothelial interactions were studied under defined laminar of the assay; in others, the blocking Abs were added to the PMN only or flow in a parallel plate flow chamber as described previously (30). Briefly, the HUVEC only and washed off before transmigration. In cases where two confluent HUVEC monolayers grown on 25-mm glass coverslips (Carolina ␮ Biological Supply) coated with 5 ␮g/ml fibronectin (Sigma-Aldrich) were blocking Abs were used in conjunction, 20 g/ml of each individual block- ␣ ing Ab was used. To ensure equal Ab concentration in parallel experimen- stimulated with 25 ng/ml TNF- for 4 h before use in transmigration as- tal conditions, the nonblocking hec1 was added to the other sample to bring says. HUVEC monolayers were also incubated with blocking or nonblock- ␮ ing control Abs before being placed in the flow chamber. Neutrophils were the final IgG concentration in each well to 40 g/ml. Nonadherent PMN ϫ 5 were removed using several washes with PBS, and the remaining adherent suspended to 5 10 cells/ml in Dulbecco’s PBS containing 0.2% HSA and incubated at room temperature for 15 min with various Abs before and transmigrated cells were fixed along with the endothelial monolayer by 2 incubation overnight in 2.5% glutaraldehyde (Electron Microscopy Sci- being drawn through the flow chamber at a constant rate of 0.76 dynes/cm . ences) in 0.1 M sodium cacodylate buffer (pH 7.4). HUVEC and neutrophil Abs were present in the flow buffer for the duration of the assay. Images were recorded in real time by videomicroscopy, and adhesion and trans- cocultures were differentially stained with Wright-Giemsa and/or silver ϫ stain, the collagen gel with the monolayer was removed and placed be- migration were determined from multiple fields using a 20 objective. Neutrophils moving beneath the EC layer were observed to darken and tween a glass slide and coverslip, and adherent and transmigrated cells become “phase-dense” (i.e., in a different plane of focus), and these were were counted in multiple fields from six replicates of each condition tested. scored as transmigrated cells. Total adhesion was calculated as the total number of cells, both adherent and transmigrated, per high-powered field (hpf). For each condition tested, a minimum of 500 PMN were counted from these individual replicate Results cultures. Transmigration data are expressed as the mean percentage of the Neutrophils freshly isolated from peripheral blood express far total cells that transmigrated below the endothelial layer (%TEM). Statis- less CD99 than monocytes tical significance was determined by one-way ANOVA with Tukey- Kramer and Newman-Keuls Multiple Comparison Test or Student’s t test, Homophilic interaction between CD99 on monocytes and CD99 where appropriate, using PRISM software (GraphPad). on EC borders is critical for TEM of monocytes (23). However, 1138 CD99 IN TEM OF PMN staining of CD99 by mAb hec2 on freshly isolated, unactivated neutrophils from peripheral blood was low in contrast to its expression on monocytes (Fig. 1). Staining of PMN by other anti- CD99 mAb, 12E7, O62, D44, and YG32 resulted in similar staining profiles (data not shown). It was therefore not immediately suggestive that CD99 would play the same critical role in neutrophil TEM as it does in monocyte transmigration.

Anti-CD99 Abs block neutrophil transmigration The presence of anti-CD99 mAb did not affect adhesion of PMN to the EC monolayers (Fig. 2A). However, anti-CD99 mAb had a major effect on PMN transmigration across IL-1␤-treated HUVEC, arresting neutrophils on the apical surface of the EC monolayer (Fig. 2C). This was despite the low expression of CD99 on unactivated PMN. Blocked PMN were observed to lodge within EC junctions, particularly at tricellular corners. To confirm that the PMN were arrested at EC junctions, we stained the PMN-EC cocultures after transmigration with silver nitrate, a procedure that defines EC junctions by the deposition of silver precipitate in EC Downloaded from junctions (Fig. 2B). The anti-CD99 mAb hec2 generally blocked Ͼ80% of PMN from transmigrating across IL-1␤-activated HUVEC in our in vitro system (Fig. 2C). An equivalent blockade was achieved with YG32, another anti-CD99 mAb (Fig. 2C). Quantitatively, the block obtained with anti-CD99 mAb was generally greater than http://www.jimmunol.org/ that obtained with 177, an anti-PECAM polyclonal rabbit Ab Ј (Figs. 2–4). Treating PMN with Fab or F(ab )2 of hec2 was com- parable to using whole Ig as a blocking agent, indicating that the blockade of PMN transmigration by anti-CD99 reagents was independent of PMN FcR engagement (Fig. 2). To demonstrate that blockade of TEM is due to specific interference with CD99 and not merely steric hindrance by junctional Ab, we used nonblocking anti-VE-cadherin mAb (hec1), which stains endothelial junctions similarly to hec2 and 177, as a control in the transmigration assay. by guest on September 30, 2021 hec1 at the same or higher concentration did not block PMN adhesion or transmigration (Fig. 2, compare anti-VE-cadherin to no Ab), demonstrating that the block in transmigration was not due to the physical presence of Abs “jamming up” the junction, or form- ing immune complexes for PMN to ingest.

Blocking CD99 on either neutrophil or EC reduces transmigration, suggesting a homophilic interaction FIGURE 2. Reagents against CD99 block TEM of neutrophils. PMN Previous studies showed that CD99 on monocytes and CD99 on were allowed to transmigrate for 20 min in the presence of 20 ␮g/ml of the EC interacted by a homophilic mechanism (23). We sought to indicated agents. A depicts the average number of adherent PMN per hpf; 20–30 hpf were examined for each condition. B illustrates the arrest of CD99-blocked PMN on IL-1␤-stimulated HUVEC stained with silver nitrate before ﬁxation. Note that some of the CD99-blocked PMN are arrested at tricellular junctions (white arrows). C shows the average percent- Ј age transmigration. The block by anti-CD99 Fab or F(ab )2 is equivalent to that by anti-CD99 whole IgG, indicating that this block is independent of Fc␥R engagement. All blocks are signiﬁcantly different from the “no Ab” and nonblocking VE-cadherin (anti-VE-cad) conditions (p Ͻ 0.001). The data are shown as mean adhesion or transmigration Ϯ SEM from six replicates of each condition, and the data are representative of a total of 5 Ј (YG32, CD99 Fab, F(ab )2, and CD99-Fc chimera) to 20 (VE-cadherin, PECAM, and CD99) independent experiments with different blood donors.

determine whether CD99 on PMN would also interact homophili- FIGURE 1. Low expression of CD99 on circulating PMN compared with monocytes. PMN (A) or monocytes (Mo) (B) were stained with hec2- cally with CD99 on EC and whether the CD99 expressed on PMN FITC or hec7-FITC to assay cell surface expression of CD99 or PECAM, was as important during TEM as EC CD99. Blocking CD99 on respectively. The histograms represent cells gated by both CD18 expres- either the PMN or the EC reduced TEM to the same extent (Fig. sion and forward and side scatter. The unﬁlled curves represent an isotype- 3), which was equivalent to blocking both sides of the PMN-EC matched (IgG1-FITC) staining control. interaction simultaneously. This is consistent with a requirement The Journal of Immunology 1139

FIGURE 5. Anti-CD99-treated PMN are blocked at a location downstream of anti-PECAM-blocked PMN. PMN and EC were pretreated with anti-VE-cadherin (negative control), anti-PECAM, or anti-CD99 before FIGURE 3. Blocking CD99 on EC, PMN, or both EC and PMN inhibits transmigration. Collagen gels were ﬁxed along with the PMN-EC cocul- transmigration equivalently. Anti-CD99 or anti-PECAM was added for 20 tures with 2% paraformaldehyde and stained with anti-VE-cadherin-Alexa f Ⅺ min to PMN ( )orEC( ), then unbound Ab was washed off before 488 (green) to visualize EC junctions and anti-CD11b-Alexa 546 (red) to ϩ p transmigration. For PMN EC samples ( ), blocking Abs were added to identify PMN. The images shown are along the X-Z plane (orthogonal to both EC and PMN and allowed to remain during the transmigration. After

monolayer). The green channel was intentionally amplified to mark the Downloaded from transmigration, monolayers were washed, fixed, and stained with Wright- location of the endothelial monolayer. VE-cadherin staining appears inter- Giemsa before PMN were counted under Nomarski optics. The data are mittent, since it is only present where the plane of imaging crosses a cell Ϯ shown as mean transmigration SEM (as for Fig. 2C), and the data are junction. In the VE-cadherin sample, the PMN is distinctly under the representative of five independent experiments. All blocks with anti-CD99 monolayer. White arrows in the CD99 samples pinpoint the contact be- Ͻ or anti-PECAM differed significantly (p 0.001) from the nonblocking tween PMN and EC, emphasizing that the PMN is partway through the VE-cadherin condition. Percentage of transmigration did not differ signif- junction. These images are representative of Ͼ250 cells examined for each icantly (p Ͼ 0.05) when the Ag was blocked on the neutrophil side, the condition in multiple experiments. http://www.jimmunol.org/ endothelial side, or both, for any of the Ab conditions. Block in transmigration by anti-CD99 differed significantly (p Ͻ 0.05) than the block by anti-PECAM under all conditions. percentage of PMN blocked from transmigrating was significantly greater than with either Ab alone (Fig. 4). On average, over 90% for a homophilic interaction between neutrophil CD99 and endo- of PMN were halted from transmigrating over the 20-min assay thelial CD99 during TEM. More direct evidence was provided by when treated with both blocking agents. This suggested they acted the fact that we could block TEM using a soluble CD99-Fc decoy at different steps in TEM. To characterize the spatial relationship protein (Fig. 2), which interacts with CD99 in a homophilic of CD99-blocked PMN to EC, we used confocal microscopy (Fig. manner. 5). PMN were incubated with HUVEC monolayers in the presence by guest on September 30, 2021 CD99 blocks downstream of PECAM during neutrophil of mAb as for the quantitative transmigration assays (Figs. 2–4). diapedesis The monolayers were then fixed and stained for VE-cadherin (to define EC junctions) and CD11b (to stain PMN) to characterize the When optimal blocking concentrations of anti-CD99 and anti- location of PMN blockade in the junction by confocal microscopy. PECAM Abs were added together during the in vitro assay, the X-Z orthogonal sections revealed that, in the presence of anti- PECAM Ab, PMN are arrested on top of the apical surface of the EC monolayer as described previously (14). In contrast, CD99- blocked PMN appeared lodged within the EC junctions (Fig. 5), similar to the appearance of monocytes blocked from transmigrating by anti-CD99 mAb (23). We conclude that CD99 inhibition of neutrophil transmigration occurs at a point later in the transmigration process than the PECAM block.

Effects of CD99 blockade are independent of the presence and type of cytokine stimulation While all of the data presented thus far show PMN interactions with IL-1␤-activated HUVEC monolayers, the effects of blocking CD99 and PECAM were seen whether HUVEC monolayers were cytokine activated or not, and whether they were activated with IL-1␤ or TNF-␣ (Fig. 6). Due to low levels of basal chemokine FIGURE 4. Blocking both PECAM and CD99 has an additive effect in secretion (see Discussion), PMN adhere at low levels to unstimu- blocking PMN transmigration. The transmigration assay was conducted lated HUVEC monolayers. Adhesion increased when HUVEC under standard conditions. Before transmigration, the EC monolayer was were treated with either cytokine. As previously shown for mono- ␮ treated for 30 min at 37°C with 20 g/ml of each of the Abs indicated. cytes (14, 23), anti-CD99 and anti-PECAM did not block adhesion Blocking PECAM, CD99, or both PECAM and CD99 resulted in transmi- of PMN to EC monolayers, while anti-CD18 (used as a positive gration percentages that differed significantly (p Ͻ 0.001) from no Ab or was sig- control) did block adhesion (Fig. 6A). Very few cells adhere in (ء) the nonblocking VE-cadherin (VE-cad) control. The CD99 block nificantly better than the PECAM block (p Ͻ 0.001). Blocking both PECAM the presence of anti-CD18 (Fig. 6A), but of those that do adhere, resulted in a block of transmigration significantly better (p Ͻ virtually 100% transmigrate (Fig. 6B). The low numbers of PMN (ءء) and CD99 0.01) than blocking either PECAM or CD99 alone. These data are represen- that adhere in the absence of cytokine stimulation transmigrate tative of five independent experiments from different blood donors. efficiently, and this TEM is blocked effectively with anti-CD99, 1140 CD99 IN TEM OF PMN Downloaded from

FIGURE 7. CD99-blocked PMN interact dynamically with EC mem- branes but are prevented from transmigrating across junctions. The montage shown represents the movement of a single representative PMN treated with 20 ␮g/ml anti-CD99 before transmigration. The montage represents a total of 420 s from the time the PMN landed onto the EC mono-

layer, with each frame representing a 20-s lapse. The polar ends of the http://www.jimmunol.org/ PMN, the leading edge and the uropod, are indicated by arrowheads. Three arrowheads represent a landing or a bifurcation. Differential contrast images were captured at ϫ40 magniﬁcation every 20 s for a total transmigration time of 20 min. To aid the reader, the EC junctions have been identiﬁed using white dotted lines in every fourth panel of the montage (time points 0, 80, 160, etc.) The images are selected from a video repre- FIGURE 6. Adhesion and percentage of transmigration of PMN across sentation (data not shown). unstimulated HUVEC and HUVEC stimulated with IL-1␤ or TNF-␣. Be- s

fore transmigration, HUVEC were left untreated ( ) or treated with 50 by guest on September 30, 2021 pg/ml IL-1␤ (f) or 25 ng/ml TNF-␣ (Ⅺ). The transmigration assay was conducted under standard conditions, with blocking Abs added to both CD99 plays a role in neutrophil diapedesis in an in vitro model PMN and EC. A shows an average of the total number of PMN adhering of fluid shear stress per hpf examined. B shows the corresponding average percentage of transmigration under the same conditions. Values shown are mean Ϯ SEM for In the vasculature, EC are subjected to biomechanical forces gen- six replicates for each condition. Value of p Ͻ 0.001 for %TEM of all erated by blood flow. EC can respond to these forces, modifying blocked conditions when compared with the “no Ab” or VE-cadherin con- their physical structure as well as up-regulating or down-regulating Ͼ ditions. For all conditions, there was no significant difference (p 0.05) in particular sets of genes (31). Hence, we investigated whether the ␤ the %TEM when the HUVEC were unstimulated or stimulated with IL-1 variable of laminar fluid shear stress would affect the ability of or TNF-␣. Values of p Ͻ 0.001 for all CD99 blocks in comparison with all blocking Abs against CD99 to inhibit neutrophil transmigration. PECAM blocks, and p Ͻ 0.01 for CD99ϩPECAM blocks in comparison with CD99 blocks for HUVEC stimulated with IL-1␤ or TNF-␣. The parallel plate flow chamber used for these experiments has been described previously (30). HUVEC grown to confluence on thin glass plates were prestimulated with TNF-␣ and set into the flow chamber. Neutrophils were perfused in the presence of CD99- anti-PECAM, or both. Anti-CD99 mAb was equally effective at blocking Abs at a steady shear rate of 0.76 dynes/cm2, parallel to blocking the relatively low numbers of PMN binding to unacti- the monolayer to provide laminar shear stress. The perfusion of vated HUVEC as it was at blocking PMN TEM across IL-1␤-or neutrophils was followed by perfusion of buffer containing block- TNF-␣-activated HUVEC (Fig. 6B). ing Ab to ensure exposure to the Abs as the neutrophils were transmigrating. Imaging of live neutrophil transmigration Under these experimental conditions, half the PMN were pre- We directly observed neutrophil interactions with EC during trans- vented from transmigrating by the presence of the anti-CD99 mAb, migration by time-lapse videomicroscopy to study the effect of hec2 (Fig. 8 and data not shown). In contrast, mAb against JAM-A CD99 blocking agents. As a control, neutrophils in the absence of had no significant effect, as reported previously (11). The live im- blocking Ab rapidly migrated across the EC monolayer as evi- aging of PMN transmigration under shear stress closely resembled denced by a visual change in phase density. In contrast, neutrophils the studies under static conditions (Fig. 7 and data not shown). in the presence of anti-CD99 mAb did not transmigrate, but instead CD99-blocked neutrophils demonstrated a wide range of morphol- were highly dynamic, moving laterally along EC junctions (Fig. 7 ogies. Whether under static or fluid shear conditions, some CD99- and data not shown). Often, the CD99-blocked neutrophils could blocked neutrophils appeared flattened and spread laterally along a be seen deforming the junctional membrane and pushing aside the junction whereas others would appear to be diving perpendicular cytoplasm of the EC, further evidence that the neutrophil was in to the EC monolayer, but with a pseudopod or a significant pro- fact stuck in the junction and had not transmigrated. portion of the cytoplasmic body lodged in the EC junction. The Journal of Immunology 1141

diapedesis of monocytes (23), and as demonstrated here, neutrophils. In agreement with the earlier study in monocyte transmigration (23), we demonstrate here that CD99 likewise comprises a “later step” in PMN transmigration. We show that CD99 homophilic interactions between EC and PMN occur downstream of PECAM- PECAM interactions (Fig. 5). Analysis by Nomarski optics (Fig. 2) and confocal microscopic imaging (Fig. 5) show that while PE- CAM-blocked PMN arrest on the EC surface overlying the junctions, CD99-blocked neutrophils are lodged within the intraendo- thelial junctional space. The mechanism by which CD99-blocked PMN become lodged downstream of those blocked by PECAM remains to be elucidated. FIGURE 8. Anti-CD99 mAb block PMN TEM in an in vitro model Still images of CD99-blocked PMN reveal a diverse morphol- incorporating shear stress. Blocking with anti-CD99 results in a transmi- ogy, with some blocked neutrophils stretched out and lodged lat- gration percentage signiﬁcantly different from the other TEM bars (p Ͻ erally in the junction whereas others have the bulk of their cell 0.01). HUVEC monolayers were stimulated with 25 ng/ml TNF-␣ for4h bodies in the luminal or abluminal side, but with a portion of their before transmigration. Neutrophils (5 ϫ 105 cells/ml in Dulbecco’s PBS) cytoplasm in the junctional space (data not shown). Live imaging

2 Downloaded from were drawn through the flow chamber at a constant rate of 0.76 dynes/cm . (Fig. 7 and data not shown) reveals that CD99-blocked PMN are ␮ PMN and EC were incubated with 20 g/ml of the indicated Abs before in fact not firmly arrested, but moving dynamically within EC transmigration, and Abs were present in the flow buffer for the duration of junctions, sometimes back and forth along the same junctional the assay. Images were taken by videomicroscopy every 15 s for a total of 10 min of transmigration. Fifteen to 20 cells were analyzed for each con- space as if looking for a place to transmigrate. This is reminiscent dition. The percentage of PMN transmigrating in each condition was de- of the appearance of monocytes blocked on TEM by anti-PECAM, termined from multiple fields using a ϫ20 objective. Neutrophils moving which “walked” along the tops of the EC borders (10). The live beneath the EC layer (i.e., in a different plane of focus) were scored as imaging illustrates the capacity of the neutrophil to deform the http://www.jimmunol.org/ transmigrated cells. Data represent the mean Ϯ SEM of three separate membrane of the EC to accommodate its movement (data not -Significantly different from no Ab or JAM-A (p Ͻ 0.01). shown). In our studies, transmigration was observed to occur para ,ء .experiments cellularly (between EC) as opposed to transcellularly (through the body of the cell) (32, 33). It is not known whether CD99 plays a Discussion role in transcellular migration, which is thought to comprise a small percentage (10% at most), of transmigration events. We demonstrate here that CD99 is a key molecule mediating the We primarily studied neutrophil transmigration in an assay de- diapedesis of neutrophils in an in vitro model of inflammation.

void of shear stress. Our static assay, which employs HUVEC by guest on September 30, 2021 CD99 expression on circulating neutrophils varies according to plated atop collagen gels set in 96-well plates (25), allows the EC donor but is consistently low in comparison to its expression on to form a physiologic basal lamina and physiological matrix into monocytes (Fig. 1). Nevertheless, we find that blocking agents which they can migrate. Findings from this assay have all been against CD99 inhibit over 80% of adherent neutrophils from mi- grating across HUVEC treated with the proinflammatory cytokine, verified in vivo (34, 35). The diapedesis step, unlike earlier steps ␤ ␣ Ј of rolling and adhesion, occurs largely independently of shear (36, IL-1 (Fig. 2), or TNF- (Fig. 6). Fab and F(ab )2 of blocking mAb, as well as CD99-Fc chimera selectively inhibit TEM as ef- 37). However, there are reports that TEM efficiency is enhanced in ficiently as intact IgG. Engaging leukocyte PECAM with Abs of- the presence of shear stress in vitro (38–40). Therefore, we sought ten enhances their adhesion to endothelial monolayers via inside- to verify our findings in a different in vitro model under flow con- ␤ ditions and found that under fluid shear stress as well, blocking out activation of 2 integrins (Fig. 2A and Ref. 21); they are nonetheless prevented from transmigrating by the same Ab (Fig. CD99 inhibits neutrophils from transmigrating (Fig. 8). Our mAb 2C and Ref. 14). No effect on adhesion is observed with anti-CD99 hec2 was quantitatively less effective at blocking TEM under these reagents. conditions. This is not likely to be due to the fact that these EC ␣ Preliminary experiments (data not shown) indicate that there is were activated by TNF- , since in the static assay, hec2 blocks a large pool of CD99 within neutrophils that becomes exteriorized equivalently under all conditions tested (Fig. 6). The differences when the PMN are activated. Thus, a PMN attached to the surface could be due to the assay conditions (static vs shear), and/or the of the EC has significantly higher CD99 surface expression than differences in preparation of the PMN. In the static assay all PMN the unactivated PMN in circulation. The exact conditions and sig- are subjected to transmigration at the same time-as soon as they are naling mechanisms that regulate this process are complex and are isolated. In the shear assay, since only one condition can be tested currently under investigation. at a time, the isolated PMN are held in suspension for longer pe- We find that binding of mAb to either neutrophil CD99 or en- riods. We have found that PMN tend to become activated upon dothelial CD99 results an equivalent quantitative block of trans- standing and this may affect their sensitivity to Ab blockade. migration, suggesting a homophilic interaction for CD99 (Fig. 3). Blocking CD99 alone in our model prevents Ͼ80% of PMN The fact that a CD99-Fc soluble chimera blocks PMN transmigra- from transmigrating; blocking PECAM generally prevents tion (Fig. 2C) lends additional support for a homophilic interaction 70–75% of PMN from transmigrating. We do not know why some between the chimera and CD99 on the EC or PMN. CD99 ho- PMN manage to transmigrate despite the presence of blocking mophilic adhesion was previously demonstrated by increased cell Abs. Perhaps PMN in different stages of their lifespan or different aggregation among L cells transfected with CD99; this increased states of activation have a different sensitivity to CD99 or PE- aggregation could be specifically blocked with the anti-CD99 CAM. Blocking both PECAM and CD99 does have an additive blocking mAb, hec2 (23). Thus it appears that homophilic PE- effect, blocking Ͼ90% of PMN (Figs. 4 and 6). The residual PMN CAM-PECAM and CD99-CD99 interactions are involved in the that transmigrate in the presence of the combined PECAM plus 1142 CD99 IN TEM OF PMN

CD99 blockade do not represent PMN that would have transmi- J. I. Gallin and R. Snyderman, ed. Lippincott Williams & Wilkins, Philadelphia, grated in the absence of cytokine stimulation. The low numbers of pp. 585–592. 3. Vestweber, D., and J. E. Blanks. 1999. Mechanisms that regulate the function of PMN that do adhere to HUVEC monolayers in the absence of the selectins and their ligands. Physiol. Rev. 79: 181–213. cytokine stimulation (due to low levels of chemoattractants made 4. Kishimoto, T. K., and R. Rothlein. 1994. Integrins, ICAMs, and selectins: role by EC and in serum that accumulate in the collagen gels (25)) and regulation of adhesion molecules in neutrophil recruitment to inflammatory sites. Adv. Pharmacol. 25: 117–169. transmigrate efficiently, and their transmigration is effectively 5. Rosen, S. D. 2004. Ligands for L-selectin: homing, inflammation, and beyond. blocked by Abs against PECAM and CD99 (Fig. 6). It is possible Annu. Rev. Immunol. 22: 129–156. that some functional redundancy exists between PECAM and 6. Johnston, B., and E. C. Butcher. 2002. Chemokines in rapid leukocyte adhesion triggering and migration. Semin. Immunol. 14: 83–92. CD99; however, our data are more suggestive of a sequential re- 7. Alon, R., and S. Feigelson. 2002. From rolling to arrest on blood vessels: leu- lationship with CD99 functioning downstream of PECAM. The kocyte tap dancing on endothelial integrin ligands and chemokines at sub-second temporal and spatial relationship between these molecules and contacts. Semin. Immunol. 14: 93–104. 8. Springer, T. A. 1990. Adhesion receptors of the immune system. Nature 346: their signaling will be an important area for research. 425–434. Other molecules have recently been implicated in diapedesis. 9. Springer, T. A. 1994. Traffic signals for lymphocyte recirculation and leukocyte The existence of PECAM- and CD99-independent pathways is emigration: the multistep paradigm. Cell 76: 301–314. 10. Schenkel, A. R., Z. Mamdouh, and W. A. Muller. 2004. Locomotion of mono- supported by our data, because even with our blocking Abs at cytes on endothelium is a critical step during extravasation. Nat. Immunol. 5: saturating conditions, a small proportion of neutrophils or mono- 393–400. cytes are capable of transmigrating. Several members of the JAM 11. Shaw, S. K., S. Ma, M. B. Kim, R. M. Rao, C. U. Hartman, R. M. Froio, L. Yang, T. Jones, Y. Liu, A. Nusrat, et al. 2004. Coordinated redistribution of leukocyte family have been implicated in transmigration. Several reports LFA-1 and endothelial cell ICAM-1 accompany neutrophil transmigration. suggest JAM-A mediates TEM of leukocytes (41, 42), although J. Exp. Med. 200: 1571–1580. Downloaded from the transmigration of human neutrophils (11) and monocytes (10) 12. Muller, W. A. 2002. Leukocyte-endothelial cell interactions in the inflammatory response. Lab. Invest. 82: 521–534. in vitro is unaffected by anti-JAM-A mAb. JAM-B and -C have 13. Muller, W. A. 2003. Leukocyte-endothelial-cell interactions in leukocyte trans- been implicated in leukocyte extravasation in vitro and in vivo migration and the inflammatory response. Trends Immunol. 24: 326–333. (43–45). Poliovirus receptor is a member of a related family of 14. Muller, W. A., S. A. Weigl, X. Deng, and D. M. Phillips. 1993. PECAM-1 is required for transendothelial migration of leukocytes. J. Exp. Med. 178: junctional proteins on EC and binds to CD226 on leukocytes. 449–460.

Blocking either poliovirus receptor or CD226 reduced monocyte 15. Newman, P. J., M. C. Berndt, J. Gorski, G. C. White II, S. Lyman, C. Paddock, http://www.jimmunol.org/ TEM in an in vitro assay (46). Thus, it appears that diapedesis is and W. A. Muller. 1990. PECAM-1 (CD31) cloning and relation to adhesion molecules of the immunoglobulin gene superfamily. Science 247: 1219–1222. regulated by a number of molecular interactions, some homophilic 16. Muller, W. A. 1995. The role of PECAM-1 (CD31) in leukocyte emigration: and some heterophilic. It will be important to determine whether in studies in vitro and in vivo. J. Leukocyte Biol. 57: 523–528. vivo these molecular gatekeepers act in series or in concert, and 17. Vaporciyan, A. A., H. M. Delisser, H.-C. Yan, I. I. Mendiguren, S. R. Thom, M. L. Jones, P. A. Ward, and S. M. Albelda. 1993. Involvement of platelet- whether they overlap in function or whether each one is required endothelial cell adhesion molecule-1 in neutrophil recruitment in vivo. Science to move a leukocyte across EC junctions. 262: 1580–1582. Inflammation is a critical part of the innate immune response 18. Christofidou-Solomidou, M., M. T. Nakada, J. Williams, W. A. Muller, and H. M. Delisser. 1997. Neutrophil platelet endothelial cell adhesion molecule-1 against infection; at the same time, a degree of inflammation is participates in neutrophil recruitment at inflammatory sites and is down-regulated present in most disease processes. For CD99 and PECAM to be after leukocyte extravasation. J. Immunol. 158: 4872–4878. by guest on September 30, 2021 considered as anti-inflammatory drug targets, it is important to 19. Gumina, R. J., J. E. Schultz, Z. Yao, D. Kenny, D. C. Warltier, P. J. Newman, and G. J. Gross. 1996. Antibody to platelet/endothelial cell adhesion molecule-1 re- understand the relative importance of each molecular interaction in duces myocardial infarct size in a rat model of ischemia-reperfusion injury. Cir- the diapedesis of different types of leukocytes. Neutrophils dom- culation 94: 3327–3333. inate in acute inflammatory responses (such as those in response to 20. Murohara, T., J. A. Delyani, S. M. Albelda, and A. M. Lefer. 1996. Blockade of platelet endothelial cell adhesion molecule-1 protects against myocardial isch- infection or following ischemia), whereas monocytes/macro- emia and reperfusion injury in cats. J. Immunol. 156: 3550–3557. phages take over in chronic inflammation. We demonstrate here 21. Berman, M. E., Y. Xie, and W. A. Muller. 1996. Roles of platelet/endothelial cell adhesion molecule-1 (PECAM-1, CD31) in natural killer cell transendothelial that CD99 plays a profound role in the transmigration of neutro- ␤ migration and 2 integrin activation. J. Immunol. 156: 1515–1524. phils. Similar to findings in monocyte TEM, we find that CD99 22. Chiba, R., N. Nakagawa, K. Kurasawa, Y. Tanaka, Y. Saito, and I. Iwamoto. appears to act at a step downstream of PECAM during diapedesis. 1999. Ligation of CD31 (PECAM-1) on endothelial cells increases adhesive ␣ ␤ ␤ Blocking both molecules has an additive effect in blocking almost function of v 3 integrin and enhances 1 integrin-mediated adhesion of eosinophils to endothelial cells. Blood 94: 1319–1329. all neutrophils and monocytes from transmigrating. It will be im- 23. Schenkel, A. R., Z. Mamdouh, X. Chen, R. M. Liebman, and W. A. Muller. 2002. portant to extend these studies into in vivo models. A recent study CD99 plays a major role in the migration of monocytes through endothelial demonstrated that polyclonal Ab against mouse CD99 inhibited junctions. Nat. Immunol. 3: 143–150. 24. Nathan, C. 2006. Neutrophils and immunity: challenges and opportunities. Nat. homing of T cell clones into inflamed skin (47). Generating re- Rev. Immunol. 6: 173–182. agents to block CD99 function in vivo and the generation of 25. Muller, W. A., and S. Weigl. 1992. Monocyte-selective transendothelial migra- CD99-deficient mice should shed further insight into CD99 function: dissection of the binding and transmigration phases by an in vitro assay. J. Exp. Med. 176: 819–828. tion in the inflammatory response. 26. Muller, W. A., C. M. Ratti, S. L. McDonnell, and Z. A. Cohn. 1989. A human endothelial cell-restricted, externally disposed plasmalemmal protein enriched in Acknowledgments intercellular junctions. J. Exp. Med. 170: 399–414. 27. Barton, E. S., J. C. Forrest, J. L. Connolly, J. D. Chappell, Y. Liu, F. J. Schnell, We thank Ron Liebman for excellent technical assistance, and Drs. Alan A. Nusrat, C. A. Parkos, and T. S. Dermody. 2001. Junction adhesion molecule Schenkel and Oliver Florey for comments on the manuscript. is a receptor for reovirus. Cell 104: 441–451. 28. Gimbrone, M. A., Jr. 1976. Culture of vascular endothelium. Prog. Hemost. Disclosures Thromb. 3: 1–28. The authors have no financial conflict of interest. 29. Krieger, M. 1986. Isolation of somatic cell mutants with defects in the endocy- tosis of low-density lipoprotein. Methods Enzymol. 129: 227–237. 30. Goetz, D. J., D. M. Greif, J. Shen, and F. W. Luscinskas. 1999. Cell-cell adhesive References interactions in an in vitro flow chamber. Methods Mol. Biol. 96: 137–145. 1. Burns, A. R., D. C. Walker, E. S. Brown, L. T. Thurmon, R. A. Bowden, 31. Topper, J. N., and M. A. Gimbrone, Jr. 1999. Blood flow and vascular gene C. R. Keese, S. I. Simon, M. L. Entman, and C. W. Smith. 1997. Neutrophil expression: fluid shear stress as a modulator of endothelial phenotype. Mol. Med. transendothelial migration is independent of tight junctions and occurs preferen- Today 5: 40–46. tially at tricellular corners. J. Immunol. 159: 2893–2903. 32. Feng, D., J. A. Nagy, K. Pyne, H. F. Dvorak, and A. M. Dvorak. 1998. Neutro- 2. Muller, W. A. 1999. Leukocyte-endothelial cell adhesion molecules in transen- phils emigrate from venules by a transendothelial cell pathway in response to dothelial migration. In: Inflammation: Basic Principles and Clinical Correlates. FMLP. J. Exp. Med. 187: 903–915. The Journal of Immunology 1143

33. Carman, C. V., and T. A. Springer. 2004. A transmigratory cup in leukocyte 41. Ostermann, G., K. S. Weber, A. Zernecke, A. Schroder, and C. Weber. 2002. ␤ diapedesis both through individual vascular endothelial cells and between them. JAM-1 is a ligand of the 2 integrin LFA-1 involved in transendothelial migration J. Cell Biol. 167: 377–388. of leukocytes. Nat. Immunol. 3: 151–158. 34. Liao, F., J. Ali, T. Greene, and W. A. Muller. 1997. Soluble domain 1 of platelet- 42. Martin-Padura, I., S. Lostaglio, M. Schneemann, L. Williams, M. Romano, endothelial cell adhesion molecule (PECAM) is sufficient to block transendothe- P. Fruscella, C. Panzeri, A. Stoppacciaro, L. Ruco, A. Villa, et al. 1998. Junc- lial migration in vitro and in vivo. J. Exp. Med. 185: 1349–1357. tional adhesion molecule, a novel member of the immunoglobulin superfamily that distributes at intercellular junctions and modulates monocyte transmigration. 35. Bogen, S., J. Pak, M. Garifallou, X. Deng, and W. A. Muller. 1994. Monoclonal J. Cell Biol. 142: 117–127. antibody to murine PECAM-1 (CD31) blocks acute inflammation in vivo. J. Exp. 43. Ludwig, R. J., T. M. Zollner, S. Santoso, K. Hardt, J. Gille, H. Baatz, Med. 179: 1059–1064. P. S. Johann, J. Pfeffer, H. H. Radeke, M. P. Schon, et al. 2005. Junctional 36. Smith, C. W., R. Rothlein, B. J. Hughes, M. M. Mariscalso, H. E. Rudloff, adhesion molecules (JAM)-B and -C contribute to leukocyte extravasation to the F. C. Schmalstieg, and D. C. Anderson. 1988. Recognition of an endothelial skin and mediate cutaneous inflammation. J. Invest. Dermatol. 125: 969–976. determinant for CD18-dependent human neutrophil adherence and transendothe- 44. Chavakis, T., T. Keiper, R. Matz-Westphal, K. Hersemeyer, U. J. Sachs, lial migration. J. Clin. Invest. 82: 1746–1756. P. P. Nawroth, K. T. Preissner, and S. Santoso. 2004. The junctional adhesion 37. Muller, W. A. 2001. Migration of leukocytes across endothelial junctions: Some molecule-C promotes neutrophil transendothelial migration in vitro and in vivo. concepts and controversies. Microcirculation 8: 181–193. J. Biol. Chem. 279: 55602–55608. 38. Kitayama, J., A. Hidemura, H. Saito, and H. Nagawa. 2000. Shear stress affects 45. Johnson-Leger, C., M. Aurrand-Lions, N. Beltraminelli, N. Fasel, and migration behavior of polymorphonuclear cells arrested on endothelium. Cell. B. A. Imhof. 2002. Junctional adhesion molecule-2 (JAM-2) promotes lymphocyte transendothelial migration. Blood 100: 2479–2486. Immunol. 203: 39–46. 46. Reymond, N., A. M. Imbert, E. Devilard, S. Fabre, C. Chabannon, L. Xerri, 39. Cuvelier, S. L., and K. D. Patel. 2001. Shear-dependent eosinophil transmigration C. Farnarier, C. Cantoni, C. Bottino, A. Moretta, et al. 2004. DNAM-1 and PVR on interleukin 4-stimulated endothelial cells: a role for endothelium-associated regulate monocyte migration through endothelial junctions. J. Exp. Med. 199: eotaxin-3. J. Exp. Med. 194: 1699–1709. 1331–1341. 40. Cinamon, G., V. Shinder, and R. Alon. 2001. Shear forces promote lymphocyte 47. Bixel, G., S. Kloep, S. Butz, B. Petri, B. Engelhardt, and D. Vestweber. 2004. migration across vascular endothelium bearing apical chemokines. Nat. Immunol. Mouse CD99 participates in T cell recruitment into inflamed skin. Blood 104: 2: 515–522. 3205–3213. Downloaded from http://www.jimmunol.org/ by guest on September 30, 2021