(A) A representative image of a venular convergence and corresponding straight region (as defined in METHODS section), which were lumenally perfused with fluorescently tagged anti-PECAM-1 antibody (10μg/ml) to visualize EC arrangement. Outlined in white are the regions (6000μm²) where the number of tri-cellular junctions and the total junctional length were quantified. All analyses were performed on real time recordings of each vessel where the focal plane was brought up or down as necessary, in order to focus on the specific, analyzed part of the vessel wall. The squares within the projections outline the tri-cellular junctions to demonstrate the apparent higher number of these junctions in venular convergences. (B and C) Quantification of tri-cellular junctions in straight and converging/diverging regions of sampled venules (B) and arterioles (C) respectively. For all groups bars are mean+SE, n=3-4 mice, 11 straight and converging regions.* Significantly different from each other (p<0.05), & Significantly different from same region in venules (Fig 2B, p<0.05).

EC junctions and interacting leukocytes were immunofluorescently labeled using anti-PECAM-1 conjugated to Alexa 488 (10μg/ml, local perfusion) and anti-CD11a conjugated to Alexa 488 (3μg/ml, i.v). Time lapsed microscopy (×90) was used to track crawling leukocytes in straight and in converging regions. (A) The fraction of adhered leukocytes that crawled in each region was not different. (B,C) Leukocyte displacement (B), and the mean velocity (C, presented as frequency distribution plot) were less in convergences compared to straight regions. (D) Representative leukocyte crawling trajectories (n=11) from two straight (upper panel) and two converging (lower panel) venular regions. All starting positions were aligned to the same origin. Black arrow indicates flow direction. Axes, distance in micrometers. (E) The direction of leukocyte crawling with respect to blood flow. Directions were defined as follows: Parallel to blood flow (within ± 45 degrees (parallel)); perpendicular to blood flow (between ± 45 and 90 degrees (perpend)); past the origin in the negative direction of the blood flow (against). For all groups in panels A-C and E, n=188 leukocytes in 9 straight and n=148 leukocytes in 7 converging regions in 5 mice were tracked. Bars are mean+SE. * Significantly different from each other (p<0.05).

In separate experiments ICAM-1 distribution and leukocyte crawling velocities near EC-junctions were quantified. (A,B) Venular ECs were stained for ICAM-1 using anti-ICAM-1 (YN/1.7.4, 50μg/ml, local perfusion) and fluorescently tagged secondary (anti-rat Alexa 488, 50μg/ml, local perfusion) antibodies in sequence. (A) Representative raw and pseudocolored images (on the right) depict ICAM-1 enrichment near tri-cellular junctions post fMLP superfusion (second from the top) and following 4 hours TNFα activation (third from the top), but not under control conditions (upper images). Corresponding ICAM-1 fluorescence intensity profiles (on the left) were obtained along the projected white line shown on the raw images. The lines were placed on the ECs that had the best focus in the field of view, and that encompassed one or more tri-cellular junctions. White arrows on the images of venules from fMLP and TNFα activated tissue indicate ICAM-1 enriched regions near tri-cellular junctions. The broken line on each intensity plot is the mean relative intensity, the patterned areas indicate the width of the junction and the gray regions represent 5μm wide regions extending on both sides of the outlined junction. Similarly, selected venules were stained for P-selectin (in green, RB40.34, 50μg/ml followed by anti-rat Alexa 488, 50μg/ml) and PECAM-1 (in red, ER-MP12-PE, Ebioscience, 10μg/ml) after fMLP superfusion. Bottom panel (Fig 6a) depicts a representative image and a corresponding intensity plot (out of n=28 cells), where the relative, corresponding intensities of junctional PECAM-1 (red) and surface P-selectin (green) were obtained along the projected white line. Unlike ICAM-1, P-selectin is not enriched near Ec-junctions. (B) Measurements of relative ICAM-1 expression (as a function of relative fluorescence intensity) near tri-cellular junctions (tri-junc), bi-cellular junctions (bi-junc) and random regions on EC-surface away from EC-junctions (EC-surface) were performed as described in Methods. Intensities that were 2 SD higher than the mean intensity of randomly measured ROIs on EC surface (above the dotted line) were defined as enriched regions. The insert shows percent of enriched tri- and bi-cellular junctions out of the total number of sampled junctions (n=58 tri- and 55 bi-cellular junctions, 5 venules from 3 mice). Approximately 43% of all tri-cellular junctions and only 23% of all bi-cellular junctions were enriched in ICAM-1. (C-E) EC-junctions and interacting leukocytes were immunofluorescently labeled using anti-PECAM-1 conjugated to Alexa 488 (10μg/ml, local perfusion) and anti-CD11a conjugated to Alexa 488 (3μg/ml, i.v). Time lapsed microscopy (×90) was used to track crawling leukocytes in straight regions and venular convergences 50 minutes post fMLP superfusion. (C) Leukocyte crawling velocities over tri-cellular junctions (tri-junc), bi-cellular junctions (bi-junc) including the 5μm wide region on both sides of the junction were measured and compared to crawling velocities on the non-junctional regions (non-junc). (D) In the bottom panel, crawling velocity of a representative leukocyte was tracked in consecutive 5-μm lengths of EC surface while the leukocyte remained in focus. The trajectory of the crawling leukocyte was traced (white dashed line, E, upper left panel) and the intensity plot along that line (left y-axis) was superimposed on the velocity plot (right y-axis). The patterned areas on the intensity plot show the locations of the two tri-cellular junctions that the leukocyte encountered on this path and the gray areas are regions of 5-μm width on both sides of the outlined junction; these are regions of the lowest leukocyte crawling velocity. (D) upper panel shows average crawling velocity of all tracked cells (n=7, 5 mice) in each of the regions shown in the representative trace (bottom panel). Crawling velocities over tri-cellular junctions were significantly slower compared to bi-cellular junctions or non-junctional regions, consistent with the ICAM-1 enrichment shown in panel B. (E) Image sequence demonstrating the initial location of a representative leukocyte and its crawling route during 863 seconds. The white arrows track the leukocyte displacement. Long arrow above the image indicates the direction of flow. The bar is 10μm. **/ˆ Significantly different from each other (p<0.01).

The intact tissue was stimulated by superfusion of fMLP (10μM, for 10 minutes). All data were collected 40 minutes after the completion of fMLP application. Leukocyte rolling (A) adhesion (B) and TEM (C) were quantified in venular convergences (regions 1 and 2) and compared to straight regions (region 3). For adhesion and TEM (panels B and C) each data point in region 1 represents an average of leukocytes on both sides of the inflow venule (regions 1 and 1*, Fig1C). Regions 1, 1*, 2 and 3 were defined as described in the Methods section. All TEM counts were normalized to 10000μm². (D) Image of a representative venular convergence, where leukocyte adhesion and TEM (white and black dotted lines respectively) were quantified in regions 1, 1* and 2. To quantify leukocyte adhesion and TEM in the straight region (region 3) the field of view had to be moved away from the convergence, thus it is not illustrated. The white arrows show the direction of flow. (E) Sequence of images over 20 minutes showing leukocyte TEM in the converging region (white ellipse) vs straight venular region (white rectangle). The bar is 25μm. (F) Time lapsed images were summed to display the origin and path of migrating leukocytes. Both leukocyte adhesion and leukocyte TEM are significantly enhanced in the regions of venular convergences. For all groups bars are mean+SE, n=4 mice, 8 venules. ** Significantly different from other groups (p<0.01)

To visualize EC-junctions and circulating leukocytes, venules were locally perfused with anti-PECAM-1 conjugated to Alexa 488 (10μg/ml) followed by intravenous injection of anti-CD11a conjugated to Alexa 488 (3μg/ml) antibodies. (A) The number of firmly adhered leukocytes at bi-cellular (bi) and tri-cellular (tri) junctions, as well as on the non-junctinal regions (non) were quantified. The number of adhered leukocytes at each location is presented as % of the total population. (B) Simulated leukocytes (circles with 8μm diameter) were positioned using randomly generated coordinates within the images of PECAM-1 stained venules that were previously obtained and used to quantify leukocyte adhesion as shown in (A). Similarly to (A), the number of leukocytes that was found at each of the junctional locations was quantified. For all groups bars are mean+SE, n=4 mice, 11 straight and converging regions. ** Significantly different from each other (p<0.01). & Significantly different from others in same group (p<0.01). (C) For illustrative purposes the endothelium of a blood perfused venule was stained for PECAM-1 (anti-PECAM-1 conjugated to Alexa 488, 10μg/ml) and leukocytes were stained for CD11a (Phycoerithrine anti-CD11a, 3μg/mouse, red). Representative images (which are z-stack projections of multiple focal planes) depict adhered leukocytes in straight and converging venular regions. The zoom-in images are a single slice of a Z-stack, and show firmly adhered leukocytes at the non-junctional (arrow head), tri-cellular (short arrow) and bi-cellular junctional regions, thus the colocalizing yellow signal. Due to the higher abundance of tri-cellular junctions in convergences (compared to straight regions), most adhered leukocytes in these regions are found on the tri-cellular junctions.

EC-junctions and interacting leukocytes were immunofluorescently labeled using anti-PECAM-1 conjugated to Alexa 488 (10μg/ml, local perfusion) and anti-CD11a conjugated to Alexa 488 (3μg/ml, i.v). Time lapsed microscopy (×90) was used to quantify the number of transmigrating leukocytes at each location. (A) % of total leukocytes that underwent TEM at the junctional region (junc) or via transcellular route (non-junc). When the route could not be clearly assigned to either pathway it was assigned to “un-assg”. (B) The fraction of adhered leukocytes that underwent TEM was determined in straight and converging regions. (C) Leukocyte TEM at bi-cellular (bi) and tri-cellular (tri) junctions, as well as at non-junctional (non) regions as percentage of the total leukocytes that underwent TEM in straight and converging regions. In panels A-C white bars represent straight and black bars represent converging regions. For all groups in panels A-C n=188 leukocytes were tracked in 9 straight and 148 leukocytes in 7 converging regions, in 5 mice. Bars are mean+SE. ***/ˆ Significantly different from each other or others in same group (p<0.001), ** Significantly different from each other (p<0.01). (D) Representative image sequence demonstrating leukocyte initial adhesion location, the crawling route, and TEM during 125 sec time period. The white arrows track leukocyte displacement. The black arrow above the images indicates the flow direction. The bar is 12μm.

(A-C) VE-Cadherin distribution. (A) Venules were stained for VE-Cadherin (anti-VE-Cadherin, BV13, 30μg/ml local intraluminal perfusion). The formation of gaps in VE-Cadherin per field of view was quantified with or without fMLP (after 50 minutes, 10μM, superfusion) and ICAM-1 tail peptide (100μg/ml, local cannulation) as indicated. All measurements were normalized to 1000 μm². (B) The location of the formed gaps in VE-Cadherin was quantified with respect to EC-junctions in the presence or absence of ICAM-1 tail peptide. Formation of “gaps” in VE-Cadherin was significantly attenuated in the presence of ICAM-1 tail peptide, but did not alter the distribution of those that formed, as the majority (∼70%) remained at tri-cellular junctions. (C) EC-junctions stained for VE-Cadherin were observed immediately upon completion of the cannulation protocol (a representative image is shown in the upper panel, control conditions) and then again 50 minutes post fMLP application (a representative image is shown in the middle panel). Control images of VE-Cadherin and 50 minutes post fMLP application in the presence of either ICAM-1 tail peptide (a representative image is shown in the bottom panel) or control-penetratin peptide (not shown) were collected. As shown by white arrows (middle panel), fMLP application to the tissue resulted in formation of a large number of gaps in VE-Cadherin. When used, the penetratin-peptides were added immediately following control data collection, but prior to fMLP superfusion. For all groups in panels A-C n=4 mice, 7 venules for each condition. To quantify leukocyte-EC interactions (panels D- H) EC junctions and leukocytes were immunofluorescently labeled using anti-PECAM-1 conjugated to Alexa 488 (10μg/ml, local perfusion) and anti-CD11a conjugated to Alexa 488 (3μg/ml, i.v). All observations were made 50 minutes post fMLP application (10μM, superfusion) in the presence of either ICAM-1 blocking antibody (YN/1.7.4, 50 μm/ml), ICAM-1 tail peptide (100μg/ml), or control peptide (100μg/ml), which were introduced intraluminally by local cannulation. (D) Leukocyte adhesion (leukocytes/80μm vessel length), crawling (fraction of total adhered) and TEM (leukocytes/10000 μm² extravascular tissue). n=4 mice, 7 venules. The residual fraction of crawling leukocytes in ICAM-1 Ab blocking and ICAM-1 tail peptide experiments was quantified for directionality (E), distance (F) and crawling velocity (G). In F and G the black arrows indicate the average crawling distance and crawling velocity (respectively) as shown in Figure 4 in the absence of peptide treatment. ** Significantly different from each other or others in the same group (p<0.001). & Significantly different from values shown in figure 4 in the absence of peptide treatment (p<0.05). Control peptide did not prevent the formation of gaps in VE-Cadherin and did not decrease leukocyte TEM (not shown), suggesting role for ICAM-1 signaling in leukocyte TEM. The bar in panel C is 10μm.