All mice were bred under pathogen-free conditions in the animal facility of the Max Planck Institute for Molecular Biomedicine and were of the C57BL/6 genetic background. All experiments were carried out as approved by the local state authorities [Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen (LANUV)].

In order to generate an endothelial-specific conditional Csk knock-out mouse line, Csk^lox/lox mice (36) were mated with transgenic mice expressing a tamoxifen-inducible form of the Cre recombinase in vascular endothelial cells using a phage artificial chromosome (PAC) containing the Pdgfb gene (Pdgfb-iCreER mice) (37). The following primers were used for genotyping: CSK forward (5′-TGG GGT TGA ATG GTA TGA ACA CTC-3′) and CSK reverse (5′-TGC CAT GTG GAG AAG AGA ATC AGC-3′). This generated a 500-base-pair PCR product for wild-type (WT) mice and a 600-base-pair product for mice carrying the loxP sites (Csk^lox).

These mice were further mated with cortactin knock-out mice (17) or with VE-cadherin-Y685F mice (26), both previously described by our group.

Blood vessel permeability in the skin was analyzed by a modified Miles assay as previously described (38). For each assay, four to five 8- to 12-week-old Csk^lox/lox or Csk^iECKO mice were i.p. injected for 5 consecutive days with 2 mg tamoxifen (#T5648-5G, Sigma-Aldrich, St. Louis, MO, USA) in sunflower oil. Mice were weighed, and their dorsal skin was shaved. Evans blue dye (Sigma-Aldrich) was injected into the tail vein [10 µL/1 g body weight of a 1% solution in phosphate-buffered saline (PBS)]. A total of 15 min later, PBS, 90 ng mVEGF, or 175 ng histamine was injected intradermally into the dorsal skin. Another 30 mins later, the mice were sacrificed, skin areas were excised, and leaked dye was extracted with formamide for 5 days. The optical densities of the samples were measured at 620 nm using a spectrophotometer (UV-1900i; Shimadzu, Kyoto, Japan).

Mice at 11–20 weeks of age were used. General anesthesia was administered using ketamine (125 mg/kg body weight) and xylazine (12.5 mg/kg body weight), and mice were closely monitored during anesthesia. Surgical preparation of the cremaster muscle and intravital microscopy were performed as previously described (26, 38, 39). Mice received an intrascrotal injection of 50 ng of recombinant IL-1β (Biomol, Hamburg, Germany) in 150 µL PBS. After 4 h, the cremaster muscle was prepared, and intravital images were recorded using an upright Axio Examiner microscope (Axiocam MRM, 40x Plan Neofluar, NA 1.30, Carl Zeiss, Oberkochen, Germany). For each animal, three to eight single unbranched postcapillary venules with a diameter of 20 to 30 µm were analyzed. Blood flow centerline velocity was measured using a dual-photodiode sensor system (Circusoft, Hockessin, DE, USA).

Recorded images and videos were analyzed using ImageJ. The leukocyte rolling flux fraction was determined as a percentage of total leukocyte flux. Leukocyte rolling velocity was determined in each vessel segment, and the total number of adherent leukocytes was analyzed for each vessel segment (100 µm) and is depicted per 10⁴ µm² vessel surface area. Transmigrated cells were counted in an area extending 75 µm outside of each vessel over a distance of 100 µm vessel length (corresponding to 1.5 × 10⁴ µm² tissue area). Newtonian wall shear rate (s⁻¹) and mean blood velocity were determined as previously described (38).

Human umbilical vein endothelial cells (HUVECs) were isolated as described (40) and cultured in an EBM-2 medium with SigleQuot supplements (Lonza, Basel, Switzerland) and used for experiments between passages 3 and 6 (Ethics Committee of Münster University Clinic, approval 2009-537-f-S). Primary endothelial cells from the lungs of Csk^iECKO mice were isolated and cultured as described (17). For Csk knock-out induction, murine primary endothelial cells were treated for 5 consecutive days with 1 µM (Z)-4-hydroxytamoxifen (#sc-3542A, Santa Cruz Biotechnology, Dallas, TX, USA) or equivalent amounts of solvent (ethanol) in the medium. Murine polymorphonuclear neutrophils (PMNs) were isolated from bone marrow by density gradient centrifugation using Histopaque 1077 and 1119 (Sigma-Aldrich). The granulocyte-containing phase was collected and washed twice in Hanks balanced salt solution, 25 mM HEPES (pH 7.3), and 10% fetal calf serum. PMNs were cultured overnight in Dulbecco’s modified Eagle medium, 20% fetal calf serum, 1% glutamine, 1% penicillin/streptomycin, and 10% WEHI-3B culture supernatant. Before experiments, human umbilical vein endothelial cells (HUVECs) and murine primary endothelial cells were treated with TNF-α (recombinant human TNF-α #300-01A, PeproTech, Cranbury, NJ, USA; recombinant murine TNF-α #315-01A, PeproTech).

For the knock-down of Csk and VE-cadherin in HUVECs siRNAs, the following sequences were used: 5′-TACGCGCCTCATTAAACCAAA-3′ (target sequence for Csk siRNA, Hs_CSK_3, Qiagen, Valencia, CA, USA) and 5′-GGUUUUUGCAUAAUAAGCtt-3′ (sense strand for CDH5 siRNA, 10696, Ambion, Austin, TX, USA). Routinely, dishes with 70% confluent HUVECs were transfected with 40 nM (Csk) or 20 nM (VE-cadherin) of siRNA using Lipofectamine RNAiMAX (#13778075, Invitrogen, Carlsbad, CA, USA), according to the manufacturer’s instructions.

For in vitro RNA interference with cortactin in murine endothelial primary cells, the following sequences were used: 5′-GAGAUGUGCUAGUGGCUUATT-3′ (sense strand for cortactin siRNA, Mm_Cttn_5, Qiagen) and 5′-CCAACAUAGAAAUGAUUGATT-3′ (sense strand for cortactin siRNA, Mm_Cttn_7, Qiagen). Primary endothelial cells at 100% confluence were transfected with 48 nM siRNA per reaction by nucleofection according to the manufacturer’s instructions (Amaxa Biosystems, Cologne, Germany).

As a negative control, an siRNA not targeting any known mammalian genes was used (AllStars negative control, 1027281, Qiagen). Experiments were carried out 72 h after transfection.

To investigate the role of tyrosine 685 of VE-cadherin, HUVECs were transfected with either a VE-cadherin-WT-EGFP or a VE-cadherin-Y685F-EGFP construct as described previously (26). Cells were incubated for 24 h and subsequently used in the experiments.

To analyze the transendothelial migration of murine PMNs across primary endothelial cell monolayers, 2 × 10⁴ cells were seeded on fibronectin-coated Transwell filters (6.5 mm, 5 µm pore size, Corning, New York, NY, USA), grown to confluence, and stimulated with 5 nM murine TNF-α for 4 h before the experiment; 2 × 10⁴ murine PMNs were allowed to transmigrate towards a 40 ng/mL CXCL-1 chemokine gradient (recombinant mouse CXCL-1/KC aa 29-96, #1395-KC-025, R&D Systems, Minneapolis, MN, USA) for 40 min. Transmigrated PMN numbers were determined using a CASY Cell Counter TT+ (Roche, Basel, Switzerland).

Murine primary endothelial cells were grown to confluence in 96-well plates coated with 100 μg/ml fibronectin and were subsequently stimulated with 5 nM TNF-α for 4 h. PMNs that were isolated 1 day prior were stained with Vybrant DiD dye (V22887, Invitrogen) according to the manufacturer’s instructions; 1 × 10⁵ PMNs in 200 µL medium [Hanks’ Balanced Salt Solution (HBSS), 25 mM HEPES (pH 7.0)] were added to each well and were allowed to adhere for 30 min at 37°C in 5% CO₂. Adherent cells that did not firmly interact with the endothelial cell monolayer were removed by washing three to five times with pre-warmed PBS. The remaining PMNs were subsequently fixed with 4% paraformaldehyde (PFA), and images of each well were taken using an Axiovert 200M microscope (20x Plan Apochromat, NA 0.80, Carl Zeiss, Oberkochen, Germany). The number of adherent cells was determined using an ImageJ plug-in.

Polyclonal rabbit antibodies VE42 against VE-cadherin (41) and C5 against VE-cadherin (42) and monoclonal antibodies mp685 against phosphorylated tyrosine 685 of murine VE-cadherin and mp731 against phosphorylated tyrosine 731 of VE-cadherin (26) and YN1/1.7 against ICAM-1 (43) have been described previously.

The following monoclonal antibodies were purchased: antibodies against phosphotyrosine (4G10, #05-321, Millipore, Billerica, MA, USA), antibodies against Csk (52, #610080, BD Biosciences, San Jose, CA, USA), antibodies against Src (GD11, #05-184, Millipore; or 327, #ab16885, Abcam, Cambridge, UK), antibodies against cortactin (4F11, #05-180, Millipore), antibodies against VE-cadherin (F-8, #sc-9989, Santa Cruz Biotechnology; or 75/Cadherin-5, #610251, BD Biosciences), antibodies against α-actin (B-12, #sc-166524, Santa Cruz Biotechnology), and antibodies against α-tubulin (B-5-1-2, #T6074, Sigma).

Furthermore, the following polyclonal antibodies were purchased: antibodies against Csk (H-75, #sc-1307, Santa Cruz Biotechnology), antibodies against the phospho-Src family (Tyr416) (#2101, Cell Signaling, Danvers, MA, USA), antibodies against phospho-SRC (Tyr529) (#44662G, Thermo Fisher, Waltham, MA, USA), antibodies against Src (SRC2, #sc-18, Santa Cruz Biotechnology), antibodies against phospho-Cortactin (Tyr421) (#C0739, Sigma), antibodies against phospho-Cortactin (Tyr421) (#4569, Cell Signaling), antibodies against VE-cadherin (C19, #sc-6458, Santa Cruz Biotechnology), and antibodies against GFP (#ab6673, Abcam).

Selective labeling of F-actin in immunostainings was achieved with phalloidin directly coupled to Alexa Fluor 647 (Invitrogen).

Horseradish peroxidase-coupled secondary antibodies were obtained from Dianova (Geneva, Switzerland), and IRDye680- and IRDye800-coupled secondary antibodies were purchased from LI-COR (Lincoln, NE, USA). Goat anti-mouse Alexa Fluor-coupled antibodies were acquired from Invitrogen.

Cells were cultured until they reached confluency on 8-well µ-slides (ibidi, Gräfelfing, Germany) as described in the “Cell culture” section. They were pretreated for 30 min with 1 mM Na₃VO₄, washed twice with PBS-MC, and fixed with 4% PFA for 10 min followed by permeabilization with 0.5% Triton X-100 for 5 min. Blocking was carried out for 1 h at room temperature (RT) with 3% bovine serum albumin (BSA), and antibody staining processes were performed for 1 h at RT in blocking buffer. Stained cells were either kept in PBS with 0.04% NaN₃ or covered in a fluorescence mounting medium (Dako Omnis, Agilent Technologies, Santa Clara, CA, USA).

Z-stack images were acquired using a confocal laser scanning microscope (LSM 880, 63x Plan-Apochromat, NA 1.40, Carl Zeiss). ImageJ was used for the quantification of the phospho-Src signal. Briefly, the cadherin signal was used to create a mask of the endothelial cell junctions. The phospho-Src signal at the junctions was then calculated as a percentage of the total phospho-Src signal in the image.

ICAM-1 recruitment to transmigrating neutrophils was determined largely as described (17). Murine leukocytes were added to TNF-α-activated confluent primary endothelial cell monolayers. To this end, 2 × 10⁵ PMNs were added to each well of an 8-well µ-slide (ibidi, Gräfelfing, Germany) and allowed to transmigrate for 15 min. Wells were subsequently washed two times to remove unbound/non-transmigrating leukocytes and fixed with 4% PFA for 15 min. Further staining was performed corresponding to the abovementioned protocol for immunofluorescence staining processes of cells. Z-stack images were acquired using a confocal laser scanning microscope (LSM 780, 20x Plan-Apochromat, NA 0.8, Carl Zeiss). The mean fluorescence intensity of the ring-like ICAM-1-rich structures was determined in one single z-slice depicting the largest expansion of the ring using an ImageJ Macro. Values are depicted relative to the ones observed in control endothelial cells expressing cortactin (cells treated with solvent instead of tamoxifen).

Cells were lysed in lysis buffer (10 mM NaPi, 150 mM NaCl, 1% NP-40, 2 mM EDTA, 1 mM Na₃VO₄, 1x cOmplete EDTA-free Proteinase Inhibitor Cocktail, Roche, Basel, Switzerland). In order to detect phosphorylation signals after immunoprecipitations, cells were lysed in 20 mM Tris–HCl (pH 7.4), 150 mM NaCl, 2 mM CaCl₂, 1.5 mM MgCl₂, 1% Triton X-100, 0.04% NaN₃, 1 mM Na₃VO₄, and 1x cOmplete. Lysates were centrifuged at 20,000 g at 4°C for 30 min before aliquots for immunoblot analysis were taken.

Murine lungs were homogenized using an ULTRA Turrax (IKA-Werke, Staufen, Germany) in radioimmunoprecipitation assay buffer containing 10 mM NaPi, 150 mM NaCl, 1% NP-40, 2 mM EDTA, 0.1% sodium dodecyl sulfate (SDS), 1% sodium deoxycholate, 1 mM Na₃VO₄, and 2x cOmplete. Tissue was lysed for 1 h at 4°C and centrifuged at 20,000 g at 4°C for 30 min before aliquots for immunoblot analysis were taken.

For immunoprecipitation of VE-cadherin, lysates were pre-cleared for 2 h at 4°C with protein A-Sepharose beads coated with an IgG control antibody. Subsequently, the immunoprecipitation was carried out for 2 h at 4°C with protein A-Sepharose beads coated with 4 µg (cell lysates) or 5 µg (lung lysates) of VE-cadherin C5 antibody. Immunocomplexes were washed five times with lysis buffer and analyzed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE).

Total cell or organ lysates were separated using SDS–PAGE in 8% gels and immunoprecipitated material in 10%–12% gels. Proteins were then transferred to a nitrocellulose membrane (Schleicher & Schuell, Keene, NH, USA) by wet blotting. For the detection of phosphorylated tyrosine residues, a blocking buffer containing 2% BSA and 200 µM Na₃VO₄ was used. Immunoblot signals were quantified using ImageJ or ImageStudio (LI-COR).

Statistical significance was analyzed using a Mann-Whitney test, unpaired t-test, one-way analysis of variance (ANOVA) or two-way ANOVA for independent samples. The ROUT method was used to identify outliers, with a maximum false discovery rate of 0.1%. The GraphPad Prism 10 software was used for this analysis. Data are shown as mean ± standard error of the mean (SEM). P-values below 0.05 (*), 0.01 (**), 0.001 (***), and 0.0001 (****) were considered statistically significant.

In order to investigate the role of Csk in the control of endothelial barrier function, we generated Csk inducible endothelial-specific KO mice (Csk^iECKO) by breeding Csk^lox/lox mice (36) with PDGFb-CreERT2 mice (37) ( Supplementary Figure 1A ). First, we analyzed the effect of Csk gene inactivation on the activation of SFKs. To this end, primary lung endothelial cells were isolated from Csk^iECKO mice and treated in culture with and without tamoxifen. As shown in Figures 1A, B , Csk expression was strongly reduced by 70% (± 6%) upon tamoxifen treatment. This was accompanied by a strong decrease in the phosphorylation of the inactivating tyrosine 529 residue of SFKs and a strong increase in the activating pY418 signal ( Figures 1C, D ). In agreement with this, phosphorylation of Y685 of VE-cadherin was also clearly increased, which is known to be an SFK target ( Figures 1E, F ). In contrast, no effect was seen for the phosphorylation level of Y731 ( Figures 1G, H ), which we have shown before to be constitutively phosphorylated. Dephosphorylation of this residue is involved in the leukocyte diapedesis step (26).

This result was verified by analyzing lung lysates prepared from Csk^lox/lox and Csk^iECKO mice after treating the mice daily for 5 days with tamoxifen. As shown by immunoblotting ( Figure 1I ), Csk expression was strongly reduced by 70% (± 19%) ( Figure 1J ). In agreement with our in vitro results, the phosphorylation level of Y685 of VE-cadherin was clearly increased ( Figures 1K, L ), while the phosphorylation level of Y731 of VE-cadherin was unaffected ( Figures 1K, M ).

Next, we analyzed whether Csk gene inactivation would influence vascular permeability ( Figure 2 ). To this end, we performed Miles assays with Csk^lox/lox and Csk^iECKO mice. We found that local intradermal injections of histamine or VEGF in mice that had been i.v. injected with Evans blue induced vascular leaks in Csk^lox/lox to the same extent as in Csk^iECKO mice ( Figure 2 ). Furthermore, the baseline levels of vascular permeability as we determined them upon intradermal injection of PBS were also similar in mice of both genotypes ( Figure 2 ). Thus, endothelial-specific gene inactivation of Csk neither increases baseline levels of vascular permeability nor augments the effects of histamine or VEGF on vascular leak formation. At first glance, this may appear surprising since an increase in VE-cadherin-Y685 phosphorylation is known to correspond to an increase in vascular leakage. However, since it has been recently shown that activation of Src and Yes has opposite effects on endothelial barrier integrity with respect to plasma leaks (29), inactivation of Csk may cause opposite and compensating effects on junction integrity via these two SFK family members.

The lack of an effect of Csk deficiency on vascular permeability prompted us to test whether Csk is relevant for leukocyte extravasation in vivo. To this end, we stimulated tamoxifen-treated Csk^lox/lox and Csk^iECKO mice i.s. with IL-1β for 4 h before preparing the cremaster muscle for intravital microscopy and analyzing neutrophil adhesion and extravasation in postcapillary venules. We found that while neutrophil rolling velocity was decreased by 14% (± 4%), neutrophil adhesion increased significantly by 20% (± 6%) and extravasation by 20% (± 4.8%) in Csk^iECKO mice when compared to Csk^lox/lox mice ( Figures 3A–C ; illustratory videos: Supplementary Videos 1 , 2 ). Parameters such as the number of rolling neutrophils, rolling flux fraction, and hemodynamic parameters were not affected by Csk gene inactivation ( Supplementary Figures 1B–H ).

Next, we decided to verify these effects in in vitro transmigration assays. We isolated primary lung endothelial cells from Csk^iECKO and treated them in culture for 5 days with either tamoxifen or solvent alone. After stimulation with TNF-α for 4 h, we analyzed the adhesion of mouse neutrophils. We found that Csk gene inactivation caused an 18% (± 4.6%) increase in adherent neutrophils ( Figure 3D ). In separate assays, we cultured lung endothelial cells (ECs) from Csk^iECKO on Transwell filters, and we analyzed neutrophil transmigration toward a gradient of CXCL1. Again, we found that tamoxifen-induced Csk gene inactivation led to a 26% (± 2.6%) increase in neutrophil transmigration, reproducing our in vivo results ( Figure 3E ).

We have previously shown that Csk binds to the phosphorylated Y685 of VE-cadherin (35). Interestingly, knock-in mice expressing the point-mutated VE-cadherin-Y685F instead of WT VE-cadherin showed a similar slight but significant increase in neutrophil extravasation (26) as we show here for the Csk^iECKO mice. This raised the question of whether Y685 of VE-cadherin might be relevant for this effect. To answer this, we analyzed the transmigration of mouse neutrophils through primary isolated lung endothelial cells from VE-cadherin-Y685F/Csk^iECKO mice. We treated the endothelial cells prior to the assays with either solvent or tamoxifen. We found that even when Csk expression was unaffected, the Y685F mutation of VE-cadherin was sufficient to increase neutrophil transmigration efficiency by 21% (± 5.6%) ( Figure 3E ). Csk gene inactivation in the double-mutant endothelial cells had no significant additive, increasing effect on transmigration [28% (± 7.4%)] ( Figure 3E ). The efficiency of in vitro Csk gene inactivation was verified by immunoblotting endothelial cell lysates for Csk ( Figure 3F ). Collectively, these results suggest that Csk dampens neutrophil transmigration in a way that is linked to the presence of Y685 of VE-cadherin.

Next, we tested whether Y685 of VE-cadherin is relevant for the control of SFK activation by Csk. Since transduction efficiency was low in primary mouse endothelial cells, we switched to HUVECs for these experiments. We treated HUVECs with siRNA against VE-cadherin (targeting a site in the upstream coding region) and siRNA targeting Csk or ctrl siRNA. Subsequently (48 h later), we transduced cells with either VE-cadherin-WT-EGFP or VE-cadherin-Y685F-EGFP. As revealed by immunoblotting of cell lysates, silencing of Csk led to reduced phosphorylation of the inactivating Y529, whereas signals for the activating pY418 were increased. These effects were identical in HUVECs expressing WT or mutant VE-cadherin. Simply expressing VE-cadherin-Y685F in the presence of endogenous Csk had no significant effect on SFK activation as analyzed from cell lysates ( Figure 4A ). The quantification of the pY418 and pY529 signals are shown in Figures 4B, C . This argues for a general control function of Csk for the activation of total cellular SFKs, which goes beyond the relevance of VE-cadherin-pY685 as an anchor for Csk.

Then, the relevance of VE-cadherin-pY685 for the control of SFK activation by Csk at endothelial junctions was tested. HUVECs were silenced for endogenous VE-cadherin and treated with either Csk siRNA or ctrl siRNA, followed by transduction with VE-cadherin-WT-EGFP or VE-cadherin-Y685F-EGFP, as described above. Cells were analyzed by staining for SFK-pY418, EGFP, and actin. As shown in Figure 4D , silencing Csk increased the level of activated SFK at endothelial junctions. This was also achieved simply by expressing VE-cadherin-Y685F, even when endogenous Csk was still expressed ( Figure 4D ). The level of activated SFK was not further increased by Csk silencing in cells that expressed VE-cadherin-Y685F ( Figure 4D ). The quantification of junctional SFK-pY418 is shown in Figure 4E . These results suggest that VE-cadherin-Y685 is relevant for the regulation of SFK activity at junctions by Csk.

Cortactin is one of the first identified Src substrates. We found previously that it is required in endothelial cells for neutrophil extravasation (17). Therefore, we tested whether cortactin is involved in the mechanism by which Csk controls neutrophil diapedesis. First, we analyzed whether silencing of Csk in HUVECs would increase the phosphorylation of an activating tyrosine residue of cortactin and whether the Csk binding site of VE-cadherin would be relevant for this effect. HUVECs were silenced for endogenous VE-cadherin and treated with either Csk siRNA or ctrl siRNA, followed by transduction with WT VE-cadherin-EGFP or VE-cadherin-Y685F-EGFP, as described above. By immunoblotting cell lysates, we analyzed the pY421 levels of cortactin as well as the expression levels of cortactin, Csk, tubulin, and EGFP-fusion protein ( Figure 5A ). The quantification of the Csk blot signals is shown in Figure 5B and that of the pY421 activation epitope of cortactin in Figure 5C . We found that silencing of Csk increased pY421 levels of cortactin. Similarly, the expression of VE-cadherin-Y685F was sufficient to increase cortactin-pY421 levels even when Csk was expressed at endogenous levels. Silencing of Csk in cells expressing the VE-cadherin mutant had no additional stimulating effect on cortactin phosphorylation. These results suggest that Csk limits the activation level of cortactin, and the need for Y685 of VE-cadherin for this effect suggests that Csk recruitment to pY685 of VE-cadherin is involved in the regulation of cortactin activity.

Next, we tested whether the effect of Csk on neutrophil diapedesis depends on the regulation of cortactin activity. To this end, we treated primary lung endothelial cells from Csk^iECKO mice with tamoxifen or solvent for 5 days, followed by siRNA treatment with either two different siRNAs for cortactin or with control siRNA. We verified the efficiency of in vitro Csk gene inactivation and cortactin siRNA treatment by immunoblotting the endothelial cells for Csk ( Figure 6A ). We then analyzed the CXCL1-stimulated migration of mouse neutrophils through TNF-α-activated monolayers of these cells. In this way, we found that Csk gene inactivation increased neutrophil transmigration when the endothelial cells expressed normal cortactin levels; however, this increase was blocked when cortactin expression was silenced ( Figure 6B ). Thus, Csk modulates neutrophil transmigration through a mechanism that requires cortactin.

To verify this in vitro finding in vivo, we analyzed neutrophil extravasation by intravital microscopy of the cremaster muscle of tamoxifen-treated double-mutant cortactin^−/−/Csk^lox/lox and cortactin^−/−/Csk^iECKO mice after i.s. stimulation with IL-1β 4 h before analysis (illustratory videos: Supplementary Videos 3 , 4 ). Hemodynamic and leukocyte rolling parameters are shown in Supplementary Figure 2 and were not affected by gene inactivation of Csk. We found that neutrophil rolling velocity, adhesion, and neutrophil extravasation were not significantly increased by gene inactivation of Csk in cortactin^−/− mice ( Figures 6C–E ). This was in contrast to our results with mice that expressed cortactin ( Figures 3A–C ). Thus, cortactin is required for the effect of Csk on neutrophil extravasation in vivo.

Since we and others have shown before that cortactin supports the recruitment and clustering of ICAM-1 around transmigrating neutrophils (14, 17) and that this requires tyrosine phosphorylation of cortactin (15), we tested whether gene inactivation of Csk would affect ICAM-1 interaction with neutrophils. To this end, we treated primary mouse lung endothelial cells from Csk^iECKO mice with tamoxifen or solvent, exposed them overnight to TNF-α, and allowed mouse PMNs to adhere and transmigrate. We stained fixed cells for ICAM-1, and we evaluated the intensity of ICAM-1 signals surrounding neutrophils. A representative picture of ICAM-1 clusters surrounding neutrophils is shown in Figure 7A . Note that the ICAM-1 structures supported by Csk gene inactivation resembled “docking structures” or “transmigratory cups” as they were described for transmigrating leukocytes before (8, 9). The quantification of 114 ICAM-1 positive ring-like structures for each condition revealed a more than 50% increase in ICAM-1 signal intensity around neutrophils that interacted with endothelial cells gene-inactivated for Csk ( Figure 7B ). In agreement with this, we found that mouse neutrophil transmigration through monolayers of mouse primary lung endothelial cells grown on Transwell filters increased upon Csk gene inactivation. This increase was not detected anymore when ICAM-1 was blocked with an anti-ICAM-1 antibody ( Figure 7C ). Thus, Csk gene inactivation augments ICAM-1 clustering around neutrophils and supports increased neutrophil diapedesis in an ICAM-1-dependent way.