Vav1 Is Essential for Mechanotactic Crawling and Migration of Neutrophils out of the Inflamed Microvasculature¹

Recruitment of circulating leukocytes to the site of inflammation occurs via complex interactions with endothelium; leukocytes tether to, roll along, and firmly adhere to the endothelium before transmigrating out of the vasculature (1, 2). In the case of neutrophils, rolling is dependent on selectins, adhesion occurs mainly through β₂ integrins, and emigration is mediated by integrins as well as PECAM-1, CD99, and junction adhesion molecules (3, 4, 5, 6, 7). Schenkel et al. (8) recently documented another step in the leukocyte recruitment cascade, namely crawling to junctions before transendothelial migration. Using an in vitro system to visualize monocytes, these investigators reported that cells firmly adhered, flattened, and then rapidly sent out pseudopods and crawled via integrins to junctions where they subsequently transmigrated. The crawling appeared to be random but was ultimately necessary for emigration. More recently, a number of laboratories reported that intraluminal crawling also occurred in vivo (9, 10, 11), and this allowed cells to reach optimal emigration sites at endothelial junctions (10). The crawling of neutrophils did not appear to be biased in the direction of flow, suggesting this was not simply a dragging motion due to hydrodynamic force displacement. However, whether the neutrophil crawling was truly a random process or followed an inherent pattern remained unclear, as was the question of whether the crawling was chemotactic, chemokinetic, haptotactic (following chemokine attached to extra cellular matrix), or mechanotactic (driven by mechanical forces).

Neutrophil adhesion and crawling are mediated by two separate molecular mechanisms. Firm adhesion and stabilization (spreading) occurred via α_Lβ₂ integrin (LFA-1), followed immediately by crawling via α_Mβ₂ integrin (Mac-1). Inhibition of LFA-1 prevented adhesion while inhibition of Mac-1 had no effect on adhesion but prevented all subsequent crawling (10). These distinct, molecular steps for LFA-1-induced adhesion followed by Mac-1-induced crawling in vivo would require exquisite communication (e.g., outside in signaling from LFA-1) for these molecules to function in a sequential, coordinated fashion under flow conditions. Moreover, these in vivo observations bestow added physiologic importance to various in vitro studies that demonstrated that cross linking of LFA-1 leads to outside in signaling and cytoskeletal rearrangements (12, 13) as well as Mac-1 activation (14) presumably leading to crawling. However, intracellular signaling molecules that might be important downstream of adhesion to impact on crawling and resultant emigration have not been identified.

Vav1, a guanine exchange factor for the Rho family GTPases Rac and Cdc42, is a major regulator of the organization of the actin cytoskeleton during leukocyte polarization and migration (15). Importantly, Vav1 regulates activation downstream of LFA-1 in all leukocytes tested, including neutrophils, lymphocytes, and NK cells (13, 16, 17, 18). Absence of Vav1 under static conditions in vitro leads to impaired neutrophil spreading, membrane ruffling, cytoskeletal rearrangement, and polarization. However, the role for Vav1 in crawling becomes equivocal under static conditions in vitro. One group has reported no role for Vav1 in crawling on protein coated coverslips (19), another group observed reduced crawling speed for Vav1^−/− cells on plastic (20), whereas a third group reported impaired crawling to some but not other chemoattractants (21). To date, no one has examined the role of Vav1 for crawling under the very dynamic shear-dependent in vivo leukocyte recruitment paradigm. In fact, to our knowledge, only one group has visualized leukocyte behavior in vivo in Vav1-deficient mice by adding a brief chemotactic stimulus and examined neutrophil rolling and adhesion for the subsequent 10 min in real time (19). The data revealed no obvious impairment in rolling or adhesion in this very acute setting in Vav1^−/− neutrophils; however, events downstream of adhesion were not examined.

Using intravital light and multichannel fluorescence spinning disk confocal microscopy in real time and time-lapse, we systematically examined the transition from adhesion to crawling to emigration in wild-type (WT)⁴ and Vav1^−/− mice. A very striking and consistent Mac-1-dependent crawling behavior was noted in WT mice, which was often perpendicular to blood flow but sometimes changed when an endothelial junction was encountered. In vitro, similar perpendicular crawling was observed in the presence of shear and absence of chemokine gradients on protein-coated coverslips, suggesting mechanotaxis rather than chemotaxis for the inherent perpendicular crawling behavior. In striking contrast, Vav1^−/− neutrophils were unable to crawl perpendicular to the direction of blood flow in vitro and in vivo under physiological shear rates. In fact, adherent Vav1^−/− neutrophils appeared to be stretched and dragged in the direction of blood flow. This movement in Vav1^−/− mice was Mac-1 independent. Vav1^−/− cells also failed to realign and follow any junctions that were not aligned in the direction of flow. This resulted in a greater interval of time for adhesion and crawling in Vav1^−/− mice, leading to more detachment and reduced emigration. Finally, these neutrophil recruitment defects translated into significant impairment in recruitment of neutrophils in an infectious disease relevant model of peritonitis.

All procedures were approved by the University of Calgary Animal Care Committee and conformed to Canadian Council for Animal Care guidelines. Male Vav1^+/+ mice (C57BL/6) and Vav1-deficient mice (C57BL/6 background; provided by Dr. J. M. Penninger, Institute of Molecular Biotechnology, Vienna, Austria), were anesthetized with 10 mg · kg⁻¹ xylazine (Bayer) and 200 mg · kg⁻¹ ketamin hydrochloride (Biomeda-MTC) i.p. Additional anesthetic was administered through a jugular vein.

The behavior of leukocytes in the microcirculation and adjacent tissue was studied in the cremaster muscle preparation through an intravital microscope (Optiphot-2; Nikon with a 25/0.35 (E. Leitz) or 40/0.75W (Carl Zeiss MicroImaging objective) connected to a video camera (5100 HS; Panasonic) (22, 23).

The number of rolling, adherent, and emigrated leukocytes in single unbranched venules (25–40 μm in diameter) before and after addition of the chemokine MIP-2 (5 nM) to the superfusate was determined using video playback analysis (22, 23). By using MIP-2, >90% of the emigrated cells were neutrophils (24).

Crawling cells within inflamed vessels (0.5 nM MIP-2 superfusion or 0.5 μg TNF-α in 200 μl of saline injected intrascrotally 4 h before experiments (23)) were tracked using time-lapse video microscopy. For the best resolution, postcapillary venules ranging from 18 to 30 μm were imaged in the cremaster, and the ×40 water immersion objective was used. The distance of cells crawling within the vessel was measured from when they first adhered until they emigrated or detached and crawling velocity was calculated. Centerline RBC velocity (Vrbc) was measured online using an optical Doppler velocimeter (Microcirculation Research Institute, Texas A&M University). Mean RBC velocity (Vmean = Vrbc/1.6) assuming cylindrical geometry was used to calculate venular wall shear rate based on the Newtonian definition γ = 8(Vmean/diameter). For blocking experiments, mAbs against Mac-1 (30 μg per mouse; eBioscience) or LFA-1 (30 μg per mouse; eBioscience) were given i.v. during experiments. Isotype controls have previously been shown not to affect the measured parameters (10).

Endothelial junctions were labeled with monoclonal anti-PECAM-1 (50 μg per mouse; Fitzgerald Industries) conjugated to Alexa Fluor 555 (Molecular Probes), and neutrophils were stained with FITC-labeled GR-1 (40 μg per mouse; eBioscience), a dose previously shown not to affect neutrophil recruitment (10). Images were acquired with an upright microscope (BX51; Olympus) using a ×20/0.95W NA water dipping XLUM Plan F1 objective. The microscope was equipped with a confocal light path (WaveFx; Quorum) based on a modified Yokogawa CSU-10 head (Yokogawa Electric), and 488- or 561-nm laser excitation (Cobalt) was used with the appropriate long-pass filters (Semrock). An EMCCD camera (C9100–13; Hamamatsu) was used for fluorescence detection. Volocity Acquisition software (Improvision) was used to drive the confocal microscope.

Peptidoglycan (PGN) from Staphylococcus aureus (Fluka) was sonicated for 1 h and injected i.p. After 4 or 24 h, mice were anesthetized with Isoflurane (inhaled anesthetic; Bimeda-MTC), and blood was collected via cardiac puncture with a heparinized syringe for circulating leukocyte counts. Next, the peritoneum was lavaged with 3 ml of PBS. Exudate cells were recovered following a 60-s manual massage and counted with a hemacytometer. Leukocyte differentials were determined from cytospins and Wright-Giemsa staining (25).

Blood was collected as described above and centrifuged (500 × g, 4°C, 15 min). The resulting plasma was diluted to 10% in HBSS and stored at 4°C until needed. Mice were then euthanized, and the ends of the femurs and tibias were resected. The bone marrow was removed by perfusion of 5 ml of ice-cold PBS and then suspended through a 20-gauge needle. Marrow cells were pelleted in a centrifuge (250 × g, 4°C, 12 min) and resuspended in 2 ml of PBS. The neutrophils (1.0 × 10⁶ cells ml⁻¹) were isolated over a discontinuous Percoll gradient (stock: 90 ml Percoll, 10 ml of 10× HBSS, diluted to 72, 64, and 52% in PBS) by centrifugation (1100 × g, 4°C, 30 min), resulting in a band of neutrophils between the 72 and 64% layers.

Glass coverslips were coated with murine plasma by first marking an area in the center of the slide 15 × 25 mm in size, with a hydrophobic pen. Two hundred microliters of the 10% murine plasma in HBSS was pipetted into the marked area, and then, the slides were incubated for 4 h at room temperature. The slides were then washed twice with room temperature HBSS, where after 100 μl of isolated murine neutrophils plus 100 μl of fresh HBSS were placed onto the plasma-coated region. The slides were incubated (15 min in a 5% CO₂, 37 °C), followed by 5 min stimulation either by 10 nM MIP-2, or with vehicle (PBS).

To assay adhesive strength, the slides were mounted in a parallel-plate flow chamber (interior dimensions: 11 × 22 × 0.17 mm) connected to a pump, and placed on a microscope (Axiovert, Zeiss) with a heated stage, and imaged using a ×20 lens (Zeiss 20/0.30). HBSS (37°C) was perfused through the chamber at 10 dynes/cm⁻², and the results were recorded by a DVD recorder for 5 min. Still images of the video were captured every 30 s using PowerDVD software, imported into ImageJ (U.S. National Institutes of Health, http://rsb.info.nih.gov/ij/), and the number of neutrophils in each frame was quantified using the “Analyze Particles” feature. To quantify the rate of neutrophil detachment, the number of cells in each frame was standardized to the number of cells present in the same field of view before flow was initiated.

To assay the role shear plays in mediating intravascular crawling, the preparation was recorded for 5 min without flow. Flow (2 dynes cm⁻²) was then initiated, and the cells in the chamber were recorded for 5 min. The flow was then stopped and the cells were recorded for an additional 5 min. The videos were time-lapsed 600 × (1 frame = 20 s), and data were split into three parts, corresponding to the preflow, flow, and postflow periods.

Average speed was calculated by measuring the total distance traveled by the cell and then dividing that by the length of each time period (5 min). The displacement was calculated by determining the location of each cell at the beginning and end of each time period. The total distance moved along the x-axis (parallel to flow, Δx) and along the y-axis (perpendicular to flow, Δy) was then calculated for each cell. The displacement from flow (DFF) was calculated by dividing Δy by Δx, resulting in a DFF of 1.0 for randomly moving cells; DFF <1.0 for cells moving with flow, and DFF >1.0 for cells moving perpendicular to flow.

All data are presented as mean ± SEM. ANOVA, single-factor, non-repeated measures followed by Fisher protected least significant difference test was performed for multiple comparisons. For the in vitro data, single factor ANOVA was used with Tukeys test. Value of p < 0.05 was deemed statistically significant.

Table I shows that no neutrophil parameter upstream of adhesion was affected by Vav1 deficiency. In WT mice, ∼80 cells min⁻¹ rolled in unstimulated blood vessels at a rolling velocity of ∼35–40 μm · sec⁻¹ (Table I). Vav1^−/− mice had similar basal values for rolling flux and rolling velocity, and rolling flux decreased in both groups to ∼40 cells min⁻¹ following exposure to MIP-2, a ligand for CXCR2 (Table I). Under basal conditions, very few neutrophils adhered in blood vessels in either strain of mice. However, upon exposure to MIP-2, WT neutrophils adhered rapidly and, within the first minute postadhesion, began to crawl (Fig. 1^,A). Although it would be reasonable to expect the shear forces to induce neutrophil crawling in the direction of flow, it became apparent that in most instances neutrophils crawled perpendicular to blood flow (Fig. 1^,A, supplemental video 1).⁵ In fact, only ∼20% of neutrophils initially crawled in the direction of blood flow (Fig. 1^,C). By contrast, Vav1^−/− neutrophils almost always moved in the direction of blood flow, rarely deviating from this path (Fig. 1, B and C). The Vav1^−/− neutrophils but not WT neutrophils were greatly elongated, with a uropod often longer than the rest of the body of the neutrophil (Fig. 1, D and E). The Vav1^−/−neutrophils were stretched in the direction of flow. Rather than the amoeboid crawling seen in WT mice, in Vav1^−/− mice, a motion more reminiscent of dragging was noted (supplemental video 2).⁵

To determine whether the shear was mediating the direction of movement of neutrophils, we examined neutrophil crawling in low and high shear. Physiologic shear in postcapillary venules of the cremaster ranges between 200 and 1000 s⁻¹. The mean value of the observed high shear rate venules (WT 787 ± 106 s⁻¹ and Vav1^−/− 860 ± 218 s⁻¹, calculated from the mean RBC velocity) and the low shear rate venules (WT 279 ± 16 s⁻¹ and Vav1^−/− 271 ± 86 s⁻¹) was not significantly different among the two groups. The percentage of WT neutrophils that crawled only in the direction of blood flow was ∼20% in both high and low shear vessels (Fig. 1 C). By contrast, ∼100% of Vav1^−/− neutrophils oriented with blood flow in high shear vessels. In low shear vessels, some Vav1^−/− neutrophils did appear to crawl in directions other than that of blood flow; however, the majority of cells still oriented in the same direction as flow. To investigate whether neutrophil crawling behavior would be influenced by endothelial activation, WT and Vav1^−/− mice were pretreated with TNF-α, which has previously been shown to reduce velocity of RBC and venular shear (23). No difference in crawling behavior compared with MIP-2 superfusion in low shear venules could be detected, because only 22 ± 2% of WT neutrophils oriented with flow, but still the majority of Vav1^−/− cells (54 ± 6%) crawled only in the direction of blood flow. These data suggest that at sufficiently low shear, some Vav1^−/− neutrophils can crawl perpendicularly, but at physiologic shear, these cells mostly oriented in the direction of flow.

The total displacement was also different among the two experimental mouse strains. WT neutrophils crawled ∼25 μm, whereas Vav1^−/− neutrophils were displaced ∼50 μm in high shear venules (Fig. 2,A). The total distance WT neutrophils crawled was unaffected by shear (Fig. 2,A). By contrast, the distance that Vav1^−/− neutrophils moved in the low shear vessels was greatly reduced and comparable to WT cells (Fig. 2,A). The speed of the displacement did not differ between the WT and Vav1^−/− neutrophils (Fig. 2,B). Because of the substantial phenotypical difference in motility between WT and Vav1^−/− mice, with a more passive less amoeboid type crawling in the latter, we examined whether the Mac-1-dependent molecular mechanism was being used in Vav1^−/− neutrophils. Surprisingly, addition of Mac-1 Ab to Vav1^−/− neutrophils had less effect upon crawling (Fig. 2,C) than in WT neutrophils; >50% of already crawling cells were induced to stop. This strongly supports the view that unlike WT neutrophils, Vav1^−/−neutrophils do not engage Mac-1 to crawl. Addition of anti-LFA-1 Ab caused an ∼80% decrease in the number of adherent WT and Vav1^−/− neutrophils. This was entirely due to prevention of new adhesion in both groups of mice (Fig. 2 D). The few cells that still adhered continued to crawl (WT) or moved in the direction of blood flow (Vav1^−/−), suggesting a very limited role for LFA-1 in crawling in these cells.

To further examine the mechanism(s) underlying crawling behavior, we examined Vav1^−/− and WT neutrophil displacement in vitro in parallel-plate flow chambers in which the shear stress could be reproducibly applied to adherent neutrophils in the absence of endothelium to eliminate any potential chemotactic gradient on the surface of this substratum. WT neutrophils were first placed on a plasma protein-coated coverslip, which we previously demonstrated has ligands for LFA-1, Mac-1, and α₄-integrin (26). Under static (nonflow) conditions, WT (Fig. 3,A) and Vav1^−/− (data not shown) neutrophils migrated randomly in all directions. When shear was applied all WT neutrophils, almost in unison, began to crawl perpendicularly to flow (Fig. 3,B). On average, the neutrophils crawled at 62° relative to flow direction. Since no chemotactic or haptotactic gradient was present in this assay, the perpendicular movement was clearly a mechanotactic response. Perpendicular crawling in response to flow was not seen in the Vav1^−/− neutrophils, which instead were oriented in the direction of flow when shear was applied (Fig. 3,C), in accordance with our in vivo observation (Fig. 1, B and C). On average, Vav1^−/− neutrophils deviated from flow by only 19°.

These data are quantitatively summarized in Fig. 3, D and E, as deviation from the direction of flow and the crawling speed, respectively. We approximate the displacement in the flow chambers as occurring in two dimensions within the plane of the flow, where the displacement in x and y, Δx and Δy, are defined as the displacement parallel and perpendicular to the direction of flow, respectively. Fig. 3,D demonstrates that before flow, neutrophils crawled equally in the y and x direction (Δy/Δx = 1). Upon introduction of flow in the x direction, the cells reoriented perpendicular to flow such that the majority of crawling occurred in the y direction (Δy/Δx > 1). In fact, WT neutrophils were four times more prone to crawl in the y direction than the x direction. Once flow was terminated, the cells returned to random migration in all directions. Although, the Vav1^−/− neutrophils also migrated randomly under static conditions (Fig. 3,D), application of shear caused the exact opposite effect; the cells now moved with much greater propensity in the x direction (with flow) than in the y direction (perpendicular to flow). Once flow was terminated, the cells returned to a random crawling behavior. Fig. 3 E demonstrates that WT and Vav1^−/− neutrophils crawled with essentially the same speed in static conditions. However, when shear was applied, the speed with which WT neutrophils crawled was altered, whereas Vav1^−/− neutrophils did not decrease their speed. Interestingly, although random migration was restored upon cessation of shear, the reduced speed was not immediately reversed in WT neutrophils.

Next, we examined the importance of the mechanotaxic program of the crawling cells. Using intravital multichannel spinning disk microscopy, we noted that 90% of crawling neutrophils did follow endothelial junctions at some point during their recruitment (Fig. 4^,A, supplemental video 1),⁵ accounting for sudden radical changes in direction that were observed. If neutrophils adhered at junctional regions, they often began following the junction regardless of direction. In WT mice, the cells found an emigration site quickly and rarely crossed additional junctions (Fig. 4,B). By contrast, Vav1^−/− neutrophils often crossed junctions (Fig. 4,B) that would have required radical directional changes for the neutrophils to initiate junctional crawling. Vav1^−/− neutrophils eventually followed junctions but always in the direction of flow (Fig. 4 A).

Intravital microscopy of the cremaster muscle microcirculation revealed that adhesion increased 10-fold in WT mice in response to MIP-2 (5 nM; Fig. 5,A). Although Vav1^−/− cells also adhered, they were not able to sustain adhesion for 30 s. Since 30 s is by definition deemed adhesion, the overall adhesion was significantly impaired for Vav1^−/− neutrophils (Fig. 5,A) as a result of detachment. Fig. 5,B summarizes that during the adhesion and crawling process leading ultimately to emigration, ∼20% of WT neutrophils detached, whereas close to 60% of Vav1^−/− neutrophils detached (Fig. 5,B). To determine whether detachment was also increased in vitro, we used a parallel-plate flow chamber, coated with a multi-integrin ligand substratum (plasma proteins). WT and Vav1^−/− neutrophils adhered equally well under static conditions. When high shear (10 dynes/cm²) was applied, unstimulated WT and Vav1^−/− neutrophils both adhered poorly, with ∼50% of neutrophils detaching during 5 min of flow (Fig. 5,C). MIP-2 stimulation resulted in an enhanced ability of WT but not Vav1^−/− neutrophils to resist high shear (Fig. 5,C). Vav1^−/− neutrophils detached significantly more than WT neutrophils, entirely consistent with the Vav1^−/− neutrophils detaching more in vivo (Fig. 5 B).

Under these flow conditions, neutrophils from C57BL/6 and Vav1^−/− mice displayed strikingly different structural phenotypes. Whereas neutrophils from C57BL/6 mice maintained a flattened round or slightly polarized morphology at 10 dynes/cm² flow (Fig. 5^,D), Vav1^−/− neutrophils took on an unusual phenotype where the cell body was attached to the site of adhesion by a narrow prolonged uropod (Fig. 5^,E). Often these prolonged uropods were two to three cell diameters in length. In addition, frequently the cell body would detach from the substratum and further elongate. As already mentioned, the elongated Vav1^−/− phenotype was also seen in vivo (Fig. 1 D and supplemental video 2).⁵ Occasionally, WT neutrophils would form a similar structure, but these structures were rare, very short-lived, not as pronounced as in Vav1^−/− neutrophils, and resulted very quickly in cell detachment.

Ultimately, the biggest impact of the impaired mechanotaxis of Vav1^−/− neutrophils was on the process of emigration. In WT mice, ∼70 neutrophils emigrated out of the vessel during 60 min with MIP-2 superfusion (Fig. 6^,A). In the Vav1^−/− mice, only 10 cells emigrated during the same time period with MIP-2 superfusion (Fig. 6^,A). Spinning disk confocal microscopy and PECAM-1-labeled junctions revealed that the majority of WT neutrophils emigrated at junctions (Fig. 6 B). The few Vav1^−/− neutrophils also emigrated at junctions, suggesting that when the Vav1^−/− cells were able to engage junctions and follow them, they would then emigrate. Interestingly, in the few Vav1^−/− neutrophils where the neutrophils began emigrating, the uropod would not detach causing cells to struggle to emigrate (supplemental video 3).⁵ Once detachment occurred, these same cells migrated efficiently in the interstitium (supplemental video 3),⁵ suggesting not all crawling is impaired in Vav1^−/− cells.

To determine whether the impaired crawling and emigration in Vav1 deficiency could impact on a clinically relevant innate immune response, WT and Vav1^−/− mice were challenged with Staphylococcus aureus-derived PGN (5 mg · kg⁻¹ i.p.), and leukocyte recruitment into the peritoneal cavity was examined (after 4 and 24 h). As shown in Fig. 7,A, treatment of WT or Vav1^−/− mice with either PGN or saline did not result in any significant changes in circulating leukocyte counts. WT mice exhibited a significant leukocyte recruitment in peritoneum following challenge with PGN (Fig. 7,B). Interestingly, peritoneal leukocyte counts were lower (12 × 10⁶) in the Vav1^−/− mice than in WT mice (22 × 10⁶) following 4 h PGN challenge. Nearly all of the recruited cells in the WT and Vav1^−/− mice were neutrophils, as determined by leukocyte differentials, and Vav1^−/− mice had less neutrophil recruitment into the peritoneal cavity than WT mice in response to PGN (Fig. 7 C).

Leukocyte recruitment was also studied in WT and Vav1^−/− mice for 24 h after challenge (PGN; 5 mg · kg⁻¹ i.p.) to determine whether neutrophil emigration was just delayed in Vav1^−/− mice. A decreased number of accumulated leukocytes in the peritoneum was detected at 24 h compared with 4 h in both WT (7.4 ± 0.8 10⁶) and Vav1^−/− (4.8 ± 01.4 10⁶) mice (Fig. 7,B), as previously described for WT mice (25), and the vast majority of recruited cells were again identified as neutrophils (Fig. 7,C). Fewer Vav1^−/− neutrophils (3 ± 0.8 10⁶) compared with WT (5.2 ± 0.5 10⁶) were recruited also at this delayed time point (Fig. 7 C; p = 0.058). This indicates that the crawling defect observed in the Vav1^−/− neutrophils, concomitant with the decrease in emigration reduces the capacity for an innate immune response and not just a delayed response in a model of peritoneal inflammation.

There is a growing body of evidence that interposed between leukocyte adhesion and emigration exists another step in the recruitment cascade, namely intraluminal crawling (8, 9, 10, 11). This step eluded investigators partly because the crawling required time-lapse video microscopy, a technique generally not applied to the microvasculature where dynamic events need to be imaged in real time. In this study, we report that an inherent characteristic of adherent neutrophils is to crawl perpendicularly to fluid shear. This behavior can be replicated on protein-coated coverslips in vitro but only in the presence of shear and entirely independent of the presence of a chemotactic gradient. In vivo, the neutrophils crawl perpendicularly to blood flow until they overlay a junction. At this stage, the shear-induced perpendicular crawling can be superseded by junctional proteins and/or haptotactic gradients and the neutrophils often change direction sometimes radically and begin to follow the path of the junction and ultimately emigrate out of the vasculature. To date, no one has assessed any of the intracellular-signaling pathways that lead to the crawling in blood vessels under flow conditions. Herein we report for the first time that the guanine exchange factor Vav1, a major regulator of the actin cytoskeleton, is critical for the shear-induced perpendicular crawling seen in WT neutrophils. The underlying molecular impairment appears to be an inability for Vav1^−/− neutrophils to crawl via Mac-1. Indeed, much of the crawling in WT neutrophils was Mac-1 dependent, whereas Mac-1 Ab did not impact on Vav1^−/− neutrophil behavior in vivo.

Vav1 is found only in hemopoietic cells (27) and is a very important signaling molecule downstream of LFA-1 (13, 16, 17, 18), and cross-linking LFA-1 does indeed induce Vav1-dependent cytoskeletal changes (15) leading to in vitro phenotypes such as neutrophil spreading, membrane ruffling, and polarization (12, 13). It has previously been shown that RhoA and actin-myosin bundles (molecules activated downstream of Vav1) localize to the uropod when neutrophil motility is initiated (28). In this study, we extend these observations by identifying severely defective Mac-1-dependent mechanotactic crawling after LFA-1-dependent adhesion in Vav1^−/− cells under shear conditions. Although WT neutrophils crawled via Mac-1, the downstream movement of Vav1^−/− neutrophils was not dependent on Mac-1 under shear conditions, suggesting a failure to activate this molecule. The sequential LFA-1 activation followed by Mac-1 activation has been documented in vitro by Simon and colleagues (14), further supporting our in vivo observation.

Neutrophils initiate lamellipodia, which dictate direction, while deadhesion occurs in the uropod allowing for a smooth forward motion. Previous work in vitro under static conditions in lymphocytes revealed that LFA-1 affinity is increased at the leading edge of a migrating cell (29) reaching peak levels in the centre of the cell (30). Although the adhesion at the uropod is also LFA-1 dependent, deadhesion is necessary at this site to allow for effective movement forward. Accordingly, Tohyama et al. (31) have shown that while affinity changes regulated LFA-1 binding at the front of the cell, avidity changes in LFA-1-mediated adhesion in the uropod, perhaps allowing for more effective detachment. It should be noted that all the aforementioned work was done under static in vitro conditions. Crawling under flow conditions would require both the molecular complexities of migration observed in tissues, but also the necessary molecular mechanisms resulting in tethering to endothelium that would prevent detachment due to shear. We would hypothesize that an exquisite communication between LFA-1 and Mac-1 must exist where LFA-1 binds first for firm adhesion and would perhaps disengage only after Mac-1 bound its ligand. It would seem intuitive that LFA-1 would remain activated until Mac-1 was engaged. In our in vivo imaging, it was clear that Vav1^−/− cells were able to adhere in blood vessels via LFA-1, but the leading edge and body often detached leaving the uropod as the only point of contact for extended periods. This reduced surface area of Vav1^−/− neutrophils in contact with endothelium frequently resulted in detachment both in vivo and in vitro perhaps due to lack of engagement of Mac-1. Indeed, Mac-1 is localized to the leading edge and body of the crawling neutrophil (32).

Although the neutrophils were able to emigrate at junctions in both strains of mice, the number of Vav1^−/− cells that emigrated was dramatically reduced. This is quite interesting when one considers that Mac-1-deficient neutrophils are incapable of crawling but emigrate at similar levels as WT neutrophils albeit with some delay (10). Clearly, Vav1^−/− neutrophils cannot emigrate as effectively as Mac-1-deficient cells, suggesting a more severe defect in the former. Indeed, Mac-1^−/− neutrophils did not detach and were able to extend pseudopods into the endothelium for both paracellular and transcellular migration. By contrast, detachment of the leading edge and cell body would prevent pseudopod extension at the leading edge of Vav1^−/− neutrophils reducing transmigration. In addition the Vav1^−/− neutrophils detach more than WT neutrophils also likely contributing to the lack of emigration. Interestingly, increased detachment of neutrophils has been reported previously but only in Vav1/3 double knockout but not Vav1^−/− mice (19). However, these investigators examined detachment only over 90 s after very brief administration of a chemokine. Indeed, we observed substantial adhesion in Vav1^−/− neutrophils over this period of time with detachment only occurring at a delayed time point. To our knowledge, neither Gakidis et al. (19) nor anyone else has examined crawling and emigration in Vav1^−/− mice.

Endothelial cells are cobblestone when grown under static conditions, but when exposed to shear as is the case in vivo, these cells are elongated such that they can be up to 200 μm in length (tapering at the ends) but only 30–50 μm in width. When taking into consideration this elongated shape of venular endothelial cells, the closest junction and by default the fastest way for a neutrophil to find a junction would be to move left or right. Therefore, intuitively, it makes sense that WT neutrophils immediately begin migrating perpendicularly to blood flow. This process was unlikely to require a chemotactic gradient, as even in vitro in the absence of chemokine gradients, all neutrophils began crawling in straight lines perpendicular to applied shear forces but immediately returned to random migration when flow was terminated, suggesting that the neutrophil has the capacity to detect and react to shear. Our data, however, do not dismiss the possibility that in vivo the neutrophils can detect chemotactic gradients in blood vessels, further modifying their crawling directionality, depending on the source of the chemoattractant (i.e., left or right).

The ability to detect mechanical forces (mechanoception) has been identified in a large variety of cells. In fact, it is well-known that a variety of single-cell organisms as well as motile cells in multicellular organisms can respond to shear. For example, Dicytostelium aligns in the direction of flow (33). Neutrophils behave very similarly to Dicytostelium and in many respects have similar molecular processes to permit crawling. Interestingly, multiple integrins have been identified as mechanoreceptors, capable of detecting shear. Indeed, studies have shown shear-dependent regulation of integrins on other myeloid cells (34). Shear-dependent regulation of CD18 on neutrophils has been reported (35), suggesting that the activity of integrins on neutrophils can be regulated by shear. We submit that the activation of integrin mechanoreceptors induced the activation of signaling molecules such as Vav1, which plays a central role in mediating neutrophil motility (28, 36, 37, 38, 39, 40). Clearly, mechanosensing is an essential cue, leading to effective neutrophil emigration thereby responding in a timely fashion to infection and immunity.

Haptotaxis (following stationary signals) is suggested to be important in leukocyte recruitment during inflammation (41, 42, 43). We found that when the perpendicularly crawling neutrophils reached an endothelial junction, some changed direction and started to follow the junctions, indicating a switch from mechanotaxis to haptotaxis as stationary junctional-specific signals were followed. Our data thus demonstrates that crawling involves two distinct stages. In the initial stage, the cell moves perpendicular to flow, using the mechanotactic signal of flow to determine direction. This mechanotactic movement is terminated when the cell detects that it is crossing an endothelial junction. At this point, the cell ignores the mechanotactic signal provided by flow and begins to follow the haptotactic signal, provided by junctional proteins, to the site of emigration.

In conclusion, an inherent behavior of neutrophils exposed to shear was detected, because the neutrophils migrated perpendicular to flow in vitro as well as in vivo. In vivo this led the neutrophils to the closest endothelial junction, which they often followed for periods of time before transmigrating out of the vasculature. This feature is absent in Vav1^−/− neutrophils, which suggests Vav1 is downstream of adhesion, and dependent upon LFA-1 outside-in signaling required for optimal Mac-1-dependent crawling under physiological shear. Finally, in a model of peritonitis, we found that fewer Vav1^−/− cells were recruited to the peritoneal cavity, indicating the importance of Vav1-dependent intravascular crawling to optimal sites for emigration in innate immunity.

We thank Krista McRae and Sara Massena Santos for excellent technical assistance, the Microscopy and Imaging Facility and The Live Cell Imaging Facility at the Faculty of Medicine, Calgary University.

The authors have no financial conflict of interest.