Different Isoforms of the Neuronal Guidance Molecule Slit2 Directly Cause Chemoattraction or Chemorepulsion of Human Neutrophils Free

Chemoattraction is the directed movement of cells toward an attractant and is the mechanism by which, for instance, immune cells move to sites of infection and inflammation (1, 2). Although much is known about chemoattraction in eukaryotic cells, relatively little is known about how eukaryotic cells move away from a repellent (chemorepulsion). Slit was originally identified as an extracellular neuronal chemorepellent protein sensed by Robo receptors in Drosophila and has since been found in Caenorhabditis elegans and vertebrates (3, 4). Mammals have Slit1, Slit2, and Slit3, which are cleaved into ∼110–140-kDa N-terminal and ∼55-kDa C-terminal fragments (3, 5). The ∼140-kDa Slit2 N-terminal fragment (Slit2-N) contains four N-terminal leucine-rich repeat domains and five EGF-like domains and shows 96.8% identity (98.8% similarity) between mice and humans. A truncated ∼110-kDa Slit2 N-terminal fragment (Slit2-S) containing only the four N-terminal leucine-rich repeat domains is 97.2% identical (98.9% similar) between mouse and human. Altered fragmentation patterns of Slit2 have been observed in cancer, inflammation, fibrosis, and obesity (6–11), and we found abnormally low levels of Slit2 fragments in the bronchoalveolar lavage of mice with pulmonary fibrosis (12).

Immune cells such as neutrophils also express Robo receptors (13–15). Slit2 inhibits the chemotaxis of a variety of immune cells toward attractants, but whether Slit can act as a repellent for immune cells is unknown (12–21). In this report, we show that, for neutrophils, Slit2-N is a chemoattractant and Slit2-S is a chemorepellent and, although the two fragments use similar receptors, they use different signal transduction pathways.

Human venous blood was collected from healthy volunteers who gave written consent, with approval from the Texas A&M University Institutional Review Board. Neutrophils were isolated as previously described (22), with the exceptions that Polymorphprep gradients (Axis-Shield; Oslo, Norway) were used following the manufacturer’s instructions, and centrifugation of gradients was done for 40 min. Cells were then resuspended in RPMI 1640 (Lonza, Walkersville, MD) containing 2% BSA (Amresco, Solon, OH) (RPMI-BSA). To check the purity of the neutrophil isolation, cell spots were prepared as described previously (22) and stained with eosin and methylene blue. Isolated cell preparations were 95.6 ± 0.5% neutrophils, 2.2 ± 0.4% monocytes, 1.6 ± 0.4% eosinophils, and 0.6 ± 0.2% lymphocytes (mean ± SEM, n = 16). In addition, cells attached to fibronectin-coated coverslips were air dried, fixed in methanol, and stained with methylene blue and eosin. These were 95.8 ± 1.2% neutrophils, 2.0 ± 1.0% monocytes, 1.8 ± 0.8% eosinophils, and 0.3 ± 0.2% lymphocytes (mean ± SEM, n = 5).

Insall chambers (23) were used to generate gradients of Slit2 and other compounds to observe the movement of neutrophils on glass coverslips coated with 10 μg/ml human plasma fibronectin (Corning, Bedford, MA), as previously described (22, 24, 25). Cells were allowed to adhere to coverslips for 30 min before the coverslip was placed onto the Insall chamber. Recombinant mouse Slit2 Gln26-Gln900 (∼110 kDa; R&D Systems, Minneapolis, MN), recombinant human Slit2 Gln26-Val1118 (120–140 kDa; PeproTech, Rocky Hill, NJ), and recombinant human Slit2 C-terminal fragment Thr1122-Ser1529 (∼50 kDa; R&D Systems) were resuspended in RPMI-BSA. Slit2 proteins were checked for purity and size by PAGE and silver staining, as described previously (12, 26). The neutrophil chemoattractant fMLF (Alfa Aesar, Ward Hill, MA) was diluted in RPMI-BSA and used at 10 nM.

Rho GTPase Cdc42 inhibitor ML141, Ras inhibitory peptide (RIP), and Rac inhibitor NSC 23766 (all from Cayman Chemical, Ann Arbor, MI) and the PI3K inhibitor LY294002 (BioVision, Milpitas, CA) were reconstituted to 10 mM in DMSO (Amresco) according to the manufacturer’s directions. The DMSO stocks were then diluted in RPMI-BSA and used at 10 μM. Cells were incubated with inhibitors for 30 min at 37°C in a humidified 5% CO₂ incubator before placing on the coverslip. Where indicated, cells were incubated with 10 μg/ml sheep anti-human Roundabout homolog 1 (Robo1) Abs (R&D Systems) or 3 μg/ml sheep anti-human Syndecan-4 Abs (R&D Systems) for 30 min as described above. After placing coverslips with adhered neutrophils on the Insall chamber, we waited 20 min for gradients to form before tracking neutrophils as previously described (22, 24). At least 10 neutrophils per experiment were tracked for 40 min. Only cells that could be tracked for the full 40 min were analyzed. For each donor, neutrophils were tracked in a no-gradient control. Neutrophils were never used past 5 h from the end of the isolation step. We used two Insall chamber/microscope/camera setups in parallel, allowing six to eight experimental conditions to be measured per set of donor neutrophils (Supplemental Videos 1–6). The results are expressed as the mean ± SEM of the movement of neutrophils from three or more different donors. We never used the same donor twice for a given experiment.

Preparation of cytoskeletons and gel electrophoresis to visualize F-actin was done as previously described (27–29). Briefly, 5 × 10⁵ cells in 100 μl RPMI-BSA were incubated at 37°C in the presence or absence of 500 ng/ml Slit2 or 10 nM fMLF. At the indicated time points, 1.4 ml of ice-cold PBS was added to the cells, and cells were collected by centrifugation at 500 × g for 5 min at 4°C. Cells were lysed with 100 μl of 100 mM PIPES (pH 6.8), 1 mM MgCl₂, 2.5 mM EGTA, and 0.5% Triton X-100 with 4× protease and phosphatase inhibitor (Cell Signaling Technology) on ice for 20 min. Triton-insoluble cytoskeletons were collected by centrifugation at 12,000 × g for 5 min at 4°C. Cytoskeleton pellets were resuspended in SDS/DTT sample buffer, heated to 98°C for 5 min, separated by PAGE, and then stained with Coomassie as described previously (29).

1 × 10⁶ cells in 200 μl of RPMI-BSA were allowed to attach to fibronectin-coated 8-well slides (Corning) for 30 min at 37°C. Two microliters of prewarmed RPMI-BSA (control), prewarmed RPMI-BSA plus 100 μg/μl Slit2-N or Slit2-S, or prewarmed RPMI-BSA plus 200 nM fMLF was then added to the corner of the well, taking care not to disturb the cells or the medium. After 10 min, the cells were fixed by carefully adding 200 μl of prewarmed 4% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA) in PBS to the wells for 20 min at 37°C. Cells were then washed with PBS and permeabilized with PBS/0.1% Triton X-100 (PBS-T) for 5 min at room temperature. Cells were then incubated with 1:1000 anti–phospho-AKT substrate (proteins containing phospho-serine or phospho-threonine preceded by arginine at the -3 position; RXXS*/T*) (30) Ab (rabbit IgG, no. 9614; Cell Signaling Technology, Danvers, MA) in PBS-T containing 2% BSA or 5 μg/ml anti-CD11b (mouse IgG, ICRF44; BioLegend, San Diego, CA) in PBS-T containing 2% BSA overnight at 4°C. Cells were washed with PBS-T before the addition of either 1 μg/ml goat anti-rabbit IgG Alexa 647 (Thermo Fisher Scientific, Waltham, MA) or donkey anti-mouse IgG Alexa 488 (Jackson ImmunoResearch, West Grove, PA) for 30 min at room temperature. Cells were washed with PBS-T, and coverslips were mounted with fluorescent mounting medium containing DAPI (Vector Laboratories). Alternatively, after fixation, cells were incubated with 1:2000 phalloidin-Alexa 555 (ab176756; Abcam) in PBS-T for 30 min at room temperature. Cells were washed with PBS-T, and coverslips were mounted with fluorescent mounting medium containing DAPI (Vector Laboratories). Immunofluorescence images were captured on an Olympus FV1000 confocal microscope and analyzed using Olympus Fluoview and ImageJ software as described previously (12, 26). Cells were scored as having polarized phalloidin staining (one region of staining, either toward or away from the corner of the well where material was added), multiple staining (having two or more regions of phalloidin staining), or circumferential/uniform staining (in which the majority of the cell was stained with phalloidin).

96-well polystyrene plates (353072; BD Biosciences) were coated with 10 μg/ml human plasma fibronectin in PBS for 1 h at 37°C. The plate was then washed three times with PBS, and 1 × 10⁵ neutrophils in 100 μl of RPMI-BSA were allowed to attach to the coated wells for 30 min at 37°C. A total of 100 μl of 10 nM fMLF, 500 ng/ml Slit2-N, 500 ng/ml Slit2-S, or 10 ng/ml TNF-α (BioLegend) was then added to the wells and incubated for a further 30 min at 37°C. Plates were then washed three times with PBS, and adherent neutrophils were counted in three 900-μm-diameter fields of view per well as described previously (31). In addition, neutrophils were preincubated with 500 ng/ml Slit2-N or Slit2-S for 15 min at 37°C before incubation with buffer, fMLF, or TNF-α for an additional 15 min. Cells were then added to fibronectin-coated wells, and cell adhesion was measured as described above.

10 × 10⁶ neutrophils in 0.4 ml of RPMI-BSA were incubated at 37°C in the presence or absence of 500 ng/ml Slit2 or 10 nM fMLF. For phosphatidylinositol 4,5-bisphosphate (PIP2) and phosphatidylinositol 3,4,5-triphosphate (PIP3) assays, after 5 min, 0.4 ml of ice-cold 1000 mM TCA was added to the cells and then incubated on ice for 5 min. PIP2 and PIP3 were extracted and detected using ELISA kits following the manufacturer’s instructions (PIP2 K-4500, PIP3 K-2500s; Echelon, Salt Lake City, UT). For Ras activation, cells were incubated with Slit2 or fMLF for 6 min, then 1.2 ml of ice-cold PBS was added to the cells, and the cells were then collected by centrifugation at 500 × g for 5 min at 4°C. Cells were then resuspended in 0.2 ml of kit lysis buffer, and the active GTP-bound form of Ras proteins was isolated and detected following the manufacturer’s instructions (BK008-S; Cytoskeleton, Denver, CO). Briefly, eluate from beads bearing the Ras-binding domain of the Ras effector kinase Raf1, which isolates GTP-Ras, was analyzed by Western blotting using 4–20% Tris/glycine gels (Lonza, Allendale, NJ) (12, 26). After blotting to PVDF membranes (Immobilon P; MilliporeSigma, Burlington, MA), protein transfer was assessed by Ponceau red staining (32). Blots were blocked and then incubated with anti-Ras Abs following the manufacturer’s instructions (BK008-S; Cytoskeleton). Cell lysates were also analyzed by Western blotting with 200 ng/ml anti-GAPDH mouse mAb (Proteintech, Rosemont, IL) to confirm effective protein transfer.

Statistical analyses with t tests or one-way or two-way ANOVA with Dunnett posttest were done using GraphPad Prism 7 (GraphPad, San Diego, CA). Significance was defined as p < 0.05.

In the developing nervous system, Slit2 acts as a chemorepellent for neurons (3–5). Slit2 concentrations between 400 ng/ml and 100 μg/ml inhibit the chemotaxis of human and murine neutrophils toward the chemoattractants fMLF, CXCL12, IL-8, and the complement component C5a (15, 16, 18). An intriguing possibility is that Slit2 might act directly as a chemorepellent for neutrophils. To test this, we used a gradient chamber to examine the effect of Slit2 gradients on human neutrophil movement. We first examined the effects of a 110-kDa fragment of mouse Slit2, henceforth designated Slit2-S. Compared with cells in no gradient and gradients of 0–5 and 0–50 ng/ml, Slit2-S did not significantly affect neutrophil movement (Fig. 1A, 1E, Supplemental Fig. 1A, 1B, 1G, and Supplemental Video 1). A gradient of 0–500 ng/ml (0–5 nM) Slit2-S caused neutrophils to move away from the source of the Slit2-S (Fig. 1B, 1E, Supplemental Fig. 1G, and Supplemental Video 2), with a forward migration index (FMI) comparable to the FMI induced by the chemorepellent DPPIV for neutrophils (22) or AprA for Dictyostelium cells (24). We previously observed that ∼17% of cells in DPPIV or AprA chemorepellent gradients move toward the source of the chemorepellent (22, 24), and we observed that 29 ± 5% of cells (mean ± SEM, n = 6 donors) of the neutrophils in the 0–500 ng/ml Slit2-S gradient moved toward the source. A gradient of 0–5000 ng/ml Slit2-S, however, did not cause chemorepulsion (Fig. 1E). As previously observed (15, 33), 1 nM fMLF caused chemoattraction of neutrophils (Fig. 1E, Supplemental Fig. 1G, and Supplemental Video 3). In the fMLF gradient, only 6 ± 4% (mean ± SEM, n = 6 donors) of neutrophils moved away from the fMLF source. This is significantly lower than the percentage of cells going the “wrong way” in the Slit2-S gradient (p < 0.05, t test), suggesting that the fMLF gradient affects more neutrophils than the Slit2-S gradient. An fMLF gradient alongside a 0–500 ng/ml Slit2-S gradient showed a dominance of the fMLF chemoattraction over the Slit2-S chemorepulsion (Fig. 1E, Supplemental Fig. 1G). Some chemoattractants affect the speed and/or directionality (the distance between the starting and ending point of a cell divided by the distance along the cell’s path) of cells (34–36). fMLF with 0 and 50 ng/ml Slit2-S increased the speed of the neutrophils along their tracks (Supplemental Fig. 1A). fMLF and fMLF plus 500 ng/ml Slit2-S caused the directionality to increase compared with control (Supplemental Fig. 1B). Other than 0–5 ng/ml Slit2-S decreasing speed for female neutrophils (Supplemental Fig. 2C), there were no significant differences in the response of neutrophils from male and female donors to Slit2-S (Supplemental Fig. 2). Together, these data suggest that Slit2-S can act as a chemorepellent for human male and female neutrophils but that this can be overridden by an fMLF chemoattraction gradient.

We also examined whether gradients of a commercially available ∼140-kDa fragment of human Slit2, henceforth designated Slit2-N, can affect neutrophil movement. Gradients of 0–5 or 0–50 ng/ml Slit2-N did not significantly affect neutrophil movement (Fig. 1F, Supplemental Fig. 1H), 0–500 ng/ml (0–3.6 nM) gradients caused neutrophils to move toward the Slit2-N (Fig. 1C, 1F, Supplemental Fig. 1H, and Supplemental Video 4), and 0–5000 ng/ml gradients caused female but not male cells to move away from the Slit2-N (Fig. 1D, 1F, Supplemental Fig. 2B). In the 0–500 ng/ml gradient of Slit2-N, 14 ± 6% (mean ± SEM, n = 6) of the neutrophils moved away from the source; this was not significantly different from the percentage of wrong-way cells in an fMLF gradient (t test). In the 0–5000 ng/ml gradient of Slit2-N, 38 ± 6% (mean ± SEM, n = 6) of the neutrophils moved toward the source; this was significantly higher (p = 0.0042, one-way ANOVA, Dunnett test) than the percentages of wrong-way cells in fMLF or 0–500 ng/ml Slit2-N gradients (p = 0.0309; one-way ANOVA, Dunnett test). None of the Slit2-N concentrations significantly affected neutrophil speed or directness (Supplemental Fig. 1C, 1D). Other than males not responding to the 0–5000 ng/ml Slit2-N gradients, there were no significant differences in the response of neutrophils to Slit2-N from male and female donors (Supplemental Fig. 2B, 2D, 2F). The data suggest that 500 ng/ml Slit2-N acts as a chemoattractant.

Although gradients of 0–50, 0–500, or 0–5000 ng/ml of a 55-kDa human Slit2-C fragment did increase neutrophil speed (Supplemental Fig. 1E), they did not significantly affect the direction or directionality of neutrophil movement (Fig. 1G, Supplemental Fig. 1F, 1I).

To induce chemorepulsion of growing axons, neurons use the Robo1 receptor to sense Slit2 (3–5). Robo1 also mediates the ability of Slit2 to inhibit neutrophil chemoattraction (15, 18). Neutrophils preincubated with anti-Robo1 Abs did not respond to either a 0–500 ng/ml Slit2-S gradient or a 0–500 ng/ml Slit2-N gradient (Fig. 2A). The anti-Robo1 Abs did not significantly affect cell speed (Supplemental Fig. 3A), but anti-Robo1 Ab pretreatment with a Slit2-S gradient increased directness (Supplemental Fig. 3B). These data suggest that Robo1 is necessary for Slit2 chemorepulsion or chemoattraction of human neutrophils.

Syndecans are proteoglycans that can act as coreceptors for Robo1 signaling (37, 38). Of the four human syndecans, Syndecan-4 appears to mediate human neutrophil chemotaxis (39). Preincubation of human neutrophils with anti–Syndecan-4 Abs inhibited the response of neutrophils to 0–500 ng/ml gradients of both Slit2-S and Slit2-N (Fig. 2B). In the presence of Slit2-S or Slit2-N gradients, the anti-Syndecan Abs increased cell speed (Supplemental Fig. 3C) but did not significantly affect directness (Supplemental Fig. 3D). These data suggest that Syndecan-4 is necessary for Slit2 chemorepulsion or chemoattraction of human neutrophils.

Neutrophil migration to chemoattractants leads to cytoskeletal reorganization and formation of F-actin at the leading edge of the cell, which can be visualized by phalloidin staining (2, 40). To determine if Slit2-induced migration induces F-actin reorganization, neutrophils were allowed to adhere to fibronectin-coated 8-well slides and were then stimulated with a point source of buffer, Slit2, or fMLF for 10 min and then fixed, permeabilized, and stained with phalloidin. Cells were then analyzed for the location of the F-actin. The addition of buffer to one corner of the well did not cause any significant asymmetry of F-actin localization toward or away from the source of the buffer, many cells had F-actin around the periphery of the cell, and some cells had two or more (multiple) spots of F-actin at the edges of cells (Fig. 3A, 3E). Neutrophils in gradients of fMLF or Slit2-N showed a decrease in the percentage of cells with a uniform distribution of F-actin at the edge of the cell and an increase in the percentage of cells with F-actin localized to the edge of the cell toward the attractant (Fig. 3B, 3C, 3E). Gradients of the chemorepellent Slit2-S also decreased the percentage of cells with F-actin all around the cell and increased the percentage with multiple regions of F-actin at the edge but did not significantly affect the percentage of cells with F-actin at the edge of the cell toward or away from the source of Slit2-S (Fig. 3D, 3E). As previously observed (27, 28, 41), fMLF transiently increased total F-actin polymerization at 90 s (Fig. 3F). Slit2-S and Slit2-N did not cause significant changes in F-actin over 300 s (Fig. 3F). Compared with control cells, both fMLF and Slit2-N induced phosphorylation of myosin L chain 2 (MLC2) at 60 s, and Slit2-N also led to reduced pMLC2 levels at 5 min (Fig. 3G). Slit2-S had no significant effect on pMLC2 levels (Fig. 3G). These data suggest that Slit2-S affects the cytoskeleton to induce chemorepulsion in a manner that is different from how fMLF and Slit2-N affect the cytoskeleton to induce chemoattraction.

Cell migration requires binding to a surface to allow traction but not with too low or too high an adhesive force, which prevents migration (34–36). To determine if the Slit2 isoforms affect cell–substratum adhesion, we allowed neutrophils to adhere to fibronectin-coated plates before the addition of Slit2 proteins, the chemoattractant fMLF, or TNF-α, a cytokine that promotes neutrophil adhesion (31). Compared with unstimulated neutrophils, cells incubated with fMLF, Slit2-N, or TNF-α had significantly increased adhesion (Fig. 4A), whereas the chemorepellent Slit2-S did not significantly affect adhesion (Fig. 4A). Preincubation with Slit2 isoforms inhibits the ability of fMLF or TNF-α to increase adhesion of neutrophils to activated endothelial cells (19, 42, 43). Both Slit2-N and Slit2-S significantly inhibited fMLF-induced adhesion to fibronectin, and Slit2-S but not Slit2-N inhibited TNF-α–induced adhesion (Fig. 4B). Cell migration leads to a change in cell shape, with an elongation of the cell with pseudopods forming at the leading edge of the cell and a more flattened cell as it attaches to the underlying matrix (1, 2). Compared with unstimulated neutrophils, cells in gradients of fMLF or Slit2-N had an increase in cell length and cell area, whereas cells in Slit2-S gradients were not significantly different from control cells (Fig. 4C, 4D). These data suggest that, unlike the chemoattractants fMLF and Slit2-N, which increase adhesion to fibronectin, cell elongation, and cell flattening, the chemorepulsive Slit2-S does not affect these parameters.

Neutrophil recruitment to an inflamed tissue involves initial attachment to endothelial cells, followed by rolling across the endothelial cell surface, and then firm adhesion to the endothelium, transendothelial migration, and chemotaxis through the tissue to the inflammatory source (1, 44). Adhesion of neutrophils to either endothelial cells or tissue extracellular matrix is mediated by integrins, including CD11b/CD18 (45, 46), which relocate to the front and rear of the cell during adhesion and migration (47, 48). To determine if Slit2 proteins affect integrin relocation, neutrophils were incubated on fibronectin in Slit2 gradients, fixed, and stained for CD11b, and the staining intensity was then scanned along an axis running through the middle of the cell parallel to the gradient. Compared with unstimulated cells, which show peaks of CD11b staining at the edges of the cell, fMLF did not significantly alter the staining of CD11b at the edge of the cell closest to the stimulus, increased staining in a region immediately behind the edge, and decreased staining at the rear (Fig. 5A). Cells incubated in a Slit2-S gradient had significantly less CD11b staining at the side of the cell closest to the Slit2-S and had decreased staining throughout the cell (Fig. 5A). Slit2-N decreased staining of CD11b in the part of the cell farthest from the Slit2-N.

Cell migration is driven by the detection of chemostimulants through a diverse number of receptors, including G protein–coupled receptors, such as the fMLF receptor (49), and single-pass transmembrane receptors, such as Robo (3, 5). Although the intracellular domains of the fMLF and Robo receptors bind different signaling molecules, both receptors appear to regulate pathways downstream from the AKT kinase (14, 50–53). To determine if the isoforms of Slit2 differentially regulate AKT activity, we used an Ab that detects phosphorylation of AKT substrates (30). Compared with unstimulated cells, cells exposed to a gradient of fMLF had no discernable difference in pAKT substrate staining (Fig. 5B). Cells incubated with Slit2-N had increased staining at the front and rear of the cell, whereas cells incubated with Slit2-S had reduced staining in the middle of the cells (Fig. 5B). These data suggest that the Slit2 fragments differentially reorganize cell surface integrins and differentially regulate AKT activity.

PI3K mediates the movement of neutrophils toward some chemoattractants (35). LY294002 is an inhibitor of PI3K (54) and has been used at 10–50 μM to inhibit PI3K signaling and migration in human neutrophils and endothelial cells (35, 55). To determine if the effects of Slit2 on neutrophils are mediated by PI3K, neutrophils were preincubated with LY294002 and then observed in Slit2 gradients. LY294002 had no significant effect on the ability of a 0–500 ng/ml gradient of Slit2-S to induce chemorepulsion (Fig. 6A). However, LY294002 caused neutrophils in a 0–500 ng/ml Slit2-N gradient to be repelled by the Slit2-N (Fig. 6A, Supplemental Video 5). LY294002 caused neutrophil speed to increase in a Slit2-S gradient (Supplemental Fig. 4A), but LY294002 did not significantly affect directness (Supplemental Fig. 4B). To confirm the role of PI3K in Slit2-N mediated chemoattraction, PIP2 and PIP3 levels were measured in neutrophils incubated with Slit2-N, Slit2-S, or fMLF. After 5 min of incubation, none of the treatments significantly affected PIP2 levels, but compared with unstimulated cells, Slit2-N and fMLF increased levels of PIP3, whereas Slit2-S did not significantly affect PIP3 levels (Fig. 7A). These data suggest that Slit2-S does not increase PIP3 and does not need PI3K to cause chemorepulsion, whereas Slit2-N increases PIP3 and needs PI3K to cause chemoattraction.

Cdc42 is a small GTPase that mediates the effects of many chemoattractants on the cytoskeleton (56). In the developing nervous system, Slit2 activation of Robo causes an inactivation of Cdc42, and this inactivation is necessary for the ability of Slit2 to cause chemorepulsion of neuronal cells (57). Cdc42 also mediates the ability of Slit2 to inhibit neutrophil chemotaxis toward stromal cell–derived factor 1α (18). ML141 is an allosteric inhibitor of Cdc42 and has been used at 10 μM to inhibit Cdc42-mediated chemotaxis in human neutrophils (58). ML141 caused 500 ng/ml Slit2-S to act as a chemoattractant rather than as a chemorepellent (Fig.6B, Supplemental Video 6) and blocked the ability of 500 ng/ml Slit2-N to act as a chemoattractant (Fig. 6B). For unknown reasons, the 0–500 ng/ml Slit2-S gradient increased the speed of neutrophils from the donors used for this experiment (Supplemental Fig. 4C). In the presence or absence of Slit2, ML141 also increased the speed of cells (Supplemental Fig. 4C) and, in the absence of Slit2, slightly increased the directionality of neutrophils (Supplemental Fig. 4D). These results suggest that blocking Cdc42 activity reverses Slit2-S chemorepulsion and inhibits Slit2-N chemoattraction.

Rac is another small GTPase involved in neutrophil chemotaxis (15, 59). The Rac inhibitor NSC23766 inhibits Rac activation (60). Although 10 μM NSC23766 only causes a partial inhibition of Rac in neutrophils (61), we used it at 10 μM because higher concentrations of this inhibitor have off-target effects (62). NSC23766 blocked the ability of Slit2-S to act as a chemorepellent (Fig. 6C). Although NSC23766 appeared to reduce the ability of Slit2-N to act as a chemoattractant, the effect was not statistically significant. NSC23766 in the presence or absence of Slit increased cell speed (Supplemental Fig. 4E), and the combination of NSC23766 and a Slit2-N gradient increased directionality (Supplemental Fig. 4F). Together, these results suggest that Rac may mediate the ability of Slit2-S to act as a chemorepellent.

Son of sevenless (Sos) is a guanine nucleotide exchange factor that can activate small GTPases such as Ras (63). In Drosophila, Slit activation of Robo in neurons causes recruitment of Sos to the plasma membrane (63). Ras mediates neutrophil chemotaxis (1, 2, 64). Guanine nucleotide-releasing factor hSos1 binds to Grb2 and links receptor tyrosine kinases to Ras signaling (65, 66). RIP corresponds to a region within human Sos1 that interacts with an SH3 domain of Grb2, preventing an interaction that is essential for Ras activation (65), and RIP has been used at 10 μM to inhibit Ras in neutrophils (66). RIP blocked the ability of Slit2-S to act as a chemorepellent but did not appear to affect the chemoattraction in a 0–500 ng/ml Slit2-N gradient (Fig. 6D). RIP alone or in the presence of Slit2-S, but not in the presence of Slit2-N, increased neutrophil speed (Supplemental Fig. 4G), and RIP had no significant effect on directness (Supplemental Fig. 4H). To confirm the role of Ras in Slit2-S mediated chemorepulsion, GTP-bound Ras was measured in neutrophils incubated with Slit2-N, Slit2-S, or fMLF. After 6 min, compared with unstimulated cells, Slit2-S and fMLF increased levels of GTP-bound Ras, whereas Slit2-N did not significantly affect levels of GTP-bound Ras (Fig. 7B, 7C). Together, these data suggest that Slit2-S increases levels of GTP-bound Ras and needs Ras to cause chemorepulsion, whereas Slit2-N does not significantly affect levels of GTP-bound Ras and does not need Ras to cause chemoattraction.

We found that two isoforms of the N-terminal fragment of Slit2 differentially regulate neutrophil chemotaxis, with Slit2-N directly acting as a chemoattractant, and Slit2-S directly acting as a chemorepellent. Both Slit2 fragments appeared to require Robo1 and the coreceptor Syndecan-4 to direct neutrophil movement. Slit2-N chemoattraction appeared to not require activation of Ras and Rac, was blocked by a Cdc42 inhibitor, increased PIP3 levels, and was reversed by a PI3K inhibitor. In contrast, Slit2-S–induced chemorepulsion was independent of PI3K, activated Ras, was blocked by Ras and Rac inhibitors, and was reversed by a Cdc42 inhibitor, indicating that, unexpectedly, the two Slit2 fragments activate different signal transduction pathways. Similar to the canonical neutrophil chemoattractant fMLF, a Slit2-N gradient induced a single zone of F-actin at the leading edge, increased cell adhesion, and altered cell morphology, whereas Slit2-S gradients caused multiple zones of F-actin accumulation and did not increase adhesion or alter cell morphology. Slit2-S thus causes chemorepulsion using an unusual effect on the cytoskeleton. Although Slit2-C did not directly induce chemotaxis, it increased cell speed, indicating that Slit2-C can modulate neutrophil chemotaxis.

Slit2-S induces chemorepulsion using a pathway involving Ras and Rac but not PI3K. In agreement with the observation that Slit2-S does not appear to activate PI3K, Slit2S decreases the phosphorylation of AKT substrates. The ability of a Cdc42 inhibitor to cause Slit2-S to act as a chemoattractant suggests that Slit2-S can activate an unknown chemoattraction pathway but preferentially activates a chemorepulsion pathway that requires Cdc42. This chemoattraction pathway may be the Ras- and Rac-independent pathway used by Slit2-N. Conversely, the ability of a PI3K inhibitor to cause Slit2-N to act as a chemorepellent suggests that Slit2-N can activate a chemorepulsion pathway but preferentially activates a chemoattraction pathway that requires PI3K.

The ability of Slit2-N to act as a neutrophil chemoattractant at low concentrations but act as a chemorepellent at high concentrations (5000 ng/ml; ∼36 nM) is similar to the effects of many other chemotactic factors, which act as chemoattractants at low concentrations but either prevent migration (by inducing increased adhesion) or act as chemorepellents (fugetaxis) at high concentrations (36, 67, 68). At concentrations of 1000–5000 ng/ml, Slit2 inhibits directed cell migration and activation of intracellular signaling pathways in many cell types (13, 15, 69, 70). An intriguing possibility is that a gradient of Slit2-N that starts at a high concentration may attract cells to tissues but create an exclusion zone near the source of the Slit2-N.

We thank the volunteers who donated blood, the phlebotomy staff at the Texas A&M Beutel Student Health Center, and Ramesh Rijal for helpful discussions.

The authors have no financial conflicts of interest.