Inflammation Differential Srgap1 Expression During Lung Repulsive Neutrophil Chemotaxis Through Slit2 Regulates Attractive Eosin

The Journal of Immunology

Slit2 Regulates Attractive Eosinophil and Repulsive Neutrophil Chemotaxis through Differential srGAP1 Expression during Lung Inﬂammation

Bu-Qing Ye,* Zhen H. Geng,† Li Ma,† and Jian-Guo Geng†

Directional migration of leukocytes is an essential step in leukocyte trafficking during inflammatory responses. However, the molecular mechanisms governing directional chemotaxis of leukocytes remain poorly understood. The Slit family of guidance cues has been implicated for inhibition of leuocyte migration. We report that Clara cells in the bronchial epithelium secreted Slit2, whereas eosinophils and neutrophils expressed its cell-surface receptor, Robo1. Compared to neutrophils, eosinophils exhibited a significantly lower level of Slit-Robo GTPase-activating protein 1 (srGAP1), leading to activation of Cdc42, recruitment of PI3K to Robo1, enhancment of eotaxin-induced eosinophil chemotaxis, and exaggeration of allergic airway inflammation. Notably, OVA sensitization elicited a Slit2 gradient at so-called bronchus–alveoli axis, with a higher level of Slit2 in the bronchial epithelium and a lower level in the alveolar tissue. Aerosol administration of rSlit2 accelerated eosinophil infiltration, whereas i.v. administered Slit2 reduced eosinophil deposition. In contrast, Slit2 inactivated Cdc42 and suppressed stromal cell-derived factor-1a–induced chemotaxis of neutrophils for inhibiting endotoxin-induced lung inflammation, which were reversed by blockade of srGAP1 binding to Robo1. These results indicate that the newly identified Slit2 gradient at the bronchus–alveoli axis induces attractive PI3K signaling in eosinophils and repulsive srGAP1 signaling in neutrophils through differential srGAP1 expression during lung inflammation. The Journal of Immunology, 2010, 185: 6294–6305.

eukocytes are recruited to the site of infection or tissue of cell adhesion molecules (4), whereas chemotaxis is mediated by injury as part of the inflammatory response of the innate several large families of chemokines and chemoattractants (5–7). L immune system. Directional migration of leukocytes Slit2, a member of the Slit family of secreted migratory cues, involves cellular interactions that are precisely regulated by tem- binds to Robo1, a prototype of the Roundabout family (Robo1–4) poral and spatial presentation of molecules on the surface of mi- of transmembrane cell-surface receptors. Engagement of Robo1 grating cells and their substrates. The cellular interactions that by Slit2 functions as a repellent in axon guidance and neuronal define the different steps of leukocyte recruitment include tether- migration (8–11), an endogenous inhibitor of leukocyte chemo- ing (initial attachment), rolling, weak and firm adhesion (arrest), taxis (12–17), and a chemoattractant of vascular endothelial cells transendothelial migration, and chemotaxis (1, 2). Among these, during vasculogenesis and angiogenesis (18–21). In addition, Slit2 tethering, rolling, and weak adhesion of leukocytes are thought to modulates migration of malignant tumor cells (22–24). be mediated mainly by the binding of selectins (CD62) to their Intracellular molecules downstream of Robo signaling have cognate glycoprotein ligands (3). Firm adhesion is mediated emerged as key regulators for determining the repulsive or at- mainly by the interaction of integrins with the Ig superfamily tractive effect of Slit on the targeting cells. Robo utilizes a variety of different signaling components in different cell types, such as Mena/Ab1 (25) and Calmodulin and Son of sevenless (26). The particular repertoire of intracellular signaling components deter- *Laboratory of Molecular Cell Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shang- mines the unique migratory response to Slit2 in cells of the same hai, China; and †Vascular Biology Center and Division of Hematology, Oncology and or different type. One signaling component that has emerged re- Transplantation, Department of Medicine, University of Minnesota School of Med- cently is the family of small Rho GTPases, particularly RhoA, icine, Minneapolis, MN 55455 Rac1, and Cdc42, which critically regulates actin cytoskeleton in Received for publication May 18, 2010. Accepted for publication September 11, 2010. guiding the directional migration of mammalian cells (27). The This work was supported by grants from the National Science Foundation of GTP-bound forms of Rho GTPases are active, whereas the GDP- China (30901302 to B.-Q.Y.), the Ministry of Science and Technology of China bound forms are inactive. The activities of Rho GTPases are (2010CB529702 to B.-Q.Y.), and the National Institutes of Health (RO1AI064743 modulated by GTPase-activating proteins (GAPs), which increase and RO1CA126897 to J.-G.G.). intrinsic GTPase activities, or guanine nucleotide exchange fac- Address correspondence and reprint requests to Dr. Jian-Guo Geng at the current address: Department of Biologic and Materials Sciences, University of Michigan tors, which exchange the GDP on a GTPase for GTP. Upon en- School of Dentistry, A323 MSRB III, 1150 West Medical Center Drive, Ann Arbor, gagement of Robo1 with Slit2, srGAP1, the prototype of a GAP MI 48109-0632. E-mail address: [email protected] family that includes srGAP1, -2 and -3, directly binds to the in- The online version of this article contains supplemental material. tracellular CC3 motif of Robo1, which inactivates Cdc42 and Abbreviations used in this paper: Aml, Aml14.3D10; AS, 5-(2,2-difluoro-benzo[1,3] RhoA, but does not affect the activity of Rac1. This molecular dioxol-5-ylmethylene)-thiazolidine-2,4-dione; BALF, bronchial airway lavage fluid; mechanism induces repulsion of migratory neuronal cells from the EPO, eosinophils peroxidase; GAP, GTPase-activating protein; HA, hemagglutinin; mIgG, mouse IgG; NE, neutrophils elastase; SDF-1a, stromal cell-derived factor-1a; anterior subventricular zone of the forebrain (28). SH, Src homology; srGAP1, Slit-Robo GTPase-activating protein 1; Tg, transgenic; Another downstream effector of Robo is the family of PI3Ks. WT, wortmannin. Upon recruitment to the inner leaflet of the plasma membrane, p110 Copyright Ó 2010 by The American Association of Immunologists, Inc. 0022-1767/10/$16.00 phosphorylates phosphatidylinositol 4,5-bisphosphate on the D3 www.jimmunol.org/cgi/doi/10.4049/jimmunol.1001648 The Journal of Immunology 6295 position to yield phosphatidylinositol 3-5-trisphosphate. There is verse primer 59-GCAGCGGCCGCTCAGTGATGATGATGATGATGATC- a diverse set of proteins with pleckstrin homology domains that TGCC ATTTCTCCAGGACC-39. Postdigestion with XbaI/NotI, the insert bind to phosphatidylinositol 3-5-trisphosphate and, consequently, was ligated into the pVL1393 vector (BD Pharmingen, San Diego, CA), which was then verified by DNA sequencing. Recombinant human Slit2 are recruited to the plasma membrane upon activation of PI3Ks with an His tag was expressed in Sf9 insect cells and purified by Talon (29). These molecules are responsible for activation of a cohort of metal affinity chromatography (BD Clontech, Mountain View, CA), as different signal transduction pathways controlling cell growth, dif- described (33–35). Eluted Slit2 was concentrated with Ultracon (10 kDa ferentiation, proliferation, apoptosis, metabolism, migration, and cutoff; Millipore, Bedford, MA) and further purified by gel-filtration chromatography (A¨ KTA FPLC; GE Healthcare Life Sciences, Piscat- intracellular trafficking. We have previously shown that PI3K in- away, NJ) on a Superdex 200 column (GE Healthcare Life Sciences), using hibitors, WT and LY294002, inhibit Slit2-induced attractive mi- PBS (pH 7.4) as the running buffer. Isolated Slit2 was eluted as a single gration of vascular endothelial cells in a dose-dependent manner sharp peak that was homogenous when subjected to silver staining (data (18). not shown). Contaminated endotoxin in our preparations of Slit2 was Although the steps of leukocyte recruitment are well defined, the routinely removed by Detoxi-Gel Endotoxin Removing Gel (Thermo Scientific, Rockford, IL) until they were ,0.03 EU/ml determined by the molecular mechanisms underlying attractive versus repulsive mi- Limulus amebocyte lysate method. More than three separate Slit2 prepa- gration and directional, nondirectional, or random migration re- rations were used in our experimentation. main obscure (5–7). Much effort has been directed at identifying exogenous factors that can be used therapeutically to control in- Cell lines flammation (30, 31). Notably, Slit2 has been shown to inhibit HL-60 cells (CCL-240, American Type Culture Collection, Manassas, VA) migration of neutrophils, lymphocytes, and macrophages in in- and Aml14.3D10 (Aml) cells (kindly provided by Dr. Arne Slungaard, flammatory responses (12–17). In this study, we unexpectedly University of Minnesota, Minneapolis, MN) were cultured in RPMI found that Slit-Robo signaling activated not only repulsive 1640 medium (Invitrogen, Carlsbad, CA) supplemented with 10% heat- chemotaxis of neutrophils during endotoxin-induced lung in- inactivated FBS, 4 mM L-glutamine, 50 mM 2-ME (Aml cells; 36), 100 U/ml penicillin, and 100 mg/ml streptomycin at 37˚C in the presence of flammation, but also attractive chemotaxis of eosinophils during 5% CO2. allergic airway inflammation. To understand whether and how distinctive intracellular pathways downstream of Slit-Robo sig- Isolation of murine and human leukocytes naling differentially modulate leukocyte migratory responses to Mouse neutrophils and eosinophils were isolated from bone marrows or directional stimuli, we investigated the molecular mechanisms of peripheral blood of C57 and IL-5-Tg mice as described (32, 34, 37–39). how Slit-Robo signaling regulates directional migration of eosi- Human neutrophils were isolated from peripheral blood of healthy vol- nophils and neutrophils during allergic airway inflammation and unteers as described (34). Human circulating eosinophils were isolated by endotoxin-induced lung inflammation. negative magnetic selection from peripheral blood according to the man- ufacturer’s protocol. The purity of mouse neutrophils and eosinophils (∼90–92%) and human eosinophils and neutrophils (∼95%) was determined by Wright’s-Giemsa staining (Supplemental Fig. 1). The endotoxin Materials and Methods levels in all buffers used were ,0.03 EU/ml determined by the Limulus Reagents amebocyte lysate method. The use of human blood was approved by the Institutional Review Board of the Shanghai Institute for Biological Sci- The Slit2 and Robo1 mAbs and polyclonal Abs were prepared and char- ences, Chinese Academy of Sciences. acterized as described before (18). The Abs against Cdc42, srGAP1, GFP, hemagglutinin (HA), CC10, eosinophils peroxidase (EPO), and neutrophils elastase (NE) were purchased from Santa Cruz Biotechnology (Santa Cruz, Immunoblotting and immunoprecipitation CA). The Abs of His tag and a-tubulin and bacterial LPSs (serotype 055: For comparing the levels of srGAP1 expression, human Aml and HL- B5) were purchased from Sigma-Aldrich (St. Louis, MO). The GAPDH 60 cells (both at 4 3 106 cells/aliquot), mouse eosinophils and neutrophils Ab was purchased from Proteintech Group (Chicago, IL). The p85 pAb (both at 7 3 106 cells/aliquot), or human eosinophils and neutrophils (both was purchased from Upstate Biotechnology (Lake Placid, NY). The anti- at 2 3 106 cells/aliquot) were lysed with the ice-cold lysis buffer (50 mM Akt and anti-phosphorylated Akt Abs were purchased from Cell Signaling HEPES [pH 7.4], 150 mM NaCl, 1% Triton X-100, 5 mM EDTA, 10% Technology (Danvers, MA). The primers and premix reagents for quanti- glycerol, 2 mM PMSF, 20 mg/ml aprotinin, 20 mg/ml leupeptin, 10 mg/ml tative PCR were purchased from SA Biosciences (Frederick, MD). The pepstanin A, and 150 mM benzamidine) on ice for 30 min. After brief anti-rabbit IgG-Cy5 and anti-goat IgG-Cy3 were purchased from Jackson centrifugation, the lysates were immunoblotted with the srGAP1 Ab. ImmunoResearch Laboratories (West Grove, PA). Recombinant mouse Alternatively, human Aml and HL-60 cells or mouse eosinophils and SDF-1a was purchased from RDI (Fitzgerald Industries, Concord, MA). neutrophils were starved with serum-free RPMI 1640 medium at 37˚C for Recombinant mouse eotaxin and the mouse CXCL1/CXCL2 quantitative 1 h. They were then stimulated with Slit2 (0.4 mg/ml in PBS) or PBS alone ELISA kit were purchased from R&D Systems (Minneapolis, MN). The at 37˚C for 2 min for immunoprecipitation (5 3 106 cells/aliquot) and for mouse IgE quantitative ELISA kit was purchased from Bethyl Laborato- 5 min for determining Akt phosphorylation (3 3 106 cells/aliquot), without ries (Montgomery, TX). Wortmannin (WT) and 5-(2,2-difluoro-benzo[1,3] or with preincubation with R5 or isotype-matched irrelevant mouse IgG dioxol-5-ylmethylene)- thiazolidine-2,4-dioe (AS605240, or AS) were (mIgG; both at 0.5 mM) for 30 min. The lysates were used for coimmu- purchased from EMD Calbiochem (Gibbstown, NJ). The EasySep human noprecipitation followed by immunoblotting with their respective Abs, as eosinophil enrichment kit was purchased from StemCell Technologies previously described (34). (Vancouver, British Columbia, Canada). In addition, mouse lung tissues were homogenized in 1 ml RIPA lysis buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 0.5% deoxycholid acid, Mice 0.1% SDS, 5 mM EDTA, 2 mM PMSF, 20 mg/ml aprotinin, 20 mg/ml C57BL6/J (C57) mice were purchased from The Jackson Laboratory (Bar leupeptin, 10 mg/ml pepstanin A, 150 mM benzamidine, and 1% Nonidet Harbor, ME). IL-5-transgenic (Tg) mice (NJ1638 strain; 32) were gener- P-40) in a Dounce tissue homogenizer, followed by centrifugation at ously provided by Dr. James J. Lee (Division of Pulmonary Medicine, 12,000 3 g at 4˚C for 10 min to remove tissue debris and immunoblotting Department of Biochemistry and Molecular Biology, Mayo Clinic Ari- as described above. zona, Scottsdale, AZ). The Slit2-Tg mice were generated according to standard procedures and characterized as described (33). Mouse experi- TAT peptide construction and purification ments were approved by the Institutional Animal Committees of Shanghai Institutes for Biological Sciences (Shanghai, China). We designed polypeptides encoding the Src homology (SH) 3 region of wild-type srGAP1 (amino acid residues of 674–1022) and its mutant, 758 760 761 Slit2 construction and expression in which the amino acid residues corresponding to Ser , Arg , Glu , Trp780, and Leu791 in the P3 loop of the SH3 domain were mutated to Ala The human Slit2 cDNA (1–2670 bp) was amplified using the forward (40). They were expressed and isolated using methods identical to those primer 59-ACCTTCTAGAATGCGCGGCGTTGGCTGGC-39 and the re- described before (34). 6296 Slit2 DIRECTS LEUKOCYTE CHEMOTAXIS

GST pulldown assay (Stratagene, La Jolla, CA). For comparing the level of srGAP1 expression, Aml and HL-60 cells (5 3 106 cells/aliquot), human eosinophils and Recombinant human Robo1 CC3 region (encoding the aa sequence neutrophils (2 3 106 cells/aliquot), or mouse eosinophils and neutrophils 1454–1657) fused with GST (GST-Robo1-CC3) and GST alone in the (1 3 107 cells/aliquot) were lysed and 2 mg (Aml and HL-60 cells or vector of pGEX-4T-1 (GE Healthcare Life Sciences) were expressed in mouse eosinophils and neutrophils) and 50 ng (human eosinophils and Escherichia coli and purified using Glutathione Sepharose 4B beads (GE neutrophils) total RNA per aliquot were used for quantitative and real-time Healthcare Life Sciences). Isolated GST-Robo1-CC3 or GST was in- RT-PCR analysis. cDNAs were synthesized using the ThermoScript RT- cubated with the polypeptides of TAT-SH3 and TAT-SH3M. After washing PCR system (Invitrogen). Quantitative PCR analysis and data collection extensively, the beads were boiled in the SDS sample loading buffer and were performed on the Mx3000 qPCR System (Stratagene) using the subjected to 12% SDS-PAGE followed by immunoblotting. primer pairs and the RT2 SYBR Green PCR Master Mix from SA Bio- Chemotactic assay sciences. All quantitations were normalized to an endogenous b-actin or GAPDH control. The relative quantitation value for each target gene The microfluidic chamber (m-slide VI, Ibidi, Mu¨nchen, Germany) for compared with the calibrator for that target is expressed as 2 2 (Ct 2 Cc) generation and maintenance of a stable gradient of any chemokines for up (Ct and Cc are the mean threshold cycle differences after normalizing to to 30 min was employed to monitor neutrophil and eosinophil chemotaxis b-actin/GAPDH). The relative expression levels of samples are presented in real time (41). The condition medium of stable HEK293 cell lines using a semilog plot. expressing human Slit2 or the plain vector (28) was collected by culturing these confluent cell monolayers with serum-free DMEM medium at 37˚C Allergic airway inflammation for 12 h. Following precoating with 10% murine plasma for 30 min and washing three times with HBSS, mouse neutrophils or eosinophils (1 3 C57 and Slit2-Tg mice (∼7 wk old) were sensitized with 20 mgOVA 6 10 /aliquot) were loaded into each channel. The Slit2 medium or the (Sigma-Aldrich) and 2 mg alum i.p. on days 0 and 5. Sham-immunized control medium was added into the input port, with or without an aliquot mice were received alum alone. From days 12 to 14, mice were aerosol (1 ml) of mouse SDF-1a (25 mg/ml) or eotaxin (100 mg/ml) in the presence challenged with 1% OVA in saline (each time for 1 h, twice per day of mIgG or R5 (both at 0.5 mM), DMSO, WT, or AS (both at 150 nM). The separated for 4 h) for 3 d (43), with or without i.v. injection of R5 or mIgG chamber was transferred to a multidimensional live cell imaging work- or i.p. injection of WT or AS (all at 1 mg/g mouse body weight) prior to station (Leica AS MDW, Leica Microsystems, Wetzlar, Germany) equip- aerosol challenge. Alternatively, C57 mice were exposed to Slit2 aerosol (8 ped with a heated enclosure to keep the chamber at 37˚C. The migration mg/ml in PBS) for 1 h prior to OVA aerosol challenge for 3 d or given i.v. of cells was recorded at 340 original magnification with images taken Slit2 (5 ng/g mouse body weight) prior to the OVA aerosol challenge on every 5 s over a period of 20 min for neutrophils and 30 min for eosino- day 13 (day 2 of aerosol challenge; 13). On day 15 (24 h after the last phils. The resulting images were imported into Image J software (National aerosol challenge), mice were anesthetized followed by surgical exposure Institutes of Health, Bethesda, MD) and processed using a cell-tracking of the lungs and hearts. Tracheas were cannulated, and each lung was protocol. lavaged with 1.5 ml PBS. Leukocytes in BALFs and blood were counted Transwell assay and stained with Wright’s dye, and cell differentials were enumerated based on morphology and staining profile. For histologic studies, mice A 24-well Transwell plate (5 mm in pore size; Corning, Corning, NY) was were perfused with 10 ml PBS through the right ventricle to remove all coated with 2.5 mg/ml mouse fibrinogen and incubated at 37˚C for 1 h. blood followed by fixation with 3 ml 10% formalin, paraffin embedding, Freshly isolated mouse neutrophils (4 3 106/ml) were suspended in 50% and sectioning (5-mm thick). DMEM and 50% M199 medium supplemented with 5% heat-inactivated FBS. An aliquot (0.1 ml) of neutrophils, with or without preincubation Endotoxin-induced lung inflammation with the polypeptide of His-SH3, TAT-SH3, or TAT-SH3M (all at 50 mg/ ml) at 37˚C for 20 min in the presence of 5% CO2, was transferred into C57 and Slit2-Tg mice (8–10 wk) were aerosol challenged with 300 mg/ml the insert. The inserts were then placed into the wells containing 0.6 ml LPS in saline for 20 min (44), with or without prior i.v. injection of R5 or medium and SDF-1a (50 ng/ml), Slit2 (0.4 mg/ml), WT or AS (both at mIgG (1 mg/g mouse body weight). Eight hours after aerosol challenge, 150 nM). Alternatively, the plate was pretreated with 0.1% BSA in the mice were anesthetized, cannulated, and lavaged. Leukocytes and their RPMI 1640 medium at 37˚C for 1 h. An aliquot (0.1 ml) of mouse eosi- differentials were counted as described above. nophils (1 3 107/ml) suspended in the same medium was transferred into the insert. The inserts were then placed into the wells containing 0.6 ml Statistical analysis medium and eotaxin (0.5 mg/ml), Slit2 (0.4 mg/ml), WT, or AS (both at 150 nM). Postincubation at 37˚C for 2 h (for neutrophils) or 4 h (for Statistical significance was determined by Student t test. For multiple eosinophils), cells that had migrated through the filter into the lower wells comparisons, ANOVA test (Bonferroni post hoc) was employed. The p were collected and counted (12, 42). values ,0.05 and ,0.01 were considered statistically significant and very Cdc42 activity assay significant, respectively. An aliquot of cells (3 3 106 cells) were incubated with PBS alone or Slit2 (0.4 mg/ml in PBS) at 37˚C for 20 min and then lysed with 500 ml lysis buffer on ice for 30 min. Postcentrifugation at 12,000 rpm at 4˚C for 10 Results min, supernatants were incubated with GST-PBD fusion protein that had Differential expression of srGAP1 and their effects on Cdc42 bound to the Glutathione Sepharose beads at 4˚C for 4 h. After washing and PI3K activities three times, the bound proteins were boiled in the SDS sample loading buffer for 5 min followed by immunoblotting with the Cdc42 Ab (28). Slit2 reportedly inhibits chemotaxis of neutrophils, macrophages, and T lymphocytes (12–17). Although the mechanism by which Immunofluorescence staining Slit2 attenuates chemotaxis of these leukocytes is unknown, Slit2 Mouse leukocytes in bronchial airway lavage fluids (BALFs) were coated to has been shown to induce repulsion of migratory neuronal cells glass slides. Mouse lung tissues were fixed with 10% formalin, paraffin within the developing brain through a pathway involving Robo1, embedded, and sectioned (5-mm thick). Prior to immunostaining, samples were incubated with 10% BSA in 50 mM Tris-HCl (pH 7.6) and 0.15 M srGAP1, and Cdc42. We therefore investigated the activity of this NaCl (TBS) at 37˚C for 30 min to reduce nonspecific binding. They pathway during leukocyte chemotaxis using neutrophils and were then incubated with appropriate primary Abs in TBS containing 1% eosinophils. Using RT-PCR and quantitative RT-PCR, we found BSA at 4˚C overnight in a humidified chamber, followed by incubation with that human eosinophilic Aml cells and mouse eosinophils freshly respective fluorescence-conjugated secondary Abs. The nuclei were stained isolated from IL-5-Tg mice (32) expressed almost undetectable with DAPI (blue). All incubations were followed by washing three times in TBS over 15 min. The immunofluorescent staining was observed under srGAP1 mRNA (Fig. 1A,1B) and protein (Fig. 1C) compared with a Leica-SP2 laser scanning confocal microscope (Leica Microsystems). human promyeloid HL-60 cells and primary mouse neutrophils. Consistently, a significantly lower level of srGAP1 mRNA and RT-PCR protein expression was also detected in human circulating eosi- Total RNAs from mouse lung tissues, isolated primary leukocytes, and nophils as compared with human circulating neutrophils (Fig. 1D, cultured cell lines were extracted with the Absolutely RNA miniprep Kit 1E). As predicted, recombinant human Slit2 induced srGAP1 The Journal of Immunology 6297

FIGURE 1. srGAP1 expression and Robo1 downstream signaling in eosinophils and neutrophils. RT-PCR (A) and quantitative real-time RT-PCR (B) analysis of srGAP1 mRNA in human eosinophilic Aml cells, human promyeloid HL-60 cells, mouse eosinophils, and mouse neutrophils. C, Immuno- blotting of srGAP1 and a-tubulin in Aml cells, HL-60 cells, mouse eosinophils, and neutrophils. D and E, The quantitative real-time PCR and immunoblotting analysis of srGAP1 expression in human eosinophils and neutrophils. F, Effects of Slit2 on srGAP1 binding to Robo1. Endogenous srGAP1 was coimmunoprecipitated with the Robo1 Ab from the lysates of HL-60 cells preincubated with PBS (–) or Slit2, followed by immunoblotting with the Abs srGAP1 or Robo1. G, Measurement of Cdc42 activity in Aml and HL-60 cells preincubated with PBS or Slit2. H, Effects of Slit2 on p85 binding to Robo1. Endogenous p85 was coimmunoprecipitated with the Robo1 Ab from the lysates of Aml cells, HL-60 cells, mouse eosinophils, and neutrophils that had pretreated with PBS (–) or Slit2, followed by immunoblotting with the Abs to p85 and Robo1. I, Immunoprecipitation of p85–Robo1 complex. Endogenous Robo1 was coimmunoprecipitated with the p85 mAb from the lysates of Aml cells preincubated with PBS or Slit2, followed by immunoblotting with the Abs to Robo1 and p85. J, Determination of Slit2-induced PI3K activation. Total and phosphorylated Akt in Aml and HL-60 cells preincubated with PBS or Slit2 was detected by immunoblotting with their respective Abs. K, The time course of Slit2-induced Akt phosphorylation in Aml cells. L, Akt phosphorylation in mouse eosinophils and neutrophils preincubated with PBS or Slit2. M, Inhibitory effect of R5 or mIgG (both at 0.5 mM) on Slit2-induced Akt phosphorylation in Aml cells. N, Measurements of Cdc42 activity and srGAP1 expression in DMSO or WT-pretreated Aml cells. Data represent two or three experiments or the mean 6 SD of two to four independent experiments. pp , 0.05; ppp , 0.01. binding to Robo1 (Fig. 1F) and resulted in inactivation of Cdc42 elicited Ser473 phosphorylation of Akt in mouse eosinophils (Fig. (Fig. 1G) in HL-60 cells, but not in Aml cells. 1L, upper left panel). In contrast, Slit2 failed to induce Ser473 We have previously shown that the PI3K inhibitors, WT and phosphorylation of Akt in HL-60 cells (Fig. 1J, upper right panel) LY294002, inhibit Slit2-induced attractive migration of vascular or mouse neutrophils (Fig. 1L, upper right panel). Total Akt was endothelial cells in a dose-dependent manner (18). We thus hy- also determined as a sample loading control (Fig. 1J–L, lower pothesized that Slit2 might also regulate eosinophil migration panels). Preincubation of Aml cells with R5, a mouse mAb to the through PI3K signaling. Indeed, Slit2 induced the Robo1 Ab co- first Ig domain of Robo1 that neutralizes the interaction of Robo1 immunoprecipitation of the p85 subunit of PI3K in Aml cells or with Slit2 (18–21), but not its isotype-matched irrelevant control mouse eosinophils, but not in HL-60 cells or mouse neutrophils mIgG, prevented Slit2-induced Ser473 phosphorylation of Akt in (Fig. 1H,1I). We next examined whether Slit2 might activate Akt, Aml cells (Fig. 1M). Interestingly, the srGAP1 level in Aml cells as evidenced by Akt phosphorylation at the amino acid residue of was apparently so low that inhibition of PI3K by WT was not Ser473. Preincubation of Aml cells with Slit2 triggered Ser473 enough to rescue the srGAP1-mediated Cdc42 inactivation (Fig. phosphorylation of Akt (Fig. 1J, upper left panel), which occurred 1N). It is thus apparent that due to the differential level of srGAP1 in a time-dependent manner (Fig. 1K, upper panel). Slit2 also expression in leukocytes—that is, a higher level of srGAP1 6298 Slit2 DIRECTS LEUKOCYTE CHEMOTAXIS expression in neutrophils versus a lower level of srGAP1 ex- leukocyte chemotaxis depending on the intracellular machinery, pression in eosinophils—Slit2 recruits srGAP1 to Robo1 for in- with Slit2 signaling through PI3K for increasing eotaxin-mediated activation of Cdc42 in neutrophils while recruiting PI3K to Robo1 eosinophil chemotaxis, whereas through srGAP1 for decreasing for activation of PI3K signaling in eosinophils. As Cdc42 and SDF-1a–mediated neutrophils chemotaxis. PI3K activities are known to critically determine Slit2-induced cell migration (18, 28), our results implicate that Slit2, by mod- srGAP1 involvement in Slit2 modulation of neutrophil ulation of Cdc42 and PI3K activities, may differentially alter chemotaxis chemotaxis of eosinophils and neutrophils in response to chemo- To verify our hypothesis regarding to srGAP1, we decided to test kines and chemoattractants. whether blockade of srGAP1 binding to the intracellular CC3 motif of Robo1 could neutralize the inhibitory effect of Slit2 on SDF-1a– Slit2 modulation of chemotaxis mediated chemotaxis of neutrophils. The crystal structure of the On the basis of above biochemical studies, we employed a micro- srGAP1-Robo1 complex indicates that the srGAP1 amino acid fluidic chamber (m-slide, Ibidi), which generates a stable gradient residues of Ser758, Arg760, Glu761,Trp780, and Leu791 in the P3 for up to 30 min, for real-time monitoring of chemotactic mi- loop of the SH3 domain are essential for recognition of the Robo1 gration of leukocytes in response to a Slit2 gradient (41). As ex- intracellular CC3 motif (40). We thus designed the polypeptides pected, conditioned medium from 293 cells overexpressing human encoding the wild-type SH3 domain (674–1022 aa residules) of Slit2 (the Slit2 medium), but not conditioned medium from 293 srGAP1 (TAT-SH3) or an SH3 domain mutant, srGAP1S758A, cells transfected with the plain vector (the control medium) (28), R760A, E761A, W780A, L791A (TAT-SH3M), both of which were fused polarized mouse and human eosinophils, causing them to spread with a TAT sequence to facilitate incorpotation into the cytoplasm and migrate in a random and nondirectional manner (Fig. 2A, of mammalian cells (Fig. 3A) (34). The polypeptides were also Supplemental Table I, Supplemental Videos 1, 3). In contrast, fused with the His tag for purification with the nickle beads. We mouse and human neutrophils retained their quiescent state when also designed a polypeptide encoding the wild-type SH3 domain exposed to Slit2 (Fig. 2A, Supplemental Table I, Supplemental without TAT (His-SH3) as a negative control. Following expres- Videos 2, 3). sion and isolation of these polypeptides, we found that, compared We next tested whether Slit2 could act with certain well- with the GST beads, the GST-Robo1 CC3 motif protein-bound characterized chemokines for modulation of leukocyte chemo- beads pulled down TAT-SH3, but not TAT-SH3M, as detected taxis. Indeed, mouse eosinophils migrated toward an eotaxin by the anti-His tag Ab (Fig. 3B, lower panel). The loaded poly- gradient, whereas mouse neutrophils migrated toward an SDF-1a peptides of TAT-SH3 and TAT-SH3M were also detected by the gradient (Fig. 2B, Supplemental Table I, Supplemental Videos 1, anti-His tag Ab (Fig. 3B, upper panel). These polypeptides did 2). Notably, the speed of eotaxin-mediated eosinophil migration not affect the survival of isolated mouse neutrophils as measured was slower than that of SDF-1a–mediated neutrophil migration. by the MTT assay (data not shown) (18, 21, 46). As expected, Importantly, the Slit2 medium, but not the control medium, in- TAT-SH3, but not TAT-SH3M, inhibited srGAP1 binding to creased eotaxin-mediated eosinophil migration while decreasing HA-Robo1 (Fig. 3C) and prevented Slit2-induced Cdc42 in- SDF-1a–mediated neutrophil migration. activation (Fig. 3D). Direct immunoblotting of HA-Robo1, GFP- To determine the net results of leukocyte chemotaxis over srGAP1 (Fig. 3C), and total Cdc42 (Fig. 3D) was used as sample a period of 1 h, we used a traditional Transwell assay consisting of loading controls. These polypeptides were directly conjugated two chambers separated by a semipermeable membrane. Eosino- with FITC followed by incubation with mouse neutrophils. phils or neutrophils were added to the upper chamber while eotaxin Compared to the FITC-conjugated His-SH3, the cytoplasmic lo- or SDF-1a was added to the lower chamber. Leukocytes that had calization of FITC-conjugated TAT-SH3 and TAT-SH3M was migrated from the upper chambers to the lower chambers were evident (Fig. 3E). Importantly, TAT-SH3, but not TAT-SH3M or then analyzed. As expected, eotaxin augmented eosinophil che- His-SH3, prevented Slit2 from inhibiting SDF-1a–mediated neu- motaxis (Fig. 2C), whereas SDF-1a enhanced neutrophil chemo- trophil chemotaxis (Fig. 3F). Our results indicate that, upon Slit2 taxis (Fig. 2D). Although it alone did not induce chemotaxis of binding to Robo1, srGAP1 inactivates Cdc42 and consequently eosinophils or neutrophils, Slit2 significantly enhanced eotaxin- inhibits SDF-1a–mediated chemotaxis of neutrophils. induced eosinophil chemotaxis and mitigated SDF-1a–induced neutrophil chemotaxis. These in vitro findings support to our hy- Slit2 expression in Clara cells and Robo1 expression on pothesis that srGAP1, which is downstream of Slit-Robo signal- eosinophils and neutrophils ing, may determine chemokine-mediated directional migration of As eosinophils and neutrophils are essential to the pathogenesis of eosinophils and neutrophils in response to Slit2. OVA-induced allergic airway inflammation and endotoxin-induced We next tested whether PI3K inhibitors, WT (a pan-inhibitor; lung inflammation (43, 44), we examined the expression profiles 18) or AS (a selective inhibitor for p110g; 45), might suppress of Slit2 and Robo1 in normal and inflammatory murine tissues and Slit2-induced eosinophil polarization and chemotaxis. We found leukocytes. Consistent with the previous reports of Slit2 expres- that in addition to R5, WT and AS, but not mIgG and DMSO, also sion in developing mouse lung (47, 48), we detected Slit2 ex- attenuated Slit2-induced eosinophil activation, such as spreading, pression on nonciliated secretory Clara cells (CC10-positive), but motility, and random and nondirectional migration (Fig. 2E, Sup- not on cilia cells (tubulin IV-positive), of normal bronchial epi- plemental Table I, Supplemental Video 4). Additionally, WT and thelium (Fig. 4A). Robo1 expression was detected on the cell AS neutralized of the ability of Slit2 to augment eotaxin-mediated surfaces of freshly isolated mouse eosinophils (positive for EPO) chemotaxis of mouse eosinophils (Fig. 2F). In contrast, WT and and neutrophils (positive for NE) (Fig. 4B). Importantly, the ex- AS failed to prevent the ability of Slit2 to inhibit SDF-1a– pression of Slit2 mRNA and protein was dramatically upregulated, mediated chemotaxis of mouse neutrophils (Fig. 2G). Our results peaking at 12–36 h following aerosol challenge with OVA (Fig. indicate that, upon Slit2 binding to Robo1, PI3K is recruited to 4C,4E). In contrast, no obvious upregulation of Slit2 mRNA and Robo1, which phosphorylates Akt and consequently increases protein expression was detected in lung tissues of endotoxin- eotaxin-mediated eosinophil chemotaxis. Taken together, these induced lung inflammation, even though the basal level of Slit2 in vitro results suggest that Slit2 has opposite biological effects on mRNA and protein expression was clearly visible (Fig. 4D,4F). In The Journal of Immunology 6299

FIGURE 2. Chemotactic modulation of Slit2 on eosinophils and neutrophils. A, Real-time monitoring of activated mouse eosinophils (30 out of 34) and mouse neutrophils (2 out of 25) in response to a Slit2 gradient. B, Real-time monitoring of migrating mouse eosinophils (15 out of 29 for eotaxin alone and 8 out of 12 for eotaxin plus Slit2) and mouse neutrophils (16 out of 22 for SDF-1a alone and 3 out of 26 for SDF-1a plus Slit2) in a m-Slide chemotaxis assay in response to the indicated gradients. Eotaxin and SDF-1a gradients were higher in the left side and lower in the right side of the image. Scale bars, 20 mm for A and B. Migration of eosinophils (C) and neutrophils (D) in a Transwell chamber. E, Preventive effects of R5 (2 out of 13), WT (2 out of 10), mIgG (10 out of 20), DMSO (9 out of 15) or AS (3 out of 18) on Slit2-triggered activation of mouse eosinophils. Scale bar, 20 mm. Effects of WT and AS on eotaxin-mediated eosinophil chemotaxis potentiated by Slit2 (F) and SDF-1a–mediated neutrophil chemotaxis suppressed by Slit2 (G) in a Transwell chamber. Data represent two or three experiments or the mean 6 SD of two to four independent experiments. pp , 0.05; ppp , 0.01. addition, no clear upregulation of Robo1 was detected in eosi- Slit2 gradient at the bronchus–alveoli axis nophils isolated from OVA-sensitized mice or neutrophils isolated As nonciliated secretory Clara cells in the bronchial epithelium from LPS-challenged mice (data not shown). Given that Slit2 acts secrete Slit2 (Fig. 4A), and aerosol challenge with OVA markedly synergistically with eotaxin for promoting eosinophil chemotaxis increases Slit2 expression (Fig. 4C,4E), we suspected that OVA- while suppressing SDF-1a–mediated neutrophil chemotaxis in sensitization might elicit a Slit2 gradient at the bronchus–alveoli vitro (Fig. 2), our findings of Slit2 expression in nonciliated se- axis—that is, the amount of secreted Slit2 might be higher in the cretory Clara cells and Robo1 expression on eosinophils and bronchial epithelium and lower in the alveolar tissues and their neutrophils suggest that Slit-Robo signaling may regulate chemo- supplying blood vessels. To test this hypothesis, we examined taxis of eosinophils and neutrophils in vivo, especially during al- Slit2 expression in OVA-sensitized lung tissues. Immunofluores- lergic airway inflammation when Slit2 expression is upregulated. cent staining of Slit2 was elevated at ∼12 h, peaking at ∼36 h and 6300 Slit2 DIRECTS LEUKOCYTE CHEMOTAXIS

FIGURE 3. Role of srGAP1 in Slit2 regulation of neutrophils chemotaxis. A, Design of the polypeptides encoding the srGAP1 SH3 domain and its mutant fused with the His tag and TAT sequence. B, GST-Robo1-CC3 pulldown of TAT-SH3, but not TAT-SH3M. C, Inhibitory effects of TAT-SH3 or TAT-SH3M on GFP-srGAP1 binding to HA-Robo1. Neutralizing effect of TAT-SH3, TAT-SH3M, or His-SH3 on Slit2-induced inactivation of Cdc42 (D) and inhibition of SDF-1a–mediated chemotaxis (F) of mouse neutrophils. E, Cytoplasmic incorporation of the FITC-conjugated TAT-SH3, TAT-SH3M, or His-SH3 in mouse neutrophils. Data are the mean 6 SD of three independent experiments. ppp , 0.01. gradually declining at ∼60 to ∼80 h (Fig. 4G), which is consistent pCMV promoter for efficient, but nonselective, expression of with the expression profile of Slit2 determined by quantitative human Slit2 in mice (33). The availability of Slit2-Tg mice allows RT-PCR (Fig. 4C) and immunoblotting (Fig. 4E). Importantly, us to investigate the functional importance of Slit2 in systemic Slit2 expression was higher in the bronchi and lower in sur- inflammation in vivo. Compared to C57 mice, the Flag tag in the rounding alveolar tissues, providing experimental evidence for the lung extracts of Slit2-Tg mice was detected by PCR, using its suspected Slit2 gradient at the bronchus–alveoli axis following specific primer pair (Supplemental Fig. 2A, left upper panel) (33). aerosol challenge with OVA (Fig. 4G). b-actin was used as a sample loading control (Supplemental Fig. 2A, left lower panel). For detection of the Slit2 transgene protein, Characterization of Slit2-Tg mice detergent extracts of whole lungs from C57 mice and Slit2-Tg Slit2 knockout mice are either embryonic lethal or die within 1 or mice were immunoprecipitated with S1 (an anti-Slit2 mAb) or 2 wk postbirth (49). We therefore used Slit2-Tg mice with the mIgG and immunoblotted with M2 (an anti-Flag mAb). Compared

FIGURE 4. Expression of Slit2 and Robo1 in lung tissues and leukocytes. A, The expression of Slit2 in CC10-positive Clara cells, but not tubulin IV-positive cilia cells, of the bronchial epithelium. B, Robo1 expression (green) in mouse eosinophils (EPO-positive; red) and mouse neutrophils (NE- positive; red). DAPI (blue) was used for staining of cell nuclei. qRT-PCR (C, D) and immunoblotting (E, F) analysis of Slit2 mRNA and protein. Lysates were extracted from lung tissues after aerosol challenge with OVA (C, E) or LPS (D, F) at the indicated time points. G, Immunofluorescent staining of Slit2 in OVA-sensitized mouse lung tissues at the indicated time points. A, B,andG, Original magnification 3630; scale bar, 10 mm. Data are repre- sentative of two to three experiments. The Journal of Immunology 6301 to C57 mice, the expression of Flag-Slit2 fusion protein was bronchial tissues was negative unless mice were aerosol challenged detected in the lung extracts of Slit2-Tg mice (Supplemental Fig. with OVA (Fig. 5A, right upper panels). Compared to C57 mice, 2B, right panel). These data attest to the successful overexpression OVA-sensitized Slit2-Tg mice had profoundly more accumulation of the Slit2 transgene in the lungs of Slit2-Tg mice. of mucus (dark blue staining) in the bronchi (Fig. 5A, right middle panels), elevated levels of IgE in BALFs (Fig. 5F) and serum Slit2 modulation of eosinophil chemotaxis in allergic airway (Fig. 5G), and increased amounts of CXCL1 and 2 mRNAs and inflammation proteins (Supplemental Fig. 3). These findings provide evidence To investigate the roles of Slit2 in eosinophil chemotaxis in vivo, that Slit-Robo signaling exacerbates allergic airway inflammation we compared leukocyte infiltration in C57 mice and Slit2-Tg mice and is associated with eosinophil recruitment to lung tissue. subjected to OVA-induced allergic airway inflammation. Aerosol To further support our conclusion, we tested whether blockade of challenge of C57 mice with OVA (43) clearly triggered the in- Slit2 binding to Robo1 suppresses allergic airway inflammation filtration of leukocytes, mainly eosinophils, into the peribronchial in vivo. Indeed, i.v. injection of R5, but not mIgG, potently and perivascular regions of the lungs (Fig. 5A, left upper panels) inhibited eosinophil deposition in the peribronchial and peri- and BALFs (Fig. 5B). Slit2-Tg mice manifested massive in- vascular regions of the lungs (Fig. 5A, left lower panels) and filtration of eosinophils into the peribronchial and perivascular BALFs (Fig. 5D) of OVA-sensitized C57 mice, without affecting regions (Fig. 5A, left middle panels) and displayed significantly the number of total leukocytes or leukocyte subpopulations in increased deposition (∼3-fold) of eosinophils in BALFs as com- circulation (Fig. 5E). R5 treatment also significantly decreased pared with C57 mice (Fig. 5B). Identical data were obtained from mucus accumulation within the bronchial lumens (Fig. 5A, right experiments using founder B of Slit2-Tg mice (data not shown). lower panels) and reduced the level of plasma IgE (Fig. 5H). Our Notably, total leukocytes and leukocyte subpopulations in blood results that Slit2 potentiates eosinophil chemotaxis, promotes remained unchanged (Fig. 5C), ruling out possibility of a hema- mucus deposition, and elevates IgE collectively indicate that Slit- topoietic mechanism by which eosinophils more aggressively in- Robo signaling exaggerates allergic airway inflammation. Nota- filtrate lung tissues in OVA-sensitized Slit2-Tg mice. bly, Slit2-Tg mice displayed accelerated neutrophil infiltration To assess the importance of Slit2 in the pathological features (Fig. 5A) and elevated CXCL1 and -2 (Supplemental Fig. 3), of allergic airway inflammation, we compared mucus deposition whereas R5 treatment attenuated neutrophil deposition (Fig. 5D), in bronchial tissues of C57 and Slit2-Tg mice. Mucus staining of suggesting that the promoting effects of Slit2 in allergic airway

FIGURE 5. Effects of Slit2 on eosinophil chemotaxis and allergic airway inflammation. A, H&E (left panels) and periodic acid-Schiff (Diagnostics Biosystems, Pleasanton, CA; right panels) staining of lung tissues obtained from C57 and Slit2-Tg mice without (Sham) or with (Model) aerosol challenge with OVA in the absence or presence of mIgG or R5 treatment. B, Infiltration of eosinophils in BALFs in C57 and Slit2-Tg mice following OVA sensitization. Leukocyte counts in blood (C) and IgE levels in BALFs (F) and blood (G) of C57 and Slit2-Tg mice with or without aerosol challenge with OVA. Effects of mIgG or R5 on eosinophil deposition in BALFs (D), circulatory leukocytes (E), and serum IgE (H) of OVA-sensitized C57 mice. A, Left panels, Original magnification 3100; scale bar, 20mm. A, Right panels, Original magnification 3200; scale bar, 5mm. Data represent the mean 6 SD of two to three independent experiments (n = 5–10 for each group). pp , 0.05; ppp , 0.01. 6302 Slit2 DIRECTS LEUKOCYTE CHEMOTAXIS inflammation may culminate in amplification of nonallergic in- OVA sensitization, which is consistent with a previous report that flammation. WT and LY294002 exhibited low potency against eotaxin-induced chemotactic responses (6). In sharp contrast, both agents potently Slit2 regulation of neutrophil chemotaxis in endotoxin-induced abolished Slit2-potentiated eosinophil accumulation in Slit2-Tg lung inflammation mice to the levels comparable to those in C57 mice. Our results To examine the effects of Slit2 on neutrophil chemotaxis in thus indicate the functional importance of PI3K signaling, a vivo, we employed a murine model of endotoxin-induced lung downstream effector of Slit-Robo signaling, for enhancing eosi- inflammation to compare the pneumonia-like phenotypic changes nophil chemotaxis in vivo. between C57 mice and Slit2-Tg mice (44). Aerosol challenge with LPS triggered a dramatic infiltration of leukocytes, mainly neutrophils, in BALFs of C57 and Slit2-Tg mice (Fig. 6A). Impor- Discussion tantly, Slit2-Tg mice manifested a decreased deposition of Engagement of Robo1 by Slit2 induces srGAP1 binding to Robo1, neutrophils (∼50%) as compared with C57 mice. In contrast, which inactivates Cdc42 for repulsive migration of neuronal cells neutralization of Slit2 binding to Robo1 by R5, but not mIgG, (28). In this study, we show that Slit2, by differential regulation ∼ of Cdc42 activity, potentiates eotaxin-mediated chemotaxis of increased neutrophil infiltration by 3-fold in BALFs of C57 mice low subjected to LPS challenge (Fig. 6C). Again, total leukocytes and eosinophils (srGAP1 ) for exaggeration of allergic airway inflammation while suppressing SDF-1a–mediated chemotaxis of leukocyte subpopulations in blood remained relatively unaltered high (Fig. 6B,6D). The observed inhibitory effect of Slit2 on neutrophil neutrophils (srGAP1 ) for attenuation of endotoxin-induced chemotaxis in vivo is fully consistent with our in vitro findings lung inflammation. During leukocyte chemotaxis, srGAP1 sig- (Fig. 2), further supporting the notion that Slit-Robo signaling naling acts as a default or dominant mechanism, whereas PI3K induces a chemorepellent response on neutrophils in endotoxin- signaling acts as an alternative or supplementary mechanism. The induced lung inflammation. balance between repulsive srGAP1 signaling and attractive PI3K signaling thus appears to critically determine directional chemo- Slit2 gradient at the bronchus–alveoli axis enhances eosinophil taxis of leukocytes during allergic and endotoxin-induced lung chemotaxis inflammation (Supplemental Fig. 4). Notably, although Slit2 mediates axon guidance and neuronal migration and induces mi- To mimic the effects of the Slit2 gradient at the bronchus–alveoli gration of vascular endothelial cells and cancer cells, it does not axis (Fig. 4G), we hypothesized that, following OVA-sensitization, directly exert any appreciable action on the chemotactic activity of aerosol inhalation of Slit2 would increase the bronchial pool of leukocytes, attesting to the functional role of Slit2 as a chemo- Slit2, whereas i.v. administration of Slit2 would decrease it by tactic regulator, but not as a chemokine itself, for leukocytes. To increasing the circulatory pool of Slit2. As predicted, aerosol the best of our knowledge, this is the first example of an endog- administration of Slit2 markedly enhanced eosinophil infiltration enous factor that regulates both attractive and repulsive chemo- into BALFs of OVA-sensitized lungs (Fig. 7A). In contrast, i.v. taxis among different subtypes of leukocytes. administration of Slit2 drastically mitigated eosinophil accumu- The formation of a Slit2 gradient within the subventricular zone lation in BALFs (Fig. 7B). Taken together, these functional find- of the brain is essential to directional migration of neuronal ings demonstrate that this newly identified Slit2 gradient at the progenitor cells, the direction of cilia, and the flow of cerebrospinal bronchus–alveoli axis accelerates eosinophil chemotaxis during fluid (50). In addition, solid tumors exhibit a Slit2 gradient, which allergic airway inflammation. is high in the center and low in the periphery, for tumor-induced angiogenesis (18). In the current study, we show that nonciliated PI3K required for Slit2-potentiated eosinophil chemotaxis secretory Clara cells within normal bronchial epithelium secrete To examine the in vivo importance of PI3K signaling in eosinophil Slit2. Although the biological function of Slit2 secreted by infiltration in response to Slit2, C57 and Slit2-Tg mice were aerosol Clara cells in the absence of inflammatory stimulation remains to challenged with OVA in the absence or presence of WT (18) or AS be elucidated, we speculate that it may participate in the main- (45). Both WT (Fig. 8A) and AS (Fig. 8B) only partially sup- tenance of physiological homeostasis of leukocytes and bronchi pressed eosinophil infiltration in BALFs of C57 mice following (12, 38, 45, 47, 50). Interestingly, aerosol challenge with OVA, but

FIGURE 6. Effects of Slit2 on neutrophil chemotaxis and endotoxin-induced lung inﬂammation. A, Inﬁltration of neutrophils in BALFs of C57 and Slit2- Tg mice following aerosol challenge with LPS. B, Leukocyte counts and differentials in the blood of C57 and Slit2-Tg mice without (Sham) or with (Model) aerosol challenge with LPS. C, Effects of mIgG or R5 on neutrophil deposition in BALFs of C57 mice following aerosol challenge with LPS. D, Leukocyte counts and differentials in the blood of C57 mice pretreated with R5 or mIgG. Data represent the mean 6 SD of two to three independent experiments (n = 6 to 7 for each group). pp , 0.05; ppp , 0.01. The Journal of Immunology 6303

inflammation. Our experimental findings are fully consistent with previously published reports (12–17), demonstrating a pathological role for Slit-Robo signaling in modulating chemotaxis of neutrophils, macrophages, and T lymphocytes. These results collectively indicate the pathological importance of Slit-Robo signaling in inflammatory responses in vivo. As IL-5 critically mediates eosinophil maturation, survival, activation, and chemotaxis (32), IL-5-Tg mice have been used extensively for a variety of studies, such as eosinophil migration FIGURE 7. Slit2 gradient at bronchus–alveoli axis modulates eosinophil and chemotaxis (52) and phagocytosis (53). In this study, we chemotaxis Eosinophil infiltration in BALFs of OVA-sensitized C57 mice have isolated mouse eosinophils from IL-5-Tg mice for in vitro following aerosol inhalation of Slit2 (A) or i.v. injection of Slit2 (B). Data represent the mean 6 SD of two independent experiments (n = 3–7 for experiments of eosinophil polarization, migration, and chemo- each group). pp , 0.05; ppp , 0.01. taxis. We have also performed OVA-induced allergic airway inflammation for in vivo investigation of eosinophils functionality. Notably, mouse eosinophils isolated from IL-5-Tg mice pheno- not LPS, dramatically elevated Slit2 expression, which is char- copy human eosinophils in srGAP1 expression and Slit2 en- acteristically high in the bronchi and low in the alveoli, thus hancement of chemotaxis, attesting to the suitability of mouse forming a Slit2 gradient at the so-called bronchus–alveoli axis. eosinophils isolated from IL-5-Tg mice to replace human eosi- These results collectively suggest that by adhering to heparan nophils isolated from asthma patients in our present study. sulfate proteoglycans on the cell surface of neighboring cells (51), In the current study, we have shown that nonsecretory Clara cells secreted Slit2 forms a concentration gradient for mediating di- express Slit2, whereas neutrophils and eosinophils express Robo1. rectional migration of neuronal cells and vascular endothelial cells In addition, rSlit2 inhibits SDF-1a–induced chemotaxis of neu- and for regulating directional chemotaxis of leukocytes. trophils while enhancing eotaxin-induced eosinophil chemotaxis As Slit2 regulates chemokine-mediated chemotaxis of eosino- in vitro. Furthermore, Slit-Robo signaling potently exaggerates phils and neutrophils in vitro, we suspected that the newly dis- eosinophil chemotaxis during allergic airway inflammation while covered Slit2 gradient at the bronchus–alveoli axis may play an drastically suppressing neutrophil chemotaxis during LPS-induced essential role in guiding leukocyte chemotaxis during allergic lung inflammation in vivo, as demonstrated by the studies using airway inflammation. To test this hypothesis, we reasoned that Slit2-Tg mice, R5 neutralizing mAb, and rSlit2 protein. These manipulation of the Slit2 gradient would alter the direction of results collectively indicate a direct effect of Slit2 on chemotaxis eosinophil chemotaxis in vivo. Remarkably, increasing the Slit2 of neutrophils and eosinophils during lung inflammation, which is gradient by aerosol administration of isolated recombinant human fully consistent with previous reports (12–17). Slit2 markedly enhanced infiltration of eosinophils into BALFs, However, our findings do not eliminate the possibility of an in- whereas decreasing the Slit2 gradient by i.v. administration of direct effect of Slit2 on chemotaxis of neutrophils and eosinophils purified Slit2 significantly reduced it. Our results thus indicate the in vivo. As Slit2 potently inhibits neutrophil infiltration in the biological significance of the Slit2 gradient at the bronchus– model of endotoxin-induced lung inflammation (Fig. 6), it appears alveoli axis for governing eosinophil chemotaxis during allergic quite unexpected to observe an increased deposition of neutrophils airway inflammation. These data further demonstrate the feasi- in the BALFs of OVA-sensitized Slit2-Tg mice (Fig. 5B,5D). To bility, for the first time to the best of our knowledge, of adopting address this parodoxal finding, we reasoned that Slit-Robo sig- mouse lungs as a powerful assay for in vivo study of leukocyte naling may possibly have other unknown direct and indirect tar- chemotaxis. gets during lung inflammation. In this regard, although asthma is Although this study has been focused on modulation of leu- traditionally considered as an eosinophilic disease in the airways, kocyte chemotaxis by Slit-Robo signaling and its underlying increased deposition of neutrophils in the lungs (54), exaggerated molecular mechanisms, we have systematically assessed the sig- accumulation of neutrophils in the sputum (55), and elevated nificance of Slit-Robo signaling in the pathogenesis of allergic and neutrophil-associated chemokines, such as IL-8 in patients and endotoxin-induced lung inflammation. For instance, in vivo ex- keratinocyte chemoattractant (CXCL1) and MIP-2 (CXCL2) in pression of exogenous Slit2 exaggerated allergic airway in- mice (39, 56, 57), have been well documented in allergic lung flammation, but mitigated endotoxin-induced lung inflammation, inflammation. We thus speculate that the attractive effects of Slit2 as evidenced by a comparison between Slit2-Tg and C57 mice. on eosinophil chemotaxis may overwhelm its repulsive effects on Additionally, neutralization of endogenous Slit2 binding to Robo1 neutrophil chemotasis in the murine model of OVA-induced al- by R5, but not isotype-matched irrelevant mIgG, decreased allergic airway inflammation, culminating in amplification of the lergic airway inflammation, but increased endotoxin-induced lung entire inflammatory responses. The findings of increased neutrophil infiltration (Fig. 5B) and elevated CXCL1 and CXCL2 (Supplemental Fig. 3) in OVA-sensitized Slit2-Tg mice but decreased neutrophil deposition following R5 treatment (Fig. 5D) appear to lend a support to this hypothesis. Further studies along this line of investigation are therefore promising to identify other direct and indirect targets of Slit-Robo signaling during the pathogenesis of lung inflammation. In response to Slit2 binding, leukocytes must send signals FIGURE 8. PI3K dependence of Slit2-potentiated eosinophil infiltration. from the cell surface to the cytoplasm, which dynamically and Eosinophil infiltration in BALFs of OVA-sensitized C57 and Slit2-Tg mice reversibly control the cellular machinery for leukocyte chemotaxis. in the absence or presence of DMSO, WT (A), or AS (B). Data represent In this study, we show that the balance between repulsive srGAP1 the mean 6 SD of two independent experiments (n = 3–7 for each group). signaling and attractive PI3K signaling may determine the di- ppp , 0.01. rectional chemotaxis of leukocytes during allergic and endotoxin- 6304 Slit2 DIRECTS LEUKOCYTE CHEMOTAXIS induced lung inflammation. However, it remains to be determined 20. Urbich, C., L. Ro¨ssig, D. Kaluza, M. Potente, J. N. Boeckel, A. Knau, F. Diehl, J.-G. Geng, W. K. Hofmann, A. M. Zeiher, and S. Dimmeler. 2009. HDAC5 is whether the signal transduction pathways involved in leukocyte a repressor of angiogenesis and determines the angiogenic gene expression chemotaxis are the same as those downstream of Robo and/or other pattern of endothelial cells. Blood 113: 5669–5679. potential receptors in neurons, endothelial cells, and cancer cells. 21. Zhang, B., U. M. Dietrich, J.-G. Geng, R. Bicknell, J. D. Esko, and L. Wang. 2009. Repulsive axon guidance molecule Slit3 is a novel angiogenic factor. Further pursuit of this line of investigation should enhance our Blood 114: 4300–4309. understanding of the biological functions of guidance cues and the 22. Schmid, B. C., G. A. Rezniczek, G. Fabjani, T. Yoneda, S. Leodolter, and molecular mechanisms driving the polarization and the directional R. Zeillinger. 2007. The neuronal guidance cue Slit2 induces targeted migration and may play a role in brain metastasis of breast cancer cells. Breast Cancer Res. migration of these fascinating cells. Treat. 106: 333–342. 23. Mertsch, S., N. Schmitz, A. Jeibmann, J.-G. Geng, W. Paulus, and V. Senner. 2008. Slit2 involvement in glioma cell migration is mediated by Robo1 receptor. Acknowledgments J. Neurooncol. 87: 1–7. We thank James J. Lee (Mayo Clinic Arizona, Scottsdale, AZ) for IL-5-Tg 24. Yuasa-Kawada, J., M. Kinoshita-Kawada, Y. Rao, and J. Y. Wu. 2009. Deubi- quitinating enzyme USP33/VDU1 is required for Slit signaling in inhibiting mice and Michael J. Franklin (University of Minnesota, Minneapolis, MN) breast cancer cell migration. Proc. Natl. Acad. Sci. USA 106: 14530–14535. for editing the manuscript. 25. Bashaw, G. J., T. Kidd, D. Murray, T. Pawson, and C. S. Goodman. 2000. Re- pulsive axon guidance: Abelson and Enabled play opposing roles downstream of the roundabout receptor. Cell 101: 703–715. Disclosures 26. Fritz, J. L., and M. F. A. VanBerkum. 2000. Calmodulin and son of sevenless dependent signaling pathways regulate midline crossing of axons in the Dro- The authors have no financial conflicts of interest. sophila CNS. Development 127: 1991–2000. 27. Hall, A. 1998. Rho GTPases and the actin cytoskeleton. Science 279: 509–514. 28. Wong, K., X. R. Ren, Y. Z. Huang, Y. Xie, G. Liu, H. Saito, H. Tang, L. Wen, S. M. Brady-Kalnay, L. Mei, et al. 2001. Signal transduction in neuronal mi- References gration: roles of GTPase activating proteins and the small GTPase Cdc42 in the 1. Springer, T. A. 1994. Traffic signals for lymphocyte recirculation and leukocyte Slit-Robo pathway. Cell 107: 209–221. emigration: the multistep paradigm. Cell 76: 301–314. 29. Ko¨lsch, V., P. G. Charest, and R. A. Firtel. 2008. The regulation of cell motility 2. Wong, K., H. T. Park, J. Y. Wu, and Y. Rao. 2002. Slit proteins: molecular and chemotaxis by phospholipid signaling. J. Cell Sci. 121: 551–559. guidance cues for cells ranging from neurons to leukocytes. Curr. Opin. Genet. 30. Ly, N. P., K. Komatsuzaki, I. P. Fraser, A. A. Tseng, P. Prodhan, K. J. Moore, and Dev. 12: 583–591. T. B. Kinane. 2005. Netrin-1 inhibits leukocyte migration in vitro and in vivo. 3. Laudanna, C., J. Y. Kim, G. Constantin, and E. Butcher. 2002. Rapid leukocyte Proc. Natl. Acad. Sci. USA 102: 14729–14734. integrin activation by chemokines. Immunol. Rev. 186: 37–46. 31. Hjorthaug, H. S., and H. C. Aasheim. 2007. Ephrin-A1 stimulates migration of 4. Furie, B., and B. C. Furie. 1995. The molecular basis of platelet and endothelial CD8+CCR7+ T lymphocytes. Eur. J. Immunol. 37: 2326–2336. cell interaction with neutrophils and monocytes: role of P-selectin and the 32. Lee, N. A., M. P. McGarry, K. A. Larson, M. A. Horton, A. B. Kristensen, and P-selectin ligand, PSGL-1. Thromb. Haemost. 74: 224–227. J. J. Lee. 1997. Expression of IL-5 in thymocytes/T cells leads to the de- 5. Malech, H. L., R. K. Root, and J. I. Gallin. 1977. Structural analysis of human velopment of a massive eosinophilia, extramedullary eosinophilopoiesis, and neutrophil migration. Centriole, microtubule, and microfilament orientation and unique histopathologies. J. Immunol. 158: 1332–1344. function during chemotaxis. J. Cell Biol. 75: 666–693. 33. Yang, X.-M., H.-X. Han, F. Sui, Y.-M. Dai, M. Chen, and J.-G. Geng. 2010. Slit- 6. Mishra, R. K., J. E. Scaife, Z. Harb, B. C. Gray, R. Djukanovic, and G. Dent. Robo signaling mediates lymphangiogenesis and promotes tumor lymphatic 2005. Differential dependence of eosinophil chemotactic responses on phos- metastasis. Biochem. Biophys. Res. Commun. 396: 571–577. phoinositide 3-kinase (PI3K). Allergy 60: 1204–1207. 34. Wang, H.-B., J.-T. Wang, L. Zhang, Z. H. Geng, W.-L. Xu, T. Xu, Y. Huo, 7. Sallusto, F., and M. Baggiolini. 2008. Chemokines and leukocyte traffic. Nat. X. Zhu, E. F. Plow, M. Chen, and J.-G. Geng. 2007. P-selectin primes leukocyte Immunol. 9: 949–952. integrin activation during inflammation. Nat. Immunol. 8: 882–892. 8. Kidd, T., K. Brose, K. J. Mitchell, R. D. Fetter, M. Tessier-Lavigne, 35. Nguyen-Ba-Charvet, K. T., and A. Che´dotal. 2002. Role of Slit proteins in the C. S. Goodman, and G. Tear. 1998. Roundabout controls axon crossing of the vertebrate brain. J. Physiol. Paris 96: 91–98. CNS midline and defines a novel subfamily of evolutionarily conserved guidance 36. Baumann, M. A., and C. C. Paul. 1998. The AML14 and AML14.3D10 cell lines: receptors. Cell 92: 205–215. a long-overdue model for the study of eosinophils and more. Stem Cells 16: 16–24. 9. Brose, K., K. S. Bland, K. H. Wang, D. Arnott, W. Henzel, C. S. Goodman, 37. Boxio, R., C. Bossenmeyer-Pourie´, N. Steinckwich, C. Dournon, and O. Nu¨sse. M. Tessier-Lavigne, and T. Kidd. 1999. Slit proteins bind Robo receptors and have 2004. Mouse bone marrow contains large numbers of functionally competent an evolutionarily conserved role in repulsive axon guidance. Cell 96: 795–806. neutrophils. J. Leukoc. Biol. 75: 604–611. 10. Li, H. S., J. H. Chen, W. Wu, T. Fagaly, L. Zhou, W. Yuan, S. Dupuis, 38. Reynolds, S. D., and A. M. Malkinson. 2010. Clara cell: progenitor for the Z. H. Jiang, W. Nash, C. Gick, et al. 1999. Vertebrate slit, a secreted ligand for bronchiolar epithelium. Int. J. Biochem. Cell Biol. 42: 1–4. the transmembrane protein roundabout, is a repellent for olfactory bulb axons. 39. Terakawa, M., Y. Tomimori, M. Goto, and Y. Fukuda. 2006. Mast cell chymase Cell 96: 807–818. induces expression of chemokines for neutrophils in eosinophilic EoL-1 cells 11. Wu, W., K. Wong, J. Chen, Z. Jiang, S. Dupuis, J. Y. Wu, and Y. Rao. 1999. and mouse peritonitis eosinophils. Eur. J. Pharmacol. 538: 175–181. Directional guidance of neuronal migration in the olfactory system by the protein 40. Li, X., Y. Chen, Y. Liu, J. Gao, F. Gao, M. Bartlam, J. Y. Wu, and Z. Rao. 2006. Slit. Nature 400: 331–336. Structural basis of Robo proline-rich motif recognition by the srGAP1 Src ho- 12. Wu, J. Y., L. Feng, H. T. Park, N. Havlioglu, L. Wen, H. Tang, K. B. Bacon, mology 3 domain in the Slit-Robo signaling pathway. J. Biol. Chem. 281: Jiang Zh, Zhang Xc, and Y. Rao. 2001. The neuronal repellent Slit inhibits 28430–28437. leukocyte chemotaxis induced by chemotactic factors. Nature 410: 948–952. 41. Heit, B., L. X. Liu, P. Colarusso, K. D. Puri, and P. Kubes. 2008. PI3K accel- 13. Kanellis, J., G. E. Garcia, P. Li, G. Parra, C. B. Wilson, Y. Rao, S. Han, erates, but is not required for, neutrophil chemotaxis to fMLP. J. Cell Sci. 121: C. W. Smith, R. J. Johnson, J. Y. Wu, and L. Feng. 2004. Modulation of in- 205–214. flammation by slit protein in vivo in experimental crescentic glomerulonephritis. 42. Ebihara, S., H. Kurachi, and Y. Watanabe. 2007. A simple preparation method Am. J. Pathol. 165: 341–352. for mouse eosinophils and their responses to anti-allergic drugs. Inflamm. Res. 14. Guan, H., G. Zu, Y. Xie, H. Tang, M. Johnson, X. Xu, C. Kevil, W. C. Xiong, 56: 112–117. C. Elmets, Y. Rao, et al. 2003. Neuronal repellent Slit2 inhibits dendritic cell 43. Xia, Y.-F., B.-Q. Ye, Y.-D. Li, J.-G. Wang, X.-J. He, X. Lin, X. Yao, D. Ma, migration and the development of immune responses. J. Immunol. 171: 6519– A. Slungaard, R. P. Hebbel, et al. 2004. Andrographolide attenuates in- 6526. flammation by inhibition of NF-k B activation through covalent modification of 15. Prasad, A., Z. Qamri, J. Wu, and R. K. Ganju. 2007. Slit-2/Robo-1 modulates the reduced cysteine 62 of p50. J. Immunol. 173: 4207–4217. CXCL12/CXCR4-induced chemotaxis of T cells. J. Leukoc. Biol. 82: 465–476. 44. Arndt, P. G., S. K. Young, J. G. Lieber, M. B. Fessler, J. A. Nick, and 16. Altay, T., B. McLaughlin, J. Y. Wu, T. S. Park, and J. M. Gidday. 2007. Slit G. S. Worthen. 2005. Inhibition of c-Jun N-terminal kinase limits lipo- modulates cerebrovascular inflammation and mediates neuroprotection against polysaccharide-induced pulmonary neutrophil influx. Am. J. Respir. Crit. global cerebral ischemia. Exp. Neurol. 207: 186–194. Care Med. 171: 978–986. 17. Tole, S., I. M. Mukovozov, Y. W. Huang, M. A. O. Magalhaes, M. Yan, 45. Wang, Z. L., X. H. Wu, L. F. Song, Y. S. Wang, X. H. Hu, Y. F. Luo, Z. Z. Chen, M. R. Crow, G. Y. Liu, C. X. Sun, Y. Durocher, M. Glogauer, and J. Ke, X. D. Peng, C. M. He, et al. 2009. Phosphoinositide 3-kinase g inhibitor L. A. Robinson. 2009. The axonal repellent, Slit2, inhibits directional migration ameliorates concanavalin A-induced hepatic injury in mice. Biochem. Biophys. of circulating neutrophils. J. Leukoc. Biol. 86: 1403–1415. Res. Commun. 386: 569–574. 18. Wang, B., Y. Xiao, B.-B. Ding, N. Zhang, X.-B. Yuan, L. Gui, K.-X. Qian, 46. Oez, S., E.Platzer, and K.Welte. 1990.A quantitative colorimetricmethod toevaluate S. Duan, Z. Chen, Y. Rao, and J.-G. Geng. 2003. Induction of tumor angio- the functional state of human polymorphonuclear leukocytes. Blut 60: 97–102. genesis by Slit-Robo signaling and inhibition of cancer growth by blocking Robo 47. Anselmo, M. A., S. Dalvin, P. Prodhan, K. Komatsuzaki, J. T. Aidlen, activity. Cancer Cell 4: 19–29. J. J. Schnitzer, J. Y. Wu, and T. B. Kinane. 2003. Slit and robo: expression 19. Wang, L. J., Y. Zhao, B. Han, Y. G. Ma, J. Zhang, D. M. Yang, J. W. Mao, patterns in lung development. Gene Expr. Patterns 3: 13–19. F. T. Tang, W. D. Li, Y. Yang, et al. 2008. Targeting Slit-Roundabout signaling 48. Greenberg, J. M., F. Y. Thompson, S. K. Brooks, J. M. Shannon, and inhibits tumor angiogenesis in chemical-induced squamous cell carcinogenesis. A. L. Akeson. 2004. Slit and robo expression in the developing mouse lung. Dev. Cancer Sci. 99: 510–517. Dyn. 230: 350–360. The Journal of Immunology 6305

49. Plump, A. S., L. Erskine, C. Sabatier, K. Brose, C. J. Epstein, C. S. Goodman, interleukin 5 (IL-5) transgenic mice: eosinophil activity and impaired clearance C. A. Mason, and M. Tessier-Lavigne. 2002. Slit1 and Slit2 cooperate to prevent of Schistosoma mansoni. Parasite Immunol. 19: 291–300. premature midline crossing of retinal axons in the mouse visual system. Neuron 54. McKinley, L., J. Kim, G. L. Bolgos, J. Siddiqui, and D. G. Remick. 2005. CXC 33: 219–232. chemokines modulate IgE secretion and pulmonary inflammation in a model of 50. Sawamoto, K., H. Wichterle, O. Gonzalez-Perez, J. A. Cholfin, M. Yamada, allergic asthma. Cytokine 32: 178–185. N. Spassky, N. S. Murcia, J. M. Garcia-Verdugo, O. Marin, J. L. R. Rubenstein, 55. Monteseirıń, J. 2009. Neutrophils and asthma. J. Investig. Allergol. Clin. et al. 2006. New neurons follow the flow of cerebrospinal fluid in the adult brain. Immunol. 19: 340–354. Science 311: 629–632. 56. John, A. E., Y. M. Zhu, C. E. Brightling, L. Pang, and A. J. Knox. 2009. Human 51. Hu, H. 2001. Cell-surface heparan sulfate is involved in the repulsive guidance airway smooth muscle cells from asthmatic individuals have CXCL8 hyperse- activities of Slit2 protein. Nat. Neurosci. 4: 695–701. cretion due to increased NF-k B p65, C/EBP b, and RNA polymerase II binding 52. Borchers, M. T., T. Ansay, R. DeSalle, B. L. Daugherty, H. Shen, M. Metzger, to the CXCL8 promoter. J. Immunol. 183: 4682–4692. N. A. Lee, and J. J. Lee. 2002. In vitro assessment of chemokine receptor-ligand 57. Coward, W. R., H. Sagara, S. J. Wilson, S. T. Holgate, and M. K. Church. 2004. interactions mediating mouse eosinophil migration. J. Leukoc. Biol. 71: 1033–1041. Allergen activates peripheral blood eosinophil nuclear factor-kappaB to generate 53. Dent, L. A., G. H. Munro, K. P. Piper, C. J. Sanderson, D. A. Finlay, granulocyte macrophage-colony stimulating factor, tumour necrosis factor-a and R. K. Dempster, L. P. Bignold, D. G. Harkin, and P. Hagan. 1997. Eosinophilic interleukin-8. Clin. Exp. Allergy 34: 1071–1078.