Rac GTPase Isoform-specific Regulation of NADPH Oxidase and Chemotaxis in Murine Neutrophils in Vivo

The Rho family GTPase Rac acts as a molecular switch for signal transduction to regulate various cellular functions. Mice deficient in the hematopoietic-specific Rac2 isoform exhibit agonist-specific defects in neutrophil chemotaxis and superoxide production, despite expression of the highly homologous Rac1 isoform. To examine whether functional defects in rac2–/– neutrophils reflect effects of an overall decrease in total cellular Rac or an isoform-specific role for Rac2, retroviral vectors were used to express exogenous Rac1 or Rac2 at levels similar to endogenous. In rac2–/– neutrophils differentiated from transduced myeloid progenitors in vitro, increasing cellular Rac levels by expression of either exogenous Rac1 or Rac2 increased formylmethionylleucylphenylalanine- or phorbol ester-stimulated NADPH oxidase activity. Of note, placement of an epitope tag on the N terminus of Rac1 or Rac2 blunted reconstitution of responses in rac2–/– neutrophils. In rac2–/– neutrophils isolated from mice transplanted with Rac-transduced bone marrow cells, superoxide production and chemotaxis were fully reconstituted by expression of exogenous Rac2, but not Rac1. A chimeric Rac1 protein in which the Rac1 C-terminal polybasic domain, which contains six lysines or arginines, was replaced with that of the human Rac2 polybasic domain containing only three basic residues, also reconstituted superoxide production and chemotaxis, whereas expression of a Rac2 derivative in which the polybasic domain was replaced with that of Rac1 did not and resulted in disoriented cell motility. Thus, the composition of the polybasic domain is sufficient for determining Rac isoform specificity in the production of superoxide and chemotaxis in murine neutrophils in vivo.

Neutrophils are important effector cells in the host response to bacterial and fungal pathogens and are endowed with che-motactic and microbicidal functions that are activated by signal transduction pathways downstream of receptors for inflammatory and microbial stimuli (1). Migration from blood vessels to inflamed sites is initiated by various chemoattractants, including chemokines secreted by host cells and by bacterial products such as fMLP. 1 Superoxide produced by the neutrophil NADPH oxidase is the precursor to toxic reactive oxidants important for normal microbial killing.
Rac GTPases, members of the Rho family of small GTPases, play a central role in regulating neutrophil chemotaxis, superoxide production, and other neutrophil functions (2). Rho GTPases act as molecular switches in signaling pathways, alternating between activated GTP-bound and inactive GDPbound states. In unstimulated cells, the GDP-bound form of Rac is bound to Rho guanine-nucleotide dissociation inhibitor, which inhibits nucleotide exchange and maintains Rac-GDP in the cytoplasm. Upon stimulation, Rac translocates to the membrane where agonist-activated guanine nucleotide exchange factors mediate the exchange of GDP for GTP, resulting in a conformational change that permits Rac binding to downstream target proteins (3). Rac has three isoforms, the ubiquitously expressed Rac1, a hematopoietic cell-specific isoform Rac2, and Rac3, which appears to be expressed in a variety of tissues but not phagocytes (4 -8). The amino acid sequence of murine Rac1 is identical to human Rac1, and murine Rac2 differs by only two amino acids from human Rac2, with a conservative substitution of aspartate instead of glutamate at position 148 and a proline instead of an alanine at position 188 (5,6). Rac1 and Rac2 have 92% identical amino acid sequences. These two isoforms have an identical effector domain (amino acids 26 -45), which is a critical site of interaction with both guanine nucleotide exchange factors and downstream protein targets. However, long range effects exerted by distant isoformdivergent amino acids may alter the flexibility of the effector domain to produce different affinities of Rac1 and Rac2 for these proteins (9). The greatest sequence divergence between Rac1 and Rac2 is in the C-terminal polybasic domain (residues 183-188), which is adjacent to a prenylated cysteine residue that can insert into cellular membranes. Rac1 has six adjacent basic residues in the polybasic domain (KKRKRK), whereas Rac2 has only three basic residues interspersed by neutral amino acids (RQQKR(A/P)) (Fig. 1A). The polybasic domain mediates differential localization of overexpressed Rac1 and Rac2 to fibroblast and epithelial cell membranes (10) and contributes to differential interactions with at least one downstream target of activated Rac GTPases, the serine-threonine kinase PAK1 (11). 2 Studies in murine Rac2-deficient neutrophils and neutrophils from a patient with a dominant-negative mutation in Rac2 suggest that this isoform is essential for normal neutrophil function (8,(12)(13)(14)(15). Although murine neutrophils have similar amounts of Rac1 and Rac2 (8), genetic deletion of Rac2 results in severely impaired F-actin polymerization and chemotaxis in response to chemoattractants, decreased L-selectinmediated adhesion, and reduced superoxide production in response to fMLP, phorbol ester, and IgG-opsonized sheep red blood cells (8,12,15). Heterozygous rac2 ϩ/Ϫ murine neutrophils have intermediate activity between rac2 ϩ/ϩ and rac2 Ϫ/Ϫ neutrophils in F-actin formation, chemotaxis, and superoxide production (8). In human neutrophils, which have predominantly Rac2 (16), expression of a dominant-negative Rac2 results in cellular defects similar to those seen in rac2 Ϫ/Ϫ murine neutrophils (13,14). Although Rac1-null murine neutrophils also exhibit decreased actin cytoskeleton assembly and a modest decrease in directed migration, superoxide production is normal in contrast to Rac2-null counterparts (17). These findings suggest that Rac1 and Rac2 play distinct roles in regulating neutrophil functions. Alternatively, the defects in Rac2deficient neutrophils could reflect effects of an overall reduction in cellular Rac levels.
The goal of the current study was to examine whether the impaired functions of Rac2-null neutrophils result from a relative cellular deficiency of total Rac or an isoform-specific role for Rac2 that is mediated by specific sequences. Retroviral vectors were used to express exogenous Rac1 or Rac2 in rac2 Ϫ/Ϫ neutrophils at levels similar to endogenous ones. Increasing Rac levels by expression of either exogenous Rac1 or Rac2 enhanced NADPH oxidase activity in rac2 Ϫ/Ϫ neutrophils differentiated in vitro with growth factors. However, we found that only expression of exogenous Rac2, but not Rac1, fully reconstituted defects in NADPH oxidase activity and chemotaxis in neutrophils harvested from rac2 Ϫ/Ϫ mice transplanted with transduced rac2 Ϫ/Ϫ marrow cells. This result could be recapitulated using a derivative of Rac1 in which the polybasic domain was substituted with the corresponding region from human Rac2, and conversely, expression of Rac2 chimeric for the polybasic Rac1 domain failed to rescue defects in rac2 Ϫ/Ϫ neutrophils and resulted in disoriented motility. Thus, the polybasic domain is sufficient to confer isoform-specific Rac functions in regulating neutrophil chemotaxis and NADPH oxidase activity.
Mice-Mice were maintained under specific pathogen-free conditions. Wild type (WT) mice used in this study were 8-to 10-week-old male or female C57/Bl6J mice purchased from Jackson Laboratory Inc. Rac2 Ϫ/Ϫ mice were previously generated (15) by targeted homologous recombination to disrupt the Rac2 gene. Rac2 Ϫ/Ϫ mice used here were 8to 10-week-old male or female mice that had been backcrossed onto the C57/Bl6J for more than 11 generations.
Retroviral Vectors for Expression of Rac Proteins-An MSCV retroviral backbone with a linked expression cassette for puromycin-Nacetyl-transferase (PAC) (19), MSCV-pac, was a kind gift from R. Hawley (American Red Cross, Rockville, MD). The MIEG3 retroviral vector was derived from the MSCV series, and contains an internal ribosome entry sequence linked to a cDNA for the enhanced green fluorescent protein (EGFP) instead of an antibiotic-resistance expression cassette (13). Murine Rac1 and Rac2 cDNAs flanked by BamHI and XbaI restriction sites were originally generated using reverse transcription-PCR from RNA prepared from the RAW 264.7 mouse macrophage cell line. The flanking restriction sites were changed to XhoI, along with creation of a Kozak sequence at the initiator ATG, using oligonucleotide primers and PCR, and cloned into the XhoI site in MSCV-pac and MIEG3 (Fig. 1B). The murine Rac1 or Rac2 cDNAs with an N-terminal hemagglutinin (HA) or FLAG epitope tag (13,20), respectively, were also cloned into MSCV-pac (Fig. 1B). The human Rac1 cDNA in which the Rac2 polybasic domain was replaced with that of Rac1 (Rac1-2) and the human Rac2 cDNA in which the polybasic domain was replaced with that of Rac1 (Rac2-1) were described previously (11). PCR was used to place EcoRI and XhoI linkers at the 5Ј and 3Ј ends, respectively, and each was subcloned into the corresponding sites in MIEG3 (Fig. 1). Details of oligonucleotide primers used for PCR amplifications are provided upon request. For all PCR-amplified cDNAs, DNA sequencing was performed to verify the fidelity of amplification. Vector plasmids were transfected using calcium phosphate into ecotropic Phoenix packaging cells (provided by Gary Nolan, Stanford University; ATCC number SD3444), and retroviral vector supernatants were collected in ␣-MEM with 20% fetal calf serum.
Transduction of Murine Bone Marrow Cells with Retroviral Vectors for Expression of Rac Proteins-Retroviral transduction of WT or rac2 Ϫ/Ϫ murine BM was generally as described previously (21), with the following modifications. Three days following injection of 150 mg/kg 5-fluorouracil into bone marrow (BM) donors, BM cells were taken from femurs and tibias by flushing with ␣-MEM. The day of BM harvest was considered Day 0. BM cells were pelleted and resuspended in 10 ml of red blood cell lysis solution (155 mM NH 4 Cl, 10 mM KHCO 3 , 100 M EDTA, in H 2 O, pH 7.4) in each tube and kept at room temperature for 3 min. Cells were then centrifuged and resuspended in either TD medium A (␣-MEM with 20% serum, 100 ng/ml rat SCF, 100 ng/ml human megakaryocyte-derived growth factor, 100 ng/ml human G-SCF and 2% penicillin/streptomycin) or TD medium B (␣-MEM with 20% serum, 100 ng/ml rat SCF, 100 units/ml human IL-6, and 2% penicillin/ streptomycin) for prestimulation for 48 h. On days 2-4, cells were transduced every 24 h with retrovirus supernatant on 6-well nontreated culture plates that were coated with CH296 (4 g/cm 2 ) by overnight incubation at 4°C prior to virus infection. For each round of transduction, cells in suspension were spun down, resuspended in fresh retrovirus supernatant containing either TD medium A or B, and returned to the corresponding well and incubated for 24 h. MSCV-pac vectors were used to transduce cells for in vitro differentiation in the presence of puromycin, and MIEG3 vectors were used to transduce cells that were sorted for EGFP fluorescence and transplanted into lethally irradiated recipient mice. There was no difference between the TD media A or B in either the transduction efficiency or subsequent in vitro differentiation and functional assays. Medium without virus was used for mock transduction. On day 5, the combined suspended and adherent cells were collected using Cell Dissociation Buffer (Invitrogen) and then differentiated ex vivo or sorted prior to transplantation into lethally irradiated recipients.
In Vitro Differentiation of Granulocytes following Retroviral Transduction of Murine Bone Marrow-Neutrophils were differentiated in vitro from mock transduced or transduced BM cells by culturing in ␣-MEM with 20% serum, 50 IU/ml murine IL-3, 0.3 ng/ml (ϭ 30 IU/ml) human granulocyte-colony stimulating factor, and 2% penicillin/streptomycin in T75 flasks. Every other day (on days 6, 8, 10, 12, and 14 post harvest) cell number was counted, and cell density was adjusted to 5 ϫ 10 5 cells/ml 2 A. Yamauchi and M. Dinauer, unpublished data. using fresh differentiation media. On days 8 -12, cells were selected by adding 2 g/ml puromycin, and by day 12, all mock-transduced cells exposed to puromycin had died. On day 14, cells were harvested for NADPH oxidase assays, using either transduced cells selected with puromycin or mock-transduced cells cultured in the absence of puromycin. Diff-Quik-stained slides prepared by Cytospin (Thermo Electron Corp., Waltham, MA) on day 14 showed ϳ80 -90% mature neutrophils, with no differences seen between WT, Rac2-null, or Rac-transduced cells. The fraction of Gr-1-positive cells as analyzed by flow cytometry was also similar between different groups. Cell lysate was prepared on days 14 -16 for SDS-PAGE and Western blotting.
Transplantation of Retrovirus-transduced Murine Bone Marrow Cells-To generate neutrophils differentiated in vivo from MIEG vector-transduced BM cells, EGFP-positive cells were selected on day 5 of the transduction protocol by sorting with a FACStar instrument (BD Bioscience), and 0.5-1.0 ϫ 10 6 sorted cells were injected via tail vein into Rac2-null mice that had been irradiated with 1100 cGy (split dose). Following hematopoietic recovery, neutrophils from peripheral blood and BM were assessed by flow cytometry for EGFP marking at 4 -6 weeks after transplantation. Neutrophils were then purified from BM as described previously (12,15) for analysis of NADPH oxidase activity and motility. In some experiments, BM from mice transplanted with sorted, MIEG vector-transduced rac2 Ϫ/Ϫ cells were used for secondary transplants into additional 1100-cGy-irradiated rac2 Ϫ/Ϫ mice to generate additional in vivo differentiated neutrophils expressing exogenous Rac. For rac2 Ϫ/Ϫ controls, a group of rac2 Ϫ/Ϫ mice was transplanted in parallel with freshly isolated rac2 Ϫ/Ϫ BM and used as a source of rac2 Ϫ/Ϫ neutrophils in subsequent functional assays. However, NADPH oxidase activity and chemotaxis in neutrophils harvested from these mice were no different from neutrophils harvested from non-transplanted rac2 Ϫ/Ϫ mice.
Immunoblotting-Preparation of neutrophil lysates, SDS-PAGE, and immunoblotting were performed as described before (12,15,22). Briefly, neutrophils differentiated in vitro or purified from BM were treated with diisopropylphosphofluoridate prior to lysis with IP buffer (1% Triton X-100, 150 mM NaCl, 10 mM Tris-Cl, 1 mM EDTA, 1 mM EGTA, pH 8.0) containing 20 g/ml chymostatin, 2 mM phenylmethylsulfonyl fluoride, 10 M leupeptin, and 1 mM 4-(2-aminoethyl)benzenesulfonyl fluoride-HCl. Following SDS-PAGE and transfer to nitrocellulose membranes, blots were sequentially probed with either an anti-Rac1 monoclonal mouse antibody, an anti-Rac2 rabbit polyclonal antibody, and an anti-p38 MAPK rabbit polyclonal antibody, followed by horseradish peroxidase-conjugated anti-mouse or anti-rabbit IgG. Proteins were visualized using an ECL detection kit (Amersham Biosciences). Bands were scanned, and the densities were analyzed by National Institutes of Health image software (rsb.info.nih.gov/nih-image/). Multiple exposures were analyzed to ensure that relative signal intensities measured were in the linear range.
Measurement of NADPH Oxidase Activity-For quantitative assays, either isoluminol chemiluminescence or a colorimetric assay based on reduction of cytochrome c was used to detect reactive oxygen species (8,23). Under these conditions, superoxide dismutase inhibited the chemiluminescence signal by ϳ97.5% and cytochrome c reduction by ϳ100%. Chemiluminescence in fMLP-stimulated cells was detected as relative luminescence units (RLUs) by fast kinetic mode for 100 s using an Lmax microplate luminometer and SoftMax Pro software (Molecular Devices, Sunnyvale, CA). The rate of cytochrome c reduction was measured as V max using SpectraMax340c microplate reader and the SoftMax PRO software (Molecular Devices).
Chemotaxis-Neutrophil chemotaxis assays using 1 and 10 M fMLP were performed using a modified Boyden chamber (48-well microchemotaxis chamber, Neuro Probes, Gaithersburg, MD) and 3-m diameter polycarbonate filter membranes as described previously (8,15). The number of migrated cells per high power view field (400ϫ) were counted for a minimum of three fields per well, and an average estimation for individual sample was calculated from data of replicate wells. Values were compared relative to the number of migrated cells in the WT group.
Time-lapse Video Microscopy and Motility Analysis-Neutrophil motility in response to fMLP stimulation was recorded using a Dunn chemotaxis chamber (24). BM neutrophils in Hanks' balanced salt solution were allowed to adhere to clean glass coverslips for 15 min at 37°C. The coverslips were mounted on the Dunn chamber with a gradient of 0 -10 M fMLP between the inner and outer wells of the chamber. The chamber was mounted on the microscope stage, and the temperature was maintained at 37°C with a stage heater (Instec Instruments, Boulder, CO). The chamber was allowed to equilibrate for 25 min prior to image collection to allow a stable gradient to develop.
Images were recorded at 10-s intervals on a Nikon Diaphot 300 microscope with differential interference contrast optics using a SPOT cooled charge-coupled device camera (Diagnostic Instruments, Sterling Heights, MI). Cell positions were tracked using the particle tracking capabilities in Metamorph 6.1 software (Universal Imaging, Brandywine, PA). Cell trajectories were analyzed using the horizon method (24,25), in which the position of the cell is marked once it reaches a "horizon" at 30 m from the initial position, and used to define the trajectory of the cell relative to the orientation of the gradient. This methodology can only be applied to cell types that move a significant distance over the time period analyzed, and therefore directional data could not be obtained for rac2 Ϫ/Ϫ or Rac1-transduced rac2 Ϫ/Ϫ cells. Directionality was evaluated relative to the null hypothesis of uniformity using the Rayleigh test (25). In all cases over 350 cells of each type were analyzed.
Statistical Analysis-The two-tailed Student's t test (either paired or unpaired, as indicated) was performed using Microsoft Excel software (Redmond, WA). Statistical comparisons of the distribution of rates of FIG. 1. C terminus of Rac proteins and retroviral vectors for expression of Rac. A, sequence alignment of the C terminus of wild type Rac1, Rac2, and derivatives. Hyphens indicate sites where amino acid residues are identical to wild type murine Rac1, whereas differences are as indicated. The polybasic domain is indicated by a rectangle. Mu, murine; hu, human. Note that the entire murine Rac1 and human Rac1 amino acid sequence are identical. B, MSCV-pac and MIEG3 retroviral vectors. The genes between long terminal repeats (LTR) are shown. PGK, phosphoglycerate kinase; pac, puromycin-N-acetyltransferase; H, HA epitope tag; F, FLAG epitope tag; IRES, internal ribosome entry site; EGFP, enhanced green fluorescent protein.

FIG. 2. Rac expression and NADPH oxidase activity in neutrophils differentiated in vitro.
After transduction with MSCV-pac retrovirus vectors, or mock transduction, murine bone marrow cells were cultured with IL-3 and G-CSF for differentiation in vitro as described under "Materials and Methods." MSCV-pac retrovirus-transduced cells were selected in puromycin. Wild type, rac2 Ϫ/Ϫ , or rac2 Ϫ/Ϫ cells expressing exogenous Rac proteins are as indicated. Emp, empty (MSCV-pac lacking a Rac cDNA); R1, Rac1; HR1, HA-tagged Rac1; R2, Rac2; FR2, FLAG-tagged Rac2. A, immunoblot of extracts prepared from in vitro-differentiated neutrophils, using anti-Rac1, anti-Rac2, or anti-p38 MAPK, as indicated. Note that epitope-tagged Rac proteins migrated slightly slower than untagged Rac proteins, and Rac1-transduced rac2 Ϫ/Ϫ cells contain both endogenous and exogenous Rac1. Representative of four experiments. B and C, are representative time courses of ROS production by in vitro differentiated neutrophils detected by chemiluminescence (for fMLP-stimulated cells; B, or reduction of cytochrome c (PMA-stimulated superoxide production (C)). RLU, relative luminescence units. mOD, optical density (ϫ 10 Ϫ3 ). D and E, quantitative analysis of NADPH oxidase activity in in vitro differentiated neutrophils. Superoxide dismutase-inhibitable luminescence (integrated RLUs over 100 s) is shown for movement between cell types were made using the Wilcoxon-Mann-Whitney rank sum test with KaleidaGraph software (Synergy Software, Reading, PA). Oriana software (Kovach Computing Services, Pentraeth, Wales) was used to plot circular histograms and for statistical analysis of directional data.

Expression of Recombinant Rac Proteins in Neutrophils
Differentiated in Vitro-In initial studies, recombinant Rac proteins were expressed in rac2 Ϫ/Ϫ neutrophils differentiated in vitro. Rac2 Ϫ/Ϫ myeloid progenitor cells were transduced with MSCV-based retroviral vectors designed to express Rac1 or Rac2. In preliminary studies using MIEG-based retroviral vectors for transduction followed by in vitro differentiation, recombinant Rac proteins were expressed at levels that were 3-to 6-fold greater than endogenous levels in wild-type neutrophils. 3 Therefore, subsequent in vitro studies utilized the MSCV-pac backbone, which contains a linked antibiotic resistance gene, puromycin-N-acetyl-transferase (pac) (Fig.  1B), where exogenous Rac proteins were not overexpressed (see below). Vectors included those in which an epitope tag was placed on the N terminus of Rac (an HA tag for Rac1 or a FLAG tag for Rac2) as well as those for expression of Rac1 or Rac2 without an epitope tag. Following transduction, bone marrow cells were differentiated in vitro with granulocytecolony stimulating factor and IL-3 with or without puromycin, as described under "Materials and Methods," with ϳ80 -90% morphologically mature neutrophils present after 12-14 days. In the absence of selection, no differences in expansion or neutrophil differentiation were observed between WT or rac2 Ϫ/Ϫ BM that was either mock transduced or transduced with a control vector containing only the puromycin cassette. There were also no significant differences in expansion or neutrophil differentiation between puromycin-selected WT and rac2 Ϫ/Ϫ cells transduced with the control vector containing only the puromycin cassette, or in rac2 Ϫ/Ϫ cells transduced with Rac-containing vectors.
To evaluate expression of Rac isoforms in neutrophils differentiated in vitro, we performed immunoblotting using antibodies specific for either Rac1 or Rac2 ( Fig. 2A). Recombinant Rac proteins were expressed from the MSCV-pac vectors at ϳ60% of endogenous levels, based on densitometric analysis of blots from four independent experiments (not shown). Placement of an epitope tag retarded the migration of the recombinant Rac proteins, as expected ( Fig. 2A), but did not affect the relative level of expression. Endogenous Rac1 levels in the in vitro differentiated rac2 Ϫ/Ϫ neutrophils were not significantly increased from wild-type levels ( Fig. 2A).
NADPH Oxidase Activity in Wild Type and rac2 Ϫ/Ϫ Neutrophils Differentiated in Vitro-Quantitative assessment of NADPH oxidase activity in neutrophils differentiated in vitro was performed using a chemiluminescence assay to detect fMLP-induced production of reactive oxidants and the cytochrome c reduction assay to measure superoxide production in cells stimulated with phorbol myristate acetate (PMA). Rac2 Ϫ/Ϫ neutrophils had substantially reduced NADPH oxidase activity compared with WT neutrophils (Fig. 2), as previously reported for neutrophils harvested either from the bone marrow storage pool or from peritoneal exudate (8,12,15,26). For both genotypes, there was no significant difference in oxi-dant production between neutrophils derived from mock transduced cells compared with puromycin-selected cells transduced with the control MSCV-pac vector (not shown). Expression of either exogenous Rac1 or Rac2 increased NADPH oxidase activity in Rac2-null cells stimulated with either fMLP or PMA. The time course of oxidant production in representative experiments is shown in Fig. 2 (B and C), and aggregate data are shown in Fig. 2 (D and E). Of note, Rac derivatives with an epitope tag had a reduced ability to rescue NADPH oxidase activity in rac2 Ϫ/Ϫ neutrophils compared with exogenous Rac proteins lacking a tag, although both epitope-and non-epitopetagged exogenous Rac proteins were expressed at similar levels ( Fig. 2A).
Expression of Rac2 was consistently slightly more effective than exogenous Rac1 in enhancing NADPH oxidase activity in rac2 Ϫ/Ϫ neutrophils (e.g. Fig. 2B), although the difference in activity between the Rac2-and Rac1-transduced groups did not reach statistical significance. Only partial reconstitution of NADPH oxidase activity was observed even in Rac2-transduced rac2 Ϫ/Ϫ neutrophils, which may reflect the fact that exogenous Rac proteins were expressed at only ϳ60% of wildtype levels using the MSCV-pac vectors (see above). Neutrophils isolated from heterozygous rac2 ϩ/Ϫ BM, which have approximately one half the level of Rac2 protein compared with rac2 ϩ/ϩ neutrophils, also have reduced superoxide production compared with wild-type neutrophils (8). Similarly, NADPH oxidase activity in rac2 ϩ/Ϫ neutrophils differentiated in vitro as above is reduced by ϳ25% or ϳ50% following PMA or fMLP stimulation, respectively. 3 Retrovirus-mediated Expression of Exogenous Rac1 and Rac2 in Neutrophils Differentiated in Vivo-We next examined whether expression of exogenous Rac1 or Rac2 in rac2 Ϫ/Ϫ neutrophils differentiated in vivo could reconstitute NADPH oxidase activity and motility. The above results using rac2 Ϫ/Ϫ neutrophils differentiated in vitro with hematopoietic growth factors indicated that exogenous expression of either isoform improved NADPH oxidase activity, with expression of exogenous Rac2 only slightly more effective than Rac1. Granulocyte macrophage-colony stimulating factor or tumor necrosis factor-␣ treatment of neutrophils isolated from rac2 Ϫ/Ϫ mice can partially correct defective NADPH oxidase activity and chemoattractant-induced actin formation (15), raising the question that exposure to pharmacologic concentrations of growth factors during the in vitro differentiation process may have influenced these results. Thus, we transplanted Rac-transduced rac2 Ϫ/Ϫ bone marrow cells into lethally irradiated rac2 Ϫ/Ϫ mice to study function of neutrophils produced in vivo. Rac derivatives without epitope tags were utilized for these studies.
FIG. 3. Rac expression and NADPH oxidase activation of neutrophils isolated from WT, rac2 ؊/؊ , and rac2 ؊/؊ mice transplanted with Rac1-or Rac2-transduced BM. Rac2 Ϫ/Ϫ BM was transduced with MIEG3 retroviral vectors for expression of either Rac1 or Rac2, sorted for EGFP-positive cells, and transplanted into rac2 Ϫ/Ϫ mice as described under "Materials and Methods." Immunoblots and NADPH oxidase assays were performed on freshly isolated neutrophils from transplanted mice. Rac2 Ϫ/Ϫ control mice were transplanted with untransduced rac2 Ϫ/Ϫ BM. Wild type, rac2 Ϫ/Ϫ , or rac2 Ϫ/Ϫ cells expressing exogenous Rac proteins are as indicated, and abbreviations are as in Fig. 2. A, immunoblot of extracts prepared from BM neutrophils, using anti-Rac1, anti-Rac2, or anti-p38MAPK, as indicated. Representative of three experiments. Note that the antibody for Rac2 has some cross-reactivity with Rac1 (15), which accounts for the faint band in Rac1-transduced rac2 Ϫ/Ϫ extracts probed for Rac2. B and C, representative time course of neutrophil ROS production detected by chemiluminescence (for fMLP-stimulated cells, B) or reduction of cytochrome c (PMA-stimulated superoxide production, C). RLU, relative luminescence units. mOD, optical density (ϫ 10 Ϫ3 ). The fraction of EGFP-positive neutrophils for Rac1-and Rac2-transduced rac2 Ϫ/Ϫ neutrophils was 84 and 92%, respectively. D and E, quantitative analysis of transplantation likely reflects provirus silencing in a subset of transduced repopulating cells, as observed in other studies where transduced BM cells are sorted ex vivo prior to transplantation (e.g. Ref. 27). As shown in a representative immunoblot in Fig. 3A, expression levels of exogenous Rac1 and Rac2 in neutrophils harvested from rac2 Ϫ/Ϫ mice receiving transduced marrow were similar to endogenous levels in rac2 ϩ/ϩ neutrophils. An exception was one recipient of MIEG3-Rac2transduced BM, in which neutrophil expression of exogenous Rac2 was only ϳ35% of wild-type levels (not shown), which was excluded from the analyses of NADPH oxidase activity and chemotaxis below.
NADPH Oxidase Activity of rac2 Ϫ/Ϫ Neutrophils Produced in Vivo following Transduction for Expression of Rac1 or Rac2-We examined NADPH oxidase activity in neutrophils harvested from rac2 Ϫ/Ϫ mice hematopoietically reconstituted with Rac-transduced rac2 Ϫ/Ϫ BM cells and that had 60 -92% EGFP-positive neutrophils. As shown in Fig. 3, expression of exogenous Rac2 in rac2 Ϫ/Ϫ neutrophils restored fMLP-and PMA-stimulated activity to wild type levels. However, increased expression of Rac1 resulted in a much smaller increase in oxidant production, and the time course of fMLP-stimulated oxidant production resembled that of rac2 Ϫ/Ϫ neutrophils, lacking the rapid onset of oxidant production characteristic of fMLP-stimulated rac2 ϩ/ϩ neutrophils (Fig. 3B). Taken together, these results indicate that the Rac2 isoform is essential for normal NADPH oxidase activity in neutrophils produced in vivo and that increased expression of Rac1 cannot compensate for its absence in rac2 Ϫ/Ϫ neutrophils.
NADPH Oxidase Activity in rac2 Ϫ/Ϫ Neutrophils Expressing Rac1 and Rac2 Derivatives Chimeric for the Polybasic Domain-Rac1 and Rac2 have the greatest sequence divergence in the polybasic domain adjacent to the prenylation site (Fig.  1A). To determine whether this domain could mediate isoform specificity in NADPH oxidase activity in neutrophils produced in vivo, we utilized cDNAs in which the sequence encoding the polybasic domain in Rac1 was replaced with the corresponding sequence from the human Rac2 (referred to as Rac1-2), and vise versa (referred to as Rac2-1) (11) (Fig. 1A). These cDNAs were cloned into MIEG3 retroviral vectors (Fig. 1B), which were used to transduce rac2 Ϫ/Ϫ BM cells. Lethally irradiated rac2 Ϫ/Ϫ mice were transplanted as above with transduced cells sorted for EGFP expression. EGFP-positive neutrophils in transplanted mice ranged from ϳ75-100%, and expression levels of Rac1-2 and Rac2-1 proteins were similar to endogenous Rac1 and Rac2 (Fig. 4A). The mobility of the chimeric Rac proteins was also similar to endogenous Rac, consistent with normal prenylation of the chimeras. Of note, the polybasic domain appears to be the dominant epitope recognized by the Rac1 and Rac2 antibodies used in this study, because the Rac1-2 protein was detected by the Rac2 antibody and the Rac2-1 protein was detected by the Rac1 antibody (Fig. 4A).
Analysis of NADPH oxidase activity in neutrophils harvested from rac2 Ϫ/Ϫ mice transplanted with Rac1-2-or Rac2-1-transduced marrow cells showed that the polybasic domain is sufficient to determine isoform specificity (Fig. 4, B and C). fMLP-or PMA-stimulated NADPH oxidase activity in Rac1-2expressing rac2 Ϫ/Ϫ neutrophils was similar to that observed for wild-type neutrophils, including the time course of oxidant production, whereas expression of Rac2-1 produced only a very small, if any, increase in NADPH oxidase activity. Interest-ingly, NADPH oxidase activity in rac2 Ϫ/Ϫ neutrophils expressing exogenous Rac1 was greater than in rac2 Ϫ/Ϫ cells expressing Rac2-1 (compare Figs. 3 and 4).
Wild type murine neutrophils have similar amounts of Rac1 and Rac2, and total cellular levels of Rac are diminished by ϳ2-fold in rac2 Ϫ/Ϫ neutrophils (8). We found that increasing cellular Rac levels by retrovirus-mediated expression of Rac2 in rac2 Ϫ/Ϫ neutrophils in vivo restored NADPH activity and chemotaxis, whereas expression of additional Rac1 only partially corrected these defects. Moreover, isoform specificity could be determined solely by sequences derived from the C-terminal polybasic domain that is adjacent to the prenylation site. A chimeric Rac1-2 protein in which the Rac1 polybasic domain was replaced with that of human Rac2 functioned in a manner similar to Rac2 in reconstituting the NADPH oxidase and chemotaxis in rac2 Ϫ/Ϫ neutrophils, whereas expression of an analogous Rac2-1 chimera did not.
A variety of molecular mechanisms could potentially account for distinct roles of Rac1 and Rac2. Although these two isoforms have an identical amino acid sequence in the effector domain, differences in the flexibility of this loop identified in molecular modeling studies, presumably due to long-range effects of divergent residues, may result in differential binding to Rac guanine nucleotide exchange factors or downstream targets (9). In addition, differences in a proline-rich region just N-terminal to the polybasic domain determine relative interactions with the adaptor protein Crk (29) and those at positions 90, 107, 147, and 151 influence the ability of the Rho-inactivating toxin, Escherichia coli cytotoxic necrotizing factor 1, to preferentially degrade Rac1 compared with other Rac isoforms (30). However, the results of the current study indicate that the polybasic domain (residues 183-188) is sufficient to determine the relative ability of Rac 1 and 2 to regulate superoxide production and chemotaxis in neutrophils.
The Rac polybasic domain has previously been shown to be an important determinant in a variety of cellular processes. In fibroblast and epithelial cell lines, the polybasic domain directs localization of Rac1 and Rac2 (upon their activation or when expressed in excess of Rho guanine-nucleotide dissociation inhibitor) to either the plasma membrane or to perinuclear endosomal membranes, respectively (10). Deletion of the polybasic domain in Rac2 impairs its biologic activities in neutophils, which in conjunction with the CAAX motif directs membrane localization (31). The Rac1 polybasic domain contains a nuclear localization signal, and Rac1 expression in the nucleus of cultured fibroblasts and epithelial cells has been reported, which may regulate Rac1 levels via proteasome-mediated degradation (10,32,33). The polybasic domain can also influence interactions with proteins that regulate Rac-GTP levels or are downstream targets of Rac. For example, Rac1 binds to and activates the serine-threonine kinase PAK1 more efficiently than Rac2, with the converse observed in chimeric proteins in which the polybasic domains were swapped (11). Phosphoinositide 5-kinase activation in epithelial cell lines is dependent on the C-terminal polybasic domain of Rac1, to which it binds more efficiently compared with the Rac2 polybasic domain (29). Co-immunoprecipitation of Rac1 with the formin/ diaphanous-related protein FHOS is also mediated by its polybasic domain (34). Differences in the analogous hypervariable domain in the C terminus of mammalian Ras isoforms have also been shown to determine specifically in localization and interaction partners (35). Hence, the non-redundant role of Rac2 in regulating neutrophil responses could reflect polybasic domain-mediated differences in the localization of activated Rac to subcellular membrane compartments or micro domains, the relative affinities with proteins that regulate Rac activation or serve as downstream targets, or a combination of these factors.
Although expression of Rac2 or Rac1-2 was required for full reconstitution of superoxide production in rac2 Ϫ/Ϫ neutrophils in vivo, expression of exogenous Rac1 did produce small improvements. In rac2 Ϫ/Ϫ neutrophils, Rac1 is already activated to a ϳ3-fold greater extent compared with wild-type neutrophils (8), and the current results suggest that increased expression of this isoform can improve phorbol ester-or fMLP-stimulated NADPH oxidase activity in rac2 Ϫ/Ϫ neutrophils. Rac1 was also recently shown to be the predominant form of Rac in human monocytes, where it was suggested to function as the main isoform for superoxide production elicited by PMA and other agonists (36). Rac1 translocates to monocyte membranes in response to PMA, fMLP, or opsonized zymosan, and coimmunoprecipitates with the p47 phox and p67 phox NADPH oxidase subunits upon opsonized zymosan stimulation (36). Of note, murine neutrophils with a combined deficiency of both Rac1 and Rac2 have extremely low superoxide production (28).
Also noteworthy is that expression of the Rac2-1 chimera containing the Rac1 polybasic domain did not increase NADPH oxidase activity in rac2 Ϫ/Ϫ neutrophils, in contrast to the small improvements seen with expression of exogenous Rac1. In addition, rac2 Ϫ/Ϫ neutrophils expressing Rac2-1 exhibited disoriented cell motility. The underlying basis for these observations remains to be determined. However, the different responses when expressing exogenous Rac1 compared with Rac2-1 are not easily understood in terms of a mechanism for isoform specificity that is based solely on polybasic domain-directed localization of activated Rac.
Placement of an epitope tag on proteins is a useful strategy FIG. 5. Chemotaxis of neutrophils isolated from WT, rac2 ؊/؊ , and Rac2 ؊/؊ transplanted with Rac-transduced BM. Experimental design and abbreviations are as described in Fig. 3 and 4. Chemotaxis assays performed on freshly isolated neutrophils, using a modified Boyden chamber in response to 1 or 10 M fMLP as described under "Materials and Methods." The ligand was loaded in the lower wells and the cells (2 ϫ 10 5 ) were loaded in upper wells separated by a 3-m pore size filter membrane, followed by incubation at 37°C for 45 min. The number of migrated cells was subsequently counted by microscopic examination. Results are expressed as the percentage of migrated cells per field compared with WT control neutrophils. Means Ϯ S.E. are shown. *, p Ͻ 0.002 (versus rac2 Ϫ/Ϫ ); **, p Ͻ 0.0005 (versus rac2 Ϫ/Ϫ ). p values are based on unpaired Student's t test (n Ն 3).
for distinguishing an exogenous protein from an endogenous one. Although not the focus of this report, it is of interest that placement of either an HA or FLAG tag on the N terminus of Rac1 and Rac2, respectively, blunted their ability to enhance NADPH oxidase activity in rac2 Ϫ/Ϫ neutrophils. This suggests that these tags impaired Rac activation and/or downstream signaling. To our knowledge, this is the first time epitope tagging of a small GTPase has been reported to interfere with FIG. 6. Videomicroscopy analysis of fMLP-induced motility in neutrophils isolated from WT, rac2 ؊/؊ , and rac2 ؊/؊ transplanted with Rac-transduced BM. Freshly isolated neutrophils were exposed to a linear gradient of 0 -10 M fMLP in a Dunn chemotaxis chamber, and the positions of cells were recorded at 10-s intervals over a 15-min period. See also accompanying movies S2-S6 and supplementary Fig. S1. A, histogram of the mean speed distribution of WT, rac2 Ϫ/Ϫ , and Rac1-and Rac2-transduced rac2 Ϫ/Ϫ cells. Data were derived from three independent experiments. B, histogram of the mean speed distribution of WT, rac2 Ϫ/Ϫ , and Rac1-2-and Rac2-1-transduced rac2 Ϫ/Ϫ cells. Data were derived from three independent experiments. C, circular histograms showing the distributions of cell trajectories of WT cells or rac2 Ϫ/Ϫ cells transduced with Rac2, Rac1-2, or Rac2-1 relative to the orientation of the gradient (0°). Plots show the mean direction and 95% confidence limit for the trajectories where there was a significant non-uniform distribution. Each plot shows a representative experiment from a total of three independent experiments with similar results. its function, although most studies typically involve overexpression in heterologous cell systems where such effects might be easily missed.
Results from this and our previous (8,12,15) studies indicate that Rac2 is a key regulator of oxidant production and motility. The assembled NADPH oxidase complex includes activated Rac, which binds to p67 phox and also likely to flavocytochrome b 558 (2,37). In the cell free NADPH oxidase system, recombinant prenylated Rac1 and Rac2 have equal potential for activating NADPH oxidase using purified flavocytochrome b 558 and recombinant cytosolic phox proteins (38). Although NADPH oxidase activity elicited by opsonized zymosan is unaffected in rac2 Ϫ/Ϫ neutrophils, these cells exhibit decreased superoxide production in response to fMLP, PMA, or IgG-opsonized particles (8,12,15). The agonistselective defects in NADPH oxidase activity in rac2 Ϫ/Ϫ neutrophils suggest that Rac2 regulates upstream events and/or is more efficiently incorporated into the oxidase complex compared with Rac1 in response to these agonists. The first scenario is supported by the observation that Rac2-null neutrophils have other functional defects in response to agonists that are associated with reduced NADPH oxidase activity (8,12,15). The parallel effects of exogenous Rac1, Rac2, and chimeric Rac proteins on chemotaxis and superoxide production by rac2 Ϫ/Ϫ neutrophils in this study are also consistent with roles in shared signaling pathways.
Our observations also support a critical role for Rac2 in cell orientation in chemotaxis. We previously showed that rac2 Ϫ/Ϫ cells, in addition to their impaired ability to move, are not polarized in the direction of the chemotactic gradient (15). We show here that Rac1-2, as well as Rac2 itself, fully rescued the ability of cells to move and, moreover, to properly orient their motility in a chemotactic gradient. Rac2-1 cells displayed a greater defect in chemotaxis in Boyden chamber assays than the frequency and speed of migrating Rac2-1 cells might predict, but the discrepancy can easily be explained by the remarkably disoriented nature of their motility. This is consistent with the accepted critical role for Rac in chemotactic signaling pathways (39 -41) and illuminates possible isoform-specific roles in this process that have not previously been recognized. The Rac GTPases are proposed to be a component of a positive feedback loop involving the GTPase and phosphatidylinositol 3-kinase that is important for establishment of cell polarity and the maintenance of a "front-ness" signal at the leading edge (39,40). Although we were unable to definitively determine whether Rac1-expressing cells showed directed migration (due to their poor motility and consequent failure to reach the 30-m horizon required for the analysis), the disoriented nature of the motile response displayed by Rac2-1-expressing cells is highly suggestive of a critical role for factors that specifically interact with the Rac2 polybasic domain in establishing cell polarity and orientation. Our results suggest the need for more work to determine interactions that specify distinct roles for Rac1 and Rac2 in regulating the cytoskeleton in hematopoietic cells.
While this manuscript was in preparation, a study by Filippi and colleagues (42) reported findings generally consistent with those described here, with a notable difference with respect to reconstitution of neutrophil motility. Similar to our results, migration of fMLP-stimulated rac2 Ϫ/Ϫ neutrophils in Boyden chambers was reconstituted by expression of Rac2. However, rescue by a Rac1-2 chimera additionally required replacement of a glycine at position 150 with aspartic acid, the residue at the corresponding position in Rac2 (42). This contrasts with our findings, where expression of a Rac1-2 chimera was sufficient to reconstitute fMLP-stimulated chemotaxis in rac2 Ϫ/Ϫ neutrophils (Figs. 5 and   6). However, the study by Filippi et al. used the murine Rac2 polybasic domain, which differs from the human Rac2 sequence used here, in having a proline instead of an alanine at position 188 (Fig. 1A). Other factors that may have contributed to the different results are that Filippi et al. used neutrophils differentiated in vitro and exogenous Rac proteins that were HA-tagged and relatively overexpressed (42).
Filippi and coworkers (42) also reported differences in localization of EGFP-tagged Rac proteins in murine neutrophils, which may contribute to their different functions. In resting cells, EGFP-Rac1 was dispersed compared with a more central cytoplasmic and perinuclear localization for EGFP-Rac2. Upon fMLP stimulation both forms were also detected at more peripheral locations, with Rac1 co-localizing with cortical actin and Rac2 being somewhat more interior. Moreover, in fMLPstimulated rac2 Ϫ/Ϫ neutrophils, EGFP-Rac1 was poorly organized, suggesting that there may be cross-talk between the two Rac isoforms. Because deficient F-actin formation is a notable feature of fMLP-stimulated rac2 Ϫ/Ϫ neutrophils (15), Filippi et al. proposed that Rac1 distribution may be correlated to F-actin organization and that Rac1 dysfunction in rac2 Ϫ/Ϫ neutrophils might contribute to their abnormal chemotaxis. We are currently undertaking studies to examine Rac1 and Rac2 localization in neutrophils isolated from bone marrow or peripheral blood. In preliminary studies on unstimulated murine neutrophils stained with Rac1 and Rac2 antibodies or expressing small amounts of exogenous EGFP-Rac1 or EGFP-Rac2, both isoforms were distributed diffusely in the cytosol. 3 These results are consistent with studies on resting human neutrophils, where Rac2 is cytosolic and in a complex with Rho guaninenucleotide dissociation inhibitor (43) but differ from the localization described by Filippi and coworkers (see above) (42). However, Rac proteins were relatively overexpressed in the latter studies, possibly exceeding Rho-guanine-nucleotide dissociation inhibitor binding capacity (10) and also studied in neutrophils differentiated in vitro using relatively high concentrations of growth factors. Neither we 3 nor Fillipi et al. (42) detected a significant pool of nuclear Rac1 in neutrophils.
In conclusion, these results establish that Rac2 plays a nonoverlapping role with Rac1 to regulate neutrophil chemotaxis and NADPH oxidase activity and demonstrate that the Cterminal polybasic domain is sufficient for conferring this specificity. Future studies will address underlying mechanisms that discriminate between Rac1 and Rac2, which may include differences in subcellular targeting as well as binding to interacting regulatory or downstream target proteins.