Different Roles of G Protein Subunits β1 and β2 in Neutrophil Function Revealed by Gene Expression Silencing in Primary Mouse Neutrophils*

Neutrophils play important roles in host innate immunity and various inflammation-related diseases. In addition, neutrophils represent an excellent system for studying directional cell migration. However, neutrophils are terminally differentiated cells that are short lived and refractory to transfection; thus, they are not amenable for existing gene silencing techniques. Here we describe the development of a method to silence gene expression efficiently in primary mouse neutrophils. A mouse stem cell virus-based retroviral vector was modified to express short hairpin RNAs and fluorescent marker protein at high levels in hematopoietic cells and used to infect mouse bone marrow cells prior to reconstitution of the hematopoietic system in lethally irradiated mice. This method was used successfully to silence the expression of Gβ1 and/or Gβ2 in mouse neutrophils. Knockdown of Gβ2 appeared to affect primarily the directionality of neutrophil chemotaxis rather than motility, whereas knockdown of Gβ1 had no significant effect. However, knockdown of both Gβ1 and Gβ2 led to significant reduction in motility and responsiveness. In addition, knockdown of Gβ1 but not Gβ2 inhibited the ability of neutrophils to kill ingested bacteria, and only double knockdown resulted in significant reduction in bacterial phagocytosis. Therefore, we have developed a short hairpin RNA-based method to effectively silence gene expression in mouse neutrophils for the first time, which allowed us to uncover divergent roles of Gβ1 and Gβ2 in the regulation of neutrophil functions.

Neutrophils are the most abundant leukocytes in the blood and play an essential role in the early stages of the innate immune responses by ingesting and killing invading pathogens. In response to inflammatory stimuli, neutrophils first adhere to and extravasate through blood vessels and then migrate through the interstitial tissue toward the site of inflammation. Although neutrophils play an important role in the host defense, uncontrolled inflammatory reactions are associated with a variety of pathological conditions, including ischemia-reperfusion injury during heart attack and strokes, arteriosclerosis, rheumatoid arthritis, and allergic reactions (1)(2)(3). Therefore, understanding how neutrophil recruitment and function is reg-ulated is critical for developing potential treatments for a number of disorders.
Neutrophils are also a fine model system to study directional cell migration and chemotaxis because of their ability to migrate rapidly and directionally under a shallow gradient of chemoattractants. Chemotaxis is a fundamental biological process used by a variety of cell types and underlies a wide range of developmental, physiological, and pathophysiological events. It consists of two basic components, directionality and motility (4 -8). Most of the neutrophil chemoattractants, including formyl-Met-Leu-Phe (fMLP), 2 bind to their specific cell surface receptors that are coupled to heterotrimeric G proteins, which upon activation regulate numerous downstream effectors. Studies in Dictyostelium cells demonstrated that the G␤ subunit plays a critical role in signal transduction to chemotaxis regulation (9,10). However, the importance of G␤ subunits in neutrophil chemotaxis has yet been investigated.
Despite basic scientific and clinical importance of neutrophil research, it has been hindered by the fact that these cells are terminally differentiated, short lived, and thus not amenable to in vitro manipulations, including transfection. Targeted gene inactivation in mice has been the only approach to provide primary neutrophils for loss of function studies. Although these studies have provided novel insights into neutrophil biology, the approach is costly and time-consuming. Therefore, it is imperative to develop a more efficient approach to study neutrophils.
RNA interference has been shown to be a rapid and powerful tool for knocking down gene expression in a sequencespecific fashion. Compared with chemically synthesized small interfering RNA, short hairpin RNAs (shRNAs) can be stably expressed in hard-to-transfect primary cells and in whole organisms (11). Recently, the miR30-shRNA cassette has been reported to yield a higher level of shRNA and more efficient knockdown than a simple shRNA design (12). Using the same miR30-shRNA cassette, Zhu et al. (13) have silenced multiple target genes simultaneously to overcome isoform redundancy issues (14), demonstrating the powerful capabilities of this microRNA-based shRNA design. Although shRNA-mediated gene silencing was successful in neutrophil-like cell lines, such as differentiated HL-60, these cells may not faithfully recapitu-late the biology of primary neutrophils. In addition, these cell lines cannot be used for in vivo studies. By taking advantage of pluripotent differentiation potentials of hematopoietic stem cells, we describe the development of a novel method to silence gene expression in primary neutrophils in mice using a retrovirus expressing shRNAs. Using this new method, we demonstrated that G␤ 2 is primarily involved in regulating directionality rather than motility of neutrophil chemotaxis, and G␤ 1 is involved in bacterial killing by neutrophils.

MATERIALS AND METHODS
Plasmids-The initial retroviral vector MIGR1 (mouse stem cell virus-internal ribosome entry site-green fluorescent protein-retrovirus-1) and the packaging vector pCL-ECO were generous gifts from Dr. Diane Krause (Yale University). For the LTR-YFP-shG␤ 2 vector, we replaced GFP in MIGR1 with YFP-miR-shG␤ 2 , which was amplified by PCR from pSLIK-miR-shG␤ 2 and carries a targeting sequence of TGCT-CATGTATTCCCACGACAA (14,15). For the CMV-YFP-shRNA vectors, the CMV promoter was amplified from the pAAV-MCS vector (Stratagene (Cedar Creek, TX)) by PCR and inserted between the LTR and YFP-miR-shRNA. miR-shRNA sequences for G␤ 1 knockdown were designed by using the RNA interference Codex algorithm (available on the World Wide Web). Four targeting sequences were tested (A, GCGACTCT-TTCTCAGATCACAA; B, ATCTGGGACAGTTATACCA-CAA; C, AACATTATCTGTGGTATCACAT; D, GGCCGAG-CAACTGAAGAACCAA). The D sequence was used for the knockdown studies in neutrophils. To generate plasmid containing tandem shRNAs, miR30-shG␤ 2 was amplified by PCR and inserted downstream of miR-G␤ 1 .
Td-Tomato was amplified and inserted into pGEX-3X (GE Healthcare) for the bacterial killing assay. All constructs were verified by sequencing.
Cell Culture-We used PHE (Phoenix Ecotropic) cells, which are capable of producing viral gag-pol and envelope proteins, as the packaging cell line (16). PHE and NIH 3T3 cells were maintained in 90% Dulbecco's modified Eagle's medium, 10% fetal bovine serum, 100 units/ml penicillin, and 0.1 mg/ml streptomycin at 37°C under 5% CO 2 in humidified air.
Retrovirus Production and Infection-One day prior to transfection, PHE cells were seeded at a density of 3 ϫ 10 6 cells/T 75 flask. Retroviral vector and pCL-ECO were co-transfected into cells using Lipofectamine Plus (Invitrogen) according to the manufacturer's instructions. Two days later, the medium containing viruses was concentrated using Amicon Ultra columns (Millipore, Billerica, MA). For viral titer determination, serial dilutions of virus supernatants were added to NIH 3T3 cells in a 24-well plate, followed by centrifugation at room temperature to enhance the transduction efficiency. Twenty-four hours later, cells were analyzed by a flow cytometer.
Flow Cytometry Analysis-Allophycocyanin (APC)-CD11b, PerCP-B220, and APC-CD3⑀ antibodies were purchased from eBioscience (San Diego, CA). APC-Gr1 and Percp-Ly-6G antibodies were from BD (San Jose, CA). The single color flow analysis was done in the Guava EasyCyte Mini Base System (Millipore), and the multiple color flow analysis was performed in an LSR II fluorescence-activated cell sorting (FACS) analyzer (BD Biosciences). Cell sorting was performed by a FACS Aria sorter (BD Biosciences). Results were analyzed by FlowJo software (Treestar, Ashland, OR).
Generation of Neutrophils from Virus-infected Bone Marrow Cells-C57BL/6 mice, aged 8 -12 weeks, were obtained from Taconic (Germantown, NY). The donor mice were treated with 5-fluorouracil (Sigma) at 150 mg/kg to enrich hematopoietic stem cells in the bone marrow. Three days later, bone marrow cells were harvested from femurs and tibias and cultured in Iscove's modified Dulbecco's medium containing 20% endotoxin-free fetal bovine serum (Invitrogen), 50 ng/ml recombinant murine stem cell factor, 10 ng/ml recombinant murine interleukin-3, and 10 ng/ml recombinant human interleukin-6 (PeproTech, Rocky Hill, NJ) for 2 days. Cells were then infected twice with viruses in the presence of 4 g/ml Polybrene (Millipore). The transduced bone marrow cells were transplanted back into lethally irradiated recipient mice (9.5 grays; ␥-rays) by retro-orbital injection. Eight weeks later, the transplanted mice were euthanized for the further analysis. The use and care of animals were approved by the Institutional Animal Care and Use Committee at Yale University.
Neutrophil Preparation and Dunn Chamber Chemotaxis Assay-Neutrophils were isolated from bone marrows by using discontinuous Percoll gradient as described previously (17). The purity of the preparation was verified by Ly6G staining, which is above 95% pure for neutrophils. fMLP was purchased from Sigma. The chemotaxis assay using a Dunn chamber was carried out as described previously (18) with some modifications. To minimize inconsistency between assays, we monitored chemoattractant gradients using free fluorescein isothiocyanate dye. Only cells under certain gradient characteristics were analyzed and included in our statistical analysis (supplemental Fig. S1A). Time lapse image series acquired in the aforementioned chemotaxis experiments were analyzed using the MetaMorph image analysis software. The software calculates the x,y coordinate of the centroid of a cell that is designated for tracking in each of the images. Several parameters that reflect the chemotactic behaviors of a cell are obtained from these timed coordinate series. Because the cells in general do not move with a constant velocity, they often move a distance that is less than the noise level of the cell tracking algorithm in the short time interval between two consecutive frames. To exclude those meaningless tracking results from the data analysis, we set a cut-off of 7.5 m, which represents onehalf of an average body length of a polarized neutrophil, as the minimal distance that a cell has to travel in order for the coordinate to be included in the following analyses.
We computed the following parameters to quantify the chemotactic behaviors of a cell (supplemental Fig. S1B). Assuming that a cell migration trace consists of n points (p 1 , p 2 , . . . p n ) in that order so that they have coordinates (x i ,y i ) i ϭ 1,2. . . . n , respectively, let ⌬x i ϭ x i ϩ 1 Ϫ x i ,⌬y i ϭ y i ϩ 1 Ϫ y i . Then the distances between consecutive points are as follows.
The directionality error angle (␣ i ) was calculated as follows, This measures the angle between the cell migration direction and the gradient direction. We take the average directionality angle for all p i values. A smaller value indicates that the cell is more closely following the gradient. Motility is calculated as follows, motility ϭ ͉p n Ϫ p 0 ͉/elapsed time. This measures the overall cell migration speed.
Air Pouch Model of Carrageenan-induced Synovitis--Carrageenan was purchased from Sigma. Air pouches were generated by subcutaneously injecting 3 ml of sterile-filtered air on day 0 and day 3. At day 6, carrageenan suspension (5 mg in 0.5 ml of sterile, pyrogen-free saline) was injected into the air pouch. 4 h later, pouch exudates were recovered with 1 ml of cold phosphate-buffered saline containing 3 mM EDTA (19). The total cell count was measured using the Guava EasyCyte Mini Base System (Millipore). Cells were then stained with APC-CD11b and Percp-Ly-6G and subjected for flow cytometric analysis using an LSR II FACS analyzer. Neutrophils were defined as Ly-6G ϩ CD11b ϩ cells.
Phagocytosis and Bacterial Killing Assay-Escherichia coli (DH10B) transformed with pGEX-Td-Tomato plasmid was induced by isopropyl 1-thio-␤-D-galactopyranoside to express Td-Tomato, the red fluorescence protein. Bone marrow was isolated from mouse tibia and femur of hind legs and then incubated at 37°C in the presence of bacteria expressing red fluorescence at a ratio of 1:10 for 15 min. Non-internalized bacteria were then removed by washing three times by centrifugation at 100 ϫ g for 5 min in Hanks' buffer with 1% bovine serum albumin, and fluorescence intensity at this time point was used as a measurement of phagocytosis. Following this initial bacterial loading, cells were incubated at 37°C for the time periods indicated prior to fixation with 2% paraformaldehyde. Bone marrow cells were then labeled with APC-conjugated anti-mouse Gr1 in order to identify the neutrophil population, and fluorescence was then assessed by flow cytometry using a BD LSRII FACS analyzer.

RESULTS
To develop an efficient method for loss-of-function study of mouse primary neutrophils, we tested whether the lentiviral miR30-embedded shRNA production system developed by Simon and colleagues (15,20) would be able to silence gene expression in neutrophils in a scheme outlined in supplemental Fig. S2. Briefly, lentiviruses expressing GFP were used to infect mouse bone marrow cells, which were subsequently used to generate neutrophils after transplantation into lethally irradiated recipient mice. We found that this lentivirus-based system did not infect mouse bone marrow cells at a high efficiency (data not shown). We decided to switch to a mouse stem cell virus-based system because mouse stem cell virus infects hematopoietic stem/progenitor cells at high efficiencies (21). The insert encoding Venus fluorescence protein (a variant of YFP) and an shRNA targeting the ␤ 2 subunit of G protein embedded within the miR30 sequence was subcloned from the lentiviral system into MIGR1, a mouse stem cell virus-based retroviral vector (Fig. 1A). We refer to this vector as LTR-YFP-shG␤ 2 . The reasons for choosing the G␤ 2 shRNA in this study were as follows: 1) the shRNA was shown to be effective and specific (15); 2) G␤ 2 is one of the two G␤s abundantly expressed in mouse neutrophils based on our expression microarray analysis (Fig. 1B); and 3) the role of G␤ in neutrophil chemotaxis has not been investigated.
We infected bone marrow cells isolated from C57BL/6 mice with LTR-YFP-shG␤ 2 virus and transplanted them into lethally irradiated C57BL/6 recipients. After 8 weeks of recovery and repopulation, neutrophils were isolated from the transplanted mice and analyzed for YFP expression. As shown in Fig. 1C, 44.92% of isolated neutrophils were YFP-positive, suggesting that this retroviral vector provides reasonable infection efficiency. We subsequently sorted the YFP-positive neutrophils (Fig. 1D) and analyzed them for shRNA-mediated knockdown efficiency. Because YFP and the shRNA were expressed from the same transcript, we expected the YFP-positive cells to express the shRNA. We observed about a 50% reduction in G␤ 2 expression in neutrophils expressing YFP (Fig. 1E). However, little effect on chemoattractant (fMLP)-induced phosphorylation of Akt at Ser-473 was observed. This phosphorylation event is known to depend on G␤␥ (14,22). Therefore, we hypothesized that the LTR promoter-based vector may not produce sufficient shRNA to allow effective knockdown of G␤ 2 and result in the expected functional defects.
To improve the shRNA production, we decided to add a strong transcriptional promoter/enhancer by inserting a cytomegalovirus (CMV) promoter with a ␤-globin intron between the 5Ј-LTR and the YFP-miR30-shRNA expression cassette ( Fig. 2A). The ␤-globin intron enhances transcriptional activities as an enhancer-like element (23). A second construct with the luciferase shRNA embedded in the miR30 backbone in place of the G␤ 2 shRNA was made as a control for our studies described below. These two new constructs are referred to as CMV-YFP-shG␤ 2 and CMV-YFP-shLuc. We first tested these new constructs by infecting NIH 3T3 cells and found that cells infected with viruses generated from the new constructs (CMV-YFP-shLuc or -G␤ 2 ) expressed higher levels of YFP than those generated from LTR-YFP-shG␤ 2 at the same multiplicity of infection (Fig. 2, B and C). Importantly, despite the same G␤ 2 shRNA sequence in both LTR-and CMV-driven vectors, CMV-YFP-shG␤ 2 yielded much more efficient knockdown of G␤ 2 expression than LTR-YFP-shG␤ 2 (Fig. 2D). These results suggest that the CMV promoter coupled with the ␤-globin intron sequence leads to higher expression levels than the viral LTR and that higher shRNA expression translates into more effective gene expression knockdown.
Next, we tested whether the modified vector would be more efficient in silencing gene expression in mouse neutrophils. The CMV-YFP-shLuc and CMV-YFP-shG␤ 2 viruses were used to infect mouse bone marrow cells, which were subsequently transplanted into lethally irradiated recipient mice. Fig. 3A shows a representative set of flow cytometric analyses of neutrophils isolated from these transplanted mice. In this experiment, 30% of CMV-YFP-shG␤ 2 neutrophils were YFP-positive, whereas close to 60% of CMV-YFP-shLuc neutrophils were YFP-positive. The YFP-positive neu-trophils were isolated by FACS, and their purity is shown in Fig. 3B. The sorted YFP-positive neutrophils were then subjected to Western analysis. As shown in Fig. 3C, the level of G␤ 2 was markedly reduced in sorted YFP-positive neutrophils expressing shG␤ 2 compared with neutrophils from non-transplanted mice or neutrophils expressing shLuc. In all of the experiments performed, there was 70 -85% reduction in G␤ 2 expression levels in cells expressing shG␤ 2 compared with the controls. Importantly, there was a clear reduction in fMLP-induced Akt phosphorylation in cells expressing YFP-shG␤ 2 compared with the controls (Fig. 3C). Of note, silencing G␤ 2 expression in neutrophil did not affect the expression of G␤ 1 or G␣ i2 (Fig. 3C).
Because G␤ 1 is also highly expressed in murine neutrophils (Fig.  1B) and it may have a redundancy function as G␤ 2 , we also tried to silence G␤ 1 expression using our new vector. Four different shRNAs targeting G␤ 1 were designed, and the most potent shRNA (shG␤ 1 -D) was identified by significantly suppressing endogenous G␤ 1 expression in NIH 3T3 cells (supplemental Fig. S3). This shRNA was used in the following study. After bone marrow transplantation, we sorted out the YFP-positive neutrophils transduced with G␤ 1 shRNA or control luciferase shRNA (Fig. 4A) and performed Western analysis. As shown in Fig. 4B, the level of G␤ 1 was significantly reduced in sorted YFP-positive neutrophils expressing shG␤ 1 compared with control neutrophils from nontransplanted mice or neutrophils expressing shLuc.
One of the main powers of the miR30-shRNA system is to simultaneously knock down multiple different gene targets (13). To exploit this possibility in our system, we tried a double knockdown of G␤ 1 and G␤ 2 in neutrophils. The second miR-shRNA was inserted downstream of the first miR-shRNA, and both shRNAs were driven by the enhanced CMV promoter (Fig. 4C). Similarly, we sorted out the YFP-positive neutrophils transduced with G␤ 1 shRNA and G␤ 2 shRNA (double knock- down) or control luciferase shRNA (Fig. 4D), and we observed the efficient suppression of G␤ 1 and G␤ 2 expression in neutrophils (Fig. 4E), suggesting the successful knockdown of multiple gene targets in vivo.
Next, we performed three functional assays for transduced neutrophils. First, we tested the effect of G␤ knockdown on in vitro neutrophil chemotaxis using a Dunn chamber, in which a shallow fMLP gradient was established and cell migration was tracked using time lapse videomicroscopy. Two key parameters were obtained from analysis of the time lapse image series: directionality error (reflecting how well a cell follows the chemoattractant gradient) and motility (see "Materials and Methods" for details). Significant portions of the neutrophils isolated from transplanted mice were YFP-negative. Supplemental Fig. S4A shows bright field and fluorescence images of neutrophils in Dunn chambers, in which YFP-positive and -negative neutrophils can be readily recognized and tracked. These YFP-negative cells have little or no shG␤ 2 expression and therefore normal levels of G␤ 2 protein (supplemental Fig. S4, B and C). Thus, these YFP-negative cells served as excellent internal controls for the Dunn chamber chemotaxis assay, which are known to be variable. In addition, paired statistical analyses were used to identify small differences in chemotaxis parameters caused by gene silencing. Analysis of chemotactic parameters of YFP-positive cells and negative cells revealed that YFP-positive neutrophils isolated from CMV-YFP-shG␤ 2 virus-infected mice showed impaired directionality compared with YFP-negative cells (Fig. 5A). Interestingly, no significant directionality defect was detected in G␤ 1 -silenced neutrophils compared with control cells (Fig. 5A), and no further directionality defect was observed in G␤ 1 and G␤ 2 double knockdown neutrophils as compared with G␤ 2 -silenced cells (Fig. 5A), suggesting that G␤ 2 is the major isoform responsible for neutrophil directionality. The effect of G␤ 1 knockdown or G␤ 2 knockdown on motility appeared to be insignificant (Fig. 5B). Neutrophils suppressing both G␤ 1 and G␤ 2 expression (Fig. 5B) showed a modest but significant reduction in motility. In addition, ϳ30% of G␤ 1 /G␤ 2 double knockdown neutrophils failed to respond to fMLP and were immobile, whereas knockdown of either G␤ subunit had no significant effects on the number of responding cells (Fig. 5C). As important controls, there were no differences between YFP-positive and -negative neutrophils isolated from mice transplanted with CMV-YFP-shLuc virus-infected bone marrow cells (Fig. 5, A  AUGUST 6, 2010 • VOLUME 285 • NUMBER 32 and B). Therefore, these results together indicate that G␤ 2 , but not G␤ 1 , has an important role in neutrophil directionality regulation, whereas both G␤ subunits are involved in motility regulation.

Roles of G␤ 1 and G␤ 2 in Neutrophil Function
To examine neutrophil recruitment in vivo, we used the air pouch model (24). In this model, subcutaneous injection of air into dorsal surface of mice results in the formation of an air pouch, which has a lining morphologically similar to the synovium. Carrageenan, polysaccharide extracted from red seaweeds, was injected into the pouch to induce robust inflammation, as indicated by an increase in total neutrophil number and higher neutrophil percentage in the pouch exudates compared with those in mice injected with saline (data not shown). Among these neutrophils recruited into the pouch, two or three populations (YFP-low, YFP-medium, and YFP-high) were observed based on their YFP expression (Fig. 6A). In our analysis, we refer to YFP-high cells as YFP-positive cells and YFPlow cells as YFP-negative cells. As indicated in supplemental Fig. S4, YFP-negative cells also served as internal controls for YFP-positive cells in this assay. For each animal, the ratio between YFP-positive neutrophils and YFP-negative neutrophils in the pouch exudates (Fig. 6A, bottom), which reflects the migration ability of transduced neutrophils, was normalized based on the same ratio in the blood (Fig. 6A, top). As expected, neutrophils expressing luciferase shRNA (shLuc) migrate normally in vivo (the ratio of YFP ϩ /YFP Ϫ is 0.97 Ϯ 0.05; Fig. 6B). However, the infiltration of G␤ 2 -silenced neutrophils was markedly inhibited (the ratio of YFP ϩ /YFP Ϫ is 0.47 Ϯ 0.12). Although we observed the inhibited migration in G␤ 1silenced neutrophils (the ratio of YFP ϩ /YFP Ϫ is 0.72 Ϯ 0.22; Fig. 6B), the difference is not statistically significant compared with that in shLuc group. Importantly, the more severe migration defect was detected in G␤ 1 and G␤ 2 double knockdown neutrophils (the ratio of YFP ϩ /YFP Ϫ is 0.29 Ϯ 0.11; Fig. 6B).
These data suggest that both G␤ subunits play important roles in neutrophil recruitment in vivo, whereas G␤ 2 appears to be a more dominant isoform than G␤ 1 in regulating neutrophil migration in vivo.
Finally, we performed phagocytosis and bacterial killing assays using neutrophils isolated from transplanted mouse bone marrow. We developed a flow cytometry-based assay to assess these two processes. We incubated neutrophils with E. coli expressing Td-Tomato, a red fluorescence protein (25), at 37°C for 15 min to allow phagocytosis to occur. After differential centrifugation to remove the free bacteria, the levels of Td-Tomato were determined by flow cytometry at varying time points. The initial level of red fluorescence reflects phagocytosis, whereas the reduction of red fluorescence reflects bacterial killing. To validate this assay, we compared this new method with the one that counts actual bacterial numbers, and we observed the similar results between this flow-based method and the conventional colony-forming unit method (supplemental Fig. S5B). In the presence of diphenyliodonium, an inhibitor of NADPH oxidase complex, no decrease in red fluorescence was observed over the 30-min experimental time  course, whereas we did detect a significant reduction of red fluorescence in the absence of diphenyliodonium inhibitor (supplemental Fig. S5C). Moreover, we did not observe the decrease of red fluorescence when fluorescence-labeled bacteria were incubated with neutrophils in ice for 30 min (supplemental Fig. S5A). These data further demonstrate that the reduction of red fluorescence measured by flow cytometry is due to neutrophil-mediated bacterial killing.
Knockdown of G␤ 1 did not alter the ability of cells to phagocytose bacteria when the fluorescence intensity of YFP Ϫ and YFP ϩ cells was compared after the 15-min ingestion period (Fig. 7, A and C). Knockdown of G␤ 2 caused a slight but statistically insignificant decrease (Fig. 7E). However, knockdown of both G␤ subunits led to a significant reduction (Fig. 7G), suggesting that both subunits are involved in the regulation of phagocytosis. With regard to bacterial killing, only knockdown of G␤ 1 demonstrated a moderate, but significant, inhibition over the time course indicated (Fig. 7D). In contrast, G␤ 2 -silenced neutrophils and neutrophils expressing shLuc control showed no significant changes in bacterial killing (Fig. 7, B and F). These data suggest that signaling events downstream of G␤ 1 may facilitate the efficient destruction of phagocytosed bacteria in neutrophils. Although the mechanisms for this G␤ 1 -mediated bactericidal effect are not known, G␤ 1 may regulate this effect via regulating superoxide production or degranulation. The lack of effect of G␤ 1 /G␤ 2 double knockdown on the apparent killing (Fig. 7H) may be due to fewer bacteria that were phagocytosed by the double knockdown cells, obscuring the moderate inhibition that resulted from G␤ 1 knockdown. Nevertheless, these results are consistent with the conclusion that the lack of effect of G␤ 2 knockdown on bacterial killing may not be due to the compensation by G␤ 1 .

DISCUSSION
In this study, we have developed a method by which gene expression can be effectively silenced in primary mouse neutrophils. This method uses a combination of bone marrow transfer and microRNA-embedded shRNA production to achieve highly efficient gene expression knockdown in primary neutrophils. Using this method, we successfully silenced G␤ 1 , G␤ 2 , or both G␤ 1 and G␤ 2 in neutrophils and uncovered unexpected roles of these G protein subunits in regulating neutrophil biology.
Retroviral transduction following transplantation of genetically manipulated hematopoietic stem cells is a powerful tool for studying hematopoietic cells in vivo. This method has been successfully used in not only myeloid cells, such as macrophage (26) and dendritic cells (27), but also lymphoid cells (28 -30). However, most of studies applied gain-of-function strategy. Our new method has several distinct advantages. First, it allows loss-of-function studies of neutrophils and will provide an efficient alternative to the time-consuming gene targeting approach, which thus far had been the only way to study loss of function of neutrophils. Advancement in the field of neutrophil biology has been in part limited by the long time frame to generate genetically modified primary neutrophils. In contrast to more than 1 year needed for generation of a knock-out mouse line, our method only takes 3 months to produce the mice for the study. In addition, the method allows studying neutrophil functions in various in vivo models, as suggested in our air pouch model (Fig. 6), which is a clear advantage over knockdown studies in neutrophil-like cell lines.
The second advantage of this method is that cells unaffected by gene knockdown are generated simultaneously serving as excellent internal assay controls. This is particularly beneficial in experiments subjected to individual mouse and systematic variations, including the in vivo studies. In this study, we use YFP-negative neutrophils as internal controls for an in vitro chemotaxis assay (Fig. 5), in vitro bacterial killing assay (Fig. 7) and in vivo air pouch model (Fig. 6).
The third advantage is that this method may be adapted to express proteins in neutrophils. By generating MIGR-CMV FIGURE 5. Chemotaxis assay for neutrophils from transplanted mice. Neutrophils isolated from transplanted mice were assayed for their chemotactic responses to an fMLP gradient in a Dunn chamber. As described under "Materials and Methods," neutrophils were stimulated by fMLP and observed under a time lapse video microscope to monitor the migration of neutrophils using the Metamorph software. The directionality (A) and motility (B) of neutrophils infected with the indicated virus were determined in YFP-positive and YFPnegative fractions. The percentages of motile YFP ϩ cells relative to motile YFP Ϫ neutrophils are shown in C. *, p Ͻ 0.05 (paired Student's t test) between the YFP-positive and YFP-negative group. Error bars, S.E. vectors with different fluorescence proteins with differing emission spectra, the method expands possibilities for multiple labeling neutrophils for both loss-of-function and gain-offunction studies.
Because hematopoietic stem cells can differentiate into multiple-lineage hematopoietic cells, our approach will very likely be applicable to the other hematopoietic lineages as well. To address this question, we did FACS analysis in spleen cells of MIGR-CMV-miR-shLuc transplanted mice and wild type mice. Using B220 as a B cell marker and CD3⑀ as a T cell marker, we detected a B220 ϩ YFP ϩ double positive population and a CD3⑀ ϩ YFP ϩ double positive population in spleen of MIGR-CMV-miR-shLuc transplanted mice but not in wild type mice (supplemental Fig. S6), suggesting that the method of retroviral transduction followed by bone marrow transplantation can transduce myeloid and lymphoid cells as well as other hematopoietic cells. Therefore, our new vectors will provide important tools to study gene functions in primary hematopoietic cells in vivo and in vitro.
This study demonstrates that G ␤ 2 knockdown primarily affected directionality of neutrophil chemotaxis in a shallow fMLP gradient, whereas knockdown of G␤ 1 had little effect. On the other hand, both G␤ subunits are involved in motility regulation, because knockdown of either had insignificant effects, whereas knockdown of both resulted in a significant reduction in cell motility and the number of immotile cells. These conclusions were supported by observations in vivo using an air pouch model and therefore suggest that G␤ 1 and G␤ 2 may regulate differing aspects of some neutrophil biology. This conclusion is further supported by the observation that G␤ 1 knockdown suppressed neutrophil-mediated bacterial killing, whereas G␤ 2 knockdown was ineffective in this regard. Meanwhile, both G␤ subunits appear to be involved in bacterial phagocytosis regulation. Indeed, previous studies in a macrophage-like cell line have suggested that the roles of G␤ 1 and G␤ 2 in cell migration differ (14). This study therefore extends the idea that G␤ 1 and G␤ 2 are not entirely redundant in their functions.
Genetic studies in Dictyostelium suggest that G␤ is required for phagocytosis and chemotaxis (31). However, our knockdown of two major G␤ subunits in neutrophils led to partial effects on chemotactic responses and phagocytosis. The lack of robust effects in some of the assays may be due to incomplete inactivation of the target genes. Because the expression levels and phenotype outcomes are not linear, an apparent efficient knockdown of the expression of a gene may fail to lead to a robust phenotype. This is an obvious drawback of this type of method. In addition, other G␤ subunits may also be expressed in neutrophils although they were not detected at high levels by our microarray analyses, which may not be reliable for estimating protein levels. Nevertheless, the methods outlined in this paper allow quick determination of a given gene for its roles in regulating a wide array of neutrophil activities and lay the groundwork for more thorough future studies.