Rac2 Functions in Both Neutrophils and Macrophages To Mediate Motility and Host Defense in Larval Zebrafish

Rho GTPases are key regulators of innate immune cell functions, including cell migration, reactive oxygen species (ROS) production, phagocytosis, and degranulation (1). Within the Rho GTPases, the Rac subfamily includes the following: Rac1, 2, and 3 (2, 3), as well as RhoG. Rac1–3 are remarkably similar at the amino acid level, differing mainly at their C termini. However, their expression patterns are different: Rac2 is largely restricted to hematopoietic cell lineages, and Rac3 is thought to be brain or CNS specific, whereas Rac1 is ubiquitously expressed (2, 4–7).

A still unanswered question is whether these different Rac isoforms have distinct functions in vivo. Both rac1^−/− and rac2^−/− murine neutrophils have migration defects (8–10), but it is difficult to track cell migration in mice, and in vitro findings on the respective functions of Rac1 and Rac2 in this process remain unconfirmed in vivo (11). In vitro studies have also found differences in the roles of Rac1 and Rac2 depending on the stimulus used or the substrate on which cells are migrating. For example, the requirement for Rac2 for in vitro ROS production in neutrophils and macrophages seems to depend on whether the stimulant is serum opsonized or IgG opsonized (12–14). In vitro macrophage migration studies have also shown mixed results, with defects reported for Rac2-null macrophages in haptotaxis and Transwell assays, but not migration on plastic (15, 16).

Zebrafish larvae are an attractive system to answer these questions in vivo as they are highly amenable to live imaging and analysis of leukocyte function in response to a variety of tissue wounding and infection models. Zebrafish have a largely conserved innate immune system to humans, including having Rac1, 2, and 3, expressed similarly to humans (17–19). We previously modeled a human neutrophil immunodeficiency disorder resulting from a dominant-negative mutation in Rac2 (Rac2^D57N) in zebrafish and found that these neutrophils have motility defects resulting in increased susceptibility to Pseudomonas aeruginosa (20) and Aspergillus fumigatus (21). However, this model only expressed Rac2^D57N in neutrophils, and dominant-negative Rac2 can also inhibit Rac1 activity (20, 22, 23). These obstacles make it difficult to discern both distinct functions of Rac1 versus Rac2 and the role of Rac2 in other hematopoietic cells, and we therefore have generated rac2^−/− zebrafish to begin to address these questions.

In this work, we report that rac2^−/− zebrafish larvae have defects in both neutrophil and macrophage basic motility, leading to decreased recruitment to both tissue wounds and bacterial infections, two inflammatory environments characterized by complex mixtures of stimuli. These larvae are highly susceptible to infection with the fungus A. fumigatus and the bacteria P. aeruginosa. We find that re-expression of Rac2 in either neutrophils or macrophages can partially rescue the susceptibility of rac2^−/− larvae to Pseudomonas infection. Unexpectedly, overexpression of Rac1 in neutrophils in rac2^−/− larvae can also rescue neutrophil recruitment to a wound and partially rescue survival of larvae postinfection. Additionally, several phenotypes induced by expression of a dominant-negative Rac2 in neutrophils are not found in neutrophils in rac2^−/− larvae, suggesting that these phenotypes are caused by inhibition of other Rho GTPases, such as Rac1.

All zebrafish were maintained according to protocols approved by the University of Wisconsin-Madison Institutional Animal Care and Use Committee, as described previously (24). To prevent pigment formation, 0.2 mM N-phenylthiourea (Sigma-Aldrich) was used. Previously published zebrafish lines were used (Table I). The irf8 mutant was a gift of C. Shiau (Boston College) and was genotyped, as described previously (25). To construct a mpeg1:mCherry-2A-rac2 line, Tol2-mpeg1 (26) and the mCherry-2A-rac2 insert (20) were ligated together. To construct a mpx:mCherry-2A-rac1 line, the coding sequence for zebrafish rac1a was cloned into a Tol2 vector with mCherry and 2A peptide sequence, as described previously (20). Microinjection of 1 nl 30 ng/μl Tol2 vector and 20 ng/μl transposase mRNA into embryos from a rac2^+/− outcross was performed to establish founders.

Left and right transcription activator–like effector nuclease (TALEN) constructs were generated by a modified Golden Gate assembly (27) and cloned into pCS2TAL3-DD and -RR, respectively (Addgene 37275 and 37276) (28). Target sequences are indicated in Fig. 1B. mRNA was in vitro transcribed from pCS2 vectors, and 1 nl 100 ng/μl (exon 3) or 400 ng/μl (exon 5) each left and right mRNA was microinjected into F₀ embryos. At 1–2 d postfertilization (dpf), genomic DNA from individual larvae was isolated in 50 mM NaOH (29), and TALEN cuts were detected by high resolution melt analysis in a Roche LightCycler using High Resolution Melting Master (Roche) and exon 3 and 5 genotyping primers (Supplemental Table I). Additional F₀-injected larvae were grown up and incrossed. F₁s were fin clipped, and those carrying mutations in rac2 were identified by high resolution melt analysis. Mutations were cloned into pCR4-TOPO (Invitrogen) and sequenced. Selected mutant lines (Fig. 1B, 1C) were maintained as heterozygotes by outcrossing to wild-type or various transgenic lines (Table I) (20, 25, 30–32).

Subsequent genotyping was done by RFLP analysis. All experiments were done on either rac2^+/− incrosses or rac2^+/− × rac2^−/− crosses, and larvae were then genotyped by RFLP analysis at the end of the experiment. DNA was isolated in 50 mM NaOH (29), the mutated region amplified with GoTaq (Promega), and restriction enzyme targeting specifically either the mutant or wild-type copy (Fig. 1B) was directly added with buffer. Digests were incubated for 10 h and run on a 2.5% agarose gel to evaluate the presence of wild-type and mutant bands. Restriction enzymes used were MboII (NEB), NlaIII (NEB), HindIII (Promega), and DdeI (Promega). For adults and larvae also carrying a rac2 transgene, a reverse genotyping primer that annealed to intronic sequence and therefore was not present in the transgene (Supplemental Table I) was used.

Larvae were prescreened on a zoomscope (EMS3/SyCoP3; Zeiss; Plan-NeoFluar Z objective) for fluorescence. For longer-term imaging, larvae were mounted in a glass-bottom dish or a glass-bottom two-well μ-Slide (Ibidi) with 1% low-melting-point agarose. Images were acquired on a spinning disk confocal microscope (CSU-X; Yokogawa) with a confocal scanhead on a Zeiss Observer Z.1 inverted microscope and a Photometrics Evolve EMCCD camera. Images were analyzed, maximum intensity projections were made, and Tiffs were exported, with Zen 2012 (blue edition) software (Carl Zeiss). To track cell motility, time series were analyzed in Fiji and two-dimensional velocity was calculated using the MTrackJ plugin (33). To count total neutrophils and quantify neutrophil distribution, five to six overlapping images were acquired along the length of the larvae, assembled in Adobe Photoshop CS5, and counted with Cell Counter plugin (Fiji). To quantify neutrophil shape, neutrophils in the caudal hematopoietic tissue were manually identified in Fiji and shape descriptors were calculated (circularity = 4π(area/perimeter²); roundness = 4 × area/π × major axis²). For circulation imaging, neutrophils were manually counted from a single z-slice time lapse.

Anesthetized 3 dpf larvae were wounded by tail transection with a no. 10 Feather surgical blade. To visualize neutrophil recruitment, larvae were fixed 1 h postwounding with 4% formaldehyde in 1× PBS overnight at 4°C and sudan black staining was performed, as described previously (34). In larvae carrying a mpx:mCherry, mpeg1:Dendra2, or mpeg1:mCherry transgene, larvae were fixed at 1, 3, and 3 h postwounding, respectively, with 1.5% formaldehyde in 0.1 M PIPES (Sigma-Aldrich), 1 mM MgSO₄ (Sigma-Aldrich), and 2 mM EGTA (Sigma-Aldrich) overnight at 4°C.

Larvae 2 dpf were infected with A. fumigatus TJMP131.5 spores expressing GFP, as previously described (21). CFUs/injection were monitored by plating on glucose minimal media and are noted in each of the figure legends. Survival was monitored for 7 d postinjection (dpi). To score hyphal appearance, larvae were individually placed into wells of a 48-well plate and anesthetized and imaged every day for 5 d on a spinning disk confocal microscope, as described above (without mounting in agarose). Hyphae had to be apparent in both the GFP and bright-field channels to be scored positively.

Larvae 3 dpf were infected with P. aeruginosa PAK (pMF230) (expresses GFP) or PAK (pBad-mKalama1), as previously described (35, 36). PAK (pMF230) was a gift of S. M. Moskowitz (University of Washington). pBad-mKalama1 (a gift from R. Campbell, 14892; Addgene) (37) was transformed into the PAK strain, as described elsewhere (38). PAK (pMF230) was used for survival and neutrophil recruitment experiments, and PAK (pBad-mKalama1) was used for macrophage recruitment experiments. A single colony was inoculated overnight in LB-Amp. In the morning, the culture was diluted 1:5 and grown for additional 1.5–2.5 h, and the OD was measured (600 nm). To prepare the final inoculum, the bacterial suspension was pelleted by centrifugation for 1 min and resuspended in PBS to achieve the desired bacterial density. Phenol red dye was added to the suspension to a final concentration of 0.5% to visualize injection success. CFUs were monitored by plating on LB-Amp and are noted in each of the figure legends. For survival analysis, infected larvae were placed into wells of a 96-well plate, and survival was monitored 1–2 times per day for 5 dpi. For neutrophil recruitment experiments, larvae were fixed 2 h postinjection with 4% formaldehyde in 1× PBS overnight at 4°C, and sudan black staining was performed (34). To visualize macrophage recruitment, larvae carrying the mpeg1:Dendra2 transgene were fixed 6 h postinjection with 1.5% formaldehyde in 0.1 M PIPES (Sigma-Aldrich), 1 mM MgSO₄ (Sigma-Aldrich), and 2 mM EGTA (Sigma-Aldrich) overnight at 4°C.

All data plotted comprise at least three independent experimental replicates. For cell recruitment, velocity, number, distribution, shape, or rate in the circulation, pooled data from all replicates were compared between experimental conditions using ANOVA. The results were summarized and plotted in terms of least squares adjusted means and SEs, or individual data points were displayed and color coded by replicate. Survival data from all replicates were also pooled and analyzed using Cox proportional-hazard regression analysis, as previously described (21), and hazard ratios (HRs) are indicated in the text. Statistical analyses were done in R version 3 (R Development Core Team, 2013), and graphical representations were done in GraphPad Prism version 6.

Translating ribosome affinity purification was performed, as previously described (32). RNA was extracted from translating ribosome affinity purification samples or whole larvae using TRIzol reagent (Invitrogen), and cDNA was synthesized with SuperScript III RT and oligo-dT (Invitrogen). Using this cDNA as a template, either PCR with GoTaq (Promega) (35 cycles) or quantitative PCR (qPCR) with FastStart Essential DNA Green Master (Roche) and a LightCycler96 (Roche) was performed. For qPCR experiments, data were normalized to ef1α using the ∆∆Ct method (39). Primers are listed in Supplemental Table I. For qPCR experiments from rac2^+/− incross single embryos, genomic DNA was also isolated from the remaining TRIzol fraction, according to the manufacturer’s protocol, and genotyping was performed, as described above.

We generated rac2^−/− zebrafish lines using TALEN technology, targeting exon 3 or 5 of the rac2 gene (Fig. 1A). Mutations were determined by cloning and sequencing, and four different lines, each containing one to seven deleted bases, were established (Fig. 1B). These deletions result in frameshifts and premature stop codons, disrupting regions that are required for nucleotide coordination (Fig. 1C). qPCR of cDNA from individual larvae from a rac2^+/− incross both confirmed the loss of rac2 mRNA in rac2^−/− larvae and showed no significant compensatory upregulation of other Rho GTPases (Supplemental Fig. 1). rac2^−/− larval progeny had no observable developmental defects (Fig. 1D, 1E). However, at 3–4 mo postfertilization, rac2^−/− adults were significantly smaller than rac2^+/− clutch mates (Fig. 1F, 1G). rac2^−/− adults also appeared sickly with fin damage and a shorter life span, whereas rac2 adult heterozygotes were indistinguishable from their wild-type clutch mates (data not shown). These phenotypes have persisted through at least three generations of outcrossing, suggesting that they are not due to off-target effects of the TALEN. Rac2 is also expressed in epithelium in larval zebrafish (Supplemental Fig. 2A), and it is unclear whether these phenotypes are due to immune deficiency or some other unknown role of Rac2. We were never able to successfully incross rac2^−/− adults, and all experiments were performed on larvae from rac2^+/− incrosses or rac2^−/− × rac2^+/− crosses; all analyses were therefore blinded, and larvae were genotyped at the completion of the experiment. Mutant lines were maintained as heterozygotes by outcrossing or crossed to established lines (Table I).

We first tested the ability of neutrophils in rac2^−/− zebrafish larvae to migrate to a tail transection wound. In all four mutant lines, directed neutrophil migration to the wound was nearly abolished in rac2^−/− larvae (Fig. 2A, 2B). This rac2^−/− phenotype was completely rescued by re-expression of Rac2 specifically in neutrophils under the mpx promoter in the exon 5-B mutant line (Fig. 2C). We have previously reported that expression of Rac2 under the mpx promoter is approximately equal to the endogenous level of neutrophil expression of Rac2 (20). This rescue experiment confirms that the migration defect is due to a disruption in rac2, and all further experiments were done using this mutant line.

We next measured the ability of rac2^−/− neutrophils to randomly migrate in vivo. In zebrafish larvae, mCherry-expressing extravascular neutrophils in the head region can be live imaged and tracked over time to calculate their mean velocity. Evaluating both the speed of individual neutrophils (Fig. 2D) and the mean velocity per larvae of neutrophils (Fig. 2E), rac2^−/− neutrophils have significantly impaired random migration, with essentially no movement at all, as compared with rac2^+/− neutrophils (Supplemental Video 1). Therefore, in vivo, Rac2 is required in neutrophils for both directed migration to a tissue injury and random motility.

While imaging these rac2^−/− larvae with mCherry-labeled neutrophils, we noticed that they did not recapitulate several phenotypes we previously found in zebrafish larval neutrophils expressing a dominant-negative form of Rac2 identified in human patients with immune deficiency (mpx:rac2^D57N) (20). Although both rac2^−/− and mpx:rac2^D57N larvae have significantly fewer total neutrophils than rac2^+/− larvae (∼80%) (Fig. 3A), mpx:rac2^D57N larvae have a large alteration in the distribution of neutrophils throughout the body, with fewer in the head and more in the pericardium, which was not as severe in rac2^−/− larvae (Fig. 3B, 3C). Another striking phenotype of neutrophils expressing Rac2^D57N is their altered polarity and shape (20). mpx:rac2^D57N neutrophils are significantly more circular and round than rac2^+/− neutrophils, but, interestingly, this phenotype is not present in rac2^−/− neutrophils (Fig. 3D–F). A greater number of neutrophils is also observed in the circulation in mpx:rac2^D57N larvae (20), as in humans with this mutation (22, 40); however, this phenotype is not observed in rac2^−/− larvae (Fig. 3G). This dominant-negative form of Rac2 can also inhibit Rac1 (20, 22), and we hypothesize that Rac2^D57N inhibits the activity of other GTPases to confer these phenotypes. Neutrophils in larval zebrafish express rac1a, rac1b, rac3a, and rac3b genes (Supplemental Fig. 2A), any of which might be inhibited by Rac2^D57N. Supporting this hypothesis, expression of mpx:rac2^D57N confers this phenotype regardless of the endogenous rac2 genotype of larvae (Fig. 3H).

Rac2 contributes to the response to infection, and we next investigated the susceptibility of rac2^−/− larvae to pathogens that infect immunocompromised patients. One such pathogen is the fungus A. fumigatus, to which rac2^−/− mice are susceptible (8). We infected larvae with Aspergillus spores via hindbrain injection, a previously established model (21). rac2^−/− larvae succumbed to Aspergillus infection at a significantly greater rate than rac2^+/− (HR = 2.4) or rac2^+/+ (HR = 5.1) larvae, with ∼45% of rac2^−/− larvae dying by 7 dpi (Fig. 4A). No death was seen in PBS-injected larvae (Supplemental Fig. 3A). As zebrafish larvae allow for imaging of the infection over the course of multiple days, we observed that this susceptibility was correlated with increased fungal growth, as measured by the appearance of hyphae (Fig. 4B, 4C, Supplemental Video 2). Hyphal appearance was significantly more likely to be found in rac2^−/− larvae than rac2^+/− (HR = 3.3) or rac2^+/+ (HR = 2.7) larvae.

Another opportunistic pathogen of immunocompromised patients is P. aeruginosa. Using a localized Pseudomonas infection model (20, 41), we found that rac2^−/− larvae are highly susceptible to this infection, with virtually 100% of larvae dying after only 1 d of infection (Fig. 4D). Conversely, rac2^+/− and rac2^+/+ larvae had ∼0% death (HR rac2^−/− versus rac2^+/− = 65.1, versus rac2^+/+ = 131.1). No death was seen in PBS otic-injected larvae (Supplemental Fig. 3B). Together, these data support the conclusion that Rac2 is required for resistance to multiple pathogens in vivo in larval zebrafish.

As this bimodal phenotype of survival or susceptibility after Pseudomonas infection was so striking, we used this model to further define the role of Rac2 in different cell types in response to infection. Neutrophils are thought to be the primary cell type required for resistance to Pseudomonas infection (42). Significantly fewer neutrophils were recruited to a Pseudomonas otic infection in rac2^−/− larvae compared with rac2^+/− and rac2^+/+ larvae (Fig. 5A, 5B), mirroring the rac2^−/− defect in neutrophil recruitment to a wound. Neutrophils are also recruited to a PBS injection, as this creates a wound, but the additional cues provided by Pseudomonas significantly increased neutrophil recruitment in control larvae, but not in rac2^−/− larvae (Fig. 5A). Importantly, reconstituting Rac2 activity exclusively in neutrophils almost completely restored survival after Pseudomonas infection (Fig. 5C). Only ∼10% of rac2^−/−; mpx:rac2 rescue larvae succumbed to infection with Pseudomonas, significantly less than rac2^−/− larvae with no rescue (HR = 9.3) (Fig. 5C).

To what extent Rac1 and Rac2 have distinct functions in vivo in response to infection is still unclear. Although larval zebrafish neutrophils also express rac1, their expression of rac2 is ∼1.5-fold higher (Supplemental Fig. 2B), and rac1 transcripts are not upregulated in rac2-deficient larvae (Supplemental Fig. 1). To determine whether ectopic Rac1 expression can restore any neutrophil function in rac2^−/− larvae, we introduced a mpx:rac1 transgene to specifically overexpress Rac1 in neutrophils. This expression of Rac1 in neutrophils also significantly improved the survival of rac2^−/− larvae to Pseudomonas infection (HR rac2^−/− versus rac2^−/−; mpx:rac1 = 3.8) (Fig. 5D). However, Rac1 does not provide as complete a rescue as Rac2: ∼25% of these larvae still succumb to infection, significantly more than rac2^−/−; mpx:rac2 larvae (HR = 5.6). We next measured the effect of Rac1 overexpression on the directed migration of rac2^−/− neutrophils and found that Rac1 almost completely restored neutrophil numbers at a tail wound (Fig. 5E). Together, these data suggest that Rac1 and Rac2 have partially redundant functions in neutrophils, and that the level of this redundancy may vary for different Rac-dependent functions.

Although rescue of Rac expression in neutrophils significantly restored survival of rac2^−/− larvae after Pseudomonas infection, ∼10–25% of these larvae still died, significantly more death than seen in heterozygous controls (HR rac2^−/−; mpx:rac2 versus rac2^+/−; mpx:rac2 = 8.7, rac2^−/−; mpx:rac1 versus rac2^+/−; mpx:rac1 = 9.1) (Fig. 5C, 5D). These observations suggest that Rac2 also functions in other cell types in the host response to Pseudomonas in vivo. We therefore predicted that rac2^−/− larvae that lack functional Rac2 throughout the whole animal would be more susceptible to Pseudomonas infection than mpx:rac2^D57N larvae that express dominant-negative Rac2 only in neutrophils (20). Indeed, at a Pseudomonas dose in which nearly 100% of rac2^−/− larvae succumbed to infection, only ∼50% of mpx:rac2^D57N larvae did (HR = 3.7) (Fig. 6A), indicating that, in 50% of infections, Rac2 function in another cell type besides neutrophils is sufficient for survival.

rac2 mRNA is translated in macrophages at nearly the same level as in neutrophils at this stage of zebrafish development (Supplemental Fig. 2). To determine whether the presence and function of macrophages account for the difference in survival between rac2^−/− and mpx:rac2^D57N larvae, we crossed the mpx:rac2^D57N transgenic line with a recently published irf8 mutant line that prevents the development of macrophages during early larval stages (25). Although the irf8 mutation alone had no effect on survival after Pseudomonas infection, irf8^−/−; mpx:rac2^D57N larvae were nearly 100% susceptible to Pseudomonas infection, recapitulating the phenotype of rac2^−/− larvae (HR irf8^−/−; mpx:rac2^D57N versus irf8^+/+; mpx:rac2^D57N = 3.6) (Fig. 6B), and supporting a role for Rac2 in macrophages in the absence of functional neutrophils. In fact, when we rescued Rac2 expression specifically in macrophages using the mpeg1 promoter (26), only ∼30% of rac2^−/−; mpeg1:rac2 larvae succumbed to Pseudomonas infection, significantly less than rac2^−/− larvae (HR = 4.0) (Fig. 6C). These data indicate that Rac2 function in macrophages also mediates resistance to Pseudomonas infection, although these larvae still have significantly more death than control rac2^+/− larvae (HR = 19.5).

We next wondered whether macrophages in rac2^−/− larvae have motility defects. In vitro, rac2^−/− murine bone marrow–derived macrophages have defects in some migration assays, but not in others (15, 16). In vivo, rac2^−/− mice were found to have fewer macrophages in a peritoneal exudate (13), but no decrease in the number of macrophages recruited to a cutaneous wound (43). It is therefore unclear to what extent Rac2 is required for macrophage migration in vivo.

To directly measure the effect of Rac2 disruption on macrophage migration in vivo in larval zebrafish, we used the same tail transection assay previously used for neutrophils. Significantly fewer macrophages migrated to a tail transection wound in rac2^−/− larvae compared with rac2^+/− or rac2^+/+ larvae (Fig. 7A). However, this defect was more modest than the defect in neutrophil migration in rac2^−/− larvae; ∼80% of macrophages still reached the wound. Rescue of Rac2 expression in macrophages (Fig. 7B) but not in neutrophils (Fig. 7C) restored wild-type macrophage migration to the wound, indicating that rac2^−/− macrophages have cell-intrinsic migration defects, and that this decrease is not due to a lower level of neutrophil-produced chemoattractants at the wound. Also supporting this conclusion, rac2^−/− macrophages had defects in random motility in tissue (Supplemental Video 3). Imaged and tracked extravascular macrophages in the head of larvae had significantly lower mean velocities than macrophages in control larvae (Fig. 7D). The mean velocity of all macrophages per fish was also lowered in rac2^−/− larvae, although the difference was not significant (Fig. 7E). A similar migration defect was also observed in macrophages responding to infection: lower numbers of macrophages in rac2^−/− larvae were recruited to the site of Pseudomonas infection compared with rac2^+/− larvae (Fig. 7F, 7G). These data indicate that, as in neutrophils, at least one mechanism through which Rac2 in macrophages promotes host resistance to Pseudomonas infection is likely through a general role in basic motility that is required for directed migration to the site of infection.

Although Rac GTPases have been studied for >25 y, the exact in vivo role and relative contribution of Rac2 versus other Rho GTPases such as Rac1 in Rho GTPase-dependent processes in different cell types have remained unclear. In this work, to our knowledge and using larval zebrafish, we present for the first time a comparison of the in vivo role of Rac2 in neutrophil and macrophage basal migration and response to tissue damage and infection.

A previously developed rac2^−/− mouse model has shown reduced numbers of neutrophils and macrophages recruited to inflammatory stimuli (8, 13, 44); however, due to the difficulty of live imaging in mice, the exact nature of these migration deficits in vivo was unknown. Using a larval zebrafish model that is highly amenable to live imaging, we find that rac2^−/− leukocytes have defects not just in directed migration to an inflammatory stimulus—either a tissue wound or an infection—but also in basal random migration, suggesting a deficit in basic motility, not chemokine sensing. This agrees with a previous in vitro study showing that Rac2 is required in neutrophils for chemokinesis, not just chemotaxis (11). The fact that leukocytes have migration deficits to both tissue wounds and infections, both inflammatory stimuli that create a vast array of chemokinetic signals, also underlines the role of Rac2 in basic motility versus chemotaxis. Additionally, in comparing the effect of rac2 disruption on neutrophils versus macrophages, we find that Rac2 has a larger role in neutrophil motility in vivo. In rac2^−/− larvae, macrophage numbers at a wound (∼80% of wild type) or infection (∼50% of wild type) and the mean velocity of basally migrating macrophages (∼70% of wild type) are significantly diminished. However, in neutrophils both directed migration and random migration are essentially abolished. This difference in the effect of Rac2 deficiency on neutrophil versus macrophage motility could either be due to a difference in the relative importance of Rac2 versus other Rho GTPases or due to a difference in the mode of migration used and the dependence of that mode on Rac function.

We have also presented data contributing to the understanding of whether Rac1 and 2 have distinct roles in neutrophils in vivo. Larval zebrafish are a good model in which to interrogate this question, as zebrafish neutrophils more closely recapitulate the expression pattern of Racs in human neutrophils, with higher levels of Rac2 compared with Rac1 (Supplemental Fig. 2), whereas in mice the level of these two proteins in neutrophils is approximately equal (19, 45). Different hypotheses have been postulated for why different isoforms of Rac exist, including the idea that hematopoietic cells require higher levels of actin polymerization and therefore express a second hematopoietic-specific copy of Rac2 in addition to the ubiquitously expressed Rac1 (8). Supporting this idea, we find that overexpression of Rac1 in neutrophils of rac2^−/− larvae can almost fully rescue neutrophil recruitment to a tissue wound and can partially rescue susceptibility to Pseudomonas infection. However, Rac1 and Rac2 proteins also have differences in their C-terminal amino acid sequences that have been reported to lead to differential localization and activity (46–50), and these differences are thought to confer some isoform-specific functions. One of these sequence differences (which is conserved in zebrafish) (150D/G) has been implicated in Rac2-specific function in in vitro migration of neutrophils to fMLP (51). We hypothesize therefore that Rac1 can complement in vivo migration to a wound because a wound produces a complex mixture of signals, not just fMLP, not all of which require this Rac2-specific amino acid–mediated function. Indeed, stimulus-specific differences in the requirement for Rac2 in leukocytes have been reported before (12–14).

We also find that several phenotypes present in larval neutrophils expressing a dominant-negative form of Rac2 (20) are not present in rac2^−/− neutrophils, including altered distribution, a rounded shape correlated with loss of polarity, and increased release into the circulation. It is therefore likely that these phenotypes are due to inhibition of Rac1 or a different Rho GTPase. The Rac2^D57N mutant exerts its dominant-negative effect through sequestration of guanine nucleotide exchange factors (22, 40), and any other Rho GTPase that requires the activity of a Rac2^D57N-sequestered guanine nucleotide exchange factor could be inhibited by this mutant. Supporting this hypothesis, Rac2^D57N expression can confer these phenotypes in the absence of endogenous Rac2. Accordingly, expression of other dominant-negative RhoGTPases, Rac1 (52) or Cdc42 (53), has conferred a similar cell-rounding phenotype in other studies. We cannot fully discount the possibility that ectopic expression of Rac2^D57N under the mpx promoter causes overexpression artifacts, although this transgene is expressed at a level approximately equal to endogenous expression in neutrophils and these phenotypes are not induced by ectopic expression of Rac2^WT (20).

The release of neutrophils into the circulation from the caudal hematopoietic tissue in larval zebrafish is functionally similar to hematopoietic stem cell/progenitor (HSC/P) mobilization into the blood in humans and mice. As in mpx:rac2^D57N zebrafish larvae, human patients in which this Rac2 mutation has been found have also had neutrophilia (22, 40, 54). In mice, Rac2 deficiency alone does cause neutrophilia and increased HSC/P mobilization (8, 55); however, the combination of Rac2 and Rac1 deficiency in hematopoietic cells causes massive HSC/P mobilization, much higher than in either single mutant (56, 57), supporting a role for both of these GTPases in this process. Interestingly, the retroviral re-expression of Rac1 in these HSC/Ps can partially reverse this defect (57). These observations and our data therefore demonstrate that both Rac1 and Rac2 play a role in HSC/P mobilization; however, it is still unclear whether these roles are overlapping or distinct.

In the response to infection, we find that Rac2 function in either neutrophils or macrophages can partially rescue larval susceptibility to Pseudomonas. Although macrophages are not necessary for survival after Pseudomonas infection (100% of irf8 mutants lacking macrophages survive), consistent with experiments in mice indicating that neutrophils are the predominant cell type required for resistance to Pseudomonas infection (42), in the absence of Rac2 function in neutrophils, Rac2 function in macrophages is sufficient for survival in 50–60% of infections. It is unclear whether the major function that Rac2 is playing in macrophages in response to infection is in migration. Although our data demonstrating that macrophages in rac2^−/− larvae have reduced migration to inflammatory stimuli are consistent with experiments in mice (13), data on rac2^−/− macrophage motility in vitro have been less clear, with studies showing defects in actin dynamics, migration through Transwell filters, and haptotaxis, but not invasion through Matrigel or migration speed on plastic (15, 16). Rac2 has also been shown to be involved in phagocytosis (13, 58), ROS production (13), and gene expression (59, 60) in macrophages in vitro, and this rac2^−/− model will be useful in the future to investigate the function of Rac2 specifically in these other Rac-dependent processes that can be live imaged in zebrafish larvae in vivo.

We thank members of the Huttenlocher laboratory for useful discussion and critiques, as well as help with zebrafish care and maintenance; Jens Eickhoff for advice on statistical analyses; and Elizabeth A. Harvie for generation of the mpeg1:mCherry-2A-rac2 Tol2 vector. We also thank C. Shiau for providing the irf8 mutant and S. Moskowitz for providing the PAK (pMF230) strain.

The authors have no financial conflicts of interest.