Knockout mouse models reveal the contributions of G protein subunits to complement C5a receptor–mediated chemotaxis

G protein–coupled receptor signaling is required for the navigation of immune cells along chemoattractant gradients. However, chemoattractant receptors may couple to more than one type of heterotrimeric G protein, each of which consists of a Gα, Gβ, and Gγ subunit, making it difficult to delineate the critical signaling pathways. Here, we used knockout mouse models and time-lapse microscopy to elucidate Gα and Gβ subunits contributing to complement C5a receptor-mediated chemotaxis. Complement C5a-mediated chemokinesis and chemotaxis were almost completely abolished in macrophages lacking Gnai2 (encoding Gαi2), consistent with a reduced leukocyte recruitment previously observed in Gnai2−/− mice, whereas cells lacking Gnai3 (Gαi3) exhibited only a slight decrease in cell velocity. Surprisingly, C5a-induced Ca2+ transients and lamellipodial membrane spreading were persistent in Gnai2−/− macrophages. Macrophages lacking both Gnaq (Gαq) and Gna11 (Gα11) or both Gna12 (Gα12) and Gna13 (Gα13) had essentially normal chemotaxis, Ca2+ signaling, and cell spreading, except Gna12/Gna13-deficient macrophages had increased cell velocity and elongated trailing ends. Moreover, Gnaq/Gna11-deficient cells did not respond to purinergic receptor P2Y2 stimulation. Genetic deletion of Gna15 (Gα15) virtually abolished C5a-induced Ca2+ transients, but chemotaxis and cell spreading were preserved. Homozygous Gnb1 (Gβ1) deletion was lethal, but mice lacking Gnb2 (Gβ2) were viable. Gnb2−/− macrophages exhibited robust Ca2+ transients and cell spreading, albeit decreased cell velocity and impaired chemotaxis. In summary, complement C5a-mediated chemotaxis requires Gαi2 and Gβ2, but not Ca2+ signaling, and membrane protrusive activity is promoted by G proteins that deplete phosphatidylinositol 4,5-bisphosphate.

The study of immune cell chemotaxis dates back to 1888 (1), when the German ophthalmologist Th. Leber (Theodor Karl Gustav Leber), using a model of mycotic keratitis, observed the migration of leukocytes to sites of inflammation, where they accumulated in large numbers, reminiscent of the swarming of neutrophils to injured tissue described more recently by Lämmermann et al. (2). Gradients of chemoattractant ligands emanating from sites of inflammation or injured tissue guide cells by binding to G protein-coupled receptors (GPCRs), but how spatiotemporal ligand-GPCR signaling accomplishes this has not been delineated. Dictyostelium discoideum has proven to be a good model system for the study of chemotaxis (3), leading to the development of the LEGI (local excitation, global inhibition) model (4) and more complex excitable network hypotheses (5). In contrast to Dictyostelium amoebae, mammalian chemoattractant receptors, such as the complement C5a receptor (C5aR), may activate more than one type of heterotrimeric G protein (6 -8), which adds considerable complexity to the signal transduction. Nevertheless, the G␣ i/o family has been strongly implicated in chemotactic signaling because pertussis toxin (PTX; previously known as lymphocytosis-promoting factor) from Bordetella pertussis, which blocks the activation of G␣ i/o subunits by GPCRs, inhibits the chemotaxis of macrophages and other phagocytes (9 -11). Moreover, the recruitment of neutrophils and macrophages to inflamed lung or peritoneum is decreased by around 50% in mice lacking Gnai2, which encodes G␣ i2 (12,13).
Genetic deletion of Gna15, a hemopoietic-specific gene coding for the G␣ q /G␣ 11 family member G␣ 15 , markedly decreases complement C5a-induced Ca 2ϩ transients in macrophages (14), but the role of G␣ 15 in complement C5a-mediated chemotaxis has not been determined. The other members of the G␣ q /G␣ 11 family are encoded by Gnaq (G␣ q ) and Gna11 (G␣ 11 ), respectively, in mouse, but the roles of these subunits in chemotaxis are unclear. In cotransfection studies, the chemokine (chemotactic cytokine) receptors for CCL2 (chemokine (CC motif) ligand 2; also known as monocyte chemoattractant protein-1) and CCL5 (also known as RANTES (regulated on activation, normal T cell expressed and secreted)) were shown to couple to G␣ q (6). In contrast, CXCL8 (chemokine (CXC motif) ligand 8; also known as interleukin-8) and complement C5a did not activate G␣ q (6). However, complement C5a may indirectly couple to G␣ q and/or G␣ 11 through C5aR-induced autocrine ATP signaling that activates purinergic receptors (15,16).

cro ARTICLE
In this study, we used knockout mouse models to explore the roles of different G protein subunits in macrophage chemotaxis. More specifically, we used time-lapse microscopy to visualize the navigation and movement of macrophages, isolated from WT or various knockout mice, in a chemotactic complement C5a gradient, and, in parallel, we imaged the Ca 2ϩ signaling and lamellipodial membrane dynamics in individual cells upon C5aR stimulation.

Gnai2, but not Gnai3, is critical for complement C5a-mediated chemotaxis
RNA sequence analysis of purified resident peritoneal F4/80 ϩ cells (macrophages) revealed that Gnai2 (G␣ i2 ) and Gnai3 (G␣ i3 ) of the G␣ i/o family of G␣-subunits, as well as Gnas (G␣ s ) and members of the G␣ q /G␣ 11 (Gnaq, Gna11, and Gna15) and G␣ 12 /G␣ 13 (Gna12 and Gna13) families, are expressed, as well as complement C5a receptor 1 (as known as CD88), encoded by C5ar1 (Fig. 1, A and B). Notably, the controversial C5aR, complement C5a receptor 2 (17), encoded by C5ar2, was negligibly expressed (see the inset with interrupted y axis in Fig. 1A). We used knockout mouse models to explore the roles of the various G␣ and G␤ subunits shown in Fig. 1B, except Gnas (G␣ s ), in transducing gradients of the chemoattractant complement C5a into stimulated motility (chemokinesis) and directed cell migration (chemotaxis) using the -Slide Chemotaxis chamber (16,18) and time-lapse, phase-contrast microscopy. Macrophages isolated from WT mice migrated robustly along chemotactic gradients of complement C5a ( Fig.  1C and Videos S1 and S2). Incubation with PTX (1 g/ml), Figure 1. Expression of G␣ subunits in macrophages and roles of Gnai2 and Gnai3 in complement C5a-mediated chemotaxis. A, expression levels of G␣ subunits in mouse resident peritoneal F4/80 ϩ cells (macrophages). RNA-Seq analysis was performed using RNA isolated from resident peritoneal F4/80 ϩ cells purified by cell sorting (n ϭ 3 mice). Inset (superimposed graph with an interrupted y axis), expression levels of receptors for complement components 3a and 5a. Error bars, S.E. B, schematic diagram showing C5aR, a member of the G protein-coupled receptor superfamily, and a heterotrimeric G protein in which the subunits are color-coded blue (G␣), green (G␤), and white (G␥). The four G␣ families (G␣ i/o , G␣ s , G␣ q/11 , and G␣ 12/13 ) are listed below the blue ␣-subunit (G␣) together with the names of the corresponding genes investigated with knockout mouse models, including two genes encoding ␤-subunits: Gnb1 (G␤ 1 ) and Gnb2 (G␤ 2 ). C, migration plots of WT, PTX-treated WT, Gnai2 Ϫ/Ϫ , and Gnai3 Ϫ/Ϫ macrophages in a chemotactic complement C5a gradient. D, summary box plots of cell velocity and chemotactic efficiency (chemotaxis index), calculated by dividing the displacement along the y axis by the cumulative distance migrated. The chemotaxis index is also known as the y-forward migration index and has a range of Ϫ1 to ϩ1. *, p Ͻ 0.05; Kruskal-Wallis test and post hoc Mann-Whitney U test with Bonferroni correction (n ϭ 75 for each group, except n ϭ 50 for the WT ϩ PTX group; three independent experiments, except two independent experiments for the WT ϩ PTX group).

G proteins and chemotactic signaling
which ADP-ribosylates a serine residue and inhibits G␣ i2 and G␣ i3 as well as other G␣ i/o family members (19), abolished complement C5a-mediated chemokinesis and chemotaxis (Fig.  1C). Similarly, genetic deletion of Gnai2 markedly impaired cell velocity and chemotactic navigation in a complement C5a gradient ( Fig. 1 (C and D) and Videos S3-S6), whereas deletion of Gnai3 only marginally impaired cell velocity. The extent of in vitro chemotaxis impairment in Gnai2 Ϫ/Ϫ macrophages was greater than expected when compared with WT versus Gnai2 Ϫ/Ϫ mouse in vivo inflammation models (12,13).

Complement C5a-induced Ca 2؉ transients and lamellipodial cell spreading are intact in Gnai2-deficient macrophages
We have previously shown that complement C5a induces robust Ca 2ϩ transients (20), which are monophasic at high agonist concentrations, and lamellipodial cell spreading in mouse macrophages (21). In the following experiments, we simultaneously imaged intracellular [Ca 2ϩ ], using the fluorescent Ca 2ϩ indicator Cal-520, and membrane dynamics, assessed by brightfield microscopy (Fig. 2). Macrophages, like neutrophils, migrate on a two-dimensional surface in an amoeboid fashion in which the dominant lamellipodial membrane protrusion steers the cell (22,23) and in the case of chemotaxis needs to be directed toward higher concentrations of chemoattractant (24), as evident, for example, in Video S1. We therefore tested whether complement C5a-induced lamellipodial membrane protrusive activity is defective in Gnai2 Ϫ/Ϫ macrophages. Surprisingly, similar to WT macrophages, complement C5a induced robust monophasic Ca 2ϩ transients and lamellipodial membrane protrusions in the absence of Gnai2 ( Fig. 2A and Videos S7 and S8). Deletion of either Gnai2 or Gnai3 did not significantly decrease the peak levels of complement C5a-induced Ca 2ϩ transients or cell spreading, indexed as the peak projected cell area normalized to that preceding stimulation (Fig. 2B). The Ca 2ϩ transient peak, but not peak cell spreading, was decreased by PTX treatment (Fig. 2B).
The temporal relation between complement C5a-induced Ca 2ϩ transients and cell spreading suggested that rapid increases in intracellular [Ca 2ϩ ] may promote branched actin polymerization, which drives lamellipodial membrane extension. To test this notion, we loaded the cytosol of macrophages with the Ca 2ϩ chelator EGTA, using its cell-permeable acetoxymethyl (AM) ester form (EGTA/AM). Intracellular loading of cells with EGTA almost completely abolished complement Ca 2ϩ -induced Ca 2ϩ transients (Fig. 3A), but cell spreading was not abrogated (Fig. 3B). This implies that the second messenger Ca 2ϩ is not important for the formation of complement C5ainduced lamellipodial membrane protrusions.

Double Gnaq/Gna11 or Gna12/Gna13 knockout macrophages have robust complement C5a-mediated chemotaxis
Next, we investigated the roles of the G␣ q /G␣ 11 and G␣ 12 / G␣ 13 families of G␣ subunits (schematically illustrated in Fig.  4A) in complement C5a-mediated chemotaxis using triple mutant mice in which one gene is deleted and the second one is conditionally knocked out in myeloid cells, which includes macrophages. The median velocity and chemotaxis index of macrophages isolated from Gnaq fl/fl /LysM-Cre/Gna11 Ϫ/Ϫ (Gnaq/Gna11 dKO (double knockout)) mice, in which Gna11 is constitutively deleted and Gnaq is conditionally deleted, did not differ from the values for WT macrophages (Fig. 4B). Similarly, macrophages isolated from Gna12 Ϫ/Ϫ /Gna13 fl/fl /LysM-Cre (Gna12/Gna13 dKO) mice, in which Gna12 is deleted and Gna13 is conditionally deleted, clearly navigated well along chemotactic complement C5a gradients, although the median velocity of Gna12/Gna13 dKO macrophages was modestly increased (Fig. 4, B and C). The polarized morphology of Gnaq/ Gna11 dKO macrophages migrating in a complement C5a gradient was unremarkable (Videos S9 and S10), whereas Gna12/Gna13 dKO macrophages exhibited modestly elongated trailing ends ( Fig. 4D and Videos S11 and S12), reminiscent of the phenotype of conditional Rhoa knockout macrophages (25), but much less extreme than Rhoa/Rhob double knockout macrophages (25). Measurements of maximal tail length (Fig. 4, D and E), assessed over a 6-h migration period, confirmed that Gna12/Gna13 dKO macrophages developed elongated trailing ends (Fig. 4E). The modest phenotypes of Gna12/Gna13 dKO macrophages suggest that complement C5a coupling to G␣ 12 and G␣ 13 contributes to the activation of the Rho subfamily of Rho GTPases and retraction of the trailing end in migrating cells. Consistent with this interpretation, we found that application of complement C5a increased the levels of active RhoA (RhoA-GTP) in WT mouse bone marrow-derived macrophages, measured by G-LISA assays (Fig. 4F).

UTP-induced Ca 2؉ transients are largely abolished in Gnaq/Gna11 double knockout macrophages
Extracellular UTP and ATP induce large Ca 2ϩ transients in mouse macrophages by binding to P2Y 2 receptors (P2Y 2 Rs), although ATP additionally induces Ca 2ϩ influx via P2X recep-tors, ATP-gated, nonselective cation channels (20). P2Y 2 Rs are thought to signal to G␣ q and G␣ 11 (26,27), functionally similar G proteins that are typically not distinguished by receptors (8). Indeed, we found that UTP-induced Ca 2ϩ signaling was almost completely abrogated in Gnaq/Gna11 dKO macrophages, whereas subsequent application of complement C5a induced robust Ca 2ϩ transients (Fig. 5, A and B). Note that the decaying Ca 2ϩ hump following large complement C5ainduced Ca 2ϩ peaks can be explained by store-operated Ca 2ϩ entry because it is absent in Ca 2ϩ -free media and can be evoked by reintroduction of Ca 2ϩ -containing media (20). The large complement C5a-induced Ca 2ϩ peak persists in Ca 2ϩ -free media and is produced by endoplasmic reticular (ER) Ca 2ϩ release (20).

Gna15 ؊/؊ macrophages have intact chemotaxis, but complement C5a-induced Ca 2؉ signaling is largely abolished
To determine whether the hematopoietic-restricted ␣-subunit G␣ 15 , a member of the G␣ q /G␣ 11 family (as illustrated in Fig. 7A), is required for macrophage migration toward complement C5a, we performed real-time chemotaxis assays. Similar to WT cells, Gna15 Ϫ/Ϫ macrophages migrated efficiently along complement C5a gradients ( Fig. 7B and Videos S13 and S14), and cells showed normal polarized morphologies in a chemotactic gradient ( Fig. 7C and Videos S13 and S14). There were no significant differences in the measured velocity and chemotactic efficiency of WT and Gna15 Ϫ/Ϫ macrophages migrating in a complement C5a gradient (Fig. 7D).
Davignon et al. (14) showed that Gna15 is not required for normal hematopoiesis but found that complement C5a-induced Ca 2ϩ transients, averaged from at least five cells, were markedly decreased in thioglycolate-elicited peritoneal macrophages from Gna15 Ϫ/Ϫ mice, whereas responses to ATP or UTP were similar to WT macrophages. In accord with the findings of Davignon et al. (14), we found that WT resident peritoneal macrophages produced robust Ca 2ϩ transients upon application of complement C5a, as well as 10 min later following the application of UTP (Fig. 8A), whereas most individual Gna15 Ϫ/Ϫ macrophages did not respond to complement C5a and those that responded gave a weak signal (Fig. 8, B and C). The peak intracellular Ca 2ϩ response to

G proteins and chemotactic signaling
UTP 10 min after complement C5a application was significantly weaker in WT macrophages compared with Gna15 Ϫ/Ϫ macrophages (Fig. 8C). This could possibly be explained as follows. The large cytosolic Ca 2ϩ signal induced by complement C5a in WT cells (which is virtually absent in Gna15 Ϫ/Ϫ cells) causes inactivation of ER Ca 2ϩ release channels (IP 3 receptors), which has been shown to occur after ϳ30 s in the sustained presence of IP 3 (28). Thus, the weaker response to the second agonist (UTP) probably reflects insufficient resensitization of these channels and/or other factors, such as insufficient time for replenishment of ER Ca 2ϩ stores and membrane PIP 2 levels.

Complement C5a-induced lamellipodial membrane protrusions are preserved in Gna15 ؊/؊ macrophages
We next tested whether impaired lamellipodial cell spreading accompanied the loss of Ca 2ϩ signaling in Gna15 Ϫ/Ϫ macrophages. Instead of using transmitted light on the spinning disk confocal microscope to assess membrane dynamics, as in earlier experiments, we used the fluorescent plasma membrane stain CellMask Orange. This fluorophore nicely stained the membrane of macrophages, but it also weakly labeled the surface of fibronectin-coated -Slide I chambers (Fig. 9A). Application of complement C5a induced rapid and robust lamellipodial membrane protrusions in WT ( Fig. 9A and Video G proteins and chemotactic signaling S15) and Gna15 Ϫ/Ϫ macrophages ( Fig. 9A and Video S16). Fig.  9A shows snapshots of CellMask Orange recordings, whereas the supporting videos (Videos S15 and S16) show an overlay of fluorescent Cal-520 and CellMask Orange signals. A sharp increase in intracellular [Ca 2ϩ ] precedes cell spreading in WT macrophages (Video S15), whereas no increase in intracellular [Ca 2ϩ ] is seen in Gna15 Ϫ/Ϫ macrophages (see the Cal-520 fluorescence intensity trace in Fig. 9A and Video S16). Notably, in Fig. 9A and in the accompanying video (Video S15), the introduction of complement C5a-containing medium to WT macrophages caused a marginal shift of the focal plane in the z axis. Quantification of the images confirmed that complement C5a induced significant increases in projected cell area in both WT and Gna15 Ϫ/Ϫ macrophages (Fig. 9B), and the effects could not be explained by effects of shear stress (Fig. 9C), which typically causes ruffling with minimal net cell spreading. There was no difference, WT versus Gna15 Ϫ/Ϫ macrophages, in the magnitude of complement C5a-induced cell spreading (Fig. 9D).

Gnb1 and Gnb2 knockout mouse models generated using CRISPR/Cas9
Impaired inflammation-mediated phagocyte recruitment in Gnai2 Ϫ/Ϫ mice (12, 13) and real-time chemotaxis assays using macrophages isolated from Gnai2 Ϫ/Ϫ mice (see Fig. 1) strongly implicate G␣ i2 as a central signal transducer in chemotactic complement C5a signaling. In addition to the inhibitory effect of G␣ i subunits on adenylyl cyclase, the G␤␥ subunits released upon activation of members of the G␣ i/o family are known to regulate various signal pathways, including PLC-␤ isoforms, phosphoinositide 3-kinases (PI3Ks), and ion channels (8). PI3Ks catalyze the conversion of PIP 2 to phosphatidylinositol 3,4,5-trisphosphate (PIP 3 ), a key phospholipid implicated in chemotactic signaling (3). G␤ subunits are encoded by five genes in mice (Gnb1-5), and RNA sequence analysis revealed that Gnb1 and Gnb2 are predominantly expressed in macrophages (Fig. 10A). Gnb1 and Gnb2 (schematically illustrated in Fig. 10B) have been implicated in chemotaxis in knockdown studies using the mouse macrophage cell line J774A.1 (29,30). Recently, the Knockout Mouse Phenotyping Program (KOMP 2 ) (31) generated Gnb1 and Gnb2 knockout mice using CRISPR/Cas9. Homozygous deletion of Gnb1 was lethal, assessed at postnatal days 13-14 ( Fig. 10C), consistent with the study by Okae and Iwakura (32) in which Gnb1 was disrupted by gene trap mutagenesis. Targeted deletion of Gnb2 was not lethal (Fig. 10D). The deletion allele, the location of primer binding sites, and an example of PCR genotyping are shown in Fig. 10E. In addition to the loss of exons 3-4, the 359-bp deletion mediated by guide RNAs and Cas9 introduces a frameshift and premature termination codon.
Gnb2 Ϫ/Ϫ macrophages exhibited robust lamellipodial membrane spreading in response to the application of complement C5a ( Fig. 11A and Video S17). Moreover, the peaks of complement C5a-induced cell spreading and Ca 2ϩ transients (examples are shown in the bottom panels of Fig. 11A), indexed as Cal-520 fluorescence, were not significantly different compared with controls (see summary box plots in Fig. 11B).
Finally, we looked at the role of G␤ 2 -containing ␤␥-subunits (schematically illustrated in Fig. 12A) in complement C5a-mediated chemotaxis. Compared with WT macrophages migrating in a complement C5a gradient, Gnb2-deficient macrophages showed reduced velocity and impaired chemotactic G proteins and chemotactic signaling navigation (Fig. 12B). Gnb2 Ϫ/Ϫ macrophages polarized in the presence of complement C5a (Fig. 12C), but lamellipodial membrane protrusions were less efficiently directed toward higher concentrations of complement C5a (Videos S18 and S19). Summary data are plotted in Fig. 12D. Both cell velocity and chemotactic efficiency were significantly decreased in Gnb2 Ϫ/Ϫ macrophages compared with WT controls.

Discussion
Intravital imaging, dating back to 1888, provides insight into the behavior of immune cells in their natural environment, but using this approach, it has been difficult to study the chemotactic response of cells to a specific molecule. This problem was solved by the introduction of the Boyden chamber (33), consisting of two reservoirs separated by a thin membrane. However, in Boyden-type transwell assays, the cells only move across a thin membrane, typically with a thickness of about 10 m, making it difficult to assess cell morphology and to distinguish chemokinesis from chemotaxis. Using knockout mouse models and real-time chemotaxis assays, which allow visualization of cell morphology and measurement of cell velocity and chemotactic efficiency, we investigated the roles of specific G protein subunits (G␣ i2 , G␣ i3 , G␣ q /G␣ 11 , G␣ 12 /G␣ 13 , G␣ 15 , G␤ 1 , and G␤ 2 ) in transducing complement C5a gradients to directed migration. The specific inhibitor of G␣ i/o subunits pertussis toxin abrogated complement C5a-mediated chemokinesis and chemotaxis, which was largely recapitulated in Gnai2 (G␣ i2 ) knockout macrophages, whereas Gnai3 (G␣ i3 ) mutants exhibited robust chemotactic efficiency but marginally reduced cell velocity. Thus, G␣ i2 is a cornerstone of complement C5a-mediated chemotaxis, schematically illustrated in Fig. 13. However, Gnai2 Ϫ/Ϫ macrophages, as well as WT macrophages treated with pertussis toxin, still induced robust Ca 2ϩ transients and lamellipodial membrane protrusions upon stimulation with complement C5a. Thus, G␣ i2 is not essential for the generation of membrane protrusions, but it is indispens-

G proteins and chemotactic signaling
able for the spatial and temporal regulation Rho GTPases and the biasing of protrusions toward higher complement C5a concentrations.
The role of Ca 2ϩ transients in chemotaxis is controversial (see Artemenko et al. (3)). We found that complement C5amediated lamellipodial membrane protrusions were not impaired by sequestration of intracellular Ca 2ϩ . Moreover, we confirmed that C5aR-G␣ 15 signaling, which activates PLC-␤, almost completely accounts for complement C5a-induced Ca 2ϩ signaling, although both lamellipodial cell spreading and chemotaxis were intact in macrophages lacking G␣ 15 . Similar to complement C5a, UTP, an endogenous P2Y 2 R ligand, induced large Ca 2ϩ transients and lamellipodial membrane protrusions. However, we found that P2Y 2 R-induced Ca 2ϩ transients strictly required G␣ q /G␣ 11 and not G␣ 15 . Notably, ATP and UTP are poor long-range chemoattractants for macrophages and neutrophils (15,34,35), possibly due to lack of activation of G␣ i2 or other G␣ i/o family members by P2Y 2 Rs and/or rapid degradation of the ligands by ectonucleotidases, such as CD39 and CD73 (36). Although we previously found that a stable (slowly hydrolyzed) analogue of ATP had no chemotactic activity for mouse macrophages (34), Collins et al. (35) observed that PLB-985 cells, a human neutrophil-like cell line, migrated transiently (for about 5 min) toward higher concentrations of ATP in a radial gradient generated by UV lightinduced uncaging of caged ATP. Thus, ATP may transiently evoke directed migration, whereas complement C5a acts as a durable chemoattractant. We have previously shown that complement C5a induces ATP release, which probably acts as a positive feedback loop because the presence of potato apyrase, which completely degraded ATP, ADP, and adenosine, inhibited complement C5a-mediated chemotaxis (16). Similarly, apyrase was shown to inhibit neutrophil chemotaxis to the chemoattractant fMLP, a formylated tripeptide (15).
C5aR-G␣ 15 signaling as well as ATP/UTP-P2Y 2 R signaling probably induce lamellipodial membrane protrusions independent of G␣ i/o family subunits via activation of PLC-␤ isoforms and depletion of PIP 2 , as illustrated in Fig. 13. Depletion of PIP 2 may cause dissociation of Rac-GAPs and Cdc42-GAPs from the membrane, along the lines described by Li et al. (37), thereby promoting increased activity of the Rac subfamily of Rho GTPases, especially Rac1 and Rac2, and Cdc42.
We found that stimulation of WT macrophages with complement C5a increases the levels of active RhoA. However, we did not test whether this effect is lost in Gna12/Gna13 dKO macrophages, and therefore further experiments are required

G proteins and chemotactic signaling
to determine whether complement C5a receptors couple to G␣ 12 /G␣ 13 . In any case, G␣ 12 /G␣ 13 signaling does not appear to play a critical role in complement C5a-mediated chemotaxis. Macrophages lacking G␣ 12 /G␣ 13 exhibited intact chemotaxis, but increased cell velocity and modestly impaired tail retraction. This observation fits in with our previous findings that Rhoa/Rhob (RhoA/RhoB) dKO macrophages exhibit the same behavior (intact complement C5a-mediated chemotaxis and increased velocity), except Rhoa/Rhob dKO cells develop much more markedly elongated trailing ends (25). Likewise, human monocytes treated with inhibitors of Rho or its major downstream effector Rho kinase exhibit both normal chemotaxis and elongated trailing ends when migrating on a 2D surface (38). On the contrary, when Rho activity is increased, as in the case of macrophages from mice lacking the RhoGAP Myo9b (21), chemotaxis is impaired, and cell velocity is decreased. The unconventional myosin Myo9b contains an atypical C1 domain (which does not bind to diacylglycerol) (39) and a RhoGAP

G proteins and chemotactic signaling
domain, but not a pleckstrin homology domain, and probably acts as an actin-binding, motorized inhibitor of Rho at the front of migrating macrophages.
Using lentiviral delivery of siRNA into J774A.1 cells (mouse macrophage cell line), Hwang et al. (30) found that deletion of both G␤ 1 and G␤ 2 eliminated G protein-mediated receptor signaling by all four G␣ families. Interestingly, in an earlier study, the authors showed that knockdown of G␤ 2 alone in J774A.1 cells completely inhibited migration toward complement C5a in transwell assays, and cells showed absent or weak Ca 2ϩ responses to complement C5a (29). The authors also found that PTX treatment prevented Ca 2ϩ responses. In contrast, we found that complement C5a-induced Ca 2ϩ signaling was intact in Gnb2 Ϫ/Ϫ macrophages and only modestly reduced in WT cells treated with PTX, whereas Gna15 Ϫ/Ϫ macrophages exhibited mostly absent or weak Ca 2ϩ responses. However, we observed moderately impaired chemotaxis by Gnb2 Ϫ/Ϫ macrophages in a complement C5a gradient. Presumably, G␤ 1 partially compensates for loss of G␤ 2 in Gnb2 Ϫ/Ϫ macrophages. Further studies using macrophages from mice with myeloid-restricted deletion of Gnb1 would be helpful to explore the relative functions of G␤ 1 and G␤ 2 in complement C5a-mediated chemotaxis.

Mice
Gnb1 knockout (Gnb1 em1(IMPC)Bay ) mice were generated by the Knockout Mouse Phenotyping Program (KOMP 2 ) (31) at the Department of Molecular and Human Genetics (Baylor College of Medicine, Houston, TX) using the CRISPR/Cas9 gene-targeting technique. Gnb2 knockout (C57BL/6NJ-Gnb2 em1(IMPC)J /Mmjax) mice were generated by KOMP 2 at the Jackson Laboratory (Bar Harbor, ME) using CRISPR/Cas9. The knockout allele was produced by injecting Cas9 mRNA and four single guide RNAs (TCCCATTC-TTCAGTGCCCCA, ATGGGCAGAATGATAGTACA, TC-CCATTCTTCAGTGCCCCA, and ATGATGGGCAGTGC-AAGAGA) into C57BL/6NJ-derived fertilized eggs, which were subsequently transferred to pseudopregnant females. The guide RNAs, in combination with the (RNA-guided) DNA endonuclease Cas9, were designed to delete 359 bp in the targeted gene (Gnb2), leading to a deletion including exons 3 and 4 and a frameshift after amino acid 19. The frameshift produces a stop codon after coding a further 19 residues. Pups were genotyped using the primer pair (sequences 5Ј 3 3Ј) Gnb2-F (CCCAAATCCTCTCAGGA-TGA) and Gnb2-R (TGCTTCCCTTTGACCTGAGT), which resulted in a 639-bp PCR product for the WT allele and a 280-bp product for the mutant allele. The generation of Gnai2 (40) and Gnai3 (41) knockout mice, which were backcrossed onto a C57BL/6J genetic background (13), have been described previously. Gnai2 Ϫ/Ϫ and Gnai3 Ϫ/Ϫ mice Below is the intracellular Ca 2ϩ signaling corresponding to the Gna15 Ϫ/Ϫ macrophage labeled M⌽1. Complement C5a and UTP were added as indicated. Intracellular [Ca 2ϩ ] is indexed as relative Cal-520 fluorescence intensity (F/F 0 ), where the measured fluorescence intensity (F) is divided by the resting fluorescence intensity (F 0 ) after subtracting the background fluorescence intensity at each time point. B, box plots of projected cell area before and after application of complement C5a to WT or Gna15 Ϫ/Ϫ macrophages. *, p Ͻ 0.05; Mann-Whitney U test. C, box plots of relative peak projected cell area after application of ligand-free medium (Sham) and complement C5a-containing medium to WT and Gna15 Ϫ/Ϫ macrophages. *, p Ͻ 0.05; Mann-Whitney U test (n ϭ 34 for WT and n ϭ 29 for Gna15 Ϫ/Ϫ (three independent experiments)). D, box plots of the changes in cell area (prestimulation cell area subtracted from the peak poststimulation cell area) after stimulating WT and Gna15 Ϫ/Ϫ macrophages with complement C5a (n.s., not significant; Mann-Whitney U test).
All animal experiments were performed in accordance with the German Animal Welfare Act (Tierschutzgesetz), approved by the local ethics committee of the University of Münster, and conformed to the Guide for the Care and Use of Laboratory Animals of the United States National Institutes of Health.

Isolation of resident peritoneal macrophages
The method, including a video, for isolation of mouse resident peritoneal macrophages has recently been described (45). Mice were killed by an overdose of isoflurane in air, and the peritoneal cavity was lavaged via a 24-gauge plastic catheter (B. Braun, Melsungen, Germany) using 2 ϫ 4.5 ml of ice-cold Hanks' balanced salt solution without Ca 2ϩ or Mg 2ϩ (14175-046, Gibco). After centrifugation (300 ϫ g for 6.5 min), cells were resuspended in bicarbonate-free RPMI 1640 medium containing 20 mM Hepes (R7388-500ML, Sigma-Aldrich, Steinheim, Germany) and modified by supplementation with 10% heat-inactivated fetal calf serum (FCS), 100 units/ml penicillin, and 100 g/ml streptomycin (pH 7.4). The cells were seeded into fibronectin-coated -Slide I chambers or -Slide Chemotaxis chambers (Ibidi, Martinsried, Germany) and placed in a humidified incubator (37°C). After 2 h, -Slide I chambers were filled with 1 ml of RPMI 1640 medium containing sodium bicarbonate (R8758-500ML, Sigma-Aldrich), which was modified by the addition of 10% heat-inactivated FCS and antibiotics and incubated overnight at 37°C with 5% CO 2 . Experiments were performed on the stage of an inverted microscope (LSM 510, Zeiss) equipped with a temperaturecontrolled incubator (incubator XL S, Zeiss) using the same bicarbonate-free RPMI 1640 medium as above.

Time-lapse 2D chemotaxis assays
The 2D chemotaxis assay has been described previously (16,25), including a video demonstration (47). In brief, cells obtained by peritoneal lavage of a single 3-4-month-old mouse were resuspended in 150 -200 l of medium and diluted to a concentration of 10 ϫ 10 6 cells/ml, and 10 l of the suspension was seeded into the narrow (1 mm ϫ 2 mm ϫ 70 m) channel of an uncoated (IbiTreat) -Slide Chemotaxis chamber (Ibidi). The narrow channel (observation area) connects two 40-l reservoirs. Notably, the bottom of the chamber is formed by a thin gas-permeable plastic sheet, which has the same thickness and optical properties as a standard No. 1.5 glass coverslip. After 2 h, the chemotaxis chamber was filled with bicarbonate-free RPMI 1640 medium containing 20 mM Hepes, which was sup- Figure 10. Homozygous deletion of Gnb1 is lethal, whereas Gnb2-deficient mice are viable. A, expression levels of G␤ and G␥ subunits in mouse resident peritoneal F4/80 ϩ cells (macrophages). RNA-Seq analysis was performed using RNA isolated from resident peritoneal F4/80 ϩ cells purified by cell sorting (n ϭ 3 mice). Error bars, S.E. B, schematic diagram highlighting the genes encoding the G␤-subunits G␤ 1 (Gnb1) and G␤ 2 (Gnb2). C, homozygous deletion of Gnb1 is lethal. D, homozygous Gnb2 mutant mice are viable. E, schematic diagram of the WT allele and the Gnb2 knockout allele, generated by CRISPR/Cas9. The forward (Gnb2-F) and reverse (Gnb2-R) primers are indicated by horizontal red arrowheads, where the arrow tip marks the start of binding. Below is an image of an agarose gel depicting the genotyping.

G proteins and chemotactic signaling
plemented with 10% heat-inactivated FCS, 100 ng/ml lipopolysaccharide, and antibiotics. Next, 15 l of medium containing chemoattractant (complement C5a) and 0.003% Patent Blue V (blue dye) was drawn into one of the reservoirs. The final concentration of complement C5a was 20 nM. Complement C5a (2150-C5-025, R&D Systems, Abingdon, UK) was reconstituted in PBS containing 0.1% BSA, which had been filtered using a 0.2-m cellulose acetate membrane (723-2520, Thermo Fisher Scientific). Aliquots were stored at Ϫ80°C. An aluminum heating block maintained at 37°C was used to keep media and the -Slide Chemotaxis chamber warm during the filling procedure. The observation area was imaged by phase-contrast microscopy via a ϫ10/0.3 objective lens. The blue dye served as a visual indicator of gradient formation. Images were captured every 2 min for 14 h, and cell migration tracks between 6 and 12 h were analyzed with ImageJ (National Institutes of Health) using a cell-tracking plugin and the chemotaxis and migration tool from Ibidi. Twenty-five cells that remained in the field of view were tracked for each chemotaxis assay.

Time-lapse imaging of intracellular [Ca 2؉ ] and lamellipodial membrane dynamics
After overnight incubation of resident peritoneal cells, seeded in fibronectin-coated -Slide I chambers, in modified RPMI 1640 medium containing sodium bicarbonate, the medium was switched to bicarbonate-free RPMI 1640 medium containing 20 mM Hepes, 1 mM N-(2-mercaptopropionyl)glycine (MPG; an antioxidant), 10% heat-inactivated FCS, and antibiotics. The -Slide I chambers were placed on an anodized aluminum rack in a dry incubator maintained at 37°C. Similar to -Slide Chemotaxis chambers, the bottom of -Slide I

G proteins and chemotactic signaling
chambers is formed by a gas-permeable plastic coverslip with the same thickness and optical properties as a No. 1.5 coverslip. To image intracellular Ca 2ϩ , cells were loaded with the nextgeneration fluorescent Ca 2ϩ indicator Cal-520, using its AM ester form Cal-520/AM (21130, AAT Bioquest, Sunnyvale, CA). A 10 mM stock of Cal-520/AM in anhydrous DMSO (276855-100ML, Sigma-Aldrich) containing 20% (w/v) Pluronic F-127 (Sigma-Aldrich) was prepared by adding 4.53 l of a 20% solution of Pluronic F-127 in DMSO to a 50-g aliquot of Cal-520/AM. To prepare a 10 M loading solution, the 10 mM Cal-520/AM stock solution was diluted 1:1000 in RPMI 1640-Hepes medium containing 1 mM MPG, 10% FCS, antibiotics, and additionally 1 mM probenecid, added from a 200 mM stock solution of the water-soluble sodium salt (P36400, Thermo Fisher Scientific), a competitive blocker of organic anion transporters that reduces the efflux of intracellular anionic fluorescent Ca 2ϩ indicators (46). Cells were loaded with Cal-520 by 20-min incubation with 10 M Cal-520/AM at 37°C and subsequently washed with RPMI 1640-Hepes medium containing MPG and probenecid.
Lamellipodial membrane dynamics were imaged by either brightfield or fluorescence microscopy. In the latter case, the plasma membrane was stained with CellMask Orange (C10045, Thermo Fisher Scientific), diluted 1:1000 (from a 5 mg/ml stock solution in DMSO) in modified RPMI 1640-Hepes medium or in the above Cal-520 loading solution. Cells were incubated for 10 -20 min at 37°C.
Cells were imaged via the Apochromat TIRF ϫ60/1.49 (oil-immersion) objective lens of a Nikon Eclipse Ti inverted microscope, which was connected to a spinning-disk confocal system (Ultra-VIEW Vox 3D live cell imaging system). The system included a Yokogawa (Japan) CSU-X1 spinning-disk scanner, a Hamamatsu (Japan) C9100-50 EM-CCD camera (1000 ϫ 1000 pixels), and Volocity software. Cal-520 was excited with a 488-nm laser, whereas CellMask Orange was excited via a 561-nm laser. Images were captured (at a fixed focus level) using 2 ϫ 2 binning (giving 500 ϫ 500 pixels/image) at a rate of 2 s/time point. Alternatively, cells were alternately imaged by brightfield and fluorescence microscopy at a slower rate of 5 s/time point. Focus drift during long recordings was prevented using the Nikon Perfect Focus System, which maintains the position of the coverslip in the z axis by reflecting near-IR light (870 nm) and detecting it via a CCD (charge-coupled device) line sensor. The temperature was maintained at 37°C using an Okolab all-in-one stage incubator (Okolab, Ottaviano, Italy) and an objective lens heating mantle (Scientific Instruments, Farmingdale, NY).

G proteins and chemotactic signaling
The ligands complement C5a and UTP were applied to one of the two reservoirs of a -Slide I chamber to induce rapid medium exchange in the 100-l channel seeded with macrophages. Highly purified (Ͼ99%) UTP (R0471, Thermo Fisher Scientific) was added from a 100 mM aqueous solution (pH 7.3-7.5). After application of a ligand or ligand-free medium (in a sham experiment), a second ligand could be applied after aspirating excess medium from the reservoirs flanking the 100-l channel.

Flow cytometry and cell sorting
Freshly isolated mouse resident peritoneal cells were stained with Alexa Fluor 488 -conjugated anti-F4/80 antibodies and resuspended in autoMACS running buffer (Miltenyi Biotec, Bergisch Gladbach, Germany), which contains PBS, 2 mM EDTA, and 0.5% BSA. Purification of F4/80 ϩ cells (mouse macrophages) was performed using a BD FACSAria II (or FAC-SAria III) cell sorter (BD Biosciences). Recovered cells were centrifuged at 300 ϫ g for 5 min, and the supernatant was removed before proceeding to RNA isolation.

RNA isolation and RNA-Seq analysis
Total RNA of purified mouse resident peritoneal F4/80 ϩ cells (macrophages) was isolated by solid-phase extraction using a Direct-zol TM RNA MicroPrep kit (Zymo Research, Freiburg, Germany) according to the manufacturer's instruc- Figure 13. Tabular summary and schematic diagram of G protein subunits involved in transducing complement C5a gradients into directed migration. A, tabular summary of results. B, schematic summary. C5aRs couple (i) directly to at least two heterotrimeric G proteins formed by G␣ 15 and G␣ i2 subunits and possibly also G␣ 12 /G␣ 13 and G␣i 3 (not shown) subunits and their respective G␤␥ subunits and (ii) indirectly to G␣ q /G␣ 11 -containing heterotrimeric G proteins via autocrine ATP signaling, which stimulates P2Y 2 Rs. The G␣ i2 subunit is indispensable for chemotaxis and associates with G␤ 2 -containing, or possibly also G␤ 1 -containing, G␤␥ subunits. G␣ i2 /G␤ 2 ␥ x heterotrimeric G proteins, where x is unknown, dissociate into active (GTP-bound) G␣ i2 subunits and G␤ 2 ␥ x dimers following receptor activation by complement C5a. The G␤ 2 ␥ x (or possibly G␤ 1 ␥ x ) dimers activate PI3Ks, which catalyze the conversion of PIP 2 to PIP 3 . PIP 3 is known to recruit pleckstrin homology domain-containing Rac-and Cdc42-GEFs to the membrane. Activation of G␣ 15 -containing heterotrimeric G proteins directly by complement C5a, as well as indirect activation of G␣ q /G␣ 11 -containing heterotrimeric G proteins via autocrine ATP and UTP signaling, increases the activity of PLC-␤ isoforms, which catalyze the hydrolysis of PIP 2 to inositol IP 3 and diacylglycerol. IP 3 induces Ca 2ϩ release from the endoplasmic reticulum, but this Ca 2ϩ signal is largely redundant for lamellipodial membrane protrusions and chemotaxis. However, we speculate that depletion of PIP 2 by PLC-␤ isoforms and PI3Ks contributes to the formation of lamellipodial membrane protrusions by promoting the dissociation of Rac-and Cdc42-GTPaseactivating proteins (GAPs). We speculate that activation of G␣ 12 /G␣ 13 by complement C5a-C5aR signaling, which remains to be confirmed, increases the activity of the monomeric (small) G proteins RhoA and RhoB via RhoGEFs. Activated (GTP-bound) RhoA and RhoB promote actomyosin-dependent retraction of the trailing end of migrating cells, whereas the RhoGAP Myo9b is thought to inhibit RhoA and RhoB at the front of cells. Extracellular ATP and UTP stimulate P2Y 2 Rs. ATP, but not UTP, additionally activates P2X receptors (not shown), ligand-gated cation channels. ATP and UTP are rapidly degraded by surface ectonucleotidases, such as CD39, to form ligands for other purinergic receptors (not shown).

G proteins and chemotactic signaling
tions. Briefly, we lysed the pelleted cells with 300 l of TRIzol (Thermo Fisher Scientific), added 300 l of analytical (100%) ethanol, and transferred the mixture into a Zymo-Spin TM IC column inserted into a collection tube. Following wash steps, as well as DNase treatment for 15 min, involving several centrifugations at 12,000 ϫ g, purified total RNA was captured in a silica column. Using RNase-free water, concentrated RNA was eluted from the silica column and collected in a DNase-/RNase-free safe-lock tube. Isolated RNA samples were tested for integrity using RNA ScreenTape (Agilent Technologies, Santa Clara, CA) and stored at Ϫ80°C.
Next-generation sequencing was performed using a NextSeq 500 Sequencing System (Illumina, San Diego, CA). Samples were prepared using a TruSeq ϫRNA sample preparation kit (Illumina), which involved the following steps: purification and fragmentation of mRNA, first-and second-strand cDNA synthesis, end repair, adenylation of 3Ј ends, ligation with adaptors, and PCR amplification. RNA-Seq data were analyzed using the Tuxedo suite, an open access set of applications for ultrafast alignment of short reads to the genome, recognition of splice junctions, and differential expression analysis.

G-LISA assay for active RhoA
Levels of active RhoA (RhoA-GTP) were measured using a colorimetric RhoA G-LISA activation assay kit (catalog no. BK124, Cytoskeleton, Denver, CO), according to the elaborate instruction manual. The microplate was shaken at a speed of 400 rpm using Fisherbrand TM microplate shakers (15504070, Fisher), one of which was placed in a cool room (4°C). Mouse bone marrow-derived macrophages were used instead of peritoneal macrophages to provide sufficient lysates for the assays. Femurs from each mouse were fractured with a scalpel blade at the mid-diaphysis level, and bone marrow cells were flushed out of each fragment using 5 ml of RPMI 1640-Hepes medium containing 10% heat-inactivated FCS and antibiotics, injected via a bent (L-shaped) 23-gauge syringe needle. The isolated cells were pipetted up and down to reduce clumping and transferred into a second 50-ml polypropylene tube via a 70-m cell strainer. Next, the tube was centrifuged at 300 ϫ g for 10 min, and the pellet was resuspended in red blood cell lysing buffer (R7757, Sigma) for 5 min, followed by centrifugation at 300 ϫ g for 5 min. After aspiration of the supernatant, the pellet was resuspended in 30 ml of RPMI 1640-Hepes medium containing 10% heat-inactivated FCS and antibiotics, which was supplemented with nonessential amino acids and 15 ng/ml recombinant mouse macrophage colony-stimulating factor (416-ML-010, R&D Systems). The suspension was pipetted into a Teflon bag (PermaLife PL30, OriGen, Biomed Europe), such that each Teflon bag contained cells from a single mouse, and incubated at 37°C with 5% CO 2 for 6 days. The Teflon bag was placed in ice for 30 min, and the cells, resuspended by shaking and massaging, were poured into a 50-ml tube. The tube was centrifuged at 300 ϫ g for 10 min, and the pellet was resuspended in RPMI 1640-Hepes medium containing 10% heat-inactivated FCS and antibiotics and seeded into 35-mm round culture dishes. After a 1.5-h incubation at 37°C in a CO 2 -free incubator, the medium was switched to RPMI 1640 medium containing sodium bicarbonate, 10% heat-inactivated FCS, and antibi-otics and incubated overnight at 37°C with 5% CO 2 . Stimulation of cells and the harvesting of lysates were performed after switching back to Hepes-containing medium and allowing at least 1 h of equilibration time.

Statistics
The nonparametric Mann-Whitney U test or Kruskal-Wallis one-way analysis of variance on ranks was used to test for statistical differences at the 0.05 level of significance. Post hoc multiple comparisons were made using the Bonferroni correction. Statistical analyses were performed using Origin 2020 (OriginLab) or earlier versions, and data are presented as box plots or mean Ϯ S.E., unless stated otherwise.