Control of local Rho GTPase crosstalk by Abr

Summary

Background

The RhoGTPases—Rho, Rac and Cdc42—regulate the dynamics of F-actin (filamentous actin) and myosin-2 with considerable subcellular precision. Consistent with this ability, active Rho and Cdc42 occupy mutually exclusive zones during single cell wound repair and asymmetric cytokinesis, suggesting the existence of mechanisms for local crosstalk, but how local Rho GTPase crosstalk is controlled is unknown.

Results

Using a candidate screen approach for Rho GTPase activators (Guanine nucleotide exchange factors; GEFs) and Rho GTPase inactivators (GTPase activating proteins; GAPs), we find that Abr, a protein with both GEF and GAP activity, regulates Rho and Cdc42 during single cell wound repair. Abr is targeted to the Rho activity zone via active Rho. Within the Rho zone Abr promotes local Rho activation via its GEF domain and controls local crosstalk via its GAP domain, which limits Cdc42 activity within the Rho zone. Depletion of Abr attenuates Rho activity and wound repair.

Conclusions

Abr is the first identified Rho GTPase regulator of single cell wound healing. Its novel mode of targeting by interaction with active Rho allows Abr to rapidly amplify local increases in Rho activity using its GEF domain while its ability to inactivate Cdc42 using its GAP domain results in sharp segregation of the Rho and Cdc42 zones. Similar mechanisms of local Rho GTPase activation and segregation enforcement may be employed in other processes that exhibit local Rho GTPase crosstalk.

Introduction

Dynamic processes powered by actin filaments (F-actin) and myosin-2 such as cell migration and cell division entail a high degree of local regulation, ensuring that assembly and disassembly of F-actin and myosin-2 filaments are spatially coordinated with each other and with other events such as adhesion and de-adhesion. At least some of this local coordination must be exerted at the level of the Rho GTPases—Cdc42, Rac, and Rho—which regulate F-actin and myosin-2 with considerable specificity [1]. For example, Cdc42 and Rac can promote assembly of highly dynamic, branched F-actin networks, while Rho can stimulate myosin-2 and formation of unbranched F-actin networks [2]. The Rho GTPases, in turn, are activated by guanine nucleotide exchange factors (GEFs), inactivated by GTPase activating proteins (GAPs), and held inactive in the cytoplasm by GDP-dissociation inhibitors (GDIs) [3, 4, 5].

It is now clear that at least some subcellular specificity in regulation of F-actin and myosin-2 is achieved by localized zones of Rho GTPase activity. For example, single cell wound healing in Xenopus oocytes entails formation of concentric zones of Rho and Cdc42 activity around the wound, with the Cdc42 zone circumscribing the Rho zone [6]. The Rho zone directs local activation of myosin-2 while the Cdc42 zone directs local accumulation of dynamic F-actin [6, 7]. Similarly, during polar body emission, a form of asymmetric cell division, a disc-like zone of Cdc42 forms at the PM (plasma membrane) immediately over the meiotic spindle and is surrounded by a ring-like zone of Rho [8]; as with wound healing, each zone makes distinct contributions to the cytokinetic event [9]. The idea of localized, complementary Rho, Rac and Cdc42 zones can be extended to events at the leading edge of crawling cells, where spatially distinct, closely spaced bursts in Rho, Cdc42 and Rac are linked to different events required for cell protrusion and retraction [10].

The existence and segregation of complementary Rho GTPase activity zones is suggestive of subcellular GTPase crosstalk. That the Rho GTPases engage in crosstalk is amply demonstrated by studies employing biochemical approaches or analyses of the morphology of fixed cells. Crosstalk may work through the Rac/Cdc42 effector, PAK, which can negatively regulate Rho GEFs [11, 12, 13]. Other mechanisms include signaling via reactive oxygen species [14], phosphorylation and competitive binding of RhoGDI [15, 16], and binding of GEFs to actomyosin [17].

To date, however, a mechanism for local (i.e. subcellular) crosstalk has not been directly explored. Here, we identify Abr, a dual GEF-GAP, as a Rho GTPase regulator in single cell wound repair, and a mediator of local crosstalk between the Rho and Cdc42 activity zones.

Results

A candidate screen identifies Abr as a potential regulator of Rho and Cdc42 during the single cell wound response

To identify potential regulators of Rho and Cdc42 during the single cell wound response, a candidate screen approach was employed. A small pool of Xenopus GEFs and GAPs (Figure S1A; Movie S1) was selected, cloned, tagged with eGFP and assessed for localization to wounds, effects on Rho and Cdc42 zones following overexpression, and effects on Rho and Cdc42 zones following dominant negative expression (Figure 1A). Active Rho was detected with eGFP-rGBD (eGFP fused to the GTPase binding domain of the Rho effector rhotekin), while active Cdc42 was detected with mRFP-wGBD (mRFP fused to GTPase-binding domain of the Cdc42 effector N-WASP) [6].

Figure 1. A screen of GEFs and GAPs during wound healing identifies the GEF-GAP Abr, which colocalizes with active Rho.

(A) Schematic of candidate regulator screen.

(B) Oocytes expressing mRFP-wGBD (red) and endogenous Abr (green) as detected by antibody staining (top panel) but not in oocytes stained with secondary antibody alone (bottom panel).

(C) Oocyes expressing Abr-3XeGFP and mRFP-wGBD mRNA.

(D) Oocyes expressing 3XmCherry-Abr and eGFP-rGBD mRNA.

(E) Oocytes expressing eGFP-farnesyl and 3XmCherry-Abr mRNA. Z-view is shown before and after wounding.

Abr, a protein with both GEF and GAP activity for the Rho GTPases [18], emerged as the only candidate to satisfy all three criteria (Figure 1; S1A). Moreover, its ability to disrupt the Rho and Cdc42 zones was evident at concentrations of mRNA 20-40 times lower than those effective for the other candidates (see Figure 3 and Figure S1). Abr is expressed in oocytes (Figure S1B;D) and both endogenous Abr as well as 1X-, 3XeGFP- or 3XmCherry-Abr localized to wounds (Figure 1B-D; Movie S2). To further test the observed localization pattern and the Abr antibody, cells expressing untagged, exogenous Abr were immunostained after wounding; this manipulation resulted in a significant increase in the Abr signal at the wound (Figure S1C).

Figure 3. Abr inhibits the Cdc42 zone and broadens the Rho zone.

(A) Cells expressing mRFP-wGBD and eGFP-rGBD alone (top panel) or 100 μg/ml Abr mRNA (bottom panel).

(B) The Cdc42 zone is shown with increasing concentrations of Abr mRNA.

(C) The intensity of the Cdc42 zone was quantified at 60 s post-wounding with increasing concentrations of Abr mRNA (n=9; **p<0.01, ***p<0.005; Tukey’s multiple comparison test).

(D) Rho zone width was quantified at 90 s post-wounding with increasing concentrations of Abr mRNA (n=6; *p<0.05, **p<0.01;Tukey’s multiple comparison test).

Live-cell imaging revealed that Abr was rapidly recruited to the wound edge (Figure 1C-D; see also below). Comparison of the distribution of Abr to Cdc42 or the Rho zone in live or fixed samples revealed that Abr consistently concentrated within the Rho zone, with a slight enrichment at the trailing edge (Figure 1B-D). Z-view comparisons to a plasma membrane marker, farnesylated eGFP revealed that wounding resulted in Abr recruitment from the cytoplasm to the plasma membrane at regions flanking wounds (Figure 1E). In addition to Abr itself, Xenopus has a closely related homologue. Abr homologue also localizes to the Rho zone (Figure S1E) and otherwise behaves identically to Abr (see Figure S3A). Thus, except where mentioned, the rest of the results are concerned with Abr itself.

Abr localizes to wounds via GEF and GAP domain-dependent interaction with active Rho

Abr consists of a DH-PH GEF domain, a C2 domain, and a GAP domain (Figure 2A). To assess which domains of Abr are important for its localization, the localization of WT Abr-3XeGFP, AbrΔDH-3XeGFP and AbrΔGAP-3XeGFP were compared (Figure 2A-B). While the WT Abr protein localizes at the wound edge, deletion of either the DH domain or the GAP domain abrogates localization (Figure 2B). Recruitment failure could not be explained by reduced expression of the mutants in that eGFP signal was clearly evident for both AbrΔDH-3XeGFP and AbrΔGAP-3XeGFP (Figure 2B) and immunoblotting demonstrated that AbrΔDH-eGFP and AbrΔGAP-eGFP are expressed at least as well as WTAbr-eGFP (Figure S2A). The C2 domain could potentially contribute to localization, however, it lacks the residues required for calcium binding, and an AbrC2-eGFP fusion failed to localize to wound, while the C2 domain of Protein Kinase C-β displays robust localization to wounds (not shown but see [19]).

Figure 2. Abr requires its substrate-binding domains and active Rho for localization.

(A) Schematic showing Abr domain structure and 3XeGFP-tagged constructs used for localization studies.

(B) Oocytes expressing Abr-3XeGFP, AbrΔDH-3XeGFP, or AbrΔGAP-3XeGFP. The localization of each at 60 s post-wounding is shown.

(C) Kymographs from cells expressing 3XmCherry-Abr and eGFP-rGBD. W=wound.

(D) A vertical line was drawn in (C) through the region where Abr and active Rho are recruited starting after wounding through 46 s post-wounding. The intensity of each was plotted over time.

(E) Cells expressing mRFP-wGBD, Abr-3XeGFP and C3 exoenzyme where indicated.

(F) PM Z-views of cells expressing eGFP-farnesyl and 3XmCherry-Abr alone or with either CA Cdc42, CA Rho, or GEF-H1 as shown. Cells were incubated in latrunculin A (lat) where indicated.

Because the GEF and GAP GTPase interacting domains are required for localization and because Abr colocalizes with the active Rho zone, we hypothesized that Abr is recruited to wounds by binding active Rho. Consistent with this hypothesis, kymograph analysis revealed a tight spatial and temporal correlation between active Rho recruitment and Abr recruitment (Figure 2C-D) but not Abr and active Cdc42 (Figure S2B-C). This localization pattern cannot be explained by transport of Abr via contraction-powered cortical flow, as Abr localizes normally in cells pretreated with the lectin WGA (Figure S2D), which blocks cortical flow [20].

To directly test the role of active Rho in Abr localization, cells were microinjected with C3 exotransferase, which inactivates Rho. C3 completely eliminated Abr recruitment to wounds (Figure 2E), showing directly that active Rho is necessary for Abr recruitment to wounds. To determine whether active Rho is sufficient for Abr recruitment, Abr localization was assessed after manipulation of GTPase activity in unwounded cells. In control cells, Abr is predominantly cytoplasmic (Figures 1E; 2F) as it is following expression of constitutively active (CA) Cdc42. In contrast, expression of CA Rho causes recruitment of Abr to the PM (Figure 2F). Likewise, expression of the RhoGEF GEF-H1 elevates PM Rho activity (Figure S1A) and causes recruitment of Abr to the PM (Figure 2F). This recruitment occurs in the presence of the F-actin inhibitor latrunculin (Figure 2F), indicating that Abr is not recruited via interaction with actomyosin. These results, taken with those showing that recruitment is dependent on the GEF and GAP domains, indicate that Abr is recruited to wounds via interaction with active Rho.

Abr positively regulates Rho while negatively regulating Cdc42

To test the functional role of Abr during wound healing, the effects of Abr expression on the Rho and Cdc42 activity zones were monitored. Microinjection of oocytes with Abr significantly increased the breadth of the Rho zone at the expense of the Cdc42 zone (Figure 3A; Movie S3). To determine whether these effects are concentration dependent, Abr was microinjected at 5, 10, 50, 100 and 500 μg/ml (needle concentration). As the concentration of Abr increases, the intensity of the Cdc42 zone decreases (Figure 3B-C), while the Rho zone intensity is not significantly affected (Figure S3A). However, the zone of active Rho widens with respect to controls as the concentration of Abr is increased (Figure 3D). Similar results were observed with the Abr homologue (Figure S3B).

GAP-dead Abr prevents segregation of the Rho and Cdc42 activity zones

The above results suggested that the GAP domain of Abr might locally suppress Cdc42 activity, in keeping with the fact that the Abr GAP domain is active toward Rac and Cdc42 but not Rho [18]. To directly test this idea, we made an Abr mutant, Abr RN/AA (Figure 4A), which renders it GAP-deficient without compromising GTPase binding [21]. Like WT Abr, this mutant localizes to the Rho zone (Figure 4B). However, when expressed at the same concentration of WT Abr previously found to sharply reduce the intensity of the Cdc42 zone, Abr RN/AA had no significant effect on the Cdc42 zone intensity (Figure 4C; Movie S4), consistent with the prediction that Abr suppresses local Cdc42 activity via its GAP domain. Perhaps more remarkably, Abr RN/AA not only significantly broadened the Rho zone, it also broadened the Cdc42 zone (Figure 4D), causing the two zones to bleed into each other (Figure 4E; Movie S4). This result indicates that the GAP activity of Abr normally participates in local Rho and Cdc42 crosstalk.

Figure 4. GAP-dead Abr prevents Cdc42 inhibition and promotes zone overlap.

(A) Domain structure of GAP-dead mutant, Abr RN/AA, and Abr RN/AA-3XeGFP.

(B) Abr RN/AA 3X-eGFP was injected along with mRFP-wGBD; 60 s post-wounding.

(C) Cdc42 zone intensity was quantified in controls and Abr RN/AA-expressing cells (n=12; p=0.6941;Unpaired T-test).

(D) Cdc42 and Rho zone width were quantified in control and Abr RN/AA-expressing cells (for Rho: n=12; ***p< 0.0001; for Cdc42: n=12, *p< 0.05; Unpaired T-test).

(E) Cells injected with mRFP-wGBD and eGFP-rGBD alone (top panel) and with 500 μg/ml Abr RN/AA (bottom panel); 90 s post-wounding.

GEF-dead Abr blocks Rho activity and Cdc42 activity

Since the Abr GAP domain is not required for increased Rho zone width and since Abr locally activates Rho but not Cdc42 (Figure 3A) it follows that Abr might locally stimulate Rho activity through its GEF domain. To test this hypothesis, S104A and R244A mutations were made in the DH domain of Abr (Figure 5A).

Figure 5. GEF-dead Abr inhibits the Rho zone but not the Cdc42 zone.

(A) Domain structure of GEF-dead mutant, Abr SR/AA, and Abr SR/AA-3XeGFP.

(B) Oocytes expressing Abr SR/AA-3XeGFP and mRFP-wGBD; 60 s post-wounding (scale bar=20 μm).

(C) Rho zone intensity was quantified in control cells and those expressing either WT Abr or Abr SR/AA (n=10; *p<0.05; Tukey’s multiple comparison test).

(D) Cdc42 zone intensity was quantified in control cells and those expressing either WT Abr or Abr SR/AA (n=10; ***p<0.0001; Tukey’s multiple comparison test).

(E) Cells expressing eGFP-rGBD and mRFP-wGBD alone (top panel) or with 500 μg/ml Abr SR/AA (bottom panel).

These mutations correspond to T506A and R634A mutations in Dbl, which result in dramatic reduction of GDP/GTP exchange [22]. Abr SR/AA localized to wounds (Figure 5B), and dramatically reduced Rho activity relative to uninjected and WT Abr-expressing controls (Figure 5C&E; Movie S5). Cdc42 activity was also inhibited relative to controls, consistent with the role of the GAP domain in promoting Cdc42 inactivation independent of GEF activity (Figure 5D-E). Together, our findings indicate that Abr negatively regulates Cdc42 through its GAP activity and positively regulates Rho through its GEF activity.

Abr localizes to and regulates healing of embryo wounds

The above results indicate that Abr locally activates Rho in the Rho zone while enforcing zone segregation via Cdc42 inhibition. If this model is correct, Abr depletion would be predicted to suppress local Rho activity while broadening the Cdc42 zone at the expense of the Rho zone. Efforts to deplete Abr in oocytes were unsuccessful, so we turned to Xenopus embryos which have a robust healing response [19] and permit morpholino (MO)-mediated depletion of target proteins following fertilization [23, 24]. Consistent with results obtained in oocytes, Abr localizes to the Rho zone around single cell embryo wounds as well as to cell-cell junctions near wounds (Figure 6A-B) which correspond local hotspots of Rho activity [19]. Further, WT Abr expression in embryos expands the Rho zone at the expense of the Cdc42 zone (Figure 6C), an effect accompanied by formation of highly developed stress folds around the wound, presumably a consequence of excess, Rho-mediated contractility.

Figure 6. Abr localizes with active Rho in embryos and Abr depletion inhibits Rho and perturbs wound healing.

(A) Embryos expressing eGFP-rGBD and 3XmCherry-Abr (scale bar=20 μm).

(B) Embryos expressing Abr-3XeGFP and mCherry-UtrCH to label F-actin were wounded near a cell border. Top panel=merge, cell border labeled by arrowheads. Bottom panel=Abr, Abr accumulation at the cell border indicated by an arrow (scale bar=20 μm).

(C) Embryos expressing 3XmCherry-wGBD and eGFP-rGBD alone (top panel), or with WT Abr (bottom panel). Stress folds indicated by arrows (scale bar=20 μm).

(D) Embryos were either uninjected, injected with 2 mM control MO, or 1 mM Abr MO and 1 mM Abr homologue MO (AbrMO 1+2) and homogenized 18 h post-fertilization. Abr and tubulin were detected by immunoblotting.

(E) Embryos expressing eGFP-rGBD, mCherry-wGBD and Wee1 with either control MO or Abr MO 1+2 and imaged 18 h post-fertilization. Top panel shows control MO phenotype. Second and third panels represent wounds from Abr MO 1+2 -injected embryos.

(F) Rho and Cdc42 zone intensity from cells in (E) was quantified after 48 s (Rho: n=25, ***p<0.0001; Cdc42; n=17, p=0.2907; Unpaired T-test).

(G) Rho zone width was quantified from cells in (E) (n=17; ***p<0.001; Unpaired T-test).

To determine the functional role of Abr in single cell wound healing, we used a MO approach to knockdown endogenous Abr in Xenopus embryos. MOs designed to target the 5′ end of both Abr mRNA and Abr homologue mRNA were microinjected into embryos (see Methods). Western blotting reveals a decrease in Abr and Abr homologue protein levels in MO-injected embryos relative to controls at 18 h post-fertilization (Figure 6D). Abr depletion did not prevent cell division but did consistently inhibit gastrulation (data not shown). Consistent with the oocyte results, Abr depletion resulted in a significant reduction in Rho, but not Cdc42 activity around single cell wounds (Figure 6E-F). Further, Abr depletion also promoted the narrowing of the Rho zone (Figure 6G), the precise opposite effect of that produced by Abr overexpression. Finally, while embryos injected with control MO healed properly, Abr MO-injected embryos displayed several other phenotypes including stalled or delayed healing (Figure 6E; S4). In some cases wounded cells were completely unable to mount a healing response and ultimately lysed (data not shown). These results support the notion that Abr positively regulates Rho activity and reveal that Abr is required for proper wound healing.

Discussion

The results of this study show that Abr, a dual RhoGTPase GEF-GAP, is a critical regulator of Rho and Cdc42 during the single cell wound response and provide what is, to the best of our knowledge, the first characterization of a subcellular RhoGTPase crosstalk mechanism in vivo. Specifically, the results indicate that Abr is recruited to the incipient Rho zone by interaction with active Rho, where it locally amplifies Rho activity via its GEF domain. Simultaneously, Abr locally suppresses Cdc42 via its GAP activity (Figure 7). We do not know whether Abr binds directly to active Rho, although the results are consistent with this possibility. This novel mechanism provides a simple explanation not only for zone segregation but also for how the initially broad and dilute distribution of active Rho [6] is rapidly converted into a tight, intense zone inside the Cdc42 zone: Assuming that concentration of active Rho is slightly higher near the wound edge than at regions distal to the wound, the initial asymmetry in active Rho and Abr would be rapidly amplified via positive feedback.

Figure 7. Model for Abr at wounds.

Abr is recruited to the Rho zone, where it interacts specifically with active Rho through its DH and GAP domains. Once recruited, Abr positively regulates Rho activity via the GEF domain. Simultaneously, the Abr speeds Cdc42 inactivation in the Rho zone through its GAP activity, maintaining zone segregation.

Additional features of Rho GTPase regulation are also revealed by what does not happen upon WT Abr expression: the intensity of the Rho zone does not increase even at high levels of overexpression, nor does the zone spread beyond the area normally occupied by the Cdc42 zone. This result indicates that the positive feedback is somehow antagonized even in the near total absence of active Cdc42. While these limits could be imposed by the availability of Rho itself or some other component of the system such as the GDIs, we favor the idea that localized Rho activation is normally limited by simultaneous inactivation via Rho GAP activity, a hypothesis previously described as the “GTPase flux” model [23, 25]. If this model is correct, positive feedback between active Rho and Abr may be restrained by one or more Rho GAPs, at least one of which would be concentrated at the trailing edge of the expanded Rho zone that results from WT Abr overexpression.

Are Abr and its homologue the only GEFs activated during wound healing? After all, both the GEF-dead Abr and Abr depletion severely curtail Rho activation. However, we suspect that there is at least one additional, non-Abr Rho GEF involved that would account for the small amount of Rho activity observed even after a high level of GEF-dead Abr expression. A Rho GEF that acts immediately after wounding would serve the role of “priming” the Rho zone by providing the initial pool of active Rho needed for Abr recruitment. Further, for a process as fundamental as cell wound repair, it makes sense that redundant mechanisms would be employed.

To what extent can the current results be extended to other systems? A general role for Abr in cellular wound repair is consistent with the fact that Abr is particularly abundant in brain and muscle tissues [26], as these contain very large cells which are especially prone to mechanical damage. More generally, a role as a regulator of local crosstalk could explain the participation of Abr in cell migration and spreading. That is, studies from mouse macrophages lacking both Abr and Bcr (the only other known dual GEF-GAP for Rho GTPases) indicate that loss of Abr and Bcr results in excessive cell spreading [21]. Similarly, in human pluripotent stem cells, Abr depletion suppresses cell rounding after dissociation and stimulates spreading [27]. Because FRET-based analyses of local Rho, Rac and Cdc42 activity during cell adhesion and spreading indicates that local hotspots of Rho activity are spatially complementary to local hotspots of Cdc42 and Rac activity [10], we suggest that the deficits observed in cells lacking Abr or Abr and Bcr reflects loss of local, Abr-enforced Rho-Cdc42/Rac crosstalk.

Finally, the possibility that the role played by Abr revealed here could potentially be played by complexes of Rho GTPase GEFs and GAPs in other contexts should be considered. For example, cytokinesis is dependent on interaction of a Rho GEF (Ect2) with a Rho GAP (MgcRacGAP)[28]. If this GEF-GAP complex can localize via interaction with active Rho, both positive feedback via Ect2 as well as cross talk via the GAP domain of MgcRacGAP could potentially occur.

Experimental Procedures

Plasmids

eGFP-rGBD, mRFP-wGBD, mCherry-UtrCH, and pCS2+3XeGFP were made as described [6, 7, 23]. 3XmCherry-pCS2+ was developed by inserting three consecutive mCherry sequences between BamH1 and BspE1 in the pCS2+ vector. 3XmCherry-wGBD was constructed by inserting the wGBD fragment into 3XmCherry-pCS2+ with Xho1 and Xba1.

The plasma membrane marker eGFP-farnesyl pCS2+ was obtained from M.V. Danilchik (Oregon Health and Science University, Portland, Oregon). The CA Rho and CA Cdc42 constructs were obtained from Tim Gomez (University of Wisconsin). X. laevis GEF-H1 (BC079763) was obtained from ATCC and subcloned into pCS2+ with Xho1 and SnaB1. XGEF (AY095313) was obtained from Laura Hake (Boston College) and cloned into eGFP-pCS2+ with BspE1 and Xba1. X. laevis Chimaerin (BC046676) was amplified from cDNA and inserted into eGFP-pCS2+ with BspE1 and Xba1 and into pCS2+-eGFP with BamH1 and Xba1. X. laevis Ect2 was obtained from Toru Miki (NIH) and subcloned into pCS2+ with BamH1 and Xba1. pCS2+MgcRacGAP R384A (DN MgcRacGAP) was made as previously stated [23]. X. laevis Wee1A (BC081031) was obtained from Open Biosystems and, via the Gateway system (Invitrogen), recombined with pDONR-221, followed by recombination with pCS2+-DEST (recombination sites between Xho1 and Xba1).

The X. laevis Abr gene (BC042307) and Abr homologue (BC080423) were obtained from Open Biosystems. Endogenous X. laevis Abr was obtained by amplifying the coding region of Abr from Xenopus oocytes cDNA. The PCR product was then subcloned into pCS2+, pCS2+-3XeGFP, and pCS2+-eGFP with Cla1 and Xba1. 3XmCherry-Abr was made by inserting Abr into 3XmCherry-pCS2+ with Xho1 and Xba1. AbrΔDH-3XeGFP and AbrΔDH-eGFP were made by amplifying Abr from amino acids 286-882 and inserting it between Cla1 and Xba1 restriction sites of the respective plasmids. AbrΔGAP-3XeGFP and AbrΔGAP-eGFP were made similarly by amplifying Abr from amino acids 1-627. Abr R686A and N798A mutations were made by separate Quickchange (Stratagene) reactions with Abr-pCS2+ and Abr-3XeGFP plasmids, where R666 was changed by mutating AGA to GCG. The N778A mutation was made by changing AAT to GCT. Abr S104A and R244A were created by changing AGT to GCT and CGT to GCT, respectively.

The Abr homologue 5′ mRNA sequence (IMAGE 6641556) was obtained by a Blast search of BC080423. The Abr homologue coding region was amplified by PCR and the product was inserted into 3XmCherry-pCS2+ with Cla1 and Xba1 or pCS2+ with Xho1 and Xba1.

mRNA preparation and oocyte injection

All mRNA was transcribed in vitro using the mMessage mMachine SP6 kit (Ambion). Oocytes were obtained as described previously [29]. A 40 nl volume of mRNA was injected 24 hours prior to imaging. Abr-3XeGFP, 3XmCherry-Abr, the Abr-3XeGFP mutants and mCherry-UtrCH were injected at 50 μg/ml needle concentration. CA Cdc42, CA Rho and GEF-H1 were injected at 1 mg/ml each. Abr RN/AA 3XeGFP and Abr SR/AA 3XeGFP were injected at 100 μg/ml and untagged Abr RN/AA and Abr SR/AA were injected at 500 μg/ml. Screen candidate mRNAs (XGEF, Chimaerin, GEF-H1, Ect2, DN-MgcRacGAP) were injected between 1-2 mg/ml. For Western blotting, the eGFP-tagged Abr mutants were injected at 1 mg/ml each. Oocytes were incubated with 10 μM latrunculin A for 30 min to 1 h. C3 exoenzyme was injected to a final concentration of 0.08 μg/μl 20-40 min before imaging.

Embryo injection and morpholinos

Embryos were fertilized in vitro and injected with a 5 nl volume of mRNA at the 2-cell stage. eGFP-rGBD, Abr-3XeGFP, 3xmCherry-Abr and mCherry-UtrCH were injected at 500 μg/ml. 3XmCherry-wGBD was injected at 750 μg/ml. Embryos were imaged after 5-9 h.

For morpholino (MO) experiments, an Abr MO with a sequence of TGTCTTGGTGGCTGACGGGTTCCAT targeting the first 25 nucleotides of Abr coding sequence, and an Abr homologue MO GAACTCCTCCGGGCCCACATGTCA, targeting 4 nucleotides of the 5′UTR and 21 bases of coding sequence were ordered from GeneTools. The morpholinos were mixed and injected at the 2-cell stage at a needle concentration of 1 mM each. Standard control MO (GeneTools) was injected at 2 mM. Embryos were injected with eGFP-rGBD (75 μg/ml), mCherry-wGBD (30 μg/ml) and Wee1 (50 μg/ml) mRNA at the 4-cell stage and wounded and imaged 18 h post-fertilization. Wee1 injection was necessary to yield larger cells at 18 h post-fertilization, allowing for wounding of single cells.

Immunofluorescence and Western blotting

Oocytes were injected with mRFP-wGBD, and after 24 h laser wounded and immediately placed in fix buffer (10 mM EGTA, 100 mM KCl, 3 mM MgCl₂, 10 mM Hepes, 150 mM sucrose at pH 7.6 with 4% PFA, 0.1% gluteraldehyde, 0.1% Triton X-100) and incubated O/N at room temperature. Cells were washed with 1X PBS, quenched with 100 mM NaBH₄ for 4 h, washed 2X with 1X PBS, bisected and incubated in TBSN-BSA (5 mg/ml BSA and 0.1% NP-40 in 1X TBS) O/N at 4°C. α-Abr (BD Biosciences) was then added at 1:200 and incubated O/N at 4°C. Cells were washed with TBSN-BSA 4X for 1 h and O/N at 4°C. Oregon green goat α-mouse (Promega) was added at 1:200, and the incubation steps were repeated.

For mutant expression analysis, 10 oocytes were washed 3X with 1X PBS and homogenized by pipetting in homogenization buffer (250 mM sucrose, 10 mM Hepes, 1 mM EGTA, 2 mM MgCl_2, 0.1% Triton X-100, 10 μM E-64, 4 mM pefabloc, 60 μg/ml chymostatin, 5 μg/ml leupeptin, 1 μg/ml pepstatin, and 3.75 μg/ml aprotinin). Homogenates were fractionated at 12,000 rpm at 4°C for 2 min, and the cytoplasmic layer was extracted and spun again. Laemmli sample buffer was added to the cleared cytoplasmic fraction and lysates were loaded on a 10% SDS-polyacrylamide gel. The gel was transferred onto nitrocellulose membrane, blocked and incubated with JL-8 α-GFP antibody (BD Biosciences) at 1:2000 O/N at 4°C. Membranes were incubated with α-mouse IgG HRP at 1:5000 (Promega), and signal was detected with Pierce ECL reagent (Thermo Scientific).

For homogenization of embryos, 25 uninjected, control MO-injected, and Abr MO-injected embryos were washed in buffer and lysed as above with additional protease inhibitors (50 μM calpeptin, 50 μM ALLN) at 18 h post-fertilization. Lysates were loaded on an 8% polyacrylamide gel and transferred to nitrocellulose. α-Abr (BD Biosciences) was added at 1:500 O/N at 4°C and α-mouse IgG HRP at 1:5000 (Promega). SuperSignal West Femto kit (Thermo Scientific) was used to detect Abr. Anti-α tubulin (clone DM1A-Sigma) was used at 1:10,000 and the secondary was used as above. Tubulin was detected with Pierce ECL reagent (Thermo Scientific).

Microscopy and data analysis

4D movies were produced as described [6]. Cells were wounded with a Micropoint pulse nitrogen-pumped dye laser (Laser Science, Inc.). Movies were analyzed using Volocity 3.7 software. Kymographs were produced from single optical plane movies using ImageJ1.41 and a 5 pixel wide box over the wound. Brightest point projections were created by Z-projection of maximum intensity using ImageJ1.41. Intensity was quantified with ImageJ1.41 by encircling the entire zone to obtain the mean gray value and subtracting the mean gray value of a similar sized box from a region away from the wound. Zone width was quantified using ImageJ1.41 by drawing 8 symmetrically spaced radial lines spanning the zone and averaging their length. Data analysis and graphing was conducted with Prism 5 for Mac OSX.