The Arf GAP Asap promotes Arf1 function at the Golgi for cleavage furrow biosynthesis in Drosophila

Drosophila embryo cleavage requires the conserved Arf GAP Asap. Asap seems to recycle Arf1 to the Golgi from post-Golgi membranes for optimal Golgi output and cleavage furrow biosynthesis.

Abstract

Biosynthetic traffic from the Golgi drives plasma membrane growth. For Drosophila embryo cleavage, this growth is rapid but regulated for cycles of furrow ingression and regression. The highly conserved small G protein Arf1 organizes Golgi trafficking. Arf1 is activated by guanine nucleotide exchange factors, but essential roles for Arf1 GTPase-activating proteins (GAPs) are less clear. We report that the conserved Arf GAP Asap is required for cleavage furrow ingression in the early embryo. Because Asap can affect multiple subcellular processes, we used genetic approaches to dissect its primary effect. Our data argue against cytoskeletal or endocytic involvement and reveal a common role for Asap and Arf1 in Golgi organization. Although Asap lacked Golgi enrichment, it was necessary and sufficient for Arf1 accumulation at the Golgi, and a conserved Arf1-Asap binding site was required for Golgi organization and output. Of note, Asap relocalized to the nuclear region at metaphase, a shift that coincided with subtle Golgi reorganization preceding cleavage furrow regression. We conclude that Asap is essential for Arf1 to function at the Golgi for cleavage furrow biosynthesis. Asap may recycle Arf1 to the Golgi from post-Golgi membranes, providing optimal Golgi output for specific stages of the cell cycle.

INTRODUCTION

The highly conserved small G protein Arf1 plays a ubiquitous role in regulating the flow of materials through the Golgi apparatus. Guanine nucleotide exchange factors (GEFs) produce Arf1-GTP, which induces two forms of vesicle budding from the Golgi. Arf1-GTP recruits the COPI coat for retrograde trafficking through the Golgi and from the cis-Golgi to the endoplasmic reticulum (ER), and its interaction with the clathrin adaptor AP-1 complex organizes budding from the trans-Golgi network (TGN) to other membrane systems and the plasma membrane (PM; Lorente-Rodriguez and Barlowe, 2011; Guo et al., 2014; Papanikou and Glick, 2014). GTPase-activating proteins (GAPs) convert Arf1-GTP into Arf1-GDP and thus reverse the direct effect of GEFs (Gillingham and Munro, 2007; Donaldson and Jackson, 2011). However, GAPs are not simply off switches. Arf GAPs are typically multidomain proteins and couple their GAP activity with other molecular interactions to coordinate vesicle maturation during biosynthesis (Gillingham and Munro, 2007; Spang et al., 2010; Donaldson and Jackson, 2011; East and Kahn, 2011).

Although biochemical and subcellular analyses have identified specific mechanisms of action for Arf1 GAPs at the Golgi, loss-of-function studies indicate that these mechanisms are redundant for cell viability. This redundancy was first discovered in single- and double-mutant analyses of the yeast Arf GAPs Gsc1 and Glo3, which act at the cis-Golgi (Poon et al., 1999), and was then found among three Arf GAPs in mammalian cell culture; ARFGAP1 (homologue of Gsc1) and ARFGAP2 and ARFGAP3 (homologues of Glo3; Frigerio et al., 2007). Redundancy may also explain the viability of Drosophila mutants for Gap69C (homologue of Gsc1/ARFGAP1; Frolov and Alatortsev, 2001) and is evident for homologues of Glo3 in Arabidopsis (Min et al., 2013). These Arf1 GAPs may thus play redundant, constitutive roles for Golgi function. Here we report an alternate paradigm in which a distinct Arf GAP, Asap (ArfGAP with SH3 domain, ankyrin repeat, and PH domain), is essential for general Golgi activity.

Arf1 is the preferred substrate of ASAP-1 (Asap in Drosophila; Furman et al., 2002). In mammalian cell culture, ASAP-1 localizes at the PM and promotes cytoskeletal remodeling and focal adhesion activity for cell spreading and protrusion (Randazzo et al., 2000; Furman et al., 2002; Liu et al., 2005; Bharti et al., 2007). In the developing Drosophila eye, Asap localizes to the PM and affects cell arrangement (Johnson et al., 2011). Asap is also found at the PM of the Drosophila embryonic ectoderm and contributes to epithelial development (Shao et al., 2010). In frog photoreceptor cells, ASAP-1 has a distinct role at the TGN, where it acts with Arf4 and Rab small G proteins for the targeting of rhodopsin to cilia (Mazelova et al., 2009; Wang et al., 2012). Overall these studies suggest that ASAP-1/Asap influences both the PM and the Golgi, although the extent of its Golgi activities is unknown. Similarly, Arf1 promotes endocytosis at the PM in addition to its Golgi functions (Donaldson and Jackson, 2011).

Membrane trafficking and cytoskeletal activities are central to cleavage of the early Drosophila embryo. The early embryo is a syncytium. The first nuclear divisions occur deep within the embryo, and then most nuclei migrate to the embryo periphery just beneath the PM (Foe and Alberts, 1983; Sullivan and Theurkauf, 1995). As part of this migration, a centrosomal microtubule array individualizes an endomembrane system for each nucleus (Frescas et al., 2006). In addition, overlap between neighboring microtubule networks dictates where furrows ingress from the embryo surface (Sisson et al., 1999). For each synchronous nuclear division cycle, these furrows are temporary; they form during interphase, reach a maximum depth at metaphase to anchor and separate mitotic spindles, and then regress (Foe and Alberts, 1983; Sullivan and Theurkauf, 1995; Sisson et al., 1999; Figure 1A). This cycle repeats several times to produce ∼6000 nuclei, after which full cellularization forms the blastoderm (Mazumdar and Mazumdar, 2002; Lecuit, 2004; Lee and Harris, 2014). For full cellularization, membrane biosynthesis from the Golgi has a well-documented role in fuelling the ∼25-fold PM growth required for the process (Lecuit and Wieschaus, 2000; Sisson et al., 2000; Frescas et al., 2006). Similarly, the repeated furrow ingressions of the preceding nuclear divisions require Arf1 (Lee et al., 2015a) and exocytic machinery (Holly et al., 2015; Mavor et al., 2016). This outward trafficking is coupled with actin-based cytoskeletal networks and PM recycling routes to support furrow ingression (Lee and Harris, 2014). How furrows regress with each division is less clear and requires greater understanding of ingression mechanisms and how they could be reversed.

FIGURE 1:

Asap is necessary for early-embryo cleavage furrows. (A) Schematic of PM furrow behavior (green) during the peripheral nuclear divisions of the syncytial Drosophila embryo. Two nuclei (purple) are shown to divide into four nuclei in a small cross-section of the embryo periphery. Note the furrow ingression from interphase to metaphase, the maximum furrow depth at metaphase, and the furrow regression from metaphase to the following interphase. (B) 3D reconstructions of ∼10–12 peripheral cell compartments in syncytial Drosophila embryos at metaphase, when furrows are normally deepest. Furrows marked with Amph (green) and metaphase chromosomes detected with pH3 (magenta). A control mCherry^shRNA embryo with intact furrows is compared with asap^shRNA1 embryos with partial or severe furrow loss. (C) xy-sections from the embryo surface to below the nuclei for the embryos shown in B. (D) Quantifications of the furrow structure of control and Asap RNAi embryos (two distinct asap shRNAs were analyzed). N, number of metaphase embryos counted for each genotype (compiled from two or more experiments, each with similar results). Constructs were expressed maternally with matαTubGAL4::VP16.

We discovered an essential role for Asap in forming early-embryo cleavage furrows. Our experiments revealed that Asap is necessary and sufficient for Arf1 accumulation at the Golgi and that both proteins, as well as their conserved binding site, are required for proper Golgi structure and general output to the PM. Remarkably, Asap translocates to the nuclear region before each cycle of furrow regression, a shift accompanied by altered Golgi structure. Thus an Asap-Arf1-Golgi pathway appears to fuel cleavage furrow ingression and may be regulated by Asap sequestration during the cell cycle.

RESULTS

Asap is required for cleavage furrow ingression

To perturb Asap in the early embryo, we pursued a maternal RNA interference (RNAi) strategy. We designed short hairpin RNA (shRNA) constructs that target distinct sequences, expressed them in the female germline, and examined embryos produced from these females. To assess cleavage furrows, we stained embryos with amphiphysin (Amph), a PM marker enriched at furrow tips, and phospho–histone H3 (pH3), a marker of mitotic chromosomes that allows selection of embryos at metaphase, when furrow ingression is greatest (Foe and Alberts, 1983; Sullivan and Theurkauf, 1995; Sisson et al., 1999; Figure 1A). Control mCherry^shRNA embryos displayed full furrow ingression at metaphase (Figure 1, B–D), whereas two distinct asap shRNA constructs induced furrow loss, asap^shRNA1 more severely than asap^shRNA2 (Figure 1, B–D). Consistent with other reports (e.g., Holly et al., 2015), embryos with furrow loss often displayed abnormally arranged and fused nuclei (see later discussion of Figure 5), suggesting a failure to separate mitotic spindles in the absence of furrows. However, at the earliest peripheral nuclear division cycle, Asap RNAi embryos were indistinguishable from controls (Supplemental Figure S1), indicating that early mitoses deep within the embryo were unaffected by Asap depletion. The stronger asap shRNA, asap^shRNA1, also induced partial female infertility (Table 1). Coexpression of an Asap–green fluorescent protein (GFP) construct with control and asap shRNA constructs confirmed that Asap protein levels were substantially reduced by asap^shRNA1 and to a lesser extent by asap^shRNA2 (unpublished data), consistent with their relative effects on furrows.

FIGURE 5:

Asap affects early-embryo Golgi organization. Side views and single cross-sections at or below the nuclei of control, Asap, Arf1, and AP-2α RNAi embryos expressing histone-mRFP for visualization of nuclei and stained for GM130 and LqfR for markers of the cis- and trans-Golgi, respectively (embryos at interphase). cis- and trans-Golgi compartments abnormally aggregate in Asap and Arf1 RNAi embryos (brackets in side views and inserts of single sections). AP-2α RNAi embryos display nuclear disorganization similar to that of Asap and Arf1 RNAi embryos but not the Golgi disorganization. Quantification of Golgi disruptions is presented below. N, embryo numbers (compiled from two or more experiments, each with similar results). Constructs expressed as in Figure 1.

TABLE 1:

Perturbations of Asap or Arf1 affecting female fertility.

UAS constructs	Expression level	Female infertility
asapshRNA1	Higha	Full
	Mediumb	Strong
asapshRNA2	High	Undetectable
arf1shRNA2 + GFP	Medium	Strong
arf1shRNA2 + arf1[N52A]	Medium	Undetectable
arf1shRNA2 + arf1[I46D]	Medium	Strong
arf1[N52A]	High	Undetectable
	Medium	Undetectable
arf1[I46D]	High	Strong
	Medium	Partial

^aExpression with Maternal Triple Driver.

^bExpression with matαTubGAL4::VP16.

To examine what step of furrow formation is disrupted with asap RNAi, we performed live imaging of a PM marker in individual embryos over several cleavage cycles. At interphase, asap RNAi embryos could form PM caps similar to controls, but these caps became abnormally scattered over several division cycles (Figure 2; compare apical surface sections). After interphase, asap RNAi embryos typically failed to ingress furrows from the PM caps (Figure 2, side views; four of six embryos formed no furrows, and two of six embryos showed partial ingression), in contrast to control embryos displaying furrow ingression (Figure 2, arrowheads; four of four embryos). Overall these results show that Asap is required specifically for cleavage furrow ingression in the early embryo.

FIGURE 2:

Asap is necessary for cleavage furrow ingression. Live imaging of the plasma membrane marker GAP43-mCherry during the four cycles of syncytial division preceding cellularization. Apical surface views show the formation of PM caps above nuclei at interphase of each cycle. Side views show the subsequent ingression of furrows from these caps. Caps form with either control or Asap RNAi, although caps become more scattered with Asap RNAi. Furrows ingress with control RNAi (arrowheads; four of four embryos) but not with Asap RNAi (four of six embryos showed no ingression, and two of six embryos showed partial ingression).

Asap exists at multiple subcellular sites and relocalizes during the cell cycle

To understand how Asap promotes cleavage furrows, we imaged GFP-tagged Asap constructs (Shao et al., 2010), as well as antibody staining for endogenous Asap (Johnson et al., 2011). Previously these probes revealed Asap localization to cell–cell junctions in later Drosophila tissues (Shao et al., 2010; Johnson et al., 2011). However, Asap antibody staining was at background levels in the early embryo (unpublished data), suggesting a relatively low expression level, a disperse protein distribution, or a masked antibody epitope. Thus we analyzed overexpressed, GFP-tagged Asap constructs in the early embryo and assessed whether their localization was relevant to endogenous Asap by comparing their distributions to that of antibody staining in slightly later embryos.

Live imaging of GFP-Asap (N-terminally tagged) revealed a dramatic relocalization of the protein during the cleavage cycles of the early embryo. During interphase, GFP-Asap localized in a punctate distribution over the PM (Figure 3A, blue arrows) and to early PM furrows (Figure 3A, white arrows), with a cytoplasmic pool apparent by nuclear exclusion (Figure 3A, asterisks). As the nuclear envelope broke down just before metaphase, GFP-Asap lost its PM association and nuclear exclusion and quickly became enriched in the nuclear region relative to the surrounding cytoplasm (Figure 3A, yellow arrows). This nuclear localization was gradually lost from metaphase to telophase, and the protein relocalized to the PM by the next interphase (unpublished data). Live imaging of Asap-GFP (C-terminally tagged) revealed similar patterns of PM localization and nuclear exclusion at interphase followed by nuclear enrichment at prometaphase (Supplemental Figure S2)

FIGURE 3:

Asap localizes to multiple subcellular sites and relocates during the cell cycle. (A) A GFP-Asap–expressing embryo imaged live over a syncytial cell cycle. A side view and a projection of the full peripheral nuclear layer are shown as a mitotic wave spreads across the embryo. At interphase (top left), GFP-Asap is enriched at punctate structures resembling apical microvilli (blue arrows) and at early PM furrows (white arrows) and has a cytoplasmic pool detectable by exclusion from nuclei (yellow asterisks). At the onset of mitosis (bottom right), GFP-Asap is depleted from the PM and becomes enriched at nuclear regions surrounding condensed chromosomes (yellow arrows). After mitosis, GFP-Asap returns to the PM (not shown). (B) Single cross-section at the apical PM surface at interphase. GFP-Asap puncta closely colocalize with F-actin–rich microvilli (arrows). (C) A side view and single section at the nuclear level at interphase shows GFP-Asap most prominently at the PM, more weakly in the cytosol, with slight enrichment around the nuclear periphery, and exclusion from the nucleus. Staining with the Golgi marker AP-1 reveals no enrichment of GFP-Asap at Golgi puncta. (D) GFP-Asap becomes enriched within the nuclear region at prometaphase. RFP-KDEL marks the periphery of the nuclear region. Inset, separation of GFP-Asap from condensed chromosomes (marked with histone RFP). (A–D) Constructs expressed as in Figure 1. (E) Ectoderm of a postsyncytial, stage 8–9 histone-GFP embryo stained for endogenous Asap and the Golgi marker AP-1. The oblique section from the embryo surface (right) inward (left) reveals endogenous Asap strongly at the PM of each cell and weakly in the cytoplasm, with no enrichment at AP-1–positive Golgi elements. (F) Mitotic domain within the ectoderm of a stage 8–9 histone-mRFP embryo stained for endogenous Asap and the PM marker Dlg. At metaphase, endogenous Asap localizes to the nuclear region (yellow arrows) surrounding condensed mitotic chromosomes (histone-mRFP). Endogenous Asap is excluded from the nucleus in interphase cells (yellow asterisks). Bars, 5 μm.

Next we compared GFP-Asap with several membrane and cytoskeletal markers. At interphase, GFP-Asap puncta colocalized with actin-rich PM protrusions over the apical surface and to early furrows of the cell compartments (Figure 3, B, arrows, and C, side view). Intracellularly, GFP-Asap was distributed as weaker puncta throughout the cytoplasm, with patchy enrichment around the nuclear periphery but with no specific overlap with AP-1–positive Golgi elements (Figure 3C) and with nuclear exclusion (Figure 3C, asterisk). At prometaphase, GFP-Asap became enriched within the nuclear region but was excluded from condensed chromosomes (Figure 3D, arrow, inset), a pattern resembling that reported for markers of the nucleoplasm (Trieselmann and Wilde, 2002) and nuclear matrix (Walker et al., 2000).

To test whether endogenous Asap displayed similar localization patterns, we analyzed antibody staining in the ectoderm of embryos 2–3 h after completion of the syncytial divisions. In the ectoderm of stage 8–9 embryos, Asap antibody staining revealed strong PM localization and weaker cytoplasmic localization, with no enrichment at AP-1–positive Golgi elements (Figure 3E). At interphase, endogenous Asap displayed nuclear exclusion (Figure 3F, asterisk). At metaphase, the staining was enriched within the nuclear region surrounding condensed chromosomes (Figure 3F, arrow). Together these probes suggest that Asap could function at the PM or within the cytoplasm, and no specific associations with the Golgi were detected. In addition, all three probes revealed a relocalization of Asap to the nuclear region at prometaphase.

Furrow ingression seems largely independent of Asap effects on the PM

To distinguish how Asap promotes cleavage furrow ingression, we investigated cellular processes affected by Asap in other systems known to promote furrow ingression in the early Drosophila embryo and implicated by our Asap localization analyses. First, we considered the actin cortex. Actin networks stabilize the furrows (Lee and Harris, 2014), and a PM Arf-GEF, Steppke, has been shown to antagonize these networks to restrain furrow growth (Lee and Harris, 2013). For example, Steppke overexpression weakens furrow actin networks and induces furrow loss through its GEF activity (Lee and Harris, 2013). Thus we hypothesized that asap RNAi might eliminate furrows because Steppke was no longer counteracted. To test this hypothesis, we examined whether the effects of asap RNAi could be suppressed by steppke RNAi through coexpression of shRNA constructs. steppke shRNA expression induced abnormal formation of basal membranes (Figure 4A) due to elevated actomyosin activity around the base of the furrows (Lee and Harris, 2013). Double RNAi revealed that the effects of Asap depletion were dominant over the effects of Steppke depletion (Figure 4B), suggesting that the asap RNAi defects were due to a distinct, and perhaps upstream, dysfunction in furrow formation.

FIGURE 4:

Furrow ingression seems independent of possible roles for Asap at the PM. (A) A steppke^shRNA metaphase embryo with furrow tips expanded to form abnormal basal membranes (shown as a 3D reconstruction of ∼10–12 peripheral cell compartments and as xy-sections from the embryo surface to below the nuclei; compare to the shRNA control embryo in Figure 1, B and C). (B) Quantifications of Asap-Steppke and control double RNAi experiments. Note that the furrow loss with Asap RNAi dominates over the furrow tip expansion that otherwise occurs with Step RNAi. N, number of metaphase embryos counted for each genotype (compiled from two or more experiments, each with similar results). (C) Top, cross-sections of the equatorial somatic region reveal intact furrows with control shRNA and loss of furrows with Asap, Arf1, and AP-2α RNAi (embryos at interphase). Middle, projections of the posterior pole show the formation of PGCs with control, Asap, and Arf1 RNAi but not with AP-2α RNAi. Amphiphysin marks the PM. The PGCs are round and protrude from the posterior pole (arrows). Bottom, quantifications of the presence of PGCs. N, embryo numbers (compiled from two or more experiments, each with similar results). Constructs expressed as in Figure 1.

To further examine the possible effect of Asap at the PM, we evaluated the first cells formed in the syncytial embryo, the primordial germ cells (PGCs). The furrow formation required for these posterior cells to divide individually from the syncytial soma is also inhibited by abnormally high Steppke GEF activity (Lee et al., 2015b) or loss of actomyosin activity (Cinalli and Lehmann, 2013; Lee et al., 2015b). Strikingly, asap RNAi embryos could form PGCs, despite strong furrow disruption in the later soma (Figure 4C; arrows show PGCs). Thus Asap seems to promote furrows via a mechanism that is distinct from Steppke and actomyosin regulation and is specifically required in the soma during later peripheral divisions.

Next we considered PM endocytosis, since transcytosis from the apical PM is required for furrow ingression (Pelissier et al., 2003). Consistent with an endocytosis/transcytosis route, disruption of the PM clathrin adaptor complex AP-2 eliminates furrows (Lee and Harris, 2013), as does RNAi of the Golgi and endocytic regulator Arf1 (Lee et al., 2015a). Because the Asap-dependent mechanism was specifically required for somatic but not PGC cleavage, we evaluated PGC formation with RNAi of AP-2α or arf1. RNAi of either gene resulted in strong furrow disruption in the soma of the syncytial embryo, but only AP-2α RNAi disrupted PGC formation (Figure 4C). These results suggest that AP-2–dependent endocytosis is critical for the cleavage of both somatic cells and PGCs, whereas Asap and Arf1 are more critical for later cleavage in the soma and thus may contribute mostly to a distinct cellular process.

Asap is required primarily for Golgi organization

Another major contributor to furrow ingression is biosynthetic trafficking from the ER and Golgi (Lecuit and Wieschaus, 2000; Sisson et al., 2000; Frescas et al., 2006; Holly et al., 2015; Mavor et al., 2016). To probe whether Asap and Arf1 participate in the biosynthetic pathway, we analyzed the Golgi morphology of asap RNAi and arf1 RNAi embryos. In the Drosophila early embryo, the Golgi is composed of dispersed tubulovesicular elements that form stacks but lack the ribbon-like organization of other cell types (Ripoche et al., 1994; Rabouille et al., 1999). To probe for Golgi structure, we stained for a cis-Golgi marker, GM130 (Lorente-Rodriguez and Barlowe, 2011), and a trans-Golgi marker, Liquid facets-related (LqfR; Leventis et al., 2011). In control RNAi embryos, each marker labeled evenly distributed puncta, and GM130 puncta were often adjacent to LqfR puncta (Figure 5). In contrast, both asap RNAi and arf1 RNAi produced abnormal aggregation of Golgi elements, with LqfR-positive structures at the core of the aggregates and GM130 puncta around their periphery (Figure 5). To assess whether this abnormal Golgi morphology occurred in response to general furrow and nuclear disruption, we analyzed the soma of AP-2α RNAi embryos, but these embryos displayed seemingly normal Golgi organization despite severe nuclear disorganization (Figure 5). Moreover, embryos expressing the weaker asap shRNA (asap^shRNA2) displayed moderate Golgi disorganization with normal nuclear organization (Supplemental Figure S3). Thus the main effect of asap RNAi seems to be on the Golgi.

Asap promotes Arf1 localization at the Golgi

Because both Asap and Arf1 are required for a similar aspect of Golgi organization, we hypothesized that Asap acts through Arf1 to control the Golgi. To test this idea, we probed for Arf1 localization using an Arf1-GFP construct that localizes to Golgi puncta (Shao et al., 2010; Lee et al., 2015a; Figure 6A) with dependence on the Golgi Arf GEF Gartenzwerg (Armbruster and Luschnig, 2012). In control RNAi embryos at interphase, Arf1-GFP localized strongly to intracellular puncta and also displayed a general cytosolic distribution with nuclear exclusion (Figure 6B). In contrast, strong asap RNAi (asap^shRNA1) depleted all pools of Arf1-GFP, and weaker asap RNAi (asap^shRNA2) did so to a lesser degree (Figure 6, B and D). To assess the specificity of the asap RNAi effect, we also probed for Arf4-GFP, which has a distribution similar to that of Arf1-GFP (Lee et al., 2015a; Figure 6C) and found that its levels at the Golgi were increased by asap RNAi versus control (Figure 6, C and E). Thus Asap has a specific and essential role in maintaining Arf1 at the Golgi, although it also affects total Arf1 levels through an unknown effect.

FIGURE 6:

Asap is necessary and sufficient for Arf1-GFP accumulation at the Golgi. (A) A syncytial embryo at interphase expressing Arf1-GFP and stained with the Golgi marker LqfR. Note the colocalization of the Arf1-GFP puncta with the Golgi marker. (B) Side views and subnuclear xy-sections of syncytial embryos expressing Control^shRNA, asap^shRNA1, and asap^shRNA2 in combination with Arf1-GFP (stained for Amph for staging; embryos at interphase). Images collected and adjusted with the same settings. Note the loss of Arf1-GFP levels at Golgi puncta with Asap RNAi and accompanying general loss of Arf1-GFP. (C) Side views and subnuclear xy-sections of syncytial embryos expressing Control^shRNA and asap^shRNA2 in combination with Arf4-GFP (stained for Amph for staging; embryos at interphase). Images collected and adjusted with the same settings. Note the increase of Arf4-GFP levels at Golgi puncta with Asap RNAi. (D) Quantification of relative Arf1-GFP fluorescence at Golgi puncta for Control^shRNA, asap^shRNA2, and asap^shRNA1. Each dot represents the average levels of one embryo. Black bars indicate median values. Asterisks indicate significant differences (p < 0.0001). (E) Quantification of relative Arf4-GFP fluorescence at Golgi puncta for Control^shRNA and asap^shRNA2 as in D. Asterisk indicates significant differences (p < 0.0001). (F) Quantification of relative Arf1-GFP fluorescence at Golgi puncta for mCherry and Asap-mCherry–overexpressing embryos as done in D. Asterisk indicates significant differences (p < 0.0001). (A–F) Constructs expressed as in Figure 1. (D–F) Data compiled from two or more experiments, each with similar results. (G) Comparison of the localization of Asap-mCherry and Arf1-GFP in a syncytial embryo at interphase. Asap-mCherry localizes most strongly to the PM, with no enrichment at Golgi elements, where Arf1-GFP is most strongly enriched (compare the proteins in the section 3.6 μm below the apical surface). Intermediate pools of Arf1-GFP are found around the nuclear periphery and at the PM (forming concentric circles in the section 1.2 μm below the apical surface). The intermediate pool of Arf1-GFP at the PM colocalizes with Asap-mCherry (arrows). Other than the accumulation difference at Golgi elements, the localization of Arf1-GFP was similar when coexpressed with mCherry (unpublished data).

To further address how Asap affects the localization of Arf1, we investigated the consequences of Asap overexpression. Specifically, we coexpressed Arf1-GFP with an Asap-mCherry fusion protein or mCherry alone. Compared to the mCherry control, Asap-mCherry overexpression significantly increased Arf1-GFP levels at the Golgi at interphase (Figure 6F). To understand how Asap-mCherry could elevate Arf1-GFP levels at the Golgi, we compared the localization patterns of the proteins at interphase. Consistent with the localization data of the GFP-tagged Asap constructs and the antibody detection of endogenous Asap (Figure 3), Asap-mCherry localized most prominently to the PM and displayed no enrichment at Arf1-GFP–positive Golgi elements (Figure 6G). Intriguingly, Asap-mCherry did colocalize with an intermediate pool of Arf1-GFP at the PM furrows (Figure 6G, arrows). Together these results indicate that Asap is necessary and sufficient for Arf1 accumulation at the Golgi and may affect Arf1 at non-Golgi sites.

Asap affects the organization of Arf1 effectors at the Golgi

To address whether Asap affects the activity of endogenous Arf1 at the Golgi, we examined the localization of two Arf1 effectors—the COPI coat and the clathrin adaptor AP-1 complex (Lorente-Rodriguez and Barlowe, 2011; Guo et al., 2014; Papanikou and Glick, 2014). To probe for effects on the COPI coat, we examined a γCOP-GFP construct. In contrast to control RNAi, strong asap RNAi or arf1 RNAi significantly reduced the levels of γCOP-GFP at Golgi puncta, and both perturbations did so to a similar degree (Supplemental Figure S4A). Thus Asap and Arf1 have a common role in promoting COPI coat localization at the Golgi. In contrast, strong asap RNAi had no apparent effect on AP-1γ levels at the Golgi, although AP-1γ localized to abnormally aggregated Golgi elements (Supplemental Figure S4B), as observed for the other trans-Golgi marker, LqfR (Figure 5).

Golgi organization and furrow ingression require an Arf1-Asap binding site

To test whether Arf1 is affected by a direct interaction with Asap, we generated Drosophila versions of two mouse Arf1 variants shown to be differentially regulated by ASAP1 with in vitro assays of purified Arf1 and purified Arf-GAPs (Luo et al., 2005). Specifically, the GTPase activity of Arf1^I46D is highly insensitive to activation by ASAP1 and weakly insensitive to activation by AGAP1 or Arf GAP1, whereas the GTPase activity of Arf1^N52A is weakly insensitive to ASAP1, AGAP1, or Arf GAP1 (Luo et al., 2005), thus providing a control for effects unrelated to strong Asap insensitivity. Because Drosophila and mouse Arf1 have 97% amino acid residue identity and the region of mouse ASAP1 assayed by Luo et al. (2005) has 48% amino acid identity with Drosophila Asap, we generated corresponding Drosophila Arf1 constructs with additional silent mutations for resistance to the arf1 shRNA. Coexpression of Arf1^N52A with arf1 shRNA substantially rescued the cleavage furrow and Golgi defects that otherwise occur with arf1 RNAi, but coexpression of the Asap-insensitive Arf1^I46D failed to rescue the defects (Figure 7, A and B). Of note, arf1 RNAi, like asap RNAi, also led to partial female infertility, which was rescued by the Asap-sensitive Arf1 construct but not the Asap-insensitive Arf1 construct (Table 1). Expression of the Asap-insensitive Arf1 construct in an otherwise wild-type background induced female infertility (Table 1), indicating that the transgene does produce an active protein. Overall these data argue that Asap acts directly on Arf1 for Golgi organization and furrow growth in the embryo.

FIGURE 7:

An Asap-insensitive Arf1 construct is relatively ineffective for furrow ingression and Golgi organization. (A) 3D reconstructions of metaphase syncytial embryos expressing Arf1 RNAi with RNAi-resistant Arf1^[N52A] (left) or RNAi-resistant Arf1^[I46D] (right). Amph shows furrows, and pH3 shows cell cycle stage. Control Arf1^[N52A] considerably rescues Arf1 RNAi embryos, but Asap-insensitive Arf1^[I46D] does not. Below, xy-sections from the embryo surface to below the nuclei. Quantified to the right with Arf1 RNAi embryos coexpressing GFP as a control for the number of UAS constructs. N, number of metaphase embryos (compiled from two or more experiments, each with similar results). (B) Single cross-sections below the nuclei of arf1^shRNA2 Arf1^[N52A] (left) and arf1^shRNA2 Arf1^[I46D] (right) embryos stained for GM130 and LqfR (embryos at interphase). More abnormal Golgi aggregation was observed with the Asap-insensitive Arf1^[I46D] than the control Arf1^[N52A]. (A, B) Constructs expressed as in Figure 1.

Subtle changes to Arf1 localization and Golgi structure over the syncytial cell cycle

As mammalian cells enter mitosis, inhibition of Golgi Arf GEFs leads to the loss of Arf1 from the Golgi and Golgi disassembly (Altan-Bonnet et al., 2003). The essential role of Asap for Arf1 function at the Golgi suggests that it could also be a control point, and the relocalization of Asap to the nuclear region at mitosis suggests that such control could be cell-cycle regulated. Thus we examined changes to Arf1 and the Golgi over the cell cycle in the early Drosophila embryo.

We began by probing for Arf1-GFP localization in fixed tissues with markers of the cell cycle, Golgi, ER, and PM (Figure 8A). At all stages of the cell cycle, strong Arf1-GFP puncta were present and juxtaposed with the trans-Golgi marker, LqfR (Figure 8A, single- and double-channel images). However, at metaphase, we observed an increase of Arf1-GFP at both the PM (Figure 8A, blue arrow) and the ER (Figure 8A, yellow arrow, inset, and Supplemental Figure S5), with Arf1-GFP exclusion from the nuclear region (Supplemental Figure S5). These results indicate that both Arf1–Golgi association and Golgi stacking are largely maintained through the cell cycle but that Arf1 partially relocalizes to the PM and ER at metaphase.

FIGURE 8:

Changes to Arf1 localization and Golgi structure over the cell cycle. (A) Single sections of the subapical and nuclear regions of fixed Arf1-GFP–expressing embryos at interphase, metaphase, and telophase. At interphase, Arf1-GFP localizes to Golgi puncta and at the PM to a lesser extent and is excluded from the nucleus. At metaphase, Arf1-GFP increases at the PM (blue arrows) and ER (yellow arrows; inset, RFP-KDEL as a coexpressed ER marker) but is excluded from the spindle region within the ER. By telophase, the Arf1-GFP distribution resembles that at interphase. Side images are single sections below the nuclei of Arf1-GFP embryos at interphase, metaphase, and telophase stained for LqfR to show Golgi ministacks throughout the cell cycle. (B) 3D reconstructions of Arf1-GFP puncta from time-lapse imaging of a single embryo at different stages of the cell cycle. Arf1-GFP puncta are color coded according to volume. (C) Histograms showing Arf1-GFP (left) and GM130 (right) puncta volume distributions at different stages of the cell cycle (color coded). From interphase to telophase, puncta volumes decrease for both Arf1-GFP and GM130. The Arf1-GFP histogram is for time-lapse imaging of a single embryo, and each dot represents the percentage of Arf1-GFP puncta within a volume range (the same trend was observed for three embryos; unpublished data). The GM130 histogram is from imaging of fixed embryos at different cell cycle stages, and each dot represents a mean ± SD of the percentage of GM130 puncta within a volume range for the embryo at that cell cycle stage (embryo numbers in inset). Insets quantify total numbers of puncta per cell cycle stage (color coded). For Arf1-GFP, each bar represents one time frame (30 s/time frame) from time-lapse imaging of one embryo (the same trend was observed for three embryos), and for GM130, the bars show means ± SD. An increase in total puncta number was evident for both Arf1-GFP and GM130 (*p < 0.05). N, embryo numbers (compiled from two or more experiments, each with similar results). (A–C) Constructs expressed as in Figure 1.

For more precise examination of the Arf1-positive Golgi elements, we examined Arf1-GFP live, with coexpressed histone–monomeric red fluorescent protein (mRFP) for staging over the cell cycle. Time-lapse, three-dimensional (3D) analyses of individual embryos revealed that the Arf1-positive Golgi puncta become smaller and more numerous from interphase to metaphase to telophase (Figure 8, B and C). To test whether these changes were due to changes in general Golgi structure, we analyzed fixed embryos stained for the cis-Golgi marker GM130 and discovered the same trend (Figure 8C). Thus the Golgi undergoes subtle reorganization over the syncytial cell cycle in Drosophila.

DISCUSSION

Our results identify an essential and specific role for Asap in promoting Arf1 activity at the Golgi. Asap promotes Arf1 but not Arf4 localization to the Golgi. Both Asap and Arf1, as well as a conserved residue responsible for their interaction, are required for proper Golgi structure and the biosynthesis of cleavage furrows. Our genetic analyses argue that the requirements of Asap and Arf1 for furrow ingression are due primarily to effects at the Golgi rather than possible effects of either at the PM. Of note, we also discovered a translocation of Asap to the nuclear region before each cycle of furrow regression and that this change coincides with a transient alteration to Golgi structure, as well as to accumulation of Arf1 at the PM and ER. Overall our data reveal an Asap-Arf1-Golgi pathway for cleavage furrow biosynthesis and that this pathway may be regulated over the cell cycle.

Although we cannot exclude the possibility of Asap acting at the Golgi, the lack of Asap enrichment at Golgi elements suggests that Asap promotes Golgi Arf1 accumulation indirectly, perhaps by recycling the small G protein from other sites. Asap could recycle Arf1 in two ways. First, Asap might inactivate Arf1 to release it from coat proteins (Gillingham and Munro, 2007; Spang et al., 2010; Donaldson and Jackson, 2011; East and Kahn, 2011), thereby resupplying Arf1 after vesicle budding from the Golgi. Second, Asap could function within a kinetic scaffold. This effect increases the supply of GDP-bound G proteins for reaction with GEFs and has been modeled to explain the ability of GAPs to promote the activity of G protein–coupled receptors (Ross, 2008). For Asap, the effect could be coupled with the uncoating of Golgi-derived vesicles or supply Arf1 from more distant membranes (such as the PM, where Asap–Arf1 colocalization is evident).

What is the relationship between Asap and other Arf GAPs that affect Arf1 at the Golgi? mRNA for each of the annotated Golgi Arf GAPs is present in cleavage stage Drosophila embryos (FlyBase, http://flybase.org/). This expression includes Arf GAPs that affect Arf1 at the Golgi in other systems and do so redundantly for cell viability (Gillingham and Munro, 2007; Spang et al., 2010; Donaldson and Jackson, 2011; East and Kahn, 2011). Because Asap’s promotion of Arf1 at the Golgi seems essential for cleavage furrow formation, its specific mechanism of action may be unique, perhaps due to the larger complexes it forms and/or to the subcellular location where it acts.

How does mitotic Golgi reorganization compare among different systems? Mitosis of mammalian cells involves strong loss of Golgi Arf1 (Altan-Bonnet et al., 2003), disassembly of Golgi ribbons (Altan-Bonnet et al., 2003), and destacking of Golgi elements (Lucocq et al., 1989; Misteli and Warren, 1995; Zaal et al., 1999). Such disassembly may allow Golgi membranes to partition evenly between daughter cells, and, in addition, loss of Arf1 leads to release of cytoskeletal regulators from the Golgi for roles in furrow formation (Altan-Bonnet et al., 2003). In the early Drosophila embryo, mitotic changes to Golgi organization seem quite distinct. First, extended Golgi ribbons are never present, and at each cell cycle stage, the Golgi exists as numerous ministacks that would presumably partition evenly to daughter compartments. Second, Arf1 only partially redistributes from the Golgi at metaphase (to the ER and PM), and only subtle changes to Golgi elements are detected (to their sizes and numbers). Third, this metaphase Golgi reorganization is followed by cleavage furrow regression rather than ingression. There may also be a major regulatory difference between these systems. Specifically, the Arf1 loss in mammalian cells is primarily due to reduced Arf GEF activity (Altan-Bonnet et al., 2003). In contrast, we speculate that the relocalization of Asap to the nuclear region at prometaphase could reduce both Arf1 function at the Golgi and the biosynthesis of furrows. Although definition of the nuclear localization mechanism is required for understanding this or other effects at mitosis, sequestration of an Arf GAP, rather than loss of an Arf GEF, could allow subtle Golgi regulation for cycles of furrow regression in rapid succession with extensive furrow growth. Significantly, furrow regression could also involve sequestration of other players, as nonmuscle myosin II relocalizes to the nuclear region at mitosis (Lee and Harris, 2013), as does dynamin (Rikhy et al., 2015). Perhaps sequestration of several factors temporally halts the fuelling and maintenance of cleavage furrows for these natural cycles of furrow regression.

Overall our results show that Asap is essential for Arf1 to function at the Golgi for cleavage furrow biosynthesis. Asap may recycle Arf1 to the Golgi from post-Golgi membranes, providing optimal Golgi output for specific stages of the cell cycle.

MATERIALS AND METHODS

Drosophila stocks and genetics

Established stocks.

The following fly lines were used: maternal-α4-tubulin-GAL4::VP16 (gift of M. Peifer, University of North Carolina, Chapel Hill, NC), Maternal Triple Driver (MTD; Bloomington Drosophila Stock Center [BDSC], stock 31777), UASp-mCherry-shRNA (P[VALIUM20-mCherry]attP2, BDSC, stock 35785), UASp-GFP-Asap, UASp-Arf1-GFP (Shao et al., 2010), UASp-Arf4-GFP (Lee et al., 2015a), UASp-arf1-shRNA2 (Lee et al., 2015a), UASp-RFP-KDEL (BDSC, stock 30910), UASp-yCOP-GFP (BDSC, stock 29711), Histone2Av-mRFP1 (BDSC, stock 23650), histone-GFP (gift of A. Wilde, University of Toronto, Toronto, Canada), and GAP-43-mCherry (gift of S. Simoes, University of Toronto). UASp-Asap-GFP and UASp-Asap-mCherry were generated as UASp-GFP-Asap (Shao et al., 2010), with the exception of destination vectors for C-terminal GFP or mCherry tagging.

asap shRNA constructs.

The shRNAs were designed based on the algorithm by Vert et al. (2006). For asap1, they targeted two unique sequences: shRNA1 top strand, 5′-ctagcagtTAAG­TTCAGCGTTGTCACCAAtagttatattcaagcataTTGGTGACAACGCTGAACTTAgcg-3′; and shRNA2 top strand, 5′-ctagcagtAAGACTGTGATCCGATATGTAtagttatattcaagcataTACATATCGGATCACAGTCTTgcg-3′. The control shRNA construct targeted the top strand sequence 5′-ctagcagtCCGAGTGTGTTTCTGA­TCCTAtagttatattcaagcataTAGGATCAGAAACACACTCGGgcg-3′. The asap1 shRNA1 and control shRNA constructs were ligated into the pValium22 vector, and the asap1 shRNA2 construct was ligated into the pValium20 vector (gifts from the Transgenic RNAi Resource Project, www.flyrnai.org/TRiP-HOME.html/), using the restriction enzymes NheI and EcoRI, confirmed by PCR, sequenced, and targeted to the attp40 recombination site on chromosome 2 for transgenic flies (BestGene, Chino Hills, CA).

RNAi-resistant Arf1 constructs.

Gene synthesis (GenScript Biotech, Piscataway, NJ) produced the nucleotide sequence of the arf79f cDNA (LD24904) flanked by NotI and XbaI restriction sites (5′-GCGGCCGC[TTTGGCAGCA…TTTAATTTTT]TCTAGA-3′) with the shRNA target region mutated from 5′-CTGAGGGATGCAGTCTTACTA-3′ to 5′-TTACGTGACGCCGTACTCTTG-3′ (maintaining the encoded protein sequence) and mutations to produce the amino acid change I46D or N52A. The vector was cut by NotI and XbaI and cloned into the pUASp-attB vector (with no protein tag). The vectors were inserted into the genome at the attP2 recombination site on the third chromosome for transgenic flies (BestGene).

Crosses.

For UAS transgene expression in the early embryo, we analyzed progeny of mothers heterozygous for maternal-α4-tubulin-GAL4::VP16 or MTD and for the transgene.

Embryo staining and imaging

Embryos were fixed for 20 min in 1:1 3.7% formaldehyde in phosphate-buffered saline (PBS):heptane and then devitellinized in methanol. Blocking and staining were in PBS containing 1% goat serum, 0.1% Triton X-100, and 1% sodium azide. Antibodies used were as follows: rabbit, amphiphysin (1:2000; gift of G. Boulianne, Hospital for Sick Children, Toronto, Canada), Asap (1:500; gift of R. Johnson, Wesleyan University, Middletown, CT), and GM130 (1:700; Abcam, Cambridge, United Kingdom); mouse, AP-1γ (1:500; gift of R. Le Borgne, University of Rennes, Rennes, France), Dlg (1:100; Developmental Studies Hybridoma Bank [DSHB]), and phospho–histone H3 (1:2000; Cell Signaling Technology, Beverly, MA); and rat, LqfR (1:500; gift of G. Boulianne). Secondary antibodies were conjugated to Alexa Fluor 488, 568, and 647 (Life Technologies, Burlington, Canada). Embryos were mounted in Aqua Polymount (Polysciences, Warrington, PA).

For tubulin and actin staining, embryos were washed with 0.1% Triton X-100, dechorionated with 50% bleach, washed with 0.1% Triton X-100, fixed for 10 min in 1:1 10% formaldehyde in PBS:heptane, devitellinized by hand peeling, and stained. Tubulin was stained with mouse anti-tubulin (E7; 1:100; DSHB). F-actin was stained with Alexa Fluor 568–conjugated phalloidin (1:200; Life Technologies).

For live imaging, dechorionated embryos were glued to a coverslip using tape adhesive dissolved in heptane and mounted in halocarbon oil (series 700; Halocarbon Products, River Edge, NJ). The coverslip, with the embryos facing up, was set into the bottom of a glass-bottom culture dish with its original coverslip removed.

Immunofluorescence images were collected by a spinning-disk confocal system (Quorum Technologies, Guelph, Canada) at room temperature using 40× Plan Neofluar numerical aperture (NA) 1.3 and 63× Plan Apochromat NA 1.4 objectives (Carl Zeiss, Toronto, Canada) with a piezo top plate and a Hamamatsu C9100-13 electron-multiplying charge-coupled device camera (Hamamatsu Photonics, Hamamatsu, Japan), with z-stacks of 1-μm and 300-nm step sizes for the 40× and 63× objectives, respectively.

Postacquisition image analyses and quantification

Basic image analysis and preparation.

Images were analyzed with Volocity software (PerkinElmer, Waltham, MA). Photoshop (Adobe, Mountain View, CA) was used for figure preparation. Except where noted, input levels were adjusted so the main signal range spanned the entire output grayscale. Images were resized by bicubic interpolation without noticeable changes at normal viewing magnifications.

Quantifications of Golgi Arf1-GFP and Arf4-GFP levels.

For each embryo, ImageJ (National Institutes of Health, Bethesda, MD) was used to measure mean fluorescence intensity of the GFP signal within a circle with a diameter of 0.6 μm covering each of five puncta most strongly stained with the Golgi marker LqfR. When the raw fluorescence signal was too weak to be clearly seen by the naked eye, deconvolved and adjusted images of the LqfR staining indicated the position of the Golgi to measure the raw Arf1-GFP signal at these sites (the Arf1-GFP data were not manipulated before the quantification). Background values were measured within the nuclear region nearest to the Golgi measurement and subtracted from the Golgi measurements individually. For each embryo, these corrected values were then averaged to calculate a single fluorescence intensity measurement, which was then plotted and compared with those of other embryos.

Quantification of furrow defects.

Syncytial embryos were staged at metaphase by chromosome morphology seen by phospho–histone H3 staining. Only embryos in cell cycle 11, 12, or 13 were analyzed. Cycle 10 embryos were excluded by nuclear density. Embryos with ≤50 nuclei per field of view (65.3 μm [x] × 65.3 μm [y]) were discarded. Furrow integrity was analyzed with Amph staining. Embryos with nuclei completely surrounded by ingressing furrows were categorized as having intact cleavage furrows. Embryos with nuclei that were only partially enclosed by furrows were categorized as having partial furrow defects. Embryos in which >50% of nuclei did not have any ingressed furrows surrounding them were categorized as having severe furrow defects. Only somatic nuclei at the equator of the embryo were imaged for analysis.

Three-dimensional surface rendering of Golgi puncta.

Imaris 6.2 software (Bitplane AG, Zurich, Switzerland) was used for 3D surface rendering of Arf1-GFP and GM130 puncta. Parameters were optimized such that 3D surfaces closely resembled the raw data by eye. Data sets for time points of Arf1-GFP live imaging were quantified with a volume of 65.3 (x) × 65.3 (y) × 18 μm (z), including the full peripheral cell compartments. Data sets for the fixed GM130 images were quantified with a volume of 65.3 (x) × 65.3 (y) × 6 μm (z) just below the nuclei of the peripheral cell compartments.

Statistics.

Comparisons were done using Student’s t tests (independent samples, two-tailed). Means are shown with SDs. Scatterplots prepared with an available template (Weissgerber et al., 2015).

Abbreviations used:

Amph

amphiphysin

Asap

ArfGAP with SH3 domain, ankyrin repeat, and PH domain

BDSC

Bloomington Drosophila Stock Center

DSHB

Developmental Studies Hybridoma Bank

ER

endoplasmic reticulum

GAP

GTPase-activating protein

GEF

guanine nucleotide exchange factor

LqfR

Liquid facets-related

MTD

Maternal Triple Driver

pH3

phospho–histone H3

PM

plasma membrane

PGC

primordial germ cell

shRNA

short hairpin RNA

TGN

trans-Golgi network.