TRAPPII is required for cleavage furrow ingression and localization of Rab11 in dividing male meiotic cells of Drosophila

In animal cell cytokinesis, the dynamic processes of contractile ring constriction and concomitant cleavage furrow ingression serve to physically partition the nascent daughter cells during the late stage of cell division. To provide the surface area required for cleavage furrow ingression, many animal cell types insert new membrane into the furrow (Aimar, 1997; Bluemink and de Laat, 1973; Danilchik et al., 2003; Shuster and Burgess, 2002; Skop et al., 2001) or modify the organization of their pre-existing plasma membrane by rounding up or flattening microvilli (Blose, 1979; Boucrot and Kirchhausen, 2007; Erickson and Trinkaus, 1976; Knutton et al., 1975). These strategies for accommodating cleavage furrow ingression are probably governed by cell shape, surface membrane organization, and internal membrane stores.

Consistent with these modifications of the plasma membrane, cytokinesis depends on vesicle trafficking components that include the membrane fusion machinery, regulatory small GTPases and multi-protein complexes involved in both exocytosis and endocytosis (for reviews, see Albertson et al., 2005; Finger and White, 2002; Montagnac et al., 2008; Prekeris and Gould, 2008). For example, vesicle-targeting syntaxins have been shown to be required for cytokinesis in embryos of the sea urchin and C. elegans (Conner and Wessel, 1999; Jantsch-Plunger and Glotzer, 1999), as well as for the cognate process of cellularization in Drosophila embryos (Burgess et al., 1997). Additionally, in dividing cultured cells, the final separation of daughter cells following furrow ingression is blocked by reduced function of the proteins syntaxin-2, endobrevin, soluble N-ethylmaleimide sensitive factor attachment protein (α-SNAP), the exocyst complex, and the GTPases ARF6, Rab35 and Rab11 (reviewed in Montagnac et al., 2008). Many of these components highlight the increasingly prominent role of endosomal trafficking in cytokinesis and cellularization (Albertson et al., 2005; Montagnac et al., 2008; Prekeris and Gould, 2008).

Cytokinesis in male meiotic cells (spermatocytes) of Drosophila melanogaster also depends on a number of vesicle-trafficking components. These include the intra-Golgi trafficking conserved oligomeric Golgi (COG) complex subunit 5 encoded by four way stop (fws) (Farkas et al., 2003), the endoplasmic reticulum (ER)-to-Golgi vesicle-docking protein encoded by Syntaxin 5 (Xu et al., 2002), the GTPases Rab11 (Giansanti et al., 2007) and Arf6 (Dyer et al., 2007), the phosphatidylinositol 4-kinase β (PI4Kβ) encoded by four wheel drive (fwd) (Brill et al., 2000), and the phosphatidylinositol transfer protein encoded by giotto (gio) (Gatt and Glover, 2006; Giansanti et al., 2006). Strikingly, mutations in fws, Rab11, Arf6, fwd and gio cause defects in actomyosin ring constriction and concomitant failure of cleavage furrow ingression in spermatocytes, suggesting an intimate relationship between membrane trafficking and contractile ring behavior during cytokinesis in this cell type.

Here, we show that function of the membrane trafficking transport protein particle (TRAPP) II complex is required for cleavage furrow ingression during cytokinesis in Drosophila spermatocytes. Mutations in the gene brunelleschi (bru), which encodes the Drosophila ortholog of the yeast TRAPPII TRS120p subunit, cause failure of both actomyosin ring constriction and cleavage furrow ingression. In addition, we found that bru genetically interacts with Rab11 and is required for the localization of Rab11 to the cleavage furrow of dividing spermatocytes. Furthermore, we found that bru genetically interacts with fwd, which has also been shown to be required for the localization of Rab11 to the cleavage furrow in spermatocytes (Giansanti et al., 2007). These genetic interactions suggest that bru also functions in membrane trafficking in Drosophila, possibly as a subunit of a TRAPPII complex. Thus, we propose that bru mediates the flow of membrane through the Golgi that is required to support cleavage furrow ingression in dividing spermatocytes.

Two mutant alleles of bru, bru^Z3358 and bru^Z0704, were previously isolated from a collection of male-sterile mutants that was screened for mutations causing failure of cytokinesis in spermatocytes (Giansanti et al., 2004). Our initial characterization of fixed spermatocytes from bru males raised at 25°C revealed defects in contractile ring constriction during telophase (Giansanti et al., 2004). We subsequently found that the defect in bru mutant males is temperature sensitive, because the failure of meiotic cytokinesis observed at 25°C was strongly exacerbated at 29°C and suppressed at 18°C. At 29°C, early spermatids from bru^Z3358/+ testes appeared normal and contained a single phase-light nucleus associated with a similarly sized, phase-dark mitochondrial derivative (MD) (Fig. 1A). By contrast, ∼80% of early spermatids from bru^Z3358/Df(2L)pr2b mutant males raised at 29°C contained four nuclei associated with a single, enlarged MD (Fig. 1A and 1B), reflecting cytokinesis failures during both meiotic divisions (Fuller, 1993). The vast majority of spermatids from bru^Z3358/Df(2L)pr2b mutant males raised at 18°C appeared normal (Fig. 1B). The phenotype of bru^Z0704 exhibited a similar temperature sensitivity in bru^Z0704/Df(2L)pr2b males (C.C.R., unpublished).

Our previous characterization of actin localization in fixed spermatocytes from bru mutants raised at the intermediate temperature of 25°C suggested that contractile rings formed normally at the onset of cytokinesis in early telophase, but were frequently unconstricted or broken in later stages of division (Giansanti et al., 2004). To analyze contractile rings in vivo, we examined the behavior of the myosin regulatory light chain fused to the green fluorescent protein (RLC-GFP) in live spermatocytes (Royou et al., 2002). RLC-GFP marks the contractile ring throughout telophase, allowing visualization of ring constriction (Dyer et al., 2007). In wild-type spermatocytes (n=10), RLC-GFP rings initiated constriction immediately after assembly and completed constriction within 25 minutes (Fig. 1C, supplementary material Movie 1). In all bru^Z3358/Df(2L)pr2b mutant spermatocytes raised at 29°C (n=9), seemingly normal RLC-GFP rings assembled in mid-anaphase. However, in eight of these cells, RLC-GFP rings did not constrict, or constricted minimally, throughout the observation period (30 minutes) (Fig. 1C, supplementary material Movie 2). In addition, the unconstricted RLC-GFP rings in these spermatocytes eventually fragmented. To quantify the defect in constriction, we scored a population of telophase spermatocytes from wild-type and bru mutant flies raised at 29°C and categorized the rings as either constricted or unconstricted based on their apparent size. In bru^Z3358/Df(2L)pr2b, unconstricted RLC-GFP rings were observed in 65% of telophase spermatocytes (n=83), whereas unconstricted rings were observed in only 10% of the wild-type telophase spermatocytes (n=117). Together, these results indicate that bru is required for ring constriction and concomitant furrow ingression.

The requirement for bru in cytokinesis appeared to be specific to spermatocytes. The bru mutants were fully viable through adulthood and female fertility was not grossly affected by loss of bru function. The quantity, viability and hatching frequency of eggs laid by bru^Z3358/Df(2L)pr2b females raised at 25°C and mated to wild-type Oregon R males were indistinguishable from that observed for their heterozygous bru ^Z3358/+ and Df(2L)pr2b/+ siblings (C.C.R., unpublished). Additionally, we did not detect polyploid cells in larval brains from bru mutant larvae, indicating that neuroblast cytokinesis is not affected (M.G.G., unpublished).

The bru locus was mapped to an ∼24-kbp interval (see the Materials and Methods), and several annotated candidate genes in this interval were sequenced. Each of the two mutant alleles of bru was found to have a single base-pair change that introduced a premature stop codon in the annotated gene CG2478, which is predicted to encode a 1320 amino acid polypeptide (http://flybase.org) (Wilson et al., 2008). bru^Z3358 and bru^Z0704 would truncate this polypeptide after residues 514 and 754, respectively (Fig. 2A). Identification of bru as CG2478 was confirmed by transgenic rescue of the male meiotic cytokinesis defect using a 5.9-kbp genomic fragment that contains only the complete wild-type CG2478 gene. Northern blot analysis revealed expression of the expected ∼4.9-kbp message in adults of both sexes and in embryos (data not shown), indicating that bru expression was not restricted to the male germline.

Sequences predicted to encode homologs of the Bru protein were found in human, Dictyostelium discoideum, nematode, plant and yeast by tBlastn queries (Fig. 2B). The Bru homolog in Saccharomyces cerevisiae is Trs120p, an essential subunit of the TRAPPII complex that is required for membrane trafficking at the late Golgi (Cai et al., 2005; Sacher et al., 2000; Sacher et al., 2001). Bru was the only homolog of Trs120p identified in the D. melanogaster genome by tBlastn query (Fig. 2C), indicating that bru encodes the ortholog of Trs120p. In yeast, TRAPPII is a decameric complex composed of the seven TRAPPI subunits, plus three additional proteins, Trs120p, Trs130p and Trs65p (Sacher et al., 2001). Drosophila appears to have an orthologous TRAPPII complex because its genome encodes single, significant-scoring polypeptide orthologs for all of the S. cerevisiae TRAPPII subunits except Trs65p and Trs85p (Fig. 2C), which is consistent with a recent phylogenetic analysis (Cox et al., 2007). Although an ortholog of S. cerevisiae Trs85p was not identified, a tBlastn query with the Schizosaccharomyces pombe Trs85p ortholog revealed a Drosophila homolog (Fig. 2C). As expected, we did not find a Drosophila homolog of the TRAPPII-specific Trs65p, which is conserved only among fungi and some unicellular eukaryotes (Cox et al., 2007). To our knowledge, the bru mutations we describe are the first reported for a TRAPPI/II subunit in Drosophila and the first for a TRAPPII-specific subunit in animals.

Amino acid sequence conservation between the Bru homologs identified by ClustalW alignment indicated that the region of highest conservation between Bru and Trs120p is a novel 50-residue motif spanning residues 243-293 (Fig. 2B), which corresponds to Trs120p domain II (Cox et al., 2007). Consistent with the possibility that one or more of the highly conserved charged residues of this motif are important for Bru function, simultaneous replacement of K244, R249, K252, D256, E276 and D283 with alanines (Fig. 2B) in the bru genomic rescue construct abolished rescue activity: none of the eight transgenic lines carrying the multiply mutated construct rescued the bru male meiotic cytokinesis defect, whereas all five transgenic lines carrying either wild-type bru or bru-GFP rescued fully.

Because TRAPPII function is required for membrane trafficking at the Golgi in S. cerevisiae, we asked whether meiotic cytokinesis in cultured larval testes is sensitive to the Golgi membrane trafficking inhibitor brefeldin A (BFA) (Klausner et al., 1992). Treatment of cultured testes with BFA caused disruption of Golgi organelles in spermatocytes expressing EYFP-Golgi, a derivative of galactosyltransferase (LaJeunesse et al., 2004) (Fig. 2D). Furthermore, wild-type testes cultured with BFA exhibited a marked failure of cytokinesis in meiosis I and II. Whereas early spermatids from control testes contained a single nucleus paired with a normal-sized MD, early spermatids from BFA-treated testes frequently contained four nuclei associated with an enlarged MD (Fig. 2E). These data indicate that BFA-sensitive components essential for Golgi membrane trafficking are required for male meiotic cytokinesis. Similarly, cytokinesis in C. elegans embryos, cellularizing Drosophila embryos and HeLa cells is also sensitive to BFA, although the sensitivity of cytokinesis varies between cell types (Albertson et al., 2005; Montagnac et al., 2008).

To determine the subcellular localization of Bru in male germline cells, we generated a bru-GFP transgene. This transgene fully rescued the bru mutant phenotype, indicating that the encoded Bru-GFP fusion protein is functional. Live prophase spermatocytes expressing bru-GFP exhibited a general cytoplasmic fluorescence plus a number of fluorescent bodies that colocalized with EYFP-Golgi (Fig. 3A). Immunofluorescent localization of Bru-GFP in fixed spermatocytes confirmed the presence of the fluorescent bodies and showed that they were closely associated with the Golgi marker p120 (Fig. 3B), which is consistent with the subcellular localization of epitope-tagged Trs120p to the late Golgi compartment in yeast (Cai et al., 2005). However, in yeast there has been no report of a cytoplasmic pool of Trs120p. The distribution of Bru-GFP was similar to that of the Drosophila TRAPPI/II subunit ortholog, dBet3, which we detected as a fusion of YFP to the Bet3p ortholog encoded by CG3911. In live prophase spermatocytes, dBet3-YFP was distributed as a haze throughout the cytoplasm and was also enriched at fluorescent bodies that colocalized with the COG5 ortholog Four-way stop fused to GFP (GFP-Fws) (Farkas et al., 2003) (Fig. 3C). During meiotic division, Bru-GFP was distributed throughout the cytoplasm, without consistent enrichment at any subcellular compartment (Fig. 3D).

Similarly to bru, wild-type Rab11 function is required for contractile ring constriction during Drosophila male meiotic cytokinesis (Giansanti et al., 2007), and in yeast, TRS120 genetically interacts with the Rab11 orthologs, YPT31/32 (Zhang et al., 2002). This prompted us to ask whether bru genetically interacts with Rab11. Flies that were wild type for bru and homozygous for Rab11^93Bi or transheterozygous for Rab11^93Bi/Rab11^E(To)3 produced viable adults (Giansanti et al., 2007; Jankovics et al., 2001). Strikingly, however, Rab11^93Bi was fully lethal in combination with bru. Flies homozygous for both bru^Z3358 and Rab11^93Bi died in early larval stages at 25°C, as did flies homozygous for bru^Z3358 and transheterozygous for Rab11^93Bi/Rab11^E(To)3. By contrast, bru^Z3358 homozygotes that were wild type for Rab11 were fully viable through adulthood at 25°C.

Consistent with a strong genetic interaction between bru and Rab11, the defect in male meiotic cytokinesis caused by loss of bru function was exacerbated by introducing a single copy of the recessive lethal Rab11^E(To)3 allele. Testes from bru^Z3358/Df(2L)pr2b;Rab11^E(To)3/+ males raised at 25°C exhibited a 4.5-fold increase in the percentage of tetranucleate early spermatids relative to testes from bru^Z3358/Df(2L)pr2b;+/+ males (Fig. 4A). Similarly, bru^Z3358/+ males that were homozygous for the weak allele Rab11^93Bi displayed a fourfold increase in the total number of multi-nucleate spermatids compared with homozygous Rab11^93Bi males that lacked mutations in bru (Fig. 4B).

bru also genetically interacted with fwd, a non-essential gene that encodes a PI4Kβ and is required for contractile ring constriction in spermatocytes (Brill et al., 2000; Giansanti et al., 2004). Testes from bru;fwd double-mutant males produced higher percentages of early spermatids containing four, eight or more nuclei associated with enlarged MDs than did testes from their siblings that contained one mutant and one wild-type copy of either bru or fwd (Fig. 4C,D). This indicated that simultaneous loss of both bru and fwd wild-type functions causes more frequent failure of cytokinesis during meiosis I and II, as well as a dramatic increase in the failure of cytokinesis during the mitotic amplification divisions that precede meiosis. In addition, the elongation of later-stage spermatids was defective in bru;fwd double-mutant testes (Fig. 4C). Thus, the function of bru appears to cooperate with the functions of fwd in the execution of cleavage furrow ingression and spermatid cyst elongation.

By contrast, bru did not interact genetically with pbl, which encodes a Rho guanine nucleotide (GTP) exchange factor that is required for assembly of the contractile ring (Giansanti et al., 2004; Prokopenko et al., 1999). The percentage of tetranucleate early spermatids in testes from bru;pbl double-mutant males was similar to that of testes from pbl mutants that were heterozygous for bru. There was no indication of cytokinesis failure during the preceding mitotic amplification divisions, and elongation of later-stage spermatids was similar to that of the bru mutant (Fig. 4E,F).

To determine whether simultaneous loss of either bru and fwd or and Rab11 affects mitotic division in somatic cells, we examined brains from third-instar larvae of bru;fwd double mutants, and the brains of the rare surviving second- or third-instar larvae from bru;Rab11 double mutants. In all cases, we did not detect polyploid cells in ∼200 metaphases scored, suggesting that mitotic cytokinesis in somatic cells is not defective in these double mutants.

Because bru genetically interacted with Rab11, we examined the localization of Rab11 in bru mutant spermatocytes during telophase. In the wild type, endogenous Rab11 localized to small puncta at the cell poles and to the cleavage furrow at the cell midzone (Fig. 5A) (Giansanti et al., 2007). By contrast, Rab11 did not show a specific pattern of localization in bru mutant spermatocytes undergoing telophase (Fig. 5A).

To confirm this result, we examined functional Rab11-GFP in dividing spermatocytes at different stages of the first meiotic division. Rab11-GFP exhibits dynamic localization behavior during the meiotic divisions, mirroring the localization of endogenous Rab11 (Giansanti et al., 2007). During the extended meiotic prophase that precedes cytokinesis, Rab11-GFP was dispersed throughout the cytoplasm and enriched at the multiple Golgi organelles in both wild-type and bru mutant spermatocytes (Fig. 5B). At anaphase of the first meiotic division in wild-type spermatocytes, Rab11-GFP fluorescence became fragmented and accumulated in puncta at opposite poles of the cell and at the parafusorial membranes (Fuller, 1993) that envelop the meiotic spindle (Fig. 5C, Fig. 6; supplementary material Movie 3). At telophase, fluorescent puncta persisted at the cell poles, and Rab11-GFP also accumulated in the cell midzone at the cleavage furrow (Fig. 5D and Fig. 6). In bru mutant spermatocytes, we did not observe distinct puncta during anaphase or telophase, and the overall distribution of Rab11-GFP appeared more diffuse than in the wild type (Fig. 5C,D, Fig. 6; supplementary material Movie 4). Additionally, Rab11-GFP did not accumulate at the cell midzone in bru mutant spermatocytes during telophase. Thus, bru function is required for proper Rab11 localization during spermatocyte cytokinesis.

Following completion of the meiotic divisions, Bru-GFP was dispersed throughout the cytoplasm of early spermatids and also enriched in a perinuclear region corresponding to the acroblast (Fig. 7A). The acroblast is a developmentally regulated Golgi derivative that is formed by the coalescence of Golgi fragments following the completion of meiosis and is marked by GFP-Fws (Fig. 7A) and the Golgi protein Lava lamp (Lva) (Farkas et al., 2003; Sisson et al., 2000; Tates, 1971). The acroblast is situated adjacent to the acrosomal granule to which it secretes exocytic cargo that will support function of the acrosome in mature sperm (Wilson et al., 2006). Consistent with the perinuclear localization of Bru-GFP, acroblast formation failed in bru mutant spermatids. Analysis of Lva localization revealed that 35% (n=120) of early spermatids from bru^Z3358/Df(2L)pr2b mutant testes exhibited Golgi fragments that had not coalesced into a single acroblast but rather segregated into numerous small Golgi organelles (Fig. 7B). By contrast, 100% (n=96) of wild-type early spermatids formed a normal acroblast. The bru defect in acroblast formation was not simply due to the multinucleate nature of the spermatids, because acroblasts were well formed in 100% (n=84) of tetranucleate early spermatids from the male meiotic cytokinesis mutant pebble (pbl) (Fig. 7B). fws, gio and Rab11 mutants also exhibit failure of both acroblast formation and cleavage-furrow ingression, implying a mechanistic link between these processes (Farkas et al., 2003; Giansanti et al., 2007; Giansanti et al., 2006).

To better understand the molecular mechanisms of cytokinesis, we characterized brunelleschi (bru), a gene that we had previously identified as being required for male meiotic cytokinesis in Drosophila. We have shown that bru encodes the Drosophila ortholog of the yeast protein TRS120p, a subunit of the TRAPPII vesicle-trafficking protein complex. Additionally, we found that bru genetically interacts with the membrane-trafficking genes Rab11 and fwd.

In yeast, TRAPPII performs membrane-trafficking functions at the Golgi and is required for viability. The complex is composed of the seven TRAPPI complex subunits plus three TRAPPII-specific subunits encoded by the essential genes TRS120 and TRS130, and the non-essential gene TRS65 (Sacher et al., 2000; Sacher et al., 2001). The fact that all of the yeast TRAPPII subunits except Trs65p (Cox et al., 2007) have orthologs in Drosophila raises the possibility that flies have a TRAPPII complex that functions in membrane trafficking at the Golgi. Whereas yeast TRAPPI facilitates ER-to-Golgi transport (Barrowman et al., 2000; Sacher et al., 2001; Sacher et al., 1998), TRAPPII appears to regulate trafficking events at the late Golgi because mutations affecting Trs120p disrupt trafficking from the early endosome to the Golgi and mutations affecting Trs130p cause a block in general secretion from the Golgi (Cai et al., 2005; Sacher et al., 2001). Biochemically, both TRAPPI and TRAPPII in yeast act as GDP-GTP exchange factors (GEFs) that activate Rab homologs: TRAPPI acts as a GEF for the Rab1 ortholog Ypt1p, whereas there is debate as to whether the GEF activity of TRAPPII is directed toward Ypt1p or the Rab11 othologs Ypt31p/Ypt32p (Cai et al., 2008; Jones et al., 2000; Sacher et al., 2001; Sacher et al., 2008). While we were writing this manuscript, Yamasaki and co-workers (Yamasaki et al., 2009) reported that mammalian cells possess only a single, high molecular weight TRAPP complex that is analogous to the TRAPPII complex of yeast. This mTRAPPII complex was found to act as a GEF for Rab1 but not Rab11 (Yamasaki, 2009). In contrast to yeast, mTRAPPII is associated with markers of the early Golgi where it appears to function in early intra-Golgi trafficking. Thus, the functions of TRAPPII might differ between yeast and metazoans (Yamasaki et al., 2009).

Given the greater evolutionary conservation between the Drosophila and mTRAPPII subunits (Cox et al., 2007), we anticipate that the activity of Drosophila TRAPPII will be more similar to that of the mammalian complex. Several observations are consistent with this possibility. Bru and dBet3 are both enriched at the Golgi while also distributed throughout the cytoplasm, similarly to the pattern exhibited by mTrs130 and mTrs23, although the latter have a more punctate cytoplasmic distribution (Yamasaki et al., 2009). Also, TRAPPII subunits from both organisms contribute to the organization of Golgi membranes. bru wild-type function is required for formation of the acroblast, a Golgi-derivative in spermatids that has an extensive cisternal organization (Farkas et al., 2003; Tates, 1971), and mTrs130 is required for the maintenance of Golgi stacks (Yamasaki et al., 2009).

We have observed genetic interactions between bru and Rab11. bru;Rab11 double mutants are synthetically lethal, and a single copy of bru exacerbates the failure of male meiotic cytokinesis in a Rab11 hypomorphic mutant background. Additionally, loss of bru wild-type function abrogates the localization of Rab11 to polar puncta and the cleavage furrow in dividing spermatocytes. These findings are consistent with observations in yeast, where overexpression of the Rab11 ortholog YPT31 suppresses lethality resulting from deletion of either TRS120 or TRS130 (Zhang et al., 2002), and loss of TRS130 function via a temperature-sensitive mutation causes redistribution of the Rab1 and Rab11 orthologs, Ypt1p, Ypt31p/Ypt32p (Morozova et al., 2006). Rab11 has been ascribed roles in both exocytic and endocytic pathways in different cell types (Chen et al., 1998; de Graaf et al., 2004; Satoh et al., 2005; Ullrich et al., 1996; Urbe et al., 1993; Wilcke et al., 2000). For example, Drosophila photoreceptor cells lack a recycling endosome (RE), and Rab11 localizes to the trans-Golgi where it is required for proper trafficking of rhodopsin to the plasma membrane (Satoh et al., 2005). Similarly, in spermatocytes, which also lack a Rab11-enriched RE, Rab11 localizes to the Golgi and is likely to function in the Golgi-to-plasma membrane trafficking required for cytokinesis (Giansanti et al., 2007). Rab11 also performs an essential role in furrow formation during the cellularization of Drosophila embryos via the RE (Riggs et al., 2003), and in many cell types, Rab11 and/or endosomal components contribute to the late stages of cytokinesis (Montagnac et al., 2008; Prekeris and Gould, 2008). However, the functional requirement for the Golgi-specific Syntaxin 5 and the COG5 ortholog fws (Farkas et al., 2003; Xu et al., 2002), as well as the sensitivity to BFA, imply that cytokinesis in spermatocytes depends on membrane trafficking through the Golgi. Thus, the genetic interaction between bru and Rab11, taken together with the localization of Bru to the Golgi, would be in agreement with the possibility that bru functions in early intra-Golgi traffic whereas Rab11 promotes the trafficking of Golgi-derived membrane to the cell surface.

bru also genetically interacts with fwd, a PI4Kβ-encoding gene that is also required in spermatocytes for contractile ring constriction and the accumulation of Rab11 at polar puncta and the cleavage furrow (Brill et al., 2000; Giansanti et al., 2004; Giansanti et al., 2007). Here, we have shown that that bru;fwd double mutants display a cytokinesis phenotype that is much stronger than either single mutant. Interestingly, genetic and biochemical interactions have been reported between Rab11 and PI4Kβ, or their orthologs, in several systems (Meyers and Cantley, 1997; de Graaf et al., 2004; Sciorra et al., 2004; Giansanti et al., 2007). In mammalian cells, the efficiency of Golgi to plasma membrane transport correlates with the level of PI4Kβ-mediated recruitment of GTP-bound Rab11 to the Golgi (de Graaf et al., 2004). Previously, we have shown that fwd and Rab11 are in the same epistasis group and therefore function in the same pathway in the context of spermatocyte cytokinesis (Giansanti, 2007). That bru genetically interacts with both fwd and Rab11 is consistent with this finding. Thus, we propose that bru cooperates with a parallel fwd-Rab11 pathway in supporting membrane addition to the cleavage furrow.

Although TRS120 is an essential gene in yeast, we do not know whether bru is essential in Drosophila. Because the two alleles of bru were recovered from a collection of homozygous-viable, male-sterile mutants (Giansanti et al., 2004; Wakimoto et al., 2004), lethal alleles of bru could not have been found in our screen. The bru mutations we characterized each introduce a premature stop codon that would remove the C-terminal half of the encoded polypeptide. However, if the mutant mRNAs were translated, the regions of highest amino acid sequence conservation in the Bru N-terminal half would be retained. In this case, the temperature sensitivity of the bru allele might reflect the degree to which the truncated Bru protein physically associates with TRAPPII, with association being more stable at 18°C. In this regard, a number of yeast TRS120 mutations that cause loss of the Trs120p C-terminus are viable and exhibit temperature-sensitive phenotypes in membrane trafficking (Cai et al., 2005). It is also possible that the process of membrane trafficking in Drosophila spermatocytes is itself sensitive to temperature, such that higher temperatures generally exacerbate any defects in the membrane trafficking machinery.

We have shown that bru mutant spermatocytes form a regular contractile ring, but that the ring fails to constrict properly; this phenotype is shared with mutations in other membrane trafficking genes such as fwd, fws, gio and Rab11 (Brill et al., 2000; Farkas et al., 2003; Gatt and Glover, 2006; Giansanti et al., 2004; Giansanti et al., 2006; Giansanti et al., 2007). In other cell types, the lack of membrane-trafficking components can interfere with the initial assembly of the contractile ring (Somma et al., 2002; Niswonger et al., 1997). Although the mechanisms underlying the interplay between ring behavior and membrane trafficking are currently unclear, there are two non-mutually exclusive possibilities. Lack of membrane addition to the advancing cleavage furrow might result in plasma membrane tension that counteracts the force of actomyosin ring constriction (Gatt and Glover, 2006; Giansanti et al., 2007). Alternatively, there might be vesicle-associated contractile ring components and actin-remodeling factors that fail to be transported to the cell midzone as observed for actin-vesicle transport in cellularizing Drosophila embryos (Giansanti et al., 2007; Albertson et al., 2008). Either of these possibilities might account for the defect in ring constriction and furrow ingression that we observe in dividing spermatocytes of the bru mutant.

We have not observed any phenotypes for bru mutants outside the male germline. However, the synthetic lethality resulting from simultaneous loss of both bru and Rab11 wild-type functions, in conjunction with the expression of bru message in females and embryos, suggests that bru serves an important role in membrane trafficking homeostasis in other cell types. Thus, the specific requirement for a TRAPPII subunit in spermatocyte cleavage furrow ingression raises the possibility that dividing spermatocytes have an elevated requirement for fast and efficient vesicle transport during cytokinesis. This notion is consistent with the low surface-area-to-volume ratio of these cells and the challenge of producing four daughter cells via two meiotic divisions within a period of 194 minutes (Cenci et al., 1994; Dyer et al., 2007; Tates, 1971). The observation that spermatocytes lack membrane protrusions and microvilli (Tates, 1971) suggests that existing plasma membrane cannot be reorganized to provide the additional surface are required for cleavage furrow ingression. Thus, the success of male gamete formation in Drosophila might rely on a level of membrane-trafficking efficiency that is not essential in other cell types of this organism.

Viable, male-sterile alleles bru^Z3358, bru^Z0704 and pbl^Z4836 are described (Giansanti et al., 2004). fwd^Z0453, Rab11 alleles and Rab11-GFP have been described (Giansanti et al., 2007). The Spaghetti squash (Sqh)-GFP transgene was used to express RLC-GFP (Royou et al., 2002). Balancer chromosomes are described (Lindsley and Zimm, 1992). EYFP-Golgi (LaJeunesse et al., 2004), pbl⁵ (Lehner, 1992) and chromosomal deficiencies Df(2L)pr2b, Df(3R)pbl-NR and Df(3L)7C (Giansanti et al., 2004) were from the Bloomington Drosophila Stock Center at Indiana University.

Squashed adult or larval testes were imaged by phase-contrast on a Zeiss Axioskop or by fluorescence on a Zeiss Axiophot and processed as described (Farkas et al., 2003). For multinucleate spermatid counts, live testes were lightly squashed and imaged at multiple focal planes through the sample to prevent disruption of the large multinucleate cells while counting the number of nuclei. For RLC-GFP counts during meiosis I, live testes were squashed to flatten the cells so that the rings could be more easily viewed. Immunofluorescence staining of fixed testes used rabbit anti-Lava lamp (a gift from John Sisson, University of Texas, Austin, TX), anti-Rab11 and anti-Centrosomin (a gift from Tim Megraw, Florida State University, Tallahassee, FL) (Giansanti et al., 2007); rabbit anti-GFP (a gift from Gianluca Cestra, Università di Roma `La Sapienza', Rome, Italy) and mouse anti-p120 (Calbiochem, anti-Golgi, Drosophila, 345867) were used at 1:100 to stain testes fixed as per (Giansanti et al., 2007). Live unsquashed spermatocyte cysts in Schneider's Drosophila Medium (Gibco) were imaged on a Leica SP2 AOBS confocal laser-scanning microscope. Serial excitation of fluorophores prevented signal bleed-through. EGFP was excited at 488 nm and 489-512 nm emissions collected. EFYP was excited at 514 nm and 520-576 nm emissions collected. All spermatocytes are primary spermatocytes in stages of meiosis I.

Time-lapse imaging of RLC-GFP (Royou et al., 2002) or Rab11-GFP (Dollar et al., 2002) in live primary spermatocytes was carried out as described (Giansanti et al., 2007). Cells were examined with a Zeiss Axiovert 20 microscope equipped with a 63× (for RLC-GFP) or 100× objective (for Rab11-GFP) and a filter wheel combination (Chroma). Images were acquired with a CoolSnap HQ camera (Photometrics) controlled through a Metamorph software package (Universal Imaging). Images were collected at 1- or 1.5-minute intervals, and 10 fluorescent optical sections were captured at 1-μm z steps. Movies were created using the Metamorph software; each fluorescent image shown in Figs 1 and 6 is the maximum-intensity projection of all the sections. Wild-type spermatocytes contained no mutant alleles.

bru was mapped to 38B5-E1 (Giansanti et al., 2004), and additional deficiency complementation tests placed the locus between the proximal breakpoints of Df(2L)pr-A16 and Df(2L)pr2b (Lindsley and Zimm, 1992). Single-embryo PCR analysis of homozygous deficiency embryos (protocol from C.C.R. available upon request) delimited the bru interval to 24 kbp. Annotated protein-coding genes in the interval were sequenced on the mutant chromosomes and compared with the wild-type Zuker background in which the mutations were made. Each bru mutant allele revealed a mutation in annotated gene CG2478. A 5.8-kbp HindIII genomic fragment containing the CG2478 coding sequence, 5′ UTR and 700 bp of 5′ DNA that included the 500 bp of the upstream gene CG2614, but no other known genes, was subcloned from the Drosophila P1 clone DS02109 and fused to a PCR product containing an additional 100 bp of the 3′ untranslated region of CG2478. This 5.9-kbp fragment was transferred into pCaSpeR4, and the resulting pCaSpeR4-bru rescue construct was used to generate transgenic flies by standard methods as per (Farkas et al., 2003). The transgene was then crossed into the bru mutant background to test for phenotypic rescue of male meiotic cytokinesis.

For bru-GFP, the EGFP coding sequence (Clontech) was inserted immediately before the stop codon of the bru rescue construct. For dBet3-YFP, the EYFP coding sequence (Clontech) was inserted immediately before the stop codon of the cDNA RE68712, which corresponds to the message from CG3911, and flanked with 2 kbp of 5′ and 3′ CG3911 genomic sequences from BAC RPC1-98 E20. For mutagenesis of the Bru motif I, a 475-bp NheI fragment was subcloned from the bru rescue fragment, and six selected codons were sequentially mutated (XL site-directed mutagenesis kit, Stratagene) to encode alanines. The mutated fragment was sequence-verified and subcloned back into pCaSpeR4-bru to generate transgenic flies.

White pre-pupae larval testes were cultured in 25 μl Schneider's Drosophila medium with penicillin/streptomycin and 30 μg/ml of BFA (Sigma) from a 10 μg/ml stock in ethanol or an equal volume of ethanol on open slides in a humidified chamber at room temperature. Observation of multinucleate spermatids in wild-type Oregon R testes required incubation in BFA for 14 hours, consistent with the observation that one cyst of spermatocytes would be expected to execute meiotic cytokinesis every 10 hours (Lindsley and Tokuyasu, 1980).