QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

Vol. 18, Issue 12, 5034-5047, December 2007

This Article Abstract Full Text (PDF) Supplemental Materials All Versions of this Article: E07-05-0415v1 18/12/5034    most recent Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Giansanti, M. G. Articles by Gatti, M. Search for Related Content PubMed PubMed Citation Articles by Giansanti, M. G. Articles by Gatti, M.

Rab11 Is Required for Membrane Trafficking and Actomyosin Ring Constriction in Meiotic Cytokinesis of Drosophila Males

Istituto Pasteur-Fondazione Cenci Bolognetti and Istituto di Biologia e Patologia Molecolari del Consiglio Nazionale delle Ricerche, Dipartimento di Genetica e Biologia Molecolare, Università di Roma "La Sapienza," 00185 Rome, Italy

Submitted May 7, 2007; Revised September 19, 2007; Accepted September 25, 2007
Monitoring Editor: Fred Chang

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Rab11 is a small GTPase that regulates several aspects of vesicular trafficking. Here, we show that Rab11 accumulates at the cleavage furrow of Drosophila spermatocytes and that it is essential for cytokinesis. Mutant spermatocytes form regular actomyosin rings, but these rings fail to constrict to completion, leading to cytokinesis failures. rab11 spermatocytes also exhibit an abnormal accumulation of Golgi-derived vesicles at the telophase equator, suggesting a defect in membrane–vesicle fusion. These cytokinesis phenotypes are identical to those elicited by mutations in giotto (gio) and four wheel drive (fwd) that encode a phosphatidylinositol transfer protein and a phosphatidylinositol 4-kinase, respectively. Double mutant analysis and immunostaining for Gio and Rab11 indicated that gio, fwd, and rab11 function in the same cytokinetic pathway, with Gio and Fwd acting upstream of Rab11. We propose that Gio and Fwd mediate Rab11 recruitment at the cleavage furrow and that Rab11 facilitates targeted membrane delivery to the advancing furrow.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Cytokinesis is the process by which two daughter cells separate at the end of mitosis or meiosis. In animal cells, cytokinesis is mediated by an actomyosin-based contractile ring that assembles just beneath the equatorial cortex of the dividing cell. Ring constriction leads to the formation of a furrow in the plasma membrane, which invaginates until the two daughter cells remain connected by a thin cytoplasmic bridge, called the midbody. This bridge is ultimately cleaved during the final step of cytokinesis, named abscission, which results in the complete separation of daughter cells (Glotzer, 2001; Schweitzer and D'Souza-Schorey, 2004). Recent work has shown that both cleavage furrow ingression and abscission require substantial membrane remodeling. Studies in a variety of organisms indicate that cleavage furrow invagination is accompanied by targeted membrane addition from internal membrane stores (Strickland and Burgess, 2004; Albertson et al., 2005; Matheson et al., 2005). There is also evidence that formation of new furrow membrane requires vesicle delivery through both the secretory and the endocytic pathways. The secretory pathway involves vesicle transport from the endoplasmic reticulum (ER) to the Golgi and then to the plasma membrane. In the endocytic pathway, plasma membrane-derived vesicles proceed through the early endosome and the recycling endosome (RE), which directs them back to the plasma membrane (Strickland and Burgess, 2004; Albertson et al., 2005).

In Drosophila, Golgi-based vesicle delivery is crucial for both cytokinesis and cellularization of syncytial embryonic nuclei, a process with features equivalent to those that occur during cleavage furrow ingression (Albertson et al., 2005). The Golgi-associated proteins required for cellularization include the Lava Lamp (Lva) golgin (Sisson et al., 2000) and the integral membrane protein called Strabismus (Lee et al., 2003). In vivo studies have shown that Lva is enriched at vesicles that move toward the apex of the advancing cellularization furrow, suggesting that they are a source of new membrane for furrow progression (Sisson et al., 2000). Golgi-associated proteins have been also implicated in cytokinesis of Drosophila spermatocytes. Four way stop (Fws), a protein homologous to the Cog5 subunit of the conserved oligomeric Golgi complex, is enriched at the Golgi stacks and Golgi-derived vesicles and it is required for cytokinesis (Farkas et al., 2003). Another protein required for meiotic cytokinesis of males is the Drosophila orthologue of Syntaxin 5, a conserved Golgi-associated soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) (Xu et al., 2002). Golgi-associated proteins involved in cytokinesis have been also identified in mammals and Caenorhabditis elegans. A proteomic analysis of purified mammalian midbodies revealed that approximately a quarter of the midbody components are Golgi-derived proteins; RNA interference (RNAi) experiments showed that the homologues of many of these proteins are required for C. elegans cytokinesis (Skop et al., 2004).

Successful cytokinesis also relies on the endocytic pathway. Cleavage furrow-specific endocytosis has been observed in zebra fish embryos from early to late stages of cytokinesis (Feng et al., 2002). In HeLa cells, endocytic vesicles internalized from the polar regions during anaphase are subsequently trafficked to the midbody where they are thought to contribute to completion of cytokinesis (Schweitzer et al., 2005). Clathrin and dynamin, two proteins that promote endocytic vesicle budding from the plasma membrane have been implicated in cytokinesis in several systems, including Dictyostelium discoideum, C. elegans, zebra fish, and mammalian cells (Niswonger and O'Halloran, 1997; Gerald et al., 2001; Thompson et al., 2002; Schweitzer et al., 2005). There is also evidence that Drosophila dynamin encoded by the shibire gene is required for cellularization (Swanson and Poodry, 1980; Pelissier et al., 2003). Furthermore, recent studies have shown that Rab35, which is enriched at endocytic clathrin-coated pits and vesicles, is essential for cytokinesis in both Drosophila and human cells (Kouranti et al., 2006). Another protein involved in the endocytic pathway and required for cytokinesis is Rab11, a small GTPase that regulates membrane trafficking through the RE (Ullrich et al., 1996; Matheson et al., 2005). An involvement of Rab11 in cytokinesis was first demonstrated in C. elegans (Skop et al., 2001). Subsequent studies showed that Rab11 and its interacting partner Nuclear-fallout (Nuf) are both required for cellularization of Drosophila embryos (Pelissier et al., 2003; Riggs et al., 2003). In mammalian cells, Rab11 and its binding partners FIP3/Arf1 and FIP4/Arf2, which share homology with Nuf, mediate the delivery of endosomes to the cleavage furrow and they are essential for completion of cytokinesis (Fielding et al., 2005; Wilson et al., 2005). These results suggest that Rab11 and Nuf/FIP3/FIP4 work in concert to play a conserved function required for cytokinesis.

Targeted vesicle fusion at the advancing cytokinetic furrow is likely to be regulated by the unique lipid composition of the cleavage furrow membrane. The outer leaflet of the plasma membrane at the equator of dividing mammalian cells is enriched in phosphatidylethanolamine (Emoto and Umeda, 2000). In sea urchin embryos, the equatorial membrane domain is enriched in ganglioside G_M1 and cholesterol (Ng et al., 2005). Sterol-rich membrane domains essential for cytokinesis have been also found at the Schizosaccharomyces pombe cleavage site (Wachtler et al., 2003; Takeda et al., 2004). Another relevant component of the furrow membrane is phosphatidylinositol 4,5 biphosphate [PtdIns(4,5)P2], which is synthesized from phosphatidylinositol (PtdIns) molecules through sequential phosphorylation events mediated by PtdIns-4-kinases (Fwd in Drosophila) and PtdIns(4)P-5-kinases. Recent work has shown that at least in some species PtdIns(4,5)P2 and the kinases involved in its generation are enriched at the furrow membrane and required for cytokinesis (Emoto et al., 2005; Field et al., 2005; Wong et al., 2005; for review, see Logan and Mandato, 2006).

Here, we have analyzed the role of Rab11 during cytokinesis of Drosophila spermatocytes. We show that Rab11 is enriched at the equator of ana-telophase cells and required for cleavage furrow ingression. Our results strongly suggest that Rab11 mediates membrane addition at the advancing furrow and that this process is essential for actin ring constriction. In addition, we show that Rab11 functions in a common pathway with Gio (phosphatidylinositol transfer protein; PITP) and Fwd (PtdIns 4-kinase) to control formation of new membrane during Drosophila cytokinesis.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Fly Strains and Genetics
The P element insertion rab11^EP3017 and the ethyl methanesulfonate (EMS)-induced rab11^93Bi mutant alleles were obtained from the Bloomington Stock Center (Indiana University, Bloomington, IN). We sequenced the rab11^93Bi mutant allele; it carries an ArgTrp missense mutation at amino acid 104, a conserved residue between species. The EMS-induced rab11^E(To)11 and rab11^E(To)3 mutant alleles, and Df(3R)e-R1 that removes rab11⁺, were kindly provided by M. Erdélyi (Hungarian Academy of Sciences Szeged, Hungary; Jankovics et al., 2001). The strain expressing Rab11-green fluorescent protein (GFP) was a gift of R. S. Cohen (University of Kansas, Lawrence; Dollar et al., 2002). The flies expressing β-tubulin-enhanced GFP (EGFP) were kindly provided by M. Savoian and D. Glover (University of Cambridge, UK; Inoue et al., 2004). The gio^EP513, fwd³, and fwd^Z0453 mutant alleles have been described previously (Brill et al., 2000; Giansanti et al., 2004, 2006). Previous studies have shown that the fwd³ allele carries an early nonsense mutation, resulting in a truncated protein of 309 amino acids; the wild-type Fwd protein contain 1338 amino acids (Brill et al., 2000). We have sequenced the fwd^Z0453 mutant allele; it carries a SerLeu missense mutation at amino acid 1318, a conserved residue of the kinase catalytic domain (our unpublished results). In addition, we compared fwd^Z0453/fwd^Z0453, fwd³/fwd^Z0453, and fwd^Z0453/Df(3L)7C males for the frequencies of aberrant spermatids, and we found no significant differences. Thus, both the fwd^Z0453 allele and the fwd³/fwd^Z0453 heteroalleic combination are genetically null. Double mutants were generated by recombination by using standard methods. All mutations were maintained over the TM6B third chromosome balancer, and mutant larvae were identified based on their non-Tubby phenotype.

The strain expressing GFP-protein disulfide isomerase (Pdi) (P{PTT-GA}Pdi^G00198) was obtained from the Bloomington Stock Center. The flies of this strain contain a GFP sequence inserted into the single intron of the protein disulfide isomerase-coding gene (CG6988), and they express a GFP-Pdi chimera that can be detected by fluorescence microscopy (Bobinnec et al., 2003). Genetic markers and special chromosomes are described in FlyBase (http://www.flybase.org/).

Immunostaining and Microscopy
Spermatid morphology in live material was examined in testes from third instars as described by Giansanti et al. (2004). Fixed cytological preparations were made with third instar larvae testes, which were dissected in testis buffer (183 mM KCl, 47 mM NaCl, and 10 mM Tris-HCl, pH 6.8), gently squashed in the same buffer, and frozen in liquid nitrogen. After removal of the coverslip, preparations were fixed with either of the following procedures. For visualization of the Rab11-GFP fluorescence, simultaneous detection of Rab11-GFP and Lva or DSas-4 immunostaining, and actin and tubulin double staining, testes were fixed with methanol-free formaldehyde (Polysciences, Warrington, PA) as described by Gunsalus et al., (1995). However, this fixation procedure does not work well for immunostaining with anti-Ra11 antibodies. Thus, for all the other immunostaining procedures with multiple antibodies, testes were fixed with 3.7% formaldehyde (containing 10% methanol) in 1x phosphate-buffered saline (PBS) and then squashed in 60% acetic acid according to Giansanti et al. (1999). This procedure, which does not preserve GFP fluorescence, is henceforth referred to as methanol/formaldehyde fixation. After this type of fixation, preparations were generally incubated with PBT (1x PBS containing 0.1% Triton-X 100) for 30 min. However, in double immunostaining experiments for Rab11 and anillin, Rab11 and Nuf, Rab11 and Gio and for staining with fluorescent wheat germ agglutinin (WGA), preparations were treated with PBT for a maximum of 5 min. The short incubation in PBT helps to visualize the plasma membrane-associated proteins but not the structures in the interior of the cell, such as the Golgi-derived fragments.

For immunostaining with rabbit and rat primary antibodies, testis preparations were incubated overnight at 4°C with the antibodies diluted in 1x PBS. For tubulin immunostaining, testis preparations were incubated for 1 h at room temperature with an anti- tubulin monoclonal (Sigma-Aldrich, St. Louis, MO) diluted in 1x PBS. The following dilutions were used: anti- tubulin (Sigma-Aldrich), 1:300; anti-anillin (gift from C. Field, Harvard University, Boston, MA; Field and Alberts, 1995), 1:200; anti-Lva (gift from J. Sisson, University of Texas, Austin; Sisson et al., 2000), 1:500; anti-Giotto (Giansanti et al., 2006), 1:2000; anti-Nuf (gift from W. Sullivan, University of California, Santa Cruz; Rothwell et al., 1998), 1:200; rabbit anti-Cnn (gift from T. Megraw, University of Texas, Dallas; Megraw et al., 2001), 1:300; rabbit anti-DSas-4 (gift from R. Basto and J. Raff, Gurdon Institute, Cambridge, UK; Basto et al., 2006), 1:300; rabbit anti-Rab11 (gift from M. Gaitan, Max-Planck Institute, Dresden, Germany; Emery et al., 2005), 1:25; and rat anti-Rab11 (gift from R. S. Cohen, University of Kansas, Lawrence; Dollar et al., 2002), 1:300. Primary antibodies were detected by incubation for 1 h at room temperature with secondary antibodies obtained by Jackson ImmunoResearch Laboratories (West Grove, PA), diluted according to the supplier's instructions. F-actin was stained with rhodamine-labeled phalloidin (Invitrogen, Carlsbad, CA) diluted 1:2 in 1x PBS (Gunsalus et al., 1995). For staining with fluorescein-conjugated WGA (Invitrogen) spermatocytes immunostained with anti-Rab11 antibodies were incubated for 2 h at room temperature with fluorescent WGA diluted 1:10 in PBS. In all cases, slides were mounted in Vectashield medium H-1200 with 4,6 diamidino-2-phenylindole (DAPI) (Vector Laboratories, Burlingame, CA) to stain DNA and reduce fluorescence fading. It should be noted that the rat and rabbit anti-Rab11 antibodies produced identical staining patterns in Drosophila testes (data not shown). Thus, in double immunofluorescence experiments, the choice of the anti-Rab11 antibody depended on the other primary antibody used for immunostaining.

Images were captured using a CoolSnap HQ charge-coupled device camera (Photometrics; Tucson, AZ) connected to a Zeiss Axioplan fluorescence microscope equipped with an HBO 100-W mercury lamp as described previously (Giansanti et al., 2004, 2006). Gray scale digital images were collected separately, converted to Photoshop format, pseudocolored, and merged.

Time-Lapse Imaging
Time-lapse imaging of living spermatocytes was carried out according to the protocol described by Inoue et al. (2004). Testes isolated from third instars were dissected under 10S voltalef oil (Elf Atochem North America, Inc., Philadelphia, PA) onto a clean coverslip attached to the underside of an aluminum slide. In most cases (except Supplemental Movie S4 generated using confocal microscopy), cells were examined with a Zeiss Axiovert 20 microscope equipped with a 63x objective and a filter wheel combination (Chroma Technology, Brattleboro, VT). Images were acquired with a CoolSnap HQ camera (Photometrics) by using a 2 x 2 bin. Image acquisition was controlled through a MetaMorph software package (Molecular Devices, Sunnyvale, CA). Images were collected at one-minute intervals; 10 (for β-tubulin-EGFP), or 14 (for Rab11-GFP) fluorescence optical sections were captured at 1-µm z-steps. Movies were created using the MetaMorph software; each fluorescent image shown is the maximum-intensity projection of all the sections.

Confocal time lapse imaging (Supplemental Movie S4 and Figure 2A) was carried out using a BD CarvII spinning disk confocal system (BD Biosciences, Rockville, MD) attached to a Nikon Eclipse TE 2000 S inverted microscope, using a 63x lens and 2 x 2 bin. The movie was created with the MetaMorph software using maximum-intensity projections of z-series recorded at 1-min intervals.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
The Dynamic Behavior of Rab11 during Drosophila Male Meiosis
To determine Rab11 behavior during male meiosis, we examined spermatocytes expressing Rab11-GFP. These cells were obtained from flies bearing one copy of a rab11-GFP transgene with an endogenous rab11 promoter (Dollar et al., 2002) and two copies of the wild-type rab11 gene. This rab11-GFP transgene rescues the lethality and the meiotic phenotype associated with rab11 mutations (see below), indicating that the Rab11-GFP protein is functional.

To avoid possible GFP photobleaching problems, we imaged spermatocytes from late prophase to mid-anaphase and from mid-anaphase to the end of cell division (Figure 1, A–C, and Supplemental Movies S1–S3). Time-lapse imaging revealed a highly dynamic behavior of the protein. During late prophase and prometaphase I, Rab11 is in part diffuse in the cytoplasm and in part concentrated at both the Golgi stacks and the nuclear envelope (Figure 1A; see below for Golgi stacks identification). At prometaphase/metaphase I, the cytoplasmic pool of Rab11 and the Golgi stacks progressively concentrate at the cell poles. When the Golgi stacks disassemble at metaphase (Farkas et al., 2003; Giansanti et al., 2006), the resulting Rab11-containing breakdown products accumulate at the cell poles, which remain enriched in Rab11 throughout anaphase and telophase (Figure 1, A–C, and Supplemental Movies S1–S3). At early telophase I, Rab11 starts to concentrate at the cleavage furrow (Figure 1, B and C, and Supplemental Movies S2 and S3). This process does not seem to be mediated by vesicle fusion events at the furrow membrane, because we do not see evident Rab11-enriched vesicles moving toward the cell equator at this stage of meiotic division. However, a clear movement of Rab11-containing vesicles does occur in mid/late telophase (Figure 1C and Supplemental Movie S3). We followed the movement of 50 of these vesicles; 12% of them displayed a poleward or an oscillatory movement, whereas the remaining 88% traveled from the poles to the cell equator. We speculate that the latter vesicles are transported along microtubule tracks to the cleavage site, where they fuse with the furrow membrane, resulting in both membrane addition and further local enrichment in Rab11.

View larger version (132K): [in this window] [in a new window]   Figure 1. Rab11 dynamic behavior during the first meiotic division of Drosophila spermatocytes. (A–C) Selected frames from three time lapses of primary spermatocytes expressing Rab11-GFP. The numbers at the bottom of each frame indicate the minutes elapsed from the beginning of imaging. (A) Frames from Supplemental Movie S1 showing a spermatocyte undergoing prophase (0), prometaphase (12, 18) metaphase (24), and anaphase (32). Note the Rab11 concentration at the Golgi stacks and at the nuclear envelope from prophase to metaphase, and its subsequent accumulation at the cell poles and the spindle envelope. (B) Frames from Supplemental Movie S2 showing a spermatocyte from early telophase (0) through mid- (6, 9) and late telophase (17); Rab11 starts to accumulate to the cleavage furrow during mid-telophase (6, arrowhead). (C) Frames from Supplemental Movie S3 showing a spermatocyte undergoing telophase. At mid-telophase, Rab11 starts to concentrate at the cell equator (2). During mid/late telophase Rab11-containing vesicles (arrow) travel from the poles to the middle of the cell and seem to reach the Rab11-enriched equatorial band (7–9). (D) Prophase (Pro), metaphase (Meta), anaphase (Ana) mid-telophase (Midtelo), and late telophase (Late telo) figures from primary spermatocytes expressing GFP-Pdi. Note that Pdi localizes to the cell poles and the spindle envelope like Rab11. However, Pdi does not concentrate at the Golgi stacks of prophase spermatocytes and at the equatorial region of telophase figures. Bars, 10 µm. (E–G) Diagrams of primary spermatocytes undergoing metaphase (E), early telophase (F), and late telophase (G) based on electron microscopy studies by A. D. Tates (Tates, 1971; also see Fuller, 1993) (printed with permission of the author). The diagrams show an entire metaphase cell (E) and only one of the two presumptive daughter cells that comprise a telophase figure (F and G). Major structures include the centrioles (C), astral membranes (AM), parafusorial membranes (P), and mitochondria (M). I indicates the cleavage site. Note the absence of parafusorial membranes in late telophase cells.

View larger version (82K): [in this window] [in a new window]   Figure 2. Rab11 localization in ana-telophase primary spermatocytes. (A) Selected frames from confocal images (Supplemental Movie S4) of primary spermatocytes expressing Rab11-GFP. The numbers at the bottom of each frame indicate the minutes elapsed from the beginning of imaging. Note that Rab11-GFP accumulates at both the spindle envelope (in all cells) and the equatorial plasma membrane (in 2 cells; arrows). (B) Oregon-R (wild-type control) and Rab11-GFP–expressing spermatocytes fixed with methanol/formaldehyde and immunostained for Rab11, anillin (red), and centrosomin (red). (C) An Oregon-R spermatocyte fixed with methanol/formaldehyde and stained with anti-Rab11 antibodies and fluorescent WGA (red). Note that Rab11 colocalizes with both WGA and anillin at the cell equator. (D–F) spermatocytes expressing Rab11-GFP (D and E) and Oregon-R spermatocytes (F) stained for tubulin (green) and DNA (blue). Rab11-GFP–expressing spermatocytes were either fixed with methanol-free formaldehyde (D) and examined for GFP fluorescence (red), or they were fixed with methanol/formaldehyde (E) and immunostained for Rab11 (red). (F) Oregon-R spermatocytes fixed with methanol/formaldehyde and immunostained for Rab11 (red). Note that the Rab11-GFP signals are fully comparable to the signals produced by Rab11 immunostaining. Bars, 10 µm.

To demonstrate that Rab11 colocalizes with the furrow membrane, we stained dividing spermatocytes with anti-Rab11 antibodies and either anti-anillin antibodies or fluorescent WGA. Anillin is a well-known contractile ring component that binds both actin and myosin, and interacts with the plasma membrane (Field and Alberts, 1995; Field et al., 2005a); WGA is a good marker for equatorial plasma membrane (Ng et al., 2005). We found that both anillin and WGA colocalize with Rab11 at the equator of telophases (Figure 2, B and C), supporting the conclusion that Rab11 is enriched at the furrow membrane. In summary, Rab11 is concentrated in the Golgi stacks and the nuclear envelope during prophase/prometaphase, it becomes enriched at the ER compartment and at some, but not all (see below), Golgi-derived vesicles during metaphase and ana-telophase, and accumulates at the cleavage furrow of telophase cells.

Our initial experiments with anti-Rab11 antibodies showed that fixation with methanol-free formaldehyde, which preserves GFP fluorescence, does not work well for Rab11 immunostaining. Conversely, the methanol/formaldehyde fixation, which allows efficient immunostaining with anti-Rab11 antibodies, disrupts GFP fluorescence. Thus, we wondered whether antibody staining and direct observation of Rab11-GFP result in identical localization patterns. To address this question, we examined the staining patterns of Rab11-GFP–expressing spermatocytes that were either fixed with methanol-free formaldehyde and scored for GFP fluorescence, or fixed with methanol/formaldehyde and immunostained for Rab11. These patterns were fully comparable and very similar to the staining pattern of wild-type spermatocytes immunostained with anti-Rab11 antibodies after methanol/formaldehyde fixation (Figure 2, D–F).

We next examined the Rab11 localization pattern in fixed spermatocytes and compared it with that of Lva, a Golgi marker that shares homology with mammalian golgins (Sisson et al., 2000). Testes expressing Rab11-GFP were fixed with methanol-free formaldehyde, and then they were immunostained for Lva. As shown in Figure 3A, Lva and Rab11 are both enriched at the Golgi stacks of prophase primary spermatocytes. However, although Rab11 and Lva largely colocalize, their staining patterns do not completely overlap, suggesting a differential accumulation of the proteins within subcompartments of the Golgi apparatus (Figure 3A). Rab11-GFP localization in fixed metaphase and ana-telophase figures does not coincide with Lva distribution. In metaphase and early anaphase, Rab11 is concentrated at the polar regions of the cell and at the spindle envelope, whereas most Lva-enriched vesicles are just outside these regions (Figure 3B). In telophase figures, Rab11 is still concentrated at the cell poles, but it is also enriched at vesicles that are distant from the poles. Cytological analysis of ana-telophases revealed three types of Golgi breakdown products: those that contain either Rab11 or Lva and those that contain both proteins. In addition, this analysis showed that Rab11 is highly enriched at cleavage furrow, whereas Lva is excluded from this site (Figure 3C; data not shown).

View larger version (63K): [in this window] [in a new window]   Figure 3. Rab11-GFP localization in fixed Drosophila spermatocytes stained for either Lva or DSas-4. (A–C) Primary spermatocytes expressing Rab11-GFP stained for Lva. In prophase cells (A), Rab11 (green) largely colocalizes with Lva (red) in the Golgi stacks; the dark areas correspond to the nuclei. In metaphase (B) and telophase (C) cells, Rab11 and Lva colocalize only in a fraction of the Golgi-derived vesicles. (D) An early prophase primary spermatocyte expressing Rab11-GFP (green) immunostained for DSas-4 (red; arrow). Note that Rab11 does not accumulate around the centrioles. Bars, 10 µm.

To further characterize the subcellular distribution of Rab11, fixed spermatocytes were immunostained for Rab11 and either Nuf or Giotto. Rab11 and Nuf colocalized in the Golgi stacks of prophase and prometaphase spermatocytes (data not shown) and showed very similar distributions in cells undergoing metaphase and ana-telophase (Figure 4, A and B; data not shown), consistent with the idea that Rab11 and Nuf function as a complex (Riggs et al., 2003). Gio and Rab11 displayed comparable subcellular localizations in ana-telophase spermatocytes (Figure 4, C and D), consistent with our previous observations on Gio distribution (Giansanti et al., 2006).

View larger version (51K): [in this window] [in a new window]   Figure 4. Localization Rab11, Nuf and Gio in Drosophila primary spermatocytes. (A–D) Wild-type primary spermatocytes in early anaphase (A and C) and telophase (B and D) stained for Rab11 (green) and either Nuf (red) or Gio (red). Note that Rab11 colocalizes with both Nuf and Gio at the cell equator. Bar, 10 µm.

View larger version (36K): [in this window] [in a new window]   Figure 5. rab11 mutations disrupt spermatocyte cytokinesis. (A) Complementation analysis among rab11 mutant alleles. V, viable; SL, semilethal; EL, early lethal; and DS, defective spermatids. (B–E) Abnormal spermatids observed in living testes of rab11 mutant. (B) Wild-type spermatids with nuclei (n) and nebenkern (nk) of similar sizes. (C–E) multinucleated spermatids from rab11 males containing two (C), three (D), or four nuclei (E) of similar sizes associated with a single large nebenkern. Bar, 10 µm. (F) Frequencies (% ± SE) of spermatids with 2 (2nu, gray), 3 (3nu, black), and 4 (4nu, dark gray) nuclei observed in rab11 mutant males. Mutant males also exhibit very low frequencies of multinucleated spermatids with differently sized nuclei (ir, empty columns). In wild type, the frequency of abnormal spermatids is virtually zero.

View larger version (76K): [in this window] [in a new window]   Figure 6. rab11 mutations affect ring constriction (A–D) Primary spermatocyte telophases stained for tubulin (green), actin (red) and DNA (blue). (A) Wild-type late telophase with a constricted actin ring. (B–D) Mutant late telophases displaying an incompletely constricted ring (B), a broken ring (C), or a fully constricted but abnormally thick ring (D). (E and F) Primary spermatocyte late telophases stained for tubulin (green), anillin (red), and DNA (blue). (E) Wild-type telophase with a fully constricted ring. (F) Mutant telophase with an incompletely constricted ring and a defective central spindle. Bars, 10 µm.

View larger version (75K): [in this window] [in a new window]   Figure 7. Spindle formation and behavior during the first meiotic division in rab1193Bi/rab11E(To)11 males expressing β-tubulin-EGFP. Time, in minutes, is relative to the initiation of cleavage furrow ingression (0). In the wild-type cell (A; Supplemental Movie S5) the central spindle midzone reaches a maximum compaction at 30 min (arrow). In the mutant cell (B; Supplemental Movie S6) that undergoes a successful cytokinesis the maximum compaction of the central spindle midzone (arrow) is the same as in the wild-type telophases and occurs at 33 min. In the mutant cell (C; Supplemental Movie S7), furrowing halts at 15 min; at this time, the central spindle midzone still exhibits a regular compaction, but it decreases in microtubule density during furrow regression; at 39 min, the central spindle has completely disappeared. Bar, 10 µm.

View larger version (82K): [in this window] [in a new window]   Figure 8. Mutations in rab11 affect Golgi-derived vesicle behavior during primary spermatocyte telophase. Wild-type (A, C, and E) and rab11 mutant (B, D, and F) spermatocytes undergoing prophase (A and B), anaphase (C and D), and late telophase (E and F) were stained for Lva (red), tubulin (green), and DNA (blue). Note that rab11 mutations only affect vesicle localization in telophase cells; vesicles are excluded from the cell equator in wild-type spermatocytes (E) but they accumulate at the center of rab11 mutant cells (F). (G and H) Wild-type (G) and rab1193Bi/rab11E(To)11 mutant (H) telophases immunostained with anti-Rab11 and anti-Lva (red) antibodies. Note that the mutant protein produced by the rab1193Bi missense mutation is detected by anti-Rab11 antibodies, and it is enriched at structures that are partially coincident with the Lva-positive vesicles accumulated at the cell equator. Bars, 10 µm.

We also examined larval brain squash preparations from rab11 mutants for the presence of polyploid cells. Highly polyploid mitotic figures are commonly observed in mutants defective in mitotic cytokinesis, such as those in the citron kinase, diaphanous, or twinstar genes (Castrillon and Wasserman, 1994; Gunsalus et al., 1995; D'Avino et al., 2004; Naim et al., 2004; Shandala et al., 2004). However, rab11^93Bi/rab11^E(To)3 and rab11^93Bi/rab11^E(To)11mutant brains did not exhibit polyploid mitoses (n = 500), suggesting that Rab11 is either not required for cytokinesis or residual Rab11 activity is sufficient to mediate cytokinesis in these cells.

Functional Relationships among Rab11, Gio, and Fwd
Mutations in rab11 cause cytokinesis defects that are virtually identical to those observed in fwd and gio mutants (Brill et al., 2000; Giansanti et al., 2004, 2006; Gatt and Glover, 2006). These defects include incomplete constriction of the actin and anillin rings, central spindle disorganization in late telophases, and accumulation of Golgi-derived vesicle at the cell equator. This prompted us to construct and analyze rab11 fwd, rab11 gio, and fwd gio double mutants to ask whether the three genes function in the same pathway. Pathway analysis can be reliably performed when both mutations are null, or when one mutation is null and the other hypomorphic; the two genes belong to the same epistasis group if the phenotype of the double mutant is the same as the phenotype of the strongest mutant. When pathway analysis is performed with two hypomorphic mutations, there are two possibilities. If the double mutant has the same phenotype as the strongest hypomorphic mutant, then the genes are likely to be in the same epistasis group. If the double mutant has a phenotype stronger than either single mutant, it cannot be established whether the genes function in the same linear pathway or in two different pathways. For pathway analysis, we used the null fwd³/fwd^Z0453 heteroallelic combination, the gio^EP513 hypomorphic allele and the rab11^93Bi/rab11^E(To)11 hypomorphic combination (see Materials and Methods). We found that in each double mutant the cytokinesis defect was qualitatively and quantitatively identical to that observed in animals homozygous for the stronger mutation used to construct the double mutant (Figure 9, A–C; data not shown). Specifically, we found that fwd³/fwd^Z0453 single mutants are not significantly different from rab11^93Bi fwd³/rab11 ^E(To)11 fwd^Z0453 and fwd³ gio^EP513/fwd^Z0453 gio^EP513 double mutants for the frequency of aberrant spermatids (Figure 9, A and C). Comparable frequencies of aberrant spermatids were also observed in gio^EP513/gio^EP513 single mutants and in rab11^93Bi gio^EP313/rab11^E(To)11 gio^EP313 double mutants (Figure 9B). Collectively, these results indicate that fwd, gio and rab11 belong the same epistasis group. This suggests that fwd, gio, and rab11 function in the same linear pathway controlling membrane addition at the cleavage furrow and ring constriction during spermatocyte cytokinesis.

View larger version (65K): [in this window] [in a new window]   Figure 9. Functional relationships between Rab11, Nuf, Gio, and Fwd. (A–C) Frequencies (%) of multinucleated spermatids, containing two (2nu), three (3nu), four (4nu), or more than four (>4nu) nuclei, observed in rab1193Bi/rab11E(To)11 and fwd3/fwdZ0453 single mutants and in rab1193Bi fwd3/rab11 E(To)11 fwdZ0453 double mutants (A); in rab1193Bi/rab11E(To)11 and gioEP315/gioEP315 single mutants and rab1193Bi gioEP313/rab11E(To)11 gioEP313 double mutants (B); and in gioEP513/gioEP513 and fwd3/fwdZ0453 single mutants and in fwd3 gioEP513/fwdZ0453 gioEP513 double mutants (C). (D) Rab11 fails to concentrate at both the cell poles and the cleavage furrow of primary spermatocyte telophases from gioEP513/gioEP513 and fwd3/Df(3L)7C mutants. Telophases were also stained for Cnn (red) and DNA. Rab11 staining at the cell poles and the cleavage site was detected in all wild-type telophases (n = 128), none of the fwd mutant telophases (n = 56) and 4% of the gio mutant telophases (n = 83). (E) Gio localization in telophases I from rab1193Bi/rab11E(To)11 mutants is the same as in wild type. The merges show only tubulin (green) and DNA staining (blue). Gio staining at the cell poles and the cleavage site was detected in all wild-type (n = 70) and rab11 mutant (n = 71) telophases. (F) Nuf localization in wild-type and mutant telophases. In wild type telophases, Nuf concentrates at the cell poles and the cleavage site (we observed this staining pattern in all telophases examined; n = 50). This localization is disrupted in all the rab1193Bi/rab11E(To)11 (n = 60), fwd3/Df(3L)7C (n = 36) and gioEP513/gioEP513 (n = 38) mutant telophases. In the merges, Nuf is red, tubulin green and DNA blue. Bars, 10 µm.

Rab11 Is Required for Acroblast Formation
Immunostaining of testis preparations revealed that Rab11 accumulates to a conical structure at the anterior side of spermatid nuclei, resembling the acroblast (Figure 10A). Drosophila acroblasts are Golgi-derived structures enriched in the Lva and Fws proteins (Farkas et al., 2003; Giansanti et al., 2006). Double immunostaining of wild type spermatids for either Rab11 and Lva or Rab11 and Nuf revealed extensive colocalization of the three proteins, demonstrating that Rab11 and Nuf are both enriched at the acroblast (Figure 10, A and B).

View larger version (65K): [in this window] [in a new window]   Figure 10. Mutations in rab11 disrupt acroblast formation. (A) Wild-type spermatids expressing Rab11-GFP (green) stained for Lva (red). Note that the acroblasts are enriched in both Rab11 and Lva. (B) Rab11 (green) colocalizes with Nuf (red) in the acroblasts of early elongating spermatids. (C and D) Spermatids from wild-type (C) and rab11 mutant (D) males stained for Lva (red), tubulin (green), and DNA (blue). Note that both round (left arrow) and elongating wild-type spermatids (right arrow) display compact acroblasts; elongating spermatids from rab11 mutants do not exhibit an acroblast at the anterior side of the spermatid nuclei (arrowheads), but they show multiple Golgi vesicles dispersed along the growing sperm tails. Bar, 10 µm.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Rab11 Localizes to Multiple Membrane Compartments of Dividing Drosophila Spermatocytes
Studies in mammalian cells have shown that during interphase Rab11 localizes to the Golgi apparatus, the Golgi-derived vesicles and the RE, suggesting an involvement of this GTPase in both the secretory and the endocytic pathway (Ullrich et al., 1996; Chen et al., 1998; for review, see Zerial and McBride, 2001). In addition, Rab11-enriched endosomes accumulate at the cleavage furrow of dividing mammalian cells and they are required for completion of cytokinesis (Horgan et al., 2004; Fielding et al., 2005; Wilson et al., 2005). We have shown that Rab11 exhibits a highly dynamic behavior during Drosophila male meiosis. In prophase primary spermatocytes, Rab11 is enriched in a subcompartment of the Golgi stacks, which is likely to correspond to the trans-Golgi network (TGN). A preferential localization of Rab11 in this Golgi compartment is suggested by studies indicating that Rab11 associates with the TGN and TGN-derived vesicles in both mammalian cells and Drosophila photoreceptors (Urbe et al., 1993; Chen et al., 1998; Satoh et al., 2005). When Drosophila spermatocytes enter the first meiotic division and the Golgi stacks disassemble, Rab11 concentrates at the ER and remains associated with this compartment throughout meiotic division. In addition, Rab11 concentrates in several vesicle-like structures that reside at the cell poles during anaphase and early telophase. During mid- and late telophase, these vesicles move toward the cell equator where they seem to fuse with the advancing cleavage furrow.

The precise origin of these Rab11-enriched vesicles is currently unclear. Specifically, it is not clear whether these vesicles are recycling endosomes. In prophase spermatocytes, Rab11 is not concentrated in pericentriolar RE, but it is enriched at the Golgi stacks. In dividing spermatocytes, pericentriolar regions are enriched in both Rab11 and Pdi and they do not appear as distinct REs. It is thus likely that Drosophila spermatocytes do not possess a canonical pericentriolar RE. Previous studies have shown that different Drosophila cell types differ markedly in the organization and function of Rab11-containing structures. In embryonic cells, Rab-11 is enriched at both the Golgi apparatus and a pericentriolar RE, and membrane traffic to the advancing cellularization furrow is mediated by this RE (Pelissier et al., 2003; Riggs et al., 2003). In contrast, in differentiating photoreceptor cells, which do not contain a pericentriolar RE, rhodopsin-containing vesicles move directly from the TGN to the apical membrane to form the rhabdomere (Satoh et al., 2005). Studies on external sensory organ precursor (SOP) cells have shown that the formation of a pericentriolar RE is subject to a striking regulation. The SOP cells divide into a posterior pIIa cell and an anterior pIIb cell, which will give rise to the outer (hair and socket) and inner (neuron and sheath) sensory organ cells, respectively. During mitosis, Rab11 is equally distributed between the pIIa and pIIb daughter cells, but after completion of cytokinesis, only the pIIb cell forms a pericentriolar Rab11-enriched RE. In the pIIa cell, the formation of an RE is inhibited and Rab11 remains dispersed in the cytoplasm (Emery et al., 2005). Collectively, these studies indicate that the formation of a pericentriolar RE is cell specific and tightly controlled. Thus, it is likely that Drosophila spermatocytes, like the photoreceptor and the SOP pIIb cells, do not possess a pericentriolar RE. This implies that the Rab11-enriched vesicles generated by mitotic fragmentation of the Golgi apparatus can traffic directly to the equatorial plasma membrane.

The Cytokinetic Phenotype of rab11 Mutants
Our results have shown that rab11 mutant spermatocytes display two main cytokinetic defects. First, they exhibit an abnormal accumulation of Lva-enriched vesicles at the equator of telophase cells. Second, although they form regular contractile rings, these ring fail to constrict properly, leading to cytokinesis failures.

Several studies suggest that the presence of Lva-containing vesicles at the cell equator reflects a defect in membrane–vesicle fusion at the cytokinetic furrow. In wild-type spermatocytes undergoing telophase, Lva-enriched vesicles are concentrated at the cell poles and they are excluded from the equatorial region. An aberrant localization of these vesicles near the cleavage furrow has been observed previously in males homozygous for gio and fwd, two genes that encode proteins involved in membrane traffic (Giansanti et al., 2006). However, this vesicle phenotype is not a general feature of mutants that disrupt spermatocyte cytokinesis; Lva-positive vesicles were not observed at the equator of spermatocyte telophases from fws and pebble (pbl) mutants, which identify the Cog5 subunit of the conserved oligomeric Golgi complex and a Rho GEF, respectively (Farkas et al., 2003; Giansanti et al., 2006). There are several additional examples of abnormal vesicle accumulations caused by inhibition of functions required for vesicular traffic. For example, in Rab11-depleted Drosophila photoreceptor cells, rhodopsin-containing vesicles fail to fuse with the rabdomere and accumulate in the cytoplasm (Satoh et al., 2005). Vesicle accumulation near the cleavage site has been also observed after inactivation of the exocyst, a multiprotein complex that targets Golgi-derived vesicles to the plasma membrane (Hsu et al., 2004). Exocyst disruption results in abnormal secretory vesicle clustering near the cleavage site in both Saccharomyces cerevisiae and S. pombe (Salminen and Novick, 1989; Wang et al., 2002) and at the midbody of mammalian cells (Gromley et al., 2005). Collectively, these results strongly suggest that in rab11 mutant spermatocytes a fraction of the Lva-enriched vesicles fails to fuse with the invaginating furrow membrane, resulting in an abnormal accumulation of these structures at the cell equator.

We have shown that disassembly of the Golgi stacks generates Lva-positive vesicles that are either enriched in Rab11 or devoid of this GTPase. In addition, we have found that in rab11 mutants that carry a missense mutation, the mutant protein associates with many of the Lva-containing vesicles that accumulate at the telophase equator. These results lead us to suggest that in wild-type telophase spermatocytes Rab11-associated vesicles generated by Golgi disassembly move from either pole toward the cleavage site and fuse with the furrow, simultaneously delivering Rab11 and providing plasma membrane required for the completion of cytokinesis.

In addition to defects in membrane–vesicle fusion at the cleavage furrow, rab11 mutant spermatocytes also exhibit an incomplete actin ring constriction. There is precedent indicating that proper membrane trafficking is essential for actomyosin ring remodeling during animal cell cytokinesis. In cellularizing Drosophila embryos, mutant for syntaxin1, a target membrane-SNARE involved vesicle targeting, all sections of plasma membrane lack detectable cortical actin and associated furrow canals (Burgess et al., 1997). Consistent with these results, S2 cells depleted of syntaxin1 by RNAi fail to form a regular actin ring (Somma et al., 2002). Similarly, Dictyostelium mutants that lack the vesicle-coating protein clathrin fail to assemble a robust actin ring during cytokinesis (Niswonger and O'Halloran, 1997). Furthermore, disruption of the GM1 and cholesterol-enriched domain at the equatorial membrane of sea urchin eggs does not affect actomyosin ring formation but blocks furrow ingression and completion of cytokinesis (Ng et al., 2005). Finally, recent work has shown that PtdIns(4,5)P2 mediates actomyosin ring formation and stabilization in several systems, including crane fly, Drosophila and mammalian cells (Saul et al., 2004; Field et al., 2005b; Wong et al., 2005; for review, see Logan and Mandato). Collectively, these studies indicate that actomyosin ring formation and constriction during animal cell cytokinesis is intimately related with proper membrane traffic. However, the mechanisms underlying the interplay between the equatorial membrane and the contractile ring are still poorly understood. Specifically, it is not clear why failure of membrane-vesicle fusion at the cleavage furrow can block ring constriction in Drosophila spermatocytes. One possibility is that lack of vesicle addition to the advancing furrow results in membrane tension that counteracts ring constriction, ultimately leading to actin ring disassembly. Alternatively, the vesicles that normally fuse with the advancing cleavage furrow might include actin-remodeling factors essential for ring constriction and stability. A strong support for the latter alternative comes from recent studies on the control of ring constriction in budding yeast. In yeast exocyst mutants, the actin ring forms but fails to constrict properly and often disassembles, as occurs in rab11 mutant spermatocytes. This phenotype has been attributed to a failure of vesicle-mediated delivery at the bud neck of Chs2, a chitin synthase that is thought to control both ring stability and septum formation (VerPlank and Li, 2005). Additional insight into the mechanisms underlying defective ring constriction in rab11 mutant spermatocytes comes from our own phenotypic analysis. In mutant telophases with a high degree of ring constriction, the actin rings are often substantially thicker than in wild-type telophases at the same stage. This phenotype has been previously observed in fwd and gio mutants (Brill et al., 2000; Giansanti et al., 2004, 2006), and, in an exacerbated form, in mutants in the twinstar (tsr) gene, which encodes the actin-severing factor cofilin (Gunsalus et al., 1995). Based on these results, we favor the hypothesis that the thick rings observed in rab11 mutant telophases are caused by a failure of vesicle-mediated delivery of critical actin remodeling factors. However, we cannot exclude the possibility that mutant rings become thicker to overcome membrane tension.

Functional Relationships among Gio, Fwd, and Rab11
We have recently reported that the Gio PITP is enriched at the furrow membrane and that it is required for Drosophila cytokinesis (Giansanti et al., 2006). Here, we have shown that the furrow membrane is also enriched in Rab11 and that Rab11 localization at the equatorial membrane requires the wild-type activity of both gio and fwd. In addition, we have shown that the wild-type functions of gio, fwd, and rab11 are all required for membrane–vesicle fusion during cytokinesis, because mutations in these genes result in an abnormal accumulation of Golgi-derived vesicles at the equator of telophase cells (Giansanti et al., 2006; this study). Finally, our results strongly suggest that gio, fwd, and rab11 function in the same cytokinesis pathway. These observations suggest a model for the mechanisms underlying membrane addition to the cleavage furrow during spermatocyte cytokinesis. We propose that Gio mediates transfer of PtdIns monomers to the furrow membrane, causing a local enrichment in PtdIns molecules. The association of Gio with this membrane domain may facilitate recruitment of the PtdIns-4-kinase encoded by fwd, which would mediate phosphorylation of PtdIns to PtdIns(4)P, allowing their further phosphorylation to PtdIns(4,5)P2. Fwd may also mediate Rab11 recruitment at the cleavage furrow, allowing targeted Rab11-dependent vesicle fusion events necessary for completion of cytokinesis. We realize that this is a rather speculative model. Its major drawback is that the subcellular localization and the molecular interactions of the Drosophila Fwd protein are currently unknown. However, studies in S. pombe have shown that one of the PtdIns-4-kinases present in this organism interacts with Cdc4p, a contractile ring protein essential for cytokinesis (Desautels et al., 2001). This finding indicates that, at least in fission yeast, one of the PtdIns-4-kinases is associated with the cleavage furrow. In addition, a recent study has shown that one of the mammalian PtdIns-4-kinases interacts physically with Rab11 and is required for Rab11 localization in the Golgi complex. The same study has also shown that recruitment of this kinase to the Golgi does not require Rab11 (de Graaf et al., 2004). These results are consistent with our findings, and they lead us to believe that Gio, Fwd, and Rab11 are all enriched at cleavage furrow, where they work in concert to ensure proper vesicle docking and fusion.

Is Rab11 Specifically Required for Meiotic Cytokinesis?
We have shown that mutations in rab11 cause frequent failures in meiotic cytokinesis of males without affecting cytokinesis of larval brain neuroblasts. The mutations we analyzed are obviously hypomorphic as they cause lethality at the larval and pupal stages, whereas rab11 null alleles result in embryonic lethality. Thus, it is possible that the rab11 mutants we analyzed retain a residual Rab11 activity that is sufficient for neuroblast cytokinesis but not meiotic cytokinesis. Alternatively, Rab11 may not be required for mitotic cytokinesis. A strong support for a specific involvement of Rab11 in meiotic cytokinesis comes from recent RNAi screens that have shown that Rab11 has little or no role in S2 cell cytokinesis (Eggert et al., 2004; Echard et al., 2004; Kouranti et al., 2006).

Previous studies have shown that null mutations in fwd and fws disrupt spermatocyte cytokinesis but that they have no observable effects on larval neuroblast mitosis (Brill et al., 2000; Farkas et al., 2003; Giansanti et al., 2004). Thus, at least three proteins involved in membrane traffic, Rab11, Cog5, and a PtdIns-4-kinase, seem to be specifically required for meiotic cytokinesis. This specificity is unlikely to depend on the peculiar features of the final steps of spermatocyte cytokinesis. In male meiotic cells, the cytoplasmic bridges generated by ring constriction are not severed by a canonical abscission process, as occurs in larval neuroblasts; they instead persist and are stabilized by the formation of a specialized structure called ring canal (Hime et al., 1996; Giansanti et al., 1999). Mutations in rab11, fws and fwd inhibit ring constriction and furrow ingression during early telophase and block cytokinesis well before the formation of a cytoplasmic bridge. These observations rule out the possibility that the spermatocyte-specific effects of these mutations reflect problems in the final step of cytokinesis when ring canals are assembled.

The specific role of Rab11, Cog5, and Fwd in spermatocyte cytokinesis may reflect a specifically high requirement for formation of new membrane at the advancing cleavage furrow. To fulfill this requirement, male meiotic cells may exploit all the extant pathways for membrane addition. These pathways would be redundant in mitotic cell where the requirements for membrane expansion at the advancing furrow are relatively low. Alternatively, the specific requirement of membrane trafficking functions for spermatocyte cytokinesis may reflect the organization of membrane stores within these cells. Spermatocytes contain a large ER that includes astral and parafusorial membranes, and they do not