|Home | About | Journals | Submit | Contact Us | Français|
The fusome plays an essential role in prefollicular germ cell development within insects such as Drosophila melanogaster. Alpha-spectrin and the adducin-like protein Hu-li tai shao (Hts) are required to maintain fusome integrity, synchronize asymmetric cystocyte mitoses, form interconnected 16-cell germline cysts, and specify the initial cell as the oocyte. By screening a library of protein trap lines, we identified 14 new fusome-enriched proteins, including many associated with its characteristic vesicles. Our studies reveal that fusomes change during development and contain recycling endosomal and lysosomal compartments in females but not males. A significant number of fusome components are dispensible, because genetic disruption of tropomodulin, ferritin-1 heavy chain, or scribble, does not alter fusome structure or female fertility. In contrast, rab11 is required to maintain the germline stem cells, and to maintain the vesicle content of the spectrosome, suggesting that the fusome mediates intercellular signals that depend on the recycling endosome.
In Drosophila and many other insects, early gamete development takes place within cysts of interconnected germ cells that are generated prior to the onset of meiosis by multiple rounds of synchronous division (reviewed in de Cuevas and Spradling, 1997). Developing cyst cells (“cystocytes”) undergo poorly understood cell cycles characterized by altered spindle behavior, asymmetric division, and incomplete cytokinesis that generate a fixed pattern of cellular interconnections. The production of a normal cyst and of functional gametes depends on the fusome, an organelle highly enriched with small vesicles and without a delimiting membrane (Fig. 1). Consequently, obtaining a better understanding of the structure and function of the fusome is fundamentally important for furthering our understanding of gamete development.
In the female, a major role for the fusome during early oogenesis (Fig. 1A) is to mediate the transfer of materials from the 15 pro-nurse cells into the pro-oocyte, in association with microtubules. Microtubule motors (Dhc, KLP61F) and microtubule interacting proteins (Lis1, Orbit/Mast, Shot, Deadlock) are needed to generate normal fusome structure and to specify oocytes (McGrail and Hays, 1997; Wilson, 1999; Liu et al., 1999; Grieder et al., 2000; Máthé et al., 2003; Röper and Brown, 2004; Wehr et al., 2006). Proteins and mRNAs become enriched on the large segment of fusome associated with the future oocyte (reviewed by Huynh and St. Johnston, 2004). Centrioles, mitochondria and other organelles move along the fusome prior to entering the oocyte to form the Balbiani body (Grieder et al., 2000; Bolívar et al., 2001; Cox and Spradling, 2003).
In the male, an asymmetric fusome forms with the same branching pattern as in females (Fig. 1B). Branching continues during the two meiotic divisions, and large segments remain within each developing spermatid until the time of individualization (Hime et al., 1996). In contrast to oogenesis, the interconnected spermatocytes have equivalent fates, and differential transport in male cysts has not been observed. Nonetheless, the male fusome plays an essential role in spermatogenesis, as dispersal of its contents causes sterility (Wilson, 2005).
Previous studies have identified several major protein components of the Drosophila female fusome. These include the membrane skeleton proteins Hu-li tai shao (Hts) (Lin et al., 1994; Petrella et al., 2007), Alpha- and Beta-spectrin (de Cuevas et al., 1996), the spectrin-microtubule linking protein Spectraplakin (Röper and Brown, 2004), and stabilized microtubules (Röper and Brown, 2004). More recent studies have identified the endoplasmic reticulum (ER) proteins Reticulon I (Rtnl1) and Sec61α(Snapp et al., 2004; Röper, 2007) as fusome-enriched. Several other proteins associate with fusomes only transiently, or less specifically, including Bam (McKearin and Ohlstein, 1995), TER94 (León and McKearin, 1999), KLP61F (Wilson, 1999), Cyclin A (Lilly et al., 2000), PAR-1 (Cox et al., 2001; Huynh et al., 2001), Orbit/Mast (Máthé et al. 2003), Protein disulfide isomerase (PDI) (Röper, 2007) and Rab11 (Bogard et al., 2007). These studies suggest a model in which the fusome consists of a stable core of microtubules associated with endoplasmic reticulum-derived vesicles and organized by a meshwork of membrane skeleton proteins.
The nature and function of fusome membranes have remained poorly understood. The presence of ER-resident proteins including Rtnl1, Sec61-alpha, PDI, and an ER-trapped GFP fusion protein (Lys-GFPKDEL) suggest an origin from the endoplasmic reticulum (Snapp et al., 2004; Röper, 2007). Fusome membranes form a largely continuous tubular network in developing cysts, similar to cytoplasmic ER (Snapp et al., 2004). This continuity may be important for the ability of the fusome to synchronize cystocyte divisions. Whether the specialized organization of ER within the fusome serves any other functions remains unknown.
A role for the fusome in maintaining germline stem cells (GSCs) has not yet been established. GSCs remain present and active despite dispersal of the fusome in hts mutants (Lin et al., 1994). However, recent work has shown that Rab11, a marker of the recycling endosome, is enriched in the fusome (Bogard et al., 2007). Germline clones of a rab11 null allele increased GSC loss fourfold, and reduced adherens junction components that normally link GSCs to their niche (Bogard et al., 2007). Consequently, Rab11 function, but not an intact fusome, may be needed to maintain GSCs in their niche.
In order to better understand the fusome, and especially its vesicular compartment, we identifed and mutated genes encoding novel fusome components. Our studies provide new insights into the dynamic behavior of the fusome, and reveal striking differences between male and female fusomes. Moreover, we find that Rab11 and the recycling endosome maintain fusome membranes, and act to retain GSCs by a novel, adhesion-independent mechanism.
All strains were maintained at 22–25°C on standard medium. Flies were fed wet yeast paste for 1–2 days prior to dissections and heat shocks. Stocks from the Carnegie Protein Trap Library (Buszczak et al., 2007) are listed in Table 1. Stocks for clonal analysis were generated by recombination of the alleles listed in Table 2 onto w; neo FRT82B (Bloomington stock). Fer1HCHPZ00451, scribKG04161, and tmodPZ00848 were generated by the Berkeley Drosophila Genome Project (Bellen et al., 2004). Fer1HCHd00634, Fer1HCHf01958, rab11d04643, rab11e03152, scribc00019, scribf02351, tmodf00674, and tmodf04997 were obtained from the Harvard Exelixis Collection (Thibault et al., 2004). rab11E(to)3 (Jankovics et al., 2001) and scrib1 FRT82B (Bilder and Perrimon, 2000) were described previously. Further details about these alleles and the other stocks used in this work are available at FlyBase (http://flybase.bio.indiana.edu/).
To generate negatively marked clones, male flies bearing control and recombinant FRT82B chromosomes were crossed to hs-FLP; FRT82B Ubi-GFP/TM3 virgins. Progeny of these crosses were collected and fed wet yeast paste for 2–4 days prior to clone induction. Clones were induced in adults by two 45-minute heat shocks at 37°C in a circulating water bath. A one- to six-hour recovery period at room temperature separated the heat shock treatments. Subsequent to heat shock, flies were placed in fresh vials with wet yeast paste, yeast paste was changed every 1–2 days. To generate clones using the dominant-female sterile technique, males of genotype hs-FLP; FRT82B ovoD1/TM3 were crossed to females bearing control and recombinant FRT82B chromosomes. Clones were induced in the adult progeny of this cross by a one-hour heat shock treatment. Heat shocked females were then crossed to control (y, ry506) males and scored for fertility. Non-heat shocked females served as a control for clone induction.
Ovaries were fixed and immunostained as described previously (Kai and Spradling, 2003). Primary antibodies were used at the following dilutions: rabbit anti-GFP (1:1000–1:5000; Torrey Pines); mAb 1B1 (anti-Hts) (1:20 – 1:50; Developmental Studies Hybridoma Bank, DSHB); anti-Scribble (1:1000) (Bilder and Perrimon, 2000); anti-Tropomodulin (1:1000; a gift from Hugo Bellen); anti-Trailer Hitch (1:1000) (Wilhelm et al., 2005); anti-Rab11 (1:500) (Dollar et al., 2002); anti-Alpha Spectrin (1:500) (Dubreuil et al., 1987); anti-E-cadherin (1:40, DSHB); anti-Armadillo (1:40, DSHB). AlexaFluor-conjugated secondary antibodies (Molecular Probes) were used at a concentration of 1:200. Nuclei were labeled with 4′, 6-diamidino-2-phenylindole (DAPI). Samples were mounted in Vectashield (Vector Laboratories). Images were sampled on a Leica SP2 confocal microscope.
Temporal expression was scored by comparing the GFP channels separately with the Hts channels for each line. Germline stem cell analysis was performed by determining the number of germaria in a sample with at least one marked (GFP−) germline stem cell as a percentage of the total number of germaria in the sample.
To identify novel protein constituents of the fusome, we screened the Carnegie Protein Trap Library (Buszczak et al., 2007). Each library strain harbors a different single gene that has been fused at its normal chromosomal location to GFP. The exact location of the inserted GFP exon, and often the structure of the fusion RNA, is known. We initially examined GFP expression in live ovarian tissue, but immunofluorescence staining of fixed material ultimately proved to be more sensitive. We selected lines whose GFP expression was enriched in the fusome even if other cellular compartments were also labeled, or if the fusome expression was confined to only one developmental period. To verify that expression was associated with the fusome, double-label immunofluorescence experiments were performed using antibodies specific for GFP and for Hts. Ultimately, twenty of 244 unique protein trap lines (8.2%) were identified with fusome-enriched expression (Table 1, Fig. 2).
We identified six proteins that were known previously to localize to the fusome, strongly supporting the validity of our approach. These proteins are: PAR-1, PDI, Sec61alpha, Rtnl1, TER94, and Rab11 (Figs. 2B, F, I, K, M, and Q). 14 lines trapped Drosophila proteins with close orthologs in other organisms, but which had not been associated previously with the fusome. We used specific antibodies to verify the accuracy of the GFP expression patterns and fusome enrichment in a sample of these lines, including tropomodulin (tmod), trailer hitch (tral) and scribble (scrib) (Figs. 2A, D, and N, insets). In each case the GFP pattern and antibody staining pattern corresponded closely. Our results show that the female fusome contains a larger and more diverse set of proteins than previously known.
Several of the fusome-enriched proteins regulate cytoskeletal structure and polarity (Table 1). Tropomodulin, for example, serves as an actin pointed-end capping protein within the vertebrate erythrocyte membrane skeleton (reviewed in Bennett and Baines, 2001). Orthologs of several other membrane skeleton proteins, including Hts, Alpha-Spectrin, Beta-Spectrin and Ankyrin, also reside within the fusome (reviewed in de Cuevas et al., 1997). Two other fusome-trapped proteins, the PAR-1 kinase, and the Drosophila GSK3 homolog Shaggy (Sgg), regulate microtubule-associated proteins (MAPs) (Gögel et al., 2006; reviewed in Munro, 2006). Scribble is a putative scaffolding protein and tumor suppressor that localizes to cell junctions and maintains epithelial polarity (reviewed in Bilder, 2004). The identification in the fusome of microtubule regulatory proteins is not surprising, given the fusome’s role in controlling microtubules and cyst polarity.
Half of the fusome-associated proteins are either known or predicted to localize to the endoplasmic reticulum. Calreticulin (Crc) and PDI are involved in protein folding within the ER lumen (reviewed in Ni and Lee, 2007). Sec61alpha, Sec63, and the gene product of l(1)G0320 (TRAP-alpha/SSR) function in translocating proteins into the ER (reviewed in Johnson and van Waes, 1999). Orthologs of TER94 (reviewed in Patel and Latterich, 1998), Rtnl1 (Voeltz et al., 2006), and Tsp42Ee (reviewed in Hemler, 2003) play roles in the biogenesis of various ER domains. Trailer hitch (Tral) facilitates targeted secretion and localizes to subdomains of the ER, near ER exit sites (Wilhelm et al., 2005). The function of one predicted ER resident protein (Surf4) is unknown. These results strongly support the view that many fusome vesicles are a form of endoplasmic reticulum (Snapp et al., 2004).
Six of the fusome-localized proteins are characteristic of endosomal compartments, components not previously associated with this organelle (Table 1). For example, Cp1 and CG12163 are the Drosophila homologs of the lysosomal cysteine proteases cathepsin L and F, respectively (Kocks et al., 2003). Sap-r is a homolog of a saposin class lysosomal protein (reviewed in Vaccaro et al., 1999). Rab11 is a small GTPase known to associate with and to regulate recycling endosomes (Ullrich et al., 1996). Vha16 is a subunit of the vacuolar ATPase that localizes to subcellular vesicles such as lysosomes and also to gap junctions (Bohrmann et al., 2000). Fer1HCH is part of the ferritin complex, which localizes to the Golgi and secretory vesicles in insects (reviewed in Nichol et al., 2002). The localization of these tagged proteins strongly suggests that in addition to ER, vesicles from the post-Golgi endosomal pathway also are found within the fusome.
The expression patterns revealed that the composition of the fusome changes during follicle development. Four general categories of expression were observed (Table 1). Tropomodulin protein represents one class, with constitutive expression from GSC fusomes (termed spectrosomes), to the fusome remnants in budding egg chambers (Fig. 2A). Three “early” class proteins, PAR-1, Shaggy, and Scribble, are expressed predominantly in spectrosomes and in the fusomes of growing cysts (Figs. 2B–D). They downregulate sharply in sixteen-cell cysts and can no longer be detected in fusomes after region 2a, although Shaggy and Scribble continue to be present elsewhere in the cyst. The density of ER-like vesicles is known to increase as stem cell daughters leave the niche (Lin et al., 1994). Early fusome genes may participate in processes limited to stem cells and the initial stages of germ cell development.
A third class, the “middle” proteins, including many ER-associated proteins, show a peak of fusome enrichment in newly completed 16 cell cysts (Figs. 2E–L and O–R). By the time cysts have reached region 2b (Fig. 1) the levels of these proteins have fallen. The final class of “late” components include the proteasome component TER94 and the lysosomal proteins Sap-r and CG12163 (Cathepsin F) (Figs. 2M, S, T). Stage-specific proteolysis is likely to be important for fusome and ring canal development. Recent work demonstrated that the essential fusome protein Ovo-Hts is cleaved to produce a major fusome component and the ring canal protein HtsRC, which is stabilized only within older cysts (Petrella et al., 2007). These observations suggest that differential protein association with the fusome may mediate the alterations in fusome size and structure that occur within region 2b cysts.
To compare the organization of male and female fusomes we analyzed the expression of the fusion proteins within Drosophila testes. Fusome enrichment was determined as before by comparing the level of GFP labeling to the level of Hts staining (Fig. 3 and Table 1). Tropomodulin and the early proteins are also present in male fusomes. The temporal expression patterns of these genes during male germ cell development closely match those seen in the female. Thus, Tmod is found in fusomes throughout male development, while PAR-1 expression is restricted to fusomes of mitotic germ cells (Figs. 3A–B). These results emphasize the similar processes by which male and female germline cysts form.
The putative endosomal components of the fusome behaved differently. None of these proteins were present in male fusomes (Fig. 3C, Table 1). However, an intriguing pattern of localization was observed in two such proteins, Saposin-related protein, and CG12163 (Cathepsin F). While absent from the fusome proper, both proteins were found in large cytoplasmic puncta directly adjacent to it. Some of these puncta were partially surrounded by fusome material (Fig. 3D–D′, insets).
Males contained fewer endoplasmic reticulum-associated fusome components than females. For example, the ER lumen-localized chaperone proteins PDI and Crc, as well as the components of the translocon and its associated proteins, were all enriched in the male fusome (Fig. 3E, and data not shown), but other ER proteins were not. Most notably absent from the male fusome was Rtnl1, a marker for smooth ER tubules that is highly fusome-specific in females (compare Fig. 2K and Fig. 3F). These results suggest that while male fusomes may house functional ER components, they lack proteins that, in female fusomes, are simply being transported toward the oocyte, a process that does not take place in males.
In order to address the function of the newly identified fusome genes in adult germ cells, we utilized the FLP-FRT system to generate homozygous mutant clones (Figure 4 and Table 2). Multiple, independently derived alleles of each gene were recombined onto FRT-bearing chromosomes and tested. Phenotypes consistent between alleles were likely to be characteristic of the gene in question, rather than derived from background chromosomal lesions. In addition, the dominant-female sterile technique (Chou and Perrimon, 1996) was used to address the ability of mutant germ cells to produce fertile eggs (Table 2).
We selected the tmod gene for initial studies because it encodes a membrane skeleton protein ortholog we expected would function like Hts or Alpha-Spectrin in the fusome. We tested clones of three lethal tmod alleles that fail to complement each other or a deletion of the region (Table 2 and data not shown). Clones produced from all three alleles demonstrated a lack of Tmod antibody staining (data not shown), but did not display any defects in fusome structure (Fig. 4A). Moreover, tmod mutant germ cell cysts contain a full complement of fifteen nurse cells and a single oocyte (Fig. 4B), and development appeared normal at all later stages of oogenesis. In addition, all three tested tmod alleles produced homozygous germline clones that could give rise to fertile eggs (Table 2). Thus, despite its requirement in the erythrocyte membrane skeleton (Chu, et al. 2003), Tmod is dispensible for oogenesis.
The scribble gene encodes a protein that acts to maintain the polarity and integrity of epithelia, and it is essential in follicle cells for these purposes (Bilder et al., 2000). The temporal restriction of Scribble expression to early fusomes resembles that of PAR-1, another epithelial polarity gene (Doerflinger et al., 2003). Loss of PAR-1 does not interfere with cyst formation but is required later to maintain oocyte determination (Cox et al., 2001). We analyzed four different scrib alleles in germline and somatic clones (Table 2, Figs. 4C–D). Although all alleles tested displayed the previously characterized follicle cell phenotype (Bilder et al., 2000), germ cells mutant for scrib developed into fertile eggs (Table 2). Germline clones of all tested scrib alleles give rise to cysts containing a normal fusome (Fig. 4C), and produce follicles with 15 nurse cells and one posteriorly positioned oocyte (Fig 4D). Thus, unlike follicle cells, which require scrib to prevent tumor formation, germ cells develop normally in its absence.
The ferritin 1 heavy chain homolog (Fer1HCH) gene encodes the heavy chain of a conserved protein complex that functions in iron storage (reviewed in Hentze et al., 2004). In Drosophila, Ferritin is highly expressed in the hemolymph, in special cells in the midgut that accumulate and store iron, and at lower levels in many other tissues, including the ovary (Georgieva et al., 2002). Fer1HCH is enriched in the Golgi and ER within the midgut (Missirlis et al., 2007). We analyzed four recessive lethal insertions within the gene (Table 2), all of which failed to complement each other or a deficiency in the region (data not shown). Germ cells lacking each allele had normal fusomes, cyst structure and oocyte determination (Fig 4E-F). Fertile eggs were recovered from germline clones of two tested alleles using the ovoD germline clone technique (Table 2). Thus, at least some prominent cytoskeletal, polarity, and membrane-associated fusome proteins are not required for fusome morphology or function.
The recycling endosomal GTPase Rab11 functions during cellularization (Riggs et al., 2003; Pellisier et al., 2003), Notch signaling (Emery et al., 2005), cadherin recycling (Langevin et al., 2005), and oocyte polarization (Dollar et al., 2002; Jankovics et al., 2001). We examined mitotic clones of three different recessive lethal rab11 alleles in order to better understand the recycling endosomal component of the fusome. All alleles caused a marked reduction in Rab11 expression within the fusome in homozygous clones (Fig 5A). In contrast to the previously studied genes, rab11 mutant germ cells showed striking defects beginning eight days after clone induction. Mutant cysts have a reduced number of cells. Rather than the normal sixteen cells found in wild type cysts, mutant cysts in region 2 of the germarium at twelve days after clone induction average 12.4 cells per cyst (N=25). In addition, rab11-deficient cysts are much more rounded than wild type, and fail to move normally out of the germarium (Figs. 5B–C). Their fusomes are often morphologically defective, with reduced branching and the accumulation of rounded aggregates of fusome material (Figs. 5B–C, arrow). Thus, rab11 is required for normal fusome structure and cyst formation.
We noticed that rab11 mutant GSCs were often lost, so we measured their half life compared to wild type controls and to GSCs lacking non-essential fusome proteins. Four days after clone induction, both rab11 and control GSCs were found at a similar frequency (Fig. 5D). However, over the next 2 weeks, the rab11 GSCs were lost about fourfold faster than either wild type GSCs or GSCs homozygous for any of the other tested genotypes (Fig. 5D). The decreased proportion of ovarioles with rab11 mutant GSCs was due to stem cell loss because many ovarioles with only wild type GSCs continued to contain downstream mutant cysts and follicles (Fig. 5C).
Two potential mechanisms might explain the instability of rab11 mutant GSCs. Given Rab11’s known role in cadherin recycling to the plasma membrane in Drosophila (Langevin et al., 2005), the adherens junctions that anchor GSCs to the somatic cap cells (Song, et al., 2002) might be compromised (Bogard et al., 2007). However, we observed no discernable change in the amount of E-cad at the GSC-cap cell border detectable by antibody staining in 96% (n=25) of rab11 GSCs examined seven or more days after clone induction (Fig. 6A). Armadillo (Beta-catenin) localization was also unperturbed in rab11 GSCs (Fig. 6B). Thus, if altered adhesion is the cause of GSC loss, it must occur suddenly, because all examined GSCs retained normal junctions to cap cells.
The recycling endosome plays a critical role in intercellular signaling through its action on membrane ligands and receptors. We examined GSCs in mosaic germaria using electron microscopy to look for defects in membrane trafficking that may be indicative of perturbed recycling endosomal function. In germaria mosaic for clones of two different rab11 alleles, we found distinctive defects in the fusome membranes of many GSCs. The number of tubulovesicular structures in these fusomes was markedly reduced (Fig 6D–E), a phenotype that was not observed in control germaria (Fig 6C). These observations suggest that active membrane recycling is required to maintain the normal membrane content of the spectrosome. The membrane-depleted spectrosomes may fail to support signals necessary to maintain GSCs.
The first molecular component of the fusome (Hts), was identified as a mutation in an unbiased forward genetic screen (Yue and Spradling, 1992), and several related fusome components were subsequently identified by antibody staining (deCuevas et al. 1996). Recently, an increasing number of Drosophila proteins have been fused to GFP in vivo in protein trap lines (Morin et al., 2001; Buszczak et al., 2007; Quiñones-Coello et al., 2007). Screening such lines provides a general method of identifying the molecular components of any subcellular structure that can be visualized in the fluoresence microscope. Several new fusome proteins were recently discovered using this approach (Röper, 2007). We have screened the largest collection of protein trap lines currently available, and as a result, have more than doubled the previously known number of fusome proteins. The high frequency with which trapped proteins demonstrate fusome enrichment, 20/243 (8%), contrasts with the rarity of mutations affecting the fusome, but suggests that this organelle is more complex than previously supposed, and likely contains hundreds of different proteins. As expected from the rarity of fusome mutations, only a small fraction of the identified fusome proteins were found to be essential for fusome structure and/or function.
Previous studies of the fusome have focused on its critical roles in germline cyst formation, cell synchronization, cytoskeletal polarization, and oocyte determination. Events crucial to all these processes occur while the fusome is still growing in region 1. Our studies support the view that growing and completed fusomes differ in structure. Proteins likely to be involved in the cytoskeletal and microtubule-organizing activities of the fusome are expressed early, while proteins associated with proteolysis and oocyte determination begin to be enriched in the fusomes of newly completed cysts. This suggests that the fusome plays distinct roles during cyst formation and during subsequent stages of germ cell development.
The upregulation of lysosomal proteases such as Sap-r and cathepsin F in the fusome following cyst completion argues that developmentally regulated proteolysis mediates at least some of these changes in fusome structure. First, key components of the fusome and ring canals, such as Hts protein isoforms, are proteolytically processed from a large precursor in a manner that changes after cyst completion (Petrella et al., 2007). Second, partial degradation of the fusome, especially in the vicinity of the ring canals, may open these intercellular bridges and allow specific molecular cargos and organelles to begin flowing toward the oocyte. This process may not only remove proteins blocking the ring canals, but may also enable microtubule motors to access the underlying core of stable, polarized microtubules within the fusome (Grieder et al., 2000; Röper and Brown, 2004).
We observed the protein constituents of multiple vesicular compartments within the fusome. These include transmembrane and luminal markers for ER membranes, lysosomes, secretory vesicles, and for recycling endosome components. The fusome may be a locus of post-Golgi vesicle trafficking within early germ cells, perhaps analogous to the subapical compartment of epithelia (Hoekstra et al., 2004). Golgi elements themselves are located near but are excluded from the fusome (King, 1970; data not shown).
Our experiments suggest two possible reasons for the non-random localization of vesicle compartments to the fusome. First, these vesicles may participate in carrying out the functions associated with the fusome during germ cell development. An intact fusome is required for cell cycle synchrony, and localization might ensure that lyosomal, endocytic, and secretory events are synchronized throughout the cyst. While the onset of pre-meiotic S phase in the newly formed 16-cell cyst may be synchronous, subsequent cell cycle events do not appear to be as tightly coordinated. The way synchrony is lost may correspond to the changes in fusome structure that take place following cyst completion.
The second reason why multiple vesicle compartments are associated with the fusome may be related to the role of the later fusome as a major pathway for transporting materials to the oocyte. Many vesicular components, especially those enriched at later stages, may simply be in transit. Centrosomes (Mahowald and Strassheim, 1970; Grieder et al., 2000; Bolívar et al., 2001), mitochondria (Cox and Spradling, 2003), and Golgi elements (Cox and Spradling, 2003) are all transported along the fusome in region 2 cysts. Our studies now suggest that post-Golgi endocytic compartments are also in transit, and we find that many of these are not essential for the completion of oogenesis. The identification of Fer1HCH within the fusome, a conserved protein involved in iron storage and transport, generally supports such an interpretation. Rather than functioning in the fusome, it may sequester iron for eventual storage and use within the oocyte.
A molecular model of the vertebrate erythrocyte membrane skeleton has long served as a starting point for thinking about the structure of the fusome’s cytoskeletal component. Transmembrane proteins associate with both the plasma membrane and with the membrane skeleton, linking the two together and allowing the underlying protein skeleton to shape the plasma membrane and provide rigidity. The same mechanism is thought to control the shape of Golgi elements. Consequently, it was postulated that a fusome skeleton exists and that individual vesicular elements might be linked to these proteins in a similar manner. The drastic disruption of fusome integrity caused by mutations in genes encoding membrane skeleton orthologs encouraged this view (Lin et al., 1994; de Cuevas and Spradling, 1996). The studies described here on the role of tropomodulin now call this model into question.
Tropomodulin, like Adducin, is an actin capping protein (Weber, et al., 1994). While Adducin caps the barbed ends of actin filaments in the erythrocyte cortical cytoskeleton, Tropomodulin localizes to and binds the pointed ends (reviewed in Bennett and Baines, 2001). Knocking out tmod greatly reduces membrane stability in mouse erythrocyte precursor cells (Chu et al., 2003). In contrast, we found that loss of Tmod from the fusome resulted in no detectable changes in fusome stability or in other aspects of cystocyte and follicle development.
There are several ways to rationalize these differences. The role of Tmod may differ in Drosophila and mammals. Perhaps proteins other than Tmod regulate actin pointed ends in junctional complexes in Drosophila. Alternatively, the basic organization of membrane skeleton proteins in the fusome may differ from that in the mammalian erythrocyte. Whatever the reason, it is likely to apply not only to the fusome, but also to the membrane skeleton of epithelial cells. Disrupting tmod within follicle cells did not affect their membrane skeletons as evidenced by anti-Hts staining, or compromise cell viability. Despite its dispensability in both germ cells and the follicular epithelia, tmod mutations in Drosophila are lethal, indicating that this protein plays an essential role in other tissues.
Our studies also revealed that male and female fusomes differ significantly in composition. All known early components of the female fusome are also found in males, including the newly identified proteins in that category described here (Tmod, Scribble, and Shaggy). This presumably reflects the similar mechanisms by which male and female fusomes are formed during synchronous cystocyte divisions and the strong similarity of their initial structure and branching pattern.
Male fusomes also express proteins characteristic of endoplasmic reticulum. Markers for rough endoplasmic reticulum such as the translocon components Sec61alpha and Sec63 are enriched, as are ER luminal proteins involved in protein folding such as PDI and Crc (Table 1). However, not all ER-localized proteins within female fusomes were found in the male version of the organelle. Rtnl1, a protein associated with smooth ER (Röper, 2007), is absent from male fusomes. Rtnl1 is highly enriched in the female fusome and later accumulates heavily in the oocyte. This also is true of Trailer hitch, which along with Cup, Me31B and other proteins is involved in translationally regulating mRNAs in transit to the oocyte (Wilhelm et al., 2005). Loss of Rtnl1 in mutant clones has no effect on fusome structure or fertility (Röper, 2007). Male cyst cells have equivalent fates, whereas female cystocytes produce one oocyte and fifteen nurse cells. The differential composition of the fusomes in male and female cysts is likely a reflection of this fundamental difference. Therefore, we suggest that fusome constitutents in the female that are simply in transit to the oocyte are likely to be absent in males, where such transport appears absent. Such proteins are also frequently unnecessary to complete egg formation.
Endosomal and lysosomal proteins found in the female fusome were also absent from the male organelle. These components become prominent in mid- to late- fusomes, and may remodel fusome structure in order to activate directional transport. The absence of lysosomal components within male fusomes may ensure that it is not chewed back to expose its underlying microtubules. Instead, a different process of modification may occur in males, as we observed that lysosomal proteases were located very close to the male fusome at some stages.
Our experiments suggest that the recycling endosome functions within the fusome. Rab11 was known previously to be necessary for proper oocyte polarization in maturing stage 8–10 follicles (Jankovics et al., 2001; Dollar et al., 2002). We found that Rab11 was required to maintain the normal vesicular structure of the fusome. Previously, only strong bam mutations had been observed to deplete spectrosomal membranes (McKearin and Ohlstein, 1995). Our observations suggest that Bam might normally function to promote endocytic recycling within the fusome.
Recently, Rab11 was localized to the fusome and a role in GSC maintenance was detected (Bogard et al., 2007). In contrast to this previous study, we did not observe a concomitant loss of cell adhesion molecules at the GSC-cap cell border, despite the fact that the alleles tested in both studies reduced GSC lifetime by a similar amount (fourfold). Changes in adhesion may be secondary to reduced endosome recycling within the fusome, but the most significant effect of lowering Rab11 may be on signaling. Rab11 plays an essential role in Notch signaling in the sensory organ precursors of Drosophila (Jafar-Nejad et al., 2005; Emery et al., 2005), and the Notch pathway is required both for GSC niche formation (Song et al., 2007), and GSC maintenance (Ward et al., 2006). JAK/STAT signaling is also under endocytic control, but this pathway requires trafficking mediated by rab5, rather than Rab11 (Devergne et al., 2007). Our results suggest that the fusome normally mediates a Notch signal that is essential for GSC maintenance.
The fusome may operate as a locus of membrane recycling that allows it to function as a regulatory center for multiple events important for early oogenesis, in addition to GSC maintenance. The recycling endosome is thought to control important aspects of mitosis, such as plasma membrane growth (Pelissier et al., 2003; Boucrot and Kirchhausen, 2007). The completion of cytokinesis depends on the Rab11-mediated transport of endocytic vesicles to the cleavage furrow (Wilson et al., 2005, Giansanti, et al., 2007). Additionally, the recycling endosome, like the fusome, is frequently associated with the cell’s microtubule organizing center (Riggs et al., 2007). Therefore, the defects we observed in rab11 germline cyst formation may represent defects in fusome-mediated membrane growth during cystocyte divisions, or from defects in MT organization.
The recycling endosome is also critical to generate signals that can act within or between cells. Signaling pathways such as Notch, that are dependent on endocytic recycling, may accelerate G1/S transitions (Baonza and Freeman, 2005). Fusome-mediated signals might synchronize the cystocyte cell cycles and coordinate microtubule organization and other aspects of cystocyte development. Such a mechanism might explain how the amount of fusome within a cell could influence its developmental fate (de Cuevas and Spradling, 1998). By better understanding the structure and function of the fusome, we are likely to gain further insights into these important issues.
We thank Shelley Paterno, Julia Bachman, Stephanie Owen, and all the other members of the Spradling lab who contributed to generating and analyzing the Carnegie Protein Trap collection (Buszczak et al., 2007). We are grateful to Mike Sepanski for assistance with electron microscopy; Hugo Bellen, David Bilder, Robert Cohen, Daniel Kiehart, and Jim Wilhelm for antibodies and fly stocks; and Anna Allen, Rachel Cox, and Todd Nystul for comments on the manuscript. M. Buszczak was a fellow of the American Cancer Society.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.