The Ovhts polyprotein is cleaved to produce fusome and ring canal proteins required for Drosophila oogenesis

Germline development, like many other developmental programs, involves a preliminary stage of mitotic expansion, followed by differentiation into mature cell types. One hallmark of germline development is that the proliferative phase occurs in clusters of cells interconnected by cytoplasmic bridges (Matova and Cooley,2001). In some systems, such as oogenesis in mammals, cytoplasmic bridges disappear early and cells differentiate separately. However, in most known forms of spermatogenesis and in insect oogenesis, cytoplasmic bridges persist and contribute to differentiation. In insect oogenesis, one cell in each cluster becomes the oocyte and is nourished by the nurse cells through cytoplasmic bridges. In all cases, germline cells must undergo specific changes in gene expression, morphology and function when making the transition from proliferation to differentiation.

Drosophila oogenesis is an attractive model to study the transition from germline cell expansion to differentiation and the formation of intercellular bridges called ring canals (RCs). In the most anterior compartment of the ovary, the germarium, germline stem cells (GSCs) undergo division to form another GSC and a daughter cystoblast(Huynh and St Johnston, 2004). The cystoblast undergoes exactly four mitoses with incomplete cytokinesis forming a 16-cell cyst in which cells are connected by arrested cleavage furrows. Once mitosis is complete the cells differentiate into 15 nurse cells and one oocyte. The proliferating cells and differentiated cells can be easily identified through subcellular structures. During proliferation, an organelle called the fusome is present. After mitosis is complete, the fusome begins to disappear and RCs form at the arrested cleavage furrows.

The fusome is derived from a precursor structure called a spectrosome(Lin et al., 1994), which is present in GSCs. A portion of the spectrosome is inherited by the daughter cystoblast, after which it grows and branches during cystoblast divisions to form the fusome (de Cuevas and Spradling,1998). Spectrosome- and/or fusome-like structures have been described in the germline of insects, Xenopus and mouse(Pepling et al., 1999). A potentially analogous structure is in mammalian lymphocytes(Dubielecka et al., 2003). The spectrosome and fusome in Drosophila are areas of highly condensed vesicles including modified endoplasmic reticulum(de Cuevas et al., 1997; Snapp et al., 2004). Fusome membranes are associated with a cytoskeleton that includes the Adductin homolog Hu-li tai shao (Hts), α-Spectrin, β-Spectrin, Filamin(Cheerio - Flybase), a Spectraplakin homolog named Short Stop (Shot), and Ankyrin (de Cuevas et al.,1996; Lin et al.,1994; Roper and Brown,2004; Sokol and Cooley,2003). Hts and α-Spectrin are necessary for fusome structure because in mutants the fusome is absent (de Cuevas et al., 1996; Lin et al., 1994) causing dramatic defects in nurse cell number and oocyte specification.

RCs allow the movement of essential proteins and RNA from nurse cells into the oocyte; thus, mutations that result in their disruption are female sterile. Formation of RCs at arrested cleavage furrows follows an organized pattern of RC protein accumulation. F-actin, Filamin and Ovhts-RC, a novel product of the hts locus, begin to accumulate on RCs starting immediately after mitosis ends (Robinson et al., 1994). Mutations in the gene for Filamin(cheerio) or hts result in a failure of F-actin accumulation and arrested RC development (Robinson et al., 1997).

hts was isolated as a female sterile mutant(Ding et al., 1993; Yue and Spradling, 1992), and products of the hts gene are present on both the fusome and RCs,making Hts an attractive candidate for participating in the transition from fusome-containing proliferating cells to RC-containing differentiating cells. Subsequent work showed that there are four distinct proteins made from splice variants of the hts gene in the ovary(Whittaker et al., 1999) (this article). All predicted Hts proteins share identical N-terminal Head and Neck domains homologous to mammalian Adducin proteins(Fig. 1), but have unique C-termini. Add1 and Add2 contain the Adducin Tail and MARCKS domains. ShAdd protein has 23 novel C-terminal amino acids and no MARCKS domain. Ovhts contains 80% of the Adducin Tail domain, and a large C-terminal domain unique to Drosophila Adducins that we call the Ring Canal domain.

Antibodies made against Hts proteins label the fusome and RCs in the germline as well as the follicle cell membranes(Fig. 1)(Robinson et al., 1994; Zaccai and Lipshitz, 1996). Mutant phenotypes are consistent with expression data and include loss of oocyte specification, too few nurse cells, and RC deformities(Yue and Spradling, 1992). Protein localization and mutant phenotype analyses were done prior to knowing that the hts gene produces multiple proteins; therefore, further characterization of hts is needed to understand its roles during oogenesis.

We carried out detailed characterization of the localization of Hts proteins, focusing on the germline-specific proteins Ovhts and ShAdd. Remarkably, Ovhts is a polyprotein that is cleaved to form two functional products we call Ovhts-Fus and Ovhts-RC. Each Ovhts protein is stable during different developmental stages, with Ovhts-Fus present in proliferating cells and Ovhts-RC in post-mitotic differentiating cells. Although other polyproteins have been described, this is the first example where the products are maintained in different developmental time points and subcellular locations. Further experiments with both wild-type and uncleavable Ovhts proteins demonstrate that there is a developmental link between the degradation of the fusome and RC establishment. Finally, we show that Ovhts-RC is functionally required in RCs throughout oogenesis.

Drosophila cultures were maintained using standard procedures(Ashburner, 1989). The wild-type control was w¹¹¹⁸(Lindsely and Zimm, 1992). hts^ΔG was from the Zuker collection (Koundakjian et al.,2004) and hts^W532X was from Trudi Schüpbach (Princeton University, Princeton, NJ). Expression of P{UASH} transgenes was done by crossing to one of the following lines: P{bamP-bam5-Gal4:VP16}(Chen and McKearin, 2003), P{tub-Gal4}LL7 (Lee and Luo,1999), P{nos-Gal4-VP16}(Van Doren et al., 1998) or a maternal triple driver, MTD-Gal4, containing the P{Gal4-nos.NGT}40 (Tracey et al.,2000), P{COG-GAL4:VP16}(Rorth, 1998) and P{nos-Gal4-VP16} (Van Doren et al., 1998) germline drivers.

The hts gene was amplified from homozygous mutants by PCR, and the products were sequenced (W. M. Keck Foundation Biotechnology Resource Laboratory, Yale University).

The pCOH vector was made by removing the K10 3′ UTR fragment from pCOG (Robinson and Cooley,1997) and replacing it with the first 301 nucleotides of the ovhts 3′ UTR (representing base pairs 3868 to 4168)(Whittaker et al., 1999). The pUASH vector was made using the same methodology to replace the K10 3′UTR of pUASP (Rorth, 1998)with that of ovhts.

The coding region of Ovhts from the cDNA (bp 397-3864) was subcloned into the pCOH or pUASH vectors. The GFP (S65T) gene was modified from the pCS2^*mt-GFP (gift of Michael Klymkowsky, University of Colorado,Boulder) vector by PCR to add an 8 Ala linker on the N-terminus of GFP and cloned into Bluescript KS. GFP (S65T) was then subcloned into P{COH-Ovhts} and P{UASH-Ovhts} to make P{COH-Ovhts::GFP} and P{UASH-Ovhts::GFP}. P{COH-Ovhts-Δ3::GFP} was made by subcloning the StuI to BglII restriction fragment from pMT-Ovhts-Δ3(see below) into P{COH-Ovhts::GFP}. To make P{COH-Cer::Ovhts}, the Cerulean gene was modified using PCR from the mCerulean-C1 vector (Rizzo et al.,2004) to add flanking EcoRI sites, and then subcloned in-frame directly upstream of Ovhts in P{COH-Ovhts}.

To make P{COH-ShAdd::Ven}, ShAdd was cloned into pCOH. Venus(Nagai et al., 2002) was modified using PCR to add a 5′ XhoI site and 3′7×His tag followed by a stop the another XhoI site. Venus was then subcloned in-frame directly downstream of ShAdd in P{COH-ShAdd}.

Embryo injections to generate transgenic animals were either performed as previously described (Robinson and Cooley,1997), or were done at Duke University, Model System Genomics.

htsM antibody was modeled after the peptide antibody raised against the MARCKS domain of human β-Adducin(Matsuoka et al., 1998) using a peptide representing amino acids 689-718 of Add1, GSPKKDKKK -KKGLRTPSFLKKKKEKKKAEA (W. M. Keck Foundation Biotechnology Resource Laboratory, Yale University). Peptide conjugation, injection and antibody production was done at Cocalico Biological, Inc. (Cocalico Biological, Inc.,Reamtown, PA). htsM antibody purification was done using the SulfoLink Kit(Pierce #44895).

All S2 cell vectors were made in the pMT V5 His copper-inducible system(Invitrogen). Full-length pMT-Ovhts contained the entirety of the Ovhts coding region (bp 397-3864). Nucleotides were removed by two-step PCR in each deletion construct resulting in the loss of the following amino acids from Ovhts: Ovhts-Δ1 amino acids 632 to 651, Ovhts-Δ2 amino acids 639-658, Ovhts-Δ3 amino acids 650-669 and Ovhts-Δ4 amino acids 785 to 794.

S2 cells were cultured, transfected with 1 μg of DNA using Lipofectin(Invitrogen #50503) and induced with 0.7 mM CuSO₄.

Protein samples were prepared as previously described(Robinson et al., 1997). Primary antibodies were diluted in PBT with 5% milk as follows: polyclonal rat htsF at 1:5000 (Lin et al.,1994; Robinson et al.,1997), monoclonal htsRC at 1:10(Robinson et al., 1994)(Developmental Studies Hybridoma Bank), monoclonal 1B1 at 1:10(Zaccai and Lipshitz, 1996)(Developmental Studies Hybridoma Bank), polyclonal rabbit anti-GFP antibody 1:1000 (Torrey Pines TP401). Horseradish-peroxidase-conjugated secondary antibodies were used at 1:30,000 (Pierce #31430, #31460 and #31470). Proteins were detected using SuperSignal West Dura ECL Substrate (Pierce #34076).

Ovaries were dissected and fixed as described previously(Robinson and Cooley, 1997). To visualize actin, egg chambers were incubated with 2 U rhodamine-conjugated phalloidin in PBT. For antibody labeling, ovaries were incubated either overnight at 4°C or 2 hours at room temperature with htsRC antibody at 1:10, 1B1 antibody at 1:7, htsF antibody at 1:500, purified htsM antibody at 1:250 or anti-GFP antibody at 1:750 (Molecular Probes A11122). Secondary antibodies conjugated to Alexafluor 488 or 568 (Molecular Probes) were used at 1:500 and were incubated with ovaries for 2 hours at room temperature. Ovaries were stored in Antifade (0.23% DABCO in 0.1 M Tris-HCL 90% glycerol) overnight at 4°C and then mounted. All images were taken on either a ZEISS LSM-510 or a ZEISS LSM-510 META microscope (Center for Cell and Molecular Imaging,Yale University School of Medicine), and images were processed using Adobe Photoshop 7.

Because previous examination of Hts protein localization was done before there was a complete understanding of the hts locus, we characterized Hts antibody localization in more detail. We used four antibodies directed against different Hts protein domains (see Fig. 1), three of which were published previously. htsF (Lin et al.,1994; Robinson et al.,1997) recognizes ShAdd, Add1 and Add2 (Add½) and Ovhts. In germaria, htsF antibody labeled the fusome in the germline(Fig. 2A, arrow) and plasma membranes in follicle cells (Fig. 2A, arrowhead). 1B1 antibody(Zaccai and Lipshitz, 1996),which recognizes Ovhts and Add½, had an identical germariumlabeling pattern to htsF (Fig. 2B). htsM antibody, which recognizes only Add½, labeled follicle cell membranes(Fig. 2C) and showed no labeling of the germline. In later-stage egg chambers, htsF, 1B1 and htsM antibodies continued to show specific labeling of lateral follicle cell membranes but no germline labeling (Fig. 2E-G). Consistent with RNA in-situ data(Whittaker et al., 1999),Ovhts is germline-specific and Add½ are follicle-cell-specific. As shadd mRNA is also exclusively found in the germline, and the htsF antibody only labels the fusome in the germline, ShAdd is likely a fusome component (see below). Although both shadd and ovhts mRNAs are expressed in the germline throughout oogenesis(Whittaker et al., 1999),their protein products detected with 1B1 and htsF antibodies were only present in the germarium. This suggests that the proteins are either not translated or are not stable once egg chambers are formed.

Because we were unable to make a useful peptide antibody specific for ShAdd, we made a ShAdd transgene expressing ShAdd fused to Venus, a modified EYFP (Nagai et al., 2002). When ShAdd::Ven was expressed with the ovarian tumor (otu)promoter in the germline of wild-type flies, it localized specifically to spectrosomes and fusomes (Fig. 2I). As the fusome began to degrade in Region 2, the localization of ShAdd::Ven also became more dispersed(Fig. 2I). Thus, ShAdd::Ven provided additional evidence that ShAdd is a fusome protein.

HtsRC antibody (Robinson et al.,1994), which recognizes the C-terminus of Ovhts, had a completely different localization pattern. Starting in Region 2a of germaria, htsRC labeled discrete puncta, which resolved into RCs in Region 2(Fig. 2D). htsRC antibody labeled RCs throughout the rest of oogenesis(Fig. 2H).

In order to verify the different labeling patterns of antibodies against the N- and C-termini of Ovhts, we made tagged Ovhts transgenes that were expressed specifically in the germline. In separate constructs containing the native ovhts UTRs, the N-terminus of Ovhts was tagged with Cerulean,a modified ECFP, and the C-terminus was tagged with GFP. The Cer::Ovhts transgene did not produce a fluorescent product; however, upon labeling with anti-GFP antibodies, Cer::Ovhts was detected on the fusomes in germaria(Fig. 2J). Like the N-terminus of Ovhts as seen by antibody labeling, Cer::Ovhts localized to both spectrosomes and branched fusomes. Co-staining with 1B1 showed that as the fusome began to break down in Region 2, Cer::Ovhts became dispersed and lost colocalization with 1B1.

GFP fluorescence from Ovhts::GFP localized specifically to RCs(Fig. 2K). As with htsRC antibody, protein was first detected in Region 2a as puncta that appeared to be near, although not within the fusome(Fig. 2K,L′, arrowhead). By Region 2b Ovhts::GFP was in rings (Fig. 2K,L′, arrow). Ovhts::GFP was seen on RCs in all subsequent stages until stage 13 (data not shown). Thus, localization of tagged Ovhts transgenes recapitulated antibody labeling, with the N-terminus present on fusomes and the C-terminus localizing to RCs.

Although the predicted size of the Ovhts protein is 128 kDa, ovary extract never contained a protein of this size. Instead, htsF(Fig. 3A) and 1B1 (data not shown) antibodies detected a doublet of ∼90 kDa, which are the Add½proteins. When protein extracts were analyzed from virgin female ovaries,which contain fewer late-stage egg chambers and therefore are enriched for germarial tissue, a band of ∼80 kDa was detected(Fig. 3A). As reported previously (Robinson et al.,1994), the htsRC antibody detected a ∼60 kDa doublet(Fig. 3B). These results suggested that Ovhts is cleaved into two smaller proteins. Moreover, when ovary extracts of flies expressing Ovhts::GFP or Cer::Ovhts were blotted with anti-GFP antibody, only bands that would represent the cleavage products (with the added GFP tag) were seen (Fig. 3C). These results, along with protein localization data, show that Ovhts is cleaved. Based on the fact that antibodies to the N-terminus of Ovhts label fusomes and antibodies to the C-terminus label RCs, we designated the cleavage products Ovhts-Fus and Ovhts-RC, respectively.

To further investigate Ovhts cleavage, we expressed Ovhts in S2 cells,which do not express endogenous Ovhts. Western analysis with either 1B1 to visualize the N-terminus or htsRC to visualize the C-terminus, revealed 80 kDa and 60 kDa proteins like those detected in ovary extracts(Fig. 3D,E). Additionally, an∼140 kDa band that represents full-length Ovhts was apparent in S2 cell extract (Fig. 3D,E). As the full-length protein can only be detected when Ovhts is overexpressed in S2 cells and not in ovary extracts, the cleavage process in ovaries must be very efficient.

In order to elucidate the functions of the individual htsproteins, new alleles of hts were characterized. All previously described alleles of hts were P-element insertions or imprecise excisions that reduce expression of all hts transcripts. We examined two new EMS-induced hts alleles. hts^W532X (gift from Trudi Schüpbach) contains a single nonsense mutation, W532X, in the tail domain(Fig. 1B). hts^ΔG was in the Zuker collection (Koundakjian et al.,2004), and DNA sequencing showed a deletion of a single G in the last part of the Tail domain (G2346 of the ovhts transcript). This results in a frame shift followed by six novel amino acids and a stop codon. Conceptual translation of hts^ΔGresults in a truncated protein that does not contain any of the normal C-terminal domains (Fig. 1B). These mutations are downstream of the entire ShAdd coding sequence.

The phenotypes of these truncation alleles were indistinguishable from the P-element alleles. Both were female sterile and showed a loss of oocyte specification, too few nurse cells, and no actin on RCs (data not shown). However, labeling of hts^ΔG and hts^W532X with Hts antibodies and western analysis showed a distinct difference between the alleles. Even though both truncation alleles should encode the epitope for the htsF antibody, protein was detected only in hts^ΔG (see Fig. S1E,I in the supplementary material). Western analysis showed that whereas hts^ΔG expressed a single truncation product, hts^W532X produced no detectable protein (see Fig. S1M in the supplementary material) and is therefore a null allele. Additionally in hts^ΔG,antibodies 1B1 and htsF labeled a cytoplasmic protein that persisted in egg chambers after they emerged from the germarium, which is never seen in wild type (see Fig. S2 in the supplementary material; data not shown). Therefore,the mutant truncated protein was aberrantly stable in germline cells that normally do not have Ovhts-Fus. Mutant follicle cells labeled with 1B1 antibody showed a significant, but not complete loss of Add½localization to lateral membranes (see Fig. S2D in the supplementary material).

To determine the functional requirements of the different Hts proteins in the germline, we crossed tagged hts transgenes expressed from the otu promoter into both hts^ΔG and hts^W532X mutant backgrounds for rescue experiments. Both P{Ovhts::GFP} and P{Cer::Ovhts} rescued recruitment of Ovhts-RC and actin on RCs (Fig. 4B,C, data not shown). However, other hts phenotypes were not rescued. Labeling with htsF, 1B1 or α-spectrin antibodies showed no fusome-like structure (Fig. 4B,C, data not shown). Anti-GFP labeling in mutants expressing P{Cer::Ovhts} only showed cytoplasmic labeling(Fig. 4A). Additionally, the egg chambers still had too few cells and degenerated. We tested whether the addition of ShAdd would improve rescuing activity. When P{ShAdd-Ven}was expressed alone in hts^ΔG,Venus fluorescence was diffuse in the cytoplasm, and hts phenotypes were not rescued (Fig. 4D). Expression of P{ShAdd::Ven} with either P{Cer::Ovhts} or P{Ovhts::GFP} in a hts^ΔG background showed the same phenotype as the single rescue (Fig. 4E,F): only RCs were rescued, but not the fusome or any of the phenotypes resulting from the loss of the fusome.

Recent work has shown that the fusome precursor, the spectrosome, first begins to form during stage 11 of embryogenesis(Wawersik and Van Doren,2005). As the fusome develops from the spectrosome, it was possible that the otu promoter was not providing Ovhts at an early enough stage. However, earlier expression of Ovhts by driving P{UASH-ovhts::GFP} with either P{nos-GAL4} or P{tub-GAL4} produced the same result as the otu promoter(data not shown).

To determine whether there is a somatic contribution to the htsphenotype that results in our inability to rescue the fusome, we performed clonal analysis with hts^ΔG. Germline clones showed the same phenotype as homozygous mutants, whereas egg chambers that had only follicle cell clones were normal (data not shown). Therefore, the loss of the fusome and RCs is caused solely by the loss of full-length Ovhts in the germline.

During the rescue experiments with P{Ovhts::GFP}, we noticed that localization of Ovhts-RC::GFP to RCs was delayed. In wild-type flies expressing this transgene, GFP was always present in Region 2 of germaria(Fig. 5A). In contrast, in either hts^ΔG(Fig. 5C) or hts^W532X (Fig. 5B) flies expressing P{Ovhts::GFP}, GFP was often absent from germaria and only appeared later. Quantitation of this phenotype revealed that in hts^W532X there was a delay in 30% of germaria, and in hts^ΔG there was a delay in 91% of germaria (Fig. 5D). Thus, in hts mutants the recruitment of Ovhts-RC and actin, and therefore the establishment of RCs, did not occur at the correct developmental stage, suggesting that the fusome is necessary for the timing of RC development.

We took advantage of the ability of S2 cells to cleave Ovhts to identify amino acids necessary for its cleavage. We made a series of small in-frame deletion mutations in the region of the predicted cleavage site, four of which are shown in Fig. 6A, and expressed them in S2 cells. The Δ1 deletion, which removes 20 amino acids in the Tail domain, was cleaved at a wild-type level(Fig. 6B). Deletion Δ2 removed the last 20 amino acids of the Tail domain, and Δ3 removed the last nine amino acids of the Tail and the first 11 amino acids of the RC domain. The Δ2 protein was cleaved, although less efficiently than wild-type protein, whereas the Δ3 protein was not cleaved(Fig. 6B). This result demonstrates that the first 11 amino acids (ALVSQLAQKYA) of the RC domain are required for cleavage.

To study the effect of uncleavable Ovhts in the ovary, we made a P{Ovhts-Δ3::GFP} transgene that was expressed from the otu promoter. Except for the 20 amino acid deletion, this transgene was identical to the P{Ovhts::GFP} transgene. Western analysis of Ovhts-Δ3::GFP from ovary extracts demonstrated that, as in S2 cells,this protein is not cleaved (Fig. 6C). When expressed in wild-type flies, Ovhts-Δ3::GFP was present not only on RCs, but also on the fusome(Fig. 6E,E′). Additionally, the 1B1 antibody, which normally only labels the fusome, now also labeled RCs (Fig. 6E,E″). Therefore, the uncleaved protein localized to both structures where the cleavage products are normally found.

Although wild-type flies expressing Ovhts-Δ3::GFP were fertile and produced apparently normal egg chambers, uncleaved Ovhts did cause a subtle,but completely penetrant dominant defect in the disappearance of the fusome. In wild-type germaria, the fusome begins to disappear where the RCs are starting to form (Fig. 6D,D″,F). This results in unobstructed RCs with fragmented fusome material between them, but not through them. In flies expressing P{Ovhts-Δ3::GFP}, fusome material was present within the RCs as late as stage 2 and 3 egg chambers(Fig. 6G). In some cases,GFP-positive rings were occluded with GFP-negative fusome material that could be visualized with 1B1 antibody (Fig. 6G′,G″ arrowhead). Therefore, at least some of the aberrant fusome contained only wild-type Ovhts-Fus protein and not the N-terminal portion of P{Ovhts- Δ3::GFP}. Additionally,RC rims were thicker than normal, less organized and misshapen(Fig. 6, compare F with G). These results suggest that the cleavage and proper maintenance of the Ovhts domains may play a role in the transition from a fusome to RCs.

We also tested whether the P{Ovhts-Δ3::GFP}transgene could rescue hts mutants, hts^ΔG and hts^W532X (data not shown). As with expression of P{Ovhts::GFP}, P{Ovhts-Δ3::GFP} rescued RCs but not the fusome. The rescued RCs were also labeled with 1B1 antibody indicating that the N-terminus of Ovhts was present.

Our ability to rescue RCs with Ovhts transgenes provided an opportunity to investigate the function of Ovhts-RC. To determine when full-length Ovhts needs to be expressed for RC localization, stage-specific induction of Ovhts expression was done. Wild-type flies carrying a P{UASH-Ovhts::GFP}transgene were crossed to two different Gal4 lines: MTD-Gal4 (see Materials and methods) that induces strong expression throughout oogenesis or P{bam-Gal4} that induces expression only in Region 1 of the germarium. In ovaries from P{UASH-Ovhts::GFP}, MTD-Gal4 flies,Ovhts::GFP was on RCs in all stages of oogenesis(Fig. 7A). However, in ovaries from P{UASH-Ovhts::GFP}; P{bam-Gal4}, Ovhts::GFP was only in the germaria, and on rare occasions on RCs in stage 2 egg chambers(Fig. 7B). Thus, continuous expression of Ovhts-RC protein is needed to maintain localization to RCs.

The continual expression and localization of Ovhts-RC throughout oogenesis indicates that it may be important for actin maintenance at the RC. To test this possibility, we expressed P{UASH-Ovhts::GFP} in hts^ΔG flies using the P{nos-GAL4} driver whose expression is high in the germarium, low during stages 2-6, and then high again starting approximately at stage 7. This allowed rescue of RCs when they form in the germaria of mutants, followed by about a day where little or no new Ovhts protein is produced. As expected Ovhts::GFP was present on RCs in the germaria, absent in mid-stage egg chambers, and present again in later egg chambers within the same ovariole(Fig. 7C). Both the amount of F-actin and its organization at RCs mirrored the presence of Ovhts::GFP(Fig. 7C′). When Ovhts::GFP was present, RCs appeared wild type. In egg chambers lacking Ovhts::GFP, there were no clear F-actin-containing RC rims. There were,however, actin-rich areas that may be disintegrating RC rims(Fig. 7C′,D). Thus,continued expression of Ovhts is needed for the recruitment and/or maintenance of F-actin on RCs.

The Drosophila hts gene is known to play key roles in early oogenesis (Lin et al., 1994; Yue and Spradling, 1992); yet its exact functions have not been characterized completely. We provide evidence that Ovhts is made as a full-length precursor protein that is cleaved, allowing the N-terminal Adducin-like domain (Ovhts-Fus) to localize to the fusome and the C-terminal unique domain (Ovhts-RC) to localize to RCs. Ovhts-Fus is stable only in proliferating cells whereas Ovhts-RC is present only after mitosis is complete, which further emphasizes the importance of post-translational regulation for this protein. We created an uncleavable form of the protein and show that it can dominantly delay early egg chamber development. Finally, expression of transgenes at specific developmental time points shows that Ovhts-RC is necessary for F-actin localization to RCs during all stages of oogenesis.

Our data demonstrate that Ovhts is made as a full-length precursor protein,which is cleaved to form two mature proteins with separate functions. The detection of cleavage products by western blotting and the differential localization of the cleaved proteins in egg chambers provide strong support for this conclusion. We have ruled out the alternative mechanism of producing the downstream protein, Ovhts-RC, by the use of an internal ribosomal entry site (IRES). There is no evidence of an IRES in the sequence of the ovhts mRNA. Furthermore, both of the new hts alleles described in this paper have point mutations 5′ to any potential IRES for Ovhts-RC translation, yet neither mutant produces Ovhts-RC.

Endoproteolytic cleavage is a common method of protein regulation. A number of proteins are made as inactive pro-proteins that are activated by cleavage. In these cases part of the protein is removed and degraded to allow the now mature protein to function. In contrast, both mature Ovhts proteins are functional after cleavage, making Ovhts more analogous to a polyprotein, which are translated as a single polypeptide and subsequently cleaved into different proteins. Two types of polyproteins have been extensively described: those of some RNA viruses and prohormones such as vasopressin (for a review, see Luke and Ryan, 2001). Viral polyproteins are translation products of whole viral genomes, including both replication enzymes and structural proteins(Stanway et al., 2000). Viral polyproteins undergo a number of different cleavages, some of which use polyprotein-encoded proteases, and some of which occur through autoproteolysis(Palmenberg, 1990). The Vasopressin polyprotein includes vasopressin, neurophysin (NP) and a third protein of unknown function. The products of the Vasopressin polyprotein are released by at least two different uncharacterized cellular endoproteases(Acher and Chauvet, 1988).

The Ovhts polyprotein is likely to be cleaved by an endoprotease rather than undergoing autoproteolysis. The active site for autoproteolysis contains one of three amino acids: Ser, Thr and Cys(Perler et al., 1997). The Ovhts cleavage region contains one Ser residue, which when mutated to either a Cys or Ala did not affect Ovhts cleavage (L.N.P. and L.C., unpublished). This suggests Ovhts is not cleaved autoproteolytically. Therefore, the more likely mechanism of Ovhts cleavage is through a cellular endoprotease. Because cleavage can occur in S2 cells as well as the ovary, the protease is likely to be widely expressed; however, the cleavage region does not contain any known protease cleavage sites.

The arrangement of the Ovhts polyprotein is conserved in Drosophila. Twelve Drosophila species have been sequenced to date (Grumbling and Strelets,2006). We annotated the hts homolog from each species,and found that the overall gene organization was highly conserved (L.N.P. and L.C., unpublished). In all cases, an exon encoding the novel RC domain was similarly positioned near the 3′ end of the gene. Whereas the Adducin domains are greater than 90% identical, the RC domain is only 43% conserved,indicating that it is diverging much more quickly. The RC domain is not present as an independent gene or as part of the Adducin gene of other sequenced insect genomes. Thus, it appears that the Ovhts polyprotein was acquired after the divergence of Drosophila.

To explain the existence of an Ovhts polyprotein, we propose a model in which the Ovhts polyprotein functions to provide Ovhts-RC at a critical time and place and in sufficient abundance through its linkage to Ovhts-Fus. At all stages of oogenesis the Ovhts full-length precursor is translated and cleaved. In Region 1 of germaria, the Ovhts-Fus product is maintained and localizes to the fusome. However, the Ovhts-RC product cannot be detected in Region 1,indicating that it is rapidly degraded. Starting in Region 2, the Ovhts-RC domain begins to be maintained and localizes to RCs. This coincides with the completion of mitotic cell cycles, and perhaps also the expression of an Ovhts-RC binding partner required for its stability. Ovhts-RC first appears as puncta around the fusome, suggesting that the cleavage of the polyprotein may take place on or near the fusome. Another possibility is that the Ovhts-RC stabilizing factor is itself associated with the fusome. In either case the Ovhts-RC protein is in close proximity to the arrested cleavage furrows where it is needed to initiate RC development.

An open question is why the transgenes we tested do not restore the fusome in hts mutants. We ruled out the effects of secondary mutations by using several combinations of hts alleles from different backgrounds(i.e. P-element insertions, point mutations and deficiencies), and we deem the existence of another protein isoform unlikely as we performed 3′ RACE and found only the known transcripts. The most likely possibility is that expression of ovhts is required at a specific time early in development for the formation and stability of spectrosomes in germline progenitor or stem cells, and we did not recapitulate this expression pattern in the rescue experiments.

The presence of the uncleavable Ovhts-Δ3::GFP protein affects the transition from a fusome to an RC, even though it localizes to both structures. Therefore, cleavage is not required for localization but is necessary for the full function of the Ovhts polyprotein products. The expression of uncleavable Ovhts-Δ3::GFP causes the aberrant persistence of fusome material in RCs. Strikingly, the fusome material includes endogenous wild-type Ovhts-Fus protein that has failed to degrade. As it is known that mammalian Adducins function as heterotetramers, a direct interaction of wild-type Ovhts-Fus or ShAdd with Ovhts-Δ3::GFP may occur and result in pulling fusome material into the lumen of the RCs. This hypothesis is supported by the fact that RCs in hts mutants that lack wild-type Ovhts-Fus are rescued by Ovhts-Δ3::GFP and do not have fusome plugs or misshapen rims. The dominant-negative effect of Ovhts-Δ3::GFP on wild-type RCs disappears by stage 3, suggesting that uncleaved Ovhts is tolerated in egg chambers that have degraded the fusome. Therefore, it is the transition from fusomes to RCs that is particularly sensitive to the presence of uncleaved Ovhts.

RC formation has been described as a multi-step process beginning with the arrested cleavage furrow followed by the recruitment of Filamin, Ovhts-RC and F-actin, and completed with the addition of other actin binding proteins(Hudson and Cooley, 2002). We found that the recruitment of Ovhts-GFP to RCs is delayed in htsmutants lacking a fusome, supporting a role for the fusome (and possibly prelocalization of Ovhts-RC) in the initiation of RC development. Furthermore,without continual expression of Ovhts-RC(Fig. 7), F-actin was lost from the inner rim of RCs. As the F-actin cytoskeleton of RCs is highly dynamic(Kelso et al., 2002), the absence of Ovhts-RC in hts mutant RCs may tip the balance of actin dynamics toward depolymerization, resulting in the depletion of RC F-actin. Thus, Ovhts-RC is needed continuously either to stabilize polymerized actin or to promote actin polymerization.

We thank Howard Lipshitz for providing us with hts cDNA constructs. We thank Trudi Schüpbach for the hts^W532Xallele, and Charles Zuker and Edmund Koundakjian for the hts^ΔG allele. We also thank Vann Bennett for mammalian MARCKS domain antibody. We are grateful to all members of the Cooley lab for help with editing and insightful conversation. This work was supported by a NSF predoctoral fellowship to L.N.P., a Burroughs Wellcome Fund Postdoctoral Fellowship of the Life Sciences Research Foundation to T.S.,and a National Institutes of Health grant (GM52702) to L.C.