The localization of oocyte-specific determinants in the form of mRNAs to the pro-oocyte is essential for the establishment of oocyte identity. Localization of the Bicaudal-D (Bic-D) protein to the presumptive oocyte is required for the accumulation of Bic-D and other mRNAs to the pro-oocyte. The Bic-D protein contains four well-defined heptad repeat domains characteristic of intermediate filament proteins, and several of the mutations in Bic-D map to these conserved domains. We have undertaken a structure-function analysis of Bic-D by testing the function of mutant Bic-D transgenes (Bic-DH) deleted for each of the heptad repeat domains in a Bic-D null background. Our transgenic studies indicate that only the C-terminal heptad repeat deletion results in a protein that has lost zygotic and ovarian functions. The three other deletions result in proteins with full zygotic function, but with affected ovarian function. The functional importance of each domain is well correlated with its conservation in evolution. The analysis of females heterozygous for Bic-DH and the existing alleles Bic-DPA66 or Bic-DR26 reveals that Bic-DR26 as well as some of Bic-DH transgenes have antimorphic effects. The yeast two-hybrid interaction assay shows that Bic-D forms homodimers. Furthermore, we found that Bic-D exists as a multimeric protein complex consisting of Egl and at least two Bic-D monomers.
IN early Drosophila oogenesis, a series of highly controlled divisions initiate the developmental pathway that leads to the formation of an oocyte (Spradling 1993). A cluster of 16 interconnected cystocytes is formed through four successive mitotic divisions of a cystoblast. One cystocyte develops as an oocyte and the other 15 become polyploid nurse cells. Although the mechanisms underlying oocyte determination are unclear, the initial establishment of oocyte identity is dependent on the accumulation of oocyte-specific transcripts and proteins.
The first requirement for Bicaudal-D (Bic-D) function is in the initial establishment of oocyte identity. In the strong hypomorphic alleles Bic-DR26 and Bic-DPA66, the polarized microtubule network does not reorganize, and oocyte-specific transcripts such as Bic-D, orb, osk, and fs(1)K10 fail to differentially accumulate in the presumptive oocyte; all 16 cystocytes become polyploid nurse cells. Egg chambers with 16 nurse cells grow to about stage 6 and then degenerate (Ephrussiet al. 1991; Suter and Steward 1991; Lantzet al. 1992; Theurkaufet al. 1993; Ranet al. 1994; Mach and Lehmann 1997). Another gene, egalitarian (egl), has the same loss-of-function phenotype as Bic-D; in egl ovaries, determinants also fail to localize to the prospective oocyte. The phenotypes of both mutants clearly indicate that the genes are involved in the localization of determinants in the form of mRNA and protein. Bic-D and Egl proteins are found in a complex (Mach and Lehmann 1997) and are likely to function together, but how they do it remains to be elucidated.
Sequence analysis of Bic-D protein reveals that it is similar to the myosin heavy chain tail and to intermediate filament proteins such as lamin and desmin (Suteret al. 1989; Wharton and Struhl 1989). These proteins contain domains of extensive heptad repeats in which the hydrophobic residues are preferentially located at positions 1 and 4 (McLachlan and Karn 1983). These heptad repeats usually mediate the packaging of one helix against another, forming coiled-coil structures made up of two or three protein molecules and resulting in homo- or heterodimers, or in multimers. The formation of coiled-coil structures can also result from the folding of one molecule in such a way that two heptad repeat domains come to lie parallel to each other (Cohen and Parry 1986).
Proteins containing coiled-coil domains have many functions and have been identified in all cell compartments. In most of these proteins, the coiled-coil domains are flanked by protein domains that control the protein's distribution or specific function. More than half of the Bic-D protein consists of heptad repeats distributed throughout, and no other protein motifs are apparent. The repeats may be classified into four well-defined regions, each greater than seven heptad repeats. Interestingly, two dominant and two recessive mutations map to these heptad domains (Wharton and Struhl 1989; Suter and Steward 1991).
To test the functional importance of the heptad repeat domains in Bic-D, and to further understand its function, we have undertaken a structure–function analysis of the Bic-D protein. We generated mutant Bic-D proteins in which each of the conserved heptad repeat domains was deleted. We assayed the functional importance of each of the four domains by expressing the mutant proteins in transgenic flies lacking endogenous Bic-D function. We found that the first and last heptad repeat domains of the Bic-D protein are essential for function, but the two centrally located domains, the second and third heptad repeat domains, are functionally less important or dispensable. Complementation tests reveal that one of the existing hypomorphic alleles, Bic-DR26, and some of the heptad-repeat-deleted transgenes have a strong antimorphic effect. Furthermore, we found that Bic-D can form homodimers and Bic-D protein exists in a heteromeric protein complex. Our results confirm that the N-terminal half of the protein may be required exclusively for function during oogenesis, while the C-terminal domain contains the sequences required for all functional aspects. The functional importance of each domain is well correlated with its conservation during evolution.
MATERIALS AND METHODS
Fly strains: The Bic-Dnull alleles (Bic-Dr5 and Bic-Dr11) are described by Ran et al. (1994). The Bic-DR26 allele and Df(2L) TW119 (Df119), which uncovers the Bic-D locus, are described by Mohler and Wieschaus (1986). The Bic-DPA66 allele is described by Schüpbach and Wieschaus (1991).
Construction of Bic-DH mutants: The classification of the four heptad repeat domains was performed by Coils programs (http://www.ch.embnet.org/software/COILS_form.html) and confirmed by visual inspection. The first heptad domain extends from amino acid (aa) 163 to aa 242, the second from aa 327 to aa 382, the third from aa 392 to aa 444, and the fourth from aa 610 to aa 743. All deletion constructs were derived from a 4-kb Bic-D genomic promoter fragment fused in the 5′ untranslated region to the full-length Bic-D cDNA in Bluescript SK+ (K. Baksa and R. Steward, unpublished results). To delete each heptad repeat domain, Not I sites were created adjacent to the sequences to be deleted by in vitro mutagenesis using the Bio-Rad (Hercules, CA) kit (Figure 1B). The mutagenized DNA was digested with NotI and religated. The deletions were checked by DNA sequencing. The ~8-kb minigenes were subcloned into the pCaSpER transformation vector and were introduced into the Drosophila genome by P-element-mediated transformation (Robertsonet al. 1988). The transgenic lines are named according to the position of the deletion—Bic-DH1, Bic-DH2, Bic-DH3, and Bic-DH4. Bic-DH refers to all four transgenes. At least two independently transformed lines for each deletion construct were tested.
To add the triple hemagglutinin (HA) epitope to the wild-type Bic-D protein, a unique NotI site was introduced at the C terminus of the Bic-D cDNA by insertional mutagenesis. The HA epitope, a 117-bp NotI fragment, was ligated into the NotI site and cloned into the pCaSpER vector. The same endogenous Bic-D promoter that was used for the heptad-repeat-deleted Bic-D transgenes was utilized.
Yeast two-hybrid interaction assay: The full-length Bic-D cDNA was inserted in frame into the pAS1-CYH2 vector (containing amino acids 1–147 of Gal 4; GalBD) or the pACTII vector (containing amino acids 768–881; Gal4AD). As host, strain CBY 14 of Saccharomyces cerevisiae, carrying LacZ (β-galactosidase) and HIS (imidazole glycerol phosphate dehydratase) reporter genes, was used. Yeast was transformed by the lithium acetate method of Gietz et al. (1992). To ensure expression of fusion protein produced from the vectors, yeast lysates from colonies transformed with the fusion constructs were analyzed by anti-Bic-D Western blotting. For quantitative β-galactosidase assays in solution, three independently derived transformants containing appropriate plasmids were selected and grown in selective medium overnight at 30° in a shaking incubator. β-Galactosidase assays were carried out using O-nitrophenyl-β-d-galactopyranoside as a substrate according to the protocols provided by Clontech (Palo Alto, CA). The β-galactosidase activity (reported as Miller units) was determined by calculating the mean from three independently isolated transformants.
Germline clones: Bic-DH4 was introduced into the Bic-Dnull FRT40A background. The Bic-Dnull FRT40A; P[Bic-DH4] females were crossed with hsFLP;ovoD FRT40A; P[Bic-DH4] males. As a negative control, Bic-Dnull FRT40A females not carrying P[Bic-DH4] were used. The progeny were collected for 24 hr and then heat shocked twice for 2 hr at 37° over a period of 2 days. Bic-Dnull FRT40A/ovoD FRT40A; P[Bic-DH4]/P[Bic-DH4], Bic-Dnull FRT40A/ovoD FRT40A; P[Bic-DH4] or Bic-Dnull FRT40A/ovoD FRT40A virgins were selected, crossed to wild-type males, and allowed to lay eggs for 5 days. Neither genotype laid any eggs and the morphology of ovaries was indistinguishable. For the analysis of germline clones in a wild-type background, the Bic-Dnull FRT40A; P[Bic-DH4] females were crossed with hsFLP; P[arm-LacZ] FRT40A males.
Immunoprecipitation: For each immunoprecipitation, 80 pairs of ovaries of appropriate genotypes were dissected in 1× PBS and homogenized with a Teflon pestle in 200 μl of extraction buffer [20 mm Hepes (pH 7.6), 300 mm NaCl, 1 mm EDTA, 0.5% Triton-X 100] supplemented with protease inhibitors. The homogenized ovary extracts were centrifuged at 13,000 rpm after incubation on ice for 30 min. The pellet was discarded, and the supernatant was recentrifuged two times. The resulting supernatant was incubated with 30 μl of a 1:1 suspension of protein A-Sepharose beads (Pharmacia, Piscataway, NJ) on a rotating wheel at 4° for 1 hr as a preclearing step. The beads were spun down and discarded. The precleared supernatant was incubated with 20 μl of anti-HA antibodies 12CA5, or 40 μl of a 1:1 mixture of hybridoma supernatants anti-Bic-D 1B11 and 4C2 antibodies, or 10 μl of rabbit anti-Egl antibodies (a kind gift from Ruth Lehmann), or 10 μl of monoclonal anti-α-tubulin antibodies (Sigma, St. Louis) on a rotating wheel at 4° for 1 hr, followed by the addition of 10 μl of a 1:1 suspension of protein A-Sepharose (Pharmacia) and incubation for 1 hr. The beads were washed three times with extraction buffer for 15 min, washed two times with HNET wash buffer [50 mm Hepes (pH 7.6), 250 mm NaCl, 5 mm EDTA, 0.1% Triton-X 100], washed once in HNE wash buffer [10 mm Hepes (pH 7.6), 150 mm NaCl, 1 mm EDTA), and resuspended in 40 μl of 2× SDS loading buffer.
Densitometric analyses: Densitometry of chemiluminographs of immunoblots was performed with a laser scanning chromosome 3 densitometer (Joice Loebl). For quantitative densitometric analysis, 2 μl of ovary extracts was subjected to immunoblotting using anti-Bic-D as the primary antibody and sheep anti-mouse horseradish-peroxidase-conjugated antibody as the secondary antibody. Bic-D proteins were detected using a chemiluminescence detection system (Pierce Chemical, Rockford, IL). The relative levels of Bic-DHA40, Bic-DH1, and Bic-DH4 were determined by comparing the densitometric values obtained for Bic-D from one copy of a transgene to the densitometric value obtained for one copy of endogenous Bic-D gene in the same lane. The ratio of the densitometric values obtained for transgenic Bic-D to the densitometric values obtained for endogenous Bic-D remained constant, with minor standard deviation when serially diluted ovary extracts were analyzed (data not shown), indicating that the densitometric values for transgenic Bic-D proteins reflect differences in the levels of these proteins. Data from four independent experiments were used to determine the relative transgenic Bic-D protein levels.
The molecular characteristics of heptad-repeat-deleted Bic-D proteins Bic-DH1, Bic-DH2, Bic-DH3, and Bic-DH4: A schematic representation of the Bic-D protein and the location of the heptad repeats, as well as existing mutations, are shown in Figure 1A. To investigate the functional importance of the four distinct heptad domains, each domain was deleted by in vitro mutagenesis (for details see materials and methods). The extent of the deletions is shown in Figure 1B with the nomenclature of the transgenic deletion lines. The deleted Bic-D genes were expressed under the control of an endogenous genomic Bic-D promoter. At least two lines for each deletion were established.
The protein product of each of the deleted transgenes was determined by Western blot of ovary extracts from heterozygous females carrying one copy of wild-type Bic-D and one copy of a transgene. Figure 4A shows a representative Western blot of extracts from wild type (lane 1) and one line of each transgene (lanes 2–5). Despite the large size of the deletions, all four proteins were expressed and the migration of each protein agrees well with the size of the deletion.
The first and last heptad repeat domains are essential for Bic-D function: Hypomorphic Bic-D alleles or germline clones of null alleles are female sterile, and homozygous Bic-D null mutants die as late pupae or soon after eclosion (Ranet al. 1994). To investigate the function of our deletion mutants, we first examined whether they can rescue the zygotic lethality of a Bic-Dnull mutation. Each transgene was crossed into a Bic-Dr5/Df119 background. Bic-DH1, Bic-DH2, and Bic-DH3 rescue viability at rates ranging from 85 to 95%, but Bic-DH4 fails to rescue the zygotic lethality (Table 1), indicating that only the fourth heptad repeat domain is required for the zygotic function of Bic-D.
We also examined whether the deleted proteins can rescue the female-sterile phenotype of a Bic-Dnull mutation. In Bic-Dr5/Df119 egg chambers, oocyte differentiation fails to occur and all 16 cystocytes in the egg chamber become polyploid nurse cells (Ranet al. 1994). Bic-DH1 fails to complement because Bic-DH1 (P[Bic-DH1]; Bic-Dr5/Df119) females show a similar phenotype as Bic-Dnull females, although an egg is produced very occasionally (one over the entire life span of 10 females). These rare eggs show strongly ventralized chorion phenotypes and do not develop (Table 1, Figure 2D).
Bic-DH2 displays some complementation; the fertility of the females is ~20% that of wild type (Table 1). A total of 21% of eggs laid by Bic-DH2 (P[Bic-DH2]; Bic-Dr5/Df119) females display a variety of ventralized chorion phenotypes ranging from partially fused appendages to no appendages. In rare cases, small eggs, collapsed eggs, and eggs with an open chorion are observed (Table 1, Figure 2). Abnormal and strongly ventralized eggs are not fertilized, but the remaining eggs develop normally.
Bic-DH3 complements strongly because Bic-DH3 (P[Bic-DH3]; Bic-Dr5/Df119) females display almost wild-type fertility, but ~1% of the eggs show a weakly ventralized dorsal appendage phenotype (Tables 1 and 3). The rescue of zygotic lethality as well as sterility indicates that Bic-DH3, despite missing 70 amino acids, surprisingly retains nearly full function.
The function of Bic-DH4 during oogenesis was analyzed in two ways. First, the Bic-DH4 transgene was crossed with the two hypomorphic alleles Bic-DPA66 or Bic-DR26. These alleles are viable, but produce a strong female-sterile, 16-nurse-cell phenotype (Suter and Steward 1991). One or even two copies of Bic-DH4 did not improve the phenotype of hemizygous females of both genotypes. These results support that Bic-DH4 also loses its ovarian function.
Second, we tested if Bic-DH4 could rescue the phenotype of homozygous Bic-Dnull germline clones. If Bic-DH4 protein provides ovarian function, the germline clone of homozygous Bic-Dnull alleles should improve oocyte development and possibly result in the formation of oocytes or mature eggs in the ovoD background. We generated mosaic clones homozygous for the Bic-Dnull allele in the presence or absence of a Bic-DH4 transgene, using the ovoD/FRT technique (Chouet al. 1993; Chou and Perrimon 1996). Both mosaic egg chambers failed to produce mature eggs, and the dissected ovaries of both genotypes were indistinguishable.
To clearly identify mutant egg chambers, we induced germline clones in a wild-type background, using armlacZ as a marker. Mosaic ovaries were double stained with anti-β-galactosidase antibodies, identifying germline clones by the absence of LacZ staining, and the DNA stain Yo-Pro, allowing the examination of the ploidy of nuclei in mutant egg chambers. Control egg chambers (total >80) of the genotype (hsFLP; + FRT40A/arm-LacZ FRT40A) always contained 15 polyploid nurse cells and 1 diploid oocyte (Figure 3, A and B). However, Bic-Dnull germline clones (total >80) in the presence of one or two copies of Bic-DH4 showed no difference in phenotypes when compared to Bic-Dnull germline clones (>70) in the absence of Bic-DH4 (Figure 3, C–E). The failure to complement Bic-DPA66 and Bic-DR26, as well as the analysis of Bic-Dnull egg chambers in the presence or absence of Bic-DH4, show that Bic-DH4 protein does not retain any ovarian function.
Bic-D protein levels and function: Western blot analysis shows that protein levels of Bic-DH2 and Bic-DH3 are similar to those expressed by one copy of the endogenous gene, but the level is decreased in Bic-DH1 and more severely decreased in Bic-DH4 (Figure 4A). Because the expression levels of Bic-DH1 and Bic-DH4 are significantly lower than that of the endogenous Bic-D, the loss of function of Bic-DH1 and Bic-DH4 could be caused by either the functional importance of the deleted heptad domains or the lower levels of the deleted proteins. To distinguish between these possibilities, we needed to determine the lowest level of Bic-D that supports normal function in the zygote and the ovary. We took advantage of a hemagglutinin (HA)-tagged wild-type transgene (Bic-DHA40; see materials and methods) that produced low levels of protein but rescued both the zygotic and ovarian phentoype.
We performed Western blots of ovaries from females heterozygous for Df119, which uncovers the Bic-D locus, carrying one copy of a transgene (Figure 4B). The relative levels of Bic-DHA40, Bic-DH1, and Bic-DH4 were determined by comparing the densitometric values obtained for Bic-D expressed by the transgene to the values obtained for the endogenous Bic-D. Densitometric analysis showed that the amount of Bic-DHA40 expressed by one copy of a transgene was ~12% of wild-type Bic-D protein expressed by one copy of endogenous Bic-D. The amount of Bic-DH1 was 14% and the amount of Bic-DH4 was 9% of wild-type Bic-D protein. Two copies of Bic-DH1 or Bic-DH4 resulted in an approximately twofold increase in abundance of the deleted proteins, as determined by densitometric analyses (Figure 4C).
Two copies of Bic-DH4 did not rescue the zygotic and ovarian Bic-Dnull phenotype, and two copies of Bic-DH1 failed to rescue the ovarian-null phenotype. Because two copies of these transgenes produce significantly more protein than one copy of Bic-DHA40 that has wild-type function, the loss of function of Bic-DH1 or Bic-DH4 is most likely caused by the deletion of an essential domain, not by insufficient levels of proteins.
Bic-DR26 acts as an antimorph: The rescue of female sterility of a Bic-Dnull mutation by Bic-DH2 and Bic-DH3 indicates that these transgenes retain some function. To investigate if they show intragenic complementation, we examined the interactions between Bic-DH2, Bic-DH3, Bic-DPA66, and Bic-DR26.
Both the fertility and the frequency of ventralized eggs were the same in heteroallelic Bic-DH2/Bic-DPA66 (P[Bic-DH2]; Bic-DPA66/Df119) females as in hemizygous Bic-DH2 (P[Bic-DH2]; Bic-Dr5/Df119) females. However, when Bic-DR26 was introduced into the Bic-DH2 background, strong enhancement of the Bic-DH2 phenotype was observed (Figure 5A). First, the heteroallelic Bic-DH2/Bic-DR26 (P[Bic-DH2]; Bic-DR26 /Df119) females rarely lay eggs, while the hemizygous Bic-DH2 females lay 20% of expected eggs. Second, the few eggs laid are virtually all strongly ventralized in that the eggs have either a single appendage or remnants of appendage materials, while the eggs laid by hemizygous Bic-DH2 females range from wild type to strongly ventralized, with ~20% showing a ventralized phenotype (Figure 2). Third, the occurrence of abnormal eggs, such as collapsed eggs, small eggs, and eggs with open chorion, is increased. In Bic-DH2, abnormal eggs are observed at a low frequency (2%), but in Bic-DH2/Bic-DR26, abnormal eggs can reach up to 20% of eggs laid (data not shown).
We also examined interaction of Bic-DH3 with Bic-DPA66 and Bic-DR26 (Figure 5B, Table 2). The heteroallelic Bic-DH3/Bic-DPA66(P[Bic-DH3]; Bic-DPA66/Df119) females exhibit similar fertility and eggshell phenotype as hemizygous Bic-DH3 (P[Bic-DH3]; Bic-Dr5/Df119). The Bic-DH3 /Bic-DR26 (P[Bic-DH3]; Bic-DR26/Df119) females display similar fertility as Bic-DH3 hemizygotes. However, the frequency of ventralized eggs is significantly increased in the Bic-DH3/Bic-DR26 females, although the degree of ventralization remains weak: almost 100% of the eggs from transheterozygous females are ventralized, while only ~1% of the eggs laid by the Bic-DH3 females display ventralized eggshells.
The observed strong phenotypic enhancement seen in the Bic-DH2/Bic-DR26 and Bic-DH3/Bic-DR26 females suggests that Bic-DR26 is antimorphic because it interferes with the function of Bic-DH2 and Bic-DH3. If this is true, females carrying two copies of Bic-DR26 in the presence of one copy of Bic-DH3 should show a more severe phenotype. As shown in Table 2, more severe phenotypes are indeed observed. First, the fertility of such females was decreased to ~19%, compared to 98% observed in females carrying one copy of each allele. Also, the occurrence of abnormal eggs, such as collapsed eggs and small eggs, is increased. For example, the frequency of collapsed eggs is up from 2% in the Bic-DH3/Bic-DR26 females to 43% in females carrying two copies of Bic-DR26 and one copy of Bic-DH3.
The antimorphic effect of Bic-DR26 in combination with Bic-DH2 and Bic-DH3 could be due to a specific interaction between the deleted proteins and Bic-DR26 because Bic-DR26/+ females do not show a female-sterile phenotype. Alternatively, Bic-DR26 could also interact with wild-type Bic-D, with one copy of wild-type Bic-D producing protein levels high enough to suppress the antimorphic effect of Bic-DR26. To address this issue, we investigated whether Bic-DR26 shows an antimorphic effect in the presence of low levels of wild-type protein. We introduced one copy of Bic-DR26 into Bic-Dnull females carrying one copy of Bic-DHA40 (P[Bic-DHA40]; Bic-Dr5/Df119) that behave like wild type despite a low level of Bic-DHA40 expression. Bic-DR26/Bic-DHA40 females lay normal numbers of eggs, but virtually all eggs are ventralized, which is similar to eggs laid by Bic-DH3/Bic-DR26 females (Figure 5, B and C). These results suggest that Bic-DR26 has a general antimorphic effect.
Interactions among the Bic-DH transgenes: The antimorphic effect of Bic-DR26 suggests that some of the heptad-repeat-deleted transgenes may also act in a similar fashion. Therefore, we tested for complementation between the heptad-repeat-deleted Bic-D transgenes and found that Bic-DH1 and Bic-DH4 have antimorphic effects.
Bic-DH1 enhances the phenotype of Bic-DH2 and Bic-DH3 females (Table 3). The Bic-DH1/Bic-DH2 females lay fewer eggs than hemizygous Bic-DH2 females, and all eggs are ventralized. The degree of ventralization becomes more severe: 55% of ventralized eggs laid by Bic-DH1 /Bic-DH2 females display a strong eggshell phenotype, while only 19% of ventralized eggs laid by Bic-DH2 females exhibit a strong eggshell phenotype. The Bic-DH1 /Bic-DH3 females lay fewer eggs (69% of expected) than Bic-DH3 females (97% of expected), and the frequency of ventralized eggs is increased as the phenotype becomes stronger.
Like Bic-DH1, Bic-DH4 also produces antimorphic effects in hemizygous Bic-DH2 or Bic-DH3 backgrounds (Table 4). For example, the Bic-DH4/Bic-DH2 females do not lay eggs, while hemizygous Bic-DH2 females lay 27% of expected eggs. The Bic-DH4/Bic-DH3 females lay fewer eggs (18% of expected) than hemizygous Bic-DH3 females.
Bic-D exists in a multimeric protein complex: Given the presence of extensive heptad repeat domains in Bic-D, which usually serve as protein-protein interaction domains, as well as the antimorphic effect of Bic-DR26, Bic-DH1, and Bic-DH4, we postulated that Bic-D may exist in a multimeric protein complex. To test this possibility, we performed coimmunoprecipitation experiments of ovary extracts containing both wild-type and HA-tagged Bic-D proteins. Ovary extracts of flies containing one copy of endogenous Bic-D and one copy of HA-tagged Bic-D were immunoprecipitated with anti-HA antibodies. Immunoprecipitated proteins were analyzed by Western blot and probed with anti-Bic-D antibodies. The extracts from the transgenic flies revealed two bands, one of the size expected for Bic-D, and the other, a novel band just above the endogenous Bic-D, corresponding to HA-tagged Bic-D (Figure 6, lane 1). Control wild-type extracts without a transgene contained only the Bic-D band (Figure 6, lane 2). Reprobing of the same immunoblot with anti-HA antibodies detected only the novel band in the transgenic extracts and did not show any band in the wild-type extracts (Figure 6, lanes 1′ and 2′). These results show that the anti-HA antibody is capable of immunoprecipitating the tagged Bic-D as well as the endogenous Bic-D, indicating that Bic-D forms a multimeric complex.
The ability of Bic-D to form a multimeric complex is not affected in Bic-DR26. Anti-HA antibodies precipitate both HA-tagged Bic-D and Bic-DR26 proteins in ovary extracts from HA-Bic-D/Bic-DR26 females (Figure 7A, lane 5). Bic-D and Bic-DR26 have been shown to be coimmunoprecipitated with anti-Egl antibodies (Mach and Lehmann 1997). Consistent with this, we found that Egl is also present in immunoprecipitates of anti-HA antibodies (Figure 7B, lanes 4 and 5). These results suggest that the Bic-D/Egl complex contains at least two Bic-D monomers.
Our genetic and immunoprecipitation results suggest that Bic-D exists in vivo as a heteromeric protein complex that contains Egl and at least two Bic-D monomers, but it is unclear whether Bic-D self-associates. We tested whether Bic-D can form homodimers by yeast two-hybrid interaction. Full-length Bic-D cDNA was fused to the DNA-binding domain (BD) and the transcriptional activation domain (AD) of Gal4 to produce Gal4BD-Bic-D and Gal4AD-Bic-D fusion proteins. Cotransformation of Gal4BD-Bic-D and Gal4AD-Bic-D resulted in growth of the yeast host on histidine-minus medium and high activation of β-galactosidase activity, 100- to 120-fold above baseline (Table 5). Neither Gal4BD-Bic-D nor GalAD-Bic-D alone showed growth on the selective medium, nor did they show detectable β-galactosidase activity. The Bic-D-Bic-D interaction is apparently specific because Bic-D failed to interact with either Gal4BD or Gal4AD alone. These results indicate that Bic-D can interact with itself, forming homodimers.
The functional importance of the heptad repeat domains and their conservation: The transgenic studies of the deleted Bic-D proteins show that the two terminal heptad repeat domains of Bic-D protein are essential for function and that the two centrally located domains are functionally less important or dispensable (Table 1). The first heptad domain is required exclusively for ovarian function: Bic-DH1 rescues the zygotic lethality of a Bic-Dnull mutation, but fails to complement female sterility. The last heptad domain is essential for zygotic as well as ovarian function: Bic-DH4 fails to rescue the zygotic lethality and the female sterility of a Bic-Dnull mutation. The second heptad domain retains only partial function: Bic-DH2 rescues the zygotic lethality, but females lay only 20% as many eggs as wild type and show eggshell phenotypes. The third heptad repeat deletion retains almost full wild-type function in both the ovary and the zygote. Interestingly, we did not identify any domain that functions specifically in the zygote, suggesting that the ovarian function of Bic-D may represent a special adaptation.
We determined that levels as low as 12% of wild-type Bic-D protein are enough to provide full zygotic and ovarian function. All the deleted transgenes produce in two copies ≥18% of wild-type protein levels, which are high enough to rescue the mutant phenotypes if the proteins were fully functional. We conclude, therefore, that the phenotypes observed by the four deleted transgenes are not due to low levels of protein, but to the essential function of the deleted domains.
The functional importance of each heptad repeat domain is well correlated with its conservation during evolution (Figure 1B). A Bic-D homologue has been sequenced in humans (Baens and Marynen 1997). The first and last heptad domains show the highest homology in flies and humans, 80 and 89%, respectively. The second heptad domain shows an intermediate homology (62.5%), while the third heptad domain is the least conserved (49%). The high conservation of heptad repeats 1 and 4 suggests that human Bic-D may also have functional homology and that it may have, like the Drosophila gene, two distinct functions.
Structural organization of Bic-D protein: Existing alleles of Bic-D agree with our observation that the two functionally essential domains of Bic-D reside at each end of the protein, with the center possibly having a spacer function. The mutation in Bic-DPA66 maps to the N-terminal domain. It is an alanine-to-valine change at aa 40 and eliminates the phosphorylation of a nearby serine. This phosphorylation is essential only for the ovarian function of the protein (Suter and Steward 1991).
Bic-Dr11, a complete loss-of-function allele, has a single aa change of lysine730 to methionine located in the fourth heptad domain. The importance of this domain is further supported by the phenotype of Bic-D71.34. In this dominant mutation, a single aa change of isoleucine689 to proline (Wharton and Struhl 1989) may result in a novel interaction of Bic-D with other proteins, specifically in late oogenesis and early embryogenesis.
Another maternal effect allele, Bic-DR26, was isolated as a revertant of the dominant Bic-D71.34 allele and contains a deletion of four amino acids in the second heptad domain in addition to the single aa change in the fourth heptad repeat domain. In Bic-DR26, the dominant embryonic phenotype is completely suppressed, but it has a much stronger ovarian phenotype than Bic-DH2. This difference may be due to our deletion eliminating the entire conserved heptad repeat region, leaving no partial repeats, unlike the deletion in Bic-DR26, which leaves behind three amino acids of a heptad repeat. It has been suggested that a deletion of four residues in heptad repeats could cause the local overwinding of the coiled-coil structure, resulting in a global effect (Brownet al. 1996).
No alleles that map to the third heptad repeat domain have ever been identified, supporting the notion that this domain has a largely redundant function.
On the basis of all these observations, it appears that the two termini of Bic-D serve as effector domains of the protein, with the C-terminal domain being essential for all functional aspects of Bic-D. Given that two ovary-specific mutations and the first and second heptad deletions affect only the ovarian function, a large part of the N-terminal half of the protein may specifically function in the ovary by interacting with ovary-specific factors. The second and third heptad repeat domains seem to mainly act as a linker that connects effector domains at both ends. Deleting part of this linker apparently can be tolerated.
A functionally similar protein organization is found in the yeast protein Zip1p, a component of synaptonemal complex (SC; Sym and Roeder 1995). Sequence analysis predicts that it encodes a protein consisting of an extensive helical coiled-coil domain (from aa 180 to aa 748) flanked by a globular domain at both termini (aa 1–179 and aa 749–875). Deletion analyses of the protein suggest that the coiled-coil domain may act as a spacer between the two effector domains at both termini because the protein is able to tolerate a large internal deletion and retain its function (Sym and Roeder 1995; Tung and Roeder 1998). A proposed model for the organization of Zip1p protein within the SC is that Zip1p protein forms a dimer and, in turn, two dimers form a tetramer that traverses the width of SC from one lateral element to the other lateral element (Tung and Roeder 1998).
Like Zip1, Bic-D has its functional domains at the ends of the protein, even though they are not structurally distinct. Given that Bic-D is required in oogenesis for the localization of mRNAs, it may function as a linker between the cytoskeletal network and the RNA localization machinery. Further experiments are necessary to investigate if indeed the central domain of Bic-D functions as a linker, or if it is largely redundant for the function of the protein.
Bic-D exists in a heteromeric protein complex: We found that Bic-DR26, Bic-DH1, and Bic-DH4 show antimorphic effects in hemizygous Bic-DH2 or Bic-DH3 backgrounds. The anti-HA antibody is capable of immunoprecipitating the HA-tagged Bic-D as well as endogenous Bic-D and Egl from ovary extracts, suggesting the existence of a multimeric Bic-D protein complex containing Egl and at least two Bic-D monomers. Because HA-Bic-D, Bic-DR26, and Egl can be coimmunoprecipitated, the antimorphic effect of Bic-DR26 protein is most likely caused by the formation of nonfunctional Bic-D/Egl complex. By yeast two-hybrid interaction analyses, we found that Bic-D can form homodimers in vivo. Consistent with this, we also found that Bic-D was eluted from ovary extracts in fractions with an estimated molecular mass of 200 kD, the size of a dimer, on sucrose density gradients (data not shown); a similar elution profile was also reported for bacterially purified Bic-D protein (Stuurmanet al. 1999).
It is likely that Bic-D functions as a dimer in the zygote as well as during oogenesis; therefore, Bic-DH1, Bic-DH2, and Bic-DH3 may retain the capability of forming homodimers. Because Bic-DH4 enhances the phenotype of hemizygous Bic-DH2 or Bic-DH3, it probably also retains its capacity to dimerize. These observations suggest that Bic-D does not contain a defined sequence responsible for dimerization; rather, it appears that the coiling of the protein occurs along the entire length of the protein, and the protein may form a rod.
Although our results support that Bic-D forms a heteromeric protein complex, many questions remain concerning the Bic-D/Egl complex and its exact function. The structure of the Bic-D protein suggests that it may represent a linker between an RNA-binding protein and the cytoskeleton or between an RNA-binding protein and a motor. Alternatively, Bic-D may be an integral part of the cytoskeletal network essential for the integrity of many different cell types. Further biochemical purification and characterization of the Bic-D/Egl complex from ovary extracts will help us to address these questions.
A role of Bic-D in the formation of dorsal-ventral eggshell axis: The distribution and level of Gurken (Grk) activity in the oocyte is central to establishing the dorsal-ventral pattern of the egg and eggshell (Schüpbach and Roth 1994; Ray and Schüpbach 1996). Many genes function to localize Grk activity to the dorsalanterior side of the growing oocyte, and there are many that control the translation of grk mRNA. It appears that there is a close correlation between the levels of Bic-D and grk function: the fertility of females directly correlates with the severity of the ventralized egg phenotype. In Bic-DH1 that fails to complement the female sterility of a Bic-Dnull allele, the occasional eggs laid are strongly ventralized. In Bic-DH2 that shows intermediate complementation, a range of ventralized eggshell phenotypes are observed in ~20% of eggs laid. In Bic-DH3 that strongly complements, only a small fraction of eggs (<1%) display ventralized eggshell phenotypes and the extent of ventralization is weak. The correlation suggests that Bic-D influences the localization and possibly the translation of grk mRNA within the oocyte.
We thank Beat Suter for the Bic-Dnull alleles and Ruth Lehmann for anti-Egl antibodies. We also thank Ananya Bhattacharia, Girish Deshpande, Kim McKim, Kirsteen Munn, David Norris, and Cordelia Rauskolb for helpful comments on the manuscript and Le Nguyen for the fly food. J.O. is supported by a Busch fellowship. This work was supported by grants from the National Science Foundation and the Horace W. Goldsmith Foundation.
Communicating editor: T. Schüpbach
- Received July 22, 1999.
- Accepted October 13, 1999.
- Copyright © 2000 by the Genetics Society of America