bicaudal encodes the Drosophila beta NAC homolog, a component of the ribosomal translational machinery*

The bicaudal (bic) mutation was the first Drosophila mutation identified as having a clear effect on embryonic pattern formation (Bull, 1966). At its most extreme, the mutation produces bicaudal embryos, that is, embryos in which the head and thorax are missing and are replaced with mirror-image duplications of abdominal segments. Milder phenotypes with asymmetrical small duplications of the posterior in the anterior or simply with missing head structures are also always associated with the mutation. A number of other mutations with a strikingly similar range of phenotypes have since been identified, of which the best characterized are dominant mutations at the loci Bicaudal D (BicD) and Bicaudal C (BicC) (Mohler and Wieschaus, 1986; Suter et al., 1989; Wharton and Struhl, 1989; Mahone et al., 1995; Saffman et al., 1998).

Detailed molecular genetic analyses of embryonic pattern formation have provided insight not only into the normal process of anteroposterior axis formation but also the molecular defects that underlie the formation of bicaudal embryos (for reviews see Lasko, 1992; Macdonald and Smibert, 1996; Rongo and Lehmann, 1996). The anteroposterior pattern has its origins in gradients of the Bicoid and Nanos proteins, which emanate from the anterior and posterior embryonic poles, respectively, soon after fertilization. Nanos accumulation in the posterior prevents translation of maternally derived mRNA for Hunchback, a transcriptional regulator that represses the gap class genes required for posterior development. Thus a region in the embryonic posterior is created in which posterior development can proceed.

Nanos is generated from maternally derived mRNA and like most maternal components of the mature egg, nanos (nos) mRNA is synthesized by the germline sister cells of the developing oocyte (the nurse cells) and transported into the oocyte through cytoplasmic bridges at its anterior end. Further, this nos mRNA is translationally active in the nurse cells throughout their existence (Wang et al., 1994). Thus, in order to prevent the appearance of Nanos protein within the developing oocyte and to limit its appearance to the extreme posterior embryonic tip at fertilization, mechanisms that prevent translation of nos mRNA within the oocyte have developed. A privileged compartment of cytoplasm termed the pole plasm is generated at the extreme posterior tip of the late oocyte/embryo and nos mRNA translation is strictly limited to this compartment.

The gene oskar (osk) plays a critical role in both the local formation of the pole plasm and the translation of nos mRNA within the pole plasm. This role in turn demands precise spatial regulation of Oskar function which is again achieved by limiting the translation of osk mRNA. The pattern of osk mRNA localization and translation differ from those of nos, however (Ephrussi et al., 1991; Kim-Ha et al., 1991, 1995; Webster et al., 1997). osk mRNA is not translated at its site of origin, the nurse cells, and shows a changing pattern of localization in the oocyte in the early stages of oogenesis. By stage 9, the mRNA appears concentrated in a tight rim of cytoplasm at the posterior oocyte pole and Oskar protein accumulates at this site from this point onward. This tight localization of both mRNA and protein persists into the mature egg.

Although at fertilization both nos and osk mRNAs are highly concentrated at their shared local site of translation (the posterior pole) most of the mRNA from both genes is diffusely distributed throughout the cytoplasm of the mature egg in a translationally inactive state (Bergsten and Gavis, 1999). For both mRNAs, translational repression requires sequences present in their 3′ UTRs (Kim-Ha et al., 1993; Gavis et al., 1996; Smibert et al., 1996; Dahanukar and Wharton, 1996). Proteins that bind to these 3′ UTR sequences have been identified. Smaug, a protein detected in very early embryos, binds to the 3′ UTR sequences required for repression of nos mRNA translation (Smibert et al., 1996) and Bruno, a protein encoded by the locus arrest, binds to the equivalent sequences of osk mRNA during oogenesis (Kim-Ha et al., 1995; Webster et al., 1997). Bruno is absent from early embryos, however (Webster et al., 1997), and thus some other mechanism must silence osk mRNA present in the bulk cytoplasm of the early embryo.

The critical defect required for production of bicaudal embryos has proved to be the ectopic translation of nos mRNA in the anterior embryonic compartment. Expression of Nanos in this compartment not only prevents maternal Hunchback action in this region but also represses translation of mRNA for the major anterior determinant Bicoid. As a result, anterior development is lost and replaced by the posterior program. Mutations at four loci (BicD, BicC, bullwinkle and Klp38B) that produce bicaudal embryos all result in ectopic expression of Nanos in the anterior (Wharton and Struhl, 1989; Mahone et al., 1995; Rittenhouse and Berg, 1995; Ruden et al., 1997). Interestingly, all of these mutations are believed to cause ectopic nos mRNA translation indirectly as a result of abnormal retention/accumulation of osk mRNA at the anterior of the mature oocyte/embryo. Three of these genes (BicC, BicD and Klp38B) encode proteins with predicted roles in RNA localization and transport. Mutations of bullwinkle have pleiotropic effects on development, including defects in migration of the somatically derived follicle cells surrounding the nurse cell/oocyte complex. bullwinkle may thus cause abnormal osk mRNA accumulation/translation by a more indirect route (Wilson et al., 1996). A number of molecular constructs that remove repressor binding sites for Smaug and Bruno from nos and osk mRNAs, respectively, leading to ectopic expression of nos mRNA in the anterior compartment, also yield efficient production of embryos with posterior duplications in the anterior (Gavis and Lehmann, 1992; Kim-Ha et al., 1995; Dahanukar and Wharton, 1996).

Analysis of the original bic mutation has been hindered by the low penetrance of the bic phenotype and its sensitivity to suppression by both genetic and environmental factors. Little work has been performed on the mutation since its original discovery. We undertook to investigate bicaudal further by cloning the affected gene and by molecular and genetic analysis of the defects associated with the mutation. We have also investigated the relationship between bic and a mutation referred to here as Enhancer of Bic (E(Bic), previously 49Da¹), which increases production of bicaudal embryos by some mutations (see below). We report our findings that (i) both mutations map to a ∼3.5 kb region containing the Drosophila homolog of beta NAC, a known component of the translational machinery and (ii) that both mutations are rescued by a 4.0 kb genomic DNA fragment containing only one complete transcription unit – that encoding beta NAC. We have further established that the bic mutation results in ectopic translation of nos mRNA during early embryogenesis but without concomitant ectopic mislocalization of osk mRNA. These findings suggest that beta NAC, a ubiquitous component of the translational machinery, is an essential element in the control mechanism that regulates nos mRNA translation.

The bicaudal phenotype was recovered by recombining the second chromosome in the bic strain M245 (Nüsslein-Volhard laboratory) with an isogenized twl bw (iso twl bw) chromosome (and later, an isogenized cn bw sp chromosome), recovering individual twl bic? or cn bic? recombinant chromosomes and testing these for production of bicaudal embryos when hemizygous with the vgB deficiency, using the rearing conditions described by Nüsslein-Volhard (1977, and below). Five recombinant bic chromosomes (bic 3, 9, 10, 14 and 54) thus generated were used here. In situ hybridization established that region 2R of the original E (Bic) chromosome (Lasko and Pardue, 1988) contained a P element at 49D, thus providing the first indication that the E(Bic) phenotype is indeed caused by a P insertion. Six further P insertions, all but one of which were proximal to the insertion at 49D, were also present on 2R. Recombination with an isogenized twl bw chromosome (see above) removed all proximal P elements. One of the twl E(Bic) recombinants generated, 40 K, was used for most of the studies here. This chromosome carries the 49D P element and the single distal P element at 60A-B. Viable, fertile revertants of the E(Bic) phenotype generated from the 40K chromosome by P element mobilization had always lost the 49D P element, while often retaining the 60A-B P element. Thus the P element at 49D causes the E(Bic) phenotypic effects and the 60A-B P element has no detectable effects on viability or fertility.

In an attempt to isolate further alleles of the E(Bic) locus, mutations were generated using ethyl methane sulfonate mutagenesis (Lewis and Bacher, 1968), followed by (i) selection for lethal mutations in the overlap region of vgB and vgC, and (ii) complementation testing with E(Bic) and alleles of other loci in this region. From a screen of >15,000 mutant chromosomes, 26 lethal mutations were isolated, some of which proved to be alleles of known loci within the vgB/vgC overlap (e.g. l(2)49Db and l(2) 49Dc, see Table 2). None however proved to be allelic to E(Bic). The vgB, vgC and vg¹³⁶ deficiency stocks were from P. Lasko. The DfSu(z)3^1.iy deficiency was provided by C. T. Wu and the nos^L7 allele by R. Lehmann. The wild-type strain used here was Ore-R. Other mutations and balancer chromosomes are described in Lindsley and Zimm (1992).

To express the bicaudal phenotype, developing flies were kept at 18°C or room temperature until 24 hours prior to eclosion, at which time they are transferred to 29°C. Virgin females of appropriate genotypes eclosing during the first 48 hours after transfer were selected and mated to wild-type OreR males. Embryos for cuticle analysis were collected every 24 hours, and aged 24 hours. Embryos for in situ hybridization and immunolocalization were collected and aged as necessary at 25°C.

Cuticle preparations (Roberts, 1986) were made from embryos that failed to hatch within 24 hours, sometimes after dechorionation with bleach (Ashburner, 1989).

Overlapping cosmids and phage genomic DNA clones extending proximally from vestigial (Williams and Bell, 1988) were isolated by standard methods from phage libraries in vector Charon 4 and cosmid libraries provided by J. Tamkun and M. Scott. Restriction enzyme site polymorphisms in various genotypes were identified by genomic DNA Southern blots. Reverse northern analysis using radiolabeled cDNA prepared from ovarian poly(A)⁺ RNA was used to identify chromosome walk regions that generate ovarian transcripts. Genomic fragments from these regions were used to isolate cDNAs from ovarian lgt10 libraries kindly provided by M. Champe and L. Kalfayan. Standard molecular analyses of these cDNA clones established that they represented two transcription units, 11A and 5A.

Cosmid G1.8 contains the genomic insert indicated in Fig. 1 in vector CosPer (Pirrotta, 1988). C42 was generated from G1.8 by deletion of an internal BamHI fragment. Techniques for injection of P element constructs were as previously described (Roberts, 1986). The recipient strain for the G1.8 and C42 constructs was w; Ki △2-3 (provided by H. Bellen). The Nco4 construct was generated by cloning a 4.0 kb NcoI fragment carrying the 5A transcription unit into vector pCaSpeR4 (Pirrotta, 1988). The helper P element used for injections was wings-clipped △ 2-3 (Pirrotta, 1988) and the recipient strain was yw^67c23.

DNA sequencing was performed by the Molecular Genetics Core facility, UT Medical School, Houston. The longest cDNAs of the 11A and 5A classes were sequenced throughout on both strands. Sequences were processed using the GCG software package (Genetics Computer Group, Madison, WI). Homologs in the GenBank database were identified using GCG software and the BLASTN and BLASTP programs (National Center for Biotechnology Information). These sequence analyses established that the 11A gene encodes a protein with homology to the unique N-terminus of synexin (Burns et al., 1989) and that the 5A gene encodes Drosophila beta NAC.

Commercially prepared primers (Genosys, DNA International, Gibco-BRL) were used to amplify genomic DNA from twl bw, bic, twl bic recombinant, and E(Bic) mutant chromosomes (Ballinger and Benzer, 1989). Amplified DNA was cloned into TA vector pCR2.1 (Invitrogen). Some amplified clones were sequenced manually (Sequenase 2.0 kit, USB). All sequence differences between strains were verified by sequencing multiple clones.

RNAs were extracted using Trizol (Life Technologies). Electrophoresis of denatured RNA was as previously (Kovalick and Beckingham, 1992). RNA was transferred to GeneScreen (NEN Research Products) and fixed to the membrane by UV cross-linking (Stratalinker, Stratagene). Hybridization was as recommended for GeneScreen. Probes were prepared by random hexanucleotide priming (BMB). Hybridization to the ribosomal protein rp49 transcript (O’Connell and Rosbash, 1984) was used as a loading control.

Dissected ovaries were transferred immediately to 500 μl fixation solution (4% formaldehyde in 120 mM Hepes, pH 6.9, 2.67 mM MgSO₄, 1.33 mM EGTA) under 600 μl heptane and gently agitated for 30 minutes. Ovaries were then washed three times, 5 minutes, in PBS, then dehydrated and stored in 9:1 100% ethanol:DMSO at

–20°C. A digoxigenin-labeled cRNA probe for the beta NAC 3′UTR was prepared using reagents and protocols from BMB.

Hybridizations were performed largely after the procedure of Tautz and Pfeifle (1989). Fixed ovaries were rehydrated in PBST (PBS+ 0.1% Tween 20), then digested with proteinase K (25 μg/mL) for a time empirically determined to give optimum probe penetration and preservation of structure. Ovaries were then washed three times for 5 minutes in PBST, fixed in 4% formaldehyde in PBST for 20 minutes, washed four more times (total time 90 minutes) in PBST, once in 1:1 PBST: hybridization solution (HS: 50% formamide, 5× SSC, 100 μg/ml heparin, 0.1% Tween 20, 100 μg/ml single-strand DNA) for 10 minutes and finally in HS for 10 minutes. Pre-hybridization was for 1 hour at 55°C in HS. Probe was denatured at 80°C for 10 minutes in 100 μl HS, cooled on ice, and used for overnight hybridization at 55°C. Ovaries were washed for 1 hour at 55°C in HS, 1 hour at 55°C in 1:1 PBST: HS, 30 minutes at 22°C in PBST and 10 minutes at 22°C in PBST. Hybridized probe was detected with an anti-digoxigenin antibody. Ovaries were dehydrated in ethanol, then xylene and mounted in Permount (Fisher).

Embryos were fixed as described above for ovaries and hybridized to the beta NAC 3′UTR cRNA probe using a published protocol (O’Neill and Bier, 1994).

Detection of nos and osk RNAs was as previously (Ephrussi et al., 1991; Gavis and Lehmann, 1992) using DNA probes. cDNA clones of osk and nos and rabbit anti-Nanos serum were provided by R. Lehmann. Nanos localization was as previously (Gavis and Lehmann, 1992). The secondary antibody was goat anti-rabbit IgG conjugated to biotin, with signal amplification by horseradish peroxidase-conjugated avidin-biotin complexes (Vectastain, Vecta). The pigment-producing substrate was diaminobenzidine; blue color was generated by addition of 4 μl of 8% NiCl₂ to the 500 μl detection reaction. For Nanos immunolocalization, embryos in the 7th/8th nuclear cycles were examined (Wang and Lehmann, 1994).

Nüsslein-Volhard (1977) characterized bicaudal as a recessive, hypomorphic mutation that lies within the overlap region of deficiencies vestigial B and vestigial C (vgB and vgC) on chromosome 2R. Attempts to isolate further maternal effect alleles of the locus proved unsuccessful, however, suggesting that the mutation is an unusual allele of a vital gene. Mohler and Wieschaus (1986) hypothesised that this vital gene might be a locus termed 49Da identified by Lasko and Pardue (1988) within the vgB/vgC deficiency overlap. 49Da was defined by a single zygotic lethal allele, 49Da¹. When homozygous, 49Da¹ produces 100% embryonic lethality, with most embryos dying in late embryogenesis with variable defects mainly in the anterior of the embryo (K. B., unpublished observations). This zygotic lethality of 49Da¹ never involves formation of bicaudal embryos. However, in the heterozygous condition in adult females, 49Da¹, like the vgB and vgC deficiencies, will enhance the maternally derived bicaudal phenotype of dominant mutations of BicD and a multiply-mutant line, YC67 (Mohler and Wieschaus, 1986). We refer here to the 49Da¹ mutation as Enhancer of Bic, E(Bic).

For both bic and E(Bic) it was necessary to generate new recombinant mutant chromosomes before further analysis could be undertaken. In the case of bic, none of the stocks that we obtained from previous workers (A. Bull and C. Nüsslein-Volhard) had been under active study for years and all proved suppressed for the bicaudal phenotype. Recombination to replace DNA proximal to bic on 2R (Nüsslein-Volhard, 1977) was therefore necessary to restore penetrance (see Materials and Methods). E(Bic) was generated by hybrid dysgenesis using P elements present in the Pi-2 strain (Lasko and Pardue, 1988), but no characterization of the P elements present on the E(Bic) chromosome had been performed. We determined that the mutation is indeed caused by a P element insertion (see Materials and Methods) and used recombination to remove several additional, potentially mutagenic, P elements from the E(Bic) chromosome. As described in the Materials and Methods, recombination for both bic and E(Bic) used the same isogenized twl bw chromosome. As a result, polymorphisms derived from the twl bw chromosome were introduced onto the recombinants that proved useful in mapping the mutations (see below).

We tested putative bic recombinant chromosomes, by generating females hemizygous for the locus (bic/vgB). The vgB deficiency was used since Nüsslein-Volhard (1977) reported greatest bicaudal phenotype penetrance with this deletion. However, in addition to three classes of bicaudal embryos (symmetrical, asymmetrical and “headless-bicaudal”) and a class of headless embryos, all of which she viewed as deriving from a single primary defect, Nüsslein-Volhard reported other lesions in embryos from bic/vgB mothers (Nüsslein-Volhard, 1977). These included abnormal heads and aberrant segmentation – defects that were not clearly related to the bicaudal phenotype but were always present, often in a high proportion of the eggs. In addition, failure of more than 50% of the eggs to develop was repeatedly recorded.

In the course of our work, we discovered that mothers heterozygous for the vgB deficiency alone (vgB/+) produce all of these additional defects, including a variable (and occasionally very high) proportion of undeveloped eggs (see, for example, Table 3B). In our studies therefore, we have used only the defects clearly identifiable as components of the bicaudal phenotype (that is bicaudal and headless embryos) to score for presence or absence of the mutation and its effects. We generated and used five recombinant chromosomes (bic 3, 9, 10, 14 and 54) that produce bicaudal embryos as 3-16% of the unhatched eggs, a percentage consonant with previous studies of the mutation (Bull, 1966; Nüsslein-Volhard, 1977).

Given that E(Bic) and bic both lie within the vgB/vgC deficiency overlap, and that E(Bic) was also known to lie close and proximal to vestigial within this region, we extended proximally a previously generated chromosome walk in the vestigial region (Williams and Bell, 1988) by a further 65 kb and sought to establish the location of both mutations within this region. We determined that bic also lies proximal to vestigial by mapping with the deficiency vg¹³⁶ which removes part of vestigial and all the more distal DNA within the vgB/vgC overlap (Lasko and Pardue, 1988; Williams and Bell, 1988) (Table 1). We further determined that a more recently identified deficiency in the region, DfSu(z)3^1.iy (Wu and Howe, 1995), which also removes the distal region of the vgB/vgC overlap, did not delete the location of either E(Bic) or bic (see Table 1). We identified the proximal breakpoint for the DfSu(z)3^1.iy deficiency within our chromosome walk and established that this deficiency extends further proximally than vg¹³⁶ (Fig. 1). Thus, the DfSu(z)3^1.iy proximal breakpoint provided a distal limit to the DNA that could contain both the E(Bic) and bic mutations (see Fig. 1).

Restriction enzyme site polymorphisms derived from the isogenized twl bw chromosome were identified on some of the twl bic and twl E(Bic) recombinants used here (see above) and these permitted further delineation of the regions that could contain the mutations. By this route, the E(Bic) and bic mutations were initially mapped to lie within ∼40 kb and 7 kb, respectively, of the DfSu(z)3^1.iy proximal breakpoint (see Fig. 1A,B). Southern blot analysis using DNA fragments for the entire 65 kb walk further established that the P element associated with the E(Bic) mutation lies within the 7.0 kb region containing the bic mutation (Fig. 1B).

Reverse northern analysis was used to identify DNA fragments within the 40 kb region of interest that are transcribed in the ovary. Transcribed fragments were then used to isolate cDNA clones from an ovarian library (see Materials and Methods). Further analysis revealed that these clones all derived from two transcription units, one proximal (termed 11A) and one distal (termed 5A) to the E(Bic) P element insertion site (Fig. 1). Three genomic DNA fragments were then used in rescue experiments to determine which of these two transcription units could rescue the E(Bic) and bic mutations. At least two different insertion lines for each DNA fragment were used in rescue experiments. Initially a large cosmid construct (G1.8) that encompasses both transcription units was assayed for rescue and then a derivative (C42) that lacks protein coding exons for the 11A transcript (see Fig. 1).

Finally a 4.0 kb NcoI fragment that spans the insertion site of the E(Bic) P element and the proximal breakpoint of deficiency DfSu(z)3^1.iy, but which contains only a single complete transcription unit (the distal 5A gene) was assayed (see Fig. 1).

As shown in Table 2, all three fragments rescue the E(Bic) mutation to full viability. Rescued females were fertile, producing few unhatched eggs that never contained bicaudal embryos. The large rescue DNA fragment G1.8 was also tested for its ability to rescue lethal mutations from two further complementation groups in the vicinity of E(Bic) termed 49Db and 49Dc (Table 2). Mutations for neither of these loci were rescued by this DNA fragment, in accord with recent mapping studies (Stitzinger et al., 1999). For the bic rescue crosses, given the variability and susceptibility to suppression of the mutation, the control mothers for assessment of rescue were sisters of the females carrying the rescue construct under study. As shown in Table 3A,B, all three DNA fragments fully rescue the defects uniquely associated with the bicaudal mutant phenotype. The defects in segmentation and head structure (listed as “other defects” in Table 3A,B) that are dominantly associated with the vgB deficiency (see earlier discussion) are unaffected by the rescue constructs (no statistically significant change). The strong and variable dominant effect of vgB on hatch rate still persists (Table 3B), and the rescue constructs have no statistically significant effect on hatch rates for embryos from vgB/+ mothers or from bic/+ mothers. However, all of the rescue constructs produce a statistically significant increase in hatching for embryos from bic/vgB mothers when compared to the appropriate controls (see legend Table 3). Thus we conclude that the single transcription unit present in the 4.0 kb NcoI fragment rescues both the bic and E(Bic) mutations.

The complete sequence of the 4.0 kb NcoI rescue fragment is shown in Fig. 2. One complete transcription unit is present in this DNA, which has a single small intron separating the 5′UTR and the initiator ATG. Genomic and cDNA sequence analysis indicate the generation of two transcripts, differing in poly(A) site usage, from this gene (see legend Fig. 2). The insertion site of the E(Bic) P element within this 4.0 kb rescue fragment was investigated by amplification and sequencing of the DNA adjacent to its terminal repeats (see Materials and Methods). This established that the P element is inserted into the 5′UTR of the gene, 46 residues upstream of the small intron (Fig. 2).

Sequence studies for (i) bic-carrying chromosomes derived from Nüsslein-Volhard’s laboratory, (ii) the iso twl bw chromosome used for recombination, and (iii) the twl bic recombinant chromosomes, established that the bicaudal mutation does not affect the protein coding region of this gene. In addition, DNA sequencing identified a single base change (G to T) on the iso twl bw chromosome that is absent from all original bic-carrying chromosomes examined and from the cDNA clones sequenced (see legend Fig. 2). This polymorphism had been recombined onto the bic 54 recombinant chromosome in addition to the twl marker and the restriction fragment length polymorphism shown in Fig. 1B. The presence of this single base change in the C-terminal protein coding sequences of the bic 54 recombinant chromosome thus indicates that the bic lesion lies distal to this position, in DNA 3′ to the beta NAC coding sequence. Altogether 1640 residues of the beta NAC transcribed region, including 544 residues of the 3′UTR, were sequenced on various bic chromosomes without identification of the mutant lesion.

The protein encoded by the rescuing transcription unit is 61% identical, 77% similar to a human protein termed beta NAC, and thus almost certainly encodes the Drosophila beta NAC homolog. Two yeast homologs (EGD1 and BTT1) with less similarity to the Drosophila and human proteins have been studied (Parthun etal., 1992; Hu and Ronne, 1994) and incomplete cDNAs from mouse and rice have also been identified (Fig. 3). In humans, the gene shows alternative splicing to produce two isoforms that differ in initiator ATG useage such that one has an amino-terminal extension of 44 amino acids relative to the other. These two forms were initially identified in work implicating beta NAC as a general transcription factor for polymerase II, and were thus termed BTF (Basic Transcription Factor) 3a and 3b (Zheng et al., 1990). Further transcription work has failed to substantiate this role for the protein, however (Flores et al., 1992; Montcollin et al., 1992). The shorter isoform BTF3b has since been identified as the smaller subunit of an abundant heterodimeric complex that can be cross-linked to short nascent polypeptide chains on ribosomes (Wiedmann et al., 1994). This isoform has thus been renamed beta NAC (Nascent Polypeptide Associated Complex). The 5A transcription unit encodes the shorter, beta NAC, isoform of the protein, as do the genes and cDNAs identified in yeast, mouse and rice (Fig. 3). The intron position in the 5A transcription unit corresponds to that used to generate the BTF3b/beta NAC isoform in humans. No sequences are present in the 4.0 kb NcoI fragment that could generate the alternatively spliced BTF3a isoform.

Our northern blot analysis indicates that beta NAC mRNA is expressed widely in Drosophila. Transcripts are extremely abundant in the ovary (Fig. 4) but high levels of transcripts were also detected in embryos, male gonads and non-gonadal tissues from both adult male and female tissues (data not shown). Polyadenylated transcripts of the two size classes predicted by our genomic and cDNA sequence studies (1.2 and 1.7 kb, see Fig. 2) were detected in all RNA samples examined. The 1.2 kb size class transcripts proved far more abundant however, perhaps reflecting the use of a perfect polyadenylation signal in their generation (see Fig. 2).

No additional transcripts were detected in RNA from either bic hemizygotes or E(Bic) heterozygotes. Scanning densitometry of the hybridization signals, using the signal from a ribosomal protein probe (rp49) for normalization, indicated that vgB heterozygotes contain significantly reduced levels of beta NAC transcripts, consistent with deletion of the gene in this deficiency. Transcript levels were also consistently lower in RNA from E(Bic) heterozygotes, but quantitation by this procedure was not reproducible enough to identify any effects of the bic mutation on RNA levels.

The spatiotemporal pattern of expression of the beta NAC gene was examined during oogenesis and embryogenesis. In embryos, beta NAC transcripts are ubiquitous and abundant prior to gastrulation, but are excluded from the pole cells. mRNA levels decline steeply at gastrulation and then increase again at full germ-band extension, showing high levels in the mesoderm. By late embryogenesis, the central nervous system and pharyngeal and body wall muscles show intense expression (data not shown). This expression pattern suggests maternally derived supplies of beta NAC, which may explain the late embryonic zygotic lethality of the E(Bic) allele (see earlier). The ovarian expression pattern is shown in Fig. 5A. Within the individual units, or ovarioles, of the ovary, cysts of germline cells are continuously generated from stem cells at the anterior tip region termed the germarium. In each ovariole, two waves of beta NAC transcript expression are detected. The first occurs in region 2B of the germarium, and reflects accumulation of beta NAC transcripts in the somatically derived follicle cells that migrate to surround the cysts of germline cells in this part of the ovary (Fig. 5A). However, these transcripts are transiently expressed and, as stage 1 egg chambers leave the germarium, no beta NAC transcripts are detectable in either the somatic or germline cells. The second burst of transcript accumulation is initiated in stage 3 egg chambers and is limited to the germline nurse cells of the follicles (Fig. 5A). From this point onwards, transcripts are present in the nurse cells until the last stages of oogenesis when they are transported to the oocyte compartment and appear initially in the anterior cortical cytoplasm. This pattern of transcript accumulation suggests the activity of two transcriptional enhancer elements during oogenesis, one within the somatic follicle cells and one within the germline cells. beta NAC transcript accumulation was examined in ovaries of bic/vgB females and as controls, in vgB/+ and bic/+ females, and compared to the wild-type pattern. Given the variable penetrance of the bic mutant effects, the fraction of bicaudal embryos produced by sisters of the females used for the in situ hybridization studies was determined. 13.2% of the unhatched embryos from the bic/vgB sister females (i.e. 7% of all eggs laid in a 24 hour period) showed the bicaudal phenotype; none of the vgB/+ or bic/+ sisters produced bicaudal embryos. In a significant fraction of the bic/vgB ovaries the second burst of beta NAC transcript accumulation in the germline cells appeared defective and therefore careful quantitative analysis of the pattern was performed, as detailed in the legend to Fig. 5. This established that, in 19% (Fig. 5C) of the ovarioles from bic/vgB ovaries, the second burst of transcription in stage 3 egg chambers is significantly reduced, delayed or undetectable (Fig. 5B). In contrast, low levels (1-2.5%) of vgB/+, bic/+, and wild-type ovarioles showed these defects. The relative representation of these defects in bic/vgB ovaries is comparable to that of the bicaudal phenotype in the embryos produced and thus could indicate a causal relationship.

In wild-type embryos, a fraction of nos mRNA is highly concentrated in a rim of cytoplasm at the extreme posterior pole of the embryo with a diffuse uniform level of transcripts throughout the rest of the embryo. Translation of the posteriorly localized nos mRNA results in the gradient of Nanos protein emanating from the posterior pole toward the center of the embryo after fertilization. We examined both nos mRNA and protein localization in embryos from vgB/+, bic/+ and bic/vgB mothers. In 32% of bic/vgB-derived embryos defects in nos mRNA distribution were detected (Table 4A). These defects included a more diffuse, reduced distribution of nos mRNA at the posterior pole (Fig. 6B) and strong, irregularly distributed levels of nos mRNA in more anterior regions (Fig. 6B-D). These defects were occasionally seen for embryos from vgB/+ and bic/+ mothers but at much reduced frequencies (Table 4A). An additional rarer type of mislocalization of nos mRNA was detected only in bic/vgB embryos; the nos mRNA at the posterior pole was not localized at the extreme posterior, but rather formed an “annulus” of hybridization displaced away from the pole itself, and usually tilted relative to the long axis of the embryo (Fig. 6C,D).

Examination of Nanos protein localization revealed several defects. In general Nanos levels appeared considerably higher throughout the embryos than in control embryos stained in parallel (Fig. 7C). The posterior Nanos gradient extended far into the anterior, sometimes emanating from a focus of very high protein concentration that was displaced from the posterior pole and located at the periphery of the embryo (Fig. 7C). This altered topology for the Nanos protein distribution appeared to reflect the altered topology for nos mRNA at the posterior pole (see above). In addition bic/vgB mothers produced some embryos with gradients of Nanos protein emanating from both the anterior and posterior poles (Fig. 7B). As for the mRNA localization, defects in Nanos localization were considerably reduced in embryos from vgB/+ and bic/+ mothers, and none of these embryos showed Nanos protein localization at both embryonic poles (Table 4B).

In experiments in which either nos mRNA or Nanos protein localization were examined individually, the fraction of embryos showing defects in Nanos protein distribution seemed considerably higher than those with mRNA localization defects. Both mRNA and protein localization defects also appeared higher than the fraction of bicaudal embryos expected in the population. In a further experiment therefore, nos mRNA and protein localization were performed on the same batch of embryos and a sample of the embryos was allowed to develop to assess penetration of the bicaudal phenotype. As shown in Table 4C, a far greater fraction of embryos showed defects in localization of nos gene products than those that ultimately produced bicaudal embryos. This could reflect early death of some embryos and/or a capacity for self-regulation within the embryo that will correct the ensuing defects and permit development ultimately to proceed normally.

No defects in osk mRNA localization were detected in embryos from bic/vgB mothers (data not shown). In particular, a sample from the same batch of embryos that was tested in parallel for nos mRNA/protein localization and bicaudal embryo production (see above), was also examined for osk mRNA localization. In contrast to the obvious defects in nos gene product distribution, osk mRNA localization was found to be completely wild type in these embryos (see Table 4C).

Our data indicate that bicaudal is a mutation of the beta NAC locus. Our evidence derives from both genetic mapping and genetic rescue studies. Despite the low and variable penetrance of the bicaudal mutant effects, in carefully controlled experiments we have mapped the mutation, with deficiencies and polymorphisms, to a small (∼2 kb) region 3′ to the protein coding sequence of the beta NAC gene. In addition, we have shown that a genomic DNA fragment containing only one complete transcription unit (beta NAC) rescues the distinctive embryonic phenotypes uniquely associated with the mutation. Our ovarian in situ hybridization experiments indicate a subtle defect that could underlie the mutant phenotype: transcript accumulation in the germline is undetectable or reduced and developmentally delayed in a fraction of the developing egg chambers of bic/vgB mothers. Thus the maternal effect specificity of the mutation appears to reflect reduced activity of an ovarian germline-specific enhancer element or mRNA stabilization sequence. This type of defect is consistent with our determination that the beta NAC protein coding sequence is unchanged in the bic mutation. beta NAC is likely to be widely distributed in the organism and most mutations of the protein coding sequence would probably affect many tissues. Interestingly, mutations in other genes with universal functions that show a maternal effect phenotype have proved to be ovary-specific promoter mutations (e.g. chickadee, Cooley et al., 1992).

Our findings also demonstrate that E(Bic) is a mutation of the locus encoding the protein beta NAC. Thus, (i) the P element responsible for the mutant effects is inserted into the 5′ UTR of the locus, (ii) a transgene of the beta NAC genomic sequence fully rescues the lethality associated with the mutation, and (iii) transcript levels for the locus are reduced by the mutation. Despite this molecular data linking both bic and E(Bic) to the beta NAC gene, we have found that the E(Bic) mutation complements the maternal effects of bic (data not shown), indicating that E(Bic) is not a null mutation. The enhancement by E(Bic) of bicaudal-embryo producing mutations has proved to be very mutation specific: none of the BicC alleles that produce bicaudal embryos are enhanced (Mohler and Wieschaus, 1988) and very different levels of enhancement are seen for the two dominant BicD alleles, BicD¹ and BicD²(K. M. G. and K. B., unpublished observations). Given that the bic and E(Bic) mutations do not affect the beta NAC protein coding region and that our mRNA localization data indicate distinctive patterns of expression of the gene in the somatic and germline cells of ovarian follicles, a different tissue of action for the E(Bic) mutation (presumably the somatic cells of the ovary as opposed to the germline) could underlie its complementation of bic and its variable interaction with other mutations. Clearly, a null mutation for beta NAC would be of great value in analysis of the locus but attempts to generate such a mutation by standard chemical mutagenesis (see Materials and Methods) or P element imprecise excision (data not shown) have proved unsuccessful.

The bicaudal mutation has not been examined in many years and the studies presented here represent the first analyses of its action in terms of molecular events in the anterior-posteror embryonic patterning pathway. As described earlier, bicaudal embryo production by other maternal effect mutations has proved in all cases to involve ectopic expression of Nanos protein, leading to activation of the posterior pattern gap class genes in the anterior compartment. Our localization of nos gene products shows that for bic, as for these other mutations, ectopic nos mRNA translation is the immediate cause of the bicaudal phenotype. This conclusion is further substantiated by genetic experiments showing (i) that nos is epistatic to bic and (ii) that the penetrance of the bic mutation is enhanced two-three fold for bic/vgB mothers carrying an extra (third) copy of the wild-type nos gene (D. C. M. and K. B., unpublished observations).

The combined effects of the bic mutation on the patterns of nos mRNA, Nanos protein and osk mRNA localization in the early embryo are distinctly different from any reported previously. For other mutations that produce bicaudal embryos, localization of nos mRNA and protein in the anterior is always associated with a similar mislocalization pattern for osk mRNA (Mahone et al., 1995; Rittenhouse and Berg, 1995; Wilson et al., 1996; Mach and Lehmann, 1997; Saffman et al., 1998), indicating, as in the wild-type situation, that local accumulation of nos mRNA and its release from translational repression is dependent on osk activity. In contrast, we find the bic mutation produces aberrant nos mRNA localization and widespread high level accumulation of Nanos protein, without alterations to the osk mRNA distribution pattern. Given that more than 95% of nos mRNA is distributed throughout the oocyte cytoplasm (Bergsten and Gavis, 1999), these findings suggest a failure to repress nos mRNA translation outside the posterior pole, resulting in widespread global production of Nanos protein.

An element (the TCE) in the 3′ UTR of nos mRNA has been identified that mediates translational repression of nos mRNA outside the posterior pole (Dahanukar and Wharton, 1996; Gavis et al., 1996; Smibert et al., 1996). However, the nos mRNA and protein patterns produced by the bic mutation are also different from those reported for a nos construct lacking the TCE (△TCE) (Dahanukar and Wharton, 1996). For △TCE, most nos mRNA remains evenly distributed throughout the oocyte cytoplasm, with a slight reduction in accumulation at the posterior pole. Most strikingly, however, ectopic Nanos protein is undetectable in early cleavage embryos carrying △TCE, whereas bic produces high Nanos levels in these early embryonic stages (compare figure 2j, Dahanukar and Wharton (1996) with our Fig. 7B,C).

As discussed earlier, in the wild-type situation, nos mRNA is translationally active in the nurse cells but becomes translationally repressed on transfer to the oocyte in the later stages of oogenesis. One possible explanation for the different levels of Nanos protein produced by the bic mutation and △TCE in the early embryo is that bic normally acts in the translational repression of nos mRNA during the final stages of oogenesis (and possibly also in early embryogenesis) whereas the TCE is used only in the early embryo. Studies for osk mRNA set a relevant precedent for this suggestion: Bruno, a protein implicated in repression of osk mRNA translation in oogenesis is not present in the early embryo (Kim-Ha et al., 1995; Webster et al., 1997), indicating that other components repress osk mRNA after fertilization. Although translational repression of nos and osk mRNAs clearly differ in some ways (see Introduction), it is possible that for nos mRNA, as for osk mRNA, different factors regulate translation before and after fertilization.

The distinctive patterns of nos mRNA and protein mis-localization produced by the bic mutation in early embryos could be secondary to the ectopic widespread translation of the mRNA. In early embryogenesis, translation and localization of nos mRNA are coupled phenomena. Bergsten and Gavis (1999) have demonstrated that the pattern of nos mRNA translation seen in wild-type embryos is the result of direct competition for binding to nos mRNA between (i) factors localized in the pole plasm that alleviate translational repression, and (ii) globally distributed repressor(s). It is unclear what the consequences of a high level, global, release of nos mRNA from repression might be upon its localization, particularly if this were to occur during oogenesis when the architecture of posterior pole cytoplasm is being established and the factors that bind nos mRNA are being assembled in position. As noted in the Results, defects in Nanos protein levels and localization are more prevalent than those in nos mRNA localization, perhaps suggesting a secondary effect on mRNA localization.

The distinctive annulus of nos mRNA seen at the posterior pole of embryos from bic/vgB mothers and the intense focus of Nanos protein sometimes seen at the periphery of the embryonic posterior are provocative however. The possibility that the bic mutation is causing mislocalization and/or overproduction of nos mRNA binding factors normally found only at the posterior pole must be considered. This defect could be occurring in addition to loss of global repression of nos mRNA translation.

How can the defects produced by the bic mutation be related to the function of beta NAC, the protein affected by the mutation? Most available information on the function of beta NAC comes from in vitro biochemical studies of translational regulation. The protein was identifed as the beta subunit of the heterodimeric Nascent polypeptide Associated Complex (NAC), an abundant ribosome-associated complex that can be cross-linked to very short nascent polypeptide chains on the ribosome (Wiedmann et al., 1994). Binding of NAC to relatively extensive regions of nascent chains (up to 17-100 amino acids from the peptidyl transfer site) suggests NAC may have some role in processing whole protein domains as they emerge from the ribosome (Wang et al., 1995). NAC association with ribosomes has also been shown in vitro to prevent non-specific ribosomal association with the endoplasmic reticulum (ER) (Lauring et al., 1995a,b; Powers and Walter, 1996).

Although beta NAC has also been found associated with transcription-related proteins, little evidence for any in vivo role of these interactions has been generated (Parthun et al., 1992; Hu and Ronne, 1994; Shi et al., 1995). The association of the protein (and a longer isoform) with mammalian RNA polymerase II led to the erroneous designation of the protein as general transcription factor BTF3 (Zheng et al., 1990). Currently this interaction with polymerase II is not known to have any significance (Montcollin et al., 1992; Flores et al., 1992). In contrast, the association of NAC with nascent polypeptide chains on the ribosome is a firmly established role for the complex (Lauring et al., 1995c).

If the ribosome is the major in vivo site of action for beta NAC, this clearly has implications for the molecular basis of the ectopic translation of nos mRNA produced by the bic mutation. The most interesting possibility is that the release of nos mRNA from translational repression is a direct consequence of loss of beta NAC function at the ribosome and thus that, in vivo, beta NAC not only binds nascent chains but, through this interaction, may also actually regulate further translation of growing polypeptides. Repression of nos mRNA translation would thus be a ribosome-related event, involving a block in translational elongation rather than initiation.

The issue of whether repression of nos mRNA translation involves ribosome-associated transcripts has received little attention. However several findings support the prediction that nos mRNA translational repression occurs on the ribosome. nos mRNA is translationally active in the early stages of oogenesis in the nurse cells (Wang et al., 1994) and thus much of the nos mRNA that is transferred to the oocyte cytoplasm has a history of association with active ribosomes. Further, Clark and Gavis (personal communication) have recently demonstrated that the nos mRNA present in newly laid eggs is associated with polysomes.

The suggestion that the ectopic translation of nos mRNA is a direct and specific consequence of loss of beta NAC function may appear to conflict with the presumably universal ability of NAC to interact with nascent polypeptide chains. However, a weak hypomorphic allele like bic can be expected to affect the process most susceptible to loss of gene function, and the robust translational repression required to prevent nos function throughout the embryo could represent such a highly sensitive process. The poor penetrance and variability of the bic mutation may reflect the fact that only a critical range of decreased beta NAC function will produce bicaudal embryos without major effects on vital processes. Similarly the limitation of the effects of the mutation to anteroposterior pattern formation could reflect the central role of translational control in the development of this embryonic axis, as opposed to the dorsal-ventral axis.

Although a direct effect of the bic mutation on nos mRNA translation is suggested by the biochemical function of beta NAC, clearly it is also possible that the mutation acts to produce aberrant nos activity through a more indirect route. As discussed earlier, aberrant production/localization of nos mRNA binding factors that are normally concentrated at the posterior pole may be produced by the bic mutation and the effects on nos mRNA translation may depend primarily on such a defect.

To date, the roles of NAC and its component proteins have mainly been addressed through in vitro biochemical methodologies. Our demonstration of a role for beta NAC in the translational regulation events of anteroposterior embryonic pattern formation provides the first indication that more complex regulatory functions are associated with the protein in the whole organism.

We thank Augusta Gray, Gregg Helt and John Thomas for help in the early phases of this work, and Elizabeth Gavis for helpful comments on the manuscript. These studies were supported by grant #NP-689 from the American Cancer Society and grant C-1119 from the Robert A. Welch Foundation.