nanos is an evolutionarily conserved organizer of anterior-posterior polarity

As the mechanisms of early patterning events in Drosophila development become increasingly well understood, it is of great interest to determine their generality in other organisms. The extraordinary evolutionary conservation of the homeotic complex genes reveals that some features of the Drosophila body plan are widely shared with other segmented animals (Kenyon, 1994). However, the morphological diversity of different embryos suggests that the earliest events of axis formation and the establishment of a segmented body plan might occur by a variety of mechanisms (Gurdon, 1992).

In Drosophila, establishment of the embryonic longitudinal axis is under the control of two localized determinants, nanos (nos) and bicoid (bcd) (for review, see St. Johnston and Nüsslein-Volhard, 1992). nos and bcd RNAs are synthesized maternally and are localized during oogenesis to the posterior and anterior poles of the egg, respectively. Bcd protein translated at the anterior pole diffuses to form an anterior to posterior gradient, and Nos protein translated at the posterior pole forms a posterior to anterior gradient. Despite these similarities in the initial establishment of the Bcd and Nos protein gradients, the two proteins function by strikingly different mechanisms. While Bcd activates zygotic transcription of several genes, including hunchback (hb), in the anterior half of the embryo (Driever and Nüsslein-Volhard, 1988; Struhl et al., 1989), Nos represses translation of the uniformly distributed maternal hb RNA in the posterior, leading to an anterior to posterior Hb protein gradient complementary to that of Nos (Tautz, 1988; Tautz and Pfeifle, 1989). Thus, using distinct mechanisms, both nos and bcd regulate hb and establish qualitatively similar gradients of Hb protein. Hb in turn plays a central role as a transcription factor in the subdivision of the embryo into defined body regions along the anterior-posterior axis (Hülskamp et al., 1990; Struhl et al., 1992), reviewed in (Hülskamp and Tautz, 1991).

Conservation of the hb DNA binding domain in insects, molluscs and annelids suggests that hb may have a widespread role in embryonic patterning (Sommer et al., 1992). In contrast, bcd activity, as tested by cytoplasmic transplantation, and bcd gene homologues have so far been identified only in species closely related to D. melanogaster within the Brachyceran suborder of the Diptera (Macdonald, 1990; Schröder and Sander, 1993; Sommer and Tautz, 1991). Since the Hb gradients resulting from the activities of bcd and nos are very similar, only one of these two means of establishing a gradient is, in principle, necessary. Transcriptional activation of anterior genes by bcd has been shown to require synergistic activation by hb (Simpson-Brose et al., 1994), leading to the suggestion that the role of bcd as a primary determinant of polarity evolved recently as bcd acquired some of the patterning functions performed by nos and maternal hb in other species. Experimental manipulations of eggs from different insect orders suggest a widely conserved role for localized posterior, but not anterior determinants in insects (Sander, 1976). Thus it is possible that the nos-hb regulatory interaction might be the fundamental mechanism for initiating the anterior-posterior axis in insects.

The mechanism by which nos regulates hb mRNA translation is not yet known. Sequences in the 3′ untranslated region (3′UTR) of the hb RNA called nanos response elements (NREs) have been shown to mediate nos-dependent translational repression (Wharton and Struhl, 1991), but sequences in the Nos protein that are important for function have not yet been identified (Wang and Lehmann, 1991). To determine if nos is part of an evolutionarily conserved mechanism of axis determination, and to use phylogenetic sequence comparisons as a tool to identify important structural features of the Nos protein, we have investigated nos structure and function within the Dipteran insect order. Genes with sequence similarity to nos were isolated from Drosophila virilis, the housefly Musca domestica and the midge Chironomus samoensis, which are separated from D. melanogaster by approximately 60, 100 and 200 million years of evolution, respectively. Early embryonic development in these Dipteran species is morphologically very similar, and the expression patterns of the nos homologues are conserved. We find that the nos homologues retain function despite divergence of surprisingly large regions of the protein sequences. The carboxy-terminal region is highly conserved among the nos homologues and defines a novel zinc finger, critical for function. Our results show that nos is a conserved organizer of anterior-posterior patterning in the Diptera, and suggest it should perform a similar function in other insects as well.

Cytoplasmic transplantation was carried out as described by Lehmann and Nüsslein-Volhard (1991). M. domestica pupae were obtained from Carolina Biological Supply and raised according to the provided instructions. Eggs were collected on fresh chicken liver, and were used as donors as described for Drosophila. C. samoensis embryos were collected from cultures maintained by Dr. Klaus Kalthoff at the University of Texas at Austin. At about 1.5 hours of development, one of the four cleavage nuclei reaches the posterior pole to form the first pole bud, which divides to form two pole cells (Kuhn et al., 1987). C. samoensis donors were either at the two pole cell stage or younger. Because of differences in the sizes of donor embryos, the ratio of number of poleplasm donors to recipients was 1:1 for Drosophila, 1:3 for Musca, and 2.5:1 for Chironomus donors.

After injection, recipient embryos were allowed to develop for 2 days at 18 °C. The resulting larvae were mounted in a 1:1 mixture of Hoyer’ s mountant and lactic acid (Wieschaus and Nüsslein-Volhard, 1986). Cuticles were scored for abdominal segmentation by counting ventral setal belts or dorsal hairs. Since rescue was frequently asym-metric either laterally or dorsoventrally, any part of a segment was counted as one segment.

A D. virilis genomic library was obtained from John Tamkun and Mary Prout, and a C. samoensis genomic library from Klaus Kalthoff. A M. domestica genomic library was prepared using genomic DNA from adult flies (Carolina Biological Supply). The D. virilis library was screened using a 2.2 kb full length nos cDNA fragment, [³²P]dCTP labeled by random hexamer priming, in 30% formamide, 5× SSPE, 1% SDS, 1× Denhardt’ s, 0.1 mg/ml sonicated salmon sperm DNA, 10% dextran sulfate at 42 °C. Washes were in 2× SSPE, 0.5% SDS at 42 °C. Phages containing inserts with similarity to nos were not obtained by low stringency screening from M. domestica or C. samoensis libraries under the conditions described above, or by using a probe corresponding to the conserved C-terminal region of nos in 25% formamide conditions.

To amplify Md nos and Cs nos sequences, the following PCR primers were used (written in 5′ to 3′ orientation with redundant nucleotides in parentheses; the corresponding Nos amino acid sequences are in brackets):

J2: TG(T,C)GTGTT(T,C)TG(T,C)GA(A,G)AA(T,C)AA [CVFCENN]

J5: GC(T,C)TT(T,A,G)ATGGC(A,G)TC(T,C)TCCAT [MEDAIKA].

PCR conditions were, 50 mM KCl, 10 mM Tris, pH 8.2, 2 mM MgCl₂, 0.2 mM each dNTP, 0.005 mM each primer, 500 ng genomic DNA, 5 units Taq polymerase (Perkin Elmer-Cetus), in a 50 μl volume with mineral oil overlay. Cycling was 94 °C for 1 minute, 50 °C for 1 minute, 72 °C for 2 minutes, for 40 cycles. Amplified fragments were isolated from agarose gels using the Mermaid kit (Bio101), treated with T4 polymerase to blunt the ends, and cloned into EcoRV-cut pBluescript SK+ (Stratagene).

A 4442 nucleotide XbaI to BamHI fragment of Dv nos, a 4592 nucleotide NsiI to ClaI fragment of Md nos, and a 4552 nucleotide EcoRV to PstI fragment of Cs nos (Fig. 4) were sequenced using standard methods. All three genomic sequences have been submitted to the EMBL and Genbank databases.

Whole-mount in situ RNA hybridization with digoxigenin-labeled RNA probes (Tautz and Pfeifle, 1989) was performed as described by Gavis and Lehmann (1992). Embryos from all species were treated in the same way. RNA probes were transcribed from the following nos subclones (see Fig. 4): Dm 2.2 kb N5 cDNA (Wang and Lehmann, 1991); Dv 4.5 kb XbaI-BamHI fragment; Md 1.1 kb EcoRI fragment; and Cs 4.5 kb EcoRV to PstI fragment.

Antibody staining was carried out as described (Gavis and Lehmann, 1992) using Nos antiserum prepared against a peptide antigen corresponding to the C-terminal 13 amino acids of Nos (Wang et al., 1994).

Cs nos cDNAs were synthesized by reverse transcription (RT) of prepole cell stage embryo total RNA, followed by PCR amplification (Kawasaki et al., 1988). The primers were designed to amplify sequences corresponding to the fourth amino acid through the stop codon of the Cs Nos open reading frame. Md nos cDNAs corresponding to Md Nos amino acids 16 to 399 were synthesized from female and 0- to 2-hour embryo total RNA. The cDNAs were completely sequenced by standard techniques.

To generate a Cs nos transcription template suitable for injection rescue experiments, the Dm nos cDNA clone N5 (Wang and Lehmann, 1991) was first modified as follows to create pNB40-N5:Nde tag RI. The sequence around the initiator methionine was altered to create an NdeI site at the ATG followed by the nine amino acid hemagglutinin epitope tag (Kolodziej and Young, 1991), incorporating the Dm nos codon bias. Following the tag, a single nucleotide change was introduced into the third nos codon, which does not change its coding potential, to create a unique BspEI restriction site. The new sequence reads, beginning at nucleotide 254 (Wang and Lehmann, 1991): TTTTCCAT ATG TAC CCC TAC GAT GTG CCC GAT TAC GCC TTC CGG AGC AAC. An EcoRI restriction site was introduced by PCR immediately following the nos stop codon (Gavis and Lehmann, 1992). A BspEI to EcoRI cassette encoding Cs Nos amino acids 4 through the stop codon was then introduced into the pNB40-N5:Nde tag RI vector backbone to create pNB40-CH. A version of this clone called pNB40-CH3 contains a 4 nucleotide frameshift deletion in the fifth amino acid of the Cs nos open reading frame. This control RNA shows no nos rescuing activity when injected at a concentration of 3 mg/ml (n=25).

A Md nos cDNA template encoding Md nos amino acids 4 to the stop codon was engineered essentially as described above for Cs nos. BspEI and EcoRI restriction sites were introduced at the fourth codon and after the stop codon, respectively, by PCR from cloned genomic DNA. An internal fragment containing the introns was replaced by the corresponding fragment of the Md nos cDNA, and the BspEI-EcoRI cassette was cloned into the pNB40-N5:Nde tag RI vector. The RNAs transcribed from the Md and Cs nos templates share the same Dm nos 5′ and 3′UTRs, and differ only in their protein coding sequences.

RNA was transcribed in vitro with SP6 polymerase from linearized templates, precipitated, resuspended in water and quantitated by UV absorption. Three-fold serial dilutions of RNAs were made in water and were injected as described (Wang and Lehmann, 1991). Injected embryos from mothers of the genotypes nos^L7, nos^BN, or nos^RC/nos^BN all gave equivalent results, and the results were pooled.

To determine whether a nos-like activity is present at the posterior pole of other Dipteran insect species, posterior poleplasm from Drosophila, Musca or Chironomus donor embryos was transplanted into D. melanogaster nos mutant embryo hosts (Fig. 1). Embryos from nos mutant females develop with a normal head, thorax and telson (posterior terminal structures), but completely lack abdominal segmentation (Lehmann and Nüsslein-Volhard, 1991). Embryos injected with posterior poleplasm from each of the species assayed developed with partial or complete abdominal segmentation (Fig. 1C). Rescue frequencies were scored either as overall rescue, which includes embryos with any abdominal segmentation, or strong rescue, which includes only those embryos with 5 or more abdominal segments. The frequency of overall rescue is similar for each donor source, varying between 54% and 75%. The frequency of strong rescue, in contrast, decreases with increasing evolutionary distance of the donor, from 40% for D. pseudoobscura to 7% for C. samoensis (Fig. 1C). We conclude from these results that each of the Dipteran species tested contains a posterior activity that can functionally replace nos. To test whether this activity was nos, we used molecular techniques to isolate nos gene homologues.

A D. melanogaster nos cDNA was used as a probe on genomic libraries under low stringency hybridization conditions (Materials and Methods), and a single class of clones with sequence similarity to nos was isolated from a D. virilis library. Comparison of the predicted D. virilis nos (Dv nos) and D. melanogaster nos (Dm nos) protein coding sequences revealed relatively low overall similarity, with a region of high conservation near the carboxy (C) terminus (see below). The C-terminal sequences were used to design degenerate oligonucleotides that allowed the amplification of DNA fragments with sequence similarity to nos from M. domestica and C. samoensis genomic DNA by the polymerase chain reaction (PCR) (Materials and Methods). Subsequently, complete nos genes were obtained from M. domestica and C. samoensis genomic libraries by screening with the PCR fragment probes.

If genes with sequence similarity to nos function as posterior determinants, their expression patterns should be similar to that of Dm nos. The nos clones were used as probes on developmental RNA blots, and they detect single transcripts of 2.4, 2.5 and 1.8 kb in RNA prepared from D. virilis, M. domestica, and C. samoensis, respectively. As in D. melanogaster (Wang and Lehmann, 1991), the RNAs are strongly expressed in females and in early embryos (data not shown). To determine if these RNAs were posteriorly localized, the nos clones were used as probes for whole-mount in situ hybridization to embryos (Fig. 2). In early embryos of each species, nos RNA is highly concentrated at the posterior pole, while a low level of RNA is distributed throughout the embryos. Each nos RNA is taken up with the posterior poleplasm into the pole cells, the germ cell precursors. The RNAs in each species are detected continuously in the pole cells as these cells migrate through the embryo and are incorporated into the gonads. Expression is not detected in tissues other than the germ cells in any of the species. The similar temporal and spatial expression patterns of these candidate nos RNAs and Dm nos RNA suggests that we have isolated true nos homologues.

The shape of the Nanos protein gradient is critical for the determination of pattern within the embryo (Ephrussi and Lehmann, 1992; Gavis and Lehmann, 1992; Smith et al., 1992). In D. melanogaster embryos this distribution is in part achieved by translational regulation, such that only posteriorly localized nos RNA is translated while unlocalized nos RNA is not translated (Gavis and Lehmann, 1994). To determine the distribution of Nos protein in the other insect species, we stained D. virilis and M. domestica embryos using a polyclonal antiserum directed against a short C-terminal peptide of the Dm Nos protein (Wang et al., 1994). This peptide sequence is conserved in Dv Nos and M. domestica Nos (Md Nos), but not in C. samoensis Nos (Cs Nos) (see below). The antiserum detects a posterior to anterior gradient of Nos in D. virilis and M. domestica embryos comparable to the gradient of Nos observed in D. melanogaster (Fig. 3). The conserved RNA and, at least in two cases, protein expression patterns support the conclusion that Nos functions as a localized determinant of posterior pattern in these species.

To determine the extent of nos sequence similarity between the four species, the nos homologues were sequenced. Since all sequences required for proper expression and function of the Dm nos gene are contained in a 4.4 kb genomic fragment (Gavis and Lehmann, 1992), genomic DNA fragments of similar length were sequenced for Dv, Md, and Cs nos (Fig. 4A). The intron/exon structures of the Md and Cs nos transcripts were determined by cDNA sequencing (Materials and Methods). Overall sizes of the predicted Nos proteins are similar, but large regions have diverged extensively in sequence. Dot matrix comparisons between the predicted Nos protein sequences show that the region of greatest sequence similarity is near the C terminus of the protein (Fig. 4B). An alignment of the predicted Nos protein sequences is presented in Fig. 5. Pairwise comparisons reveal a correlation between overall sequence divergence and time of evolutionary separation (Table 1, Fig 1B). Protein sequence similarity to Dm Nos ranges from 63% for Dv Nos to 30% for Cs Nos. Only 19% amino acid similarity, or 14% identity, is shared among all four Nos proteins.

Primary sequence alignment among all of the Nos homologues is restricted to two regions of the protein. The most extensive sequence conservation is in a 72 amino acid region near the C terminus (Dm nos residues 317 to 388). Pairwise protein sequence comparisons of this 72 amino acid region show that 72% to 97% of the residues are similar (Table 2). Among the completely conserved amino acids in this region are a series of cysteine and histidine residues (Fig. 5). They occur in the order CCHC, CCHC, and are likely to constitute a novel zinc-binding domain (see Discussion). Among sequences that have been reported in the databases, only the Xenopus Xcat-2 gene has significant sequence similarity to nos (Mosquera et al., 1993). Similarity to Xcat-2 is found exclusively in the C-terminal region of Nos, and the alignment is included in Fig. 5. Xcat-2 is 51% to 54% similar to each of the Nos homologues in the 72 amino acid region (Table 2), or 62% to 65% similar when only the first 60 amino acids of this region are compared. The precise spacing and conservation of Nos and Xcat-2 residues in this region, including the CCHC amino acids, is striking and argues that the region may form a discrete protein domain with a conserved function.

The second region of nos sequence similarity is a short stretch of 6 out of 11 amino acids (residues 170 to 181). In addition, a region rich in serine and threonine residues and a region rich in the basic amino acids asparagine and lysine are present in each species, and the location of these regions within Nos is conserved (Fig. 5).

The significant divergence of the Nos protein sequence raises the question of whether the rescuing activity detected in the posterior poleplasm of each species is solely due to nos activity. The transferred cytoplasm contained not only nos RNA and protein, but also potentially other factors that may have contributed to nos activity. To demonstrate that the nos homologues are indeed responsible for the observed rescue, their activities were tested directly. P element transformation was used to establish transgenic D. melanogaster flies carrying the heterologous nos genomic DNA fragments shown in Fig 4. This assay tests not only for conservation of protein function but also for the presence of regulatory sequences required for correct transcription, RNA localization and translation of the nos genes in D. melanogaster.

RNA in situ analysis on embryos from females carrying the Dv nos transgene demonstrates that the Dv nos RNA is localized, like the endogenous Dm nos RNA, to the posterior pole (data not shown). Antibody staining shows that the transgenic Dv Nos protein is distributed in a posterior to anterior gradient comparable to that of the Dm Nos protein (data not shown). The quantities of the Dv nos RNA and protein, as judged by whole-mount staining intensity, appear equivalent to those of the endogenous Dm nos, and to the levels observed in D. virilis. Thus, all features of nos regulation including transcription during oogenesis, proper RNA localization at the posterior pole and restricted posterior translation are conserved between the Drosophila species. Indeed, small islands of sequence similarity are found between Dv and Dm nos both in the upstream region, which presumably contains transcriptional regulatory sequences, and in the 3′ untranslated region (3′UTR) which has been shown to regulate posterior localization and translation of Dm nos RNA (data not shown; Gavis and Lehmann, 1992, 1994).

To determine whether the Dv nos transgene can functionally substitute for Dm nos, it was crossed into mutant nos females. nos is required for two processes, axis determination during embryogenesis, and production of egg chambers during oogenesis (Wang et al., 1994). The transgene complements both phenotypes, demonstrating that amino acid sequences required for both known functions of nos are conserved between the Dv and Dm Nos proteins (Tables 3, 4). However, the Dv nos transgene is less effective in complementing the nos abdominal phenotype than a Dm nos transgene (Table 4). Because D. virilis posterior poleplasm is likewise less effective than D. melanogaster posterior poleplasm in rescuing the abdominal phenotype (Fig. 1C), and because the transgenic RNA and protein distribution patterns appear equivalent to the endogenous nos patterns, the reduced activity of the Dv nos transgene is probably due to reduced Dv Nos protein activity in the D. melanogaster embryos rather than to a failure of the Dv nos transgene to be regulated properly.

P elements carrying the Md or Cs nos genes were unable to complement the D. melanogaster nos phenotype. In situ hybridization detected only trace levels of Md or Cs nos RNAs in the transgenic embryos, and the RNAs are unlocalized (data not shown). These results suggest that Md and Cs nos regulatory elements have so diverged that they are unable to function in D. melanogaster. In fact, no DNA sequence conservation is found in the non-coding regions of these genes. An RNA injection assay was therefore used to test the function of the Md and Cs nos genes. This assay circumvents the requirement for correct regulation of nos transcription, localization or translation.

The protein coding sequences of a Dm nos cDNA were precisely replaced by the predicted Md or Cs nos coding sequences while maintaining the flanking Dm nos 5′ and 3′UTR sequences (Materials and Methods). Like in vitro synthesized Dm nos RNA (Wang and Lehmann, 1991), in vitro synthesized Md nos RNA and Cs nos RNA are fully able to rescue the nos abdominal segmentation phenotype after injection into nos mutant embryos (Fig. 6). These experiments demonstrate that nos function is conserved between these species despite the fact that the Md and Cs Nos proteins share only 44% and 30% amino acid sequence similarity with Dm Nos. The results further suggest that nos RNA (and/or protein) is the active rescuing component in the cytoplasmic transplantation assay, and demonstrate that other species-specific factors are not required for the activity of these heterologous Nos proteins in D. melanogaster.

To compare the abilities of Dm, Md and Cs nos to rescue the nos mutant phenotype, the response to varying concentrations of injected RNAs was measured. For each RNA, an approximately ten-fold range was observed between the minimal concentration of RNA required for rescue and the concentration that achieved maximal rescue (Fig. 6). However, a 5-fold higher concentration of Md nos RNA, or a 7-fold higher concentration of Cs nos RNA than of Dm nos RNA is required to achieve 50% overall rescue (Fig. 6). Md and Cs nos RNAs are therefore 5- to 7-fold less active in the injection rescue assay, which is consistent with the weaker nos activity of transplanted M. domestica and C. samoensis poleplasm (Fig. 1C). Because the assay compared RNAs that differ only in their protein coding sequences, the difference in activities is likely to reflect a reduced ability of the Cs and Md Nos proteins to function in repressing translation of the D. melanogaster hb RNA.

We have used cytoplasmic transplantation assays to demonstrate that nos activity is present in posterior egg cytoplasm of five Dipteran species, and we have isolated clones with sequence similarity to nos from three of the species, D. virilis, M. domestica and C. samoensis. These clones meet three criteria indicating that they are true nos homologues, or orthologues: first, the expression patterns of the nos RNAs are indistinguishable from that of Dm nos. The RNAs are maternally expressed, localized to the posterior pole of early embryos and present exclusively in germ cells throughout embryogenesis. Second, the genes are similar in size, genomic organization and protein sequence. Using low stringency hybridization conditions for genomic DNA blots and library screens, as well as PCR with degenerate oligonucleotides, no other sequences with similarity to nos were detected in D. melanogaster or in the other three species. Thus, nos does not appear to be a member of a large gene family, and the isolated genes are the sequences most closely related to nos in their respective genomes. Finally, the isolated nos genes are able to substitute functionally for Dm nos. These results suggest that the nos homologues function like Dm nos as posterior determinants in their respective species and that a common nos-dependent mechanism is used to establish embryonic polarity in the Diptera.

Previous studies have identified hb homologues in D. virilis and M. domestica, and analysis of the expression patterns demonstrated that the maternal Hb protein in each species is expressed in an anterior to posterior concentration gradient from uniformly distributed maternal RNA (Lukowitz et al., 1994; Sommer and Tautz, 1991; Treier et al., 1989). This pattern is consistent with a role for nos in establishing the Hb maternal protein gradient. In addition, bcd gene homologues have been isolated from these species and exhibit similar RNA expression patterns to Dm bcd (Macdonald, 1990; Sommer and Tautz, 1991). Thus, the key components of anterior-posterior axis determination defined by genetic analysis in Drosophila seem to be conserved in the Brachyceran suborder of the Diptera. However, differences in the regulation of anterior and posterior determinants may exist in species of the Nematoceran suborder. In the Nematoceran C. samoensis, bicaudal embryos with two complete abdomens of opposite polarity can be obtained by UV irradiation of the anterior pole of the egg (Kalthoff, 1983). Anterior UV treatments of D. melanogaster, in contrast, do not result in bicaudal embryos, but rather in embryos lacking head and thorax that resemble bcd mutant embryos (Bownes and Sander, 1976). A model proposed by Kalthoff to explain the C. samoensis bicaudal embryos suggested that posterior determinants may be present through-out the embryo, while anterior determinants are enriched in the anterior (Kalthoff, 1983). On the basis of the posterior localization observed for Cs nos RNA, it appears that posterior determinant function is similar between this species and the other Dipterans. This implies that the difference in the developmental properties of C. samoensis and D. melanogaster embryos is not due to a difference in the distribution of the posterior activity. In D. melanogaster, bicaudal embryos result from the removal or inactivation of both the bcd and hb mRNAs at the anterior. This can occur by a variety of mechanisms, including genetic mutation of bcd and hb (Hülskamp et al., 1990), leakage of anterior cytoplasm from the embryo (Frohnhöfer et al., 1986), and translational repression of bcd and hb by nos. The latter mechanism is possible because nos can translationally repress bcd, like hb, through NRE sequences in the bcd 3′UTR (Wharton and Struhl, 1991), and because nos can be ectopically expressed at the anterior by anterior localization of nos RNA or by unregulated translation of unlocalized nos RNA (Ephrussi et al., 1991; Gavis and Lehmann, 1992). Within the framework of the D. melanogaster model, the C. samoensis bicaudal phenotype could be interpreted as a result of a high UV sensitivity of anterior determinant RNAs, possibly bcd and hb, relative to D. melanogaster, or to high UV sensitivity of a repressor of nos mRNA translation in the anterior. Anterior determinant RNA could also be distributed in a more shallow gradient relative to D. melanogaster, which might explain why a bcd-like activity has not been detected in Nematoceran species.

Molecular studies have begun to explore the conservation of Drosophila patterning genes in non-Dipteran insects such as moths (Lepidoptera), beetles (Coleoptera) and locusts (Orthoptera), (for review see Nagy, 1994). These insects differ in the degree to which the body plan is determined at the time the germ band is first established. Dipterans are long-germ band insects; at the cellular blastoderm stage the germ band includes primordia for all of the body segments. In contrast, short-germ band insects generate some or all of their segments sequentially during a growth phase after cellularization. Fundamental differences must exist in the mechanisms by which a segment is formed within a field of syncytial blastoderm nuclei versus within the context of a growth zone of dividing cells. However, a common feature of insect development is that the initial determination of polarity occurs in the early syncytial egg cell. Thus a similar mechanism of early axis patterning could occur in embryos of different germ band types. A role for posteriorly localized factors in patterning the egg has been demonstrated in many diverse species (Sander, 1976). The finding that nos is localized and functions as a posterior determinant in both suborders of the Diptera suggests that it also does so in closely related insect orders. Whether posterior determinant activities detected in more distantly related short-germ insects correspond to a nos-like gene activity remains to be tested.

In addition to its role in axis determination, nos is required for oogenesis, and females lacking nos produce very few eggs (Wang et al., 1994), (Table 3). The roles of nos in oogenesis and in pattern formation are separable by mutation in D. melanogaster, since some nos alleles with strong abdominal phenotypes have no effect on oogenesis (Wang et al., 1994). In addition, nos does not act through hb during oogenesis and thus may act through a different RNA target at that time (Wang et al., 1994). While the relationship between the two functions of nos is not understood, we find that the nos homologue RNAs in each species are present in germ cells throughout embryogenesis. At least one homologue, Dv nos, can substitute for Dm nos during oogenesis, and the similar expression patterns of Md and Cs nos suggest that in these species nos is also essential for oogenesis. Although the two nos functions may be separable and one might prove to be more ancient in evolution, our data point to conservation of both functions in the Diptera.

A number of D. melanogaster gene homologues have been isolated from D. virilis. Protein sequence identity between cognate genes from these two species is typically ∼80%, but can vary widely (for examples, see O’ Neil and Belote, 1992, and references therein). The 58% overall Dm-Dv Nos identity is relatively low, indicating that much of the Nos protein sequence is not highly constrained. Further comparisons including the more divergent Md and Cs nos sequences shows that only 19% of the Nos amino acids are identical or conserved across all four species. The ability of the heterologous Nos proteins to function in D. melanogaster indicates that the four conserved regions within the protein are likely to be functionally important. Two of these regions are similar only in amino acid content, while a short region of 11 amino acids, and a longer carboxy-terminal 72 amino acid region have substantial primary sequence identity.

The 72 amino acid C-terminal region of Nos is highly conserved, with 67% sequence similarity across all 4 species. The first 60 amino acids of this region are also conserved with the Xenopus gene Xcat-2 (Mosquera et al., 1993). In each of these sequences eight cysteine and histidine residues are conserved with invariant spacing. If it is correct to group these residues into two sequential zinc coordinating sets, as suggested by Mosquera (1993), then the order of the residues, CCHC, makes the Nos sequences most similar to the retroviral nucleocapsid class of zinc finger proteins (Schwabe and Klug, 1994). The nucleocapsid proteins specifically bind and package single stranded genomic viral RNA (Dannull et al., 1994). The conservation of the Nos CCHC region, together with the role of Nos in regulating translation of hb RNA, invites the hypothesis that the Nos and Xcat-2 C-terminal sequences define a new class of zinc-binding proteins with RNA-binding properties. Preliminary evidence that Nos C-terminal protein produced in bacteria contains divalent cations and binds RNA in vitro supports this proposal (D. C., A. Hannaford and R.L. unpublished data). The two Nos motifs, C-X₂-C-X₁₂-H-X₁₀-C and C-X₂-C-X₇-H-X₄-C, differ in spacing both from each other and from the invariant nucleocapsid motif C-X₂-C-X₄-H-X₄-C; thus a detailed analysis will be required to determine the structure of these novel protein elements.

The function of the Xcat-2 gene is unknown, but Xcat-2 RNA is localized to the vegetal pole of the developing Xenopus oocyte. This, together with its structural similarity to nos, suggests that Xcat-2 may act as a region-specific translational regulator in Xenopus. When tested for function by RNA injection assay in D. melanogaster, neither the entire Xcat-2 RNA, nor a hybrid Dm nos RNA in which the conserved 53 amino acids of the Dm nos zinc finger domain (amino acids 319 through 371) have been replaced by the corresponding Xcat-2 residues, can rescue the nos phenotype (C. Wang, M. L. King, and R. L. unpublished data). Apparently the few additional amino acid differences introduced by the Xcat-2 sequences, as compared to the tolerated differences present in the Dipteran sequences, renders the hybrid Nos protein unable to regulate hb. This experiment, as well as the finding that three nos point mutations map to the C-terminal region (D. C. and R. L., unpublished data), confirms that sequences in the C-terminal region are essential for nos function. The functional importance of the other regions conserved among the Dipteran Nos sequences is not yet known. One possible role for the basic asparagine/lysine rich region might be to facilitate or contribute to an RNA-binding function of the adjacent C-terminal region. Another possibility is that some of the conserved Nos sequences might engage in protein-protein interactions with pumilio, a gene required for nos-dependent hb repression (Barker et al., 1992), or with components of the translational machinery.

Translational control is an important mechanism of gene regulation in the egg cell, which must store many maternal transcripts in an inactive form. nos is an example of a spatially restricted, specific translational regulator acting during early development to initiate embryonic patterning. Localized translational repression plays a similar role in the generation of asymmetry in the early nematode embryo (Evans et al., 1994), and may thus have a widespread role in early embryonic patterning. Future work will determine if a nos-like molecule is involved in this process in other species, and how nos exerts its biochemical effect.

We are grateful to Klaus Kalthoff for hosting D. C. in Austin and for making the work with C. samoensis embryos possible, and to members of the Kalthoff laboratory: Michael Rebagliati for assistance with the transplantation experiments, and Michele Dick and Myrtha Laurel for helping with embryo collections and for providing the staged embryos used for C. samoensis RNA preparations. We thank Mary Prout and John Tamkun for the D. virilis library, K. Kalthoff for the C. samoensis library, and A. Hannaford and K. Wills for technical assistance. We appreciate comments on this manuscript from Liz Gavis, Charlotte Wang, Phil Zamore, Jamie Williamson and Steve Burden. D. C. was supported by a postdoctoral fellowship from the American Cancer Society and by the HHMI. R. L. is an Associate Investigator of the HHMI.