Abstract
A small number of maternal effect genes determine anterior-posterior pattern in the Drosophila embryo. Embryos from females mutant for the maternal gene bicoid lack head and thorax, bcd mRNA becomes localized to the anterior tip of the egg during oogenesis and is the source for the morphogen gradient of bcd protein.
Here we show that in vitro transcribed bicoid mRNA that has its own leader sequences substituted by the Xenopus β-globin 5′ untranslated sequences is translated more efficiently than bicoid mRNA with the natural 5′ mRNA leader when tested in vitro and in Drosophila Schneider cells. When injected into bicoid mutant embryos, only the bcd mRNA with the ′-globin leader sequence, substituted for the natural leader, is able to induce anterior development. We used P-transformation to show that sequences in the 5′ leader are neither necessary for localization of the transcript nor for the translational block of the bcd mRNA during oogenesis.
For our injection experiments, we used only one of the identified splicing forms of bcd mRNA. The bcd protein species derived from this mRNA is able to induce anterior development at any position along the anterior-posterior axis. Thus bicoid protein can induce development of head and thorax independent of any other specifically localized morphogenetic factor. Our findings further support the notion that the concentration gradient of bcd protein, and not the existence of different forms of bcd protein, is responsible for specifying subregions of the embryo.
Introduction
Development of the anterior half of the Drosophila embryo depends on the maternal gene bicoid (bcd) (N ü sslein-Volhard et al. 1987). Embryos from females that are homozygous mutant for strong bcd alleles lack head and thorax, and instead develop posterior terminal structures (the telson) at the anterior (Fig. 5) (Frohnhöfer and Nüsslein-Volhard, 1986). Cytoplasmic transplantation experiments revealed the presence of an activity located at the anterior tip of embryos from wild-type but not from bcd mutant females. This activity, when transplanted to any position along the anterior-posterior axis of a recipient embryo, was able to induce anterior development and suppress the formation of posterior structures at the site of injection (Fohnhöfer and Nüsslein-Volhard, 1986). However, these experiments could not distinguish whether bcd is the only anterior determinant in the Drosophila embryo, as other activities might have been cotransplanted.
The bcd gene is transcribcd maternally and bcd mRNAs become localized at the anterior tip of the egg during oogenesis (Frigerio et al. 1986; Berleth et al. 1988). At least four distinct phases of bcd mRNA localization can be distinguished during oogenesis (St. Johnston et al. 1989). Between stages 6 and 9 of oogenesis (staging according to King, 1970), bcd mRNA accumulates in a ring at the anterior end of the oocyte. In stage 9–10a follicles, bcd mRNA also localizes to the apical regions of the nurse cells, but as the nurse cells contract during stage 10b-ll, all the bcd mRNA becomes localized to the cortex at the anterior end of the oocyte. Finally, between stage 12 of oogenesis and egg deposition, bcd mRNA becomes localized to a spherical region of the egg that occupies a slightly dorsal position at the anterior pole. The gene products of the maternal genes exuperantia (Schüpbach and Wieschaus, 1986), swallow (Gans et al. 1975), and staufen (Schüpbach and Wieschaus, 1986) are involved in the process of bcd mRNA localization (Berleth et al. 1988; St. Johnston et al. 1989). Macdonald and Struhl (1988) demonstrated that sequences in the 3′ untranslated region of the bcd mRNA are responsible for a proper RNA localization. During oogenesis, no bcd protein is detectable (Driever and Nusslein-Volhard, 1988a). This finding and the lack of rescue activity of nurse cell cytoplasm when transplanted into embryos from bcdEl females indicate that bcd transcripts are translationally regulated during oogenesis. Translational control has also been proposed for other maternal transcripts in Drosophila (e.g. caudal:Macdonald and Struhl, 1986).
The analysis of bcd cDNAs revealed the existence of several differentially spliced forms of bcd mRNA (Berleth et al. 1988; see Fig. 1A). Type a cDNAs appear to be most abundant. Type b splice products use an alternate splice acceptor site for the third exon and thus have 5 additional codons inserted just in front of the homeo domain. A corresponding protein could not be detected immunologically in embryonic extracts (Driever and Nusslein-Volhard, 1988a). A third minor splice variant, type c, fused the first with the fourth exon. A specific biological role for individual splice products has not been investigated so far.
Translation of bcd protein begins after egg deposition; by the blastoderm stage, the protein is distributed in a concentration gradient along the anterior-posterior axis with a maximum at the anterior tip and extending over about two thirds of the syncytial embryo (Driever and Nusslein-Volhard, 1988a). Genetic experiments allowed the modification of both the shape and height of the bcd protein gradient, and the resulting changes in the blastoderm fate map indicate that bcd protein concentration determines position in the anterior half of the embryo. For example, increasing the dosage of bcd+in the mother led to an increased level of bcd protein in the embryo and a concomitant enlargement of anterior primordia, detectable as a posteriorward shift of the cephalic furrow (Driever and Nusslein-Volhard, 19886).
The bcd protein contains a homeo domain and has been shown to act as a regulator of at least one zygotic target gene, the gap gene hunchback (hb) (Tautz et al. 1987; Tautz, 1988; Schroder et al. 1988). Transcription of hb in the anterior half of the embryo is activated by binding of bcd protein to sequence elements in the upstream region of the zygotic hb promoter (Driever and Nusslein-Volhard, 1989; Driever et al. 1989a,b;Struhl et al. 1989). Other zygotic target genes might bind bcd protein with different affinities (Driever et al. 1989a), thus converting the information supplied by the bcd protein gradient into distinct domains of zygotic gene expression.
In this paper, we address the following main questions. Is bcd sufficient to induce anterior development in the Drosophila embryo or is there a need for additional localized morphogenetic factors ? Is a single bcd protein species sufficient to induce anterior development or is this function performed by several different forms of the protein, arising from the various spliced forms of bcd mRNA that have been identified ? What might be the basis for the translational control of bcd mRNA during oogenesis ?
Materials and methods
Plasmids and in vitro transcription
bcd mRNAs were transcribcd from three different transcription vectors; the additional vector sequences at the 5′ end are: pGem 1 vector (Promega) 58 bases, pGem 3 vector (Pro-mega) 62 bases and Bluescript vector (Stratagene) 73 bases. The constructs are describcd by Berleth et al. (1988) and Driever and Nusslein-Volhard (1989). Transcripts from all three vectors had similar translation rates (data not shown). bcd TN3 transcripts were generated using a pSP 64 vector (Melton et al. 1984), and include additional 12 bases of vector sequences. The construct bcdTN3 was generated by inserting the Ndel (end repaired with Klenow polymerase) - EcoRV fragment from pARbcdNB and the EcoRV - EcoRI fragment from c53.46.6c (both describcd in Driever and Nusslein-Volhard, 1989) into the Ncol (blunt ended with mung bean nuclease) - EcoRI digested pSPBP4 vector, a pSP 64 derivative containing a slightly modified Xenopus β-globin mRNA leader fragment (Siegel and Walter, 1988). The pSPBP4 vector is based on the pSP64T vector (D. A. Melton, Harvard University).
In vitro transcription was performed according to a protocol slightly modified from D.A. Melton et al. (1984). Template plasmids were linearized with EcoRI, extracted with phenol, precipitated and dissolved at 0.5μg; μl11 in TE (10mM Tris-HCl pH7.5; 2ntM EDTA). Setup of the transcription reaction: 5.5 μl diethyl pyrocarbonate (DEPC) treated distilled water; 6μl 5× salts (200mM Tris-HCl pH7.5, 30ITIM MgCl2. 10mM spermidine); 3μl 0.1M dithiothreitol; 2μl human placental ribonuclease inhibitor (Promega 10–20 unitsμl-1); 6 μl capped NTP mix 5 × (2.5 mM each ATP, UTP, CTP, 2.5 mM G(5 ′)ppp(5 ′)G, 0.5 mM GTP); 6 μl linearized DNA in TE 0.5; μ g μl -1 ; 1.5 pl Sp6 RNA Polymerase (10 – 20units; μ l-1). Transcription was performed for 60min at 40°C, then 1.5 μ l 10 mM GTP were added and incubation continued for 15min. The RNA was extracted with phenol and chloroform, precipitated with ethanol in the presence of 0.3 M NaOAc, the pellet washed with cold 70 % ethanol, dried and dissolved in 20 μl DEPC-treated water (typical yield 1.5 μ g μl -1).
In vitro translation and expression in Drosophila Schneider cells
The mRNAs were translated in vitro using reticulocyte lysate or wheat germ extract (Amersham) according to the manufacturers’ protocol. The relative efficiencies of in vitro translation were determined by cutting out the bcd protein bands from SDS-PAGE like the one shown in Fig. 2 and measuring the incorporated radioactivity by liquid scintillation counting. Transient expression in Drosophila Schneider cells, the preparation of extracts and the immunoblot analysis were as describcd (Driever and Nusslein-Volhard, 1989). We constructed the plasmids pMetbcdTN3 and pMetbcdEE by inserting the HindIII Fragment from pbcdTN3 and, respectively, the EcoRI fragment from c53.46.6c, into the pRmHa-3 metallothionein promoter expression vector (Bunch et al. 1988).
Injection into embryos
The RNA was injected into pre-polecell-stage embryos at various dilutions (using DEPC-treated water) according to standard procedures (Frohnhöfer and Nüsslein-Volhard, 1986). The biological activity of the injected mRNA was scored by analyzing the cuticular phenotypes of the injected animals after 48 h of development at 18°C. Photography was performed using dark-field or phase-contrast optics.
P-Transformation and immunohistochemical analysis
The Htridlll-EcoRI fragment with the globin leader, the bcd ORF and the 3 ′ untranslated region of the bcd transcript was cloned into the P-vector pCaSpeRbcdBglII, which was kindly provided by David Stein, Tübingen. This vector, a pCaSpeR derivative (Pirrotta, 1988) carries the 2 kb maternal bcd promoter fragment from the BamHI site to the PstI site at position 1244 and the genomic fragment from 4292 to the EcoRI site at 5.9 kb (pPbcdTN3); numbering of the genomic sequence according to Berleth et al. 1988. P transformation was as previously describcd (Driever et al. 1989a). Immunohistochemistry using monoclonal anti-bcd protein antibody (Driever and Nüsslein-Volhard, 1988a) was performed using the Vectastain ABC elite kit peroxidase (Vector Laboratories). Ovaries were dissected in BSS (Chan and Gehring, 1971), frozen in liquid nitrogen and melted again while adding twofold concentrated SDS/mercaptoethanol sample buffer with 8 M urea and sonicating. Extracts from embryos were prepared using the same procedure.
Results
Injection of bcd mRNA into embryos from bcdE1mutant females
Injection of poly(A)+ mRNA isolated from young wild-type embryos into early embryos from mutant females was shown to rescue the mutant phenotype of several maternal effect genes that regulate pattern along the dorsoventral axis (Anderson and Nüsslein-Volhard, 1984). This approach has not been successful for rescue of the bcd mutant phenotype (Berleth, 1989). Similar to experiments perfomed with the maternal gene easter (Chasan and Anderson, 1989), we tried to substitute the bcd activity by injection of in vitro synthesized bcd mRNA into the anterior tip of embryos from females homozygous for the strong bcdE1 allele. We were not able to induce the formation of any anterior wild-type cuticular structures upon injection of the mRNAs at a wide range of concentrations (0.1 to 5 μ g μ l-1) (Berleth, 1989; Table 1 A and data not shown). The bcd cDNA of the most abundant splice type a (Fig. 1A) that we used as a template for the transcription begins a few bases downstream of a TATA box, contains a polyadenylation signal and is at both ends only a few bases longer than a number of other cDNAs (Berleth et al. 1988). Thus we believe that our template is a full-length cDNA, though the transcription start site has not been mapped. All missense mutations in the cDNA with respect to the genomic sequence have been corrected (Driever and Niisslein-Volhard, 1989).
Translation efficiency of bcd mRNA
We tested the mRNAs by in vitro translation in a wheat germ and a reticulocyte lysate system and found that they were translated very inefficiently (Fig. 2, lanes WG 1 and RL 1). In addition, expression of the same cDNA in Drosophila Schneider cells from the metallo-thionein promoter resulted in low levels of bcd protein (Fig. 2, lane DSC 1). In an attempt to circumvent putative translational control mechanisms that would reduce the efficiency of translation, we exchanged the 5 ′ non-translated sequence of the abundant splice type a of bcd mRNA (Fig. 1) for that of the Xenopus figlobin mRNA (bcdTN3 expression construct). This sequence was chosen because it has neither upstream AUGs nor a strong secondary structure. Tests in all three translation systems showed that bcdTN3 mRNA with the globin leader is translated about 50 times more efficiently than the native bcd mRNAs (Fig. 2). From wheat germ translations, we recovered 41 (±3) times more activity for bcdTN3 mRNA than for bcd mRNA templates. Using reticulocyte lysates, the factor was 51 (±4; average of three determinations each).
Rescue of the bcd mutant phenotype by the injection of bcdTN3 mRNA
The development of anterior structures can be induced by the injection of bcdTN3 mRNA at appropriate concentrations into the anterior tip of embryos from bcdE1 mutant females (Table 1A, Fig. 4 and 5B; see also Driever et al. 1989b). By autoradiography and by in situ hybridization, we find that injected mRNAs maintain a high point at the site of injection for more than one hour (V.S. and D. Ferrandon, unpublished). When we analyse the distribution of bcd protein 90 min after injection, using a monoclonal antibody that detects only the full-length bcd protein derived from the injected mRNA in whole-mount immunostain reactions (Fig. 3 A and B), we detect the formation of a protein concentration gradient similar to the one in embryos from wild-type females.
Injection of bcdTN3 mRNA at a concentration of 0.4 μ g μ l-1 suppressed the formation of telson structures at the anterior in nearly all of the injected embryos, and induced the development of all the anterior structures present in wild-type larvae (acron, head and thorax) in about 30% of these embryos (Table 1 A). About 10% of the larvae hatched, and some developed into adults. Embryos that failed to hatch frequently exhibited fused abdominal segments (A3, A4 and A5 were affected in most cases). Since similar phenotypes can be observed in embryos from females that carry 6 or 8 functional copies of the bcd gene in their genome (Berleth, 1989; our unpublished data), we interpret these defects as being caused by the injection of excessive amounts of 6c<fTN3 mRNA into the embryo. We propose that the defects in abdominal segmentation we obtain are an indirect effect: increased levels of bcd protein lead to a posteriorward extension of the hb protein gradient and thereby might inhibit knirps expression, which is necessary for abdomen formation (H ü lskamp et al. 1989; Irish et al. 1989; Struhl, 1989).
Our findings support the model in which bcd protein concentration directly determines position; an elevated level of mRNA and thus protein results in an enlargement of anterior and a compression of posterior pattern, giving rise to defects in abdominal segmentation. Indeed, during gastrulation in some cases we observed that the cephalic furrow was shifted back as far as 40 % egg length (the normal position of the cephalic furrow is at 65 % egg length; 0 % is at the posterior pole; data not shown).
Induction of anterior development at ectopic positions
Confirming and extending the results from cytoplasmic transplantations (Frohnhöfer and Nüsslein-Volhard, 1986), we found that in vitro transcribcd bcdTN3 mRNA can induce anterior development at any ectopic position along the anterior-posterior axis (Fig. 5). Injection into the posterior pole of embryos from bcd – (not shown) or wild-type females (Table IB and Fig. 5E) suppresses posterior development and induces the formation of head and thoracic structures at the posterior end. Injection of bcdTN3 mRNA at 1.7 μ g μ l-1 into the posterior pole in one case resulted in the development of a complete second head and thorax (Fig. 6). Fig. 3 demonstrates that bcd protein gradients similar to the one in the anterior of wild-type embryos are formed upon the injection of the RNA. The requirement for a 4 to 5 times greater amount of bcd mRNA for the induction of a head at the posterior as opposed to the anterior end of the embryo may be due to the negative effect of the activity of posterior group genes on bcd mRNA stability (Frohnhöfer and Nüsslein-Volhard, 1986; Wharton and Struhl, 1989). The formation of duplicated abdominal segments with mirror-image symmetry (Al in Fig. 5E) appears to be an indirect organizing influence of bcd on abdominal pattern formation, as bcd on its own is not able to induce abdominal structures (Fr ö hnhofer and Nusslein-Volhard, 1986; Lehmann and Frohnhöfer, 1989).
Injection into the middle of embryos from bcd – females results in the formation of gnathal and thoracic structures at the site of injection (Table 1C, Fig. 5 C and D, results summarized in Fig. 4): we were able to identify pro-, meso- and metathoracic structures, antennal and maxillary sense organs, mouth hooks, cirri and other sclerotic structures that could not be assigned unambiguously. The structures usually appear pairwise and are arranged in a bilateral symmetric way transverse to the anterior-posterior axis of the embryo. None of the involuted parts of a wild-type head or acronal structures were identified. This result is consistent with data obtained from the analysis of maternal genes that affect terminal development in the embryo. The formation of acronal structures requires, in addition to bcd, the torso gene group, whose activity is restricted to the termini of the embryo (N ü sslein-Volhard et al. 1987; Klingler et al. 1988).
Is the bcd mRNA leader involved in translational control during oogenesis ?
To investigate the influence of the 5 ′ sequences on translational control, we used a P-transformation vector that allows the expression of bcdTN3 mRNA during oogenesis. The CaSpeR-derived plasmid pPbcdTN3 carries a 2 kb bcd promoter fragment as well as bcd 3 ′ sequences and has the bcdTN3 cDNA sequences inserted at the PstI site just a few basepairs downstream from the putative start site of transcription. We generated 12 transgenic lines in a w1 stock, 4 of which carried insertions on the second chromosome. Those were crossed into bcdE1/TM3 flies. Females homozygous for the strong allele bcdE1 that carry one pPbcdTN3 insertion give rise to progeny with a phenotype similar to that of weak bcd alleles (Fig. 7A; Frohnhöfer and N üsslein-Volhard, 1986). The analysis of the expression pattern of even-skipped (Frasch et al. 1987) reveals that the gnathal and thoracic region is expanded and shifted slightly more anteriorwards than in embryos from mothers heterozygote for a deficiency (Fig. 7B; Fr öhnhofer and Nüsslein-Volhard, 1987). Two copies of pPbcdTN3 in the germline rescue the bcdEl mutant phenotype completely and give rise to fertile adults. We conclude that our construct provides slightly less bcd+ activity to the embryo than a wild-type copy of the gene does. For the in vivo function of the gene, in our assay the 169 bp leader sequence seems not to contain any essential elements.
To test whether the observed translational control of bcd mRNA during oogenesis is due to sequence elements in the leader, we analysed ovaries from females carrying two pPbcdTN3 insertions for the presence of bcd protein. In whole mounts of ovaries, immunohisto-logically stained for bcd protein with a monoclonal antibody, we were not able to detect any specific, bcd-dependent nuclear staining in the nurse cells where the mRNA is synthesized and present at distinct apical patches during early stages of development (data not shown). Neither could we detect any specific staining in the oocyte. Though we used the same method of immunodetection that gives strong signals for bcd protein during syncytial blastoderm stages, we might have failed to detect very low levels of expression during oogenesis. Therefore, we prepared extracts for immunoblot analysis from ovaries of females homozygous for bcdE1 and pPbcdTN3 (Fig. 8). Asa control, we analysed on the same Western blot extracts of blastoderm stage embryos from females of the genotypes pPbcdTN3/pPbcdTN3 +; bcdE1bcdE1, pPbcdTN3/+; bcdEl/bcdE1 and from a strain carrying 8 functional copies of the bcd gene, for which we also prepared ovarian extracts. We estimate that stages of oogenesis where bcd mRNA is detectable contribute to more than 80% of the ovary of a well-fed female. In none of the ovarian extracts were we able to detect bcd protein, although we obtained strong signals for bcd protein from the embryonic extracts. Thus we conclude that, like wild-type bcd mRNA, bcdTN3 mRNA is not translated during oogenesis. The bcd mRNA leader sequence seems not to be required for translational control during oogenesis.
Discussion
Translational control of bcd mRNA
Previous investigations have suggested that bcd mRNA is subject to translational control. Though the bcd mRNA is already present in late previtellogenic follicles (Frigerio etal. 1986; Berleth etal. 1988), bcd protein has not been detected during any stage of oogenesis (Driever and N ü sslein-Volhard, 1988a). Thus bcd mRNA appears not to be efficiently translated before egg deposition. Further, cytoplasm from mature stage 14 oocytes but not from earlier stages of oogenesis, when the bcd mRNA is already detectable, has the ability to rescue the bcd mutant phenotype when transplanted into embryos from bcd– mutant females (Sander and Lehmann, 1988).
Selective translational repression of maternal tran-scripts during oogenesis has been reported in several systems, and different transcripts seem to be regulated by different mechanisms. For example, in Xenopus oocytes some transcripts are bound by specific proteins and are thus inaccessible to the translational machinery, a phenomenon known as maternal masking (Raff, 1980; Richter and Smith, 1984). In other cases, the translational machinery itself (e.g. eIF-4F) is altered, either in activity or availability (Audet et al. 1987; Huang et al. 1987). Finally it has been shown that polyadenylation of mRNAs can affect their translatability (Vassalli et al. 1989; Fox et al. 1989).
When we analysed the translation efficiency of bcd. mRNA and bcdTN3 mRNA in test systems from plant, vertebrate and insect we obtained similar results. Thus we suspected that the primary structure of the bcd mRNA is responsible for the low efficiency of translation. A small open reading frame of 14 codons (including one additional in frame AUG) is located at the 5′ end of the mRNA. Such minicistrons have been proposed to be involved in translational regulation (Kozak, 1984; Mueller and Hinnebusch, 1986). In addition, secondary structure prediction programs (Zucker and Stiegler, 1981) suggest that the bcd mRNA leader might contain GC-rich stem-loop structures just upstream of the start AUG that might be responsible for the inefficient translation (Pelletier and Sonnenberg, 1985). Several alternative structures can be proposed that have the GC-rich region from nucleotide 149 to 155 (sometimes extending till nucleotide 161) base paired with other parts of the molecule (e.g. the region from nucleotide 172 to 185; the AUG starts at nucleotide 170). We think it unlikely that the higher translation rate of the bcd TN3 mRNA may simply be caused by a better translation initiation context, as the natural bcd mRNA has the greater homology to the Drosophila translation initiation consensus (Cavener, 1987).
Replacement of the bcd upstream leader sequence with that of β -globin did not relieve its translational repression during oogenesis. This finding argues against an important role of the leader for the translational block. Which part of the bcd mRNA confers translational control still remains to be elucidated. It is possible that the translational block is closely linked to the different phases of bcd mRNA localization during oogenesis and early embryogenesis (St. Johnston et al. 1989).
bcd protein determines anterior development independent of any other localized, anterior specific morphogenic activity
Though a large body of information on the function of bcd during early embryogenesis has accumulated (Frohnhöfer and Nüsslein-Volhard, 1986; Driever and Nüsslein-Volhard, 1988a,b; Driever et al. 1989a,b;Struhl et al. 1989), some major questions concerning the function of bcd as the morphogen for anterior development had remained. First, it is unclear whether the presence of different spliced forms of the bcd transcript is functionally significant. For example, the different transcripts could code for functionally distinct proteins which would then be responsible for subdividing the embryo into domains. Our data argue against this, since they demonstrate that a single bcd protein species derived from one splice form of bcd mRNA (Type a in Fig. 1) can exert all morphogenetic functions of the bcd gene. The minor splice variants may result from aberrant processing of the mRNA and may not fulfill a specific function during embryonic development.
Second, we could not previously exclude the presence of other, so far unidentified localized anterior activities in the egg, which might cooperate with, or modify bcd activity, possibly in a region-specific manner. The transplantation experiments performed by Frohnhöfer and Nüsslein-Volhard (1986) demonstrate that the localized bcd activity is necessary for the induction of anterior development, but they can not prove that it is sufficient, as other anterior activities might have been cotransplanted.
The induction of anterior structures by the injections of in vitro synthesized bcd mRNAs at ectopic positions reveals that bcd is the only localized morphogenic maternal factor specific for anterior development. Independent of any other localized activity, bcd can induce the formation of gnathal structures and thorax at any position along the anterior-posterior axis. These findings rule out that bcd protein might be subject to region-specific (anterior) modifications in order to generate differentially active bcd protein species. We were able to induce acronal structures at the termini of the embryo only, consistent with previous data (Nüsslein-Volhard et al. 1987; Klingler et al. 1988) demonstrating that terminal structures depend on the activity of the torso group of maternal effect genes. The presence of bcd activity in the terminal region results in the development of anterior terminal structures (the acron) instead of posterior terminal structures (the telson).
Our analysis strongly supports the idea that bcd acts as a morphogen, specifying subregions of the embryo in a concentration-dependent manner (Driever and Nüss-lein-Volhard, 19886). There is no evidence indicating the existence of other, anterior-specific maternal morphogens; indeed all the identified maternal genes affecting anterior development appear to act by modifying bcd mRNA localization or stability and thus bcd protein distribution (exuperantia, swallow, and staufen:Berleth et al. 1988; Driever and N ü sslein-Volhard, 19886; St Johnston et al. 1989; bicaudal: Wharton and Struhl, 1989), rather than presenting a parallel maternal input into anterior pattern formation.
ACKNOWLEDGEMENTS
We thank F. Sprenger for help with the RNA secondary structure analysis, G. Thoma for excellent technical assistance, W.B. Hansen for the pSPBP4 vector and D. Stein for the pCaSpeRbcdBglll vector. We also thank H. Doyle, D. Ferrandon, R. Schnabel, D. Stein and S. Roth for critically reading earlier versions of the manuscript. V.S. is a fellow of the Jane Coffin Childs Memorial Fund for Medical Research. The work was supported by the DFG (Leipniz Programm) and a grant from the Jane Coffin Childs Memorial Fund for Medical Research.