Restricted spatial and temporal expression of G-protein α subunits during Drosophila embryogenesis

Given the importance of inductive cellular interactions in the specification of cell fate during development, it is likely that many of the genes expressed during development encode molecules important for intercellular signal transduction events that ultimately lead to changes in gene activity. Consistent with this expectation, many Drosophila genes controlling developmental events have been found to encode proteins involved in cell signaling processes (for reviews see Nusslein-Volhard et al. 1987; Akam, 1987; Ingham, 1988; Banerjee and Zipursky, 1990). For example, knirps (Nauber et al. 1988) and seven-up (Mlodzik et al. 1990) are members of the steroid hormone receptor superfamily. Receptor tyrosine kinases have also been implicated in developmental processes in Drosophila. The torso (Casanova and Struhl, 1989; Klingler et al. 1988) and sevenless (Simon et al. 1989) genes appear to code for transmembrane proteins with structural similarities to the growth-factor receptor kinases.

Of the known cellular signal transduction mechanisms, the most evolutionarily ancient is mediated by a family of heterotrimeric guanine nucleotide binding proteins or G proteins (for reviews see, Gilman, 1984; Stryer and Bourne, 1986; Spiegel, 1987). In this form of signal transduction, a diverse set of membrane-bound receptors with a characteristic seven transmembrane domain architecture modulate the activity of intracellular second messenger generating systems through the activation of intermediary GTP-binding proteins or G proteins. G proteins are membrane-bound heterotrimers consisting of α, βand γ subunits. The a subunit is unique to each class of G protein and is responsible for guanine nucleotide binding and hydrolysis. The βand γ subunits are highly conserved between different G proteins and are required for the inactivation of the a subunit. Additional functions for the β/ γ subunits have also been suggested (Kim et al. 1989; Kurachi et al. 1989). In its inactive form, the α subunit of the G protein is complexed with GDP and β/ γ subunits. Interaction with the appropriate ligand-receptor complex results in the activation of the a subunit by the exchange of GTP for GDP and the dissociation of the β/ γ subunits. The activated α subunit is then able to modulate the activity of the appropriate intracellular second messenger generating systems. A GTPase activity intrinsic to the α subunit subsequently hydrolyzes the bound GTP and returns the αsubunit to the inactive state. Within this scheme, the α subunit is responsible for the specific interaction with both receptor and the effector. Thus, the different classes of G protein have been historically defined on the basis of the functional interaction of specific α subunits with a number of effector proteins. For example, Gs α and Gi α mediate the effects of ligand – receptor complexes, which stimulate and inhibit, respectively, the activity of adenylyl cyclase (Gilman, 1984). Go α is an abundant G protein of both vertebrate and invertebrate nervous systems (Stemweis and Robishaw, 1984; Thambi et al. 1989). One of the functions of Go α appears to be the modulation of the activity of voltage-sensitive Ca²⁺ channels following receptor activation (Holz et al. 1986; Harris-Warrick et al. 1988; McFadzean et al. 1989).

Studies in simple organisms, such as yeast and Dictyostelium, point out the potential role of G-protein-coupled receptor pathways in mediating the transduction of signals leading to the activation of developmental programs in responsive cells. For example, in yeast, developmental processes initiated by exposure to mating pheromones are mediated by G-protein-coupled signal transduction events (Dietzel and Kur-san, 1987). In Dictyostelium, cAMP receptors responsible for initiating a program of cellular aggregation and differentiation have been shown to be coupled to developmentally regulated G proteins (Firtel et al. 1989). In both of these systems, G-protein-coupled sensory transduction is exclusively used to convey signals of developmental consequence. In metazoan organisms, the developmental role of G-protein-mediated signal transduction is likely to be more difficult to elucidate due to the wide variety of signals (peptides, small molecules, fight, odorants, hormones) that use this form of sensory transduction. Given this diversity of ligand –receptor systems, it has been impossible, a priori, to predict which ligand-receptor systems are used to convey developmentally important information. The common component of these systems, however, is their use of a limited number of G proteins to mediate their intracellular effects. The spatial and temporal expression pattern of G proteins should indicate then, specific developmental processes that involve the participation of G-protein-coupled signal transduction.

In this report, we present the first detailed analysis of the developmental expression of G proteins by following the expression of G-protein αsubunits in Drosophila embryogenesis. We have chosen Drosophila for these studies since its development has been well described and it is accessible to experimental manipulation by classical and molecular genetic technologies allowing us to test directly predictions arising from our observations. Although much of the molecular machinery of Drosophila development has been revealed through mutational analysis, this approach may not be feasible for G-protein systems. G proteins mediate many types of intercellular interactions in the adult organism and, probably, during development. It is likely, then, that mutations in G-protein components would be highly pleiotrophic in the adult or result in embryonic lethality and therefore would be difficult to identify specifically by mutational analysis alone. Our approach has been to use vertebrate molecular probes to identify specifically and characterize the developmental expression of the known G-protein α subunits expressed in Drosophila. The spatial and temporal expression pattern of each G α protein has been investigated by molecular techniques and immunocytochemistry using antibodies directed to peptides specifically found in each class of α subunit. Our results indicate that G-protein-coupled sensory transduction may mediate developmental interactions during both early and late stages of embryogenesis and may participate in a variety of specific developmental processes such as the establishment embryonic position, the development of the nervous system and organogenesis.

Canton S flies were grown at 19 °C. Egg collections were made at 25°C on apple juice plates (Wieschaus and N üsslein-Volhard, 1986). For staged embryos, flies were allowed to lay on fresh plates for 1 h and then a fresh plate introduced for the actual collection. Embryos developed at 25°C and were staged according to Campos-Ortega and Hartenstein (1985).

Female abdomens were frozen in OCT and 8 μm cryosections collected on slides coated with 0.1 % gelatin and processed for mRNA localization as described in Quan et al. (1989). RNA probes were generated in vitro (Quan et al. 1989) and labeled using [alpha-[³⁵S]thio]UTP.

Depending on the tissue and antibody to be used, a variety of fixation protocols were followed. Ovaries were dissected out of females and fixed in periodate – lysine-paraformaldehyde (PLP) (Mclean and Nakane, 1974) for 1h at room temperature for detection of Gs α and Go α subunits, or fixed in 100 % ethanol at – 70°C for 1 day (Parks and Spradling, 1987) for localization of Gia subunits. Embryos were fixed by the two-phase method in a manner similiar to that described by Wieschaus and N üsslein-Volhard (1986). Mass-collected and dechorionated embryos were shaken for four minutes in a vial containing a 1:1 mixture of heptane:fixative. For localization of Gs α, the fixative used was PLP. For localization of Go α and Gi α, the fixative was 15% picric acid, 4% paraformaldehyde, 0.1M phosphate pH 7.4. This fixation gave a much improved morphological preservation over PLP without affecting the antigenicity of Go α or Gi α. By contrast, it greatly reduced Gs α antigenicity so consequently for Gs αlocalization we used the weaker PLP fixative with a concomitant reduction in morphological preservation of the tissue.

After fixation, either 6 μm sections collected on gelatin-coated slides or whole embryos were washed and blocked for at least 1h in 10% normal horse serum in PBS. Both sectioned and whole embryos were incubated overnight at 4°C primary antibody diluted into 10% horse serum to a working concentration of 0.25 μgml^-1 anti-RM, 0.77 μgml^-1 anti-GO, and either 0.13 or 0.66 μgml^-1 anti-LD for whole mounts or sections respectively. These affinity-purified antibodies (Goldsmith et al. 1988a,b) were made against vertebrate sequences which are completely conserved in Drosophila (for a list see Wolfgang et al. 1990). Sections were washed and incubated in biotinylated second antibody (7.5 μgml^-1 vector laboratories) for 0.5 h while whole mounts were incubated for 1 h in second antibody diluted to 3.25 μgml^-1 which had been preabsorbed with an excess of embryos. The antibody complexes were localized in sections with peroxidase-ABC ‘elite’ diluted 1/100 (Vector) and in whole mounts with streptavidin-peroxidase diluted 1/1000 (Boehringer-Mannheim). Sections were reacted in 0.01% H₂O₂, lmgml^-1 diaminobenzidine (DAB) (Sigma), 0.1M phosphate buffer pH 7.4 and subsequently intensified with 0.1% OSO₄. For whole mounts, the OsO₄ was omitted and the H₂O₂ used at 0.001 %.

Staged embryos, larvae and pupae were frozen and ground to a powder in liquid nitrogen. The powder was solubilized in 5 M guanidine isothiocyanate, 10mM EDTA, 50mM Tris –HCl (pH7.5), 8% β-mercaptoethanol using a Polytron homogenizer at low speed. RNA was precipitated overnight with 5 – 7 volumes of 4M LiCl. Following centrifugation, the pellet was resuspended in 3ml of 50mM Tris – HCl (pH7.5), 5HIM EDTA, 0.5% SDS and proteinase K at a concentration of 150 μgml^-1. Samples were incubated at 43°C for 45min, adjusted to 150 mM NaCl and extracted 3 times with an equal volume of phenol:chloroform:isoamyl alcohol (50:50:1). RNA was precipitated twice with 2.5 volumes of ethanol and resuspended in DEPC-treated water. RNA was fractionated on 1 % agarose-formaldehyde gels containing 50 mM Hepes (pH7.8), 1 mM EDTA and transferred to Nytran membranes. Restriction fragments were nick-translated using ³²P-dCTP. Hybridizations were done using 50% formamide, 5% SDS, 0.4M NaPO₄ (pH7.2), 1mM EDTA, 1mgml^-1 bovine serum albumen.

Drosophila G α proteins were expressed in E. coli as fusion proteins with the bacteriophage T7 major capsid protein using pET3 vectors (Rosenberg et al. 1987). In these vectors, the expression of fusion proteins is placed under the control of a promoter for T7 RNA polymerase. The construction of a Gi α expression plasmid has been previously described (Wolfgang et al. 1990). In order to construct a Drosophila Gs α expression plasmid, a 1.5 kb Fnu4HI – BglII fragment containing the complete coding region from a cDNA for the long form of Drosophila Gs α (Quan and Forte, 1990) was subcloned into the Bam HI site of pET3a. pET3a/DGs α+ codes for a Gs α protein with an amino-terminal extension of 4 amino acids (Arg-Gly-Cys-Ala), in addition to those encoded by the vector. A restriction fragment containing Drosophila Go α-coding sequences was generated from λDGo59, a cDNA coding for an alternately spliced form of Drosophila Go α (Thambi et al. 1989), using the polymerase chain reaction (PCR). PCR was performed using the sense oligonucleotide 5 ’-CACGGATCCATGGGCTGCACCACATCCGC-3 ’, which incorporates a BamHI site upstream of the Go α-translation initiation codon (underlined), and an opposing, vector-specific primer. After PCR, the Goo-fragment included the BaznHI site of the Bluescript polylinker and was subcloned into the BamHI site of pET3a. pET3a/DGo α+ codes for a Go α-protein with an amino-terminal extension of 2 amino acids (Gly-Ser) in addition to the amino acids of the T7 capsid protein.

Recombinant G α-proteins were expressed in E. coli BL21 (DE3)pLysE and samples prepared for SDS – polyacrylamide gel electrophoresis as described in Wolfgang et al. (1990). Ovaries were hand dissected from adult female flies and homogenized in 100°C SDS sample buffer. Embryos (0 – 2 h) were collected, dechorianated and homogenized in 100°C SDS sample buffer. Approximately 75 μg protein from each was then separated on 10% SDS-polyacrylamide gels, western blots performed and blots probed with the affinity-purified rabbit antibodies all as described in Wolfgang et al. (1990). The LD antibody was generated to a peptide representing residues 159 – 168 of mammalian Gi α1, and RM to a peptide representing residues 385 – 394 of mammalian Gs α and Go to a peptide representing residues 345-354 of mammalian Goa. The sequences are absolutely conserved in the corresponding Drosophila G α proteins.

In Drosophila, three genes have been identified encoding five proteins with high homology to the major classes of vertebrate G α subunits. Two Gs α proteins differing by the addition or deletion of four amino acids are produced by alternate splicing of a single Gs α gene (Quan and Forte, 1990). Two Go α proteins have been identified that are produced by three distinct transcripts from a single Go α gene. The two Go α proteins are produced by use of alternate first exons and differ in 7 of 21 N-terminal amino acids (Thambi et al. 1989; deSousa et al. 1989; Yoon et al. 1989; Schmidt et al. 1989). Finally, a single Gi α gene produces two transcripts (Provost et al. 1988) which result in the production of a single Gi α protein.

Initially, the expression of each G α transcript during development was followed by northern analysis of total RNA prepared from staged embryos, larvae and pupae. Blots were probed with restriction fragments obtained from Drosophila G α cDNAs. These blots were also hybridized with a probe for ribosomal protein 49 to control for equal loading of RNA (data not shown).

In adults, the Drosophila Gs α subunit is encoded by a single alternately spliced transcript of approximately 1.9 kb that is preferentially expressed in the central nervous system (Quan et al. 1989). The 1.9 kb band is absent or present at extremely low levels in early embryos (0 – 2 h) and first appears between 2 and 6h of embryogenesis (Fig. 1A), a time coinciding with the transcriptional activation of the zygotic genome. The embryonic Drosophila Gsatranscript therefore appears to be primarily, if not exclusively, derived from zygotic transcription. Subsequently, the 1.9 kb Gs α transcript is found at relatively constant levels in all developmental stages.

Two Gia transcripts, an abundant 2.3 kb and minor 1.7 kb transcript, have been previously described (Provost et al. 1988). These transcripts are expressed at only low levels in adults, primarily in the abdomen. The highest levels of expression occur in embryos and pupae (Provost et al. 1988). Consistent with these results, the 2.3 kb Gi α transcript is found at peak levels early in embryogenesis (0 – 2h, Fig. 1B), declining thereafter. The greatest decline occurs 10h after egg deposition.

Whole-mount in situ analysis (data not shown) demonstrated a similar time course of expression with message uniformly distributed throughout the embryo. The levels of the 2.3 kb transcript are lowest in larvae and increase slightly during pupation. The minor 1.7 kb Gi α transcript is seen at constant low levels in 0 – 10h embryos and also in pupae. This pattern of developmental expression is consistent with a substantial maternal contribution for Gi α transcripts early in embryogenesis. The zygotic Gi α gene is then expressed at low levels at later stages of development (Fig. 1B).

At least three distinct Go α transcripts are found in adult flies (Thambi et al. 1989; Yoon et al. 1989; de Sousa et al. 1989). A 6.0 kb transcript is specific to adult heads, a 3.5 kb transcript is specific to adult bodies and a 4.2 kb transcript is found in both heads and bodies. These transcripts are each differentially expressed during development (Fig. 1C). The 3.5 kb transcript appears to be exclusively maternal since it is found only in 0 – 2 h embryos. The 4.2 kb transcript is found at variable levels in all stages. Levels of this transcript are extremely low in early embryos (0 – 2 h) and peak in late stage embryos (10 – 18h) and pupae suggesting that this transcript is likely to be derived primarily from zygotic transcription with only a minor maternal contribution. The 6.0 kb transcript, specific to adult heads, appears to be exclusively zygotic in origin as it first appears in 10 – 14 h embryos. This transcript is also found at relatively high levels in pupae.

In order to confirm the maternal origin of the Gi α and Go α transcripts seen early in embryogenesis (0 – 2h), ³⁵S-labeled sense and antisense RNAs were made in vitro from restriction fragments representing each cDNA and hybridized to tissue sections of adult female abdomens. Hybridizations using Gs α probes failed to reveal a specific signal above background in ovaries (data not shown), consistent with the northern results described above suggesting primarily zygotic expression of the Gs α gene (Fig. 1A). As expected for maternally derived transcripts, hybridizations with probes specific for Go α and Gi α revealed high levels of these transcripts in ovaries. Abundant Gi α mRNA was detected at all stages in nurse cells and oocytes (Fig. 2). The distribution of Go α transcripts was similar to that observed for Gi α but transcript levels were reduced (data not shown; de Sousa et al. 1989).

The GO, RM and LD antibodies used in this study are directed to peptide sequences specifically found in mammalian Go α, Gs α and Gi αl, respectively (Goldsmith et al. 1987, 1988a,b ;Simonds et al. 1989). These peptide sequences are also completely conserved in the corresponding Drosophila homologs. Previous western blot analysis of Drosophila head membranes showed that the GO and RM antibodies recognize proteins that correspond to Go α and Gs α. The LD antibody specifically recognized the Drosophila Gi α protein expressed in E. coli (Wolfgang et al. 1990).

In order to further demonstrate the ability of the LD, GO and RM antibodies to discriminate between the Drosophila G-protein family members, cDNAs encoding each of the Drosophila proteins were expressed in E. coli. Although two forms of Drosophila Gs α and Go α protein have been described (Thambi et al. 1989; deSousa et al. 1989; Yoon et al. 1989; Schmidt et al. 1989; Quan and Forte, 1990), the alternate forms do not differ in the regions recognized by the GO and RM antibodies. Thus, only one form of Drosophila Gs α and Go α was expressed in E. coli. These fusion proteins have variable amino terminal extensions as a result of plasmid construction (see Materials and methods). As shown in Fig. 3, each peptide antibody reacts specifically with a single protein in extracts from E. coli expressing the appropriate G α fusion protein. Immunoreactive material is not seen in extracts prepared from cells containing G α coding sequences in the antisense orientation (Fig. 3). In addition, these bands are not present when the antibodies have been preincubated with the cognate peptide or when the antibody is omitted (data not shown).

The expression of each G αprotein during oogenesis was examined immunohistochemically using the RM, GO and LD antibodies in sections of isolated ovaries. No Gs α immunoreactivity was observed in oocytes or nurse cells but was easily detected in the cortex of follicle cells and presumably is associated with cell membranes (Fig. 4A). Despite the relatively high levels of Go α transcript detected in ovaries by in situ hybridization (deSousa et al. 1989), no Go α immunoreactivity was observed in any ovarian cells (data not shown). Gi α immunoreactivity was absent in ovarian follicles until stage 10b (staging according to King, 1970) (Fig. 4B) when high levels are detected in a-few granules in the oocyte (arrowhead) and in the dorsal anterior follicle cells (arrow). In the follicle cells, this is the same time and place where chorion mRNA begins to be expressed (Parks and Spradling, 1987). By stage 12 (Fig. 4C), more Gi α-positive granules have appeared in the ooplasm and all the follicle cells express high levels of the protein. Granules continue to accumulate in the oocyte and are uniformly distributed and very abundant by the end of oogenesis (Fig. 4D).

The distribution of Gs α and Go α during embryogenesis was examined immunohistochemically using the RM and GO antibodies, respectively, in 8 μm sections of staged embryos (Fig. 5). In cleavage stage embryos, prior to the appearance of zygotic transcripts, Gs α and Go α immunoreactivity is at low or undetectable levels compared to controls. Following gastrulation, Gs α reactivity increases and, by germband extension, is associated with all cell membranes (data not shown). Go α remains at low levels during these stages. These levels persist until stage 13 (Fig. 5A,B), when just after the completion of germband retraction, elevated levels of both Gs α and Go α are first detected in the forming neuropil of the brain and ventral ganglion. This pattern persists for the duration of embryogenesis (Fig. 5C,D show a stage 17 embryo). The neuropil staining for both Gs α and Go α persists through adult life (Wolfgang et al. 1990).

Giα distribution throughout embryogenesis was markedly different from that observed for Gs α and Go α. Western blots of proteins present in female ovaries (Fig. 6, lane 3) and 0 – 2 h embryos (Fig. 6, lane 4) indicate that high levels of Gi α protein of the expected relative molecular mass are maternally expressed. Immunocytochemically, high levels of granular Gi α reactivity is detected in three distinct patterns in cleavage stage embryos. Gi α-reactive granules were either uniformly distributed about the periphery of the embryo (Fig. 7A), organized into longitudinal rows in the periphery and partially localized to the posterior pole (Fig. 7B) or, densely packed and entirely localized to the posterior pole (Fig. 7C). All 3 stages occur prior to nuclear migration to the periphery of the embryo. Because staining of nuclei deep inside whole embryos has been problematic to determine the timing and order of these events more precisely, nuclei were counted in serial sections of a number of embryos stained for Gi α. A uniform, peripheral distribution of Gi α was found in embryos up to cleavage stage 3 (4 nuclei). Embryos with partially polarized granules were evident from cleavage stage 2 (2 nuclei) to cleavage stage 4 (8 nuclei). Embryos with fully polarized granules were present from cleavage stage 4, about 36min after fertilization, through the syncytial blastoderm stage when the polarized granules become localized at the yolk cytoplasm boundary. By the end of the blastoderm, the granules can no longer be detected. Thus, very early in development, Gi α is rapidly restricted to the posterior end of the embryo where it persists only through the blastoderm stage.

Gi α immunoreactivity reappears at stage 14 in prospective cardioblast cells at the leading edge of the advancing mesoderm during dorsal closure (Fig. 8A). These cells will form the tube of the dorsal vessel or heart. Staining persists through closure and into the mature embryo (Fig. 8B) and larval stages (data not shown). Staining is restricted to the cardioblasts and is not detected in the pericardial cells or the alary muscles associated with the heart. Near the end of dorsal closure, Gi α appears in two groups of cells at the anterior end of the two rows of cardioblasts. These cells are subsequently incorporated into the forming ring gland in the position corresponding to the prothoracic gland (Fig. 8B,C) (Aggarwal and King, 1969).

Gi α staining is also seen in the chordotonal organs (data not shown) and their precursors in stage 15 and older embryos. The ventral ganglion of stage 17 embryos contains 11 pairs of large dorsal midline cells stained with each pair connected by fibers in each of the thoracic and abdominal neuromeres (Fig. 9A). 1 or 2 fibers descend ventrally from the bundle of fibers connecting the 2 dorsal cells through the neuropil (Fig. 9B) where they approach a small pair of ventral midline cells which also stain for Gi α (Fig. 9C). Additionally, in each abdominal hemiganglion, 1 or 2 additional lateral ventral cells stain (Fig. 9C). A thoracic nerve also showed consistent high levels of staining in late embryos (Fig. 9D). Additionally a number of unidentified cell bodies in the brain and suboesophageal ganglion were stained. The staining in the fibers, but not the cell bodies in the ventral ganglion, could be detected throughout larval life.

In this report, we provide the first detailed analysis of the developmental expression of the three major classes of G-protein α subunit. We expect that the pattern of expression of individual αsubunits during development (summarized in Table 1) indicates the existence of G-protein-coupled sensory transduction systems that potentially convey important developmental signals at these times and places.

Northern blot analysis demonstrates that each of the Drosophila G α homologs is expressed in a restricted and unique temporal pattern during development. Transcripts for the Gs α homolog appear upon activation of the zygotic genome and remain high throughout development. Each of the three Goa transcripts is expressed in a temporally distinct manner. A 3.5 kb Goa transcript appears to be exclusively maternal while a 4.2 kb transcript is found at variable levels in all developmental stages examined. The 6.0 kb Go α transcript, found specifically in the adult nervous system (Thambi et al. 1989), is first expressed at 10 to 14 h of development coincident with the expression of high levels of Go α protein in developing neurons as assessed immunocytochemically. Gi α is predominantly expressed as a maternal transcript with highest levels found in 0 – 2 h embryos.

The spatial and temporal expression of each of the Drosophila G α proteins during embryogenesis has been followed immunocytochemically. Gs α and Go α proteins display the most straightforward pattern. Each is expressed at low levels up through germ band shortening. Gs α is present uniformly, in all cell membranes following gastrulation. Immediately after germ band shortening, expression of each protein is detected in the neuropil as it begins to form. This observation suggests a role for these two G proteins in axonal outgrowth and the establishment of appropriate neuronal connections within the forming nervous system. In other systems, in vivo and in vitro observations support such a role. For example in other invertebrates, serotonin, and thus presumably serotonin receptor – G protein interactions in target neurons, have been shown to modulate axonal outgrowth and the pattern of neuronal connections (Haydon et al. 1984; Haydon et al. 1987; McCobb et al. 1988). Similar observations have been made in vertebrate systems (Lankford et al. 1987). In PC-12 cells and the brains of neonatal rats, Goa has been detected in growth cones and appears to interact with the growth-cone-specific protein GAP-43 (Strittmatter et al. 1990). Also in PC-12 cells, the neural adhesion molecules L1 and N-CAM can modulate phosphotidylinositol turnover by a pertussis-toxin-sensitive mechanism (Schuch et al. 1989).

The spatial and temporal expression pattern of the Drosophila Gi α protein during embryogenesis is most striking. In mature oocytes, Gi α is distributed uniformly in the ooplasm, but upon fertilization becomes concentrated in granules at the periphery of the embryo. This pattern persists through the 2 – 4 nuclei stage. By the 8 nuclei stage, the granules become restricted to the posterior pole of the embryo. This restriction may result from movement or differential loss of granules. The appearance of longitudinal bands does suggest the movement along some cytoskeletal elements. Subsequently, the granules are lost at the time of blastoderm formation despite the continued presence of mRNA, implicating some translational block. During these stages of Drosophila development, many of the products of maternally contributed transcripts are expressed in distinct spatial and temporal patterns that correlate with their participation in the establishment of embryonic polarity and position. (Akam, 1987; Nüsslein-Volhard et al. 1987; Ingham, 1988). One clear interpretation of these observations is that Gia protein may be involved in sensory transduction processes that participate in early developmental events. In contrast to other maternal products such as bicoid (Berleth et al. 1989) or vasa (Lasko and Ashbumer, 1990), Gi α immunoreactivity is absent by the completion of blastoderm formation. Finally, the progressive restriction of Gi α protein to the posterior pole of the embryo indicates that other maternally encoded factors are responsible for the correct spatial localization of the Gi α protein. Examination of Gi α expression pattern in other maternal mutants provide an easy test for such additional factors.

Subcellular fractionation studies in other systems have shown that Gi α can be localized to specific granules (Rotrosen et al. 1988) and the cytosol (Bokoch et al. 1988) as well as the plasma membrane. Light level and electron microscopical immunocytochemistry has demonstrated that Go α-(Gabrion et al. 1989; Peraldi et al. 1989), Gi α (Lewis et al. 1990) and Gz α-(Hinton et al. 1990) are found in intracellular accumulations in a variety of vertebrate cell types. These intracellular accumulations of Gi α-are thought to represent a surface translocatable pool of G proteins that, when mobilized, provide a mechanism by which cells can regulate receptor activity.

At later stages of development, high level of Gi α-are associated with advancing cardioblasts during dorsal closure, in the developing the chordotonal organs, a proprioceptive structure and in a subset of cell bodies in the ventral ganglion. Gi α is also expressed in the prothoracic gland portion of the ring gland which is the source of the molting hormone ecdysone. The prothoracic glands release ecdysone in response to the polypeptide hormone prothoracicotropic hormone (PTTH), which stimulates changes in the levels of Ca²⁺ and cAMP in prothoracic gland cells (Bollenbacher and Granger, 1985). Gi αmay then either couple directly to the PTTH receptor or to receptors that modulate the effect of PTTH during the molting cycle.

This report demonstrates that the spatial and temporal restriction of G-protein α subunits expression during Drosophila embryogenesis is consistent with a role for this form of sensory transduction in mediating the effect of signals important for both early embyronic determination and the organogenesis of specific tissues such as the nervous system and heart. Interestingly, mutation and molecular analysis have already resulted in the identification of a G α-like protein, the product of the concertina gene, which appears to play a specific role in a signal transduction pathway used during gastrulation (Parks and Wieschaus, 1991). Clearly, the biochemical and genetic manipulations possible in the Drosophila system will allow us to test whether the G proteins that we have identified and the signal transduction systems that they modulate also have roles in specific aspects of development.

We thank Drs Allen Spiegel and Paul Goldsmith for then-generosity in providing the affinity purified antibodies used in this study and Dr Anthony Mahowald for critical reading of the manuscript. F.Q. has been awarded a Canadian MRC postdoctoral fellowship. This work was supported by grant no. NS 27684 from the NIH to M.F.