Evolution of neuroblast identity: seven-up and prospero expression reveal homologous and divergent neuroblast fates in Drosophila and Schistocerca

A central goal in developmental biology is to understand how development has evolved to generate diversity among organisms. We are addressing this question by comparing early neurogenesis in Drosophila and Schistocerca (grasshopper) embryos. These evolutionarily distant insects are most similar at completion of a segmented germband during embryogenesis, the phylotypic stage for these organisms, while both earlier and later events are different (Sander, 1976; Patel et al., 1989, 1992; Dawes et al., 1994). In Drosophila, the formation and fate of early CNS precursors (neuroblasts; NBs) are controlled by the pair-rule class of segmentation genes, including fushi tarazu (ftz) and even-skipped (eve) (Skeath et al., 1992). In Schistocerca, homologues of eve and ftz have been cloned, and neither is expressed in pair-rule stripes (Patel et al., 1992; Dawes et al., 1994). The differences in pair-rule gene function between Drosophila and Schistocerca raise an interesting question in CNS evolution: do differences in pair-rule gene function lead to different cell fates among early NBs, or can different patterning mechanisms produce homologous NBs?

In both Drosophila and Schistocerca, a similar geometrical array of approximately 60 NBs forms in each segment (Bate, 1976; Doe and Goodman, 1985; Doe, 1992). Although position is one aspect of a NB’s identity, equivalent cell fate or cell lineage cannot be inferred based on positional correspondence alone. Here, we present a molecular comparison of the NB pattern in Drosophila and Schistocerca using position, time of formation and expression of four molecular markers to assay NB identities.

In Drosophila, patterns of gene expression in subsets of NBs have provided the basis for unambiguous identification of individual NBs (Doe, 1992). Expression of homologous genes in Schistocerca NBs offers a more accurate method for comparing NB identity. Known gene expression patterns can be used as molecular markers for NBs in both insects. The genes involved include engrailed (en; Condron et al., 1994) and a ftz-related gene, Dax (Dawes et al., 1994). To increase our number of NB markers, we have cloned the Schistocerca seven-up (svp) gene, which is dynamically expressed in NBs in both insects (Doe, 1992; Broadus et al., 1995). In addition, we use an antiserum to the prospero (pros) protein (Matsuzaki et al., 1992), which labels the MP2, MP1 and median NB in Drosophila (Doe, 1992; Broadus et al., 1995); this antiserum cross-reacts with the Schistocerca pros protein and labels the same CNS precursors.

svp, named for its role in Drosophila eye development, encodes a predicted steroid receptor transcription factor that shows high homology to the human COUP-TF (Mlodzik et al., 1990). Homologues to svp have been identified in several other organisms including grasshopper (this study), zebrafish (Fjose et al., 1993), sea urchin (Chan et al., 1992), chick (Lutz et al., 1994) and mouse (Jonk et al., 1994). A common feature in all organisms examined is svp expression in the developing nervous system. While svp is expressed in all Drosophila and Schistocerca NBs, it is especially useful as a marker for identifying individual NBs because the onset of svp expression is precisely regulated in each NB; some NBs are svp-positive at formation, whereas other NBs show svp expression at a specific time after formation. In contrast, the pros, en, and ftz/Dax genes show early and persistent expression in identified subsets of NBs.

Using these four molecular markers (seven-up, pros, en, and ftz/Dax), we find that some early NBs are homologous between insects: they have similar position, time of formation, and time of gene expression. These data show that different embryonic patterning mechanisms (with or without the pair-rule functions of ftz and eve) can generate similar NB fates. In contrast, several later-forming NBs show differences in position and/or gene expression, suggesting that their identity has diverged. Therefore, evolution of the insect CNS has occurred in part through altering NB pattern and fate.

The Schistocerca cDNA library (provided by Michael Bastiani, University of Utah) was constructed from embryos at 40% of development using the λZAP vector (Stratagene). A 2 kb DNA fragment, which included the entire Drosophila svp coding region and short flanking 3′ and 5′ untranslated regions, was radioactively labeled and used as probe. 600,000 plaques were screened at low-stringency conditions (McGinnis et al., 1984). Following secondary screening, selected positives were excised from the λZAP vector as PBS-KS clones according to the Stratagene protocol. Excised clones were sequenced on both strands using automated fluorescent cycle sequencing with Taq DNA polymerase (University of Illinois DNA Sequencing Center). GenBank accession number: BankIt 16159 U36622.

Schistocerca embryos were provided by Michael Bastiani (University of Utah) and Melody Seigler (Emory University). Embryos were staged according to Bentley et al. (1979). Embryos at 20-30% development were removed from egg chambers into 1× PBS, and the amnion surrounding the embryo was dissected away. Embryos were fixed 50 minutes in 1× PBS, 50 mM EGTA, 9.25% formaldehyde. Embryos not intended for immediate use were dehydrated through methanol series, and stored at −20°C in ethanol. For use, embryos were rehydrated in a 1:1 mixture of methanol and 5% formaldehyde in PTw (1× PBS, 0.1% Tween-20), refixed for 25 minutes in 5% formaldehyde in PTw, and then washed in PTw 5× 5 minutes. Embryos were incubated 10 minutes in a 1:1 mixture of PTw and prehybridization solution (50% formamide, 4× SSC, 250 μg/ml tRNA, 0.1% Tween-20, 1× Denhardt’s solution, 5% dextran sulfate) followed by 10 minutes in prehybridization solution. Freshly denatured herring sperm DNA was added at a concentration of 500 μg/ml, and prehybridzation was allowed to proceed for >1 hour at 55°C.

Digoxigenin-labeled RNA probes were prepared according to Boehringer Mannheim’s Genius System protocol. Alkaline hydrolysis of the probe in 0.1 M carbonate buffer (pH 10) for 40 minutes at 60°C was used to reduce probe size. Embryos were hybridized with probes for >24 hours at 55°C. Posthybridization washes were done in 50% formamide, 4× SSC, 0.1% Tween-20 over 12-16 hours at 55°C. Hybridization and posthybridization washes were performed without rocking. Embryos were transferred to PBT (1× PBS, 0.2% bovine serum albumin, 0.1% Triton-X) and allowed to equilibrate over 30 minutes. All subsequent steps were performed at room temperature with rocking. Alkaline phosphatase-conjugated anti-digoxigenin (Boehringer Mannheim) was diluted 1:2000 in PBT, and the embryos were incubated in the diluted antibody solution for 1-2 hours. Unbound antibody was removed by washing embryos 6× 10 minutes in PBT. Embryos were transferred to AP buffer (100 mM NaCl, 50 mM MgCl₂, 100 mM Tris, pH 9.5, 0.1% Tween-20) and allowed to equilibrate over 5 minutes. The alkaline phosphatase conjugate was detected by addition of 4.5 μl of 75 mg/ml NBT and 3.5 μl of 50 mg/ml X-phosphate per ml AP buffer. Development of the AP reaction product was stopped by several rinses in PBT. Embryos were cleared in glycerol and flattened under a coverslip to bring most NBs into one focal plane prior to photography.

Embryos were dechorionated in 50% Clorox for 2 minutes, rinsed with water, and then fixed for 30 minutes in a 1:1 mixture of heptane and 1× PBS, 50 mM EGTA, 9.25% formaldehyde. Embryos were devitellinized by cracking in methanol. Embryos not intended for immediate use were transferred to ethanol prior to storage at −20°C. In situ hybridization of RNA probes was identical for Schistocerca and Drosophila embryos. Following staining, embryos were cleared in glycerol and flat-mounted.

The svp transcript can be detected earlier than the lacZ gene product (β-galactosidae) expressed from a svplacZ enhancer trap, and therefore this report documents svp expression in some NBs one stage prior to that described for svplacZ (Doe, 1992; Broadus et al., 1995).

Schistocerca embryos were dissected in 1X PBS and then fixed for 50 minutes in PEM-FA (3.7% formaldehyde, 0.1 M Pipes (pH 6.9), 1 mM MgSO₄, 2 mM EGTA). Embryos not intended for immediate use were dehydrated in methanol and stored at −20°C. Immediately prior to use, embryos were rehydrated in PBT (1× PBS, 1% BSA, 0.1% Triton-X). Embryos were blocked for 1 hour in PBT + 5% normal goat serum (NGS). pros protein was detected using a 1:100 dilution of rabbit polyclonal serum generated against the 14 amino acids at the carboxyl terminus of the Drosophila pros protein (Matsuzaki et al., 1992). Prior to use, primary antibody was preadsorbed as a diluted solution by incubation with an excess of embryos for 1 hour at room temperature. Preadsorbed primary antibody was incubated with embryos for staining either overnight at 4°C or for 2 hours at room temperature. Embryos were then washed in PBT 6 × 10 minutes at room temperature. Biotinylated secondary antibody was diluted 1:200 in PBT + 5% NGS and incubated with embryos for 1 hour at room temperature. For histochemical detection of the antibody, we used the Vectastain ABC kit and SG chromagen (Vector Laboratories, Inc.) according to the vendor’s protocols. Following staining, embryos were cleared by transfer through 50%, 70%, 90% glycerol in 50 mM Tris (pH 8). Embryos were viewed on a Zeiss Axioplan compound microscope.

This paper uses a revised nomenclature for Drosophila NBs (Broadus et al., 1995), which is modified from Doe (1992). The changes include new names for: NB 1-1, to reflect the fact that this NB makes the aCC/pCC lineage; NB 1-2, to optimize alignment of gene expression between Schistocerca and Drosophila; and NBs 2-2 and 2-3.

A single Schistocerca svp clone was isolated in a screen of 600,000 plaques (see Materials and Methods). This 3 kb clone contained 363 nucleotides of open reading frame (Fig. 1A) and about 2.6 kb of 3′ untranslated sequence. The open reading frame aligned with Drosophila svp in the ligand-binding domain, which encodes the C terminus of the protein. Conceptual translation of the Schistocerca DNA sequence shows 95% amino acid identity compared to Drosophila svp, and 91% amino acid identity compared to human COUP-TF (Fig. 1B). The high amino acid identity between the Schistocerca predicted protein, Drosophila svp, and human COUP-TF strongly suggests that this clone corresponds to the Schistocerca svp cDNA.

We used this clone to prepare RNA probes with which to examine svp expression in the Schistocerca embryo. In both Drosophila and Schistocerca, svp is expressed in the central and peripheral nervous systems, in a patterned subset of cells behind the morphogenetic furrow of the developing eye, and in the fat body (Figs 2, 3; Mlodzik et al., 1990; Hoshizaki et al., 1994). These data provide additional evidence that our Schistocerca clone is a bona fide homologue of Drosophila svp.

To compare individual NB fates between insects, we used three criteria that contribute to a NB’s unique identity: time of formation, position and gene expression. The position and timing of NB formation are stereotyped, and have been previously documented for Schistocerca and Drosophila (Bate, 1976; Doe and Goodman, 1985; Campos-Ortega and Hartenstein, 1985; Doe, 1992). In both insects, NBs form a grid-like array of rows and columns. Schistocerca NBs are aligned in orderly rows and columns, while Drosophila NBs are more compacted. NBs have been given number designations according to position within the segment. For example, NB 5-2 lies in row 5 and column 2 in the final NB pattern. Between insects, number identities imply positional similarity, but not necessarily molecular or lineage homology. svp is transiently expressed in NBs of both Drosophila and Schistocerca, and is a particularly useful marker for individual NBs because the onset of svp expression is precisely regulated during the generation of each NB lineage.

At approximately 2% of development after an individual segment is formed, svp expression is observed in a subset of NBs (Fig. 3A,C). The first svp-positive NBs (3-2, 4-1, 5-2, 7-4) delaminate from svp-positive ectodermal clusters (Fig. 3C inset). The later-arising NB 7-2 is also observed to delaminate from a svp-positive ectodermal cluster. Each of these NBs is svp-positive at formation and therefore prior to the first mitotic division. The amount of svp transcript is variable according to position, with NB 5-2 and the overlying ectodermal cluster consistently showing most intense expression (Fig. 3C).

NBs 2-5, 3-5, 4-4, 5-3, 6-2, and 7-1, which are among the first NBs to form, show delayed onset of svp expression (Fig. 3B,D). The NB progeny, GMCs, can be identified by cell morphology, position immediately dorsal to NBs, and nuclear expression of pros protein (Fig. 5B). These assays show that Schistocerca NBs initiate mitosis immediately following delamination. Therefore, we conclude that NBs which do not delaminate from svp-positive ectodermal clusters express svp only after at least one cell division is complete. Eventually, all NBs and some GMCs transiently express svp. Soon after the NB array is complete, svp expression is extinguished in most NBs (Fig. 3E,F). Approximately 5-8% of development later, many NBs reinitiate svp expression. This second burst of svp expression in NBs persists until at least 50% of development, although we have not characterized this later expression in detail.

svp expression during Drosophila neurogenesis has a spatial and temporal pattern similar to that observed in Schistocerca. Early in embryonic stage 9 (staging according to CamposOrtega and Hartenstein, 1985) the first 10 (S1) NBs form in each hemisegment (Doe, 1992). Among this first group, NBs 3-2, 5-2, and 7-4 express svp at formation, each having delaminated from a svp-positive ectodermal cluster (Fig. 4A,C). Each of these NBs expresses svp at formation and therefore at the beginning of their cell lineage. The level of transcript in each NB and the overlying ectodermal cluster is stereotyped, with 5-2 consistently showing most intense expression (Fig. 4C).

At mid-stage 9 there are a total of 16 NBs (10 S1 NBs and 6 S2 NBs). The S2 NB 7-2 is also observed to delaminate from a svp-positive ectodermal cluster. In contrast, three of the S1 NBs (2-5, 5-3, and 7-1) initiate svp expression during the S2 stage of neurogenesis (Fig. 4B,D), at which time they have completed one or more cell divisions (Hartenstein et al., 1994). During the final three stages of NB formation (S3-S5), all NBs transiently express svp (Fig. 4E,F). Some NBs delaminate from a svp-positive ectodermal cluster, whereas other NBs express svp only after dividing one or more times. As svp-positive NBs divide, some GMCs also express svp. By stage 12 most CNS expression has been extinguished (data not shown). The pattern of svp expression in several early-forming Drosophila NBs is similar to that of Schistocerca; the comparison will be presented below, following a description of three additional NB markers: pros, en, and ftz/Dax.

In Drosophila, nuclear pros expression identifies a small subset of CNS precursors: MP1, MP2 and the median NB (Doe and Technau, 1993; Broadus et al., 1995). Here we show that an antiserum raised against the C-terminal region of Drosophila pros (Matsuzaki et al., 1992) cross-reacts with Schistocerca pros, and labels the same subset of Schistocerca CNS precursors. In both insects, pros is also detected in the nucleus of all newly-born GMCs (Fig. 5; Matsuzaki et al., 1992; Vaessin et al., 1991; Spana and Doe, 1995).

In Schistocerca, the earliest pros expression in the CNS is in the MP2 precursor (Fig. 5A). We identify the pros-positive precursor as MP2 based on its characteristic position and symmetrical cell division. MP2 is positioned in what will be the medial NB column; it forms before any NBs have delaminated, just as ectodermal segmentation becomes visible in that segment. At approximately 2% of development following its formation, MP2 divides symmetrically to produce a pair of neurons that also express pros: dMP2 and vMP2 (Fig. 5B,C; Bate and Grunewald, 1981). The dMP2/vMP2 neurons lie immediately dorsal to the newly formed NB 4-1; this is exactly the position of these neurons in Drosophila (Broadus et al., 1995). pros also shows nuclear expression in the unpaired MP1 precursor and the median NB. The early-forming MP1 precursor is identified by its position precisely between the pros-positive MP2 precursors and MP2 progeny (Fig. 5). The later-forming median NB is identified as the unpaired midline NB in row 7 (data not shown).

The en protein is known to be expressed in a similar pattern of NBs in Schistocerca and Drosophila (Condron et al., 1994; Broadus et al., 1995). In both insects, en is expressed in all NBs in rows 6 and 7 and in a single NB in the first row of the next posterior segment. en is also expressed in GMCs and neurons lying just dorsal to the en-positive NBs.

The ftz/Dax proteins are also known to be expressed in a similar pattern of CNS precursors in Schistocerca and Drosophila (Dawes et al., 1994; Broadus et al., 1995). In both insects, ftz/Dax is expressed in MP1 and its progeny, MP2 and its dMP2/vMP2 progeny, and a stereotyped subset of GMCs and neurons. In the MP2 lineage, ftz/Dax is expressed just prior to division of MP2 (Fig. 6E; Dawes et al., 1994); in Drosophila, division of MP2 is delayed for several hours following its formation, and expression of ftz/Dax is similarly delayed (Fig. 6F; Doe, 1992; Spana et al., 1995).

Using molecular markers for identified NBs in both Drosophila and Schistocerca, we find that the early pattern of NBs is highly conserved. In contrast, several later-forming NBs show differences in position or timing of gene expression. In Drosophila, NBs segregate from the ectoderm in 5 groups or waves (Doe, 1992). In contrast, Schistocerca NBs delaminate in a continuous sequence (Doe and Goodman, 1985). The two Schistocerca NB arrays shown in Fig. 6A,B were selected as the closest comparative stages to Drosophila S1 and S2 based on the number of NBs and patterns of gene expression. We find that the pattern of svp RNA is highly conserved in NBs of Drosophila and Schistocerca; in addition, the precise timing of svp expression within the cell lineage of identified NBs is often conserved between the two insects. For example, NBs 5-2 and 7-4 express svp at formation and therefore at the onset of their cell lineage. In both insects, NBs 5-2 and 7-4 have the same relative position (5-2 is medial and just anterior to the en stripe; 7-4 is lateral and within the en stripe), delaminate from a similar svp-positive cluster (5-2 delaminates from the ectodermal cluster with the highest level of svp expression, whereas 7-4 delaminates from an ectodermal cluster with a low level of svp expression), and have similar en expression profiles (5-2 is en-negative, 7-4 is en-positive). In contrast, NBs 5-3 and 7-1 complete at least one cell division prior to initiating svp expression. In both Drosophila and Schisotcerca, NBs 5-3 and 7-1 delaminate as S1 NBs, are similarly positioned (5-3 forms just lateral to NB 5-2; 7-1 forms at the posterior medial corner of the hemisegment), show similar en expression (5-3 is en-negative; 7-1 is en-positive), and delay svp expression until the S2 NB stage.

In addition to svp, the pros, en, and ftz/Dax genes show virtually identical expression between insects (see previous sections, Fig. 6). For example, in the Drosophila CNS, nuclear pros protein is observed in MP2, the unpaired MP1, the unpaired median NB, and all newly-born GMCs (Spana and Doe, 1995; Broadus et al., 1995). MP2 delaminates with the earliest forming NBs (at late embryonic stage 8); and expresses ftz just before it divides nearly symmetrically at stage 11 to produce two pros-positive neurons: dMP2 and vMP2 (Spana and Doe, 1995). In Schistocerca, MP2 forms and divides earlier than in Drosophila, but similarly expresses pros at formation and Dax just before dividing to produce the dMP2/vMP2 neurons (Figs 5, 6).

Using several criteria to assay NB fate (position, time of formation and temporally accurate gene expression), Drosophila and Schistocerca NBs 1-2, 5-2, 5-3, 7-1, 7-2, 7-4, and MP2 appear to be genuine homologues. The fates of other NBs have seemingly diverged or could not be assayed with the molecular markers used.

Although there are clear similarities between gene expression in Drosophila and Schistocerca NBs, there are some differences in timing and/or position of expression (Fig. 6). For example, NB 3-2 expresses svp at the time of its formation in both insects, but the NB initially delaminates at the anterior edge of the hemisegment in Drosophila and in the middle of the hemisegment in Schistocerca. In addition, pairs of adjacent NBs (6-2/7-1, 4-1/5-2) express svp in Schistocerca, whereas in Drosophila there is only one svp-positive NB at each position (NBs 7-1 and 5-2 respectively). Additional molecular markers, or ideally lineage analysis, will be necessary to determine the precise relationship between these NBs.

The phylotypic stage for arthropods is based on morphological comparisons of development in diverse species. Molecular analysis has supported the idea that complete segmentation of the embryonic germband is a highly conserved developmental stage: ectodermal expression of en foreshadows the formation of the posterior segmental boundary in all arthropods examined, while earlier gap and pair-rule divisions are not strictly conserved (Patel et al., 1989, 1994; Dawes et al., 1994). Do nonectodermal tissues also show a conserved developmental stage? This study presents a molecular comparison of early neurogenesis in Drosophila and Schistocerca, and we find that the pattern of early NBs is highly conserved. Therefore, the completion of segmentation and the formation of early NBs are not only temporally linked events, and both may be similarly constrained during evolution.

In Drosophila, the pair-rule genes regulate expression of the proneural genes in neuroectodermal clusters that foreshadow the sites of NB formation (Skeath et al., 1992). Schistocerca homologues of the Drosophila pair-rule genes ftz and eve have been cloned, and neither is expressed in a pair-rule pattern (Patel et al. 1992; Dawes et al., 1994). How is the stereotyped pattern of NB formation initiated in Schistocerca? Despite the absence of eve and ftz pair-rule function in Schistocerca, the position of NB delamination and early gene expression in NBs is highly conserved. In both insects, svp is expressed transiently at reproducible timepoints in identified NB cell lineages. Between insects, some NBs appear to be genuine homologues in that they occupy similar positions and share timed expression of several developmentally important genes: svp, pros, en, and ftz/Dax. Perhaps most impressive is the conserved dynamics of svp expression in individual NB cell lineages. In both insects, NBs 5-2, 7-2 and 7-4 delaminate from svp-positive ectodermal clusters and therefore express svp during the earliest part of their cell lineage. NBs 5-3 and 7-1 in both insects express svp only after completing one or more cell divisions or midway through their lineage. Therefore, both a conserved pattern of early NBs, as well as ectodermal expression of en in each segment, are characteristic of the phylotypic stage of embryogenesis (Fig. 7). The genes controlling early neurogenesis in Schistocerca are entirely unknown, but are clearly different than those used in Drosophila (Fig. 7).

How does neurogenesis unfold in different insects as they develop past the phylotypic stage? Previous studies have led to the proposal that there is a conserved plan for embryonic neurogenesis in arthropods (Thomas et al., 1984 but see Whitington et al., 1993). Thus, the origin of species-specific CNS differences among insects have been attributed to different modes of embryonic development: holometabolous insects undergo metamorphosis while hemimetabolous insects continue embryonic development as a miniature adult form (Truman et al., 1993). In the holometabolous Drosophila, NBs arrest mitotic cycles near the end of embryogenesis, and then reactivate to produce additional neurons during larval stages (Prokop and Technau, 1991). In contrast, the hemimetabolous Schistocerca embryo shows continuous development such that embryonic NBs undergo many more divisions than observed in Drosophila embryonic neurogenesis (Bate, 1976; Shepherd and Bate, 1990). Using molecular markers for individual NB identities, we detect differences in position and/or timing of gene expression, especially among later-forming NBs. Therefore, while the pattern of early NBs is highly conserved, several later-forming NBs appear to have divergent fates. Changes in embryonic NB pattern and fate demonstrate evolution of CNS differences prior to the dramatic changes that occur at metamorphosis.

Aspects of the mature embryonic CNS are known to be conserved between Drosophila and Schistocerca: gene expression in identified neurons (Bastiani et al. 1987; Dawes et al., 1994; Doe et al., 1988a,b; Patel et al. 1992; Zinn et al., 1988), neuronal position, and axon morphology (Thomas et al., 1984; Kuwada and Goodman, 1985). Differences are also observed. For example, within each hemisegment, eve is expressed in a single laterally positioned cluster of neurons (EL cluster) in Drosophila, and is expressed in two lateral neuronal clusters in Schistocerca (J. B., unpublished results). Differences in the pattern of neurons may arise due to changes in how individual neurons are specified. Alternatively, or in addition, differences in the pattern of neurons may reflect earlier alterations in NB identity. Lineage tracing studies of the median NB in Drosophila and Schistocerca demonstrate that some differences in the CNS are attributable to changes in the lineage generated by identified NBs (Bossing et al., 1994; Condron and Zinn, 1994). We have uncovered differences in gene expression between positionally homologous Drosophila and Schistocerca NBs, which may result in differences in NB lineages between insects.

We have compared early neurogenesis in Drosophila and Schistocerca using several criteria to assay CNS precursor identity: time of formation, position, and gene expression. Although the pattern of early NBs is highly conserved between insects, the pattern shows increasing dissimilarity as additional NBs delaminate. These data highlight two important differences between Drosophila and Schistocerca NBs: first, while many early NBs appear homologous between insects, the pattern of early NBs is established by different mechanisms; and second, the identities of some later-forming NBs appear to have diverged, demonstrating that evolution of the insect central nervous system can occur through changes in embryonic neuroblast identity.

We thank Michael Bastiani for Schistocerca embryos and the λZAP Schistocerca cDNA library; Yash Hiromi for the Drosophila svp cDNA; Fumio Matsuzki for anti-prospero; Melody Seigler and Barry Yedvobnick for the opportunity to work in their laboratories at Emory University; and Lisa Nagy, Jim Skeath and Jon Henry for comments on the manuscript. This work was supported by an NIH Systems and Integrative Biology predoctoral traineeship (J. B.) and NSF PYI and HHMI grants (C. Q. D.); C. Q. D. is an Assistant Investigator of the Howard Hughes Medical Institute.