Developmental Biology 347 (2010) 271–278

Contents lists available at ScienceDirect

Developmental Biology

j ourna l homepage: www.e lsev ie r.com/deve lopmenta lb io logy

The Drosophila gap gene giant regulates ecdysone production through specification ofthe PTTH-producing neurons

Arpan Ghosh a, Zofeyah McBrayer a, Michael B. O'Connor a,b,⁎a Department of Genetics, Cell Biology and Development, University of Minnesota, Minneapolis, MN 55455, USAb Howard Hughes Medical Institute, Minneapolis, MN 55455, USA

⁎ Corresponding author. Department of Genetics, CUniversity of Minnesota, Minneapolis, MN 55455, USA.

E-mail address: moconnor@umn.edu (M.B. O'Conno

0012-1606/$ – see front matter © 2010 Elsevier Inc. Aldoi:10.1016/j.ydbio.2010.08.011

a b s t r a c t

a r t i c l e i n f o

Article history:Received for publication 14 May 2010Revised 7 August 2010Accepted 10 August 2010Available online 9 September 2010

Keywords:GiantEcdysoneProthoracicotropic hormone (PTTH)Axon guidanceRing glandProthoracic glandDevelopmental delay

In Drosophila melanogaster, hypomorphic mutations in the gap gene giant (gt) have long been known to affectecdysone titers resulting in developmental delay and the production of large (giant) larvae, pupae and adults.However, the mechanism by which gt regulates ecdysone production has remained elusive. Here we show thathypomorphic gtmutations lead to ecdysone deficiency and developmental delay by affecting the specification of thePG neurons that produce prothoracicotropic hormone (PTTH). The gt1 hypomorphic mutation leads to random lossof PTTHproduction inoneormoreof the4PGneurons in the larval brain. In caseswherePTTHproduction is lost in allfour PG neurons, delayed development and giant larvae are produced. Since immunostaining shows no evidence forGt expression in the PG neurons once PTTH production is detectable, it is unlikely that Gt directly regulates PTTHexpression. Instead,wefind that innervation of the prothoracic gland by the PGneurons is absent in gthypomorphiclarvae that do not express PTTH. In addition, PG neuron axon fasciculation is abnormal in many gt hypomorphiclarvae. Since several other anteriorly expressed gap genes such as tailless and orthodenticle have previously beenfound to affect the fate of the cerebral labrum, a region of the brain that gives rise to the neuroendocrine cells thatinnervate the ring gland, we conclude that gt likely controls ecdysone production indirectly by contributing thepeptidergic phenotype of the PTTH-producing neurons in the embryo.

ell Biology and Development,Fax: +1 612 625 5095.r).

l rights reserved.

© 2010 Elsevier Inc. All rights reserved.

Introduction

The giant (gt) locus codes for a zinc finger containing transcriptionfactor that is widely known for its role in specifying early anterior/posterior pattern in the blastoderm embryo (Capovilla et al., 1992;Eldon and Pirrotta, 1991; Reinitz and Levine, 1990; Stanojevic et al.,1991). Amorphic gt alleles lead to embryonic lethality due to the lossof posterior abdominal segments 5–7 and sometimes 8, as well aslabral and labial structures in the head region. (Gergen andWieschaus, 1985; Mohler et al., 1989; Petschek et al., 1987). However,the original gt alleles were discovered and partially characterized asmutations that produced large larvae as a result of developmentaldelay (Bridges and Gabritschevsky, 1928). These mutants played animportant role in the early history of Drosophila genetics because theyalso produced larger than normal polytene chromosomes that aidedearly cytogenetic studies (Bridges, 1935).

The viable gt alleles exhibit variable penetrance for the develop-mental delay phenotype (~25% females, 13% males) that can beenhanced in females when placed over a deficiency, suggesting thatthey are hypomorphic mutations (Schwartz et al., 1984). These

phenotypically giant larvae exhibit pronounced developmental delayespecially during the third instar stage and pupate approximately5 days later than wildtype (Schwartz et al., 1984).

Post-embryonic development in holometabolous insects is charac-terized by defined molting periods followed by metamorphosis. Theprecise timingof these events is regulatedat a systemic level in responsetomultiple cues suchasnutritional status, body size, organdevelopmentand environmental conditions (Edgar, 2006; Menut et al., 2007; Mirthand Riddiford, 2007; Nijhout, 2003). These cues likely regulatedevelopmental timing in several ways, but ultimately they impingeupon the production and secretion of the insect steroid hormoneecdysone. In gt hypomorphic larvae, the protracted third instar stageappears to result from a delay in the rise of ecdysone titer that precedesthe initiation of metamorphosis, since feeding these animals 20-hydroxecdysone (20-E) reverts the delay phenotype leading to normalsize larvae that pupate at the appropriate time (Schwartz et al., 1984).

Themolecularmechanismbywhichgt controls ecdysoneproductionhas been a long-standing mystery. In many insects, the regulation ofecdysoneproduction in larvae involves twomajor components: a pair ofbilaterally symmetric neurons (PG neurons) located in the cerebrallabrum portion of the brain, and the prothoracic gland, the endocrineorgan that actually produces and secretes ecdysone (Gilbert et al., 2002).In Drosophila, the PG neurons directly innervate the prothoracic gland(Siegmund and Korge, 2001) and induce production and secretion ofecdysone by releasing an adenotropic peptide hormone called

272 A. Ghosh et al. / Developmental Biology 347 (2010) 271–278

prothoracicotropic hormone (PTTH) (McBrayer et al., 2007). PTTHsignals through the receptor tyrosine kinase Torso to activate a RAS/ERKcascade that ultimately stimulates transcription of ecdysone biosyn-thetic enzymes (Rewitz et al., 2009). Intriguingly, elimination of PTTHsignaling delays the rise in ecdysone titer and the onset of pupation byapproximately 5 days resulting in large pupae and adults, similar tothose produced by gt hypomorphs (McBrayer et al., 2007; Rewitz et al.,2009). The similarity in phenotype between gt hypomorphs and loss ofPTTH signaling prompted us to investigate whether gt in some waycontrols PTTH signaling. Here we report that rather than directlyregulating PTTH production in the PG neurons, gt indirectly controlsPTTH and subsequent ecdysone production by influencing the devel-opment of the PTTH-producing PG neurons.

Materials and methods

Drosophila stocks and husbandry

The gt1 and gtE6 lines were obtained from the BloomingtonDrosophila stock center. gt1; ptth-HA stocks were generated by standardgenetic methods. Genomic ptth-HA-50A line (yw; ptth-HA) and the yw;

Fig. 1. Hypomorphic giant mutants are phenotypically similar to PG neuron ablated flies andprolonged third instar stage in about 25% of females and 13% of males. This phenotype, althoThis prolonged third instar stage gives rise to larvae and pupae that are significantly larger thand fused using Photoshop). Consistently, ptth expression, as observed by in situ hybrid(C). Normally developing gt hypomorphic larvae show a variable stochastic loss of ptth expresall four PG neurons similar to wild type animals (E) and most of these animals show no de

Feb211-Gal4; UAS-GFP (Feb211-GFP) line were described previously(McBrayer et al., 2007; Siegmund and Korge, 2001). gt1;UAS-GFP;Feb211-Gal4 larvae were obtained by crossing Feb211-GFP males to gt1

females and selectingmale larvae. Flieswere raised at 25 °C on standardfood in vials.

Immunohistochemistry

The following antibodies were used at the indicated dilutions forimmunohistochemistry: rat anti-HA 3F10 (Roche) 1/500, rat anti-Gt(generous gift from Dr. Vincenzo Pirrotta) 1/500 and mouse anti-CSP(Iowa Hybridoma Bank) 1/100. The Alexa series (Invitrogen) ofsecondary antibodies were used for immunofluorescence at 1/500dilution. CNSs from third instar wandering larvae were dissected outand fixed in 4% paraformaldehyde in PBS for 20 min at roomtemperature for anti-HA and anti-CSP staining. Antibody stainingand washes of larval CNSs were conducted in 0.1% Triton-X100 in 1×PBS (PBST). Primary antibody treatments of CNSs were done at 4 °Cfor 24 h. Embryos were dechorionated in 50% bleach, fixed in 4%paraformaldehyde in PBS for 15 min and all subsequent antibodyreactions were in PBS+0.1% Tween-20. Samples were mounted in

show variable loss of PTTH expression in the PG neurons. gt1 and gtE6 larvae exhibit augh much less penetrant, is similar to PG neuron ablated larvae (ptthNGal4/UAS-grim).an yw control animals (A and B, pictures were taken at same magnification and settingsization, is lost from all four PG neurons in the developmentally delayed gt1 larvaesion ranging from one-three of the four PG neurons (D–E). Many of them express ptth invelopmental delay.

273A. Ghosh et al. / Developmental Biology 347 (2010) 271–278

80% glycerol in PBS and visualized on a Zeiss Axioplan 2 with a CARVunit for confocal microscopy.

Results and discussion

Hypomorphic mutations in giant show stochastic elimination of PTTHexpression in PG neurons

The phenotypes seen in gt hypomorphic alleles (Schwartz et al.,1984), although less penetrant, are remarkably similar to thosedescribed for loss of the PTTH-producing PG neurons or PTTH signaltransduction components. These include a developmental delay andlow ecdysone titers during the prolonged third instar stage,production of large larvae, pupa and adults, and the ability to rescuethese phenotypes by feeding larvae 20HE (Fig. 1A and B, (McBrayeret al., 2007; Rewitz et al., 2009; Schwartz et al., 1984). To examine ifloss of gt affected the expression of PTTH in the PG neurons, we carriedout in situ hybridization using a ptth probe on brains of hypomorphicgt1 or gtE6 mutants prepared from wandering third instar or largefeeding larvae. Similar in situ hybridization experiments with wildtype CNSs show four distinct ptth producing PG neurons, two in eachbrain lobe (McBrayer et al., 2007). However, gt mutant larvacontaining hypomorphic alleles revealed varying degrees of abnormalptth expression depending on the individual larvae and mutant allele.Hemizygous gt1 male mutant larva display the most variability withapproximately one third of the larvae exhibiting a developmentaldelay phenotype that correlates with either one or no cell expressingptth (Fig. 1C and D). Other larvae that were not excessively large,exhibit relatively normal ptth expression (Fig. 1E). Individualscontaining the gtE6 allele frequently exhibited expression withinonly one cell of the pair of bilateral PTTH-expressing neurons withineach brain hemisphere instead of the usual two (Fig. 1F). We conclude

Fig. 2. Giant is expressed in the embryonic brain including two neurons adjacent to the PG neGt show that Gt is expressed in the developing embryonic brain even at late embryonic stageof each brain lobe, positioned ideally to be either the PG neurons themselves or precursorsneurons in the embryo express Gt, PTTH-HA expressing embryos were co-stained withembryogenesis by which time Gt expression fades. However, on rare occasions two Gt expneurons (green arrowheads) (D).

that the similarity in phenotype between ablation of PG neurons andthe gt hypomorphic mutants and the semi-penetrant phenotype ofthe gt hypomorphic allele is likely caused by the partial loss of ptthexpression in the PG neurons. As shown previously (McBrayer et al.,2007), the loss of the ptth neurons leads to a defect in the expressionof many ecdysone biosynthetic enzymes which likely accounts for thelow and delayed peak in ecdysone titer and a delayed onset ofmetamorphosis in gt hypomorphs (Schwartz et al., 1984).

Gt is not expressed in PTTH-producing neurons or the prothoracic gland

Since gt encodes a zinc finger-containing DNA binding factor, onepossible explanation for loss of ptth expression in the PG neurons of gthypomorphic mutants is that Gt directly controls ptth transcription.Alternatively, Gt could indirectly regulate ptth transcription by severalpossible mechanisms. For example, in the late embryo, gt has beenreported to be expressed in the ring gland (Capovilla et al., 1992). Thering gland expression is intriguing since one way to restrict ptthexpression to only the two neurons that innervate the prothoracicgland is by delivery of a required retrograde signal from the targettissue. Such a mechanism is involved in regulating the peptidergicphenotype of the FMRF producing Tv neurons (Allan et al., 2003;Marques, 2003), and perhaps gt expression in the ring gland couldplay an analogous role in regulating a retrograde signal that controlsptth expression.

To address these possibilities, we used immuno-staining toexamine Gt expression in both larvae and embryos. We found noevidence for Gt expression in larval or embryonic prothoracic glands(data not shown) or in the larval PG neurons. In the embryo we detectdynamic expression of Gt in various portions of the developing brain(Fig. 2A, B and D). At stage 14 there are two prominent Gt-expressingneurons in the lateral portion of each brain hemisphere that could be

urons. Co-staining wild-type embryos with anti-Elav (a pan-neuronal marker) and anti-s (A). At stage 14 there are two prominent Gt-expressing neurons in the lateral portionto PG neurons (B and C (red channel from B), yellow arrowheads). To check if the PGanti-HA and anti-Gt antibodies (D). PTTH-HA expression is seen very late during

ressing cells (red arrowheads) could be seen right next to the PTTH-HA expressing PG

274 A. Ghosh et al. / Developmental Biology 347 (2010) 271–278

either the PG neurons or perhaps their precursors (Fig. 2B and C,yellow arrowheads). To examine this issue, we attempted to doublestain embryonic brains for Gt and PTTH using a transgenic line thatexpresses an HA-tagged form of PTTH (McBrayer et al., 2007).However, embryonic expression of PTTH-HA in the PG neurons doesnot begin until approximately stages 17–18, and at this time we nolonger detect consistent Gt expression in the brain. In a few rareembryos we do see simultaneous Gt and PTTH staining within thebrain but they are in adjacent non-overlapping cells (Fig. 2D, red andgreen arrowheads showing Gt and PTTH-HA expressing cells,respectively). We conclude that Gt is unlikely to directly regulateptth transcription in the PG neurons or indirectly regulate ptthexpression via production of a retrograde signal from the prothoracicgland since neither the prothoracic glands nor the PG neurons co-

Fig. 3. Hyomorphic mutations in gt affects expression of an independent PG neuron markerIn a gt1 background Feb211-GFP expression is lost in a stochastic manner identical to ptth inexpression. A comparison of Gt immunostaining intensity between control and gt1 embryoC (1.693 s) and D (1.754 s). At higher magnification, and increased exposure time, the gt1 emb(E and F, yellow arrowheads). Loss of Gt is also seen in a cluster of three cells positioned a

express Gt and PTTH. We suspect that the previously reportedexpression of Gt in the embryonic ring gland (Eldon and Pirrotta,1991) was a misidentified portion of the dorsal brain.

gt1 affects the development of PTTH-producing neurons

In the absence of any evidence supporting a role for Gt inregulating ptth transcription, we sought to determine if loss of Gtaffects the specification of PG neurons. Besides ptth, the only otherdescribedmarker for PG neuron fate is the Feb211-Gal4 enhancer trapline (Siegmund and Korge, 2001) that contains an insertion into anunknown gene on chromosome 3. Analysis of expression from thisenhancer line in gt1 mutant larvae revealed a similar stochastic loss ofGFP expression in different numbers of PG neurons as seen for ptth

Feb211-GFP and may affect PG neuron development by an expression threshold effect.Fig. 1A and B indicating that the fate of the PG neurons may be affected and not just ptths indicates that Gt expression is reduced in the gt1 embryo (C and D, exposure time:ryo in D shows loss of Gt expression in several bilaterally symmetric Gt-expressing cells

nteriorly (E and F, blue arrows).

275A. Ghosh et al. / Developmental Biology 347 (2010) 271–278

expression itself (Fig. 3A and B). The all or none response observed forboth ptth and Feb211-Gal4 expression in gt hypomorphs is consistentwith a stochastic loss of PG neurons in these mutants.

The gt1 mutation has been shown to be associated with twospontaneous insertions, one near the 5′ region of the gene and theother in the 3′ region (Mohler et al., 1989). We predict that theseinsertions likely affect gt expression levels during embryogenesis, andaltered gt expression may affect the specification of different neuronsubtypes within the brain including precursors that give rise to the PGneurons. To examine this issue in more detail, we sought to determineif Gt expression is reduced or if fewer cells express Gt in gt1 mutantanimals compared to wild type embryos. Fig. 3C–F shows acomparison between control and gt1 stage 13 embryos. We observethat under identical staining and exposure conditions, Gt stainingintensity in the control embryo is stronger compared to the gt1

embryo (Fig. 3C and D). The primary staining is in an anterior medialposition that is anatomically close to or overlapping with the parsintercerebralis (PI) and pars lateralis (PL) region of the brain thatgives rise to a number of neurosecretory cells including severalneurons that innervate the corpus cardaicum and corpus allatum, twoother portions of the ring gland (de Velasco et al., 2007). The PI

Fig. 4. gt1 affects fate of the PTTH-producing PG neurons. gt1 and yw control larvae expressing P(cysteine string protein) antibodies. PTTH-HA staining is clearly seen in the PG neurons, and sitethat develop normally (A and D). CSP co-localizes to the sites of innervation by the PG neuronsphenotype do not show any PTTH-HA staining either on the brain lobes where the PG neuronsglands of these larvae indicating a loss of innervation by the PGneurons (H and I). However, Cspby a different set of neurons than the PG neurons (H, yellow arrowheads).

placode derives from neuroepithilium that expresses tailless andorthodenticle, two anteriorly expressed gap genes. The exact origin ofthe PG neurons has not been established, but they may be derivedfrom two other placodes that reside more posterior to the PI region(de Velasco et al., 2007). Interestingly, we note a prominent cluster ofapproximately 5 bilateral posterior midline neurons that express Gt instage 13 embryos (Fig. 3E, yellow arrowheads). In equivalently stagedgt1 mutant embryos (Fig. 3F), the number of cells in this cluster thatexpress Gt is reduced to two to four cells (Fig. 3F, yellow arrowheads).Similarly, staining of a cluster of three cells positioned anteriorly onthe midline axis (Fig. 3E and F, blue arrows) is also dramaticallyreduced in the gt1 sample.

These results suggest that the specification of multiple neuronsubtypes in the brain is likely affected in the gt1 mutant animals. Sincewe have no lineage tracers available to directly determine if the PGneurons are derived from earlier precursors that express Gt, we usedan indirect assay to determine if PG neurons are mis-specified in gt1

hypomorphs. Previous axon tracing experiments have revealed thatthe PG neurons are the only neurons that innervate the prothoracicgland (Siegmund and Korge, 2001). To determine if gt affects thespecification of PG neurons, we examined cysteine string protein

TTH-HA under the endogenous ptth promoter were co-stainedwith anti-HA and anti-CSPs of innervation on the prothoracic gland in the yw control larvae and in gt1; ptth-HA larvaeon the prothoracic glands of these larvae (B, C, E and F). gt1 larvae thatmanifest the “giant”are located or on the prothoracic glands (G). Csp staining is also absent on the prothoracicstaining is still visible on the corpus cardiacum and the corpus allatum that are innervated

276 A. Ghosh et al. / Developmental Biology 347 (2010) 271–278

(Csp) distribution on ring gland cells. Csp is enriched in synapticboutons. As shown in Fig. 4A–C Csp co-localizes with PTTH in axonterminals and boutons on the surface of wildtype prothoracic glandcells as well as in gt1 mutant larvae that still show PTTH expression(Fig. 4D–F). In contrast, developmentally delayed gt1 larvae in whichPTTH expression is absent from all 4 PG neurons, no Csp-containingboutons are observed on prothoracic gland cells (Fig. 4G–I). In thesesame larvae however, Csp-containing axons and boutons are still seenwithin the corpus cardiacum and corpus allatum, two regions of thering gland that are innervated by different sets of neurons (Fig. 4H,yellow arrowheads (Siegmind and Korge, 2001). We conclude thatloss of Gt affects the development of the PG neurons since its absenceleads to a loss of prothoracic gland innervation. At this point, wecannot distinguish if Gt directly affects the specification of PG neurons,or if it affects PG neuron development in a cell non-autonomousmanner, perhaps by affecting cell–cell interactions during early cortexdevelopment. Nevertheless, these experiments add gt to the list ofanteriorly expressed gap genes that affect the specification of theproto-cerebrum (Younossi-Hartenstein et al., 1997).

PG neurons provide a tropic signal to the prothoracic gland

In Manduca sexta PTTH is believed to have a tropic effect on thelarval prothoracic gland as it has been shown to induce generalprotein synthesis (Rybczynski and Gilbert, 1994). Similar to Manducasexta, prothoracic gland cells in Drosophila are mitotically quiescentduring larval stages. Nevertheless, the gland cells exhibit substantialgrowth during the three larval stages and this growth is characterizedby the formation of polytene chromosomes and an increase in sizeof the gland cells (Aggarwal and King, 1969). We observed thatgt1 mutant larvae exhibiting unilateral innervation of the prothoracicglands consistently produced an asymmetrically sized gland in

Fig. 5. gt1 hypomorphs reveal that the PG neurons relay a tropic signal to the prothoracic g(A, yellow arrowhead) mostly innervate only one of the prothoracic glands. The innervated pneurons. DAPI staining of these samples show that the prothoracic gland that is not innervatePG neurons provide a tropic signal to the prothoracic glands. The cells in the non-innervatedsynthesis. To quantify the difference in size, we determined nuclei diameter for 12 nuclei frepresenting PG size. PG size was similarly calculated for innervated and non-innervated PGsis plotted (C) and shows that nuclei within non-innervated PGs are significantly smaller (peffect of variations in PG size between the three samples.

which the innervated portion was significantly larger than the non-innervated side (Fig. 5). Measuring the diameter of DAPI stainednuclei revealed that cells on the non-innervated side contained nucleithat are significantly smaller compared to the innervated side (Fig. 5Band C). This difference was consistently observed in all samples thatfailed to innervate one of the prothoracic glands indicating that DNAsynthesis is likely reduced in absence of prothoracic gland innerva-tion. Curiously, as previously reported (McBrayer et al., 2007), whenboth sides lacked innervation, the ring gland did not appearsubstantially smaller than wild type (compare Fig. 4C and Fig. 4I).However these glands are from developmentally delayed larvae inwhich the extra growth time likely enables them to “catch up” to thewildtype in terms of prothoracic gland size. Ultimately wewill need toexamine PTTH null mutants to prove that PTTH, and not some otherfactor, is the tropic signal secreted from the PG neurons. However, therecent finding that PTTH signals through the Drosophila receptortyrosine kinase (RTK) Torso is certainly consistent with the idea thatPTTH is the tropic factor since the Torso signal is transduced throughthe canonical Ras-Raf-ERK pathway (Rewitz et al., 2009) which isknown to regulate cell proliferation in many systems (reviewed inMcCubrey et al., 2007).

Residual PG neurons show enhanced axon misrouting in gt1 mutantlarvae

In addition to the absence of prothoracic gland innervation inmany gt1 hypomorphic larvae, we noted that there is an enhancedfrequency of axon misrouting in gt mutant larvae that still show ptth-HA expression in one or more of their PG neurons. In wild type larvae,the polarized PG neurons in the left lobe of the brain send out theiraxons from the cell body across the central axis of the CNS to the rightbrain lobe. There the axon forms a loopwith a left hand twist and then

lands that promotes growth. gt1; ptth-HA CNSs containing only one set of PG neuronsrothoracic gland serves as an internal control for any growth promoting effect of the PGd is consistently smaller than the control prothoracic gland (A and B) indicating that theprothoracic gland havemuch smaller nuclei (B red arrowhead) indicating impaired DNArom one PG and considered the average of these measurements as a single data pointfrom three independent brain ring glands complexes. Themean of these measurements-value=0.005). Significance was calculated using a paired student t-test to negate the

Fig. 6. PG neurons in gt1 hypomorphs exhibit an increased frequency of axon misrouting. Anti-HA staining (red) of yw; ptth-HA CNS shows that axon projections from PG neuronsfollow a well defined path with the PG neurons in the left lobe of the brain innervating the right prothoracic gland and vice versa (A, modeled in B). Similar staining of gt1; ptth-HACNSs show that gt1 mutants containing one or two PG neurons in one of the brain lobes often show branching of axons at the base of the ring gland (C, yellow arrowhead). Suchbranching is also observed in gt1; ptth-HA CNSs that retain all the four PG neurons and, in certain cases, leads to excessive innervation of one of the prothoracic glands at the cost ofthe other (D, red arrowheads). An enlargement of the base of the ring gland in C clearly shows the branching (E, yellow arrowhead). For a better understanding of the branchingpattern a model of gt1 CNS showing axon branching is provided (F). While the model shows axons from only the red PG neurons cross innervating the PGs, either side or both sets ofaxon bundles are capable of branching at the base of the PG in gt mutant larva.

277A. Ghosh et al. / Developmental Biology 347 (2010) 271–278

projects anteriorly to innervate the prothoracic gland cells on theright half of the ring gland (Fig. 6A, B and Siegmund and Korge, 2001).Similarly the neurons in the right lobe extend their axons into the leftlobe and innervate the left half of the ring gland. This innervationpattern is most clearly revealed in gt1mutant larvae retaining one pairof the bilateral PG neurons. For example, in a gt1 mutant larva thatretains the right side set of PG neurons, there is innervation onlywithin the left prothoracic gland (Fig. 5A). In wild type larvae, theaxon tracts from each pair of PG neurons are almost parallel to eachother at the base of the ring gland and rarely exhibit cross (only 1 outof 23 CNSs from wt controls showed branching). However, in the gt1

CNSs containing one or two PG neurons in only one brain lobe, weoften see the axons branching at the base of the ring gland andinnervating both prothoracic glands (Fig. 6C, yellow arrowhead).Interestingly we observed similar branching events in gt1 samples thathave all four PG neurons (Fig. 6D and E, yellow arrowhead). Thissuggests that the cross innervations are not likely to be caused by amechanism that tries to compensate for the lack of innervation on oneside of the prothoracic gland. Consistent with this view, we find thatin certain cases such branching events caused excessive innervation ofone of the prothoracic glands at the cost of the other (Fig. 6D, redarrowheads). Approximately 24% of gt1 CNSs that retained at least onePG neurons showed cross innervation events with clear branching atthe base of the ring gland.

These results suggest that gt is required not only for correctspecification of the PG neurons, but also influences the projection ofPG neurites to their target tissue. At present, we cannot distinguish ifthese axon guidance defects represent reduction in the expression ofintrinsic factors within the PG neurons that respond to guidance cuesor whether Gt not only affects the specification of the PG neuronsthemselves, but also surrounding neurons that might provideguidance cues. Ultimately, lineage tracing experiments will berequired to determine which neurons are descendent from Gt-expressing cells in order to address these issues.

Acknowledgments

We thank Naoki Yamanaka, Aidan Peterson and Mary Jane Shimellfor comments on the manuscript. MBO is an investigator with theHoward Hughes Medical Institute.

References

Aggarwal, S.K., King, R.C., 1969. Comparative study of the ring glands from wild typeand 1(2)gl mutant Drosophila melanogaster. J. Morphol. 129, 171–199.

Allan, D.W., St Pierre, S.E., Miguel-Aliaga, I., Thor, S., 2003. Specification of neuropeptidecell identity by the integration of retrograde BMP signaling and a combinatorialtranscription factor code. Cell 113, 73–86.

Bridges, C.B., 1935. Salavary chromosome maps. J. Hered. 26, 60–64.Bridges, C.B., Gabritschevsky, E., 1928. The giant mutation in Drosophila

melanogaster. Part I. The heredity of giant. Z. Indukt. Abstamm. Vererbungsl.46, 231–247.

Capovilla, M., Eldon, E.D., Pirrotta, V., 1992. The giant gene of Drosophila encodes a b-ZIP DNA-binding protein that regulates the expression of other segmentation gapgenes. Development 114, 99–112.

de Velasco, B., Erclik, T., Shy, D., Sclafani, J., Lipshitz, H., McInnes, R., Hartenstein, V.,2007. Specification and development of the pars intercerebralis and parslateralis, neuroendocrine command centers in the Drosophila brain. Dev. Biol.302, 309–323.

Edgar, B., 2006. How flies get their size: genetics meets physiology. Nat. Rev. Genet. 7,907–916.

Eldon, E.D., Pirrotta, V., 1991. Interactions of the Drosophila gap gene giant withmaternal and zygotic pattern-forming genes. Development 111, 367–378.

Gergen, J.P., Wieschaus, E.F., 1985. The localized requirements for a gene affectingsegmentation in Drosophila: analysis of larvae mosaic for runt. Dev. Biol. 109,321–335.

Gilbert, L.I., Rybczynski, R., Warren, J.T., 2002. Control and biochemical nature of theecdysteroidogenic pathway. Annu. Rev. Entomol. 47, 883–916.

Marques, G., 2003. Retrograde Gbb signaling through the Bmp type 2 receptor WishfulThinking regulates systemic FMRFa expression in Drosophila. Development 130,5457–5470.

McBrayer, Z., Ono, H., Shimell, M., Parvy, J., Beckstead, R., Warren, J., Thummel, C.,Dauphinvillemant, C., Gilbert, L., Oconnor, M., 2007. Prothoracicotropichormone regulates developmental timing and body size in Drosophila. Dev.Cell 13, 857–871.

McCubrey, J.A., Steelman, L.S., Chappell, W.H., Abrams, S.L., Wong, E.W., Chang, F.,Lehmann, B., Terrian, D.M., Milella, M., Tafuri, A., Stivala, F., Libra, M., Basecke, J.,

278 A. Ghosh et al. / Developmental Biology 347 (2010) 271–278

Evangelisti, C., Martelli, A.M., Franklin, R.A., 2007. Roles of the Raf/MEK/ERKpathway in cell growth, malignant transformation and drug resistance. Biochim.Biophys. Acta 1773, 1263–1284.

Menut, L., Vaccari, T., Dionne, H., Hill, J., Wu, G., Bilder, D., 2007. A mosaic genetic screenfor Drosophila neoplastic tumor suppressor genes based on defective pupation.Genetics 177, 1667–1677.

Mirth, C.K., Riddiford, L.M., 2007. Size assessment and growth control: how adult size isdetermined in insects. Bioessays 29, 344–355.

Mohler, J., Eldon, E.D., Pirrotta, V., 1989. A novel spatial transcription patternassociated with the segmentation gene, giant, of Drosophila. EMBO J. 8,1539–1548.

Nijhout, H.F., 2003. The control of body size in insects. Dev. Biol. 261, 1–9.Petschek, J.P., Perrimon, N., Mahowald, A.P., 1987. Region-specific defects in l(1)giant

embryos of Drosophila melanogaster. Dev. Biol. 119, 175–189.Reinitz, J., Levine, M., 1990. Control of the initiation of homeotic gene expression by the

gap genes giant and tailless in Drosophila. Dev. Biol. 140, 57–72.

Rewitz, K.F., Yamanaka, N., Gilbert, L.I., O'Connor, M.B., 2009. The insect neuropeptidePTTH activates receptor tyrosine kinase torso to initiate metamorphosis. Science326, 1403–1405.

Rybczynski, R., Gilbert, L.I., 1994. Changes in general and specific protein synthesis thataccompany ecdysteroid synthesis in stimulated prothoracic glands of Manducasexta. Insect Biochem. Mol. Biol. 24, 175–189.

Schwartz, M.B., Imberski, R.B., Kelly, T.J., 1984. Analysis of metamorphosis in Drosophilamelanogaster: characterization of giant, an ecdysteroid-deficient mutant. Dev. Biol.103, 85–95.

Siegmund, T., Korge, G., 2001. Innervation of the ring gland of Drosophila melanogaster.J. Comp. Neurol. 431, 481–491.

Stanojevic,D., Small, S., Levine,M., 1991. Regulationofa segmentationstripebyoverlappingactivators and repressors in the Drosophila embryo. Science 254, 1385–1387.

Younossi-Hartenstein, A., Green, P., Liaw, G.J., Rudolph, K., Lengyel, J., Hartenstein, V.,1997. Control of early neurogenesis of the Drosophila brain by the head gap genestll, otd, ems, and btd. Dev. Biol. 182, 270–283.