The Zic family member, odd-paired, regulates the Drosophila BMP, decapentaplegic, during adult head development

The adult head of Drosophila is a complex sensory and feeding structure. It is largely formed from paired eye/antennal imaginal discs, which fuse with the clypeo-labral and labial discs during metamorphosis to form a complete head (Bryant, 1978). The cells that form each disc are set aside during embryogenesis and undergo proliferation and pattern formation during larval life, before differentiating during pupation.

The origin of the eye/antennal disc is complex, arising from multiple embryonic segments (Jurgens and Hartenstein, 1993). Its establishment of cell lineage restrictions differs from other discs, with the dorsoventral boundary arising before that of the anteroposterior (Baker,1978; Morata and Lawrence,1978; Morata and Lawrence,1979). Multiple functionally distinct structures, such as the eye,antenna, maxillary palpus and head cuticle, arise from the eye/antennal disc. The primordia for these structures appear to be specified within the developing disc by localized patterns of signaling molecules and regionally restricted expression of transcription factors(Cavodeassi et al., 1999; Kenyon et al., 2003; Pai et al., 1998; Pichaud and Casares, 2000; Royet and Finkelstein, 1996; Royet and Finkelstein,1997).

Discs are comprised of two epithelial layers; a cuboidal disc proper and an overlying squamous epithelium called the peripodial membrane or peripodial epithelium. While the traditional view has been that head structures arise primarily from the disc proper (J. L. Haynie, PhD thesis, University of California, 1975) (Haynie and Bryant,1986), recent data suggest that the peripodial epithelium plays a significant role in the development of the eye/antennal disc(Cho et al., 2000; Gibson et al., 2002; Gibson and Schubiger, 2000). At metamorphosis, the paired eye/antennal discs fuse into a single vesicle and undergo complex morphogenetic movements inside the pupal body cavity to form the head capsule, which everts to form the final adult head(Fristrom and Fristrom, 1993; Milner and Haynie, 1979). How these morphogenetic movements come about and their relationship to the underlying pattern elements is not understood.

We have undertaken a genetic analysis of adult head formation in Drosophila. Our entrée into this was a specific class of decapentaplegic (dpp) cis-regulatory mutations that affect only the adult head capsule (Stultz et al., 2005). dpp is the Drosophila homolog of Bone morphogenetic proteins (BMPs) 2 and 4, and the major TGFβ-like protein in the fruitfly. The enhancer elements disrupted in these mutations direct expression of Dpp in the lateral peripodial epithelium of eye/antennal discs, and loss of this expression in third instar eye/antennal discs results in defects in the ventral head capsule(Stultz et al., 2006). We carried out an extensive genetic screen to recover genes that interact with dpp to form the ventral adult head (D.A.H., unpublished). Here we describe that one interacting gene resulting from this screen is odd-paired (opa). opa is a pair-rule gene(Jurgens et al., 1984; Nusslein-Volhard et al., 1985)that encodes a zinc finger protein with homology to a family of mammalian transcription factors, the `Zinc finger protein of the Cerebellum', or Zic family (Aruga et al., 1996; Benedyk et al., 1994; Cimbora and Sakonju, 1995). Zic family members have roles in neurogenesis, myogenesis, skeletal patterning and left-right axis formation. In addition, a major role appears to be controlling region-specific morphogenesis of the brain (reviewed by Aruga, 2004; Grinberg and Millen,2005).

In humans, mutations in Zic genes cause several congenital cerebellar and head malformations, such as the Dandy-Walker malformation(Grinberg et al., 2004) and holoprosencephaly (Brown et al.,1998). In the fly, opa is required for the parasegmental subdivision of the embryo, where it activates wingless and engrailed in all parasegments(Benedyk et al., 1994). opa is also required for the formation of all constrictions of the embryonic midgut, and it is negatively regulated by dpp in this tissue (Cimbora and Sakonju,1995). However, the postembryonic role of opa is completely unknown.

Here we investigate the role of opa in adult head formation, and its connection to dpp in this process. We find that opa is expressed in the eye/antennal disc, primarily in the peripodial epithelium of this structure. Loss of opa function during eye/antennal disc development results in defects in ventral head structures identical to those observed with loss of dpp. Expression of a peripodial-specific dpp β-galactosidase reporter constructed from DNA from the cis-regulatory region disrupted in dpp head capsule mutations is lost in cells that do not express Opa, indicating that opa positively regulates the peripodial dpp expression associated with ventral head development. Targeted misexpression of Opa causes ectopic expression of peripodial dpp, and dramatic head malformation. These data indicate that dpp is regulated by opa, either directly or indirectly, in a previously unknown pathway of head morphogenesis,carried out in the peripodial epithelium, and suggests that the Zic family role in head formation may be part of a conserved function also seen in insects. Interestingly, holoprosencephaly caused by Zic mutations in humans is autosomal dominant and has incomplete penetrance. This behavior has been postulated to be caused by digenic inheritance, with modifier genes enhancing the penetrance of the Zic holoprosencephaly defect(Ming and Muenke, 2002). The dominant genetic interaction we have observed between opa and dpp, both of which have vertebrate homologs implicated in holoprosencephaly, suggest that Drosophila head development may be a model for this complex developmental genetic defect.

The following Drosophila melanogaster strains were used for this paper: opa^12.3, opa^32.3, opa^Q,opa^M (this work), opa^3D246(Cimbora and Sakonju, 1995)(provided by Shigeru Sakonju), opa⁷(opa^IIC, provided by Steve DiNardo, University of Pennsylvania Medical School, PA), opa⁸(opa^2P32, provided by Trudi Schupbach, Princeton University, NJ), opa^ts125 (Bloomington Stock Center), Df(3R)6-7, Df(3R)Z1, Df(3R)110 (provided by Steve Wasserman,University of California, San Diego, CA), Df(3R)63 (provided by Steve DiNardo), dpp^s-hc1, dpp^TgR46.1, Df(2L)DTD2, P20(Stultz et al., 2005), SH53(dpp^s-hc-lacZ) (Stultz et al., 2006), ey-FLP (Bloomington Stock Center), hsFLP;Sco/Cyo (provided by Mark Mortin, NICHD/NIH, MD), FRT82B opa⁷ (this work), FRT82B Ubi-GFP (Bloomington Stock Center), FRT82B Sb[63b] M(3) 95E Pr Bsb (provided by Jim Kennison,NICHD/NIH, MD), UAS dpp (provided by William Gelbart, Harvard University, MA), UAS opa (this work), MS1096-Gal4(Milan et al., 1998) (provided by Patrick Callaerts), c309-Gal4 (Bloomington Stock Center), and ey-Gal4 (provided by Francesca Pignoni, Harvard Medical School, MA). Flies were maintained on standard media and crosses were maintained at 25°C. Temperature shift experiments were done using opa^ts125, in combination with opa LOF alleles. Crosses were temperature-shifted after first instar larval stage from 18 to 29°C. Adult heads were fixed in 70% ethanol and mounted in Euperol.

Scanning electron microscopy was carried out using standard method as described (Stultz et al.,2006).

β-Galactosidase activity was detected by X-Gal staining as previously described (Blackman et al.,1991). Discs were mounted in Aquamount (Gurr), and examined using DIC. Adult heads were fixed in 1% gluteraldehyde for 10 minutes, washed in PBT before staining. Heads were incubated at 37°C in staining solution, plus 0.12% X-Gal dissolved in N,N-dimethyl formamide and 0.15% Triton X-100. Immunohistochemistry was performed according to Carroll and Whyte(Carroll and Whyte, 1989). Mouse anti-β-galactosidase and anti-Dachshund (Developmental Studies Hybridoma Bank) were used at 1:25 and rabbit anti-Odd-paired (provided by Steve DiNardo) at 1:300. Secondary antibodies conjugated to Alexa Fluor 488 or Alexa Fluor 555 (Molecular Probes) were used at 1:1000. Nuclei were visualized with DAPI. Discs were mounted in Vectashield (Vector Laboratories) and imaged with a Radiance confocal microscope.

RNA probes were labeled with digoxigenin (DIG) using the DIG RNA Labeling Kit (Boehringer Mannheim). Antisense probe was transcribed by T7 RNA polymerase from pKSII-opa (see Plasmid constructs), hydrolyzed to an average length of 200-500 bp (Cox et al.,1984), precipitated in ethanol and dissolved in hybridization solution. Imaginal discs were treated with ethanol, xylene and acetone, as described (Nagaso et al.,2001). RNA was detected with anti-DIG antibody (Roche Diagnostics), and signals were detected using fluorescent secondary antibodies(Molecular Probes) and imaged as above.

opa cDNA was synthesized by PCR using an embryonic cDNA library(Brown and Kafatos, 1988) as a template. PCR products were inserted into pBluescript II KS(+) (Stratagene). This pKSII-opa was used for DIG-labeled RNA probe. An EcoRI fragment of opa cDNA was cloned into the pUAST vector(Brand and Perrimon, 1993). Transgenic flies were created by standard protocols(Spradling, 1986).

Homozygous mutant opa somatic clones were generated by the FLP/FRT technique (Xu and Rubin,1993). An opa⁷ mutation was recombined with a P[neo-FRT]82B chromosome to obtain P[neo-FRT]82B opa⁷. Recombinant chromosomes were selected by resistance to G418 and lethality over opa alleles. To produce opa clones in the imaginal discs, P[neo-FRT]82B opa⁷ was mated with eyFLP; P[neo-FRT]82B Ubi-GFP. Discs were dissected and analyzed by confocal microscopy. opa clones were marked by loss of GFP. To obtain opa clones in the adult head, P[neo-FRT]82B opa⁷ was crossed to hsFLP; P[neo-FRT]82B Sb[63b] M(3) 95E Pr Bsb. Clones were induced at 72, 96 or 120 hours after egg-laying by 1 hour heat shock at 38°C. opa clones were identified by loss of markers for Stubble,Prickly and Blunt short bristle.

We carried out a screen to recover mutations at dpp that specifically affected the adult head capsule(Stultz et al., 2005). Several mutations unlinked to dpp were recovered that behaved as dominant enhancers of the dpp head capsule phenotype. Four of these mutations were mapped to the third chromosome and, in addition to causing a dominant interaction with dpp, also caused recessive lethality. Cytologically visible breakpoints of three of these were localized to region 82 on the polytene chromosomes, while the fourth was mapped meiotically to a similar region (Table 1). Deficiencies in this region were used to confirm that their recessive lethality maps to this position and suggested that odd-paired was a likely candidate. All four mutations fail to complement each other and known opaalleles (Table 1). They all displayed embryonic lethality and pair rule defects when crossed to each other and to known opa alleles (data not shown).

The lethality and dominant interaction with dpp are caused by mutations in a single locus. These four mutations, as well as known opa loss-of-function (LOF) alleles and deficiencies that remove opa, all interact dominantly with dpp head capsule mutations to cause head defects (Table 1). In dpp head capsule mutations, indicated as dpp^s-hc, the ventral head is disrupted, the eye is round rather than oval, the sensory vibrissae on the ventral side of the eye are bunched, the maxillary palps are missing, reduced or duplicated, and gena tissue appears to be missing (Fig. 1B) (Stultz et al.,2005). This phenotype is also seen in mutant combinations of opa LOF alleles with dpp^s-hc mutations(Fig. 1C). In addition, when a temperature-sensitive opa allele, opa^ts125, is used to remove opa function postembryonically, similar defects of the ventral head capsule are seen (Fig. 1D), along with occasional antennal defects (see Fig. S1 in the supplementary material). These data indicate that opa plays a postembryonic role in the formation of the adult ventral head capsule, and that dpp and opa function together in this process.

We performed RNA in situ hybridization to the eye/antennal disc using labeled opa as a probe. In the antennal disc proper, opa RNA is expressed in a ring roughly coincident with the primordium of the first antennal segment (Fig. 2A). It is also found on the lateral side of the peripodial epithelium in the antennal disc, overlying the primordia of the maxillary palpus and rostral membrane (J. L. Haynie, PhD thesis, University of California, 1975)(Bryant, 1978; Haynie and Bryant, 1986)(Fig. 2D). Peripodial expression extends into the eye disc on the medial and lateral side. Additional expression may also exist in the peripodial epithelium of the eye disc, but it is not seen consistently. An enhancer trap from the 3′ end of the opa gene, opa^3D246(Cimbora and Sakonju, 1995),shows similar β-galactosidase expression in the ring corresponding to the presumptive antennal disc proper, and expression in the palps area of the peripodial epithelium (Fig. 2B,C). We have made five large β-galactosidase reporter constructs from across the opa gene, extending from approximately 5 kb upstream from the protein-coding region to approximately 6 kb past the end of the most 3′ coding exon (Fig. 3A). Three of these demonstrate significant expression in the eye/antennal disc (Fig. 3B-D). This expression is similar to that seen with our RNA localization and approximates cumulatively to the expression of the opa^3D246 enhancer trap, with a ring of expression in the disc proper, and significant peripodial expression along the lateral side of the disc. The expression of such constructs is first detectable at the late second instar larval stage, and increases subsequently (data not shown).

dpp expression responsible for the formation of the ventral head capsule is limited to the lateral peripodial epithelium of the eye/antennal disc (Stultz et al., 2006)(Fig. 4A; Fig. 5A); thus opaexpression in the eye/antennal disc is in the same epithelial layer as dpp head capsule-related expression and on the same side of the disc as the primordia of the future ventral adult head, although according to the fate map these structures derive from the disc proper and not the peripodial epithelium (J. L. Haynie, PhD thesis, University of California, 1975)(Haynie and Bryant, 1986)(Fig. 2D). These data place opa, a gene related to zinc-finger transcription factors, in the correct location to be involved in the regulation of peripodial-specific dpp expression.

To assess the effect of opa on dpp transcription, we examined the expression of a dpp β-galactosidase reporter construct, SH53, in the presence of opa mutations. For simplicity, we will refer to this reporter as dpp^s-hc-lacZ. The DNA in this reporter comes from the dpp head capsule enhancer region, which resides in the 5′ cis-regulatory region of the gene. It most accurately reflects the dpp expression in the lateral peripodial epithelium of the eye/antennal disc (Fig. 4A) that is responsible for the role of dpp in ventral head formation (Stultz et al.,2006). In genotypes in which opa was removed postembryonically, using opa^ts125 at non-permissive temperature, we observe a reduction in reporter expression, indicating that opa is a positive regulator of dpp expression in this location (Fig. 4B). This reduction was also observed in dpp^s-hc/+; opa/+ mutant combinations(Fig. 4C,D), in agreement with the dominant interaction between opa and dpp^s-hcdescribed above (Fig. 1C; Table 1). Removal of peripodial dpp expression via homozygous dpp^s-hc mutants produces an identical result (Stultz et al., 2006). dpp has previously been observed to autoregulate its own expression in several different tissues, including the eye/antennal disc proper (Biehs et al.,1996; Chanut and Heberlein,1997; Hursh et al.,1993; Pignoni and Zipursky,1997; Wiersdorff et al.,1996). These data indicate that the transcription of dppin the lateral peripodial epithelium requires positive inputs from opa and the dpp signal transduction pathway, and that the disruption in ventral head formation observed in opa,dpp^s-hc or dpp^s-hc/+; opa/+mutant combinations all correlate to reduction of dpp expression in the lateral peripodial epithelium. Expression was most strongly reduced in a region near the lateral junction of the antennal and eye discs in all these mutant combinations, midway along the band of dpp peripodial expression. It is noteworthy that loss of opa function has no effect on other dpp expression in the eye/antennal disc, as monitored by the BS3.0 β-galactosidase reporter construct(Blackman et al., 1991) (data not shown). The BS3.0 reporter construct reflects dpp expression controlled by the 3′ cis-regulatory region of the dppgene. Mutants from this region of the dpp gene(dpp^disk alleles) do not interact with opa or dpp^s-hc alleles(Stultz et al., 2005),indicating the regulatory autonomy of the lateral peripodial expression domain. This indicates that opa positively regulates dppspecifically in the lateral peripodial epithelium, and is involved with dpp signal transduction related to ventral head capsule formation,and not with other contributions of dpp to eye/antennal disc morphogenesis.

As we were working with dpp^s-hc/+; opa/+ combinations, and a temperature-sensitive opaallele, which might have residual opa function, we wished to see if complete loss of opa function would produce similar results on the dpp reporter construct. Homozygous opa⁷ clones,lacking endogenous opa activity, were generated by using the FLP/FRT system (Xu and Rubin, 1993). Somatic clones in the eye/antennal disc were produced in eyFLP/+;P{neoFRT}82B P{Ubi-GFP}83 Ubi-GFP / P{neoFRT}82B opa⁷ larvae and detected by loss of GFP. The eyFLP construct has considerable expression in the peripodial epithelium of the eye/antennal disc(Gibson et al., 2002). The expression of dpp^s-hc-lacZ was lost in all opa⁷ LOF clones (Fig. 5B-D). There was complete correlation between LOF opaclones, and loss of dpp^s-hc-lacZ expression, no matter what region within the expression pattern the clone appeared, unlike our experiments produced with the opa conditional allele, or with dpp^s-hc/+; opa/+ mutant combinations,where the midpoint of expression seemed most sensitive to loss of dppor opa function. Both large clones(Fig. 5B,C) and small clones(Fig. 5D) displayed this correlation. These data suggest that opa is required cell-autonomously for dpp expression in the lateral peripodial epithelium. We did, however, recover larger clones at higher frequencies at the junction of the eye/antennal disc. This is the same region that exhibited the most pronounced decline in dpp^s-hc-lacZ expression in opa temperature-shift experiments and dpp^s-hc/+; opa/+ mutant combinations. Clones at the posterior end of the dpp^s-hc-lacZ expression were recovered somewhat less frequently, and clones in the most anterior region of dpp^s-hc-lacZ expression were recovered rarely, and were quite small (Fig. 5E). We believe this is due to the expression pattern within the peripodial epithelium of the eyFLP construct. It is most heavily expressed in the ventral peripodial epithelium at the connection between the eye and antennal disc,overlapping dpp^s-hc-lacZ expression, but has more limited expression in the peripodial epithelium of the ventral antennal disc, and does not significantly overlap with dpp^s-hc-lacZ in that region. These data extend our observation that opa is a positive factor required for dpp expression in peripodial epithelium of the eye/antennal disc, and further indicate that this effect behaves in a cell autonomous manner.

We also wished to look at the effect of LOF opa clones on the morphology of the adult head to see if the mutant phenotypes observed with LOF clones resembled the data obtained with the opa conditional allele,or with dpp^s-hc/+; opa/+ mutant combinations. Somatic clones in adults were induced 72-120 hours after egg-laying by 1 hour heat shock at 38°C of flies with the genotypes hsFLP/+; P{neoFRT}82B P{piM} Sb[63b] M(3)95E Pr Bsb / P{neoFRT}82B opa⁷. Defects in the adult head were observed only in clones produced in the ventral portion of the adult head(Fig. 6A,B). LOF opaclones in the dorsal head, identified by their bristle phenotype, never displayed abnormal morphology (Fig. 6C). Results were similar whether recombination was induced using eyFLP or hsFLP constructs. Loss of ventral eye and rostral membrane tissue was observed, as well as bunched vibrissae, and missing or misplaced maxillary palps (Fig. 6A,B). Missing or malformed antennae were often seen in adult clones, as had been seen in some opa conditional allele experiments, but rarely in dpp^s-hc/+; opa/+ mutant combinations and never with homozygous dpp^s-hc mutants. With the exception of antennal defects, this spectrum of defects is similar, although more extreme, to that observed with opa^ts125 in temperature shift-experiments and dpp^s-hc/+; opa/+mutant combinations. LOF opa clones more closely resemble defects observed in strong dpp^s-hc mutants(Fig. 1B). Antennal defects,however, seem to be specific to opa alone. In both temperature shifts and LOF opa clones, multiple segments in the antennae are affected(Fig. 6A,B and see Fig. S1 in the supplementary material), although the defects are more severe in LOF opa clones, which often manifest complete loss of antennal structures. We attribute antennal defects to the expression of opa in the antennal disc proper, although the defects extend further than the fate map would predict for the discreet ring of opa expression. However,the function of opa in antennal development does not appear to be part of the opa/dpp pathway involved in ventral head development. These data further support the postembryonic role of opain forming the ventral adult head, and the interaction of opa and dpp in ventral head formation.

We wished to see if ectopic expression of Opa was capable of inducing dpp expression as monitored by the dpp^s-hc-lacZreporter. A full-length opa cDNA in a UAS expression construct was ectopically expressed in imaginal discs, using the Gal4 expression constructs MS1096 and c309. The MS1096-Gal4 driver expresses in the peripodial epithelium and margin cells of the lateral and medial sides of the eye/antennal disc (Fig. 7B) (Bessa and Casares,2005). Its ventral expression overlaps with that of the dpp^s-hc-lacZ reporter (compare Fig. 7A with B). Ectopically expressing Opa with the MS1096-Gal4 driver results in modest ectopic expression of the dpp^s-hc-lacZ reporter in the peripodial epithelium on the medial side of the disc(Fig. 7C). We compared the areas of ectopic dpp^s-hc-lacZ reporter expression with the presence of Opa, using an Opa antibody that is capable of detecting only overexpressed Opa. Areas where ectopic reporter expression is detected by cytoplasmic expression of β-galactosidase were associated with the presence of nuclear Opa protein (Fig. 7C,D,E). Nuclear Opa expression was also seen in areas with no ectopic β-galactosidase expression (white arrows, Fig. 7C,D). These data suggest that Opa can induce dpp reporter expression in the peripodial epithelium in a cell-autonomous fashion, but that it either requires other factors that are not present in all cells, or that the presence of repressors prevents Opa activation in some areas. The c309-Gal4 driver expresses primarily in the disc proper, in the antennal portion of the disc, and behind the morphogenetic furrow (Fig. 7G). Modest ectopic expression of the dpp^s-hc-lacZ reporter is also seen with this driver. This expression is limited to the peripodial epithelium, and is associated with a small region of peripodial-specific expression of Opa, as identified by Opa antibody (Fig. 7I,J). Note that in both cross sections (Fig. 7F,J), Opa-positive nuclei in the disc proper are not associated with β-galactosidase expression. Taken together, the results from both drivers suggest that Opa has some ability to induce the expression of the dpp^s-hc-lacZ reporter ectopically, but this ability is limited to the peripodial epithelium. Additionally, not all areas that are Opa-positive within the peripoidal epithelium are capable of expressing dpp^s-hc-lacZ, while Opa expressed in the disc proper does not seem to induce dpp^s-hc-lacZ expression in any area. We conclude that Opa is not sufficient to induce reporter expression in all cells that express the protein, either through lack of secondary factor(s), or due to active repression in most areas of the disc. However, these results are consistent with data obtained by the LOF clones, suggesting that opais a cell-autonomous activator of peripoidal dpp expression in certain regions of the peripodial epithelium of the eye/antennal disc.

Adult heads of flies in which Opa was ectopically expressed using either the MS1096 or c309 drivers were morphologically normal (data not shown). We also used an ey-Gal4 driver to ectopically express Opa to see what effect this had on dpp^s-hc-lacZ. The ey-Gal4>Opa eye/antennal disc is dramatically altered in shape,making it hard to assess the effect of Opa ectopic expression on the reporter. The antennal disc is duplicated, as indicated by markers of antennal structures such as dachshund (Fig. 8F), and cut (data not shown), and the eye disc is eliminated, with only a small amount of tissue remaining(Fig. 8F,G). dpp^s-hc-lacZ also appears bifurcated in the antennal region, but this may to be due to the fate change caused by Opa ectopic expression rather than by induction of additional dpp^s-hc-lacZ expression(Fig. 8G). The gross malformations in disc structure are reflected by defects in the adult head. ey-Gal4>Opa eliminates the eye. In the antennae, there are obvious duplications of aristae, and antennal segments(Fig. 8B). Maxillary palps are also duplicated (Fig. 8B,C). To determine if these dramatic head malformations were primarily attributable to the induction of ectopic Dpp expression in the peripodial epithelium by Opa,we expressed Dpp using the same driver. ey-Gal4>Dpp also causes a dramatic alteration in the antennal disc. dpp^s-hc-lacZexpression is broader, extending medially, and dachshund expression indicates partial antennal duplication(Fig. 8H,I). However, unlike ey-Gal4>Opa, ectopic expression of Dpp on this driver does not eliminate the eye disc, although it is reduced in size, and the retinal field is enlarged at the expense of the rest of the disc. These disc alterations are reflected in the defects seen in adult heads. The head is significantly reduced in size, with misplaced and duplicated antennae and maxillary palps(Fig. 8D). However, the eyes,while reduced, are not eliminated, and protrude from the head. The dorsal head, although also reduced in size, retains much of its normal appearance. The ey-Gal4 driver has a band of expression in the peripodial epithelium extending from the extreme anterior portion of the antennal disc into the eye disc (see Fig. S2A in the supplementary material). However, it also has extensive eye disc proper expression, particularly behind the morphogenetic furrow, in the developing retina (see Fig. S2B in the supplementary material). The difference between ey-Gal4>Opa and ey-Gal4>Dpp may be attributable to different effects of these two proteins in the eye disc proper. Opa appears to eliminate the eye disc when expressed in the disc proper, while Dpp promotes retina formation, a well-described function of dpp in eye development(Dominguez and Casares, 2005). However, the effects observed for both Opa and Dpp on the antennal disc are similar, thus may be attributed to ectopic expression in the peripodial epithelium, suggesting that the major target of Opa in this tissue layer is Dpp.

The expression of the ey-Gal4 driver is limited to the eye/antennal disc, but we have observed that Opa causes morphological defects in all discs when misexpressed with Gal4 drivers that express in all discs. Opa appears to be a protein with potent morphological abilities.

The primordia of the majority of the adult head structures are reported to arise from the disc proper (Haynie and Bryant, 1986). This would imply that the effects seen by disruption of peripodial opa or dpp expression are caused solely by disruption of the signaling required to support morphogenesis of structures derived from the disc proper. To ask if the effect of opaand dpp on the adult head was a secondary consequence of disrupted signaling to the disc proper, or whether cells in the peripodial layer contributed directly to structures in which we saw defects in opa and dpp^s-hc mutants, we examined pharate adult heads for expression from the dpp^s-hc-lacZ and our opa-lacZ reporter constructs. As monitored by histochemical detection of β-galactosidase activity, cells from the peripodial epithelium must persist in the adult head cuticle. In dpp^s-hc-lacZ, the expression of which is limited to the peripodial epithelium, significant β-galactosidase activity is seen in the ventral head, including the anterior rostral membrane, maxillary palps and a distinct area ventral to the prefrons, which abuts the clypeus(Fig. 9B). Light expression in the third antennal segment, and base of the second antennal segment is seen with this construct. We cannot currently explain this expression, as we do not see antennal defects in dpp^s-hc mutations alone. The primordia of the Proximal rostral sensilla (Prst) have been placed by the fate map in the lateral peripodial epithelium(Haynie and Bryant, 1986)within the domain of dpp^s-hc-lacZ reporter expression.β-galactosidase expression is seen in this region in posterior adult heads bearing the dpp^s-hc-lacZ construct(Fig. 9C). The expression of the opa02 and opa03 lacZ reporter constructs is also limited to the peripodial epithelium (Fig. 3B,C), and both show ventral head expression similar to the dpp^s-hc-lacZ reporter, with punctate expression in the maxillary palps (Fig. 9D,E). These two constructs have almost no expression in the antennae. The opa04 lacZ construct, which has more extensive lateral expression in third instar discs (Fig. 3D), has significant β-galactosidase expression in the ventral head and maxillary palps (Fig. 9F). This construct also expresses in the antennal portion of the disc proper in the third instar, and has dark expression in the adult third antennal segment, as well as light expression in the distal portion of the second antennal segment. Heads from yw flies without reporter constructs did not show β-galactosidase expression, indicating that there is limited endogenous β-galactosidase activity in the adult head, and an engrailed enhancer trap showed a pattern of expression similar to that previously described (data not shown)(Hama et al., 1990), which is distinct from the patterns generated by our dpp and opaconstructs. Thus we believe the expression we observe accurately reports the contribution of peripodial cells to the adult cuticle.

We recovered mutations in the gene opa as dominant enhancers of a dpp mutant phenotype affecting ventral head development. In this work we have established that this interaction arises due to the requirement of opa for the expression of dpp in the peripodial epithelium of the eye/antennal disc. In doing this, we have identified a role for the pair-rule gene, opa, in the development of the eye/antennal disc and subsequent ventral head morphogenesis.

Our work demonstrates that opa is an upstream activator of dpp in the peripodial epithelium, and acts in a cell-autonomous fashion. We do not know whether this role is direct, with Opa acting as a transcription factor for dpp, or through other proteins. This ability to activate dpp appears limited to the peripodial epithelium of the eye/antennal disc, as misexpression of Opa in the disc proper does not induce expression. Furthermore, Opa acts only on a dpp reporter that has expression restricted to the peripodial epithelium of the eye/antennal disc. With the exception of antennal defects, loss-of-function clones of opa produce identical head defects to homozygous dpp^s-hc mutants, and ectopic expression of either Dpp or Opa in the peripodial epithelium produces a similar spectrum of misplaced sensory structures. These data suggest that dpp is the major target of opa in the peripodial epithelium.

Both opa and dpp are involved in embryonic midgut development, where dpp is a negative regulator of opa in the visceral mesoderm (Cimbora and Sakonju,1995). In addition, BMP2 and BMP4 are negative regulators of Zic proteins in zebrafish (Grinblat et al.,1998; Rohr et al.,1999), but the exact mechanism of this regulation is unclear. Thus, Zic family proteins are often seen in regulatory networks with BMP proteins, but there does not seem to be a canonical regulatory relationship. Our data indicates that during eye/antennal disc development opaexerts a positive effect on peripodial dpp.

Both opa and dpp exert their role on ventral head development through expression limited to the peripodial epithelium of the eye/antennal disc. The structures affected in ventral head capsule mutations,such as palps and vibrissae, are reported to arise from the disc proper in the fate map of the eye/antennal disc (Haynie and Bryant, 1986); thus the effect of Opa-Dpp signal transduction could be to cross epithelial layers, from the peripodial epithelium to the disc proper. We have also shown that loss of lateral peripodial Dpp expression results in apoptosis in the underlying disc proper(Stultz et al., 2006), which further suggests a role for peripodial signaling to support disc proper cell viability and morphogenesis. However, when the descendants of peripodial cells are followed by the perdurance of β-galactosidase expression through metamorphosis, significant contributions of lateral peripodial cells are found in areas of the ventral head where we observe defects in dpp^s-hc or opa mutations, suggesting that the ventral adult head is formed from descendants of both disc proper and peripodial cells. Adult head expression has also been seen with the MS1096-Gal4 driver, of which expression in the eye disc is limited to the lateral and medial peripodial epithelium and margin cells(Bessa and Casares, 2005). These data provide further support to the idea that the peripodial epithelium provides more than passive or purely mechanical functions during disc development. The role of the peripodial epithelium in imaginal disc development has begun to receive more attention, and there is evidence that peripodial-specific signaling affects the patterning of the eye(Cho et al., 2000), growth control (Gibson et al., 2002; Gibson and Schubiger, 2000)and the fusion of discs at metamorphosis(Agnes et al., 1999; Zeitlinger and Bohmann, 1999). It now seems likely that in addition to providing such support to cells of the disc proper, peripodial cells contribute directly to the cuticle of the adult head.

In mice and humans, Zic genes are associated with holoprosencephaly, a congenital head defect the extreme manifestation of which is cyclopia. In holoprosencephaly there is variable loss or disruption in the development of the ventral forebrain, and midline facial structures (reviewed by Muenke and Beachy, 2000; Wallis and Muenke, 1999). Holoprosencephaly is a common defect in humans, and genes in the TGF-βand hedgehog pathways are also associated with both the human and mouse condition (Hayhurst and McConnell,2003; Petryk et al.,2004; Zakin and De Robertis,2004). Relevant to our work, a significant number of holoprosencephaly cases result from autosomal dominant inheritance, and often,obligate carriers of these autosomal dominant pedigrees are clinically normal(Ming and Muenke, 2002; Nanni et al., 1999). This incomplete penetrance suggests extreme dose sensitivity and the presence of multiple modifying loci. The ability of our genetic screen to recover multiple dominant enhancers of the dpp ventral head defect, including opa, suggests that this may be a model for the kind of digenic inheritance seen with holoprosencephaly. The hedgehog pathway is known to be crucial to adult head development in Drosophila(Royet and Finkelstein, 1996; Royet and Finkelstein, 1997),and our work adds TGF-β and opa to this process in the fruitfly. It will be of interest to see how many other connections exist between vertebrate and fly head malformations.

We thank Steve DiNardo for sharing information and reagents during this study. We thank Steve Wasserman, Trudi Schupbach, Shigeru Sakonju, Patrick Callaerts, Francesca Pignoni, Jim Kennison, Mark Mortin and the Bloomington Stock Center for fly stocks, and the Developmental Studies Hybridoma Bank for antibodies. We are very grateful to Fernando Casares for sharing his protocol for staining adult heads. We thank Tom Talbot for expert scanning electron microscopy. We thank Judy Kassis, Mark Mortin, Mary Lilly and Brent McCright for insightful comments on the manuscript. This work was funded by the Center for Biologics Research and Review. This paper does not present an official position of the FDA.