Distinct functional specificities are associated with protein isoforms encoded by the Drosophila dorsal-ventral patterning gene pipe

The determination of Drosophila embryonic dorsal-ventral (DV)polarity is mediated by the ventral activation within the embryo of a receptor, Toll, that is uniformly distributed in the embryonic membrane(Anderson and Nüsslein-Volhard,1986; Govind and Steward,1991; Morisato and Anderson,1995; Schüpbach and Wieschaus, 1989). Gastrulation-defective (Gd)(Konrad et al., 1998), Snake(DeLotto and Spierer, 1986) and Easter (Chasan and Anderson,1989) are secreted serine proteases present in the perivitelline space between the eggshell and the embryonic membrane. Gd cleaves and activates Snake, which then cleaves and activates Easter(Dissing et al., 2001; LeMosy et al., 2001). Easter then processes an inactive precursor form of Spätzle, another perivitelline space protein. One of the cleavage products of Spätzle is the activating ligand for Toll (Morisato and Anderson, 1994; Schneider et al., 1994). It has been suggested that a ventral cue, produced during oogenesis, is deposited ventrally within the egg, where it leads to ventral activation of the protease cascade(Anderson, 1998; Sen et al., 1998). Activated Toll transmits the ventralizing signal into the embryonic cytoplasm, leading to the formation of a nuclear gradient of the transcription factor Dorsal(Roth et al., 1989; Rushlow et al., 1989; Steward, 1989) that is responsible for differential transcription, along the DV axis, of zygotic genes responsible for the formation of correct pattern(Stathopoulos and Levine,2002).

The expression of pipe in the ventral cells of the follicle layer is responsible for the localized activity of the dorsal group serine protease cascade (Sen et al., 1998). pipe encodes ten distinct protein isoforms, generated by alternatively spliced mRNAs, all of which exhibit significant sequence homology with the vertebrate enzymes heparan sulfate 2-O-sulfotransferase(HS2ST) and dermatan/chondroitin sulfate 2-O-sulfotransferase (D/CS2ST), two Golgi resident proteins that mediate the sulfation of glycosaminoglycan (GAG)side chains of proteoglycans (Kobayashi et al., 1997; Kobayashi et al.,1999; Sen et al.,1998). HS2ST and D/CS2ST mediate the transfer of a sulfate group to the 2-O position of the uronic acid residues in their respective GAG substrates (Kobayashi et al.,1996; Kobayashi et al.,1999). The similarity between the Pipe isoforms and HS2ST and D/CS2ST initially suggested that Pipe-ST2 might act in follicle cells to sulfate a proteoglycan carrying HS or D/CS side chains. Although subsequent studies have confirmed that Pipe-ST2 acts as a sulfotransferase in ventral follicle cells (Zhu et al.,2005), the nature of the putative carbohydrates modified by Pipe-ST2 remain unknown. Genes involved in GAG synthesis need not be expressed in the follicle cell layer of female flies in order for progeny embryos with normal DV polarity to be produced (Zhu et al., 2005; Zhu et al.,2007). Similarly, the production of a pipe-dependent sulfated molecule in the embryonic salivary gland does not require the expression of genes needed for GAG synthesis. Moreover, purified Pipe-ST2 fails to exhibit sulfotransferase activity on HS, CS or DS in vitro(Xu et al., 2007) (A. Amiri and D.S., unpublished). In contrast to these data suggesting that GAGs do not represent the carbohydrate substrate of Pipe, a recent report indicates that HS exhibits an altered pattern of sulfation in pipe mutant ovaries(Park et al., 2008). Regardless of the specific structure of its target, the current model for Pipe protein action suggests the existence of a glycoprotein-associated carbohydrate molecule that is synthesized in all follicle cells and sulfated by Pipe protein(s) in ventral cells of the epithelium. Following modification,this molecule is secreted, becoming associated with either the oocyte membrane or the vitelline membrane of the eggshell, where it remains until embryogenesis, when it somehow activates the serine protease cascade ventrally in the perivitelline space.

Among the Pipe isoforms, only Pipe-ST2 has been shown to be essential for embryonic DV polarity (Zhu et al.,2005) or to be capable of directing the formation of lateral and ventral pattern elements in the progeny of transgene-expressing females(Sen et al., 1998). Nevertheless, the complex pattern of splicing of the pipe primary transcript has led us to examine the functional basis for the existence of the multiple protein isoforms encoded by the locus. In one potential scenario, the various Pipe protein isoforms could be associated with distinct enzymatic activities, with a subset of the proteins being required in the ovary for the generation of the progeny embryo DV pattern, and other isoforms functioning in the embryonic salivary gland or in other tissues. Alternatively, all isoforms might share a common activity despite their distinct amino acid sequences. To distinguish between these possibilities, we have examined the expression of the individual isoforms and have observed that all are highly expressed in the embryo, whereas several isoforms, in addition to Pipe-ST2, are expressed in the ovary. Among the various isoforms, only the expression of Pipe-ST2 in the ovary is required for the production of embryos with DV polarity. None of the other isoforms expressed transgenically in the follicle cell layer can substitute for Pipe-ST2. These observations suggest that Pipe-ST2 provides a unique enzymatic specificity that is distinct from those associated with the other isoforms.

We have extended the investigations of Pipe isoform functional specificity using a fortuitously identified antibody that detects a sulfated, pipe gene-dependent epitope that is present in the lumen of embryonic salivary glands. By individually expressing Pipe isoforms in the salivary glands of embryos otherwise lacking all Pipe activity, we have observed that three of the Pipe isoforms are capable of restoring the presence of the antigen, whereas Pipe-ST2 cannot. This provides further evidence that proteins encoded by the pipe locus differ in their enzymatic and functional specificities.

Finally, in a separate set of experiments, we have observed that uniform follicle cell expression of Pipe-ST2 in females that otherwise completely lack Pipe protein expression leads to the production of embryos with residual DV polarity that is appropriately oriented with respect to the eggshell. This suggests the existence of a second polarizing input, in addition to the ventral transcription of the pipe gene, that influences the orientation of the embryonic DV axis.

All stocks were maintained employing standard conditions and procedures. The wild-type D. melanogaster stock used was Oregon R. Mutant alleles of pipe have been described as follows: pipe¹ and pipe² (formerly pipe³⁸⁶ and pipe⁶⁶⁴) (Anderson et al., 1985); pipe⁹, pipe³ and pipe^C14 (Zhu et al.,2005); pipe^ZH1(Sergeev et al., 2001); Df(3L)pipe^A13 (Sen et al., 1998). Stocks carrying the following mutations have been described previously: Papss²(Zhu et al., 2005); sll^7E18 (Luders et al., 2003); wind^M88(Konsolaki and Schupbach,1998). The following transgenic P-element insertions used in this study have been described previously: UAS-pipe-ST2(Sen et al., 1998); 55B-Gal4 (Brand and Perrimon,1994); T155-Gal4, CY2-Gal4 and UAS-λtop(Queenan et al., 1997); fkh-Gal4 (Henderson and Andrew, 2000). Laboratory strains of the Drosophilaspecies D. ananassae, D. erecta, D. immigrans, D. pseudoobscura, D. sechellia, D. simulans, D. virilis, D. willistoni and D. yakubawere obtained from the Drosophila Species Stock Center at the University of Arizona in Tucson (current location, University of California at San Diego).

Total RNA was isolated from embryos or dissected ovaries using Trizol reagent (Invitrogen, Carlsbad, CA, USA). cDNA for RT-PCR was prepared using the SuperScript III First-Strand Synthesis System (Invitrogen).

Thirty cycles of PCR amplification (94°C for 30 seconds, 59°C for 30 seconds, 72°C for 60 seconds) were carried out for the detection of reverse-transcribed cDNAs. The primer 5′-GGATGTCTCTGAACGCCGAGCG-3′, which starts at the -2 position within the 5′ UTR of all spliced pipe mRNA isoforms, served as the 5′ oligo in all polymerization reactions. The 3′-end primers specific to each isoform were derived from sequences within the unique fourth exon; for Boxes 1 through 10, respectively, they were (5′ to 3′):GAGCAACTCCATCATCTTCTC, CATCAGCTCCATCAGCGATTG, CATCAACTCGATCAGCGTTTC,GATGAGTCGCGTCATTGACTG, CATAAGCTCAATCAGCTGCAT, AAAGAGCTCCAGCATAGATTC,CATAAACTCCGTAAGCGATTC, GGAGAGTTCCATTAGAGCTTC, GAAAAGTGGGACCATTTGCTC and AAGGAGCTCCATAAACGTCTG. As a control for the quality of the mRNA and reverse transcription, the same cDNA preparations were used to amplify a 797 bp fragment of the windbeutel transcript using the 5′-end primer 5′-AGTCTCTGGTTTTTGCCACG-3′, which corresponds to nucleotides +42 to +61 of the transcript, and 5′-TCACAGTTCCTCCTTTTCCG-3′, which corresponds to nucleotides +819 to +838.

pipe isoforms were cloned following 30 cycles of PCR amplification as above with high-fidelity Pfu DNA polymerase (Stratagene, La Jolla, CA, USA)using cDNA produced with mRNA from wild-type embryos as the amplification template. The 5′-end primer 5′-CCGAGCGCGGCCGCAAAATGAAACTACGCGATGTGG-3′ starts at the -4 position of the 5′ untranslated region shared by all spliced isoforms,with the addition of a NotI restriction site upstream of the initiation codon. For Boxes 1 through 10, respectively(Sergeev et al., 2001), the isoform-specific 3′ primers used for amplification were (5′ to 3′): GGATCCT CTAGACTAATCCCACAGCTTGGGATC,GGATCCTCTAGACTAATAGTCCTCATTGTATTC, GGATCCTCTAGATTAGAAGTGCAATTGCTTATTTAT,GTACTATCTAGACTAAAATTCGTTCTTATGGAC, GGATCCTCTAGACTAGTTGTTGATGGCCATTTC,GTACTATCTAGACTAGTTTAACAGACCATGCAC, GGATCCTCTAGATTAGGAAAAATTCAGTCGTTCCAA,GGATCCTCTAGATTAAGTAGTCCCTGAAACATTTTT, GGATCCTCTAGATTAGTGAAGATTGTTCTCGAG and GTGAAGTCTAGACTTCAATTGCCAATCAGGCCG. The 3′ oligonucleotides contained the stop codon for each particular isoform, as well as an XbaI restriction site downstream of the stop codon. PCR products corresponding to isoform cDNAs were digested with NotI and XbaI and ligated to similarly digested pUAST (Brand and Perrimon, 1993).

For the construction of pHS-CaSpeR-pipe-ST2,pBluescriptSK-pipe-ST2(Sen et al., 1998), which contains a cDNA corresponding to pipe-ST2, was digested at the ApaI site 3′ of the cDNA insert. This site was then made blunt-ended using Klenow enzyme and an XbaI linker was added. The pipe-ST2 cDNA was then excised as an XbaI fragment and ligated to XbaI-digested phs-CaSpeR(Bang and Posakony, 1992). A plasmid carrying the pipe-ST2 cDNA downstream of the hsp70promoter in the sense orientation was identified and designated pHS-CaSpeR-pipe-ST2.

For expression studies, a transgenic insertion of pHS-CaSpeR-pipe-ST2 was introduced into pipe¹/pipe² mutant females. These females were subjected to two rounds of heat shock (1.5 hours at 37°C, 3 hours apart). Embryos were collected at 6-hour intervals. Heat-shocked females produced few eggs, most of which were collapsed and failed to develop until 18 hours after heat shock. Embryos undergoing development, many of which exhibited restoration of lateral and ventral pattern elements, were first observed during the interval 18-24 hours after heat shock. The number of embryos exhibiting phenotypic rescue decreased dramatically in the 30-36 hour interval, with most embryos exhibiting a completely dorsalized phenotype. Based on these results, subsequent analysis of hsp70promoter-directed expression of Pipe-ST2 was carried out using embryos collected 18-30 hours after heat shock of transgene-bearing mutant females.

Whole-mount staining of embryos and egg chambers was carried out according to Macdonald and Struhl (Macdonald and Struhl, 1986). The rabbit polyclonal anti-Twist antiserum(Roth et al., 1991) was used at a dilution of 1:2000, and the affinity-purified rabbit polyclonal anti-Engrailed antiserum (Myat and Andrew,2002) was used at 1:1000. The monoclonal anti-Engrailed antibody(4D9, Developmental Studies Hybridoma Bank) was used at 1:25. The rabbit polyclonal antibody directed against an N-terminal peptide present in all Pipe protein isoforms (Zhu et al.,2005) was pre-absorbed against ovaries from pipe^C14/Df(3L)pipe^A13 females at a dilution of 1:100, and used at a final dilution of 1:2000. Primary antibodies were used in conjunction with biotinylated goat anti-rabbit or anti-mouse secondary antibodies (pre-absorbed against wild-type embryos, final dilution 1:500 or 1:2000, respectively) and visualized with avidin-HRP complex (Vector Laboratories, Burlingame, CA, USA).

Embryos from stocks carrying mutations balanced over either CyO, Krüppel-Gal4 UAS-GFP or TM3, Sb Krüppel-Gal4UAS-GFP were collected on apple juice agar plates overnight, dechorionated in 50% sodium hypochlorite, then transferred to a glass plate and covered with hydrocarbon 27 oil (Sigma). Stage 12-16 embryos were separated into groups containing fluorescent wild-type or non-fluorescent homozygous mutant embryos using a Leica MZFLIII dissecting microscope equipped for detection of GFP. Embryos were stained with Alcian Blue according to Zhu et al.(Zhu et al., 2005), or were processed for whole-mount immunohistochemical staining using the Engrailed^* antibody as outlined above.

To examine the function of the various Pipe protein isoforms in embryos,stocks that carry the pipe³ allele together with each Pipe protein isoform cloned into pUAST were individually crossed to the stock pipe^C14 fkh-Gal4/TM3, Sb Krüppel-Gal4 UAS-GFP, and progeny embryos were collected overnight on apple juice agar plates. As described above, stage 12-16 embryos were separated into fluorescent and non-fluorescent populations. Non-fluorescent embryos expressed an individual Pipe protein isoform in the salivary glands of embryos that otherwise lacked pipe expression. Embryos were stained with Alcian Blue or subjected to whole-mount staining with the Engrailed^* antibody as described above.

Gastrulation patterns were examined under halocarbon 27 oil by conventional light microscopy (Wieschaus and Nüsslein-Volhard, 1986). Larval cuticles were prepared according to van der Meer (van der Meer,1977). DV phenotypes of embryonic cuticles were classified according to Roth et al. (Roth et al.,1991) with the modifications described in the text.

Ten distinct Pipe protein isoforms can be generated through differential splicing of the primary pipe transcript(Sergeev et al., 2001)(http://flybase.org/reports/FBgn0003089.html). All mature mRNAs share the first three exons, which encode the N-terminal 95 amino acids of the protein including the putative membrane-spanning,Golgi-targeting determinant (Sen et al.,2000). The putative sulfotransferase domain of each isoform is encoded by a distinct set of isoform-specific exons. The regions encoding the isoform-specific exons were initially referred to as Box 1 through 10, based on their proximity to the common 5′ exons, with the Box 1 exons proximal and the Box 10 (Pipe-ST2) exons distal to the 5′-end of the gene(Sergeev et al., 2001). Outside of the common region, the ten ORFs share 37-50% amino acid sequence identity with one another, and 21-32% identity with the orthologous regions of the vertebrate enzymes HS2ST and D/CS2ST. An alternate nomenclature has been adopted by FlyBase based on the annotation of the sequenced Drosophila genome, with the predicted pipe mRNA isoforms referred to as pipe-RK, RL, RC, RF, RD, RG, RH, RI, RJ and RA (pipe-ST2), which correspond to the three common exons joined to the exons represented by Boxes 1 through 10, respectively. Corresponding proteins are referred to as Pipe-PK through PA. According to the FlyBase annotation, Box 2 forms a portion of pipe-RL and of a second predicted mRNA, denoted pipe-RE. We have never detected expression of the predicted pipe-RE isoform, suggesting that it does not represent a bona fide mRNA. Similarly, an additional isoform is predicted by FlyBase, pipe-RM, that carries segments of both Box 9 and Box 10 and generates a protein that corresponds to the first 292 amino acid residues of Pipe-PJ. As with pipe-RE, we have never detected cDNA corresponding to pipe-RM, suggesting that it is not expressed in vivo.

The results of RT-PCR using isoform-specific oligonucleotide primers indicated that all predicted Pipe isoforms (except pipe-RE and pipe-RM, see above) are abundantly expressed in wild-type embryos(Fig. 1A), presumably in the salivary glands. In addition to pipe-ST2 (Box 10), we found that the mRNAs corresponding to Boxes 1 through 7 are also expressed in wild-type ovaries, although most of them, with the exception of Box 4, are expressed at low levels (Fig. 1B) as compared with their level of expression in embryos. As a control for the sensitivity of this assay, RT-PCR analysis of ovarian mRNA from mutant females of the RNA-null mutant genotype pipe^C14/Df(3L)pipe^A13 showed a complete absence of pipe mRNA expression(Fig. 1C).

Among 11 available EMS-derived mutant alleles, ten are associated with lesions affecting pipe-ST2-specific exons(Zhu et al., 2005). One exception, pipe³, carries a stop codon at the end of the common third exon and leads to the formation of a truncated protein of 94 amino acids by all mRNAs. No mutations associated with exons specific to any of the other Pipe protein isoforms have been identified. The detection of multiple pipe mRNA isoforms in the ovary led us to revisit the question of whether any of the other isoforms participate in embryonic DV patterning. Overexpression of other isoforms that participate in DV patterning might be expected to lead to the production of ventralized embryos. Similarly,isoforms that are not normally expressed in the ovary, but which have the same enzymatic activity as Pipe-ST2, should to be able to functionally substitute for Pipe-ST2. To test these possibilities we individually expressed cDNAs corresponding to each of the Pipe isoforms throughout the follicle cell layer of pipe³/+ and pipe¹/pipe³females, under the control of the follicle cell-specific Gal4 enhancer-trap elements CY2-Gal4 and T155-Gal4(Queenan et al., 1997),respectively. As negative controls, CY2-Gal4/+; pipe³/+females produced normal embryos (Fig. 2A), whereas embryos from T155-Gal4 pipe¹/pipe³ females were completely dorsalized(Fig. 2B). Whereas expression of the Pipe-ST2 isoform was capable of directing the formation of ventralized embryos by pipe³/+ females(Fig. 2C), none of the other isoforms was capable of doing so (Fig. 2E). Similarly, whereas expression of Pipe-ST2 led to the formation of ventral pattern elements in embryos from pipe¹/pipe³ females(Fig. 2D), the progeny of pipe¹/pipe³ females expressing any of the other isoforms were dorsalized (Fig. 2F). Thus, despite the expression of multiple pipe mRNA isoforms in the wild-type ovary, only Pipe-ST2 is associated with an activity that is necessary and sufficient for the production of ventral and lateral pattern elements along the embryonic DV axis.

Myat and Andrew (Myat and Andrew,2002) described an embryonic salivary gland luminal epitope that is detected by an antiserum (raised in the laboratory of Dr Patrick O'Farrell)against Drosophila Engrailed protein. In addition to the conventional segmental staining pattern exhibited by Engrailed, this polyclonal antiserum detects the salivary gland antigen, which we refer to as En^*(Fig. 3A). This antigen does not correspond to Engrailed protein, as the anti-Engrailed monoclonal antibody 4D9 (Patel et al., 1989) does not stain embryonic salivary glands (Fig. 3B). We tested for the presence of En^* staining in embryos homozygous for several pipe alleles. The En^*antibody did not detect the salivary gland antigen in embryos homozygous for either pipe^C14 or pipe³(Fig. 3D,F), both of which lead to the elimination of all Pipe protein isoforms. By contrast, En^*antigen was detected in embryos homozygous for the alleles pipe^ZH1 and pipe²(Fig. 3H,J). pipe^ZH1 is a deletion mutant that eliminates the Box 9 and Box 10/pipe-ST2-specific exons as well as part of the Box 8-specific region, but leaves intact the exons specific to Box 1 through 7. pipe² is a mutation that interferes with the correct splicing of the mRNA encoding Pipe-ST2. These findings indicate that the presence of the En^* epitope in the embryonic salivary gland is dependent on the expression of the pipe gene and that Pipe isoforms distinct from Pipe-ST2 are capable of supporting its production.

Embryos homozygous for mutations affecting the windbeutel(wind; wbl) gene also lacked En^* staining(Fig. 3L). windencodes a homolog of the vertebrate endoplasmic reticulum protein Erp29(Konsolaki and Schupbach,1998), and we have previously shown that Wind protein is required for the correct subcellular localization and function of Pipe-ST2 and probably of all other Pipe isoforms as well (Sen et al., 2000; Zhu et al.,2005).

To confirm that the En^* epitope represents a sulfated species,we stained for its presence in the salivary glands of embryos homozygous for mutations in Paps synthetase (Papss)(Jullien et al., 1997; Zhu et al., 2005), which generates the universal sulfate donor PAPS, and in slalom(sll), which encodes the PAPS Golgi transporter(Kamiyama et al., 2003; Luders et al., 2003). Embryos homozygous for mutations in either of these genes lacked En^*staining (Fig. 4B,D). Collectively, our results are consistent with the premise that the En^* epitope is sulfated in the Golgi of embryonic salivary gland cells by one or more of the Pipe protein isoforms, but that Pipe-ST2 is dispensable for its formation. The En^* epitope exhibits characteristics similar to those of sulfated material present in the lumen of embryonic salivary glands that we have previously detected using the histochemical stain Alcian Blue (Zhu et al., 2005).

Determining whether individual Pipe protein isoforms are capable of producing the En^* epitope provides another way of distinguishing their functional specificities. To test this, we used the embryonic salivary gland-specific Gal4 enhancer-trap line fkh-Gal4(Henderson and Andrew, 2000)to direct expression of each Pipe isoform individually in the salivary glands of pipe³/pipe^C14 mutant embryos and stained for the presence of En^*. pipe³/pipe^C14 mutant embryos do not produce functional versions of any of the Pipe isoforms and lack En^* antigen (Fig. 5B). However, the three Pipe isoforms Box 2(Fig. 5D), Box 4(Fig. 5F) and Box 6(Fig. 5H), individually supported the production of the En^* epitope. Pipe-ST2, by contrast,was not capable of producing the En^* epitope(Fig. 5L).

We also tested whether the Pipe isoforms were capable of individually restoring Alcian Blue staining to otherwise pipe-null mutant embryos. None of the Pipe protein isoforms was capable of doing so (data not shown). Based on this result, we suspect that Alcian Blue staining requires wild-type levels of pipe expression owing to the low sensitivity of the stain. Presumably, the En^* antibody can detect its sulfated target at concentrations that are too low for detection by Alcian Blue.

Based on these results, the various Pipe protein isoforms can be distinguished into three functional classes. Pipe-ST2 (Box 10) expression in follicle cells can rescue the pipe mutant phenotype and mediate the formation of lateralized/ventralized embryos, but its expression in embryonic salivary glands cannot generate the En^* epitope. Boxes 2, 4 and 6 cannot direct the formation of ventral or lateral pattern elements in embryos,but they can facilitate the formation of the En^* epitope. Boxes 1,3, 5, 7, 8 and 9 can neither ventralize/lateralize embryos nor produce the En^* epitope (Fig. 5C,E,G,I-K). Interestingly, despite the fact that Boxes 2, 4 and 6 share the ability to generate the En^* epitope, they do not represent members of a phylogenetically related subgroup among the Pipe protein isoforms (Fig. 6).

Our suggestion that the En^* antigen represents a specific structural component of the Alcian Blue-staining material is supported by tests for its presence in a number of other species of Drosophila,all of which are predicted to express multiple Pipe protein isoforms based on an examination of genomic sequence. We found that Alcian Blue-staining material was present in all of the additional Drosophila species that we tested (data not shown). By contrast, En^* epitope was present in species of the melanogaster group (D. melanogaster, D. simulans,D. sechellia, D. yakuba, D. erecta, D. ananassae)(Fig. 7A-F), but was not present in the salivary glands of the more evolutionarily distant species D. pseudoobscura, D. willistoni, D. virilis and D. immigrans(Fig. 7G-J). The most parsimonious explanation for these findings is that the Alcian Blue-staining material represents a heterogeneous population of epitopes, generated through the action of multiple Pipe isoforms, which is required for the correct development or function of the embryonic salivary gland. En^*represents a specific structural component of that material. Differences in the structure of the Alcian Blue-staining material generated over the course of evolution of the genus Drosophila have led to the presence of the En^* antigen in species of the melanogaster group and its absence from more distantly related species.

The ventrally restricted expression of pipe in the follicle cell layer, and the ability of ectopic expression of Pipe-ST2 in dorsal follicle cells to invert the polarity of progeny embryos, led to the initial assumption that ventral expression of Pipe-ST2 in the follicular epithelium is the primary, and perhaps only, determinant of embryonic DV polarity(Sen et al., 1998). However,when the Gal4/UAS system was used to direct uniformly high levels of Pipe-ST2 expression throughout the follicular epithelium of pipe¹/pipe² mutant females, the progeny embryos exhibited some residual polarity that could be detected in the pattern of gastrulation movements (D.S., unpublished). If the ventral transcription of Pipe-ST2 in the follicle cell layer were the only polarizing factor influencing the embryonic DV axis, then the progeny of females experiencing uniform follicle cell expression of Pipe-ST2 would be expected to exhibit apolar gastrulation movements. These results suggest the existence of a second polarizing input, the influence of which can be observed when Pipe-ST2 expression in the follicle cell layer is uniform.

The pipe¹ and pipe² mutant lesions specifically affect the Pipe-ST2 isoform(Zhu et al., 2005). The polarity observed when Pipe-ST2 is expressed uniformly in this mutant background could result from residual Pipe-ST2 enzymatic activity associated with the expression of one of these alleles in ventral follicle cells. Alternatively, other Pipe protein isoforms that are expressed ventrally in the pipe¹/pipe² mutant background could partially circumvent the effects of uniform Pipe-ST2 expression. Although the results reported above indicate that only Pipe-ST2 is essential for the production of lateral and ventral embryonic pattern elements, those results do not rule out the possibility that other Pipe isoforms influence the amount, subcellular distribution or activity of Pipe-ST2. If any of those mechanisms do operate,uniform expression of Pipe-ST2 in females otherwise completely lacking pipe mRNA or protein would lead to the formation of apolar lateralized or ventralized embryos. By contrast, the production by these females of embryos with residual polarity would be evidence of the existence of another factor that, in addition to ventral transcription of the pipe gene, contributes to the asymmetry of the DV signal transduction pathway.

To distinguish between the possibilities outlined above, we expressed Pipe-ST2 uniformly along the DV circumference of the follicle cell layers in egg chambers from several types of pipe mutant females and systematically examined the phenotypes of the progeny embryos, focusing on the extent of rescue of the lateral and ventral pattern elements, and on whether the rescued embryos exhibited appropriately polarized gastrulation movements. We expressed Pipe-ST2 either under the control of the heat-inducible hsp70 promoter or under the control of the 55B-Gal4enhancer-trap insertion, which expresses Gal4 in a band of follicle cells surrounding the anterior end of the developing oocyte(Goentoro et al., 2006; Mantrova and Hsu, 1998)(Fig. 8A). Although the anti-Pipe antibody that we have generated is not sufficiently sensitive to detect endogenous Pipe expression in wild-type ovaries(Fig. 8B) or Pipe-ST2 expressed under the control of 55B-Gal4 (data not shown), we were able to confirm that hsp70 promoter-mediated expression of Pipe-ST2 in otherwise pipe-null mutant[pipe^C14/Df(3L)pipe^A13] females leads to high and apparently homogeneous levels of protein expression throughout the follicle cell layer of heat-shocked females(Fig. 8C).

The following pipe mutations were used in testing the ability of Pipe-ST2 to restore DV pattern elements: pipe¹ (V123D, a missense mutation affecting Pipe-ST2), pipe⁹ (R123Stop,affecting Pipe-ST2), pipe^ZH1 (a deletion that eliminates Pipe-ST2, Box 9 and a portion of Box 8), pipe³ (Q95Stop,affecting all Pipe isoforms), pipe^C14 (pipeRNA-null mutation) and Df(3L)pipe^A13 (a large deficiency that uncovers the pipe locus).

Following hsp70- or 55B-Gal4-directed expression of Pipe-ST2, the patterns of gastrulation movements and the cuticles of progeny embryos were examined. The classification of embryonic cuticular phenotypes was carried out according to Roth et al.(Roth et al., 1991) with some modifications. Depending on the severity of the phenotype, dorsalized embryos were classified into one of four groups: completely dorsalized (D0, dorsal epidermis and amnioserosa only), strongly dorsalized (D1, Filzkörper or Filzkörper material present, but lacking ventral denticles), moderately dorsalized (D2, Filzkörper and ventral denticle bands of narrowed width present) and weakly dorsalized (D3, twisted or tail-up phenotype or head defects observed). Ventralized (V) embryos were classified into two groups:weakly ventralized (V1, ventral denticle bands of normal or slightly larger than normal width present, together with a reduction or loss of Filzkörper material) and strongly ventralized (V2, exhibit an expansion of non-cuticular tissue that corresponds to mesoderm, together with patches of ventral denticles compressed to one side of the embryo). Based on the presence of pattern elements derived from different regions of the DV circumference,the D1, D2, D3, V1 and V2 embryos were polar. In addition to D0 embryos, two classes of lateralized embryos exhibited apolar phenotypes in which Dorsal group signaling appeared to have occurred uniformly all around the DV circumference of the embryos. L1 embryos were encircled by bands of small denticles, like those found laterally in wild-type larvae, and L2 embryos were encircled by bands of larger denticles, like those produced ventrally by wild-type larvae.

Expression of Pipe-ST2 under the control of hsp70 or 55B-Gal4 led to the production of embryos that exhibited a range of DV phenotypes in each of the pipe mutant backgrounds tested(Table 1; data not shown). The variation in the extent of phenotypic rescue observed upon expression of Pipe-ST2 in the various pipe mutant backgrounds precludes useful deductions as to how the presence of the ventrally expressed endogenous pipe transcript, or of the other Pipe isoforms, might influence the action of the uniformly expressed Pipe-ST2. However, the ability of Pipe-ST2 expressed uniformly in pipe^C14/Df(3L)pipe^A13 RNA/protein-null egg chambers to direct the formation of normally polarized embryos, as assessed by cuticular pattern elements(Fig. 9G,J), gastrulation movements (Fig. 9H,K), or expression of the ventral marker Twist(Fig. 9I), provides strong evidence that another polarizing input, in addition to the ventral transcription of the pipe gene, contributes to the formation of the embryonic DV axis.

The restriction of Pipe-ST2 to ventral cells of the follicle cell layer in wild-type egg chambers is mediated by the repression of pipetranscription in dorsal follicle cells as a result of activation of the receptor tyrosine kinase Torpedo (Egfr - FlyBase)(Sen et al., 1998). In order to determine whether the second polarizing input is, like pipetranscription, controlled by signaling through the Torpedo receptor, we simultaneously expressed both λtop(Queenan et al., 1997), a constitutively active version of Torpedo, and Pipe-ST2 throughout the follicle cell layer, under the control of the CY2-Gal4 enhancer-trap line. The majority of embryos produced by pipe^ZH1/pipe^C14 females expressing Pipe-ST2 under the control of CY2-Gal4 exhibited a polar ventralized phenotype (V2) in which there was a massive expansion of mesoderm as well as a small region of cuticular material bearing ventral denticles that was often pushed to a position adjacent to the dorsal side of the egg(Fig. 9L). By contrast,expression of λtop under the control of CY2-Gal4 led to the formation of dorsalized (D0) progeny contained within eggshells that were also dorsalized (Fig. 9M). When the CY2-Gal4 driver was used to direct uniform transgenic expression of Pipe-ST2 in follicle cell layers in which Torpedo signaling was also uniformly active, the resultant progeny embryos, also contained within dorsalized eggshells, produced cuticle that was encircled by faint bands of lateral denticles (L1) (Fig. 9N). Thus,the constitutive activation of Torpedo had two effects upon the consequences of uniformly expressed Pipe-ST2 that were not the result of a decrease in pipe transcription. First, constitutive Torpedo activation led to a decrease in the ventral character of the most ventral pattern elements produced, leading to the formation of lateral epidermis rather than mesoderm. Second, it led to the loss of polarity. From these results we can conclude that the second polarizing input that we have observed depends upon the pre-existing polarity imposed by the Gurken/Torpedo signaling cassette and that Torpedo signaling in the dorsal follicle cells of wild-type egg chambers has a dorsalizing effect that is distinct from its repressive effects upon the transcription of the pipe gene.

pipe is a highly conserved gene in the insects. In insects other than the cyclorrhaphan (true) flies, only a single Pipe isoform is normally present. Moreover, the individual Pipe isoforms that are predicted to be expressed in mosquito (Anopheles gambiae), silkworm (Bombyx mori), honeybee (Apis mellifera) and flour beetle (Tribolium casteneum) exhibit greatest sequence similarity to the Pipe-ST2 isoform,among the multiple isoforms expressed by Drosophila melanogaster. By contrast, the genomes of all Drosophila species that have been sequenced (D. ananassae, D. erecta, D. grimshawi, D. mojavensis, D. persimilis, D. pseudoobscura, D. sechellia, D. simulans, D. virilis, D. willistoni and D. yakuba) predict multiple Pipe protein isoforms. Taken together, these observations suggest that an insect ancestral to the holometabolous Diptera, Coleoptera, Hymenoptera and Leptidoptera expressed a single Pipe isoform, the role of which was to regulate the formation of the embryonic DV axis. Pipe protein may also regulate DV axis formation in more basal genera of insects and in other arthropods, as the genome of the water flea (Daphnia pulex), a crustacean, apparently encodes an orthologous protein that is more similar to Pipe-ST2 than to HS2ST or D/C2ST(http:/wfleabase.org).

The existence of multiple protein isoforms in the genomes of the various Drosophila species indicates that several duplications of pipe exons have occurred during the evolution of the members of the dipteran suborder Brachycera, perhaps in the line leading to the family Drosophilidae. Some of the isoforms might have arisen as a means of modulating the effects of the essential Pipe-ST2 isoform during oogenesis, perhaps by modifying the carbohydrate target of Pipe-ST2 to make it a more avid substrate. This pre-modification of the Pipe target would not be essential for Pipe-ST2 action, but would make the developmental process more robust and reproducible from egg chamber to egg chamber, an adaptation to the rapid oogenesis in the Drosophilid species. Alternatively, the multiple Pipe protein isoforms in the Drosophilids might have arisen to provide a function that is totally unrelated to embryonic DV patterning. We have previously shown that the absence of expression of all isoforms of Pipe leads to the formation of structurally aberrant salivary glands and to a decrease in fly viability and growth (Zhu et al., 2005). Although these effects are correlative, it is reasonable to assume that the decrease in viability and growth results from an alteration in feeding behavior that is attributable to a perturbation of the development or function of the salivary gland owing to the absence of all Pipe catalytic activity. As such, the presence of multiple isoforms of the Pipe protein in the Drosophilids might represent an adaptation that is specifically related to the feeding behavior and physiology of those species. Ultimately, an understanding of the basis for the multiple Pipe proteins and their functions will require a precise determination of the distinct catalytic activities of the various proteins, a goal that will first require the identification of the class of molecule(s) on which these enzymes act.

The production of appropriately polarized, partially rescued embryos by females expressing Pipe-ST2 in an otherwise pipe-null background strongly suggests the existence of a second signal that can impart polarity,even in a situation in which Pipe-ST2 activity is not ventrally restricted. Carneiro et al. (Carneiro et al.,2006) have proposed a model in which maternally directed activation of the Decapentaplegic (Dpp) pathway facilitates degradation of the Cactus protein on the dorsal side of embryos in a Toll-independent manner,thus representing a second mechanism for polarizing the DV axis of the embryo. Although this pathway could, in principle, correspond to the second polarizing input, we do not favor this explanation. The findings of Carneiro et al.(Carneiro et al., 2006) are difficult to reconcile with the existence of mutant gain-of-function alleles of easter and Toll that cause uniform activation of Toll around the DV circumference of the embryo, and the consequent production of apolar lateralized or ventralized embryos. The production of such apolar embryos would not be expected were the Dpp pathway acting independently of Toll signaling to establish embryonic DV axis formation. Based on the existence of these unusual apolar alleles of easter and Toll, we favor a model in which the pipe-independent signal that we have observed contributes to the polar activation at a step upstream of Easter-mediated cleavage of Spätzle.

We propose two possible explanations for the second polarizing influence. According to the current model, Pipe protein action in ventral follicle cells leads to the sulfation of a secreted glycoprotein that is deposited ventrally in the egg, where it mediates the activation of the dorsal group serine protease cascade. If any of the genes that encode proteins involved in the synthesis of the direct carbohydrate target of Pipe or its predicted carrier protein were expressed preferentially in ventral follicle cells, then the ventrally enriched production of the Pipe substrate could account for the residual polarity of progeny embryos under conditions of uniform Pipe-ST2 activity. The ability of Pipe-ST2 to invert the polarity of progeny embryos when expressed ectopically in dorsal follicle cells would result from the ability of Pipe-mediated sulfation to overcome the polarizing effects of ventral enrichment of the Pipe target.

An alternative basis for residual polarity in the presence of uniform Pipe-ST2 activity might be the existence of an inhibitor of Gd or Snake that is enriched dorsally within the perivitelline space, or of a ventrally enriched, positively acting regulator of the processing or activity of Gd,Snake or Easter, distinct from Pipe-ST2. Although a serpin inhibitor of Easter, Spn27A, has been shown to operate in the perivitelline space, this molecule is uniformly distributed(Hashimoto et al., 2003; Ligoxygakis et al., 2003). Moreover, available data suggest that Spn27A acts to restrict Easter activity to the ventral side of the embryo following its activation, and does not have any intrinsic ability to orient the activity of Easter.

The geometry of egg formation might also introduce an asymmetry that could influence the effects of Pipe-ST2 along the embryonic DV axis. Over the course of oogenesis, the egg chamber and egg assume characteristic shapes that are polarized along the DV axis. The morphogenetic processes that govern egg formation and the physical properties of the mature egg might influence the distribution of the modified target of Pipe or the spatial constraints of serine protease activation and in this way impart a weakly polarizing influence that is detectable when pipe expression is uniform in the developing egg chamber.

The extent to which this second polarizing input is required for the formation of embryonic DV polarity in wild-type embryos is difficult to assess. The second input cannot impart DV polarity to embryos in the absence of Pipe-ST2 activity. This suggests that the role of the second input might be to buffer the effects of rare perturbations in the normal ventral pattern of pipe expression. The power of this second input to influence the DV axis is demonstrated by our generation of viable stocks in which Pipe-ST2 is expressed uniformly along the DV axis under the control of 55B-Gal4in otherwise pipe-null mutant backgrounds that lack endogenous ventral expression of Pipe-ST2.

In conclusion, our investigations of the functions of the various Pipe protein isoforms has demonstrated the existence of additional layers of complexity in the mechanism generating embryonic DV polarity, including the presence of a previously unappreciated mechanism that reinforces the effects of the ventral transcription of the pipe gene. The elucidation of the structure of the substrate of Pipe-ST2 and its function in regulating the dorsal group serine protease cascade is likely to provide insights into this alternate polarizing input and into the specific roles of the sulfotransferases encoded by the pipe locus.