Complex Organization of Promoter and Enhancer Elements Regulate the Tissue- and Developmental Stage-Specific Expression of the Drosophila melanogaster Gld Gene

Corresponding author: Douglas R. Cavener, Department of Biology, 208 Mueller Laboratory, The Pennsylvania State University, University Park, PA 16802., drc9{at}psu.edu (E-mail)

Communicating editor: W. F. EANES

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The Drosophila melanogaster Gld gene has multiple and diverse developmental and physiological functions. We report herein that interactions among proximal promoter elements and a cluster of intronically located enhancers and silencers specify the complex regulation of Gld that underlies its diverse functions. Gld expression in nonreproductive tissues is largely determined by proximal promoter elements with the exception of the embryonic labium where Gld is activated by an enhancer within the first intron. A nuclear protein, GPAL, has been identified that binds the Gpal elements in the proximal promoter region. Regulation of Gld in the reproductive organs is particularly complex, involving interactions among the Gpal proximal promoter elements, a unique TATA box, three distinct enhancer types, and one or more silencer elements. The three somatic reproductive organ enhancers each activate expression in male and female pairs of reproductive organs. One of these pairs, the male ejaculatory duct and female oviduct, are known to be developmentally homologous. We report evidence that the other two pairs of organs are developmentally homologous as well. A comprehensive model to explain the full developmental regulation of Gld and its evolution is presented.

GLUCOSE dehydrogenase (GLD), a rare flavo-selenoenzyme that generates reactive oxygen species, participates in an eclectic set of seemingly unrelated functions in Drosophila (CAVENER and MACINTYRE 1983 ; CAVENER 1992 ; COX-FOSTER and STEHR 1994 ). During Drosophila melanogaster development GLD is expressed in specific sensory organs including the antenno-maxillary complex as well as several ectodermally derived tissues including the somatic reproductive organs (COX-FOSTER et al. 1990 ). GLD is critically required to modify the puparial operculum so that the newly developed adult fly can escape from its puparium case (CAVENER and MACINTYRE 1983 ). At the adult stage, GLD participates in two unrelated functions: female fertility (our unpublished data) and immunity (COX-FOSTER and STEHR 1994 ; LOVALLO and COX-FOSTER 1999 ). A single Gld gene (CAVENER et al. 1986 ) is responsible for all of these diverse expression patterns and functions; consequently the transcriptional regulation of Gld is exceedingly complex. Various Drosophila species and other insects appear to share these patterns of expression during preadult development but display different patterns in the adult somatic reproductive organs (KRASNEY et al. 1990 ; SCHIFF et al. 1992 ; ROSS et al. 1994 ).

Gld transcription is regulated in part by two major regulatory pathways: sex determination (FENG et al. 1991 ) and ecdysone, an insect steroid hormone (MURTHA and CAVENER 1989 ). Temporal expression of Gld mRNA during embryogenesis, larval development, and metamorphosis is controlled by ecdysone via the Gld proximal promoter region. The sex-determination pathway controls the tissue-specific expression of Gld mRNA in the somatic reproductive organs of the adult by repressing Gld expression in specific reproductive organs. The expression of GLD is particularly intriguing in the reproductive organs. In D. melanogaster, as well as most other Drosophila species, Gld is expressed in a specific subset of the male and female somatic reproductive organs during their development including the male ejaculatory duct and ejaculatory bulb and the female oviduct, seminal receptacle, spermathecae, parovaria, and vaginal plate (SCHIFF et al. 1992 ; ROSS et al. 1994 ). Shortly after eclosion Gld is expressed in a more limited subset of the mature adult somatic reproductive organs. In contrast to the conservation of Gld expression patterns during development, the patterns of expression at the adult stage vary dramatically among the 54 species characterized in the genus Drosophila (SCHIFF et al. 1992 ). Among these species, five distinct male expression pattern types and six distinct female expression pattern types are observed in the adult reproductive tract. Although GLD is abundantly expressed in the male ejaculatory duct of D. melanogaster and a few other closely related species, it is not expressed in any reproductive organ in adult males in 20% of the other species. Expression of GLD in the adult female spermathecae is highly conserved among species but expression in other female organs (e.g., seminal receptacle, parovaria, and oviduct) is highly species specific.

Previously we showed that a somatic reproductive organ enhancer complex (SREC) activates the Gld promoter fused to a lacZ reporter gene in the developing and mature reproductive tracts similar to the pattern observed for the normal Gld gene in D. melanogaster (KEPLINGER et al. 1996 ). However, in combination with a heterologous promoter, SREC activates expression in a much smaller subset of the reproductive organs, suggesting that the Gld promoter in combination with SREC plays an important role in reproductive tract expression. Herein, we show that dispersed proximal promoter elements, denoted Gpal, interact synergistically with the SREC to activate transcription in several somatic reproductive organs. A nuclear protein, denoted GPAL, was identified that binds specifically to the Gpal elements and is present in tissues known to require the Gpal elements for Gld expression. Further mapping of the 3.3-kb region containing SREC revealed three distinct reproductive organ enhancers and an embryonic labial enhancer unrelated to reproductive organ expression. The three reproductive organ enhancers activate expression in three different pairs of male and female organs that are developmentally homologous. In addition we present evidence that the male ejaculatory bulb medulla is homologous to the female spermathecae and the ejaculatory bulb cortex is homologous to the female uterus.

Although some details remain to be elucidated, we show herein a comprehensive explanation for the complex tissue-specific expression of Gld mRNA throughout development. In addition these data provide an evolutionary model for the sequential acquisition of regulatory elements and their subsequent diversification with respect to controlling the expression of Gld in the somatic reproductive organs among the species in the genus Drosophila.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Plasmid construction:
A P-element vector containing unique EcoRI, BamHI, and KpnI sites between the SREC and a lacZ reporter gene was constructed to facilitate insertion of various Gld promoter fragments examined herein. The 3.3-kb XbaI-EcoRI SREC fragment from Gld intron I was blunt ended with T4 DNA polymerase (Promega, Madison, WI) and inserted in the reverse orientation into the blunt ended EcoRI site of pCaSpeR-AUG-ßGal (THUMMEL et al. 1988 ), creating SREC/LacZ.

The -104 to +84 and -39 to +84 Gld promoter fragments were PCR amplified from SREC/-425/LacZ template. The PCR products were digested with BamHI and KpnI and inserted in the unique BamHI/KpnI sites of SREC/LacZ, creating SREC/-104/lacZ and SREC/-39/lacZ. The entire PCR products were sequenced in the final constructs for SREC/-104/LacZ, SREC/-39/LacZ, and all transgenes described herein.

The -425dGpal^- Gld promoter was PCR amplified from genomic DNA isolated from pWZ170 transgenic flies, which carry the -425dGpal^-/LacZ transgene. The -425dGpal^- PCR product was digested with EcoRI and KpnI and inserted into the unique EcoRI/KpnI sites in SREC/LacZ, creating SREC/-425dGpal^-/LacZ.

To construct the SREC/dGpal/hsp70/LacZ transgene, the 3.3-kb XbaI-EcoRI SREC fragment from Gld intron I was isolated as an EcoRI fragment subsequently blunt ended with T4 DNA polymerase. It was then inserted in the reverse orientation into the unique blunt-ended KpnI site of dGpal/hsp70/LacZ (GUNARATNE et al. 1994 ) to create SREC/dGpal/hsp70/LacZ. To construct the SREC/-425mTATA/LacZ transgene, the 3.3-kb XbaI/EcoRI SREC fragment from Gld intron I was isolated as an EcoRI fragment from pEG25 and inserted in the reverse orientation into -425mTATA/LacZ (pW^AG183). -425m TATA/LacZ contains the TATA mutant PstI/BclI -425 Gld promoter fragment (a single T A transversion at the second position of the TATA element).

SREC 2/-425/lacZ, SREC 2.1/-425/lacZ, SREC 2.2/-425/lacZ, and SREC 2.3/-425/lacZ were constructed by PCR amplification of the appropriate intronic regions of Gld and inserted in the -425/lacZ P-element vector.

Site-directed mutagenesis:
Site-directed mutagenesis was performed using the mega-primer PCR mutagenesis method (PICARD et al. 1994 ). SREC/-425dGpal^-/LacZ, which already contained four single nucleotide substitutions in the dGpal element, was used as a template. The LacZrev oligonucleotide (AAAGGGGGATGTGCTGCAAG), corresponding to LacZ reporter gene sequences, was used as the downstream (3') primer. The PD6 oligonucleotide (TTGAGTTCGAtCgaTTCG TCAGTT) served as the mutagenic primer to incorporate three single nucleotide substitutions into Gpal-1 (lowercase letters). The PD5 oligonucleotide (TAGGATCCGACCAAGTTTCACAGAGCGCAtCgaTGCGGCCA) served as the upstream (5') primer. PD5 also incorporated three single-nucleotide substitutions into Gpal-4 and adds a convenient BamHI restriction site (underlined). The resulting PCR product, -104 Gpal^- was digested with BamHI and KpnI and inserted into the unique BamHI/KpnI sites of SREC/LacZ, creating SREC/-104Gpal^-/LacZ. The -104 Gpal^- PCR product fragment was subsequently sequenced to confirm mutation of the Gpal elements 1–4 and to confirm the integrity of the flanking Gld sequences.

Linker-scanning mutagenesis:
Linker-scanning mutagenesis was performed using the PCR method developed by GUSTIN and BURK 1993 . Oligonucleotide primers used were as follows: For PCR product 1, Casper Gld.1 (CTAGAATTCCTGCAGCCGTTCG) and either LS1A (taggatccgacctcaAATCGGGAGCTTTTGAGTG), LS2A (taggatccgacctcaCTGCGCTCTCTGAAAC), or LS3A (taggatccgacctcaACAGCTTAAAGCTGGCCGCAAA). For PCR product 2, Droadh.1 (GGTCAAAGTAAACGACATG) and either LS1B (taggatccgacctcaGCAGCTTTGCGGCCAG), LS2B (taggatccgacctcaTGTCTTTCGTTGAGTTCGAG), or LS3B (taggatccgacctcaTTCGTCAGTTTAAAAAGACT). For each of the LS oligonucleotides, the BamHI sequence is underlined and Gld homologous sequence is in uppercase. The PCR products were digested with EcoRI and KpnI and inserted into the unique EcoRI/KpnI sites of SREC/LacZ, creating SREC/LS1/LacZ, SREC/LS2/LacZ, and SREC/LS3/LacZ. Each PCR product was sequenced to confirm the presence of the LS mutations and the integrity of the flanking Gld promoter sequences.

P-element-mediated transformation and isolation of transgenic strains:
P-element-mediated gene transformation of Drosophila was performed essentially as described by RUBIN and SPRADLING 1982 . Recombinant plasmids in the pW5-hsp70-ßgal or pCaSpeR-AUG-ßgal P-element transformation vectors were injected into preblastula stage embryos of the D. melanogaster 2-3(99B) strain (ROBERTSON et al. 1988 ). Crosses were conducted to localize each transgene to a particular chromosome, to replace the genetic background with a ßgal null mutant, to remove the endogenous transposase activity, and to make the transgenes homozygous.

ß-Galactosidase histochemical assays:
Dissected Drosophila tissues were assayed for ß-galactosidase (ßGAL) activity using a modified histochemical stain developed by GLASER et al. 1986 . Three to five individuals for each transgenic strain were assayed at room temperature in the dark for 15 min to overnight as necessary. In addition, each strain was analyzed at least twice in separate experiments. ßGAL expression levels for each strain were visually quantitated on a + to +++++ scale, with + representing just detectable expression and +++++ representing very high level expression.

Detection of sequence specific binding proteins:
To characterize nuclear proteins that bind to Gpal four complementary oligonucleotide pairs were synthesized. The two oligonucleotide pairs, dPalGld and mdPalGld, correspond to the sequence of the Gld promoter from -82 to -46. mdPalGld has four mutations in the dPal, which destroys the palindromic nature of this sequence. The oligonucleotide pair dPalN contains the Gpal sequence flanked by 19 nucleotides unrelated to the Gld promoter. The oligonucleotide pair mdPalN is identical to dPalN except for four mutations in the dGpal sequence.

Protein extracts were isolated from pelleted nuclei of embryos, larvae, pupae, or adults after methods of QUINE 1991 . To extract GPAL-1 from dissected tissues, the tissues of interest were dissected from the appropriate individuals in phosphate-buffered saline solution, pH 7, and nuclear proteins were isolated using the same methods for whole animal homogenates.

Nuclear extracts were added to the assay buffer to give 30 µl total volume and allowed to incubate at room temperature for 30 min. The assay buffer was composed of 20 mM Hepes (pH 8), 100 mM KCl, 1 mM EDTA, 1 mM EGTA, 1 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride, 20% glycerol, 1 mg/ml BSA, 1 µg poly(dI)poly(dC)/1 µg GPAL-1 extract, 1 ng gel-purified ³²P-end-labeled oligonucleotide probe per reaction. The binding reactions were electrophoresed in a 5% nondenaturing Tris-glycine polyacrylamide gel. Best results were obtained with 0.5–1.5 µg GPAL-1 extract. Fold differences among lanes were estimated by scanning laser densitometry (LKB Ultroscan XL).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Interaction between the Gld promoter and somatic reproductive organ enhancer:
We previously defined an SREC within a 3.3-kb XbaI-EcoRI enhancer fragment from Drosophila Gld intron I (KEPLINGER et al. 1996 ). SREC can activate the -425- to +84-nucleotide (nt) Gld promoter in all the somatic reproductive organs where Gld is normally expressed, both in the developing reproductive tract (during metamorphosis) and in the mature adult reproductive tract (Table 1). In addition, SREC activates expression of the -425/+84 Gld promoter in the adult oviduct and seminal receptacle where Gld is not normally expressed at the adult stage (but where Gld is expressed in preadult development). SREC can also activate the heterologous Drosophila hsp70 promoter, but only in the adult male ejaculatory duct and to a lesser extent in the developing male ejaculatory duct and female spermathecae. These observations suggested that Gld expression in most of the reproductive tract requires the SREC as well as proximal promoter elements.

Delimiting the Gld promoter region necessary for reproductive tract expression:
To identify the promoter region required in combination with the SREC for reproductive tract expression, we constructed a 5' deletion series of the Gld promoter with a common 3' terminus located at +84 nt relative to the transcription start site (Fig 1). The deletion series was designed to separate putative regulatory elements previously identified by us (KRASNEY et al. 1990 ). Truncation of the region from -425 to -104 nt removes the TTAGA elements, a dispersed repetitive family of elements. The TTAGA elements, when tested independent of the Gld, are capable of activating a heterologous hsp70 promoter at very low levels in the reproductive tract (QUINE et al. 1993 ), but their requirement for Gld promoter activity had not been tested. Truncation of the region from -104 to -39 nt removes a highly conserved 13-nt element (CAGCTTTAAGCTG) at -68 nt exhibiting dyad symmetry (Fig 1). This element is sufficient to activate expression of a heterologous hsp70 promoter in the anterior spiracular glands of third instar larvae, in the wings and halteres of developing adults during metamorphosis, and in the adult male ejaculatory bulb (GUNARATNE et al. 1994 ). This element was previously named Gpal but will be denoted herein as dGpal (dyad Gpal) and Gpal will denote the AGCTTT core element. Two Gpal elements compose dGpal whereas other Gpal elements are dispersed as single elements in the proximal promoter region of Gld.

View larger version (39K): In this window In a new window Download PPT slide   Figure 1. Transgenic constructs to map regulatory elements of the proximal promoter region of D. melanogaster Gld. SREC, somatic reproductive organ enhancer complex contained with a 3.3-kb region of intron I of Gld; +1 indicates the transcriptional start site of the Gld or hsp70 genes. Arrowheads below the sequence indicate the presence of wild-type Gpal element (AGCTTT) in either the sense or antisense strand. Broken arrowheads indicate the presence of a wild-type degenerate Gpal element composed of AGYYYY (Y = C or T). Asterisks indicate that site directed mutations were introduced with the mutant sequence shown below the asterisk. All DNA sequences shown are from the Gld proximal promoter region with the exception of italicized sequences from the hsp70 promoter. All constructs were cloned in the pCaSpeR-AUG-ßGal P-element transformation vector. Three or more independent transgenic Drosophila strains were isolated for each of these constructs.

Each 5' deletion promoter fragment was combined with the SREC and placed upstream of the LacZ report-er gene in the P-element vector pCaSpeR-AUG-ßgal (THUMMEL et al. 1988 ), creating SREC/-104/LacZ and SREC/-39/LacZ. Expression patterns were analyzed by histochemical analysis of ßGAL expression in flies carrying the transgenes in a genetic background deficient for the endogenous ß-galactosidase gene. Three or more transgenic strains were analyzed for each transgene to rule out chromosomal background effects of the random insertion site of the transgenes.

Removal of the -425- to -104-nt region in SREC/-104/LacZ results in the loss of expression in the adult female seminal receptacle but otherwise its pattern was identical to the SREC/-425/LacZ transgene (Table 1). Deletion of the -104 to -39 region (SREC-39/LacZ) resulted in a dramatic and broad attenuation of reproductive tract expression of the lacZ reporter gene. Nonetheless, this transgene is expressed at moderate levels in the adult male ejaculatory duct. Much lower levels are observed in the developing and adult spermathecae and developing oviduct. Expression of SREC/-39/LacZ in the reproductive tract is similar to that of the SREC/hsp70/LacZ transgene (Table 2) where only the adult ejaculatory duct shows substantial expression (Fig 2B). In summary, the proximal promoter region from -104 to -39 immediately upstream of the TATA box is crucial for activation of the Gld promoter in the majority of tissues in which Gld is normally expressed. However, for some tissues this region is neither necessary nor sufficient for Gld expression.

View larger version (122K): In this window In a new window Download PPT slide   Figure 2. Expression of the lacZ reporter controlled by Gld transcriptional regulatory elements. All transgenic lines bearing lacZ reporter constructs are homozygous for a null mutation of the endogenous Drosophila ß-galactosidase gene and thus the X-GAL staining patterns (blue) are solely due to the lacZ reporter. The patterns depicted in this figure and summarized in Table 1 Table 2 Table 3 were shown in three or more independent transgenic lines. (a) SREC/-425/lacZ transgenic exhibits reporter expression in the embryonic antenno-maxillary complex (upper arrow) and labium (lower arrow). The pattern is identical to the pattern of Gld mRNA and protein expression. (b) SREC/-39/lacZ transgenic exhibits reporter expression in the adult male ejaculatory duct (upper arrow) but not in the ejaculatory bulb. This pattern is identical to Gld mRNA and protein. (c) SREC 2.1/-425/lacZ transgene exhibits reporter expression in the ejaculatory bulb cortex (lower arrow) but not the ejaculatory duct. (d) -425/lacZ transgene exhibits expression in the embryonic antenno-maxillary complex (arrow) but lacks expression in the labium. The LBE/-39/lacZ transgene exhibits expression in the labium but not antenno-maxillary complex (not shown). (e) SREC/-425/lacZ transgene exhibits expression in the spermathecal caps (upper arrow) and seminal receptacle (lower arrow). (f) SREC 2.1/-425/lacZ transgene exhibits reporter expression in the adult female uterus.

The unique Gld TATA and Gpal elements are required for expression in a subset of reproductive organs:
The TATA and dGpal elements are the most highly conserved elements within several hundred nucleotides upstream of the Gld transcription start site (ROSS et al. 1994 ). The Gld TATA element is positioned within the 28- to 31-nt range from the transcription start site as typically seen for Drosophila promoters (CHERBAS et al. 1986 ); however, its sequence (TTTAAAAA) is atypical; T at the second position is rare in eukaryotic TATA boxes (BREATHNACH and CHAMBON 1981 ). Nonetheless, the unusual Gld TATA sequence is highly conserved among the 11 Drosophila species examined (KRASNEY et al. 1990 ; ROSS et al. 1994 ). We speculated that this unusual TATA sequence may provide tissue specificity.

To examine the potential regulatory function of Gld unusual TATA box, we chose to change the T at the second position to an A, thereby creating a prototypical TATA consensus sequence (Fig 1). This mutation (mTATA) was examined in the fusion gene SREC/-425mTATA/LacZ, otherwise identical to SREC/-425/LacZ. Analysis of the resultant transgenic strains showed that the TATA element mutation abrogates ßGAL expression in the adult female seminal receptacle and developing female uterus (Table 2). Surprisingly, the mutant TATA transgene activates expression in the male seminal vesicle, an organ for which Gld expression has never been observed in any Drosophila species or in any other of the numerous Gld transgenic strains we have studied. Therefore we speculate that a major function of the atypical Gld TATA is to repress Gld in the seminal vesicle.

We performed two experiments to analyze the highly conserved dGpal element. Four nucleotide substitutions were introduced into dGpal creating SREC -425dGpal^-/LacZ (Fig 1). The resultant transgenic strain exhibited a lack of expression in the adult female seminal receptacle as compared to the nonmutated SREC/-425/LacZ control (Table 2); however, expression of ßGAL was observed in all other reproductive tissues for which the control transgene is expressed. To test the sufficiency of the dGpal element to interact with SREC in isolation of other Gld promoter elements, it was inserted upstream of the hsp70 promoter in SREC/hsp70/LacZ, creating SREC/dGpal/hsp70/LacZ (Fig 1). Analysis of the resultant transgenic strains as compared with SREC/hsp70/LacZ showed that the addition of dGpal resulted in activation of the reporter gene in only the adult male ejaculatory bulb (Table 2). Previously we showed that dGpal is capable of activating the heterologous hsp70 promoter in the male ejaculatory bulb (GUNARATNE et al. 1994 ). However, this expression did not require SREC. The full Gld promoter (-104 to +84) containing dGpal does not support expression in the adult ejaculatory bulb. Taken together these results indicate that other elements in the Gld promoter act to repress activation of Gld expression in the adult ejaculatory bulb. In addition, elements within the SREC enhancer region also repress Gld expression in this organ at the adult stage (see below).

Multiple segments of the promoter region are required for reproductive tract expression:
A comparison of the 5' deletion analysis and the mutagenesis of the dGpal and TATA elements suggested that other elements must be present in the 65-bp region between -104 and -39 nt that are required for expression in most of the reproductive organs that normally express GLD during metamorphosis and in mature adults. A series of linker-scanning mutations in the -104- to -39-nt region were constructed to identify such elements. Each mutation replaces a 20-nt block of Gld sequence with random sequence (Fig 1). The LS1 mutation extends from -106 to -87 nt, LS2 from -82 to -63 nt and LS3 from -59 to -40 nt.

In the adult reproductive tract, the LS1 mutation abolished ßGAL expression in the seminal receptacle (Table 2). Expression levels in other adult reproductive tract tissues were unchanged relative to the control SREC/-425/LacZ. The LS2 mutation abolished expression in the spermathecae and vaginal plate and resulted in highly variable expression in the seminal receptacle. The LS3 mutation abolished expression in the seminal receptacle and resulted in highly variable expression in the spermathecae and vaginal plate. ßGAL expression in the adult ejaculatory duct and oviduct were unaffected for all three linker-scanning mutations. The adult expression patterns of the linker-scanning mutations do not reveal distinct Gld promoter regions required for expression in distinct subsets of adult reproductive tract tissues, but rather multiple regions that collectively contribute to expression in several tissues. For example, linker-scanning regions 1, 2, and 3 are all required for adult seminal receptacle expression. Linker-scanning regions 2 and 3 are also required for adult spermathecae and vaginal plate expression. No single sequence within the -104- to -39-nt region is necessary for adult oviduct expression, suggesting the presence of redundant elements.

Dispersed Gpal elements interact synergistically with the SREC and among themselves to produce the complex Gld reproductive tract expression pattern:
We noted that a dispersed sequence element (AGCTTT) in the Gld proximal promoter region was identical to the core repeat element of dGpal and denoted these AGCTTT sequences as Gpal elements. We speculated that these elements collectively interact with SREC to activate the Gld promoter in the reproductive tract. Four Gpal elements, including the two within dGpal, are present in the -104- to -39-nt region (Fig 1). Two additional Gpal elements are located in the -425- to -104-nt region. To test the role of the Gpal elements in reproductive tract expression, we employed a combination of deletion and site directed mutagenesis to sequentially remove all six Gpal elements in the -425-nt Gld promoter (Fig 1). The two upstream Gpal elements were removed in the previously described deletion that removes the -425- to -104-nt region. We performed site directed mutagenesis on the four remaining Gpal elements in the -104-nt Gld promoter, creating SREC/-104 Gpal^-/LacZ.

In the adult reproductive tract, the SREC/-104 Gpal^-/LacZ transgene expresses ßGAL only in the male ejaculatory duct, consistent with previous findings that ejaculatory duct expression is not dependent upon Gld promoter elements (KEPLINGER et al. 1996 ). However, as was seen for the SREC/-39/lacZ transgene, obliteration of the Gpal elements reduces the expression in the male ejaculatory duct indicating that the Gpal elements are necessary for the normally high level of Gld expression in this organ. Expression of the SREC/-104 Gpal^-/LacZ transgene is undetectable in other adult reproductive tract tissues (female spermathecae, oviduct, and vaginal plate), in contrast to SREC/-104/LacZ, with four intact Gpal elements, or SREC/-425/LacZ, with six intact Gpal elements. In the developing reproductive tract, ßGAL is expressed in the male ejaculatory duct (Table 3). However, in comparison to SREC/-104/LacZ and SREC/-425/LacZ, expression is greatly diminished in many developing reproductive tract tissues (female parovaria, seminal receptacle, uterus, and oviduct) and is both diminished and highly variable in others (male ejaculatory bulb and female spermathecae and vaginal plate). Thus, the repetitive Gpal elements are required for synergistic interactions between the SREC and the Gld promoter and for activation in all adult reproductive tract tissues except the ejaculatory duct. Gpal elements also play a key role in expression in the developing reproductive tract, although they are not strictly required in all tissues.

Two additional sequences (AGTTTC at -99 nt and AGTTCG at -50 nt) present in the -104- to -39-nt region are very similar to the Gpal elements and may account for the residual expression in the reproductive tract seen in the SREC/-104 Gpal^-/LacZ transgene. To examine whether these two Gpal-like elements were responsible for the residual expression, these sites were mutated in the SREC/-104 Gpal^-/lacZ plasmid yielding SREC/-104 allGpal^-/lacZ. These mutations had only a small additive effect by eliminating expression in the preadult ejaculatory bulb and by increasing line-to-line variability among most of the other remaining tissues in which the SREC/-104 Gpal^-/lacZ transgene was expressed.

Collectively, the three linker-scanning mutations eliminate the Gpal sequences in the -104- to -39-nt region. LS1 is mutated for the Gpal-like elements at -99 nt; LS2 removes Gpal elements 2, 3, and 4. LS3 removes Gpal-1 and the Gpal-like element at -50 nt (Fig 1). Assuming that most, if not all, of the effects of the linker scanning mutations are due to altering the Gpal elements, it is clear that no single Gpal element is required for all tissues. Most tissues are relatively insensitive to loss of Gpal elements if the other Gpal elements in the -104- to -39-nt region are intact. However some tissues (e.g., adult female seminal receptacle and vaginal plate) are particularly sensitive to mutations of one or more Gpal elements. LS2, which bears mutations in three Gpal elements including the highly conserved dGpal elements at -72 nt, has the single most potent effect of eliminating expression in the developing uterus and adult female spermathecae as well as reducing the expression in vaginal plate (Table 2). Expression in the developing uterus requires the unusual TATA box sequence as well. However, elimination of the dyad Gpal elements alone does not abrogate expression in the developing uterus.

Proximal promoter regulation of Gld expression in nonreproductive organs:
Previously we reported that the dGpal dyad repeat could drive expression of the heterologous hsp70 promoter in the anterior spiracular gland of larvae as well as the adult male ejaculatory bulb, indicating that the Gpal elements may have regulatory functions in both reproductive and nonreproductive tissues. To further test this hypothesis we examined the expression of the Gld promoter deletion transgenes described above. As with GLD, the -425/lacZ and SREC/-425/lacZ transgenes both show very high levels of reporter gene expression in the larval anterior spiracular glands (ASG) and pupal developing wings, halteres, and hypoderm (Table 1). Reporter gene expression is dramatically reduced in the SREC/-39/lacZ transgene in the ASG, wings, and halteres but is largely unaffected in the hypoderm. These data are consistent with the hypothesis that the Gpal elements are necessary for expression in the ASG, wings, and halteres but that only the -39 to +84 region of the Gld promoter is required for hypoderm expression. Hypoderm expression is still maintained in the TATA mutant transgene (Table 2). We speculate that the region between +1 and +20 contains a hypoderm activation element inasmuch as it is the most highly conserved region flanking the transcription start of the Gld gene among nine Drosophila and one Scaptodrosophila species analyzed to date (KRASNEY et al. 1990 ; ROSS et al. 1994 ; MERRITT 1998 ).

A nuclear protein binds selectively to the dyad Gpal element:
Nuclear extracts from Drosophila embryos or specific tissues were isolated and probed with a radiolabeled double-stranded 30-bp oligonucleotide (dPalGld) corresponding to the dyad Gpal (dGpal) and flanking sequences. All reactions were done in the excess of the unrelated poly dIdC competitor and other specific and nonspecific competitors as indicated. Protein-oligonucleotide complexes were electrophoretically fractionated to observe potential mobility shifts in the probe as a consequence of binding to specific proteins. Nuclear proteins isolated from the 100–300 mM NaCl nuclear fractionation revealed the presence of a prominent protein that binds the 30-bp dPalGld (e.g., Fig 3A, lane 1 and 3b, lane 1). We denoted this protein as GPAL. To determine the specificity of binding GPAL we performed a number of competition binding studies where the ability of mutant and nonspecific DNA fragments were tested for their ability to inhibit the binding of dPalGld to GPAL. A mutated form of dPalGld, denoted mdPalGld, contains four substitutions in the 13-bp region containing the dPal sequence. These substitutions destroy the regulatory function of dPal (GUNARATNE et al. 1992). Unlabeled dPalGld was able to compete effectively with labeled dPalGld when present in only a 20-fold excess (Fig 3A, lane 2) whereas a 450x molar excess of mdPalGld (Fig 3A, lane 6) was required to compete at an equivalent level. Therefore these four substitutions also ablated GPAL binding as well as dGpal regulatory function. To further examine the importance of the flanking sequences surrounding dGpal two additional 33-bp oligonucleotides, dPalN and mdPALN, were constructed. dPalN contains the native 13-bp dPal sequence but has a randomized flanking sequence unrelated to the Gld promoter. mdPALN is identical to dPalN but contains the same four nucleotide substitutions in mdPalGld. Binding affinity was reduced somewhat when the sequences flanking dGpal were substituted with unrelated sequences (Fig 3B, lanes 7–12) indicating that these flanking sequences are important for increasing specificity. The binding affinity was further reduced when four substitutions were introduced in the 13-bp region containing the dGpal element (Fig 3B, lanes 13–18). In summary, the dPal sequence is the most important determinant of binding GPAL to dPalGld, but the flanking sequences also somewhat influence the binding.

View larger version (67K): In this window In a new window Download PPT slide   Figure 3. Identification of GPAL, a Drosophila nuclear protein that specifically binds the dGpal element. Gel mobility shift experiments were conducted using nuclear protein extract from one or more Drosophila developmental stages or tissues and double-stranded oligonucleotides corresponding to regions of the Gld promoter containing Gpal elements, isolated Gpal elements, mutant Gpal elements, or nonspecific DNA fragments. The following double-stranded oligonucleotides were used in these experiments: dPalGld, 30 bp containing the 13-bp dGpal and native Gld flanking sequences; mdPalGld, 30 bp containing four substitutions in dPal but otherwise identical to dPalGld; dPalN, 33 bp containing the 13-bp dPal sequence flanked by random sequences; mdPalN, 33 bp identical to dPalN but containing the same four substitutions in the dPal element as mdPalGld; EF-1, 22 bp containing the binding site of the EF-1 mammalian transcription factor and poly(dI) poly(dC) noncompetitive inhibitor. (a) The competitive ability of a multiply substituted mdPalGld fragment was compared with that of native dPalGld sequence. Lanes 1–6, radiolabeled dPalGld. Nonradiolabeled competitors were added to the reactions as follows: lanes 1–3 contained 10x, 20x, and 40x molar excess of unlabeled dPalGld DNA competitor, respectively; lanes 4–6 contained 50x, 150x, and 450x molar excess of unlabeled mdPalGld, respectively. (b) Lanes 1–20, radiolabeled dPalGld. Nonradiolabeled competitors were added to the reactions as follows: lanes 2–6 contained 5x, 15x, 45x, 135x, and 405x molar excess of unlabeled dPalGld DNA competitor, respectively; lanes 8–12 contained 5x, 15x, 45x, 135x, and 405x molar excess of unlabeled dPalN DNA competitor, respectively; lanes 14–18 contained 5x, 15x, 45x, 135x, and 405x molar excess of unlabeled mdPalN DNA competitor, respectively; lanes 19 and 20 contained 1000x and 4000x molar excess of EF-1 competitor, respectively. In addition all reactions contain 1000 ng of poly(dI) poly(dC) as nonspecific competitor. (c) Developmental stage- and tissue-specific expression of GPAL. Radiolabeled dPalGld, in the presence of unlabeled nonspecific competitor poly(dI) poly(dC) (1000x molar excess), was used to probe nuclear fractions of various developmental stages and isolated tissues. Lane 1, 0- to 12-hr embryos; lane 2, 12- to 24-hr embryos; lane 3, 12- to 24-hr embryos; lane 4, third instar larvae; lane 5, 24- to 48-hr third instar larvae; lane 6, spiracular glands dissected from third instar larvae; lane 7, 12- to 24-hr embryos (repeat of lane 2); lane 8, adult males; lane 9, adult females; lane 10, adult male abdomens; lane 11, adult male head and thorax; lane 12, adult male abdomens without ejaculatory bulbs; lane 13, ejaculatory bulbs dissected from adult males.

To further examine the specificity of binding of GPAL, three unrelated and unlabeled competitor DNAs were tested for their ability to act as competitors with dPalGld for binding GPAL. These competitors, including EF-1 (Fig 3B, lanes 19 and 20), calf thymus DNA (not shown), and plasmid DNA (not shown), were found to be 50- to 100-fold less efficient competitors as compared to DNA fragments containing the wild-type sequence of dPal.

On the basis of our Gld promoter mapping experiments (QUINE et al. 1993 ; GUNARATNE et al. 1994 ) and experiments described herein, we predicted that the GPAL protein should be present in several tissues. Two of these tissues, which can be readily dissected, are the anterior spiracular glands of third instar larvae and the male ejaculatory bulb. In addition we predicted that the level of expression of GPAL should be high in embryos as compared to third instar larvae and should be substantially higher in adult males than in adult females. Mobility shift experiments were conducted (as above) using nuclear extracts of developmentally staged animals and hand-dissected tissues (Fig 3C). As predicted GPAL is at least an order of magnitude higher in embryos as compared to third instar larvae (Fig 3C, lanes 1–3 vs. lanes 4 and 5). However, the anterior spiracular glands of third instar larvae are highly enriched for GPAL (Fig 3C, lane 6). Adult male extracts and extracts isolated from adult male ejaculatory bulbs are also enriched for GPAL (Fig 3C, lane 13). A tissue subtraction experiment also shows that most of the GPAL protein in the abdomen of adult males is present in the ejaculatory bulb (lane 13 vs. lanes 10–12). A nuclear protein in D. virilis was found to comigrate with GPAL of D. melanogaster and was also found to be highly enriched in the male ejaculatory bulb (not shown). A high level of GLD is present in the male ejaculatory bulb of D. virilis (SCHIFF et al. 1992 ).

A 361-bp fragment of SREC contains two separate enhancers and a silencer:
We previously reported the presence of redundant enhancer elements dispersed in the 3.3-kb SREC that could activate reproductive organ expression (KEPLINGER et al. 1996 ). Results from two overlapping fragments suggested that a strong adult ejaculatory duct-oviduct enhancer mapped in a 361-bp region as well as a more general somatic reproductive organ enhancer that activates expression during the metamorphic development of these organs. This 361-bp fragment (denoted SREC 2) as well as three small overlapping subfragments (denoted SREC 2.1, SREC 2.2, and SREC 2.3) were inserted upstream of the -425 Gld promoter/lacZ P-element expression vector (Fig 4). Replicate transgenic lines were obtained for each of the four constructs and the expression of lacZ was determined in adult reproductive tissues (Table 4). As predicted, the 361-bp SREC 2 activated the Gld promoter in the adult male ejaculatory duct and female oviduct as well as the developing ejaculatory duct, the cortex of the ejaculatory bulb, oviduct, and spermathecae. Surprisingly, the 5' most 110-bp subfragment (SREC 2.1) activated the Gld promoter in the adult male ejaculatory bulb cortex and female uterus, two tissues lacking expression in the SREC 2 transgenic lines. In addition SREC 2.1 activates reporter gene expression in the developing ejaculatory bulb, uterus, and seminal receptacle. SREC 2.2 (middle one-third of SREC 2) did not activate the Gld promoter in any reproductive tissue. SREC 2.3 activated the Gld promoter in the adult and developing ejaculatory duct. Within SREC 2.3 is a 9-bp element (TTTTGAATG) similar to an ejaculatory duct enhancer element present in the Drosophila Adh gene (FANG et al. 1991 ; FANG and BRENNAN 1992 ). Similar sequences are present in regulatory regions of two other genes, andropin and Est-6, that are expressed in the ejaculatory duct (KAROTAM and OAKSHOTT 1993; SERGEEV et al. 1995 ).

View larger version (15K): In this window In a new window Download PPT slide   Figure 4. Mapping the labial and somatic reproductive organ enhancers in the Gld gene. Transcript map of the Gld gene. Solid rectangles represent exons I–IV. Thin line represents 5' flanking sequences, introns, and 3' flanking sequences. B, BamHI; BII, BglII; BI, BglI; C, ClaI; HII, HincII; P, PstI; R, EcoRI; and X, XbaI. Open rectangles below the transcript map represent fragments examined for enhancer activity. The labial enhancer (LBE) is localized to the XC fragment. The family of SREC enhancers are located in the XR fragment and are composed of the BMSE (fragment 34), the BCUE (fragment 2.1), and the EOE (fragment 2.3).

Overall the 361-bp SREC 2 is composed of two distinct tissue-specific enhancers, one for the cortex of the ejaculatory bulb and the uterus (BCUE) and one for the oviduct and ejaculatory duct (EOE; Table 4; Fig 4). In addition, the SREC 2.3 fragment appears to repress enhancer activity of SREC 2.1 in the uterus (at both stages) and in the ejaculatory bulb cortex at the adult stage. This repressor element may correspond to an element distinct from the ejaculatory duct-oviduct enhancer or the activity of the EOE may block the enhancer activity of the BCUE.

The developmental homology of the female and male somatic reproductive organs is correlated with the Gld enhancer tissue specificity:
The genital imaginal discs contain primordia for both male and female reproductive organs and the major function of the sex-determination pathway is to repress the opposite sex primordia (EPPER and NOTHIGER 1982 ). A variety of fate mapping and transplantation experiments of these primordia have also shown that specific male and female organs are serially homologous (DOBZHANSKY 1930 ; LITTLEFIELD and BRYANT 1979 ; DOBENDORFER and NOTHIGER 1982 ; CHEN and BAKER 1997 ). Evidence for serial homology is most compelling for the male ejaculatory duct and female oviduct. Under appropriate experimental conditions the primordia for each can be redirected to differentiate into the organ normally found in the opposite. We show for the first time that a single enhancer (EOE) specifically activates gene expression in these two homologous organs. Similar fate mapping experiments are somewhat less clear about the developmental homology of the female uterus and spermathecae and male ejaculatory bulb. We present herein new evidence showing that (1) the female uterus is homologous to the cortex of the male ejaculatory bulb, (2) the paired female spermathecae are homologous to the two hemispheres of the ejaculatory bulb medulla, and (3) the Gld gene contains two distinct enhancers that correspond to these two newly discovered developmental homologies.

BELOTE and BAKER 1982 isolated and characterized temperature-sensitive mutants of the transformer (tra) gene, a gene required for female sexual development. They showed that if chromosomal female (XX) tra^ts mutants were reared throughout development at the restrictive temperature (29°) they would develop as males with normal male somatic reproductive organs whereas if they were reared at the permissive temperature (18°) they would develop as females. We showed that if such females were reared at the restrictive temperature for the first few days of development until the decision to select the appropriate primordia for differentiation and then shifted to the permissive temperature, Gld expression was misregulated in the resultant male reproductive organs (FENG et al. 1991 ). Specifically, Gld expression was derepressed in the ejaculatory bulb medulla. Reciprocal temperature shift experiments resulted in depression of Gld expression in female oviduct. We report herein that if these temperature shift experiments are performed during the period when the decision to select the sex-appropriate primordia, elements of both male and female reproductive tissues are produced. In several cases, we observed formation of hybrid reproductive organs composed of fused spermathecae surrounded by uterine tissue yielding a structure very similar to the male ejaculatory bulb medulla and cortex, respectively (Fig 5, b–e) In these structures, the spermathecal caps exhibit aberrant lateral extensions similar to the horns of the ejaculatory bulb medulla. Most intersexual reproductive tracts generated were composed of both male ejaculatory duct and female oviduct that were both attached directly to the spermathecae. The spermathecae were partially transformed to ejaculatory bulb medullar hemispheres. In normal males, the ejaculatory duct directly connects to the ejaculatory bulb medullar hemispheres (Fig 5A). We propose that this hybrid organ is created by simultaneous activation of the male and female genital disc primordia resulting in the development of pseudoejaculatory bulb composed of the two homologous female tissues, namely the uterus forming the cortex and the paired spermathecae forming the two hemispheres of the medulla. Previous fate mapping experiments (LITTLEFIELD and BRYANT 1979 ; DOBENDORFER and NOTHIGER 1982 ) are consistent with this hypothesis although such studies did not consider the homology of the ejaculatory bulb cortex and medulla separately.

View larger version (97K): In this window In a new window Download PPT slide   Figure 5. Modulating the activity of the tra-2 sex-determination gene during prepupal development results in adult male-female hybrid reproductive organs. (a) Normal adult female reproductive tract shows expression of GLD in the proximal and distal terminal segments of the spermathecal ducts of the spermatheca (SP), which is attached to the anterior end of the uterus (UT). (b) Normal adult male ejaculatory duct (ED) shows expression of GLD but the ejaculatory bulb medulla (right upper arrow) and cortex (right middle arrow) do not. Right lower arrow, apodeme projecting from the medulla. (b–e) XX tra-2[ts1] were reared at the permissive temperature until the early prepupal stage and then shifted to restrictive temperature for metamorphic development. Often hybrid male-female organs were generated in such individuals, indicating that both male and female primordia of the genital imaginal disc had undergone differentiation. In such animals the male ejaculatory duct and female oviduct are typically present as separate organs. However, the female spermathecae and uterus and male ejaculatory bulb were typically absent and in their place male/female hybrid organs were found to be connected to the ejaculatory duct and oviduct, anteriorly, and to the external genitally, posteriorly. The various hybrid organs we observed from these animals represented a continuum from female-like organs to male-like organs, c through f. (c) The upper arrow shows a paired structure unmistakably similar to the paired spermathecae but also shows lateral extensions similar to the paired ejaculatory bulb medulla. The middle-right arrow shows tissue similar to the uterus and ejaculatory bulb cortex. The lower-left arrow shows tissue similar to the spermathecal ducts. (d) Hybrid organ with projecting ejaculatory duct and oviduct. The upper arrow indicates tissues similar to the male ejaculatory bulb medulla and female spermathecae. The lower arrow indicates tissue similar to male ejaculatory bulb cortex and female uterus. (e) This hybrid organ is more male-like than are those in b and c. The upper arrow shows tissue similar to the ejaculatory bulb medulla but is still cap-like similar to the female spermathecae. The middle arrow shows tissue similar to the male ejaculatory bulb cortex and female uterus. The lower arrow shows a rudimentary male ejaculatory bulb apodeme projecting from the hybrid bulb cortex/spermathecae. (f) XX, tra-2 [ts1] pseudomale reproductive organs. The fly from which these organs were isolated was raised at the permissive temperature until late prepupal stage (a few hours later than c, d, and e) and then shifted to the restrictive temperature. In this case, the adult exhibits normal male ejaculatory duct and ejaculatory bulb. However, the expression of GLD is depressed in the ejaculatory bulb. This latter result was previously shown by us (FENG et al. 1991 ).

The BCUE activates expression in the male ejaculatory bulb cortex and uterus, organs which we propose are developmentally homologous. Previously we had identified an adjacent enhancer, downstream in SREC (denoted segment 34 in Fig 4) that activates expression in male ejaculatory bulb and female spermathecae. Reanalysis of these data and transgenic lines now show this segment activates expression specifically in the male ejaculatory bulb medulla and female spermathecae. We denote this enhancer as the ejaculatory bulb medulla and spermathecae enhancer (BMSE). The BMSE, similar to the BCUE and EOE, activates expression in homologous male and female tissues.

An embryonic labial enhancer is present in the first intron of Gld adjacent to SREC:
During embryonic development GLD is expressed in the antenno-maxillary complex and labium. Previously we showed that the -425 to +84 promoter region of Gld was sufficient to drive expression in the embryonic antenno-maxillary complex immediately anterior to the labium (Fig 2D) as well as all other somatic tissues throughout preadult development. The embryonic labium is the only known nonreproductive organ in which this minimal promoter region is not sufficient for Gld expression. In the process of deletion mapping the SREC, we discovered an enhancer that activates expression in the embryonic labium (Fig 2A). The labial enhancer (LBE) was identified in a 1.3-kb fragment located 639 bp downstream of the Gld promoter. When present with either the Gld promoter or a heterologous hsp70 promoter, LBE activates a highly specific and restricted expression in the embryonic labium. Deleting all of the Gpal elements from the Gld promoter does not affect the ability of this enhancer to activate the Gld promoter in this tissue. Thus despite the coincident expression of Gld in the antenno-maxillary complex and the labium, their transcriptional regulation is quite distinct, the former controlled by the proximal promoter elements and the latter controlled by a downstream enhancer.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Gpal elements drive expression in both reproductive and nonreproductive tissues:
We previously showed that the dGpal element alone is sufficient to drive expression of a heterologous hsp70 promoter in the anterior spiracular glands of third instar larvae and in the adult male ejaculatory bulb of Drosophila (GUNARATNE et al. 1994 ). GPAL, a nuclear protein that binds specifically to dGpal, is readily detectable by mobility shift assays in the ASG and ejaculatory bulb. GLD expressed in the ASG is secreted and performs an essential developmental function (CAVENER and MACINTYRE 1983 ; COX-FOSTER et al. 1990 ). Although GLD is not normally expressed in the adult male ejaculatory bulb of melanogaster, we found that a D. pseudoobscura Gld transgene is expressed in this organ at the adult stage when inserted into the genome of D. melanogaster (SCHIFF et al. 1992 ). Therefore it is not surprising that the dGpal element, taken out of context of other Gld regulatory elements, can activate expression in the melanogaster adult ejaculatory bulb. Presumably, silencer elements (present in D. melanogaster but not D. pseudoobscura Gld) normally block the activation of Gld in the adult ejaculatory bulb. This hypothesis is supported by the fact that the D. melanogaster Gld gene is derepressed in this organ upon partial derepression of the female sex-determining factors in males (FENG et al. 1991 ). Despite the importance of both SREC and the dGpal element to reproductive tract expression, it is interesting to note that dGpal is apparently incapable of productively interacting with SREC to expand expression of the heterologous hsp70 promoter beyond the reproductive organs that are independently activated by SREC and dGpal. Moreover, selectively mutating the dGpal element within its normal context flanked by other Gpal elements has a negligible impact on the interaction of SREC and the Gld proximal promoter region. In contrast, selective mutation of the dGpal element obliterates expression in the larval anterior spiracular gland, a developmentally essential site of GLD expression.

We speculate that over the course of evolution of Gld function and expression, dGpal initially functioned solely for expression in tissues essential for GLD function in cuticle metabolism and cellular immunity. By the process of slipped-strand mutagenesis, the number of Gpal elements was expanded and SREC was transposed to the first intron, thus adding expression of Gld to the reproductive organs.

Synergism among Gpal elements:
Mutation of all six Gpal elements within the -425-nt Gld promoter results in a dramatic attenuation of expression in all reproductive tract tissues except the ejaculatory duct. Adult expression patterns are perturbed by progressive mutation of subsets of Gpal elements. Adult female seminal receptacle expression is abrogated first, followed by spermathecae and vaginal plate expression and finally oviduct expression.

The observed functional interactions among the Gpal elements may arise from cooperative DNA binding or from simultaneous contact of different rate-limiting components of the transcription initiation complex. We suggest that synergism among Gpal elements is likely at the level of cooperative DNA binding to GPAL since Gpal elements are closely linked as required for cooperative DNA binding (JIANG and LEVINE 1993 ; SZYMANSKI and LEVINE 1995 ). Several studies have identified the number and quality of binding sites as key determinants reaching the critical threshold for transcriptional activation (STRUHL et al. 1989 , STRUHL et al. 1992 ; JIANG et al. 1991 ). Alteration of binding sites reprograms the critical threshold required for expression in a specific tissue. Various Gld promoter mutations that we have generated and tested alter the number and/or quality of binding sites. Some tissues are considerably more sensitive to the loss of one or more Gpal elements than others. The differential tissue sensitivity to alterations in Gpal elements may reflect the level of GPAL in each tissue. Some of the reproductive tract expression patterns displayed in Gpal mutants of D. melanogaster are reminiscent of the expression patterns found in other Drosophila species, which show markedly different patterns of expression. We suggest that species-specific variation in adult expression patterns could be due to the unique threshold response of each species Gld promoter to GPAL, determined by its unique number and quality of Gpal elements. GPAL has also been detected in the reproductive organs of two other Drosophila species, pseudoobscura and virilis.

Enhancer-promoter synergism:
While the Gld somatic reproductive organ enhancer complex can activate a heterologous promoter in the male ejaculatory duct similar to many other intronically localized eukaryotic enhancers (e.g., LIU et al. 2000 ), we found that the other reproductive tissues require a combination of proximal promoter elements and the SREC. This interdependence may be the result of one or more mechanisms. SREC bound factors may make productive contacts with a Gld promoter specific transcriptional initiation complex, but not with a canonical transcriptional initiation complex (such as that of the hsp70 promoter). Such a mechanism has been suggested for the muscle-specific enhancer of the mouse myoglobin gene (WEFALD et al. 1990 ) and the Drosophila gooseberry and gooseberry neuro enhancers (LI and NOLL 1994 ). This is an attractive model because, as previously noted, Gld has a highly conserved atypical TATA box sequence, TTTAAAA, which might direct the assembly of a Gld promoter specific transcriptional initiation complex. We tested this hypothesis by converting the atypical Gld promoter to canonical TATA box and found that the reproductive tract expression was indeed abrogated in two female tissues. However, in all of the other tissues that depend upon specific interaction of the Gld promoter and SREC, expression was unaltered. Herein we have identified a cluster of repetitive sequence elements upstream of the TATA box denoted Gpal elements as indispensable for expression in this larger subset of reproductive tissues. Moreover, there is a strong functional synergism between these elements and the SREC. This synergism may be direct, wherein the respective factors interact to bind DNA in a cooperative manner (GINIGER and PTASHNE 1988 ; JANSON and PETTERSSON 1990 ; XIAO et al. 1991 ), or indirect, wherein the respective factors simultaneously contact different rate-limiting factors in the transcription initiation complex (CAREY et al. 1990 ; LIN et al. 1990 ; JIANG and LEVINE 1993 ; MANNERVIK et al. 1999 ).

Why is SREC activation in most reproductive tract tissues strictly dependent on Gpal elements whereas adult ejaculatory duct activation is not? One possibility is that the male ejaculatory duct is activated by a completely separate set of elements present in SREC. Indeed we have shown that SREC is composed of several dispersed enhancer elements that activate expression in different subsets of the somatic reproductive organs (KEPLINGER et al. 1996 ). Further mapping experiments reported herein show that a 140-bp region contains an ejaculatory duct and oviduct enhancer. All of the various SREC subfragments that activate the Gld/lacZ fusion gene in the ejaculatory duct also activate expression in the oviduct. Only when these fragments are tested with the heterologous hsp70 promoter do we observe expression restricted to the ejaculatory duct. However, it should be noted that the Gld promoter/SREC transgene activates a substantially higher level of expression than does the hsp70/SREC transgene. Together these findings suggest that EOE trans-activating factors may be sufficiently high in the adult ejaculatory duct to obviate the need for Gpal elements in order to activate a moderate level of expression but requires those elements for the normally very high level seen in this organ. An unsolved paradox is why the Gld gene is not normally expressed in the adult female oviduct. We speculate that although the oviduct requires both Gld promoter and enhancer elements, the presence of other elements in the Gld gene act to block the expression of the Gld mRNA in the oviduct at the adult stage. We have been able to relieve the repression of Gld expression in the oviduct by transiently expressing male sex-determining factors immediately before differentiation of the oviduct occurs, showing that the sex-determination pathway is responsible for the repression of Gld in the oviduct.

Somatic reproductive organ enhancers activate expression in homologous organ pairs in the males and females and exhibit epistatic interactions:
We have identified three distinct enhancers that activate reporter gene expression in specific pairs of somatic reproductive organs. Remarkably, each of these enhancers activates expression in one female organ and one male organ. These male and female organ pairs were previously shown or shown herein to be developmentally homologous. To our knowledge such enhancers that activate expression in developmentally homologous reproductive organs between the two sexes have not been reported. The EOE has an additional function to silence the activity of the nearby BCUE in the ejaculatory bulb cortex and uterus and the BCUE in the ejaculatory bulb medulla at the adult stage. The close distance of the EOE to the BCUE may be important for repression as was shown to be important for the activity of the giant repressor in Drosophila (HEWITT et al. 1999 ). It is important to note that the EOE does not negatively effect these other enhancers during preadult development. The silencing activity of the EOE at the adult stage is in part controlled by the sex-determination pathway, inasmuch as expression of Gld in the ejaculatory bulb medulla is derepressed in males that have received a brief pulse of feminizing factors by manipulation of the sex-determination pathway. We speculate that this occurs by specifically disrupting the silencing activity of the EOE on the BMSE.

A summary of the tissue-specific regulatory elements controlling the Gld gene and a model of Gld transcriptional regulation is shown in Fig 6. Control of Gld transcription in most of the nonreproductive tissues (antenno-maxillary complex (AMC), ASG, hypoderm, and wings) is exerted through the Gpal elements independent of the Gld enhancers. The temporal expression of Gld in these tissues is controlled by the ecdysone as mediated through the proximal promoter elements (MURTHA and CAVENER 1989 and our unpublished data). However for the embryonic labium, tissue specificity is controlled exclusively through an enhancer (LBE) and does not require Gld proximal promoter elements or Gld's unique TATA box. Regulation in the somatic reproductive organs is much more complex. All of these organs require to varying degrees the Gpal elements with possible exception of the ejaculatory duct at the adult stage. The spermathecae and ejaculatory bulb medulla are at one extreme since they are the only reproductive organs for which Gpal elements are sufficient in the absence of Gld enhancers. At the other extreme, the ejaculatory duct requires only an EOE. However, despite their differences in regulation, all three of these organs require both the Gpal proximal promoter elements and intronic enhancers for high levels of expression. The seminal receptacle and uterus are highly sensitive to mutations in Gpal elements or the TATA sequence and the composition of the enhancer elements. Although certain enhancer regions are required for seminal receptacle expression, we have not been able to pinpoint a specific element that it requires. Regulation of Gld in the male ejaculatory bulb and female oviduct are the most complex and intriguing. These organs normally express Gld mRNA during their development but not at their sexually mature state. Repression of Gld expression at the adult stage in these organs is predetermined during late larval-early pupal development by the sex-determination pathway. The repression of adult expression in the ejaculatory bulb medulla is mediated by the EOEs that both activate expression in the oviduct and ejaculatory duct and silence the activity in the adult ejaculatory bulb medulla and female uterus. Of the five tissues that are activated by the EOE, BMSE, and BCUE during preadult development, only the ejaculatory duct and spermathecae express GLD at the adult stage; therefore an additional silencing activity is required to specifically repress the EOE in the adult female oviduct. The elements responsible for repression of Gld mRNA in this organ (denoted sex silencers, SEXS) remain to be mapped within a 17.5-kb fragment that we showed contained all of the Gld regulatory elements including SEXS. Preliminary data indicate that SEXS is located in the distal half of the first intron downstream of SREC.

View larger version (70K): In this window In a new window Download PPT slide   Figure 6. A summary of the regulatory elements controlling the transcriptional regulation of Gld and a model of their interactions. (Top) Gld regulatory elements controlling Gld tissue-specific regulation are shown in the first row. A plus (+) indicates that the element is required to activate expression in the organ shown in the first column. A minus (-) indicates that the element blocks expression. With the exception of activation of expression in the seminal receptacle and repression of oviduct expression, the activity of the SREC complex is composed of the three reproductive organ enhancers, EOE, BCUE, and BMSE, and their interaction among themselves and with the Gld promoter elements. (Bottom) Model of the interaction of the Gld intronic enhancers, silencers, the proximal promoter elements, and the promoter. The locations of these elements have been mapped as described herein with the exception of the SEXS element which is only known to map within the 17.5-kb fragment defining the total Gld gene.

With the exception of the somatic reproductive organs, tissue-specific expression of Gld is highly conserved among the genus Drosophila (SCHIFF et al. 1992 ). Functional conservation has also been conserved as demonstrated by the ability of the D. pseudoobscura Gld gene to rescue D. melanogaster Gld null mutants (KRASNEY et al. 1990 ; SCHIFF et al. 1992 ). This interspecific gene transformation experiment was particularly revealing with respect to the pattern differences in the adult somatic reproductive organs observed between these two species. The D. pseudoobscura Gld gene when introduced into the genome of D. melanogaster was expressed in two organs, the ejaculatory bulb medulla and seminal receptacle, in which the host (D. melanogaster) Gld gene is not expressed at the adult stage. However, the host D. melanogaster Gld gene can be expressed ectopically in these two organs if these females are subjected to a short burst of male sex-determining factors immediately prior to their differentiation. In addition, the heterospecific D. pseudoobscura Gld gene was expressed in the male ejaculatory duct of the host D. melanogaster males, a tissue in which the Gld gene of D. pseudoobscura is not normally expressed. Introducing the entire male ejaculatory duct of D. pseudoobscura into a D. melanogaster host via imaginal disc transplantation methods, however, showed that the donor ejaculatory duct did not express the Gld gene. Collectively these experiments indicate that the D. pseudoobscura gene contains elements that can activate expression in the adult ejaculatory duct but that the D. pseudoobscura ejaculatory duct lacks the necessary cell autonomous regulatory factors to activate these elements. On the other hand, the D. pseudoobscura Gld gene appears to lack the silencer elements observed in the D. melanogaster Gld gene that block expression in the female seminal receptacle and male ejaculatory bulb medulla. D. teisseri, a species much more closely related to D. melanogaster, lacks GLD expression in the ejaculatory duct and oviduct at both the preadult and adult stages but otherwise shows the same pattern of expression as D. melanogaster in other somatic reproductive organs (CAVENER 1985 ; ROSS et al. 1994 ). Future experiments will focus on identifying the regulatory elements and/or factors that underlie D. teisseri unique pattern differences in the somatic reproductive organs.

	ACKNOWLEDGMENTS

We thank Anna Rabetoy for technical assistance. This work was supported by National Institutes of Health grant GM-34170.

Manuscript received May 23, 2000; Accepted for publication November 2, 2000.

	LITERATURE CITED

TOP ABSTRACT MA