The Gene Search System: A Method for Efficient Detection and Rapid Molecular Identification of Genes in Drosophila melanogaster

Corresponding author: Toshiro Aigaki, Department of Biological Sciences, Tokyo Metropolitan University, 1-1 Minami-osawa, Hachioji-shi, Tokyo 192-0397, Japan., aigaki-toshiro{at}c.metro-u.ac.jp (E-mail)

Communicating editor: T. C. KAUFMAN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We have constructed a P-element-based gene search vector for efficient detection of genes in Drosophila melanogaster. The vector contains two copies of the upstream activating sequence (UAS) enhancer adjacent to a core promoter, one copy near the terminal inverted repeats at each end of the vector, and oriented to direct transcription outward. Genes were detected on the basis of phenotypic changes caused by GAL4-dependent forced expression of vector-flanking DNA, and the transcripts were identified with reverse transcriptase PCR (RT-PCR) using the vector-specific primer and followed by direct sequencing. The system had a greater sensitivity than those already in use for gain-of-function screening: 64% of the vector insertion lines (394/613) showed phenotypes with forced expression of vector-flanking DNA, such as lethality or defects in adult structure. Molecular analysis of 170 randomly selected insertions with forced expression phenotypes revealed that 21% matched the sequences of cloned genes, and 18% matched reported expressed sequence tags (ESTs). Of the insertions in cloned genes, 83% were upstream of the protein-coding region. We discovered two new genes that showed sequence similarity to human genes, Ras-related protein 2 and microsomal glutathione S-transferase. The system can be useful as a tool for the functional mapping of the Drosophila genome.

GENOME sequencing and expressed sequence tag (EST) projects are rapidly progressing in various organisms. The next step in exploiting genomics requires an efficient method to detect and identify genes for functional mapping of the genome. Genetic approaches in Drosophila melanogaster have defined many genes that have been informative for understanding the function of their counterparts in vertebrates, including humans (SIDOW and THOMAS 1994 ; BANFI et al. 1996 ). It is desirable to develop a method for efficient detection of genes and for mapping them on the genome on the basis of their function. In Drosophila, P-element insertional mutagenesis appears to be the most suitable method for this purpose, because the sequence information of vector-flanking DNA can be obtained relatively easily (COOLEY et al. 1988 ; BELLEN et al. 1989 ; BIER et al. 1989 ). In fact, the gene disruption project with P elements is an integral component of the Berkeley Drosophila Genome Project (BDGP); the insertion lines contribute by serving as the materials for obtaining DNA markers and for studying gene function (SPRADLING et al. 1995 ).

Mutations caused by P-element insertion are principally loss of function. One possible limitation of a loss-of-function screen is the sensitivity of phenotype detection. Genes that are not essential for normal development would not be detectable on the basis of easily scorable phenotypes, such as viability or visible phenotypes (MIKLOS and RUBIN 1996 ). Another potential problem with P-element mutagenesis is that a significant fraction of P-element-induced mutations are not associated with the insertions (DÉAK et al. 1997; SALZBERG et al. 1997 ). In other words, molecular information obtained from P-element-flanking DNA is not always relevant to the phenotype because it might be caused by background mutations.

Gain-of-function mutagenesis is an alternative approach to identifying genes. Misexpression of genes using transgenic technology has been widely used to assess gene functions, especially since the GAL4-UAS system was introduced into Drosophila (BRAND and PERRIMON 1993 ). There are genes whose loss-of-function mutations do not show any obvious phenotype, but whose misexpression causes obvious phenotypes that are suggestive of the genes' normal functions (CHIBA et al. 1995 ; CHEN et al. 1997 ). Thus, gene detection based on a gain-of-function phenotype can identify genes that are not uncovered by loss-of-function phenotypes. Genetic screens will be more efficient with a P element that is capable of inducing forced expression of a gene, in addition to having the potential to disrupt a gene. The GAL4-UAS system that allows conditional gene expression appears to be suitable for a systematic generation of gain-of-function mutations. Insertion of a P-element vector containing upstream activating sequence (UAS) into the Drosophila genome allows GAL4-dependent forced expression of genes flanking the inserted P element. Although a gain-of-function phenotype alone is not sufficient to define the normal function of a gene, the method can be a powerful tool for functional mapping of the genome, along with providing rapid molecular identification of affected genes; genes would be detected based on an easily scorable phenotype, and identified molecularly through sequencing of misexpressed transcripts derived from an insert. Because background mutations occurring during P-element mobilization would not affect the GAL4-dependent phenotype, molecular information derived from misexpressed transcripts is closely associated with phenotypic information caused by forced expression of a gene. To obtain maximum efficiency, it is necessary to have a P-element vector that induces GAL4-dependent phenotypes at high frequency.

Two versions of P elements containing UAS for gain-of-function mutagenesis have been reported already (RORTH 1996 ; CRISP and MERRIAM 1997 ). Here we report a new version of the P-element vector containing UAS, which has a greater sensitivity in detecting genes than those previously reported. We also established a method for molecular identification of induced transcripts using reverse transcriptase PCR (RT-PCR) followed by direct sequencing. The system can be a useful tool for the functional mapping of the Drosophila genome.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Construction of gene search vector:
pCaSpeR3 P-element transformation vector (THUMMEL et al. 1988 ) was modified as follows: 5'P end of pCaSpeR3 was PCR mutagenized to create an EcoRI site using a primer corresponding to the 5'P end (5'GCCGAAGCTTACCGAAGTATACACTTAAAT) and an EcoRI site-flanked primer corresponding to the P transposase coding region (5'GCAGAATTCGACTAGTTTCATTTTTTTTTATTCCACGTAAGGG); the mutagenized product was digested with EcoRI and HindIII; the 140-bp fragment was used to replace a 4.7-kb EcoRI-HindIII fragment of pCaSpeR3, resulting in a 3.2-kb plasmid (pT1) containing the P-element ends and a multiple cloning site. The pCaSpeR3 was cleaved with EcoRI and EcoT22I, and the 4.0-kb fragment containing the mini-white gene and a multiple cloning site was subcloned into the EcoRI/PstI site of pUC19 (TOYOBO, Osaka, Japan), generating a 6.6-kb plasmid (pT2). Five tandem repeats of UAS for GAL4 with a core promoter derived from the Hsp70Bb gene were excised from pUAST (BRAND and PERRIMON 1993 ) with BamHI and EcoRI to obtain a 400-bp fragment, and with HindIII and XhoI to obtain a 440-bp fragment. The 400-bp fragment was subcloned into the BglII/EcoRI site located upstream of the mini-white gene in the pT2, producing a 7.0-kb plasmid (pT3). A 4.4-kb EcoRI-HindIII fragment (UAS/core promoter/mini-white gene) from the pT3 and the 440-bp HindIII-XhoI fragment (UAS/core promoter) from the pUAST were inserted together into the EcoRI/SalI site of the pT1. The resulting construct contains a P-element-based gene search (GS) vector composed of 5254 bp, which has a mini-white gene as a marker in the middle and two copies of UAS adjacent to a core promoter, one copy near the terminal inverted repeats at each end of the vector, and oriented to direct transcription outward. The GS vector was introduced into flies using P-element-mediated transformation (RUBIN and SPRADLING 1982 ) with the Df(1)w stock as a recipient. For genetic nomenclature, refer to FlyBase (http://flybase.bio.indiana.edu).

Generation and screening of GS vector insertion lines:
The GS vector inserted on the second chromosome of the Df(1)w stock was mobilized onto a CyO chromosome using Delta2-3 transposase (ROBERTSON et al. 1988 ). Then the vector was further mobilized onto various chromosomes using Delta2-3 transposase to generate a collection of new insertion lines. The 613 GS lines obtained include 147 on the X, 226 on the second, 237 on the third, and 3 on the fourth chromosomes. Insertions were kept in homozygous state or balanced with Binsinscy, SM1, or TM3 for the X, second, and third chromosome, respectively. To induce forced expression of vector-flanking DNA, GS lines were crossed to GAL4-expressing lines and the F₁ individuals carrying both a GAL4 transgene and a GS insert were screened for lethality and visible phenotypes. The following GAL4 lines were used: 29BD-GAL4 (P{GawB}29BD; A. BRAND and N. PERRIMON, personal communication) obtained from A. Brand; c355-GAL4 (P{GawB}c355; HARRISON et al. 1995 ); dpp-GAL4 (P{GAL4-dpp.blk1}40C.6; STAEHLING-HAMPTON et al. 1994 ); and sev-GAL4 (P{GAL4-Hsp70.sev}2; BRUNNER et al. 1994 ) obtained from the Bloomington Drosophila Stock Center (Indiana). The GAL4 expression pattern was examined using UAS-GFP (P{UAS-GFP.S65T}T2; constructed by B. Dickson and obtained through the Bloomington Drosophila Stock Center) as a reporter. Flies were reared at 25° using standard fly culture medium.

Identification of induced transcripts:
To analyze the GAL4-induced transcripts, GS vector insertion lines were crossed to hs-GAL4 stock (P{GAL4-Hsp70.PB}89-2-1; constructed by A. Brand and obtained through the Bloomington Drosophila Stock Center). The F₁ third instar larvae were transferred into a 1.5-ml microfuge tube (20–30 individuals/tube), and heat-shocked at 37° for 1 hr. Poly(A)⁺ RNA was isolated from the larvae using the QuickPrep Micro mRNA purification kit (Amersham Pharmacia Biotech, Arlington, IL). mRNA was reverse-transcribed using the first-strand cDNA synthesis kit (Amersham Pharmacia Biotech) with a NotI site-flanked oligo(dT) primer (5'AACTGGAAGAATTCGCGGCCGCAGGAATTTTTTTTTTTTTTTTTT, Amersham Pharmacia Biotech). A total of 1 µl of the reaction was used to amplify both 5'P and 3'P transcripts by PCR using ELONGASE enzyme mix (GIBCO BRL, Gaithersburg, MD) in a total volume of 50 µl with the upstream common primer (5'CTGAATAGGGAATTGGGAATTCG) and the NotI site-flanked oligo(dT) primer. To amplify the transcripts of the 5'P or 3'P element ends separately, 1 µl of the first PCR reaction was reamplified in a total volume of 50 µl using either the 5'P-specific primer (5'GTGTATACTTCGGTAAGCTTCG) or the 3'P-specific primer (5'ATTGCAAGCATACGTTAAGTGGA) as an upstream primer together with a downstream primer (5'AGAACTGGAAGAATTCGCGG). PCR was carried out using a Perkin-Elmer (Norwalk, CT) gene amp PCR system 2400 or 9700 with the following thermal cycling program: 94° (60 sec), 16 cycles of 94° (15 sec) –65° (10 min), 12 cycles of 94° (15 sec) –65° (10 min with 15-sec increment for every cycle), and 72° for 10 min, then held at 4°. The resulting PCR products were electrophoresed on a 1.0% agarose (Type II; Sigma, St. Louis) gel, and the amplified bands were excised with a razor blade and subsequently purified using the QIAEX II gel extraction kit (QIAGEN, Chatsworth, CA). The purified DNA fragments were used as a template for sequencing reactions with the dRhodamine terminator cycle sequencing FS ready reaction kit (Perkin-Elmer) using the 5'3'P common primer (5'CGACGGGACCACCTTATGTTA). Sequencing was carried out using a Perkin-Elmer ABI PRISM genetic analyzer 310. Sequence similarity searches were performed using the BLASTN or BLASTX program (ALTSCHUL et al. 1997 ) with the NCBI nonredundant nucleic acid database and dbest, or NCBI nonredundant protein database.

Subcloning:
For subcloning of cDNA derived from the misexpressed transcripts of Rap2l and Mgstl, RT-PCR products obtained were blunt-ended using T4 DNA polymerase (TOYOBO), digested with NotI, and ligated into the NotI/EcoRV site of pBluescript SK+ (Stratagene, La Jolla, CA). At least three clones were sequenced to determine the structure of the cDNAs.

RACE:
The 5' end structure of wild-type transcripts of Rap2l was determined using the 5' rapid amplification of cDNA ends (RACE) system, version 2.0 (GIBCO BRL), according to the manufacturer's protocol. Poly(A)⁺ RNA was isolated from wild-type (Canton-S) larvae as described above. First-strand cDNA for Rap2l was synthesized using a gene-specific primer (R-1: 5'CTATAAAAGCGTACAACAA). A poly (C) tail was added to the 3' ends of the cDNA using terminal deoxynucleotidyl transferase (GIBCO BRL) and dCTP (GIBCO BRL). Tailed cDNA was amplified by PCR using a nested, gene-specific primer (R-2: 5'CGAACGATGGTGGCGAATACTT) and a poly (G)-containing anchor primer (GIBCO BRL). The cDNA was reamplified using a nested, gene-specific primer (R-3: 5'GGGTGCTGGCTGACTTCCTTT) and the anchor primer. R-3 was used as a primer for direct sequencing. Similarly, the 5' end structure of the Mgstl transcript was determined using gene-specific primers M-1 (5'AAGGTCTAGACCTATGTGCTC) for reverse transcription, M-2 (5'CGTTCGGATCGTCGAACTT) for the first PCR, and M-3 (5'CCTCTAGAAGACGGGATTGGAG) for the second PCR and for direct sequencing.

The 3' end structure of Rap2l transcript was determined by 3' RACE. The first strand cDNA was synthesized from poly(A)⁺ RNA from wild-type larvae using the NotI site-flanked oligo(dT) primer. The cDNA was amplified by PCR using a gene-specific primer (R-4: 5'TCGTCTCGGGATGCTTTATTGA) and the downstream primer used for the amplification of misexpressed transcripts described above. The cDNA was reamplified using a nested, gene-specific primer (R-5: 5'GCACAGAGCAATTCGCATCCAT). R-5 was used as a primer for direct sequencing. Similarly, 3' RACE for the Mgstl transcript was carried out using M-4 (5'GCGAATTCAAACACATACAATGGCC) as a gene-specific primer, which was also used for direct sequencing.

Analysis of Mgstl gene structure:
The genomic region containing Mgstl was amplified by PCR using primers M-1 and M-4. The amplified fragments were directly sequenced as described above using the PCR primers as a sequencing primer. The 5'- and 3'-flanking regions were obtained by inverse PCR; genomic DNA isolated from wild-type (Canton-S) flies was digested with HindIII, self-ligated, and PCR amplified using primers M-3 and M-5 (5'CTCGAATTCTTCGTGGCCTTGG). The amplified products were blunt-ended by T4 DNA polymerase (TOYOBO) and digested with HindIII. The resulting two fragments were subcloned into the HincII/HindIII site of pBluescript SK+, and at least three clones were sequenced. All restriction enzymes used in this study were purchased from TOYOBO.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Scheme of the GS system:
The GS system consists of three steps: (1) the generation of fly lines with single inserts that allow conditional forced transcription of genomic sequence, (2) induction of forced expression and screening for lines with a detectable phenotype, and (3) the molecular identification of transcribed sequences (Figure 1). We constructed a P-element-based GS vector utilizing the GAL4-UAS ectopic expression system (BRAND and PERRIMON 1993 ). The GS vector is similar to the EP element constructed by RORTH 1996 , but it contains two copies of the UAS adjacent to the core promoter from the Hsp70Bb gene, one copy near the terminal inverted repeats at each end of the GS vector and oriented to direct transcription outward. Thus, upon GAL4 activation the vector integrated into the genome will induce transcription toward the flanking DNA on both sides of its integration site. The induced transcripts may cause phenotypic changes if the products affect biological pathways operating in the organism. An insertion upstream of the protein-coding region will cause over- or ectopic expression of the gene, while an insertion downstream of the protein-coding sequence produces antisense RNA, which may interfere with translation of the wild-type mRNA (MCGARRY and LINDQUIST 1986 ; NICOLE and TANGUAY 1987 ). Both sense and antisense strands will be transcribed if the GS vector is inserted in the middle of a gene. Observation of a phenotypic change for a given GS insert would indicate that the insertion site is near or within a gene that is capable of altering phenotype; the transcripts are readily and rapidly identifiable using RT-PCR followed by direct sequencing.

View larger version (39K): In this window In a new window Download PPT slide   Figure 1. Schematic representation of the GS system. (A) Structure of the GS vector. The GS vector contains UAS and a core promoter derived from the Hsp70Bb gene near the inverted terminal repeats at both P-element ends. The mini-white gene is included as a marker. A collection of transgenic flies, each with a single insertion of the GS vector, were generated by mobilizing the vector in the genome using 2-3 transposase (see MATERIALS AND METHODS). (B) Screening of GS vector insertion lines (GS lines) for phenotypes. Flies from GS lines were crossed to flies bearing GAL4 drivers to induce forced expression of the vector-flanking sequences in the F1. The F1 were screened for lethality and visible phenotypes. (C) Molecular analysis of induced transcripts. Upon GAL4 activation, transcription occurs toward the flanking genomic sequences through the P-element ends. (1) GS lines were crossed to hs-GAL4 and poly(A)+ RNA was prepared from heat-shocked F1 larvae and (2) reverse-transcribed using an oligo(dT) primer. (3) cDNAs corresponding to the induced transcripts were amplified by two rounds of nested PCR using the vector-specific primers. (4) Finally, 5' end sequences of the cDNAs were determined by direct sequencing.

Generation and screening of GS vector insertion lines:
We generated a total of 613 GS vector insertion lines (GS lines for short) and screened for dominant synthetic phenotypes, such as lethality, semilethality (<50% viability), or visible phenotype in the adult structure using four GAL4-expressing lines as drivers. The frequency of producing any phenotype with forced expression of flanking DNA depends on the extent and level of GAL4 expression. Two P{GawB} enhancer-trap lines, 29BD-GAL4 and c355-GAL4, express GAL4 in all imaginal discs at high level (Figure 2). These were selected because they are likely to produce phenotypes at high frequency. The two other GAL4 drivers have more specific expression patterns: dpp-GAL4 is expressed along the anterior/posterior compartment boundary of each imaginal disc, and sev-GAL4 is expressed mainly in the eye imaginal discs (Figure 2). dpp-GAL4 allows us to assess the variability of effects caused by forced expression of flanking DNA in different body parts, while with sev-GAL4, the effects of forced expression are assessed in the eye, in which subtle perturbations of gene regulatory networks are detectable (XU and RUBIN 1993 ).

View larger version (89K): In this window In a new window Download PPT slide   Figure 2. Expression pattern in the imaginal discs of GAL4 lines used as drivers. (A–C) 29BD-GAL4. (D–F) c355-GAL4. (G–I) dpp-GAL4. (J) sev-GAL4. (A, D, G, J) Eye-antennal disc. (B, E, H) Wing disc. (C, F, I) Leg disc. 29BD-GAL4, a P{GawB} enhancer trap-line, expresses GAL4 in all imaginal discs at high level more or less ubiquitously. c355-GAL4 also expresses GAL4 in all imaginal discs, with a high level of expression in the wing pouch. dpp-GAL4 is expressed along the anterior side of anterior/posterior boundary of each imaginal disc. sev-GAL4 is expressed mainly in the eye-antennal discs. Driver expression pattern was examined using UAS-GFP as a reporter.

Figure 3 shows the frequency of phenotypes obtained with four distinct GAL4 lines. As expected from the expression pattern, a high frequency of phenotypes was obtained with the ubiquitously expressing GAL4 transgenes, 57% for 29BD-GAL4 and 48% for c355-GAL4. Frequency of lethality was correlated with the total frequency of phenotype (33, 29, 14, and 7% for 29BD-GAL4, c355-GAL4, sev-GAL4, and dpp-GAL4, respectively), while visible phenotypes were obtained approximately at the same frequency (20%), except for that with c355-GAL4 (13%). Figure 4 represents the number of lines that showed visible phenotypes in each body part, which roughly corresponds to where GAL4 is expressed. 29BD-GAL4, c355-GAL4, and dpp-GAL4 induced visible phenotypes in various body parts, because these express GAL4 in all imaginal discs. Phenotypes caused by sev-GAL4 are seen principally in the eye, as expected on the basis of expression of this driver. Overall, 394 lines (64%) showed a detectable phenotype in combination with at least one GAL4 line. 29BD-GAL4 appeared to be the most efficient driver to detect genes on the basis of a misexpression phenotype; it induced phenotypic changes in 88% of the GS lines that showed phenotypes with any of the GAL4 drivers.

View larger version (34K): In this window In a new window Download PPT slide   Figure 3. Frequency of phenotypes for 613 GS lines crossed to four different GAL4 drivers. Flies carrying a GS insert were crossed to flies carrying each of the four GAL4 drivers, and the F1 progenies carrying both transgenes were screened for lethality, semilethality, and visible adult phenotypes. Semilethal: viability was <50%. Visible phenotypes were scored if the penetrance was >50%. Lines showing both semilethality and visible phenotypes were included in the visible category. Flies were reared at 25°.

View larger version (26K): In this window In a new window Download PPT slide   Figure 4. Summary of visible phenotypes by body part. The number of lines showing any visible phenotypes in each body part is indicated for each GAL4 driver. Note that the phenotype frequency is correlated roughly with the extent and level of GAL4 expression in each body part; phenotypes appeared in various body parts in 29BD-GAL4, c355-GAL4, and dpp-GAL4, while sev-GAL4 induced phenotypes mainly in the eyes.

Molecular analysis of forced-expression transcripts:
To identify genes whose expression was forced in GS lines, we performed molecular analysis of the induced transcripts for 170 insertions of randomly selected GS lines among those that showed a phenotype upon forced transcription of flanking DNA. GS lines were crossed to the hs-GAL4 line, poly(A)⁺ RNA was isolated from heat-shocked F₁ larvae, and the transcripts derived from the vector insertion site were amplified with RT-PCR using vector-specific and oligo(dT) primers. The amplified cDNA fragments were subjected to single-pass sequencing using a vector-specific primer corresponding to the P-element end (see MATERIALS AND METHODS). In most of the cases (146 of 170 inserts), we obtained two distinct transcripts derived from a single insert, indicating that the GS vector was indeed capable of inducing transcription bidirectionally. Database searches of the obtained sequences revealed that 47% of insertions were in known sequences (Table 1). A total of 21% showed similarity to sequences of cloned genes, 18% matched reported EST sequences (BDGP/Howard Hughes Medical Institute Drosophila EST Project; D. HARVEY, L. HONG, M. EVANS-HOLM, J. PENDLETON, C. SU, P. BROKSTEIN, S. LEWIS and G. M. RUBIN, unpublished results), and 4% matched reported sequence tagged site (STS) sequences (BDGP; G. M. RUBIN, unpublished results; European Drosophila mapping Consortium; M. ASHBURNER, unpublished results). We investigated the insertion sites relative to the transcription start site for those that matched cloned genes or ESTs (Table 2). The 5'-most ends of mRNA reported so far were defined as +1. More than 50% of insertions were found between -150 and +100, most frequently in between -100 and -1 (Figure 5). With respect to the insertions in cloned genes, 83% were upstream of the protein-coding region (data not shown), suggesting that most of the phenotypes detected in this screen were caused by over- or ectopic expression of full-length products.

View larger version (33K): In this window In a new window Download PPT slide   Figure 5. Insertion sites of GS vector in 47 lines with a forced expression phenotype. The insertion sites relative to the transcription start site were investigated for the insertions in cloned genes or ESTs. The 5' most ends of mRNA reported so far were defined as +1. Insertions mapped between -500 and +500 are indicated. Insertions in an intron were not included. More than 50% of insertions were found between -150 and +100, most frequently in between -100 and -1.

Identification of a gene similar to human Rap2:
Molecular analysis of misexpressed transcripts revealed that two of the GS insertions were in new genes that showed sequence similarity to human genes (Table 1). We subcloned the RT-PCR products into a plasmid and sequenced the entire cDNA. The sequence of the misexpressed transcripts derived from the insert in line GS2069 showed a similarity to human Ras-related protein 2 (Rap2; PIZON et al. 1988 ); thus we named it Ras-related protein 2-like (Rap2l). Rap2l was detected as an insertion that showed a lethal phenotype when combined with 29BD-GAL4 or c355-GAL4, and a rough eye phenotype when driven by sev-GAL4 (Figure 6). Rap2l gene was localized to cytological map 60B by chromosomal in situ hybridization. It is indeed an active gene in the wild-type flies, since a cDNA encoding RAP2L was obtained by RT-PCR using poly(A)⁺ RNA prepared from the wild-type larvae as a template. A full-length cDNA encoding RAP2L was obtained by 5' and 3' RACE, which was 661 bases long with 5'- and 3'-untranslated regions of 47 and 65 bases, respectively. DNA sequence analysis revealed that the protein-coding region and 3'-untranslated region (UTR) were identical to those of the misexpressed transcript (Figure 7A). Based on the difference of 5'-UTR between the wild-type and the misexpressed transcripts, the insertion site of the GS vector was determined to be 339 bp upstream of the transcription start site (Figure 7A). Recently, the genome sequencing of this region (P1 clone DS00543) has been completed by BDGP (GenBank accession number AC004642), and has revealed that the Rap2l protein-coding region was interrupted by three introns. Sequence comparison between the genomic region and cDNA demonstrated that the misexpressed transcript was spliced and polyadenylated at exactly the same sites as the wild-type transcript (Figure 7A). Note that a single-pass sequencing read of 602 bases for the misexpressed transcript was long enough to reach the second exon that is separated from the first exon by 1160 bases of an intron.

View larger version (80K): In this window In a new window Download PPT slide   Figure 6. Examples of phenotypes caused by forced expression of vector-flanking DNA. (A, C, E, G) Wild-type eye, aristae (indicated by arrows), leg, and wing, respectively. (B) A rough eye phenotype of sev-GAL4/GS2069 fly. (D, F, H) Phenotypes observed in GS1051/+;dpp-GAL4/+ adults: (D) missing aristae (indicated by arrows), (F) leg with fused tarsal segments, and (H) wing with complex defects, such as fused, missing, or disrupted veins and notched margin.

View larger version (52K): In this window In a new window Download PPT slide   Figure 7. (A) Diagram of genomic organization of Rap2l and the structure of misexpressed transcript. The sequence of cDNA encoding RAP2L was obtained by 5' and 3' RACE using poly(A)+ RNA prepared from the wild-type larvae as a template. The exon-intron boundaries were determined on the basis of a comparison of sequences between a full-length cDNA and the genomic region containing Rap2l (P1 clone DS00543, GenBank accession number AC004642). Exons are represented by boxes with protein-coding sequence (dark gray) and UTR (light gray). Misexpressed transcript starting from the GS vector (indicated by a triangle) was spliced and polyadenylated as the wild-type transcript. The sequence corresponding to the hatched region was obtained by a single-pass sequencing, and used as a query for an initial search of databases. B, E, H, and S on the solid line at the bottom represent the restriction sites of BglII, EcoRI, HindIII, and SpeI, respectively. (B) Aligned amino acid sequences of RAP2L, human Rap2, and Drosophila RAP1. Identical residues for all three proteins are reverse-contrasted and those shared by two of them are highlighted. In residues of RAP2L, 68% (124/182) and 55% (100/182) were identical to human Rap2 and Drosophila RAP1, respectively.

Figure 7B shows the deduced protein sequence of the RAP2L compared to those of human Rap2 and Drosophila RAP1, whose gain-of-function mutation is known as Roughened (HARIHARAN et al. 1991 ). RAP2L protein is 182-amino-acid residues long and contains a GTP-binding domain shared by the Ras family proteins. The amino acid sequence was 68 and 55% identical to human Rap2 and Drosophila RAP1, respectively. The N-terminal residues including the GTP-binding domain were highly conserved among the three proteins, while the similarity was less for the carboxy termini.

Identification of a gene similar to human mGST:
Line GS1051 had a GS vector insertion in a gene whose sequence is similar to human microsomal glutathione S-transferase (mGST), which encodes an enzyme involved in the detoxification defense system (DEJONG et al. 1988 ). Misexpression of the vector-flanking DNA resulted in semilethality in combination with each of the four GAL4 drivers. All viable flies showed visible phenotypes in various body parts: a strong rough eye phenotype was induced by 29BD-GAL4 and sev-GAL4; a mild rough eye phenotype was produced by c355-GAL4 and dpp-GAL4; and missing aristae, fused tarsal segments, reduced size of scutellum, some missing macrochaetae, and notched wings were induced by 29BD-GAL4, c355-GAL4, and dpp-GAL4. Examples of such visible phenotypes caused by dpp-GAL4 are shown in Figure 6.

On the basis of sequence similarity, the gene was named Microsomal glutathione S-transferase-like (Mgstl) and localized to 19E by chromosomal in situ hybridization. A cDNA encoding MGSTL was amplified by RT-PCR using mRNA prepared from the wild-type larvae, and a full-length cDNA sequence was determined by 5' and 3' RACE. Analysis by 5' RACE revealed that the GS vector was inserted 39 bp downstream of the transcription start site, and 60 bp upstream of the first ATG codon for translation (Figure 8A). The 675-bases-long wild-type full-length cDNA contained an open reading frame encoding MGSTL. Sequences of the protein-coding region and 3'-UTR were identical to those of the misexpressed transcript (Figure 8A). The genomic region containing Mgstl was obtained by PCR and inverse PCR, and revealed that there is only one intron (378 bases) within this gene. Comparison of the sequences between the wild-type and the misexpressed transcript for Mgstl demonstrated that they were spliced and polyadenylated at exactly the same sites (Figure 8A). Figure 8B shows the deduced amino acid sequence of MGSTL consisting of 152 residues compared with that of human mGST consisting of 155 residues. The Mgstl intron position corresponded to the second intron in the human mGST gene, which contains three introns (KELNER et al. 1996 ). Although the sequence identity to human mGST was 45%, the hydrophobicity profile was very similar (data not shown). A single-pass sequencing was sufficient to obtain sequence information for the second exon, which contains a region with a high similarity to human mGST.

View larger version (34K): In this window In a new window Download PPT slide   Figure 8. (A) Diagram of genomic organization of Mgstl and the structure of misexpressed transcript. The sequence of cDNA encoding MGSTL was obtained by 5' and 3' RACE using poly(A)+ RNA prepared from the wild-type larvae as a template. The genomic region was obtained by PCR and inverse PCR, subcloned into plasmid, and sequenced. Exons are represented by boxes with protein-coding sequence (dark gray) and UTR (light gray). Misexpressed transcript starting from the GS vector (indicated by a triangle) was spliced and polyadenylated as the wild-type transcript. The sequence of the hatched region was obtained by a single-pass sequencing, and used for an initial search of databases. B, E, H, and P on the solid line at the bottom represent the restriction sites of BglII, EcoRI, HindIII, and PstI, respectively. (B) Aligned amino acid sequences of MGSTL and human mGST. In residues of MGSTL, 45% (68/152) were identical to human mGST, which is reverse-contrasted.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Gain-of-function screening based on misexpression phenotypes is an alternative to a loss-of-function screening approach to discover new genes (MIKLOS and RUBIN 1996 ; RORTH 1996 ; CRISP and MERRIAM 1997 ; RORTH et al. 1998 ). In a loss-of-function screen, COOLEY et al. 1988 showed that 15% of new P-element insertions caused phenotypes. This frequency may not correspond to the frequency of gene disruption by P elements. There must have been insertions that disrupted a gene, but phenotypic defects were not detected, because the defects were compensated by functionally overlapping genes, or phenotypes were too subtle for detection (MIKLOS and RUBIN 1996 ). Considering that the P elements are frequently inserted upstream of transcription start sites (SPRADLING et al. 1995 ), some P-element insertions might be near a gene but not disrupt their function. These two categories of insertions could be potentially detected with the gain-of-function mutagenesis; ectopic expression of genes in various tissues would increase the probability of producing a phenotype, and the site preference of P-element insertions is in favor of causing misexpression of a gene.

A high-efficiency vector in terms of inducing phenotypic changes would be valuable as a tool for functional mapping of the genome through discovery of genes on the basis of phenotypes and obtaining sequence information associated with them. The GS vector used in this study appeared to be very efficient in terms of phenotype frequency. The EP element constructed by RORTH 1996 was the first vector used for a systematic gain-of-function mutagenesis. CRISP and MERRIAM 1997 constructed another version of misexpression vector with the yellow gene as a marker, which is convenient for identification of flies carrying the misexpression vector and a GAL4 driver containing the white gene as a marker. We showed that the frequency of GAL4-dependent phenotypes was extremely high with the GS vector, 10-fold higher than those obtained with the EP element (RORTH 1996 ; RORTH et al. 1998 ). For instance, the frequency of phenotypes with sev-GAL4 was 38% for GS inserts, while the same driver induced phenotypic changes in 4% of EP inserts (RORTH 1996 ). Likewise, 32% of GS inserts showed phenotypes with dpp-GAL4, while only 2% of EP inserts had phenotypes with the same driver (RORTH et al. 1998 ). The EP element has the UAS enhancer/core promoter near the 3' end of the P element only, while the GS vector contains the UAS enhancer/core promoter at both ends of the vector. Although this modification increases by 2-fold the probability of inducing forced expression of genes, this cannot account for a 10-fold difference. The frequency of phenotypes also depends on the criteria of mutant phenotype, especially for visible phenotype. However, this is unlikely to be the cause of the 10-fold difference in the frequency of phenotypes, because the frequency of lethal phenotype (which is unambiguous) was extremely high for GS inserts compared to EP inserts (7.3 vs. 0.3% for dpp-GAL4 and 14 vs. 0.3% for sev-GAL4, respectively). Comparable data for the same GAL4 drivers are not available for the vector constructed by CRISP and MERRIAM 1997 . The high frequency of mutant phenotypes involving the GS vector must be attributed to its unique structure. The mechanism for the high efficiency is not clear, but it may be due to its insertion frequency near genes, or more likely due to the efficiency of forced expression of flanking DNA.

In the GS system, GAL4-dependent phenotypic changes simply indicate the presence of a gene near the vector insertion site, and this is sufficient for rapid detection and identification of new genes. For the functional mapping of the genome, it is important to obtain reliable molecular information from the insertion site that is associated with a phenotype. We have established a procedure for obtaining the sequence of misexpressed transcripts derived from an insertion. We used RT-PCR using vector-specific and oligo(dT) primers, followed by single-pass sequencing. Although it requires more steps compared to inverse PCR using the genome DNA as a template, mRNA sequences are more informative than genomic sequences, which might contain noncoding sequences. In fact, it was indeed the case for Rap2l and Mgstl genes, which we characterized in this study. The misexpressed transcripts of these genes were spliced correctly, and a single-pass sequencing of the RT-PCR products was sufficient to reach the second exon of each gene.

On the basis of the sequence similarity to a human Rap2, we identified a new gene, Rap2l. The amino acid sequence of RAP2L was also similar to that of Drosophila RAP1, the only member of the Rap family known in Drosophila. Dominant mutations of Rap1 have been shown to genetically interact with fat facets in eye development (LI et al. 1997 ). Human Rap2 has been thought to be a modifier of the Ras signaling pathway (PIZON et al. 1988 ), and it has been shown to have GTP-binding activity and a low intrinsic GTPase activity (LEROSEY et al. 1991 ). However, the function of human Rap2 has not been clearly demonstrated. Overexpression of Rap2 in cultured cells had no effects on cellular proliferation or transformation induced by the ras oncogene (JIMENEZ et al. 1991 ). Nevertheless, a high conservation of amino acid sequences between humans and Drosophila suggests that Rap2 has some important functions. In vivo studies using Drosophila mutants should facilitate understanding the function of Rap2 proteins. Loss-of-function studies are especially necessary to define cellular function. The P-element insert in the Rap2l locus is useful for generating loss-of-function alleles by local transposition (TOWER et al. 1993 ) or by excising the vector from the chromosome, which occasionally deletes flanking DNA (SALZ et al. 1987 ). The same is true for Mgstl, which encodes a protein similar to human mGST. The identity was 45%, but the hydrophobicity profile was very similar, suggesting that they share a functional similarity. Studies on mGST using Drosophila mutants should provide evidence for its in vivo function.

The progress of genome sequencing and the EST project is important for the functional mapping of the genome on the basis of gain-of-function screens. A partial sequence of cDNAs derived from misexpressed transcripts would be sufficient for identifying genomic DNA clones or cDNA clones that are available from BDGP through commercial vendors, which facilitates further analysis of individual genes. We found that 18% of the insertions with forced-expression phenotypes showed sequence similarity to ESTs. Since the EST data are rapidly growing, the GS system will identify many more genes corresponding to ESTs. Gain-of-function phenotypes obtained by forced expression of EST-corresponding genes might provide a clue as to their functions. The system may also identify genes that may not be found as an EST, such as those expressed normally at very low levels, expressed in a few cells, or expressed only transiently during development. The GS system should contribute to functional genomics as a method for easy detection and rapid molecular identification of genes in the Drosophila genome, and the obtained inserts will serve as materials to start loss-of-function studies on the new genes.

	ACKNOWLEDGMENTS

We thank S. Kawasaki, M. Matsuno, and T. Umemiya for contributing to the screening, A. Nose, Y. Fuyama, J. Merriam, M. Wolfner, and K. White for comments on the manuscript, and the Bloomington Stock Center for providing fly stocks. This work was supported in part by a Human Frontier Science Program (HFSP) grant (RG-377/93 B).

Manuscript received September 4, 1998; Accepted for publication November 3, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ALTSCHUL, S. F., T. L. MADDEN, A. A. SCHÄFFER, J. ZHANG, and Z. ZHANG et al., 1997  Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25:3389-3402[Abstract/Free Full Text].

BANFI, S., G. BORSANI, E. ROSSI, L. BERNARD, and A. GUFFANTI et al., 1996  Identification and mapping of human cDNAs homologous to Drosophila mutant genes through EST database searching. Nat. Genet. 13:167-174[Medline].

BELLEN, H. J., C. J. O'KANE, C. WILSON, U. GROSSNIKLAUS, and R. K. PEARSON et al., 1989  P-element-mediated enhancer detection: a versatile method to study development in Drosophila. Genes Dev. 3:1288-1300[Abstract/Free Full Text].

BIER, E., H. VAESSIN, S. SHEPHERD, K. LEE, and K. MCCALL et al., 1989  Searching for pattern and mutation in the Drosophila genome with a P-lacZ vector. Genes Dev. 3:1273-1287[Abstract/Free Full Text].

BRAND, A. H. and N. PERRIMON, 1993  Targeted gene expression as a means of altering cell fates and generating dominant phenotypes. Development 118:401-415[Abstract].

BRUNNER, D., K. DÜCKER, N. OELLERS, E. HAFEN, and H. SCHOLZ et al., 1994  The ETS domain protein pointed-P2 is a target of MAP kinase in the Sevenless signal transduction pathway. Nature 370:386-389[Medline].

CHEN, F., M. BARKETT, K. T. RAM, A. QUINTANILLA, and I. K. HARIHARAN, 1997  Biological characterization of Drosophila Rapgap1, a GTPase activating protein for Rap1. Proc. Natl. Acad. Sci. USA 94:12485-12490[Abstract/Free Full Text].

CHIBA, A., P. SNOW, H. KESHISHIAN, and Y. HOTTA, 1995  Fasciclin III as a synaptic target recognition molecule in Drosophila. Nature 374:166-168[Medline].

COOLEY, L., R. KELLEY, and A. SPRADLING, 1988  Insertional mutagenesis of the Drosophila genome with single P elements. Science 239:1121-1128[Abstract/Free Full Text].

CRISP, J. and J. MERRIAM, 1997  Efficiency of an F1 selection screen in a pilot two-component mutagenesis involving Drosophila melanogaster misexpression phenotypes. Dros. Inf. Serv. 80:90-92.

DEÁK, P., M. M. OMAR, R. D. SAUNDERS, M. PÁL, and O. KOMONYI et al., 1997  P-element insertion alleles of essential genes on the third chromosome of Drosophila melanogaster: correlation of physical and cytogenetic maps in chromosomal region 86E-87F. Genetics 147:1697-1722[Abstract].

DEJONG, J. L., R. MORGENSTERN, H. JORNVALL, J. W. DEPIERRE, and C. P. TU, 1988  Gene expression of rat and human microsomal glutathione S-transferases. J. Biol. Chem. 263:8430-8436[Abstract/Free Full Text].

HARIHARAN, I. K., R. W. CARTHEW, and G. M. RUBIN, 1991  The Drosophila Roughened mutation: activation of a rap homolog disrupts eye development and interferes with cell determination. Cell 67:717-722[Medline].

HARRISON, D. A., R. BINARI, T. S. NAHREINI, M. GILMAN, and N. PERRIMON, 1995  Activation of a Drosophila Janus kinase (JAK) causes hematopoietic neoplasia and developmental defects. EMBO J. 14:2857-2865[Medline].

JIMENEZ, B., V. PIZON, I. LEROSEY, F. BERANGER, and A. TAVITIAN et al., 1991  Effects of the ras-related rap2 protein on cellular proliferation. Int. J. Cancer 49:471-479[Medline].

KELNER, M. J., M. N. STOKELY, N. E. STOVALL, and M. A. MONTOYA, 1996  Structural organization of the human microsomal glutathione S-transferase gene (GST12). Genomics 36:100-103[Medline].

LEROSEY, I., P. CHARDIN, J. DE GUNZBURG, and A. TAVITIAN, 1991  The product of the rap2 gene, member of the ras superfamily. Biochemical characterization and site-directed mutagenesis. J. Biol. Chem. 266:4315-4321[Abstract/Free Full Text].

LI, Q., I. K. HARIHARAN, F. CHEN, Y. HUANG, and J. A. FISCHER, 1997  Genetic interactions with Rap1 and Ras1 reveal a second function for the Fat facets deubiquitinating enzyme in Drosophila eye development. Proc. Natl. Acad. Sci. USA 94:12515-12520[Abstract/Free Full Text].

MCGARRY, T. J. and S. LINDQUIST, 1986  Inhibition of heat shock protein synthesis by heat-inducible antisense RNA. Proc. Natl. Acad. Sci. USA 83:399-403[Abstract/Free Full Text].

MIKLOS, G. L. and G. M. RUBIN, 1996  The role of the genome project in determining gene function: insights from model organisms. Cell 86:521-529[Medline].

NICOLE, L. M. and R. M. TANGUAY, 1987  On the specificity of antisense RNA to arrest in vitro translation of mRNA coding for Drosophila hsp23. Biosci. Rep. 7:239-246[Medline].

PIZON, V., P. CHARDIN, I. LEROSEY, B. OLOFSSON, and A. TAVITIAN, 1988  Human cDNAs rap1 and rap2 homologous to the Drosophila gene Dras3 encode proteins closely related to ras in the `effector' region. Oncogene 3:201-204[Medline].

ROBERTSON, H. M., C. R. PRESTON, R. W. PHILLIS, D. M. JOHNSON-SCHLITZ, and W. K. BENZ et al., 1988  A stable genomic source of P element transposase in Drosophila melanogaster.. Genetics 118:461-470[Abstract/Free Full Text].

RØRTH, P., 1996  A modular misexpression screen in Drosophila detecting tissue-specific phenotypes. Proc. Natl. Acad. Sci. USA 93:12418-12422[Abstract/Free Full Text].

RØRTH, P., K. SZABO, A. BAILEY, T. LAVERTY, and J. REHM et al., 1998  Systematic gain-of-function genetics in Drosophila. Development 125:1049-1057[Abstract].

RUBIN, G. M. and A. C. SPRADLING, 1982  Genetic transformation of Drosophila with transposable element vectors. Science 218:348-353[Abstract/Free Full Text].

SALZ, H. K., T. W. CLINE, and P. SCHEDL, 1987  Functional changes associated with structural alterations induced by mobilization of a P element inserted in the Sex-lethal gene of Drosophila. Genetics 117:221-231[Abstract/Free Full Text].

SALZBERG, A., S. N. PROKOPENKO, Y. HE, P. TSAI, and M. PÁL et al., 1997  P-element insertion alleles of essential genes on the third chromosome of Drosophila melanogaster: mutations affecting embryonic PNS development. Genetics 147:1723-1741[Abstract].

SIDOW, A. and W. K. THOMAS, 1994  A molecular evolutionary framework for eukaryotic model organisms. Curr. Biol. 4:596-603[Medline].

SPRADLING, A. C., D. M. STERN, I. KISS, J. ROOTE, and T. LAVERTY et al., 1995  Gene disruptions using P transposable elements: an integral component of the Drosophila genome project. Proc. Natl. Acad. Sci. USA 92:10824-10830[Abstract/Free Full Text].

STAEHLING-HAMPTON, K., P. D. JACKSON, M. J. CLARK, A. H. BRAND, and F. M. HOFFMANN, 1994  Specificity of bone morphogenetic protein-related factors: cell fate and gene expression changes in Drosophila embryos induced by decapentaplegic but not 60A. Cell Growth Differ. 5:585-593[Abstract].

THUMMEL, C. S., A. M. BOULET, and H. D. LIPSHITZ, 1988  Vectors for Drosophila P-element-mediated transformation and tissue culture transfection. Gene 74:445-456[Medline].

TOWER, J., G. H. KARPEN, N. CRAIG, and A. C. SPRADLING, 1993  Preferential transposition of Drosophila P elements to nearby chromosomal sites. Genetics 133:347-359[Abstract].

XU, T. and G. M. RUBIN, 1993  Analysis of genetic mosaics in developing and adult Drosophila tissues. Development 117:1223-1237[Abstract].

This article has been cited by other articles:

				T. D. Southall, D. A. Elliott, and A. H. Brand The GAL4 System: A Versatile Toolkit for Gene Expression in Drosophila CSH Protocols, July 1, 2008; 2008(8): pdb.top49 - pdb.top49. [Abstract] [Full Text]

				M. Shindo, H. Wada, M. Kaido, M. Tateno, T. Aigaki, L. Tsuda, and S. Hayashi Dual function of Src in the maintenance of adherens junctions during tracheal epithelial morphogenesis Development, April 1, 2008; 135(7): 1355 - 1364. [Abstract] [Full Text] [PDF]

				F. Bejarano, C. M. Luque, H. Herranz, G. Sorrosal, N. Rafel, T. T. Pham, and M. Milan A Gain-of-Function Suppressor Screen for Genes Involved in Dorsal-Ventral Boundary Formation in the Drosophila Wing Genetics, January 1, 2008; 178(1): 307 - 323. [Abstract] [Full Text] [PDF]

				C. M. Miller, A. Page-McCaw, and H. T. Broihier Matrix metalloproteinases promote motor axon fasciculation in the Drosophila embryo Development, January 1, 2008; 135(1): 95 - 109. [Abstract] [Full Text] [PDF]

				S. J. Araujo, C. Cela, and M. Llimargas Tramtrack regulates different morphogenetic events during Drosophila tracheal development Development, October 15, 2007; 134(20): 3665 - 3676. [Abstract] [Full Text] [PDF]

				E. N. Mindorff, D. D. O'Keefe, A. Labbe, J. P. Yang, Y. Ou, S. Yoshikawa, and D. J. van Meyel A Gain-of-Function Screen for Genes That Influence Axon Guidance Identifies the NF-{kappa}B Protein Dorsal and Reveals a Requirement for the Kinase Pelle in Drosophila Photoreceptor Axon Targeting Genetics, August 1, 2007; 176(4): 2247 - 2263. [Abstract] [Full Text] [PDF]

				J. Kirchner, S. Gross, D. Bennett, and L. Alphey Essential, Overlapping and Redundant Roles of the Drosophila Protein Phosphatase 1{alpha} and 1{beta} Genes Genetics, May 1, 2007; 176(1): 273 - 281. [Abstract] [Full Text] [PDF]

				J. Jiang, E. Benson, N. Bausek, K. Doggett, and H. White-Cooper Tombola, a tesmin/TSO1-family protein, regulates transcriptional activation in the Drosophila male germline and physically interacts with Always early Development, April 15, 2007; 134(8): 1549 - 1559. [Abstract] [Full Text] [PDF]

				T. Sasamura, H. O. Ishikawa, N. Sasaki, S. Higashi, M. Kanai, S. Nakao, T. Ayukawa, T. Aigaki, K. Noda, E. Miyoshi, et al. The O-fucosyltransferase O-fut1 is an extracellular component that is essential for the constitutive endocytic trafficking of Notch in Drosophila Development, April 1, 2007; 134(7): 1347 - 1356. [Abstract] [Full Text] [PDF]

				N. C. Grieder, I. Charlafti, U. Kloter, H. Jackle, U. Schafer, and W. J. Gehring Misexpression Screen in Drosophila melanogaster Aiming to Reveal Novel Factors Involved in Formation of Body Parts Genetics, April 1, 2007; 175(4): 1707 - 1718. [Abstract] [Full Text] [PDF]

				N. Sasaki, T. Sasamura, H. O. Ishikawa, M. Kanai, R. Ueda, K. Saigo, and K. Matsuno Polarized exocytosis and transcytosis of Notch during its apical localization in Drosophila epithelial cells. Genes Cells, January 1, 2007; 12(1): 89 - 103. [Abstract] [Full Text] [PDF]

				C. Molnar, A. Lopez-Varea, R. Hernandez, and J. F. de Celis A Gain-of-Function Screen Identifying Genes Required for Vein Formation in the Drosophila melanogaster Wing Genetics, November 1, 2006; 174(3): 1635 - 1659. [Abstract] [Full Text] [PDF]