The Drosophila melanogaster ade5 Gene Encodes a Bifunctional Enzyme for Two Steps in the de novo Purine Synthesis Pathway

Corresponding author: Denise V. Clark, Department of Biology, Bag Service 45111, University of New Brunswick, Fredericton, New Brunswick E3B 6E1, Canada., clarkd{at}unb.ca (E-mail)

Communicating editor: V. G. FINNERTY

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Steps 6 and 7 of de novo purine synthesis are performed by 5-aminoimidazole ribonucleotide carboxylase (AIRc) and 4-[(N-succinylamino)carbonyl]-5-aminoimidazole ribonucleotide synthetase (SAICARs), respectively. In vertebrates, a single gene encodes AIRc-SAICARs with domains homologous to Escherichia coli PurE and PurC. We have isolated an AIRc-SAICARs cDNA from Drosophila melanogaster via functional complementation with an E. coli purC purine auxotroph. This cDNA encodes AIRc yet is unable to complement an E. coli purE mutant, suggesting functional differences between Drosophila and E. coli AIRc. In vertebrates, the AIRc-SAICARs gene shares a promoter region with the gene encoding phosphoribosylamidotransferase, which performs the first step in de novo purine synthesis. In Drosophila, the AIRc-SAICARs gene maps to section 11B4-14 of the X chromosome, while the phosphoribosylamidotransferase gene (Prat) maps to chromosome 3; thus, the close linkage of these two genes is not conserved in flies. Three EMS-induced X-linked adenine auxotrophic mutations, ade4¹, ade5¹, and ade5², were isolated. Two gamma-radiation-induced (ade5³ and ade5⁴) and three hybrid dysgenesis-induced (ade5⁵, ade5⁶, and ade5⁸) alleles were also isolated. Characterization of the auxotrophy and the finding that the hybrid dysgenesis-induced mutations all harbor P transposon sequences within the AIRc-SAICARs gene show that ade5 encodes AIRc-SAICARs.

THE de novo purine synthesis pathway involves 10 enzymatic steps for the synthesis of inosine monophosphate (IMP), followed by one of two paths to produce either adenosine monophosphate (AMP) or guanosine monophosphate (GMP). In bacteria, the de novo purine synthesis enzymes are encoded by separate genes, while their eukaryotic counterparts are often encoded by genes for multifunctional polypeptides (HENIKOFF 1987 ; ZALKIN and DIXON 1992 ). One example of such variation in gene organization is found with the enzymes operating at steps 6 and 7 in the de novo synthesis of purines (Fig 1), 5-aminoimidazole ribonucleotide carboxylase (AIRc) and 4-[(N-succinylamino)carbonyl]-5-aminoimidazole ribonucleotide synthetase (SAICARs). Three types of organization for these two enzymes can be found in bacteria, yeasts, and vertebrates. First, Escherichia coli AIR carboxylase and SAICAR synthetase are encoded in the operons purEK and purC, respectively (TIEDEMAN et al. 1989 , TIEDEMAN et al. 1990 ; WATANABE et al. 1989 ). Here, AIR carboxylase function requires both PurE and PurK, where PurK is required to assist PurE in the formation of the 5'-phosphoribosyl-5-aminoimidazole carboxylate (CAIR) product (see below) (MUELLER et al. 1994 ). Bacillus subtilis has a single pur operon containing 12 cistrons, including purE, purK, and purC, which encode the three proteins as found in E. coli (EBBOLE and ZALKIN 1987 ). The second type of organization has the PurE and PurK AIR carboxylase functions fused. In the yeast Saccharomyces cerevisiae, the fused PurK-PurE-homologous sequences are encoded by the ADE2 gene (STOTZ and LINDER 1990 ), with ~23 linker amino acids between the two domains (ZALKIN and DIXON 1992 ). Yeast SAICAR synthetase is encoded by the ADE1 gene (MYASNIKOV et al. 1991 ). The fusion of PurK and PurE domains is also found for the ADE6 gene of fission yeast Schizosaccharomyces pombe (SZANKASI et al. 1988 ). The third type of organization has a SAICAR synthetase domain fused with an AIR carboxylase domain to form a bifunctional enzyme encoded at a single locus (AIRc-SAICARs). This arrangement has been found for chickens (Gallus gallus; GAVALAS et al. 1993 ), humans (MINET and LACROUTE 1990 ), and rats (IWAHANA et al. 1995 ). Interestingly, in these cases the AIR carboxylase domain lacks a PurK-homologous region (ZALKIN and DIXON 1992 ). The nematode Caenorhabditis elegans also appears to have a locus encoding a bifunctional enzyme as found in the vertebrates (Genome Sequencing Center, personal communication; GenBank accession no. U39848).

View larger version (12K): In this window In a new window Download PPT slide   Figure 1. Pathway for de novo purine synthesis. Names of Drosophila genes known to correspond to pathway enzymes are shown in italics. Pathway intermediate abbreviations are: PRPP, 5'-phosphoribosyl-1-pyrophosphate; PRA, 5'-phosphoribosyl-1-amine; GAR, 5-phosphoribosylglycinamide; FGAR, 5'-phosphoribosyl-N-formylglycinamide; FGAM, 5'-phosphoribosyl-N-formylglycinamidine; AIR, 5'-phosphoribosyl-5-amino-imidazole; CAIR, 5'-phosphoribosyl-5-aminoimidazole carboxylate; SAICAR, 5'-phosphoribosyl-4-(N-succinocarboxamide)-5-aminoimidazole; AICAR, 5'-phosphoribosyl-4-carboxamide-5-aminoimidazole; FAICAR, 5'-phosphoribosyl-4-carboxamide-5-formamidoimidazole; IMP, inosine monophosphate; SAMP, adenylosuccinate; XMP, xanthosine monophosphate; AMP, adenosine monophosphate; GMP, guanosine monophosphate. Pathway enzymes are: (1) PRAT, phosphoribosylamidotransferase; (2) GARS, GAR synthetase; (3) GART, GAR transformylase; (4) FGARAT, FGAR amidotransferase; (5) AIRS, AIR synthetase; (6) AIRC, AIR carboxylase; (7) SAICARS, SAICAR synthetase; (8) and (12) adenylosuccinate lyase; (9) AICAR transformylase; (10) inosinate cyclohydrolase; (11) SAMP synthetase; (13) IMPDH, IMP dehydrogenase; (14) GMPS, GMP synthetase. Figure after HENIKOFF 1987 .

Along with these three alternate forms of gene organization, there is variation in the mechanism employed by AIR carboxylase (Fig 2). In chickens, the conversion from AIR to CAIR proceeds in one step directly from substrate to product with the use of the AIR carboxylase (PurE-like) domain (FIRESTINE and DAVISSON 1994 ). In E. coli, the conversion from AIR to CAIR proceeds in two steps via an unstable intermediate called N⁵-CAIR (N⁵-carboxyaminoimidazole ribonucleotide). This intermediate is the result of the conversion of AIR by PurK in the presence of a bicarbonate ion fueled by the hydrolysis of ATP. The N⁵-CAIR is then converted to CAIR by PurE (FIRESTINE et al. 1994 ; MUELLER et al. 1994 ). The PurK and PurE functions have fused into a single polypeptide in fungi, although it appears that they also have the bacterial two-step mechanism (FIRESTINE et al. 1998 ). Thus, there are two AIR carboxylase mechanisms with different substrate, energy, and cofactor requirements, and the E. coli and fungal de novo purine biosynthesis pathways are one step longer than for vertebrates.

View larger version (17K): In this window In a new window Download PPT slide   Figure 2. Divergent catalytic mechanisms for AIR carboxylase. (A) Chicken mechanism. (B) E. coli mechanism (FIRESTINE and DAVISSON 1994 ; FIRESTINE et al. 1994 , FIRESTINE et al. 1998 ). Abbreviations: AIR, 5'-phosphoribosylaminoimidazole; CAIR, 5'-phosphoribosyl-5-aminoimidazole carboxylate; SAICAR, 5'-phosphoribosyl 4-(N-succinocarboxamide)-5-aminoimidazole; N5-CAIR, N5-carboxyaminoimidazole ribonucleotide, ribose-P, 5'-phosphoribosyl moiety. Figure after FIRESTINE and DAVISSON 1994 .

With this difference in catalytic mechanism, the question arises as to whether different AIR carboxylases are interchangeable in vivo. In some cases this was shown to be true; a chicken cDNA and a methanobacterial gene were cloned by functional complementation of an E. coli purK mutant, yet their sequences have PurE- but no PurK-homologous sequences (HAMILTON and REEVE 1985 ; CHEN et al. 1990 ). These same sequences can also complement E. coli purE mutants. A possible explanation for why the chicken and methanobacterial AIR carboxylases can complement a purK mutant is that, rather than providing a functional version of PurK, they provide a completely different route for the production of CAIR. On the other hand, a human AIR carboxylase cDNA, cloned by functional complementation of a S. cerevisiae ade2 mutant defective in its PurE domain (MINET and LACROUTE 1990 ; SCHILD et al. 1990 ), was unable to complement a S. cerevisiae ade2 mutant defective in its PurK domain (MINET and LACROUTE 1990 ). The divergence in AIR carboxylase function has been speculated to be a reflection of the level of intracellular CO₂ in the organism, where those organisms lacking the PurK function have higher CO₂ levels (FIRESTINE et al. 1998 ).

The existence of multifunctional enzymes for de novo purine synthesis in eukaryotes may serve two related functions. First, the proximity of two enzyme domains in a single polypeptide may allow for substrate channeling and, second, enzymes can be expressed stoichiometrically (HENIKOFF 1987 ). An extension of this idea is the prediction that expression of different purine genes is somehow coordinated. Coordinated expression of purine genes has been found in microbial organisms. Extensive analysis of purine gene regulation in E. coli, B. subtilis, and S. cerevisiae has resulted in the identification of mechanisms that coordinate transcription in response to cellular concentrations of purines. In E. coli, transcriptional regulation is carried out by the PurR repressor, which represses expression of seven pur operons in response to increased purine concentrations (ZALKIN and DIXON 1992 ). In B. subtilis, a PurR repressor controls expression of the single pur operon (WENG et al. 1995 ). In S. cerevisiae, purine starvation results in the transcriptional activation of several ADE genes by the transcription factors GCN4 (ROLFES and HINNEBUSCH 1993 ), BAS1, and BAS2 (DAIGNAN-FORNIER and FINK 1992 ; DENIS et al. 1998 ).

There is little evidence for coordinated expression of purine genes in multicellular organisms; however, one case has been found for coordinated expression of the two genes encoding phosphoribosylamidotransferase (PRAT; Fig 1) and AIRc-SAICARs in vertebrates. PRAT performs the first committed step in de novo purine synthesis and is considered a rate-limiting point in the pathway (WYNGAARDEN and KELLEY 1983 ). These two genes are divergently transcribed from common promoter regions, ranging from 230 to 625 bases, in chickens, rats, and humans (GAVALAS et al. 1993 ; BRAYTON et al. 1994 ; IWAHANA et al. 1995 ). The use of bidirectional promoters occurs at other loci in eukaryotes, but this type of promoter is usually shared by evolutionarily conserved genes (GAVALAS et al. 1993 ). This purine gene arrangement is the only known instance where such a bidirectional promoter has been found between two genes that function at different points in the same pathway. Very little is known about the mechanisms for transcriptional regulation of these two genes in eukaryotes. In a recent study of the human promoter region, binding sites for nuclear respiratory factor-1 (NRF-1) and Sp1 transcription factors were identified and shown to be required in part for transcription of both genes in vitro (CHEN et al. 1997 ). However, these two transcription factors are not sufficient to explain the regulation of these genes and further studies are required to confirm their activities in vivo. While there are still many questions to be answered regarding the nature of the PRAT-AIRc-SAICARs bidirectional promoter, the unique nature of this structural feature and involvement of the rate-limiting enzyme for de novo synthesis of purines makes this a very intriguing point in the genetic regulation of the pathway.

Here we describe the isolation of a D. melanogaster AIRc-SAICARs cDNA through functional complementation of an E. coli purine auxotroph. This cDNA rescued an E. coli purC (SAICAR synthetase) mutant but was unable to rescue a purE mutant even though, as determined by sequence alignments, it contains both SAICAR synthetase and AIR carboxylase domains as found in vertebrates. This raises interesting questions about the mechanism of AIR carboxylase action in Drosophila in comparison to the chicken and bacterial mechanisms.

We have also determined the structure of the AIRc-SAICARs gene (ade5) from Drosophila and have identified potential sites for its transcriptional regulation. In addition, we show that the close linkage between the PRAT and AIRc-SAICARs genes found in vertebrates is not conserved in Drosophila. To learn more about the function of the ade5 gene in Drosophila and to identify sequences important for its expression, we have generated several purine auxotrophic and lethal alleles of this gene using EMS, X rays, and P-element mutagenesis and have linked a subset of these mutations with structural defects in the gene.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

E. coli functional complementation screen:
NK6051 ([gpt-lac]5 purEK79::Tn10 -relA1 spoT1 thi-1) and NK6056 ([gpt-lac]5 purC80::Tn10 -relA1 spoT1 thi-1) were obtained from the E. Coli Genetic Stock Center. These purine auxotrophs were grown on minimal media, described by VOGEL and BONNER 1956 as Medium E, supplemented with 0.5% glucose, 0.1% acid hydrolyzed casein, 0.2 µg/ml thiamin, 125 µg/ml tetracycline, and 100 µg/ml adenine. For functional complementation tests of Drosophila cDNAs, nonpermissive growth conditions were employed by omitting the adenine supplement.

Two D. melanogaster 0- to 24-hr embryonic cDNA libraries were generously supplied by Carl Thummel. One library was constructed using random-primed cDNA synthesis while the other was oligo(dT) primed. Both were housed in Stratagene's ZAPII expression vector, where cDNAs were directionally inserted, with the EcoRI restriction site at the 5' end and the XhoI site at the 3' end of the pBluescript SK plasmid (CARL THUMMEL, personal communication).

The ZAPII screening method was modified to facilitate the functional complementation screen being performed. The in vivo excision protocol (SHORT et al. 1988 ) was followed until the cDNAs had excised to yield Drosophila libraries in the pBluescript phagemid packaged in filamentous phage particles. The particles were then concentrated via polyethylene glycol (PEG) precipitation (SAMBROOK et al. 1989 ) from 300 CFU/ml to 2 x 10⁵ CFU/ml. The titer of the phagemid library stocks was determined using a number of hosts: SOLR [e14^- (mcrA), (mcrCB-hsdSMR-mrr) 177, sbcC, recB, recJ, umuC::Tn5 (kan^R), uvrC, lac, gyrA96, relA1, thi-1, endA1, ^R, [F', proAB, lacI^q ZM15] Su^-], XL1 Blue-MRF' {end A1, hsp R17, sup E44, thi-1, recA1, gyrA96, relA1, lac, [F', proAB, lacI^q ZM15, Tn10 (tet^R)]}, NK6051, and NK6056; and plating on 100 µg/ml ampicillin containing media for phagemid selection.

The phagemid libraries were used to infect NK6051 and NK6056, in the absence of helper phage, directly followed by platings under nonpermissive conditions to test for complementation. An AIRc-SAICARs chicken cDNA (kindly supplied by Howard Zalkin; CHEN et al. 1990 ) was used as a positive control for this procedure.

Genomic library screening, subcloning, and sequence analysis:
Genomic phage library screening, DNA isolation, restriction digests, and subcloning followed standard protocols (SAMBROOK et al. 1989 ). The D. melanogaster genomic library (MANIATIS et al. 1978 ) was obtained from the American Type Culture Collection. DNA fragment subcloning used the plasmid vector pVZ1 (HENIKOFF and EGHTEDARZADEH 1987 ). Purified plasmid DNA was obtained using QIAGEN (Valencia, CA) plasmid preparation kits. DH5 (F^-, endA1, hsdR17 (r^-_k, m⁺_k), supE44, thi-1, recA1, gyrA96, relA1, (argF-laczya)U169, 80dlacZM15, ^-) cells were made competent by standard calcium chloride method and used for transformations (SAMBROOK et al. 1989 ).

Hybridization probes were labeled using the random primer fluorescein-12-dUTP labeling kit (DuPont NEN) and detection followed the nucleic acid chemiluminescence reagent kit (DuPont NEN) protocol.

DNA sequencing was performed using the ABI 310 prism genetic analyzer (PE Applied Biosystems, Foster City, CA) with dideoxynucleotide triphosphate terminator sequencing chemistry (SANGER et al. 1977 ). In this particular method the dRhodamine terminator cycle sequencing ready reaction DNA sequencing kit (PE Applied Biosystems) was employed using the manufacturer's protocol, except that the suggested reaction volumes were halved. To facilitate the sequencing of the large genomic subclones, 5.5EL(d) and 5.5EL(r), Exonuclease III nested deletions were generated (CLARK and HENIKOFF 1994 ). SeqEd software (PE Applied Biosystems) was used to edit, align, and produce consensus sequences from raw sequence data. Sequence alignments were performed using MULTALIN software with default parameters (CORPET 1988 ).

In situ hybridization to polytene chromosomes:
Canton-S third instar larval salivary gland polytene chromosomes were hybridized (ASHBURNER 1989 ) with a probe made using the 12.6-kb insert from the LibD genomic clone as a template. The probe was labeled with biotin-conjugated dUTP [Life Technologies (Rockville, MD) Bioprime kit], and biotin detection followed the procedure described by ASHBURNER 1989 using a Detek-1-hrp kit (Enzo Diagnostics, Farmingdale, NY).

RNA methods:
Total RNA was extracted from ~1.5 g of frozen adult flies of various genotypes using TRIzol reagent (Life Technologies). RNA yields were assessed using ethidium bromide staining in agarose gels and optical density measurements from GeneQuant RNA/DNA calculator (Amersham Pharmacia Biotech, Piscataway, NJ). Polyadenylated RNA [poly(A)-RNA] was purified from 0.5 mg total RNA using the poly(AT) tract kit (Promega, Madison, WI).

For Northern blots, 2–5 µg of poly(A)-RNA was fractionated on a 1% agarose, 0.66 M formaldehyde gel (SAMBROOK et al. 1989 ) and transferred to GeneScreen Plus positively charged nylon membrane (Dupont NEN).

For primer extension, the AIRc-SAICARsPE primer was 5'-GGTTGTGGTGGTGGACATCTCGTTAATCGC-3'. This primer was radiolabeled by 5' phosphorylation with [-³²P]ATP using the primer extension kit protocol (Promega). The primer extension reaction was performed using 10 µg of total RNA from adult Canton-S flies and 100 fmol of the ³²P-labeled AIRc-SAICARsPE primer. Size markers were a ³²P-labeled X174 ladder standard (Promega) and a ³²P-labeled sequencing ladder made using the AIRc-SAICARsPE primer and p5.5EL(d) plasmid (see RESULTS) as template according to the Silver Sequencing protocol (Promega).

Drosophila strains and mutant screens:
Unless otherwise indicated, flies were cultured at 25° on "standard medium," which was Instant Drosophila medium (Carolina Biological) supplemented with Fleischmann's yeast. Canton-S was used as the wild-type strain wherever needed. Genotypes of strains carrying EMS-induced ade4 or ade5 mutations were y cv v ade4¹ f/In(1)FM6, y^31d sc⁸ dmB, y cv v ade5¹ f/In(1)FM6, y^31d sc⁸ dmB, y cv v ade5²⁽⁷⁾ f/In(1)FM6, y^31d sc⁸ dmB, y cv v ade5²⁽²⁵⁾ f/In(1)FM6, y^31d sc⁸ dmB, and y cv v H23-14(44)f/In(1)FM6, y^31d sc⁸ dmB. Strains carrying gamma-radiation-induced mutations were ade5³/In(1)FM6, y^31d sc⁸ dmB and ade5⁴/In(1)FM6, y^31d sc⁸ dmB. Strains carrying hybrid dysgenesis-induced mutations were v ade5⁵/In(1)FM6, y^31d sc⁸ dmB, v ade5⁶/In(1)FM6, y^31d sc⁸ dmB, and v ade5⁸/In(1)FM6, y^31d sc⁸.

For EMS-induced mutations, isolation of X chromosome-linked mutations was as described previously (NASH and JANCA 1983 ; JANCA et al. 1986 ). This screen was designed to locate lethal mutations on the X chromosome in the 9E1-10A11 region. At the same time, all strains carrying nonlethal mutagenized X chromosomes (y cv v f/FM6) were screened for a reduction in red eye pigmentation, and these were then tested for auxotrophy on Sang's defined medium (SANG 1956 ) under axenic culture conditions. From this screen, three mutants that responded to supplementation of the medium with adenosine were recovered (ade4¹, ade5¹, and ade5²). For nutritional supplementation experiments on ade4¹ and ade5¹, five ade/FM6 females and three to five hemizygous ade mutant males were mated on a yeast-sugar medium (NASH and BELL 1968 ) for 3 days and then transferred to a defined medium for 5 days. Additives to Sang's medium (SANG 1956 ) were all at 3.2 mM except RNA, which was at 4 mg/ml or 8 mg/ml (Table 1).

Gamma-ray mutagenesis screens were performed to isolate noncomplementing alleles of ade5¹ and, in doing so, ade5³ (C. SHERF and D. NASH, unpublished results) and ade5⁴ (S. TIONG, unpublished results) were produced.

A P-M hybrid dysgenesis screen was carried out to isolate P-element insertion-induced alleles of ade5¹. In this screen, P strain 2 males, carrying many P elements, were crossed with females of an isogenic vermilion (v) M strain. The v dysgenic male progeny, with germ-line P-element mobilization, were mated with y cv v ade5¹ f/FM6 females. The females from this cross were screened for mutations that failed to complement ade5¹. Failure of complementation was scored on the basis of "purine syndrome" characteristics (TIONG and NASH 1990 ), such as reduced red eye pigmentation and wing defects. Thus, these females had the genotype v*/y cv v ade5¹ f (where * represents the P-element-induced mutation). For each line, the mutagenized X chromosome was balanced by crossing with FM6-bearing males. Females of genotype v*/FM6 were then mated with FM6 males again to obtain a stock isogenic for the X chromosome. For each mutant, two healthy and stable lines were retained and the v* X chromosomes were then retested for failure to complement ade5¹. Three independently isolated alleles of ade5 were isolated from this screen: ade5⁵, ade5⁶, and ade5⁸.

Viability and complementation tests:
To determine the viability of each ade mutant strain on Instant Drosophila Medium (Carolina) supplemented with yeast, one of the following crosses was performed: (1) 5 adeX males x 30 adeX/FM6 virgin females or (2) 5 FM6 males x 30 adeX/FM6 virgin females, where X represents 1 of the 10 ade mutations listed above. In the first cross, female progeny were scored on the basis of the dominant Bar-eyed marker on FM6, where the ratio of adeX/adeX non-Bar-eyed females to adeX/FM6 Bar-eyed females provided a measure of viability for X chromosome homozygotes. In the second cross, male progeny were scored on the basis of Bar-eyed phenotype, where the ratio of adeX males to FM6 males provided a measure of viability for X chromosome hemizygotes.

For complementation tests between ade4¹ and various alleles of ade5, 10 ade4¹ males and 30 ade5^X/FM6 virgin females were crossed and cultured on standard media, where ade5^X refers to the ade5 alleles 1, 3, 4, 5, 6, and 8. Progeny were scored based on the Bar-eyed marker.

Amplification of AIRc-SAICARs segments in ade5 strains:
PCR using the eLONGase amplification system (Promega) was used to amplify regions of the AIRc-SAICARs gene from genomic DNA isolated from the P-element-containing ade5⁵, ade5⁶, and ade5⁸ strains and the wild-type Canton-S strain. The AIRc-SAICARs gene primers were: 5'-ATGAAGTTGCCCAGC CCCAGAATGG-3' (primer 4), 5'-CACTTGCGGGCGATGA AGGCCTTGG-3' (primer 6), 5'-AGGAACAGCGGCGGATAG GACGC-3' (primer 11), and 5'-TCGGCCATTTGGGAACT GCAGGG-3' (primer 12). The position and orientation of these four primer sequences in the gene are indicated in Fig 4. The P-element inverted repeat (PIR) primer was 5'-CGACGGGACCACCTTATGTTATTTCATCATG-3'.

View larger version (109K): In this window In a new window Download PPT slide   Figure 3. Multiple sequence alignment of AIRc-SAICARs. Standard single-letter amino acid abbreviations were employed. Identical and conservative amino acids are boxed in black and gray, respectively. The approximate division between the SAICAR synthetase and the AIR carboxylase domains is indicated with an asterisk above the Drosophila sequence. The chicken and human sequences were derived from cDNAs (CHEN et al. 1990 ; MINET and LACROUTE 1990 ). It should be noted that the E. coli and S. cerevisiae sequences used in this alignment are not continuous, but were derived from two separate sequences. For E. coli, these were purC (TIEDEMAN et al. 1990 ) and purE (TIEDEMAN et al. 1989 ; WATANABE et al. 1989 ), respectively. For S. cerevisiae, these were ADE1 (MYASNIKOV et al. 1991 ) and ADE2 (STOTZ and LINDER 1990 ), respectively. The S. cerevisiae sequences are entered on two lines, as there is a small region of overlap between the end of the ADE1 sequence (first) and the beginning of the ADE2 sequence (second). For purposes of this alignment, the PurK-homologous domain [amino acids 1~402 (STOTZ and LINDER 1990 )] was removed from the S. cerevisiae ADE2 sequence. The Drosophila sequence represented here has GenBank accession no. AF102579.

View larger version (19K): In this window In a new window Download PPT slide   Figure 4. Characterization of the AIRc-SAICARs gene. (A) Restriction map of the 12.6-kb LibD genomic DNA clone. EcoRI sites at either end are part of the Charon4A vector (MANIATIS et al. 1978 ). Hatching indicates the two EcoRI fragments that hybridized with the cDNA. E, EcoRI; B, BamHI; H, HindIII; K, KpnI; S, SstI; X, XbaI. (B) Sequence organization of the AIRc-SAICARs gene. The transcribed region lies between the vertical lines marked as 5' UTR (untranslated region) and 3' UTR. Within the transcribed region, boxes represent exons and connecting lines represent introns. Arrows below the gene map indicate primers used in PCR analysis of the P-element insertions. Triangle below the gene map points to the site of P-element insertions within the 5' UTR. The P-element inverted repeat (PIR) primer would anneal to sequences in either orientation within the region marked "P." Light gray represents the SAICAR synthetase domain; dark gray represents the AIR carboxylase domain. The sequence has GenBank accession no. AF102579.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Isolation of a Drosophila AIRc-SAICARs cDNA by functional complementation of an E. coli purC mutant:
Previously, a chicken cDNA had been isolated by complementation of E. coli purC and purEK mutants (CHEN et al. 1990 ). It therefore seemed reasonable that this would be a route for isolation of a corresponding Drosophila cDNA. Two 0- to 24-hr embryonic cDNA expression libraries were used in the functional complementation screens. These libraries were made with the ZAP vector (SHORT et al. 1988 ). Typically, with this vector, the library would be excised, infected into a standard host strain, and subsequently purified as a double-stranded plasmid for further use in a functional complementation screen. In this case, the double-stranded plasmid would have been transformed into the E. coli purine auxotrophs to assess complementation. This procedure requires a number of time-consuming manipulations that could reduce library complexity. To circumvent some of these steps, the filamentous phage generated by in vivo excision from the library were concentrated via PEG precipitation and used to directly infect the E. coli purine auxotrophs.

Following infections of purC and purEK mutants, cells were plated on restrictive media lacking a purine source. As a positive control for this novel procedure, the pZD1 phagemid containing the chicken AIRc-SAICARs cDNA (CHEN et al. 1990 ) was used to infect the same strains in parallel. We screened 2.8 x 10⁵ oligo(dT)-primed and 2.8 x 10⁵ random-primed cDNAs for rescue of purEK, and 1.5 x 10⁴ oligo(dT)-primed and 3.5 x 10⁴ random-primed cDNAs for rescue of purC. Of these, a single colony was isolated from the purC strain infected with the oligo(dT)-primed cDNAs after 96 hr of growth at 37°. This phagemid, named pAlly1, was further tested by transforming its double-stranded DNA form into both the purC strain and the purEK strain. Growth under nonpermissive conditions was observed for purC, confirming that pAlly1 is capable of complementing the purine auxotrophy of the purC strain. However, pAlly1 failed to rescue the purEK defect. Therefore, pAlly1 supplied E. coli with SAICAR synthetase function but could not rescue the purEK AIR carboxylase function.

The 1.8-kb cDNA sequence from pAlly1 (GenBank accession no. AF102579) was conceptually translated and aligned to AIRc-SAICARs sequences from other organisms (Fig 3). Despite its inability to complement the purEK mutation, it appears to contain the entire coding region for both AIR carboxylase and SAICAR synthetase. Overall, the Drosophila cDNA encodes a protein more closely related to the vertebrate sequences (57 and 58% amino acids identical with the human and chicken proteins, respectively) than those for E. coli or S. cerevisiae (25 and 23% amino acids identical, respectively). On the basis of the alignment, the fly cDNA appears to contain the entire coding region for both AIR carboxylase and SAICAR synthetase. The first 262 amino acids correspond to the SAICAR synthetase domain, which has 58 and 28% amino acids identical with the chicken domain and E. coli PurC, respectively. Amino acids 263–429 correspond to the AIR carboxylase domain, which has 58 and 23% amino acids identical with the chicken domain and E. coli purE, respectively. The two domains are defined here based on mutational and sequence alignment analyses of a chicken AIRc-SAICARs cDNA (CHEN et al. 1990 ).

Isolation and structural characterization of the AIRc-SAICARs gene from Drosophila:
The pAlly1 cDNA fragment was used as a probe to screen a D. melanogaster genomic DNA library (MANIATIS et al. 1978 ) housed in the vector, Charon 4A. Approximately 3.6 x 10⁵ phage were screened and, of these, one clone called LibD showed intense and reproducible hybridization when hybridization was carried out at high stringency. DNA isolated from the LibD clone was mapped using several restriction enzymes and the location of the hybridization signal was narrowed down to 5.5- and 1.3-kb EcoRI fragments (Fig 4A). The 5.5-kb EcoRI fragment from LibD was subcloned in both orientations into the plasmid vector pVZ1 (HENIKOFF and EGHTEDARZADEH 1987 ) to produce p5.5EL(d) and p5.5EL(r). A series of nested deletions on both p5.5EL(d) and p5.5EL(r) was generated and the complete sequence for 4.69 kb of both strands was produced (GenBank accession no. AF102579). In addition, the sequence of a single strand was determined for 0.71 kb at the 3' end of the 5.5-kb EcoRI LibD subclone and 1.18 kb beyond the 5' end of this subclone. Thus, in total, a region that spans 6.58 kb of the LibD clone was sequenced.

On the basis of sequence comparisons between the cDNA and the genomic sequence, several structural features of the AIRc-SAICARs gene have been identified (Fig 4B). There are four introns of varying length, 447, 100, 427, and 68 bases, respectively, all of which are contained in the SAICAR synthetase domain. In addition, a polyadenylation consensus sequence, AAUAAA (SHEETS et al. 1990 ), is located 19 bases prior to the poly(A) addition site in the cDNA.

We detected a single AIRc-SAICARs mRNA of ~1.9 kb in wild-type adult polyadenylated mRNA (Fig 5A). This corresponds to the length of the AIRc-SAICARs cDNA. To identify the transcription initiation site(s) for AIRc-SAICARs, primer extension mapping was performed using the AIRc-SAICARsPE primer and total RNA from adult Canton-S flies as the template for extension (Fig 5B). Two transcription initiation sites were detected; the major site occurs 243 bases upstream and a minor site occurs 220 bases upstream of the initiator methionine. In addition, the AIRc-SAICARs cDNA 5' end lies in this vicinity as well, at 226 bases upstream of the initiator methionine. The cDNA 5' end sequence is identical to the genomic DNA sequence in this region, indicating there are no introns in the 5' UTR of this gene. To date, 14 AIRc-SAICARs cDNA 5' ends have been sequenced as a part of the Berkeley Drosophila Genome Project Expressed Sequence Tag (EST) Database. The 5' ends of these cDNAs fall within a 25-base interval from 201 to 225 bases upstream from the initiator methionine (not shown). These sites are all in the vicinity of the two transcription initiation sites determined by primer extension mapping for AIRc-SAICARs, but none extends back to the major initiation site.

View larger version (47K): In this window In a new window Download PPT slide   Figure 5. Transcription of AIRc-SAICARs. (A) Northern blot of 1 µg of Canton-S adult polyadenylated RNA. (B) Primer extension identifies major and minor transcription initiation sites. A sequencing reaction primed with the same primer used for primer extension is shown as a marker. A and B indicate 5' ends within a few nucleotides of resolution. Initiator sequence downstream from the major start site is boxed. Asterisk indicates the 5' end of the cDNA isolated by functional complementation.

A search for promoter consensus sequences was performed for the AIRc-SAICARs gene. The two most common promoter element sequences in Drosophila are the TATA-box, found in approximately half of Drosophila promoters, and the initiator element (Inr), found in about one-third of Drosophila promoters. The TATA-box consensus sequence in Drosophila is ^A/G/_C TAT AAA ^{G/A G/C} and it is typically found at the -25 to -30 position with respect to the transcription initiation site (ARKHIPOVA 1995 ). The AIRc-SAICARs gene has the sequence CCATAAAGT, beginning at position -23 with respect to its major transcription initiation site. This sequence matches seven out of nine bases in the TATA-box consensus sequence. The Inr consensus sequence in Drosophila is ^T/A/_G TCA ^G/_T T ^T/_C G and is typically found in the -10 to +10 interval with respect to transcription initiation. The AIRc-SAICARs gene has the sequence CTCAGTTG, beginning at position +3 with respect to the major transcription initiation site, which matches the Inr consensus sequence at seven of eight positions. Given the locations of these two sequences and their similarities to the consensus sequences, they are candidates for a TATA-box and initiator element in the AIRc-SAICARs gene.

Localization of the AIRc-SAICARs gene:
The AIRc-SAICARs gene was localized within the D. melanogaster genome by in situ hybridization of a biotinylated DNA probe to Canton-S polytene chromosome squashes. The AIRc-SAICARs gene gave a single site of hybridization on the X chromosome at 11B4-14. The consistently diffuse nature of this region of the X chromosome (LEFEVRE 1976 ) did not allow us to localize the hybridization signal further.

The ade4 and ade5 genes colocalize with the AIRc-SAICARs gene:
The ade4 and ade5 genes were originally identified in a screen for EMS-induced X chromosome-linked recessive mutations associated with reduced red eye pigmentation and adenine auxotrophy. Recombination mapping placed both ade4¹ and ade5¹ in the v-f interval (D. NASH, unpublished data), corresponding to sections 10 through 15 on the cytogenetic map (LEFEVRE 1976 ). Colocalization of ade4¹ and ade5¹ with the AIRc-SAICARs gene on polytene chromosomes made one or both of these mutations candidate alleles of the gene.

The ade5² allele was isolated in the same screen, but the initial strain (H23-14) was only partially responsive to an RNA dietary supplement (data not shown). It appeared that there were two classes of flies in this strain, one carrying a mutation with reduced red eye pigmentation and adenine auxotrophy and the other carrying a mutation with normal red eye pigmentation and auxotrophy where only slowly developing survivors occur on an RNA supplement. Two ade5² stocks, ade5²⁽⁷⁾ and ade5²⁽²⁵⁾, were derived from the original H23-14 strain, and these both failed to complement ade5¹ and have both reduced red eye pigmentation and adenine auxotrophy (data not shown). These two strains most likely carry the same mutation. The nature of the other mutation in the original H23-14 stock, which was subsequently isolated in a stock called H23-14(44), has not been resolved.

The two gamma-radiation-induced ade5 alleles, ade5³ and ade5⁴, and the three hybrid dysgenesis-induced alleles, ade5⁵, ade5⁶, and ade5⁸, were all isolated as mutations that failed to complement ade5¹, as indicated by a purineless phenotype that had been associated with mutations in other de novo purine synthesis genes (TIONG et al. 1989 ; TIONG and NASH 1990 ; CLARK 1994 ). This phenotype is seen in the presence of standard Drosophila medium where there is an exogenous supply of adenine in the form of yeast or yeast extract and features adults with reduced pteridine (red) eye pigments and wing and leg defects. The null phenotype is pupal lethality.

Nutritional supplementation of ade4¹ and ade5¹ auxotrophic mutants:
Supplementation of Sang's defined medium (SANG 1956 ) with various nutrients was done to further characterize the auxotrophic phenotype of ade4¹ and ade5¹ (Table 1). Only the data for females could be considered here since, for unknown reasons, the FM6-bearing males were only responsive to the yeast-sugar medium. Both ade4¹ and ade5¹ responded positively to supplementation with adenine, adenosine, and RNA, as has been found for auxotrophic mutations in ade2 and ade3 (NASH and HENDERSON 1982 ), which affect steps in the pathway prior to IMP. Likewise, ade4¹ and ade5¹ did not respond to supplementation with cytidine, guanine, uridine, or hypoxanthine. The data for supplementation with guanosine were equivocal. Since guanosine is not converted to adenine nucleotides in flies (NASH and HENDERSON 1982 ), guanosine must somehow act indirectly, perhaps through a mechanism for balancing nucleotide pools. Such indirect action may be unusually sensitive to experimental variation and this could explain these results.

The above results suggest that both ade4¹ and ade5¹ affect de novo purine synthesis at steps prior to IMP. The AICA and AICAR supplementation data help to localize the point in IMP synthesis that is affected in ade4¹ and ade5¹. Both mutants responded to supplementation with AICAR. AICAR is synthesized from SAICAR by adenylosuccinate lyase, which performs the eighth step in the pathway (Fig 1). AICA also rescues the mutants, consistent with the observation that AICA can be converted to AICAR by adenine phosphoribosyltransferase (APRT; THOMAS et al. 1973 ). These data place the function affected in these mutants at or before the eighth step in the IMP synthesis pathway.

Complementation test crosses were performed reciprocally between ade4¹ and ade5¹ on both Sang's medium (SANG 1956 ) and yeast-sugar medium (NASH and BELL 1968 ). The data indicate that the mutations complement with respect to the auxotrophic phenotype when grown on Sang's medium (Table 2). Thus, although ade4¹ and ade5¹ are adenine auxotrophic mutations mapping to the same region of the genome, it appears that they affect different functions in IMP synthesis. The ade5² alleles fail to complement ade5¹ with respect to the auxotrophic phenotype (D. NASH, unpublished data).

Complementation tests between ade4 and ade5 alleles:
The two gamma-radiation-induced alleles, ade5³ and ade5⁴, and the three hybrid dysgenesis-induced alleles, ade5⁵, ade5⁶, and ade5⁸, were isolated as mutations that failed to complement ade5¹, giving a purine syndrome phenotype on standard medium. Although ade4¹ and ade5¹ complement, they are mutations with similar phenotypes and map positions and thus could be complementing alleles of the same gene. Therefore, the gamma-radiation- and hybrid dysgenesis-induced ade5 alleles were also tested for complementation with ade4¹. In Table 3, the proportion of ade4¹/ade4¹ females to total females represents the viability of the ade4¹ chromosome on standard medium. If a cross produces a higher proportion than for ade4¹, then it shows improved viability and, therefore, the mutation must at least partially complement ade4¹. If a cross produces the same or a lower proportion than for ade4¹, then it shows a decrease in viability and, therefore, the mutation fails to complement ade4¹. Again, in comparing relative viabilities, the assumption is there is no secondary recessive mutation on the ade4¹ chromosome that reduces viability.

From Table 3, it can be seen that all of the alleles tested except ade5³ complement ade4¹. One possible explanation for this result is that intragenic complementation is occurring, where ade4¹ affects the enzymatic function of either the AIRc or SAICARs domain but not both, while the ade5 alleles that complement it affect the other domain. ade5³ would then represent a mutation that affects both the AIRc and SAICARs domains and, therefore, fails to complement both ade4¹ and ade5¹. An alternative explanation is that, since ade5³ is a gamma-radiation-induced mutation, it could be a double mutant or deficiency affecting both ade5 and ade4. Thus, ade4 and ade5 may still represent two separate genes.

Viability of ade4 and ade5 mutants:
To estimate viability of each ade4 and ade5 allele on standard medium, mutant homozygotes and/or hemizygotes were scored relative to siblings carrying the balancer FM6. The parental chromosomes for the ade4 and ade5 mutants were not available for this experiment; therefore, relative viabilities for the mutant alleles were examined rather than their viabilities with respect to the parental chromosome. We recognize that the reduced viability of any of the ade4 and ade5 homozygotes could be in part the result of a secondary recessive mutation on the X chromosome.

The chromosomes carrying the two gamma-radiation-induced alleles, ade5³ and ade5⁴, are completely recessive lethal. On the basis of relative survival of ade/ade and ade/FM6 siblings (data not shown), the following mutant chromosomes can be placed in order from least to most viable: ade5¹, ade4¹, ade5²⁽⁷⁾, [ade5⁵, ade5⁶, ade5⁸]. Thus, in general, the chromosomes carrying EMSinduced mutations are associated with lower viability than those carrying hybrid-dysgenesis-induced mutations.

Polymorphisms in the AIRc-SAICARs gene are associated with the three hybrid dysgenesis-induced ade5 alleles:
To search for a link between ade5 and the AIRc-SAICARs gene, we searched for restriction fragment polymorphisms in the hybrid dysgenesis-induced ade5 strains using the 5.5-kb EcoRI fragment (Fig 4A) as a probe for Southern blots. All three of the hybrid dysgenesis-induced alleles (ade5⁵, ade5⁶, and ade5⁸) have larger AIRc-SAICARs EcoRI and BamHI restriction fragments in comparison to wild type (only BamHI data are shown in Fig 6A).

View larger version (51K): In this window In a new window Download PPT slide   Figure 6. Analysis of the AIRc-SAICARs gene in hybrid dysgenesis-induced ade5 mutants. (A) Southern blot of genomic DNA isolated from females homozygous for ade5 alleles, and Canton-S, and digested with BamHI. Blot was probed with a 5.5-kb EcoRI LibD fragment that contains the AIRc-SAICARs gene (Fig 4A). AIRc-SAICARs-hybridizing bands that are altered between fly strains are indicated with an arrowhead. Fragment sizes are indicated in kilobases. (B) PCR amplification of the AIRc-SAICARs region of the three ade5 hybrid dysgenesis-induced mutations. Genomic DNA of ade55, ade56, and ade58 homozygous and Canton-S females was amplified using primers 11 and 4 (see Fig 4B). DNA size standard is the Life Technologies 1-kb ladder. (C) Wild-type DNA sequence of the region around the P-element insertions for ade56 and ade58, showing the site of insertion (arrow) and the sequence duplicated in the mutants (boxed and to the right of the arrow). The site of insertion for ade55 is the same as for ade56 but the extent of the duplication is unknown (see text).

P elements vary in size, and the complete P element is typically 2.9 kb in length. Thus, one would expect a P-element insertion in the AIRc-SAICARs region to produce an increase of up to 2.9 kb in the size of a restriction fragment. This appears to be the case for the ade5⁵, ade5⁶, and ade5⁸ alleles, as substantial increases in the size of AIRc-SAICARs restriction fragments with respect to Canton-S are observed for all three alleles. However, the increases in restriction fragment size are not always consistent for the two restriction enzymes, indicating that rearrangements or new restriction sites were introduced or deleted in the region for some of the alleles. The wild-type EcoRI fragment hybridizing to the probe is 7.4 kb while the major wild-type BamHI fragment hybridizing to the probe is ~4.8 kb. The ade5⁵ allele is associated with a 9.9-kb EcoRI fragment (2.5 kb longer) and a 8.7-kb BamHI fragment (3.9 kb longer). This inconsistency is due either to a rearrangement, gain of an EcoRI site, or loss of a BamHI site. The ade5⁶ allele is associated with a 7.6-kb EcoRI fragment (0.2 kb longer) and a 6.4-kb BamHI fragment (1.6 kb longer). A full-length 2.9-kb P element has an EcoRI site but no BamHI site. Again, this is an inconsistency that could be explained in the same way as for ade5⁵. The ade5⁸ allele is associated with an 8.6-kb EcoRI fragment (1.2 kb longer) and a 5.9-kb BamHI fragment (1.1 kb longer). Within the resolution of this Southern blot, ade5⁸ appears to have a partial P-element insert that did not create new restriction sites.

The idea that the AIRc-SAICARs polymorphism observed for ade5⁵, ade5⁶, and ade5⁸ is the result of P-element insertions is supported by the fact that, when the Southern blot was stripped and reprobed with a P-element sequence, the same polymorphic AIRc-SAICARs EcoRI and BamHI fragments were detected (not shown). Taken together, these polymorphisms are a strong link between the ade5 mutations and the AIRc-SAICARs gene.

Localization of P-element insertions in the 5' untranslated region of the AIRc-SAICARs gene:
A PCR-based approach was taken to further localize P-element insertions in the ade5⁵, ade5⁶, and ade5⁸ alleles. Primers 11 and 4 (Fig 4B) were first used to amplify the entire AIRc-SAICARs region from the three P-element ade5 mutants and wild-type flies. The wild-type 4-kb AIRc-SAICARs PCR product was increased to 5.8 and 5.4 kb for ade5⁶ and ade5⁸, respectively (Fig 6B). Several attempts were made to generate this PCR product from ade5⁵ genomic DNA without success. Thus, the ade5⁶ and ade5⁸ appeared to have P-element insertions of 1.8 and 1.4 kb, respectively. Since ade5⁵ would not amplify with these primers, we suspect there is either a rearrangement or deletion of the region surrounding one of the primer binding sites. Alternatively, based on the Southern blot analysis, the predicted PCR product would be at least 6.5 kb, and so it may not have been generated due to inefficient amplification of a larger product.

Two other primer pairs were used in PCRs to localize the P element in the 4-kb region between primer sites 4 and 11. PCR with the PIR primer and primer 4 gave a 2.3-kb product for ade5⁵, ade5⁶, and ade5⁸, indicating all three alleles had P-element sequences in the same region. PCR with primers 12 and 6 generated 3.0, 2.6, and 1.2 kb products for ade5⁶, ade5⁸, and Canton-S, respectively (not shown). For ade5⁶ and ade5⁸, these sizes are consistent with those generated using primers 11 and 4, indicating single P-element insertions of 1.8 and 1.4 kb in this 1.2-kb interval. As found with primers 11 and 4, the PCR results with primers 12 and 6 were inconclusive for ade5⁵.

Sequencing of the PCR products across the insertion sites shows that all three alleles have P-element insertions that lie very close together in the 5' untranslated region of the AIRc-SAICARs gene (Fig 6C). For ade5⁵ and ade5⁶, the insertions are 63 bp downstream from the major transcription start site. For ade5⁸, the insertion is 75 bp downstream. P-element insertions typically generate an 8-bp duplication (O'HARE and RUBIN 1983 ). Both ade5⁶ and ade5⁸ P-element insertion sequences have the typical 8-bp duplication. With only sequence data for the 3' side of the ade5⁵ insertion, the size of the duplication could not be determined.

Thus, for ade5⁶, there appears to be a 1.8-kb P element in the 5' untranslated region (UTR) of AIRc-SAICARs. The inconsistency between the EcoRI restriction fragment size (only 0.2 kb longer than wild type) might be explained by the addition of an EcoRI site in the P element; however, we did not determine the sequence far beyond the P-element inverted repeats. For ade5⁸, the data are consistent with a 1.4-kb P-element insertion in the 5' UTR of AIRc-SAICARs.

Altered AIRc-SAICARs mRNAs in ade5 mutants:
With the structure of these P-element insertions in mind, we examined polyadenylated mRNA isolated from ade5⁵, ade5⁶, ade5⁸, and Canton-S females in a Northern blot using the 1.6-kb BamHI AIRc-SAICARs cDNA fragment as a probe (Fig 7). The most notable result is that ade5⁸ produces a major transcript that is ~1 kb longer than the wild-type transcript at 3 kb. The discrepancy between the transcript length and the length one would predict from the P-element insertion (3.3 kb) is likely due to limitations of gel resolution. Alternatively, there could be an additional or aberrant splicing event or a different transcription initiation site from within the P element. In any case, the reading frame in this mRNA may be altered. The ade5⁵ transcript is wild type in length, but may not be present at wild-type levels, taking into account the RNA loading differences between ade5⁵ and Canton-S RNA (Fig 7B). No transcript is detectable in the ade5⁶ females. Thus, the analysis of the AIRc-SAICARs transcripts from ade5⁵, ade5⁶, and particularly ade5⁸ mutants provides evidence for the link between the ade5 gene first identified in a purine auxotrophy screen and the AIRc-SAICARs gene. The viability of flies carrying these alleles suggests that there must be some low level of AIRc-SAICARs produced from these alleles with P-element insertions in the 5' UTR rather than in the open reading frame.

View larger version (49K): In this window In a new window Download PPT slide   Figure 7. AIRc-SAICARs mRNA in ade55, ade56, and ade58 mutants. (A) A Northern blot of polyadenylated RNA isolated from a mixed population of adult males and females probed with the AIRc-SAICARs cDNA. One lane of the blot image was omitted between the ade58 and Canton-S lanes. (B) Ethidium bromide staining of the gel prior to blotting.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

A Drosophila cDNA encoding AIRc-SAICARs was cloned by functional complementation of an E. coli purC (SAICARs) adenine auxotroph. Using the same assay, this cDNA was unable to complement an E. coli purE mutant. Sequence alignments with AIRc-SAICARs from other organisms showed that this cDNA contains the entire coding region of both SAICARs and AIRc. Since the chicken cDNA can complement the purE mutant (CHEN et al. 1990 ), there is a possibility that there are functional differences between the chicken and Drosophila AIR carboxylases.

The AIRc enzymes from E. coli and chickens operate by two distinct mechanisms (FIRESTINE et al. 1994 ). The Drosophila AIRc domain has 58% amino acid identity with the chicken AIRc and only 24% amino acid identity with E. coli PurC. Therefore, based on amino acid identities, it is likely that the Drosophila AIRc enzyme functions in a similar fashion to the chicken AIRc. It has been proposed that the chicken AIRc is capable of substituting for both PurE and PurK because it offers a completely separate path to go from the AIR substrate to the CAIR product (CHEN et al. 1990 ; FIRESTINE and DAVISSON 1994 ; MUELLER et al. 1994 ). If this is the case, then it is not clear why the Drosophila AIRc does not provide AIRc function in the E. coli purEK mutant. There are a number of possible explanations for this discrepancy. First, the Drosophila AIRc could function by the same mechanism as E. coli PurE and, as such, requires a functional version of PurK. However, our BLAST sequence searches with E. coli PurK have not identified any PurK-like homology in Drosophila or vertebrates, so there is no support for this explanation. Second, the Drosophila cDNA could contain a mutation in the AIRc domain. This possibility was ruled out when the exon sequences obtained for the AIRc-SAICARs gene were found to align exactly with the cDNA sequence. As there are several Drosophila AIRc-SAICARs cDNAs available from the Berkeley Drosophila Genome Project, we could test these cDNAs for consistency with the functional complementation behavior of our cDNA. A third explanation is that the AIRc mechanism is identical in Drosophila and chickens but, for some reason, the Drosophila enzyme may not be as efficient. Perhaps a post-translational modification unavailable in E. coli is required for Drosophila AIRc to function.

Transcription of the AIRc-SAICARs and the PRAT genes in vertebrates is divergent and controlled by common promoter elements in the intergenic region (BRAYTON et al. 1994 ; GAVALAS and ZALKIN 1995 ). We found that the AIRc-SAICARs gene maps to the X chromosome in section 11B 4-14. On the other hand, Prat maps to the right arm of chromosome 3 in section 84E1-2 (CLARK 1994 ). A second Prat gene maps to the left arm of chromosome 3 (D. V. CLARK, unpublished observations; D. HARVEY, L. HONG, M. EVANS-HOLM, J. PENDLETON, C. SU, P BROKSTEIN, S. LEWIS and G. M. RUBIN, unpublished results). Thus, although the AIRc-SAICARs and Prat genes in Drosophila are not closely linked and divergently transcribed, this finding does not rule out the possibility that they have similar regulatory elements and their transcription is somehow coordinated.

The AIRc-SAICARs gene promoter region was identified by primer extension of adult RNA. A TATA-box and initiator (Inr) were found at typical positions relative to the major transcription start site, between -25 to -30 and -10 to +10, respectively (ARKHIPOVA 1995 ). A second, minor transcription start site was also identified; however, no TATA-box or Inr consensus sequences were located in proximity to this initiation site. In contrast to the AIRc-SAICARs gene, Prat has a promoter lacking both TATA and Inr sequences with multiple transcription start sites (CLARK et al. 1998 ). There are no obvious sequences shared between the Prat and AIRc-SAICARs gene promoter regions. Further functional studies on their regulatory sequences and the expression patterns must be done to resolve this issue.

Two complementing EMS-induced adenine auxotrophic mutations, ade4¹ and ade5¹, mapped to the same region of the Drosophila genome as AIRc-SAICARs. The EMS-induced alleles have the nutritional supplementation behavior typical of de novo purine synthesis genes. Furthermore, the mutants are rescued by compounds (AICA and AICAR) that place the defect at a point at or prior to step 8 in the pathway, which is consistent with a defect in AIRc-SAICARs at steps 6 and 7 in the pathway.

Two gamma-radiation-induced and three hybrid dysgenesis-induced alleles of ade5 were isolated in subsequent screens. The P-element insertion alleles were isolated as independent noncomplementers of the EMS-induced ade5¹ allele on the basis of the "purine syndrome" phenotype. In previous work, the first alleles of ade2 and ade3 were isolated in the same adenine auxotrophy screens (NASH and JANCA 1983 ; JANCA et al. 1986 ). More severe alleles were subsequently isolated as lethals or semilethals; these alleles show a purine-syndrome phenotype when flies are grown on standard medium (TIONG et al. 1989 ; TIONG and NASH 1990 ). The phenotypes of various ade2 and ade3 alleles served as the basis for the design of the gamma-radiation and hybrid dysgenesis screens for ade5 alleles: new alleles would display a purine-syndrome phenotype in combination with ade5¹ on standard medium.

Characterization of the lesions harbored in the independently isolated hybrid dysgenesis-induced ade5 alleles indicates that the AIRc-SAICARs and ade5 gene are synonymous. Evidence supporting this conclusion is threefold. First, Southern blot and PCR analyses of the ade5⁵, ade5⁶, and ade5⁸ hybrid dysgenesis-induced mutants showed polymorphisms associated with AIRc-SAICARs sequences in all three mutants. Second, sequencing of the regions flanking P-element inserts in ade5⁵, ade5⁶, and ade5⁸ localized P elements to the 5' end of the transcribed but untranslated region of the AIRc-SAICARs gene. The ade5⁶ and ade5⁸ P elements are inserted at different sites in this region. Third, Northern blot analysis of the AIRc-SAICARs transcripts produced from the three mutants indicates a shift in the size of transcript produced by one of the alleles, ade5⁸. Reductions in mRNA levels appeared to occur for the other two alleles, but these were not quantified. The position of the P element within the transcribed region, coupled with the increased transcript size for ade5⁸, indicate that the P element in this strain is probably being transcribed.

Although all three hybrid dysgenesis-induced alleles have P-element sequences inserted in the 5' untranslated region, the insertions are different sizes. The size variation could be due to insertion of different nonautonomous P elements already harboring deletions or due to insertion of a full-length P element followed by imprecise excision in subsequent generations before the mutation was backcrossed sufficiently to remove all sources of P-element transposase. This variation in insert size and sequence could explain the different behavior of the three alleles. For ade5⁸, its longer 5' end could interfere with the initiation of translation of AIRc-SAICARs or could result in initiation and termination of translation within the P-element sequence itself. Alternatively, initiation and translation through the P-element sequence into the AIRc-SAICARs sequence would result in additional amino acids or a reading frame shift. Any of these scenarios could impede the AIRc-SAICARs enzyme function, producing the mutant phenotype observed in ade5⁸. The absence of detectable AIRc-SAICARs mRNA in ade5⁶ could be due to reduced transcription initiation as a result of the P-element insertion, or the presence of the P element may reduce mRNA stability.

On the basis of our nutritional studies, gene mapping, and molecular characterization of the hybrid dysgenesis-induced alleles, we conclude that the AIRc-SAICARs corresponds to the ade5 gene. Further evidence could be provided by showing the ade5 mutations could be rescued by transformation with a wild-type AIRc-SAICARs gene. The 5.5-kb EcoRI fragment from the LibD clone would be an ideal starting point for this type of experiment, since it contains the entire coding region of the gene and a considerable amount of flanking sequence.

As for the ade4¹ mutation, further study is required to determine if it represents a separate gene or is a complementing allele of ade5. The latter scenario is reasonable, considering that AIRc-SAICARs is a bifunctional enzyme with independent domains (FIRESTINE and DAVISSON 1994 ). The gamma-radiation-induced allele ade5³ fails to complement ade4¹ completely; no ade4¹/ade5³ females were observed when cultured on standard medium. This result is unexpected, since males hemizygous for ade4¹ are viable on standard medium. Perhaps this observation can be explained by a dosage compensation effect, where ade5³ is a null allele and females hemizygous for ade4¹ have less transcription of this gene relative to hemizygous males. This finding is in contrast to the fact that ade5³ was isolated as a noncomplementing allele of ade5¹, where this heteroallelic combination is viable. Since ade4 and ade5 map in the same region of the X chromosome, and ade5³ is a gamma-radiation-induced mutation, it may be a deletion that encompasses both ade4 and ade5. Further characterization of ade5³ may help to resolve the relationship between ade4 and ade5.

If ade4¹ is not an allele of ade5, the question becomes, what gene is it? Based on its behavior in the nutritional supplementation experiments, it is a good candidate for another de novo purine biosynthesis gene that operates at or prior to the eighth step in the pathway. Genes encoding steps 1 (Prat); steps 2, 3, and 5 (ade3); step 4 (ade2); and now steps 6 and 7 (ade5) have been identified. Since ade4¹ can be rescued with AICAR, it may be a mutation in the gene encoding the enzyme responsible for step 8 in the pathway, adenylosuccinate lyase, converting SAICAR to AICAR with the release of fumarate. One caveat to this interpretation is that adenylosuccinate lyase also catalyzes a very similar reaction in step 12 (Fig 1), converting SAMP to AMP with the release of fumarate (HENIKOFF 1987 ). If ade4¹ corresponds to the gene encoding this enzyme, then the prediction is that it should not be supplementable with inosine or any intermediate prior to step 12, including AICAR (NASH and HENDERSON 1982 ). cDNAs encoding this enzyme as well as those for the other uncharacterized genes in the pathway are now available through the Berkeley Drosophila Genome Project (HARVEY et al. 1997). Thus, it will be straightforward to clarify this issue using these cDNAs as probes for in situ hybridization to polytene chromosomes to localize the remaining purine pathway genes in the Drosophila genome.

	FOOTNOTES

¹ Present address: Department of Biochemistry and Molecular Biology, Faculty of Medicine, Dalhousie University, Halifax, Nova Scotia B3H 4H7, Canada.
² Present address: Department of Genetics, Exelixis Pharmaceuticals, Inc., South San Francisco, CA 94080-2770.

	ACKNOWLEDGMENTS

The authors thank the Genome Sequencing Center, Washington University, St. Louis, for communication of DNA sequence data prior to publication.