Isolation and characterization of the glucose-6-phosphate dehydrogenase gene of Drosophila melanogaster.

To investigate the molecular basis of dosage compensation in Drosophila, a recombinant lambda phage containing the Drosophila melanogaster glucose-6-phosphatase dehydrogenase (G6PD) gene was isolated by differential screening of a Drosophila genomic lambda library with poly(A)+RNA obtained from polyribosomes enriched for or depleted of G6PD mRNA sequences. Of 44 000 plaques screened, a single phage, lambda DmG21, showed hybridization with the enriched poly(A)+RNA but not the depleted one. Confirmation that the Drosophila DNA fragment cloned in lambda DmG21 contains the G6PD gene sequence is based on the following observations. lambda DmG21 DNA shows hybridization only to the 18D region of the salivary gland X-chromosome, which is the known cytological locus for the G6PD gene. In vitro translation of the poly(A)+mRNA selected by hybridization to lambda DmG21 DNA sequences shows a polypeptide product of apparent Mr 55 000, identical to that of the monomeric unit of G6PD. When the putative coding sequence of G6PD is cloned into the expression vector lambda gt11, recombinant plaques are recognized by anti-G6PD immunoglobulin. A transcriptional map of the G6PD gene shows that it is divided into two exons, 0.9 kb (exon I) and 1.8 kb (exon II) long, which are separated by a 2.4-kb intron. The G6PD mRNA is 2.0 kb in length and the steady-state level of the mRNA is similar in both sexes. Measurement of the copy number of the G6PD gene in males and females shows the gene to be present once per X-chromosome in both sexes. No amplification of the gene sequence was observed in males. These results are, therefore, in agreement with the previous suggestion that dosage compensation is the result of enhanced transcription of X-linked genes in males.


INTRODUCIION
Glucosed-phosphate dehydrogenase (G6PD; EC 1.1.1.49) is widely distributed in the animal kingdom and catalyzes the first reaction in the pentose phosphate pathway. In D. melanogaster, the gene encoding G6PD has been genetically mapped to the Xchromosome at position l-63.6 and is designated Zw (Young et al., 1963). Cytogenetic analysis has further localized Zw witbin the cytological interval proximal to 18Dl-2 but distal to 18F (Stewart and Merriam, 1974).
G6PD is one of a small number of X-linked enzymes that have been extensively utilized for investigation of dosage compensation at the level of enzyme synthesis (for review, see Lucchesi, 1977;Stewart and Merriam, 1980) Dosage compensation is the equalization of X-linked gene products in males and females (Muller et al., 1931). Unlike mammals, where equalization of X-linked gene products occurs by X-chromosomal inactivation (Lyon, 1961) both X-chromosomes remain active in female Drosophila (Kazazian et al., 1965). The equalization of X-linked gene products in male Drosophila, which have one dose of X-linked genes and female Drosophila, which have two doses of such genes, is believed to be mediated by enhancing the level of transcription of X-linked genes in males (Mukherjee and Beerman, 1965;Korge 1970;Hohnquist, 1972). Although these studies suggest that the steady-state level of mRNA transcripts from dosage compensated Xlinked genes would be equivalent in males and females, no direct evidence has been published.
Recent studies have shown that the autosomallinked xanthine dehydrogenase gene (Spradling and Rubin, 1983) and dopadecarboxylase gene (Scholnick et al., 1983) show partial dosage compensation when inserted into the X-chromosome via P-element-mediated transformation Rubin and Spradling, 1982). Conversely, seven cases have been observed in which the Xlinked white gene was inserted into autosomal sites, and in each case the white gene was partially dosagecompensated (Hazehigg et al., 1984). These observations have led to the suggestion that c&acting 'receptor' sequences may mediate dosage compensation and that such sequences are numerous on the X-chromosome (Hazehigg et al., 1984).
This hypothesis suggests that other X-linked genes may have closely adjacent receptor sequences and thus retain the ability to exhibit dosage compensation when translocated to autosomal sites. To determine if this is the case, and to investigate other possible mechanisms that might mediate dosage compensation in Drosophila, we have chosen the gene encoding G6PD as a model. In this paper we report the isolation and structural organization of the G6PD gene sequence in D. melanogaster. We also present evidence that dosage compensation of the G6PD gene in males is manifested at the level of transcription.

(a) Purification of glucosed-phosphate dehydrogenase and preparation of anti4XPD immunoglobulio
Adult D. melanogaster (Oregon R + ; 10 g) was homogenized at 4' C in 80 ml of 50 mM Tris, pH 8.0, 1 mM EDTA, 7 mM 2-mercaptoethanol, 10% glycerol (TEMG; Steele et al., 1969). The homogenate was centrifuged at 16 000 x g for 30 min, and the supernatant was applied to a 2', 5'-ADP agarose (PL Biochemicals) affinity column (110 x 10 mm) equilibrated with TEMG. After washing with 45 ml of TEMG, G6PD was eluted with 9.0 ml of TEMG containing 0.1 mM NADP + (Brodelius et al., 1974). The enzyme was placed on a DE-52 cellulose column (50 x 5 mM) equilibrated with TEMG and the column was washed with 10 ml of TEMG, containing 50 mM NaCl. The enzyme was eluted with 5 ml TEMG containing 150 mM NaCl. In a typical experiment, the yield of G6PD was determined to be 24%, and the specific activity was approx. 70 u/mg protein. The purification factor was 1296, in excellent agreement with previously reported values (Hori and Tanda, 1980). Enzyme activity at each step of purification was assayed according to Steele et al. (1969).
SDS-PAGE of the purified enzyme showed a major Coomassie blue-stained band of apparent M, 55000, which is the apparent M, of the monomeric unit of G6PD (Lee et al., 1978). A minor band, representing less than 10% of the intensity of the 55-kDal band was observed at apparent it4, 65000. The enzyme preparation was free of contamination by 6PGD activity, as assayed by staining the gel with 6PGD specific stain (Hori and Tanda, 1980). Rabbit antibody against G6PD was prepared as described by DeFlora et al. (1977). Anti-G6PD IgG was purified from rabbit serum by (NH&SO, precipitation, DE-52 cellulose column chromatography (Weir, 1967), and affinity chromatography on a protein A-Sepharose CL-4B column (Goudswaard et al., 1978). Heparin (20 u/ml) was added to the antibody preparation to inhibit RNase activity.
One ml of anti-G6PD IgG, at 1 mg/ml in PBS was added to 1.0 ml of polyribosomes, and the mixture was incubated for 2 h at 4°C. The mixture was applied to a 1.5~ml protein A-Sepharose C1-4B (Pharmacia) column at 4" C and washed with 80 ml of polyribosome buffer; mRNA in bound polyribosomes was eluted with 3.0 ml of 25 mM Tris, pH 7.6,20 mM EDTA, 20 u/ml of heparin (Shapiro and Young, 1981). RNA in the unbound polyribosome fraction (i.e., depleted polyribosomes) and in the specifically bound polyribosomes (i.e., enriched polyribosomes) was further purified by phenol-chloroform extraction, and polyadenylated RNA was isolated by oligo(dT)-cellulose chromatography (Levy and Manning, 1981). This procedure was repeated until approx. 10 000 A,,, u of polyribosomes had been processed. Total poly(A) + RNA obtained from the depleted and enriched polyribosomes were 2.0 mg and 14 pg, respectively.

(c) Determination of transcription orientation
(1) Sl protection assay The 1.2-kbBgZII-AvaI fragment (see Fig. 2) within exonI1 was 32P-labeled at the 5' terminus and digested with endonuclease PstI to generate a 0.4-kb BgZII-PstI fragment and 0.8-kb PstI-AvaI fragment. These DNA fragments (approx. 1 x 10m3 pg of coding sequence) were hybridized to adult polyribo-somal poly(A) + RNA (30 pg) under conditions that allow RNA-DNA hybridization, but do not allow DNA-DNA reannealing (Casey and Davidson, 1977;Levy et al., 1982). The R,,t value achieved was 460, which is approx. ten times that required for saturation hybridization under these conditions (Levy et al., 1982). After hybridization the sample was treated with Sl nuclease and the RNA-protected DNA was electrophoresed on an alkaline agarose gel (Maniatis et al., 1982). After electrophoresis the gel was neutralized in 3 M NaCl, 0.5 M Tris, pH 7.0, blotted to nitrocelhilose, and exposed for autoradiography.

(d) Radiolabeling of DNA and RNA
DNA was nick-translated with [ a-32P]dCTP (Amersham) to a specific activity of 2-4 x lo8 cpm/pg using a nick-translation kit from Bethesda Research Laboratory (BRL). The radiolabeled DNA was then purified as described by Levy et al. (1982). RNA was 5' -end-labeled to a specific activity of 3-5 x lo7 cpm/pg using T4 polynucleotide kinase (Boehringer-Mannheim) and [ Y-~~P]ATP (Levy et al., 1982). All restriction enzymes were purchased from BRL and used as recommended. Isolation and electrophoresis of nucleic acids and Southern and Northern transfers are described elsewhere (Levy and Manning, 1981;Levy et al., 1982). Hybridization of 5'-end-labeled RNA to Southern blots and filters for library screening was according to Fouts et al. (1981). Northern blots were prehybridized and hybridized as described by Thomas (1980). The DNA clone described here was selected by in situ plaque hyb~~zation technique (Benton and Davis, 1977) from a library of D. ~el~nog~s~r genomic DNA cloned in phage I Charon 4 (Maniatis et al., 1978).

(f) Hybrid selection of G6PD mRNA
For hybrid selection of G6PD mRNA, recombinant plasmid DNA pDmG20/3.9RI containing exon II was linearized with Hind111 and co-valent& coupled to DBM-cellulose (Anderson et al., 1979). Enriched pol~bosom~ poly(A) + RNA (1 fig) was hybridized to 1.3 pg of cell~os~bo~d DNA in 10 yl of 70% formamide cont~n~g 0.1 M Tris, pH 7.8, 0.3 M NaCl, 10 mM EDTA at 46°C for 24 h. The DNA-DBM-cellulose was washed three times with 200 ~1 of 0.3 M Naacetate, pH 7.5, three times with 200 ~1 of 55% formamide containing 0.7 M NaCl, 0.2 M Tris, pH 7.5, at 25°C and twice with 200~1 of the 55% formamide solution at 42°C. Hybridized RNA was recovered by incubating the DNA-DBM-cellulose in 300 $1 of 99 % formamide for 2 min at 60 ' C followed by ethanol pr~ipitation (Anderson et al., 1979). The hybrid-selected poly(A) + RNA was translated in a rabbit reticulocyte lysate system (BRL), and products were analyzed by SDS-urea PAGE.  (Rosen, 1976). The precipitate was dissolved in 20~1 sample buffer (Laemmli, 1970), heated at 100°C for 2 min and adjusted to 5 M urea. For total product analysis 10~1 of the post-ribosomal supematant was digested with 100 pg/ml RNase for 30 min at 37°C and adjusted with sample buffer and urea. The total (lanes a, c) and anti-GGPD-IgG-precipitable products (lanes b, d) synthesized in the cell-free system were analyzed by SDS-urea PAGE (Storti et al., 1976) shown in Fig. 1. Although the major translation products of enriched and depleted poly(A) + RNA of the two RNAs are essentially identical, a minor polypeptide of apparent Mr 55 000, the size of the G6PD monomer (Lee et al., 1978), is visible only in the ~~slation products directed by the enriched poly(A) + RNA. ~unoprecipitation of these pro-ducts with anti-G6PD IgG shows a polypeptide band of Mr 55000 and a few smaller polypeptide bands (Fig. 1, lanes b and d), which may represent premature termination of synthesis of G6PD or proteolytic degradation of the G6PD monomer in the cell-free ~~sla~on system. The intensity of the 5%kDal polypeptide band in the enriched poly(A) + RNA translation products was consistently higher (three experiments) than that observed for the depleted poly(A) + RNA, indicating a detectable difference in the abundance of G6PD mRNA between the two poly(A)+ RNAs. As discussed below, the abundance of G6PD mRNA in the enriched poly(A) + RNA is approx. 67fold higher than that in the depleted poIy(A) + RNA.

(a) Isolation of the G6PD gene
To isolate the G6PD gene sequence pairs of duplicate plaque filters of the ~~~~~~~~ genomic library were hybridized with t3'P]RNA probes from the enriched and depleted poly(A) + RNAs, respectively, From an initial screening of approx. 44000 plaques a single clone, ADmG21, was observed upon successive plaque purifications to hybridize only with the euriched poly(A) + RNA probe (not shown). Fig. 2 shows the restriction map of the cloned 13,0-kb EC&I ~~~so~~~ genomic DNA fragment in JDmG21. Hyb~d~ation of enriched fs2P Jpoly(A)"RNA to Southern (1975) blots of restricted ADmG21 DNA was confined to two separate regions of the DNA, the O.Pkb PvuII-BarnHI fragment and the 1%kb BglII-PvuI fragment. Depleted [ 32P]poly(A) + RNA was also observed to hybridize with these two DNA fragments, although the intensity of the hybridization signal was notably less than that observed with the enriched poly(A)+ RNA. These results suggest that the coding region of the putative G6PD gene is contained within the I3.0-kb EcoRI fragment and is interrupted by at least one intron 2.4 kb in length.

95
That IDmG21 contains the coding sequence for G6PD was confirmed by three methods, [3H]1DmG21 DNA and the 3.Pkb EcoRI fragmem containing exonI1 were separately hybridized in situ to larval salivary gland polytene chromosomes. Hyb~d~a~on of either f3H]-labeled probe was observed only at the 1 gD region of the X-chromosome (Fig. 3), consistent with the chromosomal position of the Zw + locus (Stewart and Merriam, 1974) In the second method, poly(A) + RNA complimentary to exonI1 was hybrid-selected and translated in a rabbit reticulocyte lysate system. SDS-urea PAGE of the translational products (Fig. 4b) showed three poly peptides of apparent M,s 92 500,55 000 and 50000, The bands at &4,92 500 and M, 50000 are likely due to the presence of endogenous RNA in the reticulocyte lysate (Robson et al., 1982), since these bands are also observed in the absence of exogenous RNA (Fig. 4a). The 55ikDal polypeptide, identical in apparent Mp to the monomeric unit of GfiPD, is observed only upon addition of the hybrid selected poly(A)+ RNA. In the third method, the 3.9-kb EcoRI fragment containing exonII was partially digested with exonuclease RAL31, repaired with the Klenow fragment of DNA polymerase I and bhmtend-ligated with EcoRI linkers. Following digestion with EcoRI, the fragments were inserted into the EC&I restriction site of the expression vector lgtl I (Young and Davis, 1983), and recombinant I phages were screened with anti-G6PD IgG. Approx. 200/, of the recombinant phages showed a positive sighal, which is in reasonable agreement with the expected number of 1 in 6, assuming random orientation of the EcoRI fragments in coding phase. Based upon these results we conclude that ADmG21 contains sequences coding for G6PD.   (Levy et al.,19g2). The results show that the DNA hybridizes to the 18D region of the X-chromosome. Similar results were obtained when a 3.9-kb EcoRI fragment was used as a probe. The cytological map described by Lefevre (1976) was used for chromosomal localization.

(b) Measurement of G6PD mRNA size and relative abundancy in males and females
[3H]Uridine incorporation into the single polytenic X chromosome of male larval salivary glands and the paired X chromosomes of females is equivalent (Mukherjee and Beermann, 1965), suggesting that dosage compensation is a ~~sc~ption~ phenomenon. Since this observation suggests that the steady-state level of mRNA encoded by X-linked dosage compensated gene sequences is equivalent in males and females, we determined the relative level of G6PD mRNA in the two sexes.
Initially, the size and relative abundance of G6PD mRNA in the enriched and depleted poly(A) + RNA preparations was measured by Northern blot hybridization. As shown in Fig. 5, the length of the G6PD mRNA transcript is 2.0 kb and the abundance of G6PD mRNA in the enriched poly(A)+ RNA is greater than in the depleted poly(A) + RNA. Overexposure of the autoradio~~ of Fig. 5 showed the presence of a minor band of length 2.3 kb. Possible ~p~cations of this RNA species are discussed below (see DISCUSSION, section b), To quantitate the relative abundance of the two RNA fractions, the area of each blot containing only the intensely hybridizing band was removed and counted by liquid scinti~ation. When the 3.9-kb EcoRI fragment was used as the probe, G6PD mRNA transcripts were seen to be 67 times more abundant in the enriched poly(A) + RNA than in the depleted poly(A) + RNA (Fig. 5, lanes a-c). A similar difference (i.e., S-fold) in the abundance of G6PD mRNA transcripts between the two RNAs was observed when the 3.0-kb MI fragment was used as the hybridization probe (Fig. 5, lanes d and f).
To determine the steady-state level of G6PD mRNA transcripts in male and female Drosophila equivalent amounts of polyribosomal poly(A) + RNA from adult male and female flies were analyzed by Northern blot hybridization (Fig. 6). Quantitation of radioactivity in the hybridizing bands showed the 4.3-0.588- Fig. 6. Determination of the relative steady-state level of the G6PD mRNA level in male and female polyribosomal poly(A)+RNA. Northern blots of glyoxalated RNA were hybridized with 32P-labeled G6PD DNA (lanes a-d) or Drosophila Adh DNA (lanes e, f) and autoradiographed. Lanes a-d were exposed for six days, and lanes e and f were exposed for one day.  Table I. Size markers on left margin as in Fig. 5. level of G6PD mRNA transcripts in adult male RNA to be slightly greater than that of the female RNA. Essentially identical results were obtained when the 3.0-kb Sal1 fragment was used as the DNA probe (not shown). To determine whether the observed difference in the steady-state level of GBPD mRNA in males and females is a result of inaccuracy in RNA quantitation, the steady-state level of Adh mRNA in these two RNA samples was measured. As shown in Fig. 6 and Table I, male RNA has a slightly higher level of Adh mRNA than female RNA. The ratio of the steady-state levels of G6PD  Fig. 6 were excised out and radioactivity determined. Nitrocellulose of an equivalent area was excised where no hybridiiation was observed and counted to determine background radioactivity. b Probes used were s2P-labeled pDmG20/3.9RI (G6PD DNA) and SAC-~ (Adh DNA).
mRNA and Adh mRNA in the two sexes is found to be essentially identical, thus indicating that the steady-state level of G6PD mRNA in male and female Drosophila is equivalent.

(c) Transcription orientation
Two methods were used to determine the transcriptional orientation of the G6PD gene on the restriction map shown in Fig. 2. First, a 0.4-kb BglII-PstI fragment and a 0.8-kb PstI-AvaI fragment were 5'-end-labeled at the Bg01 and AvaI sites, respectively, and hybridized with a large sequence excess of adult polyribosomal poly(A)+ RNA. Following S 1 nuclease digestion (Fig. 7) only the 0.8-kb &I-AvaI fragment is protected, thus providing the 5'-3' orientation shown in Fig. 2. In the second method enriched poly(A) + RNA EcoRI-AvaI fragments but not with the 0.45-kb was partially hydrolyzed, 5'-end-labeled with AvaI fragment (results not shown). When 3'-end-[ y3'P]ATP (Levy et al., 1982), and chromato-enriched RNA sequences were used as a probe, graphed on an oligo(dT) cellulose column. The 5'hybridization was seen only with the 1.8-kb EcoRIend-enriched (unbound) and 3'-end-enriched AvaI and 0.45-kb AvaI fragments, but not with the (bound) sequences were hybridized to Southern 1.5-kb SalI-BamHI fragment. These results suggest blots containing the following DNA fragments in that the 1.5-kb WI-BamHI fragment is located at equimolar amounts (Fig. 2) : (a) 1.5-kb SalI-BamHI the 5' end ofthe G6PD mRNA and the 0.45-kbAva1 containing exon I; (b) 1.8-kb EcoRI-AvaI containing fragment is located near the 3' end of the G6PD exon II and (c) 0.45-kbAva1 located at the 3' end of mRNA. This leads to the 5'-3' orientation shown in exon II. The 5'-end-enriched RNA sequences hy- Fig. 2 and is consistent with the results of the Sl bridized with the 1.5-kb SalI-BamHI and 1.8-kb nuclease mapping experiment. a b

(d) Copy number of the G6PD gene in maJe and female Drosophila geaomes
As shown above, the steady-state amount of G6PD mRNA is approximately equivalent in both male and female Drosophila. One explanation for this would be a single duplication of the G6PD gene in males, thus providing G6PD mRNA equivalence between the sexes.
We therefore examined the average copy number of the G6PD gene sequence in the genomes of male and female Drosophila. Genomic DNA from adults was digested with PstI and blotted to nitrocellulose after gel electrophoresis. Included on the Southern blot was DNA from subclone pDmG20/3.9RI, containing exonI1 equivalent to 1, 2,4 and 6 copies per haploid female genome (i.e., one set of autosomes and one X-chromosome). When the Southern blot was probed with a 32P-labeled 1.35-kb &I fragment containing the 3'-distal half of exonI1, the only genomic DNA band that hybridized was of length 1.35 kb (Fig. 8). When the intensities of the hybridization in the genomic DNAs are compared with those of the various equivalents in the cloned DNA, it is apparent that both males and females have one copy of the G6PD gene sequence per X chromosome and that the G6PD gene in the male X-chromosome is not duplicated.

(a) Isolation of the C6PD gene sequence
We have been able to select from a Drosophila genomic 1 library a recombinant phage, IDmG21, containing the DNA sequence encoding G6PD. Three separate approaches were employed to confirm the presence of G6PD coding sequences in the recombinant 1 phage. First, the genomic location of the inserted Drosophila DNA fragment was shown by in situ hybridization to correspond to the chromosomal site of G6PD. Second, the in vitro translation product of hybrid-selected mRNA was identical in M, (i.e., 55000) to the monomeric unit of G6PD. Due to the low abundance of G6PD mRNA in the total RNA population, amounts of hybrid-selected RNA could not be obtained in sufficient quantities to determine whether the 55-kDal translation product is recognized by antisera directed against G6PD. However, digestion of the 3.9-kb EcoRI fragment containing exon II with BAL 3 1 nuclease, followed by insertion of the resected DNA fragments into the expression vector Iz gtll (Young and Davis, 1983), did yield recombinant phages, 20% of which reacted positively with anti G6PD IgG. Collectively, these results indicate that 1 DmG21 contains sequences that encode G6PD.

(b) WPD mRNA
As shown in Fig. 6, the size of the G6PD mRNA in adult male and female flies is 2.0 kb, and the steady-state levels of the mRNAs in the somatic cells of the two sexes are similar. This observation indicates that similar levels of G6PD enzyme activity in the two sexes (Seecof et al., 1969) is due to the steady-state level of G6PD mRNA. Also, it supports the previous conclusions, based on studies using salivary gland chromosomes (Mukherjee and Beermann, 1965), that dosage compensation is the result of increased transcriptional activity of the male X chromosome. Our results do not, however, eliminate differential RNA processing and/or RNA turnover rates as possible mechanisms for generating equivalent levels of G6PD mRNA in the two sexes.
Overexposure of the autoradiogram (Fig. 6) shows the presence of a minor RNA species of length 2.3 kb in both male and female poly(A)+ RNA. Since the 2.3-kb RNA hybridizes with exon I and exon II gene probes, it is possible that this RNA represents a minor species of G6PD mRNA. Whether this RNA results from either differential processing of the G6PD primary transcript or a different transcriptional start or stop site from that of the major 2.0-kb mRNA has not been determined. We have observed, however, the 2.3-kb RNA to be present in the poly(A) + RNA from several embryonic stages of development as well as the three larval stages and the pupal stage (R.G. and J.E.M., unpublished observation). In all cases the ratio of the 2.3and 2.0-kb mRNAs remains unaltered. Thus it is unlikely that this RNA species represents a minor component in one stage of development and a major component in another.

(c) Gene copy number
It is formally possible that the equivalent steadystate levels of G6PD mRNA in the two sexes result from duplication of the G6PD gene sequence in males. To examine this possibility, the copy number of the G6PD gene was determined in total genomic DNA from both male and female tlies (Fig. 8). The results of this experiment show that the number of G6PD gene sequences in male and female genomes is directly propo~ion~ to the number of X-chromosomes in the two sexes, and that the G6PD gene is present as a single copy sequence on the X-chromosome. Dosage compensation of G6PD is not, therefore, due to amplification of the G6PD gene in males.