INTRODUCTION

The vacuolar ATPase¹ is a multisubunit complex, related to the F₁/F₀ ATPase (1, 2, 3). The transmembrane protonophore is made of six copies of a 16-kDa proteolipid, linked by further subunits to a catalytic headgroup comprising three copies each of a 67-kDa A-subunit and a 57-kDa B-subunit. Traditionally, the A-subunit is described as catalytic, whereas the B-subunit is considered regulatory, although in reality the active sites for nucleotide binding and proton flux may lie in the interfaces between neighboring A- and B-subunits (4). Although there is known to be very high conservation within the V-ATPase family, the 57-kDa subunit is interesting as several transcripts are known, some of which are tissue-specific (5, 6). It has been argued further that choice of B-subunit transcript may affect the overall subunit composition of the holoenzyme by influencing the choice of other subunits during assembly of the V₁ headgroup (6), and the B-subunit is phosphorylated by a component of AP-2, the clathrin assembly complex (7), suggesting a role in control of vesicle trafficking.

V-ATPases, although originally defined as endosomal, are now known to energize plasma membrane transport in a variety of tissues, such as kidney, osteoclasts, and frog skin (8); and loss of kidney V-ATPase function in autoimmune disease is clinically significant (9). In an invertebrate model, a painstaking biochemical purification of particles on the goblet cell apical membranes of lepidopteran midgut (10) showed that the invertebrate K⁺ pump (11) was in fact a V-ATPase (12), driving a K⁺/H⁺ antiporter (13) to produce a remarkably potent transport system (14, 15). Monospecific antibodies against the lepidopteran plasma membrane V-ATPase were used to demonstrate plasma membrane V-ATPases immunocytochemically in salivary glands, Malpighian (renal) tubules, recta, and cuticular sensillae, suggesting an energization of transport by proton, rather than sodium, motive force in probably all insects (16, 17, 18). Physiological evidence shows that Drosophila melanogaster Malpighian tubules, in common with those of other insects, are energized by a plasma membrane V-ATPase (19, 20). Accordingly, D. melanogaster not only contains a V-ATPase, but should embody the full range of plasma membrane and endomembrane roles for V-ATPase function, and our group has embarked on the characterization of V-ATPase genes with a view to dissecting the differing requirements of these roles using the genetic tools unique to Drosophila (21). The first steps in such a procedure, in D. melanogaster or any other species, are to clone and characterize the gene, obtain a chromosomal localization, and identify or produce a genetic null (a "knockout"), which may have an informative phenotype in its own right, but which provides a genetic background for subsequent genetic intervention. These are the steps reported here for vha55, the gene encoding the V-ATPase B-subunit in D. melanogaster; they represent the first knockout of a V-ATPase gene in an animal.

EXPERIMENTAL PROCEDURES

Primers for polymerase chain reaction were based on two areas of particularly close homology in the then known sequences for the 57-kDa subunit (22), optimized for Drosophila codon preference (23). Two templates were used: 100 ng of first strand cDNA from Manduca sexta midgut and 1 µl of plate lysate from a D. melanogaster head cDNA library, constructed in ZapII. PCR products were sequenced directly to establish their identity, and a D. melanogaster cDNA library, prepared from eya adult fly heads, was screened by plaque hybridization using the D. melanogaster PCR product as a probe. Positives were obtained at 1:1000 and were observed to fall into one of three distinct strengths of hybridization. The insert of a single recombinant phage from the strongest hybridizing class of plaque was excised as a pBluescript plasmid, and the cDNA was sequenced over its entire length on both strands. Double-stranded sequencing was performed according to the Sequenase^TM II protocol (U. S. Biochemical Corp.), with the aid of synthetic oligonucleotide primers.

Total RNA was extracted from whole flies or from hand-dissected body segments by the RNAzol B method (24). Electrophoresis of RNA was carried out in 1% formaldehyde-agarose gel in MOPS buffer and followed by transfer to a nylon filter. RNA was cross-linked to the filters by UV radiation. The Northern blot filters were probed with the vha55 cDNA or a control rp49 cDNA encoding a ribosomal protein, each labeled with ³²P by random priming. Blots were prehybridized in Church buffer (7% SDS, 1% bovine serum albumin, 1 mM EDTA, Na₂HPO₄ 0.25 M, pH 7.2) at 55 °C for 2-3 h and hybridized in Church buffer overnight. The filters were then washed at 55 °C with 2 × SSC, 0.1% SDS for 30 min, then 0.5 × SSC, 0.1% SDS for 30 min, and finally 0.1 × SSC, 0.1%SDS for 30 min. Filters were then exposed to Fuji x-ray film for 1-3 days.

Genomic DNA was purified from pools of approximately 100 adult D. melanogaster (25), cleaved with a range of restriction endonucleases, Southern blotted, and probed with ³²P-labeled random-primed probes derived from the coding portion of the D. melanogaster vha55 or vha14 (26) cDNAs. Filters were hybridized and washed to high stringency. Blots were prehybridized in Church buffer at 65 °C for 2-3 h and hybridized in Church buffer overnight. The filters were then washed at 65 °C with 2 × SSPE (0.15 M NaCl, 0.01 M NaH₂PO₄, 0.001 M EDTA, pH 7.7) + 0.1% SDS for 30 min then 0.5 × SSPE, 0.1% SDS for 30 min, and finally 0.1 × SSPE, 0.1% SDS for 30 min. Filters were then exposed to Fuji x-ray film for 1-3 days.

Single flies were homogenized in extraction buffer (27) and 1-µl aliquots used as templates in a PCR reaction using primers known to bracket an intron at the 5 end of the open reading frame. Results were analyzed by agarose gel electrophoresis and the identity of products verified by Southern blotting with vha55 cDNA as described above (not shown). The products of the reactions with l(3)j2E9/TM3 DNA template were cloned using the Invitrogen TA cloning kit according to the manufacturer's instruction, and the authenticity of the products established by sequencing, incidentally allowing the position of the intron to be fixed.

Northern and Southern analyses were performed on laboratory stocks of Oregon R wild-type flies. P-element insertional mutants were obtained from the Bloomington stock center and lines carrying deficiencies from the Umea stock center or as kind gifts of Prof. Janos Gausz (Szeged, Hungary). Alleles of Polycomb (Pc) were obtained from the Umea stock center. Genetic markers were checked before use. Flies were reared on standard Drosophila medium at 25 °C in a 12:12-h photoperiod.

For crosses, virgin females were collected at 4-h intervals, mated singly to males, and tapped into fresh tubes daily for 1 week before discarding. Particular care was taken with Pc crosses to allow the progeny to grow under uncrowded conditions, as the penetrance of the phenotype is particularly sensitive to environmental conditions (28). Emerging adults were collected for several days from each tube, to reduce the risk that certain classes of phenotype might be slower growing and thus underrepresented in early collections.

Embryos were washed from staged egg plates, dechorionated, fixed, devitellinized, and stained for -galactosidase according to standard protocols (25). Larvae or adults were dissected in Drosophila saline, pinned out, fixed, and stained similarly. Staining patterns were viewed either under a Wild Stereomicroscope or a Leitz Ortholux microscope and photographed using Kodak Ektar film.

RESULTS

An 830-bp PCR product was isolated from a D. melanogaster eya head cDNA library by PCR with degenerate primers based on conserved domains in known V-ATPase B-subunits. The PCR product was sequenced directly to establish its identity and used to identify a cDNA from the same Drosophila head library. The 2.6-kb cDNA contained a long open reading frame of 1470 bp, encoding a polypeptide of 490 amino acids with deduced molecular mass close to 55 kDa (Fig. 1) Sequence identity at the amino acid level with known V-ATPase B-subunits is extraordinarily high (Fig. 2); only 11 of 490 residues differed between the M. sexta (29) and D. melanogaster sequences, while the human brain sequence is 90% identical. Peptide sequence motifs were identified using the PROSITE data base. In common with all known F-ATPase (ATP synthase) A-subunits and V-ATPase B-subunits, the sequence shows the PPVNVLPSLS motif at 370 (Fig. 3) apart from a conservative V/I substitution at 396; it is thought that this motif is essential for ATPase function. This gene thus clearly encodes a B-subunit of a V-ATPase and was accordingly named vha55. The initiation site is consistent with those known for other sequences, and there is a canonical polyadenylation signal at 2529-2534, 25 bases before the start of the poly(A) region (Fig. 2).

Northern analysis of head + body mRNA (Fig. 3A) reveals the presence of at least four bands of 2.8, 2.5, 2.0, and 1.8 kb, suggesting transcriptional complexity. This is interesting, because in vertebrates the B-subunit is thought to be the most transcriptionally diverse subunit in this highly conserved proton pump (5). The vha55 cDNA has a 5-untranslated region of 86 bp and a long 3-untranslated region of 1019 bp (Fig. 1). Separate transcripts, differing in length of their 3-untranslated regions, have been identified in kidney and brain, with brain cDNAs being 1 kb longer than those from kidney (5, 6), suggesting that, in humans, the B-subunit may be a useful target for selective therapeutic intervention against kidney or osteoclast isoforms (30). Tissue- and sex-specific Northern blots showed that, in common with other recently characterized D. melanogaster V-ATPase subunits (26, 31), vha55 was ubiquitously expressed (Fig. 3B).

D. melanogaster (Oregon R) genomic DNA was cleaved with a range of restriction endonucleases, Southern blotted, and probed with vha55 cDNA. Hybridization and washing at either high (not shown) or low stringency suggests that this is a single-copy gene (Fig. 4). In accordance with this, salivary gland polytene chromosome squashes (25) probed with vha55 cDNA revealed a single band at 87C (not shown). The availability of physical mapping clones spanning most of the D. melanogaster genome can allow genomic clones to be obtained rapidly when such localizations are known. In this case, six P1 clones spanning 87C (a kind gift of Dr. Stephen Russell) were screened by PCR with internal nondegenerate primers bracketing a 1.2-kb intron and so known to yield products of 430-bp off cDNA and 1600-bp off genomic DNA templates. Two positive clones (DS00602 and DS00681) were obtained, both previously mapped to 87C2, and their identities confirmed by Southern blotting of P1 DNA with vha55 cDNA (not shown).

To confirm the localization, Df(3R)kar^3J, a fly line carrying a deficiency spanning 87C1-87D1, was subjected to quantitative Southern analysis using a fragment of the vha55 cDNA as a probe. Fig. 5 confirms that the vha55 gene falls within the Df(3R)kar^3J deficiency.

A powerful tool in the analysis of any gene is the identification of mutant or null alleles. Two attempts to mutagenize vha55 by site-selected P-element mutagenesis (32) were unsuccessful. However, two lethal P-element insertions, l(3)j2E9 and l(3)05043 mapping to 87C2-3 and 87C6-8 respectively, were investigated on the basis that they might lie sufficiently close to the predicted location of vha55 (87C2-3) to permit mutagenesis by "local jumping," a strategy that relies on the relatively common reinsertion of excised P-elements close to their previous locations. PCR off genomic DNA prepared from single flies, using the gene-specific primers described above, revealed that the line l(3)j2E9 carried a P-element insertion within the only known intron in vha55 (Fig. 6); these positions were verified by cloning and sequencing the PCR products and place the intron between bases 160 and 161 of the sequence reported in Fig. 1.

The region covered by Df(3R)kar^3J has been subjected to saturation mutagenic analysis, in which a total of just four lethal complementation groups have been identified (33). The existence of a large number of small deficiencies spanning 87C (33) permitted an accurate localization of l(3)j2E9 by complementation. l(3)j2E9 virgins, balanced over TM3 (carrying Stubble, a dominant marker), were crossed singly to males carrying each of a number of deficiencies (also balanced over TM3) with breakpoints in 87C. If any progeny had wild-type (Sb⁺) bristles, then this would show that the lethal mutation fell outside the deficiency. As can be seen (Fig. 7), the results place l(3)j2E9 within the interval 87C2-4.

These results allow the l(3)j2E9 allele to be reconciled with known complementation groups at 87C. Despite a small discrepancy (87C2-3 cf. 87C4-5) between the documented positions of the two loci, l(3)j2E9 maps within the same combination of deficiencies diagnostic of SzA (=l(3)87Ca) (33). As three ethyl methanesulfonate-induced alleles of SzA are still extant, it was possible to confirm that l(3)j2E9 (but not l(3)05043) is allelic to SzA (Table I). Additionally, the P-element was mobilized by crossing to the 2,3 transposase source, and 100 lines were established from progeny which had lost the w⁺ marker, and so represented independent P-element excisions. None of these was homozygous viable, revealing the presence of an undocumented additional source of lethality in the l(3)j2E9 line. However, this locus could be separated from the P-element insertion by five generations of outcrossing to w¹¹¹⁸ flies, a maneuver which exchanges 97% of the genome by recombination. A selection of 10 independent P-element excision lines were able to complement both this outcrossed P-element stock and the defining deficiency spanning the locus, Df(3R)kar^Sz29, while the outcrossed P-element stock remained unable to complement Df(3R)kar^Sz29 (Table II). So the lethal locus which remains after P-element excision does not map to SzA. Therefore, the P-element insertion causes lethality at the SzA locus, and it is appropriate to name this allele vha55^j2E9.

Table I. Allelism between lethal P-element insertions within 87C and SzA Virgins heterozygous for two lethal P-element insertions within 87C, l(3)j2E9/TM3 (Sb) or l(3)05043/TM3 (Sb) (Bloomington stocks), were crossed to males heterozygous for three lethal alleles of SzA/TM3 (Sb) (kind gift of J. Gausz) and the progeny scored for stubbly bristles. Wild-type bristles imply complementation between chromosomes carrying the two lethal mutations, and thus that they are in different genes. Cross Sb+ Sb l(3)j2E9 × SzA1 0 77 l(3)j2E9 × SzA9 0 113 l(3)j2E9 × SzA12 0 137 l(3)j05043 × SzA1 21 50 l(3)j05043 × SzA9 42 72 l(3)j05043 × SzA12 8 13

Table II. Loss of the P-element in l(3)j2E9 reverts lethality at 87C Heterozygous flies from four independent lines, ex1-ex4, generated by loss of the P-element in l(3)j2E9 were crossed either to the deficiency spanning 87C4-87C9, Df(3R)karSz29, or to a stock, l(3)j2E9(7CS), derived from the original P-element line by seven generations of outcrossing to w1118 in a Canton S background. As parental lines were homozygous lethal and kept over the TM3 balancer; complementation was revealed by the presence of Sb+ progeny for crosses involving Df(3R)karSz29, and of w+Sb+ progeny for crosses involving l(3)j2E9(7CS). Complementing/total progeny ex1 ex2 ex3 ex4 Df(3R)karSz29 Df(3R)karSz29 50 /147 28 /96 37 /123 10 /29 (0) l(3)j2E9(7CS) 49 /196 5 /53 5 /44 8 /25 0 /51

On this basis, existing detailed descriptions of the phenotypes associated with the 13 lethal alleles of SzA (33) can be ascribed to genetic "knockouts" of the V-ATPase regulatory subunit, vha55. Severe ethyl methanesulfonate-induced alleles produce viable heterozygotes but a recessive embryonic lethal phenotype; whereas complete deletion of the gene produces later lethality, after the emergence of homozygous first-instar larvae (33). This suggests that subunits carrying point mutations may misincorporate into and disrupt maternally derived V-ATPase holoenzymes earlier than the enzyme would have become ineffective by dilution among the cells of the developing embryo, so producing an earlier lethal phase. As a V-ATPase presumably needs all three of its B-subunits to function, this could provide a "dominant negative" phenotype characteristic of multisubunit proteins. We found the lethal phase of the vha55^j2E9 homozygotes to be around the time of hatching, as seen with Df(3R)kar^3J homozygotes (33), suggesting that insertion of the 8-kb P-element completely disrupts the locus and is thus a genetic null.

Affected larval Malpighian tubules in SzA flies are also colorless, lacking characteristic luminal white concretion bodies (33). This can be interpreted as a failure of V-ATPases to acidify the tubule lumen and precipitate urates or mineral concretions. The apical membranes of tubule cells are packed arrays of V-ATPases (34); and D. melanogaster tubules are known to be highly sensitive to bafilomycin, a selective inhibitor of V-ATPases (19, 35), so tubules are natural sites to observe manifestations of defects in V-ATPase function. The tubule phenotype is cell autonomous in transplants of tubules to healthy flies, as would be expected for a V-ATPase mutant, and the failure of the transplanted cells to grow or thrive prompted Gausz to suggest that the SzA⁺ gene product was required for growth of most cell types (33).

The P{LacW} element in vha55^j2E9 contains a LacZ enhancer detector element, although PCR with P-element end-specific primers (not shown) revealed the orientation of the reporter to be opposite to that of transcription of vha55. Despite this, the LacZ pattern described below is plausible (Fig. 8) and is identical to those we have observed with P-element insertions into two other V-ATPase subunits.² As would be expected of a housekeeping gene, expression was reported in most tissues; however, it was particularly strong in those epithelia where V-ATPases have been demonstrated immunocytochemically on plasma membranes of other insects (36): the Malpighian tubules (Fig. 8, B and C), the rectum (Fig. 8D) and the salivary glands and the cuticular sensillae (Fig. 8E). There is also staining of the uterus and female accessory glands, again suggesting a plasma membrane transport role for V-ATPase in this tissue, possibly in eggshell formation. Interestingly, the earliest embryonic staining was observed in a region of the anterior midgut (Fig. 8A) where the pH is particularly low (37). This region of the midgut contains goblet cells structurally reminiscent of those of lepidopteran midgut, the model system in which a plasma membrane role for V-ATPases was first demonstrated (12).

Products of the Pc group of genes are thought to act cooperatively to elicit transcriptional silencing of genomic DNA in response to the initial transcription levels of homeotic genes in the embryo by binding untranscribed DNA and packing it into heterochromatin (38). They thus help to preserve the "memory" of the homeotic selector genes by shielding inactivated genes from later access by transcription factors which might otherwise activate them ectopically. Mutant phenotypes of any of the Pc group (ectopic expression of sex combs on T2 and T3 legs of males) thus resemble a homeotic misassignment of T2 and T3 legs to a T1 fate. This model for transcriptional silencing may have general significance for the long term inactivation of genes during development. A previous screen for dosage-dependent suppressors or enhancers of Pc had identified a number of homeotic genes, together with several that were mapped but unidentified (28). One of the powerful suppressing loci was shown to be allelic to l(3)87Ca (28), a locus that we have now demonstrated to encode vha55. Given that most known suppressors of Pc are homeotic genes, Kennison and Tamkun's previous assignment was confirmed by generating vha55^j2E9/Pc¹¹ flies from a vha55^j2E9/TM3(Sb) × Pc¹¹/TM3(Sb, Ser) cross and scoring the male progeny for ectopic sex combs on the T2/T3 legs. This confirmed that hemizygosity for this V-ATPase subunit suppresses the ability of Pc to elicit ectopic sex combs (Table III), although it remains technically possible that this effect is due to interference with a nearby transcription unit, rather than with vha55. Indeed, there is evidence that SzA may be a complex locus (33), and breakpoints in the 86-89E region can elicit transvection effects at the Ubx locus (39); it will require further work to exclude such possibilities.

Table III. Dosage-dependent suppression of Pc phenotype by l(3)j2E9 l(3)j2E9/TM3(Sb) virgins (Bloomington) were crossed to Pc11/TM3(Sb, Ser) males (Umea) and the three classes of surviving male progeny scored for sex combs on T2 and T3 legs. Class of progeny Legs with ectopic sex combs Total T2/T3 legs scored Number % l(3)j2E9/TM3(Sb) 0 0 44 l(3)j2E9/Pc11 5 14 36 Pc11/TM3(Sb, Ser) 14 69 32

DISCUSSION

This study reports the gene knockout of a V-ATPase in an animal. The phenotype displayed here is straightforward and confirms that these genes are essential for normal embryonic development. The enhancer detector element also displays elevated gene expression in those epithelia thought to be energized by V-ATPases. This is because, whereas the acidification of endomembrane compartments requires only a few V-ATPase holoenzymes per vesicle, the plasma membranes of V-ATPase-energized epithelia contain essentially semi-crystalline arrays of V-ATPases (40, 41). It is reasonable to suppose that so great a density of V-ATPases would require higher levels of transcription of the relevant genes, and so the extent of LacZ reporter expression can thus be considered to be at least a semi-quantitative estimate of vha55 expression. Once identified, such null alleles are powerful tools in the analysis of this and other V-ATPase genes. They both provide null backgrounds into which mutagenized genes may be reintroduced to rescue function in vivo and allow novel alleles to be generated by imprecise excision of the P-elements. Although the yeast model is ideal for the mutagenic analysis of endomembrane V-ATPases in unicellular organisms (42), it is interesting to note that disruption of V-ATPase subunits in yeast is not obligatorily lethal, but results in a pH-dependent phenotype (43). For multicellular animals, then, Drosophila may prove a more suitable model, which will allow an analysis in an organism which possesses the full spectrum of predicted endomembrane, epithelial, neuronal, and sensory phenotypes.

In summary, this work provides a first description of the expression patterns and lethal phenotypes associated with a range of alleles of a V-ATPase subunit in Drosophila. It suggests that mutations will be lethal recessive and that, for at least those subunits which are present in multiple copies in the holoenzyme, severe alleles will be antimorphs. It also predicts that epithelial dysfunction may be detectable at an early stage in those tissues where V-ATPases play a plasma membrane role. As further V-ATPase mutations are identified, either in Drosophila or in other animals, it will be interesting to establish whether they match this prototype.