The oxen Gene of Drosophila Encodes a Homolog of Subunit 9 of Yeast Ubiquinol-Cytochrome c Oxidoreductase Complex: Evidence for Modulation of Gene Expression in Response to Mitochondrial Activity
Maxim V. Frolov, Elizaveta V. Benevolenskaya, James A. Birchler


A P-element insertion in the oxen gene, ox1, has been isolated in a search for modifiers of white gene expression. The mutation preferentially exerts a negative dosage effect upon the expression of three genes encoding ABC transporters involved in pigment precursor transport, white, brown, and scarlet. A precise excision of the P element reverts the mutant phenotype. Five different transcription units were identified around the insertion site. To distinguish a transcript responsible for the mutant phenotype, a set of deletions within the oxen region was generated. Analysis of gene expression within the oxen region in the case of deletions as well as generation of transgenic flies allowed us to identify the transcript responsible for oxen function. It encodes a 6.6-kD homolog of mitochondrial ubiquinol cytochrome c oxidoreductase (QCR9), subunit 9 of the bc1 complex in yeast. In addition to white, brown, and scarlet, oxen regulates the expression of three of seven tested genes. Thus, our data provide additional evidence for a cellular response to changes in mitochondrial function. The oxen mutation provides a model for the genetic analysis in multicellular organisms of the effect of mitochondrial activity on nuclear gene expression.

WE are interested in defining a complete set of modifiers that exhibit a dosage effect upon the expression of a single target locus. Our interest in dosage-dependent modifiers of gene expression is centered on the hypothesis that these are responsible for aneuploidy syndromes and various types of dosage compensation (Birchler and Newton 1981; Guo and Birchler 1994; Birchler 1996). It is generally accepted that the expression of a structural gene is directly proportional to gene dose. However, the study of gene expression in dosage series in both maize and Drosophila revealed that this is not always the case. The expression of some genes was found to remain unchanged despite variation in the dosage of the chromosomal segment where they reside. In contrast, some genes, whose dosage has not been altered, were affected by aneuploidy of unlinked chromosomal regions (Birchler 1979; Birchler and Newton 1981; Devlinet al. 1982; Guo and Birchler 1994). To resolve this paradox, it was proposed that for any one gene, multiple trans-acting dosage-dependent modifiers exist in the genome. When the modifier is present in a varied chromosomal region, the expression of an unlinked target gene is affected. It was further hypothesized that the molecular basis of deleterious effects associated with aneuploid syndromes is due to an imbalance in the system of dosage-dependent modifiers, rather than a disproportion of structural gene products encoded by the varied chromosomal region (Birchler and Newton 1981; Guo and Birchler 1994). Hence, isolation and dissection of the function of these modifiers of gene expression is important for understanding the basis of the dosage effects.

The Drosophila white gene represents a particularly convenient model where such dosage effects can be studied. Extensive molecular genetical and biochemical studies of pigment synthesis revealed that white together with two other genes, brown and scarlet, is involved in the uptake of pigment precursors by the cell. Biochemical analysis indicates that white and scarlet participate in the transport of brown pigment precursors kynurenine, 3-hydroxy-kynurenine, and tryptophan, while white and brown are responsible for the transport of guanine, a precursor of red pigments (for review, see Ewart and Howells 1998). WHITE (for review, see Hazelrigg 1987), BROWN (Dreesenet al. 1988), and SCARLET (Tearleet al. 1989) share an extensive homology to each other and belong to a superfamily of ATP-binding cassette (ABC) transporters. ABC transporters are found in both prokaryotes and eukaryotes and their function is to translocate various molecules including sugars, amino acids, peptides, metals, ions, toxins, and antibiotics across the cellular membrane (for review, see Croop 1998). All ABC transporters share a conserved domain of ∼200 amino acid residues including an ATP-binding site and a hydrophobic domain comprising five to eight transmembrane segments. On the basis of structural organization ABC transporters are subdivided into three broad categories. WHITE, BROWN, and SCARLET define a distinct group of ABC transporters. Their hydrophilic domain contains an ATP-binding motif followed by the hydrophobic domain. Genetical, molecular, and biochemical studies indicate that these proteins form heterodimers to transport pigment precursors (for review, see Ewart and Howells 1998). In this model, heterodimers between WHITE and SCARLET perform an active transport of tryptophan while WHITE and BROWN form a heterodimer responsible for guanine uptake (Dreesenet al. 1988; Tearleet al. 1989; Ewart and Howells 1998). Thus, white mutants will have defects in transport of both pigments. A leaky allele, white-apricot, retains some white activity, which results in a light eye color. This allows one to detect a wide range of modulation of eye pigmentation, which in turn reflects changes in white expression.

Using white as a target a number of modifiers of gene expression have been isolated. Among them are Wow (Birchleret al. 1994), Mow (Bhadraet al. 1997a), Ufo (Bhadraet al. 1997b), sugarless (Benevolenskayaet al. 1998), Regena (Frolovet al. 1998), and others. Interestingly, a majority of the modifiers exert an effect not only on white but also on brown and scarlet expression (Birchleret al. 1994; Bhadra et al. 1997a,b; Benevolenskayaet al. 1998; Frolovet al. 1998). These observations indicate the existence of coordinated regulation of the expression of three ABC transporters in Drosophila.

In this article we describe the isolation and molecular characterization of the oxen gene, a dosage-dependent modifier involved in regulation of gene expression that is distinct in function from the previously identified ones. Loss-of-function oxen alleles affect the steady-state mRNA level of all three ABC transporters as well as some other unrelated genes, rudimentary, αGpdh, and P0. The gene encodes a 6.6-kD protein homologous to the yeast subunit 9 of mitochondrial ubiquinol cytochrome c oxidoreductase (bc1 complex). The yeast homolog is essential for the assembly of a functional bc1 complex and its deletion perturbs mitochondrial function (Phillipset al. 1990). Recent evidence indicates that modulations of mitochondrial functions will affect nuclear gene expression (Parikhet al. 1987; Poyton and McEwen 1996). Our data establish an additional example of an alteration of mitochondrial biogenesis that leads to changes in nuclear gene expression, thus suggesting a possible mechanism for the cell to adapt to variation in mitochondrial activities.


Fly stocks: Flies were raised on standard Drosophila media at 25°. Genetic markers used here can be found in Lindsley and Zimm (1992).

ms(2)00815, a single P-element insertion on chromosome 2 (49C1-D3), was identified in a screen for dominant autosomal mutations affecting the eye color of white-apricot flies. The same insertion was previously isolated as a male-sterile mutation and named oxen1 (ox1; Castrillonet al. 1993). To mobilize the P element, ox1/CyO; ry506/ry506 males were crossed to the Sp/CyO; Δ2-3, Sb/TM6, Ubx strain (Robertsonet al. 1988). The F1 ox1/CyO; Δ2-3, Sb/ry506 males were crossed individually to three Sp/CyO; ry506/TM6, Ubx females. The Cy, non-Sb, non-Sp progeny (ox1/CyO; ry506/ry506) were screened for rosy- flies, which were mated to Gla/SM6a, Cy balancers to establish a stock. In total, 211 derivatives were generated.

Three oxen alleles, ox44-2, ox107-2, and ox157-1, were recombined with P[ry+; hs-neo; FRT]42D. Presence of both the FRT and oxen mutation was confirmed by complementation tests and G418 resistance. To produce mitotic clones in the adult eye, y w P[ry+; hsFLP]12; P[ry+; hs-neo; FRT]42D P[mw+ NM]46F/CyO males were crossed to y w; ox- P[ry+; hs-neo; FRT]42D/SM6a and the progeny were heat-shocked during the first and second instars for 1 hr at 37° in a water bath (Xu and Harrison 1994). Mitotic mutant clones in y w P[ry+; hsFLP]12; P[ry+; hs-neo; FRT]42D P[mw+ NM]46F/; ox- P[ry+; hs-neo; FRT]42D were w-.

To produce germline clones, first ox107-2 and ox44-2 were recombined with P[mw; FRT]2R-G13. Four independent stocks carrying both FRT and the oxen allele were generated for each deletion. w P[ry+; hsFLP]12; P[mw; FRT]2R-G13 ox-/ovoD P[mw; FRT]2R-G13 females were heat-shocked twice for two hr at the late second to early third larval instar and crossed to ox-/SM6a males (Chou and Perrimon 1996).

For the developmental Northern analysis, ox1/SM6a, Cy females were crossed to T(2;3)CyO, Cy Tb ch translocation males. The F1 males containing this translocation heterozygous with ox1 were mated to Canton-S females. The Tb marker allows discrimination between +/+ and ox1/+ classes at the pupal stages, while the Cy marker allows this distinction in adults.

For P-element transformation, the cDNAs containing complete open reading frames (ORFs) of QCR9 and αNAC were cloned into the pHT4 vector that has a rosy marker gene (Schneuwlyet al. 1987) to generate P[ry+; hsp-QCR9] and P[ry+; hspNAC], respectively. The constructs were injected together with the wings-clipped helper plasmid into ry506 embryos (Spradling and Rubin 1982). One transformed line was established carrying the αNAC transgene and 11 transformants were established for the QCR9 transgene. The DGKε cDNA was cloned into the pUAST vector (Brand and Perrimon 1993) and injected with wings-clipped helper plasmid into y w1118 embryos. Two transformed lines with P[mw+; UAS-DGK] on chromosome 2 were generated. The P[mw+; UAS-DGK] transgenes were then mobilized by supplying the stable source of transposase, Δ2-3, to obtain three independent insertions on chromosome 3. P[mw+; UAS-DGK], P[ry+; hsp-QCR9], and P[ry+; hspNAC] transgenes on the third chromosome were separately recombined with ox107-2 and ox44-2 alleles for rescue experiments.

DNA and RNA techniques: All standard DNA manipulations were performed as described in Sambrook et al. (1989).

The P1 phage 05-71, containing wild-type DNA from the 49D1-2 region on the cytological map (Hartlet al. 1994), was used to obtain an overlapping set of DNA fragments for cDNA library screens and sequencing.

The cDNA library was prepared from 2-week-old male and female wild-type adults (Canton-S) in the λZAP II vector (Stratagene, La Jolla, CA). About 600,000 phage have been screened as described in Sambrook et al. (1989). To clone the QCR9 cDNA, the rapid amplification of cDNA ends (RACE) protocol was performed as described (Frolovet al. 1998). For PCR the following gene-specific primer was used: 5′-GGAACAAATCGGACGTCTTT-3′.

RNA blots, RNA probes, and Northern hybridization were performed according to Frolov et al. (1998).

For sequencing, DNA fragments were cloned into the pSP72 (Promega, Madison, WI) or Bluescript II SK (Stratagene) vector. To obtain nested clones for sequencing, a γ-δ transposon-based system was used (Strathmannet al. 1991). Sequencing was performed on a Sequi-Gen GT nucleic sequencing cell (Bio-Rad, Hercules, CA) using the Sequenase (v.2.0) kit (Amersham, Arlington Heights, IL). Homology searches were performed at the National Center for Biotechnology Information’s BLAST WWW Server.

A region between 1.07 and 1.67 kb on the restriction map (Figure 1) was found to be 99% identical to the 3′ untranslated region of the Drosophila G protein β-subunit gene, Gbe, (accession no. M76593; Yarfitzet al. 1991). Gbe has been mapped to 76C on the polytene chromosome map and was proposed to be a single copy gene. Analysis of Drosophila genomic sequence shows that no Gbe gene sequences reside within the oxen region. Therefore, we concluded that the Gbe cDNA, whose sequence was deposited into GenBank, is a cloning artifact comprising both the sequences of the Gbe gene and the sequences from the oxen region.

In situ hybridization: In situ hybridization with whole mount embryos was performed essentially as described in Tautz and Pfeifle (1989). Riboprobes were prepared by in vitro transcription with DIG RNA labeling mix and detected with alkaline phosphatase conjugated anti-DIG antibody (Boehringer-Mannheim, Indianapolis).


Identification of the oxen gene: A genetic screen was performed to identify the P-element insertions dominantly modulating the expression of a target gene, white. The hypomorphic white-apricot allele that confers an orange-yellow eye color was used. The molecular basis of the white-apricot lesion is an insertion of the retrotransposon, copia, into the second intron (Bingham and Judd 1981). This sensitized white-apricot background allows one to recognize mutations that result in both elevation and reduction of white gene expression. As a heterozygote, one of the P insertions produced a darker eye color and has previously been identified as a male-sterile mutation, oxen (ox1; Castrillonet al. 1993). To gain preliminary information about the nature of ox1 interaction with the white locus, the effect of ox1 on different white alleles was tested (data not shown). In particular, ox1 was found to darken the eye color of point mutations, wcrr and wcf, implying that copia is not required for an interaction to occur. In contrast, the Adh promoter-white structural gene construct (Birchleret al. 1990) did not respond to ox1, which indicates the necessity of the white promoter region for the interaction.

The chromosome carrying the ox1 mutation contains a single P-element insertion at the cytological position 49C1-4. The ox1 allele is sterile over the deficiency Df(2)vg135 (FlyBase, which uncovers the region from 49A4-13 to 49E7-F1 on the polytene chromosome map (Lindsley and Zimm 1992). In addition, the deficiency as a heterozygote darkens the eye color of white-apricot flies as ox1 does, suggesting that ox1 is a loss-of-function allele. To further prove that the insertion is responsible for the mutant phenotype, the P element on the ox1 chromosome was mobilized. Out of 211 established stocks, 122 were homozygous viable lines showing precise excision of the P element in each case. One line, designated as oxrev, was chosen for further study as a control of normal expression of the oxen locus. When tested for the interaction with white-apricot, oxrev shows a reversion of the eye color phenotype.

We sought molecular evidence that white gene expression is affected in the ox1 mutants. Total RNA was isolated from pupae segregating for normal or the ox1 chromosome. Pupae were chosen because the majority of pigment is deposited at this developmental stage. RNA transfers were made in triplicate and hybridized with a white antisense probe. The same blots were then probed with rRNA, which served as a gel-loading control. The phosphorimagery data are presented in Table 1. Consistent with the eye color phenotype, the mutation results in elevation of white transcript levels in mid- and late pupae. The expression of two related ABC transporters, brown (bw) and scarlet (st), was also monitored. It was found that the ox1 mutation affects the expression of both scarlet and brown but the effect is different. The steady-state level of st is increased in females but it is decreased in males. In late pupae, st mRNA is increased in females and slightly increased in males. In contrast, the expression of bw is mostly increased in pupae of both sexes and the effect is stronger compared to scarlet and white. In summary, the ox1 mutation exhibits an effect upon the expression of all three genes. The effect has a complex profile depending upon the sex and the developmental stage.

Transcriptional mapping of the oxen gene: We have previously identified a DGKε gene 123 bp from the P-element insertion site in the ox1 chromosome (M. V. Frolov, E. V. Benevolenskaya and J. A. Birchler, unpublished results). It encodes a homolog of human diacylglycerol kinase ε isoform. To further characterize the oxen region, a set of genomic DNA fragments from a P1 phage, 05-71, (Hartlet al. 1994) covering the P-element insertion site was used to probe a cDNA library. Isolated clones fall into two transcription units, designated as tafazzin and αNAC (Figure 1), based upon their homology. TAFAZZIN shows a similarity to a human family of proteins, TAFAZZINs, responsible for Barth syndrome (Bioneet al. 1996). αNAC encodes a homolog of mouse nascent associated polypeptide complex and coactivator α (Yotov and St-Arnaud 1996).

Two additional transcription units were uncovered from the expressed sequence tag (EST) database made available by the Berkeley Drosophila Genome Project/HHMI EST Project. The ESTs are located in the immediate vicinity of the P-element insertion site and the corresponding transcription units were referred to as HL (EST HL07956) and QCR9 (EST GH22548; Figure 1). The ORF derived from the HL cDNA does not share significant homologies to proteins with known function. To clone the corresponding cDNAs for the HL and QCR9 transcripts, RACE was performed. We were not able to generate a cDNA for HL. In the case of QCR9, a 0.35-kb product was amplified by PCR using a primer complementary to the sequence from the 5′ region of EST GH22548.

View this table:

Effect of ox1 on white, scarlet, and brown transcripts

Generation and analysis of deletions in the oxen region: About 10% of ox1 homozygotes survive to the adult stage, suggesting that the ox1 allele might retain a substantial level of function. Indeed, the P element apparently does not abolish the transcription of any mRNA. To generate stronger ox alleles we looked for imprecise excisions of the P element, which would remove flanking sequences. Southern analysis revealed that in 22 excision events, the sequences outside of the P-element insertion site are deleted. The deletions ranged from 0.3 to 2.6 kb and can be grouped into three classes (Figure 1). Two deletions, ox68-1 and ox86-1, remove the structural sequences of four genes, tafazzin, HL, QCR9, and DGKε, and thus presumably represent null alleles. The second class comprises 11 deletions, which eliminate QCR9 and various portions of both DGKε and HL, while leaving the 5′ sequences of tafazzin unaffected. Two representative examples from the second class, ox44-2 and ox38-1, are shown in Figure 1. Finally, 9 deletions, such as ox12-1, ox107-2, and ox157-1, are associated with the loss of a structural portion of DGKε and QCR9 and presumably do not affect HL and tafazzin. No deletions spanning into αNAC were isolated.

Figure 1.

—Molecular map of the oxen region. (A) Restriction map of the oxen locus. Sites for endonucleases EcoRI (RI), BamHI (B) and HindIII (H) are shown. Site of the P[lacZ, rosy+] insertion in the ox1 allele is designated by a triangle. (B) Positions of several oxen deletions. The sequences deleted in different oxen alleles are shown by open boxes while the positions of the endpoints are denoted with arrows. In the case of deletion ox86-1 the location of the left breakpoint was not determined. (C) The intron-exon structure and direction of the transcripts identified in the oxen region. For the HL transcription unit, no cDNA was isolated. The position of the 5′ end is therefore based upon the sequence of the EST HL07956, while the 3′ end is not mapped. Using a genomic fragment containing the sequences of HL as a probe a single 0.7-kb transcript was found on Northern blots.

The new ox alleles were tested for the lethality, male sterility, and eye color phenotypes. With no exceptions, all 22 deletions exhibited an interaction with wa and were lethal as homozygotes. To determine the stage of lethality, two alleles, ox44-2 and ox107-2, were crossed to the wild-type Canton-S stock. The F1 ox/+ progeny were intercrossed, and the numbers of F2 surviving to embryo, larvae, pupae, and adult stages were counted. For both alleles it was found that the lethality occurred at late first instar larvae with no escapers. Homozygous mutant animals do not exhibit any gross abnormalities, although they show a sluggish response to physical contact compared to the ox heterozygous larvae. Finally, males transheterozygous for ox1 and any of the ox deletions are sterile. Thus, there are no differences in the phenotype whether tafazzin and/or HL transcription units are affected by the deletion.

View this table:

Effect of deletions ox44-2 and ox107-2 on the expression of white mRNA and transcripts from the oxen region in heterozygous (ox/+) adults

To further confirm that neither tafazzin nor HL define the oxen phenotype, we examined an effect of deletions ox107-2 and ox44-2 on the expression of the white gene and transcripts from the oxen region by Northern analysis. As a control for normal expression, the revertant, oxrev, was used. Triplicate RNA transfers were hybridized with antisense probes for white, DGKε, tafazzin, and αNAC (Table 2). Consistent with the eye color phenotype, mutant alleles, ox44-2 and ox107-2, mostly upregulate the steady-state level of white transcripts. The expression of DGKε and QCR9 is decreased to one-half in both deletions relative to the respective controls and restored to normal in the revertant. This is in agreement with the Southern data showing that both deletions remove a portion of DGKε and QCR9. However, the deletions differ in their effect upon tafazzin and HL transcripts. Deletion ox44-2 removes most of the sequences upstream of tafazzin (Figure 1) and decreases its expression. On the contrary, in the case of the ox107-2 allele, tafazzin expression is unaffected. Indeed, the left breakpoint is about 1 kb upstream of the 5′ end of the tafazzin gene and therefore its upstream sequences are presumably intact. Expression of HL is reduced to one-half in the case of ox44-2, in which most of the HL sequences are deleted. On the other hand, the ox107-2 deletion does not extend into the HL transcript and the HL expression is unaffected (Table 2). Taken together these observations disfavor the possibility that tafazzin or HL are responsible for the mutant phenotypes.

Northern data indicate that both deletions increase the level of αNAC expression (Table 2). One possibility to explain this result is that some regulatory sequences of αNAC that might be located within the DGKε gene are eliminated in the mutants. To exclude the possibility that the mutant effects are due to an increased αNAC expression, a full-length αNAC cDNA was expressed under a heat-shock promoter. The overexpression of αNAC after heat shock was confirmed by Northern analysis. Transgenic animals carrying an ox mutant or wild-type alleles in a wa background were heat-shocked twice a day until the late pupal stage. The eye color of emerged flies was compared to that of flies without the transgene. The overexpression of the transgene did not exhibit any effect on the eye color. Therefore, we concluded that αNAC is not responsible for the mutant phenotype. This was further confirmed by the observation that Df(2R)vg135, which deletes αNAC, as well as ox107-2 and ox44-2, which retain αNAC, have a similar effect on wa.

Thus, the above results left DGKε and QCR9 as remaining candidates. According to the Southern analysis, each of the 22 deletions removes a portion of both DGKε and QCR9 and therefore potentially represents a null allele for both genes. Hence, the deletion tests do not discriminate between the two transcripts. However, the P-element insertion located between the 5′ ends of DGKε and QCR9 might differ in the effect upon their expression. Indeed, as revealed by sequence analysis, the insertion site is located 15 bp upstream of the 5′ end of QCR9 and 123 bp upstream of the putative transcription start site of DGKε. Therefore the expression of DGKε and QCR9 in heterozygotes segregating for the ox1 chromosome was monitored. To study the expression in the ox1 homozygotes, RNA was isolated from rare surviving flies, which do not carry the SM6a balancer marker Cy and, therefore, are homozygous for ox1. In the case of the mutation, the steady-state DGKε mRNA level is slightly decreased to 0.90 ± 0.02 and 0.81 ± 0.03 in the mutant heterozygotes and homozygotes, respectively, compared to the controls (Figure 2). On the contrary, the effect on QCR9 is much more profound. The QCR9 transcripts are decreased to 0.75 ± 0.01 in the ox1/+ mutants and to 0.45 ± 0.03 in flies homozygous for the ox1 chromosome (Figure 2). This is consistent with the fact that the P element is inserted 15 bp upstream of the the 5′ end of the QCR9 gene, while the distance between the site of the insertion and the 5′ end of DGKε is 123 bp. Thus, the level of DGKε expression is higher in ox1/ox1 homozygotes than in animals heterozygous for the oxen deletions, ox44-2 and ox107-2 (see above). This is in apparent contradiction with the stronger phenotype of ox1/ox1 flies compared to ox-/+, such as semilethality and male sterility. To directly prove that DGKε does not define the oxen function, a UAS-DGKε transgene was constructed, transformed, and expressed using a heat-shock GAL4 driver, hsp70-GAL4 (Brand and Perrimon 1993). The overexpression of the DGKε transgene does not rescue the lethality or male sterility. Because hsp70 is not efficiently expressed in testes, another GAL4 driver, 32B (Brand and Perrimon 1993), was used. This driver was able to provide a high expression level in testes in the control experiments using a UAS-lacZ transgene (data not shown). It was found that UAS-DGKε does not rescue the mutant phenotype with the 32B driver. Similar results were obtained with three different lines containing the UAS-DGKε transgene at various chromosomal locations.

Figure 2.

—Northern analysis of QCR9 and DGKε. Hybridization of a blot containing total RNA isolated from flies segregating for mutant (ox1) and wild-type (+) oxen alleles with antisense RNA probes for QCR9 and DGKε. rRNA was used as a gel-loading control.

Thus, the combined data on the expression of the tafazzin and HL transcripts in the case of the deletions and the facts that the DGKε and αNAC transgenes do not rescue the mutant phenotype disfavor the possibility that any of them represents the oxen gene. On the other hand, QCR9 is the only transcript whose expression is significantly affected by the oxen deletions and the hypomorphic allele ox1.

To unambiguously show that QCR9 represents the oxen locus, the QCR9 cDNA was expressed under the control of the heat-shock promoter in transgenic animals as described in materials and methods. Six independent insertions on chromosome 3 were tested with the oxen mutation, which is on chromosome 2. The mutant eye color phenotype of the heterozygous ox1 allele was completely rescued by three of six tested QCR9 transgenes, hsp-QCR9#106, hsp-QCR9#39, and hsp-QCR9#99, when the heat shock was administered once a day at larval and pupal developmental stages. Partial rescue was observed with the three other tested transgenes, hsp-QCR9# 1, hsp-QCR9#31, and hsp-QCR9#19. Consistently, the two strongest transgenes, hsp-QCR9 #106 and hsp-QCR9#99, rescued the lethality of the ox107-2 allele following the heat-shock treatment, while the weaker transgene hsp-QCR9#1 did not. In the control experiment without heat shock, no ox107-2 homozygotes were recovered. Taken together these data argue that QCR9 represents the oxen function.

Requirement of QCR9 activity in development: Several experiments were employed to examine the role of QCR9 in development. As mentioned above, any QCR9 null allele results in early larval lethality. To determine what the consequences of the loss of QCR9 might be in adult patterning, a mosaic analysis was performed. We employed the FRT/FLP system (Xu and Harrison 1994) to generate homozygous ox107-2 and ox44-2 mutant somatic clones. Following the induction of FLP recombinase in the first or second instar larvae, numerous duplication clones (+/+) marked with a mini-white gene were seen in adult eyes. In contrast, no mutant clones lacking the mini-white gene were found. These results indicate that QCR9 is required for cell viability.

Next we addressed the question of whether QCR9 is necessary for oogenesis. To generate homozygous ox44-2 and ox107-2 mutant germline clones, the “FLP-DFS” technique was used (Chou and Perrimon 1996). The mosaic females depleted of oxen function laid no eggs. Examination of the associated ovarian phenotype revealed that oocyte development is arrested before onset of vitellogenesis. Therefore, the presence of QCR9 is presumably required either to enter vitellogenesis and/or for cell viability in late oogenesis.

To further elucidate the QCR9 function, the pattern of its expression in development was investigated by in situ hybridization. Wild-type embryos were hybridized to a digoxigenin-labeled RNA probe of QCR9. QCR9 transcripts are ubiquitously present in early embryos at the time of cellularization (Figure 3A) and throughout germ-band extension stages (Figure 3, B and C). By the end of germ-band retraction, expression is enriched in amnioserosa and in midgut (Figure 3D). As dorsal closure proceeds, QCR9 transcripts are predominantly found in midgut, while the staining in other tissues weakens (Figure 3, E and F). During late development when midgut constrictions appear, the transcripts are present in all four midgut chambers (Figure 3G). At the end of embryogenesis, robust staining persists in the midgut (Figure 3H).

An effect of oxen deletions on gene expression: The question of whether the oxen function is restricted to the the regulation of the expression of Drosophila ABC transporters was addressed. The steady-state mRNA level of seven unrelated genes, β1-tubulin at 56D (β-tubulin), Glucose-6-phosphate dehydrogenase (Zw), α-Glycerol-3-phosphate dehydrogenaseGpdh), rudimentary (r), Alcohol dehydrogenase (Adh), ribosomal protein P0 (P0), and ribosomal protein 49 (rp49), was examined by Northern analysis in ox107-2 mutants (Table 3). As a control of normal expression, the oxrev allele was used. The expression of two genes, αGpdh and P0, is increased in the case of mutation in both sexes and is restored to normal in the revertant. On the contrary, the oxen deletion results in elevation of r transcripts in females and their reduction in males. The steady-state Adh mRNA level was found to be elevated in mutant males. However, examination of the Adh expression in oxrev flies revealed a similar effect suggesting that the observed effect on Adh transcripts is not caused by the ox107-2 allele. Finally, three genes, β-tubulin, Zw, and rp49, are not affected by the oxen deletion. The same set of genes was affected by deletion ox44-2 (data not shown).

Figure 3.

QCR9 expression pattern in embryogenesis. (A-C) QCR9 is ubiquitously expressed at early stages of development. Uniform expression in the yolk of early stage 5 (A) and during germ-band extension at stage 9 (B) and stage 11 (C) embryos. At stage 13, staining is present in amnioserosa and in the midgut (D). From the time of dorsal closure and until the end of embryogenesis, QCR9 transcripts are strongly expressed in the midgut at stage 14 (E), at stage 15 (F), at stage 16 (G), and at stage 17 (H) embryos.

View this table:

Effect of oxen deletion, ox107-2, on gene expression

QCR9 encodes a component of the mitochondrial bc1 complex: Sequence analysis of the QCR9 mRNA revealed that it encodes a 6.6-kD protein that shares extensive similarity throughout the whole sequence with the smallest protein of the mitochondrial ubiquinol-cytochrome c oxidoreductase complex (bc1 complex) from bovine (Schaggeret al. 1983) and yeast (Phillipset al. 1990; Figure 4). Drosophila QCR9 is 57 and 36% identical to bovine and yeast homologs, respectively. Inspection of the first 15 amino acids of the QCR9 product shows features necessary for the protein to be transported to mitochondria (von Heijne 1986). Indeed, there is a high content of Arg, Lys, Leu, Ser, and Thr, while no acidic residues are present.


The P-element mutation in the oxen gene was isolated in a screen for trans-acting modifiers exhibiting a dominant effect on the white-apricot phenotype. Our and previously published data (Castrillonet al. 1993) demonstrate that the oxen gene is essential for cell viability, spermatogenesis, and oogenesis. Mutations in the locus are haplo insufficient for the expression of a subset of Drosophila genes, in particular the ABC transporters, white, brown, and scarlet. We provide evidence that the oxen locus encodes a structural homolog of the smallest subunit of mitochondrial ubiquinol cytochrome c oxidoreductase (bc1 complex).

In the course of transcriptional mapping, mRNAs from five closely arranged genes, tafazzin, HL, QCR9, DGKε, and αNAC, were identified in the oxen region. Several lines of evidence indicate that the effects seen in the oxen mutants are due to a lesion in QCR9. First, the hypomorphic allele ox1, which has a P-element insertion upstream of QCR9, can be reverted by precise excision of the P element. In addition, Df(2R)vg135, which uncovers this region, exhibits a similar effect on white-apricot. Second, the mutant phenotype is retained whether tafazzin and/or HL sequences are uncovered in various deletions within the region. This excludes both tafazzin and HL as a source for oxen activity. Third, αNAC is upregulated in all three tested oxen mutant alleles; however, overexpression of the αNAC transgene does not cause the mutant phenotype. Neither does it exhibit an effect on the eye color in a white-apricot background. This leaves DGKε and QCR9 as possible candidates. Indeed, all of the oxen deletions eliminate at least the 5′ portion of both genes. This is confirmed by a Northern analysis, which shows a twofold reduction of DGKε and QCR9 expression in the case of two oxen deletions. Unlike the deletions, the P-element insertion in the ox1 allele preferentially affects the level of QCR9 rather than DGKε transcripts. This difference is most likely due to the fact that the insertion site is 15 bp upstream of the putative 5′ end of QCR9 while the distance to the DGKε 5′ end is 123 bp. Finally, the QCR9 transgene completely rescues the eye color phenotype and lethality associated with the oxen mutations. On the contrary, DGKε transgenic flies do not show the effect upon the eye color or reversion of lethality. Taken together, these data indicate that QCR9 defines the oxen function.

Figure 4.

—Multiple alignment of the amino acid sequences of Drosophila QCR9 and subunit 9 ofm yeast and subunit 11 of bovine bc1 complexes. Amino acids that are identical between QCR9 and one or more of other proteins are shaded. The alignment was made by use of the computer program MegAlign from DNA STAR.

The predicted product of QCR9 shows a sequence similarity to the smallest subunit of mitochondrial bc1 complex from yeast and bovine. The cytochrome bc1 is one of the three major respiratory enzyme complexes residing in the inner mitochondrial membrane. The enzyme oxidizes ubiquinol, which reacts from the membrane phase, reduces cytochrome c in the intermembrane space, and uses the free energy change to transport two protons across the membrane from the matrix to the intermembrane space, and releases two additional protons there (Saraste 1999). The bovine and yeast bc1 complexes contain 11 and 9 subunits, respectively. Only 3 subunits carry the redox centers. These key polypeptides are cytochrome b, the Rieske FeS protein and cytochrome c1 (Saraste 1999). Most of the other subunits are small proteins and their role is not well understood. Genetic studies in yeast identified those essential for assembling a fully functional complex. It was found that deletion of a yeast homolog of Drosophila QCR9 results in a 20-fold reduction of the activity of the bc1 complex (Phillipset al. 1990). The generation of QCR9 mutant Drosophila lines provides a tool for examining the consequences of the loss of QCR9 in a multicellular organism. Our data are consistent with the proposed indispensability of the yeast homolog in the cell. Flies homozygous for QCR9 null alleles die as first instar larvae. Moreover, no homozygous mutant clones were observed in the eye indicating that the QCR9 is required for cell viability. It is possible that the survival of homozygous animals to the first instar larval stage is due to a maternal contribution of QCR9 from heterozygous mothers. However, because of the requirement of the QCR9 function for oogenesis, the possibility of recovering homozygous individuals from a homozygous female germline lineage is eliminated and does not permit a test of this potential explanation. Oocyte development is arrested before vitellogenesis if the germline is depleted of QCR9. Thus, our data suggest that QCR9 is an essential gene in Drosophila and its function is required in every examined organ.

Several genes encoding mitochondrial proteins have been identified in Drosophila (Haywardet al. 1993; Hartensteinet al. 1997; Iyengaret al. 1999; Zhanget al. 1999). Interestingly, mutations in three of them result in nervous system defects. The sluggish-A gene encodes a mitochondrial proline oxidase involved in proline biosynthesis and depletion of its activity causes locomotory abnormalities (Haywardet al. 1993). The tamas gene was isolated in a screen for mutations affecting larval response to light. The tamas product shows a significant similarity to the mitochondrial DNA polymerase catalytic subunit (Iyengaret al. 1999). Larvae homozygous for the colt mutation exhibit a sluggish response to physical contact. colt encodes a member of the mitochondrial carrier family (Hartensteinet al. 1997). It is worth noting that oxen mutant homozygous larvae show defects similar to colt mutants. This may indicate that the nervous system is more sensitive to perturbations in mitochondrial function, reflecting a higher energy requirement. Indeed, altered mitochondria were found in all cases of hereditary motor and sensory neuropathy with optic atrophy and in several cases in a larger group of unselected neuropathies (Schroder 1993).

Why was oxen isolated in a screen for modifiers of white gene expression? As mentioned above, white together with brown and scarlet encode ABC transporters. Yeast mitochondrial ABC transporter ATM1 was proposed to be involved in the signaling from the mitochondria to the cell (Leighton and Schatz 1995), so a modulation of this gene family is potentially occurring. There are a growing number of observations that changes in mitochondria and, in particular, in the respiratory chain are followed by alterations in nuclear gene expression. This retrograde regulation has been hypothesized to be a general mechanism for the cell to monitor and adjust its functions in response to perturbations in mitochondrial biogenesis (Parikhet al. 1987; for review, see Poyton and McEwen 1996). One of the best-studied examples of retrograde regulation is the induction of the peroxisomal isoform of citrate synthase, CIT2, in cells with dysfunctional mitochondria (Liaoet al. 1991). Remarkably, CIT2 induction was observed after addition of antimycin A, which inhibits the respiratory chain (Liaoet al. 1991). Given an absolute requirement of yeast QCR9 for assembling a functional bc1 complex and cell lethality caused by the loss of QCR9 in yeast and Drosophila, it is likely that mitochondrial respiratory functions are compromised in oxen mutants. Although there is increasing evidence in favor of retrograde regulation, the molecular mechanism underlying this phenomenon remains largely unknown. Because the changes in gene expression occur when the oxen mutation is heterozygous, it is likely that mitochondrial perturbations contribute to aneuploid syndromes. The oxen mutational effects provide additional evidence for the existence of signal communication between mitochondrial and nuclear genomes and provide a model genetic system in a multicellular organism to study this phenomenon.


We are grateful to the Indiana University Drosophila Stock Center for providing fly strains, Daniel Hartl for the P1 clone, Maria Nurminskaya and Dmitry Nurminskii for help with the RACE, Steve Wasserman for sharing the results regarding the ox1 allele, and Norbert Perrimon and Walter Gehring for transformation vectors. Special thanks to Kathy Newton, Olga Karpova, Eugene Kuzmin, and members of the Birchler lab for discussion and critical comments. This study was supported by a National Science Foundation grant to J.A.B. M.V.F. was partially supported by a postdoctoral fellowship from the Molecular Biology Program at the University of Missouri-Columbia. The accession number of the sequence reported here is AF017783.


  • Communicating editor: K. Golic

  • Received March 14, 2000.
  • Accepted July 31, 2000.


View Abstract