Linking naturally occurring genotypic variation to the organismal phenotype is critical to our understanding of, and ability to, model biological processes such as adaptation to novel environments, disease, and aging. Rarely, however, does a simple mutation cause a single simple observable trait. Rather it is more common for a mutation to elicit an entangled web of responses. Here, we employ biochemistry as the thread to link a naturally occurring two amino acid deletion in a nuclear encoded mitochondrial protein with physiological benefits and costs in the fly Drosophila simulans. This nuclear encoded gene produces a protein that is imported into the mitochondrion and forms a subunit of complex IV (cytochrome c oxidase, or cox) of the electron transport chain. We observe that flies homozygous for the deletion have an advantage when young but pay a cost later in life. These data show that the organism responds to the deletion in a complex manner that gives insight into the mechanisms that influence mitochondrial bioenergetics and aspects of organismal physiology.
When Darwin proposed that natural selection operated on variation, he had no idea what material produced the observable differences. Ninety-one years later, Hershey and Chase's experiments with the T2 phage identified the source of inherited variation as DNA. Today, in the postgenomics era, biologists are still struggling to elucidate the details of the interaction between genetic variation and phenotypic differences (Gregersen 2009). One explanation for the difficulty is that the organism is more than a simple sum of its nucleotides. Indeed, it may now be argued that Darwin's entangled bank analogy applies equally to genes within a single organism as it does to species within a community. However, for this analogy to be robust we must replace birds, insects, and worms with differential splicing/expression, posttranslational protein modification, and mobile elements.
The broader goal of this study is to use biochemistry as a link between the genotype and the phenotype. This approach has provided significant evolutionary insight into adaptation at specific loci and facilitated understanding of complex biological problems of broad evolutionary interest. One notable example is the alcohol dehydrogenase (Adh) gene–enzyme system in Drosophila melanogaster. Adh presents two electromorphs that differ in their biochemical properties. Fast may be broadly described as being more active, whereas Slow is more thermostable. Adh systematically shows clines where the Fast allele is predominant at higher latitudes. This system has provided insight into important evolutionary problems including (1) an understanding of the mechanisms by which way an organism adapts to a novel environment (Chambers 1988, 1991), (2) evidence for the action of natural selection (Kreitman 1983; Hudson et al. 1987), (3) a model system for the study of eukaryotic gene expression (Chambers 1991), and (4) the natural history of fruit flies (Veuille et al. 1998). A second example, considers Colias butterflies with different allelic combinations (electromorphs) of the glycolytic enzyme phosphoglucose isomerase. In this case electromorphs exhibit distinct kinetic and thermal stability properties, leading to variation in flight performance (Watt et al. 1983) and associated fitness components (Watt et al. 1986; Watt 1992).
The specific goal of the study is to investigate the physiological and bioenergetic benefits and costs of a two amino acid deletion (ΔTrp85, ΔVal86) in the fly Drosophila simulans. The deletion occurs in subunit 7A of the mitochondrial electron transport chain protein cytochrome c oxidase (cox7A) (Melvin et al. 2008). The nuclear encoded cox7A gene produces a protein that is imported into the mitochondrion and forms a subunit of complex IV (cytochrome c oxidase, or cox) of the electron transport chain. Cox represents the terminal complex of the respiratory chain, and is hypothesized to be a control point for the rate of electron flow through the entire chain (Villani et al. 1998). Here, we extend our previous studies and show that the organism responds to the deletion in a complex manner that gives insight into the mechanisms that influence mitochondrial bioenergetics and aspects of organism physiology.
Melvin et al. (2008) modeled the tertiary structure of cox7A by homology to the bovine cox7A structure and predicted the mutation would decrease the function of complex IV. This prediction led to the hypothesis that flies with the deletion would have lower cytochrome c oxidase activity. This result was observed (Ballard et al. 2010). To test whether the mutation was linked to a specific mutation found elsewhere in the genome, the mutation was genetically placed into outbred genetic backgrounds of flies from Hawaii and Africa. Enzyme activity data gathered from the outbred lines supported the hypothesis derived from modeling that the deletion was causally responsible for a reduction in cox activity (Ballard et al. 2010). It was not clear, however, whether heterozygotes would be intermediate or more similar to one of the homozygotes so these data are included here.
We hypothesized that a reduction in cox activity in homozygous cox7A mutant flies would affect the number of offspring produced between genotypes. Fertility was chosen as an index of competitive ability because it is relatively easy to measure and changes in fertility are expected to affect population size and the outcome of competition (Aiken and Gibo 1979). Furthermore, reproduction is reported to be influenced by oxidative stress but the mechanism by which this occurs is not well established (Ballard et al. 2007; Rush et al. 2007). Plausibly, flies harboring the deletion may suffer from reduced ATP production and a lowered fertility. Mutations in mitochondrial genes are frequently characterized by dysfunction of energy production (Smeitink et al. 2001) and have been associated with disease in humans and in model organisms (Wallace 1999; Ventura et al. 2006). Alternatively, the organism may compensate for the decreased cox activity in mutant flies. Mitochondria are highly plastic and may respond to reduced OXPHOS efficiency by producing more mitochondria and more OXPHOS complexes (Hood et al. 2006). This compensation hypothesis is supported by our previous results showing that flies harboring the deletion had elevated mRNA expression of genes encoding the protein subunits of complex I, III, IV, and V (Ballard et al. 2010).
We quantified the bioenergetic efficiency of the mitochondrial electron transport chain in homozygous mutant and normal flies by measuring the quantity of oxygen consumed by mitochondria when provided with a quantity of ADP (ADP:O ratio) (Nicholls and Ferguson 2002). From each mitochondrial preparation, we measured ADP:O ratio of the organelle respiring on sn-glycerol 3-phosphate, a substrate that feeds electrons into complex III of the electron transport chain. The overall rate at which mitochondria can use complex III substrate makes it the preferred substrate when high energy yield is needed quickly as for flight (Sacktor 1976).
Mitochondria are known to influence aspects of organismal physiology including survival in Drosophila (Rand et al. 2006). Here we assay physical activity and survival of the cox7A fly lines as possible physiological costs of the mutation. We predict that flies homozygous for the cox7A mutation will have reduced physical activity compared to that of homozygous normal or heterozygote flies. In humans, running performance of trained athletes declines from young adulthood to early middle age and is correlated with decreased aerobic capacity and ATP production (Tanaka and Seals 2003; Short et al. 2005). The same general trend is found in a variety of insects. Physical function declines with age in D. melanogaster (Minois et al. 2001; Martin and Grotewiel 2006) and physical activity of D. simulans declines with age (James and Ballard 2003). Alternatively, physical activity may be maintained in the face of declining bioenergetic capacity by the amplification of energy-producing mitochondria (Melvin et al. 2007).
We predict that homozygous mutant flies will have decreased survival because the cox7A mutation creates a bottleneck in the electron transport chain that results in increased residence time of electrons at sites upstream from cox. Plausibly, this would increase the likelihood of reactive oxygen species (ROS) production. The oxygen free radical theory of aging predicts that long-lived animals accumulate damage to biomolecules from ROS at a slower rate than do shorter-lived animals (Harman 1956, 1992; Sastre et al. 1996, 2000). We assayed the rate of hydrogen peroxide (H2O2, a form of ROS) production by mitochondria respiring on complex III substrates. In D. melanogaster both complexes I and III are known to produce superoxide radicals (Miwa et al. 2003) with H2O2 produced from the dismutation of mitochondrial superoxide.
The overall efficiency of OXPHOS may also be influenced by proton leak across the inner mitochondrial membrane. Proton leak is both a major contributor to resting energy expenditure and has been proposed to play a role in thermogenesis, protection from ROS generation, and regulation of carbon fluxes (Skulachev 1996; Rolfe and Brown 1997; Brand 2000). We predict that flies homozygous for the cox7A deletion mutation will have higher proton leak. Proton leak is defined as conductance of protons from the intermembrane space across the inner mitochondrial membrane to the matrix independent of ATP production. The driving force of proton leak is the potential gradient across the inner mitochondrial membrane (membrane potential) created during respiration when protons are pumped from the matrix to the intermembrane space. When ATP production is inhibited, the rate of proton leak is proportional to the rate of respiration and the relationship of proton leak to membrane potential can be measured by titration with respiration inhibitors.
In this study, we explore the consequences of a two amino acid deletion mutation in a nuclear encoded gene that produces a protein that is imported into the mitochondrion. We observe that flies homozygous for the mutation have an advantage when young but pay a cost for harbouring the mutation later in life. The mutation causes a reduction in cox activity and this was compensated for by an increase in mtDNA copy number and ADP:O ratio that is correlated with higher early fertility. The rates of H2O2 production and proton leak were higher in flies with the deletion and this was correlated with reduced physical activity between homozygotes on day 18 and reduced mean survival. This complex response to a simple two amino acid deletion reinforces the difficulty of making causal connections between the genotype and the phenotype when biochemical, molecular, or proteomic intermediates are not studied.
Materials and Methods
The D. simulans flies were constructed from a heterozygous isofemale line, HW01, collected in Honolulu, Hawaii in November of 2004. For this study, our goal was to isolate genotypes that were either homozygous for the presence or the absence of the six-base deletion in cox7A but were as identical as possible at all other loci. To achieve this goal, we employed a six-step strategy (Ballard et al. 2010). Briefly, (1) the wild-caught isofemale HW01 fly line was inbred for five generations to reduce heterozygosity, (2) the five generation inbred line was maintained in population cages in the laboratory for 12 months to allow recombination to reduce the linkage block around cox7A deletion, (3) flies were treated with tetracycline to cure them of Wolbachia infection (Hoffmann et al. 1986), (4) we constructed five 11-generation inbred lines by sibling mating, (5) pairs of homozygous normal and homozygous mutant lines were constructed, and (6) two pairs of 12-generation inbred lines that showed 50% starvation resistance most similar to the five-generation inbred HW01 isofemale line were included. The goal of the final step was to expose any slightly deleterious mutations that may have accumulated during inbreeding or were linked with the deletion. In this study, we do not report on the cox activity following genetic transfer of the mutation to multiple other backgrounds. This important step was conducted in a previous study (Ballard et al. 2010). Below, we do report on the body protein content in three outbred fly lines.
Prior to experiments, cox7A undeleted and deleted flies were raised for two generations at a constant density in 250-mL bottles on instant Drosophila media (Carolina Biological, Burlington NC) at 23 ± 1 °C, 50% relative humidity, and 12 h light: 1 h dark daily cycle (Hercus and Hoffmann 2000). The use of a commercially available Drosophila media enables independent studies to be conducted in different laboratories. To produce flies for experiments, parent flies were released into population cages that contained solid oviposition resources (4% agar, 10% molasses supplemented with yeast paste) and were allowed to lay eggs for 4 h. Eggs were collected by the method of Clancy and Kennington (2001) and placed into 250-mL bottles containing instant Drosophila food at a density of about 200 flies/bottle.
Experimental flies homozygous for either the deletion or normal allele of cox7A were collected directly from their respective homozygote parent lines. Experimental heterozygote lines were constructed afresh for each study by crossing homozygote normal and homozygote deleted lines. To avoid female bias, experimental heterozygote flies were collected from reciprocal crosses; one in which the female was homozygote normal and another in which the female was homozygote deleted. Equal numbers of flies from the crosses were pooled for assays.
Fertility studies showed a significant difference in rates of reproduction and, as this bears an energetic cost, virgin females were used in all subsequent assays (Chapman et al. 1995; Harshman and Zera 2007). Unless otherwise stated, virgin females were assayed at four days and 18 days of age because the cox7A genotypes do not differ in survival at these ages. Flies were aged in standard demography cages (Tu et al. 2002). Equal numbers of flies were removed from five replicate cages and pooled for experiments to avoid cage effects.
CYTOCHROME C OXIDASE (COX) ACTIVITY
Previously, we have shown that cox activity was significantly lower in flies homozygous for the deletion (Ballard et al. 2010). Here we compare the cox activity of heterozygotes with that of homozygotes. Cox activity extracts from 15 replicate single thoraces of inbred virgin female flies aged four days and 18 days were tested as previously described (Ballard et al. 2010). Briefly, individual thoraces were thawed on ice and single thoraces were homogenized for 10 s in 100 μL of ice-cold homogenization buffer (50 mM NaH2PO4, pH 7.1 containing 0.05% Tween-80) using a Kontes pellet pestle motor. Homogenates were diluted to 200 μL with 50 mM NaH2PO4, pH 7.1, and cell debris was removed by centrifugation. The supernatant was diluted 1:20 and 40 μL aliquoted into four microplate sample wells. Assays of tissue extracts that interrogate the maximum rate at which the cox enzyme reduces cytochrome c were initiated by adding 160 μL of reduced horse heart cytochrome c (50 μM) to each sample well following Melvin and Ballard (2006). Data were analyzed by nested analysis of variance (ANOVA, JMP 1995).
BODY PROTEIN CONTENT
To test the potential for the mutation to influence a physiological trait, we assayed the whole organism trait of body protein content in three outbred fly lines as previously described (Ballard et al. 2010). Briefly, unmated homozygous mutant males were crossed with outbred virgin females from three fly lines (KY17 from Kenya), HW07 (from Hawaii), and HW05 (from Hawaii). The F1 heterozygotes from this cross were sibling-mated and homozygote normal and mutant flies were identified in the F2 offspring by restriction fragment length polymorphism. These lines were selected because all cox loci had previously been sequenced (Melvin et al. 2008). Thorax extracts prepared for determining cox activity were assayed for protein content using a Bio-Rad DC protein assay kit (Richmond, CA).
PHYSIOLOGICAL BENEFITS OF THE MUTATION
Changes in fertility have been shown to affect population size in interspecific Drosophila competition studies (Aiken and Gibo 1979). We hypothesized that the number of offspring may differ between genotypes. Fertility of 10 singly mated four-day female flies from each pair of 12-generation inbred lines for each cox7A genotype were assayed over a 10-day period. Fifteen newly eclosed, virgin females and males of each line were collected. The newly collected females were housed in individual vials and males were kept in one vial on instant Drosophila media. After four days, females were confirmed to be virgin by checking for the absence of larvae in the individual vials. Each of the 60 females was observed to mate once with a single male and was placed in a new vial containing fresh food each day for 10 days. Emerging flies were counted after 12 days. An alternate assay would have been to count the number of eggs. We did not take this alternate approach because egg production can be a biased estimator of fitness if a substantial number of unfertilized eggs are laid. This assay does not quantify the lifetime fitness of each genotype but does reflect the conditions routinely used in the laboratory. Data were analyzed by nested ANOVA (JMP 1995).
BIOENERGETIC BENEFITS OF THE MUTATION
Mitochondrial DNA copy number
To investigate the hypothesis that flies harboring the deletion compensated for the reduced cox activity, we measured mtDNA copy number. In humans, mtDNA copy number tends to decline with age but can be increased to meet bioenergetic demands (Moraes 2001; Short et al. 2005). Relative mitochondrial DNA copy number was inferred from the ratio of the mitochondrial gene for cytochrome c oxidase I (coI) to the nuclear gene cytochrome c oxidase subunit 5A (cox5A). Here, we infer that mtDNA copy number correlates directly with cytochrome c oxidase I expression level (Passos et al. 2007). The frequency of deleted mtDNA genomes in flies aged less than 37 days is very low (Yui et al. 2003). Previously, we have shown that cox5A expression is 13% higher in 4-day and 5% higher in 18-day mutant flies compared to normal flies of the same ages when normalized to the ubiquitously expressed housekeeping gene Actin 42A (Ballard et al. 2010).
DNA was extracted from single thoraces of flies aged four days and 18 days post emergence. Flies of the same line and age were immobilized on ice, thoraces were separated from heads and abdomens, flash frozen in liquid nitrogen, and stored at −80°C. DNA was extracted using a PureGene DNA extraction kit (Gentra, Minneapolis, MN). Real-time quantitative Taqman PCR analysis was performed on an ABI7000 Sequence Detection System (Applied Biosystems, Foster City, CA). The 20-μL reactions contained 10-μL Taqman Universal PCR Master Mix, 2-μL template DNA, 1 μL each of forward and reverse primers (each 18 μM), and 0.5 μL of the 6-FAM labeled Taqman probe (5 μM). The reaction was amplified for 40 cycles (15 s denaturing at 95°C and 1 min annealing/extension at 60°C). Each DNA sample was amplified in triplicate for both genes. The primers for the coI gene were coI-375F (AACTGTTTACCCACCTTTATCTGCTG), coI-458R (CCCGCTAAGTGTAAAGAAAAAATAGC), and Taqman probe coI-404–6FAM (TTGCCCACGGTGGAGCTTCAGTTC). No nuclear copy of coI was found in the simulans_mosaic_050411 (June 2, 2005). The nuclear gene was cox5A. As described, primers were cox5A-206F, cox5A-322R, and probe cox5A-286–6FAM (Ballard et al. 2010).
Mitochondrial DNA copy number was calculated from cycle threshold (Ct) as 2−ΔCt where ΔCt = Ct(coI)-Ct(cox5A) and the level of mtDNA relative to the mean mtDNA level of four-day old cox7A homozygous flies was analyzed by nested ANOVA (JMP 1995).
We observed differences in mtDNA density between genotypes and hypothesized that the ADP:O ratio of genotypes may differ. Intact mitochondria were isolated from flies aged four days and 18 days using the procedure of Melvin and Ballard (2006). Briefly, 50 flies of the same age and cox7A genotype were immobilized on ice and thoraces were separated from heads and abdomens and the mitochondria extracted. Thoraces were used in biochemical assays because the thorax muscles represent the major muscle mass of the adult fly and because tissue-specific differences in mitochondrial bioenergetics have been demonstrated in D. simulans (Katewa et al. 2006). All assays using intact mitochondria were conducted within 3 h of isolation. The respiratory control ratios (RCRs) of preparations were 3.8 ± 0.33 SE. RCR is the ratio of oxygen consumption in the presence of ADP to that in the absence of ADP. For the complex III substrate sn-glycerol 3-phosphate used in this study, an RCR greater than 3.0 indicated that mitochondria were intact and that oxygen consumption was linked to the phosphorylation of ADP to ATP.
From each mitochondrial preparation, we measured ADP:O ratio using the complex III substrate sn-glycerol 3-phosphate (5 mM) using the procedures of Melvin and Ballard (2006) and Katewa and Ballard (2007). Briefly, respiration rate was measured using a Clark-type oxygen electrode in a 3-mL incubation chamber (Rank Brothers, Cambridge, UK) maintained at 25°C and assuming 479 nmol atomic oxygen/mL at air saturation. ADP:O was calculated as the quantity of ADP (120 nmole) added to the respiration chamber to initiate state 3 respiration divided by the quantity of atomic oxygen consumed during state 3 respiration (Estabrook 1967). ADP:O data were analyzed by nested ANOVA (JMP 1995).
PHYSIOLOGICAL COST OF THE MUTATION
Survival of virgin female flies of each cox7A genotype was determined at 23°C. Newly emerged flies were collected at 2-h intervals, females were separated from males under light CO2 anesthesia, and placed into five replicate demography cages provided with instant Drosophila media. Deaths were counted, dead flies were removed, and fresh medium was provided every two days. Deaths during the first interval after transfer to demography cages were not included in calculations to avoid confounding deaths due to handling with “natural” deaths. Initial cohort size was calculated as the summed death observations across all ages beginning at the second two-day interval. In the first two-day interval, two homozygous cox7A deleted, zero heterozygous, and zero cox7A normal flies died. Initially, there were a total of 416 homozygote normal, 461 heterozygote, and 419 homozygote-deleted flies.
Survival functions were estimated and compared using JMP statistical software (JMP 1995). We tested the null hypothesis that survival function is the same for all three lines using a Log Rank test for homogeneity between groups. The age at 50% survival was calculated from the survival data and compared by ANOVA. The best fit of mortality rate data to models in the Logistic-Makeham-Gompertz set was determined by maximum likelihood methods for each line. The purpose of this analysis was to determine the divergence of mortality between lines. Mortality function variables were compared using WinModest software (Pletcher 1999).
Physical activity is influenced by both metabolic rate and life span (Sohal et al. 1993; Berrigan and Partridge 1997). We determined the physical activity of virgin female flies at four days and 18 days using a Trikinetics (Waltham, MA) physical activity monitor. Five female flies of each genotype were anaesthetized with humidified CO2 and placed into individual 5-cm-long glass tubes that contained 2.5 mm of instant Drosophila media at one end. After inserting a fly a cotton plug and positioned to restrict the end-to-end distance inside the tube to 4.5 cm. Tubes were placed in the activity monitor horizontally inside a 23 ± 1°C incubator with 12 h light: 12 h dark daily cycle. Flies were allowed 10 h to recover from anesthesia and acclimate to the tubes. Assays were run over an eight-day period with at least two replicates per line included on each assay day. The number of times a fly crossed an infrared light beam bisecting the glass tube was recorded by DAM software (Trikinetics) and the data were analyzed by nested ANOVA (JMP 1995).
BIOENERGETIC COSTS OF THE MUTATION
H2O2 as an estimator of reactive oxygen species production
We assayed aspects of complex III bioenergetics because we hypothesized that the cox7A deletion mutation may result in an increased residence time of electrons at sites upstream in the electron transport chain from cox. The rate of H2O2 production was measured from isolated mitochondria respiring on sn glycerol 3-phosphate in the presence of rotenone. Glycerol 3-phosphate feeds electrons into complex III (ubiquinol-cytochrome c oxidoreductase) of the electron transport chain via FAD, glycerol 3-phosphate dehydrogenase (G3PDH), and ubiquinol. Rotenone prevents back cycling of electrons to complex I effectively isolating complex III. The H2O2 produced is considered to originate from complex III and G3PDH (Miwa et al. 2003).
Assays were performed using an Amplex Red H2O2/Peroxidase Assay Kit following Melvin and Ballard (2006). Briefly, reaction of Amplex Red reagent with H2O2 in the presence of horseradish peroxidase produces the oxidation product, resorufin, which has an absorption maximum at 560 nm. Reactions contained 5 μg of mitochondrial protein, respiration buffer, 2 μM rotenone, 30 U mL−1 superoxide dismutase, sn glycerol 3-phosphate, and Amplex Red reagent as per manufacturer's instructions in a total of 100 μL. Reactions containing known concentrations of H2O2 were used to construct a standard curve and were run in each assay plate. Controls lacking mitochondrial protein, substrate, or horseradish peroxidase were included. Results were analyzed by nested ANOVA (JMP 1995).
Differences in rates of H2O2 production between genotypes were observed and we measured proton leak. Proton leak has been hypothesized to reduce the rate of ROS production by preventing a backup of electrons in the electron transport chain (Brand 2000). We isolated mitochondria as described above and measured respiration rate and membrane potential simultaneously using electrodes sensitive to oxygen and to the potential sensitive probe triphenyl methyl phosphonium cation (TPMP+) using the protocols of Katewa and Ballard (2007). Briefly, membrane potential was measured using a TPMP+-sensitive electrode constructed following the procedure of Brand (1995) and inserted into the incubation chamber. Membrane potential and respiration were progressively inhibited through successive additions of cyanide up to 110 μM. Data were collected from both electrodes simultaneously using a PowerLab data-acquisition and analysis system and Chart software (ADI Instruments, Sydney, Australia). The relationship of proton leak to membrane potential was displayed by plotting a steady-state respiration rate for each value of membrane potential. Respiration at the highest common potential was analyzed by ANOVA using JMP software (JMP 1995).
CYTOCHROME C OXIDASE (COX) ACTIVITY
Cytochrome c oxidase activity was lowest in thoraces of homozygous cox7A mutants at both four days and 18 days of age. In comparison cox activity of the homozygote normal and the heterozygote are similar on both four days and 18 days (Fig. 1). Cox activity decreased in all genotypes over the age interval of 4–18 days. Relative to enzyme activity at four days, cox activity in 18 days thoraces was reduced 59.5% in the mutant, 19.8% in heterozygote, and 19.2% in the normal lines (Fig. 1). Nested ANOVA shows that replicate lines do not differ (F3,190= 0.32, P= 0.81) and there is a significant influence of genotype (F3,190= 36.62, P < 0.001), age (F1,190= 36.07, P < 0.001), and a genotype × age interaction (F2,190= 3.12, P= 0.04) on cox activity.
BODY PROTEIN CONTENT
We assayed the whole organism trait of body protein content in three outbred fly lines. We had previously observed that flies homozygous for the cox7A deletion mutation had significantly elevated body protein content (Ballard et al. 2010). Consistent with this result, body protein content was elevated in three outbred lines that carried the cox7A mutation (Fig S1). Nested ANOVA shows that protein content was not influenced by genetic background (F4,60= 0.73, P= 0.58) but was significantly affected by cox7A genotype (F1,60= 10.63, P= 0.002).
PHENOTYPIC BENEFITS OF THE MUTATION
We assayed early fertility of the 12-generation inbred genotypes by measuring the total number of adult flies emerging after a singly mated female was allowed to lay eggs for 10 days. Over the 10-day study, homozygote mutant females produced 702 adult flies, heterozygote females produced 415 adult flies, and homozygote normal flies produced 225 adult flies. The number of homozygote mutant adults that emerged was 41% and 68% higher than that of the heterozygote and homozygote normal genotypes, respectively (Fig. 2). Nested ANOVA showed that replicate lines did not differ in fertility (F3,43= 1.17, P= 0.33) and fertility of the three genotypes differed significantly (F2,43= 3.45, P= 0.04).
BIOENERGETIC BENEFITS OF THE MUTATION
Mitochondrial DNA copy number
mtDNA copy number was highest in mutant flies at both ages assayed and increased by 28% from four days to 18 days of age (7.8 ± 0.7 and 10.0 ± 0.8, ±SE times as many mtDNA copies compared to four-day-old homozygote normal flies, respectively) (Fig. 3). In contrast, mtDNA copy number decreased with age in both the heterozygote and homozygote normal flies (50% and 32% decrease, respectively). Nested ANOVA showed that replicate lines did not differ (F3,103= 1.13, P= 0.0.34). mtDNA density was influenced significantly by cox7a genotype (F2,103= 190.88, P < 0.001) and genotype × age interaction (F2,103= 5.19, P < 0.01) but not by age (F1,103= 1.52, P= 0.22).
The cox7A deletion mutant and heterozygote lines had 61% and 55% higher ADP:O ratio, respectively than the homozygote normal lines at four days of age (Fig. 4). The ADP:O ratio measured for the mutant and heterozygote flies remained higher than that of the homozygote normal line at 18 days of age. Nested ANOVA showed that replicate lines did not differ in ADP:O ratio (F3,10= 0.25, P= 0.86) and that ADP:O ratio was influenced by cox7a genotype (F2,10= 5.45, P= 0.02) but not by age (F1,10= 1.37, P= 0.26) or genotype × age interaction (F2,10= 0.35, P= 0.71).
PHENOTYPIC COST OF THE MUTATION
As expected flies harbouring the mutation had reduced survival. The survival function of virgin, female D. simulans differed among the three genotypes (Log-Rank test χ2= 218.3, P < 0.001) (Fig. 5). Survival of mutant and heterozygote lines was significantly shorter than that of homozygote normal lines (Log-Rank test χ2= 105.8, P < 0.001 and χ2= 198.0, P < 0.001, respectively) and their survival functions differed significantly from each other (Log-Rank test χ2= 18.7, P < 0.001).
Compared with homozygote normal lines, the age of 50% survival was decreased by 18.7% in mutant lines and by 19.7% in heterozygote lines (Fig. 6). Age at 50% survival for mutant lines was 36.7 ± 1.7 days, that of heterozygotes was 36.3 ± 1.1 days, and that of normal flies was 45.2 ± 2.1 days. Nested ANOVA showed that replicate lines did not differ (F3,24= 2.24, P= 0.11) and there was a significant influence of genotype (F2,24= 8.58, P < 0.01).
Mortality rate analysis shows that the rate of aging of mutant and heterozygote flies begins to diverge from that of normal flies after 34 days (Table 1). Mortality rates of the three genotypes are low and have overlapping 95% confidence intervals up until the flies are aged 34 days. At the next interval, 36 days, and thereafter the mortality rates of the mutant and heterozygote flies begin to rise and their 95% confidence intervals no longer overlap with that of normal flies.
Table 1. Age-specific mortality rate of the three cox7A genotypes aged 34 days and 36 days with lower and upper 95% confidence intervals.
Mortality rate (ln(μx))1
Lower 95% CI1
Upper 95% CI1
1The 95% CI of the mortality rates overlap at 34 days (normal compared with heterozygote and mutant) but do not overlap at 36 days (normal compared with heterozygote and mutant).
Physical activity differences were only detected in older flies and only when the homozygotes were compared (Fig. S2). Nested ANOVA showed that replicate lines did not differ significantly (F3,152= 1.94, P= 0.13). Physical activity was significantly influenced by genotype × age interaction (F2,152= 4.67, P= 0.01) but not by cox7a genotype (F2,152= 2.20, P= 0.11) or age (F2, 152= 0.003, P= 0.97). If we compare just normal and mutant flies they had similar physical activity at four days (170.88 ± 15.79 cm/h for normal and 184.32 ± 13.18 cm/h for mutant flies). However, at 18 days the physical activity of normal flies was 25% higher than mutant females (206.73 ± 26.91 cm/h for normal and 155.13 ± 25.98 cm/h for mutant flies). Nested ANOVA shows that replicate lines do not differ (F3,103= 1.48, P= 0.23). The difference in physical activity between normal and mutant flies is significant for genotype (F1,103= 24.28, P= 0.04) and the genotype × age interaction (F1,103= 5.15, P= 0.03) but not for age F1,103= 1.29, P= 0.26). Nested ANOVA shows that replicates did not differ over the eight days of assays (F7,103= 1.31, P= 0.25) therefore days were combined in the above analysis.
BIOENERGETIC COSTS OF THE MUTATION
H2O2 as an estimator of reactive oxygen species production
As expected, the rate of H2O2 production was highest in mitochondria isolated from thoraces of homozygous mutant flies at both four days and 18 days of age (78% and 61% higher than cox7A homozygote normal flies on four days and 18 days, respectively) (Fig. 7). Rate of H2O2 production increased over the age range 4–18 days in all genotypes. The increase with age was 22% for mutant, 27% for heterozygotes, and 32% for normal homozygotes. Nested ANOVA shows that replicate lines do not differ (F3,68= 0.68, P= 0.57) and that H2O2 production is influenced by genotype (F2,68= 77.82, P < 0.001) and age (F1,68= 32.74, P < 0.001) but not by their interaction (F2,68= 0.27, P= 0.77).
As expected, flies homozygous for the cox7A deletion had highest proton leak at both four days and 18 days. At the greatest common membrane potential of 150 mV the respiration rate of cox7A mutant flies was 103% and 102% higher than that of cox7A homozygote normal flies on four days and 18 days, respectively. Heterozygote flies had intermediate proton leak that was 32% and 30% higher than that of homozygote normal flies (Fig. 8). Nested ANOVA shows that replicate lines did not differ (F3,15= 0.70, P= 0.57) and that proton leak is influenced by genotype (F2,15= 7.59, P < 0.01) but not by age (F2,15= 0.007, P= 0.93) or genotype × age interaction (F2,15= 0.005, P < 0.94).
Mitochondria are the most ubiquitous symbiont on the planet. Over evolutionary time, it is proposed that genes residing in the proto-organelle moved into the nuclear genome. The proteins from many of these nuclear encoded genes now need to be imported into the mitochondria for normal organelle functions to be performed. Mutations in these genes affect not only the individual protein subunits but also assembly of the subunits into respiratory complexes, the proper phospholipid environment of the organelle and the mtDNA replication, transcription, and translation machinery (Zeviani and Di Donato 2004; Jacobs and Turnbull 2005). To date, mutations in approximately 70 such genes have been identified in humans (Jacobs and Turnbull 2005). Clearly, however, the mitochondria are dynamic organelles that have the potential to buffer the changing bioenergetic requirements of the cell and the organism. Mitochondrial plasticity in response to contractile exercise is characterized by increased transcription of nuclear cox genes, mtDNA copy number, and increased mitochondrial protein content (Hood et al. 1989). Here we show that mitochondria can also buffer DNA mutations.
Cox activity is influenced by mutations in the genes that encode its constituent proteins and genes involved in the assembly of the cox holoenzyme (Taylor and Turnbull 2005; Fontanesi et al. 2006). To date, only one naturally occurring mutation in a nuclear encoded cox gene has been identified and associated with human mitochondrial disease (Massa et al. 2008). It is likely that highly deleterious mutations in cox subunits are rarely observed because they are fatal early in development whereas mutations that produce subclinical effects may go unnoticed (Shoubridge 2001). In D. melanogaster Canton S flies and S2 cell cultures, deletions in portions of cox subunits have been created by mutagenesis. Liu et al. (2007) produced a large deletion of cox6A, levy, in D. melanogaster by mutagenesis with ethylmethanesulfonate (EMS). The levy mutation causes severe neuropathy and decreased life span. Likewise Mandal et al. (2005) produced a 56 base pair deletion, tenured, in cox subunit 5A of Drosophila S2 cultured cells. Mutant cells produced significantly less ATP and had increased levels of phosphorylated AMP-activated protein kinase (AMPK) compared to normal cells. AMPK is conserved in eukaryotes and functions to sense intracellular ATP status. When phosphorylated, AMPK inhibits the target of rapamycin (TOR) pathway preventing growth during periods of energetic stress (Hardie 2005).
In this study, we observe that flies homozygous for the deletion have a fertility advantage when young but pay survival and physical activity costs later in life. We hypothesize that the flies responded to decreased cox function, resulting from the cox7A mutation, by upregulating mitochondrial proteins in an attempt to maintain energy homeostasis. Plausibly, this resulted in an increase in the ADP:O ratio (Fig. 4), rates of H2O2 production (Fig. 7), proton leak (Fig. 8), and total protein in thoraces (Fig. S1). However, it is also possible that the elevated ADP:O levels in heterozygote and mutant flies (Fig. 4) could be due to reduced cox activity, and hence reduced oxygen consumption. In Drosophila, cox is composed of three mtDNA encoded subunits and nine nuclear encoded subunits that are translocated into the mitochondrion (Das et al. 2004). Although no isoforms of cox7A have been detected in D. simulans (Melvin et al. 2008) it is not known if modifiers influence mRNA expression and phenotypic effects of the two amino acid deletion. Simple genetic diseases can have complex phenotypes due to genetic modifiers (Cutting 2005). We suggest that the biochemical and organismal results we observe are unlikely to be the result of genetic modifiers in HW01 flies. The mutation influences both the bioenergetic trait of cox activity (Ballard et al. 2010) and the physiological trait of protein content (Fig. S2).
Changes in nuclear genes have previously been suggested to compensate for mitochondrial dysfunction. Bai and Attardi (1998) characterized a mouse cell line (4A) carrying a mitochondrial DNA mutation in the subunit for respiratory complex I, NADH dehydrogenase, in the ND6 gene. This mutation abolished the complex I assembly and disrupted the respiratory function of complex I. Deng et al. (2006) report that a galactose-resistant clone, 4AR, was isolated from the cells carrying the ND6 mutation. These cells (4AR) still contained the homoplasmic mutation, and apparently there was no ND6 protein synthesis, but the assembly of other complex I subunits into complex I was recovered. Furthermore, the respiratory activity and mitochondrial membrane potential were fully recovered. Together these data strongly suggest that nuclear changes in 4AR have compensated for the mtDNA deficiency. Increased mitochondrial mass is also observed in mouse models of mitochondrial disease and is hypothesized to be a mechanism of compensating for reduced OXPHOS function and maintenance of ATP homeostasis (Wredenberg et al. 2002). Furthermore, an increase in mtDNA copy number has been observed as cells age and is a hallmark of human disease caused by mtDNA mutation (Butow and Avadhani 2004).
Currently, it is not clear why the expression of nuclear genes was double in the mutant flies (Ballard et al. 2010) whereas mtDNA density was observed to be eightfold higher in the current study. One possibility is that there is over replication of mtDNA in response to oxidative damage that may cause a disproportionate increase in mtDNA compared to the expression of nuclear encoded OXPHOS subunits. In human mitochondrial diseases most patients with mtDNA deletions demonstrate compensatory mtDNA overreplication (Bai and Wong 2005). What appears clear, however, is that the nuclear genome is responding to mitochondrial dysfunction by upregulating the expression of nuclear genes that produce proteins targeted to the mitochondria. It would appear most likely that this occurs as a result of a retrograde response that needs further study. The retrograde response is a signaling pathway from the mitochondria to the nucleus. Retrograde signaling activates nuclear and mitochondrial transcription factors and results in increased numbers of mtDNA, OXPHOS complexes, and whole mitochondria (Passos et al. 2007). Retrograde response is well known in the yeast Saccharomyces cerevisiae and to a lesser extent in mammalian cells (Butow and Avadhani 2004).
Life span was lower in the cox7A deletion mutant flies but a difference in mortality rate between mutant and normal flies cannot be detected before 34 days. In contrast, significant differences in H2O2 production could be detected at four days and 18 days of age. This result is consistent with the predictions of the oxygen free radical theory of aging. This theory states that aging, loss of biological function with time, is caused by accumulated oxidative damage to DNA, proteins, and lipid membranes. Mutants produce H2O2 at a higher rate than do normal flies and show age-related declines in OXPHOS efficiency. We suggest these results with virgin flies are conservative and predict that the observed differences in survival and H2O2 production rates would be more pronounced in nonvirgin females. Reproduction is energetically expensive and known to shorten female life span in Drosophila (Chapman et al. 1995).
Two lines of evidence suggest that the two amino acid deletion in complex IV is causally responsible for the phenotypic effects observed. First, a six-step strategy was employed to select duplicate fly lines for this study. Second, the mutation was placed into multiple heterozygous genetic backgrounds and a concomitant decrease in cox activity was observed (Ballard and Melvin 2010). Despite these data, it is possible that a slightly deleterious mutation that is physically linked with the deletion may be responsible for the differences in the life-history traits we observed. Moreover, we have not assayed the pattern of mutations surrounding the deletion mutation. Cox7A is located on the minus strand of chromosome 3R at positions 17,095,012 to 17,095,384 and is within 100 kilobase pairs of the genes GD20873 (orthologous to human polynucleotide kinase/3′-phosphatase, ENSP00000323511), GD18539 (orthologous to Drosophila melanogaster dipeptidyl aminopeptidase, DppIII), GD18536 (orthologous to D. melanogaster ADP ribosylation factor 84F, Arf84F), and GD18540 (orthologous to Mus musculus meiosis specific nuclear structural protein). Unlike D. melanogaster, D. simulans has a paucity of chromosomal inversions and therefore recombination rates are unlikely to be reduced in the region surrounding the deletion (Aulard et al. 2004).
In this study, we have reiterated the utility of biochemistry as a link between the genotype with the phenotype. More specifically, we have shown the physiological and bioenergetic benefits and costs of a two amino acid deletion in a nuclear encoded gene that produces a protein that is imported into the mitochondrion. Variation in nuclear encoded genes that produce proteins imported into the mitochondrion is well known (Ballard et al. 1996, 2002; Willett and Burton 2004; Melvin et al. 2008; Willett and Ladner 2009) (Jacobs and Turnbull 2005) and the influence of specific mutations can be specifically targeted using this approach. We assayed the cost and benefit of a mutation in electron transport chain complex IV of Drosophila but mutations in other complexes and other species could also be targeted. The biochemical process of OXPHOS is quite well understood and the overall performance of OXPHOS and that of each step in the process can be assayed (Kirby et al. 2007). The results from this study also have implications for studies of human mitochondrial disorders. It has recently been proposed that upregulating PGC (peroxisome proliferator-activated receptor gamma coactivator)-1α signaling could increase the total number of mitochondria within cells and may lead to a possible mechanism to treat human mitochondrial disorders in (Schon et al. 2010). The data presented here suggest that compensation for mitochondrial dysfunction plausibly through a retrograde response may lead to shorter term improvements in mitochondrial biogenesis and fertility but may also have longer term deleterious affects.
Associate Editor: M. Wayne
We wish to thank R. Brooks, D. Rand, S. Simpson, F. Clissold, J. Wolf, and K. Ballard for comments on the study. Two anonymous reviewers made constructive and valuable comments that improved the manuscript.