Expression of alternative oxidase in Drosophila ameliorates diverse phenotypes due to cytochrome oxidase deficiency

Expression of alternative oxidase in Drosophila ameliorates diverse phenotypes due to cytochrome oxidase deficiency Kia K. Kemppainen1, Juho Rinne1, Ashwin Sriram1, Matti Lakanmaa1, Akbar Zeb1, Tea Tuomela1, Anna Popplestone1, Satpal Singh2, Alberto Sanz1, Pierre Rustin3and Howard T. Jacobs1,4,∗ 1 Institute of Biomedical Technology and Tampere University Hospital, University of Tampere, FI-33014 Tampere, Finland, 2 School of Medicine and Biomedical Sciences, State University of New York at Buffalo, 206 Cary Hall, Buffalo, NY 14214, USA, 3 INSERM UMR 676, Ho ˆpital Robert Debre ´,48BdSe ´rurier, 75019 Paris, France and 4 Molecular Neurology Research Program, University of Helsinki, FI-00014 Helsinki, Finland Received August 20, 2013; Revised and Accepted November 22, 2013 Mitochondrial dysfunction is a significant factor in human disease, ranging from systemic disorders of childhood to cardiomyopathy, ischaemia and neurodegeneration. Cytochrome oxidase, the terminal enzyme of the mitochondrial respiratorychain, is a frequent target. Lower eukaryotes possess alternativerespiratory-chain enzymes thatprovidenon-proton-translocatingbypassesforrespiratorycomplexesI(single-subunitreducednicotinamide adenine dinucleotide dehydrogenases, e.g. Ndi1 from yeast) or III 1IV [alternative oxidase (AOX)], under conditions of respiratory stress or overload. In previous studies, it was shown that transfer of yeast Ndi1 or Ciona intestinalis AOX to Drosophila was able to overcome the lethality produced by toxinsor partialknockdown of complex I or IV. Here, we show that AOX can provide a complete or substantial rescue of a range of phenotypes induced by global or tissue-specific knockdownof differentcIV subunits, includingintegralsubunitsrequiredforcatalysis,as well as peripheral subunits required for multimerization and assembly. AOX was also able to overcome the pupal lethality produced by muscle-specific knockdown of subunit CoVb, although the rescued flies were short lived and had a motility defect. cIV knockdown in neurons was not lethal during development but produced a rapidly progressing locomotor and seizure-sensitivity phenotype, which was substantially alleviated by AOX. Expression of Ndi1 exacerbated the neuronal phenotype produced by cIV knockdown. Ndi1 expressed in place of essential cI subunits produced a distinct residual phenotype of delayed development, bang sensitivity and male sterility. These findings confirm the potential utility of alternative respiratory chain enzymes as tools to combat mitochondrial disease, while indicating important limitations thereof. INTRODUCTION Mitochondrial diseases affecting the respiratory complexes of the oxidative phosphorylation (OXPHOS) system are a diverse collection of pathologies, which can affect almost any tissue, at any age (1,2). They are typically progressive in nature, and no effective treatments are currently available. Where genetic causes are known, they can include lesions in any of hundreds of genes whose products are needed for the biosynthesis or function of the respiratory complexes. These genes, moreover, are distributed between nuclear and mitochondrial DNA (mtDNA), and a subset of mitochondrial diseases also results from defective communication between the cell’s two genomes. In order to understand better the pathophysiological mechanisms of mitochondrial disease, and develop a possible strategy for eventual therapy, we have exploited the fact that lower eukaryotes, including plants and many invertebrates, possess an alternative, non-proton-pumping respiratory chain in mitochondria, whose components can act as a bypass of the OXPHOS system under conditions of respiratory stress or overload (3,4). Alternative reduced nicotinamide adenine dinucleotide(NADH) dehydrogenases such as yeast Ndi1 can replace complex I (cI), ∗ To whom correspondence should be addressed at: Institute of Biomedical Technology, FI-33014 University of Tampere, Finland. Tel: +358-3-3551-7731; Fax: +358-3-3551-7710; Email: [email protected] #The Author 2013. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Human Molecular Genetics, 2014, Vol. 23, No. 8 2078–2093 doi:10.1093/hmg/ddt601 Advance Access published on November 29, 2013 at Tampere University Library. Department of Health Sciences on September 27, 2016http://hmg.oxfordjournals.org/Downloaded from while alternative oxidases (AOXs) can replace complexes III + IV (cIII +IV). The alternative enzymes are each composed of a single gene product, which together conduct electrons from NADH to molecular oxygen via ubiquinone as an intermediate electron acceptor. AOX becomes enzymatically active only when electron transfer via the OXPHOS system becomes inhibited beyond ubiquinone. The main mechanism to achieve this is believed to be that AOX has a much higher K m for its substrate, ubiquinol, than does cIII. In contrast, Ndi1 may be constitutively active, but its role in electron flow under normal physiological conditions may be limited by the tight coupling between cI and cIII in supercomplexes. Like Ndi1, AOX is absent from vertebrates, as well as arthropods (5), but we reasoned that if the gene product were transferred to higher metazoans it might be able to functionally replace the corresponding OXPHOS complexes under conditions resembling those pertaining in mitochondrial disease. When expressed transgenically in human cells (6,7), Drosophila (8) or most recently the mouse (9), AOX from the urochordate Ciona intestinalis was shown to be enzymatically functional when cIII or cIV was inhibited by specific toxins, or genetic mutation. Similarly, when yeast Ndi1 was transferred to mammalian cells or to Drosophila, it was functional, enabling NADH oxidation when cI was inhibited by rotenone (10). Remarkably, the ubiquitous presence of these alternative enzymes is tolerated by the whole organism, to which it can confer toxin resistance in vivo (8–10). In wild-type flies, no deleterious phenotype is produced by constitutive AOX (or Ndi1) expression (8,10). However, the alternative enzymes can overcome the lethality produced by partial knockdown of at least some subunits of the corresponding OXPHOS complexes, including two subunits of the membrane portion of cI (10), a nuclear-coded ‘supernumerary’ subunit of cIV [Cox6c, the cyclope gene product (8)] and a cIV assembly factor, Surf1 (8). Note, however, that AOX cannot complement the total loss of cIV, e.g. via a null mutation in Cox6c (9). The structure of metazoan cIV is relatively well understood (11). It comprises a catalytic core, composed of the three mtDNA-encoded subunits, whose functional assembly is dependent on a set of phylogenetically conserved nuclear-coded subunits (Cox4, Cox5a, Cox5b, Cox6b, Cox6c, probably Cox7c, Cox8, using the mouse nomenclature) required to produce, via a stepwise assembly pathway, a membrane-bound sub-complex with enzymatic activity (12,13). The incorporation of prosthetic groups (a-type haemes and copper) is also involved in this process. The other nuclear-coded subunits (Cox6a and Cox7a) are then incorporated to form the fully functional cIV, which also contains a recently described subunit (14) formerly believed to be a constituent of cI. Some of these subunits, notably Cox6a, are believed to be important for dimerization of the complex and, together with other specific proteins such as Rcf1 (HIG2A) and Cox7RP, may also be required for the formation ofsupercomplexes containingcIV,as well ascI and/orcIII indifferent stoichiometries (15–18). The more specific roles of the nuclear-coded subunits of cIV (often termed ‘supernumerary’ since they are absent in bacteria), for example in physiological regulation, are less clear. Null mutations in most of the above subunits are lethal in Drosophila, whereas a splice-site mutation in the levy gene, encoding subunit Cox6a, produces an adult-onset neurodegenerative phenotype (19). Cytochromecoxidase (COX) deficiency inhumans has a wide variety of pathological manifestations and a diversity of genetic causes (20,21). Organs affected can be the central nervous system, skeletal or heart muscle, the liver or a combination of these and other organs, and tissue-specificity is poorly understood. Underlying genetic defects commonly impact accessory factors for cIV biosynthesis, including assembly factors such as SURF1 (22,23) or C2ORF64 (24) or proteins involved in cofactor synthesis or transport, such as SCO2 (25) and COX10 (26). Comparatively rare mutations are also found in genes for structural subunits of cIV, such as COX6B1 (27)or COX7B (28), as well as those encoded in mtDNA (29–31), plus genes required mainly or exclusively for the biosynthesis of mtDNA-encoded cIV subunits [e.g. LRPPRC (32), TACO1 (33)]. Pathological COX deficiency has been reported in many other disorders (34,35), including neurodegenerative conditions such as Alzheimer dementia (36–38) and Huntington’s disease (39) although its aetiological significance remains unclear. COX is also a pathological target in ischaemia, sepsis and other types of toxic injury (40). Some cIV subunits are encoded by isogenes differentially expressed between tissues, where they are adapted to specific physiological conditions (35,41). Tissues also vary in their substrate dependence, reflected in the degrees to which the different OXPHOS complexes contribute to threshold effects for respiration (42). Furthermore, supercomplexes, whose formation may also vary between tissues, entrain a greater or lesser degree of channelling of respiratory electron flow, and may also limit the degree to which exogenously introduced alternative respiratory enzymes can contribute to respiration (18). In a general sense, these phenomena are believed to contribute to the bewildering tissue-diversity of mitochondrial disease, though there remain few concrete mechanistic explanations for this. The relative importance in COX-associated pathology of disturbed redox homeostasis, apoptosis induction, deranged cell signalling, adenosine triphosphate (ATP) deficiency, proteotoxic stress and other types of metabolic disturbance remains a topic of intense debate. ThepossibleuseofAOXinfuturetherapiesfor COXdeficiency will obviously be limited by these considerations, as well as by the fact that the alternative respiratory chain enzymes do not contributedirectly toATP production. Completelyreplacing thefunction of one or more OXPHOS complexes cannot completely restore ATP production capacity to wild-type, even though it can facilitate proton-pumping linked to electron transfer at OXPHOS complexes other than the one(s) it is bypassing. Because AOX bypasses two of the three proton-translocating steps, it should not restore ATP production for those substrates whose oxidation supplies electrons directly to ubiquinone via cII or other dehydrogenases. Even if AOX can restore redox balance, mitigate mitochondrial reactive oxygen species (ROS) production (6–9) and limit systemic lactic acidosis (6), tissues that depend on such substrates may still suffer a substantial loss of function that AOX cannot correct. On the other hand, where COXactivity is limiting forrespiration, AOXmay promoteasignificant restoration of ATP production (via proton pumping at other sites), as well as other metabolic benefits. Drosophila offers a convenient model for human diseases, including those affecting mitochondrial functions. The composition of the OXPHOS complexes, the overall structure and Human Molecular Genetics, 2014, Vol. 23, No. 8 2079 at Tampere University Library. Department of Health Sciences on September 27, 2016http://hmg.oxfordjournals.org/Downloaded from gene content of mtDNA and the energetic physiology of the main organ systems are all similar to those of humans. The fact that AOX expression in the fly is able to confer cyanide resistance and overcome the lethality of partial knockdown of Cox6c or Surf1 raises the question of how far it is able to replace cIV throughout the life cycle. We thus set out to use the fly as a model system to address this issue, focusing on three specific questions relevant to understanding the pathophysiology of mitochondrial disease and the possible future use of AOX in therapy. First, what phenotypes resultfrom deficiencies in different classes of cIV subunits, and can AOX alleviate these phenotypes? Secondly, in which tissues is loss ofCOX function crucial for producing such phenotypes, and their rescue by AOX? Thirdly, what residual phenotypes result from replacing cIV function with AOX? The many genetic tools available in Drosophila (including effective and specific RNAi in vivo)makeitanideal tool for such a study, in advance of translating the most prominent findings to the mouse, which is less flexible, and also takes much longer and is substantially more expensive to manipulate. In a first step, to validate use of the inducible (upstream activating sequence) UAS-AOX transgenic lines that we created previously, we first conducted an important control to verify that the rescue seen previously was not subject to any confounding, ‘promoter-dilution’ effects from the simultaneous use of transgenes dependent on the same transcription factor. We then proceeded to test the phenotypes generated by knockdown of different cIV subunits in the whole fly and in specific tissues, and the ability of AOX to rescue these phenotypes, focusing on the major differentiated cell-types affected by human mitochondrial disease (muscle and neurons). The results highlight the different phenotypes produced by deficiencies of different subunits, indicate a crucial developmental role for cIV in muscle and confirm COX deficiency as a cause of adult-onset neurodegeneration. AOX expression was able to partially rescue these phenotypes. In contrast, loss of protonpumping at cI promotes a distinct phenotype of developmental delay, bang sensitivity and male infertility. RESULTS Constitutive low-level AOX expression confers resistance to COX inhibition In previous experiments we expressed inducible AOX using the UAS/Gal4 (yeast transcription activator protein GAL4) system, which partially rescued the lethality of cyanide treatment or the disruption of COX by knockdown of the Cox6c subunit (the cyclope gene: see Table 1) or the Surf1 assembly factor. However, the UAS/Gal4 system results in a very high level of transgene expression. Furthermore, since COX knockdown by RNAi also involves the same induction system, we cannot completely exclude a contribution to the rescue from promoter competition, even though UAS-GFP (green fluorescent protein) was unable to rescue. We therefore created transgenic flies bearing the same AOX transgene, but under the control of the constitutive a-tubulin promoter (tub-AOX), with independent single insertions into non-coding DNA on chromosomes X, 2 and 3 (Supplementary Material, Fig. S1A–D). Expression of these transgenes at the RNA level was 20–50-fold lower than from UAS-AOX driven by da-GAL4 (Fig. 1A), though still much higher than UAS-AOX in the absence of a GAL4 driver. tub-AOX expression was substantially higher in larvae and adult males than females (Fig. 1A), and was maintained over the first 2weeks of adult life to a variable extent. AOX expression at the protein level was also less than when driven by GAL4, even when we combined tub-AOX insertions Table 1. Nomenclature, expression patterns and assembly of COX subunits in Drosophila Subunit name a Official gene name(s) and symbol(s) b Expression pattern(s) c Comments d Cox1 mt CoI Ubiquitous mtDNA-encoded, part of core sub-complex S2 Cox2 mt CoII Ubiquitous mtDNA-encoded, incorporated into sub-complex S3 Cox3 mt CoIII Ubiquitous mtDNA-encoded, incorporated into sub-complex S3 Cox4 CoIV Ubiquitous Two isogenes with different expression patterns, part of core sub-complex S2 CG10396 Testis-specific Cox5a CoVa Ubiquitous Part of core sub-complex S2 Cox5b CoVb Ubiquitous Incorporated into sub-complex S3 Cox6a levy Ubiquitous Incorporated at final assembly steps into mature complex IV; levy 1 splice-site mutant manifests adult-onset neurodegeneration (19) Cox6b CoVIb Ubiquitous Incorporated into sub-complex S3 Cox6c cyclope Ubiquitous Null-mutant larval lethal, incorporated into sub-complex S3 Cox7a CG9603 Ubiquitous, lowest in testis Incorporated at final assembly steps into mature complex IV; 99% of brain Cox7a expression contributed by CG9603 CG34172 Mainly muscle-specific (heart crop, hindgut, carcass, lower in head) CG18193 Testis-specific Cox7b None identified Proposed to be required for an early assembly step (28); clear orthologues identified only in vertebrates Cox7c CoVIIc Ubiquitous, but low in testis Only recently identified in Drosophila Cox8 CoVIII Ubiquitous, but low in testis Incorporated into sub-complex S3 a Using mouse nomenclature. Note that, some subunits are encoded by gene families in mouse but single genes in Drosophila, and vice versa. b From www.flybase.org. c From www.flyatlas.org. d Assembly program based on Ref. (12). 2080 Human Molecular Genetics, 2014, Vol. 23, No. 8 at Tampere University Library. Department of Health Sciences on September 27, 2016http://hmg.oxfordjournals.org/Downloaded from on all three major chromosomes into a single line of flies (Fig. 1B). tub-AOX conferred cyanide resistance to substrate oxidation by isolated mitochondria (Supplementary Material, Fig. S1E), and single tub-AOX insertions were able to confer almost as great a time of resistance to cyanide as UAS-AOX driven by da-GAL4 (Fig. 1C, compare with Fig. 5A of Ref. 8). A single copy of tub-AOX was also able to rescue the semilethality of knockdown of either Cox6c (Fig. 2A) or Surf1 (Fig. 2B and C), when RNAi was induced with different doses of RU486, in the presence of ubiquitously expressed GeneSwitch, a modified version of GAL4 that is dependent on the drug for activity. [Note that, for clarity, we use the mammalian (mouse) names for all subunits and the genes that encode them throughout the text, indicating the Drosophila gene name only upon first mention, in Table 1and in figure legends.] Knockdown of Cox6c by about 80% at the RNA level, using 10 mMRU486 (Supplementary Material, Fig. S2A), not only resulted in lethality of about three-quarters of the progeny (Fig. 2A), but those flies that were able to eclose did so with a 3-day delay (Fig. 2D), a phenotypic feature seen in many Drosophila OXPHOS mutants. The presence of tub-AOX corrected both of these phenotypes (Fig. 2A and D). tub-AOX also enabled developing flies to reach pupal stage at doses of RU486 that induced a degree of Surf1 knockdown that prevented any pupariation of control flies (Fig. 2B). At 0.1 mMof the drug, a substantial number of tub-AOX expressors even reached eclosion, whereas only a few viable adults were produced at this dose by control flies under Surf1 knockdown alone (Fig. 2C). Extent of COX deficiency rescued by AOX expression Knockdown of subunits Cox5a or Cox5b also produced an RU486 dose-dependent lethality (Fig. 2E and F). This enabled us to test how rescue is affected by the number of constitutively Figure 1. Transgenic expression of tub-AOX.(A) Quantitative real time-polymerase chain reaction (qRT-PCR) analysis of AOX mRNA expression (normalized against RpL32) in transgenic adults and larvae, as indicated. Means +SD of at least three technical replicates of each of at least three biological replicates. Comparing (upstream activating sequence) UAS-AOX expression driven by da-GAL4 (yeast transcription activator protein GAL4) with that of tub-AOX,P,0.01 for each sex/ age analysed; similarly, comparing UAS-AOX expression in the absence of driver with that of tub-AOX,P,0.01 in each case, except where indicated (#), where P, 0.05 (Student’s ttest, two-tailed, unequal variances). (B) Western blot of AOX protein and adenosine triphosphate (ATP) synthase subunit a(loading control) in 1-day-old adults of the transgenic strains indicated. tub-AOX denotes flies homozygous for each of tub-AOX 7 ,tub-AOX 35 and tub-AOX 50 transgenes (males are hemizygous for tub-AOX 35 ). Replicate batches of protein extracts from 30 females or 40 males of each genotype are shown in adjacent lanes. (C) Survival time on cyanide-impregnated agar of flies of the (homozygous) strains indicated. Means +SD of 80–100 flies of each group, in batches of 10 flies per vial. P,0.01 in each case, in comparison with w 1118 control flies of same sex (Student’s ttest, two-tailed, unequal variances). See also Supplementary Material, Figure S1. Human Molecular Genetics, 2014, Vol. 23, No. 8 2081 at Tampere University Library. Department of Health Sciences on September 27, 2016http://hmg.oxfordjournals.org/Downloaded from expressed copies of AOX, and to estimate the amount of knockdown of COX required to produce AOX-rescuable lethality. A single copy of tub-AOX was sufficient to overcome the lethality of knocking down Cox5a or Cox5b to about half of the wild-type level at larval stage L3 (Supplementary Material, Fig. S2A), which also produced a decrease of about 50% in the enzymatically measured COX activity (Supplementary Material, Fig. S2B). Flies rescued from the lethality of Cox5a or Cox5b knockdown by two copies of tub-AOX eclosed with a slight, but statistically non-significant developmental delay (Supplementary Material, Fig. S2D). The rescued flies were fertile when mated to wild-type flies of the opposite sex. Flies in which Cox5b was knocked down during development by a higher dose (3 mM) of RU486, failed to develop beyond the larval stage, but four copies of tub-AOX enabled the flies to reach pupal stage, with 12%of the pupae eclosing.The flies were tested for bang sensitivity, a sensorineural phenotype associated with mitochondrial dysfunction, but no bang sensitivity was found (Supplementary Material, Fig. S2E). Having excluded promoter-dilution effects as a reason for AOX rescue of COX deficiency, we set out to use the UAS-AOX lines, in combination with different GAL4 drivers, to analyse the developmental phenotypes produced by knockdown of COX subunits, and the degree to which AOX expression is able to complement them. AOX expression rescues developmental lethality of COX knockdown The nuclear-coded subunits of cIV are believed to confer various regulatory properties on COX, as well as being required for its proper assembly and stability. Deficiency of specific subunits leads to the accumulation of various assembly intermediates Figure 2. Rescue of COX deficiency by tub-AOX.(A)tub-AOX rescue of developmental lethality and (D) of developmental delay, from partial knockdown of Cox6c (Drosophila gene cyclope) using tub-GS driver. Proportion of eclosing progeny or eclosion day for different genotypes and concentrations of RU486 as shown, means +SD from 4 or more biological replicates. (Band C)tub-AOX rescue of developmental lethality from partial knockdown of Surf1 using tub-GS driver. The number of pupae or eclosing flies for different genotypes and concentrations of RU486 as shown, means+SD from 3 or more biological replicates. (Eand F) Rescue of developmental lethality frompartial knockdown of Cox5a or Cox5b (Drosophila genes CoVa,CoVb) using tub-GS driver, as shown. Number of eclosing progeny at different concentrations of RU486, for the indicated numbers of tub-AOX transgenes. Means +SD from 3 or more biological replicates. P,0.01 (∗)or, 0.05 (#), Student’s ttest, two-tailed, unequal variances, comparing flies with and without tub-AOX. See also Supplementary Material, Figure S2. Note that at 0 mM RU486 all knockdown lines tested were indistinguishable from wild-type flies in the assays shown, and that wild-type flies eclose on Days 10 and 11 at 258C. Two copies of tub-AOX also rescued the lethality of Cox6b knockdown (Supplementary Material, Fig. S2C). 2082 Human Molecular Genetics, 2014, Vol. 23, No. 8 at Tampere University Library. Department of Health Sciences on September 27, 2016http://hmg.oxfordjournals.org/Downloaded from and to subtly different phenotypes. Most nuclear-coded subunits of COX in Drosophila are encoded by single-copy nuclear genes (Table 1). Two notable exceptions are Cox4, encoded by a specific isogene (CG10396) in testis, with a second isogene (CoIV, CG10664) expressed ubiquitously, and subunit Cox7a, also encoded by testis-specific (CG18193) and ubiquitous (CG9603) isogenes, plus an additional, widely expressed isogene (CG34172), prominently expressed in muscle (carcass, crop, hindgut and heart). Using da-GAL4 to drive simultaneously the high-level expression of UAS-AOX and the knockdown of nuclear-coded COX subunits in the whole developing fly, we delineated the limits of AOX rescue of COX deficiency. Knockdown of the somatic Cox4 isogene produced early larval lethality (Fig. 3A), consistent with the involvement of this subunit in an essential early step in cIV assembly, and its requirement for the production of an enzymatically functional complex. Knockdown larvae survived for at least 15 days, but never developed beyond a morphologically abnormal L1 or L2 stage (Fig. 3A), whereas co-expression of UAS-AOX enabled development to proceed as far as the pupal stage (Fig. 3A), albeit with very few flies eclosing (Supplementary Material, Fig. S3A). Blue native polyacrylamide gel electrophoresis (BNE) in-gel histochemistry confirmed the functional knockdown (≥50% decrease of COX activity) at larval stage L3 in the ‘rescued’ flies (Fig. 3E), as did polarography using a cIV-specific substrate mix (Fig. 3F). Interestingly, oxygen consumption driven by cIor G3PDH-linked substrates was increased compared with Figure 3. Partial rescue of global COX deficiency by UAS-AOX.(A) Micrographs illustrating typical phenotypes produced by knockdown of Cox4 (Drosophila gene CoIV) driven by da-GAL4. In the absence of transgenic rescue, progeny arrested as abnormal L1/L2 larvae. Co-expression of UAS-AOX enabled development to proceed as far as late pupa, although few flies eclosed. See also Supplementary Material, Figure S3A. (B) Micrograph illustrating typical pupal-lethal phenotype produced by knockdown of the major Cox7a-encoding isogene CG9603, driven by da-GAL4.(C) AOX rescue of pupal lethality from Cox7a (CG9603) knockdown. Proportionof eclosingprogeny fordifferent genotypes as shown,means +SD from3 ormore biologicalreplicates.Flieswiththe da-GAL4driverhavenormalbristles, distinguishing them from those with the Sb balancer. Expression of UAS-AOX produced a full rescue whereas UAS-Ndi1 expression produced none. (D) Ascorbate/ TMPD-drivenoxygen consumptionof homogenates fromUAS-AOXexpressing L3 larvae or adultflies, with or withoutknockdown ofCox7a (CG9603).Means +SD from 3 or more biological replicates; asterisks indicate significant differences (P,0.01, Student’s ttest, two-tailed, unequal variances). (E) BNE gels of mitochondrial extracts (37.5 mg mitochondrial protein per lane) from L3 larvae of the genotypes shown, stained histochemically for cI or cIV activity (cIV activity staining performed for the indicated times). See Ref. 10, Figure 3, for migration of major bands in relation to molecular weight markers on these gels. Asterisk indicates assembly sub-complex S3 (12). (F) Oxygen consumption of homogenates from L3 larvae of the indicated genotypes, driven by different substrate mixes (pyruvate + proline, G3P and ascorbate +TMPD, respectively, as described in the Materials and methods section for cI-, G3PDHand cIV-linked respiration. For clarity, oxygen consumption is expressed as a percentage of corresponding values for controllarvae expressing AOX but without RNAi. Means +SD, ≥3 biological replicates; asterisks indicate significant differences (P,0.01, Student’s ttest, two-tailed, unequal variances). Human Molecular Genetics, 2014, Vol. 23, No. 8 2083 at Tampere University Library. Department of Health Sciences on September 27, 2016http://hmg.oxfordjournals.org/Downloaded from control larvae, suggesting a compensation for missing protonpumping activity in response to AOX ‘replacement’ of the depleted cIV. Expression of UAS-Ndi1 as a control produced no rescue of Cox4 knockdown (Supplementary Material, Fig. S3A). Knockdown of Cox5b produced a similar larval-lethal phenotype that was also partially rescued by UAS-AOX, allowing flies to develop to pupal stage, with 5–10% of pupae eclosing as viable flies (Supplementary Material, Fig. S3B). Subunit Cox7a is required for the assembly of the fully functional holoenzyme, but not for the formation of an enzymatically active sub-complex (denoted as assembly intermediate S3). Knockdown of the major isogene for Cox7a (CG9603) also resulted in lethality, but in contrast to the early larval lethality produced by knockdown of core subunits, Cox7a knockdown produced lethality only at the pupal stage (Fig. 3B). This was completely rescued by co-expression of UAS-AOX (Fig. 3C), but not by UAS-Ndi1 (Fig. 3C). Consistent with the incorporation of Cox7a at a late stage in cIV assembly and the partial redundancy of the major Cox7a-encoding isogene CG9603 with isogene CG34172 in some tissues, knockdown of CG9603 resulted in only a partial loss of respiratory capacity in the whole fly, based on polarography (Fig. 3D) or BNE in-gel histochemistry (Fig. 3E), with a clear accumulation of assembly subcomplex S3 in L3 larvae. AOX-rescuable pupal lethality due to COX knockdown is primarily a muscle phenotype To test the tissue(s) in which COX expression is critical for the completion of development, we employed a set of tissue-specific drivers to knockdown either of the enzymatically crucial subunits Cox5b (Fig. 4A) or Cox6b (Supplementary Material, Fig. S4A). No lethality was produced by Cox5b knockdown using any of five nervous system-specific drivers, two of which are active in all neurons (elav-GAL4 located on chromosome 3, Bloomington strain 8760 and nrv2-GAL4), nor were the progeny flies bangsensitive (Supplementary Material, Fig. S4B). Lethality was also not produced with driver BG57 (Fig. 4A), which expresses in larval muscles from L2 stage and in pupal and adult stages Figure 4. AOX rescue of tissue-restricted cIV knockdown. (A–C) Knockdown of Cox5b (Drosophila gene CoVb) using the drivers indicated. Flies with the CyO balancer marker are progeny from the same crosses, but without driver. (A and B) Proportion of eclosing progeny of the genotypes indicated, means+SD for 3 or more biological replicates. (C) Climbing index (defined as in the Materials and methods section) of 1-day-old UAS-AOX transgenic flies bearing the G14 driver, with and without knockdown of Cox5b. Means+SD of 10 batches of 5 flies, for each sex and genotype. Note that the AOX-rescued Cox5b-knockdown males were unable to climb at all in this experiment. (Dand E) Survival curves of flies of the indicated genotypes, following Cox5b or Cox4 (Drosophila gene CoIV)knockdown and transgene expression driven by (D) elav-GAL4 strain 458 or (E) 8760, as shown. Data are means from two independent experiments. See also Supplementary Material, Figure S4. 2084 Human Molecular Genetics, 2014, Vol. 23, No. 8 at Tampere University Library. Department of Health Sciences on September 27, 2016http://hmg.oxfordjournals.org/Downloaded from predominantly in abdominal muscle (Supplementary Material, Fig. S5A). However, knockdown using driver G14, which expresses mainly in the thoracic muscles (Supplementary Material, Fig. S5B), produced lethality (Fig. 4A and B), with flies dying at the late pupal stage or during attempted eclosion. Cox6b knockdown gave similar results (Supplementary Material, Fig. S4A). Co-expression of AOX in the same tissues via UAS-AOX rescued the lethality of Cox5b knockdown by the G14 driver (Fig. 4B). Multiple copies of tub-AOX also rescued this lethality, though less completely (Supplementary Material, Fig. S4C). Flies rescued by UAS-AOX from the muscle-specific lethality of Cox5b knockdown nevertheless manifested a severe locomotor defect (Fig. 4C), and all died within 2 weeks, as a result of becoming trapped in the food. AOX rescue of phenotypes produced by COX knockdown using neuron-specific drivers The failure to induce lethality by COX knockdown in neurons is superficially surprising. We first verified this for knockdown of a further COX subunit (Cox4), which gave a similar result (Supplementary Material, Fig. S4D). Next, we tested a second elav-GAL4 driver located on chromosome X (Bloomington strain 458). Different crosses were implemented, so as to study females heterozygous for the driver, plus males hemizygous for the driver, or control males lacking the driver completely (Supplementary Material, Fig. S4E, Supplementary Material, Table S1), as well as any rescue produced by UAS-AOX (or UAS-Ndi1). In general, elav-GAL4 driver strain 458 produced stronger effects than driver strain 8760, and stronger phenotypes for knockdown of Cox4, which was now almost completely lethal (Supplementary Material, Fig. S4E and F), than for Cox5b. Males were also more affected than heterozygous females. Expression of AOX, but not Ndi1, improved all phenotypes. Viable Cox5b-knockdown flies were transferred to fresh vials and their survival tracked over 2 weeks (Fig. 4D). All genotypes manifested locomotorimpairmentafter several days, which wasmore severe in males. Females with or without co-expression of AOX survived throughout the experiment, but half of the females co-expressing Ndi1 died by 2 weeks. Cox5b-knockdown males had a mean lifespan of just 8 days, but co-expression of AOX enabled most flies to survive at least 2 weeks. In contrast, Cox5b-knockdown males co-expressing Ndi1 all died after only a few days. When knockdown was driven by nrv2-GAL4 or the autosomally located elav-GAL4 driver (Bloomington strain 8760) the outcome of similar survival experiments was much milder. Cox4 knockdown again gave a stronger phenotype than that of Cox5b, and was again stronger in males than females. After 2 weeks, some of the flies with Cox4 knocked down by the nrv2-GAL4 driver had died (though this varied between experiments), whereas Cox5b knockdown using the nrv2-driver did not impair survival over 2 weeks atall. Cox4 knockdown by the autosomal elav-GAL4 driver (Fig. 4E) produced a similar effect as Cox5b knockdown by the stronger, X-chromosomal elavGAL4 driver (Fig. 4D). Knockdown flies were able to climb, and were observed feeding, although mostly remained motionless, whereas flies co-expressing AOX were normally active and able to fly. Most knockdown males died within a few days, a phenotype exacerbated by co-expression of Ndi1, whereas 80% of males co-expressing AOX survived at least 2 weeks. Females were less affected, though followed a similar pattern as with Cox5b knockdown driven by X-chromosomal elavGAL4, with AOX affording protection against early death. To account for the inconsistency between the effects ofknockdown using different neuron-specific drivers we re-examined the tissue-specificity of these drivers, using UAS-GFP (Stinger or mCD8) as a reporter (Supplementary Material, Fig. S5) and by immunocytochemistry and histochemistry (Supplementary Material, Fig. S6). These experiments revealed that expression driven by nrv2-GAL4 driver was indeed much weaker than that driven by elav-GAL4, while the latter gave also a very weak expression in thoracic muscles at pupal stage (see legends to Supplementary Material, Figs S5 and S6 for detailed explanations). Serial sections of the brain and thoracic muscle revealed no gross anatomical defects from Cox4 knockdown using even the strongest elav-GAL4 driver (Supplementary Material, Fig. S6G). AOX partially rescues adult neurodegeneration caused by COX deficiency In order to confirm that the degenerative phenotypes produced by elav-GAL4 driven COX knockdown and partially rescued by AOX expression were indeed neuronal, we adopted two strategies. First, we took advantage of the fact that Cox7a in muscle is predominantly encoded by isogene CG34172, allowing us to knock down the ubiquitously expressed isogene CG9603 specifically in neurons using elav-GAL4, with minimal effects on muscle. Secondly, we analysed the effects of AOX expression in the levy 1 mutant in the gene encoding Cox6a, a subunit required for dimerization of cIV. The splice-site mutation results in a frameshift early in the polypeptide sequence, and produces an adult-onset neurodegenerative phenotype (19). Neuronal depletion of Cox7a driven by elav-GAL4 resulted in locomotor dysfunction (Fig. 5A) and seizure sensitivity at 298C (Fig. 5B), both of which were partially rescued by coexpression of AOX. Neither phenotype was produced by muscle-specific knockdown using the G14 driver (Fig. 5A and B), as expected, given that Cox7a expression in muscle relies mainly on isogene CG34172. Cox7a function is also required during development (Fig. 3C), but AOX-rescue of the lethality gave an adult phenotype similar to that seen in flies specifically knocked down for Cox7a and rescued by AOX co-expression in neurons, namely a mild but progressive locomotor defect (Fig. 5C) and seizure sensitivity at 298C, which was alsoprogressive (Fig. 5D). However, the rescued flies showed normal survival at 2 weeks. When mated to wild-type flies of the opposite sex, males and females were both fertile and their progeny appeared normal. This contrasts with the phenotype exhibited by flies knocked down for subunits of cI, but rescued by Ndi1 (Supplementary Material, Fig. S7), which includes pronounced developmental delay, bang sensitivity at room temperature immediately upon eclosion, and male sterility. Ubiquitous AOX expression using either tub-AOX (Fig. 6A) or UAS-AOX driven by da-GAL4 (Fig. 6B) also alleviated the temperature-dependent seizure sensitivity exhibited by levy 1 mutant flies, although effects on lifespan at 298C were minimal (Supplementary Material, Fig. S8A–C). Expression of UAS-Ndi1 in place of UAS-AOX did not alleviate the seizure sensitivity of levy 1 flies (Supplementary Material, Fig. S8D), instead mildly exacerbated it, similar to the observation that Human Molecular Genetics, 2014, Vol. 23, No. 8 2085 at Tampere University Library. Department of Health Sciences on September 27, 2016http://hmg.oxfordjournals.org/Downloaded from Ndi1 expression enhanced the degenerative effects of Cox5b knockdown by elav-GAL4 (Fig. 6D and E). Although levy 1 was previously suggested to be a null mutation (of Cox6a), we also tested the effects of Cox6a knockdown by RNAi. Ubiquitous knockdown was semi-lethal to females at pupal stage (Fig. 6C), although the degree of male lethality varied between experiments. Eclosing flies failed to inflate their wings and died in the food shortly after eclosion. Coexpression of UAS-AOX partially rescued female lethality (Fig. 6C, Supplementary Material, Fig. S8E), but the resulting flies were still weak. Knockdown of Cox6a using the musclespecific driver G14 did not produce the uninflated wings phenotype (0 of 143 flies analysed), whereas elav-GAL4 did so (223 of 237 flies). Co-expression of AOX rescued this phenotype, with only 9 of 129 flies failing to inflate wings at eclosion (Supplementary Material, Fig. S8E). The Cox6a knockdown phenotype thus appears to be neuronal and again partially rescued by AOX. The more severe organismal phenotype produced by Cox6a knockdown than by the levy 1 splice-site mutation in Cox6a was reflected in a more severe biochemical phenotype as detected by BNE in-gel histochemistry (Fig. 6D). BNE gels revealed a multiplicity of complexes showing COX activity, corresponding to monomeric and dimeric cIV, as well as supercomplexes that were not fully characterized. Knockdown of Cox6a resulted in a clear and reproducible increase in the mobility of the monomeric complex, and a substantial decrease in the abundance of all multimeric and supercomplexes. One of these comigrated on gels with the major complex exhibiting cI activity, and cI activity based on this assay was also clearly decreased in Cox6a knockdown flies, though was hardly affected by the levy 1 mutation. Table 2shows a summary of the different phenotypes produced by knockdown of COX subunits using various drivers, and their alleviation by tub-AOX and/or UAS-AOX in the experiments described above. DISCUSSION AOX can partially replace the functions of COX in vivo Despite a much lower level of expression than with the GAL4 system, AOX expressed constitutively under the a-tubulin promoter was effective in combating diverse insults affecting cIV integrity or activity. Flies were protected from cyanide toxicity and from genetic manipulations of Cox6c (cyclope) or the cIV assembly factor Surf1, to approximately the same extent as when AOX expression was driven by ubiquitously expressed GAL4 (8). These findings indicate, furthermore, that AOX rescue driven by GAL4 or GeneSwitch is not due to promoter dilution effects, and thus validate the use of GAL4 drivers in the remainder of the experiments reported here. The phenotypes associated with knockdown of most nuclearcoded subunits of cIV were significantly alleviated by tub-AOX expression, including Cox4, Cox5a, Cox5b, Cox6a, Cox 6b and Figure5. AOXpartially rescuesneuronal phenotypesresultingfrom Cox7aknockdown.(AandC) Climbingindex and(Band D) bangsensitivityat298C offliesof the indicated genotypes, ages and sex. Means +SD generated by analysis of batches of .50 individual flies of each class. Where, indicated, P,0.001, Student’s ttest, two-tailed, unequal variances, (A and B) comparing flies knocked down for Cox7a (isogene CG9603) with and without co-expression of AOXor (C and D) comparing flies of a given sex expressing AOX, with or without concomitant knockdown of Cox7a. 2086 Human Molecular Genetics, 2014, Vol. 23, No. 8 at Tampere University Library. Department of Health Sciences on September 27, 2016http://hmg.oxfordjournals.org/Downloaded from 45. Dell’agnello, C., Leo, S., Agostino, A., Szabadkai, G., Tiveron, C., Zulian, A., Prelle, A., Roubertoux, P., Rizzuto, R. and Zeviani, M. (2007) Increased longevity and refractoriness to Ca(2+)-dependent neurodegeneration in Surf1 knockout mice. Hum. Mol. Genet.,16, 431–444. 46. So¨rensen, L., Ekstrand, M., Silva, J.P., Lindqvist, E., Xu, B., Rustin, P., Olson, L. and Larsson, N.G. (2001) Late-onset corticohippocampal neurodepletion attributable to catastrophic failure of oxidative phosphorylation in MILON mice. J. Neurosci.,21, 8082–8090. 47. Dufour, E., Terzioglu, M., Sterky, F.H., So¨rensen, L., Galter, D., Olson, L., Wilbertz, J. and Larsson, N.G. (2008) Age-associated mosaic respiratory chain deficiency causes trans-neuronal degeneration. Hum. Mol. Genet.,17, 1418–1426. 48. Burman, J.L., Yu, S., Poole, A.C., Decal, R.B. and Pallanck, L. (2012) Analysis of neural subtypes reveals selective mitochondrial dysfunction in dopaminergicneurons from parkinmutants. Proc. Natl. Acad. Sci. USA,109, 10438–19443. 49. Humphrey, D.M., Parsons, R.B., Ludlow, Z.N., Riemensperger, T., Esposito, G., Verstreken, P., Jacobs, H.T., Birman, S. and Hirth, F. (2012) Alternative oxidase rescues mitochondria-mediated dopaminergic cell loss in Drosophila.Hum. Mol. Genet.,21, 2698–2712. 50. Massie, R., Wong, L.J. and Milone, M. (2010) Exercise intolerance due to cytochrome b mutation. Muscle Nerve,42, 136–140. 51. Sanz, A., Ferna´ndez-Ayala, D.J., Stefanatos, R.K. and Jacobs, H.T. (2010) Mitochondrial ROS production correlates with but does not directly regulate lifespan in Drosophila.Aging,2, 200–223. 52. D’Aurelio, M., Gajewski, C.D., Lenaz, G. and Manfredi, G. (2006) Respiratory chain supercomplexes set the threshold for respiration defects in human mtDNA mutant cybrids. Hum. Mol. Genet.,15, 2157–2169. 53. Stefanatos, R. and Sanz, A. (2011) Mitochondrial complex I: a central regulator of the aging process. Cell Cycle,10, 1528–1532. 54. Bahadorani, S., Cho, J., Lo, T., Contreras, H., Lawal, H.O., Krantz, D.E., Bradley, T.J. and Walker, D.W. (2010) Neuronal expression of a single-subunit yeast NADH-ubiquinone oxidoreductase (Ndi1) extends Drosophila lifespan. Aging Cell,9, 191–202. 55. O’Brien, S.J. and MacIntyre, R.J. (1972) The a-glycerophosphate cycle in Drosophila melanogaster. I. Biochemical and developmental aspects. Biochem Genet.,7, 141–161. 56. Davis, M.B. and MacIntyre, R.J. (1988) A Genetic analysis of the a-Glycerophosphate oxidase locus in Drosophila melanogaster.Genetics, 120, 755–766. 57. Zhang, Y.Q., Roote, J.,Brogna, S.,Davis, A.W., Barbash, D.A., Nash, D. and Ashburner, M. (1999) stress sensitive B encodes an adenine nucleotide translocase in Drosophila melanogaster.Genetics,153, 891–903. 58. Toivonen, J.M., O’Dell, K.M., Petit, N., Irvine, S.C., Knight, G.K., Lehtonen, M., Longmuir, M., Luoto, K., Touraille, S., Wang, Z. et al. (2001) technical knockout,aDrosophila model of mitochondrial deafness. Genetics,159, 241–254. 59. Toivonen, J.M., Manjiry, S., Touraille, S., Alziari, S., O’Dell, K.M. and Jacobs,H.T. (2003)Genedosageand selectiveexpressionmodifyphenotype in a Drosophila model of human mitochondrial disease. Mitochondrion,3, 83–96. 60. Jacobs, H.T., Ferna´ndez-Ayala, D.J., Manjiry, S., Kemppainen, E., Toivonen, J.M. and O’Dell, K.M. (2004) Mitochondrial disease in flies. Biochim. Biophys. Acta,1659, 190–196. 61. Graham, B.H., Li, Z., Alesii, E.P., Versteken, P., Lee, C., Wang, J. and Craigen, W.J. (2010) Neurologic dysfunction and male infertility in Drosophila porin mutants: a new model for mitochondrial dysfunction and disease. J. Biol. Chem.,285, 11143–11153. 62. Oliva, M., De Pinto, V., Barsanti, P. and Caggese, C. (2002) A genetic analysis of the porin gene encoding a voltage-dependent anion channel protein in Drosophila melanogaster.Mol. Genet. Genomics,267, 746–756. 63. Rovio, A.T., Marchington, D.R.,Donat, S.,Schuppe, H.C.,Abel, J., Fritsche, E., Elliott, D.J., Laippala, P., Ahola, A.L., McNay, D. et al. (2001) Mutations at the mitochondrial DNA polymerase (POLG) locus associated with male infertility. Nat. Genet.,29, 261–262. 64. Baklouti-Gargouri, S., Ghorbel, M., Chakroun, N., Sellami,A., Fakhfakh, F. and Ammar-Keskes, L. (2012) The CAG repeat polymorphism of mitochondrial polymerase gamma (POLG) is associated with male infertility in Tunisia. Andrologia,44(Suppl. 1), 68–73. 65. Ruiz-Pesini, E., Lapen ˜a, A.C., Dı´ez-Sa´nchez, C., Pe´rez-Martos, A., Montoya, J., Alvarez, E., Dı´az, M., Urrie´s, A., Montoro, L., Lo´pez-Pe´rez, M.J. and Enrı´quez, J.A. (2000) Human mtDNA haplogroups associated with high or reduced spermatozoa motility. Am. J. Hum. Genet.,67, 682–689. 66. Piomboni, P., Focarelli, R., Stendardi, A., Ferramosca, A. and Zara, V. (2012) The role of mitochondria in energy production for human sperm motility. Int. J. Androl.,35, 109–124. 67. Williams, S.L., Valnot, I., Rustin, P. and Taanman, J.-W. (2004) Cytochrome c oxidase subassemblies in fibroblast cultures from patients carrying mutations in COX10, SCO1, or SURF1. J. Biol. Chem.,279, 7462– 7469. 68. Shishido, E., Takeichi, M. and Nose, A. (1998) Drosophila synapse formation: regulation by transmembrane protein with Leu-rich repeats, CAPRICIOUS. Science,280, 2118–2121. 69. Sykiotis, G.P. and Bohmann, D. (2008) Keap1/Nrf2 signaling regulates oxidative stress tolerance and lifespan in Drosophila.Dev. Cell,14, 76–85. 70. Zordan, M.A., Cisotto, P., Benna, C., Agostino, A., Rizzo, G., Piccin, A., Pegoraro, M., Sandrelli, F., Perini, G., Tognon, G. et al. (2006) Post-transcriptional silencing and functional characterization of the Drosophila melanogaster homolog of human Surf1. Genetics,172, 229– 241. 71. Dietzl, G., Chen, D., Schnorrer, F., Su, K.C., Barinova, Y., Fellner, M., Gasser, B., Kinsey, K., Oppel, S., Scheiblauer, S. et al. (2007) A genome-wide transgenic RNAi library for conditional gene inactivation in Drosophila.Nature,448, 151–156. 72. Barolo, S., Carver, L.A. and Posakony, J.W. (2000) GFP and beta-galactosidase transformation vectors for promoter/enhancer analysis in Drosophila.Biotechniques,29, 726–732. 73. Kemppainen, E., Ferna´ndez-Ayala, D.J., Galbraith, L.C., O’Dell, K.M. and Jacobs, H.T. (2009) Phenotypic suppression of the Drosophila mitochondrialdisease-likemutanttko(25t) byduplicationof themutantgene in its natural chromosomal context. Mitochondrion,9, 353–363. 74. Miwa, S., St-Pierre, J., Partridge, L. and Brand, M.D. (2003) Superoxide and hydrogen peroxide production by Drosophila mitochondria. Free Radic. Biol. Med.,35, 938–948. 75. Stefanatos, R., Sriram, A., Kiviranta, E., Mohan,A., Ayala, V., Jacobs, H.T., Pamplona, R. and Sanz, A. (2012) dj-1bregulates oxidative stress, insulin-like signaling and development in Drosophila melanogaster.Cell Cycle,11, 3876–3886. 76. Ross, J.M. (2011) Visualization of mitochondrial respiratory function using cytochrome c oxidase/succinate dehydrogenase (COX/SDH) double-labeling histochemistry. J. Vis. Exp. 2011;(57):e3266. Human Molecular Genetics, 2014, Vol. 23, No. 8 2093 at Tampere University Library. Department of Health Sciences on September 27, 2016http://hmg.oxfordjournals.org/Downloaded from