Cryostat head sections of 18 day-old adult Canton S control (A, C) and NnaD^PL90 males (B, D). Sections from Canton S control males indicate normal anatomy (A), while head sections through NnaD^PL90 males reveal extensive vacuolization and photoreceptor layer thinning in the retina (B). These differences are especially apparent at higher power, where NnaD^PL90 males display severe disorganization of the retina, with photoreceptor neurons either missing or highly vacuolated (arrows). Mild degeneration of the lamina (L) and medulla (m) is also present in NnaD^PL90 flies. (E-G) Electron micrographs of five day-old adult flies of different genetic constitutions were performed to compare the paracrystalline array of ommatidia. The stereotypical appearance of normal yellow control retina demonstrates seven photoreceptors with their rhabdomeres, surrounded by pigment and support cells (E). In NnaD^PL90 retina, this paracrystalline array is severely disrupted, with many photoreceptors lost or in the process of dying (F). Misshapen rhabdomeres of dying cells with darkened cytosol are thus present throughout the NnaD^PL90 retina. Electron micrographs of retinal sections from NnaD^PL90;eye.GAL4/UAS.NnaD-RB flies reveals restoration of normal retinal cytoarchitecture, consistent with rescue (G). (H) The Benzer countercurrent assay for phototaxis was performed on four day-old flies of the indicated genotypes. Note that a score of ‘5’ is a perfect score in this testing paradigm. LEFT: While NnaD^PL90 mutant flies performed very poorly in this assay, transgenic expression of NnaD or the anti-apoptotic protein p35, driven by eye.GAL4, yielded marked improvement in phototaxis assay performance in the NnaD^PL90 flies (P < .0005, Mann-Whitney test). RIGHT: When we tested four day-old flies carrying the NnaD^RNAi transgene, we documented very poor performance in the Benzer assay (P < .001, Mann-Whitney test). Error bars = s.e.m. Benzer assay analysis of eight day-old NnaD^PL90 flies revealed a significantly worse performance in comparison to four day-old NnaD^PL90 flies (data not shown), consistent with a progressive degenerative phenotype. [See also Figure S2]

A) We transfected Drosophila S2 cells with a CMV-NnaD-HA expression construct, and we determined mitochondrial localization by immunostaining with an anti-HA antibody (red) and an antibody directed against the Complex V ATP synthase (FF0) β-subunit (green). Cells were also stained with DAPI to detect nucleic acid. Deconvoluted fluorescent microscopy revealed co-localization of NnaD with this Complex V intra-mitochondrial protein in merged images of almost all cells expressing NnaD. Scale bar = 10 μm.
B) HEK293T cells were transfected with a CMV-NnaD expression plasmid and immunostained with anti-NnaD serum (green). Mitochondria were visualized using MitoTracker (red) and nucleic acids were stained with Hoechst dye (blue). Confocal fluorescent microscopy revealed localization of NnaD in the mitochondria (yellow). Scale bar = 5 μm.
C) Mitochondrial complex activity measurements. The activity of Complex I, II and V was measured in mitochondrial extracts prepared from yellow (y) controls, FM7 control siblings, and NnaD^PL90 third instar larvae. In larval extracts, the activity of each of these complexes is significantly decreased in NnaD^PL90 mutants. Error bars = s.e.m.
D) Cohorts of seven day-old NnaD^PL90 males and FM7 controls (n = 10 - 20 / group) were reared on 2,4-dinitrophenol (2,4-DNP) mixed into their food at a concentration of either 0.03% or 0.003%. NnaD^PL90 flies displayed significantly reduced viability at both high and low concentrations of 2,4-DNP (P < .05, Fisher's exact test).
E) Cohorts of seven day-old NnaD^PL90 males and FM7 controls (n = 10 - 20 / group) were reared on oligomycin mixed into their food at a concentration of either 0.03% or 0.003%. NnaD^PL90 flies displayed significantly reduced viability at high concentrations of oligomycin (P < .05, Fisher's exact test), but minimal lethal toxicity was noted for the low oligomycin concentration.

A) Genomic organization of the NnaD locus. The gene encoding NnaD (CG32627) is coincident with another gene (CG9940) encoded on the opposite strand. Three transcripts are predicted from the NnaD locus: CG32627-RA, CG32627-RB and CG32627-RC. The position of the mini-white-containing P element (mini-w⁺) in the NnaD^PL90 allele strain is shown.
B) GFP fluorescence in the brain, optic lobes, and ventral nerve cord of a third instar transgenic NnaD.GAL4/UAS.GFP larva.
C) NnaD^PL90 escaper males (bottom) have normal cuticles, but have droopy, opaque wings compared to controls (top).
D) NnaD^PL90 escaper males have a shortened lifespan compared to yw controls, NnaD^excision controls, and FM7 siblings.
[See also Figure S1]

A) Electron micrograph of ommatidia from a seven day-old Drosophila control retina yields normal photoreceptor neuron array and rhabdomere morphology. Scale bar = 2 μm.
B) Electron micrograph of photoreceptors from a seven day-old NnaD^PL90 male retina reveals several different ommatidia at various stages of degeneration. The cell at the upper right (black arrow) has a normal-looking rhabdomere (top right corner) and cytoplasm with normal density. However, the cell immediately below it has darkened cytoplasm, a vacuolated nucleus, and a large, ribosome-studded endoplasmic reticulum lamellar structure (white arrow). Note that the cell marked with an asterisk is severely degenerated. Scale bar = 2 μm.
C) In this higher power view of the ommatidia shown in panel A, numerous mitochondria are observed – all with well-defined cristae and of fairly uniform size. Scale bar = 500 nm.
D) In this higher power view of the most normal ommatidium shown in panel B (black arrow), ultrastructural analysis confirms that profound anomalies are not yet present. However, careful inspection of this micrograph reveals mitochondria of reduced electron density (when compared to panel C). These mitochondria also display poorly defined, irregular, swollen cristae, with occasional vacuolation, indicating that mitochondrial ultrastructural abnormalities occur early in degenerating NnaD^PL90 retina. Scale bar = 500 nm.
E) Quantification of mitochondrial ultrastructural abnormalities in the fly eye. We compared the mitochondria of NnaD^PL90 escaper males and their FM7 control siblings by grading electron density, degree of swelling, amount of vacuole formation, and extent of membrane damage, and then used the combined score to categorize an individual mitochondrion as normal, moderately abnormal, or severely abnormal (n ≥ 4 micrographs / region; n ≥ 4 regions per individual; n = 5 per genotype). The mitochondria in the retinae of NnaD^PL90 flies displayed significantly more structural abnormalities than control flies (P < 0.05, Mann-Whitney U-test). Error bars = s.d.
F) Electron micrograph of a Purkinje cell body and nucleus (Nu) from a wild-type three week-old mouse reveals a nuclear compartment of normal electron density with well-demarcated boundaries. The perinuclear region contains normal ER and abundant normal mitochondria. Scale bar = 2 μm.
G) Electron micrograph of a Purkinje cell body and nucleus (Nu) from a three week-old pcd mouse demonstrates a number of ultrastructural abnormalities. The nucleus is considerably electron-dense and has poorly defined boundaries. ER is swollen (black arrow), and many mitochondria have unusual morphologies, including occasional enlarged, vacuolated mitochondria (white arrow). Scale bar = 2 μm.
H) Electron micrograph of a Purkinje cell body and nucleus (Nu) from a different three week-old pcd mouse again reveals ultrastructrual abnormalities, including vacuole formation (asterisk) and numerous mitochondria with poorly defined cristae (black arrows). Scale bar = 2 μm.

A) We fractionated rat brain or spinal cord lysates and then immunoblotted the mitochondrial fraction (M) and the cytosolic fraction (C) with the indicated antibodies. Nna1 antibody detects Nna1 protein in both mitochondrial and cytosolic fractions in the brain, but not the spinal cord. We confirmed the integrity of the mitochondrial fraction and equivalent loading by probing with anti-Cox IV antibody, and assured the integrity of the cytosolic fraction and equivalent loading by probing with anti-Hsp 90 antibody.
B) We performed sub-mitochondrial fractionations on mouse liver and then immunoblotted the different fractions (unfractionated mitochondria, mito fr.; outer membrane, OM; inter-membrane space, IMS; inner membrane, IM; mitochondrial matrix, matrix) with the indicated antibodies. Nna1 antibody detects Nna1 protein in the mitochondrial outer membrane and the mitochondrial matrix. Boxed bands indicate expected sub-mitochondrial localization for the four controls. Immunoblots were performed in triplicate and the Nna1 protein band was confirmed by running sub-mitochondrial fractions between lanes containing brain protein lysates from pcd^5J mice and positive controls (not shown).
C) Mitochondrial complex I activity in the cerebellum of pcd^5J mice. We prepared mitochondrial fractions from cerebella dissected from four week-old wild-type (wt) and pcd^5J mice (pcd 5J) (n = 3 / group), and then measured complex I activity at 30 sec intervals upon addition of substrate. Complex I activity for pcd^5J mice (pcd 5J) was significantly reduced (P < .001, t-test), as there was decreased magnitude of substrate utilization. Error bars = s.e.m.
D) Increased phosphoglycerate mutase activity in pcd retina. Measurement of phosphoglycerate mutase activity in retinal extracts from heterozygous (het) and pcd^5J (pcd-5J) mice revealed a marked increase in phosphoglycerate mutase activity for the pcd^5J samples (P < .0001; t-test).
E) Decreased pyruvate kinase activity in pcd retina. Assay of pyruvate kinase activity on retinal extracts from wild-type (wt) and pcd^5J (pcd-5J) mice revealed a significant decrease in pyruvate kinase activity for the pcd^5J samples (P < .05; t-test).
F) Aldolase A levels and proteolytic processing are altered in pcd cerebellum. We immunoblotted cerebellar lysates from two sets of two-week old wild-type (wt) and pcd^5J mice with an anti-aldolase A antibody. While overall levels of aldolase A appear elevated in pcd^5J, proteolytic processing of aldolase A to a lower migrating isoform is markedly reduced in pcd^5J cerebellum.
[See also Figure S5]