Changes of Mitochondrial Ultrastructure and Function During Ageing in Mice and 2 Drosophila

1 Changes of mitochondrial ultrastructure and function during ageing in mice and 2 Drosophila

3 Tobias Brandt1†, Arnaud Mourier2,3,4†, Luke Tain5, Linda Partridge5,6, Nils-Göran Larsson2,7, 4 Werner Kühlbrandt1*

6 1Max-Planck-Institute of Biophysics, Department of Structural Biology, Max-von-Laue-Str. 7 3, 60438 Frankfurt am Main, Germany

8 2Max Planck Institute for Biology of Ageing, Department of Mitochondrial Biology, Joseph-

9 Stelzmann-Str. 9b, 50931 Cologne, Germany

10 3Université de Bordeaux, Institut de Biochimie et Génétique Cellulaires UMR 5095, Saint-

11 Saëns, F-33077 Bordeaux, France.

12 4CNRS, Institut de Biochimie et Génétique Cellulaires UMR 5095, Saint-Saëns, F-33077

13 Bordeaux, France.

14 5Max Planck Institute for Biology of Ageing, Department of Biological Mechanisms of 15 Ageing, Joseph-Stelzmann-Str. 9b, 50931 Cologne, Germany

16 6Institute of Healthy Ageing, Department of Genetics, Evolution, and Environment, 17 University College London, Darwin Building, Gower Street, London WC1E 6BT, UK

18 7Department of Medical Biochemistry and Biophysics, Karolinska Institutet, 171 77 19 Stockholm, Sweden

21 * Corresponding author

23 Author contributions. WK and NGL initiated and supervised the project. TB and AM 24 performed and analyzed experiments. NGL, LT and LP provided materials. WK and TB 25 interpreted the data and wrote the paper.

27 † These authors contributed equally to this work

29 Abstract

30 Ageing is a progressive decline of intrinsic physiological functions. We examined the impact 31 of ageing on the ultrastructure and function of mitochondria in mouse and fruit flies 32 (Drosophila melanogaster) by electron cryo-tomography and respirometry and discovered 33 distinct age-related changes in both model organisms. Mitochondrial function and 34 ultrastructure are maintained in mouse heart, whereas subpopulations of mitochondria from 35 mouse liver show age-related changes in membrane morphology. Subpopulations of 36 mitochondria from young and old mouse kidney resemble those described for apoptosis. In 37 aged flies, respiratory activity is compromised and the production of peroxide radicals is 38 increased. In about 50% of mitochondria from old flies, the inner membrane organization 39 breaks down. This establishes a clear link between inner membrane architecture and 40 functional decline. Mitochondria were affected by ageing to very different extents, depending 41 on the organism and possibly on the degree to which tissues within the same organism are 42 protected against mitochondrial damage.

44 Introduction

45 Mitochondria produce most of the ATP in non-photosynthetic eukaryotes, providing the 46 energy to drive a multitude of cellular processes. Mitochondria have an inner membrane, 47 which surrounds the matrix, and an outer membrane, which surrounds the inner membrane 48 and separates the mitochondrial compartments from the cytoplasm. In addition to a multitude 49 of soluble enzymes and ribosomes, the matrix houses the mitochondrial genome (mtDNA), 50 which in humans and flies encodes 13 hydrophobic subunits of the oxidative phosphorylation 51 system. Thus nearly all mitochondrial proteins are nuclear-encoded and imported into the 52 different mitochondrial compartments by a set of protein translocases (1). The oxidative 53 phosphorylation (OXPHOS) system consists of the respiratory chain complexes I to IV and

54 the mitochondrial F1Fo-ATP synthase, sometimes referred to as complex V. Complexes I to 55 IV transfer electrons from soluble electron donors to molecular oxygen. In this process, 56 complexes I, III and IV generate an electrochemical proton potential across the mitochondrial 57 inner membrane that is used by the ATP synthase to produce ATP by rotary catalysis (2). 58 Oxidative phosphorylation occurs mostly, if not entirely, in the deeply invaginated cristae of 59 the inner mitochondrial membrane (3,4). The ATP synthase forms long rows of dimers at the 60 tightly curved cristae ridges, while the respiratory chain complexes are confined to the 61 remaining membrane regions (5-7).

62 Ageing has long been linked to mitochondrial dysfunction. In mammals, the age-related 63 weakening of physiological functions frequently goes along with a decline in health. While 64 there has been considerable progress in the study of age-associated diseases, the biological 65 mechanisms of ageing remain elusive. Ageing is attributed to either ‘programmed’ or 66 ‘damage-based’ processes (8). More than fifty years ago, Harman postulated that ageing 67 results from the accumulation of molecular damage caused by oxygen radicals, often called 68 reactive oxygen species (ROS) (9). Oxygen radicals are side products of electron transfer 69 reactions in respiratory chain complexes I and III (10). Later refinements of Harman’s free 70 radical theory of ageing postulated that an impairment of respiratory chain function increases 71 ROS production, resulting in higher levels of mtDNA damage, which in turn would 72 compromise the turnover of damaged respiratory chain complexes, resulting in a vicious 73 circle of damage and decline (11). The mitochondrial DNA mutator mouse model has 74 provided experimental evidence of a causative link between mtDNA mutations and an ageing 75 phenotype in mammals (12), although mitochondrial defects in this model were found not to 76 be associated with an increase in ROS production (13). Several other key predictions of the

77 free-radical theory (14) have also not been substantiated. For example, in some model 78 organisms, such as the nematode Caenorhabditis elegans, moderately elevated levels of ROS, 79 induced either chemically or through genetically engineered defects in the respiratory chain, 80 actually increase lifespan (15-17). In the same organism, removal of the ROS-scavenging 81 superoxide dismutase (SOD) does not increase oxidative damage to mtDNA and has no 82 apparent effect on life span (18-20). In the fly, SOD knockouts do have a decreased lifespan, 83 but they do not accumulate mtDNA mutations more quickly than wildtype (21).

84 In addition to energy conversion and ATP production, the oxidative phosphorylation system 85 also has a key role in shaping the inner membrane cristae (5-7). Accumulating oxidative 86 damage and OXPHOS dysfunction might, therefore, be expected to affect mitochondrial 87 ultrastructure, which should be visible by electron microscopy. Recent analyses of yeast and 88 mouse mitochondria show that a defect in supra-molecular organisation of the ATP synthase 89 results in aberrant cristae morphology (5,22). In aged cultures of the short-lived filamentous 90 fungus Podospora anserina, the inner membrane strikingly vesiculates and ATP synthase 91 dimers break down (23,24). However, little is known about the impact of ageing on 92 mitochondrial morphology and membrane structure in metazoans. Previous studies of the 93 ultrastructure of mitochondria in mouse liver by electron microscopy of thin plastic sections 94 revealed anomalous cristae in a subpopulation of the organelles (25). A study of mouse 95 skeletal muscle mitochondria found that occasionally mitochondria from trained animals lost 96 their cristae, whereas no differences were observed in non-trained animals (26). Mitochondria 97 from aged D. melanogaster flight muscle were found to have cristae ‘swirls’ that were 98 attributed to oxidative damage (27), and mitochondria in aged Drosophila repleta heart 99 muscle were enlarged (28).

100 We chose D. melanogaster (average life span >50 days depending on environmental 101 conditions (29)) and mouse (average life span 107 weeks for females and 114 weeks for males 102 (30)) as two well-established metazoan ageing models (31) for a systematic study of 103 mitochondrial ultrastructure, respiration and ROS production in young and old animals. Our 104 results reveal clear tissue- and organism-related age-specific differences, establishing an 105 apparent link between mitochondrial ultrastructure and function.

106 Results

107 Effects of ageing on function and ultrastructure of mouse mitochondria

108 (a) Mitochondrial OXPHOS activity and ROS homeostasis

109 To find out how ageing impacts mitochondrial function and ultrastructure, mitochondria from 110 heart, liver and kidney of young (20 weeks) and old (80-96 weeks) mice were isolated for 111 electron cryo-tomography (cryoET) and high-resolution respirometry. Isolated mitochondria 112 were incubated with respiratory substrates that deliver electrons at the level of complex I 113 (pyruvate, glutamate and malate) or complex II (succinate and rotenone). The mitochondrial

114 oxygen consumption was recorded under phosphorylating (addition of ADP and Pi; state 3), 115 non-phosphorylating (addition of oligomycin to inhibit ATP synthase; state 4) and uncoupled 116 conditions (addition of CCCP). Mitochondrial respiratory rates normalised to protein content 117 were found to be highest in heart mitochondria, ~2-3-fold lower in kidney and ~5-fold lower 118 in liver (Figure 1A). In line with previous reports (32) and liver (33,34), our investigation 119 showed no significant differences in respiratory rates between mitochondria isolated from 20 120 week-old and 80 week-old heart tissue (Figure 1A), and no age-related change in the rate of 121 ATP production was observed (Figure 1C). Moreover, maximal activities of key 122 mitochondrial enzymes remained unchanged (Figure 1E). These results prompted us to 123 investigate how ageing impacts mitochondrial ROS homeostasis. We first assessed the 124 peroxide yield, defined as the hydrogen production rate normalized to the mitochondrial 125 respiration assessed under the same conditions, of isolated mitochondria (35). Surprisingly, no 126 age-dependent increase in hydrogen peroxide release relative to the amount of oxygen 127 consumed was observed in the three different tissues (Figure 1B). To investigate ROS 128 homeostasis further, we analysed the steady-state levels and activity of the anti-oxidant 129 enzymes superoxide dismutase (SOD1, SOD2) and catalase in young and old heart, liver and 130 kidney tissue by quantitative western blot electrophoresis and gel densitometry (Figure 1D). 131 This revealed a tendency towards lower levels of anti-oxidant enzymes in aged tissue. In old 132 mouse liver, catalase activity was reduced by 30% (Figure 1 F) and SOD activity was almost 133 halved in old mouse kidney (Figure 1 – figure supplement 1).

134 (b) Tomography of mouse heart mitochondria

135 Vitrified samples of the mitochondrial preparations used for high-resolution respirometry 136 were analysed by electron cryo-tomography (cryo-ET). Mitochondria isolated from young 137 mouse heart showed morphologies typical of tissues with a high energy demand. Stacks of

138 parallel, thin and flat lamellar cristae were embedded in a dense matrix (Figure 2, left, Video 139 1). Cristae frequently appeared discontinuous in 2D slices, but 3D volumes indicated that this 140 was due to fenestration of the lamellar disks rather than to disconnected cristae vesicles 141 (Figure 2 – figure supplement 1A). Crista junctions were circular (70%, average diameter 15 142 ± 2 nm) or slightly elongated (30%, 16 ± 1 x 29 ± 2 nm, dimensions ± standard deviations 143 measured in the typical tomogram of figure 2, top row, second from left). Most cristae were 144 branched and connected to the inter-membrane space by more than one junction. They were 145 also highly interconnected through narrow apertures (Figure 2 – figure supplement 1B). In 146 20% of the mitochondria examined, membranes of two or more neighbouring cristae were so 147 closely appressed that there was almost no matrix between them. These regions often showed 148 membrane ‘swirls’ of high and variable membrane curvature, involving several neighbouring 149 crista lamellae (Figure 2 – figure supplement 1C). There was no apparent difference between 150 the structures of isolated sub-sarcolemmal and interfibrillar mitochondria. Comparing 151 mitochondria from young and old mouse hearts indicated similar morphologies (Figure 2, 152 right), except that about 24% of mitochondria from aged heart tissue had some exceptionally 153 wide cristae (Table 1; Figure 2 – figure supplement 1D).

154 (c) Tomography of mouse liver mitochondria

155 The morphology of mitochondria isolated from mouse liver was conspicuously different from 156 that of mouse heart (Figure 3, left, Video 2). The cristae were more heterogeneous, less 157 regular and not arranged in parallel stacks. They were generally wider and did not span the 158 entire mitochondrion. Also, the cristae were less interconnected to one another and to the 159 inner boundary membrane. As in cardiac mitochondria, fenestration was observed, although it 160 was less frequent. The matrix was very dense, making it difficult to segment larger 161 mitochondria. In one sample, the matrix contained dense granules up to 50 nm in diameter. 162 No such granules were found in heart mitochondria (Video 2). In two instances, mitochondria 163 had central low-density compartments we refer to as voids (see below).

164 Tomograms of mouse liver mitochondria from aged animals revealed two different 165 phenotypes. While the majority of mitochondria looked similar to those isolated from young 166 mouse liver, 32% (n = 31) had conspicuous low-density compartments, or voids, in the centre 167 of the organelle (Figure 3, right, Video 3; Table 1). Segmentation of 3D volumes revealed that 168 the membrane delineating these central voids was continuous with the inner membrane. In 169 three segmented mitochondria, the voids accounted on average for 22% of the total cristae 170 surface. To estimate their volumes, voids and mitochondria were approximated as simple

171 geometrical shapes (spheres or cylinders for voids, cylinders for mitochondria). In the three 172 segmented organelles, the voids occupied on average 6% of the total mitochondrial volume. 173 The voids were not internal vesicles, but were connected to the inter-membrane space via 174 apertures of varying size. In some cases two such connections were observed on opposite ends 175 of the voids, resulting in a toroid or doughnut-shaped matrix. In other cases, the outer 176 membrane appeared to protrude into the inner-membrane voids. The cristae extending from 177 the boundary membrane into the matrix looked normal, but none protruded into the matrix 178 from the membrane defining the voids. In two tomograms, dense granules measuring up to 179 100 nm in diameter were observed in the matrix.

180 (d) Tomography of mouse kidney mitochondria

181 Kidney mitochondria resembled heart mitochondria more closely than those from liver 182 (Figure 4A). Mitochondria from young and old mouse kidney were similar, except that dense 183 matrix granules were found in 18% of samples from old animals compared to only one (3%) 184 observed for samples from young animals. Cristae were lamellar but generally wider than 185 those in cardiac tissue and less stringently arranged in parallel stacks. Occasionally cristae 186 formed membrane swirls. 18% of kidney mitochondria from old mice and 12% from young 187 mice had the characteristic morphology described for apoptotic cells (36,37). In these 188 mitochondria, the cristae were irregular, not arranged in any discernible pattern, and the 189 membrane curvature was locally inverted. Crista junctions were very wide or not discernible, 190 resembling those of prohibitin-deficient mouse mitochondria, in which OPA1 (that is essential 191 for cristae junction formation) is incorrectly processed (38). The inner membrane enclosed a 192 convoluted but apparently continuous volume. The width of the intermembrane space 193 between the inner boundary and outer membrane was largely the same as in normal heart, 194 liver or kidney (Figure 4B), indicating that the mitochondria were intact and had not suffered 195 from osmotic shock during isolation (39). Results are summarized in Table 1.

196 (e) Mutator mouse mitochondria

197 Next, we compared the function and ultrastructure of wild-type mouse mitochondria to those 198 from heart and liver of mtDNA mutator mice, a strain with a premature ageing phenotype 199 (12). A minor respiratory defect was found under phosphorylating and uncoupling conditions 200 in heart from mtDNA mutator mice (Figure 5A), consistent with previous observations 201 (12,13). The morphology of heart mitochondria from the mtDNA mutator mice resembled 202 wild-type heart mitochondria, except for a high incidence (40%) of ellipsoidal peripheral

203 voids connected to the cristae or the intermembrane space that were delineated by a double 204 membrane (Figure 5B). The matrix space between the two membranes was minimal. In one 205 mtDNA mutator mouse heart mitochondrion, the outer membrane was punctured by small 206 openings (Figure 5B, left).

207 Mitochondria isolated from mtDNA mutator mouse liver showed the same central low-density 208 voids (Figure 5C) as liver mitochondria from old wild-type mice (Figure 3), but to a larger 209 extent (40%; Table 1). In one case, cristae with apparently normal junctions extended from 210 the central void into the matrix, which was not observed in wild-type mice (Figure 5C). About 211 20% of mtDNA mutator mouse liver mitochondria contained dense matrix granules (Table 1).

212 D. melanogaster mitochondria show profound age-associated changes

213 We also examined the ultrastructure and function of mitochondria from young and old D. 214 melanogaster as a non-mammalian metazoan. The flies showed standard sigmoidal survival 215 curves, with a mean life span of 69.5 days and a maximum life span of 78.5 days (Figure 6A). 216 Life span correlated with climbing ability, as reported previously (40). Climbing ability 217 remained constant for 25 days and then dropped rapidly (Figure 6A). We investigated the 218 respiratory activity of mitochondria from young and old flies (15 and 70 days, respectively; 219 Figure 6A). The respiration rate in old flies decreased by up to 60% (Figure 6B), and the 220 peroxide yield increased by 40-100% (Figure 6C), indicating that mitochondrial function is 221 severely compromised, confirming an earlier report (41).

222 We performed cryo-ET on three sample preparations per age group to find out whether and 223 how this functional decline is reflected in mitochondrial morphology. Results are summarized 224 in Table 1. Mitochondria from young flies (n = 29; Figure 6D, Video 4) looked similar to 225 those from mouse heart, with a dense matrix and thin, lamellar, mostly parallel cristae. All 226 cristae had ridges indicative of ATP synthase dimer rows (6,42) and several junctions to the 227 intermembrane space. Wide cristae were observed in 10% of the mitochondria, similar to 228 mouse heart. A small subpopulation of mitochondria (10%) was unusually elongated (axial 229 ratio above 3:1) but looked otherwise normal. By contrast, the structure of mitochondria from 230 old D. melanogaster (n = 39) was heterogeneous (Figure 6E). Approximately 75% had the 231 standard morphology. Some had wide cristae (15%), other mitochondria were unusually long 232 and thin (18%), or branched (5%) but with well-developed cristae (Figure 6E, top left). In the 233 remaining 25% of old fly mitochondria, the inner membrane assumed a variety of non- 234 standard shapes. Several organelles identified as mitochondria by their characteristic double

235 membrane and dense matrix appeared to lack cristae entirely (Figure 6E, top right) or the 236 cristae were minimally developed (10%). Many cristae were not connected to the 237 intermembrane space and were, therefore, small vesicles completely surrounded by matrix 238 (Figure 6E, lower right, Video 5). In one mitochondrion, the cristae were spherical (Figure 239 6E, centre left). In two others they were concentric (Figure 6E, lower left, Video 6), lacking 240 the membrane ridges associated with ATP synthase dimer rows (6,42). These observations 241 establish a strong link between mitochondrial inner membrane organisation and respiratory 242 function in D. melanogaster.

243

244 Discussion

245 Our study provides the first systematic assessment of changes in mitochondrial function and 246 inner membrane structure upon ageing in two common metazoan ageing model organisms, 247 mouse and D. melanogaster. 3D volumes of entire mitochondria at an estimated resolution of 248 3 to 5 nm were generated by cryo-ET. As far as possible, mitochondria from different 249 organisms and tissues were treated in the same way. The fact that there are clear differences 250 e.g. between mitochondria from young and old mouse liver, and that these look different from 251 heart and kidney mitochondria allows us to conclude that the isolation process itself does not 252 affect mitochondrial membrane structure significantly. We conclude further that the different 253 appearance of the young mouse liver mitochondria is not an artefact of isolation, but a 254 genuine feature.

255 We found a clear correlation between OXPHOS capacity and the number of inner membrane 256 cristae per unit volume. Mitochondrial respiration was highest in mouse heart and young fly 257 mitochondria, which were entirely filled with closely stacked, parallel cristae. This 258 arrangement, which is characteristic for tissues with high respiratory activity, maximises the 259 membrane area available for oxidative phosphorylation (6). In mouse heart and young fly 260 mitochondria, we frequently observed cristae fenestration, which is a morphological 261 characteristic of tissues with high energy demand (43,44). Fenestration increases the total 262 length of cristae ridges that harbour the ATP synthase, and hence the potential for ATP 263 production. By comparison, mouse liver mitochondria had fewer cristae and the matrix was 264 both denser and more voluminous, in line with the lower respiratory activity and higher 265 metabolic activity of liver cells. Mouse kidney mitochondria were structurally more diverse. 266 Remarkably, a significant proportion of kidney mitochondria from both young and old mice 267 had a morphology typical of apoptotic cells (36,37). This correlates well with the recently 268 reported continuous turnover and short lifespan of kidney cells of about 30-60 days (45), 269 compared to the 200-400 day lifespan of rat hepatic cells (46) and the very low turnover rates 270 for cardiomyocytes of 1.3-4% per year (47). The peroxide yield correlated inversely with 271 respiratory activity and was highest in mouse liver and lowest in mouse heart, suggesting that 272 liver mitochondria may be affected by oxidative damage more severely than mitochondria 273 from cardiac tissue.

274 Unexpectedly, levels of mitochondrial respiration, respiratory chain activity, hydrogen 275 peroxide production or steady-state levels of anti-oxidant enzymes did not vary greatly with 276 age, indicating at most moderate changes in respiratory rates and hydrogen peroxide yield in

277 ageing mice. This contravenes the free-radical theory of ageing, which postulates a major age- 278 related increase in ROS production and respiratory defects. In line with our results on 279 mitochondrial function, the morphology of mouse heart mitochondria did not change much 280 with age. By contrast, one out of four mitochondria from old mouse livers had a striking, age- 281 specific phenotype characterized by a central matrix void. We found the same feature in 282 mitochondria of mtDNA mutator mouse livers, where it was considerably more prevalent. 283 Although it is at present unknown what causes the voids and how they affect mitochondrial 284 function, they evidently reflect a specific reorganisation rather than a random breakdown of 285 the inner membrane. The lack of sharp cristae ridges and crista junctions in the void 286 membranes suggests that they do not contain ATP synthase dimers and hence do not 287 contribute to oxidative phosphorylation. The fact that mitochondria with these central voids 288 have otherwise normal cristae may explain why no age-related reduction in overall respiratory 289 activity was evident. Apparently, if the majority of mitochondria are normal, they can 290 compensate for a loss of inner membrane area available for oxidative phosphorylation in 25% 291 of the total population. Similar ring- or cup-shaped mitochondria with central cavities have 292 been reported (48) and ascribed to the effect of drugs, toxins (49,50), or oxidative damage 293 (51). Although the cavities found in these earlier studies were lined by both the outer and 294 inner mitochondrial membranes, the central voids described here may likewise result from 295 such damage, as damage to mtDNA is known to accumulate in liver (52,53). If denatured 296 respiratory chain complexes are cleared from the membrane, the lipids left behind would be 297 expected to form such featureless membrane regions expanding into the matrix interior. The 298 matrix granules that we found in old liver and kidney mitochondria may consist at least in part 299 of denatured respiratory chain complexes, consistent with their absence from old heart 300 mitochondria, where we did not observe matrix granules. Indeed, it has been shown that these 301 granules contain complex IV subunits (54), in addition to lipids, glycoproteins and calcium 302 (55,56).

303 Interestingly, changes in the membrane structure of mouse mitochondria were organ- 304 dependent, in a way that suggests different degrees of resilience against ageing. Mouse 305 cardiac mitochondria showed the highest respiratory rates but appeared to be protected most 306 effectively from oxidative damage, as indicated by their low and unchanged peroxide yield. 307 Again, these findings contradict the classical free-radical theory, which would predict that 308 cardiomyocytes with their high density of highly active mitochondria produce more ROS and 309 thus age faster (14). An analysis of DNA methylation has shown that heart tissue ages more 310 slowly than would be predicted chronologically (57). This observation is in line with our

311 findings, which thus imply that mitochondria in vital organs with slow regeneration, such as 312 the heart, are protected from damage more effectively than mitochondria in other organs. The 313 effect may be genetically controlled through the expression levels of superoxide dismutase or 314 catalase. Overexpression of catalase targeted to mitochondria has been shown to attenuate 315 ageing effects in the murine heart (58). However, ROS signalling (59) requires a fine 316 equilibrium of ROS production and sequestration that is probably tissue-dependent.

317 The most striking differences in our study, both with respect to morphology and activity, were 318 found between mitochondria from young and old D. melanogaster. About half of the 319 mitochondria from old flies had lost the standard organization of the inner membrane into 320 boundary membranes with well-developed cristae and crista junctions. Many had round or 321 concentric cristae that lacked sharp ridges. In extreme cases, cristae or crista junctions were 322 completely absent. Measurements of respiratory activity in old fly mitochondria indicated that 323 oxygen uptake was decreased by a factor of almost 2 and peroxide yield was increased by

324 more than 50%, in line with increased mitochondrial H2O2 production in live ageing 325 Drosophila (60). These results suggest strongly that about 50% of the old fly mitochondria are 326 inactive, consistent with the observation that about 25% of the mitochondria in old flies lack 327 normal cristae or crista junctions and an additional 18% deviate from the standard 328 morphology. Such a drastic breakdown of mitochondrial structure and function would result 329 in the death of the organism within a short period.

330 It is interesting to compare our results on mouse and fruit fly mitochondria to similar studies 331 on the filamentous fungus P. anserina (23,24), another well-characterized ageing model (61). 332 P. anserina has an average lifespan of only 18 days, three times shorter than D. melanogaster, 333 and almost 50 times shorter than mouse. Morphological changes in P. anserina mitochondria 334 were both more homogenous and more extreme than in fruit flies, affecting about 80% of 335 organelles from senescent populations (23). Cryo-ET of aged P. anserina mitochondria or 336 inner membrane vesicles indicated that the cristae had receded into the inner boundary 337 membrane and that ATP synthase dimers dissociated into monomers (24). In terms of inner 338 membrane morphology, old P. anserina mitochondria resembled the quasi-apoptotic 339 subpopulation in mouse kidney, the tissue with the highest turnover rate in our study, 340 suggesting similar mechanisms of programmed ageing. On the one hand, the higher 341 proportion of functional mitochondria in old flies indicates that the decline is less complete 342 and slower than in P. anserina. On the other hand, it is much more rapid in D. melanogaster

343 than in mouse, suggesting a link between the complexity of an organism and the rate of 344 ageing.

345 Conclusions

346 The increasing complexity of organisms goes along with an increasingly complex ageing 347 process. For the primitive multicellular eukaryote P. anserina, a simple, clear correspondence 348 between age, mitochondrial ultrastructure and organization of the mitochondrial ATP 349 synthase has been shown (23,24). In Drosophila, we now establish a clear link between inner 350 membrane morphology and functionality, which correlates closely with age and agility. In 351 mouse, the relationship between inner membrane ultrastructure, function and age is less clear- 352 cut and evidently tissue-dependent. While mouse heart mitochondria show little if any change 353 with age, a quarter of liver mitochondria display a severe, age-related phenotype that does 354 however not seem to result in an overall reduction of oxidative phosphorylation. A relatively 355 high proportion of kidney mitochondria in young and old mice resemble those observed 356 during apoptosis, consistent with the high turnover of kidney cells. Our data thus indicate 357 major differences in how ageing relates to mitochondrial morphology and function in 358 metazoans. In mouse, we find no evidence of age-related progressive impairment of the

359 oxidative phosphorylation system or an increase of mitochondrial H2O2 production, whereas 360 both effects are evident in Drosophila.

361

362 Materials and methods

363 Mouse breeding

364 All mice in this study were on an inbred C57Bl/6N nuclear background. Mutator mice were 365 generated as previously described (62). Briefly, PolgAWT/mut mice were generated by crossing 366 a PolgAWT/WT female with a PolgAWT/mut male, and subsequently intercrossed to generate 367 PolgAmut/mut (mutator mice). Mice were maintained on a standard mouse chow diet and 368 sacrificed at different time points by cervical dislocation in strict accordance with the 369 recommendations and guidelines of the Federation of the European Laboratory Animal 370 Science Association (FELASA). Protocols were approved by the Landesamt für Natur, 371 Umwelt und Verbraucherschutz, Nordrhein-Westfalen, Germany (Permit ref: 84- 372 02.05.20.12.086).

373 Fly breeding

374 wDah wild-type flies were maintained at 25°C and fed a standard sugar/yeast/agar diet (SYA). 375 Once mated, females, raised at controlled larval densities, were used. Adult flies were kept in 376 SYA food vials containing 10 or 25 flies per vial for survival and climbing analysis, 377 respectively. Climbing ability was assessed as previously described (63).

378 Isolation of mouse mitochondria

379 Mice were sacrificed by cervical dislocation, and heart, liver and kidneys were quickly 380 collected in ice-cold DPBS (Gibco), minced and homogenized with a few strokes of a Potter S 381 homogenizer (Sartorius) in 5 ml (for heart and kidney) or 20 ml (for liver) of ice-cold 382 mitochondria isolation buffer (MIB; 310 mM sucrose, 20 mM Tris-HCl, 1 mM EGTA, pH 383 7.2). Mitochondria were purified by differential centrifugation (1200g for 10 minutes) and the 384 supernatants were then centrifuged at 12000g for 10 minutes. The crude mitochondrial pellet 385 was resuspended in an appropriate volume of MIB. Mitochondrial concentration was 386 determined using the Protein DC Lawry based assay (Bio-Rad).

387 Isolation of mitochondria from fruit flies

388 Fruit flies were homogenized with a few strokes of a loose Potter S homogenizer (Sartorius) 389 in 5 ml of ice-cold mitochondria isolation buffer (MIB; 310 mM sucrose, 20 mM Tris-HCl, 1 390 mM EGTA, pH 7.2). After filtration through a 100 µm nylon mesh filter, mitochondria were 391 further homogenized in a tight Potter S homogenizer (Sartorius) and purified by differential

392 centrifugation (800g for 10 minutes) and the supernatant was then centrifuged at 4500g for 15 393 minutes. The crude mitochondrial pellet was resuspended in an appropriate volume of MIB. 394 Mitochondrial concentration was determined using the Protein DC Lowry based assay (Bio- 395 Rad).

396 Mitochondrial respiratory assay

397 The rate of mitochondrial oxygen consumption was measured as previously described (64) at 398 37°C using 65–125 μg of crude mitochondria in 2.1 ml of mitochondrial respiration buffer

399 (120 mM sucrose, 50 mM KCl, 20 mM Tris-HCl, 4 mM KH2PO4, 2 mM MgCl2, 1 mM 400 EGTA, pH 7.2) in an Oxygraph-2k (Oroboros Instruments). Oxygen uptake was measured 401 using either pyruvate/glutamate/malate (10 mM pyruvate, 5 mM glutamate and 5 mM malate) 402 or 10 mM succinate and 10 nM rotenone. Oxygen consumption was assessed under 403 phosphorylating conditions with 1 mM ADP (state 3) or non-phosphorylating conditions by 404 adding 2.5 μg/ml oligomycin (pseudo state 4). Respiratory control ratios (65) were above 10 405 with Pyruvate/glutamate/malate and above 5 with succinate-rotenone. Respiration was 406 uncoupled by successive addition of carbonyl cyanide m-chlorophenyl hydrazone (CCCP) up 407 to 3 μM to reach maximal respiration. The same procedure was used for D. melanogaster, 408 except that isolated mitochondria were first incubated with pyruvate/glutamate/malate/proline 409 (10 mM pyruvate, 5 mM glutamate, 5 mM malate, 10 mM proline) for state 3, state 4 and the 410 uncoupled state, respectively. Succinate (10 mM), glycerol-3-phosphate (10 mM) and finally 411 rotenone (10 nM) were then added to determine the maximal respiration driven by succinate 412 and glycerol-3-phospate versus complex I driven-respiration.

413 Measurement of ATP synthesis flux (JATP)

414 Isolated mitochondria (65 µg/ml) were suspended in respiration buffer (see above). After 415 addition of 2 mM succinate, 10 nM rotenone and 1 mM ADP, oxygen consumption and ATP 416 synthesis rates were measured as previously described (66). Aliquots were collected every

417 20 s and precipitated in 7% HClO4/25 mM EDTA and were centrifuged at 16000g for 10 418 minutes and then neutralized with 2 M KOH, 0.3 M MOPS. The ATP content in these 419 samples was determined with the ATPlite 1step (PerkinElmer). In a parallel experiment, 420 oligomycin (2.5 μg/ml protein) was added to the mitochondrial suspension to determine the 421 rate of non-oxidative ATP synthesis.

422

423

424 Measurement of reactive oxygen species

425 The rate of H2O2 production was determined by monitoring the fluorescence emission at 426 590 nm upon oxidation of the indicator dye Amplex Red (5 U/ml) in the presence of 427 horseradish peroxidase (1 µM) with excitation at 560 nm. A standard curve was obtained by

428 adding known amounts of H2O2 to the assay medium in the presence of the reactants. 429 Mitochondria (65 µg protein ml-1) were incubated in respiratory medium (see above) at 37°C.

430 H2O2 production was initiated by substrate addition, and the rate was determined by 431 monitoring the fluorescence change with time (35).

432 Enzyme activities

433 Tissue proteins (15-50 μg) were diluted in phosphate buffer (50 mM KH2PO4, pH 7.4) 434 followed by spectrophotometric analysis of isolated respiratory chain complex activities at 435 37°C using a Hitachi UV-3600 spectrophotometer. Citrate synthase activity was measured at 436 412 nm (ε = 13,600 M−1cm−1) after addition of 0.1 mM acetyl-CoA, 0.5 mM oxaloacetate and 437 0.1 mM 5,5-dithiobis-2-nitrobenzoic acid (DTNB). NADH dehydrogenase activity was 438 determined at 340 nm (ε =6220 M-1cm-1) after addition of 0.25 mM NADH, 0.25 mM 439 decylubiquinone and 1 mM KCN and monitoring rotenone sensitivity. Succinate 440 dehydrogenase (SDH) activity was measured at 600 nm (ε =21000 M-1cm-1) after addition of 441 40 mM succinate, 35 μM dichlorphenol indophenol (DCPIP) and 1 mM KCN. COX activity 442 was assessed using a classical TMPD/ascorbate assay. Briefly, homogenized tissue (65 µg/ml) 443 was suspended in mitochondrial respiration buffer (see above). Oxygen consumption was 444 assessed in the presence of TMPD (0.2 mM), ascorbate (1 mM) and antimycin A (0.5 µM). 445 After a few minutes of stationary respiration, KCN (2 mM) was injected into the chamber. 446 COX activity corresponds to the KCN-sensitive respiration. Catalase activity was assessed 447 using an Oroboros oxygraph. Catalase activity of homogenized tissues (65 µg/ml) was

448 followed by recording the oxygen production in the presence of 0.01% H2O2. All chemicals 449 were obtained from Sigma Aldrich.

450 Western blot analysis

451 Proteins from tissue lysates were separated by SDS-PAGE and blotted onto PVDF 452 membranes (GE Healthcare). The following primary antibodies were used: rabbit anti- 453 superoxide dismutase 1 (1:1000, Abcam-ab16831) and rabbit monoclonal anti-superoxide

454 dismutase 2 (1:1000, Millipore-06-984). The following HRP-conjugated secondary antibodies 455 were used: donkey anti-rabbit IgG (Amersham, NA9340V) and sheep anti-mouse 456 (Amersham, NXA931). For chemiluminescence detection, samples were incubated with ECL 457 (GE Healthcare). Densitometry analyses were performed using the FIJI software.

458 Electron cryo-tomography

459 For cryo-ET, mitochondria were washed twice with 320 mM trehalose, 20 mM Tris pH 7.3, 1 460 mM EGTA. Samples were mixed 1:1 with fiducial gold markers (10 nm gold particles 461 conjugated to protein A, Aurion, Netherlands) and immediately plunge-frozen in liquid ethane 462 on Quantifoil holey carbon grids (Quantifoil Micro Tools, Germany). Single tilt series 463 (typically ±60°, step size 1-1.5°) were collected at 300 kV with an FEI Polara electron 464 microscope equipped with a post-column Quantum energy filter and an Ultrascan 4x4k CCD 465 camera (Gatan, USA), or with a post-column Tridiem energy filter and a 2x2k CCD camera 466 (Gatan). Alternatively, tilt series were recorded with an FEI Titan Krios electron microscope 467 equipped with a Quantum energy filter and a K2 summit direct electron detector (Gatan). 468 Underfocus was 8-9 µm and the magnification was chosen to give an object pixel size 469 between 4.3 Å and 7.3 Å. The total electron dose per tilt series was 120-150 e-/Å². Tilt series 470 were aligned to gold fiducial markers and tomograms were reconstructed by back-projection 471 with the IMOD software package (67). A final filtering step applying non-linear anisotropic 472 diffusion (68) was performed to increase contrast. Tomograms were manually segmented with 473 the program AMIRA (FEI).

474

475 Acknowledgments

476 We acknowledge funding from the Max Planck Society and the Bundesministerium für 477 Bildung und Forschung (SyBACol 0315893A-B) (to L.S.T.). NGL received additional 478 support from the Swedish Research Council (2015-00418) and the Knut and Alice 479 Wallenberg foundation.

480

481 Competing interests

482 WK: Reviewing editor, eLife. The other authors declare that they have no competing 483 interests.

484 Figure legends

485 Figure 1: Bioenergetic and functional analysis of mitochondria from heart, liver and kidney 486 from young (white, 20 weeks old) and old (black, 80-96 weeks old) mice. (A): Oxygen 487 consumption rate of mitochondria isolated from young (n = 7-8) or old (n = 8) animals. 488 Isolated mitochondria were incubated with electron donors to complex I (pyruvate, glutamate, 489 malate) or complex II (succinate, complex I inhibited with rotenone). Each set of substrates 490 was successively combined with ADP (to assess the phosphorylating respiration, state 3), 491 oligomycin (to measure non-phosphorylating respiration, state 4) and finally uncoupled by 492 adding increasing concentrations of CCCP. (B) Mitochondrial peroxide yield assessed in 493 mitochondria from young (n = 3-4) and old (n = 3-4) animals. (C) Mitochondrial ATP 494 synthesis rate in heart mitochondria from young (n = 4) and old (n = 3) animals. Error bars 495 indicate mean ± standard error of the mean (SEM). (D) Steady state levels of different anti- 496 oxidant enzymes in heart, liver and kidney mitochondria isolated from young (y) and old (o) 497 mice were quantified by Western blot analyses. Long (LE) and short exposure (SE) times are 498 presented for catalase and SOD1 detection. (E) Citrate synthase (CS) and respiratory chain 499 enzyme activity (complex I, II, IV) measurements in heart, liver and kidney tissue extracts 500 from young (n = 4), and old (n = 4) animals. Error bars indicate mean ± SEM. (F) Catalase 501 enzyme activity measured in heart, liver and kidney tissue extracts from young (n = 4) and old 502 (n = 4) animals.

503

504 Figure 2: Cryo-ET of heart mitochondria from young (left, 20 weeks old) and old (right, 80- 505 96 weeks old) mice. Upper panel: slices through tomographic volumes (scale bars, 250 nm). 506 Lower panels: segmented 3D volumes of two typical mitochondria with closely stacked, 507 roughly parallel cristae (blue). Grey, inter-membrane space; light yellow, outer membrane 508 (omitted for clarity in the right panel). Cristae are connected to the intermembrane space by 509 well-defined, multiple crista junctions.

510

511 Figure 3: Cryo-ET of liver mitochondria from young (left, 20 weeks old) and old (right, 80 512 weeks old) mice. Upper panel: slices through tomographic volumes (scale bars, 250 nm). 513 Lower panels: segmented 3D volumes. About 25% of the mitochondria from old animals have 514 large central voids (red). The voids were connected to the intermembrane space (IMS) by 515 openings of variable size.

516 Figure 4: Cryo-ET of kidney mitochondria from young (20 weeks old) and old (80 weeks 517 old) mice. (A) About 80% of the kidney mitochondria from young (left) or old (right) animals 518 had lamellar and locally parallel cristae, not unlike heart mitochondria, except that the cristae 519 were less tightly packed. (B) About 20% of the mitochondria from young and old kidney 520 showed an inner membrane morphology resembling that in apoptotic cells (36), without any 521 discernible pattern and wide or irregular junctions. Scale bars, 250 nm.

522

523 Figure 5: Activity and ultrastructure of mtDNA mutator mouse mitochondria. (A) Oxygen 524 consumption rate assessed in heart mitochondria from control (white, n = 8, 30 weeks old) 525 and mutator (black, n = 8, 30 weeks old) mice. Mitochondria were isolated and analysed as in 526 Figure 1. (B-C) Cryo-ET (B) Mutator mouse heart mitochondria had lamellar, parallel cristae 527 similar to wild-type (see Figure 2), but with occasional peripheral voids at the inner boundary 528 membrane. (C) About 40% of liver mitochondria from mutator mouse had low-density central 529 voids, as in old liver (see Figure 3).

530

531 Figure 6: Activity and ultrastructure of D. melanogaster mitochondria. (A) Survival rates (n 532 = 150; solid line) and climbing ability (n ≥ 25; dashed line) of wDah wild-type flies. Error bars 533 represent SEM and arrows indicate sampling points for young (15 days old) and old (70 days 534 old) flies. (B) Oxygen consumption rate assessed in mitochondria from young (white, n = 3) 535 and old (black, n = 3) flies. Mitochondria were isolated and analysed as in Figure 1. Succinate 536 and glycerol-3-phosphate (AS), and finally rotenone were added (GS) compared to complex I- 537 driven respiration (CPI). (C) Mitochondrial peroxide yield in mitochondria from young 538 (white, n = 3) and old (black, n = 3) flies. (D-E) Cryo-ET of typical mitochondria from young 539 (D) or old (E) flies. Mitochondria from young flies had lamellar, mostly parallel cristae, 540 similar to those from mouse heart. Mitochondria from old flies had highly variable shapes and 541 cristae organisation, with the following main types (clockwise from top left): elongated and 542 branched morphology; round, lacking cristae; small disconnected cristae; concentric narrow 543 cristae; irregular, wide cristae.

544 Figure 1 – supplement 1: Steady state level of different anti-oxidant enzymes in heart, liver 545 and kidney extracts, as assessed by quantitative densitometry of western blots in Figure 1D.

546 Figure 2 – supplement 1: (A-C): Young mouse heart mitochondrion. (A): Detailed view of 547 crista fenestration. (B): Detailed view of two neighbouring cristae with a narrow tubular 548 connection. (C): Cut-open view of a membrane swirl. (D): Two examples of old mouse heart 549 mitochondria with widened cristae (scale bar 250 nm).

550

551 Videos 1-6: Tomographic volume and 3D segmentation of a mitochondrion from young 552 mouse heart (1), young mouse liver (2), old mouse liver (3), young fly (4) and old fly (5-6).

553

554

555

556

557

558

559 Table 1

560 Table 1: Overview of organisms and tissues analyzed. #number of animals dissected (mouse) or number of mitochondrial 561 preparations (fly) vs number of individual mitochondria examined tomographically organism age tissue samples / mitochondria with abnormal morphology (%) mitochondria# mouse young heart 4 / 27 wide cristae: 4% cristae membrane swirls: 19% old heart 3 / 17 wide cristae: 24% cristae membrane swirls: 6% young liver 5 / 18 voids: 11% granules: 6% old liver 4 / 31 voids: 32% granules: 6% apoptotic: 3% young kidney 3 / 33 apoptotic: 12% granules: 3% old kidney 2 / 22 apoptotic: 18% granules: 18% mutator heart 2 / 10 membrane swirls: 30% membrane enclosures: 40% granules: 20% mutator liver 2 / 10 voids: 40% granules: 20% fly young whole 3 / 29 elongated (axial ratio > 3): 10% organism wide cristae: 10% old whole 3 / 39 elongated (axial ratio > 3): 18% organism wide cristae: 15% branched: 5% various other: 23% 562

563

564

565 References

566 1. Pfanner, N., van der Laan, M., Amati, P., Capaldi, R.A., Caudy, A.A., Chacinska, A., Darshi, M., 567 Deckers, M., Hoppins, S., Icho, T. et al. (2014) Uniform nomenclature for the mitochondrial 568 contact site and cristae organizing system. J Cell Biol, 204, 1083-1086. 569 2. Leslie, A.G., Abrahams, J.P., Braig, K., Lutter, R., Menz, R.I., Orriss, G.L., van Raaij, M.J. and 570 Walker, J.E. (1999) The structure of bovine mitochondrial F1-ATPase: an example of rotary 571 catalysis. Biochem Soc Trans, 27, 37-42. 572 3. Gilkerson, R.W., Selker, J.M. and Capaldi, R.A. (2003) The cristal membrane of mitochondria 573 is the principal site of oxidative phosphorylation. FEBS Lett, 546, 355-358. 574 4. Vogel, F., Bornhovd, C., Neupert, W. and Reichert, A.S. (2006) Dynamic 575 subcompartmentalization of the mitochondrial inner membrane. J Cell Biol, 175, 237-247. 576 5. Davies, K.M., Anselmi, C., Wittig, I., Faraldo-Gomez, J.D. and Kühlbrandt, W. (2012) Structure 577 of the yeast F1Fo-ATP synthase dimer and its role in shaping the mitochondrial cristae. Proc 578 Natl Acad Sci U S A, 109, 13602-13607. 579 6. Davies, K.M., Strauss, M., Daum, B., Kief, J.H., Osiewacz, H.D., Rycovska, A., Zickermann, V. 580 and Kühlbrandt, W. (2011) Macromolecular organization of ATP synthase and complex I in 581 whole mitochondria. Proc Natl Acad Sci U S A, 108, 14121-14126. 582 7. Paumard, P., Vaillier, J., Coulary, B., Schaeffer, J., Soubannier, V., Mueller, D.M., Brethes, D., 583 di Rago, J.P. and Velours, J. (2002) The ATP synthase is involved in generating mitochondrial 584 cristae morphology. EMBO J, 21, 221-230. 585 8. de Magalhaes, J.P. (2011) In Stuart-Hamilton, I. (ed.), An Introduction to Gerontology. 586 Cambridge University Press, Cambridge, pp. 21-47. 587 9. Harman, D. (1956) Aging: a theory based on free radical and radiation chemistry. Journal of 588 gerontology, 11, 298-300. 589 10. Dröse, S. and Brandt, U. (2012) Molecular mechanisms of superoxide production by the 590 mitochondrial respiratory chain. Advances in experimental medicine and biology, 748, 145- 591 169. 592 11. Harman, D. (1972) The biologic clock: the mitochondria? J Am Geriatr Soc, 20, 145-147. 593 12. Trifunovic, A., Wredenberg, A., Falkenberg, M., Spelbrink, J.N., Rovio, A.T., Bruder, C.E., 594 Bohlooly, Y.M., Gidlof, S., Oldfors, A., Wibom, R. et al. (2004) Premature ageing in mice 595 expressing defective mitochondrial DNA polymerase. Nature, 429, 417-423. 596 13. Trifunovic, A., Hansson, A., Wredenberg, A., Rovio, A.T., Dufour, E., Khvorostov, I., Spelbrink, 597 J.N., Wibom, R., Jacobs, H.T. and Larsson, N.G. (2005) Somatic mtDNA mutations cause aging 598 phenotypes without affecting reactive oxygen species production. Proc Natl Acad Sci U S A, 599 102, 17993-17998. 600 14. Stuart, J.A., Maddalena, L.A., Merilovich, M. and Robb, E.L. (2014) A midlife crisis for the 601 mitochondrial free radical theory of aging. Longev Healthspan, 3, 4. 602 15. Heidler, T., Hartwig, K., Daniel, H. and Wenzel, U. (2010) Caenorhabditis elegans lifespan 603 extension caused by treatment with an orally active ROS-generator is dependent on DAF-16 604 and SIR-2.1. Biogerontology, 11, 183-195. 605 16. Lee, S.J., Hwang, A.B. and Kenyon, C. (2010) Inhibition of respiration extends C. elegans life 606 span via reactive oxygen species that increase HIF-1 activity. Curr Biol, 20, 2131-2136. 607 17. Yang, W. and Hekimi, S. (2010) A mitochondrial superoxide signal triggers increased longevity 608 in Caenorhabditis elegans. PLoS Biol, 8, e1000556. 609 18. Doonan, R., McElwee, J.J., Matthijssens, F., Walker, G.A., Houthoofd, K., Back, P., Matscheski, 610 A., Vanfleteren, J.R. and Gems, D. (2008) Against the oxidative damage theory of aging: 611 superoxide dismutases protect against oxidative stress but have little or no effect on life span 612 in Caenorhabditis elegans. Genes Dev, 22, 3236-3241. 613 19. Gruber, J., Ng, L.F., Fong, S., Wong, Y.T., Koh, S.A., Chen, C.B., Shui, G., Cheong, W.F., 614 Schaffer, S., Wenk, M.R. et al. (2011) Mitochondrial changes in ageing Caenorhabditis 615 elegans--what do we learn from superoxide dismutase knockouts? PLoS One, 6, e19444.

616 20. Van Raamsdonk, J.M. and Hekimi, S. (2009) Deletion of the mitochondrial superoxide 617 dismutase sod-2 extends lifespan in Caenorhabditis elegans. PLoS Genet, 5, e1000361. 618 21. Itsara, L.S., Kennedy, S.R., Fox, E.J., Yu, S., Hewitt, J.J., Sanchez-Contreras, M., Cardozo- 619 Pelaez, F. and Pallanck, L.J. (2014) Oxidative stress is not a major contributor to somatic 620 mitochondrial DNA mutations. PLoS Genet, 10, e1003974. 621 22. Mourier, A., Ruzzenente, B., Brandt, T., Kuhlbrandt, W. and Larsson, N.G. (2014) Loss of 622 LRPPRC causes ATP synthase deficiency. Hum Mol Genet, 23, 2580-2592. 623 23. Brust, D., Daum, B., Breunig, C., Hamann, A., Kühlbrandt, W. and Osiewacz, H.D. (2010) 624 Cyclophilin D links programmed cell death and organismal aging in Podospora anserina. 625 Aging Cell, 9, 761-775. 626 24. Daum, B., Walter, A., Horst, A., Osiewacz, H.D. and Kühlbrandt, W. (2013) Age-dependent 627 dissociation of ATP synthase dimers and loss of inner-membrane cristae in mitochondria. 628 Proc Natl Acad Sci U S A, 110, 15301-15306. 629 25. Wilson, P.D. and Franks, L.M. (1975) The effect of age on mitochondrial ultrastructure. 630 Gerontologia, 21, 81-94. 631 26. Ludatscher, R., Silbermann, M., Gershon, D. and Reznick, A. (1983) The effects of enforced 632 running on the gastrocnemius muscle in aging mice: an ultrastructural study. Exp Gerontol, 633 18, 113-123. 634 27. Walker, D.W. and Benzer, S. (2004) Mitochondrial "swirls" induced by oxygen stress and in 635 the Drosophila mutant hyperswirl. Proc Natl Acad Sci U S A, 101, 10290-10295. 636 28. Sohal, R.S. (1970) Mitochondrial changes in the heart of Drosophila repleta, Wollaston with 637 age. Exp Gerontol, 5, 213-216. 638 29. Linford, N.J., Bilgir, C., Ro, J. and Pletcher, S.D. (2013) Measurement of lifespan in Drosophila 639 melanogaster. J Vis Exp. 640 30. Turturro, A., Duffy, P., Hass, B., Kodell, R. and Hart, R. (2002) Survival characteristics and age- 641 adjusted disease incidences in C57BL/6 mice fed a commonly used cereal-based diet 642 modulated by dietary restriction. J Gerontol A Biol Sci Med Sci, 57, B379-389. 643 31. Cho, J., Hur, J.H. and Walker, D.W. (2011) The role of mitochondria in Drosophila aging. Exp 644 Gerontol, 46, 331-334. 645 32. Mulligan, C.M., Le, C.H., Demooy, A.B., Nelson, C.B. and Chicco, A.J. (2014) Inhibition of 646 Delta-6 Desaturase Reverses Cardiolipin Remodeling and Prevents Contractile Dysfunction in 647 the Aged Mouse Heart Without Altering Mitochondrial Respiratory Function. J Gerontol A 648 Biol Sci Med Sci. 649 33. Weindruch, R.H., Cheung, M.K., Verity, M.A. and Walford, R.L. (1980) Modification of 650 mitochondrial respiration by aging and dietary restriction. Mech Ageing Dev, 12, 375-392. 651 34. Wilson, P.D., Hill, B.T. and Franks, L.M. (1975) The effect of age on mitochondrial enzymes 652 and respiration. Gerontologia, 21, 95-101. 653 35. Votyakova, T.V. and Reynolds, I.J. (2001) DeltaPsi(m)-Dependent and -independent 654 production of reactive oxygen species by rat brain mitochondria. J Neurochem, 79, 266-277. 655 36. Scorrano, L., Ashiya, M., Buttle, K., Weiler, S., Oakes, S.A., Mannella, C.A. and Korsmeyer, S.J. 656 (2002) A distinct pathway remodels mitochondrial cristae and mobilizes cytochrome c during 657 apoptosis. Dev Cell, 2, 55-67. 658 37. Sun, M.G., Williams, J., Munoz-Pinedo, C., Perkins, G.A., Brown, J.M., Ellisman, M.H., Green, 659 D.R. and Frey, T.G. (2007) Correlated three-dimensional light and electron microscopy 660 reveals transformation of mitochondria during apoptosis. Nat Cell Biol, 9, 1057-U1021. 661 38. Merkwirth, C., Dargazanli, S., Tatsuta, T., Geimer, S., Lower, B., Wunderlich, F.T., von Kleist- 662 Retzow, J.C., Waisman, A., Westermann, B. and Langer, T. (2008) Prohibitins control cell 663 proliferation and apoptosis by regulating OPA1-dependent cristae morphogenesis in 664 mitochondria. Genes Dev, 22, 476-488. 665 39. Wrogemann, K., Nylen, E.G., Adamson, I. and Pande, S.V. (1985) Functional studies on in situ- 666 like mitochondria isolated in the presence of polyvinyl pyrrolidone. Biochim Biophys Acta, 667 806, 1-8.

668 40. Rhodenizer, D., Martin, I., Bhandari, P., Pletcher, S.D. and Grotewiel, M. (2008) Genetic and 669 environmental factors impact age-related impairment of negative geotaxis in Drosophila by 670 altering age-dependent climbing speed. Exp Gerontol, 43, 739-748. 671 41. Ferguson, M., Mockett, R.J., Shen, Y., Orr, W.C. and Sohal, R.S. (2005) Age-associated decline 672 in mitochondrial respiration and electron transport in Drosophila melanogaster. Biochem J, 673 390, 501-511. 674 42. Strauss, M., Hofhaus, G., Schroder, R.R. and Kühlbrandt, W. (2008) Dimer ribbons of ATP 675 synthase shape the inner mitochondrial membrane. EMBO J, 27, 1154-1160. 676 43. Slautterback, D.B. (1965) Mitochondria in Cardiac Muscle Cells of the Canary and Some Other 677 Birds. J Cell Biol, 24, 1-21. 678 44. Smith, D.S. (1963) The Structure of Flight Muscle Sarcosomes in the Blowfly Calliphora 679 Erythrocephala (Diptera). J Cell Biol, 19, 115-138. 680 45. Rinkevich, Y., Montoro, D.T., Contreras-Trujillo, H., Harari-Steinberg, O., Newman, A.M., Tsai, 681 J.M., Lim, X., Van-Amerongen, R., Bowman, A., Januszyk, M. et al. (2014) In vivo clonal 682 analysis reveals lineage-restricted progenitor characteristics in mammalian kidney 683 development, maintenance, and regeneration. Cell Rep, 7, 1270-1283. 684 46. Macdonald, R.A. (1961) "Lifespan" of liver cells. Autoradio-graphic study using tritiated 685 thymidine in normal, cirrhotic, and partially hepatectomized rats. Arch Intern Med, 107, 335- 686 343. 687 47. Malliaras, K., Zhang, Y., Seinfeld, J., Galang, G., Tseliou, E., Cheng, K., Sun, B., Aminzadeh, M. 688 and Marban, E. (2013) Cardiomyocyte proliferation and progenitor cell recruitment underlie 689 therapeutic regeneration after myocardial infarction in the adult mouse heart. EMBO Mol 690 Med, 5, 191-209. 691 48. Ghadially, F.N. (1988), Ultrastructural Pathology of the Cell and Matrix. Butterworths, 692 London, Vol. 1, pp. 191-329. 693 49. David, H. and Kettler, L.H. (1961) [Degeneration of liver mitochondria after ammonia 694 poisoning]. Z Zellforsch Mikrosk Anat, 53, 857-866. 695 50. Stephens, R.J. and Bils, R.F. (1965) An Atypical Mitochondrial Form in Normal Rat Liver. J Cell 696 Biol, 24, 500-504. 697 51. Ding, W.X., Li, M., Biazik, J.M., Morgan, D.G., Guo, F., Ni, H.M., Goheen, M., Eskelinen, E.L. 698 and Yin, X.M. (2012) Electron microscopic analysis of a spherical mitochondrial structure. J 699 Biol Chem, 287, 42373-42378. 700 52. Barazzoni, R., Short, K.R. and Nair, K.S. (2000) Effects of aging on mitochondrial DNA copy 701 number and cytochrome c oxidase gene expression in rat skeletal muscle, liver, and heart. J 702 Biol Chem, 275, 3343-3347. 703 53. Yen, T.C., Su, J.H., King, K.L. and Wei, Y.H. (1991) Ageing-associated 5 kb deletion in human 704 liver mitochondrial DNA. Biochem Biophys Res Commun, 178, 124-131. 705 54. Hertsens, R.C., Bernaert, I., Joniau, M. and Jacob, W.A. (1986) Immunocytochemical 706 investigation of native matrix granules of the rat heart mitochondrion. J Ultrastruct Mol 707 Struct Res, 94, 1-15. 708 55. Ghadially, F.N. (2001) As you like it, Part 3: A critique and historical review of calcification as 709 seen with the electron microscope. Ultrastruct Pathol, 25, 243-267. 710 56. Jacob, W.A., Bakker, A., Hertsens, R.C. and Biermans, W. (1994) Mitochondrial matrix 711 granules: their behavior during changing metabolic situations and their relationship to 712 contact sites between inner and outer mitochondrial membranes. Microsc Res Tech, 27, 307- 713 318. 714 57. Horvath, S. (2013) DNA methylation age of human tissues and cell types. Genome Biol, 14, 715 R115. 716 58. Dai, D.F., Santana, L.F., Vermulst, M., Tomazela, D.M., Emond, M.J., MacCoss, M.J., Gollahon, 717 K., Martin, G.M., Loeb, L.A., Ladiges, W.C. et al. (2009) Overexpression of catalase targeted to 718 mitochondria attenuates murine cardiac aging. Circulation, 119, 2789-2797. 719 59. Lenaz, G. (2012) Mitochondria and reactive oxygen species. Which role in physiology and 720 pathology? Adv Exp Med Biol, 942, 93-136.

721 60. Cocheme, H.M., Quin, C., McQuaker, S.J., Cabreiro, F., Logan, A., Prime, T.A., Abakumova, I., 722 Patel, J.V., Fearnley, I.M., James, A.M. et al. (2011) Measurement of H2O2 within living 723 Drosophila during aging using a ratiometric mass spectrometry probe targeted to the 724 mitochondrial matrix. Cell Metab, 13, 340-350. 725 61. Scheckhuber, C.Q. and Osiewacz, H.D. (2008) Podospora anserina: a model organism to study 726 mechanisms of healthy ageing. Molecular genetics and genomics : MGG, 280, 365-374. 727 62. Ross, J.M., Stewart, J.B., Hagstrom, E., Brene, S., Mourier, A., Coppotelli, G., Freyer, C., 728 Lagouge, M., Hoffer, B.J., Olson, L. et al. (2013) Germline mitochondrial DNA mutations 729 aggravate ageing and can impair brain development. Nature, 501, 412-415. 730 63. Greene, J.C., Whitworth, A.J., Kuo, I., Andrews, L.A., Feany, M.B. and Pallanck, L.J. (2003) 731 Mitochondrial pathology and apoptotic muscle degeneration in Drosophila parkin mutants. 732 Proc Natl Acad Sci U S A, 100, 4078-4083. 733 64. Freyer, C., Cree, L.M., Mourier, A., Stewart, J.B., Koolmeister, C., Milenkovic, D., Wai, T., 734 Floros, V.I., Hagstrom, E., Chatzidaki, E.E. et al. (2012) Variation in germline mtDNA 735 heteroplasmy is determined prenatally but modified during subsequent transmission. Nat 736 Genet, 44, 1282-1285. 737 65. Brand, M.D. and Nicholls, D.G. (2011) Assessing mitochondrial dysfunction in cells. Biochem J, 738 435, 297-312. 739 66. Mourier, A., Devin, A. and Rigoulet, M. (2010) Active proton leak in mitochondria: a new way 740 to regulate substrate oxidation. Biochim Biophys Acta, 1797, 255-261. 741 67. Kremer, J.R., Mastronarde, D.N. and McIntosh, J.R. (1996) Computer visualization of three- 742 dimensional image data using IMOD. J Struct Biol, 116, 71-76. 743 68. Frangakis, A.S. and Hegerl, R. (2001) Noise reduction in electron tomographic reconstructions 744 using nonlinear anisotropic diffusion. J Struct Biol, 135, 239-250.

745