RCF1-dependent respiratory supercomplexes are integral for lifespan-maintenance in a fungal ageing model

Mitochondrial respiratory supercomplexes (mtRSCs) are stoichiometric assemblies of electron transport chain (ETC) complexes in the inner mitochondrial membrane. They are hypothesized to regulate electron flow, the generation of reactive oxygen species (ROS) and to stabilize ETC complexes. Using the fungal ageing model Podospora anserina, we investigated the impact of homologues of the Saccharomyces cerevisiae respiratory supercomplex factors 1 and 2 (termed PaRCF1 and PaRCF2) on mtRSC formation, fitness and lifespan. Whereas PaRCF2’s role seems negligible, ablation of PaRCF1 alters size of monomeric complex IV, reduces the abundance of complex IV-containing supercomplexes, negatively affects vital functions and shortens lifespan. PaRcf1 overexpression slightly prolongs lifespan, though without appreciably influencing ETC organization. Overall, our results identify PaRCF1 as necessary yet not sufficient for mtRSC formation and demonstrate that PaRCF1-dependent stability of complex IV and associated supercomplexes is highly relevant for maintenance of the healthy lifespan in a eukaryotic model organism.

Convincing evidence exists that mitochondrial dysfunction plays a key role in biological ageing and various age-related human pathologies [1][2][3][4][5] . It has long been suggested that the root cause for the progressive decline of mitochondrial function is the age-dependent accumulation of reactive oxygen species, which inevitably arise during oxidative phosphorylation 6 . The primary ROS superoxide anion (O 2 .− ) is produced at complexes I and III of the ETC and can give rise to the secondary ROS hydrogen peroxide which in turn can lead to the formation of the highly reactive hydroxyl radical through the Fenton reaction. Both O 2 .− and the hydroxyl radical are able to damage proteins, lipids and DNA. Consequently, all ROS are harmful to mitochondria when present in excess 7 . Recently, this narrow view of ROS solely as damaging agents has been challenged by counter-intuitive and contradictory experimental observations 8,9 . These may, at least in part, result from the fact that low levels of ROS are essential for cellular signalling and to control developmental processes 10,11 . A balanced generation and degradation of ROS is therefore highly important to keep biological systems functional over time. In this context, maintaining integrity of the ETC is crucial 5 .
In their seminal paper published in 2000, Schägger and Pfeiffer convincingly demonstrated the existence of stoichiometric assemblies of individual ETC complexes, so called mitochondrial respiratory supercomplexes or respirasomes, in yeast and mammalian mitochondria 12 . The 'plasticity model' of ETC organization is based on additional observations of considerable variations in mtRSC species and hypothesizes, that individual ETC complexes and supercomplexes can exist side by side in the inner mitochondrial membrane 13,14 . This view is supported by a growing number of studies demonstrating mitochondrial supercomplexes to be functional units of respiration assumed to be important for facilitating and directing electron flow by substrate channeling and consequently, to regulate the production of ETC-derived ROS. Additionally, there is evidence for a reciprocal dependence of supercomplex formation and the assembly and stabilization of individual ETC complexes, most importantly complex I [14][15][16][17] .
The proteins RCF1 and RCF2, members of the 'hypoxia-inducible gene 1' (HIG1) protein family, were recently identified in S. cerevisiae as being important constituents of mtRSCs [18][19][20] . ScRCF1 in particular was revealed as a crucial component for stabilization of the III 2 IV 2 supercomplex [18][19][20][21] , while ScRCF2 appears to have a less prominent function 19,20 . The Rutter group also demonstrated a loss of complex IV-containing supercomplexes in mitochondria extracted from C2C12 mouse myoblast cells after knockdown of the mammalian RCF1 homologue HIG2A 18 .
Despite these notable insights, the in vivo relevance of mitochondrial supercomplexes remains to be clarified in detail. A destabilization of mtRSCs has already been linked to the development of at least one complex human disorder known as Barth syndrome 22,23 . It seems likely, owing to the central role of mitochondria in health and disease 5 , that similar connections will be unravelled in the near future 16 . To this end, it is an important task to determine the relationship between altered supercomplexes, mitochondrial function and impact on the organism as a whole.
The filamentous ascomycete P. anserina is characterized by a limited lifespan, a clear mitochondrial aetiology of ageing and has been extensively analysed as a simple model for organismal ageing 24 . Its ETC features, in contrast to that of S. cerevisiae, both complex I as well as an alternative terminal oxidase (AOX) that can be activated to bypass complexes III and IV by directly transferring electrons from ubiquinol to oxygen 25 . In the wild type, three major supercomplexes (I 1 III 2 IV 0-2 ) can be resolved by blue native polyacrylamide gel electrophoresis (BN-PAGE) 26 . Genetic disruptions of complexes III and/or IV in P. anserina have been shown to result in absence of the corresponding supercomplexes 26,27 and enhanced AOX-dependent respiration, with mutant strains consistently displaying increased lifespans 25,[27][28][29] .
Here we describe the role of RCF1 and RCF2 in P. anserina. Deletion of PaRcf2, encoding the RCF2 homologue, has no pronounced effect on the phenotype, while ablation of PaRCF1 shifts the complex IV monomer almost exclusively to a form that migrates faster during BN-PAGE than that of the wild type. In addition, the abundance of I 1 III 2 IV 1-2 supercomplexes is strongly reduced. Concomitant with these changes, mitochondrial integrity as well as vital characteristics of the PaRcf1 deletion strain, such as growth rate and fertility, are impaired and lifespan, despite activation of the alternative respiratory pathway, is shortened. Overall, our results identify RCF1-stabilized complex IV and complex IV-containing supercomplexes as crucial ETC components in P. anserina, establishing a novel connection between their destabilization, impaired mitochondrial function and a negative impact on health and lifespan.
PaRcf1 and PaRcf2 deletion strains (Δ PaRcf1 and Δ PaRcf2) were generated according to the method described by El-Khoury and colleagues, with the respective wild-type gene's ORF replaced by a hygromycin B resistance gene 30 . The deletion strains were verified by Southern blot analysis (Fig. 1b,c and Supplementary Fig. S2).
To analyse the importance of PaRCF1 and PaRCF2 for maintaining integrity of the ETC, mitochondrial protein extracts from wild type (n = 4), Δ PaRcf1 (n = 3) and Δ PaRcf2 (n = 4) were compared using BN-PAGE (Fig. 1d, Table 1 and Supplementary Fig. S3). Deletion of PaRcf2 had no discernible effect on overall ETC composition. PaRcf1 deletion, however, led to marked changes. Specifically, relative abundance of the I 1 III 2 IV 2 (S 2 ) supercomplex (0.07 ± 0.01 AU; P = 0.02 by two-tailed Student's t-test) and of the I 1 III 2 IV 1 (S 1 ) supercomplex (0.03 ± 0.02 AU; P = 6.8E-04 by two-tailed Student's t-test) in Δ PaRcf1 was significantly reduced, while relative abundance of the I 1 III 2 IV 0 (S 0 ) supercomplex (15.30 ± 2.58 AU; P = 0.03 by two-tailed Student's t-test) was significantly increased. Free complex IV monomer was undetectable in mitochondrial protein extracts from Δ PaRcf1, though it should be noted that even in wild type and Δ PaRcf2 monomeric complex IV is detectable only by a diffuse and rather faint band after BN-PAGE and Coomassie staining. A complex IV 'in-gel' activity assay (Fig. 1e, Table 2 and Supplementary Fig. S4) was used to measure relative distribution of complex IV activity within each individual strain and confirmed the near total absence of the S 2 and S 1 supercomplexes in Δ PaRcf1. It The I 1 III 2 IV 0-2 (S 0-2 ) supercomplexes, dimeric complexes III and V (III 2 and V 2 ) as well as monomeric complexes I, IV and V were visualized by Coomassie staining. (e) Representative complex IV 'in-gel' activity assay with mitochondrial protein extracts from the indicated strains. (f) Representative western blot analysis of mitochondrial protein extracts from wild type, Δ PaRcf1 and Δ PaRcf2. A ScCOX2-specific antibody was used to detect the ~29 kDa PaCOX2 subunit of complex IV. PaPORIN (PaPOR) was detected as a loading control. (g) Quantitative western blot analysis of mitochondrial protein extracts from wild type (n = 4), Δ PaRcf1 (n = 4) and Δ PaRcf2 (n = 4). The PaCOX2 abundance was normalized to that of PaPOR and the mean wild-type abundance was defined as 1. Data given in parentheses are mean PaCOX2 abundance ± s.e.m. in arbitrary units (AU).

(AU)
Wild type ΔPaRcf1 ΔPaRcf2 also revealed that monomeric complex IV in Δ PaRcf1 is almost exclusively present in a faster migrating form (IV B ) than in the wild type or in Δ PaRcf2. Relative abundance of PaCOX2, an essential complex IV subunit encoded by the mitochondrial genome, was found to be slightly reduced in mitochondrial protein extracts from Δ PaRcf1 (0.84 ± 0.07 AU; P = 0.45 by two-tailed Student's t-test), while in contrast being increased in those from Δ PaRcf2 (1.55 ± 0.11 AU; P = 4.7E-02 by two-tailed Student's t-test) compared to the wild type (1.00 ± 0.18 AU; Fig. 1f,g and Supplementary Fig. S5).

ETC alterations in ΔPaRcf1 impair mitochondrial integrity.
To assess the functional consequences of PaRcf1 and PaRcf2 deletion for mitochondrial respiration, we measured oxygen consumption of wild type (n = 3), Δ PaRcf1 (n = 3) and Δ PaRcf2 (n = 3) mycelium with and without the addition of the specific respiratory inhibitors KCN, inhibiting complex IV, and salicylhydroxamic acid (SHAM), inhibiting AOX ( Fig. 2a and Supplementary Fig. S6). Notably, total absolute oxygen consumption of Δ PaRcf1 was tendentially increased compared to the wild type ( Supplementary Fig. S6). Similar effects have previously been observed in other P. anserina ETC mutants, likely reflecting inefficient oxygen utilization for ATP generation 31 . Sequential addition of both inhibitors in either order (i.e. first KCN and then SHAM or first SHAM and then KCN) almost completely inhibited oxygen consumption in all strains. Therefore, the majority of oxygen in P. anserina is consumed during mitochondrial respiration ( Supplementary Fig. S6). Relative oxygen consumption after inhibition of complex IV with KCN was significantly higher in Δ PaRcf1 (0.83 ± 0.07 AU; P = 5.2E-03 by two-tailed Student's t-test) than in wild type (0.25 ± 0.08 AU), i.e. Δ PaRcf1 exhibits, in accord with the results obtained by BN-PAGE (Fig. 1d,e and Tables 1 and 2), a decrease in complex IV-dependent respiration. In contrast, inhibition of AOX with SHAM resulted in significantly lower relative oxygen consumption in Δ PaRcf1 (0.43 ± 0.03 AU; P = 7.0E-03 by two-tailed Student's t-test) than in wild type (0.76 ± 0.05 AU). Interestingly, the effect of complex IV inhibition by KCN in Δ PaRcf1 was more pronounced if AOX was first inhibited by SHAM and vice versa ( Supplementary Fig. S6). This probably reflects an adaptive upregulation of either respiratory pathway after inhibition of the other pathway during oxygen consumption measurements.
In conclusion, these observations indicate an upregulation of the alternative respiratory pathway in Δ PaRcf1. Similar yet less pronounced changes in absolute and relative oxygen consumption after addition of KCN or SHAM were also observed in Δ PaRcf2 ( Fig. 2a and Supplementary Fig. S6), despite the lack of visible alterations in ETC composition (Fig. 1d,e and Tables 1 and 2).
Comparison of PaAOX abundance in mitochondrial protein extracts from wild type (n = 4), Δ PaRcf1 (n = 4) and Δ PaRcf2 (n = 4) by western blot analysis ( Fig. 2b-d and Supplementary Fig. S7) revealed significant changes only in Δ PaRcf1, where relative PaAOX abundance was considerably increased (4.89 ± 0.28 AU; P = 5.5E-03 by two-tailed Student's t-test). As no obvious deviations in Δ PaRcf2's phenotype from that of the wild type, apart from the changes in oxygen consumption (Fig. 2a), were identified (see also below), we subsequently focused on Δ PaRcf1.
Different P. anserina ETC mutants display decreased mitochondrial ROS and an increased lifespan concomitant with enhanced activation of the alternative respiratory pathway 25,28 . Consequently, the observation that respiration in Δ PaRcf1 is primarily AOX-dependent suggested that superoxide mediated damage to proteins might be reduced in this strain. Contrary to this assumption we found that Δ PaRcf1 had a slight but nonetheless significant increase in mitochondrial protein carbonylation (1.33 ± 0.07 AU; n = 6; P = 4.6E-03 by two-tailed Student's t-test), possibly reflecting impairments in ROS scavenging mechanisms and/or in the clearance of damaged mitochondrial proteins (Fig. 2e,f).
Quantitative complex IV 'in-gel' activity assay with mitochondrial protein extracts from wild type (n = 4), Δ PaRcf1 (n = 4) and Δ PaRcf2 (n = 4). As a measure for the relative distribution of activities of monomeric complex IV and complex IV-containing supercomplexes within each individual strain, optical densities of bands representing complex IV activity were quantified and normalized to total complex IV activity in the corresponding lane. Data are mean complex IV activity ± s.e.m. in percentage.
dismutase PaSOD3 (0.24 ± 0.03 AU; P = 1.2E-04 by two-tailed Student's t-test) and peroxiredoxin PaPRX (0.40 ± 0.08 AU; P = 2.3E-03 by two-tailed Student's t-test) in Δ PaRcf1, while level of the chaperone PaHSP60 (1.16 ± 0.22 AU) remained unchanged (Fig. 2g,h and Supplementary Fig. S8). In addition, relative abundance of mitochondrial aconitase PaACO2 was also found to be strongly and significantly reduced (0.05 ± 0.02 AU; P = 9.0E-03 by two-tailed Student's t-test) in the PaRcf1 deletion strain (Fig. 2g,h and Supplementary Fig. S8). Surprisingly, despite the strong reduction of PaSOD3 protein abundance in Δ PaRcf1 by almost 80% (Fig. 2g,h), visualization of PaSOD3-activity in mitochondrial protein extracts from P. anserina wild type and Δ PaRcf1 with an SOD 'in-gel' activity assay revealed no observable differences between the two strains ( Supplementary Fig. S9). This result suggests that the remaining PaSOD3 in Δ PaRcf1 is highly active to counteract oxidative stress. oxidized proteins derivatized to 2,4-dinitrophenylhydrazone (DNP-hydrazone) were detected with a DNPspecific antibody. The Coomassie stained gel after blotting served as a loading control. (f) Quantitative western blot analysis of mitochondrial protein extracts from wild type (n = 6) and Δ PaRcf1 (n = 6). The mean wild-type protein carbonylation was defined as 1. Data are mean protein carbonylation ± s.e.m. in AU. (g) Representative western blot analyses of mitochondrial protein extracts from wild type and Δ PaRcf1 using the indicated antibodies. PaACO2 and PaHSP60 were detected with antibodies directed against the human homologues (Anti-HsACO2 or Anti-HsHSP60). PaSOD3 was detected with a rat SOD2 antibody (Anti-RnSOD2). PaPOR was detected as a loading control. (h) Quantitative western blot analyses of mitochondrial protein extracts from wild type (n = 4) and Δ PaRcf1 (n = 4). Protein abundances were normalized to that of PaPOR and the mean wild-type abundances were defined as 1. Data are mean protein abundance ± s.e.m. in AU.
Scientific RepoRts | 5:12697 | DOi: 10.1038/srep12697 ΔPaRcf1 is unable to maintain a healthy lifespan. To date, consequences for organismal health following a targeted genetic disruption of mitochondrial respiratory supercomplexes are scarcely studied. P. anserina RCF1, as demonstrated in this study, is critically involved in stabilizing complex IV. Concomitant with a destabilization of complex IV in Δ PaRcf1, abundance of associated supercomplexes is also strongly reduced (Fig. 1d,e and Tables 1 and 2). Presumably as a consequence thereof, mitochondrial integrity in the PaRcf1 deletion strain is negatively affected (Fig. 2a-h). To better understand the resulting impact on the organism as a whole, several vital characteristics of Δ PaRcf1 were assessed.
It is reasonable to assume that Δ PaRcf1, showing increased oxidative protein damage (Fig. 2e,f) and a reduction of key enzymes involved in ROS scavenging (Fig. 2g,h), should be more susceptible to oxidative stressors. This was indeed true for paraquat, known to generate O 2 .− at the ETC 32 , which almost completely inhibited growth of the PaRcf1 deletion strain already at a concentration of 80 μ M. At this concentration, growth rate of wild type and Δ PaRcf2 remained nearly unaffected (Fig. 3a). Other stressors, namely H 2 O 2 and CuSO 4 , had no intensified effect on Δ PaRcf1 (Supplementary Fig. S10). Next, influence of 20 μ M paraquat on survival of the PaRcf1 and PaRcf2 deletion strains was investigated. Previous work of our laboratory showed that, similar to what has been observed in Caenorhabditis elegans 33,34 , low doses of paraquat can considerably prolong lifespan of the P. anserina wild-type strain 35 . Interestingly, this effect was completely abrogated in Δ PaRcf1, while still being preserved in Δ PaRcf2 (Supplementary Fig. S11).
As the mammalian RCF1 homologue HIG1A is known to be transcriptionally regulated 36 , we speculated that PaRcf1 expression might be induced by low doses of paraquat and that PaRCF1 is possibly involved in mediating paraquat-triggered lifespan extension. Relative PaRcf1 expression in the wild type was, however, not elevated after cultivation with 20 μ M paraquat (Supplementary Fig. S11). The PaRcf1 deletion strain was further characterized by female infertility (Fig. 3b) and a significant reduction of its growth rate on standard medium by 58% (0.25 ± 0.01 cm d −1 ; n = 69; P = 2.2E-27 by two-tailed Wilcoxon rank-sum test; Fig. 3c and Table 3). Perhaps the most remarkable characteristic of Δ PaRcf1's phenotype was a significant 40% shortening of its mean lifespan (14.5 ± 0.7 d; n = 71; P = 3.4E-19 by two-tailed Wilcoxon rank-sum test; Fig. 3d and Table 3). Deletion of PaRcf2, aside from a slightly elevated growth rate, again led to no distinct deviations from the wild type under any of the conditions tested (Fig. 3a-d and Table 3).
To determine the specificity of the observed effects, we complemented Δ PaRcf1 by introducing a C-terminally 6xHis-tagged variant of PaRcf1 under control of the native promotor and terminator. Presence of the recombinant gene in the complemented strain (Δ PaRcf1/PaRcf1-6xHis) was verified by Southern blot analysis ( Supplementary Fig. S12). Relative expression of PaRcf1-6xHis was similar to that of PaRcf1 in the wild type ( Supplementary Fig. S12) and the recombinant protein could be detected in mitochondrial protein extracts from Δ PaRcf1/PaRcf1-6xHis ( Supplementary Fig. S12). The complemented strain was again fertile and displayed a wild-type like growth rate ( Fig. 3c and Table 3) and lifespan ( Fig. 3d and Table 3).
Finally, since absence of PaRCF1 had such a dramatic negative effect on health and lifespan, we investigated whether overexpression of PaRcf1 might be beneficial for P. anserina. A strain expressing PaRcf1-6xHis under control of a constitutive promotor in the wild-type background (PaRcf1-6xHis_ OEx) was generated and verified by Southern and western blot analysis (Supplementary Fig. S13). Relative PaRcf1 expression (20-fold; P = 3.6E-02 by two-tailed Student's t-test) and relative PaRCF1-6xHis protein abundance (17-fold; P = 6.5E-17 by two-tailed Student's t-test) were significantly elevated compared with the wild type or Δ PaRcf1/PaRcf1-6xHis, confirming stable overexpression of the construct ( Supplementary Fig. S13). While paraquat resistance of PaRcf1-6xHis_OEx remained essentially unchanged ( Supplementary Fig. S13), both its growth rate (0.65 ± 0.01 cm d −1 ; n = 26; P = 8.3E-06 by two-tailed Wilcoxon rank-sum test) and lifespan (27.6 ± 0.3 d; n = 26; P = 1.1E-02 by two-tailed Wilcoxon rank-sum test) were slightly but significantly increased by + 10% and + 13%, respectively (Fig. 3c,d and Table 3). These changes were, however, not correlated with a discernible increase in supercomplexes as assayed by BN-PAGE, thereby suggesting that PaRCF1 alone is not sufficient to markedly elevate formation of mtRSCs (Supplementary Fig. S13 and Supplementary Table S1).

Discussion
The mitochondrial protein RCF1 has recently been identified in three independent studies as a novel stabilizing component of the III 2 IV 2 supercomplex in S. cerevisiae. It was further demonstrated to preferentially interact with cytochrome c oxidase and to be important for maintaining activity of this complex. In addition, ScRCF1 is also able to independently associate with complex III [18][19][20][21] . Despite these insights, the detailed nature and function of ScRCF1 are not entirely clear and the authors' interpretations are somewhat conflicting. Based on the observation that ScRCF1 maintains association with complex III subunits even in absence of assembled complex IV, one study concluded that ScRCF1 is a true supercomplex assembly factor and not a subunit of complex IV 18 . In contrast, Vukotic and colleagues described ScRCF1 as more likely being a new subunit of at least one particular complex IV isoform in which it mediates contact with complex III 20 . In yet another study, ScRCF1 was recognized as a possible cytochrome c oxidase assembly and regulatory factor but it was also noted that the ability of ScRCF1 to reliably interact with complex III is unique for a potential complex IV component 19 .
In our study, we show that RCF1 in P. anserina is absolutely crucial for the stability of complex IV and that PaRCF1 ablation leads to a strong reduction of supercomplex abundance. While it is not yet clear whether PaRCF1 directly interacts with supercomplexes to induce their formation, its impact on overall ETC organization is even more pronounced than what has been observed for RCF1 in S. cerevisiae [18][19][20] or the mammalian RCF1 homolog HIG2A in mice 18 . In contrast, deletion of the gene coding for the P. anserina RCF2 homologue led to no pronounced phenotype and did not measurably affect complex IV integrity or ETC organization. While intermediate activation of the alternative respiratory pathway in Δ PaRcf2 clearly argues for a role of PaRCF2 in maintaining standard respiration, it appears questionable whether PaRCF2 is indeed a functional homologue of ScRCF2, which was demonstrated to have partially   37,38 . In mammals, which do not possess a RCF2 homologue, complex I has been described as being stabilized by supercomplex formation with complexes III and IV and at the same time to provide a scaffold for efficient respirasome assembly 39 .
There is convincing evidence that in P. anserina the stability of complex I is not dependent on complexes III and/or IV, possibly due to such specific features as presence of the AOX and complex I dimerization 27 . It can be speculated that complex III and IV interaction in the context of supercomplex formation in turn is less reliant on complex I and mainly dependent on the presence of other factors such as PaRCF1.
Probably the most striking change in the ETC of Δ PaRcf1 is the appearance of a faster migrating complex IV variant (IV B ) which is found almost exclusively instead of fully assembled complex IV. While a comparable phenomenon has so far not been observed in S. cerevisiae, Chen and colleagues report a similar yet less pronounced increase in incomplete complex IV after knockdown of mammalian HIG2A in C2C12 mouse myoblast cells 18 . The appearance of the IV B variant seems to be highly specific for absence of PaRCF1, as it has not been detected nearly as prominently or at all in other P. anserina ETC mutants impaired in supercomplex formation 26,27 .
In light of these insights, it is logical to conclude that PaRCF1 is a subunit or a specific regulatory factor of a particular complex IV variant, whose assembly likely precedes and in fact appears to be a prerequisite for formation of the I 1 III 2 IV 1-2 supercomplexes. Thus, decreased formation of supercomplexes in Δ PaRcf1 could well be a direct consequence of incorrect complex IV assembly. To further the understanding of supercomplex assembly and regulation, it arises as an important future task to address the specialized role and heterogeneous nature of distinct ETC complex variants in different organisms, especially regarding their importance for higher order organization of the respiratory chain.
The majority of studies concerning mitochondrial respiratory supercomplexes have focused on their biochemical, structural and kinetic properties. Consequently, the contribution of mtRSCs to mitochondrial function and organismal integrity is as yet not well understood. One of the first hints at the relevance of mtRSCs in vivo has come from the insight that human Barth syndrome, a hereditary cardiomyopathy occurring exclusively in males, is linked to a destabilization of supercomplexes caused by abnormal mitochondrial cardiolipin 22,23 . In addition, a recent study in mice demonstrated the necessity of a supercomplex assembly factor for the regulation of energy metabolism in muscle 40 . To a lesser extent mtRSCs have also been implicated in more complex phenomena such as cancer progression, neurodegeneration and ageing, though causative evidence and understanding of the underlying molecular pathways is still largely missing 41 .
Our results clearly demonstrate that RCF1-dependent stability of complex IV and presence of associated supercomplexes in P. anserina is important for mitochondrial function and organismal integrity. On a physiological level, in good agreement with our biochemical observations, the PaRcf1 deletion strain displays reduced complex IV-dependent respiration and activation of the alternative respiratory pathway. Other P. anserina mutants lacking I 1 III 2 IV 1-2 supercomplexes and respiring primarily via the AOX, owing to defects in complex III and/or complex IV, very consistently display prolonged lifespans 25,27-29 . In striking contrast to these previous observations, not only are vital functions and superoxide resistance of the PaRcf1 deletion strain negatively affected but its lifespan is also markedly reduced by nearly 50%. On a molecular level, this adverse overall impact is correlated with increased oxidative damage of mitochondrial proteins. This is a further distinguishing feature of Δ PaRcf1, as a switch to AOX-dependent respiration in P. anserina mutants was generally found to result in a decreased ROS burden 25,28 . A likely explanation for these findings, aside from an increase in ROS production itself, is the bold reduction of several ROS scavenging and protein quality control components in mitochondria of Δ PaRcf1. The marked reduction of PaSOD3 protein abundance in mitochondria of the PaRcf1 deletion strain seems to be of particular significance because, quite unexpectedly, PaSOD3 activity in Δ PaRcf1 still appears almost identical to that in the wild type. Together with the observation that Δ PaRcf1 is barely able to survive any additional paraquat-induced oxidative stress, this strongly suggests that PaSOD3 in Δ PaRcf1 is working at the limits of its capacity. Interestingly, deletion of S. cerevisiae Rcf1 increased mitochondrial SOD abundance 18 . The inability of Δ PaRcf1 to likewise upregulate its mitochondrial SOD in response to elevated endogenous oxidative stress underscores the dramatic effect on respiratory chain integrity and activity following ablation of PaRCF1 and likely reflects impairment of mitochondrial protein import and/or cytoplasmic protein synthesis due to reduced availability of ATP.
Though additional studies are necessary to address these phenomena in detail, it can already be inferred that they must be linked to conditions or properties only present in Δ PaRcf1 and not in other P. anserina ETC mutants. The aforementioned unprecedented and near exclusive emergence of a faster migrating complex IV variant in absence of PaRCF1 meets this criterion. Of note, Vukotic and colleagues proposed that a specific complex IV variant, absent in mitochondria of the S. cerevisiae Rcf1 deletion strain, serves to protect the ETC from excess ROS generation and that lack of ScRCF1 might thus lead to malfunction and ROS production in a catalytic manner 20 . Despite several crucial differences in the ETC of S. cerevisiae and P. anserina, our results are readily compatible with this model. Beyond that, we identified a novel connection between destabilization of supercomplexes, impaired mitochondrial function and adverse effects on health and lifespan in a simple eukaryotic model organism. As the overall Scientific RepoRts | 5:12697 | DOi: 10.1038/srep12697 respiratory chain composition of P. anserina is comparable to that of mammals, it will be of great interest to investigate whether similar relationships exist in them as well.
To generate a PaRcf1-6xHis overexpressing strain, wild-type spheroblasts were transformed with the newly constructed plasmid pExMtterhph-PaRcf1-6xHis-OEx containing a hygromycin B resistance gene in the pExMtterhph vector backbone 46 and the full-length PaRcf1 gene under control of the strong constitutive metallothionein promoter and the metallothionein terminator, with a 6xHis-tag coding sequence added before the gene's stop codon. Transformants were selected for hygromycin B resistance and verified by Southern blot analysis. Strains with a single integration of pExMtterhph-PaRcf1-6xHis-OEx were termed PaRcf1-6xHis_OEx. To construct the vector, the PaRcf1 gene including the 6xHis-tag coding sequence was amplified by PCR using the oligonucleotides BglII-PaRcf1_for (5′ -TAAGATCTATGTCGAACGGACCCCTCTC-3′ ) and XbaI-6xHis_rev (5′ -GCTCTAGATTAATGGTGATGGTGGTGATG-3′ ) with pKO6-PaRcf1-6xHis as a template, introducing BglII and XbaI restriction sites (underlined). The amplicon was cloned into the pExMtterhph backbone (BamHI/XbaI digested) to obtain pExMtterhph-PaRcf1-6xHis-OEx.

Transformation of P. anserina spheroblasts.
The respective strain to be transformed was grown on BMM at 27 °C under constant light for 3 days and subsequently under the same conditions in liquid complete medium (CM) for 2 days (CM medium: 1 g/l KH 2 PO 4 , 0.5 g/l KCl, 0.5 g/l MgSO 4 × 7 H 2 O, 10 g/l glucose, 3.7 g/l NH 4 Cl, 2 g/l tryptone, 2 g/l yeast extract and 1 g/l ZnSO 4 , FeCl 2 and MnCl 2 ; pH 6.5). 20 g of the resulting mycelium was washed with TPS buffer (5 mM Na 2 HPO 4 , 45 mM KH 2 PO 4 , 0.8 M sucrose; pH 5.5) and TPS buffer containing 20 mg/ml 'Glucanex' (Novozymes) was added to a final volume of 100 ml. After chopping the mixture in a 'Waring Blendor' the resulting suspension was incubated for 1.5 h at 35 °C. Following filtration through gauze and glass wool, the suspension was centrifuged for 10 min at 4.000 rpm to pelletise the spheroplasts and the pellet was washed three times with TPS buffer.

Lifespan determination.
To determine the lifespan of the strains used in this study, monokaryotic ascospores were isolated from independent crosses of the respective strains (in the case of the PaRcf1 deletion strain from a cross of wild type with Δ PaRcf1) and germinated for 3 days at 27 °C in the dark on BMM supplemented with 60 mM ammonium acetate. After germination, pieces of the resulting 3 day old mycelia were placed on M2 agar race tubes (M2 medium: 0.25 g/l KH 2 PO 4 , 0.3 g/l K 2 HPO 4 , 0.25 g/l MgSO 4 × 7 H 2 O, 0.5 g/l urea and 10 g/l yellow dextrin. Addition of 2.5 μ g/l biotin, 50 μ g/l thiamine, 5 mg/l citric acid × Measurement of growth rate under stress conditions. To assess the susceptibility of the strains used in this study to paraquat-induced oxidative stress, monokaryotic ascospores were germinated as described above. After germination, pieces of the resulting 3 d old mycelia were placed on agar plates containing M2 medium supplemented with different concentrations of paraquat (0, 80, 160 or 320 μ M) and incubated at 27 °C under constant light. Growth was recorded for 4 days and growth rate was expressed as growth of the mycelia in centimetres per day.
Fertility analysis. Assessment of female fertility was essentially performed as described previously 48 .
Isolates of wild type, Δ PaRcf1 and Δ PaRcf2, in each case originating from monokaryotic ascospores, were grown on agar plates containing M2 medium at 27 °C under constant light for 13 days. Following spermatization at day 13, all plates were incubated for an additional three days at 27 °C, after which the total number of perithecia developing on each plate was counted. The mean number of perithecia developing per plate overgrown with the wild-type strain 's' at 27 °C was defined as 1.

Oxygen consumption measurement.
To measure complex IV-and AOX-dependent oxygen consumption of wild type, Δ PaRcf1 and Δ PaRcf2, monokaryotic ascospores of each strain were germinated as described above. After germination, pieces of the resulting 3 d old mycelia were grown for 2 days on M2 medium at 27 °C under constant light and subsequently under the same conditions in liquid CM for 3 days. Small pieces of mycelium (dry weight 2 to 10 mg) were then transferred into the 'OROBOROS Oxygraph-2k' (OROBOROS INSTRUMENTS) high-resolution respirometer and oxygen consumption was measured in liquid CM medium according to the manufacturer's instructions. To inhibit respiration via complex IV, KCN was added to a final concentration of 1 mM. Respiration via AOX was inhibited by adding salicylhydroxamic acid (SHAM) to a final concentration of 4 mM. Absolute oxygen consumption was measured as pmol oxygen consumed per second and milligram dry weight mycelium. To express relative oxygen consumption after addition of specific respiratory inhibitors, absolute oxygen consumption of the respective strain (wild type, Δ PaRcf1 or Δ PaRcf2) in the presence of KCN or SHAM was normalized to its total absolute oxygen consumption with no added inhibitors.

Southern blot analysis.
Total DNA of P. anserina was isolated with a well-established method for rapid extraction of nucleic acids from filamentous fungi 49 . DNA digestion, gel electrophoresis and Southern blotting were performed according to standard protocols. For Southern blot hybridization and detection, Digoxigenin-labeled hybridization probes ('DIG DNA Labeling and Detection Kit' , Roche Applied Science) were used according to the manufacturer's instructions. The PaRcf1-specific hybridization probe was amplified by PCR using the oligonucleotides PaRcf1_A2 (5′ -AGGAACCGCTCGTCCCAATC-3′ ) and PaRcf1_B2 (5′ -CCTTGCCTGAGCAGCAACAC-3′ ) and corresponded to 371 nucleotides in exon 2 of PaRcf1. The PaRcf2-specific hybridization probe was amplified by PCR using the oligonucleotides PaRcf2_for1 (5′ -AAGACGCCCACTTCAAGG-3′ ) and PaRcf2_ rev2 (5′ -TGGCTTCCGCTCAGATAC-3′ ) and corresponded to 408 nucleotides in exon 1 of PaRcf2. The hph-specific hybridization probe corresponded to the 727 bp ClaI-NcoI-fragment of the plasmid pKO7 44 . As a hybridization probe specific for the phleomycin resistance gene (ble), the 1293 bp BamHI-fragment of the plasmid pKO3 50 was used.
Western blot analysis. Mitochondrial protein extracts from P. anserina strains were isolated according to a previously developed procedure 25 and further purified by discontinuous sucrose gradient (20-36-50%) ultracentrifugation 44  (Merck Millipore) according to the manufacturer's instructions. Separation of proteins by SDS-PAGE and subsequent transfer of proteins to PVDF membranes (Immobilon-FL, Millipore) were performed following standard protocols. Blocking and antibody incubation of blotted PVDF membranes were performed according to the Odyssey 'Western Blot Analysis' handbook (LI-COR).
In all analyses, secondary antibodies conjugated with the infrared dyes IRDye 800CW or IRDye 680CW (LI-COR) were used (antibody dilution: 1:15,000-20,000). The 'Odyssey Infrared Imaging System' (LI-COR) was used for detection of western blots and densitometric quantification was performed with the image processing and analysis software ImageJ according to the developer's documentation.
BN-PAGE and complex IV 'in-gel' activity assay. BN-PAGE was performed according to the protocol described in detail by Wittig and colleagues 51 . For preparation of each sample, 100 μ g of mitochondrial protein extracts were solubilized using a digitonin/protein ratio of 3:1 (w/w). Linear gradient gels (4-13%) overlaid with 3.5% stacking gels were used for separation of the solubilised samples. Respiratory chain components were then visualized by Coomassie blue staining and assigned as described previously 26 . To measure complex IV 'in-gel' activity, Coomassie blue staining was omitted and the gel was incubated in 50 mM phosphate buffer (pH 7.4) containing 1 mg/ml 3,3′ -diaminobenzidine, 24 U/ml catalase, 1 mg/ml cytochrome c and 75 mg/ml sucrose 52 . Densitometric quantification was performed as described above.
Statistical analysis. For statistical analysis of BN-PAGE, complex 'in-gel' activity, western blot and OxyBlot data as well as oxygen consumption measurements, two-tailed Student's t-test was used. For statistical analysis of lifespan and growth rate, two-tailed Wilcoxon rank-sum test was used. If not explicitly stated otherwise, the respective samples were compared to the appropriate wild-type sample. P-values < 0.05 were considered statistically significant.