The Mitochondrial Permeability Transition Pore Activates a Maladaptive Mitochondrial Unfolded Protein Response

Mitochondrial activity determines aging rate and the onset of chronic diseases. The mitochondrial permeability transition pore (mPTP) is a pathological pore in the inner mitochondrial membrane thought to be composed of the F-ATP synthase (complex V). Oligomycin sensitivity-conferring protein (OSCP), a subunit of F-ATP synthase, helps protect against mPTP formation. How the destabilization of OSCP may contribute to aging, however, is unclear. We have found that loss OSCP in the nematode Caenorhabditis elegans initiates the mPTP and shortens lifespan specifically during adulthood, in part via initiation of the mitochondrial unfolded protein response (UPRmt). Genetic or pharmacological inhibition of the mPTP inhibits the UPRmt and restores normal lifespan. The mitochondria of long-lived mutants are buffered from the maladaptive UPRmt, partially via the transcription factor FOXO3a/daf-16. Our findings reveal how the mPTP/UPRmt nexus may contribute to aging and age-related diseases and how inhibition of the UPRmt may be protective under certain conditions.


INTRODUCTION
As mitochondrial function declines with age, the frequency of the mPTP increases (Rottenberg and Hoek, 2017). The mPTP is central to early-stage pathologies associated with several agerelated diseases, including Alzheimer's and Parkinson's disease (AD, PD) and late-stage pathologies of ischemia-reperfusion injuries including heart attack and stroke (Ong et al., 2015;Panel et al., 2018). The mPTP is a pathological channel that forms in the inner mitochondrial membrane in response to excessive cytosolic Ca 2+ or high ROS conditions. Sustained opening of the mPTP leads to outer mitochondrial membrane rupture, release of Ca 2+ into the cytosol, and cell death (Bernardi and Di Lisa, 2015). Cyclosporine A (CysA), a well-characterized mPTP inhibitor, inhibits the mPTP by binding and sequestering cyclophilin D, a mitochondriallylocalized peptidyl prolyl isomerase that helps catalyze pore formation (Basso et al., 2005;Nakagawa et al., 2005). Genetic inhibition of cyclophilin D protects against mPTP formation (Baines et al., 2005). The mPTP is also modulated by adenine nucleotide translocases (ANTs), which exchange ADP and ATP across the IMM. Genetic inhibition of ANTs also help prevent pore formation (Karch et al., 2019;Kokoszka et al., 2004). Both CysA and loss of ANTs have been shown to extend lifespan in C. elegans (Farina et al., 2008;Ye et al., 2014). Thus, the mPTP appears to be an important modulator of healthspan and lifespan.
Though the identification of the proteins that make up the mPTP is controversial, recent studies have pointed to F-ATP synthase (complex V) as the most probable candidate due to its ability to bind cyclophilin D and form Ca 2+ currents. (Bernardi and Di Lisa, 2015;Mnatsakanyan and Jonas, 2020). Some models posit that dimeric forms of F-ATP synthase open to form a pore while other models have suggested that the pore occurs via the membrane-bound proton rotor (Alavian et al., 2014;Azarashvili et al., 2014;Bonora et al., 2013;Bonora et al., 2017;Giorgio et al., 2013;Mnatsakanyan et al., 2019;Neginskaya et al., 2019;Urbani et al., 2019). ATP5O, also known as oligomycin sensitivity-conferring protein (OSCP), a subunit of the F-ATP synthase that regulates ATPase rotational activity to provide efficient ATP production (Murphy et al., 2019), has emerged as an important regulator of the mPTP. OSCP levels decrease with normal aging and loss of OSCP increases propensity for pore formation in vitro and in mouse models of AD (Beck et al., 2016b;Gauba et al., 2017;Giorgio et al., 2013). Conversely, OSCP confers protection against the pore under low pH conditions and OSCP overexpression protects from mPTP initiation in AD and cardiac dysfunction models (Antoniel et al., 2018;Beck et al., 2016b;Guo et al., 2020). Thus, OSCP appears to be an important regulator of aging and disease progression, possibly via its ability to modulate mPTP formation.
Under stress, the mitochondria attempt to repair the damage, recycle damaged mitochondria, or, under deleterious circumstances, initiate cell death. Similar to the endoplasmic reticulum unfolded protein response (UPR ER ) and the cytoplasmic heat shock response (HSR), the mitochondrial unfolded protein response (UPR mt ) is capable of initiating a broad-range transcriptional response that, among other functions, aids in the refolding of mitochondrial matrix proteins (Naresh and Haynes, 2019). Recent studies also show that a loss of mitochondrial membrane potential correlates with activation of the UPR mt , and disruption of mitochondrial processes other than protein misfolding, such as those involved in TCA cycle and lipid catabolism, also induce the UPR mt (Rolland et al., 2019). UPR mt activation is associated with longevity and improvement in neurodegenerative models (Durieux et al., 2011;Kim et al., 2016;Merkwirth et al., 2016;Sorrentino et al., 2017;Tian et al., 2016), but it has also conversely been shown to increase neurodegeneration, propagate mtDNA mutations, and exacerbate ischemic conditions (Lin et al., 2016;Martinez et al., 2017;Yung et al., 2019), underscoring its complexity. If left unmitigated, UPRs can initiate cell death (Iurlaro and Munoz-Pinedo, 2016;Munch and Harper, 2016). Thus, the context or cellular environment are important determinants of whether UPR mt induction results in beneficial or detrimental effects.
In C. elegans, mild mitochondrial perturbations early in life can extend lifespan. Loss of OSCP/atp-3 has previously been shown to extend lifespan when initiated during larval development (Dillin et al., 2002;Rea et al., 2007). In contrast, here, we have determined that loss of OSCP/atp-3 during adulthood leads to initiation of the mPTP, the UPR mt , and a shortened lifespan. Surprisingly, atfs-1, the UPR mt master transcription factor (Haynes et al., 2010;Nargund et al., 2012), helps drive the reduction of lifespan, suggesting that the UPR mt program promotes aging during adulthood. The adult UPR mt is responsive to mPTP regulators, including the immunosuppressive drug, CysA, as well as a mitochondrially-localized cyclophilin and ANTs, pointing to a previously undiscovered coordination between the UPR mt and the mPTP. We find that proton rotor subunits as well as subunits important for dimerization of the F-ATP synthase are essential for transducing the adult UPR mt . Loss of these subunits as well as pharmacological CysA treatment restores lifespan due to loss of OSCP/atp-3. These results are consistent with current models that posit that the F-ATP synthase forms the mPTP (Bernardi and Di Lisa, 2015). Remarkably, we find that long-lived mutants which have previously been shown to be protected from the mPTP (Zhou et al., 2019) expressing high FOXO3a/daf-16 activity do not initiate a UPR mt during adulthood. Overall, our findings point to a model in which loss of OSCP/atp-3 in adults induces mPTP formation and subsequent detrimental activation of the UPR mt , from which long-lived mutants are buttressed. Understanding the relationship between these two mitochondrial processes will undoubtedly further our understanding of aging as well as disparate age-related disorders, including neurodegenerative diseases, cancer, heart-attack, and stroke.

Loss of OSCP/atp-3 Induces Mitochondrial Permeability Transition Pore (mPTP) Characteristics
The opening of the mPTP is characterized by a loss of membrane potential as well as an increase in cytosolic Ca 2+ . We observed that a reduction in the abundance of OSCP/atp-3 by RNA interference (RNAi) caused a loss of membrane potential as measured by the mitochondrial dye, TMRM, in young adult worms ( Figure 1A, 1B). Strikingly, RNAi of other OXPHOS subunits from complex I, IV, and V had no effect on membrane potential. RNAi of OSCP/atp-3 during adulthood also caused an increase in cytosolic Ca 2+ as measured by the intestinal FRET-based Ca 2+ reporter, KWN190, while RNAi of other F-ATP synthase subunits, F6/atp-4 and d/atp-5, did not ( Figure 1C). Loss of OSCP/atp-3 also induced mitochondrial swelling and fragmentation ( Figure 1D, 1E). In contrast, we observed no swelling or fragmentation following the reduction in expression of the F-ATP synthase subunit d/atp-5 ( Figure 1F). These results suggest that loss of OSCP/atp-3 during adulthood uniquely induces the mPTP in C. elegans.
To verify the stability of the mitochondrial membrane potential after RNAi of other OXPHOS subunits, we checked for RNAi efficiency via qPCR; we observed efficient mRNA reduction from subunit cco-1 from complex IV as well as from several of the complex V subunits, including atp-1, atp-2, atp-4, and atp-5 (Figure S1A-S1E). We also examined protein levels for subunits in which antibodies were available and observed that RNAi of the complex I subunit NUO-2 and complex V subunits ATP-1 and ATP-2 resulted in significant knockdown of protein levels ( Figure S1F-S1H). These findings show that knockdown of OXPHOS subunits other than OSCP/atp-3 do not lead to a loss of membrane potential.

Loss of OSCP/atp-3 Induces a Unique UPR mt During Adulthood
A recent study showed that a loss of membrane potential is associated with inducing the UPR mt (Rolland et al., 2019). To determine if loss of OSCP/atp-3 selectively induces the UPR mt during adulthood due to the observed drop in membrane potential, we utilized a GFP reporter under the promoter of the mitochondrial UPR chaperone, hsp-6 (phsp-6::GFP) (Yoneda et al., 2004) and compared it to select representative OXPHOS genes encoding complex I, III, IV, and V subunits. RNAi of OXPHOS subunits induced little to no UPR mt if initiated after the last larval stage (L4), which we termed the post-developmental UPR mt (pdvUPR mt ). Only RNAi of OSCP/atp-3 induced a robust UPR mt in young adults (Figure 2A, 2B, Table S1), the timing of which corresponded with the drop in membrane potential ( Figure 1A, 1B). In contrast, RNAi of all same genes induced a robust UPR mt if initiated during the early larval stages (L1, L2, and L3) of RNAi of various OXPHOS subunits in N2 wild-type worms did not lead to changes in the membrane potential as measured by the TMRM dye, except for RNAi of OSCP/atp-3. RNAi was administered for 48 hours beginning at young adulthood. TMRM was administered via the NGM plates. (B) Quantification of TMRM intensity from (A). Data are the mean ± SEM of ≤ 15 animals combined from three biological experiments. **p ≤ 0.01 by Student's t-test. I, IV, and V correspond to OXPHOS complexes. (C) RNAi of OSCP/atp-3 leads to an increase in cytosolic calcium as measured by the FRET-based calcium indicator protein D3cpv/cameleon. RNAi was administered for 48 hours beginning at young adulthood. Data are the mean ± SEM of ≤ 15 animals combined from three biological experiments. **p ≤ .01 by Student's t-test. (D-F) Confocal micrographs of intestinal mitochondria labeled with GFP (pges-1::GFP mt ) in young adults. RNAi and was administered for 48 hours beginning at young adulthood, then worms were removed from the RNAi and aged until Day 7 of adulthood followed by collection for microscopy. Top panels, fluorescent channel; bottom panels, rendering of individual mitochondria. See Materials and Methods for details on rendering.  (Figure 2A, 2B).These results are consistent with previous reports demonstrating that the UPR mt is robustly induced during development but poorly induced during adulthood in C. elegans (Durieux et al., 2011;Labbadia and Morimoto, 2015). Loss of membrane potential per se via FCCP, a potent mitochondrial uncoupler, however, did not induce a UPR mt in young adults, suggesting that uncoupling and its associated loss in membrane potential is not sufficient to induce a pdvUPR mt ( Figure S1I). Post-developmental loss of OSCP/atp-3 increased endogenous transcript levels of hsp-6 as well as endogenous HSP-6/mtHSP70 protein levels ( Figure S2A, S2B). Post-developmental loss of OSCP/atp-3 mildly induced the mitochondrial chaperone reporter phsp-60::GFP ( Figure S2C, S2D). Neither the UPR ER nor the HSR were induced by post-developmental loss of OSCP/atp-3 ( Figure S2C). RNAi of other mitochondrial genes that are known to induce a dvUPR mt , clk-1 (coenzyme Q hydroxylase), mrps-5 (mitochondrial ribosome), and tomm-22 (translocase of outer mitochondrial membrane) Bennett et al., 2014;Houtkooper et al., 2013), did not induce the pdvUPR mt (Table S1). Importantly, we found that the pdvUPR mt was dependent on the master UPR mt transcription factor, atfs-1, as well as its previously described co-factors, the peptide exporter haf-1 and the ubiquitin-like protein ubl-5 ( Figure 2C, 2D) (Haynes et al., 2010), demonstrating that the pdvUPR mt is regulated similarly to the previously described dvUPR mt . Thus, loss of OSCP/atp-3 induces a robust and specific pdvUPR mt which is dependent on the conserved transcription factor ATFS-1.

The Post-Developmental UPR mt is Temporally Confined and Reversible
Given that the UPR mt has been studied during development in C. elegans, we sought to determine the window of the UPR mt during adulthood. We initiated RNAi of OSCP/atp-3 beginning at the last larval stage (L4 stage) and every few hours thereafter into adulthood. GFP expression was examined 48 hours after RNAi initiation ( Figure S2E). We observed that the pdvUPR mt is initiated up to 6 hours after the L4 stage, after which RNAi of OSCP/atp-3 no longer induced the UPR mt . In contrast, RNAi of COX5B/cco-1 of complex IV had no effects on the UPR mt at any of these stages ( Figure S2E). For all subsequent post-development experiments, RNAi was therefore administered at the young adult stage corresponding to 4-hours after the L4 stage. Thus, the pdvUPR mt is confined to pre-gravid stages of adulthood, corresponding with previous reports showing a global decline in stress responses at the onset of egg-laying (Labbadia and Morimoto, 2015).
Previous studies have shown that developmental RNAi of cco-1 RNAi leads to persistent activation of the UPR mt into adulthood, even after removal from RNAi (Durieux et al., 2011). Similarly, we observed that developmental RNAi of OSCP/atp-3 initiated a UPR mt that persisted into adulthood, even after removal from atp-3 RNAi ( Figure S2F). In contrast, removal from post-developmental atp-3 RNAi treatment led to a steady decline in the GFP signal ( Figure S2F), suggesting the activation of the pdvUPR mt is reversible.

Loss of ATP Synthase ATPases Induce a Post-Developmental UPR mt
F-ATP synthase is composed of a membrane-bound proton rotor (Fo), a catalytic ATPase that converts ADP to ATP (F1), and peripheral stalk and supernumerary subunits that help bridge these two portions together ( Figure 2F). OSCP/atp-3 sits atop the ATPase and helps tether it to the peripheral stalk subunits. We systematically tested via RNAi whether loss of F-ATP synthase subunits other than OSCP/atp-3 could induce a pdvUPR mt . During adulthood, loss of rotor For developmental treatment, worms were exposed to RNAi beginning from eggs for 72 hours. For post-developmental treatment, worms were exposed to RNAi beginning from young adulthood for 48 hours. CV, control vector RNAi; dev, development; post-dev, post-development.
(D) Quantification of GFP intensity from (C). Data are the mean ± SEM of ≤ 15 animals combined from three biological experiments. **p ≤ 0.01 by Student's t-test. pdv, post-development.
(E) Post-developmental RNAi of ATPase and c-ring subunits but not peripheral stalk or supernumerary subunits of the F-ATP synthase induced the phsp-6::GFP reporter. Worms were exposed to RNAi beginning from young adulthood for 48 hours. post-dev, post-development.
(G) Quantification of GFP intensity from (E). Data are the mean ± SEM of ≤ 15 animals combined from three biological experiments. *p ≤ 0.05, **p ≤ 0.01, and ***p ≤ .0001 by Student's t-test. pdv, post-development.  Table S2). Loss of the ATPase subunits (a/atp-1; b/atp-2; d/F58F12.1; e/hpo-18) induced a mild to moderate UPR mt , though none as robustly as loss of OSCP/atp-3. Subunits a and e induced a moderate UPR mt , while loss of the b and d subunits induced a limited, though significant, UPR mt ( Figure 2E, 2G, Table S2). In contrast, during development, loss of rotor subunits, ATPase subunits, or peripheral stalk and supernumerary subunits all induced a robust UPR mt ( Figure S2G, S2H). Thus, in addition to the robust induction of the pdvUPR mt caused by loss of OSCP/atp-3, loss of F-ATP synthase ATPases moderately to mildly induce the pdvUPR mt .

The Post-Developmental UPR mt is dependent on Mitochondrial Permeability Transition Pore (mPTP) Factors
To determine if the pdvUPR mt is initiated in response to the mPTP, we tested pharmacological and genetic modulators of the mPTP on induction of the pdvUPR mt . The mPTP is known to be inhibited by the immunosuppressive drug, cyclosporin A (CysA). CysA binds cyclophilins, which in the cytoplasm regulates calcineurin signaling (Liu et al., 1991;Takahashi et al., 1989), while in the mitochondria inhibits the mPTP (Nicolli et al., 1996). To parse out the mitochondrial versus cytoplasmic functions of CysA, we also tested the cytoplasmic-only immunosuppressive drug, FK506, which acts similarly to CysA in that it modulates calcineurin signaling in the cytoplasm (Liu et al., 1991). We observed that CysA strongly inhibited the pdvUPR mt but not dvUPR mt in a dose-dependent manner ( Figure 3A, 3B, Figure S3A). In contrast, we found that FK506 had no effect on the pdvUPR mt ( Figure 3A, 3B, Figure S3B), demonstrating that CysA acts in the mitochondria to suppress the pdvUPR mt . The mPTP has been shown to be regulated by adenine nucleotide translocases (ANTs) of the inner mitochondrial membrane and loss of ANTs helps prevent the mPTP (Karch et al., 2019). We tested ant-1.1, which is ubiquitously expressed in C. elegans, and ant-1.2, which is expressed predominantly in the pharynx and intestines (Farina et al., 2008). RNAi of ant-1.1 moderately suppressed the pdvUPR mt while RNAi of ant-1.2 strongly suppressed the pdvUPR mt , but not the dvUPR mt ( Figure 3C, 3D). In mammals, CysA acts in the mitochondria to inhibit the mPTP by binding and sequestering cyclophilin D, a peptidyl prolyl isomerase (Nicolli et al., 1996). C. elegans contains 17 poorly defined cyclophilins, of which 2 are predicted to be mitochondrially localized, cyn-1 and cyn-17 (Table S3). RNAi of cyn-1 did not inhibit the pdvUPR mt while cyn-17 did ( Figure 3D), suggesting that cyn-17 may act similarly to cyclophilin D in mediating a conformation change that leads to the mPTP. RNAi of cyn-17 did not affect the dvUPR mt ( Figure 3C). Together, these results show that the pdvUPR mt is regulated by canonical pharmacological and genetic mPTP factors.

Loss of Essential F-ATP Synthase Subunits Important for the mPTP Suppress the Post-Developmental UPR mt
Current models posit that the F-ATP synthase forms a pore that is capable of releasing Ca 2+ under conditions of high oxidative stress, leading to rupturing of the mitochondria and initiation of cell death cascades. Some models suggest that ATP synthase dimers are required for the mPTP ( Figure 3G) and that peripheral and supernumerary subunits are essential for pore formation (Carraro et al., 2014;Guo et al., 2019). Other models demonstrate that ATP synthase monomers ( Figure 2F) are sufficient for the mPTP and specify the c-ring proton rotor as the and FK506 had no effect on the dvUPR (A) but selectively suppressed the pdvUPR mt in the phsp-6::GFP reporter strain (B). Data are the mean ± SEM of ≤ 15 animals combined from three biological experiments. ***p ≤ .0001 by Student's t-test. (C, D) Loss of cyclophilin cyn-17 and adenine nucleotide translocase ant-1.2 via RNAi had no effect on the dvUPR mt (C) while their loss suppressed the pdvUPR mt in the phsp-6::GFP reporter strain (D). ant-1.1 partially suppressed the pdvUPR mt (D). Data are the mean ± SEM of ≤ 15 animals combined from three biological experiments. *p ≤ .05 , ***p ≤ .0001 by Student's t-test. (E) Concomitant post-developmental RNAi of OSCP/atp-3 and individual F-ATP synthase subunits inhibited the postdevelopmental UPR mt to varying degrees. (F) Quantification of GFP intensity from (E). Data are the mean ± SEM of ≤ 15 animals combined from three biological experiments. **p ≤ .001, ***p ≤ 0.0001 compared to CV + atp-3 condition by Student's t-test. pdv, post-development. (G) Schematic of ATP synthase dimers. White subunits, ATPase; black subunits, H+ rotor/c-ring; grey subunits, peripheral stalk and supernumerary subunits; red subunit, oligomycin sensitivity-conferring protein (OSCP/atp-3).  *** a n t -1 . 2 a n t -1 . 1 * B

phsp-6::GFP
phsp-6::GFP actual pore-forming component (Alavian et al., 2014;Azarashvili et al., 2014;Bonora et al., 2013;Bonora et al., 2017;Giorgio et al., 2013;Mnatsakanyan et al., 2019;Neginskaya et al., 2019;Urbani et al., 2019). To determine whether the structural integrity of F-ATP synthase subunits were required for the pdvUPR mt , we systematically knocked down OSCP/atp-3 as well as one additional F-ATP synthase subunit via RNAi. When we knocked down the c-ring subunits (color coded black) as well as peripheral and supernumerary subunits (color coded grey) via RNAi in adults, we observed nearly complete inhibition of the OSCP/atp-3 RNAi-mediated pdvUPR mt (( Figure 3E, 3F). When we knocked down the ATPase subunits in adults, we observed that loss of the b/atp-2 subunit robustly suppressed the OSCP/atp-3 RNAi-mediated pdvUPR mt , possibly due to its role in modulating Ca 2+ in the mPTP (Giorgio et al., 2017), while loss of a/atp-1 moderately inhibited the pdvUPR mt ( Figure 3E, 3F, color coded white). Loss of the ATPase subunits d/F58F12.1 or e/hpo-18 in adults did not affect the pdvUPR mt . In contrast, dual loss of subunits during development all robustly activated the dvUPR mt ( Figure S3C, S3E). Dual loss of subunits from other OXPHOS complexes (NDUFS3/nuo-2, complex I; COX5B/cco-1, complex IV) had no effect or slightly increased the dvUPR mt and the pdvUPR mt (Fig S3D,  S3F). Thus, we find subunits critical for dimerization (peripheral and supernumerary subunits) and proton translocation (c-ring rotor) are required to transduce the OSCP/atp-3 RNAi-mediated pdvUPR mt . We also find that the b/atp-2 subunit, previously found to play an important role in Ca 2+ mediated mPTP, is required to transduce the OSCP/atp-3 RNAi-mediated pdvUPR mt . Taken together, these findings support a model in which inhibition of the mPTP via deletion of critical F-ATP synthase subunits inhibits the pdvUPR mt .

Loss of OSCP/atp-3 During Adulthood Promotes Aging Partially via the UPR mt
Previous reports have shown that loss of OSCP/atp-3 initiated during development robustly increases lifespan in C. elegans (Dillin et al., 2002), but how loss of OXPHOS subunits during adulthood affects lifespan has not been well studied. We initiated RNAi during both development and post-development. As previously reported, we found that developmental RNAi treatment led to lifespan extension ( Figure 4A, Table S4). Worms continuously exposed to OSCP/atp-3 RNAi during post-development experienced a high incidence of matricide (data not shown). To circumvent this outcome, we administered OSCP/atp-3 RNAi to young adults for 48 hours of adulthood and observed approximately a 38% decrease in lifespan independent of matricide ( Figure 4A, 4B, Table S4). Surprisingly, when atfs-1 mutants were exposed to OSCP/atp-3 RNAi during adulthood for 48 hours, we observed only about a 19% decrease in lifespan, suggesting that the initiation of the pdvUPR mt via atfs-1 contributes to reduced lifespan ( Figure 4B, Table S4). Loss of other OXPHOS subunits had little or no effect on lifespan when administered during adulthood for 48 hours ( Figure 4C, Table S4). Thus, we have determined that loss of OSCP/atp-3 induces an age-accelerating, transcriptionally-dependent pdvUPR mt not previously described in C. elegans.

Pharmacological or Genetic Inhibition of the mPTP Restores Lifespan
To determine if inhibition of the mPTP could also inhibit its associated toxicity, we tested the mPTP inhibitor, CysA, in lifespans. We found that CysA treatment was sufficient to rescue the shortened lifespan caused by RNAi of OSCP/atp-3 in adults ( Figure 4D). Based on our observations that loss of peripheral stalk subunits are capable of suppressing the pdvUPR mt , we tested if their loss would also suppress the shortened lifespan. RNAi with either F6/atp-4 or  Figure 4E, 4F). Loss of d/atp-5 subunit also suppressed the mitochondrial swelling and fragmentation observed after OSCP/atp-3 RNAi ( Figure 4G, 4H). Thus, inhibition of key subunits of F-ATP synthase as well as pharmacological inhibition of the mPTP by CysA rescues the mPTP/pdvUPR mt toxicity and rescues its effects on aging.
To verify that the use of dual-RNAi did not interfere with knockdown of OSCP/atp-3, we assessed an endogenously-expressing patp-3::ATP-3::GFP translational reporter generated via CRISPR-Cas-9 ( Figure S4A). We observed efficient knockdown of OSCP/atp-3 via RNAi in the presence of either b/atp-2 or d/atp-5 RNAi, two subunits that suppress the pdvUPR mt , demonstrate that the inhibition of the pdvUPR mt is not due to ineffective RNAi of OSCP/atp-3 but rather the functional consequences of removal of additional ATP synthase subunits. To examine the effects of dual-RNAi another way, we examined how loss of ATP synthase subunits impacted the pdvUPR mt after a/atp-1 RNAi, which induced the second most robust pdvUPR mt ( Figure 2E, 2G). Remarkably, we see the same pattern of pdvUPR mt activation and inhibition as with loss of OSCP/atp-3: loss of the ATPase subunit b/atp-2 and peripheral stalk subunit d/atp-5 suppressed the a/atp-1 RNAi-mediated pdvUPR mt while loss of NDUFS3/nuo-2 and COX5B/cco-1 had no effect ( Figure S4B). Importantly, immunoblots against a/atp-1 showed similar protein knockdown under all conditions ( Figure S4B), confirming that dual-RNAi is an effective method to assess the structural components of the F-ATP synthase.
We next sought to determine the mechanism of suppression of the pdvUPR mt by mutations conferring longevity. The longevity of glp-1 and daf-2 has previously been shown to be dependent on the transcription factor FOXO3A/daf-16 (Hsin and Kenyon, 1999;Kenyon et al., 1993) and we also found that the longevity of fem-1(hc17) to be dependent on FOXO3a/daf-16 ( Figure S5A, S5B). On the other hand, eat-2 has previously been found to be dependent on FOXA/pha-4 (Panowski et al., 2007). Thus, we tested whether FOXO3a/daf-16 was required for the suppression of the pdvUPR mt and found that introducing the mutation, daf-16(mu86), partially restored the pdvUPR mt in glp-1(e2141) mutants and completely restored it in fem-1(hc17) mutants ( Figure 5C). Previous studies have suggested that an increase in mitochondrial biogenesis may protect against opening of the mPTP (Goncalves et al., 2016). To determine if fem-1(hc17) and glp-1(e2141) exhibited enhanced mitochondrial content via FOXO3a/daf-16, we examined the mitochondria in the intestines where the UPR mt is most robustly activated. We    observed significantly higher mitochondrial mass in the intestines of glp-1(e2141) and fem-1(hc17), which was dependent on FOXO3a/daf-16. To determine if the increased mitochondrial content may be due to a block in mitophagy or autophagy, we examined mitochondrial content after RNAi of pdr-1 (parkin homology) and hlh-30 (TFEB homolog). RNAi of pdr-1 or hlh-30 significantly increased the mitochondrial content in glp-1(e2141) and fem-1(hc17) mutants, suggesting that a block in mitophagy/autophagy was not the responsible mechanism. Overall, these findings suggest that glp-1(e2141) and fem-1(hc17) mutants are protected from initiating a maladaptive mPTP/pdvUPR mt , possibly due to enhanced mitochondrial biogenesis via FOXO3a/daf-16.

DISCUSSION
While loss of the ATP synthase subunit OSCP/atp-3 during development leads to lasting activation of the UPR mt and is associated with longevity (Dillin et al., 2002;Rea et al., 2007), we have discovered that loss of this subunit during adulthood induces the mPTP and activates a reversible and atfs-1-dependent UPR mt (pdvUPR mt ). Furthermore, we observed that activation of the UPR mt during adulthood helps drive aging. Suppression of the mPTP/UPR mt via genetic or pharmacological interventions is protective. Similarly, long-lived mutants from various longevity paradigms dependent on FOXO3a/daf-16 are protected from the mPTP/UPR mt during adulthood.
The fact that loss of other OXPHOS subunits or loss of membrane potential via FCCP are not capable of inducing the UPR mt during adulthood suggests a specific activation of the UPR mt due to the mPTP. In contrast, it does not appear that the UPR mt is activated in response to the mPTP during development. Thus, loss of OXPHOS subunits can have drastically different effects depending on the life stage or cellular milieu of the organism and suppression of the UPR mt can be beneficial.
While the mPTP detrimental role in health is well established, considerable evidence suggests that the UPR mt contributes to health and longevity Merkwirth et al., 2016;Mouchiroud et al., 2013;Sorrentino et al., 2017). Activating the UPR mt in neurons can activate protective cell non-autonomous signals and also epigenetically rewire C. elegans to live longer (Durieux et al., 2011;Merkwirth et al., 2016;Tian et al., 2016;Zhang et al., 2018). NAD+ boosters appear to activate the UPR mt and contribute to longevity and ameliorate AD Sorrentino et al., 2017). However, an unbiased screen that identified activators of the UPR mt found no correlation between UPR mt activation and longevity (Bennett et al., 2014) and constitutive activation of the UPR mt in dopaminergic neurons lead to increased neurodegeneration (Martinez et al., 2017). Activation of the UPR mt in our setting is distinct in that it is specifically linked to activation of the mPTP. Thus, it is possible that preemptively boosting the UPR mt may ward of aging and disease while activation in a diseased setting may exacerbate conditions, akin to instances of inflammation in disease.
We propose a model in which loss of OSCP/atp-3 induces a conformational change in F-ATP synthase that leads to pore formation and activation of the UPR mt during adulthood but not during development ( Figure 5D). Previous reports have shown that loss of OSCP increases susceptibility to Ca 2+ -induced mPTP formation and that key residues within the OSCP are required to suppress the mPTP during conditions of low pH (Antoniel et al., 2018;Giorgio et al., 2013), suggesting that a functional and intact OSCP protects against pore formation. OSCP levels have also been shown to decrease with age while concomitantly increasing its binding to amyloid b, suggesting that loss of OSCP destabilizes the remaining ATP synthase to increase pore formation (Beck et al., 2016b). However, it has also been shown that OSCP provides the binding site for cyclophilin D and thus it has been proposed to be critical in the formation of the mPTP (Giorgio et al., 2017). However, these findings have been contested as knock out studies of OSCP show that cyclophilin D still binds to ATP synthase and EM studies have identified a cyclophilin D structure positioned on peripheral stalk subunits lateral to the OSCP (Daum et al., 2013;He et al., 2017a). Thus, our findings support a model in which loss of OSCP/atp-3 induces a conformation change that is favorable for cyclophilin binding to the remaining peripheral stalk proteins or to sites independent of the F-ATP synthase, leading to destabilization of ATP synthase, pore formation with a loss of membrane potential, and subsequent activation of the pdvUPR mt .
The loss of OSCP/atp-3 may also be more impactful than other subunits. Unlike other ATP synthase subunits, OSCP/atp-3 is prominently accessible in the mitochondrial matrix and has a diverse set of binding partners, such as estradiol and p53, which can modulate ATP production (Bergeaud et al., 2013;Zheng and Ramirez, 1999). Its loss may induce a strong protein misfolding cascade, reminiscent of the He and Lemasters mPTP model first proposed in 2002 (He and Lemasters, 2002). In this model, it was proposed that exposure to activators of the mPTP, such as oxidants, causes protein misfolding of integral membrane proteins, thereby recruiting chaperones such as cyclophilin D to repair. If the protein misfolding is unable to be repaired, then Ca 2+ , along with cyclophilin D, could catalyze pore formation in a CysAdependent manner. Indeed, both paraquat and manganese have been shown to induce both the UPR mt and the mPTP in separate studies (Angeli et al., 2014;Costantini et al., 1995;Rao and Norenberg, 2004). However, more research is needed to clarify the relationship between protein misfolding and the mPTP.
Despite mounting evidence implicating F-ATP synthase as the pore-forming component of the mPTP, this remains controversial. The Walker lab has systematically deleted nearly every subunit of ATP synthase in a cell model and showed that a pore still forms (Carroll et al., 2019;He et al., 2017a;He et al., 2017b), leading some groups to suggest that in the absence of an intact F-ATP synthase, smaller low conductance CysA-dependent pores distinct from the mPTP still form (Neginskaya et al., 2019). Other groups have proposed that the mPTP can be mediated by ANT, as well as a cyclophilin D binding structure, which supports a "multi-pore model", which would explain why deletion of putative pore components may still yield pore formation (Carraro et al., 2020;Karch et al., 2019).
AD and PD both display evidence of the mPTP and in some instances, elevated UPR mt profiles have also been observed (Beck et al., 2016a;Beck et al., 2016b;Ludtmann et al., 2018;Perez et al., 2018;Sorrentino et al., 2017), but the relationship between these two mitochondrial processes have not been fully explored. Ischemic reperfusion directly causes the mPTP but little is known about the UPR mt under these conditions. Establishing a clearer understanding of the relationship between the UPR mt and the mPTP in these disease states could result in the development of new therapeutics for these and related disorders. BioResource Project. The Y82E9BR.3 RNAi clone was a generous gift from Dr. Seung-Jae V. Lee from the Seoul National University College of Medicine. This work was supported by NIH grants RFAG057358 and R01AG029631 and the Larry L. Hillblom Foundation.

DECLARATION OF INTEREST
G.J.L. is cofounder of GeroState alpha and declares no financial interest related to this work. A.A.G. is founder of Image Analyst MKII Software and declares a financial interest.

CONTACT FOR REAGENT AND RESOURCE SHARING
Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contacts, Gordon Lithgow, glithgow@buckinstitute.org, Julie Andersen, jandersen@buckinstitute.org, and Suzanne Angeli, sangeli@buckinstitute.org

Nematode Strains
The following strain was obtained from the National BioResource Project: atfs-1(tm4525). The following strains were obtained from the Caenorhabditis Genetic Center (

Nematode and Bacterial Culture Conditions
Nematodes were maintained on nematode growth medium (NGM) plates. NGM plates were seeded with E. coli OP50 obtained from CGC that was grown in LB broth at 37˚C for 18 hours shaking at 225 rpm. Plates with bacteria were dried for 48 hours before use.
For RNAi experiments, E. coli HT115 (DE3) bacteria obtained from the Ahringer and Vidal RNAi Library was used (Kamath et al., 2003;Rual et al., 2004). All RNAi clones were verified via sequencing (Eurofins™). RNAi plates were prepared by cooling NGM to 55˚C and supplementing with a final concentration of 50ug/ml carbenicillin and 1mM Isopropyl β-d-1thiogalactopyranoside (IPTG). RNAi bacteria was inoculated with one colony of RNAi bacteria into LB with 50ug/ml carbenicillin and was grown shaking overnight for 18 hours at 37˚ at 225 rpm.

Post-Developmental Timing
To achieve synchronous nematode populations, day 1 adult nematodes were allowed to lay eggs for 2 hours on seeded NGM plates. For convenience, nematodes were developed at 25˚C on E. coli OP50 until worms were visibly past the L4 stage (loss of crescent) but not yet gravid, approximately 45 hours for wild-type (although the time it takes for the worms to reach the young adult stage varies by strain). Nematodes were shifted to 20˚C once they reached adulthood.

RNA Interference (RNAi) Treatment
For developmental treatments, synchronized eggs were moved onto plates seeded with RNAi bacteria and developed at 20˚C for 72 hours. Nematodes were then either collected for analysis or for lifespans, remained on RNAi bacteria for the remainder of their life for survival analysis. For post-developmental treatments, synchronized eggs were developed on plates with E. coli OP50 at 25˚C until the young adult stage and then transferred to RNAi plates at 20˚C. Nematodes were collected after 48 hours for analysis or for lifespans, moved onto E. coli OP50 for the remainder of their life.

Quantitative RT-PCR
Approximately 300 adult nematodes were collected; nematode pellets were resuspended in 300uL RNA Lysis Buffer and frozen. Pellets was thawed, vortexed, and snap frozen three times. Zymo Research Quick-RNA MiniPrep kit was used to extract RNA. The primers used for qPCR are as follows: act-1, Forward: ACGACGAGTCCGGCCCATCC act-1, Reverse: GAAAGCTGGTGGTGACGATGGTT

Lifespans
Day 1 adult nematodes were allowed to lay eggs for 2 hours on seeded NGM plates to obtain a synchronous aging population. 5-fluoro-2-deoxyuridine (FUdR) was omitted from plates due to its potentially confounding effects (Angeli et al., 2013). Worms are transferred to freshly seeded bacterial plates every day for the first 7 days of adulthood and then as needed afterwards. Worms were scored as dead when they failed to respond to gentle prodding with a platinum wire. Worms that experienced matricide or bagging were censored.

Tetramethylrhodamine Methyl Ester (TMRM) Staining
Plates were prepared by spotting seeded NGM plates with a TMRM solution diluted in water to a final concentration of 0.1 µM in the plates. Water was used as a solvent control. Plates were allowed to dry for 24 hours before use. For developmental experiments, synchronized eggs were placed on plates for 72 hours and then nematodes were collected for analysis. For postdevelopmental experiments, young adult nematodes were placed on plates for 48 hours and then collected for analysis.

Cyclosporine A (CysA) Treatment
Plates were prepared by spotting seeded NGM plates with a CysA (stock solution in DMSO) solution diluted in 100% ethanol. Comparable amounts of DMSO and ethanol were used as solvent controls. Plates were allowed to dry for 24 hours before use. For developmental experiments, synchronized eggs were placed on plates for 72 hours and then nematodes were collected for analysis. For post-developmental experiments, young adult nematodes were placed on plates for 48 hours and then collected for analysis; for lifespans, worms were moved to regular NGM plates for the remainder of their life after 48 hours on drug-treated RNAi bacteria.

Microscopy
Worms were anesthetized with 2mM levamisole and mounted on 2% agarose pads on glass slides. Fluorescence micrographs of GFP and TMRM were taken using a Zeiss Imager A2 at 5x magnification with 600ms exposure using the ZEN software. GFP expression was enhanced using the brightness/contrast tool in Photoshop. The same parameters were used for all images.
Confocal micrographs of mitochondrially targeted GFP and the cytosolic calcium sensor D3cpv (Zhang et al., 2016) were taken using a Zeiss LSM780 laser scanning confocal microscope using a 63× Plan Apochromat NA1.4. To visualize outlines of mitochondria, Image Analyst MKII (Image Analyst Software, Novato, CA) was used. Selected rectangular regions of interest (ROI) from the worm intestine were segmented and converted to outlines by a modification of the "Segment mitochondria" pipeline. Emission ratio images of D3cpv were excited at 440nm and captured at 450-490nm and 520-560nm and analyzed in Image Analyst MKII. Images were Wiener filtered and the ratio of the 540nm over the 470nm channel, indicative of cytosolic calcium concertation was calculated, and showed in pseudo-color coding. Emission ratios were determined in ROIs in the posterior intestine by the Plot Ratio function.

Western Blot
Approximately 30 to 50 adult worms were collected in S-basal buffer. Supernatant was removed and nematodes were flash-frozen. Worms pellets were resuspended in 2% SDS sample buffer with 2.5% b-mercaptoethanol and samples were boiled for 10 minutes. Samples were subjected to SDS-PAGE using 4-12% SDS gels (Invitrogen™) and transferred to Immun-Blot PVDF Membrane (BIO-RAD™) using BIO-RAD™ western blot Criterion apparatus. Membranes were blocked with 5% non-fat dry milk blocking solution; concentrations for antibodies were 1:1000 for primary antibodies and 1:2000 for secondary antibodies.

QUANTIFICATION, STATISTICS, AND PREDICTION Statistics
Significance between control and experimental groups was determined by using two-tailed Student's t test. Asterisks denote corresponding statistical significance *p < 0.05; **p < 0.01; ***p < 0.0001. Error bars were generated using the standard error of the mean (SEM), typically from three pooled biological replicates. Graphpad Prism 7™ was used to plot survival curves. Log Rank (Mantel Cox) test in Prism™ was used to determine significance between the control and experimental group.

GFP Quantification
GFP intensity of worms was quantified using Image J 1.52A. The 'integrated density' of GFP expression and length of worms was measured using Image J tools. Integrated Density value was normalized by number of worms and average length of worms. The final value is in arbitrary units.

Mitochondrial Targeting Prediction
Mitochondrial presequences were predicted using the MitoFates tool (Fukasawa et al., 2015).    CV + atp-3 cco-1 + atp-3 nuo-2 + atp-3 (A) Germline mutant fem-1(hc17) is long-lived, which is abolished in daf-16(mu86) mutant background. Worms were developed at 25°C and shifted to 20°C beginning at young adulthood. (B) Germline mutant fem-1(hc17) is long-lived, which is abolished after treatment with daf-16 RNAi. Worms were developed at 25°C and shifted to 20°C beginning at young adulthood. (C) Germline mutants fem-1(hc17) and glp-1(e2141) with a phsp-6::GFP reporter strain background induce robust UPR mt when OSCP/atp-3 RNAi is administered during development. Worms were developed at 25°C and exposed to RNAi beginning from the egg stage and collected for microscopy after approximately 48 hours. (D, E) Germline mutants fem-1(hc17) and glp-1(e2141) with a phsp-4::GFP reporter strain background induce robust UPR ER when exposed to 4ug/ml tunicamycin. Worms were developed at 25°C and exposed to tunicamycin as young adults and collected for microscopy after approximately 24 hours. Tunicamycin was spotted onto seeded NGM plates and water was used as the solvent control.
(F) Germline mutants fem-1(hc17) and glp-1(e2141) with a phsp-16.2::GFP reporter strain background induce a robust heat-shock response when to 33°C for 2 hours. Worms were developed at 25°C and exposed to heat as young adults following immediate collection for microscopy. (G) Germline mutants fem-1(hc17) and glp-1(e2141) with a pges-1::GFP mt reporter strain background exhibit higher intestinal mitochondrial mass after RNAi with pdr-1 and hlh-30. Worms were developed at 25°C and shifted to 20°C and exposed to RNAi as young adults; worms were collected for microscopy after approximately 48 hours.