Maintenance and propagation of a deleterious mitochondrial genome by the mitochondrial UPR

mtDNA mutations and deletions are relatively common within the mtDNA population of aging post-mitotic cells but typically represent a minor percentage of the total mtDNA^6,7. However, if the deleterious mtDNA reaches a high enough percentage relative to wildtype mtDNA, it can become toxic depending on the severity of the mtDNA lesion and lead to disease⁸. As mtDNA encodes 13 (12 in C. elegans) essential components of the respiratory chain and ATP synthase, along with rRNAs and tRNAs required for their synthesis, an increase in deleterious heteroplasmy can perturb multiple cellular activities reliant on OxPhos. However, the mechanism(s) that promote tolerance to large mtDNA deletions or contribute to their propagation are unknown.

Here, we examined the role of the UPR^mt on the maintenance and propagation of a mtDNA with a 3.1 kb deletion (ΔmtDNA) encoding four essential OxPhos genes (Fig. 1a) using a heteroplasmic C. elegans strain⁵. The UPR^mt is regulated by the transcription factor ATFS-1, which is normally efficiently imported into mitochondria and degraded. However, during mitochondrial stress or OxPhos dysfunction, a percentage of ATFS-1 fails to be imported and traffics to the nucleus to activate a broad transcriptional program (over 500 transcripts) that promotes repair and recovery of mitochondrial function^3,4.

To examine the effect of ΔmtDNA on UPR^mt activation, ΔmtDNA was crossed into a transcriptional reporter worm harboring the hsp-6_pr::gfp transgene used to monitor UPR^mt activation⁹. Consistent with previous studies demonstrating that perturbation of nuclear-encoded OxPhos components activated the UPR^mt (Extended Data Fig. 1a)^4,10, ΔmtDNA also modestly activated the UPR^mt, which required atfs-1 (Fig. 1b). Consistent with a previous report⁵, ΔmtDNA made up 60% of all mtDNAs (Fig. 1c). And, ΔmtDNA caused a significant reduction in basal oxygen consumption as well as total respiratory capacity (Fig. 1d), suggesting 60% ΔmtDNA perturbs OxPhos and activates the UPR^mt.

Surprisingly, the development of worms harboring ΔmtDNA was unaffected by atfs-1(RNAi) in striking contrast to worms with hypomorphic mutations in nuclear-encoded OxPhos components (Fig. 1e)^4,11, indicating the UPR^mt is not required for development when the OxPhos defect derives from a mtDNA deletion. Thus, we examined the effect of UPR^mt inhibition on ΔmtDNA levels by impairing several components required for hsp-6_pr::gfp induction^4,12. Strikingly, atfs-1 deletion or knockdown caused a dramatic reduction in ΔmtDNA levels shifting the percentage from 60% to 7% (Fig. 1c and Extended Data Fig. 1b–d), likely explaining the normal growth rate (Fig. 1e). ΔmtDNA quantification in individual wildtype or atfs-1-deletion worms was consistent with that observed in larger worm populations and in some cases ΔmtDNA was reduced below the limit of detection in worms lacking ATFS-1 (Extended Data Fig. 1e). Therefore, ATFS-1 and UPR^mt activation are required to maintain the deleterious mtDNA.

A mechanism by which deleterious mtDNA levels can be altered involves the germline bottleneck where only a small number of mtDNAs are passed maternally to the next generation allowing shifts in heteroplasmy⁸. Because the atfs-1-deletion strain was generated via mating, it is unclear if the shift in ΔmtDNA occurred during germline transmission, somatic cell division and growth, or both. To examine the role of ATFS-1 in ΔmtDNA maintenance specifically in somatic cells, we employed glp-4(bn2) worms that lack germlines when raised at the restrictive temperature. Interestingly, exposure of worms to atfs-1(RNAi) from the L1 stage to adulthood also depleted ΔmtDNA relative to the identical worm population raised on control(RNAi) (Fig. 1f), consistent with a role for the UPR^mt in maintaining ΔmtDNA levels in postmitotic somatic cells.

Because mitochondrial autophagy (mitophagy) could potentially eliminate ΔmtDNAs when atfs-1 is inhibited, we examined the interaction between ΔmtDNA, atfs-1 and known mitophagy components. Mitophagy involves the recognition of defective mitochondria by the kinase PINK-1. Once PINK-1 accumulates, it recruits the ubiquitin ligase Parkin (PDR-1 in C. elegans) to the mitochondrial outer membrane, which directs the damaged organelle to lysosomes for degradation¹³. As described previously, pdr-1-deletion resulted in increased ΔmtDNA (Fig. 2a)¹⁴, consistent with mitophagy targeting defective mitochondria containing relatively high levels of deleterious mtDNAs¹⁵. However, the reduction of ΔmtDNA caused by atfs-1 inhibition was only partially blocked by pink-1;pdr-1-deletion or atg-18-deletion, which impairs general autophagy (Figs. 2a and Extended Data Fig. 2a). Additionally, atfs-1 inhibition did not impair development of pink-1;pdr-1-deficient ΔmtDNA worms (Fig. 2b) despite the increased UPR^mt activation (Fig. 2c). As PDR-1 inhibition did not completely restore ΔmtDNA in atfs-1-deletion worms, ATFS-1 likely promotes ΔmtDNA propagation independent of mitophagy or other Parkin-mediated activities¹⁶.

As modest UPR^mt activation was required to maintain the deletion genome (Fig. 1), we examined the effects of stronger UPR^mt activation on ΔmtDNA expansion or propagation. To further activate the UPR^mt in somatic cells, the mitochondrial protease SPG-7 was impaired by RNAi^3,4 during development in worms lacking germlines. Despite strong UPR^mt activation, ΔmtDNA levels did not increase (Extended Data Fig. 2b and Fig. 2d). However, because PINK-1 and Parkin are also activated by mitochondrial unfolded protein stress¹⁷, we wondered whether PINK-1 and PDR-1 activities limit ΔmtDNA accumulation. Remarkably, ΔmtDNA levels increased three-fold in pink-1;pdr-1-deletion worms raised on spg-7(RNAi) (Fig. 2d) suggesting that UPR^mt activation can promote propagation of deleterious mtDNAs, which is antagonized by mitophagy during strong mitochondrial stress. However, because wildtype mtDNAs were also increased, the ΔmtDNA percentage was unaffected (Fig. 2d). As both wildtype and ΔmtDNA were increased by spg-7(RNAi), ATFS-1 may promote a mitochondrial biogenesis program in response to mitochondrial dysfunction. Interestingly, mitochondrial mass¹⁸ also increased in pink-1;pdr-1-deficient worms treated with spg-7(RNAi) (Fig. 2e) consistent with ATFS-1 mediating a compensatory mitochondrial biogenesis program that maintains ΔmtDNA. Combined, these results suggest that during strong mitochondrial stress, balanced PINK-1/PDR-1 and ATFS-1 activity limits ΔmtDNA accumulation.

To determine if ATFS-1 activation is sufficient to increase mitochondrial biogenesis and ΔmtDNAs independent of mitophagy, we utilized a mutant strain with constitutive UPR^mt activation due to an amino acid substitution within the ATFS-1 mitochondrial targeting sequence¹⁹. Consistent with the UPR^mt promoting a mitochondrial biogenesis program, atfs-1(et18) animals also had a dramatic increase in mitochondria (Fig. 3a). atfs-1(et18) worms also had increased ΔmtDNA and wildtype mtDNAs (Fig. 3b), however ΔmtDNA was further increased than wildtype mtDNAs resulting in an increase of ΔmtDNA from 63 to 73%. Consistent with increased ΔmtDNAs, atfs-1(et18);ΔmtDNA worms developed significantly slower than atfs-1(et18) worms (Fig. 3c) or ΔmtDNA worms⁵. And consistent with further OxPhos impairment, atfs-1(et18);ΔmtDNA worms consumed less oxygen (Fig. 3d), had reduced mitochondrial membrane potential (Extended Data, Fig. 2c) and were sensitive to the OxPhos inhibitor rotenone (Extended Data, Fig. 2d). Combined, these results indicate that UPR^mt activation is detrimental in the presence of ΔmtDNAs.

To investigate potential causes of the OxPhos deficiency and UPR^mt activation in ΔmtDNA worms, we compared expression of a gene within the ΔmtDNA deletion (ND2), that is only expressed by the population of wildtype mtDNAs, to mtDNA-encoded genes located outside of the deletion that are encoded by both wildtype and ΔmtDNAs (ND4, ND6) (Fig. 1a). Surprisingly, ND2 transcripts were not reduced in ΔmtDNA or atfs-1(et18);ΔmtDNA animals despite the absence of the ND2 gene in ΔmtDNA (Fig. 3e). Interestingly, ND4 was expressed significantly higher in both ΔmtDNA and atfs-1(et18);ΔmtDNA relative to wildtype worms while the nuclear-encoded complex I transcript nuo-4 was unaffected (Fig. 3e). These results suggest that mitochondrial dysfunction in ΔmtDNA worms is not due to reduction of transcripts encoded by genes within the mtDNA deletion.

Next, we performed electron microscopy to examine mitochondrial morphology in atfs-1(et18) harboring 73% ΔmtDNA. Impressively, the cristae were largely absent in these mitochondria (Fig. 3f), consistent with severe mitochondrial dysfunction and reduced OxPhos (Fig. 3d). Interestingly, a number of autophagosome-like structures associated with degenerate mitochondria were observed only in atfs-1(et18);ΔmtDNA worms (Fig. 3f). And, consistent with increased mitophagy in atfs-1(et18);ΔmtDNA animals, pdr-1-deletion further increased the percentage of ΔmtDNAs (Fig. 3g). Combined, these results suggest that in the context of deleterious heteroplasmy, increased UPR^mt activation can further perturb mitochondrial function potentially leading to increased mitophagy.

To better understand the ATFS-1-mediated transcriptional program that promotes ΔmtDNA maintenance and propagation, the transcriptomes of wildtype and atfs-1(et18) worms were examined. Activated ATFS-1 induced many transcripts suggestive of mitochondrial biogenesis including the mitochondrial protein import machinery, the cardiolipin synthesis enzyme tafazzin, mitochondrial ribosome and translation factors, prohibitin complex components, mitochondrial chaperones and OxPhos assembly factors (Supplementary Table 1) consistent with ATFS-1 regulating a mitochondrial biogenesis and mitochondrial proteostasis program to recover mitochondrial function. atfs-1(et18) also increased the mtDNA polymerase polg-1, the worm ortholog of the mtDNA-binding protein TFAM (hmg-5), as well as transcripts required for mitochondrial dynamics (Supplementary Table 1, and Extended Data Fig. 3a).

Interestingly, inhibition of mitochondrial fusion or fission by fzo-1(RNAi) or drp-1(RNAi) reduced the ΔmtDNA percentage similar to atfs-1(RNAi) (Fig. 3h) in ΔmtDNA worms suggesting that organelle dynamics stimulated by the UPR^mt promote deleterious mtDNA maintenance. Because fzo-1(RNAi) inhibition had relatively little effect on mitochondrial biogenesis (Fig. 3a), only organelle morphology (Fig. 3a, lower panel), we speculate that organelle mixing mediated by drp-1 and fzo-1, which requires OxPhos function²⁰, limits the enrichment of deleterious mtDNAs in individual organelles and promotes tolerance to deleterious mtDNAs^21,22. Consistent with this idea, development of atfs-1(et18);ΔmtDNA animals was delayed when mitochondrial fusion was inhibited (Extended Data Fig. 3b). Impairment of polg-1 and hmg-5 (TFAM) also reduced the ΔmtDNA percentage in ΔmtDNA worms (Fig. 3h and Extended Data Fig. 3c), suggesting that replication and mtDNA protection is involved in maintaining the deleterious genome. Combined, these data suggest that through multiple outputs, ATFS-1 activation provides favorable conditions for ΔmtDNA proliferation.

These studies suggest an unanticipated consequence of ATFS-1 and UPR^mt activation in the context of deleterious mtDNA heteroplasmy. While the UPR^mt is protective during exposure to mitochondrial toxins²³ or mutations within nuclear-encoded OxPhos genes^4,24, in the context of mtDNA heteroplasmy, UPR^mt activation maintains the deleterious mtDNA. We propose that by inducing a mitochondrial recovery program, UPR^mt activation inadvertently propagates deleterious mtDNAs in an attempt to recover OxPhos activity. These results potentially shed light on the underlying mechanisms that lead to mitochondrial diseases and the enrichment of ΔmtDNAs found in aged cells^7,8. These results also emphasize the importance of UPR^mt regulation and suggest that prolonged UPR^mt activation is potentially harmful²⁵, as ATFS-1 activation creates an environment favorable for ΔmtDNAs.