A tRNA processing enzyme is a key regulator of the mitochondrial unfolded protein response

The mitochondrial unfolded protein response (UPRmt) has emerged as a predominant mechanism that preserves mitochondrial function. Consequently, multiple pathways likely exist to modulate UPRmt. We discovered that the tRNA processing enzyme, homolog of ELAC2 (HOE-1), is key to UPRmt regulation in Caenorhabditis elegans. We find that nuclear HOE-1 is necessary and sufficient to robustly activate UPRmt. We show that HOE-1 acts via transcription factors ATFS-1 and DVE-1 that are crucial for UPRmt. Mechanistically, we show that HOE-1 likely mediates its effects via tRNAs, as blocking tRNA export prevents HOE-1-induced UPRmt. Interestingly, we find that HOE-1 does not act via the integrated stress response, which can be activated by uncharged tRNAs, pointing toward its reliance on a new mechanism. Finally, we show that the subcellular localization of HOE-1 is responsive to mitochondrial stress and is subject to negative regulation via ATFS-1. Together, we have discovered a novel RNA-based cellular pathway that modulates UPRmt.


Introduction
Mitochondria are central to a myriad of cellular processes including energy production, cellular signaling, biogenesis of small molecules, and regulation of cell death via apoptosis (Nunnari and Suomalainen, 2012). Mitochondrial dysfunction can lead to metabolic and neurological disorders, cardiovascular disease, and cancers (Vafai and Mootha, 2012). To maintain proper mitochondrial function cellular mechanisms have evolved that respond to, and mitigate, mitochondrial stress Wang and Chen, 2015;Wrobel et al., 2015;Munkácsy et al., 2016;Tjahjono and Kirienko, 2017;Weidberg and Amon, 2018;Naresh and Haynes, 2019;Fessler et al., 2020;Guo et al., 2020).
One of the predominant mitochondrial stress response mechanisms is the mitochondrial unfolded protein response (UPR mt ). Although first discovered in mammals (Zhao et al., 2002), UPR mt has been best characterized in Caenorhabditis elegans (Naresh and Haynes, 2019). UPR mt is primarily characterized by transcriptional upregulation of genes whose products respond to and ameliorate mitochondrial stress (Yoneda et al., 2004;Nargund et al., 2012).
In C. elegans, activation of UPR mt relies on the transcription factor ATFS-1 that primarily localizes to mitochondria, but under mitochondrial-stress conditions is trafficked to the nucleus where it drives the expression of mitochondrial stress response genes (Haynes et al., 2010;Nargund et al., 2012;Nargund et al., 2015). However, it has become increasingly apparent that UPR mt is under multiple levels of control: Mitochondrial stress in neurons can activate intestinal UPR mt non-cell-autonomously via retromer-dependent Wnt signaling (Durieux et al., 2011;Berendzen et al., 2016;Zhang et al., 2018); overexpression of two conserved histone demethylases are independently sufficient to activate UPR mt (Merkwirth et al., 2016); and ATFS-1 is post-translationally modified to facilitate its stability and subsequent UPR mt activation (Gao et al., 2019). Given mitochondrial integration into many diverse cellular signaling and metabolic pathways, there are likely yet-to-be identified pathways regulating UPR mt .

hoe-1 is required for maximal UPR mt activation
We discovered that RNAi against hoe-1, a gene encoding a 3'-tRNA phosphodiesterase, attenuates hsp-6p::GFP induction-a fluorescence based transcriptional reporter of UPR mt activation (Yoneda et al., 2004). Knockdown of hoe-1 by RNAi is sufficient to attenuate UPR mt reporter activation induced by a loss-of-function mutation in the mitochondrial electron transport chain (ETC) complex I subunit NUO-6 (nuo-6(qm200)) ( Figure 1A and B).
To further interrogate the potential role of hoe-1 in UPR mt regulation, we used CRISPR/Cas9 to generate a hoe-1 null mutant (hoe-1(-/-)) by deleting the open reading frame of hoe-1 (Dokshin et al., 2018). The hoe-1 null mutants do not develop past late larval stage 3, thus the allele is maintained over a balancer chromosome, tmC25 (Dejima et al., 2018). UPR mt induced by the knockdown of both the mitochondrial protease, spg-7, and ETC complex IV subunit, cco-1, is robustly attenuated in hoe-1 null animals ( Figure 1C-E). Furthermore, UPR mt induced by nuo-6(qm200) is attenuated in hoe-1 null animals similarly to what is seen in nuo-6(qm200) animals on hoe-1 RNAi ( Figure 1F and G). Taken together, these findings suggest that HOE-1 is generally required for maximal UPR mt activation.

HOE-1 is dual-targeted to nuclei and mitochondria
To better understand the role of HOE-1 in UPR mt regulation, we sought to identify where HOE-1 functions in the cell. HOE-1 is predicted to localize to both nuclei and mitochondria and this dual-localization has been shown for HOE-1 orthologs in Drosophila, mice, and human cell lines (Dubrovsky et al., 2004;Brzezniak et al., 2011;Rossmanith, 2011;Siira et al., 2018). To determine the subcellular localization of HOE-1 in C. elegans, we C-terminally tagged HOE-1 with GFP at its endogenous locus (HOE-1::GFP). Both hoe-1::GFP homozygous and hoe-1::GFP/hoe-1(-/-) trans-heterozygous animals grow and develop indistinguishably from wildtype animals suggesting that GFP-tagging HOE-1 does not compromise its essential functions (Figure 2-figure supplement 1A). We found that HOE-1 localizes to both mitochondria and nuclei ( Figure 2A).

Mitochondrial HOE-1 is not required for UPR mt activation
Given the dual-localization of HOE-1, we questioned whether it is mitochondrial or nuclear HOE-1 that is required for UPR mt activation. To address this question, we created mitochondrial and nuclear compartment-specific loss-of-function mutants of HOE-1 ( Figure 2B). hoe-1 contains two functional start codons. Translation beginning from the first start codon (encoding methionine 1 (M1)) produces a protein containing a mitochondrial targeting sequence (MTS). Translation beginning from the second start codon (encoding methionine 74 (M74)), which is 3' to the MTS, produces a nuclear specific protein. This feature is conserved in human ELAC2 and it has been shown that mutating M1 to an alanine produces a mitochondrial-specific knockout (Brzezniak et al., 2011). Thus, we used the same approach to create a mitochondrial-specific knockout of C. elegans HOE-1 (hoe-1(ΔMTS)). This mutation is sufficient to strongly attenuate mitochondrial targeting without impacting nuclear localization ( Figure 2-figure supplement 2A). UPR mt reporter activation by spg-7 and cco-1 RNAi is not attenuated in hoe-1(ΔMTS) animals ( Figure 2C and D and Figure 2-figure supplement 3A and B). In fact, UPR mt reporter activation is slightly elevated in hoe-1(ΔMTS) animals relative to wildtype. These data suggest that mitochondrial HOE-1 is not required for UPR mt activation.
Nuclear HOE-1 is required for UPR mt activation HOE-1 is predicted to contain two nuclear localization signals (NLS). Given that hoe-1 null mutant animals are developmentally arrested and hoe-1(ΔMTS) animals are superficially wildtype we reasoned that completely ablating nuclear localization of HOE-1 may result in recapitulation of the null phenotype. In effort to disentangle the developmental effects from the effect on UPR mt we ablated only one of the nuclear localization signals of HOE-1. To compromise nuclear localization, we mutated the Figure 2. Nuclear HOE-1 is required for maximal UPR mt activation. (A) Fluorescence images of a terminal intestinal cell in a wildtype animal expressing HOE-1::GFP (green) stained with TMRE (magenta) to visualize mitochondria. GFP and TMRE co-localization shown in white in merged image. Arrow indicates nuclei. Scale bar 20 μm. Representative line segment analysis of individual mitochondrion. (B) Schematic of HOE-1 protein showing the mitochondrial targeting sequence (MTS) and nuclear localization signals (NLS). ΔMTS allele created by replacing START codon with an alanine (M1A). Transcription begins at M74 for nuclear localized HOE-1. ΔNLS allele created by compromising the most N-terminal NLS ( 636 KRPR > AAPA). (C) Fluorescence images of UPR mt reporter (hsp-6p::GFP) in L4 wildtype and hoe-1(ΔMTS) animals on control and spg-7 RNAi. Scale bar 200 μm. (D) Fluorescence intensity quantification of hsp-6p::GFP in individual L4 wildtype and hoe-1(ΔMTS) animals on control and spg-7 RNAi (n = 15,20,17, and 19 respectively, mean and SD shown, ordinary two-way ANOVA with Tukey's multiple comparisons test). (E) Fluorescence images of UPR mt reporter (hsp-6p::GFP) in L4 wildtype and hoe-1(ΔNLS) animals on control and spg-7 RNAi. Scale bar 200 μm. (F) Fluorescence intensity quantification of hsp-6p::GFP in individual L4 wildtype and hoe-1(ΔNLS) animals on control and spg-7 RNAi (n = 15 for each condition, mean and SD shown, ordinary two-way ANOVA with Tukey's multiple comparisons test). (G) Fluorescence images of UPR mt reporter in L4 nuo-6(qm200) animals in wildtype and hoe-1(ΔNLS) backgrounds. Scale bar 200 μm. (H) Fluorescence intensity of hsp-6p::GFP in individual L4 nuo-6(qm200) animals in wildtype and hoe-1(ΔNLS) backgrounds (n = 30 for each condition, mean and SD shown, unpaired t-test). (I) mRNA transcript quantification of hsp-6 in L4 wildtype and hoe-1(ΔNLS) animals on control and spg-7 RNAi normalized to ama-1 (n = 4 for each condition, mean and SD shown, ordinary two-way ANOVA with Tukey's multiple comparisons test).
The online version of this article includes the following figure supplement(s) for figure 2:     positively charged residues of the most N-terminal NLS to alanines (hoe-1(ΔNLS)). These mutations are sufficient to strongly attenuate, but not completely ablate, HOE-1 nuclear localization whilst still allowing animals to develop to adulthood (Figure 2-figure supplement 4A-C).
In contrast to loss of mitochondrial HOE-1, loss of nuclear HOE-1 robustly attenuates UPR mt reporter activation induced by spg-7 RNAi ( Figure 2E and F) and attenuates UPR mt reporter activation induced by nuo-6(qm200) ( Figure 2G and H). Furthermore, loss of nuclear HOE-1 attenuates the transcriptional upregulation of UPR mt target genes hsp-6 ( Figure 2I) and cyp-14A1.4 (Figure 2-figure supplement 5A) under conditions of mitochondrial stress. Together these data suggest that HOE-1 is required in the nucleus to facilitate UPR mt activation.
Compromising HOE-1 nuclear export is sufficient to activate UPR mt Like many nuclear localized proteins (la Cour et al., 2004), HOE-1 has both nuclear localization signals and a nuclear export signal (NES). Given that loss of nuclear HOE-1 results in UPR mt attenuation we questioned if compromising HOE-1 nuclear export, by ablating the NES of HOE-1, is sufficient to activate UPR mt . We created a HOE-1 NES knockout mutant (hoe-1(ΔNES)) by replacing the strong hydrophobic residues of the predicted NES with alanines ( Figure 3-figure supplement 1A). hoe-1(ΔNES) animals are superficially wildtype in their development but are sterile. Thus, the allele is balanced with tmC25. Homozygous hoe-1(ΔNES) animals have elevated nuclear HOE-1 accumulation relative to wildtype ( Strikingly, the UPR mt reporter hsp-6p::GFP is robustly activated in adult hoe-1(ΔNES) animals similarly to that seen in mitochondrial stressor nuo-6(qm200) and constitutive UPR mt activation in atfs-1 gain-of-function (atfs-1(et15)) mutant animals ( Figure 3A and B). hoe-1(ΔNES) also mildly induces the less sensitive UPR mt reporter hsp-60p::GFP ( UPR mt activation is characterized by the transcriptional upregulation of a suite of mitochondrial stress response genes that encode chaperone proteins, proteases, and detoxification enzymes that function to restore mitochondrial homeostasis . To interrogate the extent of UPR mt induction in hoe-1(ΔNES) animals, we measured transcript levels of a diverse set of UPR mt associated genes. We found that the UPR mt genes encoding a chaperone protein (hsp-6), stress response involved C-type lectin (clec-47), and P450 enzyme (cyp-14A4.1) are all upregulated in hoe-1(ΔNES) animals ( Figure 3C, D and E). These data support hoe-1(ΔNES) being sufficient to activate the UPR mt transcriptional response.
The online version of this article includes the following figure supplement(s) for figure 3:     HOE-1 is selectively degraded in the intestine (Figure 3-figure supplement 5A and B). This data strongly suggests that compromised nuclear export of HOE-1 activates UPR mt cell-autonomously in the intestine.
Compromising HOE-1 nuclear export specifically activates UPR mt Changes in protein synthesis rates and associated protein folding capacity can broadly activate cellular stress response mechanisms (Wang and Kaufman, 2016;Das et al., 2017;Boos et al., 2019). Given the role of hoe-1 in tRNA maturation, we questioned if the robust upregulation of UPR mt in hoe-1(ΔNES) animals may be the result of compromised cellular proteostasis in general rather than specific activation of UPR mt . One stress response mechanism that is sensitive to global proteotoxic stress is the endoplasmic reticulum unfolded protein response (UPR ER ) (Preissler and Ron, 2019). We find that the UPR ER reporter hsp-4p::GFP is not activated in hoe-1(ΔNES) animals ( Figure 3H and I), suggesting that hoe-1(ΔNES) does not cause ER stress nor cellular proteotoxic stress. Additionally, a basal reporter of GFP that has been used to proxy general protein expression (Gitschlag et al., 2016), ges-1p::GFPcyto, is only mildly upregulated in hoe-1(ΔNES) animals relative to wildtype ( Figure 3J and K). Together these findings support that impaired nuclear export of HOE-1 specifically activates UPR mt . Compromising HOE-1 nuclear export reduces mitochondrial membrane potential UPR mt is known to be triggered when mitochondrial membrane potential is compromised (Rolland et al., 2019;Shpilka et al., 2021). Thus, we assessed mitochondrial membrane potential, using TMRE staining, in adult hoe-1(ΔNES) animals where UPR mt is robustly activated. Consistent with UPR mt activation, we found that mitochondrial membrane potential is severely depleted in adult hoe-1(ΔNES) animals ( Figure 4A and B). However, hoe-1(ΔNLS) animals also exhibit compromised mitochondrial membrane potential without UPR mt activation suggesting that decreased membrane potential does not guarantee UPR mt induction ( Figure 4A and B). Compromised mitochondrial membrane potential can be both a cause and consequence of UPR mt activation (Rolland et al., 2019;Shpilka et al., 2021). Thus, we assessed whether or not compromised membrane potential in hoe-1(ΔNES) animals is atfs-1-dependent. Mitochondrial membrane potential is not rescued in hoe-1(ΔNES) animals on atfs-1 RNAi ( Figure 4C and D) suggesting that reduced mitochondrial membrane potential in hoe-1(ΔNES) animals is not a result of UPR mt activation. Taken together, these data show that compromised nuclear export of HOE-1 results in depletion of mitochondrial membrane potential. Furthermore, this depletion in membrane potential correlates with UPR mt activation, consistent with the possibility that hoe-1(ΔNES) activates UPR mt via depletion of mitochondrial membrane potential.
Given that HOE-1 catalytic activity is required for UPR mt , we further interrogated the potential role of tRNA processing as a mechanism by which HOE-1 may modulate UPR mt induction. Production of mature tRNAs begins with transcription of tRNA gene loci by RNA polymerase III followed by sequential cleavage of 5'-leader and 3'-trailer sequences from immature tRNA transcripts by RNase P and HOE-1, respectively. Following cleavage of 3'-trailer sequences, tRNAs can be transported to the cytosol by tRNA exportin (Hopper and Nostramo, 2019).
Given that HOE-1 nuclear levels are elevated in hoe-1(ΔNES) animals, we reasoned that 3'-tRNA processing should be elevated due to increased nuclear activity of HOE-1. Thus, we questioned if UPR mt induction in hoe-1(ΔNES) animals is a result of elevated 3'-tRNA processing. First, we knockeddown RNA pol III subunit rpc-1 to attenuate the production of total RNA pol III-dependent transcripts in hoe-1(ΔNES) animals. If hoe-1(ΔNES)-induced UPR mt is due to elevated processing of tRNAs we hypothesized that restriction of nascent tRNA production should attenuate UPR mt activation. Indeed, we found that rpc-1 RNAi robustly attenuates hoe-1(ΔNES)-induced UPR mt (Figure 6-figure supplement 2A and B). Interestingly, rpc-1 RNAi has little impact on mitochondrial stress-induced UPR mt (nuo-6(qm200)) ( Figure 6-figure supplement 2C and D). These data show that rpc-1 is required for hoe-1(ΔNES)-induced UPR mt and support our hypothesis that increased 3'-tRNA processing by HOE-1 activates UPR mt .
For the majority of tRNAs 5'-end processing by the RNase P complex is a prerequisite for 3'-end processing by HOE-1 (Frendewey et al., 1985;Yoo and Wolin, 1997). Thus, if increased 3'-tRNA end processing is responsible for UPR mt activation, compromising 5'-end processing by RNAi against RNAse P should attenuate hoe-1(ΔNES)-induced UPR mt . RNAi against a subunit of the RNase P complex, popl-1, attenuates UPR mt induction in hoe-1(ΔNES) animals ( Figure 6A and B). popl-1 RNAi also attenuates both UPR mt induced by nuo-6(qm200) ( Figure 6C and D) as well as basal induction of ges-1p::GFPcyto ( Figure 6E and F), albeit to a lesser extent than the attenuation seen in hoe-1(ΔNES) animals. These data suggest that popl-1 RNAi may have a broad impact on protein expression but supports that elevated 3'-tRNA processing in hoe-1(ΔNES) animals is responsible for UPR mt activation given that popl-1 RNAi strongly attenuates hoe-1(ΔNES)-induced UPR mt .
While 5'-and 3'-tRNA processing are the only steps known to be required for tRNA export from the nucleus, there are other downstream tRNA maturation processes that occur (Hopper and Nostramo, 2019). Some nascent tRNAs include introns that need to be removed and then ligated by a tRNA ligase (Englert and Beier, 2005;Popow et al., 2012). For tRNAs to be charged with corresponding amino acids, nascent tRNAs must contain a CCA sequence as part of the 3' acceptor stem. This can be achieved by a CCA-adding tRNA nucleotidyl transferase (Hou, 2010). Knockdown of both tRNA ligase, rtcb-1, and tRNA nucleotidyl transferase, hpo-31 mildly attenuate hoe-1(ΔNES)-induced UPR mt (Figure 6-figure supplement 3A-C). However, rtcb-1 RNAi also mildly attenuates nuo-6(qm200)-induced UPR mt (Figure 6-figure supplement 3D and E). Knockdown of hpo-31 severely compromised growth of nuo-6(qm200) animals and thus the impact on UPR mt could not accurately be intensity from day 1 adult wildtype and hoe-1(ΔNES) animals normalized to total protein (n = 4 for each condition, mean and SD shown, unpaired t-test).
The online version of this article includes the following source data and figure supplement(s) for figure 5: Source data 1. Blots for wildtype and hoe-1(ΔNES) animals with DVE-1::GFP ( Figure 4G and H).   Taken together, these data suggest that UPR mt induction by nuclear export deficient HOE-1 is the result of increased 3'-tRNA processing and that these tRNA species are required in the cytosol to trigger UPR mt .
Compromised HOE-1 nuclear export does not activate UPR mt via GCN2 or eIF2α Alteration to tRNA processing can activate cellular signaling pathways (Raina and Ibba, 2014). One such pathway is the integrated stress response (ISR) in which uncharged tRNAs activate the kinase GCN2 which, in turn, phosphorylates the eukaryotic translation initiation factor, eIF2α. This inhibitory phosphorylation of eIF2α leads to upregulation of a select number of proteins including the transcription factor ATF4 (Pakos-Zebrucka et al., 2016;Costa-Mattioli and Walter, 2020). Interestingly, ATF4 intensity quantification of hsp-6p::GFP in individual day 2 adult wildtype and hoe-1(ΔNES) animals on control and xpo-3 RNAi (n = 24 for each condition, mean and SD shown, ordinary two-way ANOVA with Tukey's multiple comparisons test). (I) Fluorescence images of UPR mt reporter (hsp-6p::GFP) activation in day 2 adult wildtype and nuo-6(qm200) animals on control and xpo-3 RNAi. Scale bar 200 μm. (J) Fluorescence intensity quantification of hsp-6p::GFP in individual day 2 adult wildtype and nuo-6(qm200) animals on control and xpo-3 RNAi (n = 24 for each condition, mean and SD shown, ordinary two-way ANOVA with Tukey's multiple comparisons test). (K) Fluorescence images of intestinal-specific basal protein reporter (ges-1p::GFPcyto) activation in day 2 adult wildtype animals on control and xpo-3 RNAi. Scale bar 200 μm. (L) Fluorescence intensity quantification of ges-1p::GFPcyto in individual day 2 adult wildtype animals on control and xpo-3 RNAi (n = 24 for each condition, mean and SD shown, unpaired t-test).
Nuclear HOE-1 is dynamically responsive to mitochondrial stress and negatively regulated by ATFS-1 To better understand the potential physiological implications of HOE-1 in UPR mt , we assessed hoe-1 expression and subcellular dynamics of HOE-1 during mitochondrial stress. It is predicted that two major transcripts are produced from the hoe-1 gene locus: one that contains an MTS and one that does not, which are translated into mitochondrial-and nuclear-targeted HOE-1, respectively. However, it has been shown in other systems that hoe-1 orthologs produce a single transcript that encodes both a mitochondrial targeted and nuclear targeted HOE-1 isoform via alternative translation initiation (Rossmanith, 2011). Thus, we first sought to determine which mechanism is used for hoe-1 expression. To do so, we designed two sets of primers complementary to hoe-1 mRNA one of which amplifies only mRNA that includes the MTS and the other which amplifies all hoe-1 mRNA (spans a sequence that is included in all predicted HOE-1 isoforms). If there are two independent hoe-1 transcripts, we expected there to be higher levels of hoe-1 mRNA measured by the primer pair for total transcripts than for the mitochondrial specific pair. However, we found that both primer pairs measured similar levels of hoe-1 mRNA (Figure 8-figure supplement 1A) suggesting that, like in other systems, there is a single hoe-1 transcript. Next, we assessed hoe-1 mRNA levels in non-stress versus mitochondrial stress conditions. We found, using both primer pairs, that hoe-1 mRNA levels are modestly elevated in nuo-6(qm200) animals relative to wildtype (Figure 8-figure supplement  1B and C) suggesting that hoe-1 may be transcriptionally upregulated under conditions of mitochondrial stress.
Next, we assessed the subcellular dynamics of HOE-1 in response to mitochondrial stress. We found that HOE-1::GFP nuclear levels are markedly diminished under mitochondrial stress induced by nuo-6(qm200), cco-1 RNAi, and spg-7 RNAi ( Figure 8A and B and Figure 8-figure supplement 2A and B). This observation was unexpected given that hoe-1 transcript levels are elevated during mitochondrial stress and it runs contrary to the fact that compromising HOE-1 nuclear export is sufficient to induce UPR mt ( Figure 3A and B). A common feature of signaling pathways is negative regulation. Thus, we questioned if reduced nuclear HOE-1 is a result of negative feedback rather than a direct result of mitochondrial stress. Given that mitochondrial stress activates UPR mt , we assessed HOE-1::GFP status in a mitochondrial stress background wherein atfs-1 is knocked down by RNAi. HOE-1 levels are significantly upregulated in nuclei of nuo-6(qm200) animals on atfs-1 RNAi relative to nuo-6(qm200) animals on control RNAi, as well as both wildtype animals on control and atfs-1 RNAi ( Figure 8A and B and Figure 8-figure supplement 3A-C). Moreover, total cellular HOE-1 levels are elevated under mitochondrial stress in an atfs-1 RNAi background ( Figure 8C and D and Figure  8-source data 1A-E and Figure 8-source data 2A-E). Additionally, mitochondrial HOE-1 levels are elevated under mitochondrial stress conditions irrespective of RNAi treatment (Figure 8-figure  supplement 3D). Together these data suggest that HOE-1 is upregulated and accumulates in nuclei upon mitochondrial stress. Then, nuclear HOE-1 is negatively regulated by ATFS-1 once UPR mt is activated.
To further test if nuclear HOE-1 is negatively regulated by UPR mt activation rather than by mitochondrial stress, we assessed HOE-1 localization in ATFS-1 gain-of-function animals (atfs-1(et15)). atfs-1(et15) constitutively activates UPR mt in the absence of mitochondrial stress (Rauthan et al., 2013). Thus, we asked if atfs-1(et15) is sufficient to reduce nuclear HOE-1 levels. Indeed, nuclear HOE-1 levels are markedly reduced in atfs-1(et15) animals relative to wildtype ( Figure 8E and F and Figure 8-figure supplement 4A-C) while total and mitochondrial HOE-1 protein levels are largely unperturbed ( Figure 8G and H, Figure 8-figure supplement 4D, Figure 8-source data 3A-E and Figure 8-source data 4A-E). These data further support that UPR mt activation negatively regulates nuclear HOE-1. (B) Fluorescence intensity quantification of intestinal nuclei relative to extranuclear signal in day 1 adult wildtype and nuo-6(qm200) animals on control and atfs-1 RNAi (n = 40 for each condition, mean and SD shown, ordinary two-way ANOVA with Tukey's multiple comparisons test). (C) Western blot for HOE-1::GFP and actin from day 1 adult wildtype and nuo-6(qm200) animals on control and atfs-1 RNAi. (D) Quantification of HOE-1::GFP western blot band intensity from day 1 adult wildtype and nuo-6(qm200) animals on control and atfs-1 RNAi normalized to total protein (n = 4 for each condition, mean and SD shown, ordinary two-way ANOVA with Tukey's multiple comparisons test). (E) Fluorescence images of HOE-1::GFP in day 1 adult wildtype and atfs-1(et15) animals. Scale bar 200 μm. (F) Fluorescence intensity quantification of intestinal nuclei relative to extranuclear signal in day 1 adult wildtype and atfs-1(et15) animals (n = 40 for each condition, mean and SD shown, unpaired t-test). (G) Western blot for HOE-1::GFP and actin from day 1 adult wildtype and atfs-1(et15) animals. (H) Quantification of HOE-1::GFP western blot band intensity from day 1 adult wildtype and atfs-1(et15) animals normalized to total protein (n = 4 for each condition, mean and SD shown, unpaired t-test). (I) Mitochondrial stress triggers activation of HOE-1 resulting in altered RNA processing that facilitates UPR mt via ATFS-1. Activation of UPR mt negatively regulates HOE-1.
The online version of this article includes the following source data and figure supplement(s) for figure 8: Source data 1. Blots for wildtype and nuo-6(qm200) animals on control and atfs-1 RNAi ( Figure 8C).

Discussion
Regulation of UPR mt is not completely understood and elucidating this mechanism has broad implications for understanding cellular response to mitochondrial dysfunction. Here, we describe a novel mechanism by which mitochondrial stress is transduced to activate UPR mt and how that response is regulated through a feedback mechanism ( Figure 8I).
We show that loss of hoe-1 results in varied attenuation of UPR mt depending on how UPR mt is activated. UPR mt induction by RNAi (cco-1 and spg-7) is robustly attenuated by loss of hoe-1 while nuo-6(qm200)-induced UPR mt is only modestly attenuated. RNAi by feeding works well in all tissues except neurons (Timmons et al., 2001;Kamath et al., 2003). Importantly, UPR mt can be activated non-cell autonomously in the intestine by mitochondrial stress in neurons (Durieux et al., 2011;Berendzen et al., 2016;Zhang et al., 2018). UPR mt induced cell-autonomously in the intestine by RNAi may be hoe-1 dependent while neuron-to-intestine UPR mt induction may work primarily in a hoe-1-independent manner. Consistent with this, increased nuclear accumulation of HOE-1 only activates UPR mt in the intestine. These results further exemplify the complexity of UPR mt signaling.
UPR mt is generally triggered via compromised mitochondrial membrane potential which facilitates the nuclear accumulation of ATFS-1 (Rolland et al., 2019;Shpilka et al., 2021). We find that UPR mt activation via hoe-1(ΔNES) correlates with a decrease in mitochondrial membrane potential providing a potential trigger for UPR mt induction. Furthermore, we show that the UPR mt transcription factors ATFS-1 and DVE-1 have increased nuclear localization in hoe-1(ΔNES) animals, thus likely facilitating the robust UPR mt activation.
HOE-1 functions in tRNA processing (Nashimoto et al., 1999;Mayer et al., 2000;Schiffer et al., 2002;Takaku et al., 2003;Dubrovsky et al., 2004;Brzezniak et al., 2011;Sanchez et al., 2011;Siira et al., 2018). Here, we show that increased 3'-tRNA processing by HOE-1 is likely responsible for UPR mt activation. Restricting HOE-1-dependent 3'-tRNA trailer sequence cleavage indirectly by RNAi against RNA polymerase III subunit, rpc-1, and RNase P subunit, popl-1, strongly attenuate hoe-1(ΔNES)-induced UPR mt . Moreover, these RNA species must be required in the cytosol to activate UPR mt as RNAi against tRNA exportin xpo-3 is sufficient to robustly attenuate hoe-1(ΔNES)-induced UPR mt . Our findings herein are the first reported connection between altered tRNA processing and UPR mt in C. elegans. Given the general requirement for tRNAs in protein translation on the one hand, and the mitochondria-specific nature of UPR mt on the other, our findings of a connection between the two are intriguing. However, besides performing their core housekeeping function in protein translation, tRNAs have also emerged as small RNAs with important regulatory roles inside cells (Avcilar-Kucukgoze and Kashina, 2020). Perhaps, the most well-characterized regulatory role for tRNAs is in the activation of the integrated stress response (ISR). In ISR, uncharged tRNAs activate the eIF2α kinase, GCN2, resulting in the upregulation of ATFS-1 orthologs ATF4 and ATF5 (Pakos-Zebrucka et al., 2016;Costa-Mattioli and Walter, 2020). However, we show that gcn-2 and eIF2α are not required for hoe-1(ΔNES)-induced UPR mt activation suggesting that a different mechanism is responsible. The lack of involvement of ISR in HOE-1's role in UPR mt is not too surprising as there may be a greater pool of fully mature tRNAs in the cytosol in hoe-1(ΔNES) animals due to increased 3'-end processing of tRNAs above wildtype levels. This would result in an excess of charged tRNAs in the cytosol, the opposite of what is required to trigger GCN2-dependent ISR. Instead, we can speculate on several additional possibilities for the consequences of increased levels of charged tRNAs that can explain the role of HOE-1 in UPR mt regulation. For example, the use of amino acids to charge excess tRNAs in hoe-1(ΔNES) animals may limit the pool of free amino acids available for mitochondrial import, thus affecting translation of proteins encoded by the mitochondrial genome. This may result in stoichiometric imbalance between nuclear and mitochondrial-encoded components of the electron transport chain, which is known to compromise mitochondrial membrane potential and trigger UPR mt . Alternatively, mito-nuclear imbalance in hoe-1(ΔNES) animals may result from excessive translation of nuclear-encoded mitochondrial proteins due to increased abundance of available charged tRNAs in the cytosol. In yet another scenario, UPR mt may not be the consequence of a global increase in the levels of all cytosolic tRNAs but rather, may be due to changes in the levels of specific tRNAs that preferentially impact translation of genes enriched for the corresponding codons. Such selective upregulation of tRNAs has been shown previously to have specific cellular consequences (Gingold et al., 2014;Goodarzi et al., 2016). Finally, it is possible that a tRNA-like RNA or other small RNA species such as tRNA fragments are responsible for UPR mt induction in hoe-1(ΔNES) animals (Kruszka et al., 2003;Lee et al., 2009;Bogerd et al., 2010;Siira et al., 2018). However, if this is the case, our data argue that such an RNA species would need to be transported to the cytosol by tRNA exportin. Non-tRNA transport by an ortholog of xpo-3 has not yet been reported (Hopper and Nostramo, 2019).
We show that nuclear HOE-1 is dynamically regulated by mitochondrial stress. In the presence of stress, nuclear HOE-1 levels are depleted. However, this is UPR mt dependent as HOE-1 nuclear levels under mitochondrial stress are elevated above wild-type levels when UPR mt is blocked by atfs-1 RNAi. These data, paired with the fact that compromising HOE-1 nuclear export triggers UPR mt , lead us to hypothesize that upon mitochondrial stress, nuclear HOE-1 levels are elevated. This upregulation of nuclear HOE-1 elevates 3'-tRNA processing thereby triggering a signaling cascade that results in elevated nuclear ATFS-1 and DVE-1 and subsequent UPR mt induction. Activated UPR mt then negatively regulates HOE-1 nuclear levels thus providing a feedback mechanism to tightly control mitochondrial stress response. UPR mt -negative regulation of HOE-1 is further supported by our data showing that constitutive activation of UPR mt by atfs-1(et15) is sufficient to reduce nuclear HOE-1 levels in the absence of mitochondrial stress. How it is that mitochondrial stress activates HOE-1 is still unknown. Multiple mitochondrial derived small molecules have been reported to communicate mitochondrial status including reactive oxygen species (ROS), NAD+, and acetyl-CoA Mouchiroud et al., 2013;Ramachandran et al., 2019;Tjahjono et al., 2020;Zhu et al., 2020) We look forward to further investigating whether these, or other molecules, are involved in HOE-1 regulation.
In humans, mutations in the ortholog of HOE-1, ELAC2, are associated with both hypertrophic cardiomyopathy (Haack et al., 2013;Shinwari et al., 2017;Saoura et al., 2019) and prostate cancer (Tavtigian et al., 2001;Korver et al., 2003;Noda et al., 2006). Historically, it has been suggested that mutations in ELAC2 cause disease because of a loss of mature tRNA production. Our works suggests an intriguing alternative whereby ELAC2 mutations lead to altered tRNA processing that triggers aberrant stress response signaling resulting in disease state. Our system provides a convenient opportunity to interrogate these disease causing variants.
Taken together, our findings provide a novel mechanism-involving the tRNA processing enzyme HOE-1-by which mitochondrial stress is transduced to activate UPR mt thus providing important insight into the regulation of mitochondrial stress response.

Worm maintenance
Worms were grown on nematode growth media (NGM) seeded with OP50 E. coli bacteria and maintained at 20 °C.

Mutants and transgenic lines
A complete list of C. elegans strains used can be found in Supplementary file 1. All new mutant and transgenic strains generated via CRISPR/Cas9 for this study were confirmed by Sanger sequencing.

Genetic crosses
Strains resulting from genetic crosses were generated by crossing ~20 heterozygous males of a given strain to 5-8 L4 hermaphrodites of another strain (heterozygous males were generated by first crossing L4 hermaphrodites of that strain to N2 males). F1, L4 hermaphrodites were then cloned out and allowed to have self-progeny. F2 progeny were cloned out and once they had progeny were genotyped or screened (if fluorescent marker) for presence of alleles of interest. All genotyping primers were purchased from IDT and can be found in Supplementary file 2.

Fluorescence microscopy
All whole animal imaging was done using Zeiss Axio Zoom V16 stereo zoom microscope. For all whole animal imaging, worms were immobilized on 2% agar pads on microscope slides in ~1 μl of 100 mM levamisole (ThermoFisher #AC187870100) and then coverslip applied.

Fluorescence image analysis
For whole animal fluorescence intensity quantification, total pixels (determined by tracing individual animals and summing the total number of pixels within the bounds of the trace) and pixel fluorescence intensity (pixel fluorescence intensity on 1-255 scale) were quantified using imageJ and mean fluorescence intensity for each worm was calculated (sum total of fluorescence intensity divided by total number of pixels within bounds of the trace). For DVE-1::GFP image analysis ( Figure 5E&F), brightness threshold was set to 25 in imageJ and then the number of gut cell nuclei that were saturated at this threshold were counted. For Figure 8A&B and Figure 8E&F, and Figure 8-figure supplement 2A&B, mean fluorescence intensity was calculated within the bounds of gut cell nuclei and outside of the bounds of gut cell nuclei and then graphed as the ratio fluorescence intensity of nuclear to extranuclear signal.

Quantification of gene expression
cDNA was synthesized using Maxima H Minus First Strand cDNA Synthesis Kit, with dsDNase (Ther-moFisher #K1682) according to manufacturer's directions. Lysates for cDNA synthesis were made by transferring 10, day 2 adult worms to 10 μl of lysis buffer supplemented with 20 mg/ml proteinase K and incubating at 65 °C for 10 min, 85 °C for 1 min and 4 °C for 2 min. Quantification of gene expression was performed using droplet digital PCR (ddPCR) with Bio-Rad QX200 ddPCR EvaGreen Supermix (Bio-Rad #1864034). Primers used for ddPCR can be found in Supplementary file 2.

TMRE staining
A total of 500 μl of 1 mM TMRE (ThermoFisher #T669) solution in M9 buffer (prepared from a stock TMRE solution of 0.5 M in DMSO) was supplemented on top of standard NGM plates pre-seeded with 200 ul lawn of OP50 and allowed to dry overnight in the dark. The following day young L4 animals were transferred to TMRE plates and incubated on TMRE for 16 hr. After 16 hr, animals were transferred from TMRE plates to seeded standard NGM plates for 1 hr to remove any non-specific TMRE signal from cuticle and intestinal lumen. Animals were then imaged via confocal microscopy as described below.

Confocal fluorescence imaging
Worms were grown at 20 °C and age-synchronized by timed egg-lays on NGM plates seeded with OP50 or HT115 bacteria for RNAi experiments. Before imaging, worms were immobilized with 3 μl 0.05 µm Polybead microsphere suspension (Polysciences) on a 10% agarose pad with a coverslip (1). Images were taken in the mid-or posterior intestine using a Nikon Ti2 with CSU-W1 spinning disk and Plan-Apochromat 100 X/1.49 NA objective. HOE-1::GFP was imaged by 488 nm laser excitation and ET525/36 m emission filter. 2 X integration was applied (Nikon Elements) to increase signal strength. TMRE and ATFS-1::mCherry were imaged with 561 nm laser excitation and ET605/52 M emission filter.
Image processing and analysis was performed with Nikon Elements software. Raw images were subjected to deconvolution and rolling ball background subtraction. Mitochondrial networks were segmented using the TMRE signal after excluding dye aggregates via Bright Spot Detection. To objectively set threshold parameters across groups with different TMRE intensity levels, the low threshold for segmentation was calculated based on a linear correlation with mean TMRE intensity within each group, y = 0.6411*x + 89.71 (x = mean TMRE intensity and constants derived from an initial manual validation). Regions of interest (ROIs) were manually drawn to encompass a single intestinal cell, and nuclei were identified and segmented manually using brightfield images. Mean intensities were measured within the resulting masks.
To detect localization of HOE-1::GFP in mitochondria, images of TMRE-stained intestinal cells of control and ΔMTS worms were collected and blinded. Mitochondria were segmented by TMRE signals as above. For each cell, one representative line scan was drawn manually across the mitochondrial short axis.

Western blot
Fifty adult worms were transferred into a tube containing 20 μl of M9 Buffer. Then, 20 μl of 2 x Laemmli Buffer (BioRad #161-0737) supplemented with 2-mercaptoethanol (i.e. βME) was added to worm suspension and gently pipetted up and down 5 times to mix. Worms were lysed at 95 °C for 10 min in thermocycler followed by ramp down to room temperature (25 °C). Lysates were then pipetted up and down 10 times to complete disrupt and homogenize suspension. Samples were briefly centrifuged to pellet any worm debris. 20 μl of lysate supernatant was loaded onto precast Mini-PROTEAN TGX Stain-Free Gel (BioRad #4568045). Gel was run for 30 min at 100 V and then an additional 40-45 min at 130 V in 1 x Tris/Glycine/SDS Running Buffer (BioRad #1610732). Following electrophoresis gel was activated and imaged for total protein. Gel was equilibrated in Trans-Blot Turbo Transfer Buffer (BioRad #10026938) and transferred to activated and equilibrated Trans-Blot Turbo LF PVDF Membrane (BioRad #10026934) for 7 min at 2.5 A/25 V on Trans-Blot Turbo Transfer System. Following transfer, stain-free membrane was imaged for total protein. Membrane was then blocked in 5% milk in TBST for 2 hr rocking at room temperature. Following blocking, membrane was incubated in primary antibody overnight rocking at 4 °C. Mouse monoclonal anti-β-actin (Santa Cruz Biotechnology #sc-47778) or mouse monoclonal anti-GFP (#sc-9996) were used at a dilution of 1:2,500 in 5% milk in TBST. The following day the membrane was washed three times for 5 min each with TBST and then incubated with HRP-conjugated goat anti-mouse antibody (sc-2005) at 1:2000 in 5% milk in TBST for 2 hours at room temperature. Membrane was again washed three times for 5 min each with TBST. Membranes were then incubated for 5 min in Clarity Western ECL Substrate (BioRad #1705060) and immediately imaged on a BioRad ChemiDoc MP imager. Band intensity was quantified using imageJ.

Statistical analysis
Experiment-specific details regarding sample size and statistical test used can be found in the corresponding Figure Legends. Significant p-values under 0.05 are denoted on all graphs and p-values above 0.05 are considered non-significant (ns). All statistical analysis was performed in GraphPad Prism 9. All data points for each experiment are included (no outlier exclusion was performed). For all whole animal fluorescence analysis, a sample size of 24 animals was generally used, each animal considered a biological replicate. Statistical analysis of high resolution fluorescence confocal imaging (HOE-1::GFP, ATFS-1::mCherry, and TMRE) was conducted on sample sizes between 60 and 80 animals of which animals were collected and imaged on three independent days, each animal considered a biological replicate. For western blot analysis, four independent samples were used for each condition, each sample (containing 50 worms each) is considered a biological replicate. For ddPCR analysis, a sample size of 4 was used for each condition, each sample (containing 10 worms each) is considered a biological replicate, each biological replicate was run in technical duplicate of which the average value was used for analysis.

Additional files
Supplementary files • Transparent reporting form • Supplementary file 1. C. elegans strains used in this study.
• Supplementary file 2. Oligonucleotides used in this study.

Data availability
All data generated or analyzed during this study are included in the manuscript and supporting files. Source data files have been provided for Figures 5 and 8.