Inhibition of ATR Reverses a Mitochondrial Respiratory Insufficiency

Abstract

Diseases that affect the mitochondrial electron transport chain (ETC) often manifest as threshold effect disorders, meaning patients only become symptomatic once a certain level of ETC dysfunction is reached. Cells can invoke mechanisms to circumvent reaching their critical ETC threshold, but it is an ongoing challenge to identify such processes. In the nematode Caenorhabditis elegans, severe reduction of mitochondrial ETC activity shortens life, but mild reduction actually extends it, providing an opportunity to identify threshold circumvention mechanisms. Here, we show that removal of ATL-1, but not ATM-1, worm orthologs of ATR and ATM, respectively, key nuclear DNA damage checkpoint proteins in human cells, unexpectedly lessens the severity of ETC dysfunction. Multiple genetic and biochemical tests show no evidence for increased mutation or DNA breakage in animals exposed to ETC disruption. Reduced ETC function instead alters nucleotide ratios within both the ribo- and deoxyribo-nucleotide pools, and causes stalling of RNA polymerase, which is also known to activate ATR. Unexpectedly, atl-1 mutants confronted with mitochondrial ETC disruption maintain normal levels of oxygen consumption, and have an increased abundance of translating ribosomes. This suggests checkpoint signaling by ATL-1 normally dampens cytoplasmic translation. Taken together, our data suggest a model whereby ETC insufficiency in C. elegans results in nucleotide imbalances leading to the stalling of RNA polymerase, activation of ATL-1, dampening of global translation, and magnification of ETC dysfunction. The loss of ATL-1 effectively reverses the severity of ETC disruption so that animals become phenotypically closer to wild type.

1. Introduction

The essential nature of mitochondria is underscored by the devastating effects of inherited mitochondrial diseases, which afflict approximately one in 5000 children and almost always result in pathological shortening of life [1]. Because these organelles play central roles in many cellular processes, including energy production, nucleotide metabolism, and apoptotic signaling [2,3,4], their progressive, age-dependent reduction in function means almost all of us, at some time in life, will experience the negative consequences of mitochondrial dysfunction [5]. Indeed, a growing number of age-related ailments common to western societies are either associated with, or directly caused by, mitochondrial disruption [6,7].

In Caenorhabditis elegans, as in humans, mitochondrial dysfunction is threshold-dependent, that is, a critical level of electron transport chain (ETC) dysfunction must be reached before pathology is observed [8]. Nematodes with high levels of mitochondrial dysfunction exhibit shortened lifespan and reduced fitness, whereas worms with moderate levels of mitochondrial dysfunction display an extended lifespan [9]. This response is evolutionarily conserved from yeast to mammals, and understanding the molecular mechanisms regulating lifespan extension in worms in response to mild ETC disruption may provide insight into strategies that could be applied to help keep our own cells above their critical mitochondrial dysfunction threshold.

Reduced ETC activity disrupts many downstream cellular processes, and, as might be expected, cells have evolved a variety of sensors and strategies to counter progression toward their critical ETC threshold. Multiple signaling pathways, collectively called retrograde responses, are activated within cells and these, in turn, control expression of nuclear counter-measures [10]. For example, an increase in the unfolded protein load of the mitochondrial matrix re-directs the transcription factor ATFS-1 from the matrix to the nucleus where it stimulates expression of mitochondrial chaperones [11]. Other signals, such as changes in calcium concentration, activation of mitogen-activated protein kinases (MAPKs), and even knock-on changes in the cytoplasmic unfolded protein load also result in retrograde response activation [12,13,14]. Counter-measures that are activated by cells include increased mitochondrial biogenesis, elevated mitochondrial DNA (mtDNA) replication, increased mitophagy, activation of alternate pathways of energy production (such as glycolysis), and changes in the abundance and activity of respiratory complexes or their regulatory factors [8,15,16]. The complete network of pathways that detect and respond to mitochondrial stress is, however, far from being fully understood [15].

Previously, we described a phenomenon in worms in which the loss of CEP-1, a homolog of the human p53 checkpoint protein that recognizes identical DNA sequences and is the only p53 family member present in C. elegans [16], right-shifted the mitochondrial ETC threshold, effectively allowing animals to cope with a greater degree of mitochondrial ETC disruption, and consequently mitigating the lifespan-shortening effects of severe mitochondrial dysfunction [17]. This response was due to alterations in autophagy and lipid metabolism, both of which are regulated by CEP-1 [18]. In the current study, we examined the role of other checkpoint proteins in modulating the mitochondrial threshold effect of worms, since this is still a little-explored area of investigation.

Ataxia-telangiectasia-mutated (ATM) and ‘ATM- and rad3-related’ (ATR) are two well-studied members of the phosphoinositide 3-kinase-related kinase (PIKK) family of proteins that play essential roles in transducing checkpoint signaling during the DNA damage response (DDR) [19]. In humans, both proteins together directly phosphorylate over 900 sites on some 700 proteins, including p53 [20]. ATM and ATR differ in the lesions to which they are recruited, with ATM being recruited to dsDNA breaks, and ATR having a broader specificity underscored by the presence of ssDNA, including recessed dsDNA breaks and stalled replication forks [21]. Depending on the level of damage and success of repair, checkpoint activation can lead to pro-survival or pro-apoptotic mechanisms [22]. Recently, additional functions were ascribed to both ATM and ATR. For ATM, these involve regulation of a variety of processes, including insulin signaling, the pentose phosphate pathway, and mitophagy [23,24,25,26,27]. Moreover, in yeast, ATM/Tel1p was found to act as a specific sensor of mitochondria-generated ROS [28,29]. ATR, on the other hand, was shown to be important in controlling an outer mitochondrial membrane (OMM)-localized apoptotic signal, autophagy, RNA polymerase activity, and chromatin condensation [28,29,30,31,32].

In this study, we uncover a new role for ATR in mitochondrial retrograde response signaling. Studying atl-1(tm853) loss-of-function mutants in C. elegans, we show that ATR/ ATL-1 is activated when the mitochondrial ETC is impaired. We find no evidence for increased nuclear DNA damage under these conditions, yet a significant change in ribose- and deoxyribose nucleotide pools is observed. Consistent with ATL-1 activation being tied to this alteration, we find an elevation in R-loop formation in atl-1(tm853) mutants following exposure to ETC disruption. R-loop formation is indicative of RNA pol II stalling, a known mechanism of ATR activation in human cells. Unexpectedly, we find that mutant atl-1(tm853) worms have half the amount of mitochondrial DNA of wild type worms, yet they show no reduction in the amount of oxygen they consume, neither under control- or ETC-disrupting conditions. In determining the mechanism by which ATL-1 functions to regulate mitochondrial retrograde response signaling, we show that ATL-1 inhibits translation, potentially through ICD-1, a ribosomal exit-tunnel chaperone. Finally, in the process of studying atl-1(tm853) mutants, we have discovered several other DNA damage response proteins, including MAPKAP-2 and MAPKAP-3, that function downstream of mitochondrial ETC disruption. These findings underscore what appears to be an extensive role for the DDR network in mitochondrial retrograde response signaling.

2. Methods

A detailed description of all methods is provided in File S1.

2.1. Nematode Strains and Maintenance

Genotypic information for all strains used in this study is provided in Table S4. Lines were maintained at 20 °C on standard NGM agar plates seeded with E. coli (OP50) [33]. To control for possible untoward effects derived from the nT1(IV;V) chromosomal pair in maternally-rescued atl-1(tm853) mutants, all lines, including the wild type control, were moved into the nT1(IV;V) background, and then unbalanced F1 animals were freshly collected for each relevant experiment. Further details are provided in File S1.

2.2. Bacterial Feeding RNAi

All bacterial feeding RNAi constructs were designed in the pL4440 vector backbone and maintained in HT115 bacteria [34]. Growth rate, life span, RNAi efficacy assays, and RNAi dilution methods are described in File S1.

2.3. Microscopy

Images of fluorescent worms were captured using an Olympus SZX16 fluorescence microscope connected to an Olympus DP71 CCD camera. In Figure 7C, to determine whether differences existed in the morphology of mitochondria between the four tested conditions, four randomly selected images from each condition were mixed and then grouped by a scorer who was blind to image identity. In a set of 16 images, 10 were co-grouped correctly. The probability that this grouping occurred by chance was modeled using an Excel VBA macro that randomly binned the images into four groups of four, 10,000 times (File S3). The final p-value was <0.0024 and deemed significant.

2.4. mRNA and mtDNA Quantitation

For quantitation of mRNA, fold change in each mRNA of interest was determined using the ΔΔCt method normalized to the geometric mean of cdc-42, pmp-3, and Y45F10D.4 [35]. For quantitation of mitochondrial DNA (mtDNA), PCR primer pairs targeting mtDNA-encoded ctb-1 and intron 4 of genomic DNA-encoded ama-1 were used as described [36]. The ratio of mitochondrial DNA to nuclear DNA was calculated using Real Time PCR, with ama-1 serving as the nuclear normalization quantity [37]. A complete list of qPCR primers is provided in Table S5.

2.5. Mutation Screening Assays

Full methodological details of the unc-93(e1500) and unc-58(e665) reversion assays; the eT1(III;V) and nT1(IV;V) lethal mutation assays; the Lac-Z Frameshift Assay; and the ‘terminal deoxynucleotidyl transferase dUTP nick end labeling’ (TUNEL) assay are presented in File S1.

2.6. Nucleotide Extraction and Mass Spectrometry

The quantitation of deoxyribonucleotide and ribonucleotide species in whole-worm extracts of isp-1(qm150) and wild-type (N2) animals was undertaken using HPLC-MS at the UTHSCA Mass Spectrometry Facility. Methods for nucleotide extraction, separation, quantitation, and statistical analysis are provided in File S1.

2.7. R-Loop Quantification

R-loop formation in worms was quantified using αS9.6 primary antibody (Kerafast, 1:2000 in 5% Blotto, 16 h, 4 °C), previously shown to be selective for DNA::RNA hybrids [38]. A list of R-loop reagents, and detailed methodology describing their use, is provided in File S1.

2.8. Spliceosome Reporter Assays

egl-15 and ret-1 alternate splicing reporter genes were engineered by Kuroyanagi and colleagues [39,40]. Details of reporter construction and their use are provided in File S1.

2.9. Nematode Oxygen Consumption

Nematode oxygen consumption measurements were undertaken using a Seahorse XFe24 Analyzer. Approximately 30 worms were added to each Seahorse well, with 5 replicate wells employed per test condition. We followed the procedure of Luz and colleagues [41], and only averaged oxygen consumption rates across the final four measurement phases for each well to provide a single rate value for that well. A total of four independent experiments were performed. Significance testing was undertaken using the Student’s t-test, with p < 0.05 considered significant.

2.10. Polysome Profiling

Quantitative polysome profiling was undertaken using the procedure of Steffen and colleagues [42] using a Brandel BR-188 Density Gradient Fractionation System. Several modifications were included to optimize for C. elegans polysome extraction, and these are described in detail in File S1. Two fully independent, experimental replicates were collected.

2.11. Western Analysis

Western blotting reagents and suppliers are described in File S1.

2.12. DNA Damage Response (DDR) Screen

The atm-1(gk186); atl-1(tm853) suppressor screen was undertaken using a bacterial feeding RNAi targeting 201 DDR-related genes [43]. A protocol describing our screening method, including quantitation of the RNAi effect-size and significance testing, is provided in File S1 and Tables S1–S3.

3. Results

3.1. Loss of ATL-1/ATR Desensitizes Worms to Mitochondrial ETC Stress

In C. elegans, atp-3 encodes the ortholog of the human ATP5O/OSCP subunit of the mitochondrial F₁F₀ ATP synthase, whereas isp-1 encodes the Rieske Fe-S ortholog of complex III. We have previously reported that wild type worms exposed to increasing amounts of bacterial feeding RNAi targeting either atp-3 or isp-1 reach a critical threshold, after which the concentration of RNAi causes worms to exhibit a reduction in body size and an extension of lifespan [44]. For isp-1 RNAi, life extension follows a monotonic function, whereas for atp-3 RNAi, life is first extended, and then it is pathologically shortened [44]. We previously mapped these functions using 12 point datasets across multiple rounds of testing [44]. The differing effect of atp-3 and isp-1 feeding RNAi on lifespan reflects the more potent degree of ETC inhibition that is ultimately attained following severe atp-3 knockdown.

Using these well-defined treatment regimes, we tested whether loss of the checkpoint proteins ATM-1 and ATR altered the phenotypic response of worms to mitochondrial ETC disruption (see File S1 for a full description of mutants used in this study, and the method used to the propagate lethal DDR mutations). When mutant atm-1(gk186) worms were exposed to atp-3 or isp-1 RNAi (1/10th strength and undiluted), animals matured and survived indistinguishably from similarly-treated wild type worms (Figure 1A–D). In contrast, atl-1(tm853) mutants were differentially refractory to the effects of knockdown of either RNAi. This effect was most evident in worms treated with 1/10th strength atp-3 or isp-1 RNAi, where animals showed obvious resistance to both size reduction (Figure 1A,B) and lifespan extension (Figure 1C,D) (additional examples are presented later in Figure 2A–C). atp-3 and isp-1 knockdown did eventually reduce adult size, and extend life in the mutant atl-1(tm853) background, suggesting the response to ETC disruption in these animals is right-shifted relative to wild type control (see schematic Figure 1C,D). Surprisingly, the measurement of atp-3 and isp-1 mRNA levels following RNAi treatment revealed that target gene knockdown was as efficacious in atl-1(tm853) mutants as it was in wild type worms (Figure 1E), indicating these animals were not deficient for RNAi. Furthermore, we found that atl-1(tm853) mutants were also refractory to mitochondrial ETC disruption induced using chemical inhibition (Figure S1). Based on these findings, we conclude that inhibition of the phosphoinositide 3-kinase-related kinase ATL-1 confers resistance to mitochondrial ETC disruption.

3.2. MAK-1 and MAK-2 Participate in Mitochondrial Retrograde Response Signaling

During the course of our studies, we made the unexpected observation that two independently-generated atm-1(gk186); atl-1(tm853) double mutants regained sensitivity to atp-3 and isp-1 knockdown (Figure 1). This observation provides additional support for the notion that the RNAi machinery is unaffected by the loss of atl-1. To explain this odd finding, we reasoned that ETC disruption might normally induce DNA damage that activates ATR, but not ATM. In the absence of both ATR and ATM, however, secondary forms of DNA damage might accumulate that specifically trigger a third checkpoint response that reinstates sensitivity to atp-3 and isp-1 knockdown. This hypothesis reflects the well-established redundancy in the DNA damage response network [20]. Several recent studies by us and others [13,36,37] have shown a role for MAPK signaling following ETC disruption in worms. Interestingly, the p38 MAPK-activated protein kinases MAPKAP-2 and MAPKAP-3 are checkpoint proteins known in humans to act in parallel to ATM and ATR [45,46,47]. To explore the possibility that alternate checkpoint signaling is activated in PIKK mutant worms following ETC disruption, we exposed animals to atp-3 knockdown, and simultaneously inhibited production of either MAK-1 or MAK-2, the worm orthologs of MAPKAP-2 and -3, respectively. The effect of these treatments on alt-1(tm853) and atm-1(gk186) single mutants, atm-1(gk186); atl-1(tm853) double mutants, and wild type worms was examined. Our results can be summarized as follows: (i) knockdown of mak-1 or mak-2, either alone or in combination with atp-3 knockdown, had no effect on final adult size in any of the genetic backgrounds (Figure 2A). (ii) Knockdown of mak-1 or mak-2 failed to reproducibly prevent life extension following atp-3 inhibition in the two independent atm-1(gk186); atl-1(tm853) double mutant isolates. (iii) Unexpectedly, the reduced ability of atp-3 RNAi to extend the lifespan of atl-1(tm853) mutants was further impaired when undertaken in conjunction with mak-1 or mak-2. A similar result was also observed for the atm-1(gk186) mutant. (iv) For wild type worms, only mak-1 knockdown significantly impaired atp-3 RNAi-induced life extension (Figure 2B,C). These findings suggest that at least two other checkpoint proteins, namely MAK-1 and MAK-2, are causally involved in lifespan extension (but not size determination) of worms experiencing mitochondrial ETC disruption. The role of different checkpoint proteins is, therefore, dependent upon the functional status of both ATL-1 and ATM-1. When these two checkpoint proteins are disrupted simultaneously, MAK-1 and MAK-2 become dispensable, potentially implying yet another checkpoint system awaits identification in atm-1(gk186); atl-1(tm853) double mutants. Although clearly complicated, this is not unexpected for a response that is network-based and sensitive to a variety of DNA lesions.

3.3. Knockdown of DNA Pol α Reverses the Small Phenotype of atm-1(gk186); atl-1(tm853) Mutants

To further explore the role of novel checkpoint proteins in modulating the critical ETC threshold in atm-1(gk186); atl-1(tm853) double mutants, we undertook a double RNAi screen of 201 DNA damage response-related genes. We asked whether knockdown of a target gene in this genetic background could re-confer resistance to atp-3 knockdown, similar to atl-1(tm853) mutants (for screen details, see Figure 3A). We identified two genes: scc-3, which encodes a cohesin complex subunit; and Y47D3A.29, which is orthologous to the human PolA1 catalytic subunit of DNA polymerase alpha (DNA pol α, Figure 3B,C). The effect of losing each gene was robust to three different post-hoc tests (each p < 0.001, see Tables S1–S3, File S2, and Methods for details of statistical testing). The isolation of PolA1 is particularly intriguing because it suggests that atp-3 knockdown in atm-1(gk186); atl-1(tm853) double mutants could be inducing some kind of disruption to DNA, its repair, or its replication, and which is bypassable by low-fidelity polymerases (which presumably accommodate knockdown of PolA1). This, in turn, might avoid checkpoint activation and make animals appear more atl-1(tm853)-like in our assay.

3.4. No Evidence for Permanent Nuclear DNA Damage Following Mitochondrial ETC Disruption

Nuclear DNA (nDNA) damage is the canonical signal that results in activation of ATR. In worms, resected ends of dsDNA breaks, and stalled replication forks have both been shown to activate ATL-1 [48]. Both lesions contain single-stranded DNA (ssDNA), and binding of heterotrimeric replication protein A (RPA) to these regions, in turn, recruits and activates ATL-1. To test if ETC disruption results in elevated levels of dsDNA breaks, we focused on isp-1(qm150) genetic mutants, since no viable atp-3 mutant exists, and utilized TUNEL staining on sections of paraffin-embedded worms [49]. We found no evidence for increased nDNA breakage (Figure S2A). DNA repair is sometimes inaccurate, depending on the repair machinery that is employed. To test if DNA mutation frequency was increased in worms exposed to ETC disruption, we employed four independent genetic assays. To facilitate analysis, these studies were undertaken using atp-3 RNAi. The eT1(III;V) and nT1(IV;V) reciprocal chromosomal translocations each suppress recombination against the regions they balance (~15 Mbp for both balancers). Homozygous-lethal DNA mutations, therefore, accumulate on balanced chromosomes, and they become identifiable when the balancing eT1 chromosomal pair, or the nT1 chromosomal pair, are lost because unbalanced progeny that are homozygous for the mutation are inviable and, hence, absent. Using both assays, we found no evidence for increased nDNA mutation rate following knockdown of atp-3 in wild type worms (Figure 4A). Similarly, the use of two Unc phenotypic reversion assays yielded similar conclusions (Figure 4B). Taken together, our data show there is neither an increase in DNA strand breakage nor is there an increase in nuclear DNA mutation frequency in worms experiencing mitochondrial ETC dysfunction.

3.5. Ribonucleotide and Deoxyribonucleotide Pools Are Disrupted by Mitochondrial ETC Dysfunction

To search for alternate mechanisms by which ATL-1 might be activated in worms experiencing mitochondrial ETC dysfunction, we examined changes in whole-worm nucleotide levels. As in humans, pyrimidine biosynthesis in C. elegans requires a functional mitochondrial electron transport chain. This is because formation of orotate from dihydroorotate is coupled to reduction of ubiquinone (UQ) in a reaction catalyzed by the inner mitochondrial membrane enzyme, dihydroorotate dehydrogenase (DHODH, Figure S3). In mammals, processes that slow flux through the mitochondrial ETC, downstream of UQ, also slow the rate of dihydroorotate dehydrogenase, resulting in measurable disruption of nucleotide pools [50]. Several important enzymes are sensitive to changes in nucleotide abundance, including DNA polymerases which are rate-limited by availability of their deoxyribonucleotide substrates [51]. Stalling of DNA polymerase, followed by continued DNA unwinding by the MCM helicase, results in ssDNA formation [52,53], which is a known substrate for ATR activation.

To measure nucleotide levels, we re-focused our attention on the isp-1(qm150) genetic mutant for ease of culture. We interrogated three stages of development—the final two stages of larval development (L3 and L4), and the first day of adulthood. These stages were selected because in wild type worms, there is a heavy demand for nucleotide precursors around the L3/L4 molt, and then again around the L4/adult molt. At these times, the gonad arms are expanding, and there is also a concomitant five-fold and a six-fold increase in mitochondrial DNA (mtDNA), respectively [54]. Among nine deoxyribonucleotide species that we were able to resolve and reliably detect (see Methods), we observed significant increases in dCDP and dTMP in L4 larvae of isp-1(qm150) mutants (Figure 5A). The increase in dCDP abundance was maintained into adulthood (Figure 5B). Among eleven ribonucleotide species that we were able to reliably resolve and detect, we observed significant increases in AMP, CMP, and UMP in L4 larvae of isp-1(qm150) mutants (Figure 5C). This was accompanied by a significant decrease in the level of GTP. In adult isp-1(qm150) worms, changes in ribonucleotide pools became more pronounced, with eight species being significantly underrepresented—ADP, ATP, GDP, GTP, CMP, CTP, UDP, and UTP (Figure 5D). Changes in nucleotide pools were specific to L4 and adult worms, since we did not detect any measurable differences in any quantifiable nucleotide species in L3 larvae of isp-1(qm150) mutants relative to wild type worms (Figure 5E). Finally, the absence of a corresponding increase in many of the monophosphate ribonucleotide species in adult isp-1(qm150) worms when their corresponding di- and triphosphate counterparts decreased suggests ribonucleotide pool sizes are reduced in these animals. We combined the levels of each mono-, di-, and triphosphate species, and tested if they differed significantly between strains at each developmental stage, but they did not. Error propagation, however, severely reduced our statistical resolving power (Figure 5F). Collectively, we conclude that disruption of the mitochondrial ETC in C. elegans results in measurable and significant changes in the abundance of multiple nucleotide species, notably with all ribonucleotide triphosphate species becoming underrepresented by adulthood.

3.6. Evidence for RNA Polymerase Stalling Following Mitochondrial ETC Disruption

Stalling of RNA polymerase II (RNA pol II) during transcription is sufficient to activate an ATR, RPA, and p53-dependent DNA damage response in human fibroblasts [55]. Given that mitochondrial ETC dysfunction in isp-1(qm150) mutants leads to measurable reductions in all four ribonucleotide triphosphates, and because we have previously shown a role for p53/cep-1 in the life extension of these worms [17], we hypothesized that ATL-1 activation following ETC disruption might occur as a result of RNA pol II stalling due to reduced nucleotide availability. We therefore quantified the formation of R-loops on genomic DNA [56]. R-loops are DNA::RNA hybrids that accumulate when transcription stalls (Figure 6A). We detected R-loops in purified, whole-worm genomic DNA fractions using slot blot analysis in conjunction with a DNA:RNA hybrid-specific antibody (αS9.6) [38]. To obtain sufficient material for these studies, we utilized a double feeding RNAi approach, and tested four conditions: wild type worms exposed to either vector control RNAi, 1/10th strength isp-1 RNAi, 9/10th strength atl-1 RNAi, or both 9/10th strength atl-1 RNAi with 1/10th strength isp-1 RNAi. Vector control feeding RNAi was added as balancer where appropriate. qRT-PCR analysis undertaken before the start of the experiment revealed 9/10th strength atl-1 RNAi reduced atl-1 mRNA by 50% (Figure 6B), the same amount that was observed in maternal-effect rescued atl-1(tm853) mutants (Figure 6C). The results of our R-loop studies can be summarized as follows (Figure 6D): (i) knockdown of isp-1 significantly enhanced R-loop formation relative to vector control (Student’s t-test, p < 0.05). (ii) Knockdown of atl-1 did not block R-loop formation by isp-1, as would be expected for a gene that functioned downstream of R-loop accumulation. We next tested whether any differences in DNA:RNA polymer length distinguished R-loops generated in worms exposed to isp-1 RNAi from those exposed to isp-1 and atl-1 double RNAi. Genomic fractions were separated on an agarose gel, stained with ethidium bromide, then transferred to nitrocellulose and probed with αS9.6 antibody (Figure 6E). No overt difference distinguished the two samples; however, in both instances, we noted some of the staining with αS9.6 localized to low molecular weight DNA:RNA fragments (<2 kb), in addition to the higher molecular weight structures that we expected. The identity of these smaller fragments remains uncharacterized (see Section 4). Collectively, our studies suggest that transcriptional stalling is increased in worms exposed to mitochondrial ETC disruption, identifying one route by which ATL-1 is potentially activated in these animals.

3.7. RNA Splicing Is Not Altered by Mitochondrial ETC Disruption

Having identified a possible mechanism by which ATL-1 is activated in worms experiencing mitochondrial ETC stress, we next sought to determine the mode by which ATL-1 acts to further disrupt mitochondrial function. It is well-established that pre-mRNA splicing is a co-transcriptional process [58], and it has also been reported that isp-1(qm150) mutants are sensitive to knockdown of two spliceosome factors, SFA-1 and REPO-1, fully ablating their life extension [59]. We therefore tested whether the stalling of RNA pol II altered either splicing efficiency or splicing fidelity in worms exposed to ETC disruption, and if ATL-1 played a role in this process. We used two synthetic-gene reporter assays developed by Kuroyanagi and colleagues [39,60], designed to measure alternate splicing at the egl-15 and ret-1 loci, respectively. The latter reporter was shown to be a sensitive marker of age-related splicing dysfunction [59]. Each assay has the potential to produce two fluorescent outputs that record changes in the nature of a differential splicing event, and the tissues in which these events occur (Figure S4A,B). We exposed worms carrying the relevant reporter cassette to double feeding RNAi targeting atp-3 or isp-1 (1/10th strength) in conjunction with either atl-1 or vector control (9/10th RNAi strength) (Figure S4C). Animals were followed three days into adulthood. None of the conditions that disrupted the mitochondrial ETC, neither in the presence or absence of atl-1, resulted in alteration of egl-15 or ret-1 splicing relative to vector control-treated animals.

3.8. Reduced ATL-1 Activity Does Not Enhance Hormetic Stress Response Activation

We next tested whether the loss of atl-1 affected the activity of major stress response pathways known to be activated by atp-3 or isp-1 depletion [3,13,44,61]. The loss of atl-1 did not result in constitutive activation of Phsp-6::GFP, Pgst-4::GFP, or Ptbb-6::GFP, which are markers of ATFS-1, SKN-1/NRF-2, and PMK-3/p38 activation, respectively (Figure S5A–D); and neither did the loss of atl-1 increase gene expression of sod-3, ugt-61, or hsp-6 (Figure S5E). These results indicate that the reduction of atl-1 does not affect typical mitochondrial stress response pathways.

3.9. Oxygen Consumption Remains Surprisingly Unaltered in atl-1 Mutants Exposed to ETC Disruption

We next tested whether mitochondria in atl-1(tm853) mutants were inherently different from those of wild type worms. We also analyzed mitochondria from atm-1(gk186) single mutant worms, as well as atl-1(tm853); atm-1(gk186) double mutants. Properties that we examined included mitochondrial DNA (mtDNA) content, nuclear-encoded ETC transcript abundance, mitochondrial morphology, and whole-worm oxygen consumption. Our findings can be summarized as follows: relative to wild type worms, mtDNA copy number is halved in atl-1(tm853) mutants, consistent with an earlier report [62]. mtDNA content was also halved in atl-1(tm853); atm-1(gk186) double mutants (Figure 7A). To assess changes in nuclear-encoded ETC transcript abundance, we focused on five genes: nuo-6, mev-1, isp-1, cco-1, and atp-3, representing subunits from mitochondrial complexes I through V, respectively. We found that all transcripts except nuo-6 were significantly decreased relative to wild type worms in all three mutant strains studied. The only exception was mev-1, for which mRNA levels remained unchanged in atm-1(gk186) mutants (Figure 7B).

To quantify changes in mitochondrial morphology, we employed a hexokinase-2::GFP translational reporter (HXK2::GFP) that is localized to the outer mitochondrial membrane [63]. We observed a significant increase in the number of fused mitochondria in reporter worms treated with isp-1 RNAi (p < 0.0024, see Methods for details of statistical testing). Co-knockdown of atl-1 did not significantly affect this result. Mitochondria in worms treated only with atl-1 RNAi were indistinguishable from those of wild type animals (Figure 7C). Finally, the most significant finding that we uncovered was related to alterations in oxygen consumption by atl-1(tm853) mutants. When wild type worms, atl-1(tm853) and atm-1(gk186) single mutants, as well as atl-1(tm853); atm-1(gk186) double mutants were cultured on isp-1 RNAi, the basal oxygen consumption rate of atl-1(tm853) mutants, as well as atl-1(tm853); atm-1(gk186) double mutants, significantly increased relative to wild type worms (Figure 7D). Moreover, though isp-1 knockdown uniformly decreased oxygen consumption in wild type worms, atm-1(gk186) single mutants and atl-1(tm853); atm-1(gk186) double mutants, and by the same extent, oxygen consumption in atl-1(tm853) surprisingly remained unchanged following this treatment. We conclude that though atl-1(tm853) mutants contain half as much mtDNA as wild type worms and have a reduced abundance of several nuclear-encoded ETC transcripts, these worms surprisingly maintain normal oxygen consumption, even when confronted with disruption of their mitochondrial electron transport chain.

3.10. Loss of ATR Leads to Recovery of Translational Activity in Worms Exposed to ETC Disruption

Several mechanisms could explain how atl-1(tm853) mutants maintain oxygen consumption despite ETC disruption. Any process that ultimately increases flux or efficiency of the respiratory chain could be involved. Enhanced supercomplex formation, elevated matrix Ca[2]⁺, modification of ETC subunits, or simply enhanced translational output of respiratory subunits could all be behind the unique respiratory response seen in atl-1(tm853) mutants. We focused on translational output because there is evidence from human 293T cells that ATR might directly regulate the ribosomal machinery by affecting phosphorylation of the nascent polypeptide-associated complex (NAC) [20]. In worms, ICD-1 is the sole βNAC isoform [64,65]. Although the ATR phosphorylation site of human BTF3/βNAC is not conserved in ICD-1, we observed, by western analysis, that ICD-1 protein levels are selectively elevated in atl-1(tm853) mutants following atp-3 knockdown (isp-1 knockdown was not tested due to limited quantities of antisera, see Section 2) (Figure 8A). Moreover, early analyses involving several hundred microarray datasets [66] showed icd-1 formed part of a tightly clustered gene co-expression group (Figure 8B) that is significantly enriched (hypergeometric distribution, p < 0.05) for genes encoding ribosomal and mitochondrial ETC subunits, as well as key translational regulatory factors (Figure 8C). We therefore tested whether global translational activity was downregulated by mitochondrial ETC disruption, and if the loss of atl-1 permitted its recovery. We employed polysome profiling in conjunction with RNAi targeting both isp-1 (1/10th RNAi strength) and either atl-1 or vector (both at 9/10th RNAi strength). We found that the actively translating polysome fraction of worms treated with isp-1 RNAi was decreased by half relative to vector control-treated worms. Strikingly, simultaneous removal of both atl-1 and isp-1 restored polysome activity to wild type levels (Figure 8D). Loss of atl-1 in the absence of ETC disruption did not significantly impact polysome abundance relative to wild type worms. These data imply that ATL-1, when activated by mitochondrial ETC disruption, normally acts to dampen global ribosomal translation. Whether this is through a novel phosphorylation site on ICD-1, the downregulation of factors that normally control icd-1 expression, or via some other means that globally regulates translational activity remains unclear. There must be some specificity in the translational targets of ATR-1, however, because simply elevating global rRNA production by inhibiting ncl-1, a negative regulator of FIB-1/fibrillarin, a nucleolar protein involved in the regulation and maturation of rRNA, is not sufficient to block the size reduction caused by isp-1 bacterial feeding RNAi (Figure 8E,F). Collectively, our results lead to a model for how ATL-1 acts to modulate the mitochondrial threshold effect (Figure 8G).

4. Discussion

In this study, we discovered that the PIKK kinase ATR/ATL-1 plays a specific role in modulating the mitochondrial threshold effect of C. elegans. When flux through the ETC was reduced, ATL-1 was activated. We presented data suggesting that the stalling of RNA polymerase could be the trigger for ATL-1 activation. We recorded measurable reductions in many nucleotide species in adult worms, including substrates for both RNA and DNA polymerase, and we presented biochemical evidence showing RNA polymerase stalling following ETC disruption. In addition to these findings, we showed that ATL-1 paradoxically exacerbates mitochondrial dysfunction during times of mitochondrial stress. Specifically, when worms containing reduced levels of atl-1 were challenged with ETC disruption, unlike wild type animals, these animals showed no signs of impairment in global translational output or respiratory oxygen consumption. Phenotypically, loss of atl-1 effectively reversed a mitochondrial ETC insufficiency in C. elegans. The survival benefits of further dampening ETC activity following an ETC insult are not entirely clear, but dampening of mitochondrial respiratory activity in conjunction with global translational dampening might represent an effort to slow overall metabolism or to maintain stoichiometric balance between nuclear- and mitochondria-encoded ETC proteins [67].

4.1. Mode of ATL-1 Activation Following Mitochondrial ETC Disruption

In our search for signals that triggered ATL-1 activation in response to ETC disruption, we found no evidence for elevated rates of nuclear DNA mutation or strand breakage. This does not preclude the possibility that DNA damage occurred and was accurately repaired. Such damage would have escaped detection in our assays. Nonetheless, we investigated other potential mechanisms of ATL-1 activation, and, in isp-1(qm150) mutants, found that nucleotide pools were significantly changed. This is significant because ATR activation has been linked to changes in nucleotide pools in other model systems. Specifically, in S. cerevisiae, glucose starvation induces Snf1p (AMP kinase) at the mitochondrial surface following reduction of adenosine ribonucleotide energy charge. In turn, Snf1p was shown to phosphorylate Mec1p (ATR) [32]. Whether a similar mode of ATR activation functions in C. elegans remains unclear, but Ggc1p, the protein that initially recruits Mec1p to the yeast mitochondrial membrane under low-glucose conditions, has no worm ortholog. We pursued an alternate hypothesis for nucleotide-dependent activation of ATR in worms, namely the stalling of RNA and/or DNA polymerases.

Previously, it had been shown that blocking RNA polymerase during transcriptional elongation is sufficient to activate ATR and induce a DNA damage response (DDR) [55]. Indeed, we found evidence that DNA::RNA R-loop hybrids are increased in the presence of mitochondrial ETC disruption, which is indicative of transcriptional stalling. However, it is unclear whether all the R-loops we detected were genomic DNA (gDNA)-derived transcription intermediates. When examining samples on an agarose gel in conjunction with an antibody targeting these structures (S9.6), R-loops are expected to appear at a high molecular weight in association with the gDNA band. We clearly detected an elevated signal in this region in our isp-1 RNAi-treated samples, consistent with our hypothesis. Moreover, we observed an additional smear of low molecular weight nucleic acids. At this point, it is unknown if these low molecular weight molecules are truly DNA::RNA hybrids or a contaminating nucleic acid. The S9.6 antibody has a five-fold higher affinity for DNA::RNA hybrids over RNA::RNA hybrids [38], so in anticipation of the latter, our samples were treated with RNase A, ruling out this possibility. One idea is that the low molecular weight DNA::RNA hybrids are derived from mtDNA fragments that have leaked from dysfunctional mitochondria. Consistent with this idea, in mouse embryonic fibroblasts, mitochondrial stress results in mtDNA leaking into the cytoplasm and initiating an immune response [68]. On the other hand, worms use a rolling circle replication method for their mtDNA, so precisely how R-loops would form in the absence of D-loop (theta) replication, as is used by mammalian cells, remains enigmatic [69]. Another idea is that the low molecular weight species we detected are cytoplasmic DNA::RNA hybrids produced by DNA polymerase α (DNA pol α). Starokadomsky and colleagues [70] showed that a small fraction of DNA pol α localizes to the cytoplasm in human cells, where it generates DNA::RNA hybrids. These molecules directly modulate production of type I interferons, and possibly exist to control the activation threshold of the innate immune response by foreign nucleic acids. In patients with X-linked reticulate pigmentary disorder (XLPDR), a primary immunodeficiency with autoinflammatory features, levels of the catalytic subunit of DNA pol α (DNA PolA1) are decreased enough to disrupt cytoplasmic DNA::RNA hybrid formation, but not enough to affect DNA replication [70]. Intriguingly, in our screen for DDR proteins that, when inhibited, reversed the small size phenotype of atl-1(tm853); atm-1(gk186) double mutants experiencing mitochondrial ETC stress, we identified an RNAi clone targeting DNA PolA1. Although not discussed earlier, we also isolated a second DNA pol α clone in that screen that targeted a different region of DNA PolA1, but instead resulted in synthetic lethality. This dose-dependent phenotypic response is remarkably similar to what was just discussed for humans. Obviously, more work is needed to determine the identity, provenance, and significance of the low molecular weight DNA::RNA species in worms experiencing ETC stress, and if and how they might relate to ATR activation.

4.2. Downstream Targets of ATL-1 That Modulate Mitochondrial ETC Activity

One important finding from our study is the observation that atl-1 knockdown mitigates the effects of mitochondrial ETC dysfunction. This is an intriguing finding, as it suggests the normal cellular response to ETC disruption is ostensibly further detrimental. Two possible explanations for how the loss of atl-1 could result in enhanced ETC function are: atl-1 knockdown triggers active restoration of mitochondrial function, or, alternatively, atl-1 knockdown precludes propagation of whatever negative downstream effects are normally elicited by ETC disruption. According to our data, atl-1(tm853) mutants contain half the amount of mtDNA as wild type worms, and have a significantly reduced abundance of nuclear-encoded ETC transcripts. These findings argue against the first hypothesis. Surprisingly, despite their shortcomings, we find that atl-1(tm853) mutants are still able to maintain oxygen consumption at near normal levels when confronted with ETC disruption. The precise mechanism involved is yet to be determined, but it may be as simple as atl-1(tm853) mutants being able to maintain the translation of a factor that enhances supercomplex formation, for example [71]. Alternatively, atl-1(tm853) mutants may be able to keep their cells fortified with sufficient ETC proteins to counteract any rapid turnover that is likely to occur when subunits become rate-limiting during the assembly process. Along these lines, when we examined polysome formation under conditions of ETC dysfunction, we found that animals with functional ATL-1 had reduced polysome formation, but removal of atl-1 restored this number back to pre-ETC disruption levels. Future studies will be aimed at determining the precise mRNA species present in these polysomes.

Another way in which atl-1(tm853) mutants might be able to counter ETC disruption involves the nascent protein-folding machinery. We showed that ICD-1/βNAC, a protein that acts as a ribosome-associated chaperone, is upregulated in atl-1 mutants upon ETC disruption. In yeast and plants, βNAC plays a role in targeting proteins to the mitochondria, and is required for efficient mitochondrial function [72,73,74,75,76]. βNAC was also one of the few proteins found to be specifically regulated by ATR-dependent phosphorylation in mammals [20]. The ATR-dependent phosphorylation site in mammals is not conserved in C. elegans; however, there are several other potential phosphorylation sites in the ICD-1 protein sequence [77]. It will be of particular interest in the future to ascertain if ATR directly regulates βNAC abundance on ribosomes to modulate translation of ETC transcripts in worms.

4.3. Mitochondrial Dysfunction in Mammals

Johnson and colleagues [78] successfully extended the survival of a mouse model of Leigh syndrome, which was deficient in Ndufs4 of complex I, by chronic administration of the mTOR inhibitor, rapamycin. mTOR is a positive regulator of p70S6 kinase, and a negative regulator of 4E-BP, both of which work to elevate translation. At first, this seems surprising in light of our findings, suggesting global translation needs to be elevated to counter ETC dysfunction in worms. However, though acute inhibition of mTOR indeed results in a global dampening of translation, it also results in the specific translational upregulation of mitochondrial ETC components [79,80]. Moreover, it has been reported that in contrast to acute rapamycin administration, chronic rapamycin administration does not lead to reduced ribosomal activity in vivo [81]. Whether the increase in polysomes that we observed in atl-1 animals exposed to ETC disruption represented upregulation of global mRNA translation or just a discrete subset of mRNAs only encoding ETC proteins remains to be determined.

In conclusion, we have shown that the blockade of atl-1 effectively reverses ETC disruption in C. elegans. Preventing mitochondria from transitioning into their critical ETC threshold appears, quite literally, to be the difference between life and death.