Partial loss of function in Pol III specifically impacts tRNA expression in C. elegans

Pol III transcribes 100 s of short, noncoding RNAs involved in several fundamental cellular processes [9,10]. To better understand which are involved in ageing, we sought to define the Pol III-transcribed RNAs whose levels are altered by the longevity-promoting, partial loss-of-function in Pol III. We compared expression profiles of adult Caenorhabditis elegans that were treated with RNAi against the gene encoding the largest subunit of Pol III, rpc-1, from the L4 stage to that of controls exposed to the vector alone; this rpc-1 RNAi treatment results in partial knockdown of the cognate mRNA and promotes longevity [12]. We employed a modified RNA-Seq protocol, similar to published protocols [16], including size-selection and demethylation steps to enrich for Pol III-transcribed RNAs, which are about 70 to 300 bases long and often heavily modified. The reads obtained were used to quantify transcript levels using Salmon, which is able to discriminate RNAs similar in sequence and estimate their levels [17]. Differential expression was determined with DESeq2 [18], for all transcripts shorter than 300 bases at a false discovery rate (FDR) of 10%. We obtained consistent signals for all main, expected RNAs, including tRNAs, 5S rRNA, and a number of snoRNAs known to be transcribed by Pol III in worms [19].

We found that not all Pol III-transcribed genes were equally sensitive to the partial loss of function in the polymerase, detecting widespread changes in the amounts of different tRNAs in the worms treated with rpc-1 RNAi but no substantial effects on 5S, 7SL, and others (Fig 1A, full analysis results are given in S1 Data). Indeed, we detected 35 tRNAs as differentially expressed. We independently confirmed the substantial and reproducible reduction in the levels of a tRNA^Leu and a tRNA^Arg by northern blotting (Figs 1B and S1). Interestingly, although the majority of tRNAs were repressed, 9 (26%) were induced. As these changes are a result of Pol III inhibition, the most parsimonious explanation is that a reduction in some tRNAs triggers a compensatory increase in others. Overall, our analysis indicated that the longevity phenotype of rpc-1 knockdown is associated with a disturbance in expression of tRNAs specifically, but not of other Pol III-transcribed genes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Partial inhibition of Pol III in worms alters tRNA expression specifically.

(A) Volcano plots of differential expression of short RNAs (<300 b) between worms treated with rpc-1 RNAi and vector controls, with different classes of transcripts coloured in different panels. “Other” refers to snoRNAs identified in a previous study (see main text). Horizontal line indicates the p value threshold for 10% FDR. (B) Northern blots of RNA from worms treated with the vector control or rpc-1 RNAi. Actin band from 1 wild-type sample was accidentally lost from the blot. Data underlying this figure can be found in S1 Data. Uncropped images of blots are presented in S1 Fig.

https://doi.org/10.1371/journal.pbio.3002853.g001

Pol III inhibition and impairments in tRNA metabolism can activate UPR

Reduced activity of Pol III in fruit flies results in resistance to tunicamycin [12], an agent that causes proteostatic stress in the endoplasmic reticulum (ER). To test whether this is conserved in C. elegans, we quantified the ability of control and rpc-1 RNAi-treated animals to withstand tunicamycin exposure. Similarly to flies, we observed that worms are significantly more tunicamycin resistant upon partial inhibition of Pol III (Fig 2A and S1 Table). Additionally, we found that rpc-1 RNAi also enhanced C. elegans resistance to a pulse of heat stress (Fig 2B and S1 Table), a marker of general proteostatic tolerance in worms [20]. These resistance phenotypes appeared specific for stresses that disturb proteostasis, as rpc-1 RNAi treatment did not induce tolerance to juglone, a xenobiotic thought to cause oxidative stress [21] (S2A Fig and S2 Table). Together, these results suggest that a reduction in Pol III activity causes a conserved increase in proteostatic resistance.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Pol III inhibition and interference in tRNA metabolism both trigger UPR in worms.

(A) Tunicamycin and (B) heat shock survival of worms treated with rpc-1 RNAi or the vector control. Demography and statistical analyses of data presented in A, B and additional repeats are provided in S1 Table. (C) Expression of the hsp-4-4::GFP reporter in rpc-1 RNAi- or vector control-treated worms (n = 11 rpc-1 RNAi/12 control; effect of RNAi p < 10⁻⁴, effect of age p < 10⁻⁴, RNAi by age interaction p < 10⁻⁴, LM). Fluorescence intensities are relative to control on day 2. (D) Representative images. (E) Expression of the hsp-4::GFP reporter (day 5 of adulthood) upon RNAi against genes encoding aminoacyl tRNA synthases, as well as rpc-1 RNAi and tunicamycin treatment (n = 10–25; ANOVA p < 10⁻⁴; p value from Dunnet’s test to vector control is indicated). Boxplots show quantiles with individual data points overlayed. Data underlying this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3002853.g002

Such increased proteostatic resistance may be caused by the activation of the unfolded protein response (UPR) [22]. To establish whether rpc-1 RNAi triggers UPR activation, we examined C. elegans carrying copies of the well-established hsp-4::GFP reporter [23] using quantitative fluorescence microscopy. HSP-4 is the C. elegans homologue of BiP, the ER resident chaperone whose expression increases in response to ER stress [24,25]. Indeed, we found that rpc-1 knockdown increased the expression of hsp-4::GFP as worms aged, indicating that the UPR is also induced in these animals (Fig 2C and 2D).

Given that Pol III inhibition is specifically impacting tRNA expression, we examined whether the UPR could also be activated by directly interfering with tRNA metabolism. Using RNAi, we individually knocked down 25 aminoacyl tRNA synthases, enzymes responsible for charging tRNAs with their cognate amino acids, and quantified hsp-4::GFP expression. Interestingly, we found that individually knocking down several aminoacyl tRNA synthases was sufficient to activate the expression of hsp-4::GFP to the levels similar to those achieved by rpc-1 RNAi or tunicamycin exposure (Fig 2E). One of these was dars-2, an aspartate-tRNA synthase, and indeed, a tRNA^Asp (WormBase transcript ID: F28F8.t3) was the second most repressed tRNA in rpc-1-knockdown worms (S1 Data). Hence, altering the availability of specific, charged tRNAs is sufficient to alter C. elegans physiology, namely UPR activation, consistent with alterations caused by Pol III inhibition.

The UPR consists of 3 branches each activating a specific transcription factor (TF): XBP-1, ATF-4, and ATF-6 [22]. Mutations in each TF was sufficient to reduce the expression of hsp-4::GFP and abolish the ability of rpc-1 RNAi to induce it (S2B–S2D Fig). Given that each UPR branch has been associated with health and longevity [26–34], we then tested whether reducing the activity of each individual TF was sufficient to reduce the lifespan extension caused by Pol III inhibition. As previously reported, RNAi against rpc-1 was able to extend lifespan in wild-type worms [12] and this effect was maintained in worms mutant for xbp-1 or atf-4 or when rpc-1 and atf-6 RNAi were co-administered (S2E–S2G Fig and S2 Table), suggesting that no UPR branch alone is required for rpc-1 RNAi mediated longevity. Similarly, individually knocking down the aminoacyl tRNA synthases, whose loss-of-function was able to significantly induce hsp-4::GFP (Fig 2E), was not sufficient to induce longevity (S2H Fig). In contrast, broad, pharmacological inhibition of these enzymes in yeast and C. elegans [35], or a mutation in a valyl aminoacyl tRNA synthetase (vars-2) in C. elegans [36], can all extend lifespan, possibly because they cause a more substantial loss of cellular tRNA-charging ability. Hence, while we could not ascribe the longevity of rpc-1 knockdown to a specific branch of UPR in C. elegans, our data indicate that partial inhibition of Pol III increased tolerance to proteostatic stress, likely mediated by a disruption to tRNA metabolism.

Pol III loss-of-function impacts tRNA expression and activates UPR in flies

Pol III activity limits lifespan in both worms and flies, showing evolutionary conservation [12]. We wanted to examine whether the differential effect on tRNA expression by a partial loss of function in Pol III can also be observed in flies. We compared expression profiles between wild-type and long-lived female flies carrying a heterozygous mutation (a P-element insertion) in one of the Pol III-specific subunits, Polr3D (previously named dC53, see ref. [37]) at 1 week of age. This mutation represents a substantial loss of function: homozygous flies are not viable while heterozygous females have a significant reduction in Polr3D mRNA [12]. We focused on the adult fly midgut, since Pol III activity in this organ drives fly ageing [12].

Similarly to C. elegans, we found that not all Pol III-transcribed genes were equally sensitive to the partial loss of function in the polymerase, detecting extensive changes in the amounts of tRNAs in the mutant guts, with a limited or no effects on 5S RNA and others (Fig 3A, “Others” identified in ref. [38]; full analysis given in S1 Data). Again, as observed in C. elegans, we found some tRNAs were repressed and others induced; surprisingly, however, in Polr3D^EY/+ flies the majority were induced (Fig 3A). Additionally, the magnitude of change was substantially less in flies than in worms (compare x axes in Figs 1A and 3A). The results of northern blot analyses of a tRNA^His and a tRNA^Gln expression were consistent with the RNA-Seq findings (Fig 3B with the results of Linear Model (LM) analysis presented in the caption, S3A–S3D Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Partial inhibition of Pol III in flies alters tRNA expression and induces BiP.

(A) Volcano plots of differential expression of short RNAs (<300 b) in midguts of female Polr3D^EY/+ flies relative to wild type, with different classes of transcripts coloured in different panels. “Other” refers to U6 and several snoRNAs identified in a previous study (see main text). Horizontal line indicates the p value threshold for 10% FDR. (B) Northern blots of RNA from dissected guts, representative images. (C) Northern blots, quantification of relative tRNA expression (n = 5, Polr3DEY/+ versus wild type p = 0.035 one-tailed t test on least-squares means within LM; LM summary: effect of genotype p = 0.07, no significant effect of tRNA type or the genotype by type interaction). (D) Relative expression of pre-tRNA^His assessed by qPCR (n = 7–8, p = 1.5 × 10⁻³ t test). (E) TPM estimated in the RNA-Seq data for 3′ processed region of pre-tRNA^His and mature tRNA^His (n = 3, Polr3D^EY/+ versus wild type for pre-tRNA and tRNA separately p < 0.05 post hoc t tests on least-squares means within LM; LM summary: effect of genotype p > 0.05, effect of tRNA sequence p < 10⁻⁴, genotype by sequence interaction p = 2 × 10⁻³). (F) Relative expression of BiP mRNA in whole bodies of female Polr3D^EY/+ flies (n = 10, p < 10⁻⁴ t test). Boxplots show quantiles with individual data points overlayed. Data underlying this figure can be found in S1 Data. Uncropped images of blots are presented in S3 Fig.

https://doi.org/10.1371/journal.pbio.3002853.g003

tRNAs are produced by Pol III as precursors that are subsequently processed to mature forms [39,40] and the level of these pre-tRNAs is often a good indicator of in vivo Pol III activity. The surprising finding of increases in abundance of several mature tRNAs in Polr3D^EY/+ flies prompted us to quantify pre-tRNAs in the fly gut. Using qPCR, we found the expression levels of pre-tRNA^His reduced (Fig 3C), which contrasted with the increase in the mature tRNA^His levels observed in RNA-Seq and northern blot (S1 Data and Fig 3B). This differential effect on pre-tRNA^His and mature tRNA^His was confirmed when we reanalysed the RNA-Seq data to estimate specifically the occurrence of sequences corresponding to the 3′ end of pre-tRNA^His, which is absent from the mature form, and the mature tRNA^His. As expected, transcripts per million (TPM) estimated for the precursor were reduced in Polr3D^EY/+ flies, while those corresponding to the mature tRNA were increased (Fig 3D; note that pre-tRNA^His TPMs are several orders of magnitude lower than for the mature tRNA). This indicates that compensation for partial loss of Pol III activity is occurring in the fly gut, for example by reduced tRNA degradation. This compensation is likely more extensive than seen in worms since the mutant flies had impaired Pol III function from development onwards. The differential impact on the expression of different tRNA observed by RNA-Seq may not be simply due to this feedback, as Polr3D loss of function had no significant effect on the expression levels of 2 other pre-tRNAs, pre-tRNA^Ile and pre-tRNA^Leu (S3E and S3F Fig). Hence, both in worms and flies, loss-of-function in Pol III results in a complex disturbance of tRNA expression.

Similarly to worms, partial Pol III inhibition increased the fly’s resistance to the proteotoxic stress caused by tunicamycin [12]. We wanted to examine if it also causes UPR activation. As a readout of UPR activation, we examined the transcript levels of the ER chaperone Hsc70-3/BiP (orthologue of the worm HSP-4). Expression of BiP was significantly increased in Polr3D^EY/+ females (Fig 3E), consistent with increased tunicamycin resistance. Overall, our analyses revealed that the longevity due to a partial loss of function in Pol III is associated with changes in tRNA abundance and UPR activation in flies, as in worms.

Impairing initiation on tRNA genes is sufficient to extend Drosophila lifespan

Our RNA-Seq analyses indicated complex alteration occurring specifically in tRNA levels upon partial loss of function in Pol III in both worms and flies. Pol III initiates transcription from 3 types of promoters. Each promoter type is used to recruit the polymerase to a specific class of genes and requires a particular combination of TFs [9,41] (Fig 4A): TFIIIA is solely required for the expression of 5S (type 1 promoter), TFIIIC for expression of 5S and tRNAs (types 1 and 2), while SNAPc is required for expression of other RNAs such as 7SK [9,41] (type 3). To directly examine which Pol III targets contribute to ageing, we decide to manipulate TFs that are required for initiation by the polymerase, as this allowed us to restrict the classes of genes whose expression is affected.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. TFIIIC activity promotes ageing in the fruit fly.

(A) Schematic of promoter types and TFs used by Pol III to initiate transcription. (B) Lifespans of GTF3C5^KG/+ and wild-type female flies (n = 134/4 and 124/7 dead/censored flies respectively, p = 2.3 × 10⁻¹³ log-rank test). (C) Changes in median lifespan relative to wild-type observed for the indicated genotypes in several experimental trials (see S4 Fig for details; p values from log-rank test are indicated). (D) Tunicamycin resistance of GTF3C5^KG/+ and wild-type female flies (n = 165/0 and 158/0 dead/censored flies, respectively, p = 1.3 × 10⁻⁸ log-rank test). (E) Relative expression of pre-tRNA^Leu (n = 5–8, p = 0.04 t test). (F) Lifespans of TIGS>GTF3C3^RNAi in the presence or absence of the RU486 inducer (n = 144/2 and 147/2 dead/censored flies respectively, p = 2.2 × 10⁻⁴ log-rank test). Data underlying this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3002853.g004

We employed the fly model as TFIIIA, TFIIIC, and SNAPc could be distinguished in the genome of this organism. Using Flybase, we identified the CG9609 (henceforth GTF3A) as the Drosophila orthologue of the human GTF3A gene encoding TFIIIA, l(2)37Cd (henceforth GTF3C5) as encoding a subunit of TFIIIC and Pbp49 encoding a subunit of SNAPc. We obtained P element insertions into these genes, namely GTF3A^G2025, GTF3C5^KG04199, and Pbp49^KG02425, and backcrossed them into an outbred, wild-type background. We compared the lifespans of heterozygous mutant and wild-type females, since Pol III inhibition has a pronounced effect on longevity in that sex [12]. TFIIIC heterozygous mutant flies were consistently long lived in all 4 independent, sequential experimental trials performed (Figs 4B, 4C and S4). On the other hand, TFIIIA heterozygous mutants were long-lived in 2 out of 5 trials (Figs 4C and S4) indicating TFIIIA may have a conditional or restricted effect on ageing. SNAPc mutants were not observed to be long-lived in any experimental trial (Figs 4C and S4). This partial loss-of-function in TFIIIC was also sufficient to cause tunicamycin resistance (Fig 4D) as previously observed upon Pol III inhibition [12]. This phenotype appeared specific to the UPR since TFIIIC heterozygous mutant flies were not resistant to the oxidative stress-inducing agent, paraquat [42] (S4J Fig).

We also found similarities between the effects of GTF3C5^KG04199/+ and Polr3D^EY/+ on pre-tRNA expression in the gut: both resulted in a decrease in expression of at least 1 pre-tRNA tested without affecting all pre-tRNAs equally. Specifically, the mutation in the gene encoding the TFIIIC subunit resulted in a decrease in pre-tRNA^Leu levels (Fig 4E), without significant effects on pre-tRNA^His and pre-tRNA^Ile (S5A and S5B Fig). Hence, as expected, TFIIIC heterozygous mutant flies show evidence of impaired initiation from a type 2 promoter.

Partial Pol III inhibition restricted to the adult female fly gut is sufficient for longevity [12]. To extend and further validate our TF analysis, we knocked down GTF3A, CG8950 (henceforth GTF3C3) encoding another subunit of TFIIIC or the Pbp95 subunit of SNAPc specifically in the adult female gut, using available and validated RNAi lines and the inducible, GeneSwitch driver, TIGS. The RNAi constructs were induced from day 2 of adulthood by feeding the RU486 inducer; the inducer had no significant effect on the lifespan of the driver-alone control (S5C Fig). Similar to the mutant phenotypes described above, we observed an extension of lifespan upon adult gut-restricted knockdown of the TFIIIC subunit that was not present when TFIIIA or a SNAPc subunit were knocked down (Figs 4F, S5D and S5E); GTF3A and Pbp95 knockdowns actually shortened lifespan. Hence, using independent approaches and genetic reagents we established TFIIIC activity as driving ageing in the fruit fly. Together with the transcriptomic analysis presented above, the consistent effect of TFIIIC on longevity strongly implicates tRNAs as Pol III targets limiting lifespan, with the expression of 5S potentially having a more restricted, condition- or tissue-specific role.

Mutation in a TFIIIC subunit recapitulates broad health benefits of Pol III inhibition

Akin to humans, flies display a range of age-related pathologies and functional impairments. With age, the intestinal stem cells in their gut hyperproliferate and mis-differentiate and the ability of their gut to act as a barrier becomes compromised, with these phenotypes being more prominent in females [43–45]. Our previous work showed that the inhibition of Pol III is able to delay this age-related decline in gut health [12]. We next examined whether a mutation in a TFIIIC subunit can recapitulate this, by looking at the increase in the number of flies with compromised gut barrier function (those that were unable to restrict the presence of a blue food dye to their gut—the “smurf” phenotype [44,46]). Similarly to humans, flies in our outbred, healthy background display substantial heterogeneity in ageing phenotypes (e.g., ref. [47]). Hence, to insure robust statistical analyses, we employed appropriate regression models to detect effects on the rate of ageing within the population as a significant age-by-genotype interaction. We found that both a heterozygous mutant in the GTF3C5 TFIIIC subunit and in the Polr3D Pol III subunit were able to significantly delay the age-related increase in the number of flies with a compromised gut barrier function with age (Fig 5A, statistical analysis using ordinal logistic regression and a priori contrasts is given in S6A Fig). We found no evidence that the magnitude of their effects on this phenotype was different between GTF3C5^KG04199 and Polr3D^EY/+ (S6A Fig), indicating both mutants were similarly able to delay age-associated impairments in gut health.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Pol III and TFIIIC limit fly health in late life.

(A) Cumulative proportion of partial and full smurfs (flies with impaired gut barrier function) in cohorts of Polr3D^EY/+, GTF3C5^KG/+ and wild-type females (n = 52–788, effect of genotype: mutants versus wild-type p = 0.11 and GTF3C5^KG/+ versus Polr3D^EY/+ p = 0.88, age p < 2 × 10⁻¹⁶, age by genotype interaction: mutants versus wild-type p = 9.8 × 10⁻⁴ and GTF3C5^KG/+ versus Polr3D^EY/+ p = 0.42, ordinal logistic regression). Mean and standard error of the mean shown. (B) Time spent in the central zone of the arena during exploratory walking by Polr3D^EY/+ and wild-type females (n = 14–16, effect of genotype p = 0.01, age p = 2.7 × 10⁻³, age by genotype interaction p = 9 × 10⁻³, LM). (C) Hight climbed by Polr3D^EY/+ and wild-type females in a climbing (negative geotaxis) assay (n = 99–126, effect of genotype p = 4.7 × 10⁻⁶, age p < 2 × 10⁻¹⁶, age by genotype interaction p = 2.7 × 10⁻⁶, LM). (D) Hight climbed by Polr3D^EY/+, GTF3C5^KG/+ and wild-type females in a climbing assay (n = 95–108, effect of genotype: mutants versus wild-type p = 0.09 and GTF3C5^KG/+ versus Polr3D^EY/+ p = 1.7 × 10⁻³, age p < 2 × 10⁻¹⁶, age by genotype interaction: mutants versus wild-type p = 0.046 and GTF3C5^KG/+ versus Polr3D^EY/+ p = 0.32; LM). Boxplots show quantiles with individual data points overlayed. Data underlying this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3002853.g005

As flies age, the performance of their neuronal and muscular systems also deteriorates. This deterioration can be observed as age-related changes in their spontaneous, exploratory walking behaviour, as well as in their ability to climb a vertical surface in an induced, negative geotaxis response [47–49]. Comparing the performance of Polr3D^EY/+ and wild-type females, we found that Polr3D^EY/+ delayed the age-related increase in the time spent in the centre of the arena during exploratory walking (Fig 5B, statistical analysis using a LM is given in S6B Fig), with no effect on other age-related changes in exploratory behaviour (S6B–S6E Fig). Similarly, Polr3D^EY/+ delayed age-related loss in climbing ability (Fig 5C). Hence, in addition to its effect on gut health, the reduction in Pol III activity improves the performance of the neuromuscular system with age.

Lastly, we used a climbing assay to examine whether the improvements to neuromuscular performance in old age can be recapitulated by impairing TFIIIC function. Both GTF3C5^KG04199 and Polr3D^EY/+ females showed a significant delay in the age-related decline in climbing ability, with no significant difference in the magnitude of the effects of the 2 mutations (Fig 5D, statistical analysis using LM and a priori contrasts is given in S6F Fig). Hence, Pol III inhibition has broad health benefits, improving performance of multiple organ systems in old flies, that can be recapitulated by partial loss of function in TFIIIC. This, together with our transcriptomic and lifespan analyses, further implicates tRNAs in late-age health more broadly.

The effect of partial Pol III inhibition on tRNA expression and proteostatic resilience is conserved in mice

We wanted to examine whether the differential effect of partial loss-of-function in Pol III on tRNA expression was conserved in mammals. We extended our RNA-Seq analysis to mice and used a heterozygous loss-of-function Polr3B mutant [15], profiling expression in the superior duodenum, representing a similar tissue to the fly gut, in both female and male mice. Again, we found evidence of differential expression of tRNAs in the Polr3B^-/+ mutant (Fig 6A and 6B; full analysis results are available in S1 Data). The effect was sexually dimorphic with significant changes observable in females (Fig 6A) but not in males (Fig 6B), echoing the sexual dimorphism of the lifespan effects observed in flies [12] as well as the sexually dimorphic age-related phenotypes observed in Polr3B^-/+ mice [15]. Such sexual dimorphism may reflect an underlying sex-specific physiology, as is the case for the fly gut [50,51]. Note that fewer differentially expressed tRNAs were detected in mutant mice relative to worms and flies, again consistent with mild phenotypes observed in adult Polr3B^-/+ mice [15].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Partial inhibition of Pol III alters tRNA expression in mice.

Volcano plots of differential expression of short RNAs (<300 b) in the superior duodenum between Polr3B^-/+ and wild-type mice, with different classes of transcripts coloured in different panels. (A) Female. (B) Male. “Other” refers to 7SK, RNase P, MRP, U6, Vault, and Y RNA; horizontal line indicates the p value threshold for 10% FDR. (C) Thapsigargin LD50 of primary fibroblasts in culture, isolated from female or male, wild-type or Polr3B^-/+ mice (n = 5–7, effect of genotype p = 6 × 10⁻³, effect of sex or interaction p < 0.05, LM). Boxplots show quantiles with individual data points overlayed. Data underlying this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3002853.g006

In worms and flies, Pol III loss of function and disruptions to tRNA metabolism resulted in increased organismal tolerance of proteostatic stress. To examine whether similar proteostatic resilience can be observed in mice, we isolated dermal fibroblasts and then exposed them to different doses of thapsigargin, another agent that causes ER stress [52]. The dose required to kill 50% of the cells (LD₅₀) was significantly higher for both female and male Polr3B^-/+ mutant fibroblasts compared to wild-type cells (Fig 6C), indicating increased cellular capacity to withstand proteotoxic stress. Hence, in worms, flies, and mice, a partial loss-of-function in Pol III leads to a disturbance in the composition of the tRNA pool and promotes resistance to proteostatic stress.

Partial loss of Pol III function impacts translation in worms and flies

Finally, we compared the effect of partial Pol III inhibition on the tRNA pools in different animals. Despite the overall effects on Pol III-target expression appearing similar between species, with dysregulation of tRNAs observed in worms, flies, and mice, we could not find any obvious shared pattern with respect to which tRNAs were differentially expressed. For example, tRNAs belonging to specific isotypes were not consistently affected between species (S7 Fig); a similar case was observed for the isodecoders (S8 Fig). This lack of similarity at the codon level may not be surprising considering the known differences in codon usage and tRNA gene abundance between C. elegans, D. melanogaster, and mammals [53]. Interestingly, we found that changes in tRNA levels in worms were correlated with the tRNA copy number, while those in flies were correlated with codon usage (S9 Fig). This observation highlights that different forces may shape the final tRNA levels in response to a reduction in Pol III activity, e.g., a substantial feedback is likely occurring in flies to compensate for loss of Pol III function.

Although different between species, the observed tRNA changes could result in altered decoding rates of individual codons, due to changes in abundance of both the cognate tRNA as well as the competing, near-cognate tRNAs [54]. Using computational models of the ribosomal codon decoding process [55], we confirmed that loss-of-function in Pol III is likely to lead to changes in decoding rates in each organism (Fig 7A–7C). Such alterations to decoding rates could result in changes to both bulk translation as well as to translation of specific mRNAs.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Predicted and observed changes in translation upon partial inhibition of Pol III across worms, flies, and mice.

Computationally predicted fold changes in codon decoding times resulting from changes in the tRNA pool in animals with a reduced function of Pol III compared to controls, based on the RNA-Seq data. (A) C. elegans. (B) Fruit fly. (C) Female mouse. Note differences in Y axes between panels. (D) Translation rates measured by puromycin incorporation in worms, example blot and quantification (n = 3, p = 0.033 paired t test). (E) Translation rates measured in female fly guts, example blot and quantification (n = 5, p = 0.64 paired t test). Boxplots show quantiles with individual data points overlayed. (F) Volcano plot of differential translation efficiency in whole bodies of female Polr3D^EY/+ flies relative to wild type, estimated from Ribo-/RNA-Seq analysis. Position of Clbn is indicated. Data underlying this figure can be found in S1 Data. Uncropped images of blots are presented in S10 Fig.

https://doi.org/10.1371/journal.pbio.3002853.g007

In worms, the translation rates of several codons were predicted to be substantially reduced (Fig 7A). Consistent with this, we observed a significant reduction in overall translation in worms treated with rpc-1 RNAi using a puromycin-incorporation assay (Figs 7D and S10A). In flies, the predicted change to translation rates were more subtle and balanced (Fig 7B) and indeed we found no evidence for an overall impact of Polr3D^EY/+ on translation in female fly guts (Figs 7E and S10B). To assess whether translation may be qualitatively rather than quantitatively affected in flies, we performed Ribo- and RNA-Seq, again comparing Polr3D^EY/+ to wild-type flies. To obtain sufficient biological material and as a proof-of-principle, we used whole flies (not dissected midguts as above). We sequenced ribosome protected fragments corresponding to over 19,000 open reading frames, even though we recovered a large number of fragments corresponding to rRNA, as is often the case in Drosophila [56] (S11 Fig). Combining the Ribo-Seq and RNA-Seq data, we found over 400 mRNAs with evidence of different translation efficiency in the mutant relative to wild-type flies (10% FDR, Fig 7F). Interestingly, this included evidence of increased translation of Caliban (Clbn) mRNA (Fig 7F), encoding a key component of the Ribosome-associated Quality Control (RQC) pathway that ensures co-translational proteostasis [57,58]. Overall, the tRNA dysregulation observed as a consequence of a partial loss of function in Pol III is predicted to cause a change in translation in worms, flies, and mice, with this prediction experimentally validated in worms and flies. This alteration in translation may be the mechanism underlying the increased proteostatic resilience and longevity.

C. elegans maintenance and strains

C. elegans were routinely grown and maintained on Nematode Growth Media (NGM) seeded with Escherichia coli OP50-1. The wild-type strain was Bristol N2. The following strains were used: SJ4005 (zcIs4 [hsp-4p::GFP]), SJ17 (xbp-1(zc12);zcIs4 [hsp-4p::GFP]), RE666 (ire-1(v33)), RB790 (atf-4(ok576)), RB772 (atf-6(ok551)). JMT200 (atf-4(ok576);hsp-4p::GFP) and JMT201 (atf-6(ok551);hsp-4p::GFP) were constructed by genetically crossing zcIs4 [hsp-4p::GFP] to atf-4(ok576) and atf-6(ok551), respectively.

C. elegans lifespan assays

Lifespan assays were performed as previously described [12]. Briefly, synchronous L1 animals were placed on NGM plates seeded with E. coli OP50-1 until they reached L4 at 20°C. At L4 stage, worms were shifted to plates seeded with E. coli HT1115 carrying appropriate RNAi clones obtained from the Ahringer RNAi library. The plates were supplemented with 50 μm FuDR (5-Fluoro-2′-deoxyuridine), unless otherwise specified, and lifespans were carried out at 25°C. For combinatorial RNAi knockdown, the 2 respective RNAi cultures were mixed 1:1 before seeding the plates. In these experiments, controls were also diluted 1:1 with HT115 expressing empty pL4440. Live, dead, and censored worms were recorded every 2 to 3 days by scoring movement with gentle prodding when necessary.

C. elegans hsp-4::GFP expression assays

Synchronous worms expressing the hsp-4::GFP transcriptional reporter were raised, treated with RNAi, and aged as for lifespan assay. At specific time points, animals were immobilised on 2% agarose pads using 0.06% Levamisole and imaged immediately using a Leica DMR with 10× zoom. ImageJ software [79] was used for GFP fluorescence quantification. Fluorescence intensity was calculated as pixels per unit area, and background fluorescence subtracted. The values were normalised to those of the control on day 2 of adulthood.

Worm stress assays

Worms were raised and treated with RNAi as for lifespan assays and assays carried out at 25°C on day 2 adults unless otherwise stated. For heat shock, worms were shifted to 30°C on plates for 3 h. Live, dead, and censored worms were scored daily. For tunicamycin resistance L4 animals were transferred to RNAi plates containing FuDR and supplemented with 40 μg/ml tunicamycin (dissolved in di-methyl sulfoxide, DMSO) or DMSO only. Live, dead, and censored worms were scored daily. For oxidative stress, 100 worms per group were transferred to fresh RNAi plates containing 400 μm of juglone (5-hydroxy-1,4-naphthoquinone). Dead worms were scored every hour until the population died.

C. elegans RNA extraction

Worms were raised and treated with RNAi as for lifespan assays. On day 2 of adulthood worms were washed off the plates with M9 buffer (containing protease inhibitor), allowed to settle and washed again to remove excess bacteria. The tubes containing worms were flash frozen in liquid nitrogen and thawed on ice several times. The worm lysate was centrifuged at 12,000 g for 10 min, supernatant collected in lo-bind eppendorfs and stored in −80°C. To extract total RNA, the worm samples were thawed on ice and supplemented with RNase inhibitor before commencing RNA extraction. Total RNA was extracted using Direct-Zol RNA MiniPrep Kit (Zymo Research) following manufacturer’s instructions.

Fruit fly stocks and husbandry

The wild-type stock was acquired in Dahomey (now Benin) in 1970, and it has been maintained in population cages at 25°C with a 12 h:12 h light/dark cycle. The background used here was cleared of Wolbachia several years ago. The w¹¹¹⁸ or v¹ mutations were introduced into this outbred background by extensive backcrossing. The mutations: CG9609^G2025 (called here GTF3A^G2025), l(2)37Cd^KG04199 (GTF3C5^KG04199), Pbp49^KG02425 and Polr3D^EY22749; and transgenes: TIGS, UAS-GTF3A^RNAi (TRiP.HMS01000), UAS-CG8950^RNAi (called here UAS-GTF3C3^RNAi, TRiP.HMC02421) and UAS-Pbp95^RNAi (TRiP.HMC06384) were backcrossed into the Dahomey background carrying w¹¹¹⁸ or v¹ to allow tracking of the insertion. RNAi lines were from the TRiP project [80]. Flies were reared and experiments performed at 25°C, 12 h:12 h light/dark cycle, 60% humidity, and standard sugar/yeast/agar (SYA) medium [81], unless otherwise noted.

Fruit fly lifespan assays

Lifespan experiments were carried out as described [12]: progeny of suitable crosses collected over <24 h were reared at standardised larval densities, adults permitted to mate for 48 h after emergence before being sorted by sex on CO₂ and placed into plastic vials (15 per vial) in DrosoFlippers (https://www.drosoflipper.com/). Where required, food was supplemented with RU486 (Sigma, M8046) at 200 μm, in which case the control food contained the same volume of the ethanol vehicle. Food was changed and dead/censored flies were recorded trice weekly.

Fly stress assays

Flies were prepared as for lifespans (female flies, 10 flies per vial). On day 7, the flies were transferred to experimental vials. For tunicamycin resistance flies were introduced onto food containing 5% sucrose and 1.5% agar (no yeast) and tunicamycin (2 mg/l in dimethylsulfoxide; Cell Signalling) on day 9. The equivalent volume of the vehicle alone was added to control treatments (which essentially showed no death in the assay period). Dead flies were scored once or twice daily. For paraquat resistance assays fly food (1% agar, 5% sucrose, and no yeast) was supplemented with 10 mM Paraquat (Sigma, 856177-1G METHYL VIOLOGEN DICHLORIDE HYDRATE, 98%) and dead flies were scored twice daily (beginning and end of the day, approximately 8 h apart).

Fruit fly gut permeability (smurf) assays

Smurf assays were carried out largely as described [44,47]. Prior to scoring, flies were placed on food containing 2.5% (wt/vol) of blue dye (FD & C blue dye no. 1, Fastcolors, CAS: 3844-45-9) for 48 h. Flies were scored under CO2 as partial smurfs if the dye had seeped outside the gut but had not stained the head or as a full smurfs if the entire fly was blue. The same cohort was continuously assayed.

Fruit fly negative geotaxis (climbing) assays

These were performed essentially as described [47,49]. Flies were transferred to empty vials so that they could climb 2 vial heights in DrosoFlippers (https://www.drosoflipper.com/). Flies acclimatised for 20 to 30 min, were gently tipped to the bottom and the climbing was video recorded. Stills from the same time point (when young wild-type flies start reaching maximum height, most often 15 s) were analysed in Fiji [79] and coordinates exported to Excel (Microsoft). To protect against potential outliers, top 5% and bottom 5% of measurements were removed from all conditions/time points. The same cohort was continuously assayed.

Fruit fly exploratory walking assay

Exploratory walking assay was performed as described [47,48]. Individual flies were placed in 4 cm diameter/1 cm height circular Perspex arena and video recorded for 15 min, after a 1-min rest. Videos were analysed using Ethovision XT video tracking software (Noldus) to extract the reported parameters. Naive flies were used at each time point.

Fruit fly RNA extraction and qPCR

Total RNA was quantitatively extracted with Trizol (Invitrogen) from 7- to 9-day-old whole flies (5–10) or dissected midguts (12). cDNA was generated with random hexamers and Superscript II (Invitrogen) according to manufacturer’s instructions. Quantitative PCR was performed using Power SYBR Green PCR Master Mix (ABI), on QuantStudio6 Flex real-time PCR, and quantities of each sequence were determined with the relative standard curve method. Primers used were:

Tubulin F TGGGCCCGTCTGGACCACCA,

Tubulin R TCGCCGTCACCGGAGTCCAT,

Hsc70-4 (BiP) F GCTTGATTGGTCGCAAGT,

Hsc70-4 (BiP) R CTTCTCGTCCTTGTAGGTCA,

pre-tRNAHisGTG-1 F CGTGATCGTCTAGTGGTTAG,

pre-tRNAHisGTG-1 R CCCAACTCCGTGACAATG,

pre-tRNAIleTAT-1 F CGCACGGTACTTATAATCAG,

pre-tRNAIleTAT-1 R, CCAGGTGAGGCTCGAACTC,

pre-tRNALeuCAC-1 F GCGCCAGACTCAAGATTG,

pre-tRNALeuCAC-1 R TGTCAGAAGTGGGATTCG,

U3 forward CACACTAGCTGAAAGCCAAG, and

U3 reverse CGAAGCCCTGCGTCCCGAAC.

The primers specific for pre-tRNAs were designed based on previous biochemical characterization [82] or predicted intronic sequences [83] where each primer set may recognise transcripts emerging from identical or highly similar tRNA genes in different genomic loci.

Fruit fly ribosome footprinting assay (Ribo-Seq) and analysis

Sample preparation and sequencing for Ribo-Seq and the associated RNA-Seq was performed by CD Genomics, using 300 to 500 mg of whole flies, as described [84] and using RNase I. Library was prepared with NEBNext Multiple Small RNA Library Prep Set for Illumina (catalogue no. E7300S, E7300L) and sequenced using Illumina HiSeq TM X10. Raw sequencing data have been submitted to Gene Expression Omnibus project GSE232724.

FastQC [85] was used to verify the quality of the raw sequencing reads for Ribo-seq and RNA-seq data set, and Cutadapt (v2.10) [86] was used to trim the reads with the following parameters -a CTGTAGGCACCATCA—error-rate = 0.1—overlap = 3—times = 2. Ribo-Seq data was assessed with RoboToolkit [87]. Salmon (v0.13.2) [17] was used to quantify the expression of fragments using a whole genome decoy-aware transcriptome index built BDGP6.32 including noncoding RNAs, with parameters—validateMappings—k 20 for Ribo-Seq—validateMappings for RNA-Seq. Subsequent analysis used DESeq2 [18] in R (R version 4.1.2 and 4.2.2) [88]. Initial principal component analysis (PCA) of RNA-Seq data showed that wild-type repeat 3 clustered substantially away from other samples and was therefore removed from further analyses. Differential translation efficacy was assessed as the interaction term in the DESeq2 model: ~ batch (replicate) + condition (genotype) + sequence type (Ribo/RNA) + condition: sequence type. DESeq2 analysis results are given in S1 Data.

Puromycin incorporation assay

Puromycin incorporation assays were performed essentially as described [12]. Flies were prepared as for lifespans (10 flies per vial). Midguts of 7-day-old female flies were dissected in ice-cold PBS, with wild-type and mutant flies processed in batches carried out in parallel with 10 guts per replicate. Dissected midguts were then transferred to 0.6 ml of ice-cold Schneider’s medium (Sigma, #S0146), and 0.4 ml of the same medium pre-warmed to 25°C and supplemented with puromycin (GIBCO, #A11138-02, to a final concentration of 10 mM) was added and incubated at 25°C for 30 min. TCA was added immediately to a final concentration of 12.5%. The midguts were lysed with glass beads (Sigma, #G8772) in a homogeniser (Precellys 24 Ribolyzer). For worms, the protocol was adapted from [89] with minor modifications. Worms were raised on E. coli OP50-1 until L4 larval stage. At L4 larval stage, worms were transferred to control and rpc-1 RNAi plates. On day 2 of adulthood, exactly 250 worms were picked for each condition and transferred into 100 μl S-basal buffer in a microcentrifuge tube. After 2 washes, worms were incubated in 90 ml S-basal medium supplemented10 ml heat-inactivated E. coli OP50-1 and with or without 0.5 mg/ml puromycin. Worms were incubated in microcentrifuge tubes for 3 h at 20°C with gentle manual agitation every 20 to 30 min. Finally, worms were gently centrifuged, washed with S-basal a few times, and homogenised with a pellet pestle in 20 ml SDS loading dye containing proteinase K.

The proteins were separated on gradient gels (Thermo Fisher Scientific, #NP0322BOX). The western blots were probed with anti-puromycin antibody (Milipore, #12D10) followed by HRP conjugated Anti-Mouse secondary (Abcam, #ab6789). Anti-Tubulin antibodies (Sigma, T6199 for flies and T5168 for worms) were used as the loading controls. Blots were imaged in an Amersham ImageQuant800 imager. The intensity of anti-puromycin bands between 15 and 165 kDa was quantified from chemiluminescence images in ImageJ 1.54D [79], relative to tubulin and expressed relative to the control. Batching information was used to perform paired t test.

Northern blot analysis

Relevant RNA sequences were downloaded from FlyBase (https:flybase.org) or WormBase (https:wormbase.org). The sequences were reverse complemented and the best stretch of primer worthy approximately 60 bp in the sequences were identified using Primer3 (https://primer3.ut.ee/). DNA Oligos 3’Digoxigenin (DIG)-labelled and HPLC purified were ordered from Eurofins Genomics (see below). Midguts of the 7-day-old female flies were dissected and total RNA was quantitatively isolated as described above. For worms, total mRNA was isolated from day 2 adults as described for small RNA-seq, without the size selection step. The RNA were separated in 6% TBE-Urea gel (Invitrogen, EC68652BOX) and transferred to BrightStar-Plus Positively Charged Nylon Membrane (AM10100). The RNA was fixed with UV-crosslinking in (Stratagene UV Stratalinker 1800). The probes (25 pmole) were hybridised at 64°C overnight in Roche DIG Easy Hyb Hybridization Buffer (11603558001) with gentle agitation. The membrane washing and blocking were performed with Roche DIG Wash and Block Buffer Set (11585762001, version 12). Hybridised DIG-labelled probes were detected with DIG Roche Anti-Digoxigenin-AP, Fab fragments (11093274910) and Roche CDP-Star, ready-to-use (12041677001) reagent as per manufacturer’s instructions. The membranes were imaged and quantified as described under puromycin incorporation assay. Actin was used as the loading control. Probes used for Northern Blotting are as follows:

Flies

tRNA Glutamine-TTG-1-1 FBtr0076083 GATTTGAACTCGGATCGCTGGATTCAAAGTCCAGAGTGCTAACCATTACACCATAGAACC

tRNA Histidine-GTG-1-1 FBtr0087963

GCCGTGACCAGGATTCGAACCTGGGTTACCACGGCCACAACGTGGGGTCCTAACCACTAG

Actin (Act57B) FBgn0000044

ACGGGGAGCGTCATCACCGGCGAAACCGGCCTTGCACATGCCGGAGCCATTGTCAACGAC

Worms

tRNA Leucine- TAA2.1 T22C8.t1

GTTCGAACCCACGCGGTGTTATCCACCATTGGAACTTAAGGCCAACGCCTTAACCACTCG

tRNA Arginine- TCG-2-1 F14B4.t1

GATCGAACCTGTAATCCCCAGATTCGAAGTCTGGTGGCTTATCCATTAGGCGACGCGACC

Actin (act-1) T04C12.6

GGTGACACCATCTCCAGAGTCGAGGACGACTCCGGTGGTACGTCCGGAAGCGTAGAGGGA

Mouse lines and husbandry

The mouse line C57BL/6N-Polr3b<em7(IMPC)Tcp>/Tcp was provided as part of the NorCOMM2 project funded by Genome Canada and the Ontario Genomics Institute (OGI-051) at the Toronto Centre for Phenogenomics. This allele from project TCPR0237 was generated by injecting Cas9 D10A mRNA and a single guide RNA with the spacer sequence TCATGTCCCTCAGACGGCAC. This resulted in an 8 bp insertion TCCAGTGT at Chromosome 10 positive strand 84632456 bp (GRCm38) predicted to cause a frameshift mutation with early truncation; c.305_306insTCCAGTGT; p.R103Pfs*3. The generation of founder mice was performed in compliance with the Animals for Research Act of Ontario and the Guidelines of the Canadian Council on Animal Care. The TCP Animal Care Committee reviewed and approved all procedures conducted on animals at TCP. Four male and 5 female 2-month-old heterozygous C57BL/6N-Polr3b<em7(IMPC)Tcp>/Tcp (hereafter Polr3b^+/-) founder mice were obtained and bred in-house with non-transgenic C57Bl/6N mice. To avoid any possible carry-over effect in wild-type offspring derived from Polr3b^+/- parents, experimental mice were generated by mating Polr3b^+/- or Polr3b^+/+ offspring with wild-type C57Bl/6N mice to generate experimental Polr3b^+/- or Polr3b^+/+ mice, respectively [15]. Mice were maintained as previously described [90], under a UK Home Office Project Licence (PDBDC7568), following local ethical review, and adherance to the “principles of laboratory animal care” (NIH Publication No. 86–23, revised 1985).

Mouse genotyping

Mice were genotyped from ear biopsies either in-house by PCR, or by qPCR using an external provider (Transnetyx, Cordova, Tennessee, United States of America). In-house PCR was carried out on genomic DNA using primer pairs specific for the wild-type allele (for: 5′-AGGCTGCTGTGCACTGTATT-3′; Rev: 5′-GACGGCACTGGAGCAGAAT-3′; band size = 82 bp) or the mutant allele (for: 5′-TCAGTGGGGAAAGTTCAGGC-3′; Rev: 5′-TCAGACGGACACTGGACACT-3′; band size = 110 bp).

Mouse dermal fibroblasts cultures and thapsigargin survival assays

Primary dermal fibroblasts from 4- to 6-week-old wild-type and Polr3b^+/- mice were isolated and cultured as previously described [91], and then cellular resistance to endoplasmic reticulum stress was assessed as previously described [52]. In brief, fibroblasts were exposed to thapsigargin (0–100 μm; Sigma Aldrich (Dorset, United Kingdom)) and subsequently incubated at 37°C for 24 h in DMEM. Following incubation, cells were washed with 1× PBS, incubated in DMEM supplemented with Pen-Strep, L-Glutamine, and 10% FBS at 37°C. Cell viability was measured using the WST-1 test following the manufacturer’s protocol (Roche Diagnostics, West Sussex, UK). Each cell line was exposed to thapsigargin in triplicate and LD₅₀, the dose which caused 50% cell death, calculated using probit analysis (AAT Bioquest, California, USA).

Mouse RNA extraction

Following an overnight fast, mice were culled by cervical dislocation and then duodenum was dissected, flushed with PBS, immediately frozen in liquid nitrogen, and stored at −80°C. RNA was subsequently isolated following homogenisation with Trizol (Invitrogen, Paisley, UK) following manufacturer’s instructions.

RNA-Seq of Pol III-transcribed genes

The RNA samples obtained from worm, fly, and mouse were selected for RNAs less than 200 nucleotides in length using mirVANA RNA isolation kit. The final elute containing the small RNAs was demethylated using rtStar tRF&tiRNA Pretreatment Kit following manufacturer’s instructions. Finally, the RNA cleanup was performed using Zymo-RNA Clean and Concentrator-5 cleanup. RNA quality was checked prior to Library preparation using Agilent Small RNA kit for Bioanalyzer. The typical yield at this step was 4 to 6 μg of <200 nucleotide-long, demethylated, DNA-free RNA in 8 μl of water. Library preparation and sequencing was done by Glasgow Polyomics, Glasgow, UK. Briefly, libraries were prepared using TruSeq Nano kit, analysed using Bioanlyser and sequenced on the Nextseq using paired end 2*75 bp read length to a sample average of 15 M reads. Raw sequencing data have been submitted to Gene Expression Omnibus projects: GSE232719, GSE232720, GSE232721, and GSE232723.

For each data sets, raw sequencing reads were checked using FastQC [85] and trimmed by Cutadapt (v2.10) [86] using default parameters. Transcript abundance were quantified using Salmon (v0.13.2) [17] with parameters—validateMappings, using all-RNA, decoy-aware transcriptome indexes built from WBcel235, GRCm39, or BDGP6.32 including noncoding RNAs for worm, mouse, and fly, respectively. The data for sequences shorter than 300 bases were imported into R with Tximport [92], and differential expression analysis performed with DESeq2 using a standard workflow [18]. Genomic tRNA Data Base [83] was used to obtain information on tRNAs for each organism. tRNA gene occurrence in the genome was counted based on files in the tRNA Data Base [93]. Codon usage per species was obtained from Codon Usage Data Base [94]. DESeq2 analysis results are given in S1 Data.

Codon decoding times were predicted using Gillespie models of the tRNA-dependent codon decoding process [95,96]. Modified bases for tRNA anticodons were retrieved from the tRNAdb2009 database [97] and individual tRNA: codon relationships were classified as Watson–Crick decoding cognate, wobble-decoding cognate, near-cognate, or non-cognate using base-pairing schemes as described [95]. Cellular concentrations for each tRNA were calculated assuming a total cellular tRNA concentration of 190 μm [95] and using the fractional signal for each tRNA in the RNA-Seq data. Total concentrations of the 4 different tRNA classes for each codon were then calculated by summing the concentration for all tRNAs in each class. Codon decoding times were estimated by averaging results from 10,000 repeat runs of the Gillespie model for each codon, using the tRNA class concentrations as input.

Statistical analysis

Survival differences were assessed by a log-rank test either using the online freeware OASIS (http://sbi.postech.ac.kr/oasis/surv/), in the case of worm survival, or Excel, in case of fly survival. Linear models and ordinal logistic regression were performed either in R or in JMP Pro 17.0.0 and each model had a full factorial design. ANOVA, post hoc tests, and t tests were performed in JMP Pro 17.0.0.