Previously, we showed that an m⁵C modification is introduced at position C2381 on the 26S rRNA of C. elegans large ribosomal subunit by NSUN-5 (Adamla et al., 2019; Schosserer et al., 2015), which is required to modulate animal lifespan and stress resistance (Schosserer et al., 2015). On this basis, we were interested to learn if other related rRNA methyltransferases in C. elegans might display similar properties.

Therefore, we investigated the RNA substrate of NSUN-1 (also formerly known as NOL-1, NOL-2, or W07E6.1) and its potential roles in worm development and physiology. NSUN-1 is a member of the NOP2/Sun RNA-methyltransferase family. Since there are only two known m⁵C residues on worm 26S rRNA (Sharma and Lafontaine, 2015; Trixl and Lusser, 2019), one of them at C2381, being installed by NSUN-5, we speculated that NSUN-1 might be required for introducing the second m⁵C residue at position C2982. Notably, both 26S m⁵C sites are localized close to the decoding site and peptidyl transferase center of the ribosome, and are highly conserved between yeast, worm and human (Figure 1A,B).

In order to test if NSUN-1 is involved in large ribosomal subunit m⁵C methylation, we first sought to identify a suitable model to study loss of NSUN-1. We selected the tm6081 allele which has a deletion in the 3’ untranslated region (3’ UTR) of nsun-1 (Figure 1—figure supplement 1A). After letting single hermaphrodites, which were heterozygous for tm6081 self-fertilize, we were unable to detect any viable offspring carrying the homozygous mutation (Figure 1—figure supplement 1B). This indicates that tm6081 is a recessive lethal mutation, which agrees with previous reports about lethality of nsun-1 depletion by egg-onset RNAi (Kamath et al., 2003; Piano et al., 2002). Since nsun-1 mRNA levels were only decreased by 40% in animals heterozygous for tm6081 (Figure 1—figure supplement 1C), we decided to use RNAi instead with a chance to achieve higher knock-down efficiencies. As it will become evident below, there are several other advantages of using RNAi in C. elegans. One is that it allows to deplete a factor of interest at a particular life stage only (e.g. in adult worms), another is that it allows performing tissue-specific knockdown of gene expression.

26S rRNA was purified from worms treated with siRNAs specific to NSUN-1-encoding mRNAs on sucrose gradients, digested to single nucleosides, and analyzed by quantitative HPLC. In our HPLC assay, the m⁵C nucleoside eluted at 12 min, as established with a m⁵C calibration control (data not shown). The depletion of NSUN-1 was conducted in two genetic backgrounds: N2 (wildtype), and NL2099 (an RNAi-hypersensitive strain due to mutation in rrf-3) (Figure 1—figure supplement 2). Treating N2 worms with an empty control vector, not expressing any RNAi, did not significantly reduce the levels of 26S rRNA m⁵C methylation (Figure 1C, 97% instead of 100%). Interestingly, treating N2 worms with an RNAi construct targeting nsun-1 led to a reduction of 26S rRNA m⁵C methylation by 35% (Figure 1C). In the NL2099 strain, nsun-1 RNAi treatment also led to a reduction of 26S rRNA m⁵C methylation by 26% (Figure 1D). A second independent biological replicate confirmed these findings (Figure 1—figure supplement 3).

As there are only two known modified m⁵C residues on worm 26S rRNA, and since one of them is introduced by NSUN-5 (Adamla et al., 2019; Schosserer et al., 2015), a complete loss of NSUN-1 activity was expected to result in a 50% decrease in m⁵C methylation. However, protein depletion achieved with RNAi is usually not complete. It is not clear why the level of m⁵C depletion was not higher in the RNAi-hypersensitive strain in comparison to the N2 strain; nonetheless, RNAi-mediated depletion of nsun-1 significantly reduced the levels of 26S rRNA m⁵C modification in both worm strains, thus, we conclude that NSUN-1 is responsible for 26S rRNA m⁵C methylation.

We analyzed the 26S rRNA m⁵C levels in a nsun-5 deletion strain as control (strain JGG1, Figure 1E). In this case, we observed a near 2-fold reduction in methylation (58% residual), as expected from the known involvement of NSUN-5 in modification at position C2381. When nsun-1 was additionally depleted by RNAi in the nsun-5 knockout animals, the level of 26S rRNA m⁵C was further reduced to 43%, again in agreement with our conclusion that NSUN-1 is responsible for methylating the second position, C2982.

Since our conclusion is based on depletion of nsun-1 to ~20% residual expression and not on a full gene knockout (Figure 1—figure supplement 2), we cannot exclude the formal possibility that NSUN-1 might not be the only m⁵C2982 writer in C. elegans. However, we consider this possibility to be highly unlikely because the combination of a knockout of Rcm1 with a catalytic mutation of Nop2 in yeast was sufficient to completely remove m⁵C from 25S rRNA (Sharma et al., 2013).

To further prove that NSUN-1 is not involved in C2381 modification, methylation levels at this position were specifically tested by Combined Bisulfite Restriction Analysis (COBRA) assay in animals depleted of nsun-1 or nsun-5. This method is based on bisulfite conversion of total RNA, followed by PCR amplification and restriction digest, yielding two bands in case of methylation at C2381 and three bands in case of non-methylation (Adamla et al., 2019). As expected, only nsun-5 depletion strongly reduced methylation at C2381, and there was no residual m⁵C2381 in the nsun-5 knockout strain, while nsun-1 RNAi had no effect on modification at this position (Figure 1F). Bisulfite sequencing is well-known to be sensitive to RNA secondary structure (Warnecke et al., 2002), which likely explains why, despite repeated attempts, we could not monitor modification at position C2982 by use of this technique.

In conclusion, NSUN-1 and NSUN-5 are each responsible for installing one m⁵C onto the worm 26S rRNA, with NSUN-1 being responsible for position C2982 and NSUN-5 for position C2381 under the bona fide assumption that indeed only two m⁵C positions are present as described (Sharma and Lafontaine, 2015).

Next, we investigated if knockdown of nsun-1 modulates healthy lifespan in a similar fashion as that described for nsun-5 (Schosserer et al., 2015). In order to achieve this aim, we depleted nsun-1 by RNAi in N2 wild-type animals starting from day 0 of adulthood and, quite surprisingly, we did not observe any extension of mean or maximum lifespan (Figure 2A, Table 1). Next, we evaluated the stress resistance of adult worms upon nsun-1 or nsun-5 depletion, as an increased health at an advanced age often improves resilience to adverse events (Lithgow et al., 1994). Indeed, depletion of either nsun-1 or nsun-5 increased resistance to heat stress compared to the RNAi control (Figure 2B). Furthermore, we tracked the movement of animals treated with either an empty vector control or two constructs expressing RNAi directed against nsun-1 or nsun-5 in a time course analysis, starting at day 1 of adulthood up to day 16. Interestingly, we observed increased average speed at day 8 of adulthood in both nsun-1 (+47.8%, p=0.009) and nsun-5 (+34.7%, p=0.073) depleted animals compared to the control, as well as at day 12 (nsun-1 RNAi: +10.2%, p=0.539; nsun-5 RNAi: +73.5%, p=0.008 compared to the control) (Figure 2C). Other timepoints remained unaffected. Thus, while nsun-1 knockdown does not extend lifespan, it improves two healthspan parameters, namely thermotolerance and midlife locomotion (Bansal et al., 2015; Rollins et al., 2017).

Intrigued that depletion of nsun-1 did not extend the lifespan of C. elegans in a similar fashion as that reported for nsun-5 when whole adult animals were treated with RNAi, we reasoned that performing tissue-specific depletion of nsun-1 might help us to further elucidate a possible effect on lifespan. We focused on the comparison of the germline and somatic tissues, because somatic maintenance and aging are evolutionarily tightly connected (Kirkwood and Holliday, 1979), and because signals from the germline modulate C. elegans lifespan (Hsin and Kenyon, 1999). In addition, only loss of soma- but not germline-specific eIF4E isoforms, which are central regulators of cap-dependent translation, extend nematode lifespan (Syntichaki et al., 2007). To test if nsun-1 has similar tissue-specificity, we made use of worm strains sensitive to RNAi only in either the germline or somatic tissues. This is achieved, on the one hand by mutation of rrf-1, which is required for amplification of the dsRNA signal in the somatic tissues (Kumsta and Hansen, 2012; Sijen et al., 2001), and, on the other hand by functional loss of the argonaute protein ppw-1 rendering the germline resistant to RNAi (Tijsterman et al., 2002). Interestingly, germline-specific nsun-1 RNAi had no effect on animal lifespan (Figure 2D, Table 1), but depletion of nsun-1 in somatic tissues reproducibly increased mean lifespan by ~10% (Figure 2E, Table 1).

In conclusion, both NSUN-1 and NSUN-5 m⁵C rRNA methyltransferases mildly affect thermotolerance and mobility of wild-type nematodes at midlife. Whole-animal nsun-5 depletion expands mean lifespan by 17% (Schosserer et al., 2015), and in contrast to this, a 10% lifespan extension is only detected after depletion of nsun-1 specifically in the somatic tissues.

The ‘disposable soma theory’ of aging posits that long-lived species exhibit impaired fecundity and reduced number of progenies. The proposed underlying cause is that energy is invested in the maintenance of somatic tissues rather than in rapid reproduction (Kirkwood and Holliday, 1979). In keeping with this theory, we expected that the absence of nsun-1 in somatic tissues, which increased longevity, may reduce fecundity. Therefore, we measured the brood size upon nsun-1 and nsun-5 depletion by RNAi. After reaching adulthood but prior to the egg-laying stage, worms were transferred to individual wells of cell culture plates containing NGM-agar and fed with bacteria expressing the specific RNAi or, as control, the empty vector. Egg production was impaired upon nsun-1 knockdown (reduced by 42%), but this was not the case upon nsun-5 knockdown (reduced by 2%) (Figure 3A). Egg production ceased rapidly after day one in all conditions (Figure 3—figure supplement 1A).

Thus far, all the experiments were performed on worms subjected to adult-onset nsun-1 knockdown, as animals depleted of nsun-1 during development were smaller and were infertile upon adulthood. To follow up on these observations, we measured mRNA expression levels of both m⁵C rRNA methyltransferases at different developmental stages including eggs, L1/L2 larvae, L3 larvae, L4 larvae, and young adults. RT-qPCR indicated that both nsun-1 and nsun-5 mRNA levels constantly increase during development (Figure 3B). The same observation applied to mRNA levels of nsun-2 and nsun-4 (Figure 3—figure supplement 1B), indicating that all four members of the NSUN-protein family might play important roles during development.

To further assess whether nsun-1 expression is indeed necessary for progressing faithfully through larval stages, we captured images of young adult animals subjected to larval-onset RNAi. The disparity in body size between RNAi control and nsun-1 RNAi was apparent, whereby nsun-1 depleted animals showed reduced length by approximately 20%. Interestingly, this reduced body size was not seen upon nsun-5 RNAi treatment (Figure 3C,D).

In addition, we imaged 3 day old animals using differential interference contrast (DIC) microscopy. Worms subjected to nsun-1 knockdown displayed morphological alterations; specifically, the gonad appeared severely distorted (Figure 3E). In contrast, RNAi control and nsun-5 RNAi showed comparable morphology of distal and proximal gonads (Figure 3E). Consequently, we hypothesized that nsun-1 RNAi-treated worms might be arrested in early L4 larval stage when the gonad is not yet fully developed and animals still grow, instead of normally reaching adulthood after 3 days like RNAi control or nsun-5 RNAi-treated nematodes. To test this possibility, we used the TP12 kaIs12[col-19::GFP] translational reporter strain, which expresses COL-19::GFP specifically upon reaching adulthood, but not during larval stages. Surprisingly, larval-onset RNAi against nsun-1 or nsun-5 did not reveal differences in the expression of COL-19::GFP as compared to RNAi control, suggesting that neither nsun-1 nor nsun-5 induce larval arrest (Figure 3F). Together with the reduced brood size upon adult-onset RNAi, these findings imply that loss of nsun-1 induces phenotypic changes in the reproductive organs of C. elegans.

Since knockdown of nsun-1 extended lifespan only when it was applied to somatic tissues, we hypothesized that body length might also be affected when these tissues are specifically targeted for depletion. To test this hypothesis, we depleted nsun-1 specifically in the germline or in somatic tissues using tissue-specific RNAi strains and measured body size during three consecutive days after adulthood was reached. While germline-specific knockdown of nsun-1 did not induce any changes in body size (Figure 3G), soma-specific knockdown revealed a decrease in body length by 23% on day 1, 11% on day 2 and 12% on day 3 compared to the RNAi control (Figure 3H), phenocopying nsun-1 depletion in wild-type animals after whole-animal RNAi. Similarly, no effect of nsun-1 knockdown was evident in another germline-specific RNAi strain, which was recently developed to enhance germline-specificity (Zou et al., 2019; Figure 3—figure supplement 1C).

In conclusion, nsun-1 but not nsun-5 depletion impairs body size and morphology of the gonad and leads to a significant reduction of brood size. Furthermore, these phenotypes are also observed when nsun-1 is specifically knocked-down in somatic tissues, but not when depleted in the germline only.

To further investigate the mechanisms underlying impaired fecundity upon nsun-1 knockdown, we analyzed the morphology of the gonad in nsun-1 depleted animals in more detail. The germline of adult hermaphrodites resides within the two U-shaped arms of the gonad, which contains germ cells at various stages of differentiation (Figure 4A). The gonad is sequentially developing from the proliferative germ cells near the distal tip cell, through the meiotic zone into the loop region, finally culminating in fully-formed oocytes in the proximal gonad (Pazdernik and Schedl, 2013). The limiting factor for fecundity in self-fertilizing hermaphrodites is sperm produced in the spermatheca (Hodgkin and Barnes, 1991).

Upon visualizing the germline cell nuclei with DAPI-staining, no oocytes were observed in worms after knockdown of nsun-1 in contrast to RNAi control treated animals (Figure 4B). The mitotic zone at the distal end of the gonad appeared normal in nsun-1 depleted animals, whereas oocyte production starting at the pachytene zone was hampered. Analysis of GFP::RHO-1 and NMY-2::GFP expressing worm strains, which specifically express GFP in the germline, confirmed our observations (Figure 4C, Figure 4—figure supplement 1). The gonads of control and nsun-5 RNAi-treated animals appeared normal, clearly depicting the different stages of in-utero embryo development, whereas the germline of nsun-1 RNAi-treated animals displayed a strikingly altered morphology. Importantly, this phenotype showed 100% penetrance in worms exposed to nsun-1 RNAi.

Since other phenotypes observed upon nsun-1 depletion were detected in soma- but absent from germline-specific RNAi-treated strains, we hypothesized that the somatic part of the gonad might specifically require NSUN-1 for normal oocyte production. Indeed, soma-specific depletion of nsun-1 phenocopied the distorted gonad morphology of wild-type animals exposed to nsun-1 RNAi (Figure 4D). Remarkably, upon germline-specific knockdown of nsun-1, the gonad appeared completely unaffected (Figure 4E; Figure 4—figure supplement 2).

Since the only known function of NSUN-1 and NSUN-5 is m⁵C methylation of rRNA, we reasoned that methylation-induced alterations of ribosome biogenesis and function might explain the observed phenotypes. Therefore, we tested if the presence of NSUN-1 or NSUN-5 is required for ribosomal subunit production and pre-rRNA processing. To this end, total RNA was extracted from worms treated with nsun-1 RNAi, separated by denaturing agarose gel electrophoresis and processed for northern blot analysis (Figure 5A,B). Again, two reference worm strains were used (N2 and NL2099). Upon nsun-1 knockdown, we observed a mild accumulation of the primary pre-rRNA transcript and of its immediate derivative, collectively referred to as species ‘a’ (Figure 5A; Bar et al., 2016; Saijou et al., 2004), as well as a mild accumulation of the pre-rRNAs ‘b’ and ‘c’’ (Figure 5A,B, see lane 1 and 3 as well as 5 and 6). Again, these findings were observed in both worm backgrounds tested, N2 and NL2099.

For comparison, we also analyzed rRNA processing in nsun-5 deletion worms (JGG1 strain) in presence and absence of NSUN-1 (nsun-1 RNAi in JGG1). In both cases, we noted an important reduction in the overall production of ribosomal RNAs (Figure 5B,C), with an apparent increase of rRNA degradation (seen as an increase in accumulation of metastable RNA fragments, in particular underneath the 18S rRNA). Furthermore, we observed that NSUN-1 is not required for mature rRNA production as shown by the unaffected levels of mature 18S and 26S rRNAs (Figure 5C). This was confirmed by determining the 26S/18S ratio, which was 1.0 as expected since both rRNAs are produced from a single polycistronic transcript (Figure 5C). The levels of the other two mature rRNAs (5S and 5.8S) were also unaffected (Figure 5—figure supplement 1). This behavior was shown in both worm backgrounds, N2 and NL2099, used. The overall decrease in mature ribosome production observed in nsun-5 deletion worms did not affect the ratio of mature ribosomal subunits (26S/18S ratio of 1.0) (Figure 5C). In agreement with the reduced amounts of 18S and 26S rRNA observed in nsun-5 deletion worms, total amounts of all precursors detected were reduced (Figure 5B). Analysis of low molecular weight RNAs by acrylamide gel electrophoresis revealed the absence of NSUN-5 to severely inhibit processing in the internal transcribed spacer 2 (ITS2), which separates the 5.8S and 26S rRNAs on large precursors. This was illustrated by the accumulation of 3’-extended forms of 5.8S, and of short RNA degradation products (Figure 5—figure supplement 1, see lanes 3 and 4). Depletion of nsun-1 partially suppressed the effect of nsun-5 deletion: the overall production of mature rRNA and, in particular, the amount of mature 26S rRNA was increased (ratio of 1.2) (Figure 5C). Consistently, the accumulation of 3’-extended forms of 5.8S and of short RNA degradation products was reduced (Figure 5—figure supplement 1).

In order to test if mature ribosomes of animals lacking any of the two m⁵C rRNA methyltransferases might be functionally defective, we analyzed global protein synthesis by incorporation of puromycin in N2 worms treated with either RNAi control, nsun-1 or nsun-5 RNAi. Worms were exposed to puromycin for three hours at room temperature. Following lysis, puromycin incorporation was measured by western blot with an anti-puromycin antibody (Figure 5D). Quantification of three independent experiments revealed no changes in global protein synthesis (Figure 5E). We also performed polysome profiling which provides a ‘snapshot’ of the pool of translationally active ribosomes. Comparison and quantification of profiles obtained from control and nsun-1 knockdown nematodes did not reveal any differences in the distribution of free subunits, monosomes, and polysomes (Figure 5F,G). This agrees with the absence of global protein translation inhibition in the metabolic (puromycin) labeling assay (Figure 5E).

In conclusion, NSUN-1 is neither required for pre-rRNA processing nor for global translation. On the contrary, the amounts of ribosomal subunits were reduced in the absence of NSUN-5. The ribosomal biogenesis alterations observed upon nsun-5 depletion result from a combination of processing inhibitions in ITS2 and increased rRNA intermediates turnover. However, global translation was not detectably affected.

Since depletion of nsun-1 did not affect global protein synthesis, we hypothesized that loss of 26S rRNA m⁵C methylation might modulate the translation of specific mRNAs, as was previously observed after Rcm1 (NSUN-5 homolog) depletion in yeast (Schosserer et al., 2015). To test this possibility, we isolated mRNAs contained in the polysomal fraction, systematically sequenced them by RNA-seq and compared their abundance in polysomes between animals subjected to nsun-1 RNAi versus RNAi control. Thereby, we identified 52 protein-coding mRNAs to be differentially associated with polysomes between the two conditions (p-adj: 0.05) Source data 1. From those, we selected the nine most upregulated and ten most repressed mRNAs based on the magnitude of their fold-change variation (Figure 6A,B) and, for comparison, measured their abundance also in polysomes of wild-type (N2) and nsun-5 knockout (JGG-1) animals by RT-qPCR (Figure 6C,D). In contrast to nsun-1 depletion by RNAi, knockout of nsun-5 did not significantly affect the presence in polysomes of any of these mRNAs. This suggests that the pattern of translated mRNAs upon nsun-1 depletion is highly specific and strikingly distinct from that observed upon loss of the other 26S rRNA m⁵C methyltransferase.

As changes in polysome abundance upon nsun-1 depletion can either be caused by transcriptional regulation or by specific recruitment of the respective mRNAs into translating ribosomes, we analyzed total mRNA levels in cell lysates from which then polysome fractionation was performed. We compared total intracellular mRNAs representing the transcriptome, to those contained in the polysomes representing the most actively transcribed mRNAs (in the following designated as ‘translatome’). We considered only protein-coding mRNAs with a minimum fold-change of 2 between translatome and transcriptome using an adjusted p-value cut-off at 0.05 (Figure 7A; Source data 2). We observed that many more mRNAs had their translation repressed (RNAi control: 599, nsun-1 RNAi: 536) than stimulated (RNAi control: 94, nsun-1 RNAi: 84). Since the composition of 3’ UTRs can affect translation (Tushev et al., 2018), we analyzed GC-content, length, and minimal free folding energy of all coding, promoted, and repressed mRNAs in our dataset (Figure 7B). Interestingly, all three features significantly differed between RNAi control and nsun-1 RNAi in promoted and repressed mRNAs (p<0.05), while they remained unchanged when analyzing all coding mRNAs present in our dataset. These findings suggest that loss of nsun-1 causes the translation of specific subsets of mRNAs based on the composition and length of their 3’ UTRs.

Interestingly, 3’ UTRs of mRNAs translationally repressed by nsun-1 depletion were exclusively and significantly enriched (p<0.001) for several binding motifs of ASD-2, GLD-1 and RSP-3 (Source data 3). All three RNA-binding proteins are known to play essential roles in C. elegans development (Lee and Schedl, 2010; Longman et al., 2000). Although gld-1 mRNA was significantly translationally repressed in both nol-1 RNAi and RNAi control animals by the same magnitude (log2 fold-change: −1.2, p-adj.<0.05, Source data 2), neither asd-2, rsp-3 or gld-1 mRNAs were differentially regulated between nol-1 RNAi and RNAi control in the transcriptome or in the translatome (Source data 1).

GLD-1 is particularly interesting, because levels are highest in the pachytene (also referred to as meiotic zone), where it acts as a translational repressor of mRNAs modulating oogenesis. At the transition zone between the pachytene and the diplotene, GLD-1 levels sharply decrease and previously repressed mRNAs are consequently translated (Lee and Schedl, 2010). Since the gonads of nsun-1 knockdown animals appeared defective precisely at this transition (Figure 4C) and GLD-1 target mRNAs were repressed (Source data 3), we set out to investigate the effects of nsun-1 depletion on the spatial distribution of GLD-1 expression during development. For this aim, we used a GLD-1::GFP reporter strain and exposed larvae to RNAi control and nsun-1 RNAi. Indeed, GLD-1::GFP protein expression was restricted exclusively to a small portion of the loop region in adult nematodes (Figure 7—figure supplement 1). Taken together, these findings suggest that nsun-1 is required for correct gonadal GLD-1 localization during development, which might then directly or indirectly influence the specific translation of mRNAs required for further steps in development based on motifs in their 3’ UTRs.

To further understand the mechanistic link between differential translation and the phenotypes observed upon nsun-1 knockdown, we performed GO-term enrichment analysis. Among others, GO-terms associated with collagens, structural integrity of the cuticle, and embryo development were significantly enriched amongst the mRNAs which were translationally repressed upon nsun-1 knockdown (Figure 7C, Source data 4).

Since three collagens (col-35, col-36, and col-37) were also found among the five most strongly repressed genes upon loss of nsun-1 (Figure 7A), we decided to assess whether collagen deposition is indeed altered in the animals. In order to test this possibility, we performed a specific histological staining aimed at distinguishing young collagen, detected in blue, from mature collagen, highlighted in pink/brownish-red (Herovici, 1963; Teuscher et al., 2019). While young adult worms exposed to RNAi control showed presence of both young and mature collagen, animals subjected to nsun-1 RNAi displayed a strikingly overall reduction of collagen deposition compared to the cytoplasmatic counter-stain (yellow) (Figure 7D).

Interestingly, we repeatedly observed an increased fraction of animals displaying gonad extrusion upon nsun-1 RNAi (see also Figure 4D), which might be caused by loss of cuticle structural integrity. To quantify this phenotype precisely, we classified mid-aged animals according to the grade of gonad extrusion which we defined to fall in either of three categories: (i) no visible signs of gonad extrusion, (ii) mild extrusion, or (iii) severe extrusion (Figure 7—figure supplement 2A). Upon nsun-1 depletion, 134 of 220 animals (~60%) showed mild to severe extrusion of the gonad (categories ii and iii), while no extrusion was observed in any of the 50 RNAi control nematodes tested (Figure 7E). To further evaluate the possible physiological consequences of altered collagen deposition, we tested cuticle barrier integrity. This assay is based on the principle that the Hochst33342 dye is membrane-permeable, but cuticle-impermeable. As previously described by Ewald et al., 2015, worms were grouped into four categories according to whether they were: (i) not permeable (absence of stained nuclei in the animal tail region), (ii) mildly permeable (<5 stained nuclei), (iii) permeable (5–10 stained nuclei), or (iv) highly permeable (>10 stained nuclei) (Figure 7—figure supplement 2B,C). Consistent with the reduced production of several collagens, nsun-1 RNAi caused cuticle permeability (categories ii, iii, and iv) in 26 of 46 animals (~56%) compared to only 5 of 51 RNAi control animals (~10%) (Figure 7F).

Taken together, this indicates that NSUN-1 is partially required for translation of several cuticle collagens, which may explain the loss of gonad integrity and the increased cuticle permeability observed upon nsun-1 depletion. Moreover, several mRNAs whose translation depends on NSUN-1 are associated with embryogenesis.

Worm strains used in this study are listed in the Key Resources Table. Worms were cultured following standard protocols on Escherichia coli OP50-seeded NGM-agar plates at 20°C, unless indicated otherwise (Brenner, 1974).

Genotyping was performed using the ExACT Genotyping Kit (BioCat). Single nematodes were picked from plates into 5 µL nuclease-free water in PCR tubes, taking care to minimize bacterial contamination. 1 µL Buffer A and 0.5 µL Buffer B were added to each tube and incubated at 75°C for 5 min and for 10 min at 95°C in a thermocycler with heated lid. Then, PCR was performed following the manufacture’s protocol using specific primers spanning the tm6081 allele (Supplementary file 1) in 25 µL reaction volume, 50°C annealing temperature and 35 cycles.

For inactivating nsun-1 and nsun-5, feeding of double-stranded RNA expressed in bacteria was used (Timmons et al., 2001 ). Therefore, the HT115 strain of E. coli, carrying either the respective RNAi construct or the empty vector (L4440) as RNAi control, was cultured overnight in LB medium with ampicillin and tetracyclin at 37°C. Bacteria were harvested by centrifugation, resuspended in LB medium and either 100 µL (60 mm plates) or 400 µL (100 mm plates) were plated on NGM containing 1 mM isopropyl-b-D-thiogalactoside and 25 µg/mL carbenicillin. The plates were incubated at 37°C overnight and used within one week.

Larval-onset RNAi was achieved by bleaching adult animals. Released eggs were transferred directly to plates seeded with RNAi bacteria. Adulthood was usually reached after three days and animals were used for experiments when the RNAi control strain started to lay eggs.

In case of adult-onset RNAi, eggs were transferred to plates seeded with RNAi control bacteria. Animals were raised until egg production commenced and subsequently transferred to the respective RNAi bacteria.

Worms were paralyzed using 1 M sodium azide solution and mounted on 2% agar pads. Images were acquired on a Leica DMI6000B microscope with a 10x dry objective (NA 0.3), a 20x dry objective (NA 0.7), or a 63x glycerol objective (NA 1.3) in DIC brightfield or fluorescence mode. Confocal microscopy presented in Figure 4C was performed on a Leica TCS SP5 spot scanning confocal microscope equipped with an HCX PL APO CS 40x/0.85 dry objective, HyD detector and Argon-laser. Cropping, insertion of scale bars and brightness and contrast adjustments were done with Image J (version 2.0.0-rc-65/1.51 w; Java 1.8.0_162 [64-bit]).

Animals were either synchronized by timed egg-lay (two replicates) or by hypochlorite treatment (one replicate) on RNAi control plates. When reaching adulthood, nematodes were transferred to RNAi plates. Every few days at regular intervals, plates were rocked in order to induce movement of animals and videos were subsequently recorded for one minute. Worms were transferred to fresh plates whenever necessary. At day 16 the vast majority of worms completely ceased movement, thus we did not include any later timepoints. Notably, we did not notice any obvious aversion behavior or elevated speed at young age upon nsun-1 or nsun-5 RNAi, which was previously shown to be present upon depletion of other components of the translational machinery (Melo and Ruvkun, 2012). Worm Lab version 4.1.1 was used to track individual animals and calculate the average speed.

Lifespan measurement was conducted as previously described (Schosserer et al., 2015). For lifespan assays, 90 adults per condition were transferred to plates seeded with the respective RNAi bacteria (control, nsun-1, nsun-5). Wildtype worms were pre-synchronized on NGM plates seeded with UV-killed OP50 bacteria. 50 adult worms were transferred to NGM plates and allowed to lay eggs for 15 hr; then the adult worms were removed. Synchronization by timed egg-lay was performed 72 hr after the pre-synchronization by transferring 350 gravid worms from the pre-synchronization to fresh NGM plates seeded with RNAi control bacteria and allowed to lay eggs for four hours. After 68 hr, 90 young adult worms per condition were placed on fresh NGM plates containing 5 mL NGM, 100 µL bacterial suspension and 50 μg FUdR. This day represents day 0 in the lifespan measurement. Worms were scored as ‘censored’ or ‘dead’ every two to four days. Nematodes were scored as ‘censored’ if they had crawled off the plate, were missing or died due to other causes than aging, such as gonad extrusion. Animals were transferred to fresh plates every 3–7 days depending on the availability of the bacterial food source. Lifespans were performed at 20°C. Kaplan-Meier survival curves were plotted and log-rank statistics were calculated.

Thermotolerance was assessed as previously described (Vieira et al., 2018). Animals were synchronized by hypochlorite treatment and released eggs were transferred to NGM plates seeded with RNAi control bacteria and kept at 20°C. After 48 hr, L4 animals were picked on RNAi control, nsun-1 or nsun-5 RNAi plates and exposed to RNAi for approximately three days (68 hr). Subsequently, plates were transferred to 35°C and scored every 1–2 hr for survival.

Worms were synchronized by hypochlorite treatment and incubated in liquid S-Basal medium overnight. On the following day, eggs/L1 were transferred to RNAi plates (RNAi control, nsun-1, and nsun-5 RNAi). Three days later, worms were transferred to agar pads and paralyzed using sodium azide and visualized using DIC microscopy (see above).

Worms were synchronized by treatment with hypochlorite solution and incubated in S-Basal at room temperature overnight. L1 larvae were subsequently transferred to NGM plates seeded with RNAi control bacteria. After 48 hr, L4 animals were transferred to individual wells of a 24-well plate seeded with the respective RNAi bacteria (HT115, nsun-1, nsun-5). Each well contained 1.5 mL of NGM-agar and 3 μL of bacterial suspension (1:2 dilution in S-Basal). Worms were transferred to a new well every day for four consecutive days and total progeny of individual animals was counted. Per condition and experiment, five worms were analyzed.

Puromycin incorporation was measured as previously described (Tiku et al., 2018) with minor modifications. Heat-inactivated OP-50 (75°C, 40 min) were provided as food source during pulse-labeling. As negative controls, RNAi control treated worms were used either without addition of puromycin or by pulse-labeling at 4°C instead of 20°C. Around 100 animals per condition were harvested for western blot analysis. Lysis was done directly in SDS loading dye (60 µM Tris/HCl pH 6.8, 2% SDS, 10% glycerol, 0.0125% bromophenol blue and 1.25% β-mercaptoethanol). Worms in SDS loading dye were homogenized with a pellet pestle for 1 min. Then, the samples were heated to 95°C and loaded on 4–15% Mini-PROTEAN TGX gels (BioRad) in Laemmli-Buffer (25 mM Tris, 250 mM glycine and 0.1% SDS). Protein bands were transferred to PVDF-membranes (Bio Rad) at 25 V and 1.3 A for 3 min. After blocking with 3% milk in PBS, the membrane was incubated overnight at 4°C with a mixture of anti-Histone H3 (Abcam ab1791, 1:4000) and anti-puromycin (Millipore 12D10, 1:10000). After washing and secondary antibody incubation (IRDye680RD and IRDye800CW, 1:10000), the membrane was scanned on the Odyssey Infrared Imager (LI-COR). Quantification of band intensities was performed in Image J (version 2.0.0-rc-65/1.51 w; Java 1.8.0_162 [64-bit]).

Two-day-old adult worms were used to generate polysome profiles as previously described (Rogers et al., 2011). One hundred microliter worm-pellet were homogenized on ice in 300 µL of solubilization buffer (300 mM NaCl, 50 mM Tris-HCl (pH 8.0), 10 mM MgCl₂, 1 mM EGTA, 200 µg/mL heparin, 400 U/mL RNAsin, 1.0 mM phenylmethylsulfonyl fluoride, 0.2 mg/mL cycloheximide, 1% Triton X-100, 0.1% sodium deoxycholate) using a pellet pestle. 700 µL additional solubilization buffer were added, vortexed briefly, and placed on ice for 10 min before centrifugation at 20.000 g for 15 min at 4°C. Approximately 0.9 mL of the supernatant was applied to the top of a linear 10–50% sucrose gradient in high salt resolving buffer (140 mM NaCl, 25 mM Tris-HCl (pH 8.0), 10 mM MgCl₂) and centrifuged in a Beckman SW41Ti rotor (Beckman Coulter, Fullerton, CA, USA) at 180.000 g for 90 min at 4°C. Gradients were fractionated while continuously monitoring the absorbance at 260 nm.

Trizol LS (Life Technologies) was immediately added to collected fractions and RNA was isolated following the manufacturer’s protocol. PolyA-selection, generation of a strand-specific cDNA library and sequencing on the HiSeq 4000 platform (Illumina) using the 50 bp SR mode was performed by GATC Biotech (Konstanz, Germany). At least 30 million reads were generated per sample.

FASTQ Trimmer by column (Galaxy Version 1.0.0) was used to remove the first 12 bases from the 5’ end of each read due to an obvious base bias in this region, as detected by FastQC (Galaxy Version 0.69). Filter by quality (Galaxy Version 1.0.0) was performed using a cut-off value of 20 and only reads with a maximum number of 8 bases with quality lower than the cut-off value were retained. RNA STAR (Galaxy Version 2.6.0b-1) was used to align reads to the WBcel235 reference genome using the default options. Aligned reads with a minimum alignment quality of 10 were counted using htseq-count (Galaxy Version 0.9.1).

Differential expression was analyzed using the DEseq2 package in R. The contrast 

~batch + condition (batch = biological replicate, condition = sample description)

was applied to compare the polysome fraction to the total RNA of either RNAi control or nsun-1 RNAi-treated samples. Afterwards, results were filtered in R to contain only protein-coding genes (according to ENSEMBL annotation), genes with detectable expression (base mean >1), a fold-change of >2 (log2FC > 1) and an adjusted p-value of <0.05. Volcano plots were generated in R using the Enhanced Volcano package, labeling the top five up- and down-regulated genes respectively.

GO-term enrichment using DAVID (version 6.7) was performed as previously described (Rollins et al., 2019). Only protein-coding genes with detectable expression (base mean >1), a fold-change of >2 and an adjusted p-value of <0.10 were considered. For visualization, only the broadest GO-terms of the GOTERM_BP_FAT, GOTERM_MF_FAT and GOTERM_CC_FAT categories, which were still significantly enriched (FDR < 0.05), are shown while similar terms based on the same subset of genes but lower in hierarchy were manually removed. Full results are contained in the supplements.

UTR characterization and RBP motif enrichment were performed as previously described (Rollins et al., 2019) using only protein-coding genes with detectable expression (base mean >1), a fold-change of >2 and an adjusted p-value of <0.10.

The raw and processed sequencing data are available from the Gene Expression Omnibus database (https://www.ncbi.nlm.nih.gov/geo) under accession GSE143618.

The R-script for analyzing RNA-seq data is provided as Source code 1.

All primer sequences are provided in Supplementary file 1. Samples were collected by either transferring worms individually into 1.5 mL tubes, by washing them off NGM plates using S-Basal or from polysome fractions and lysates (see above). After three washing steps with S-Basal, 300 μL TRIzol LS Reagent were added to approximately 100 µL residual S-Basal including worms. Subsequently, worms were homogenized with a pellet pestle for one minute, 600 µL TRIzol LS Reagent were added and the sample was vortexed for five minutes at room temperature. Total RNA was isolated using Direct-zol RNA MiniPrep Kit (Zymo) according to the instructions by the manufacturer. For cDNA synthesis, 500 ng RNA were converted into cDNA using the Applied Biosystems High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). cDNA was amplified from total RNA using random primers. RT-qPCR was performed on a Rotor-Gene Q (QIAGEN) using HOT FIREPol EvaGreen qPCR Mix. The absolute amounts of mRNAs were calculated by computing a standard curve and the resulting copy numbers were normalized to the housekeeping genes act-1 and tba-1.

For measuring mRNA expression during development, worms were synchronized by treatment with hypochlorite solution and the released eggs were subsequently transferred to four separate NGM plates seeded with UV-killed OP50 bacteria. Samples were taken from eggs immediately after bleaching, L1/L2 (20 hr after bleaching), L3 (32 hr after bleaching), L4 (46 hr after bleaching) and young adults (60 hr after bleaching).

The PyMOL Molecular Graphics System (Version 2.0) was used. The structure was modeled on the human 80S ribosome (PDB 6EK0).

NSUN-5 activity was measured by the COBRA assay as previously described (Adamla et al., 2019). Primer sequences are provided in Supplementary file 1.

13–15 µg 26S purified on sucrose gradient were digested to nucleosides and analyzed by HPLC. Peaks elutes at 12 min and as a control, a commercial 5-methylcytidine (NM03720, CarboSynth) was used. For quantification of m⁵C peak area, the peak was normalized to either the peak eluting at 16 min (asterisk on the Figure), or to the peak eluting at 8 min (U), with similar results. The results are shown for normalization to the peak eluting at 16 min.

For analysis of high–molecular weight RNA species, 3 µg total RNA was resolved on a denaturing agarose gel (6% formaldehyde/1.2% agarose) and migrated for 16 hr at 65 volts. Agarose gels were transferred by capillarity onto Hybond-N+ membranes. The membrane was prehybridized for 1 hr at 65°C in 50% formamide, 5x SSPE, 5x Denhardt’s solution, 1% SDS (w/v) and 200 µg/mL fish sperm DNA solution (Roche). The ³²P-labeled oligonucleotide probe (LD2648 (ITS1): CACTCAACTGACCGTGAAGCCAGTCG; LD2649 (ITS2): GGACAAGATCAGTATGCCGAGACGCG) was added and incubated for 1 hr at 65°C and then overnight at 37°C. For analysis of low molecular weight RNA species, northern blots were exposed to Fuji imaging plates (Fujifilm) and signals acquired with a Phosphorimager (FLA-7000; Fujifilm).

No explicit power analysis was used. Sample sizes estimations were partially based on our own previous empirical experience with the respective assays, as well as the cited literature.

No systematic blinding of group allocation was used, but samples were always analyzed in a random order. Nematodes were randomly assigned to the experimental groups. All lifespan, stress resistance and locomotion experiments were performed by at least two different operators.

Most experiments were performed in three independent experiments, unless stated otherwise in the figure legend. Independent experiments were always initiated at different days and thus always resemble different batches of nematodes. Some experiments (RNA isolation for RNA-seq, pre-rRNA processing analysis) were performed once with all frozen independent batches of nematodes to minimize technical variation. No outliers were detected or removed. Criteria for censoring animals for lifespan, stress resistance, and locomotion experiments are indicated in the respective chapters.

Statistical tests used, exact values of N, definitions of center, methods of multiple test correction, and dispersion and precision measures are indicated in the respective figure legends. P-value thresholds were defined as *p<0.05, **p<0.01, and ***p<0.001. For RNA-seq, statistical tests and p-value thresholds are explained in detail in the ‘RNA-seq’ chapter.