A Long Noncoding RNA on the Ribosome Is Required for Lifespan Extension

The biogenesis of ribosomes and their coordination of protein translation consume an enormous amount of cellular energy. As such, it has been established that the inhibition of either process can extend eukaryotic lifespan. Here, we used next-generation sequencing to compare ribosome-associated RNAs from normal strains of Caenorhabditis elegans to those carrying the life-extending daf-2 mutation. We found a long noncoding RNA (lncRNA), transcribed telomeric sequence 1 ( tts-1 ), on ribosomes of the daf-2 mutant. Depleting tts-1 in daf-2 mutants increases ribosome levels and signiﬁcantly shortens their extended lifespan. We ﬁnd tts-1 is also required for the longer lifespan of the mitochondrial clk-1


In Brief
Essers et al. find a long noncoding RNA, transcribed telomeric sequence 1 (tts-1), on ribosomes of C. elegans carrying the life-extending daf-2 insulin receptor mutation as well as the clk-1 mitochondrial mutant. They then demonstrate that this RNA is required for the life-extension phenotypes and that its depletion results in increasing ribosome levels.

INTRODUCTION
The aging of eukaryotes depends on a number of genetic and environmental factors. One of the best-studied pathways controlling lifespan is the highly conserved insulin/IGF-1 pathway. Within this pathway is the well-known daf-2 mutation in the insulin receptor, which in Caenorhabditis elegans results in a 2-to 3-fold extension of life (Kenyon, 2010). This increase in insulin/ IGF-1-mediated lifespan is dependent on the Forkhead transcription factor DAF-16 inducing the expression of a number of stress-resistance genes as well as an increase of the cellular self-digestion and recycling process of autophagy (Kenyon et al., 1993;McElwee et al., 2003;Melé ndez et al., 2003;Murphy et al., 2003). Other longevity pathways mechanistically distinct from insulin signaling include the mutation of mitochondrial path-ways controlling respiration and physiological timing, the inhibition of protein translation through reducing ribosomal proteins or inhibition of the mammalian target of rapamycin (mTOR) pathway, and dietary restriction that is also thought to function through mTOR (Blagosklonny, 2010;Hansen et al., 2007;Jonassen et al., 1998;Kenyon, 2010;Hekimi, 1996, 1998;Wong et al., 1995). The loss of ribosome elements such as ribosomal proteins in the mitochondria leads to decreased respiration and increased lifespan in C. elegans, while in S. cerevisiae, the depletion of ribosomal proteins in the cytoplasm also increases lifespan (Houtkooper et al., 2013;Steffen et al., 2008). Taken together, the data indicate that the reduction of ribosome levels and/or the deceleration of protein translation are ways to extend lifespan.
We and others recently determined that the C. elegans proteome of the daf-2(e1370) mutant (hereafter referred to as daf-2) reveals a dramatic reduction of ribosomal proteins, mRNA processing components, and protein metabolism factors compared to N2 wild-types (Depuydt et al., 2014;Stout et al., 2013). In turn, both studies using different experimental approaches found that the rate at which de novo proteins are synthesized is significantly reduced in daf-2 mutant cells. However, each study also found that the total amount of protein in the daf-2 mutants remains equivalent to wild-types, likely reflecting concurrent reductions in protein metabolism (Depuydt et al., 2014;Stout et al., 2013).

RESULTS
We hypothesized that the daf-2 reduction of protein synthesis would be reflected by changes in ribosome-associated RNAs. In order to compare these RNA subsets, we took a nonbiased approach by performing next-generation sequencing on the RNAs isolated from the monosomal or polysomal fractions (hereafter referred to as ''ribosomal'') of N2, daf-2, and daf-2;daf-16 strains normalized to total protein amounts and separated over sucrose density gradients (experimental setup shown in Figure S1A). The expected reduction of the ribosome profile peaks specifically in the daf-2 mutants compared to wild-type levels was consistent with our previous findings (Stout et al., 2013). Figure S1B illustrates the relative proportion of different subsets of RNAs that we found in the ribosomal fractions of N2, daf-2, and daf-2;daf-16 strains. Tables S1 and S2 reveal that the mRNAs found in the ribosomal fractions of the N2 and daf-2;daf-16 strains largely code for proteins important for biological processes such as growth, development, the cell cycle, and reproduction, while the same fractions of the daf-2 mutants reveal mRNAs that largely code for proteins involved in aging and stress response. Gene Ontology (GO) analysis of gene functions using Database Annotation, Visualization, and Integrated Discovery (DAVID) in Figure S1C confirmed these biological functions to be consistent with other studies measuring transcriptional changes in the daf-2 mutant (Halaschek-Wiener et al., 2005;Murphy, 2006).
In addition to the differential enrichment of many mRNAs on daf-2 ribosomes, we also found a long noncoding RNA (lncRNA), transcribed telomeric sequence 1 (tts-1), highly expressed in ribosomal fractions of daf-2 cells, but not in those of N2 or daf-2;daf-16. The tts-1 lncRNA is transcribed to two different isoforms of 711 or 659 bp long from a gene found on chromosome X that is not conserved in any other species and has almost no homology with other C. elegans genes. The G/C content of the tts-1 transcript is very low, calculated to be 34% for the total length of the transcript and dropping to less than 15% at the final 150 bp of the 3 0 end. A low-affinity cyclic-AMP responsive element (TGATGTCA) lies 728 nt upstream of the tts-1 transcription start site. Figure S1D illustrates the location of the RNA-seqencing mapped read densities with an Integrative Genomics Viewer (Robinson et al., 2011). Serial analysis of gene expression (SAGE) previously found that compared to wild-type strains, tts-1 is one of the most upregulated transcripts in the daf-2 mutants as well as in the developmentally arrested and longer-lived C. elegans in dauer formation (a type of stasis resulting from unfavorable environments that permits survival under harsh conditions) (Halaschek-Wiener et al., 2005;Jones et al., 2001). The expression of tts-1 is also significantly upregulated in C. elegans subject to attack by Gram-positive bacterial pathogens, a situation that slows the growth and increases constipation of the worms (O'Rourke et al., 2006). However, the function of the tts-1 lncRNA and its role in the longevity and immunity programs of C. elegans are, to date, entirely unknown.
We first validated the increase of tts-1 expression in daf-2 mutants compared to the levels in N2s and daf-2;daf-16 mutants using qPCR analysis of cDNA generated from isolated total RNA. The results of tts-1 expression are shown compared to expression of the pmp-3 housekeeping gene. This gene has been previously validated as an optimal reference gene for qPCR in C. elegans (Zhang et al., 2012) and remains unchanged in our next-generation sequencing data sets. Figure 1A shows that with the expression of pmp-3 set to 1, there is almost no detectable tts-1 expression in N2s or daf-2;daf-16 mutants. This is compared to an approximately 2-fold increase of expression of tts-1 over pmp-3 in the daf-2 mutants, confirming previous reports of high tts-1 expression in these mutants and also suggesting this expression is at least partly daf-16 dependent. Further qPCR results comparing the levels of tts-1 on N2 and daf-2 monosomes and polysomes reveal that compared to pmp-3, tts-1 is enriched over 12-fold in the daf-2 monosomal fraction and over 4-fold in the daf-2 polysomal fraction, with again almost no detectable tts-1 found in either fraction of the N2 strain (Figure 1B). These data suggest that tts-1 is not only expressed at much higher levels in daf-2 mutants but also preferentially A B C Figure 1. tts-1 lncRNA Expression Is High in the Ribosomal Fractions of daf-2 Mutants (A) qPCR analysis of tts-1 levels compared to the pmp-3 housekeeping gene in total RNA isolated from N2, daf-2, and daf-2;daf-16 strains. The expression of pmp-3 is set to 1. (B) qPCR analysis of tts-1 expression compared to pmp-3 RNA isolated from monosomal and polysomal fractions of N2 and daf-2 strains. The expression of pmp-3 is set to 1. (C) Fluorescent in situ hybridization of tts-1 probes (red) in N2 or daf-2 mutants. The white box in the top panel is enlarged in the panel beneath. DAPI (blue) is used for nuclear staining. *p < 0.01. See also Figure S1. enriched on ribosomes. Fluorescent in situ hybridization analysis (FISH) with 11 different probes against tts-1 confirmed the substantially higher expression of tts-1 in daf-2 mutants compared to N2 ( Figure 1C). Moreover, it reveals that tts-1 is uniformly expressed in the cytoplasm and nuclei of cells in the intestine of the daf-2 mutant, which is in line with the prominent role of the C. elegans intestine in the regulation of insulin/IGF-1 lifespan (Libina et al., 2003). To confirm that tts-1 is in fact ribosome bound and not merely contaminating the polysomal fractions of the sucrose gradients, we measured levels of tts-1 in the polysomal fractions of untreated daf-2 mutants compared to those treated with puromycin. The puromycin-induced dissociation of polysomes is confirmed by profiles revealing a reduction of polysome peaks and a widening of the monosome peak in Figure S1E. This dissociation of polysomes results in a reduction of both pmp-3 and tts-1 levels in the polysomal fraction, indicating that the expression of tts-1 in the polysomal fractions is not merely a contaminant ( Figure S1F).
To understand the function of the tts-1 lncRNA in the context of lifespan extension, we designed a double-stranded small interfering RNA (siRNA) construct against tts-1 in the L4440 vector that we expressed in HT115 E. coli and then fed to the worms.
Compared to daf-2 mutants fed bacteria expressing the empty L4440 vector, the bacteria expressing the tts-1 siRNA were successful at reducing the levels of tts-1 in daf-2 mutants by over 90% ( Figure S2A). This siRNA of tts-1 significantly shortened the extended lifespan of the daf-2 mutants (Figures 2A and S2B). Importantly, as opposed to inducing toxicity, tts-1 is regulating daf-2 lifespan specifically, as neither wild-type nor daf-2;daf-16 nematodes reveal any changes in lifespan upon exposure to the tts-1 siRNA (Figures 2A and 2B). BLAST results of the siRNA sequence revealed no significant homology between the construct and any other gene in the C. elegans genome except for tts-1 itself, suggesting a low probability of any off-target effect of the siRNA that may negatively affect daf-2 lifespan (Kamath et al., 2001. These results reveal a necessary role of tts-1 in the extension of the daf-2 mutant lifespan. We next examined the effect of tts-1 depletion on the polysome profiles of daf-2 mutants normalizing the lysate on the gradients to total protein levels. Consistent with what we previously reported (Stout et al., 2013), the profiles of the daf-2 mutants reveal low levels of ribosomes ( Figure 3A). These results are also in line with the reduction of the total number of ribosomal proteins that both proteomic studies of the daf-2 mutants revealed earlier ( that knocking down the expression of tts-1 in the daf-2 mutants results in the polysome peak sizes returning to more wild-type levels, suggesting an increase of ribosome levels ( Figures 3B  and 3C). In order to evaluate if tts-1 loss shifted the location of ribosomes from polysomal to nonpolysomal fractions in the den-sity gradient, which would give an indication about the overall level of protein translation, we then normalized the lysates to cytoplasmic rRNA levels. Here, we detect no difference in the polysome peak sizes ( Figure S3A). Moreover, we do not observe any differences in the area under the curves of the polysomal compared to the nonpolysomal fractions ( Figure S3B). Taken together, these data suggest that while tts-1 loss in the daf-2 mutants increases ribosome levels, it does not affect the overall rate at which these ribosomes translate protein.
In order to assess the stoichiometry of the tts-1 lncRNA relative to the number of ribosomes, we isolated monosomal RNA from daf-2 mutants and compared the expression of tts-1 to 18S rRNA by qPCR analysis. We calculated that for every tts-1 transcript in the monosomal fraction, there are 116,000 ± 10,000 (n = 3) 18S rRNA transcripts. This suggests that while tts-1 may be highly expressed in the daf-2 mutants, it is not acting simply to compete with all other mRNAs for occupancy on the ribosome.
We next asked if the increase of tts-1 expression and reduction of ribosomes is unique to the insulin/IGF-1 longevity pathway. For these experiments, we used clk-1(qm30) mutants (which carry a mitochondrial pathway mutation that reduces respiration and decreases ubiquinone biosynthesis) and eat-2(ad465) mutants (models of dietary restriction with impaired pharynxes and defective feeding behavior) (Jonassen et al., 1998;Hekimi, 1996, 1998;Wong et al., 1995). Both of these mutants have an established longer lifespan Hekimi, 1996, 1998). qPCR analysis relative to pmp-3 expression on the total RNA isolated from these mutants reveals an increase of tts-1 expression in both strains compared to N2, with much higher levels of tts-1 found in the clk-1 mutants compared to eat-2 ( Figure 4A). Correlating with these results, the profiles of both mutants compared to N2 strains reveal a far more dramatic reduction of ribosome levels in the clk-1 mutants (Figures S4A and S4B). Further in line with this, we find that the depletion of tts-1 in the clk-1 mutants results in a substantial shortening of their longer lifespan (p < 0.0001) and a marginal yet not nearly as significant shortening of the eat-2 mutant lifespan (p = 0.02) ( Figures 4B and 4C). The difference between the effects of tts-1 depletion on the lifespans of the clk-1 (and the daf-2) versus the eat-2 mutants moreover supports the specificity of the RNAi and suggests that the changes in lifespan are not due to off-target effects. All of the controls and statistical analysis for this assay are shown in Figures S4C and S4D.
Although we did attempt to construct a C. elegans strain overexpressing tts-1 in the N2 genetic background, we were unsuccessful in establishing an integrated line that expressed tts-1 at physiological levels or that did not reveal toxicity (data not shown). Thus, whether tts-1 is sufficient to extend lifespan remains to be determined.

DISCUSSION
Long noncoding RNAs were until recently thought to exist and function predominantly in the nucleus. It is now fast becoming realized that they effusively associate with cytosolic ribosomes (van Heesch et al., 2014;Wilson and Masel, 2011). Several functions for short noncoding RNAs (<20 bp) bound to ribosomes A B C Figure 4. The Effects of tts-1 Depletion on clk-1 and eat-2 Mutant Lifespans (A) qPCR analysis of tts-1 expression in clk-1 and eat-2 mutants compared to pmp-3 expression. The expression of pmp-3 is set to 1. (B) Longevity curve of clk-1 mutants fed bacteria expressing empty L4440 vector or L4440 expressing tts-1 siRNA. (C) Longevity curve of eat-2 mutants fed bacteria expressing empty L4440 vector or L4440 expressing tts-1 siRNA. *p < 0.01, **p < 0.02. See also Figure S4.
have been described, such as those that derive from both mRNAs and tRNAs and function as stress-induced inhibitors of protein translation (Ivanov et al., 2011;Pircher et al., 2014;Sobala and Hutvagner, 2013). Also recently a function for the ribosome-bound long intergenic noncoding RNA p21 (lincRNA-p21) was found to selectively repress the translation of JUNB and CTNNB1 mRNAs (Yoon et al., 2012). It is thus becoming clear that ncRNAs, both short and long, are playing roles in protein translation that are only beginning to be fully appreciated.
We are unable to definitively state that the tts-1 lncRNA does not code for protein. We do not find any protein sequences in http://www.wormbase.org or the NCBI database that corresponded to potential open reading frames of tts-1. Moreover, we do not detect any corresponding peptides in our proteomics study (Stout et al., 2013). It may of course be that any synthesized peptides are too small in size or short in half-life to be detected by current proteomic methods. Thus, it remains an open question as to the protein-coding potential of lncRNAs on the ribosome, reflected by a number of conflicting recent reports (Guttman et al., 2013;Niazi and Valadkhan, 2012;Smith et al., 2014).
The strong effect of tts-1 depletion on the longevity phenotype of the daf-2 and clk-1 mutants, but not the eat-2 mutants, is curious, especially given that the clk-1 mutation is known to be daf-16 independent (Lakowski and Hekimi, 1996). It is known that AAK-2, the C. elegans homolog of AMP-activated kinase subunit a, functions as a sensor of energy levels and is activated in conditions with high AMP:ATP ratios (Apfeld et al., 2004). Both clk-1-and daf-2-extended lifespans are dependent on AAK-2 in a pathway that is not shared by eat-2 (Curtis et al., 2006). Recently, it was shown that CRTC-1 (the sole C. elegans cyclic-AMP response element binding protein [CREB]-regulated transcriptional coactivator) interacts with the CREB homolog-1 transcription factor (CRH-1) and is directly activated by AAK-2 (Mair et al., 2011). As mentioned previously, a low-affinity cyclic-AMP responsive element lies 728 nt upstream of the tts-1 transcription start site. It is therefore possible that AMP:ATP levels are a driver of tts-1 transcription, which would also account for the high tts-1 expression in animals attacked by Gram-positive pathogens as they respond to the invasive stress and increase the AMP:ATP ratio (Hardie, 2011).
The precise mechanism of the tts-1 lncRNA remains to be determined. One intriguing possibility is that it is specifically regulating the translation of ribosomal protein mRNAs. Supporting this notion is the observation that despite the marked reduction of ribosomal proteins in the daf-2 mutant proteome, expression levels of ribosomal protein mRNAs in the daf-2 mutants are actually higher than in wild-types (Depuydt et al., 2014;Halaschek-Wiener et al., 2005). This suggests that a specific block of ribosomal protein gene expression at the level of translation is imposed in mutants undergoing lifespan extension, and we believe this will be an interesting area of future study.
In sum, we propose that the tts-1 lncRNA is able to reduce ribosome levels in a manner that is necessary for lifespan extension. Since many recent reports demonstrate that both genetic and pharmacological manipulations of the translation machinery can extend longevity in eukaryotes, our study puts lncRNAs forward as a compelling area in the field of aging research.
Puromycin Treatment daf-2 nematodes were synchronized and grown on NGM OP50 plates at 15 C until the L4 stage. They were collected and incubated overnight at 25 C in liquid NGM medium including OP50 bacteria, 0.1% Triton X-100, and 50 mg/ml puromycin (Sigma Aldrich).

RNAi
The last exon of tts-1 was synthesized by Eurofins with BglII and NcoI restriction enzymes sites on the 5 0 or 3 0 end, respectively, and cloned into the L4440 vector (Addgene) using the same enzymes. This vector was transformed into HT115 E. coli. Positive clones were selected by Sanger sequencing and grown overnight at 37 C in Luria broth with 50 mg/ml ampicillin and 0.2 M isopropyl b-D-1-thiogalactopyranoside (IPTG) before seeding.

Synchronization
For experiments, nematodes were synchronized by bleaching and allowed to hatch overnight in M9 buffer (Eisenmann, 2005). The L1 arrested larvae were plated onto NGM OP50 plates or plates inoculated with tts-1-specific RNAi bacteria and grown at 15 C until L4 stage, at which point they were shifted to 25 C overnight (for daf-2, daf-2;daf-16, and controls only).

Lifespan Analysis
Lifespan analysis was performed as previously described (Stout et al., 2013) on 35 mm NGM plates including FUdR (Sigma-Aldrich), 50 mg/ml ampicillin, and 0.2 M IPTG inoculated with the gene-specific siRNA bacteria of interest (Hansen et al., 2005). Lifespan curves and the associated statistics were analyzed using GraphPad Prism software and a Mantel-Haenszel test. The N2 and daf-2 ± tts-1 siRNA lifespan curves were performed in biological duplicate.

Peak Calculations
The polysome peak heights and monosomal peak widths were measured from the lower left most point of the peak curve using ImageJ software. Statistics were performed using a Student's t test. The area under the polysome peaks was calculated using R software (http://www.r-project.org/).

Polysome Profiling
Polysome profiling was performed as previously described (Pereboom et al., 2011). The concentration of protein in the lysates was measured with a Bradford reagent (Bio-Rad), and the cytoplasmic ribosome particles were measured by 260 nm optical density readings. Either an equal amount of total protein or equal levels of cytoplasmic rRNA were loaded onto the sucrose gradients for every experiment. Fractions were collected using a Foxy Jr Fraction Collector (Teledyne ISCO).

RNA-Sequencing Analysis
Monosomal and polysomal fractions from two experiments were pooled and RNA was extracted using TRIzol LS (Invitrogen) according to the manufacturer's protocol. For each condition, two libraries were constructed from RNA isolated from two separate experiments using the SOLiD Total RNA-Seq Kit (Life Technologies) and analyzed on the SOLiD platform. The sequence reads were mapped against the genome assembly WBcel215. Using cufflinks, we identified any possible new transcripts through reference annotated base transcript (RABT) assembly (Trapnell et al., 2013). Subsequently differential expression of transcripts was determined using cuffdiff across all pairs (Trapnell et al., 2013).

ACCESSION NUMBERS
The next-generation sequencing data reported in this paper have been deposited to the European Nucleotide Archive and are available under accession number PRJEB8029.