Nutrient control of growth and metabolism through mTORC1 regulation of mRNA splicing

SUMMARY

Cellular growth and organismal development are remarkably complex processes that require the nutrient-responsive kinase mTORC1. Anticipating that important mTORC1 functions remained to be identified, we employed genetic and bioinformatic screening in C. elegans to uncover mechanisms of mTORC1 action. Here we show that during larval growth nutrients induce an extensive reprogramming of gene expression and alternative mRNA splicing by acting through mTORC1. mTORC1 regulates mRNA splicing and production of protein-coding mRNA isoforms largely independently of its target p70S6K, by increasing the activity of the serine/arginine-rich (SR) protein RSP-6 (SRSF3/7) and other splicing factors. mTORC1-mediated mRNA splicing regulation is critical for growth, mediates nutrient control of mechanisms that include energy, nucleotide, amino acid, and other metabolic pathways, and may be conserved in humans. Although mTORC1 inhibition delays aging, mTORC1-induced mRNA splicing promotes longevity, suggesting that when mTORC1 is inhibited enhancement of this splicing might provide additional anti-aging benefits.

eTOC Blurb

Ogawa et al. show in C. elegans that the essential growth regulator mTORC1 extensively reprograms gene expression and alternative mRNA splicing. This splicing regulation is critical for growth, orchestrates growth-associated metabolism, and appears to be conserved. Maintaining mTORC1-induced splicing may be beneficial for longevity under conditions of mTORC1 inhibition.

Graphical Abstract

INTRODUCTION

During growth, cellular macromolecules and organelles must be produced in a regulated fashion in response to growth and nutrient signals. This process involves a broad remodeling of gene expression, as revealed by analyses of starved cells that are released from arrest by growth factor treatment^1–5. A cascade of gene-specific transcription factors initiates this remodeling, but it has remained undetermined whether more comprehensive levels of control or coordination with nutrient signals might be involved. This gene expression program is critical not only in normal growth and development, but also in cancer cell growth.

The evolutionarily conserved serine/threonine kinase complex mTORC1 (mechanistic target of rapamycin complex 1) controls various processes that are critical for growth^6,7. mTORC1 integrates environmental signals that include amino acid and growth factor availability, with the former relayed in large part through the Rag GTPases (RagA-RagD)^8,9. In response, mTORC1 increases the biosynthesis of proteins, lipids, nucleotides, and other macromolecules, and inhibits autophagic turnover of cellular structures and lysosomal hydrolysis of membrane lipids^6,7,10–13. While mTORC1 is essential for tissue development and regeneration, its hyperactivation has been implicated in cancer and other diseases^6,7. mTORC1 is also important in aging, with suppression of its activity increasing lifespan from yeast to mice^14–16. To understand the actions of mTORC1 in these settings, it will be necessary to gain a complete appreciation of the range of mechanisms that mTORC1 controls in vivo.

Studies in yeast and mammalian cells have identified mechanisms through which mTORC1 regulates cellular processes⁷. mTORC1 increases protein synthesis by phosphorylating eIF4E-Binding Protein (4EBP) and p70 S6 kinase (S6K)^6,7,10. Phosphorylation by mTORC1 activates S6K, which in turn phosphorylates translation regulators and other targets^6,7,17. Among the latter, S6K promotes the nuclear localization of serine/arginine-rich (SR) protein kinase 2 (SRPK2), which phosphorylates and activates some SR family proteins, mRNA splicing regulators that ensure accurate splice site selection^18–22. Through this pathway, S6K reduces intron inclusion at certain lipid biosynthesis genes and promotes lipogenesis^17,23. mTORC1 acts through the transcription factors SREBP1 (sterol regulatory-element binding protein 1), HIF1α (hypoxia-inducible factor 1), Myc, TFEB (transcription factor EB), and ATF4 (activating transcription factor 4) to regulate biosynthetic metabolic pathways^7,11,13. Illustrating the potential complexity of mTORC1 functions, phosphoproteomic analyses indicate that mTORC1 alters the phosphorylation of hundreds of proteins^24–30. However, for the most part, the extent to which mTORC1 might influence the functions of these downstream proteins and mechanisms has not been elucidated.

Here we combined genetic and bioinformatic screening in C. elegans to search for mechanisms that are controlled by mTORC1 during growth and development. We determined that the response to nutrients involves a widespread reprogramming of gene expression and alternative mRNA splicing that is orchestrated by mTORC1, and extends well beyond the functions of S6K. mTORC1-mediated mRNA splicing control is critical for growth and metabolic regulation, has implications for the relationship between mTORC1 and longevity, and appears to be evolutionarily conserved.

RESULTS

Genetic and informatic analyses implicate mRNA splicing regulation in mTORC1 function

To identify mTORC1 functions that are critical during growth, we analyzed orthologs of proteins that had been detected in mammalian cells as being phosphorylated in an mTORC1-dependent manner (mTOR phosphoproteome)^24–29, or forming mTOR-dependent complexes (mTOR interactome)^29,31 (Table S1). We screened these (mTOR proteome) and a set of additional genes through (1) RNA interference (RNAi) to detect genetic interaction with mTORC1 and (2) analysis for presence of an evolutionarily-conserved TOR signaling (TOS) motif, a site through which the mTORC1 complex binds phosphorylation targets³² (Figure 1A and Table S1). We expected that combining these strategies would increase the likelihood of identifying proteins that are regulated by mTORC1.

Figure 1. Identification of functional interactions involving mTORC1.

(A) Genetic and bioinformatic screening to identify mechanisms that are regulated by mTORC1. See also Figure S1A–S1D, and Table S1.

(B) Body size after daf-15 (Raptor) and let-363 (mTOR kinase) RNAi was performed in wild-type (WT) and raga-1(−/−) animals. EV, empty vector.

(C) KEGG pathway analysis for 187 genetic screen hits that showed strong or moderate interactions with raga-1. Fisher’s exact test.

(D) STRING analysis illustrating known or predicted protein-protein interactions among the 47 double-positive screen hits (Figure S1E). mRNA splicing and mTOR signaling genes are shown in red and blue, respectively. Confidence refers to the approximate probability of protein-protein association scored by known or predicted interactions.

(E) Genetic interaction between raga-1 and rsp-3 or rsp-6. RNAi knockdown of rsp-3 (1:9 dilution) or rsp-6 reduced brood size in raga-1(−/−) but not WT animals. One-way ANOVA with Tukey’s correction, mean ± SEM, ***P < 0.001; n.s., not significant. Three biological replicates were performed.

To uncover genetic interactions, we screened for enhancement of phenotypes deriving from a partial loss of mTORC1 function (Figure 1A). mTORC1 function is reduced but not eliminated by a null mutation in the RagA GTPase raga-1(ok386)(hereafter raga-1(−/−)), which transduces amino acid availability signals to mTORC1^13,32. While complete ablation of mTORC1 arrests larval development, raga-1(−/−) mutants exhibit slower development, decreased body and brood size, and partial embryonic lethality, phenotypes that can be explained by reduced mTORC1 activity^14,33. By identifying genes for which RNAi exacerbated these defects in raga-1(−/−) compared to wild-type (WT) animals (Figures 1A and 1B, S1A), we expected to find mechanisms that are functionally associated with mTORC1, and might function downstream of mTORC1. Strong or moderate phenotypic interactions with raga-1 were detected for 69/637 mTOR proteome genes and 118/1388 other genes that were screened (Figure 1C and S1B; Table S1). This screen detected the known mTOR pathway genes let-363(mTOR), rsks-1(S6K), and rheb-1(RHEB) (Figure 1B and 1C; Table S1), supporting its effectiveness. The TOS motif was enriched among raga-1 genetic interactors within the mTOR proteome (Figure S1C), indicating a correlation between genetic and possible physical interactions involving mTORC1. Accordingly, raga-1 genetic interactors were significantly enriched in each mTOR interactome dataset (Figure S1B). Genetic screen hits were enriched to a lesser degree across the mammalian phosphoproteome (Figure S1B), possibly because those studies identified numerous indirect phosphorylation events^24–29.

Many genes that exhibited strong or moderate interactions with raga-1 were involved in processes known to be associated with mTORC1 (growth signaling, longevity, or autophagy), but it was striking that the most highly represented functional category was mRNA splicing (Figure 1C and Table S1). Notably, 47 “double-positive” screen hits exhibited both a conserved TOS motif and genetic interaction with raga-1 (Figure 1D; Figure S1D and S1E; Table S1). The double-positives included proteins associated with mTOR signaling and various cellular processes, but the mechanism that was most prominently represented was mRNA splicing (Figure 1D and S1E). It was striking that these splicing factor hits included multiple spliceosome components, as well as three evolutionarily conserved SR proteins (rsp-3 (SRSF1/9), rsp-6 (SRSF3/7), and rsp-8 (TRA2A/B, RMBX, RMBXL1). Two of these SR proteins (RSP-3 and RSP-6) exhibited conservation of a TOS consensus at a similar position near their N-terminus, consistent with functional conservation (Figure S1F). Confirming our screen results, RNAi against rsp-3 or rsp-6 impaired fecundity and development more severely in raga-1 than WT (Figures 1E and S1I). In a previous C. elegans phosphoproteomic study an RSP-6 phosphosite was identified as showing the greatest fold-change in mTORC1-dependent phosphorylation, and could be phosphorylated by mTORC1 in vitro³⁰. This suggests that RSP-6 may be regulated directly by mTORC1, consistent with the prediction from our screen.

Nutrients broadly remodel mRNA expression and splicing through mTORC1

The prominence of mRNA splicing factors among our screen hits suggested that during development and growth mTORC1 might regulate mRNA splicing more broadly than can be explained by the action of S6K. We therefore investigated how mTORC1 and its target S6K influence mRNA production and alternative splicing during a developmental response to nutrients. When C. elegans hatch in the absence of food, they arrest growth at the L1 larval stage and enter a stable diapause state in which they remain viable for weeks^34,35. When food is provided, the arrested animals re-enter and complete development, and produce offspring. After subjecting WT, raga-1, and rsks-1 (S6K) animals to this growth arrest, we compared them by RNA sequencing (RNA-seq) to a group that was released from arrest by feeding for four hours (Figure 2A). We examined this time point in order to capture early gene expression changes while minimizing any possible effects of the L1 developmental program³⁵. Notably, L1 arrest and recovery did not reduce progeny production in raga-1 animals (Figure S2A), suggesting that their overall fitness was not impaired by diapause. Aside from two germ cell precursors L1 stage C. elegans consist of somatic tissues^34,36, minimizing possible differential effects on somatic and reproductive cells.

Figure 2. mTORC1 mediates nutrient regulation of gene expression.

(A) Scheme for RNA-seq analysis of the nutrient response.

(B) Heatmap showing genes that were differentially expressed in response to feeding or mTORC1 pathway gene mutation in WT animals. All P-values for RNA-seq data have been calculated as false-discovery rate (FDR)-adjusted. P < 0.01 was used for the cutoff. Three biological replicates were performed.

(C) Comparison of genes that were differentially up-regulated by feeding (WT Fed/WT Stv) and mTORC1 activity (WT Fed/raga-1 Fed).

(D) GO terms enriched in genes that were up-regulated by both feeding and mTORC1 (WT Fed/raga-1 Fed). Fisher’s exact test.

(E) Relative expression levels of mRNA splicing-related genes that exhibited strong genetic interactions with raga-1 (mTORC1) (Figure 1C; Table S1 and Table S2), normalized to WT levels. Mean ± SEM, ***P < 0.001, n.s., not significant.

Feeding of arrested L1 larvae induces widespread changes in mRNA expression and isoform representation³⁵. In WT animals, at four hours after feeding-induced release from growth arrest the expression levels of >4000 mRNAs were increased (Figure 2B and 2C, and Table S2). Many of these genes (2,834) were also upregulated in raga-1 and rsks-1 mutants (Figure S2B and S2C, but for most their upregulation was markedly diminished in raga-1 mutants, as revealed by the relative heatmap intensities of WT Fed/Starved (Stv) vs raga-1 Fed/Stv (Figure 2B), comparison of WT Fed/Stv with WT Fed/raga-1 Fed genes (Figure 2C), and analysis of individual genes (Figures S2D–S2E). Genes that were downregulated by feeding exhibited a similar pattern (Figure 2B; Figures S2F and S2G, and S2H–S2I), with mutation of rsks-1 in general having a smaller effect in each case (Figure 2B; Figure S2D–S2E and S2H–S2I). Genes that were upregulated by feeding and mTORC1 were enriched for similar processes, including post-transcriptional gene regulation and mRNA splicing (Figure 2D; Figure S2B and S2C), consistent with most feeding-induced gene expression being dependent upon mTORC1 (Figures 2B and 2C). Interestingly, most mRNA splicing genes that our screen identified as strong genetic interactors with raga-1 also depended upon raga-1 for full levels of expression (Figure 2E), as did many other mRNA splicing factor genes (Figure S2J–S2N). Knockdown of daf-15 (Raptor) or let-363 (mTOR kinase) reduced splicing factor gene expression similarly to raga-1 RNAi, further supporting the importance of mTORC1 for splicing factor gene expression (Figure S2O). Genes that were downregulated by feeding and expressed at higher levels in raga-1 mutants were enriched for autophagy, dauer development, and lifespan determination (Figures S2F and S2G), consistent with the increases in stress response gene expression and lifespan seen when mTORC1 is inhibited^14,37,38. Thus, the profound effects of feeding on gene expression are mediated largely by mTORC1.

We next compared how feeding and mTORC1 affected mRNA splicing. Alternative mRNA splicing is less prevalent in C. elegans than in mammals, with approximately 25% of the roughly 20,000 C. elegans protein coding genes expressing more than one mRNA isoform compared to 90% of mammalian genes^39–41. Applying high-stringency criteria to detect changes in splice junction use, we found that feeding altered the frequency of annotated alternative splicing events at 409 genes (Figures 3A–3C; Table S3), a substantial proportion of the roughly 5000 at which alternative splicing occurs ^39–41. Importantly, mTORC1 induced a major proportion of the alternative splicing changes that we identified, as indicated by the differences between WT Fed/Stv and raga-1 Fed/Stv, and between WT and raga-1 fed animals (Figures 3A–3C). The effect of rsks-1 mutation on these splicing events was much smaller, as revealed by the relative numbers of affected events and the intensities of the differences in the heatmap (Figures 3A and 3B), indicating that the bulk of mTORC1-mediated splicing regulation occurs independently of S6K.

Figure 3. Nutrients broadly regulate alternative mRNA splicing through mTORC1.

(A) rMATS analysis of alternative splicing events. FDR-adjusted P-value < 0.05. To increase stringency the minimum delta PSI (Percent Spliced In) was set at 10% instead of the rMATS default of 5%.

(B) Heatmap comparing 582 feeding-regulated splicing events detected in wild-type worms (WT Fed vs. WT Stv) to feeding effects detected in raga-1 and rsks-1 mutants.

(C) Overlap between genes that were alternatively spliced in response to feeding (WT Fed/WT Stv) and mTORC1 activity (WT Fed/raga-1 Fed).

(D) Processes that were enriched among the 133 genes for which alternative splicing was altered by altered by both feeding (WT Fed/WT Stv) and mTORC1 activity (WT Fed/raga-1 Fed). Fisher’s exact test. *P < 0.05.

(E) rsp-6 isoform transcript structure. Isoform a encodes a functional protein.

(F) RNA-seq and qPCR data indicate that raga-1 is required for feeding-induced alternative mRNA splicing events that increase levels of the functional rsp-6a isoform. Two-sided unpaired t-test, mean ± SEM, *P < 0.05, **P < 0.01. Three biological replicates were performed.

The genes at which raga-1 mutation altered splicing event frequencies are involved in a wide range of processes, including metabolism, growth and development, lifespan, and mRNA splicing (Figures 3C, 3D, and S3A; Table S3). A similar set of processes was enriched among genes at which both feeding and raga-1 mutation altered the expression of individual mRNA isoforms (Figure S3B and S3C). The representation of mRNA splicing factor genes was particularly striking because it suggested that mTORC1 may modulate alternative mRNA splicing in part by controlling the alternative splicing of splicing factor mRNAs. For example, at the double-positive screen hit rsp-6 (SR protein SRSF3/7), levels of the protein-coding and non-coding transcripts were respectively increased or reduced by feeding, effects that were blunted when mTORC1 activity was reduced by loss of RAGA-1 (Figure 3E and 3F). Similarly, mTORC1 increased the representation of coding region transcripts at the SR protein genes rsp-4 and rsp-5, and the spliceosome component uaf-1 (U2AF2) (Figures S3D–S3F; Table S4). Thus, during the nutrient response, mTORC1 was required not only for appropriate expression of numerous genes involved in growth, including mRNA splicing factor genes (Figures 2 and S2), but also for generation of protein-coding mRNA transcripts at many of these genes.

We examined how mTORC1 influences alternative splicing in vivo using a fluorescent reporter for the ret-1 gene (Figure S3G)⁴². This reporter monitors an exon skipping or inclusion choice that we had detected as regulated by feeding and raga-1 (Table S3). In L1 larvae feeding increased inclusion of this exon, largely dependent upon raga-1 (Figures S3H and S3I). In starved day-2 adults feeding induced an even more robust splicing response that was also markedly attenuated in raga-1-mutants, indicating that this response is induced by nutrient availability independently of the larval developmental program (Figure S3J–S3K). Knockdown of raga-1 and let-363 (mTOR kinase) altered ret-1 reporter splicing comparably (Figure S3L and S3M), as would be predicted from RAGA-1 acting through mTORC1. Notably, exon inclusion was decreased by reductions in the activity of either the RAG GTPase RAGC-1 or two SR proteins that were double-positives in our screen (RSP-3 and RSP-6) (Figure S3N–S3O), supporting the idea that mTORC1 promotes the activity of these mRNA splicing factors. Together with our evidence that mTORC1 increases expression and functional protein-coding mRNA isoform production at mRNA splicing factor genes (Figures 2E, 3E, 3F, and Figures S2J–S2N, S3D–S3F and Table S3), the data suggest that mTORC1 controls alternative mRNA splicing in part by increasing the expression and activity of these mRNA splicing factors.

Importance of mTORC1-mediated mRNA splicing regulation in vivo

If mRNA splicing regulation is important for mTORC1 function in vivo, then it might be possible to overcome some effects of mTORC1 impairment by increasing the activity of mTORC1-dependent splicing. To test this idea, we investigated whether transgenic overexpression (OE) of the SR protein and double-positive screen hit RSP-6 (SRSF3/7) (Figure 1D and 4A; Figure S1G) could rescue effects of a reduction in mTORC1 function (RNAi against raga-1 or other mTORC1 pathway components). RSP-6 is positively regulated by mTORC1 at the levels of mRNA splicing and expression (Figures 2E and 3F), affects ret-1 reporter splicing similarly to mTORC1 (Figure S3N and S3O), and is phosphorylated by mTORC1³⁰. This suggests that during growth mTORC1 promotes RSP-6 activity, possibly through direct phosphorylation. RSP-6 OE rescued the lethality of the rsp-6(ok798) null mutant (rsp-6(−/−)), but RSP-6 mutants in which we deleted the RNA recognition motif (rsp-6 ΔRRM) or the arginine- and serine-rich (RS) domain (rsp-6 ΔRS) did not (Figure 4A and Figure S4A). Remarkably, OE of RSP-6 but not RSP-6 ΔRRM or RSP-6 ΔRS partially rescued the developmental delay and reductions in body length, brood size, and live progeny production induced by the knockdown of mTORC1 pathway genes (Figures 4B–4E). By contrast, in control experiments RSP-6 OE did not affect RNAi phenotypes associated with other genes (Figures S4B–S4E). Thus, an increase in RSP-6 activity can partially compensate for a modest reduction in mTORC1 activity during C. elegans development.

Figure 4. Partial rescue of mTORC1-dependent phenotypes by RSP-6 overexpression.

(A) Schematic of transgenic RSP-6::GFP proteins.

(B) Developmental stage of initially synchronized animals at 72 hours after treatment with RNAi against raga-1 (RAGA), ragc-1 (RAGC), rheb-1 (RHEB) or daf-15 (Raptor). In all such experiments RSP-6 OE is compared to rol-6 transgenic marker control. N > 25, chi-square-test. ***P < 0.001, n.s.= not significant. Three biological replicates were performed.

(C) Representative images illustrating body length after 72 hours of treatment with RNAi against raga-1, ragc-1, rheb-1 or daf-15. Scale bars: 100μm. Three biological replicates were performed.

(D) Quantification of (C). Two-way ANOVA with Tukey’s correction, *P <0.05, **P <0.01, ***P <0.001, n.s.= not significant. ###P < 0.001 compared to rol-6 control animals with EV.

(E) Number of live progeny after treatment with RNAi against raga-1, ragc-1, rheb-1 or daf-15. Two-way ANOVA with Tukey’s correction *P < 0.05, **P < 0.01, ***P < 0.001, n.s.= not significant. Three biological replicates were performed.

(F) Representative images illustrating body length after 72 hours of treatment with RNAi against raga-1. Scale bars: 100μm. Three biological replicates were performed.

(G) Quantification of (F). Two-way ANOVA with Tukey’s correction, *P <0.05, **P <0.01, ***P <0.001, n.s.= not significant. ###P < 0.001 compared to rol-6 control animals with EV.

(H) Number of live progeny after treatment with RNAi against raga-1. Two-way ANOVA with Tukey’s correction *P < 0.05, **P < 0.01, ***P < 0.001, n.s.= not significant. Three biological replicates were performed.

(I) Relative expression of SR protein gene expression, detected by qPCR. Two-way ANOVA with Tukey’s correction, ***P < 0.001, #P < 0.05, ##P < 0.01, ###P < 0.001. Controls are rol-6 transgenic vector animals and EV RNAi. Three biological replicates were performed.

(J) Semi-quantitative PCR analysis indicating that RSP-6 OE overcomes the effects of reduced mTORC1 activity on the rsp-4 coding transcript. Two-sided unpaired t-test, mean ± SEM, **P < 0.01, ***P < 0.001. Three biological replicates were performed.

Given that mTORC1 increases expression and coding mRNA transcript production at multiple mRNA splicing factor genes (Figures 2E, S2J–S2O, 3B, 3F, and S3D–S3F), we considered the possibility that RSP-6 OE might have similar effects. Accordingly, RSP-6 OE increased the levels of expression and functional coding transcript production at numerous alternatively spliced genes that are regulated by feeding and mTORC1, including mRNA splicing factor genes, thereby overcoming the effects of raga-1 loss (Figures 4I and 4J; Figures S4F–S4J). Mirroring this finding, knockdown of the SR protein RSP-3 reduced the expression of splicing factor genes (Figure S4K). Together, the data suggest that the regulation of expression and alternative splicing of many of these genes may be interdependent, involving the activity of multiple mRNA splicing factors that function downstream of mTORC1. Interestingly, transgenic RSP-6 in which the mTORC1 phosphorylation site³⁰ has been altered (RSP-6 (S99A)) rescued the rsp-6 null mutation (Figures 4A and S4A), but when overexpressed failed to rescue raga-1(RNAi) phenotypes (Figures 4F–4H). A speculative explanation is that although phosphorylation of RSP-6 at this site by mTORC1 is not essential for its function, this phosphorylation may be important under conditions of reduced mTORC1 activity, in which mTORC1 may not sufficiently enhance the activity of other mRNA splicing factors.

Feeding remodels metabolism through mTORC1-mediated splicing regulation

Our results raise an important question: how does mTORC1-mediated mRNA splicing regulation contribute to growth? Genes for which feeding and mTORC1 altered splice junction and mRNA isoform representation are involved in various biological processes, notably including metabolism (Figure 3D, and Figures S3A–S3C). This raises the possibility that mTORC1 might orchestrate metabolic effects by controlling alternative mRNA splicing. To investigate how feeding and mTORC1 influence metabolism in C. elegans, we performed a steady-state non-targeted liquid chromatography-tandem mass spectrometry (LC-MS/MS) metabolomics analysis of WT and raga-1 L1 larvae (Figure 5A). We then examined our RNA-seq data to determine how feeding and mTORC1 affects alternative mRNA splicing at genes within the metabolic pathways that we identified.

Figure 5. Feeding remodels metabolism through mTORC1 regulation of mRNA expression and splicing.

(A) Scheme of metabolomics experiment.

(B, C) Volcano plots of metabolites regulated by (B) feeding (Fed EV vs Stv EV) or (C) mTORC1 (Fed EV vs Fed raga-1 RNAi). Up- and down-regulated metabolites are indicated in red and blue, respectively, with gray showing metabolites with no significant difference. Total metabolites detected are indicated in parentheses. The horizontal dotted line indicates a P-value cutoff at 0.05. two-sided unpaired t-test. Three biological replicates were performed.

(D) Venn diagram comparing feeding- and mTORC1-upregulated metabolites.

(E) Heatmap showing relative levels of metabolites that were upregulated by both feeding and mTORC1 (raga-1-dependent). P-values are shown in the right panel, with a P-value cutoff of 0.05. two-sided unpaired t-test. Asterisks indicate metabolites that were previously detected as mTORC1-regulated in mammalian cell studies^43–47.

(F) Metabolic pathway diagram illustrating regulation of metabolite levels, mRNA expression, and alternative mRNA isoform use by feeding and mTORC1. Asterisks indicate genes where mTORC1 and feeding increases expression of a functional protein-coding isoform. Red arrows indicate metabolic pathways where the direction of biosynthesis correlates with mTORC-1 (raga-1) increasing the levels of both the protein coding mRNA isoform and the metabolite immediately downstream.

(G-I) Scatter plots comparing the effects of feeding with raga-1 or rsks-1 mutation on the expression of individual mRNA isoforms at genes in the indicated metabolic pathways. Only genes that encode two or more isoforms were included. X-axis; alteration of isoform expression by feeding (WT Fed/WT Stv), Y-axis; alteration of isoform expression by raga-1 or rsks-1 (WT Fed/raga-1 Fed or WT Fed/rsks-1 Fed). Spearman correlation coefficients (R) and associated P-values are shown in colors corresponding to the data plots. Points were excluded as outliers if their Z-score exceeded ±3 units. For evaluating the statistical significance of the difference in correlations, Fisher’s z-transformation was applied based on correlation coefficients.

(J) Schematic of atic-1 isoforms. The bifunctional protein ATIC-1 (5-aminoimidazole-4-carboxamide ribonucleotide (AICAR) formyltransferase/inosine monophosphate (IMP) cyclohydrolase) mediates IMP synthesis from AICAR, connecting the pentose phosphate and purine biosynthesis pathways (F). Isoform b encodes a functional protein.

(K) Expression levels of atic-1 and mRNA isoforms are plotted from RNAseq data. two-sided unpaired t-test, mean ± SEM, *P < 0.05, **P < 0.01.

After starvation arrest at the L1 stage, four hours of feeding increased the levels of 110/198 detectable metabolites, and in fed animals 81/203 detectable metabolites were raga-1-dependent (Figure 5B and 5C, and Table S5). 61 of these feeding- or mTORC1-upregulated metabolites overlapped, indicating that most feeding-upregulated metabolites are mTORC1-dependent (Figure 5D and 5E). 21 of these mTORC1-regulated metabolites had been previously detected as such in mammalian cells and included components in the TCA cycle, glycolysis/gluconeogenesis, pentose phosphate, amino acid, and purine and pyrimidine metabolism pathways (Figure 5E, S5A, and Table S5)^43–47. We also identified 60 metabolites within these pathways that had not previously been identified as mTORC1-regulated, including 18 amino acids or amino acid derivatives, along with new metabolic categories or pathways that previously had not been associated with mTORC1. The latter groups include amino sugar and nucleotide sugar metabolism, which is required for glycosylation^48,49 and mevalonate, which mediates prenylation^50,51, and co-factors (Figure S5A and Table S5). Thus, feeding profoundly alters growth-related metabolism in C. elegans, with mTORC1 playing a role that is both evolutionarily conserved and broader than described previously.

Many metabolites that were upregulated by both feeding and mTORC1 are produced in growth-related pathways (purine and pyrimidine metabolism, glycolysis, TCA cycle, pentose phosphate) that were enriched among genes at which feeding and mTORC1 modulated mRNA splicing (Figures 5E and 5F; Figure S3C). Accordingly, we detected a significant correlation between feeding-induced (WT Fed vs WT Stv) and mTORC1-dependent (WT Fed vs raga-1(−/−) Fed) expression of individual mRNA isoforms at genes within some metabolic pathways, as well as among fatty acid metabolism genes (Figure 5G–5I; Figure S5B; Table S4). The correlations between feeding-induced and rsks-1-dependent isoform expression were either weaker or not significant (Figure 5G–5I and Figure S5B; Table S4). The proportion of genes within the metabolic pathways shown in Figure 5F that express more than one mRNA isoform (44%) is greater than the C. elegans average of 25%^39–41 (Figure S5C). Of these, 31 (60%) were regulated by feeding and mTORC1 at the level of mRNA splicing, a group that was widely distributed throughout these pathways (Figure 5F and Figure S5C). Together, our findings suggest that a substantial proportion of growth-related metabolism is regulated at the level of mRNA splicing, and that nutrients act through mTORC1 to regulate mRNA expression and alternative splicing at numerous genes within these metabolic pathways.

We examined our RNAseq data to investigate how mTORC1-regulated alternative mRNA splicing might influence metabolism. Neither feeding nor mTORC1 affected expression of the purine biosynthesis gene atic-1, but each increased the levels of its functional mRNA isoform and reduced representation of other transcripts (Figures 5F, 5J, and 5K). mTORC1 preferentially increased the levels of protein-coding transcripts at additional genes in these pathways, including fbp-1(fructose-bisphosphatase), B0001.4 (uridine-cytidine kinase), F40F8.1 (cytidine/uridine monophosphate kinase), T22D1.3 (inosine monophosphate dehydrogenase), and tald-1 (transaldolase) (Figure 5F; Figures S5D–S5H). We asked whether metabolite production within these pathways might be linked to expression of functional protein coding transcripts at metabolic genes. Importantly, at many steps upregulation of a protein-coding mRNA isoform by feeding and mTORC1 was associated with increased levels of the metabolite immediately downstream (Figure 5F), consistent with metabolite production being regulated by mTORC1 through its control of mRNA splicing. These examples likely underestimate the contribution of mTORC1-mediated splicing regulation because (1) our metabolomics captured the steady-state levels of individual metabolites, not the flow of biosynthesis through the pathway and (2) throughout our study the levels of non-coding transcripts might have been reduced by nonsense-mediated decay⁵². We conclude that mTORC1-mediated regulation of mRNA splicing plays a major role in metabolic regulation during growth.

We investigated whether the effects of reduced mTORC1 activity on metabolic regulation might be relieved when mRNA splicing is enhanced by RSP-6 OE. Steady-state metabolomics revealed that in the WT control background RSP-6 OE decreased or increased the levels of 54 and 2 metabolites, respectively, further indicating that metabolism is broadly regulated at the mRNA splicing level (Figure 6A and S6A; Table S5). A greater proportion of metabolites were increased by RSP-6 OE in raga-1(RNAi) animals (Figure 6B and S6B; Table S5), in which the landscape of alternative mRNA splicing is altered (Figure 3). RNAi against raga-1 reduced the levels of many metabolites in both WT and RSP-6 OE animals (Figures S6C–S6E). Importantly, of the 83 metabolites for which levels were decreased by raga-1 RNAi in control animals (Figure S6C), 11 were partially rescued by RSP-6 OE, in that their levels in raga-1 animals moved closer to WT levels (Figure 6C). The RSP-6 OE-rescued metabolites include components of the mTORC1-regulated purine (hypoxanthine, xanthine, and adenine), pyrimidine (L-dihydroorotate, orotate) and glycolysis (pyruvate) metabolism pathways (Figures 6C and 5F; Table S5), all of which except L-dihydroorotate were upregulated by feeding (Figure 6D). Importantly, RSP-6 OE also partially reversed the effects of raga-1 RNAi on expression of many genes in the glycolysis, pentose phosphate, and purine and pyrimidine metabolism pathways (Figure 6E–6H) and rescued coding transcript production at multiple metabolic genes (Figures 6I–6K and Figures S6F–S6L). The striking capacity of this SR protein for overcoming insufficient RAGA-1 activity strongly supports the idea that mRNA splicing regulation is a key mechanism through which mTORC1 controls metabolism and growth.

Figure 6. Importance of mTORC1-mediated mRNA splicing regulation in metabolism and longevity.

(A and B) Volcano plot showing effects of RSP-6 OE on steady-state metabolite levels in (A) EV control and (B) raga-1 RNAi animals, compared to rol-6 transgene control. The horizontal dotted line indicates a P value cutoff of 0.05. Significantly up- and down-regulated metabolites are indicated by red and blue dots, respectively. Two-sided unpaired t-test. Three biological replicates were performed.

(C) Heatmap showing levels of metabolites that were significantly decreased by raga-1 RNAi compared to EV, and rescued by RSP-6 OE in raga-1 RNAi animals (RSP-6 OE raga-1 RNAi/control raga-1 RNAi). P values are shown in the right panel; with a cutoff of 0.05. Two-sided unpaired t-test.

(D) Comparison of metabolites that were upregulated under the indicated conditions. (E-H) qPCR analysis showing effects of raga-1 RNAi and RSP-6 OE on overall expression of selected genes in the glycolysis, pentose phosphate, and purine and pyrimidine metabolism pathways. Mean ± SEM, *P < 0.05, **P < 0.01. #P < 0.05, ##P < 0.01, compared to rol-6 transgenic control animals with EV RNAi. Two-sided unpaired t-test. Three biological replicates were performed.

(I-K) Relative expression of individual mRNA isoforms at the indicated genes, assayed and illustrated as in (E-H). Mean ± SEM, **P < 0.01, ***P < 0.001. #P < 0.05, ##P < 0.01, compared to rol-6 transgenic control animals with EV RNAi. Two-sided unpaired t-test. Three biological replicates were performed.

(L-N) Lifespan analysis of the indicated genotypes at 20°C, measured from day 1 of adulthood. Genetic mutants are shown in (L), with RNAi experiments shown in (M, N). A composite of 3 or 4 independent biological replicates is shown, with complete data presented in Table S6. Log-rank test, *P < 0.05, **P < 0.01, ***P < 0.001, n.s.= not significant.

Both raga-1 and RSP-6 OE affect expression and alternative splicing at numerous mRNA splicing factor genes (Figures 3, 4, S3, and S4), suggesting that while RSP-6 may be critical for splicing regulation by mTORC1, other factors might also be important. We therefore searched genes for which splicing was altered by feeding and raga-1 for presence of consensus human SR protein binding sites (Figure S6M). The SRSF3 (RSP-6) motif was enriched at some splice junctions, suggesting possible direct targeting by RSP-6, but other motifs were also associated with raga-1 dependence (WT Fed/raga-1 Fed, Figure S6M). This suggests that while RSP-6 may have an important role linking mTORC1 to mRNA splicing regulation, multiple mRNA splicing factors may mediate mTORC1 splicing regulation of individual splice junctions. Interestingly, the motif for SRSF10 (RSP-4), was enriched at some junctions that appeared to be raga-1-independent, consistent with splicing of feeding-regulated genes that is mTORC1-independent (Figures 3A and 3B) possibly being regulated by a distinct pathway.

RSP-6 enhances longevity from low mTORC1

Inhibiting mTORC1 increases lifespan in essentially all eukaryotes^14–16, but overexpression of certain mRNA splicing factors also increases lifespan^53–56. In the latter case, lifespan extension could derive from an enhancement of gene expression fidelity during aging, or effects on specific processes that promote longevity. Moreover, although dietary restriction (DR) maintains “youthful” mRNA splicing patterns during age^53,57, an effect that requires mTORC1⁵³, it is generally accepted that DR reduces mTORC1 activity and that this is important for its beneficial effects on longevity and health^14,15,58. If RSP-6 OE simply reversed effects of mTORC1 inhibition on longevity, as we observed for development and many aspects of metabolism, this should blunt the lifespan increase resulting from low mTORC1. However, RSP-6 OE increased lifespan in long-lived rsks-1 or raga-1 loss-of-function backgrounds (by 11.7%–15.6%), and RSP-6 OE and raga-1 mutation in combination increased mean lifespan by 56% (Figures 6L–6N and S6N; Table S6). By contrast, neither RSP-6 ΔRRM nor RSP-6 ΔRS altered mean lifespan consistently or meaningfully (≥10%) across genotypes (Figure 6M–6N; Table S6). Thus, increased activity of this mTORC1-dependent mRNA splicing factor does not reverse lifespan extension from mTORC1 inhibition and instead further slows aging, suggesting that not all effects of mTORC1 inhibition are necessarily beneficial with respect to longevity.

mTORC1-mediated mRNA splicing regulation in human cells

We investigated whether mTORC1 also regulates mRNA splicing broadly in human cells. We inhibited mTORC1 in lymphangioleiomyomatosis (LAM) 621–101 (TSC2-deficient) cells, in which mTORC1 activity is elevated⁵⁹, and compared these effects to inhibition of S6K and its target SRPK2¹⁷ (Figure 7A and S7A). Inhibition of mTORC1 with Torin1⁶⁰ or rapamycin altered gene expression levels far more profoundly than did S6K inhibition (Figure 7B and Table S2), and reduced expression of mRNA splicing genes along with growth-related genes (Figure S7B). Similarly, mTORC1 inhibition altered a much greater number of annotated mRNA splicing events than did inhibition of S6K, SRPK1, or loss of the S6K target SRPK2¹⁷ (Figure 7C). Torin1 treatment altered the frequency of mRNA splicing events at mRNA splicing factor genes, and in many of the same metabolic pathways (Figure 7D and S7C). Inhibition of mTORC1 also increased the relative representation of transcript isoforms that do not encode proteins (Figure 7E). For example, Torin1 boosted the relative levels of non-coding transcripts for the SR protein genes SRSF4 and SRSF5 (Figures S7E and S7F). By contrast, S6K inhibition only modestly altered the distribution of coding versus non-coding mRNA transcripts (Figure S7D). In general, these findings were analogous to our C. elegans results, even though the number of alternative splicing events regulated by mTORC1 was far greater than observed in C. elegans (Figures 3A and 7C), in keeping with the greater prominence of alternative mRNA splicing at human genes⁴¹.

Figure 7. Conservation of mTORC1 gene regulation functions during human cell growth.

(A) Schematic of the regulatory pathways involved in mRNA splicing mediated by mTORC1, shown with inhibitors targeting its components.

(B) Heatmap showing 9357 differentially expressed genes affected by Rapamycin (Rapa), Torin1, the S6K inhibitor PF4708671 (PF), the SRPK1 inhibitor SRPIN340 (SRPIN), or SRPK2 knockout (SRPK2 KO) in LAM cells. FDR < 0.05. Three biological replicates were performed per condition.

(C) Effects of the indicated treatments on mRNA splicing events in LAM cells. SE; Skipped exon, RI; Retained intron, MX; Mutually exclusive exons, A3SS; Alternative 3’ splice sites, A5SS; Alternative 5’ splice sites. FDR < 0.05.

(D) Processes that were enriched among the 2562 genes for which alternative mRNA transcript use was affected by mTORC1 activity (control/Torin1).

(E) mTORC1 increases representation of protein-coding mRNA transcripts. Fold change of Torin1 effects is shown. The bar chart represents the number of protein-coding or non-coding (retained intron) transcripts that were affected. Two-sided chi-square test. ***P < 0.001.

(F) Heatmap showing the effects of 4h feeding and Torin1 on gene expression in HEK 293E cells. FDR < 0.05. Three biological replicates were performed per condition.

(G) GO terms for genes that are feeding-upregulated and mTORC1-dependent.

(H) mTORC1-dependence of mRNA splicing protein gene expression. Two-sided unpaired t-test, mean ± SEM, **P < 0.01, ***P < 0.001.

(I) mTORC1 dependence of feeding-induced alternative splicing events. FDR < 0.05.

(J) GO terms for feeding-induced alternative splicing events that are mTORC1-dependent.

(K) Effects of Torin1 on ATIC total expression and mRNA transcript levels. Two-sided unpaired t-test, mean ± SEM, *P < 0.05.

(L) Model for nutrient-responsive mTORC1 regulation of gene expression and alternative mRNA splicing, and its effects on growth, metabolism, and longevity.

We examined how mTORC1 influences mRNA expression and alternative splicing during a growth response in human embryonic kidney (HEK) 293E cells, in which mTORC1 is not hyperactive. We released a starvation-induced growth arrest by refeeding, then analyzed mRNA levels and alternative splicing events one and four hours later under either control or Torin1 treatment conditions. In the control one hour of feeding altered the expression of 93 genes, including the transcriptional regulators EGR1 (early growth response 1), EGR3, FOSB, JUN, and ID1 (Figure S7G and Table S2), many of which are immediately responsive to growth factors¹. Aside from upregulation of those classic immediate early genes, Torin1 blocked most of this early response (Figure S7G). Torin1 also dramatically suppressed the far broader response to four hours of feeding (Figure 7F and S7G). Thus, most serum-response genes are dependent upon mTORC1 for their activation during the feeding response. The feeding-induced genes that were inhibited by Torin1 were enriched for genes involved in mRNA metabolism and splicing, consistent with our C. elegans findings (Figures 2D, 7G, and 7H). Cell feeding also altered the frequency of thousands of alternative splicing events, as indicated by rMATS (Figure 7I and S7H). This response was broadly suppressed by Torin1, indicating mTORC1 dependence (Figure 7I). The genes at which feeding and mTORC1 modulated the same splicing events were enriched for involvement in mRNA splicing and metabolic processes that we have shown to be regulated by mTORC1 in C. elegans, including purine, pyrimidine, energy, and amino acid metabolism (Figures 5F and 7J). Many of the genes at which alternative mRNA transcript production was altered by feeding and mTORC1 in human cells were also regulated at the level of alternative mRNA splicing in C. elegans (Figure S7I). These shared genes were notable for their involvement in both mRNA processing and metabolism (Figure S7J). Also analogous to our C. elegans findings, during the feeding response mTORC1 increased the level of functional protein-coding transcripts from multiple genes (Figures 5F, 5J, 6J, 7K, and Figures S7K–S7N). These included the metabolic gene ATIC (atic-1), at which coding transcript production was also regulated by mTORC1 in C. elegans. Taken together, the data suggest that across a wide evolutionary spectrum mTORC1 controls a major proportion of the gene expression response to growth stimulation, including a broad reprogramming of alternative mRNA splicing that orchestrates metabolic gene regulation.

DISCUSSION

Starting with unbiased screening in C. elegans, here we determined that control of alternative mRNA splicing is a critical function of mTORC1, and that the scope of this regulation is considerably greater than can be accounted for by the action of S6K (Figure 7L). By modulating mRNA and protein isoform production, mTORC-1 mediated mRNA splicing regulation adds considerable complexity and range to mTORC1 function, providing a pathway through which nutrient signals can control many biological processes during growth, notably including metabolism. The effects of mTORC1 on mRNA expression and splicing are remarkably similar in human cells, suggesting that this mTORC1 function is likely to be evolutionarily conserved. Our C. elegans screening also identified numerous raga-1 genetic interactors that are involved in other mechanisms, many of which also encode a conserved TOS motif (Figures 1C, 1D, and S1E; Table 1)³², providing a resource for future work that may uncover additional processes that are controlled by mTORC1.

Cellular growth involves an extensive reprogramming of gene expression that has been classically attributed to the action of a cascade of gene-specific transcription factors^1–5. Gene regulation by mTORC1 has been thought to be more limited than what we observed. We were therefore surprised to find that in C. elegans and humans, during the nutrient response mTORC1 controls the bulk of gene regulation events, including a widespread remodeling of alternative mRNA splicing. These observations revise an understanding of the growth response that has been held for 30 years, and should revitalize interest in this area.

In this regard, an important question we have begun to address is how mTORC1 regulates alternative mRNA splicing to such a broad extent. In C. elegans phosphorylation of RSP-6 by mTORC1 might play an important role, as suggested by the effects of RSP-6 loss and OE on development, alternative mRNA splicing, and metabolism, and by our observation that mutation of its mTORC1 phosphorylation site prevents RSP-6 OE from rescuing raga-1 phenotypes. Interestingly, in mammalian cells mTORC1 hyperactivation is associated with increased exon skipping, an effect that is partially attributable to elevated levels of SRSF3⁶¹. This suggests that RSP-6 (SRSF3/7) might have a conserved role in mTORC1 splicing regulation. However, we also determined that mTORC1 is required for expression and generation of protein-coding mRNA transcripts at multiple mRNA splicing factor genes. The effects of RSP-3 loss and RSP-6 OE on expression and alternative mRNA splicing at these splicing factor genes suggest that they are regulated in an integrated and interdependent manner, possibly at transcriptional and/or post-transcriptional levels. mTORC1 control of mRNA splicing is therefore likely to be complex, and might involve multiple mTORC1 targets. In S. cerevisiae alternative splicing and intron formation are rare, but under starvation or low-TORC1 conditions excised introns accumulate and are critical for maintaining quiescence^62,63. An intriguing possibility is that the relationship we have described between growth regulation, mTORC1, and alternative mRNA splicing evolved from this ancient pathway, and that in metazoa mRNA splicing products themselves might have undescribed regulatory functions.

A number of observations illustrate the biological importance of mTORC1-mediated mRNA splicing regulation. Multiple mRNA splicing factor genes exhibited synthetic phenotypes with raga-1 during C. elegans development (Figure 1), and RSP-6 (SRSF3/7) OE partially rescued developmental and metabolic effects of raga-1 insufficiency (Figures 4, 6; S4 and S6). Each of these findings represents genetic evidence for a functional association between mTORC1 and mRNA splicing. In C. elegans and human cells mTORC1 altered the distribution of protein-coding vs non-coding mRNA transcripts, primarily favoring production of coding transcripts (Figures 3E, 3F, and 7E). We identified numerous examples in C. elegans which mTORC1 promoted generation of coding transcripts at genes encoding mRNA splicing factors and metabolic enzymes, with mTORC1 increasing levels of both an enzyme coding transcript and its metabolite product in many cases. With respect to metabolic regulation by mTORC1, our findings reveal alternative mRNA splicing control to be an important mechanism that provides an overarching level of regulation that presumably operates alongside the transcriptional mechanisms identified previously^6,10–13. The finding that many of the same genes are subject to mTORC1-mediated mRNA splicing regulation in C. elegans and human cells suggests the exciting and likely idea that this control mechanism is evolutionarily conserved. Regulation of pyruvate kinase 2 (PKM2) and other metabolic enzymes by mRNA splicing has been of considerable interest, particularly for cancer biology^64–67. By indicating that many enzymes are regulated at the mRNA splicing level in mTORC1-regulated metabolic pathways, our results may suggest strategies for development of new therapies.

An important implication of our findings is that in some settings, targeting of specific mTORC1 functions might be of greater value than inhibiting mTORC1. Among interventions that extend lifespan mTORC1 inhibition is one of the most heavily studied, and is the most extensively developed as a pharmacological anti-aging strategy^14–16. Maintenance of mRNA splicing also promotes longevity, and is important for lifespan extension^{38,53–55,57}. Importantly, here we have shown that enhancing mRNA splicing by overexpression of RSP-6, and thereby counteracting the effects of mTORC1 inhibition on mRNA splicing (Figures 4 and 6), can provide further lifespan extension in the setting of low mTORC1 activity (Figures 6L–6N and S6N). Thus, for optimal lifespan extension it may be promising to restore splicing activity in the setting of mTORC1 inhibition or to target downstream mTORC1-regulated mechanisms that have been linked to longevity, such as protein synthesis, stress resistance, autophagy, or redox protection from production of hydrogen sulfide^14–16,68. Targeted approaches might similarly be of great value for cancer therapy, in which interfering with specific metabolic pathways is a promising strategy. A deeper understanding of the complexity of mTORC1 functions may enable such targeted efforts.

LIMITATIONS OF THE STUDY

Our findings suggest that mTORC1 acts through RSP-6 to regulate mRNA splicing, but also that other mechanisms are almost certainly involved. It will be important to identify these mechanisms, and determine why the expression and alternative splicing of mRNA splicing factor genes is interdependent during growth. It will also be interesting to elucidate how the mRNA splicing-mediated metabolic regulation by mTORC1 we have uncovered might be coordinated with previously described mechanisms. With respect to mTORC1 and aging, it remains to be determined whether the benefits of maintaining mRNA splicing activity derive from an enhancement of gene expression fidelity during aging, or splicing-mediated regulation of specific processes. Our results suggest that broad regulation of alternative mRNA splicing and its actions on metabolism during growth are conserved mTORC1 functions, but to establish this unambiguously in humans it will be necessary to elucidate the mechanisms involved.

RESOURCE AVAILABILITY

Lead contact

Further Information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, T. Keith Blackwell (Keith.Blackwell@joslin.harvard.edu).

Material availability

All unique reagents and strains generated in this study are available from the lead contact without any restrictions for academic research purposes.

Data and code availability

The accession number for the RNA-seq sequencing and processed data reported in this paper is GEO: GSE272718, GSE273387, and GSE273388. The accession number for the metabolomics and processed data reported in this paper is Metabolomics Workbench: PR002071. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

STAR★METHODS

Experimental Model and Study Participant Details

C. elegans strains

N2 Bristol was used as the Wild-type (WT) strain. The following other strains were used in this study: VC222: raga-1(ok386) II, RB1206: rsks-1(ok1255) III, and VC597: rsp-6(ok798) IV/nT1[qIs51] were obtained from the Caenorhabditis Genetics Center (CGC). The strain KH2235 (lin-15(n765)ybIs2167[eft-3p::ret-1E4E5(+1)E6-GGS6-mCherry+eft-3p::ret-1E4E5(+1)E6(+2)GGS6-GFP+ lin-15(+)+ pRG5271Neo]X) was a gift from Hidehito Kuroyanagi ⁴². LD1936: raga-1(ok386) II; lin-15(n765)ybIs2167[eft-3p::ret-1E4E5(+1)E6-GGS6-mCherry+eft-3p::ret-1E4E5(+1)E6(+2)GGS6-GFP+ lin-15(+)+ pRG5271Neo] X was generated by crossing VC222 and KH2235. LD1271: N2; Ex1060[pRF4(rol-6(su1006))] was described previously ⁷¹.

LD1937: ldEx32[Prsp-6::rsp-6::gfp::let-858 3’UTR+ rol-6(su1006)], LD1938: ldEx33[Prsp-6::rsp-6::gfp::let-858 3’UTR+ rol-6(su1006)], LD1939: ldEx34[Prsp-6::rsp-6 ΔRRM::gfp::let-858 3’UTR+ rol-6(su1006)], LD1941: ldEx36[Prsp-6::rsp-6 ΔRS::gfp::let-858 3’UTR+ rol-6(su1006)], LD1943: ldIs38[rol-6(su1006)], LD1945: ldIs40[Prsp-6::rsp-6::gfp::let-858 3’UTR+ rol-6(su1006)], and LD1969: ldEx38[Prsp-6::rsp-6 S99A::gfp::let-858 3’UTR+ rol-6(su1006)] were made in this study (see below). ldEx1964 was generated by SunyBiotech. The presence of raga-1(ok386), rsks-1(ok1255), rsp-6(ok798), Prsp-6::rsp-6::gfp::let-858 3’UTR, Prsp-6::rsp-6 ΔRRM::gfp::let-858 3’UTR, Prsp-6::rsp-6 ΔRS::gfp::let-858 3’UTR, and Prsp-6::rsp-6 S99A::gfp::let-858 3’UTR were confirmed using PCR and sequencing.

Plasmid construction

An rsp-6 genomic DNA fragment (2125 bp) was amplified by PCR from the wild-type C. elegans genome using the primers Prsp-6-F-outer and rsp-6-gfpFusion-R. A gfp::let-858 3’UTR was amplified from pPD158.87 using the primers rsp-6-gfpFusion-F and let-858UTR-Out-R. Fusion PCR was performed between two amplified products using the primers Prsp-6-F-SbfI and let-858UTR-AflII-R. The amplified product generated by fusion PCR was subcloned into the backbone of plasmid pPD158.87 using the SbfI and AflII restriction enzymes. 20 μg/μl of the obtained plasmid were microinjected into the germline of N2 worms with 50 μg/μl of pRF4(rol-6(su1006)) plasmid (KB057) as a co-injection marker to generate the strains LD1937 and LD1938. For making plasmids that lack the RRM or RS domains in rsp-6, or substitutions of serine at the 99th amino acid to alanine in rsp-6, DNA was amplified by PCR from the plasmid carrying Prsp-6::rsp-6::gfp::let-858 3’UTR with the primers rsp-6(RRM) Forward and rsp-6(RRM) Reverse, rsp-6(RS) Forward and rsp-6(RS) Reverse, or rsp-6 S99A Forward and rsp-6 S99A Reverse respectively. The amplified products were ligated using the Q5 Site-Directed Mutagenesis Kit (NEB). 20 μg/μl of the plasmids KB058-KB060 were microinjected respectively into the germline of N2 worms with 50 μg/μl of KB057 plasmid as a co-injection marker. KB061 was microinjected respectively into the germline of N2 worms with 50 μg/μl of pRF4 plasmid as a co-injection marker by SunyBiotech.

Integrated strain isolation

LD1271 was used to create LD1943 and LD1937 was used to create LD1945. Ultraviolet light was used to integrate extrachromosomal arrays into chromosome, as described in WormBook ^84,85. L4 stage LD1271 or LD1937 animals were exposed to UV for 30,000 microjoules/cm² using a stratalinker 1800. The integrated strains were back crossed with N2 at least 6 times. These strains were used in the study presented in Figures 4I, 4J, S4F–S4J, 6A–6K, and S6A–S6L.

Nematode culture conditions and maintenance

Nematode strains were maintained on nematode growth medium (NGM) plates seeded with Escherichia coli strain OP50–1 as a food source using standard techniques ⁸⁶ at 20°C. The OP50 bacteria were cultured overnight in Luria-Bertani (LB) Broth medium with 10 mg/L streptomycin at 37°C, and then 1 mL of liquid culture was seeded on NGM plates to grow for 2–4 days at room temperature. E. coli HT115 (DE3) bacteria transformed with vectors expressing target genes or empty vector (pL4440)(EV) were obtained from the Ahringer ⁸⁷ or Vidal ⁸⁸ RNAi libraries. RNAi cultures were grown overnight in LB medium containing 50 μg/mL carbenicillin at 37°C. The cultures were diluted 1:5 in LB containing carbenicillin to re-enter the logarithmic growth phase (~6 hours). The cultures were then concentrated to 1/10 volume by centrifugation and seeded directly onto NGM plates containing 50 μg/mL carbenicillin and 0.2g/L isopropylthio-β-galactoside (IPTG). For diluted RNAi treatment, clones and L4440 control were grown separately in parallel and were adjusted to OD600=1.0 respectively. For controls, clones were diluted with L4440 to be the indicated percentage. Synchronization was performed using alkaline hypochlorite solution. Eggs were incubated in M9 buffer overnight at 20°C with gentle shaking. Newly hatched L1 stage-arrested worms were used, unless otherwise noted. All experiments were conducted at 20°C.

Cell lines

The human renal angiomyolipoma-derived cell line Lymphangioleiomyomatosis (LAM) 621–101, which was immortalized by human papillomavirus (HPV) E6/E7 and human telomerase reverse transcriptase (hTERT), was described previously⁵⁹. It was a gift from Elizabeth Henske’s lab and was cultured in IIA complete media (Dulbecco’s Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/F12), 10% Fetal Bovine Serum (FBS), 1 nM triiodothyronine, 10 mU/mL vasopressin, 200 nM hydrocortisone, 50 nM sodium selenite, 10 nM cholesterol, 20 ng/mL EGF, 25 mg/mL insulin, 10 mg/mL transferrin, and 1.6 mM ferrous sulfate)^59,78,89. Human embryonic kidney (HEK) 293E cells (provided by Brendan Manning) were grown in DMEM (high glucose, with L-glutamine), 10% FBS, and 1% penicillin/streptomycin supplement (all from Invitrogen).

Method Details

High-throughput RNAi screens

Our targeted RNAi libraries were prepared by using the Ahringer and Vidal C. elegans RNAi libraries^87,88. Screening of RNAi clones was performed in 96 well plates prepared with NGM and IPTG for induction of RNAi expression ⁹⁰. 10–15 synchronized eggs were incubated on NGM agar. Phenotypes were scored in sterility, larval development to adulthood, embryonic viability, vulval development and motility. The genetic interactions were scored as ‘strong’ if the effect of an RNAi clone was 100% penetrant for any of the screened phenotypes in the raga-1 mutant and less than 25% penetrant in the WT background. Genetic interactions were scored as ‘moderate’ if the effect of an RNAi clone was 75–100% penetrant for any of the screened phenotypes in raga-1 animals and 25–50% penetrant in WT. Genetic interactions were scored as ‘weak’ if the effect of an RNAi clone was 25%–75% penetrant for any of the screened phenotypes in raga-1 mutant animals and 25–50% penetrant in the WT. The genetic interactions were scored as ‘no effect’ an RNAi clone showed the same penetrance for any of the screened phenotypes in both WT and raga-1 mutant backgrounds. The genetic interactions were scored as ‘none’ if neither WT nor raga-1 mutant exhibited detectable phenotypic interaction with the RNAi clone. Phenotypes were scored three times during the screen (3rd, 4th and 6^th day) following hatching on the RNAi clone. The clones that were identified as strong/moderate were examined at least three times to confirm the results. The key RNAi clones carrying raga-1, ragc-1, rheb-1, daf-15, rsks-1, and rsp-6 were verified by DNA sequencing.

TOR signaling (TOS) motif analysis

An expanded TOS motif consensus (F-[DEAVP]-[FMILV]-[DEVRL]-[MLIEFYAR]) was generated from published TOS motif data^32,91–93, and target sequences in orthologs of the mTORC1 substrates RPS6KB1, EIF4EBP2, ULK1, EIF4EBP1, EIF4EBP3, and STAT3, and mTOR itself in H. sapiens, M. musculus, D. rerio, D. melanogaster, X. tropicalis or C. elegans (See also Table S1). The ScanProsite tool was used to search for this motif in the entire proteome (UniProt) of these 6 species. H. sapiens orthologs were obtained using the DIOPT2 database. All proteins were annotated by their corresponding total TOS motif counts in all orthologs. The proteins with conserved TOS motifs in at least 4 (TOS motif count ≥ 4) out of 6 organisms were marked as ‘conserved TOS motif’ proteins. R and Python (pandas) packages were used for this analysis.

Worm RNA-Seq samples collection

All worms were passaged on NGM with OP50 bacteria for at least two generations before the assay. Adult stage worms were bleached and the collected eggs were incubated in M9 buffer for 24hr for synchronization and induction of starvation-induced arrest. Hatched and starved larvae were placed on NGM plates with OP50 for 4h, then harvested after washing in M9 buffer. Samples were snap-frozen in liquid nitrogen and stored at −80°C until needed. After five freeze-thaw steps in a 37°C water bath and liquid nitrogen, total RNA was extracted using TRIzol, then purified with the Direct-zol^™ RNA Kit and treated with DNase I to remove possible DNA contamination. All worm samples were analyzed using paired-end 150 bp sequencing at Novogene Co., Ltd. Three biological replicates were performed.

RNA-Seq sample collection, libraries preparation and sequencing for LAM cells

80% confluent cells on 60 mm culture dishes were treated with drugs (100 nM Rapamycin, 20 μM PF4708671, 250 nM Torin1, 20 μM SRPIN340) without serum for 24 hr. Total RNA was isolated using PureLink RNA Mini kit and treated with DNase I for 30 min at room temperature. RNAseq was performed by BGI AMERICAS CORPORATION, MA. Briefly, cDNA libraries were constructed using TruSeq Stranded Total RNA Library Prep Kit and paired-end 100 bp sequencing was performed on HiSeq 4000 (Illumina). Three biological replicates were performed.

RNA-Seq samples collection, libraries preparation and sequencing for HEK293E cells

Cells were starved by incubation in DMEM with 0.1 % FBS overnight. Re-feeding was achieved by treatment with regular growth media. Total RNA was extracted from cells using the RNeasy Mini Kit. Purified RNA samples were DNase treated before sending for sequencing at Novogene Co., Ltd. All RNA samples were passed quality control before library construction. The final cDNA library is ready after a round of purification, terminal repair, A-tailing, ligation of sequencing adapters, size selection and PCR enrichment. All HEK293E samples were analyzed using paired-end 150 bp sequencing. Three biological replicates were performed.

Processing and analysis of RNA-Seq data

FASTQ output files were aligned to C. elegans WBcel269 and Homo sapiens GRCh38.98 reference genomes using STAR with a 2-pass procedure ⁹⁴. The annotated genes and transcripts (Aligned.sortedByCoord.out.bam file from STAR outputs) were quantified using RSEM ⁹⁵. Differential expression analysis of genes and transcripts was performed in R using the Bioconductor package DESeq2 ⁹⁶. A threshold of FDR (padj) < 0.01 was set to determine differentially expressed targets.

Alternative splicing analysis

Alternative splicing (AS) events involving two isoforms from an alternatively spliced region were characterized and tabulated as the five most common AS types: skipped exon (SE), retained intron (RI), mutually exclusive exon (MXE), alternative 3′ splice site (A3SS) and alternative 5′ splice site (A5SS). This analysis was performed using rMATS ⁹⁷, and the sorted bam files from STAR outputs. A threshold of FDR < 0.05 was set to determine differentially alternative splicing events.

RNA binding protein motif enrichments analysis

RNA binding protein motif enrichments for the spliced exon events in WT Fed/WT Stv and WT Fed/raga-1 Fed samples were identified by using rMAPS2 tool by analyzing the differential alternative splicing data obtained from rMATS tool. Exon skipping events with at least 5% increase in exon inclusion level (FDR < 0.05) between treatment versus control sample groups were compiled as upregulated exons. Exon skipping events with at least 5% decrease in exon inclusion level (FDR < 0.05) were treated as downregulated exons. For each motif, motif occurrences were scanned separately in exons or their upstream and downstream introns. 250bp upstream or downstream intronic sequences and the first 50bp of the 5’ or 3’ end of exonic sequences were examined. To calculate motif density, 50bp sliding window was used to count the number of nucleotides covered by a given motif. Bonferroni correction was applied by dividing the alpha (0.01) by the number of RBP motifs (114) tested by the rMAPS2 tool. All SR proteins and other RBPs with FDR < 0.01 (−log10(P-value) > 4.057) were labeled as significant enrichments.

Gene structure and domain information

C. elegans gene isoform structures were obtained from WormBase. The domain information diagram for each gene was obtained using SMART (Simple Modular Architecture Research Tool)⁷⁴.

GO term and pathway analysis

GO biological process and pathway enrichment analyses were performed using AmiGO 2 ⁹⁸. Heatmaps and gene expression profile cluster plots were obtained using Morpheus. Genes network graphs were created using STRING ⁷⁹. KEGG pathway analysis was performed using DAVID ⁸⁰. Schematics of metabolic pathways were obtained from the KEGG pathway analysis and Wormflux ⁸¹ and depicted.

Immunoblot assay

Immunoblot assay in LAM cells was analyzed as previously described5. Cell lysates were prepared by washing cells twice with ice-cold PBS and homogenizing them on ice using a lysis buffer (40 mM HEPES [pH 7.4], 1 mM EDTA, 120 mM NaCl, 0.5 mM DTT, 10 mM b-glycerophosphate, 1 mM NaF, 1 mM Na3VO4, 0.1% Brij-35, 0.1% deoxycholate, and 0.5% NP-40) supplemented with protease inhibitors (250 mM PMSF, 5 mg/ml pepstatin A, 10 mg/ml leupeptin, and 5 mg/ml aprotinin). The cell lysates were cleared by centrifugation at 13,000 rpm at 4 C for 20 min. Protein concentration was determined using the Bradford assay (Bio-rad), and the proteins were denatured by boiling for 10 min in a Laemmli sample buffer. Immunoblot signals were detected using the Odyssey imaging system (LI-COR Biosciences).

Quantitative RT-PCR (qRT-PCR)

First-strand cDNA was synthesized in duplicate from each sample by SuperScript III. SYBR green was used to perform qRT-PCR (ABI 7900). Three biological replicates were examined for each sample. Gene expression fold change was calculated using the ΔΔCt method. Other PCR reactions were performed using BioMix^™ Taq enzyme. Quantification of bands was performed using 2% agarose gel. Primer sequences are provided in GETPrime.

Semi-quantitative RT-PCR

RNA extraction was performed on synchronized second-generation worms at the L1 stage following RNAi treatment. Synchronized L1 stage worms were cultured on NGM plates seeded with RNAi bacteria for 3 days. Gravid worms were subjected to hypochlorite treatment for synchronization, and the resulting eggs were allowed to incubate overnight. Subsequently, synchronized L1 stage worms were cultured for 4 hours on NGM plates seeded with RNAi bacteria. Worms were harvested and washed three times with M9 buffer. After being centrifuged at 1000 rpm for 1 minute, the resulting pellet (100 μl) was lysed using RNA-zol and snap-frozen in liquid nitrogen. After undergoing four cycles of freeze-thawing, the lysate was resuspended in 0.12 ml of sterilized water and incubated at room temperature for 15 minutes. The lysate was then centrifuged at 12,000 × g for 10 minutes, and 300 μl of the supernatant was collected. An equal volume of isopropanol was added to the supernatant, followed by a 15-minute incubation and centrifugation at 12,000 × g for 10 minutes. The supernatant was discarded, and the precipitated RNA was washed with 0.4 ml of 75% ethanol and centrifuged at 4000 × g for 3 minutes twice. Following ethanol removal, the RNA pellet was resuspended in nuclease-free water. The extracted RNA was further purified using the RNeasy mini Kit according to the manufacturer’s instructions. The cDNA synthesis was performed using FastGene Scriptase following the manufacturer’s instructions. The synthesized cDNA was subsequently amplified by PCR using EX Premier DNA polymerase and gene-specific primers. PCR reactions were cycled for 35 cycles with an annealing temperature of 57°C.

Splicing reporter assay

Alternative splicing of the ret-1 reporter ⁴² was analyzed in WT or raga-1 animals. For assays during larval stages, synchronized L1 animals were incubated on NGM plates with/without OP50 for 4 hours and then imaged. For the assays in adults, animals were imaged daily from day 1 to day 3. Synchronized day 1 adult animals were transferred onto NGM with/without OP50 and incubated for 24 hours. Day 2 adult animals were transferred to NGM plates with OP50 and incubated for 24 hours. For assays with RNAi against mTORC1 components or SR proteins, synchronized L1 stage ret-1 reporter animals were incubated on NGM seeded with HT115 RNAi bacteria. Animals were imaged on day 2 of adulthood. For the nutrient-response assay at the post-developmental stage, animals were imaged daily from day 1 to day 3 of adulthood. Synchronized worms at the L1 stage were incubated for 3 days. On the first day of adulthood, worms were harvested using M9 buffer and washed three times with M9 buffer to remove E. coli. During each washing step, worms were incubated in M9 buffer for 5 minutes. Subsequently, the worms were placed on NGM plates without bacteria and incubated for 24 hours. The worms were then harvested and washed twice with M9 buffer, and placed on NGM plates with bacterial food lawn for 24 hours.

Metabolomics sample preparation and analysis

All worms were passaged on NGM with OP50 bacteria for at least two generations before the assay. For metabolomics, 25,000 second generation L1 animals treated with RNAi against EV or raga-1 were incubated on NGM with the corresponding food source for 4 hours. Animals were collected and washed 3 times with M9 buffer and once quickly with HPLC-grade water quickly and flash-frozen in liquid nitrogen. Metabolites were extracted using 800μL of 40:40:20 acetonitrile:methanol:water solvent at −20°C, vortexed for 10 sec, and further lysed with a TissueLyser LT (Qiagen) with stainless steel beads. The samples were then centrifuged at 16,000g for 15 minutes at 4°C, and the supernatant was passed through an Ultrafree-MC VV Centrifugal Filter, 0.1 μm (Millipore) by centrifuging at 9100g. The filtrate was concentrated and dried using a Speedvac (Thermo Scientific), and resuspended in 50:50 acetonitrile:water for LC-MS analysis. Metabolites were resolved on a Vanquish U-HPLC system coupled to a Q Exactive HF-X hybrid quadrupole-orbitrap mass spectrometer (ThermoFisher) with a HESI source operating in negative ion mode. The analytes were separated by using an iHILIC column (5μm, 150 × 2.1 mm I.D., HILICON) coupled to a Thermo Scientific SII UPLC system. The iHILIC column was used with the following buffers and linear gradient: Buffer A = water with 20 mM ammonium carbonate with 0.1% ammonium hydroxide, Buffer B = 100% acetonitrile. The gradient was run at a flow rate of 0.150 mL/min as follows: 0–23min linear gradient from 95% B to 5% B; 23–25min hold at 5%B. To waste from 25–25.5min gradient to 95% B at 0.20 mL/min, 25.5–32.5min hold at 95%B and finally 32.5–33min 95%B at 0.15 mL/min. MS data acquisition and targeted feature extraction and quantification were performed using TraceFinder 5.1 or open-source EL-MAVEN software. Peak area integration and metabolite identification were done using accurate mass and retention time curated with in-house standard library compounds. Three biological replicates were used for each condition.

Worms body size and developmental stage measurement

For these assays animals were analyzed in the second generation of RNAi treatment. Synchronized eggs were incubated with RNAi against raga-1, ragc-1, daf-15, and rheb-1 for 72 hours. The body length of at least 20 individuals was measured for each condition using imaging and ImageJ. Developmental stages were scored using an optical microscope, with approximately 80 animals scored for each condition. At least two biological experiments were performed in each case.

Brood size and progeny measurement

Synchronized L1 stage animals were incubated on NGM with HT115 RNAi bacteria. At the L4 stage ten parental worms per condition were singled onto 35 mm plates with the corresponding food source. Animals were transferred to fresh plates daily until the end of their reproductive period. The numbers of live F1 progeny and dead embryos were counted before the onset of fertility. The total number of eggs was calculated as the sum of dead embryos and live progeny. At least two biological replicates of each assay were performed.

Lifespan assays

All lifespan assays were conducted at 20°C. All worms were passaged on NGM with OP50 bacteria for at least two generations before the assay. Gravid adult worms were placed on plates with either OP50 (analyses of genetic mutants) or HT115 carrying EV (RNAi analyses) for 6 hours to lay eggs. After three days of synchronization young adults were transferred to NGM plates containing 50 μM 5-fluoro-2’deoxyuridine (FUdR). After 4–5 days, worms were transferred to fresh plates with FUdR so as not to be starved. Survival was plotted with the young adult as time-point = 0 and was scored every other day. Worms were censored when they were crawled off the plate, hatched inside, or lost vulval integrity. Survival curves were generated using GraphPad Prism 9. The P values were determined by the log-rank (Mantel-Cox) test. At least three independent replicates were examined for each data set. All lifespan data are available in Table S6.

Fluorescence microscopy and image analysis

Animals were randomly selected and immobilized by mounting on a 2% agarose pad with 0.6% tetramisole (Sigma) diluted in M9 buffer. Samples were imaged with ZEN 2012 software on an Axio Imager M2 microscope with a 10X/0.25 objective (Zeiss, Jena, Germany). Fluorescence was quantified blindly using NIH ImageJ.

Quantification and Statistical Analysis

All statistical analyses were undertaken utilizing R or GraphPad Prism. The significance of comparisons between 2 groups was determined by paired or unpaired 2-tailed t tests as indicated. The significance of comparisons between 3 groups or more was determined by one-way ANOVA with Tukey’s multiple comparison. The significance of comparison between 2 groups or more that involved 2 or more conditions was determined by two-way ANOVA with Tukey’s multiple comparison. Worms were raised together under identical conditions and randomly used in each experiment.

Supplementary Material

2

Table S1. Results of genetic and bioinformatic screening, related to Figure 1. Contains a summary of mTOR proteome studies, human orthologs of proteins identified in mTOR proteome studies, a broad TOS motif consensus, and a complete list of genes tested in phenotypic screening results.

3

Table S2. Gene expression in C. elegans, LAM cells, and HEK 293E cells, related to Figures 2 and 7.

4

Table S3. Alternatively spliced events in C. elegans and HEK 293E cells, related to Figures 3 and 7.

5

Table S4. Isoform expression in C. elegans, LAM cells, and HEK 293E cells, related to Figures 2 and 7.

6

Table S5. Metabolites in C. elegans, related to Figures 4 and 6.

7

Table S6. A data table of C. elegans lifespan assay, related to Figure 6.

8

Table S7. Primers used in this study, related to STAR Methods.

Key Resource Table.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Bacterial and Virus Strains
E. coli: Strain OP50-1	Caenorhabditis Genetics Center	#OP50-1; RRID:WB-STRAIN:OP50-1
E. coli: Strain HT115	Caenorhabditis Genetics Center	#HT115; RRID:WB-STRAIN:HT115
Chemicals, Peptides, and Recombinant Proteins
5-Fluoro-2’-deoxyuridine (FUdR)	Thermo Fisher Scientific	AC227600025
TRI Reagent	Sigma	T9424
Isopropyl β-D-1-thiogalactopyranoside (IPTG)	Thermo Fisher Scientific	BP1620-10
Ampicillin	Sigma	A9518
Kanamycin	Sigma	K1377
Tetracycline	Sigma	T7660
Tetramisole hydrochloride	Sigma	L9756
DreamTaq DNA Polymerase	Thermo Fisher Scientific	EP0712
Deoxynucleotide (dNTP) Solution Mix	New England BioLabs	N0447S
Subcloning Efficiency DH5α Competent Cells	Thermo Fisher Scientific	18265017
SbfI	BioLabs	R0642S
AflII	Thermo Scientific	FD0834
BioMix™ Taq enzyme	Meridian Bioscience	BIO-25012
Acetonitrile	Sigma	75-05-8
Methanol	Thermo Fisher Scientific	A456-500
Rapamycin	Sigma-Aldrich	R8781
Torin1	Tocris Bioscience	4247
PF4708671	Sigma-Aldrich	PZ0143
SRPIN340	Selleck Chemicals	S7270
DNase I	Sigma-Aldrich	AMPD1
Critical Commercial Assays
Q5® Site-Directed Mutagenesis Kit	NEB	E0554S
Direct-zol™ RNA Kits	Zymo Research	R2050
RNeasy Mini Kit	QIAGEN	74104
First-Strand Synthesis SuperMix	Thermo Fisher Scientific	11752050
SuperScript III	Invitrogen	18080044
SYBR Green	Thermo Fisher Scientific	11760500
Pierce BCA Protein Assay kit	Thermo Fisher Scientific	23225
PureLink RNA Mini kit	Ambion	12183018A
TruSeq Stranded Total RNA Library Prep Kit	Illumina	RS-122-2201
FastGene™ Scriptase II	Nippon Genetics	NE-LS64
Deposited Data
RNA-seq data	This paper	Gene Expression Omnibus69: GSE272718; GSE273387; GSE273388
Metabolomics data	This paper	Metabolomics Workbench70: PR002071
Experimental Models: Organisms/Strains
C. elegans: Strain N2: Wild-type	Caenorhabditis Genetics Center	N2
C. elegans: Strain VC222: raga-1(ok386) II	Caenorhabditis Genetics Center	VC222
C. elegans: Strain RB1206: rsks-1(ok1255) III	Caenorhabditis Genetics Center	RB1206
C. elegans: Strain VC597: rsp-6(ok798) IV/nT1 [qIs51]	Caenorhabditis Genetics Center	VC597
C. elegans: Strain KH2235: lin-15(n765)ybIs2167[eft-3p::ret-1E4E5(+1)E6-GGS6-mCherry+ eft-3p::ret-1E4E5(+1)E6(+ 2)GGS6-GFP+ lin-15(+)+ pRG5271Neo] X	Kuroyanagi et al, 201342	KH2235
C. elegans: Strain LD1936: raga-1(ok386) II; lin-15(n765)ybIs2167[eft-3p::ret-1E4E5(+1)E6-GGS6-mCherry+eft-3p::ret-1E4E5(+1)E6(+2)GGS6-GFP+lin-15(+)+ pRG5271Neo] X	This paper	LD1936
C. elegans: Strain LD1271: N2; Ex1060[pRF4(rol-6(su1006))]	Tullet et al., 200871	LD1271
C. elegans: Strain LD1937: N2; ldEx32[Prsp-6::rsp-6::gfp::let-858 3’UTR + rol-6(su1006)]	This paper	LD1937
C. elegans: Strain LD1938: N2; ldEx33[Prsp-6::rsp-6::gfp::let-858 3’UTR + rol-6(su1006)]	This paper	LD1938
C. elegans: Strain LD1939: N2; ldEx34[Prsp-6::rsp-6 ΔRRM::gfp::let-858 3’UTR+ rol-6(su1006)]	This paper	LD1939
C. elegans: Strain LD1941: N2; ldEx36[Prsp-6::rsp-6 ΔRS::gfp::let-858 3’UTR + rol-6(su1006)]	This paper	LD1941
C. elegans: Strain LD1969: N2; ldEx38[Prsp-6::rsp-6 S99A::gfp::let-858 3’UTR + rol-6(su1006)]	This paper	LD1969
C. elegans: Strain LD1943:N2; ldIs38[pRF4(rol-6(su1006))]	This paper	LD1943
C. elegans: Strain LD1945: ldIs40[Prsp-6::rsp-6::gfp::let-858 3’UTR+ rol-6(su1006)]	This paper	LD1945
C. elegans: Strain LD1960: raga-1(ok386); Ex1060[pRF4(rol-6(su1006))]	This paper	LD1960
C. elegans: Strain LD1961: rsks-1(ok1255); Ex1060[pRF4(rol-6(su1006))]	This paper	LD1961
C. elegans: Strain LD1962: raga-1(ok386); ldEx32[Prsp-6::rsp-6::gfp::let-858 3’UTR + rol-6(su1006)]	This paper	LD1962
C. elegans: Strain LD1963: rsks-1(ok386); ldEx32[Prsp-6::rsp-6::gfp::let-858 3’UTR + rol-6(su1006)]	This paper	LD1963
Cell line: Human renal angiomyolipoma-derived LAM 621-101 (TSC2-/-)	Yu et al., 200459	RRID: CVCL_S879
Cell line: SRPK2 knockout	Lee et al., 201717	N/A
Cell line: HEK 293E	Dr. Brendan Manning	N/A
Oligonucleotides
Primers	This paper	Table S7
recombinant DNA
Plasmid: KB057: pRF4(rol-6(su1006))	Tullet et al., 200871	KB057
Plasmid: pPD158.87	Addgene	1709
Plasmid: KB058: Prsp-6::rsp-6::gfp::let-858 3’UTR	This paper	KB058
Plasmid: KB059: Prsp-6::rsp-6 ΔRRM::gfp::let-858 3’UTR	This paper	KB059
Plasmid: KB060: Prsp-6::rsp-6 ΔRS::gfp::let-858 3’UTR	This paper	KB060
Plasmid: KB061: Prsp-6::rsp-6 S99A::gfp::let-858 3’UTR	This paper	KB061
Software and Algorithms
ScanProsite	de Castro et al., 200672	https://prosite.expasy.org/scanprosite/
UniProt	The UniProt Consortium, 202073	https://www.uniprot.org
WormBase version WS269	Alliance of Genome Resources	https://wormbase.org/#–-5
SMART	Leyunic and Bork, 201874	https://smart.embl.de
AmiGO 2 version 2.5	Ashburner et al., 2000, The Gene Ontology Consortium75,76	http://amigo.geneontology.org/amigo
ImageJ version 1.51	U. S. National Institutes of Health	https://imagej.nih.gov/ij/
SMALT version 0.7.4	Wellcome Trust Sanger Institute	http://www.sanger.ac.uk/science/tools/smalt-0
Subread version 1.4.6	Liao et al., 201477	http://subread.sourceforge.net
edgeR version 3.20.9	Anders et al., 201378	https://bioconductor.org/packages/release/bioc/html/edgeR.html
Morpheus	Broad Institute	https://software.broadinstitute.org/morpheus
STRING version 11	Szklarczyk et al., 202179	https://string-db.org
DAVID version 6.8	Huang da et al., 200980	https://david.ncifcrf.gov
Wormflux version iCEL1314	Yilmaz and Walhout, 201681	http://wormflux.umassmed.edu
GETPrime version 2.0	David et al., 201782	https://gecftools.epfl.ch/getprime/
TraceFinder 5.1	Thermo Fisher Scientific	https://www.thermofisher.com/us/en/home/industrial/mass-spectrometry/liquid-chromatography-mass-spectrometry-lc-ms/lc-ms-software/lc-ms-data-acquisition-software/tracefinder-software.html
EL-MAVEN version 0.10.0	Elucidata	https://elucidata.io/el-maven/
MetaboAnalyst version 4.0 or 5.0	Pang et al., 202183	https://www.metaboanalyst.ca
GraphPad Prism version 9	GraphPad Software Inc	https://www.graphpad.com

Highlights.

• During C. elegans growth mTORC1 broadly controls mRNA splicing and expression

• mTORC1 increases expression and activity of SR proteins and other splicing factors

• mTORC1 controls growth-related metabolic pathways by regulating mRNA splicing

• Regulating metabolic genes through mRNA splicing is a conserved mTORC1 function