- Split View
-
Views
-
Cite
Cite
Samantha C Chomyshen, Hadi Tabarraei, Cheng-Wei Wu, Translational suppression via IFG-1/eIF4G inhibits stress-induced RNA alternative splicing in Caenorhabditis elegans, Genetics, Volume 221, Issue 3, July 2022, iyac075, https://doi.org/10.1093/genetics/iyac075
- Share Icon Share
Abstract
Splicing of precursor mRNA is an essential process for dividing cells, and splicing defects have been linked to aging and various chronic diseases. Environmental stress has recently been shown to modify alternative splicing, and molecular mechanisms that influence stress-induced alternative splicing remain unclear. Using an in vivo RNA splicing reporter, we performed a genome-wide RNAi screen in Caenorhabditis elegans and found that protein translation suppression via silencing of the conserved eukaryotic initiation factor 4G (IFG-1/eIF4G) inhibits cadmium-induced alternative splicing. Transcriptome analysis of an ifg-1-deficient mutant revealed an overall decrease in intronic and intergenic reads and prevented cadmium-induced alternative splicing compared to the wild type. We found that the ifg-1 mutant up-regulates >80 RNA splicing regulatory genes controlled by the TGF-β transcription factor SMA-2. The extended lifespan of the ifg-1 mutant is partially reduced upon sma-2 depletion and completely nullified when core spliceosome genes including snr-1, snr-2, and uaf-2 are knocked down. Depletion of snr-1 and snr-2 also diminished the enhanced cadmium resistance of the ifg-1 mutant. Together, these data describe a molecular mechanism through which translation suppression inhibits stress-induced alternative splicing and demonstrate an essential role for RNA splicing in promoting longevity and stress resistance in a translation-compromised mutant.
Introduction
Posttranscriptional splicing of precursor mRNAs (pre-mRNA) is an essential step in the regulation of eukaryotic gene expression and contributes to proteome diversity (Baralle and Giudice 2017). Precision in RNA splicing is necessary for maintaining cellular homeostasis; this includes alternative splicing of select transcripts into isoform variants in specific tissues, or in response to particular developmental cues to coordinate organismal growth (Wang et al. 2008; Nilsen and Graveley 2010). Errors in RNA splicing can disrupt cellular homeostasis, driven by the assembly of aberrant mRNA molecules that lead to the production of erroneous proteins with abnormal functions and detrimental consequences (Dutertre et al. 2011). A decline in RNA splicing fidelity has been reported during aging and is prominently linked to many chronic human diseases including neurodegeneration and cancer (Tollervey et al. 2011; Scotti and Swanson 2016; Deschênes and Chabot 2017). However, underlying cellular mechanisms that influence splicing fidelity remains poorly understood. Uncovering mechanistic insights into alternative splicing regulation is intriguing as small molecules that modulate RNA splicing have been explored for their potential as chemotherapy agents (Eskens et al. 2013).
In the genetic model Caenorhabditis elegans, it was recently demonstrated that RNA splicing fidelity deteriorates with age and this dysregulation is hypothesized to be a central contributor to the decline in transcriptome and protein homeostasis (Heintz et al. 2017). Interventions that extend lifespan such as dietary restriction can improve splicing fidelity during aging; conversely, loss of splicing homeostasis nullifies dietary restriction-induced longevity, suggesting a codependency of dietary restriction and splicing homeostasis in extending lifespan. In Rhesus monkeys, caloric restriction extends lifespan and leads to modifications of RNA splicing as evident by the up-regulation of various spliceosome components at the transcriptome and proteome levels in the liver (Rhoads et al. 2018). This suggests that a causative relationship between RNA homeostasis and dietary restriction-induced longevity in C. elegans may be conserved in higher organisms. Next to dietary restriction, other interventions that significantly extend C. elegans lifespan include reduced insulin and TOR signaling (Kenyon et al. 1993; Robida-Stubbs et al. 2012), decreased mitochondrial respiration (Felkai et al. 1999; Feng et al. 2001), and inhibition of protein translation (Hansen et al. 2007; Pan et al. 2007); however, whether maintaining RNA splicing is essential for the longevity phenotype of these interventions remain to be characterized.
Recently, we showed that RNA splicing is disrupted in C. elegans upon exposure to the heavy metal cadmium (Wu et al. 2019). This demonstrated that environmental stress may promote aging by negatively influencing RNA homeostasis. In this study, we performed a genome-wide RNAi screen and found that knockdown of protein translation via reduced translation initiation factor ifg-1 (initiation factor 4G family) inhibits cadmium-induced alternative splicing. This inhibitory effect is dependent on the transcription factor gene sma-2, which is required for the expression of >80 genes involved in RNA splicing that is up-regulated in the ifg-1 mutant. Knockdown of select core RNA splicing regulators reduced the lifespan and cadmium survival of the ifg-1 mutant, implicating an essential role for RNA splicing factors in promoting longevity and stress resistance of this translation compromised mutant. Overall, our results suggest a model where reduced protein translation up-regulates the expression of RNA splicing regulatory genes via the SMA transcription factors to modify alternative splicing under stress to promote longevity and stress resistance.
Materials and methods
C. elegans strains
C. elegans strains were grown and maintained at 20°C using standard methods (Brenner 1974). The following strains were used: wild-type N2 Bristol, KH2235 lin-15(n765) ybIs2167[eft-3::ret-1E4E5(+1)E6-GGS6-mCherry+eft-3::ret-1E4E5(+1)E6(+2) GGS6-EGFP+lin-15(+)+pRG5271Neo]X, ifg-1(cxTi9279). The KX52 ifg-1(cxTi9279) II; bcIs39 V strain was used to isolate the ifg-1(cxTi9279) allele followed by 4 rounds of outcrossing with N2.
Genome-wide RNAi screen and RNAi experiments
RNAi was performed as described previously by feeding C. elegans strains of engineered Escherichia coli [HT115(DE3)] that transcribed double-stranded RNA (dsRNA) homologous to a target gene (Wu et al. 2016, 2019). Approximately 19,000 dsRNA from the ORFeome RNAi feeding library (Open Biosystems, Huntsville, AL, USA) and the MRC genomic RNAi feeding library (Geneservice, Cambridge, UK) were used for screening. Briefly, synchronized L1 larvae of KH2235 C. elegans obtained from the alkaline hypochlorite method were grown in liquid nematode growth medium (NGM) with dsRNA producing bacteria for 2 days, followed subsequently by exposure to 300 µM cadmium chloride for 24 h and manually screened for relative fluorescence of the ret-1 GFP reporter under an Olympus SZX61 stereomicroscope. Clones that resulted in the retention of ret-1 GFP signal after 24 h cadmium exposure were recorded and rescreened 3 additional times for verification. All subsequent RNAi experiments were performed on NGM agar plates in the presence of 50 µg mL−1 carbenicillin and 100 µg mL−1 isopropyl β-d-thiogalactopyranoside. E. coli expressing the pPD129.36 (LH4440) plasmid encoding 202 bases of dsRNA that are not homologous to any predicted C. elegans gene, or referred to as empty vector (EV), was used as an RNAi control.
RNA extraction and RT-PCR
Total RNA was extracted with the Invitrogen Purelink RNA mini kit (ThermoFisher, 12183020) according to the manufacturer’s instructions using a QSonica Q55 sonicator. RNA was extracted from 4 biological replicates per condition, with each sample containing 200–300 synchronized worms. RNA samples were treated with DNAse I (ThermoFisher, EN0521) before cDNA synthesis using the Invitrogen Multiscribe reverse transcriptase (ThermoFisher, 4311235). Analysis of ret-1 alternative splicing was carried out with primers previously described using the Applied Biosystems ProFlex thermocycler (Heintz et al. 2017). RT-PCR product was separated on a 2.5% agarose gel and stained with SYBR Green I nucleic acid gel stain (ThermoFisher, S7563) for band visualization using a BioRad Gel Doc EQ system.
Fluorescent microscopy
Preparation of C. elegans for fluorescent imaging was carried out as previously described (Murray et al. 2020). Fluorescent images were captured with a Retiga R3 camera mounted to a Zeiss Axioskop 50 microscope. GFP fluorescence was detected using filter sets with Ex: 460–495 nm/Ex: 510–550 nm and RFP fluorescence was detected using filter sets with Ex: 540–580 nm/Em: 610 nm. Grayscale images were converted to color images using ImageJ (Schneider et al. 2012). Worm images used to quantify relative fluorescence were prepared by mounting 8 worms per slide immobilized by 2% sodium azide on a 2% agarose pad. Relative fluorescence was calculated in ImageJ using the Measure function. Background signal was subtracted from each image by defining an area on the same image where fluorescence signal was absent, with the dimension of the defined background constant for all images. Composite images displaying both GFP and RFP fluorescence were created with the Merge Channel function in ImageJ.
Lifespan and cadmium survival assay
Lifespan assays were performed at 20°C on NGM RNAi plates seeded with HT115 E. coli with the EV dsRNA plasmid used for control. To circumvent the larval arrest phenotype of most RNAi tested in this study, synchronized L1 worms were first grown on EV dsRNA until day 1 of adulthood, after which the worms were transferred to RNAi plates seeded with the corresponding dsRNA to initiate knockdown of the gene of interest. For 2 dsRNA that did not affect larval development (sma-2, sma-3), RNAi was initiated at the L1 stage. Lifespan assays were performed without the use of FUDR with progeny separation accomplished via daily picking during the worm’s reproductive window. For cadmium survival assay, synchronized L1 worms were first grown on EV dsRNA until day 1 of adulthood followed by transfer to RNAi plates seeded with the corresponding dsRNA for 48 h to initiate gene knockdown. Three-day-old worms were then transferred to corresponding RNAi seeded NGM plates containing 300 µM of cadmium chloride. For both lifespan and survival assays, worms were considered dead if they did not respond to gentle prodding with a metal pick and censored when they displayed protruding gonad or vulva. Survival assays were scored daily while lifespan assays were scored every 2 days. Three independent trials were performed for each assay, with the number of animals used in each experiment described in the corresponding supplementary table.
Whole-transcriptome sequencing and data analysis
Wild-type and ifg-1(cxTi9279) worms were synchronized at the L1 stage and grown on EV dsRNA NGM plates until day 1 of adulthood followed by transfer to EV dsRNA NGM plates with or without 300 µM of cadmium chloride for 24 h. Total RNA from wild-type and ifg-1(cxTi9279) worms with and without cadmium treatment were extracted from 3 biological replicates for each condition with ∼2,000-3,000 worms per replicate. RNA samples were sent to Novogene (Sacramento, CA, USA) on dry ice for cDNA library preparation with oligo(dT) enrichment followed by RNA-sequencing and statistical analysis. HISAT2 (v2.0.5) was used to map sequencing reads to the WBcel235 genome, with FeatureCounts (v1.5.0-p3) used for gene quantification, DESeq2 (v1.20.0) for differentially analysis, and rMATS (v3.2.5) for alternative splicing analysis (Anders and Huber 2010; Shen et al. 2014; Kim et al. 2019). Gene ontology enrichment was performed using DAVID functional analysis (Huang et al. 2009). Fold enrichment is shown that is defined as the ratio between observed (occurrence within the provided gene list) and expected (occurrence within the transcriptome) frequency of a given gene ontology term. Heat maps showing differentially expressed genes were analyzed using Gene Cluster 3.0 with uncentered average linkage and visualized with Java TreeView (v1.2.0). Raw sequence data and processed data are deposited to the NCBI Gene Expression Omnibus with the accession number GSE184491. Raw data from previously published RNA-sequencing of sma-2(rax5) and sma-4(rax3) were retrieved from the NCBI Sequence Read Archive with the BioProject ID PRJNA395096 (Yu et al. 2017). The Galaxy (usegalaxy.org) open-source platform was used to analyze differential gene expression data using the aforementioned methods (Afgan et al. 2018).
Measurement of motility
Wild-type and ifg-1(cxTi9279) worms were synchronized at the L1 stage and grown on EV dsRNA NGM plates until day 1 of adulthood followed by transfer to EV dsRNA NGM plates with or without 300 µM of cadmium chloride for 24 or 48 h. C. elegans swimming rate (body bends per second in liquid) was measured by re-suspending worms grown on a 6-cm NGM agar plate with ∼5 ml of M9 buffer. Worms were allowed to acclimate to the liquid condition for at least 1 min, after which the swimming behavior was recorded by a Retiga R3 camera mounted to an Olympus SZX61 stereomicroscope for 30 s. The video file was processed with the wrMTrck program in ImageJ to calculate body bends per second (Nussbaum-Krammer et al. 2015).
Statistical analyses
Graphical data and statistical analysis were generated using the GraphPad Prism software (v7.04). For statistical analysis, the Student’s t-test was performed when 2 means were compared. One -way ANOVA with Tukey’s correction was used for multiple comparisons with 1 factor, and 2-way ANOVA with Sidak correction was used for multiple comparisons with 2 factors. F-test for linear regression was used to determine statistical significance when comparing relationships between 2 sets of RNA-sequencing data, Fisher’s exact test was used to determine statistical significance between Venn diagram overlap genes. For lifespan and survival analysis, the OASIS2 software (https://sbi.postech.ac.kr/oasis2/) was used to calculate mean lifespan, 95% confidence interval, and statistical significance between conditions using the Log-rank test (Han et al. 2016). For all statistical tests *P < 0.05, **P < 0.01, and ***P < 0.001, false discovery rate (FDR) correction was used for all P-values involving RNA-sequence datasets.
Results
Reduced protein translation inhibits cadmium-induced alternative splicing
Exposure to environmental stress are known to inhibit pre-mRNA splicing (Biamonti and Caceres 2009). We recently showed that the heavy metal cadmium can alter RNA splicing in C. elegans, potentially by disrupting the snRNA processing function of the Integrator complex (Wu et al. 2019). To obtain a visual biomarker to study cadmium-induced alternative splicing, we used an in vivo fluorescent alternative splicing reporter constructed with the C. elegans ret-1 gene (Kuroyanagi et al. 2013). Briefly, this C. elegans splicing reporter strain expresses a pair of integrated fluorescent ret-1 minigenes driven by the ubiquitous eft-3 promoter. The mCherry reporter is modified with a + 1 nucleotide insertion to exon 5, while the GFP reporter is modified with +1 insertion to exon 5 and +2 insertion to exon 6 (Fig. 1a). The nucleotide insertions alter the native reading frame of each fluorescent reporter, resulting in mCherry fluorescence when exon 5 is skipped or GFP fluorescence when exon 5 is included (Kuroyanagi et al. 2013; Heintz et al. 2017). Using this splicing reporter, we show that exposure to 300 µM of cadmium for 24 h results in a strong loss of GFP fluorescence in the intestinal tissue, indicating an increased incidence of exon 5 skipping (Fig. 1b). We did not observe noticeable changes to mCherry fluorescence that is primarily expressed in the nervous system and body wall muscle, this may be due to potential differences in cadmium presence in these tissues compared to the intestine that is the primary site of xenobiotic bioaccumulation and the major tissue of the C. elegans heavy metal stress response (Girard et al. 2007). As such, we focused specifically on changes to the GFP fluorescence as an indicator for cadmium-induced splicing alterations for this study.
To verify alternative splicing occurs to the endogenous ret-1 gene, we exposed wild-type worms with 0, 100, 200, and 300 µM of cadmium followed by RNA extraction and RT-PCR analysis with primer pairs that bind to exon 4 and exon 6 of the endogenous ret-1 gene (Heintz et al. 2017). Two isoforms of the ret-1 transcript are amplified and can be visualized via agarose gel electrophoresis, with the top band representing the full ret-1 transcript expressing exons 4, 5, and 6, and the bottom band representing the skipped ret-1 transcript expressing only exons 4 and 6 (Fig. 1c). Exposure to incremental concentration of cadmium increased the relative proportion of the bottom band representing the skipped isoform. The effects of cadmium on ret-1 alternative splicing appears to be stronger in the splicing reporter compared to RT-PCR. The relative variance may be attributed by the fact that GFP fluorescence in the splicing reporter is primarily expressed in the intestine tissue while RT-PCR measured the endogenous splicing patterns of ret-1 using whole worm RNA samples. As such, the effects of cadmium on ret-1 alternative splicing may be stronger in the intestine compared to the whole animal.
We then performed a genome-wide RNAi screen to identify genetic regulators of cadmium-induced alternative splicing (Fig. 1d). We screened ∼19,000 dsRNA clones and identified 74 gene knockdowns that resulted in the retention of the ret-1 GFP reporter fluorescence after cadmium exposure. Using DAVID functional analysis, we found a strong enrichment for protein translation processes among the 74 positive hits from the RNAi screen (Fig. 1e), with the majority of the RNAi targeting genes encoding small or large ribosomal proteins and various protein translation initiating factors (Supplementary Table 1). Interestingly, secondary confirmation screen revealed that knockdown of several genes encoding translation factors reduced GFP fluorescence under basal conditions compared to the EV control. However, all candidate genes tested also resulted in the retention of GFP fluorescence after cadmium exposure relative to the EV control (Fig. 1f). Overall, these findings revealed that suppression of protein translation inhibits cadmium-induced alternative splicing.
Suppression of ifg-1 enhances C. elegans survival
It is well established that reducing protein synthesis via depletion of various translation-related genes extends lifespan in C. elegans (Hansen et al. 2007; Pan et al. 2007). We selected 4 genes from the RNAi screened that demonstrated high retention of ret-1 GFP reporter fluorescence after cadmium exposure to determine their role in lifespan and cadmium resistance regulation. These 4 genes were also chosen as they function at different steps of protein translation including ribosomal structural protein, translation initiation factor, and ribosome biogenesis process. RNAi depletion initiated at day 1 of adulthood of rps-23 (human homolog: ribosome protein S23) and ifg-1 (human homolog: eIF4G) extended mean lifespan in all 3 independent trials whereas knockdown of F57B9.3 (human homolog: eIF4A2) and Y61A9LA.10 (human homolog: BMS1 ribosome biogenesis factor) extended mean lifespan in 2 out of 3 trials (Fig. 2a). Interestingly, only RNAi depletion to ifg-1 and rps-23 led to a consistent significant increase in cadmium survival relative to EV control (Fig. 2b). Knockdown of rps-23 increased mean survival by an average of 12.9% under normal conditions and by 43.4% under cadmium. Meanwhile, knockdown of ifg-1 had a smaller variation with a mean survival increase of 17.8% under normal condition and an increase of 23.5% under cadmium (Supplementary Table 2). The variable effect on cadmium survival suggests that knockdown of different translation regulators may invoke distinct downstream signals, which may include an overlapping effect on longevity but not stress survival. We decided to focus our downstream analysis on ifg-1 given that its knockdown produced the strongest inhibition to cadmium-induced alternative splicing (Fig. 1f) and consistently extended C. elegans survival under normal and cadmium conditions (Fig. 2, a and b).
Given that RNAi from the genome-wide screen was initiated at the L1 stage, knockdown of ifg-1 resulted in the larval arrest of C. elegans at the L2/L3 stage. To verify that knockdown of ifg-1 inhibition against cadmium-induced ret-1 alternative splicing is not due to RNAi-induced developmental delay, we first attempted to introduce a viable ifg-1(cxTi9279) partial loss of function mutation background into the ret-1 splicing reporter strain (Morrison et al. 2014); however, we were not able to recover viable homozygous offspring (data not shown). Given that the ret-1 splicing reporter has been shown to gradually lose a substantial amount of GFP fluorescence as early as 5 days old through aging (Heintz et al. 2017), it restricts the comparison of cadmium-induced reduction in GFP fluorescence to only young worms. To minimize developmental differences between EV and ifg-1 RNAi within an age window where ret-1 GFP is highly fluorescent, we initiated RNAi depletion of ifg-1 at the L2/L3 stage for 48 h followed by 24 h of cadmium exposure. This ensured that comparisons were made when worms are 2 days old where ret-1 GFP is still highly expressed. Worms with ifg-1 depleted via RNAi at L2/L3 were able to reach adulthood and retained high levels of ret-1 GFP fluorescence after cadmium exposure compared to the EV control (Fig. 2, c and d). To verify that alternative splicing changes are also observed to the endogenous ret-1 transcript, we extracted total RNA from wild-type and ifg-1(cxTi9279) worms with and without cadmium treatment. RT-PCR analysis showed that cadmium exposure increased the skipped/full ratio of the ret-1 transcript in the wild type, and this ratio remained unchanged in the ifg-1(cxTi9279) mutant (Fig. 2e). Overall, these findings revealed that reduction of protein translation via depletion of ifg-1 increases C. elegans resistance to cadmium and confers inhibition against cadmium-induced alternative splicing.
The transcriptome of ifg-1 mutant resembles a cadmium-induced stress response
To investigate how reduced ifg-1 expression protects against cadmium toxicity, we first analyzed transcription changes in wild-type and ifg-1(cxTi9279) worms treated with and without cadmium by whole-transcriptome RNA sequencing (Fig. 3a). RNA-sequencing revealed that ifg-1(cxTi9279) worms show a 57% reduction in the ifg-1 mRNA level (Fig. 3b), indicating that this partial loss of function mutant derives from decreased transcript abundance. This reduction is comparable to a previous study demonstrating a 42% decrease in ifg-1 mRNA in this mutant as determined via Northern blot (Morrison et al. 2014). The ifg-1(cxTi9279) worms displayed a dramatically altered transcriptome compared to the wild type, with 1,423 up-regulated genes and 4,246 down-regulated genes (>2-fold change, Padj < 0.05 Supplementary Table 3). KEGG pathway analysis revealed that ifg-1(cxTi9279) up-regulated genes enriched to antioxidant stress response pathways and protein metabolism, while down-regulated genes enriched to pathways regulating neuroactivity along with calcium and Wnt signaling pathways (Fig. 3c). Up-regulation of stress response genes in ifg-1 depleted worms was similarly demonstrated previously through translation state array analysis, which showed an increased abundance of stress genes associated with highly translated ribosome fractions (Rogers et al. 2011).
A clustered heat map analysis demonstrated that gene expression changes to the ifg-1(cxTi9279) mutant were highly similar to wild type after cadmium exposure (Fig. 3d). Genes that were up-regulated in both ifg-1(cxTi9279) mutant and wild type after cadmium exposure enrich to metabolic functions and KEGG pathways involved in antioxidant detoxification, including cytochrome P450 metabolism, glutathione activity, and ABC transporters. Meanwhile, genes down-regulated in both conditions enrich to metabolic enzyme functions including endopeptidase inhibitor and hydrolase (Fig. 3d). For cadmium-induced gene expression changes greater than 2-fold, linear regression analysis revealed a highly significant (P < 0.001) Pearson correlation coefficient of +0.65 when compared to expression changes to the same genes in the translation compromised ifg-1(cxTi9279) mutant (Fig. 3e). Previous studies have shown that suppression of protein synthesis is commonly observed in response to cellular stress as an adaptive mechanism to maintain proteostasis (Harding et al. 2003). Consistently, a previous study in human cells showed that exposure to cadmium inhibits protein synthesis (Ovelgönne et al. 1995). These results suggest that mutations to ifg-1(cxTi9279) mimic a state of cadmium stress at the transcriptome level and that the overlap in gene expression changes between ifg-1(cxTi9279) and cadmium may both be in response to attenuated translation efficiency.
Cadmium-induced alternative splicing is modified in the ifg-1 mutant
Through RNA-sequencing, we next compared the transcriptome-wide alternative splicing of the ifg-1(cxTi9279) mutant relative to wild type worms treated with cadmium. This is accomplished by first determining the inclusion level (IncLevel) of a given transcript that is defined as a ratio between 2 alternatively spliced transcript isoforms detected (Fig. 4a), followed by calculating their relative differences between the experimental conditions (IncLevel difference) to determine significant alternative splicing events (Shen et al. 2014). Cadmium exposure in wild type worms resulted in 437 significant alternative splicing events, 209 of these events were not significantly different in the ifg-1(cxTi9279) mutant (Fig. 4b). This indicated that ifg-1 mutants showed a 48% decrease in significant cadmium-induced alternative splicing events observed in the wild type. RNA-seq data also showed that cadmium treated ifg-1(cxTi9279) mutants significantly increased ret-1 exon 5 inclusion compared to the wild type by ∼18% (ifg-1 IncLevel = 0.673 ± 0.039, WT = IncLevel 0.569 ± 0.012; Supplementary Table 3), supporting the results obtained from the splicing reporter and RT-PCR (Fig. 2, c–e). Next, we performed a linear regression analysis of IncLevel difference between cadmium treated ifg-1(cxTi9279) and cadmium treated wild type for the 437 alternative splicing events. This analysis revealed a negative Pearson correlation of −0.37 (P < 0.001), resulting from majority of the data points positioned within the upper left and lower right quadrant where alternative splicing events occurs in the reverse direction in the ifg-1(cxTi9279) mutant relative to the wild type under cadmium (Fig. 4c). This suggested that the relative degree of cadmium-induced alternative splicing changes observed in wild type worms is diminished in the ifg-1(cxTi9279) mutant. Correlation analysis for each alternative splicing type revealed that while 4 out of 5 categories showed a negative correlation coefficient, only exon skipping and alternative 3′ splice site (A3SS) coefficients are statistically significant (Supplementary Fig. 1). It should be noted that exon skipping and A3SS account for 73% of all alternative splicing events induced by cadmium (320/437), and the lack of statistical significance may be due to a relatively small number of splicing events in other categories.
Next, we analyzed the relative percentage of intron and intergenic reads between the wild-type and ifg-1(cxTi9279) mutant. Given that the cDNA sequencing library was constructed with poly-A capture enrichment, these 2 parameters are indicators of aberrant splicing and have previously been shown to increase with age and after spliceosome inhibition (Heintz et al. 2017). In the ifg-1(cxTi9279) mutant, the relative percentage of both intron reads and intergenic reads are significantly lower compared to the wild type, suggesting that alternative splicing is augmented in the ifg-1 (cxTi9279) mutant (Fig. 4d).
Functional analysis of the 209 genes accounting for 48% of cadmium-induced alternative splicing observed in the wild type that are not significantly altered in the ifg-1(cxTi9279) mutant revealed enrichment toward biological functions related to muscle contraction and cell adhesion located within the actin cytoskeleton cellular compartments (Fig. 4e). No enrichment to KEGG pathway was observed likely due to the small number of genes. Given that cadmium-induced alternative splicing of genes functioning in muscle regulation, we measured the swimming behavior of wild-type and ifg-1(cxTi9279) mutant after 24 and 48 h of cadmium exposure. Cadmium exposure for 24 h led to a 21.8% decrease in wild-type swimming motility as determined by body bends per second, and this reduction is significantly improved in the ifg-1(cxTi9279) mutant where only a 10.9% decrease was observed after cadmium exposure (Fig. 4f). Similar trends were also observed after 48 h cadmium exposure. It should be noted these experiments only show an association of improved motility in the ifg-1 mutant under cadmium, as it was not determined in this study whether cadmium-induced alternative splicing of transcripts related to muscle functions are directly detrimental to motility. Together, these results revealed that cadmium-induced alternative splicing observed in wild type worms are reduced in the ifg-1(cxTi9279) mutant and that these modifications may provide resistance against cadmium-induced impairment to muscle functions.
SMA transcription factors regulate ifg-1-mediated gene expressions
To gain mechanistic insights on how reduced ifg-1 provides inhibition against cadmium-induced alternative splicing, we performed a small-scale RNAi screen to identify transcription factors that are required to maintain ret-1 GFP fluorescence after ifg-1 depletion and cadmium exposure. The rationale for this approach is that potential transcription factors may act downstream of reduced ifg-1 to modulate RNA splicing via changes to gene expression. We first initiated ifg-1 depletion via RNAi in L2/L3 worms for 48 h, followed by feeding with ∼500 dsRNA clones targeting C. elegans transcription factors for another 48 h before treatment with cadmium (Fig. 5a). In ifg-1-depleted worms subsequently fed with EV RNAi as a control, we show that these worms are able to maintain high GFP fluorescence after cadmium exposure (Fig. 5b). This suggested that a secondary RNAi treatment did not reduce the penetrance of the initial ifg-1 depletion. Using this protocol, we identified 2 dsRNA clones corresponding to the sma-2 and sma-3 genes that suppressed ifg-1 induced retention of ret-1 GFP following cadmium exposure (Fig. 5b). Quantitative analysis revealed that knockdown of sma-2 after ifg-1 depletion resulted in a significant decrease of GFP levels after cadmium exposure, which was unchanged in ifg-1 depleted worms fed with EV control. While knockdown of sma-3 also reduced GFP levels after cadmium exposure, the decrease was not statistically significant. This may be attributed by sma-3 knockdown causing a significant reduction in GFP levels under basal conditions compared to the EV control, thereby reducing the relative difference in GFP levels after cadmium exposure (Fig. 5b). The C. elegans sma-2 and sma-3 encode receptor-regulated Smad proteins within the TGF-β/BMP signaling pathway that was initially characterized for its role in body size regulation (Savage et al. 1996; Savage-Dunn and Padgett 2017). Given the defined role of SMA proteins functioning as transcription factors, we compared the transcriptome of a sma-2(rax5) mutant previously described with a defect in lipid metabolism to the transcriptome of the ifg-1(cxTi9279) mutant from this study (Yu et al. 2017). We observed a striking overlap between genes that are up-regulated in the ifg-1(cxTi9279) mutant but down-regulated in the sma-2(rax5) mutant (FDR > 0.05, no FC cut-off) (Fig. 5c), with 79% (2,899/3,680) of genes down-regulated in the sma-2(rax5) mutant found to be up-regulated in the ifg-1(cxTi9279) mutant. Functional analysis of the 2,899 overlapped genes revealed enrichment toward RNA metabolism pathways including RNA transport and the spliceosome (Fig. 5d). Other KEGG pathways enriched include proteasome, nucleotide excision repair, and ER processing. We then analyzed the 2 RNA-sequencing datasets and manually identified 88 genes functioning in various processes in RNA splicing regulation that were up-regulated (1.2- to 1.9-fold) in the ifg-1(cxTi9279) mutant but down-regulated in the sma-2(rax5) mutant (Fig. 5e and Supplementary Table 4).
To further examine the relationship between ifg-1 and sma-2, we performed linear regression analysis on the expression of 12,114 genes that are differentially expressed (FDR < 0.05) in the ifg-1(cxTi9279) mutant against their corresponding expression in the sma-2(rax5) mutant and observed a Pearson correlation coefficient of −0.27 (P < 0.001) (Fig. 5f). Reciprocal analysis of 5,840 genes differentially expressed in the sma-2(rax5) mutant also revealed similar patterns with a Pearson correlation coefficient of −0.41 (P < 0.001) (Fig. 5g). While no RNA-sequencing data was available for a sma-3 mutant, similar results were also observed when comparing the transcriptome of a sma-4(rax3) mutant to the ifg-1(cxTi9279) mutant showing an overlap of 3,699 genes (Supplementary Fig. 2a). Consistently, linear regression analysis on gene expression data between the 2 mutants revealed a Pearson correlation coefficient of −0.39 (P < 0.001) and −0.52 (P < 0.001) when ifg-1(cxTi9279) or sma-4(rax3) differentially expressed genes were analyzed respectively (Supplementary Fig. 2, b and c).
To further investigate the relationship between ifg-1 and sma genes, we found that transcript levels of sma-2, sma-3, and sma-4 genes are all up-regulated in response to cadmium, and this response is further enhanced in the ifg-1(cxTi9279) mutant compared to the wild type (Supplementary Fig. 3, a–c). This suggests that the ifg-1 mutant positively promotes the expression of sma encoding genes as a potential mechanism to influence the transcription of sma dependent splicing regulatory genes. Interestingly, in both the sma-2(rax5) and sma-4(rax3) mutants, transcript levels of ifg-1 are significantly decreased by 24% and 29%, respectively, compared to the wild type (Supplementary Fig. 3d and Supplementary Table 5). This reduction in ifg-1 mRNA level is comparable to that observed in the ifg-1(cxTi9279) mutant (Fig. 3b) and potentially explains the significant transcriptome overlap of reciprocal changes between the ifg-1(cxTi9279) mutant to the sma-2(rax5) and sma-4(rax3) mutants (Fig. 5c and Supplementary Fig. 2a). Overall, our results demonstrated that knockdown of sma-2 suppressed ifg-1(cxTi9279)-mediated inhibition toward cadmium-induced alternative splicing, which may be attributed to the requirement for sma-2 in regulating transcription of RNA splicing regulatory genes that are up-regulated in the ifg-1(cxTi9279) mutant.
Core RNA-splicing regulatory genes are required for ifg-1 longevity and stress resistance
Our data suggest that sma and RNA-splicing factors function downstream of reduced ifg-1. To test this, we used RNAi to knockdown sma-2, sma-3, and 4 RNA splicing regulatory genes that were up-regulated in the ifg-1(cxTi9279) mutant and down-regulated in either the sma-2(rax5) or sma-4(rax3) mutant to determine their effects on wild type and ifg-1(cxTi9279) lifespan and stress resistance. We had attempted to generate an ifg-1;sma-2 double mutant but the combination of these 2 mutations appears to be synthetic lethal as no homozygous worms carrying both mutations could be recovered (data not shown). The mean lifespan of ifg-1(cxTi9279) is significantly increased by ∼25% compared to the wild type; this is consistent with previous reports (Fig. 6a and Supplementary Tables 6 and 7) (Pan et al. 2007; Rogers et al. 2011). Knockdown of sma-2 and sma-3 slightly reduced ifg-1(cxTi9279) lifespan by 7% and 8%, respectively (Fig. 6, a and b; P < 0.05 in 2/3 trials for sma-2 and 3/3 trials for sma-3; Supplementary Table 6), but did not reduce the lifespan of wild type worms. Consistent with our ifg-1 RNAi data (Fig. 2b), the ifg-1 mutant also shows increased resistance against cadmium compared to the wild type; however, knockdown of either sma-2 or sma-3 did not reduce this resistance (Fig. 6, a and b and Supplementary Table 8). This suggests that sma-2 and sma-3 are not required for the ifg-1 cadmium survival and is consistent with previous evidence demonstrating that ifg-1 confers its stress resistance phenotype largely via activation of stress response genes dependent on the transcription factor skn-1 (Wang et al. 2010).
We next investigated whether 4 RNA-splicing regulatory genes up-regulated in the ifg-1(cxTi9279) mutant are required for its longevity and stress-resistant phenotypes. RNAi knockdown of small nuclear ribonucleoproteins snr-1 and snr-2 beginning at day 1 of adulthood significantly reduced wild type lifespan by an average of 15.8% and 18.5%, respectively, and completely nullified the long-lived phenotype of the ifg-1(cxTi9279) mutant with snr-1 RNAi causing a 27.2% decrease and snr-2 RNAi causing a 33.2% decrease in mean lifespan (Fig. 6, c and d and Supplementary Tables 7 and 8). Similarly, knockdown of both snr-1 and snr-2 significantly reduced the stress resistance of the wild type worms by an average of 10.5% and 22.6%, respectively, and in the ifg-1(cxTi9279) mutant by an average of 7.5% and 25.6%, respectively (Fig. 6, c and d and Supplementary Tables 7 and 8). Interestingly, knockdown of the U2 small nuclear RNA auxiliary factor uaf-2 beginning at day 1 of adulthood also completely nullified ifg-1(cxTi9279) mutant’s long-lived lifespan (22% decrease in mean lifespan) but did not affect the wild type lifespan (Fig. 6e and Supplementary Tables 7 and 8). Knockdown of uaf-2 also did not decrease stress resistance of the wild type worms and slightly reduced the ifg-1(cxTi9279) mutant stress resistance in 1 out of 3 trials (Fig. 6e and Supplementary Tables 7 and 8). Meanwhile, RNAi knockdown of the serine arginine-rich protein rsp-2 did not affect either the wild type or ifg-1(cxTi9279) mutant lifespan or stress resistance (Supplementary Fig. 4 and Supplementary Tables 7 and 8). Together, these results show that the extent to which different splicing factors are required for ifg-1 longevity and stress resistance are variable, and core spliceosomal encoding genes such as snr-1 and snr-2 that have a direct influence on transcriptome-wide splicing dynamics and gene expression regulation are completely essential for the longevity and stress-resistant phenotypes of the ifg-1(cxTi9279) mutant.
Discussion
RNA splicing is an essential step in regulating eukaryotic gene expression and is a major contributor to transcriptome and proteome diversity (Dutertre et al. 2011). Aberrant RNA splicing is an underlying factor in many human diseases, including cancer and neurodegeneration (Tollervey et al. 2011; Scotti and Swanson 2016; Deschênes and Chabot 2017). Recently, it was demonstrated that RNA splicing deteriorates with age in C. elegans, suggesting a causative link between RNA splicing homeostasis and healthy aging (Heintz et al. 2017). To date, only a handful of studies have examined the effects of environmental stress on RNA splicing homeostasis (Yost and Lindquist 1986; Biamonti and Caceres 2009; Dutertre et al. 2011; Wu et al. 2019), and molecular mechanisms that inhibit stress-induced alternative splicing have not yet been described. Here, we show that protein translation suppression in C. elegans via reduced ifg-1 influences cadmium-induced alternative splicing (Figs. 1–4), and we propose that this mechanism signals through the SMA family of transcription factors to regulate the expression of RNA splicing regulatory genes in influencing alternative splicing to promote longevity and stress resistance (Figs. 5 and 6g).
Environmental stress-induced alternative splicing
In response to stress, it is well established that cells can mount an adaptive response such as transcription of damage repair genes to elicit cellular protection. These responses can include up-regulation of chaperone genes to facilitate protein folding in response to heat stress or an increase in expression of xenobiotic detoxification genes in response to oxidative stress (An et al. 2005; Brunquell et al. 2016). However, the functional significance of stress-induced alternative splicing is not well understood, as modifications to transcript splicing could indicate an adaptive response or as a consequence of pathological events. Evidence for stress-induced alternative splicing with detrimental consequences has been reported in response to heat stress, where widespread alternative splicing of ∼1,700 genes is observed in the mouse fibroblast that is associated with the impaired translation of the alternatively spliced transcripts (Shalgi et al. 2014). We previously showed that cadmium exposure alters the Integrator complex's function, resulting in the aberrant polyadenylation and misprocessing of several snRNA transcripts that normally assemble with ribonucleoproteins in forming a functional spliceosome (Matera et al. 2007; Baillat and Wagner 2015; Wu et al. 2019). Disruption of the Integrator complex in C. elegans is deleterious and impairs RNA splicing leading to widespread transcriptome dysregulation (Gómez-Orte et al. 2019). RNAi knockdown of the Integrator complex via subunit-4 depletion results in embryonic lethality and produces an alternative splicing profile that is significantly correlated to the alternative splicing pattern observed after cadmium exposure (Wu et al. 2019). While these evidence do not rule out the possibility that some of the cadmium-induced alternative splicing events are an adaptive response, the resemblance in alternative splicing events between Integrator dysfunction and cadmium exposure suggest that cadmium-induced alternative splicing events may also be deleterious.
Influence of translational suppression on RNA splicing and aging
It is well established that reduced protein translation via mutation or knockdown of various ribosomal and translation factors can significantly extend lifespan in worm, fly, and yeast (Kapahi et al. 2004; Kaeberlein et al. 2005; Hansen et al. 2007; Pan et al. 2007; Curran and Ruvkun 2007; Steffen et al. 2008; Zid et al. 2009; Rogers et al. 2011). Mechanistically, a conserved cellular response to reduced protein translation includes the selective up-regulation and increased translation of various genes involved in stress signaling (Zid et al. 2009; Rogers et al. 2011). Consistently, suppression of protein translation in worms provides enhanced resistance to various environmental stress including hypoxia, thermotolerance, oxidative stress, heavy metal stress, and ER stress (Anderson et al. 2009; Wang et al. 2010; Howard et al. 2016). The mechanism behind lifespan extension induced by protein translation suppression is likely to be multifactorial, given that transcriptome-wide alteration to gene expression is observed (Fig. 3), and presumably, to the proteome as well. It is hypothesized that the combination of enhanced stress resistance and global reduction to protein translation both contribute to the longevity phenotype (Steffen and Dillin 2016). Enhanced stress resistance may provide an increase in cellular repair response to mitigate the progressive buildup of molecular damage accumulated through aging (Gems and Partridge 2013). Our results show that the ifg-1 mutant up-regulates various stress-responsive genes involved in xenobiotic detoxification (Fig. 3). This is consistent with a previous report showing enhanced oxidative stress resistance in ifg-1 depleted worms, which is dependent on the SKN-1 transcription factor that functions as a master regulator of antioxidant defense (An et al. 2005; Wang et al. 2010; Wu et al. 2017). Interestingly, while loss of skn-1 abolishes ifg-1’s oxidative stress resistance, it had minimal impact on its longevity (Wang et al. 2010), suggesting that enhanced oxidative stress response may not be the primary factor in promoting longevity in translation compromised mutants.
Meanwhile, reduced translation may directly decelerate the proposed hypertrophic effect of protein synthesis in later life, which is supported by evidence that translation fidelity decreases with age leading to an increase in the accumulation of translational errors (Gems and Partridge 2013; Anisimova et al. 2018). Mild inhibition of translation has also been shown to increase protein folding accuracy in mammalian cells, suggesting a positive effect of reduced translation on enhancing overall protein synthesis fidelity (Meriin et al. 2012). Similar to translation, RNA splicing efficiency also declines with age, a deterioration that can be protected through dietary restriction in C. elegans (Heintz et al. 2017). In this study, we provide evidence that translation inhibition via reduced ifg-1 also modifies RNA splicing; however, this may be accomplished via independent mechanisms compared to dietary restriction that did not show global changes to splicing factor mRNA expression (Fig. 5e). Expression of splicing regulators is a determinant factor in mouse and human longevity (Lee et al. 2016), and down-regulation of splicing factors is observed with increasing chronological age in men (Holly et al. 2013). In calorically restricted Rhesus monkeys that lived on average 20% longer than the control group, transcriptome analysis of liver biopsy samples showed up-regulation of various components of the spliceosome (Rhoads et al. 2018). Though it should be cautioned that the correlative relationship between splicing factor expression and aging likely differs between organisms and can vary across different tissues, and functional studies are required to determine their relative contribution to aging. In this study, we showed that different splicing factors had varying effects on C. elegans lifespan and stress resistance. Depleting rsp-2 encoding an SR protein had no effect on either the lifespan or stress resistance whereas knockdown of snr-1 and snr-2 encoding core spliceosome proteins reduced both of these phenotypes in the wild type and ifg-1 mutant worms. Furthermore, knockdown of uaf-2 encoding the U2 auxiliary factor only reduced ifg-1 lifespan but not the wild type. It should be noted that a previous study that introduced uaf-2 RNAi beginning at the L1 stage instead of day 1 adulthood used in this study showed a significant reduction in wild type lifespan (Heintz et al. 2017). These data suggest a differential requirement of splicing factors in lifespan and stress resistance regulation that is dependent on both the genetic background and expression during development.
Why does translational suppression influence RNA splicing? Under periods of environmental stress, cells naturally cope by suppressing global protein translation via eIF2α phosphorylation (Wek et al. 2006). This allows cells to redirect energy resources to prioritize other essential needs, including enhancing RNA processing to limit the potential stress-induced disruption to RNA splicing. While cells can protect against mis-spliced mRNA molecules via RNA surveillance mechanisms such as nonsense-mediated decay (NMD), costs associated with degradation of an erroneous transcript is energetically expensive given that it costs at least 10 ATP molecules to perform splicing of each intron and that NMD relies on the activity of various ATP-dependent hydrolysis reactions (Lynch and Marinov 2015; Serdar et al. 2016). Modifications to RNA splicing may also prevent the translation of erroneous transcripts that can cost on average 100–200 molecules of ATP to degrade (Peth et al. 2013; Lynch and Marinov 2015). Given that the combined energetic cost of RNA transcription and splicing are magnitudes lower than the cost of protein translation and degradation, these modifications may ensure the production of accurate mature transcripts before translation to avoid loss of energy investment toward the synthesis and degradation of aberrant proteins.
SMA signaling in aging
TGF-β/Sma signaling is a critical developmental pathway in C. elegans initially discovered for its role in regulating body size growth and male tail morphogenesis (Savage et al. 1996; Savage-Dunn and Padgett 2017). In the last decade, a role for TGF-β/Sma signaling in aging has also been described. Loss of function mutations to sma genes delay reproductive aging characterized by an extended offspring production window supported by improved oocyte and germline quality (Luo et al. 2009, 2010). In the oocyte of sma-2 mutants, reproductive aging is delayed via transcriptional up-regulation of genes involved in cell cycle regulation, DNA damage repair, and cell death programming, while down-regulating genes involved in oviposition, proteolysis, and cell differentiation (Luo et al. 2010). Recently, a role for SMA in regulating lipid metabolism has also been described, with sma-2 and sma-4 mutants transcriptionally up-regulating various β-oxidation genes to reduce lipid storage in C. elegans (Yu et al. 2017; Clark et al. 2018). These transcriptional programs appear to be distinct from the regulation of genes involved in RNA metabolism described in this study, suggesting a pleiotropic role for the TGF-β/Sma signaling in regulating diverse cellular functions. Our study shows that depletion of either sma-2 or sma-3 does not shorten wild type lifespan but partially reduces the longevity of the ifg-1 mutant. This suggests a differential requirement of sma genes in influencing lifespan that is dependent on the genetic background, where SMA may function as critical transcription factors in long-lived mutants to regulate the expression of genes required to potentiate longevity. This genotype-specific requirement for sma genes in longevity was also recently demonstrated in an insulin signaling mutant where loss sma-3 partially reduced the longevity of the daf-2(e1370) worms (Clark et al. 2021).
In summary, we have described a new mechanism through which suppression of protein translation provides inhibition against stress-induced alternative splicing in C. elegans. We show that reducing the translation initiation factor ifg-1 up-regulates the expression of RNA splicing regulators and inhibits stress-induced alternative splicing. Overall, these findings reveal novel mechanistic insights into RNA splicing regulation under stress and highlight its importance in promoting longevity and stress resistance.
Data availability
All datasets supporting this manuscript are found within the article and its supplementary files. RNA-sequencing data generated from this study (raw and annotated) are available on the NCBI GEO data repository GSE184491. Strains are available upon request.
Supplemental material is available at GENETICS online.
Acknowledgments
Some C. elegans strains were provided by the Caenorhabditis Genetic Centre (University of Minnesota, Minneapolis, MN, USA), which is supported by the NIH Office of Research Infrastructure Programs (P40 OD010440). We thank Dr. Hidehito Kuroyanagi (Tokyo Medical and Dental University) for providing the KH2235 strain and Dr. Brett Keiper (East Carolina University) for providing the ifg-1(cxTi9279) strain.
Funding
C.-W.W. is supported by an Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant (04486) and an Saskatchewan Health Research Foundation (SHRF) Establishment Grant (4971). S.C.C. and H.T. were supported by a Western College of Veterinary Medicine (WCVM) graduate teaching fellowship award.
Author contributions
S.C.C. and H.T. performed the experiments and S.C.C. analyzed the results. C.-W.W. designed the study, performed experiments, and analyzed results. C.-W.W. and S.C.C. wrote the manuscript.
Conflicts of interest
None declared.