Strains and Culture

The wild type strain N2 of C. elegans was grown under standard conditions at 18^°C and fed with Escherichia coli OP50 and cholesterol [33]. The worms were synchronized by killing all the larval and adult worms by immersion in sodium hypochlorite, a condition in which only eggs can survive [33]. After synchronization, worms were seeded on a Petri dish and fed with E. coli OP50 until they reached early L4 larval stage; half of the larvae were washed with M9 buffer to eliminate all bacteria, after the washes the larvae were incubated on a Petri dish without bacteria, while the other half were normally fed. After 12 hrs, both samples (well-fed and 12-hr starved) were washed with M9 buffer (42.26 mM Na₂HPO₄; 22.04 mM KH₂PO₄; 85.56 mM NaCl; 0.87 mM MgSO4).

Length measurement

One drop of M9 buffer-containing well-fed and 12-hr starved early L4 larvae were laid on a Neubauer chamber for length measurements; 30 larvae were measured for each condition, and photographed under a Nikon SMZ800 stereoscopic microscope.

Body fat assessment

Well-fed and 12-hr starved early L4 larvae were washed three times in M9 buffer. The pellets of larvae were resuspended in 120 ul of PBS 1X and an equal volume of buffer MRWB 2X (160 mM KCl; 40 mM NaCl; 14 mM Na₂EGTA; 0.2% β-mercaptoethanol), containing 2% of paraformaldehyde. The larvae were agitated at 1000 rpm for 1 hr at room temperature, washed 3 times with PBS 1X to eliminate the paraformaldehyde. Samples were suspended in isopropanol 60% to dehydrate, incubated for 15 min at room temperature and were suspended in 60% Oil-Red-O stain (prepared as follows: from 0.5 g/100 mL isopropanol stock solution equilibrated for several days, freshly diluted with 40% water, allowed to sit 10 min and filtered through 0.2 to 0.4 μm). Larvae were photographed under a 20X objective in an Olympus microscope BX51W1 coupled to a Disk Spinning Unit.

Brood size quantification

Well-fed L4 larvae or larvae subjected to starvation as described previously, were placed on individual 3-cm NGM plates and fed with E. coli OP50 strain (n = 21, three replicates for each condition). After three days, larvae produced by each worm were counted under a stereoscopic microscope (Nikon SMZ800).

Gonad size estimation

Well-fed and 12-hr starved L4 larvae were washed three times in M9 buffer; after the last wash, the supernatant was drained and 300 ul of -20°C methanol was added and left for 5 min. Then 200 ul of PBS-T (PBS 1X, 0.1% Tween 20) were added and centrifuged at 3000 rpm for 1 min. Two washes with 500 ul of PBS-T were performed. Finally, one drop containing the worms plus one drop of 100% glycerol were placed on a microscope slide and covered with a cover slip. The worms (n = 5) were observed under an Olympus Laser Scanning Confocal Microscope (100X objective). Area estimation was calculated with the ImageJ program (http://imagej.nih.gov/ij/), and reported as pixels/micron².

Lifespan assessment

Well-fed and 12-hr starved early L4 larvae were fed until they reached the adult stage, then 100 worms from each condition were seeded on a Petri dish. Dead worms were counted daily, and withdrawn with a platinum wire. To get rid of embryos and L1 larvae, the Petri dishes were washed every other day.

RNA isolation, cDNA library preparation and Illumina deep sequencing

Small RNAs were purified and size-selected by using the miRNeasy mini kit and minelute columns following manufacturer’s instructions (Qiagen). The cDNA libraries were prepared from 5 ug of RNAs <200 nt. Preparation of cDNA libraries for the Illumina deep sequencing experiment was performed using the DGE-Small RNA Sample Prep Kit ver. 1.0 (Illumina) according to the manufacturer’s instructions. Briefly, RNAs between 20 and 30 nt in size were purified and ligated to adapters and amplified by RT-PCR. Purified DNA was captured on an Illumina flow cell for cluster generation. Libraries were sequenced for 36 cycles on an Illumina Genome Analyzer IIx using Illumina’s protocols for single-end reads (DGE-Small RNA Cluster Generation Kit and 36 Cycle Solexa Sequencing Kit). The raw deep-sequencing data and processed data are available from GEO GSE67711.

C. elegans genome and non-coding RNA annotation

The full genome, the protein-coding and non-coding transcripts annotation were downloaded from WormBase version WS235 in FASTA format [34]. Sequences for all known C. elegans miRNA hairpins were obtained from miRBase version 20 [6]. Mature sequences were also obtained from this database, but a Perl script was used to extend them by 3 bases at the 5’ end and 5 bases at the 3’ end, when permitted by the length of the hairpin. This extension facilitates sRNA-seq reads to map directly to mature miRNAs. All FASTA sequences were then concatenated and converted into an index for bowtie version 0.12.9 [35].

Processing and mapping sRNA-Seq reads

An artificial hexamer (ACATCG) was present at the 5’ end of 6–8% of all reads, so a custom Perl script was used to trim these occurrences. We then used Reaper to process the FASTQ files [36], and remove the Illumina sRNA 3’ adapter sequence (ATCTCGTATGCCGTCTTCTGCTTGC). Since we were interested in miRNAs, only cleaned reads between 16 and 28 nucleotides were kept. These sequences were collapsed to unique reads using tally [36], assigning their total counts to each sequence identifier. The resulting FASTA files were mapped to the concatenated sequences described in the previous section using bowtie [35], searching for end-to-end hits with at most 2 mismatches, and allowing reads to hit up to 100 different locations. All further processing was performed with ad hoc shell and R scripts using several packages from the Bioconductor project [37]. In particular, reads were first assigned to a class with the following preference: miRBase mature microRNA > miRBase microRNA hairpin > WormBase non-coding transcript > WormBase coding transcript > WormBase intergenic. Reads assigned to a microRNA hairpin are those that mapped outside the annotated mature regions; we expect these to mostly represent reads coming from the loop region, a byproduct of miRNA biogenesis. Once a read was assigned to a class, the observed counts for each unique read were divided between any multiple locations within the same class. Any read that ended with a divided count of less than 1 was ignored. The collapsed sequences were also processed with the miRanalyzer free-web server tool [38] to predict the presence of previously unknown miRNAs candidates, using the default parameters, allowing 1 mismatches, and a threshold of the posterior probability to consider a new miRNA of 0.9.

Differential expression of miRNAs

For differential expression analysis we only considered reads that preferentially mapped to miRBase sequences. We only considered miRNA regions with at least 1 reads per-million from one of the libraries. These were tested for differential expression using the edgeR package [39], setting the common dispersion to 0.1 (since we did not have biological replicates) and normalizing with the TMM method. False Discovery Rates were calculated using the Benjamini & Hochberg procedure [40].

Predicted miRNA-mRNA target interactions

We searched for predicted miRNA targets involved in metabolism or development, using microRNA.org [41], miRanda [42] and TargetScan [43] websites. These results are presented in S2 and S3 Tables, with references indicating the involvement of the target genes in metabolism or development included in S1 File.

qRT-PCR

Stem-loop qRT-PCR was used for the quantification of miRNAs, to provide an enhanced sensitivity and specificity as compared to linear primers [44,45]. Stem-loop RT primers include a modification to provide the Universal ProbeLibrary Probe #21 sequence binding site into the primer stem region [46]. All reactions were carried out in triplicate, both biological and technical. Primers used for Reverse Transcription of miR-35-3p, miR-36-3p, miR-39-3p, miR-240-5p, miR-246-3p, and miR-58-3p were RTCEL35, RTCEL36, RTCEL39, RTCEL240, RTCEL246, and RTCEL58, respectively (S1 Table). Primers used for Reverse Transcription of lin-23, gld-1 and β-actin were LIN23-reverse, GLD-reverse and ACT-reverse, respectively (S1 Table). For Reverse Transcription of miRNAs (miR-35-3p, miR-36-3p, miR-39-3p, miR-240-5p, miR-246-3p, and miR-58-3p) and mRNAs (gld-1, lin-23 and β-actin), a total of 2000 ng and 300 ng were used, respectively. The Reverse Transcription mix contained RNA, oligonucleotide (250 fmol) and water for a final volume of 4 ul, and was incubated at 70°C for 5 min. Then, dNTPs, buffer for RT and Reverse Transcriptase (RevertAid H Minus Reverse Transcriptase, Fermentas) were added and the mix was incubated at 37°C for 5 min, 42°C for 1 hr and 70°C for 10 min.

For miRNA´s qPCR quantification, a total of 2000 ng of cDNA was used for each reaction (each reaction was done in duplicate). Primers pairs were FCEL35/UPR, FCEL36/UPR, FCEL39/UPR, FCEL240/UPR, FCEL246/UPR, and FCEL58/UPR for miR-35-3p, miR-36-3p, miR-39-3p, miR-240-5p, miR-246-3p, and miR-58-3p, respectively (S1 Table). Primers and the hydrolysis probe (Universal ProbeLibrary Probe #21, Roche) were used at 1 μM and 0.1 μM, respectively. Reactions were done in a final volume of 18 ul, using Master Mix 1X (LightCycler TaqMan Master) in a LighCycler 2.0 (Roche) equipment. The program used for qPCR was: 1 cycle (at 95°C for 2 min), 40 cycles (at 94°C for 15 secs, 45°C for 30 secs, 70°C for 30 secs), 1 cycle (at 70°C for 15 min). The expression of miR-58-3p was used as an endogenous control. The quantification of miR-35-3p, miR-36-3p, miR-39-3p, miR-240-5p, and miR-246-3p expression relative to miR-58-3p was calculated as in Pfaffl, 2001[47].

For gld-1, lin-23 and β-actin mRNA qPCR quantification, we used 100 ng of cDNA. Primer pairs were LIN23forward/LIN23-reverse, GLD-forward/GLD-reverse, ACT-forward/ACT-reverse, respectively (S1 Table). Primers were used at a concentration of 50 nM and SYBR Green PCR Master Mix (Applied Biosystems) was used. Reactions were run in a StepOne Applied Biosystems equipment. The program used for qPCR was: 1 cycle (at 95°C for 2 min), 40 cycles (at 95°C for 15 secs, 50°C for 30 secs, 70°C for 1 min), 1 cycle (at 70°C for 15 min). The β-actin gene was used as the endogenous control. The relative expression ratio of gld-1 and lin-23 mRNAs relative to β-actin mRNA expression was calculated as previously described[47].

Statistics

In this study, statistical analyses were performed using GraphPad Prism software (San Diego, CA). A Wilcoxon Signed Rank Test was used to compare length differences between well-fed and 12-hr starved worms. For brood size analysis, a Student´s t-test was performed. The significance of analysis of longevity was performed using Kaplan-Meier survival curves method and Log-rank (Mantel-Cox) test. Differential (upregulated or downregulated) expression of miR-35-3p and target mRNAs gld-1 and lin-23 (qRT-PCR) was also analyzed using Wilcoxon Signed Rank Test. A p-value ≤ 0.05 was considered to be statistically significant.

Data deposition

All the raw deep-sequencing data and processed data are available from the NCBI Gene Expression Omnibus (GEO) database under accession number GSE67711.

Fasting for 12-hrs has dramatic effects on the physical appearance, lipid accumulation, brood size, gonad size, and lifespan in early L4 larvae of C. elegans

We determined the effect of fasting for 12-hrs on C. elegans larvae. Starved-larvae were thinner and shorter in length compared to well-fed animals. The length of the animals was 0.75 ± 0.01 mm and 0.59 ± 0.01 mm (mean ± standard error of the mean) for the well-fed and starved animals, respectively (Fig 1A).

Download:

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

Fig 1. Physical aspect, length and lipid content of well-fed and 12-hr starved early L4 larvae.

(A) Length of well-fed and 12-hr starved early L4 larvae, n = 30 early L4 larvae per group, mean ± standard error of the mean, p < 0.0001 according to a Wilcoxon Signed Rank Test. (B) Lipid content of well-fed and 12-hr starved early L4 larvae stained with Oil-Red-O (20X objective).

https://doi.org/10.1371/journal.pone.0142262.g001

Because animals subjected to even 4–6 hrs of fasting can deplete their fat reserves [48], we analyzed the amount of lipids in well-fed vs. 12-hr starved animals. Fasted animals showed a significant decrease in the amount of lipids, maybe as a consequence of the catabolism of lipids as a mechanism to re-establish homeostasis (Fig 1B). A similar result has been reported for starved worms in feeding defective mutants [49]. These results show that starvation of early L4 larvae has dramatic effects on the physical appearance and the accumulated lipids in the worm. We also observed a very dramatic effect on the fertility of L4 larvae subjected to a 12-hr starvation, as the progeny diminished by 78% in contrast to the well-fed group. The brood size was 123 ± 5 and 27 ± 1 larvae, mean ± SEM) for the well-fed and 12-hr starved worms, respectively) (Fig 2A). In general, a reduction in reproductive ability has been documented as a consequence of the lack of food, since energy is allocated into cellular processes needed for survival [50]. In particular, fecundity has been found to correlate with the amount of bacterial food source in C. elegans [51]. Several pathways have been shown to be involved in such phenomena, such as the Insulin/IGF, mTOR (let-363), and cytochrome P450 (DAF-9/CYP450), steroid hormone Δ7-dafachronic acid (DA), and nuclear hormone receptor NHR-8 [52]. This result displays a large effect on fertility produced by starvation.

Download:

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

Fig 2. Effect of a 12-hr starvation on brood size and gonad formation.

(A) Total larval progeny per hermaphrodite at day 3, n = 21, p-value<0.0001 according to a Student’s t-test. (B) Representative images of the proximal gonad from 12-hr starved vs. well-fed early L4 larvae, as seen by DIC microscopy (100X objective). (C) Estimated area of the proximal gonad, n = 5 early L4 larvae per group, mean ± SEM, p-value <0.0079 according to a Wilcoxon Signed Rank Test.

https://doi.org/10.1371/journal.pone.0142262.g002

Food-deprivation of L4 larvae can result in L4 or adult arrest, adult matricide of bagging, adult reproductive diapause and eugenic germ line starvation response, causing defects in gametogenesis, gonad development, reproductive competence and longevity [53,54]. Which of the different outcomes arises depends on the precise time of starvation onset and its duration [53,54]. We found that after a 12-hr starvation treatment, the area occupied by the proximal gonad was dramatically decreased (Fig 2B and 2C). A diminished gonad could represent one of the many factors that could explain the reduction in progeny that we observed (Fig 2A). These results are consistent with previous reports, as when L4 larvae were subjected to starvation, a decrease in the size of their gonad was observed [54].

Another aspect that we wanted to examine was if starvation for 12 hrs produced changes in the worm’s lifespan. Dietary restriction, a reduction in caloric uptake without malnutrition, can increase the lifespan in different organisms such as yeasts, worms, flies, rodents, non-human primates and possibly even in humans [55]. We found that early L4 larvae fasted for 12 hrs showed an increased lifespan (16.9 d), compared to that of well-fed worms (12.4 d). The increase was 36% (Fig 3). This amount is comparable in magnitude to the lifespan extension provoked from a complete removal of food during adulthood in C. elegans [56].

Download:

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

Fig 3. Kaplan-Meir survival curves of well-fed (n = 117) and 12-hr starved (n = 112) early L4 larvae.

Starved worms showed a significantly increased lifespan (*) compared to well-fed early L4 larvae, according to a log-rank (Mantel-Cox) test (p < 0.0001).

https://doi.org/10.1371/journal.pone.0142262.g003

In summary, a 12-hr starvation period during early L4 larvae produced worms that were thinner and shorter than well-fed animals, with a decreased lipid accumulation, diminished progeny, reduced gonad size, and an increased lifespan.

Deep sequencing analysis of miRNAs from early L4 C. elegans subjected to food starvation for 12 hrs.

Purified RNA, enriched in the range of 20 to 30 nucleotides, obtained from well-fed early L4 larvae and early L4 larvae starved for 12 hrs, was used to construct two independent libraries. We obtained 30,270,857 and 29,656,856 total reads from the well-fed and fasted libraries, respectively. After removing adapter sequences and selecting trimmed reads between 16–28 bases, we kept 23,724,403 and 22,626,023 reads, of which the fraction that mapped to mature miRNAs was 92% and 89% for well-fed and starved early L4 larvae, respectively (see Fig 4A and 4B). By far, the most highly-expressed miRNAs were miR-58-3p and miR-1-3p, accounting for 50.1% and 19.8%, respectively, of the reads that mapped to miRNAs in well-fed larvae and 48.5% and 18.2%, under starvation (Fig 5). Neither miRNA changed in a significant manner in well fed versus 12-hr starved larvae (Fig 5). These expression levels were similar to a previous report for normally fed L4 larvae stage worms [57]. In particular, the expression of miR-58-3p was found to be in the range of 44–54% and that of miR-1-3p was between 22%-32%, when measured in embryo, L1, L2, L3, L4 and adult worms [57].

Download:

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

Fig 4. Functional classification of sRNA-seq reads.

Reads were classified according to the type of sequence to which they mapped (see Materials and Methods). (A) Well-fed, and (B) 12-hr starved early L4 larvae.

https://doi.org/10.1371/journal.pone.0142262.g004

Download:

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

Fig 5. The most abundant miRNAs in each library.

The most abundant miRNAs in well-fed and 12-hr starved early L4 larvae are represented as the percentage of reads that mapped to known miRNAs.

https://doi.org/10.1371/journal.pone.0142262.g005

C. elegans miR-58-3p is homologous to bantam in Drosophila that has the role of scaling dendrite growth to larval growth in epithelial cells and also controls dendrite/axon regeneration in the peripheral nervous system [58,59]. In C. elegans, miR-58-3p is a member of a highly expressed family that also includes miR-80, miR-81, miR-82 and miR-1834 [60]. While miR-58-3p single mutants show no developmental, functional or viability deficits, mutants deleted for four members of the mir-58 family are deficient in body size, egg laying, locomotion and cannot form dauer larvae [60]. Several mRNA targets have been identified, including an mRNA cap-binding factor of the eIF4E family (ife-3), proteins involved in DNA replication, repair and recombination (rpa-1), chromatin remodeling (isw-1), chromatin binding and regulation of the RNAi mechanism (gfl-1), or in transcription like RNA Polymerase II (B) subunit (rpb-7), among others [61]. It is interesting to note that other members of the mir-58 family are also amongst the top 10 most abundant in our analysis, with miR-80-3p and miR-81-3p each contributing about 1% of the miRNA-mapping reads. Comparing the abundances of miR-58-3p to that of miR-81-3p and miR-80-3p, it is clear that the first dominates, contributing more than 90% of the reads assigned to the whole family, as reported by Kato et al. [57].

The second most abundant miRNA, miR-1-3p, functions at neuromuscular junctions and is one of the most ancient animal miRNAs, with conserved expression during musculature differentiation in bilaterians [62,63]. Mutants of miR-1-3p are resistant to levamisole-induced paralysis mediated by the increased expression of its targets unc-29 and unc-63, coding for a non-alpha and an alpha subunit of the nicotinic acetylcholine receptor, respectively. Both subunits mediate fast actions of acetylcholine and bind levamisole [62]. In Drosophila, the homologue of miR-1-3p is required for proper growth of larval muscle [64].

Members of the let-7 miRNA family, as well as miR-228-5p, miR-66-5p, miR-70-3p and miR-59-3p were found to be highly expressed, but did not significantly change their expression upon starvation

The third most expressed miRNA was miR-84-5p, that together with other abundant miRNAs like miR-48-5p and let-7-5p, are members of the let-7 family, that also includes miR-7, miR-241, miR-793, miR-794 and miR-795 [65]. Let-7 was one of the first identified miRNAs, and targets the lin-41 3’UTR [66]. The name was given because of the lethal phenotype of worms with ruptured vulva that die before reaching the adult stage [67]. It was found that mutations in let-7 cause delayed temporal fates in the last larval stages and the accumulation of let-7 at the end of the third larval stage causes downregulation of LIN-41 protein expression, leading to the adoption of later larval and adult stages [66]. The let-7 miRNA is also one of the most conserved miRNAs in animals, as it has been found in all bilateria [63,68]. Although miRNAs from the same family potentially regulate the same targets, masking the phenotype of individual mutants, a single let-7 mutation was sufficient to cause developmental anomalies and lethality [66].

Let-7-5p is not expressed in embryo, L1 or L2 stages; its expression augments in L3 and continues high in L4 and adult stages [57]. Let-7 expression is regulated at both transcriptional and post-transcriptional levels. For instance, LIN-28 binds endogenous let-7 pri-miRNAs, blocking Drosha processing, leading to a reduction in mature let-7 [69]. Two transcription factors regulate let-7 expression: HBL-1, a zinc-finger transcription factor that is homologous to Drosophila Hunchback that represses let-7 expression [70] and the nuclear hormone receptor DAF-12, a regulator of dauer diapause that represses let-7 expression in the absence of its ligand, dafachronic acid [71,72]. Several transcription factor transcripts, including daf-12, emerged as let-7 targets from an RNAi screen of candidate genes that contained predicted binding sites [73]. Additionally, in C. elegans, many of the well-established let-7 target sites, including those in lin-41, daf-12, and hbl-1, were detected by CLIP (cross-linking immunoprecipitation) [74]. As the 3’UTR of the small GTPase let-60/RAS shows multiple binding sites for let-7 family miRNAs, the lethal phenotype of let-7 mutants that rupture through the vulva could be caused by abnormal regulation of let-60/RAS [75].

Another highly expressed miRNA is miR-228-5p (Fig 5). It is conserved in Caenorhabditis briggsae and Ascaris lumbricoides [76–79], but has no assigned function yet. Similarly, the functions of other abundant miRNAs like miR-66-5p, miR-70-3p and miR-59-3p are currently unknown.

miRNAs that changed their abundance under starvation conditions

Differential expression analysis of the deep sequencing results shows that 13 miRNAs and 1 miRNA hairpin were upregulated, while 2 miRNAs and two miRNA hairpins were downregulated in 12-hr starved vs well-fed early L4 larvae (Fig 6, Table 1).

Download:

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

Fig 6. Differential expression analysis of miRNAs that changed their expression in 12-hr starved compared to well-fed early L4 larvae.

MA-plot showing the absolute and relative expression of all known miRNAs with at least 1 read per million in one of the two libraries (well-fed and 12-hr starved worms). Positive (negative) log₂ fold-changes represent miRNAs with higher (lower) expression under fasting conditions. Significantly differentially expressed miRNAs (False Discovery Rate <5%) are shown as red circles. Grey text indicates three miRNA hairpins (excluding annotated mature products) also found to be differentially expressed.

https://doi.org/10.1371/journal.pone.0142262.g006

Download:

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

Table 1. Differentially expressed miRNAs and miRNA hairpins under starvation conditions.

https://doi.org/10.1371/journal.pone.0142262.t001

As can be seen in Table 1, seven members of the miR-35 family (from miR-35-3p to miR-41-3p; [80]) were upregulated from ~6- to 20-fold under starvation conditions. This result is consistent with a previous study that found up regulation of the miR-35-41 family in a genetic model of dietary restriction in C. elegans [24]. Another miRNA found to be upregulated was miR-246-3p, that has been suggested to promote longevity, since mutants in this miRNA display a shorter lifespan and its overexpression leads to increased lifespan [81]. Although the targets of miR-246-3p have not been identified, aakg-1 (one of the five gamma subunits of AMP kinase) and lin-37 (involved in larval, embryo and vulval development) have been predicted as targets.

miRNA “star” strands

We observed in our data that miR-34-3p, miR-39-5p, miR-41-5p, miR-240-5p, miR-48-13-5p were upregulated upon starvation conditions, while let-7-3p and miR-85-5p were downregulated (Fig 5, Table 1). All of them have in common that they represent the less abundant forms of the two potential miRNAs encoded in each miRNA hairpin, previously referred to as the “star” (*) strands. The miRNA star sequence is the strand complementary to the mature miRNA in the miRNA duplex. Initially, star strands were considered to be by-products of the miRNA processing steps that were rapidly degraded. But lately, star strands have been found incorporated into Argonaute complexes in Drosophila [82–84]. In one report, the star strands were found to interact with Drosophila AGO2, which is also associated with the siRNA pathway while the mature strands bound to AGO1, involved in translational repression [83]. Additionally, the star strands were shown to repress the expression of synthetic targets in an in vitro model, highlighting the capacity of star strands to act as regulatory molecules [85].

In C. elegans, star strands bind preferentially to the AGO protein RDE-1, which is required during RNA-mediated interference (RNAi) for specific removal of the passenger strand [86]. Star strands have demonstrable impact on vertebrate regulatory networks and should be considered in studies of miRNA function and their contribution to disease states. We are still lacking information about the importance of star strands in regulation, as well as the role, if any, that they play in the siRNA pathway. Star strands are predicted to have targets that are different from those of the mature strand and star strands could base pair with the mature strands and interfere with miRNA-mRNA interaction. Because an RNA molecule can base pair with other RNAs, and considering that most of the genome is transcribed, star strands may be part of a complex regulatory network in which they compete for miRNA binding and amongst each other, as has recently been postulated for competing endogenous RNAs or ceRNAs [87].

It will be interesting to discover the functions of the star strands that changed their expression under starvation conditions, whether they are bound by RDE-1 and consequently involved in the endogenous siRNA pathway, interfere with the function of their mature counterparts or have regulatory functions on their own.

miRNA hairpins

The cleavage of pre-miRNA hairpins by Dicer generates the mature miRNA, the miRNA* (star) strand and the intervening terminal loop (loop-miR). Although miRNA* and loop-miRs were largely regarded as processing by-products of miRNA biogenesis, their regulatory role has now been demonstrated. Recently, selected loop-miRs were found enriched in Argonaute complexes in multiple drosophilids and in humans. These molecules are competent for repressing the expression of mRNAs by base pairing in standard luciferase assays [88,89]. In our analysis we were able to detect three miRNA hairpins that were differentially expressed, even though their absolute counts were quite low. This is the case of mir-35, which was upregulated, and mir-79 and mir-85, both downregulated upon starvation. It remains to be seen if these loop-miRs are incorporated into Argonaute complexes and to determine their possible regulatory role.

Starvation-responsive miRNAs could target genes with metabolic and developmental functions

The differentially expressed miRNA sequences were used as queries to search for targets using microRNA.org [41], miRanda [42] and TargetScan [43] websites. We found several mRNAs that code for proteins involved in metabolism and development as putative targets of miRNAs that changed their expression under starvation conditions. For example, for those involved in metabolic processes we identified genes involved in the Insulin/IGF-1 signaling (IIS) such as ins-9, vang-1, skn-1, predicted mRNA targets of miR-39-5p, miR-41-5p, miR-240-5p, respectively (S2 Table and S1 File). As they participate in the IIS pathway, the proteins are expressed in response to nutrient availability and mutants in their coding genes show an increased lifespan. Such features could explain why L4 larvae that were starved for 12-hr showed an increased lifespan.

Other mRNAs predicted as targets of differentially expressed miRNAs are involved in lipid metabolism such as B0301.1 that encodes a protein involved in lipid accumulation and could be targeted by miR-4813-5p, and nhr-28, that codes for a nuclear hormone receptor that functions in lipid storage and could be targeted by miR-85-5p (S2 Table and S1 File). Predicted targets that are involved in lipid metabolism could help to explain the decrease of lipids shown by worms that were starved for 12 hrs (Fig 1C). These predictions seem to be consistent with the results reported by Van Gilst et al. [90], which showed that fasting produced changes in the abundance of several mRNAs from genes involved in processes such as mitochondrial β-oxidation and synthesis of mono and polyunsaturated fatty acids, whose expression could be regulated by miRNAs.

Of the genes that are involved in development we found gck-4 as a predicted target of miR-39-5p, and glh-2 and lag-1, predicted targets of miR-240-5p (S3 Table and S1 File). The proteins coded by such mRNAs play important roles in the development of the germline, cellular proliferation, fecundity, oogenesis and growth.

miR-240-5p, miR-246-3p, and the miR-35-41 cluster were upregulated under starvation conditions

We have shown that several miRNAs were differentially expressed in starved worms (Fig 5, Table 1). We validated the reliability of the RNA-seq profiling data by quantifying transcript abundances of selected miRNAs using qRT-PCR. We observed that the expression of miR-240-5p and miR-246-3p was upregulated in 12-hr starved worms, as seen in the RNA-seq experiments (Fig 7A). Also, we confirmed that the expression of three members of the miR-35-41 family (miR-35-3p, miR-36-3p, and miR-39-3p) was upregulated in 12-hr starved worms, consistent with the RNA-seq results in which the expression of all of the -3p forms of the miRNAs that constitute the mir-35-41 cluster were upregulated from 6 to 20 times under starvation conditions in early L4 larvae (Fig 7B). For these experiments, we used miR-58-3p as loading control, since its expression was not significantly changed under starvation conditions.

Download:

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

Fig 7. Expression of selected miRNAs and mRNAs involved in the fasting response as quantified by qRT-PCR.

Relative quantification of (A) miR-240-5p, miR-246-3p, and (B) miR-35-3p, miR-36-3p, miR-39-3p, and gld-1 and lin-23 mRNA abundance in well-fed and 12-hr starved early L4 larvae by qRT-PCR. For miR-240-5p, miR-246-3p, miR-35-3p, miR-36-3p, miR-39-3p quantification, miR-58-3p was used as a control. For gld-1 and lin-23 mRNAs quantification, β-actin mRNA was used as a control. Error bars represent the standard error of the mean of three independent experiments. (miR-240-5p) p = 0.03, (miR-246-3p) p = 0.03, (miR-35-3p, SEM = 37.60) p = 0.03, (miR-36-3p, SEM = 14.33) p = 0.03, (miR-39-3p) p = 0.03, (gld-1, SEM = 20.87) p = 0.03 and (lin-23, SEM = 19.74) p = 0.4, calculated by the Wilcoxon Signed Rank Test.

https://doi.org/10.1371/journal.pone.0142262.g007

The mir-35 family is composed of eight miRNAs, mir-35 to mir-42, all of them located on chromosome II. One cluster contains mir-35 through mir-41 (mir-35-41), while the other contains mir-42, mir-43 and mir-44, although the last two do not belong to the mir-35 family [80]. The mir-35 family is conserved in worms and planaria [80,91]. In C. briggsae and C. remanei it is composed by 17 members, while C. brenneri has 27 [92]. Most embryo cells express mir-35 members, first at the onset of gastrulation, with a peak at the onset of elongation [11,57,60]. A mutant with a deletion of all eight members of the mir-35 family resulted in temperature-sensitive embryonic or L1 larval lethality [60]. Since the individual expression of each of the members of the mir-35 family (but not the unrelated mir-43 and mir-44) rescued the defects caused by deletion of all family members, it appears that miRNAs of the mir-35 family act redundantly [60]. The mir-35-41 cluster genes were also found to be expressed specifically in oogenesis and not in spermatogenesis [60]. In addition to its participation in germ cell proliferation, a key role has been ascribed to mir-35 family members in the G1/S transition in intestinal cells, as loss of mir-35 shows a significant decrease of nuclei numbers in both the intestine and the distal mitotic gonad [93].

Recently, two other roles have been found for the mir-35-41 cluster. Firstly, these miRNAs are important for RNAi functions [94]. Worms lacking the mir-35-41 cluster showed a reduced expression of lin35/Rb, the C. elegans homolog of the tumor suppressor Retinoblastoma gene, which is also involved in RNAi responsiveness [94]. The lin-35/Rb gene is a member of the synthetic multivulva B (synMuv B) family, including transcriptional repressor and chromatin remodeling genes [95]. The mir-35-41 cluster inhibits the exogenous RNAi pathway by positively regulating the expression of LIN-35/Rb protein, although the mechanism is not known [94]. This result exemplifies the fact that miRNAs, besides regulating gene expression by direct binding to mRNAs, also have the ability to affect the activity of other small RNA pathways [94]. Secondly, the mir-35-41 cluster miRNAs have important roles for full reproductive capacity, as a temperature sensitive mutant (a deletion of the mir-35-41 cluster) presents reduced hermaphrodite fecundity as a result of sperm defects and somatic gonad dysfunction, and showed morphogenesis defects of male-specific mating structures [96].

In our experiments, the mir-35-41 cluster was up-regulated in early L4 larvae subjected to a 12-hr fasting. This contrasts with the normal expression of this family in the embryo stage only. Given the functions ascribed to the members of the mir-35 family, their up-regulation can have effects on RNAi sensitivity and reproduction. In the 12-hr starvation condition, up-regulation of miR-35-3p will negatively regulate G1/S transition in intestinal cells and will negatively regulate oogenesis [93]. This makes sense under starvation, since cell division and oogenesis are not essential, cells being focused on survival, not on making more cells or taking part in reproductive functions. Since mir-35-41 mutants also show an increased sensitivity to endogenous RNAi [94], we can predict that starvation should cause a reduction in sensitivity to endogenous RNAi, since the expression of mir-35-41 members are enhanced under this condition. If this is the case, it will be interesting to investigate what are the physiological consequences of having reduced sensitivity to endogenous RNAi during starvation.

Changes in the abundance of gld-1 and lin-23 mRNAs during starvation, known targets of miR-35-3p

The gld-1 and lin-23 mRNAs have been reported to be direct targets of miR-35-3p [93]. LIN-23 (abnormal cell LINeage) is the F-box component of the SCF (Skp1-Cul1-F box) complex [97,98]. This complex shows activity of E3 ubiquitin ligase involved in the degradation of CYE-1 (cyclin E-1) [99,100] and CDC-25.1 (cell division cycle related) [101], and participates in the negative regulation of the G1/S transition. Consistent with this, a mutant in lin-23 develops intestinal hyperplasia [101]. GLD-1 (defective in Germ Line Development) is an RNA-binding protein, member of the STAR (for signal transduction and activation of RNA metabolism) family that includes human/mouse QUAKING, SAM68, and Drosophila HOW [102]. The STAR proteins contain a conserved region with a maxi-KH binding domain and two conserved flanking domains (Qua1 and Qua2) [102]. The GLD family plays an important role in the determination of meiosis start, by indirectly regulating mitotic proliferation of the germline [103,104], and forms part of the regulatory network of GLP-1 (abnormal Germ Line Proliferation)/Notch signaling that promotes germ line cell divisions in the distal mitotic region of the gonad [105,106]. In the adult hermaphrodite worm, germ cells that will develop into oocytes are produced in the proliferative zone, the distal part of the gonad arm, and differentiate into oocytes as they move away [107]. Since GLD-1 is an important regulator of the start of meiosis, its expression is highly regulated at the level of mRNA stability and mRNA translation [108–110]. A mutant of GLD-1 showed a defect in oogenesis and formation of tumors in the proximal part of the gonad [102].

Taking into account the important functions of GLD-1 and LIN-23, we were interested in assessing the expression of lin-23 and gld-1 mRNAs in 12-hr starved L4 larvae. In the case of lin-23 mRNA, we did not detect a significant change in its level upon starvation (Fig 7B). This could be explained if the binding of miR-35-3p to lin-23 mRNA leads to translational repression [111,112]. In this context, it would be informative to experimentally assess if the LIN-23 protein level is affected by a 12-hr fasting.

In contrast, we found that whereas miR-35-3p expression was upregulated in 12-hr starved larvae, gld-1 mRNA was downregulated (Fig 7B), a result that is consistent with the idea that an increase of miR-35-3p would lead to the degradation of its target gld-1 mRNA. Such type of regulation in which the binding of a miRNA to its target mRNA leads to a destabilization of the message has been shown to be the predominant way by which miRNAs regulate gene expression in mammalian cells [113,114]. It is important to experimentally elucidate if mRNA destabilization is the mechanism used by miRNAs to regulate gene expression in C. elegans in general, and, in particular, if the down regulation of gld-1 mRNA observed under a 12-hr fasting is mediated by a direct binding of miR-35-3p.