Age-related micro-RNA abundance in individual C. elegans

Nano-fluidic quantitative polymerase chain reaction of miRNA samples from individual worms

To examine age-related miRNA expression changes we used the quantitative polymerase chain reaction (qPCR) to assay a panel of 69 miRNAs in RNA samples collected from 48 C. elegans animals. We used individual worms to assay the miRNA levels, an approach which has previously been used in the context of gene expression profiling via microarrays [5, 30]. Such an approach captures variation between animals with age, as it screens individuals as opposed to most conventional methods which analyze populations containing large numbers of pooled animals. Individual wild type (N2; Bristol strain) animals were examined for expression of each miRNA simultaneously using dynamic integrated nano-fluidic circuits (Fluidigm inc.) with Taqman qPCR assays (ABI) specific for each miRNA [31]. We first tested for expression of the miRNAs in 24 young (hermaphrodites on their first day of adulthood) and 24 old (hermaphrodites on their twelfth day of adulthood) animals. Young adult animals were non-gravid, while old animals were post-reproductive and were collected while the population was undergoing a high mortality rate (Figure 1A). We found that 45 miRNAs were consistently detected (in more than half of the animals) while 43 miRNAs were consistently detected in old animals (Supplemental Table 1). Of the 43 miRNAs that were consistently detected in both young and old animals, slightly more than half (56%) decreased in abundance with age, demonstrating that amongst the miRNAs we assayed, there is a trend towards most miRNAs decreasing in expression with age. As we had determined the levels of miRNAs in individual animals, we next examined the variation in abundance of each miRNA between individual worms for both young and old animals (standard deviation of the population from the mean Ct (cycle threshold)). There is a generalized increase in the variation of miRNA abundance with age, consistent with a generalized increase in stochastic dysregulation. Specifically we found that 68% of the miRNAs showed an increase in variation between individual worms with age, while if we only counted the miRNAs which were detected in all 48 animals then 74% of these showed an increase in their variation with age (Supplemental Table 1). This suggests that variability of miRNA expression generally increases with age. This increase in variability may be in part due to stochastic cell and tissue loss in aged animals [2, 32-34].

Analysis of the expression profiles in single worms lead to the identification of miRNAs which had significantly altered levels between young and old animals (Figure 1 and Table 1). Some miRNAs that showed altered levels with age had higher levels later in life. Of these, the most statistically significant increases were observed for miRNAs; mir-51, mir-44, mir-240 and the members of the mir-35-41 cluster (Figure 1B-D and Figure 2A-E). We also identified miRNAs whose levels decreased significantly with age; among these mir-235, mir-60 andmir-58 were identified as those being consistently detected and showing the most significant decrease with age (Figure 1E-G). Other than the mir-35 family members all of these miRNAs showed an increase in variation with age (Supplemental Table 1). To test these results with an independent approach we quantified green fluorescent protein (GFP) expression in aged worm strains carrying promoter-GFP fusions for the age dependent hits mir-1 (Supplemental Figure 1), mir-60 (Supplemental Figure 2) and mir-228 (Supplemental Figure 3). In agreement with our biochemical results, these strains showed decreased expression of GFP with age.

High expression of mir-35 family members in aged animals originated in the post-reproductive germline

Some of the largest and most significant changes in levels of miRNAs between first day adults and post-reproductive adults came from members of the mir-35 miRNA family and specifically came from the mir-35-41 gene cluster (Figure 2A-E and Table 1). We found that the mir-35 cluster members in our set of assays (mir-35-39) showed a dramatic increase in abundance with age. The only exception being mir-38 which included three outlier signals that shifted the overall mean towards a net decrease in levels, while still having a median increase in abundance with age (Figure 2D and Supplemental Table 1). mir-35 family members are redundantly required for viability and are highly expressed in developing embryos [35, 36]. The mir-35-41 gene cluster is also co-transcriptionally regulated [36] and it has been reported that expression from the mir-35-41 cluster originated in the germline and was likely to be mainly found in oocytes and embryos [35]. These results, along with our previous observation of the DNA increase in the post-reproductive germline [32, 33] led us to speculate that we may be observing an increase in expression from dysregulated genomes of immediate germline descent. Indeed many of the profiling candidates we identified as having altered abundance with age may be a result of changes in germline expression due to the massive increase in germline DNA copy number with age.

To test the contribution of the germline on miRNA expression we performed qPCR analysis of miRNA levels, on germline-less animals. We chose a mutant strain that allowed for genetic ablation of the germline during development. glp-1(e2141) mutants are temperature sensitive sterile and at the restrictive temperature lack a germline due to the inability of the germline stem cells to remain mitotically active. These animals are also long-lived in a DAF-16 and DAF-12 dependent manner [37], however they still respond to deficits in insulin/IGF like signaling with lifespan extension. To examine changes with age in animals lacking a germline, we collected individual glp-1 mutants spanning 4 adult ages that included both young and old animals (Figure 3A). Interestingly we found that expression from the mir-35-41 gene cluster was nearly absent in animals that lacked a gonad. Of the mir-35-41 cluster we only detected mir-37 in the germline ablated animals, but only in 30% of the individual worm samples. The levels of mir-37 we measured were much lower than those observed in wild type young adult animals (Figure 2F and compare Supplemental Tables 1 and 2). These results along with previous reports, suggests that the age-dependent increase in the mir-35 group arises from high expression in the germline of post-reproductive animals.

The presence of the germline altered the expression of several miRNAs

In addition to the mir-35-41 cluster there were other individual miRNAs whose expression profiles are distinctly different between wild type animals and germline-less glp-1 mutants. Of these, mir-51 showed a significant elevation in expression with aging in wild type animals (Figure 1C). However, in the germline ablated glp-1 mutant, mir-51 expression tended to decrease with age (Figure 2G). One interpretation of this result is that mir-51 is expressed in both somatic and germline tissues but that the trend in its expression with age is inverted for the two tissues. Another miRNA expression profile that showed a dramatic difference between wild type and germline less mutants was mir-60, which showed a significant and profound decreased expression profile with age in wild type animals (Figure 1F). However, in mutants lacking a germline, mir-60 showed an average elevated expression in all of the ages tested relative to the first day adults (Figure 2G). This suggests that the germline normally functions to suppress somatic expression of mir-60, possibly implicating it in germline regulated somatic signaling cascades [37]. However, it is not absolutely required for the resulting lifespan extension observed from germline stem cell ablation (Supplemental Table 3).

As with the wild type miRNA expression profiles, analysis of the glp-1 profiles indicated that many miRNAs show altered abundance with age (Figure 3B-F and Table 2). Nine miRNAs showed robustly significant age-dependent levels (Table 2). The most statistically significant miRNAs showing increased expression with age were mir-354 and mir-240, while mir-54 and mir-67 trended towards significance (Figure 3B-D, H and Table 2). In the germline ablated animals most of the statistically significant changes we observed showed a decrease in levels with age (78%). Of these mir-64, mir-81, mir-65, mir-53, mir-58 were all significantly down-regulated in aged animals (Figure 3E-H and Table 2). Interestingly, and analogous to our results with wild type animals, we found that the variation in individual worm miRNA levels tended to increase with age, as 63% of miRNAs assayed showed an increase in variation. However, if we only enumerated those miRNAs that were detected in all individuals 79% of miRNAs showed an increase in variability across individuals with age (Supplemental Table 2). Collectively these results indicate that the variation of miRNA between individuals is not merely an artifact of post-reproductive germline abnormalities.

Most miRNAs exhibited no or relatively minor influences on lifespan

In order to test whether miRNAs displaying age-specific abundance modulate lifespan, we tested mutant strains for several of the candidate miRNAs. These mutants were directly assayed for their lifespan and were compared to wild type animals (Supplemental Table 3). Of these, we found that one mutant strain carrying a deletion resulting in the removal of several miRNAs led to a weak lifespan extension. Specifically the deletion nDf48 that removes mir-42-44 led to a small but significant lifespan extension. (Supplemental Table 3), suggesting that individually or collectively these miRNAs may inhibit longevity in wild-type animals. We have also identified single miRNA mutations that resulted in mild or inconsistent lifespan phenotypes. The mir-80 allele nDf53 had a weak but significantly extended lifespan (Supplemental Table 3), while the mir-228(n4382) mutant trended towards an extended lifespan (Supplemental Table 3). In addition to miRNA mutants that showed an extended lifespan, we identified miRNA mutants that had shortened lifespans. We found that the mir-71 deletion n4115 had a profoundly shortened lifespan compared to wild type (Figure 4 and Supplemental Table 3), while the mir-60 deletion, n4947 trended towards shortened lifespan but was not consistently significantly different from wild type (Supplemental Table 3).

To further investigate the role of miRNAs in modulating lifespan, we analyzed the effect of expressing the missing miRNAs directly in the miRNA mutants. If the miRNAs absence in the mutants was causative of the lifespan phenotype, then these transgenic animals should be rescued for that phenotype or otherwise show a modulation of lifespan. We tested several such miRNA mutants containing transgenes and found that several of these mutants which exhibited lifespan phenotypes, showed modulation of the lifespan phenotype when the corresponding transgene was present (Supplemental Figure 4 and Supplemental Table 3). These included mir-1, mir-80 and mir-228. This further suggests that these miRNAs may influence the lifespan of the organism. We found that several of the miRNAs mutants that did not exhibit lifespan phenotypes (the mir-64 cluster, mir-81-82 and mir-235) also failed to show lifespan phenotypes when the miRNAs were added back (Supplemental Table 3). Collectively, these results indicate that mir-1, mir-80 and mir-228 may have a role in modulating lifespan.

mir-71 acts at least partly in parallel to insulin/IGF signals

We next investigated the mechanism underlying a specific miRNA's effect on lifespan. We focused on mir-71 which when mutant led to robust and significant lifespan phenotype, demonstrating its requirement for a normal long life. We examined the spatial expression pattern of mir-71 using a promoter-GFP fusion strain [36] (Supplemental Figure 4). As previously reported mir-71 is expressed from early embryogenesis through to adulthood [38]. In early development expression of GFP from the mir-71 promoter is restricted to the primordial intestine of the embryo, but after hatching appears to be expressed in most non-neuronal tissues (Supplemental Figure 4). As GFP expression was very pronounced in the pharynx (Supplemental Figure 4A-F) we addressed whether this tissue is important for mir-71 function in lifespan modulation. We found that pharyngeal expression of mir-71 could partly rescue the lifespan of mir-71 null mutants (Supplemental Figure 4G). This suggests that the pharynx is an important mediator of mir-71 function and likely contains targets of the putative regulatory RNA. We also analyzed expression of mir-71 under strong heterologous promoters and found they led to arrest and unstable extra-chromosomal arrays (data not shown). A heat shock specific promoter driven construct was more stable and exhibited little developmental delay. Transgenic worms over-expressing mir-71 were found to be long-lived (Supplemental Figure 4H). This result demonstrates that mir-71 is sufficient to extend lifespan.

Since over-expression from heterologous promoters led to developmental delay and loss of the transgene, we utilized a stably integrated over-expressor of mir-71 for further genetic analysis. We chose the promoter-GFP fusion construct for mir-71, known as maIs352 [36], which was of particular interest since it contains the entire mir-71 sequence. Consistent with this strain containing, the entirety ofmir-71 under its own promoter as well as GFP, we observed over-expression of mir-71 in the strain (Supplemental Figure 4I). Importantly we found that the transgene partly rescued a mir-71 null mutant (Figure 4A and Supplemental Figure 4I). We conclude that functional mir-71 is required for animals to survive late into normal adulthood, and that mir-71 acts in a pro-longevity manner. Along with our expression profiling results demonstrating that mir-71 levels increase with age (Supplemental Table 1-2 and Supplemental Figure 6A-B), we suggest that the normal function of mir-71 is sufficient to extend the lifespan of the adult animal.

It has previously been reported that mir-71 mutants are short-lived and that over-expression of mir-71 extended lifespan [17]. It was further suggested thatmir-71 acts with insulin/IGF signaling to modulate the lifespan of C. elegans. We found this interesting as we had identified a putative mir-71 binding site in the 3' UTR of the insulin/IGF type receptor, DAF-2 (Supplemental Figure 6C). DAF-2 is a modulator of lifespan, whose effects are attributed to its role in regulating the FOXO transcription factor DAF-16. In agreement with de Lencastre and colleagues we found that mir-71 mutants shorten the lifespan of daf-2 hypomorphic mutants (Figure 4B). Collectively these data suggested to us that mir-71 may regulate expression of the daf-2 mRNA directly. To determine whether mir-71 regulates the expression of DAF-2 we generated GFP fusion constructs with either daf-2 or control 3'UTRs. However, we were unable to detect significant GFP expression differences between these constructs in mir-71 mutant or mir-71 over-expressor backgrounds (data not shown). These data suggested that mir-71 does not directly regulate the expression of DAF-2.

We next sought to test whether the lifespan effects of mir-71 depend on a functional insulin/IGF pathway. DAF-16 is the major transcription factor downstream of DAF-2 and is required for the lifespan effects resulting from IIS in C. elegans. If mir-71 acts solely with IIS, then the effects of mir-71 should depend on functional DAF-16. We therefore tested the lifespan phenotype of mir-71; daf-16 double null mutants. We find that the mir-71(n4115); daf-16(mu86) double mutants are significantly shorter lived than either daf-16 or mir-71 mutants alone (Figure 4C). This result indicates that mir-71 acts in parallel to DAF-16 to modulate adult lifespan. This result is inconsistent with the finding of de Lencastre and colleagues who performed a similar analysis but with non-null daf-16(RNAi). To further test whethermir-71 acts independently of IIS we next tested whether DAF-16 is required for the lifespan extending effects of mir-71 over-expression. We found that the mir-71 over-expressor, maIs352 significantly extends the lifespan of daf-16 null mutants (Figure 4D). Collectively our data demonstrates that mir-71 acts at least partly in parallel to insulin/IGF type signaling to modulate lifespan in C. elegans.

Nematode Strains

Nematodes were maintained as previously described [42]. The following previously described alleles, strains and transgenics were used in the experiments presented here. N2; glp-1(e2141); daf-16(mu86); daf-2(e1368); mir-71(n4115); maIs352[Pmir-71::GFP]; mir-85(n4117); mir-42-44(nDf49); mir-84(n4037); mir-273(n4438); mir-51(n4473); mir-54-56(nDf58); mir-240(n4541); mir-60(n4947); mir-70(n4109); mir-81-82(nDf54); mir-53(n4113);mir-235(n4504);nDf63; mir-63(n4568);mir-80(nDf53);mir-64c(nDf63);mir-1(n4101); mir-58(n4640);mir-228(n4382); maIs251[Pmir-1::GFP]; maIs276[Pmir-60::GFP]; maIs188[Pmir-228::GFP].

Lifespan Assays

Lifespan assays were performed as previously described [43]. In brief, synchronized worm populations were transferred to NGM plates supplemented with 10μg/mL floxuridine (FUDR) at day one of adulthood. Animals were scored and transferred to fresh plates approximately every other day until death. Lifespans presented in Figure 1 and Figure 3 did not contain FUDR nor did any lifespan that involved the mutant glp-1. Lifespan assays containing glp-1 were shifted to 25°C immediately after the egg lay and maintained at that temperature, all other lifespan assays were performed at 20°C. Comparisons of survival were performed with GraphPad Prism 4™ software and P values were generated by the log-rank Whitney test of significance.

Preparation of samples for BioMark

To quantify changes in miRNA abundance during aging, we cultured worms under standard laboratory conditions and aged them by daily transferring the adults away from their progeny. At selected ages, we harvested individual animals and isolated total RNA from each sample using the miRNeasy (Qiagen) collection kit with automated Qiacube processing (Qiagen). cDNA synthesis was accomplished with the high capacity RT kit (Applied Biosciences). miRNA reverse transcription was performed with Applied Biosciences custom made miRNA reverse transcription primers. Reverse transcription products were subjected to a multiplex pre-amplification prior to the qPCR.

miRNA Detection by Quantitative Polymerase Chain Reaction

qPCR assays for 69 miRNAs were obtained from Applied Biosystems (ABI), at the time this represented their entire coverage of C. elegans miRNAs. Currently over 170 miRNAs are listed on miRBase (www.mirbase.org) as existing in the C. elegans genome. However, the validity of this estimate as being representative of the actual number of miRNAs is unclear. A recent report using high-throughput sequencing techniques revealed that of the approximately 170 annotated miRNAs (at the time) only 106 carried nucleotide signatures consistent with processing by the miRNA processing enzymes [44]. qPCR on the BioMark (Fluidigm) was performed according to the protocol of [31] with slight modifications for use with single C. elegans samples.

Additional methods are included with the Supplemental information.