Deletion of microRNA-80 Activates Dietary Restriction to Extend Healthspan and Lifespan

Summary

Caloric/dietary restriction (CR/DR) can promote longevity and protect against age-associated disease across species. The molecular mechanisms coordinating food intake with health-promoting metabolism are thus of significant medical interest. We report that conserved Caenorhabditis elegans microRNA-80 (mir-80) is a major regulator of the DR state. mir-80 deletion confers system-wide healthy aging, including maintained cardiac-like and skeletal muscle-like function at advanced age, reduced accumulation of lipofuscin, and extended lifespan, coincident with induction of physiological features of DR. mir-80 expression is generally high under ad lib feeding and low under food limitation, with most striking food-sensitive expression changes in posterior intestine. The acetyltransferase transcription co-factor cbp-1 and interacting transcription factors daf-16/FOXO and heat shock factor-1 hsf-1 are essential for mir-80(Δ) benefits. Candidate miR-80 target sequences within the cbp-1 transcript may confer food-dependent regulation. Under food limitation, lowered miR-80 levels directly or indirectly increase CBP-1 protein levels to engage metabolic loops that promote DR.

Introduction

The promotion of healthy aging is a goal of modern medicine, and simple interventions that protect against age-associated decline and disease are the dream of many in the general population. Genetics, environment, and stochastic factors all make substantial and complex contributions to healthspan. Single gene mutations that affect conserved pathways in model organisms can extend life and slow age-associated decline [1], [2]. Environmental factors such as diet can also have a profound effect on the quality of aging. For example, dietary restriction (DR), limitation of calorie intake with maintained vitamin and mineral support, can extend lifespan and protect against diseases of age across many species [3]. Elaboration of molecular mechanisms that control DR in simple animal models may thus inform on strategies to activate health-promoting metabolism to help address clinical challenges associated with aging.

In the nematode Caenorhabditis elegans, food limitation that results in lifespan extension can be introduced via several protocols [4], [5], [6], [7], [8], although the specific genetic requirements for longevity benefits of different DR regimens are not fully overlapping. For example, the transcription factor DAF-16/FOXO is dispensable for longevity induced in the feeding-impaired eat-2 mutant, whereas with a DR protocol in which bacterial food is diluted on plates, DAF-16/FOXO is essential for lifespan extension [4], [9]. Such observations most likely reflect highly complex regulatory loops that control the precise metabolic state.

microRNAs (miRNAs) can be metabolic regulators [10]. miRNAs are small, ∼22 nt non-coding RNAs that can bind to transcripts via partial sequence complementarity to down-regulate translation of those target mRNAs. Many miRNAs are conserved over their lengths or in the critical 5′ seed region, defining families across species [11], [12], [13]. Although the co-evolution of miRNAs and their targets is a complex process [14], some miRNA/target pairings have been molecularly and functionally conserved. For example, discovery of LET-7 miRNA regulation of target RAS in C. elegans [15] inspired anti-oncogenic therapies for mammalian lung tumors [16].

We took advantage of the powerful reagents for miRNA study in C. elegans [17], [18] and our previous characterization of a DR fluorimetric signature of endogenous gut fluorescence in these transparent nematodes (derived from lipofuscin+advanced glycation end products [19]) to identify miRNAs that might regulate DR. Here we report bantam-homolog miR-80 as a food-regulated miRNA that normally represses DR when food is abundant. Transcription factors DAF-16, HSF-1, and CBP-1 are required for mir-80(Δ) benefits. Of these, the cbp-1 transcript includes sequences that might be directly targeted by miR-80 to coordinate this circuit. Our data suggest an approach to metabolic activation of DR even under ad lib feeding that could inspire strategies for treating obesity, limiting age-associated disease, and promoting healthy aging.

Results

Deletion of microRNA-80 promotes system-wide healthy aging in C. elegans

Our previous studies revealed that age pigment levels (lipofuscin+advanced glycation end products) inversely correlate with locomotory healthspan—low age pigment levels late in life are typical of animals that age gracefully and maintain strong locomotory vigor, whereas high age pigment levels are typical of same-chronological age animals that age poorly and appear decrepit [19]. Thus, to identify C. elegans miRNAs that might impact healthy aging, we screened available C. elegans mir deletion strains [20] for differences in autofluorescent age pigment levels in old animals. We found that mir-80(nDf53) [hereafter referred to as mir-80(Δ)], exhibits low age pigment fluorescence levels late in life compared to wild type (WT) animals (Fig. 1A, ∼58% lower, p<0.0005). The low age pigment phenotype of mir-80(Δ) is rescued by a transgene array harboring a mir-80(+) gene, confirming that the low age pigment phenotype is conferred by mir-80 deletion itself. Thus, late in adult life (∼2/3 through the WT lifespan), mir-80(Δ) mutants exhibit low age pigment accumulation typical of healthy aging animals.

To test if mir-80(Δ) mutants exhibited additional healthspan phenotypes, we next measured two indicators of maintained muscle integrity and function late into adult life–pharyngeal pumping rates and swimming vigor. Pharyngeal pumping is the mechanism by which food is pulled into the gut using specialized cardiac-like muscles. Pharyngeal pumping rates decline markedly with age, such that after the first week of life, feeding capacity is greatly diminished [21], a functional decline that tracks with physical changes in muscle integrity [22], [23], [24]. We find that pumping rates are significantly higher in mir-80(Δ) late in life (day 11) as compared to WT (44% increase), a phenotype reversed by a mir-80(+) transgene (Fig. 1B, right graph; p<0.005). Importantly, 5 day old WT and mir-80(Δ) (i.e., young adult; Fig. 1B, left graph) have similar pumping rates. Thus, mir-80(Δ) mutants are not simply hyper-activated for pumping, but rather maintain pumping function better late into life. We conclude that mir-80(Δ) exerts a positive effect on the quality of cardiac-like muscle aging.

As occurs with human skeletal muscle sarcopenia (the debilitating progressive loss of muscle mass and strength that accompanies aging across species), C. elegans body wall muscle deteriorates with age, featuring sarcomere loss [22], [24]. Physical decline is correlated with loss of locomotion vigor. We compared late-age swimming (body bend frequency) in WT and mir-80(Δ) to show that mir-80(Δ) mutants are significantly more vigorous swimmers in late adulthood (Fig. 1C right panel; 69% increase at day 11, p<0.0001). Early in adult life WT and mir-80(Δ) swim similarly (Fig. 1C, left panel). We conclude that mir-80(Δ) delays locomotory aging without altering young adult swimming behavior itself.

Given that mir-80(Δ) mutants exhibit several features of extended healthspan, we examined the longevity phenotype. We find that mir-80(Δ) mutants exhibit both mean and maximum healthspan extension, subject to mir-80(+) transgene rescue (Fig. 1D, p<0.0001, individual lifespan data in Fig. S1; average age increase at 75% mortality over all lifespan studies in this paper (13) was 24.1%+/−4.7%). Thus, deletion of mir-80 confers longevity.

In summary, mir-80(Δ) confers multiple features of extended adult healthspan late in life: lowered intestinal age pigment accumulation, maintained pharyngeal pumping capacity, increased swimming vigor, and lifespan extension. Because mir-80(Δ) does not exhibit notable defects in development ([20], and our observations), it appears that mir-80 has a predominant and focused impact on aging of the adult.

mir-80(Δ) mutants appear dietary-restriction constitutive

Spectral properties of age pigments

To ask whether mir-80(Δ) might act via a DR mechanism to extend healthspan and lifespan, we tested mir-80(Δ) mutants for phenotypic features of the DR state (Fig. 2). Our previous in vivo studies of fluorescent age pigments revealed that transparent C. elegans under DR have a unique fluorimetric “signature” that is distinct from spectral properties of both WT and long-lived mutants induced by other longevity pathways [19], [25], [26]. A spectrofluorimeter excitation/emission direct scan of WT reports a characteristic excitation maximum (Ex_max) for age pigments at ∼345 nm. However, under all DR-inducing conditions we previously tested (multiple feeding-impaired mutants [27], liquid feeding of WT [7], [19], limiting bacterial concentrations for WT [4], complete removal of food [28], treatment with DR-mimetic drug metformin [25]), we noted a downward shift in Ex_max. Thus, the age pigment Ex_max shift indicates a DR-like state. We found that mir-80(Δ) consistently exhibits the DR Ex_max shift despite growth in the presence of abundant food (Fig. 2A) and has low age pigment levels, even at young age (day 4, p<0.0005, Fig. 2B, about 66% lower in these studies), the latter of which also characteristic of DR mutants. Thus, mir-80(Δ) exhibits the spectral signature of DR despite the presence of food, consistent with mir-80(Δ) being a DR constitutive mutant.

Scrawny bodies

WT animals under DR appear thin and pale [29], [30]. We found that mir-80(Δ) had a somewhat scrawny and pale appearance and on average is ∼10% shorter in length than WT (p<0.0005, Fig. 2C). Thus, the physical appearance of mir-80(Δ) mutants resembles that of DR animals.

Reduced fecundity

Reduced fecundity is associated with DR across species [31]. We find that mir-80(Δ) exhibits a significant extension of reproductive period, producing progeny through Day 8 as compared to WT Day 6, p<0.001, Fig. 2D. In addition, there is a decrease in the number of live progeny laid per day by mir-80(Δ) (p<0.05 for Day 3 and p<0.001 for days 4–6), without a significant difference in the total number of surviving progeny (not shown), a pattern of progeny production similar to that of feeding-defective DR mutant eat-2 (Fig. 2D). We conclude that mir-80(Δ) exhibits reduced fecundity, similar to animals experiencing DR.

Hypersensitivity to a DR-mimetic drug

The anti-diabetes DR-mimetic drug metformin can induce a life-prolonging DR-like state in WT animals, but administering metformin to animals already in DR (e.g., the eat-2 mutant), leads to a reduced lifespan [25]. This metformin hypersensitivity in DR animals has been suggested to result from pushing DR metabolism into a deleterious starvation-like state [25], [32]. We found that although mir-80(Δ) is long-lived relative to WT under normal growth conditions (Fig. 1D), the mir-80(Δ) lifespan is decreased relative to WT in the presence of metformin (three individual trials and combined data in Fig. S2), similar to what occurs for DR mutant eat-2. Thus, like other DR strains, mir-80(Δ) is hypersensitive to metformin, consistent with mir-80(Δ) being in a DR constitutive state.

Molecular reporter of DR: SKN-1 expression in ASI neurons

SKN-1 is a transcription factor crucial for endodermal development and response to oxidative stress [33], that must also be expressed in the pair of chemosensory ASI neurons for the longevity outcomes of some DR regimens [7]. For example, SKN-1 appears continuously expressed in the ASI neurons in the constitutive DR eat-2(ad1116);Is007[skn-1-gfp] reporter strain, whereas this reporter is not highly expressed in wild type Is007[skn-1-gfp] animals that are grown under ad lib feeding conditions (Fig. 2E). We found that SKN-1-GFP is highly expressed in the ASI neurons in the well fed mir-80(Δ) mutant (Fig. 2E). At day 7, 95% mir-80(Δ) and 92% eat-2 mutants exhibited strong signals in 1–2 ASI neurons; but 36% of WT only express weak signal in at best one ASI.

Our data support that mir-80(Δ) induces molecular features of DR.

In summary, the mir-80(Δ) mutant has a lean body, reduced fecundity, hypersensitivity to metformin, and expresses both molecular and fluorimetric markers of DR despite growth in abundant food. Importantly, unlike the eat-2 mutant, pumping rates in mir-80(Δ) are normal in young animals and are actually enhanced relative to WT later in life (Fig. 1B). Thus, mir-80(Δ) does not physically reduce the ability to eat, but rather is likely to act further downstream to influence DR metabolism. Remarkably, then, deletion of a single miRNA gene can shift C. elegans metabolism into DR to promote healthy aging.

mir-80 expression is positively regulated by the presence of food

mir-80 is broadly expressed in well-fed animals

Previous deep sequencing studies indicate that miR-80 is a relatively high abundance miRNA expressed from late embryogenesis into adulthood [34], [35]. Somewhat paradoxically, two published transgenic lines of the same P_mir-80::GFP fusion transcriptional reporter (utilizing 1741 bp 5′ to mir-80 as promoter [17]), exhibit different cellular expression patterns, an observation we confirmed (Fig. S3A,B). To address this discrepancy, we constructed a new mCherry reporter that extended mir-80 5′ sequences up to the next annotated gene (1814 5′ bp, designated P_mir-80L). 5/5 extra-chromosomal transgenic lines of this reporter exhibit a broad cellular expression pattern, somewhat similar to the published extrachromosomal array line (vulva, hypodermis, body wall muscle, head neurons, tail neurons, excretory cell, dorsal/ventral nerve cord, and weaker expression in intestine [anterior or posterior] and pharynx). The P_mir-80L pattern is distinctive in exhibiting highest expression in the two most anterior gut cells and in posterior gut under high food conditions (Fig. 3A). Interestingly, mir-80 does not appear expressed in the ASI neurons in any lines (Fig. S4A,B) and thus miR-80 most likely acts non-cell autonomously to influence skn-1::GFP expression in the ASI neurons (Fig. 2E).

P_mir-80 reporter expression is down-regulated in the absence of food

Since the genetic elimination of mir-80 results in constitutive DR phenotypes (Fig. 2, Fig. S2), we hypothesized that mir-80 expression might be reduced when food is limited. We examined multiple mir-80 transcriptional reporters for expression level 48 hours after shift from abundant food to no food (Fig. 3B–D, Fig. S3). We find that for all reporters examined, expression for mir-80 is significantly lower in the absence of food. Furthermore, food limitation by alternative diet regimens is also associated with general down-regulation of mir-80 reporter expression (Fig. S5). For the broadly expressing transgenes, it appeared that overall expression in many cells was down-regulated, although the changes in the mid and posterior intestine have the largest differential, ∼4–10× P_mir-80L::mCherry level changes food/no food (Fig. 3D). We confirmed general down-regulation in the absence of food by deep sequence analysis of miR-80: overall expression levels food/no food are 1.5 increased (data not shown, p-value<0.05). However, we emphasize that not all cells exhibit miR-80 down-regulation: expression in two anterior-most and two posterior-most gut cells, and vulval muscle expression appear maintained, and possibly enhanced, in no food. We conclude that mir-80 expression can be modulated by the presence of food: in most cells, mir-80 expression is relatively high in the presence of food and is reduced when food is limiting. The broad mir-80 expression pattern suggests a potential role for miR-80 in global regulation of metabolism; although dramatic posterior intestinal regulation raises the possibility that major changes in this tissue could provide the most critical influence on organism-wide regulation (see Discussion).

Transcription factors implicated in DR metabolism are required for mir-80(Δ)-associated fluorimetric features

To identify genes required for mir-80(Δ)-regulated DR, we used RNAi to knockdown genes previously implicated in DR lifespan benefits, hypothesizing that genes required for mir-80(Δ) DR should be needed for the Ex_max shift and low age pigment levels typical of multiple DR states. Of the 18 genes we screened, we found that RNAi knockdown of transcription factors daf-16/FOXO, heat shock transcription factor hsf-1, and CREB binding protein homolog cbp-1 modulated both the Ex_max shift and low age pigment levels of mir-80(Δ) (Tables S1, S2, Figs. 4A,B; 5A,B; 6A,B).

DAF-16/FOXO is required for fluorimetric indicators of DR age pigment and lifespan extension in mir-80(Δ)

Transcription factor daf-16/FOXO, an important modulator of longevity through insulin signaling, is also critical for lifespan extension benefits of serial dilution of bacteria on plates (sDR) and peptone dilution on plates (pDR) [4]. We found that the mir-80(Δ);daf-16(Δ) double mutant was reversed for both the DR Ex_max shift (Day 4, Fig. 4A, p<0.05) and the low age pigment levels (Fig. 4B, p<0.005) that are characteristic of mir-80(Δ). Moreover, the mir-80(Δ);daf-16(Δ) double mutant had a short lifespan, similar to that of daf-16(Δ) (Fig. 4C). These data identify DAF-16 as an required regulator of the fluorimetric DR signature and longevity benefits in the mir-80(Δ) background. Our in silico analyses did not identify candidate miR-80 target sites in the daf-16 transcript, suggesting an indirect role in the mir-80(Δ)-regulated DR pathway.

To address the relationship of mir-80(Δ) and the insulin signaling pathway further, we compared longevity phenotypes of mir-80(Δ) and mir-80(Δ) treated with daf-2 RNAi, which targets the C. elegans insulin receptor (Fig. 4D). We find that mir-80(Δ) lifespan can be further extended by daf-2(RNAi) (p<0.005). The additive effects of mir-80(Δ)+daf-2(RNAi) knockdown suggest that healthspan and longevity benefits of mir-80(Δ) may be conferred in part by a daf-2-independent pathway. However, the fact that mir-80(Δ) does not further extend daf-2(RNAi) lifespan (p<.98) is also consistent with a model in which miR-80 partially down-regulates the insulin pathway, and that daf-2(RNAi) reflects a stronger activation of the DAF-16-dependent transcriptional response, more toward an optimal healthspan signaling strength. Regardless of the details of pathway overlap, our data are definitive in establishing that daf-16/FOXO is needed for fluorimetric properties and longevity outcomes of mir-80(Δ).

hsf-1 deficiency eliminates multiple mir-80(Δ) healthspan phenotypes

hsf-1 regulates the expression of many heat-inducible target genes, modulates longevity, and is required for lifespan extension conferred by bacterial food deprivation [36] and dietary deprivation [5], [28]. In the mir-80(Δ) background, hsf-1(RNAi) reverses the Ex_max shift that typifies DR (p<0.06, Table S1, Fig. 5A) and partially restores 4 day age pigment levels (p<0.05 compared to mir-80(Δ)+empty vector RNAi, Table S2, Fig. 5B). hsf-1(RNAi) does not affect Ex_max or age pigment levels in WT (data not shown). To determine if hsf-1 is also required for mir-80(Δ) longevity, we examined survival curves for the mir-80(Δ);hsf-1(sy441) double mutant. We find that disruption of hsf-1 eliminates the lifespan extension conferred by mir-80(Δ) (Fig. 5C). We conclude that hsf-1 is required for both mir-80(Δ)-induced fluorimetric features that typify DR and for mir-80(Δ)-induced longevity. Consistent with a role for hsf-1 in mir-80(Δ)-induced benefits, HSF-1 target gene hsp-16.2 transcripts are elevated in the mir-80(Δ) mutant (Fig. S6). In silico analyses did not reveal candidate miR-80 target sites in the hsf-1 transcript, suggesting indirect regulation in the mir-80(Δ)-induced DR pathway.

The CREB-binding protein CBP-1 is required for mir-80(Δ)-dependent changes in DR fluorimetric indicators and for mir-80(Δ)-dependent longevity

C. elegans histone acetyltransferase transcriptional coactivator homolog cbp-1 is required for lifespan extension via at least three different DR regimens (growth in axenic media, growth in diluted bacteria in liquid media, and the eat-2 feeding-impaired model). In the bDR regimen, cbp-1 deficiency has been shown to disrupt expression of daf-16 and hsf-1 target genes [32], [37] and thus the action of two transcription factors that influence mir-80(Δ) benefits has been mechanistically linked to CBP-1 in DR. We find that cbp-1(RNAi) in the mir-80(Δ) mutant reverses the DR-associated Ex_max shift (p<0.05 +/− RNAi; Fig. 6A), and increases age pigment levels in day 4 animals (p<0.09, +/− RNAi, Fig. 6B) (cbp-1(RNAi) does not affect Ex_max but modestly reduces age pigment levels in WT (data not shown)). Thus, cbp-1 activity plays a role in mir-80(Δ) regulation of age pigments, and appears generally needed for mir-80(Δ) DR metabolism. Consistent with a contribution to DR benefits, we find that the lifespan extension conferred by mir-80(Δ) depends strongly on cbp-1 (p<0.003; Fig. 6C).

Sequences within the cbp-1 transcript may be direct binding targets of miR-80

Interestingly, of the three transcription factors required for mir-80(Δ) healthspan, cbp-1 is the only one for which the transcript is predicted to include potential miR-80 miRNA target sequences (Fig. 6D). One candidate miR-80 binding site is present in the cbp-1 5′UTR, and another is present within exon 8. To test whether direct CBP-1 regulation might be a mechanism by which miR-80 controls metabolic state, we constructed translational reporters in which the cbp-1 promoter drives expression of a GFP that includes either no candidate miR-80 binding sites (NBS) or both the 5′UTR and the exon 8 candidate binding sites (5+8BS) (Fig. S7A).

We compared GFP expression levels of these constructs in ad lib fed animals +/− mir-80, with a focus on the posterior gut region in which mir-80 regulation is most dramatic. We find that the NBS construct is not regulated by miR-80(Δ) (Fig. S7B, left panel); whereas the 5+8BS construct is expressed at a higher level in the absence of mir-80 (Fig. S7B, right panel). Although rigorous testing in native context will be required to validate cpb-1 as a direct miR-80 target, our data suggest binding sites in the cbp-1 transcript may contribute to cbp-1 inhibition by miR-80 when levels are high in food.

If mir-80 represses cbp-1 translation, then we would expect higher levels of CBP-1 protein in mir-80(Δ) animals. We measured CBP-1 protein levels using anti-CBP antibodies against human CREBBP for WT and mir-80(Δ) mutants (day 7). We find that CBP-1 protein levels are significantly increased in mir-80(Δ) mutants compared to WT (p<0.05, Fig. 6E). Thus, in whole animal context, mir-80(Δ) is associated with increased CBP-1 protein.

Our data are consistent with a model in which in the presence of food, cbp-1 is translationally repressed by binding of miR-80 to target sites within the cbp-1 transcript (Fig. 7). When food is lacking, miR-80 levels drop, translational repression of cbp-1 is relieved, and CBP-1+DAF-16+HSF-1-mediated transcriptional changes induce DR within the animal. Interestingly, the human CREBBP transcript might be targeted by miR-80 family members or another miRNA homologous to the exon 8 site (Fig. S7C), suggesting miRNAs could exert a conserved role in DR metabolic regulation that might be harnessed in the future to promote healthy metabolism with anti-aging applications.

Discussion

Deletion of a single C. elegans miRNA, mir-80, induces systemic healthy aging—improving cardiac muscle-like and skeletal-muscle-like maintenance and function later into life, limiting age-associated accumulation of lipofuscin-like material in the gut, and extending lifespan. Our data indicate that miR-80 acts as a negative regulator of metabolic loops that promote DR metabolism when nutrients are scarce. Acetyltransferase CBP-1 acts together with DAF-16/FOXO and HSF-1 to promote healthy metabolism in this regulatory circuit. Sequences within the Cecbp-1 transcript, the protein product of which increases in DR nematodes ([32], Fig. 6E) and in hypoglycemic mouse [38], may serve as direct targets of miR-80 down-regulation when food is abundant. Similarities between miR-80/target features in nematodes and mammals raise the possibility that miRNA manipulation of related DR metabolic loops in humans might be recruited to promote healthy aging.

mir-80 is expressed broadly and is regulated by food availability

mir-80 is an abundant, widely-expressed miRNA, and thus might be involved in global regulation of metabolism coordinated across tissues. Indeed, multiple mir-80 reporters indicate broad cellular expression and regulation by E. coli food availability. However, not all tissues/cells reflect similar magnitudes of regulation, with the largest fold food-induced change in expression in posterior intestine (Fig. 3D). The dramatic gut regulation raises the possibility that intestinal cells, well-positioned to monitor nutrient uptake, might play the most critical role in metabolic sensing and control. We speculate that miR-80 level changes in intestinal cells might initiate body-wide signaling via gut secretion of insulins and other hormones, analogous to human gastrointestinal tract and adipose tissue hormonal signaling to hypothalamus [39]. Because mir-80(Δ) induces skn-1::GFP expression in the ASI neurons (previously suggested to be similar to hypothalamic neurons [7]) but mir-80 is not expressed in ASI neurons (Fig. S4A,B), relief of miR-80 repression under food limitation could act upstream of ASI skn-1 induction via a gut-to-neuron signaling relationship.

DAF-16, HSF-1 and CBP-1 transcription factors are needed for mir-80(Δ)-induced healthspan benefits in a likely complex regulatory circuit

The requirement for daf-16, hsf-1, and cbp-1 in mir-80(Δ) DR is interesting in multiple regards. First, DAF-16/FOXO and HSF-1 can each individually bind to CBP-1 in nematodes and mammals (C. elegans DAF-16 and CBP-1; mammalian FOXO3A and CBP [40]; mammalian HSF-1 and CBP1 [41]), underscoring their capacity to co-regulate transcription. Second, previous work identified C. elegans daf-16 and hsf-1 as required for the CBP-1-dependent bDR lifespan extension [37]. In the bDR study, cbp-1(RNAi) blocked expression of DAF-16 and HSF target genes sod-1 and sip-1, respectively, rather than blocking transcriptional induction of daf-16 and hsf-1 that accompanies bDR. These data suggest that the CBP-1 cofactor couples and modifies transcriptional outputs of DAF-16- and HSF-1-dependent longevity pathways under bDR conditions, a model likely to apply for mir-80(Δ)-induced DR.

Although our study focused on DR genes that have most dramatic impact on the age pigment DR signature, we emphasize that our data support that additional genes contribute in a complex network to regulate age pigment phenotypes in mir-80(Δ). For example, knockdown of either AMPK subunit encoded by the C. elegans genome, aak-1 or aak-2, can alter age pigment levels (Tables S1 and S2), but not Exmax shift, suggesting separate regulation of lipofuscin content and levels. We thus anticipate that our data just touches the surface of a large interrelated network of metabolic genes and processes that are regulated by miR-80.

Could miR-80 directly target the cbp-1 transcript to regulate protein levels?

We fully expect that miR-80 regulates dietary restriction by binding to multiple target transcripts. An interesting candidate target, however, is the cbp-1 gene itself, which we have shown to be critical for mir-80(Δ)-induced DR benefits. The potential cbp-1 target sites for miR-80 binding are unusual, being situated in the 5′ UTR and within a highly conserved exon. Interestingly, the 5′ UTR sequences in cbp-1 are perfectly conserved in C. brenneri, C. briggsae, and C. remanei (though not in C. japonica) and the exon 8 site is somewhat conserved among all (Fig. S7D). Exon targeting by miRNAs is common in plants [42] and has been demonstrated for mammalian transcription factors Nanog, Oct4, and Sox2 [43], [44], fly DICER [45], and is now predicted in many additional genes after algorithm refinements that consider coding sequences [45], [46].

Ideally, we could test direct miR-80 targeting in vivo by manipulation of a cbp-1 transgene, +/− candidate miR-80 binding sites. Technical challenges, including the long length of the cbp-1 gene/cDNA, as well as an apparent exquisite sensitivity of CBP-1 activity levels for health and viability [47], [48], precluded direct study. Our studies of expression of a GFP transgene flanked by the 5′ UTR and the exon 8 sites from cbp-1 supported that miR-80 can down-regulate artificial construct expression in posterior gut. Although not definitive proof of direct targeting, these data, together with our findings that CBP-1 protein levels are elevated in DR (Fig. 6E; DR induction of CBP-1 also reported in [32], [37]) and miR-80 levels drop in DR (Fig. 3, Fig. S3, S5) are consistent with a model in which miR-80 mediates DR regulation by directly effecting CBP-1 levels (Fig. 7). Even if miR-80 effects are indirect, it is clear that cbp-1 is critical for mir-80(Δ)-induced age pigment and lifespan changes. Given that cbp-1 plays a role in dietary restriction associated with growth in axenic medium, growth on diluted bacteria, and eat-2 feeding impairment [37] and intersects with the insulin pathway for lifespan extension [37], and that we have noted mir-80 expression regulation under bacterial dilution and dietary deprivation, and a partial engagement of the insulin signaling pathway in mir-80(Δ)-induced longevity (Fig. 4D), the miR-80/CBP-1 regulatory loop may constitute a core mechanism by which diverse and intersecting metabolic pathways are coordinately regulated to respond to nutrient availability.

Might the conserved miR-80 microRNA family regulate metabolism across species?

C. elegans miR-80 family

The most conserved mir-80 family members encoded in the C. elegans genome are mir-80, mir-58, mir-81 and mir-82 [12], [18]. We did not find an Ex_max shift in mir-58(Δ) or in the double mir-81(Δ) mir-82(Δ) mutant (data not shown) and thus mir-80 is the sole family member that can be deleted to induce the DR Ex_max shift. Interestingly, however, the quadruple mutant mir-80; mir-58; mir-81-82 has a very small body size, more severe than the scrawny body type we documented for mir-80(Δ) (Fig. 2C), which can be rescued by a mir-80 high copy number transgene [18] suggesting some functional redundancy among mir-80 family members.

Drosophila melanogaster

mir-80 is homologous to the miRNA bantam in Drosophila melanogaster (Fig. S7E), well studied for roles in developmental growth and cell death regulation [49], [50], [51], [52], and more recently implicated in regulation of the core circadian clock [53], neuronal dendritic growth regulation by epithelia [51], and regulation of ecdysone/insulin interplay that influences body size [54]. Our data raise the question of whether developmental processes or circadian clocks might be sensitive to metabolic state, or whether modulation of metabolism might be used to regulate growth.

Human

To date, there are 3 identified human miRNAs closely related to miR-80: hsa-mir-450b-3p; hsa-mir-556-5p, and hsa-mir-3689a-5p (Fig. S7E), none of which have been well studied for function in mammalian biology. The human hCBP coding sequences corresponding to Cecbp-1 exon 8 have sufficient homology to human miR-80 family members to raise the possibility of analogous interaction and conserved regulatory mechanism. We do note, however, that hCBP coding sequences in the Ce exon 8-homologus region have a stronger match to hsa-mir-136 (more distantly related to Cemir-80) (Fig. S7C). If the negative regulatory interaction between miR-80 (or specific miRNAs) and DR targets is conserved, disruption of human regulatory miRNAs might be exploited to promote healthy aging.

Materials and Methods

We grew C. elegans under standard conditions [55] at 20°C, on E. coli strain OP50-1 or HT115 for RNAi. Age-synchronized cultures were prepared by bleach treatment egg preparation, with hatch counted as day 1. Note all presented data represent 3 independent combined trials; error bars +/− SEM. Protocol details are provided in Figure Legends and Supplemental Methods. Note all presented data in this paper represent at least 3 independent combined trials; error bars +/− SEM; we counted age with the egg hatch corresponding to day 0.

Strains and plasmids

A detailed list of strains is included as Table S3.

We grew C. elegans under standard conditions [55] at 20°C unless otherwise indicated. The food sources we used were E. coli strain OP50-1 or HT115 for RNAi feeding experiments (Caenorhabditis Genetics Center, University of Minnesota, Twin Cities, MN, USA). To generate synchronized cultures, we bleached gravid adults and starved L1 progeny. The wild-type strain was var. Bristol N2 [55]. The mir-80(nDf53) allele breakpoints are 5′- tgctttcgatgtctatactctc -3′ and 5′-tctggcgaacgaaatgagt-3′, encompassing part of the promoter region, the entire precursor sequence and ∼300 bp downstream. We genotyped mir-80(Δ) by PCR using primer pairs mir80Out-F (5′- ttcgtcgccatcaacacacg-3′)+mir80Out-R (5′- gagcgcggatagatatacagtcag-3′) that flank the deletion and mir80Flank-F (5′- caacaacgatgtgaatgctcgtc-3′)+mir80Flank-R (5′- ctcgcacacggacggactgcc-3′) that bind internal to nDf53. We worked with a 6× outcrossed line. The mir-80 deletion mutant does not exhibit gross developmental phenotypes ([20]; our observations). Developmental timing, L1 nuclei numbers, early adult locomotion, pumping rates, defecation rates, amphid neuron dye filling, and dauer entry/exit behaviors are within wild type ranges in mir-80(Δ), supporting that mir-80 does not contribute an essential role in development and basic function. Thus, mir-80 deletion primarily impacts adult maintenance and DR phenotypes.

For the P_mir-80L:mCherry transcriptional reporter, we amplified the mir-80 promoter using primers 5′-cgagatgagaagtaagaagagtgg-3′ and 5′-tccgtgtgcgagagagtgagcgag-3′ and cloned into the Pmec4::mCherry plasmid vector at the start codon of mCherry from [56] using the In-fusion cloning kit (Clontech Inc). The resulting plasmid was injected at 50 ng/ul into wild type animals along with a rol-6 co-injection marker (100 ng/ul) to generate extrachromosomal transgenic lines ZB3039-ZB3043.

• For the binding site test constructs (Fig. S7A), we amplified the cbp-1 promoter (4.4 kb until start codon) using primers 5′-gACTAGTc tcttcc atgtcg gtttaa gcgcgg aaacgg tttttt aaa-3′ and 5′-tcccCCCGGGggga caatta gtagaa aaatgt atatat ttgac-3′ containing Spe1 and Xma1 restriction sites, respectively. This product was introduced within the Spe1-Xma1 digested pKS(-) vector. The mir-80 target sites were incorporated within primers that amplified the GFP coding sequence using the primer sites (outlined below) and introduced at the Xma1 site of the above cloned P_cbp-1 vector: NBS: 5′- tccccccgggatgagtaaaggagaagaacttttcactgg-3′+5′-cggggtaccctatagttcatccatgccatgtgtaatccc-3′; 5+8 BS: 5′- tccccccgggaacagctatatctggtgatttgatgagtaaagaagaag-3′+5′-cggggtacctaagatcctcttgactgaacacttcatagttcatccatgcc-3′

Measurement of fluorescent age pigments

We grew age-synchronized animals (see above) under standard conditions (20°C, OP50-1) and scanned animals (n≥50 per strain) for age pigment accumulation (Day 4, Day 9, Day 11) using a Fluorolog 3 spectroflorimeter as in Gerstbrien et al. [19]. All graphs represent mean data from at least 3 independent trials. For Ex_max determination at Day 4, we used Datamax software (Horiba Scientific) to identify the peak excitation value. The peak for tryptophan fluorescence was also analyzed to normalize scores, as TRP levels do not change markedly with age.

qRT-PCR experiments for assaying gene expression changes

We synchronized strains by alkaline bleaching [57] and placed synchronized L1 larvae (Day 1) on NGM plates seeded with OP50-1 bacteria. On Day 4 or day 7, we moved approximately half the animals to plates containing OP50-1 with 50 uM FUdR. We used the other half for total RNA extraction using TRIZOL as described below. ∼1.5 ug of total RNA was used for cDNA synthesis using the Invitrogen SuperScript III cDNA synthesis kit and OligoDT primers to synthesize cDNA from all poly-adenylated RNA. We used 100 ng of cDNA to measure gene expression levels using the standard curve approach. Standard curves were generated from wild type cDNA by utilizing multiple dilutions of cDNA (1000, 100, 10, 1, 0.1, 0.01 ng) and probing for expression levels of the house-keeping gene, actin (act-1). Primers used were act1RT-F (5′- ttactctttcaccaccaccgctga-3′) and act1RT-R (5′- tcgtttccgacggtgatgacttgt -3′) for act-1, ama1RT-F (5′- cctacgatgtatcgaggcaaa-3′) and ama1RT-F (5′- cctccctccggtgtaataatg-3′) for ama-1, hsf1RT-F (5′-tagtaatggcagagatgcgtgcga-3′) and hsf1RT-R (5′- tggctgcatgacagagacgagaaa-3′) for hsf-1 and hsp16.2RT-F (5′- atggaacgccaatttgctccagtc-3′) and hsp16.2RT-R (5′- tccttggattgatagcgtacgacc-3′) for hsp-16.2. We plotted C_t values obtained from amplification for target DR genes against this standard curve to determine transcript levels.

Quantification of GFP fluorescence from mir-80 target site constructs

We synchronized strains (refer Fig. S7A,B, and Table S3) by alkaline bleaching [57] and placed synchronized L1 larvae (Day 1) on NGM plates seeded with OP50-1 bacteria. On Day 4, 100 mCherry(+) animals were picked and GFP fluorescence was measured in the spectrofluorimeter at 488 nm excitation and 511 nm emission. We measured fluorescence using ImageJ with a region-of-interest (ROI) that included the entire length of the body (using the line tool) and then plotted a histogram of the mean intensity along the length of the line.

Western blot analysis for detection of CBP-1 protein

We synchronized strains by alkaline bleaching [57] and placed synchronized L1 larvae (Day 1) on NGM plates seeded with OP50-1 bacteria. On Day 7, 250 animals were placed in 50 ul of RIPA buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM Na₂EDTA, 1 mM EGTA, 1% NP-40, 1% Sodium deoxycholate, 2.5 mM beta-glycerophosphate, 1 mM Na₃VO₄)+Protein sample buffer and heated at 95°C for 15 mins. 25 ul of samples was loaded onto a MiniProtean TGX gradient gel (4–20%, Bio-Rad) and transferred onto PVDF membrane following separation. Membrane was blocked using 5% non-fat milk in PBST buffer for 1 hour. Membrane was then incubated with CBP-1 and TUB-1 antibodies (Santa Cruz Biotechnology) at 1∶500 and 1∶4000 dilutions respectively in 2% non-fat milk overnight at 4°C. Protein bands were detected using the ECL reagent (Invitrogen) using horseradish peroxidase conjugated secondary antibodies (Jackson ImmunoResearch Labs) at 1∶10,000 dilutions. Band intensities were calculated using ImageJ [58].

Supporting Information

Attachment 1

Attachment 2

Attachment 3

Attachment 4

Attachment 5

Attachment 6

Attachment 7

Attachment 8

Attachment 9

Attachment 10

Attachment 11

Attachment 12

Attachment 13