The mir-84 and let-7 paralogous microRNA genes of Caenorhabditis elegans direct the cessation of molting via the conserved nuclear hormone receptors NHR-23 and NHR-25

MicroRNAs (miRNAs) constitute a large class of small (∼22 nt) noncoding RNAs present across eukaryotic phylogeny. In Caenorhabditis elegans,miRNAs were discovered through genetics(Johnston and Hobert, 2003; Lee et al., 1993; Reinhart et al., 2000),cloning (Lau et al., 2001; Lim et al., 2003) and bioinformatic prediction (Ambros et al.,2003; Grad et al.,2003; Lim et al.,2003). Similar approaches revealed hundreds of miRNAs in plants,fungi and other metazoans (Bartel,2004). Most of the few metazoan miRNAs studied to date negatively regulate the expression of protein-coding genes by binding imperfectly complementary sites in the 3′ untranslated region (UTR) of the target mRNA and inhibiting translation (Lee et al., 1993; Olsen and Ambros,1999; Wightman et al.,1993). However, the mechanism of this translational block remains unclear. Also, some miRNAs previously thought to act primarily by blocking translation cause some degradation of their target transcripts(Bagga et al., 2005).

The let-7 and lin-4 miRNAs were discovered as mutations that alter the normally invariant cell lineage of C. elegans. Mutations in let-7 cause cell divisions normally restricted to the last larval stage to recur in adults(Reinhart et al., 2000). Similarly, mutation of lin-4 causes the reiteration of cell division patterns appropriate for the first larval stage, such that vulval and hypodermal tissues characteristic of adults never form(Ambros and Horvitz, 1984; Chalfie et al., 1981). By contrast, mutations in the coding regions of target genes of let-7and lin-4 cause the precocious execution of later larval or adult-specific programs during early larval stages. The let-7 miRNA is robustly expressed at the L4 stage, when let-7 binds to the lin-41 and hbl-1 mRNAs, causing LIN-41 and HBL-1 protein levels to decline (Abrahante et al.,2003; Lin et al.,2003; Reinhart et al.,2000; Slack et al.,2000). Freed of repression by LIN-41 and HBL-1, the transcription factor LIN-29 then directs the larval-to-adult transition in the epidermis,marked by fusion of the lateral seam cells, synthesis of an adult cuticle and the cessation of molting (Bettinger et al., 1996; Rougvie and Ambros,1995).

Many miRNAs are members of paralogous families(Grad et al., 2003; Lim et al., 2003), suggesting the importance of knowing whether paralogous miRNAs typically act in the same or different pathways and whether they share targets. The C. elegansgenome specifies three paralogs of let-7: mir-48, mir-84 and mir-241, all of which are expressed in a temporally regulated manner(Lau et al., 2001; Lim et al., 2003). Genetic analysis revealed that let-7 paralogs function redundantly to specify patterns of cell division during larval development(Abbott et al., 2005).

The life cycle of C. elegans includes four molts, when animals synthesize a new cuticle and shed their old one. Mutations in let-7,lin-4 or lin-29 cause animals to continue molting after reproductive maturity. Conversely, mutations in particular precocious heterochronic genes cause animals to synthesize an adult cuticle and exit the molting cycle prematurely (Ambros,1989; Jeon et al.,1999). Although the heterochronic pathway impacts the number of molts, the molecular mechanism by which heterochronic genes affect the molting cycle has not yet been described.

Molting is the hallmark of the ecdysozoan clade, which includes nematodes and insects (Aguinaldo et al.,1997). In insects, pulses of the steroid hormone ecdysone control transitions between life stages by activating stage-specific transcriptional cascades involving several nuclear hormone receptors, including ECR and USP,which together form the receptor for 20-hydroxyecydsone, as well as DHR3 andβFTZ-F1 (Riddiford et al.,2003). For example, the prepupal pulse of ecdysone induces expression of DHR3, the product of which in turn promotes expression of βFTZ-F1 (Lam et al.,1997; White et al.,1997). Abrogation of the function of βFTZ-F1 causes a defect in the prepupal-to-pupal transition(Broadus et al., 1999). Intriguingly, expression of let-7 in Drosophila correlates with pulses of ecdysone (Bashirullah et al., 2003; Sempere et al.,2002; Sempere et al.,2003).

The C. elegans genes nhr-23 and nhr-25 encode orphan nuclear hormone receptors orthologous, respectively, to DHR3 andβFTZ-F1, which are related to mammalian ROR/RZR/RevErb and SF-1,respectively. Both receptors are essential for completion of the larval molts(Asahina et al., 2000; Gissendanner and Sluder, 2000; Kostrouchova et al., 2001),suggesting that particular functions of nhr-23/DHR3 and nhr-25/ βFTZ-F1 might be conserved and, further, that regulation by steroid hormones might be a common feature of molting in C. elegans and Drosophila. However, a steroid hormone regulating molting of C. elegans has not yet been identified and the genome lacks orthologs of ECR or USP(Sluder and Maina, 2001).

Here, we show that mir-84 works together with let-7 to direct the terminal differentiation of the epidermis and cessation of the molting cycle. We show that genes normally expressed only before the larval molts are also expressed in let-7 mir-84 mutants as they enter a supernumerary molt. Moreover, we show that mir-84 and let-7control the molting cycle by regulating known targets in the heterochronic pathway as well as the nuclear hormone receptor genes nhr-23 and nhr-25.

Cultivation and genetic manipulation of C. elegans were performed using standard techniques (Sulston and Hodgkin, 1998). The mir-84(tm1304) deletion allele was generated by the laboratory of S. Mitani, and outcrossed to wild-type (N2) C. elegans six to eight times before analysis. GR1431 was generated via eight crosses to N2.

The mir-84 gene was PCR-amplified from genomic DNA using Taq polymerase (Roche) and primers GH21 5′-AAGTTGACTGACATGACAACCGAC-3′and GH32 5′-TTGACACAAAGGCAAGAGCTTG-3′. The mir-84::gfpreporter gene was generated via single-end overlap extension PCR(Hobert, 2002), fusing the mir-84 promoter sequence and gfp from vector pPD95.75 (A. Fire). The primers used were GH32, GH107 5′-TATTCATCATACGTCTGCCTGTGCATGCCTGCAGGTCGACTAGAG-3′, GH108 5′-CTCTAGTCGACCTGCAGGCATGCACAGGCAGACGTATGATGAATA-3′, and CAW32 5′-CCGCTTACAGACAAGCTGTGACCG-3′. For both constructs, three independent PCR reactions were combined to ensure that much of the product lacked unwanted mutations. To generate mgEx671, the mir-84gene was injected into N2 animals at a concentration of 15 ng/μl along with 50 ng/μl of plasmid DNA specifying the co-injection marker tub-1::gfp, kindly provided by Ho Yi Mak. Transgenic animals were irradiated with ultraviolet light to generate four independent lines in which the transgene integrated into a chromosome. mgIs45 and mgIs47 animals were outcrossed three or more times to N2 before analysis. To generate mgEx674, the mir-84::gfp fusion gene was injected into N2 animals at 10 ng/μl along with 25 ng/μl of plasmid DNA specifying the co-injection marker ttx-3::rfp, provided by Ho Yi Mak. Transgenic animals expressing mir-84::gfp were cultivated at 15°C, whereas other animals were typically cultivated at 20°C.

The mlt-10p::gfp-pest and nas-37p::gfp-pest fusion genes were previously described (Frand et al.,2005). To generate mgIs49, the mlt-10p::gfp-pestfusion gene was injected into wild-type (N2) animals at 10 ng/μl along with 50 ng/μl plasmid DNA specifying the co-injection marker ttx-3::gfp(Hobert et al., 1997), and 20 ng/μl pBluescript. Transgenic animals were irradiated with ultraviolet light to integrate the transgene into a chromosome. One integrant was backcrossed four times to N2 to generate GR1395.

Construct 4271, specifying nhr-23::gfp, was provided courtesy of J. Rall and colleagues (Kostrouchova et al., 1998). Plasmid pCG9, specifying nhr-25::gfp, was a kind gift from C. Gissendanner and A. Sluder(Gissendanner and Sluder,2000). To generate the extrachromosomal arrays mgEx728[nhr-23::gfp] and mgEx729[nhr-25::gfp], the plasmids were injected into wildtype (N2) animals at a concentration, respectively, of 10 or 20 ng/μl, along with plasmid pRF4, specifying the co-injection marker rol-6(su1006), to a final DNA concentration of 100 ng/μl.

The following strains were used in this study.

N2: wild type

RG365: him-1(e879) I; veIs13[col-19::gfp; rol-6(su1006)]V

SP231: mnDp1(X;V)/+ V; unc-3(e151) let-7(mn112) X

JR672: wIs54[scm::gfp] V

GR1395: mgIs49[mlt-10p::gfp-pest; ttx-3::gfp]

GR1368: mgEx656[nas-37p::gfp-pest; pha-1(+)]

GR1425: mgIs46[mir-84++; tub-1::gfp]; wIs54[scm::gfp]V

GR1426: mgIs45[mir-84++; tub-1::gfp] I; let-7(mn112)unc-3(e151) X

GR1427: mgEx674[mir-84::gfp; ttx-3::rfp]

GR1428: mgIs45[mir-84++; tub-1::gfp] I

GR1429: veIs13[col-19::gfp; rol-6(su1006)] V; mir-84(tm1304) X

GR1430: wIs54[scm::gfp] V; mir-84(tm1304) X

GR1431: mir-84(tm1304) X

GR1432: let-7(mg279) X

GR1433: let-7(mg279) mir-84(tm1304) X

GR1434: wIs54[scm::gfp] V; let-7(n2853) X

GR1435: wIs54[scm::gfp] V; let-7(n2853) mir-84(tm1304)X

GR1436: mgIs49[mlt-10p::gfp-pest; ttx-3::gfp] IV; let-7(mg279) X

GR1437: mgIs49[mlt-10p::gfp-pest; ttx-3::gfp] IV; mir-84(tm1304) X

GR1438: mgIs49[mlt-10p::gfp-pest; ttx-3::gfp] IV; let-7(mg279)mir-84(tm1304) X

GR1439: mgIs47[mir-84++; tub-1::gfp]; let-7(mg279)mir-84(tm1304) X

GR1440: mgIs47[mir-84++; tub-1::gfp]; let-7(mg279) X

GR1441: mgIs47[mir-84++; tub-1::gfp]

GR1442: mgIs47[mir-84++; tub-1::gfp]; mgIs49[mlt-10p::gfp-pest;ttx-3::gfp] IV; let-7(mg279) mir-84(tm1304) X

GR1443: mgEx656[nas-37p::gfp-pest; pha-1(+)]; let-7(mg279)mir-84(tm1304) X

GR1444: veIs13[col-19::gfp; rol-6(su1006)] V; let-7(mg279)mir-84(tm1304) X

GR1445: veIs13[col-19::gfp; rol-6(su1006)] V; let-7(mg279) X

GR1446: mgIs45[mir-84++] I; lin-29(n333) sqt-1(sc13) II; mgIs49[mlt-10p::gfp-pest] IV

GR1447: mgIs45[mir-84++]/+ I; lin-29(n333) sqt-1(sc13)II; mgIs49[mlt-10p::gfp-pest] IV

GR1448: mgEx728[nhr-23::gfp; rol-6(su1006)]

GR1449: mgEx729[nhr-25::gfp; rol-6(su1006)]

GR1450: mgEx729[nhr-25::gfp; rol-6(su1006)]; let-7(mg279)mir-84(tm1304)

GR1451: mgEx728[nhr-23::gfp; rol-6(su1006)]; let-7(mg279)mir-84(tm1304)

Images were captured on a Zeiss Axioplan microscope equipped with a Hamamatsu ORCA-ER digital camera and Openlab software (Improvision).

RNAi was performed essentially as described(Fraser et al., 2000), except that our nematode growth medium (NGM) contained 8 mmol/l isopropyl-β-D-thiogalactopyranoside and 25 μg/ml carbenicillin. Bacterial clones expressing double-stranded RNA were obtained from J. Ahringer(Fraser et al., 2000; Kamath et al., 2003) and M. Vidal (Rual et al., 2004). In the case of lin-28, we cloned C. elegans genomic DNA corresponding to nucleotides 3766 to 4098 of cosmid F02E9 (Accession number:emb|Z81494) into the same vector and bacterial strain(Rual et al., 2004).

RNA extractions and northern blots were performed essentially as described(Lee et al., 1993; Reinhart et al., 2000). We used a Starfire-labeled oligonucleotide probe (Integrated DNA Technologies)with sequence complementary to mir-84(5′-TACAATATTACATACTACCTCA-3′) and incubated blots at 44°C.

Protein extractions and immunoblots were performed as described(Reinhart and Ruvkun, 2001). Approximately 4 μg total protein were loaded per lane and transferred to Hybond ECL membrane (Amersham). We obtained anti-GFP monoclonal antibody and E7 β-tubulin monoclonal antibody, respectively, from Clontech and the Developmental Studies Hybridoma Bank at the University of Iowa. The Western Lightning ECL kit (Perkin-Elmer) and X-Omat Blue XB-1 or X-Omat AR film(Kodak) were used to detect signal.

To explore the function of microRNAs paralogous to let-7, we studied the mir-84 gene of C. elegans. Of three known paralogs, mir-84 is most similar in sequence to let-7,sharing 17 out of 22 nucleotides (Fig. 1A). The mir-84 gene is expressed from the first larval stage through adulthood (Abbott et al.,2005; Esquela-Kerscher et al.,2005), consistent with a role in postembryonic development. We obtained a strain bearing a 654 bp chromosomal deletion that includes the mir-84 gene from the Mitani lab of the Japanese National BioResources Project (Fig. 1B). No mir-84 could be detected by northern analysis in total RNA prepared from the mir-84(tm1304) mutant(Fig. 1C), verifying that tm1304 eliminates expression of the gene. To test whether loss of mir-84 caused phenotypes associated with tm1304, a transgene carrying many copies of the wild-type gene, including 989 bp 5′ and 813 bp 3′ of the mature mir-84 miRNA, was integrated into the genome. The abundance of mature mir-84 miRNA increased in wild-type animals carrying any one of four independent chromosomal arrays(Fig. 1D).

Given the sequence similarity between mir-84 and let-7,we expected that these miRNAs might target the same or similar mRNAs to control the timing of developmental events in particular tissues. We therefore asked whether loss of mir-84 would enhance mg279, a partial loss-of-function mutation in let-7. The level of mature let-7 miRNA is diminished in let-7(mg279) mutants because a deletion of 27 nucleotides upstream of the let-7 precursor hairpin impedes processing of the primary transcript(Bracht et al., 2004; Reinhart et al., 2000). We anticipated that some let-7(mg279) animals would initiate a supernumerary molt after reproductive maturity, because other mutations in let-7 cause a supernumerary, fifth molt(Reinhart et al., 2000).

We observed the molting behavior of ten individual let-7(mg279)mir-84(tm1304) double mutants, let-7(mg279) and mir-84(tm1304) single mutants, and wild-type animals for one day following the fourth molt. All the let-7(mg279) mir-84(tm1304) double mutants became immobile and ceased pharyngeal pumping, behaviors characteristic of lethargus, a period of inactivity that precedes every larval molt. By contrast, only two let-7(mg279) mutants and no mir-84(tm1304) or wild-type animals entered lethargus during this time. Interestingly, let-7(mg279) mir-84(tm1304) and let-7(mg279) mutants remained lethargic for over 6 hours (data not shown), whereas wild-type larvae reduce activity for only 2 hours before ecdysis (Singh and Sulston,1978). Most miRNA mutants that became lethargic also initiated a supernumerary molt but were unable to completely shed the cuticle(Fig. 2A). The mutants died shortly thereafter, when they ceased to lay eggs and their progeny hatched internally. At the end of this experiment, half the let-7(mg279)mir-84(tm1304) animals were dead, whereas all the single mutants were alive.

In subsequent experiments, we used lethality caused by the internal hatching of progeny as an indicator of a supernumerary molt. Fig. 2B shows that 98%(n=40) of let-7(mg279) mir-84(tm1304) double mutants died within 72 hours of the fourth molt at 22°C, compared with 28%(n=39) of let-7 single mutants. We observed no lethality in mir-84(tm1304) single mutant adults. Further, restoring expression of mir-84 by introducing the mgIs47[mir-84(++)] transgene rescued the viability of let-7(mg279) mir-84(tm1304) mutants. The loss of mir-84 thus accounts for enhancement of let-7(mg279), and the 1.8 kb DNA contained in mgIs47 is sufficient to confer the function of mir-84 in the cessation of molting. Together, the data show that mir-84 and let-7 work synergistically to inhibit lethargus and shedding of the cuticle during the adult stage.

We asked whether mir-84 and let-7 regulate particular genes expressed during larval molting, including the metalloprotease gene nas-37 and the novel gene mlt-10. nas-37 specifies a collagenase essential for ecdysis that probably degrades the pre-molt cuticle(Davis et al., 2004; Frand et al., 2005). Missense alleles of mlt-10 prevent ecdysis (A.R.F. and G.R., unpublished), but the function of the gene is not yet understood. A transcriptional fusion gene between green fluorescent protein (gfp) and mlt-10 or nas-37 is expressed in epithelial cells before each of the four larval molts, but is not expressed in adults that no longer molt(Frand et al., 2005).

We monitored expression of the mlt-10p::gfp-pest reporter in populations of animals cultivated at 25°C for 68 hours following release from starvation as L1 larvae. Animals expressed a pulse of GFP and then completed the fourth molt after approximately 42 hours of cultivation. The let-7(mg279) mir-84(tm1304) double mutants expressed an extra pulse of GFP, at levels comparable to those of wild-type animals late in the fourth larval stage, as judged by visual inspection and also by western analysis(Fig. 3). The majority of double mutants expressed GFP after 54 hours of cultivation, about the same time they began to lay eggs. Many let-7(mg279) adults also expressed GFP, but later, such that the majority of animals became fluorescent after 63 hours of cultivation. By contrast, none of the mir-84(tm1304) or wild-type animals was observed to express GFP after the fourth molt(Fig. 3A). Expression of mir-84 from the mgIs47 transgene restored repression of the mlt-10 reporter to let-7(mg279) mir-84(tm1304) adults(Fig. 3A). let-7(mg279)mir-84(tm1304) adults also expressed the gfp reporter for nas-37, consistent with a supernumerary molt(Fig. 3D). Loss of mir-84 thus promotes expression of genes characteristic of larval molting in adults with reduced levels of let-7.

We expected mir-84 and let-7 to promote exit from the molting cycle by regulating known targets in the heterochronic pathway. Inactivation of any one of five precocious heterochronic genes, lin-14,lin-28, lin-42, lin-41 or hbl-1, is sufficient to suppress mutations in let-7. We found that RNA-interference (RNAi) of lin-42, hbl-1 or lin-41 fully suppressed the supernumerary pulse of expression of mlt-10p::gfp-pest as well as the inviability of let-7(mg279) mir-84(tm1304) adults(Fig. 4). Inactivation of lin-14 or lin-28 likewise abrogated expression of the gfp reporter in let-7(mg279) mir-84(tm1304) mutants, but only when animals were fed the corresponding bacterial clones continuously for two generations (Fig. 4). RNAi of the lin-14 and lin-28 genes might be less effective in a single generation because lin-14 and lin-28 are downregulated early in larval development by lin-4(Feinbaum and Ambros, 1999; Lee et al., 1993; Moss et al., 1997; Wightman et al., 1993). Thus, mir-84 and let-7 act through the heterochronic pathway to prevent molting in the adult stage. Further, we identify mlt-10 and nas-37 as targets of the heterochronic pathway, consistent with our previous report that mlt-10p::gfp-pest is expressed in adults that continue molting due to inactivation of lin-29(Frand et al., 2005), the transcription factor gene farthest downstream in the heterochronic pathway.

We hypothesized that let-7 and related miRNAs ensure the cessation of molting by directly or indirectly repressing genes that otherwise provoke a molt, including the conserved nuclear hormone receptor genes, nhr-23and nhr-25, that are key regulators of the larval molting cycle(Gissendanner and Sluder,2000; Kostrouchova et al.,2001). We therefore asked whether nhr-23 and nhr-25 were required for let-7(mg279) mir-84(tm1304) mutants to enter the supernumerary molt. Fig. 5 shows that inactivation of either nhr-23 or nhr-25 by RNAi restored viability and repression of mlt-10p::gfp-pest to the vast majority of let-7(mg279)mir-84(tm1304) adults. Likewise, among let-7(mg279)mir-84(tm1304) mutants suppressed by RNAi of nhr-25 or nhr-23, respectively, only 1 of 20 or 0 of 16 expressed nas-37p::gfp-pest as gravid adults. By contrast, 37% (n=83)of mutants fed control bacteria expressed the nas-37 reporter gene(data not shown). Double mutants fed nhr-23 or nhr-25 dsRNA typically remained active and did not shed their cuticle, in stark contrast to animals fed control bacteria not expressing dsRNA for a worm gene. Thus,inactivation of either nhr-23 or nhr-25 was sufficient to block initiation of the supernumerary molt in let-7(mg279)mir-84(tm1304) mutants.

The model that mir-84 and let-7 act via regulation of nuclear hormone receptors predicts an increase in the levels of NHR-23 and NHR-25 in adults with less of the miRNAs. Indeed, we found that fluorescence from a gfp fusion gene containing the promoter and first six exons of nhr-23 (Kostrouchova et al.,1998) was approximately fourfold brighter in the hypodermal nuclei of let-7(mg279) mir-84(tm1304) adults than in wild-type animals(Fig. 5C). The miRNAs probably regulate nhr-23 indirectly, considering that the 3′ UTR of nhr-23 lacks obvious binding sites for the let-7 family and that this particular gfp reporter lacks the native 3′ UTR of nhr-23. One possibility is that transcription of nhr-23 is repressed by LIN-29. A similar gfp reporter for nhr-25, also lacking the native 3′ UTR(Gissendanner and Sluder,2000), showed only a modest increase in expression in let-7(mg279) mir-84(tm1304) mutants compared with wild-type gravid adults (data not shown).

To address the possibility that let-7 and related miRNAs target the nhr-25 message, we searched the 3′ UTR of the nhr-25 gene for binding sites using the computer program RNAhybrid(Rehmsmeier et al., 2004). We identified one site apt to form a helix that includes the first seven nucleotides of mir-84 and lacks G:U base-pairs or bulged nucleotides within that seed region (Fig. 6), features characteristic of high-quality miRNA binding sites(Doench and Sharp, 2004; Lall et al., 2006). Two additional sites were predicted to form helices with let-7 or mir-241 and mir-48 that contain G:U base-pairs and bulged nucleotides within the seed region, features present in experimentally validated let-7 binding sites in lin-41(Vella et al., 2004b; Vella et al., 2004a). All three sites are highly conserved among the 3′ UTRs of nhr-25from the nematodes C. briggsae, C. remanei and C. elegans,consistent with interaction with the let-7 paralogs in vivo(Fig. 6B). The let-7family of miRNAs might thereby directly target the nhr-25 message,reinforcing transcriptional downregulation of nhr-25 during the adult stage (Gissendanner and Sluder,2000).

We considered that mir-84 and let-7 might target additional genes that promote molting and therefore examined the predicted 3′ UTRs (Hajarnavis et al.,2004) of 47 genes identified as essential for completion of the larval molts through a genome-wide RNAi screen(Frand et al., 2005 and Table 1 within). We identified potential binding sites for let-7 and related miRNAs in alg-1 and pan-1, the latter consistent with work by Lall and colleagues (Lall et al., 2006)(see Fig. S1 in the supplementary material). We are exploring the significance of these sites to the regulation of molting.

As a complementary strategy to identify genes acting downstream of the let-7 paralogs in genetic pathways regulating the molting cycle, we asked whether inactivation of additional genes required for the larval molts(Frand et al., 2005) would prevent the supernumerary molt of miRNA mutants. RNAi of the ribosomal protein gene rpl-27, rpl-31 or rpl-32 restored viability and repression of the mlt-10p::gfp-pest reporter to many let-7(mg279)mir-84(tm1304) adults, suggesting that protein synthesis is essential for initiation or progression of the supernumerary molt (see Table S1 and Table S2 in the supplementary material). RNAi of 89 other genes did not significantly suppress both indicators of the supernumerary molt, at the threshold of P ≤0.001 in chi-square tests. The rpl genes are unlikely to serve as direct targets of let-7 or paralogous miRNAs, because their 3′ UTRs lack obvious binding sites and contain no more than 47 nucleotides. Also, let-7(mg279) mir-84(tm1304) mutants suppressed by RNAi of an rpl gene were smaller and less active than those suppressed by RNAi of nhr-23 or nhr-25, suggesting that NHR-23 and NHR-25 are the main effectors of let-7 and mir-84in the regulation of molting.

Because we identified potential binding sites for the let-7 family members among genes essential for molting, we predicted that increased expression of mir-84 would suppress mutations in lin-29,bypassing the canonical heterochronic pathway. We examined the molting phenotypes caused by a probable null allele, lin-29(n333)(Bettinger et al., 1996; Rougvie and Ambros, 1995), in the presence or absence of excess mir-84, comparing mgIs45[mir-84++]; lin-29(n333); mgIs49[mlt-10p::gfp-pest]animals to segregants from an mgIs45 heterozygote (GR1447). Individuals expressing mlt-10p::gfp-pest were selected late in the fourth larval stage and observed several times over the next 29 hours of cultivation at 25°C. In total, 57% (17/30) of lin-29(n333)mutants carrying the mir-84++ transgene expressed GFP as adults,whereas all (30/30) animals lacking mgIs45 expressed an extra pulse of GFP (P ≤0.001, chisquare test). Moreover, only 21% (7/30) of mir-84++ animals completed a supernumerary molt, whereas 86% (25/29)of animals lacking mgIs45 molted, indicated by shedding of the cuticle (P ≤0.001, chi-square test). The observation that overexpression of mir-84 can prevent or delay the supernumerary molt of lin-29(n333) mutants supports the view that mir-84directly targets particular mRNAs, the products of which otherwise provoke a supernumerary molt.

To determine where and when mir-84 is expressed, we fused the gfp gene and the unc-54 3′ UTR to the putative mir-84 promoter, a 989 bp sequence 5′ of the mature miRNA. Transgenic mgEx674[mir-84::gfp] animals expressed GFP in the lateral hypodermal seam cells (Fig. 7)and other cells (see Fig. S2 in the supplementary material). The mir-84::gfp reporter was expressed in seam cells during early larval stages in some transgenic animals (Fig. 7B), although expression was more prevalent in L3-stage and older animals (Fig. 7A; see Fig. S2A in the supplementary material). Our results thus suggest a broader temporal expression pattern for mir-84 than previously described by Esquela-Kerscher and colleagues(Esquela-Kerscher et al.,2005), who observed expression of a mir-84::gfp fusion gene in the seam cells beginning only in the L4 stage. We saw a similar pattern of mir-84::gfp expression in three independent transgenic lines (data not shown). Further, we saw no obvious difference in expression between this particular mir-84::gfp fusion gene and one in which 8.1 kb of sequence upstream of mir-84 was fused to yellow fluorescent protein, kindly provided by A. Yoo and I. Greenwald(Fig. 7C; see Fig. S2F in the supplementary material). Consistent with synergistic functions for mir-84 and let-7, mutations in let-7 impact the development of many tissues that express mir-84::gfp, including the seam cells and vulva, and a let-7::gfp fusion gene is expressed in the seam cells beginning at the L4 larval stage(Johnson et al., 2003).

Given that mir-84 regulates termination of molting and is expressed in epithelial cells that synthesize the cuticle, we anticipated that mir-84 might promote the terminal differentiation of particular epithelial cells, including the lateral seam cells and the major body hypodermal syncytium, hyp7. Seam cells divide in a stem-cell like fashion at the larval-to-larval molts, but terminally differentiate, fuse and secrete a cuticular structure called alae at the larval-to-adult molt(Sulston and Horvitz, 1977). To examine the role of mir-84 in the seam-cell lineage, we used the temperature-sensitive mutation let-7(n2853) to sensitize the genetic background and an scm::gfp fusion gene(Terns et al., 1997) to visualize the nuclei of seam cells. let-7(n2853) mir-84(tm1304) young adults had an average of 22.4±2.7 (n=17) seam cells when cultivated at 20°C (Table 1), whereas let-7(n2853) young adults had 19.2±3.0(n=23) seam cells, a modest but significant difference(P<0.001, Student's t-test). By contrast, wildtype adults had 16.1±0.2 (n=17) seam cells, similar to all animals observed at the L4 stage. We also examined the cuticle of young adults. Animals with reduced levels of let-7 due to the mutation mg279 form less distinct alae than wild-type animals(Reinhart et al., 2000). The alae of let-7(mg279) mir-84(tm1304) double mutants were even less prominent than those of let-7(mg279) animals (data not shown).

To examine the role of mir-84 and let-7 in differentiation of the hypodermis, we used a gfp reporter for the cuticle collagen gene col-19. The fusion gene is expressed only in the adult stage, in both the hypodermis and seam cells(Abrahante et al., 1998). Only 25% (n=84) of let-7(mg279) mir-84(tm1304) young adults expressed GFP in both the hypodermis and seam cells, whereas 63%(n=70) of let-7(mg279) mutants expressed GFP in both tissues, a significant difference (P ≤0.001, chi-square test)(Fig. 8). Similarly, Abbott and colleagues (Abbott et al.,2005) show that mir-48(n4097); mir-84(n4037) double mutants fail to express a col-19::gfp reporter in the hypodermis,even though the mutants express that particular col-19::gfp in seam cells. Together, our observations indicate that loss of mir-84exacerbates defects in the terminal differentiation of the seam cells and hypodermis caused by mutations of let-7.

Given the similarities between the nucleotide sequences and the spatial and temporal expression patterns of mir-84 and let-7, we expected that mir-84 could functionally substitute for let-7. Animals with a null mutation in let-7 burst at the vulva during the L4 stage, and therefore rarely have progeny(Reinhart et al., 2000). We found that 93% (n=15) of let-7(mn112) mutants overexpressing mir-84 from the mgIs45 transgene survived and produced progeny, whereas only 3% (n=31) of let-7(mn112) animals produced progeny in the absence of auxiliary mir-84(Fig. 9). Rescue of the null allele of let-7 required robust expression of mir-84,because mgIs47, which drives a lower level of mir-84expression than mgIs45 (Fig. 1D), failed to suppress let-7(mn112) (data not shown). However, the mgIs47 transgene did reduce lethality caused by let-7(mg279) (Fig. 2B). Thus, mir-84 can substitute for let-7 when abundant. Alternatively, suppression of the let-7 null allele by mgIs45 might be attributable to precocious developmental events caused by overexpression of mir-84 (G.D.H. and G.R., PhD thesis,Harvard University, 2005) (Johnson et al.,2005), considering that mutations in precocious heterochronic genes also suppress let-7 mutations(Slack et al., 2000). Similar to our findings with mir-84, increased expression of the let-7 paralog mir-48 also suppresses lethality caused by the loss of let-7 (Li et al.,2005).

Here, we show that mir-84 functions synergistically with the paralogous miRNA let-7 to promote the transition from larval to adult developmental programs, including the terminal differentiation of particular epithelial cells and the cessation of molting. Our findings suggest that the let-7 family of miRNAs work in a combinatorial mode to repress particular targets. Indeed, animals that lack all three paralogs of let-7, but express let-7 itself, fail to repress a reporter for the let-7 target gene hbl-1 or exit the molting cycle at the appropriate time (Abbott et al.,2005).

Fig. 10 shows a genetic model for the function of mir-84 and let-7 in epithelial differentiation, as related to the molting cycle. The let-7 miRNA targets lin-41 mRNA (Slack et al., 2000) and also hbl-1 mRNA, in combination with paralogous miRNAs (Abbott et al.,2005; Abrahante et al.,2003; Lin et al.,2003). During early larval development, LIN-41 and HBL-1 together repress production of the zinc-finger transcription factor LIN-29(Abrahante et al., 2003; Rougvie and Ambros, 1995; Slack et al., 2000). Expression of let-7 and related miRNAs late in larval development represses lin-41 and hbl-1, thereby activating LIN-29. LIN-29 promotes expression of col-19 and possibly other collagen genes characteristic of an adult cuticle and also represses expression of col-17 and possibly other collagen genes characteristic of larval cuticle (Bettinger et al.,1996; Liu et al.,1995; Reinhart et al.,2000; Rougvie and Ambros,1995). LIN-29 is likely to regulate additional genes that control the molting cycle that have not yet been identified.

Here, we show that inactivation of either one of the nuclear hormone receptor genes nhr-23 or nhr-25 is sufficient to prevent the aberrant supernumerary molt caused by reduced levels of mir-84 and let-7. NHR-23 and NHR-25 thus serve as key downstream effectors of the miRNAs in regulation of the molting cycle(Fig. 10). One model is that LIN-29, or a transcription factor regulated by LIN-29, represses nhr-23 and nhr-25 following the fourth molt. Accordingly,GFP expression from an nhr-23 reporter gene increases fourfold in the hypodermis of let-7 mir-84 adults. The relationship between nhr-23 and nhr-25 in C. elegans remains to be determined; however, DHR3 stimulates transcription of βFTZ-F1 in flies (Lam et al., 1997; White et al., 1997).

The identification of sites in the 3′ UTR of nhr-25 that are complementary to let-7 family members and are also conserved in other nematodes suggests that the let-7 family targets the nhr-25message to negatively regulate production of NHR-25 in adults(Fig. 10). Consistent with this model, increasing the abundance of mir-84 partly suppresses the supernumerary molt caused by a probable null mutation in the lin-29gene. Also, in preliminary experiments we have detected RNA species attributable to cleavage of the nhr-25 message upon binding of let-7-like miRNAs in extracts from wildtype adults, using the method of Bagga et al. (Bagga et al.,2005). Steroid hormones and co-factors probably also regulate activity of NHR-23 and NHR-25 during the life cycle.

Regulation by miRNAs thus converges on transcription factors upstream in the genetic networks regulating molting. NHR-23 coordinates several aspects of larval molting by promoting expression of genes required for patterning the new cuticle and ecdysis, including, respectively, the collagen gene dpy-7 and the collagenase gene nas-37(Frand et al., 2005; Kostrouchova et al., 1998; Kostrouchova et al., 2001). Here, we show that inactivation of either nhr-23 or nhr-25abrogates the reiterated expression of gfp reporters for mlt-10 and nas-37 caused by mutation of let-7 and mir-84. NHR-25 might promote expression of the corresponding genes during larval development, even though RNAi of nhr-25 is not sufficient to abrogate expression of the gfp reporters in wild-type larvae (Frand et al., 2005). Interestingly, inactivation of nhr-23 or nhr-25 causes an earlier blockade in the molting program in let-7 mir-84 adults than in wild-type larvae, such that the mutant adults do not enter lethargus or attempt to ecdyse. Parallel pathways might drive early steps of molting during larval development.

Intriguingly, adults with reduced levels of mir-84 and let-7 are unable to shed their cuticle to complete the supernumerary molt. One possibility is that particular genes required for ecdysis are not induced. Whereas the hypodermis and seam cells retain some larval character in let-7 mir-84 adults, other cells, perhaps particular neurons or specialized epithelia, might be fully differentiated and therefore unable to coordinate with the molting program. Consistent with this idea, let-7 mir-84 adults spend an atypically long time in lethargus, suggesting a failure to exit the behavioral program. Alternatively, particular structural features of the fifth cuticle might be physically incompatible with shedding the exoskeleton.

Considering an aberrant ecdysis as the terminal phenotype of let-7 mir-84 mutants, it is intriguing to speculate that the let-7family and possibly other miRNAs regulate aspects of the larval molting cycle. Indeed, increased expression of either mir-84 or let-7causes some larvae to arrest development, trapped inside partly shed cuticle,indicating that levels of let-7-like miRNAs can impact molting of larvae (G.D.H. and G.R., unpublished).

Mechanisms that set the pace of the molting cycle are not well understood,although physiologic cues such as nutritional status(Ruaud and Bessereau, 2006)and environmental cues such as temperature impact the duration of larval stages. Interestingly, let-7 and let-7 mir-84 mutants initiate the supernumerary molt in synchrony, rather than in a stochastic fashion, relative to the time of hatching. Thus, a timing mechanism for molting persists in these particular miRNA mutants.

The let-7 gene is perfectly conserved throughout bilaterian phylogeny (Pasquinelli et al.,2000), and vertebrate genomes specify many miRNAs homologous to let-7 (Lagos-Quintana et al.,2001). Vertebrate let-7 and protein-coding genes orthologous to targets of let-7 identified in C. elegansplay crucial roles in development(Kloosterman et al., 2004; Moss and Tang, 2003). Moreover, reduced expression of human let-7 correlates with shortened survival in lung cancer patients(Takamizawa et al., 2004), and let-7 might regulate the RAS oncogene(Johnson et al., 2005). The possibility of functional conservation among homologs of let-7 in humans and worms intimates the importance of understanding how let-7and its paralogs function in C. elegans. Our work shows how analysis of double mutants can reveal how the many miRNAs that form paralogous families work together to regulate their targets.

We thank all members of the Ruvkun lab for support throughout this project. We thank the anonymous referees, I. Greenwald and J. Kim for critical review of the manuscript, and A. Pasquinelli for technical advice and discussion. We thank S. Mitani, J. Rothman, A. Rougvie, A. Yoo and I. Greenwald for sharing strains and reagents. This work was supported by NIH grant GM44619 to G.R.