Dual Regulation of the lin-14 Target mRNA by the lin-4 miRNA

microRNAs (miRNAs) are ∼22 nt regulatory RNAs that in animals typically bind with partial complementarity to sequences in the 3′ untranslated (UTR) regions of target mRNAs, to induce a decrease in the production of the encoded protein. The relative contributions of translational inhibition of intact mRNAs and degradation of mRNAs caused by binding of the miRNA vary; for many genetically validated miRNA targets, translational repression has been implicated, whereas some analyses of other miRNA targets have revealed only modest translational repression and more significant mRNA destabilization. In Caenorhabditis elegans, the lin-4 miRNA accumulates during early larval development, binds to target elements in the lin-14 mRNA, and causes a sharp decrease in the abundance of LIN-14 protein. Here, we monitor the dynamics of lin-14 mRNA and protein as well as lin-4 miRNA levels in finely staged animals during early larval development. We find complex regulation of lin-14, with the abundance of lin-14 mRNA initially modestly declining followed by fluctuation but little further decline of lin-14 mRNA levels accompanied by continuing and more dramatic decline in LIN-14 protein abundance. We show that the translational inhibition of lin-14 is dependent on binding of the lin-4 miRNA to multiple lin-4 complementary sites in the lin-14 3′UTR. Our results point to the importance of translational inhibition in silencing of lin-14 by the lin-4 miRNA.


Introduction
MicroRNA (miRNAs) are ,22-nucleotide noncoding RNAs present in plant and metazoan phylogeny that repress gene expression post-transcriptionally. In animals, miRNAs usually repress target mRNAs by base-pairing to their 39UTR, which have perfect or near-perfect sequence complementarity to nucleotides 2-7 (the ''seed region'') of miRNAs, with bulges and mismatches in other regions of the miRNA-mRNA duplex [1]. The mechanisms by which animal miRNAs repress target mRNA stability or translation are still an open question. In particular, the relative contributions of mRNA degradation and translational repression in miRNA-mediated silencing are variable depending on the biological and experimental context (for review, see [2] and [3]). There are examples of miRNAs that primarily inhibit protein synthesis of their target mRNAs without affecting mRNA stability [4,5,6]. On the other hand, some analyses of other miRNA target genes only show a modest decrease in protein synthesis rate, arguing that most of the silencing is attributable to mRNA destabilization by miRNAs [7]. Although seemingly controversial, the multiple mechanisms of miRNA-mediated silencing are not necessarily mutually exclusive. Several recent studies monitoring the temporal effects of miRNA-mediated silencing gave rise to the hypothesis that miRNAs silence their target genes by an initial phase of translational repression, followed by mRNA deadenylation and decay, which further consolidate the silencing effect [8,9,10,11,12]. Even though this hypothesis reveals a dynamic process and reconciles some of the previous controversies, it should be noted that many of the studies were in vitro assays or experiments using synthetic reporters and/or via manipulating miRNA abundance. Moreover, the conclusions of the animal miRNA field have been plagued by the difficulty in assigning bona fide target mRNAs to miRNAs; as a result, the modest responses of hundreds of candidate miRNA targets may not represent the core mechanism that is subject to evolutionary selection. Therefore, instead of monitoring the entire genome, we focus on the C. elegans lin-4 miRNA regulation of lin-14, an interaction that is biologically significant and subject to natural selection. lin-14, encoding a transcription factor, is a major target of the lin-4 miRNA and contains seven conserved lin-4 complementary sites in the 39UTR [13,14,15]. Deletion of these complementary sites in lin-14 gain-offunction mutants or a loss-of-function mutation in the lin-4 miRNA disrupts the normal down-regulation of LIN-14 protein, which normally starts at the mid-first larval (L1) stage, and results in retarded heterochronic phenotypes [14,16,17].
Here, we monitored the abundance of lin-14 mRNA and protein as well as lin-4 miRNA levels in finely staged animals over the course of early larval development in both wild-type animals and mutants in which lin-14 is relieved from repression by the lin-4 miRNA. Our results revealed complex regulation of lin-14. In wild-type animals, lin-14 mRNA levels first decline as soon as lin-4 miRNA levels rise during the early L1 stage, accompanied by subsequent decrease in the abundance of LIN-14 protein. As lin-4 miRNA levels rise during the late L1 stage, lin-14 mRNA levels fluctuate with no obvious overall decline, whereas LIN-14 protein levels continue to decrease. In contrast, no decrease in the LIN-14 protein levels was observed in lin-14 mutant animals in which the lin-4-complementary sites in the 39UTR are deleted. Together, our analyses point to two phases of regulation: a fast lin-14 mRNA destabilization phase, and long-term translational inhibition that is important in maintaining the silencing of lin-14 by lin-4 miRNA.

Experimental Procedures
Characterization of the lin-14(n355) inversion RNA samples derived from wild type and n355 were analyzed by 39 RACE (Roche). The most prominent band that was detected in n355 and not wild type corresponded to the length of the lin-14 39UTR in the n355 mutant that was deduced by Wightman et al. (1991). Sequencing of this band showed that at position 254 of the 39UTR the lin-14 sequence was fused to an intergenic region of the X chromosome corresponding to cosmid ZC373. PCR using primers homologous to lin-14 and ZC373 and DNA from the n355 mutant as template yielded bands that confirmed the chromosomal structure in the n355 mutant that was predicted by 39 RACE (data not shown).

Quantitative Western blot
Odyssey CLx Infrared Fluorescent Western Blot was performed following the vendor's protocol. Briefly, NuPAGEH LDS sample buffer (26) equal to the total volume of the worms was added and samples were boiled for 5 minutes. The lysates were pelleted in a microfuge and the supernatant loaded to a 4-12% Bis-Tris gel (Invitrogen) for electrophoresis. The proteins were transferred to nitrocellulose membrane (Bio-Rad). The membrane was blocked in blocking buffer (5% nonfat milk in PBS+0.1% TweenH 20) for 1.5 h at room temperature. Subsequently, the membrane was incubated in rabbit anti-LIN-14 (1:1000) and mouse anti-actin (1:5000, Abcam) primary antibodies at 4uC overnight. After washing four times with PBS+0.1% TweenH 20, the membrane was incubated in RDye 800 CW Goat anti-Rabbit IgG (1:10000) and IRDye 680 RD Goat anti-Mouse IgG (1:5000) at room temperature for 1 hour. The membrane was washed four times with PBS+0.1% TweenH 20, briefly rinsed with PBS and imaged with the Odyssey CLx infrared imaging system. LIN-14 and actin blots were scanned using the 800 nm and 700 nm channels, respectively. The abundance of LIN-14 was quantified by normalization to actin.
Quantitative RT-PCR analysis of lin-14 mRNA Total RNA was purified using TRIzol Reagent (Molecular Research Center), DNase treated (TURBO TM DNase, Ambion), and reverse-transcribed using oligo(dT) 20 primer (Invitrogen). The PCR amplification was monitored by SYBR green incorporation, and a corresponding threshold cycle (C T ) was obtained. The quantity of lin-14 mRNA, relative to rpl-32, was calculated using the formula 2 -DCT , where DCT = (C T lin-14 -C T rpl-32 ). The lin-14 level was then normalized to its level at 0 h after release from L1 diapause. Primers used: lin-14 forward primer spanning the last exon-exon junction (caaaaactgagagcgaaacg) and lin-14 reverse primer in the last exon (tggaccttgaagaggaggag).

Taqman miRNA Assays
Taqman miRNA Assays (Applied Biosystems) were performed following the vendor's protocol. Briefly, 100 ng total RNA was reverse transcribed using a miRNA-specific RT primer. The realtime PCR amplification was performed and a corresponding threshold cycle (C T ) was obtained. The quantity of lin-4 miRNA, relative to U6 as an internal reference gene, was calculated using the formula 2 -DCT , where DCT = (C T lin-4 -C T U6 ). The lin-4 level was then normalized to its level at 0 h after release from L1 diapause.

Northern Blot
We collected synchronized embryos and L4 staged animals of lin-14(n536n540)/szT1 heterozygotes. szT1 is a (I;X) translocation balancer chromosome that balances the slow growing and very small brood size of the lin-14(n536n540) homozygotes. While the lin-14(n536n540)/szT1 mutants segregate homozygous larval lethal szT1 homozygotes and homozygous lin-14(n536n540) animals, they constitute a minor component of the population because of the much larger brood and faster growth of the lin-14(n536n540)/szT1 heterozygotes. Total RNA was purified using guanidium isothiocyanate disruption and purification through a CsCl cushion via ultracentrifugation, polyA selected, and separated on a 1.2% formaldehyde agarose gel. RNA was transferred to Nylon membrane, UV crossed linked, then hybridized to radiolabeled DNA probes that were generated from the 3.8 kb EcoRI fragment bearing the last 7 exons of lin-14 and its 39UTR. A DNA probe for the histone gene was used for normalization of mRNA content per lane, and a probe for the vitellogenin gene was used to indicate animal stages. X-Omat films exposed to the Northern blot were scanned on an optical scanner.

Characterization of the lin-14(n355gf) mutant
The lin-14(n355) gain-of-function mutation was initially described as a possible translocation that separates the lin-14 promoter and coding region from most of its 39UTR, relieving it from repression by the lin-4 miRNA [17]. To characterize the structure of the 39UTR in the n355 mutant we performed 39 RACE (see Materials and Methods for details). We found that n355 is an inversion that causes a break at nucleotide 254 of the lin-14 39UTR and a fusion to an intergenic sequence of the X chromosome corresponding to cosmid ZC373, downstream of the gene col-176 (the sequences flanking the fusion were lin-14 to ZC373: attatccccaTCATTTCGAG). A series of adenosine residues that did not correspond to the genomic sequence, presumably indicating the polyadenosine tail, began at position 82 of the sequence derived from ZC373, suggesting that the lin-14(n355) 39UTR is ,335 nt, far shorter than the normally 1.6 kb wild-type lin-14 39UTR (data not shown). The n355 inversion removes all of the characterized complementary sites for lin-4 [14] from the lin-14 39UTR (Fig. 1). The n355 mutation causes retarded heterochronic phenotypes and continued expression of the LIN-14 protein at larval stages later than the L1 stage, similar to a lin-4 null mutant, further indicating that the n355 mutation prevents the lin-14 39UTR from responding to lin-4 [13,16].

Temporal analyses revealed two phases of regulation of lin-14 by lin-4 miRNA
To resolve the dynamics of lin-4 mediated down-regulation of lin-14, we performed temporal analyses of lin-14 mRNA and protein as well as lin-4 miRNA levels in finely staged wild-type and lin-14(n355n679) mutant animals collected at 3-hour intervals from the early first larval (L1) to early second larval (L2) stage at 20uC. The n355 allele decouples lin-14 from lin-4 repression resulting in gain-of-function retarded phenotypes and slower development rate, but the lin-14(n679) V299D missense mutation confers a temperature-sensitive suppression of the lin-14(n355) retarded phenotype, due to a reduction-in-function mutation in the encoded LIN-14 protein. Therefore, while the LIN-14 protein stays at a high level after the L1 stage, that temporally misexpressed LIN-14 protein is non-functional for specification of L1 cell fates [18]. At 20uC, the lin-14(n355n679) mutant animals grow and develop at a similar rate to wild-type animals, allowing us to compare side-by-side the lin-4 and lin-14 dynamics in wildtype and lin-14(n355n679) mutant animals in which lin-14 is relieved from repression by lin-4. More importantly, the misexpression of LIN-14 protein is decoupled from the misspecification of L1 stage cell fates, so that molecular phenotypes can be interpreted without the complication of changes in developmental fate phenotypes.
To quantitatively assay the full-length polyadenylated lin-14 mRNA, we used oligo(dT) 20 primer to reverse-transcribe mRNA isolated from synchronized animals and performed quantitative PCR (qPCR) analysis. Relative lin-14 levels were obtained by normalization to rpl-32, a large ribosomal subunit that is abundant and stably expressed throughout development. The results are shown relative to the lin-14 mRNA level at release from L1 diapause, or 0 hours of larval development. In wild-type animals, lin-14 mRNA decreased to 49% and then to 38% of the 0 hour sample at 9 hours and 12 hours of larval development, respectively ( Fig. 2A). From 12 hours to 24 hours of larval development, lin-14 mRNA levels fluctuated with little evidence of further decline comparing the 24 hour to the 12 hour time point (p = 0.23, one-tailed t-test) ( Fig. 2A). On the other hand, lin-14 mRNA levels in the lin-14(n355n679) mutants fluctuated modestly, and showed little evidence of the monotonic reduction throughout the L1 stage as seen for wild type (Fig. 2C).
We monitored LIN-14 protein levels in the same samples that were analyzed for lin-14 mRNA abundance. LIN-14 protein levels were analyzed by Odyssey CLx Infrared Fluorescent Western Blot and quantified by normalization to actin. The results are shown relative to the LIN-14 protein level at release from L1 diapause, or 0 hours of larval development. In wild-type animals, LIN-14 protein levels were stable during the first 9 hours of larval development ( Fig. 2A and 2B). Subsequently, they declined to 62% and further to 18% at 12 hours and 15 hours ( Fig. 2A and  2B), which followed the decline of lin-14 mRNA levels that took place from 6-12 hours (Fig. 2A). Since LIN-14 protein levels declineed by ,5 fold while lin-14 mRNA levels declined by ,2.5 fold, this suggests that both mRNA decay and inhibition of protein translation contribute to silencing of lin-14 at early stages. From 15 hours to 24 hours of larval development, LIN-14 protein levels decreased from 18% to 8% of the 0 hour levels ( Fig. 2A and 2B). In contrast, LIN-14 protein levels in the lin-14(n355n679) mutants did not show any significant down-regulation throughout the first larval stage (Fig. 2C and 2D), suggesting the ,10-fold downregulation of LIN-14 protein in wild-type animals is indeed mediated by lin-4 miRNA binding to lin-14 39UTR.
To track the mature lin-4 miRNA levels, we performed Taqman miRNA assays in the same samples as above. In wild-type animals, lin-4 miRNA is present at very low level in embryos, and can be detected by deep-sequencing [19] but not by Northern Blot, until 9-12 hours of larval development at 20uC [20,21]. Consistent with previous studies, our analyses showed that lin-4 miRNA levels began to rise substantially at 9 hours, and its level became ,5000 fold higher at 24 hours compared to 0 hours. Curiously, the initial decline of lin-14 mRNA levels between 6-12 hours happened prior to the significant accumulation of lin-4 miRNA. When lin-4 miRNA became abundant at later stages, lin-14 mRNA levels fluctuated yet showed little evidence of significant down-regulation; however, LIN-14 protein levels continued to decrease. Together, our analyses point to two phases of regulation: a fast lin-14 mRNA destabilization as soon as lin-4 miRNA emerges, and long-term translational inhibition that initiates and is particularly important for maintaining the silencing of lin-14 mRNA by the lin-4 miRNA.
Strikingly, we found that lin-4 miRNA levels in the lin-14(n355n679) mutants were ,50-200-fold lower compared to stage-matched wild-type animals. We note that the mutant LIN-14 V299D protein encoded by the lin-14(n355n679) locus has reduced function. Thus the failure to up-regulate the lin-4 miRNA in this lin-14 mutant suggests that LIN-14 gene activity may normally act upstream of lin-4 miRNA expression, which then feeds back on production of LIN-14 protein. The defect in LIN-14 protein activity in a lin-14(n355n679) mutant may break this autoregulatory loop. Alternatively, in C. elegans, miRNAs are protected from degradation by their target mRNAs [22,23]. Therefore, it is possible that lin-4 miRNA loses this protection in lin-14(n355n679) mutants due to the absence of lin-4 binding sites in the lin-14 39UTR. This would suggest that in contrast to models of hundreds of mRNA targets of miRNAs, in the case of lin-4, there may be just the lin-14 mRNA target that is key for accumulation of the lin-4 miRNA. The lin-14 mRNA levels of wild-type and a mutant bearing a lin-14 39UTR deletion are equal both at early and at late stages of animal development Another approach to assay whether lin-4 repression acts via mRNA destabilization or translational inhibition is to compare the levels of wild-type lin-14 mRNA to mutant lin-14 mRNA missing lin-4 complementary sites, in a heterozygous animal. To this end, we assayed lin-14 mRNA levels in the lin-14(n536n540)/szT1 heterozygous strain with one wild-type and one n536n540 allele of lin-14. lin-14(n536) bears a 607-bp deletion that removes five of the seven lin-4-complementary elements in the lin-14 39UTR (Fig. 1) and is a gain-of-function mutation causing retarded heterochronic development due to derepression of lin-14 [14,17]. The n540 allele, an amber mutation at position 280 (Lys to amber), was isolated as a recessive suppressor of the lin-14(n536gf) mutant [16,24]. We reasoned that if lin-4 miRNA silences lin-14 by enhancing mRNA decay, then lin-14(n536n540) mRNA should be stabilized relative to the wild-type lin-14 mRNA due to reduction of lin-4-mediated regulation. However, Northern blot of RNA isolated from lin-14(n536n540)/szT1 heterozygotes showed that the 3.5-kb wild-type lin-14 and 2.9-kb lin-14(n536n540) mRNA levels are similar at both the embryonic stage, before lin-4 miRNA is expressed, and at the L4 stage, after lin-4 repression has occurred (Fig. 3). This experiment indicates that there is very modest down-regulation of lin-14 mRNA between the pre-lin-4 expression stage and the L4 stage from either the wild-type allele or the lin-14 allele missing the majority of the lin-4-complementary regions. However, the two residual lin-4 complementary sites in the lin-14(n536) 39UTR are insufficient to mediate the silencing effect by lin-4 through translational repression, because the n536 mutation derepresses expression of the LIN-14 protein at the L2 and later stages. Our result is also largely consistent with a recent study [25] and further stresses the importance of translational inhibition in lin-4 miRNA-mediated silencing of lin-14.

Discussion
The results we present here consolidate the earlier studies where lin-14 was found to be silenced by lin-4 without being significantly destabilized at the mRNA level [4,14], as well as a recent ribosome profiling study showing translational control of lin-14 by lin-4 [25]. Although cases have been reported where regulation by miRNAs can cause destabilization of target mRNAs in C. elegans [26], the relative contributions of mRNA degradation and translational repression in miRNA-mediated repression are variable [5]. For example, lin-41 mRNA levels regulated by the let-7 miRNA showed the strongest reduction, whereas lin-14 mRNA showed modest reduction [26]. However, the lin-41 mRNA degradation noted by Bagga et al. [26] was not observed by Stadler et al., who instead suggested translational control of lin-41 by the let-7 miRNA [25].
Several recent studies have suggested a two-phase model for miRNA-mediated silencing that begins with translational repression, followed by mRNA deadenylation and decay, which consolidates the silencing [2,8,9,10,11,12]. Interestingly, we observed distinct dynamics for lin-4 silencing of lin-14. First, lin-14 mRNA level declines to ,40% as soon as lin-4 miRNA is first expressed during 6-9 hours of L1 development, prior to the decrease of LIN-14 protein level. Although translational inhibition may also contribute to this initial silencing of lin-14, currently this hypothesis is not supported by solid evidence. On the other hand, between 12-24 hours of larval development, lin-14 mRNA levels fluctuate with no obvious further decline, whereas LIN-14 protein levels continue to decrease. Furthermore, by the early L2 stage, the LIN-14 protein level has declined by more than 10 fold and the lin-14 mRNA level has declined by ,3 fold, suggesting that long-term translational inhibition maintains silencing of lin-14 mRNA by the lin-4 miRNA.
It remains unclear why the modes of action of miRNAmediated silencing are different depending on the biological and experimental context. In particular, it is intriguing why and how the predominant mechanism of silencing could change even for the same miRNA and its target mRNA in the same organism. Since the levels of lin-4 miRNA increase more than 5000 fold from the early L1 stage, when lin-14 mRNA just begins to decay, to the early L2 stage, when lin-14 has been stably silenced, we hypothesize that the number of lin-4 molecules binding to the lin-14 mRNA 39UTR might determine the mode of silencing. Specifically, it is possible that a single or a few lin-4 miRNA molecules binding to lin-14 39UTR can cause mRNA decay, whereas more miRNAs binding causes translational control. Because there are seven lin-4 complementary sites in the lin-14 39UTR, this would trigger a fast decline in LIN-14 protein synthesis even when the level of lin-4 is low, to achieve a sharp transition in development. To maintain silencing, translation of lin-14 mRNA is inhibited probably through binding of multiple lin-4 molecules to the lin-14 mRNA 39UTR as lin-4 massively accumulates. Supporting this hypothesis, we found that the lin-14(n536n540) mRNA missing five out of the seven lin-4complementary elements in its 39UTR is destabilized to a similar level as the wild-type lin-14 mRNA. However, lin-14(n536) is not properly silenced and causes a retarded heterochronic phenotype [14,17]. This suggests that the two remaining lin-4 complementary sites might be sufficient to cause the normal lin-4triggered decay of lin-14 mRNA, but insufficient to induce lin-4-triggered translational repression.
Finally, our fine-stage analysis showed that lin-14 mRNA drops initially during the first larval intermolt stage, but returns to a higher level during the L1-L2 molt. It is an emerging trend that mRNA levels could oscillate in animals entering and exiting the molt. Therefore, an important lesson is that looking at a single time point could be misleading, which may contribute to the differences in results and conclusions obtained from different labs. Figure 3. The levels of lin-14 mRNA derived from the wild-type and the lin-14(n536n540) mutant allele bearing a 39UTR deletion are similar. Shown are Northern blots of embryos and L4 staged lin-14(n536n540)/szT1 heterozygotes, using probes from the lin-14 (top), histone (middle) and vitellogenin genes (bottom). The levels of 3.5-kb wild-type lin-14 and 2.9-kb lin-14(n536n540) mRNA are almost equal at both the embryonic and L4 stages. Histone mRNA is used as a control for even loading. Vitellogenin mRNA, encoding yolk polypeptides, which is most abundant during oogenesis, is shown to indicate animal stages. doi:10.1371/journal.pone.0075475.g003