lin-28 Controls the Succession of Cell Fate Choices via Two Distinct Activities

lin-28 is a conserved regulator of cell fate succession in animals. In Caenorhabditis elegans, it is a component of the heterochronic gene pathway that governs larval developmental timing, while its vertebrate homologs promote pluripotency and control differentiation in diverse tissues. The RNA binding protein encoded by lin-28 can directly inhibit let-7 microRNA processing by a novel mechanism that is conserved from worms to humans. We found that C. elegans LIN-28 protein can interact with four distinct let-7 family pre-microRNAs, but in vivo inhibits the premature accumulation of only let-7. Surprisingly, however, lin-28 does not require let-7 or its relatives for its characteristic promotion of second larval stage cell fates. In other words, we find that the premature accumulation of mature let-7 does not account for lin-28's precocious phenotype. To explain let-7's role in lin-28 activity, we provide evidence that lin-28 acts in two steps: first, the let-7–independent positive regulation of hbl-1 through its 3′UTR to control L2 stage-specific cell fates; and second, a let-7–dependent step that controls subsequent fates via repression of lin-41. Our evidence also indicates that let-7 functions one stage earlier in C. elegans development than previously thought. Importantly, lin-28's two-step mechanism resembles that of the heterochronic gene lin-14, and the overlap of their activities suggests a clockwork mechanism for developmental timing. Furthermore, this model explains the previous observation that mammalian Lin28 has two genetically separable activities. Thus, lin-28's two-step mechanism may be an essential feature of its evolutionarily conserved role in cell fate succession.


Introduction
Tissue and organ formation in animals requires that diverse cell types arise in proper succession from a common pool of progenitors. Mutations in the heterochronic genes of the nematode Caenorhabditis elegans either skip or reiterate developmental events, indicating that they encode components of a cell fate succession mechanism. A lin-28 null mutant, for example, causes precocious development by skipping many second larval stage (L2) cell fates [1]. A let-7 null mutant causes retarded development by reiterating larval fates and delaying differentiation [2]. Lin-28 encodes one of twelve proteins and let-7 one of five microRNAs known to act in the heterochronic pathway [3][4][5]. The complex dynamics of activation of the microRNAs and repression of particular proteins specifies stage-appropriate behavior in progressively differentiating lineages. Genetic and molecular analyses have revealed further complexity in the form of feedback loops, oscillating regulators, and microRNA redundancy [4,[6][7][8][9][10]. Still, our knowledge of their relationships remains inadequate to explain how many of these components contribute to the cell fate succession mechanism.
While investigating the mechanism by which accumulation of the mature let-7 microRNA is blocked in pluripotent cells, Viswanathan and colleagues discovered that mammalian LIN28 protein can bind the let-7 pre-microRNA and inhibit its processing [26]. The details of this mechanism have been elucidated and the phenomenon has been confirmed for the C. elegans ortholog [27][28][29][30][31][32][33]. Prior to this finding, the direct targets of LIN-28 protein in C. elegans were unknown. Mammalian LIN28 has been reported to act on mRNAs as well, but a specific regulatory mechanism has not yet been discovered [21,[34][35][36][37][38]. Its inhibition of let-7 microRNA processing is a novel form of gene regulation and offers a molecular explanation for how lin-28 controls cell fate succession in C. elegans.
Earlier studies of the C. elegans heterochronic pathway had not addressed the issue of whether lin-28 requires let-7 microRNAs for its function [2,29,39]. Like other animals, C. elegans possess multiple let-7 family members [40][41][42][43][44]. Significantly, Abbott and colleagues discovered that three let-7 relatives-miR-48, miR-84 and miR-241-function redundantly to repress the transcription factor gene hbl-1 and cause the succession of L2 to L3 cell fates [6]. Because lin-28's primary role is to govern this same cell fate transition, it is reasonable to hypothesize that it acts via one or more of these let-7 relatives. let-7 itself has been believed to act much later in the heterochronic pathway, at the L4-to-adult transition. However, another possibility is that let-7 acts earlier together with its relatives in a previously unrecognized role, which would explain lin-28's action upon it. Our results show, however, that lin-28 does not act via any of these let-7 family members in its primary role in C. elegans development. To explain this discrepancy, we provide evidence that lin-28 acts in two-steps to control successive cell fates in a manner like that of lin-14 [45]. We speculate that the pairwise and overlapping activities of lin-14 and lin-28 reveal a ''clockwork'' logic underlying the pathway. The significance of our findings is that they explain two activities observed of mammalian Lin28 and thus may reveal an essential feature of lin-28's evolutionarily conserved role as a regulator of cell fate succession in animals.

lin-28 Represses the Accumulation of let-7 in the L1 and L2
The binding of mammalian LIN-28 to pre-let-7 leads to the degradation of the precursor and eventual loss of mature let-7 [27][28][29][30][31][32]. To determine whether C. elegans lin-28 prevents the developmental accumulation of the let-7 family microRNAs, quantitative RT-PCR assays were performed on wildtype and lin-28 mutant larvae. Relatively few worms (,200) are required to perform this assay, allowing precise staging of worms at the lethargus period prior to each larval molt.
As previously reported [2,6,48,49], mature let-7 was very low or undetectable in wildtype larvae at the L1 and L2 molts, accumulated during the L3 stage, and reached its peak by L4 ( Figure 1A, grey bars). The miR-48, -84, and -241 levels were all relatively low but detectable at the L1 molt and peaked by the L2 molt ( Figure 1B-1D, grey bars). The absence of lin-28 caused substantial premature accumulation of let-7 in both the L1 and L2 stages, higher than its peak at the L4 molt in wild type ( Figure 1A, blue bars). The removal of lin-28 caused no change in the levels of mature miR-48 and -241 in the early stages ( Figure 1C and 1D, blue bars). Only miR-84 showed a significant difference between wild type and the lin-28 mutant at the L2 molt ( Figure 1B, blue bars), as has been reported by others [29]. These findings suggest that lin-28 does not alter the accumulation of miR-48, miR-84, and miR-241 to the extent that it affects let-7, despite its ability to interact with them in the yeast three-hybrid assay. Importantly, only let-7 levels were altered at the L1 lethargus, the period immediately preceding the seam cell divisions of the L2.

lin-28 Acts Independently of let-7 MicroRNAs to Control Cell Fates
To test whether let-7 family microRNAs are required for lin-28's developmental activity, we examined mutants lacking both lin-28 and let-7 family members. The lateral hypodermal seam cells normally divide at each larval stage and differentiate as the animal becomes adult. lin-28 null mutants have fewer seam cells than wild

Author Summary
As tissues form, different cell types are generated from a common pool of undifferentiated cells. The mechanisms that control this developmental timing are largely unknown. In the nematode Caenorhabditis elegans, the heterochronic genes control a succession of cell fates in progressively differentiating tissues of the larva. Two of these genes, lin-28 and let-7, are evolutionarily conserved in animals where they have roles in pluripotency and differentiation. The LIN-28 protein is known to bind to and block the maturation of the small RNA encoded by let-7. This mechanism would seem to explain lin-28's role in development. Here we show that lin-28's primary activity in C. elegans-the proper timing of second larval stage cell fates-does not require let-7 or related genes. In explaining this discrepancy, we provide evidence that lin-28 has two distinct activities controlling successive cell fates. This situation is remarkably like that of lin-14, which acts one stage earlier. The overlap of their activities by one stage may reflect a fundamental feature of this cell fate succession mechanism. Furthermore, the two-step mechanism explains observations that mammalian Lin28 also has genetically separable activities. Therefore, lin-28's two successive activities may be essential to its evolutionarily conserved role in developmental timing.
type because they skip the one symmetric division in the seam lineage during the L2, and these cells differentiate at least one stage early, synthesizing adult cuticle alae precociously ( Table 2, lines 1 and 2) [1]. let-7 null mutants show retarded adult alae synthesis, but produced the normal number of seam cells ( Table 2, line 3) [2]. We observed that lin-28; let-7 animals had the reduced seam cell number characteristic of lin-28 mutants ( Table 2, lines 2 and 4), but as reported previously did not display precocious adult alae [2]. Thus, the let-7 null allele is epistatic to the lin-28 null allele only for the alae phenotype, not for the early seam cell division defect; the animals display both precocious and retarded characters.
The three let-7 family members mir-48, mir-84, and mir-241 act redundantly to control seam cell fates: when they are deleted together, the L2-specific symmetric cell division is reiterated, resulting in supernumerary seam cells [6]. In addition, in these triple-mutant animals, seam cell differentiation fails and they form no adult alae. A lin-28 null mutation is entirely epistatic to this retarded phenotype, having a reduced seam cell number and precocious adult alae ( Table 2, lines 5 and 6) [6]. Given that mir-48, mir-84, and mir-241 act redundantly and are related in sequence to let-7, we first wished to test whether let-7 might also be redundant with them in controlling L2 seam cell behavior. We constructed a strain lacking all four genes and assessed its seam cell phenotypes: we observed that animals lacking all four let-7 family members had the same seam cell number as those lacking only three ( Table 2, lines 5 and 7). Surprisingly, a strain lacking lin-28 and all four let-7 genes had the reduced seam cell number of a lin-28 mutant ( Table 2, line 8). Thus, lin-28 requires none of these let-7 family members to control the L2 seam cell fates. However, this strain did not make precocious adult alae ( Table 2, line 8), indicating that let-7 is required by lin-28 after the L2.

Lack of Evidence for Additional MicroRNAs Mediating lin-28 Activity
We surmised that lin-28 might act on a microRNA unrelated to let-7 to control L2 events. To test this idea we constructed strains defective in a gene needed for general microRNA function: ain-1 [50]. Removing ain-1 alone causes a slight increase in seam cell number from wild type ( Table 2, line 9), as previously reported [50]. In contrast to removing let-7, which had no effect, removing ain-1 from a strain lacking mir-48, mir-84, and mir-241 nearly doubled its seam cell nuclei number (Table 2, line 10). This increase reflects a reiteration of the L2 seam cell fate, and indeed indicates additional microRNA regulation of the L2 seam cell fate. However, removing ain-1 in a strain lacking lin-28 and the three let-7 family members did not result in an increase in seam cell number ( Table 2, line 11). This result is consistent with previous studies showing a lin-28 mutation is epistatic to ain-1 and ain-2 mutants in seam cell development [50,51]. The ain-1 mutation did substantially suppress the precocious adult alae phenotype of a lin-28 mutant, as if let-7 was fully active, demonstrating that the ain-1 mutation was able to reduce although not eliminate microRNA function in seam cell development ( Table 2, line 11).
To further test the idea that lin-28 inhibits accumulation of another microRNA, we performed a microarray analysis comparing wild type and lin-28; lin-46 double mutant animals staged during the L1 lethargus period (GEO accession: GSE35634). These double mutants develop like wild type [10], thus reducing the potential for indirect effects on microRNA abundance. We chose the L1 molt period because the first observable defect in lin-28(null) occurs shortly afterward. We observed that let-7 was upregulated 42-fold in the absence of lin-28, and that no other microRNA was affected more than 1.5-fold (Table S3). Therefore, because lin-28 regulates no other microRNA in the same manner it regulates let-7, we conclude that it possesses a different molecular activity to control L2 cell fates.

lin-28 Positively Regulates hbl-1 Expression through Its 39 UTR
hbl-1 is believed to be the most direct regulator of L2 hypodermal fates [6,52,53]. We addressed whether lin-28 affects hbl-1 expression using a hbl-1::GFP::hbl-1 39UTR reporter [54]. As previously observed, the reporter was high in hypodermal nuclei in the L1, down-regulated through the L2 and L3, and undetectable by the L4 stage ( Figure 2A, Table S4) [52][53][54]. Also as seen previously [6], in a strain lacking mir-48, mir-84, and mir-241, the reporter was constitutively expressed from L1 to L4 ( Figure 2B, Table S4). We observed that when lin-28 was also mutant, the reporter was rapidly down-regulated after the L1, earlier than it was in wild type, becoming undetectable by the L4, despite the absence of the three microRNAs ( Figure 2C, Table S4). This observation indicates that lin-28 is a positive regulator of hbl-1 expression that acts independently of the let-7 relatives. Similar results were obtained with animals lacking all four let-7 family members ( Figure S2). When the analysis was performed with a companion reporter that substitutes the hbl-1 39UTR with the unrelated unc-54 39UTR, the reporter was continuously expressed despite the absence of lin-28 ( Figure 2D). This observation indicates that lin-28 acts via the 39UTR of hbl-1, possibly directly, to temporally support hbl-1 expression and thereby promote L2 cell fates.

lin-28 Has Two Separable Activities
We were surprised that despite the evolutionary conservation of lin-28's ability to block let-7 accumulation, this activity is not required for its primary effect on C. elegans larval development, namely the normal execution of L2 cell fates. Previously, lin-28 was thought to specify L2 fates only, but the possibility that it has two activities was raised by these findings. In other words, to explain the relevance of let-7 to lin-28 function, we hypothesized that lin-28 acts in two mechanistically independent steps: first to control early fates and second to control later fates via direct action on pre-let-7.
Ambros and Horvitz documented that some seam cell lineages in lin-28 null mutants display precocious development that skips two larval stages [1,55]. In quantifying this phenotype, we found that in lin-28 null mutants 37% of seam cells differentiated at the L2 molt, two stages early (Table 3; Figure 3). Because lin-28 null mutants execute normal L1 cell lineages throughout the animal [1], we concluded these lineages skipped the L2 stage and one subsequent stage ( Figure 3). The other 63% of seam cells in these animals skipped only the L2 stage (Table 2 and Table 3; Figure 3). Although all animals contained both one-stage and two-stage precocious lineages, why some lineages skipped only the L2 fates, while others skipped two stages, is not clear.
We addressed whether any aspect of lin-28's two-stage precocious phenotype depended on let-7 family members. Comparable to lin-28 null mutants alone, 21% of the seam cells in animals that also lack mir-48, mir-84, and mir-241 displayed adult alae at the L2 molt (Table 3). By contrast, none of the lin-28; let-7 animals displayed adult alae at the L2 molt (Table 3). These observations indicate that let-7, and not its three relatives, is needed for the two-stage precocious phenotype of lin-28 null mutants.
To further address whether lin-28 possesses two genetically separable activities, we performed RNAi using bacteria not induced with IPTG (lin-28(lowRNAi)), which we expected to produce a range of weaker precocious phenotypes. Many animals displayed the same precocious phenotype observed commonly in lin-28 null mutants ( Figure 3). However, in 10% of the animals that had skipped L2 cell fates, all seam cell lineages terminally All animals examined were homozygous for null alleles of the genes indicated and carry an integrated transgene wIs78(scm::GFP; ajm-1::GFP) to mark seam cells. All alleles are null. 2 Seam cell counts were performed on L4 animals except where indicated. 3 Alae formation was assessed in the early L4 stage. 4 Strains carrying the let-7 mutation additionally contained a linked unc-3 mutant allele. They were grown at 15uC to limit constitutive dauer formation that results from the unc-3 mutation at higher temperatures in these backgrounds. 5 Seam cell fusion with no alae formation was observed in the other 85% of animals. SEM, standard error of the mean; ND, not determined. doi:10.1371/journal.pgen.1002588.t002 differentiated at the normal time ( Figure 3). We interpret these seam cell lineages as having executed L3 fates precociously as well as L3 fates at the normal time. These abnormal lineages demonstrate that a precocious phenotype early does not necessitate a precocious phenotype later, suggesting the two are separately regulated by lin-28.
In characterizing the interactions between LIN-28 protein and let-7 precursor sequences, we observed that LIN-28 could interact with the loop portion of the C. elegans pre-let-7 but not with that of Drosophila pre-let-7 (Table S2). Thus we could construct a version of let-7 that encoded the loop sequence of Drosophila pre-let-7 and thereby was insensitive to LIN-28's inhibitory activity. We generated animals carrying either a wildtype let-7 genomic transgene or a chimeric worm/fly transgene. We found that at a given concentration of DNA injected, 22% of F1 animals with the wildtype construct displayed precocious adult alae (n = 50), whereas 46% of F1 animals with the chimeric construct displayed precocious alae (n = 50). Animals receiving either transgene had an average of 16 seam cells at the L4 stage, indicating no change in the early cell fate decision (wildtype let-7, n = 47; chimeric let-7, n = 51). We established stable lines carrying each construct and found that those with the chimeric pre-let-7 expressed higher mature let-7 in early larval development than those with the wildtype pre-let-7 (Table S5). Therefore, the inhibition of mature let-7 accumulation is likely the means by which lin-28 governs seam cell development after the L2. let-7 Controls L4 Development let-7 is thought to act during the L4 stage to cause the L4-toadult transition, including the terminal differentiation of seam cells [2]. We and others have observed that let-7 accumulates in the L3 stage in wild type, a stage earlier than originally reported ( Figure 1) [2,6,48,49]. Therefore, one possibility is that let-7 mutants reiterate L3 developmental events in the L4 stage. We therefore reconsidered when let-7 has its earliest role in larval development. We examined let-7 null mutant animals in the L4 stage to see whether any defects had already occurred by this time. A confounding issue in this analysis is that the hermaphrodite seam cell lineages display exactly the same division patterns in L3 and L4 stages, so that reiteration of L3 or L4 fates cannot not be distinguished (see Figure 3). One seam cell lineage that is different in this regard is the male V5 lineage [56]. We observed a cell division in the V5 lineage that normally occurs during the L3 lethargus to be reiterated at the end of the L4 stage: 100% of animals showed a V5 lineage division in let-7 males recurring 12-13 hours after the L3 molt, in the late L4 (n = 10). Another consistent defect observed in let-7 null males was a delay in tail tip retraction that normally occurs in male tail morphogenesis during the L4 (Figure 4) [57]. All males examined displayed a marked failure of tip retraction by the mid-L4 stage (n = 10). These observations indicate that the earliest observable consequence of let-7 activity occurs long before the L4-to-adult transition, and suggest let-7 acts at the late L3 stage.

The Relative Roles of hbl-1 and lin-41
The let-7 family microRNAs have two known targets in the heterochronic pathway: hbl-1 and lin-41. We observed that lin-28 positively regulates expression of hbl-1, a regulator of L2 seam cell fates ( Figure 2) [6,52], whereas lin-41 is thought to act later to regulate the L4-to-adult transition [39]. We sought to clarify the roles of these two genes with respect to let-7 activity. In a wildtype background, reduction of hbl-1 by RNAi caused 80% of animals to display precocious adult alae formation, and reduction of lin-41 by RNAi caused 35% to have precocious alae (Table 4). In a let-7 null mutant background, seam cells divide at the L4 molt and synthesize adult alae one stage later [2]. We observed that the two let-7 target genes differed in their abilities to suppress this phenotype: penetrance of let-7's retarded defect was reduced from 100% to 80% by hbl-1(RNAi), whereas it was reduced to 6% by lin-41(RNAi) ( Table 4). These observations suggest that let-7 acts primarily through lin-41 to regulate seam cell differentiation. hbl-1 has been shown to be the primary target of let-7's relatives mir-48, mir-84; and mir-241 [6]. How the microRNAs belonging to the same family act selectively on different targets is currently unknown. Table 3. lin-28 mutants can be two stages precocious due to let-7 activity.   -7s), or animals carrying a transgene constitutively expressing lin-28 (lin-28(gf)) [62]. let-7 null mutants, whose defect in these lineages is first visible in the late L4 stage. Two types of seam cell lineages observed in lin-28 null mutants [1]. Seam cell lineages that skip L2 fates in lin-28(low RNAi) animals (see text Discussion lin-28 and let-7 had been thought to act at widely separated times in C. elegans larval development, with lin-28 controlling an early, proliferative fate of seam cells and let-7 controlling their terminal differentiation two larval stages later [3,58]. The serendipitous discovery that mammalian LIN28 binds to and inhibits let-7 precursor processing [26], and the subsequent proof that this mechanism is evolutionarily conserved in C. elegans [29,31], caused us to consider what their molecular interaction means for the regulation of cell fate succession in C. elegans. The progressively differentiating lateral hypodermal seam cells of C. elegans are often used to model cell fate succession in the analysis of heterochronic genes. These cells adopt three types of stage-appropriate fates: an asymmetric division producing one blast and one differentiated cell; a double division characteristic of the L2 stage producing two blasts and two differentiated cells; and terminal differentiation in which all cells fuse and secrete adult cuticular alae (Figure 3) [56]. Based on their null allele phenotypes, lin-28 controls the characteristic L2 proliferative division and let-7 controls the terminal differentiation. Given the redundancy of the three let-7 paralogs mir-48, mir-84, and mir-241 in regulating L2 fates, two alternatives seem likely: either lin-28 inhibits the accumulation of multiple let-7 family members, including these three let-7s known to control the L2-to-L3 transition, or let-7 is at least partially redundant with its relatives in controlling this early fate transition.
Surprisingly, we find that neither of these situations is the case. We demonstrate by using null alleles that lin-28 does not require let-7, mir-48, mir-84, and mir-241 for its control of L2 cell fates ( Table 2). It remains possible that other let-7 family members mediate lin-28's control of L2 fates, however, the LIN-28 protein interacts with none these (Table 1), and no microRNAs other than let-7 itself are dysregulated in a lin-28 null mutant (Table S3). Even in the absence of these microRNAs, we observe a marked positive effect of lin-28 on hbl-1 expression, supporting the model that lin-28 acts via hbl-1 to control the L2-to-L3 transition (Figure 2; Figure S2). Furthermore, this regulation depends on the hbl-1 39 UTR, suggesting a post-transcriptional mechanism. Our findings using the ain-1 mutant suggest additional microRNA activity controlling L2 cell fates, but are inconsistent with microRNAs mediating lin-28's role in the L2 (Table 2 and Table S3). We therefore conclude that lin-28 acts to oppose hbl-1's repression, but does so without changing microRNA abundance.
Given that the premature accumulation of mature let-7 does not account for lin-28's precocious phenotype, why then does LIN-28 inhibit let-7?
Because heterochronic genes act in succession, the actions of early-acting genes necessarily have consequences later in life. For example, the microRNA lin-4 represses the expression of lin-14, and when that repression fails, L1 cell fates are reiterated [59,60]. The fact that seam cell differentiation never occurs is not taken to mean that lin-4 directly controls that event. Rather, the reiteration of L1 fates-the direct consequence of loss of lin-4-leads to the permanent postponement of differentiation. Likewise, the precocious terminal differentiation of seam cells in a lin-28 mutant might simply be the consequence of skipping the L2 cell fates and everything else falling in line after that. In such a scenario, each factor has a single activity and an early defect leads to a cascade of wrong fate decisions directed by other factors. However, an alternate interpretation is possible. lin-14, another heterochronic gene which controls primarily the L1 cell fates, was shown to possess two separable and sequential activities [45]. These activities are termed lin-14a and lin-14b, although they do not correspond to distinct gene products [61]. lin-14a controls the L1to-L2 transition and lin-14b controls the L2-to-L3 transition. [45]. By analogy, lin-28 can be said to have two separable activities as well ( Figure 5). The first of lin-28's activities governs the L2-to-L3 transition and is independent of let-7 and the second acts via let-7 to control the L3-to-L4 transition. Thus, a parsimonious explanation for lin-28's inhibition of let-7 in C. elegans is that it constitutes the second of two activities. However, this view requires adjustments to existing models of the heterochronic pathway.
First, because LIN-28 protein is down-regulated by the L3, we must consider the time of let-7 expression. Early reports showed mature let-7 rising in the L4 stage, however as microRNA detection methods have improved, expression of mature let-7 could be seen a full stage earlier [6,49]. Our quantitative RT-PCR data indicate that mature let-7 accumulates during the L3 (Figure 1), after LIN-28 has disappeared [62].
Second, although it is impossible at present to distinguish between L3 seam cell fates and L4 seam cell fates, we must reconsider the time of let-7's activity. Because mature let-7 levels are very low at the L2 molt and nearly at their peak by the end of the L3, it is reasonable to assume that let-7 could act by the end of the L3. Thus loss of let-7 might actually cause the reiteration of L3 fates, the consequence of which would be problems in the L4. None of the previous data concerning let-7's role in seam cells decides whether it acts to control the L3-to-L4 transition or the L4-to-adult transition. However, we observed consistent abnormal cell division and morphogenesis events in the L4 male, which is in agreement with a reiteration of L3 cell fates in let-7 null mutants. Thus we propose that let-7 (and possibly other regulators believed to control the L4-to-adult transition such as lin-41) act earlier than previously thought.
Third, hbl-1 has been assigned to roles in both L2 seam cell fates and terminal differentiation [6,52,53]. Our comparison of the ability of hbl-1and lin-41-knockdown to suppress a let-7 null mutation reveals that lin-41 has a more significant role downstream of let-7. Therefore, we propose that hbl-1 is the most proximal regulator of L2 fates, being regulated by the three let-7 paralogs, and lin-41 is let-7's target for controlling later events ( Figure 5). Thus, it is LIN-28's direct action on pre-let-7 that exerts influence on those later events via lin-41.
We note that although lin-14 and lin-28 each act twice to govern successive cell fate decisions, their functions overlap by one stage, with the second lin-14 activity coinciding with the first of lin-28's ( Figure 5). We have previously proposed that the lin-14b activity is a consequence of a positive feedback loop between lin-14 and lin-28 [10]. Therefore, the second period of lin-14's action is tied to the first one for lin-28. We speculate that the pairwise and overlapping activities of these two factors reveal an underlying ''clockwork'' mechanism for cell fate succession. Each of these regulators has its first role in determining the fates expressed in a particular stage, then a second role that is linked to the next regulator in sequence. In the case of lin-14, it first determines what fates are expressed in the L1, then by positive feedback on lin-28, it governs what happens in the L2 [10,45]. Similarly, lin-28 first determines what events occur in the L2, then by its positive regulation of lin-41 via let-7, influences events of the L3. By each factor having both a cell fate determining role and a link to the next stage through the next factor in the pathway, the proper Table 4. Relative contribution of hbl-1 and lin-41 for the let-7 retarded phenotype. The let-7 mutants were identified by Unc phenotype due to the unc-3 mutation. 2 The precocious alae were assessed at the end of L3-L4 molt or in the early L4 stage of development. 3 As previously noted, hbl-1(RNAi) causes a proliferation defect in the late L4 which is not interpreted as heterochronic [53]. These divisions were not scored. ND, not determined. doi:10.1371/journal.pgen.1002588.t004 succession of cell fates is achieved. This overlap of regulators resembles, at least superficially, the ABC model for floral organ identity [63]. In each case, four developmental distinctions are specified: larval stage-specific cell fates in C. elegans and whorl organ identities in plants. Because in C. elegans the overlap is temporal rather than spatial, the cell fates progress sequentially as successive regulators are repressed in turn. We also note that for each lin-14 and lin-28, the earlier of its activities is more sensitive to reduction than the later activity ( Figure 3) [45], which may be important for the order in which the two activities occur. Most significantly, lin-28's two-stage action in C. elegans explains a split function observed of mammalian Lin28 in neural development [18]. Lin28 activity can promote neuronal differentiation and inhibit astroglial differentiation. These two activities were found to be genetically separable: a mutant form of Lin28 can block gliogenesis without affecting the number of neurons. Furthermore, changes in let-7 levels do not fully account for Lin28's activity in this system. By finding that C. elegans lin-28 has two distinct activities, we surmise that the split phenotype in mammalian neurogenesis is a consequence of a similar two-step mechanism involving let-7-dependent and let-7-independent activities. Considering the long evolutionary association of lin-28 and let-7 with cell fate succession in diverse contexts, we propose that having two sequential, mechanistically distinct activities is critical to lin-28's role in governing successive developmental transitions.

Microscopy and Phenotype Analysis
Nomarski DIC and fluorescence microscopy were used to count seam cell nuclei. Developmental stage was assessed by the extent of gonad and germ line development. In some cases where seam cell division was ongoing or just completed, the two daughter nuclei were counted as one. All images were taken with a 1006 objective on a Zeiss Axioplan2 imaging microscope equipped with a CCD camera. To analyze the V5 cell-lineage in let-7 mutant males, wIs78; him-5(e1467) males were crossed to wIs78; mnDp1(X:V)/ +;unc-3(e151) let-7(mn112) X hermaphrodites and Unc males among the cross progeny were examined for V5 seam cell divisions.

RNA Interference
Bacterially-mediated RNA-interference was performed as previously described [66]. The RNAi vectors contained a 3.5 kb region of hbl-1 genomic sequence or 740 bp of the lin-28 ORF. The I-4J11 bacterial strain from the Ahringer RNAi library that expresses lin-41 dsRNA was also used. dsRNA-expressing bacteria were induced in culture and seeded on NGM plates containing 1 mM IPTG, 50 mg/ml ampicillin and 12.5 mg/ml tetracycline. Empty vector was used as a negative control. RNAi for hbl-1 and lin-41 was done post-embryonically: gravid adults were dissected and embryos allowed to hatch on dsRNA expressing bacteria. For hbl-1 and lin-28 ''low'' RNAi, uninduced bacterial cultures were seeded on NGM plates without IPTG. Animals were propagated on lin-28(low RNAi) for analysis. L4 animals grown on hbl-1(low RNAi) were transferred to NGM plates seeded with normal food (AMA1004) for analysis.

Yeast Three-Hybrid Assay
Yeast three-hybrid assays were performed using the YBZ-1 strain as described previously [18,47]. The C. elegans lin-28 open reading frame was fused to the activation domain sequence in pACT2, and experimental RNAs were fused to the MS2 stem loop sequence in pIIIA/MS2-2. X-gal overlays were assessed after 6 hours and overnight. All RNAs that produced negative interactions were shown by RT-PCR to be expressed at a level comparable to those of RNAs that produced positive interactions. Sequences of selected RNAs tested in interaction assays are listed in Table S1.

RNA Extraction and qRT-PCR Assays
For RNA isolation, 50-200 animals in the pre-molt lethargus were collected in M9 buffer. RNA was isolated using mirVana miRNA isolation kit (Ambion) following the manufacturer's instructions with an additional sonication step performed immediately after the addition of lysis/binding buffer. The quality and concentration of the RNA were determined using a Nanodrop 1000 spectrophotometer (Thermo Scientific). The microRNA-qRT-PCR (TaqMan assay, Applied Biosystems) was performed using TaqMan probes for let-7, miR-48, miR-84, miR-241 and small nucleolar RNA sn2841 according to the manufacturer's instructions. Reverse transcriptase-free controls confirmed amplification was dependent on input RNA. Samples were analyzed on an Applied Biosystems StepOne machine. Relative changes in the microRNA levels were determined by the DDCt method using snoRNA sn2841 levels for normalization [67]. Gene copy number assessments were made using the SYBR Green assay (Applied Biosystems) and primers specific for ama-1 and let-7 on approximately 20 animals. Single amplicon SYBR Green products were confirmed by agarose gel electrophoresis. Dissociation/ melting curves were determined after each run. Samples were analyzed on an Applied Biosystems 7500 machine. Triplicate technical replicates were performed with each sample.