A let-7-to-miR-125 MicroRNA Switch Regulates Neuronal Integrity and Lifespan in Drosophila

Messenger RNAs (mRNAs) often contain binding sites for multiple, different microRNAs (miRNAs). However, the biological significance of this feature is unclear, since such co-targeting miRNAs could function coordinately, independently, or redundantly with one another. Here, we show that two co-transcribed Drosophila miRNAs, let-7 and miR-125, non-redundantly regulate a common target, the transcription factor Chronologically Inappropriate Morphogenesis (Chinmo). We first characterize novel adult phenotypes associated with loss of both let-7 and miR-125, which are derived from a common, polycistronic transcript that also encodes a third miRNA, miR-100. Consistent with the coordinate upregulation of all three miRNAs in aging flies, these phenotypes include brain degeneration and shortened lifespan. However, transgenic rescue analysis reveal separable roles for these miRNAs: adult miR-125 but not let-7 mutant phenotypes are associated with ectopic Chinmo expression in adult brains and are suppressed by chinmo reduction. In contrast, let-7 is predominantly responsible for regulating chinmo during nervous system formation. These results indicate that let-7 and miR-125 function during two distinct stages, development and adulthood, rather than acting at the same time. These different activities are facilitated by an increased rate of processing of let-7 during development and a lower rate of decay of the accumulated miR-125 in the adult nervous system. Thus, this work not only establishes a key role for the highly conserved miR-125 in aging. It also demonstrates that two co-transcribed miRNAs function independently during distinct stages to regulate a common target, raising the possibility that such biphasic control may be a general feature of clustered miRNAs.


Introduction
RNA-mediated post-transcriptional mechanisms regulate the accumulation and homeostasis of proteins not only during animal development but also during adulthood [1][2][3]. These mechanisms include regulation by microRNAs (miRNAs), a class of small non-coding RNAs that usually silence messenger RNAs (mRNAs) by binding to partially complementary sequences frequently found in the target 3' untranslated (3'UTR) sequence [4]. Some miRNAs are known to affect lifespan by post-transcriptionally silencing mRNAs that play critical, beneficial roles at early stages of the life cycle but are deleterious when expressed inappropriately at later stages [1,2,[5][6][7]. For example, loss of C. elegans lin-4, the first miRNA to be functionally characterized for its role in lifespan, leads to shortened lifespan due to the persistence of its target, lin-14 [2,8]. Similarly, the adult onset of Drosophila miR-34 promotes longevity and maintains neuronal homeostasis by repressing Eip74EF, a transcription factor required for progression through earlier life stages [1,3]. Although loss of other miRNAs like Drosophila miR-1000 lead to shortened lifespan [3], the complete repertoire of miRNAs that regulate aging processes remains uncharacterized [9].
Results let-7-Complex miRNAs modulate age-associated processes in the brain The let-7-Complex (let-7-C) locus in Drosophila encodes an evolutionarily conserved cluster of three co-transcribed miRNAs: miR-100, let-7 and miR-125, the orthologue of C. elegans lin-4 [33,34]. Although the levels of processed let-7 are known to increase with age in testes and ovaries [35,36], the relative expression levels of all three miRNAs have not been characterized in aging flies. To address this, we performed Northern blot and quantitative reverse transcription polymerase chain reaction (qRT-PCR) analyses of whole animals (Fig 1A and 1B, left). These analyses revealed an age-dependent increase in all three let-7-C miRNAs in both adult males and females, suggesting a role for this miRNA cluster in aging-related processes.
To characterize the role of this age-dependent increase in let-7-C miRNAs, we analyzed a let-7-C hypomorphic (let-7-C hyp ) strain in which let-7-C miRNAs were expressed during development [37] but not maintained during adulthood (Fig 1B, right). This hypomorphic strain was trans-heterozygous for two let-7-C null alleles but also harbored a single copy of a minimal let-7-C rescuing transgene that contained regulatory elements required for onset of pri-let-7-C (let-7-Cp 3.3kb ::cDNA) during development but lacked elements needed for its post-developmental maintenance. Consistent with our previous analysis [37], young let-7-C hyp mutant males expressed reduced levels of miR-100 (9.2±4.6% of control), let-7 (22±10.2% of control) and miR-125 (33±7.2% of control) that decreased further as the adults aged ( Fig 1B). We therefore performed survival analysis of these let-7-C hyp mutant males and found that they died prematurely relative to control males (Fig 1C, compare black and red curves; w 1118 : median survival 74d, maximum lifespan 98d; let-7-C hyp : median survival 36d, maximum lifespan 56d). Prompted by this reduced viability, we assayed the let-7-C hyp strain for additional functional and morphological age-dependent phenotypes. Young let-7-C hyp mutants climbed normally, indicating that the levels of let-7-C miRNAs they express during metamorphosis and early adulthood is sufficient for general adult function. However, aged let-7-C hyp mutants displayed a steep reduction in this ability ( Fig 1D). These results indicated that persistent expression of one or more of the three let-7-C miRNAs specifically during adulthood was required for normal adult healthspan.
Given the neural expression of let-7-C miRNAs [33,37,38], we next looked for age-associated deterioration in brain morphology. Brain degeneration has been anatomically characterized by an age-dependent increase in the number of scattered vacuoles that mark cells undergoing necrotic cell death [39]. Sections of 40-day old control and let-7-C hyp brains revealed a sharp increase in vacuole number in mutant brains (Fig 1E-1G). As with the climbing defect described above, this phenotype had an adult onset since the brains of young mutant flies contained hardly any vacuoles (0 vacuoles in w 1118 , 0.6 ± 0.9 vacuoles in let-7-C hyp , n = 5). Importantly, a let-7-C transgene that substantially restored miR-100 levels (59.4±11.1% of control), let-7 levels (50.3±24% of control) and miR-125 levels (108±28%) in 3-day old adults (let-7-C hyp rescue in Fig 1B) rescued the lifespan and age-dependent climbing defects as well as the brain deterioration of let-7-C hyp mutants (Fig 1C, 1D and 1G). Since our qRT-PCR analysis indicated that rescued let-7-C hyp mutants express a constant level of let-7-C miRNAs during adulthood (Fig 1B), we inferred that the age-dependent increase in let-7-C miRNAs detected in wildtype adults was not absolutely required for their pro-survival and neuroprotective roles. Taken together, these results confirmed a role for let-7-C miRNAs in the aging processes that occur in the brain.

miR-125 and let-7 mutants display reduced lifespan and neurodegeneration
In order to distinguish the roles of the three let-7-C miRNAs, we generated a set of rescuing transgenes with either miR-100, let-7 or miR-125 deleted. These transgenes were inserted into identical chromosomal locations using phiC31-mediated integration [40] and crossed into a trans-heterozygous let-7-C null background, yielding strains we referred to as ΔmiR-100, Δlet-7 and ΔmiR-125 single mutants, respectively (see S1 Fig for our crossing scheme that ensured Fig 1. let-7-C mutants are short-lived and display adult onset defects in motility and brain morphology. (A) Northern blots of total RNA from w 1118 (wt) males and females that were aged for 3 or 40 days and probed for let-7, miR-125, and miR-100. Blots were also probed with U6 as a loading control. (B) Quantitation of miR-100, let-7, and miR-125 in 3 or 40 day w 1118 (wt), let-7-C hyp rescue, and let-7-C hyp adult males (n = 3). Processed miRNA levels were normalized to 2S rRNA. (C) let-7-C mutant males displayed a reduced lifespan (w 1118 : median survival 74d, maximum lifespan 98d, n = 210 male flies; let-7-C null: median survival 12d, maximum lifespan 26d, n = 210 male flies; let-7-C hyp : median survival 36d, maximum lifespan 56d, n = 210 male flies; let-7-C hyp rescue: median survival 66d, maximum lifespan 110d, n = 210 male flies). P value of the lifespan curves were calculated by log-rank test. (D) Climbing activity of 3 or 30 day w 1118 , let-7-C hyp rescued and let-7-C hyp adult males. At 3 days, the climbing ability of let-7-C hyp (80 ± 5.8%) was comparable to the control w 1118 (81.33 ± 2.3%). However, at 30 days, only 10.7 ± 6.18% of let-7-C hyp males were able to climb as opposed to 68.72.5 ± 4.67% of w 1118 males. The age related climbing defects of let-7-C hyp were almost completely rescued by a single copy of the let-7-C rescue transgene (64.66 ± 3.69%). Mean ± S.D of three experiments, n = 15 male flies in each experiment, *** = p value <0.001. (E, F) Brain sections of 40d let-7-C hyp rescue (E) and let-7-C hyp mutant males (F) (arrows point at vacuoles). (G) Quantification of vacuole number. See S1 Table for  that single mutant strains were otherwise as close to identical as possible). Unlike previously generated strains with P-element rescue transgenes [33], differences between these single mutants could be attributed to loss of an individual miRNA rather than to position effects.
Quantitative RT-PCR analysis of miR-100, let-7 and miR-125 confirmed the absence of miRNA expression in each of the deletion lines (Fig 2A). However, this analysis also revealed cross-regulatory relationships between the three miRNAs: loss of let-7 resulted in reduced levels of both miR-100 (0.29 fold relative to control) and miR-125 (0.35 fold relative to control), while loss of miR-100 and miR-125 resulted in increased levels of let-7 (2.5 fold relative to control) and miR-100 (2.5 fold relative to control), respectively (Fig 2A). To assess the cause of these changes, we turned to a cell culture assay in which we could quantify the activity of each miRNA in cells transfected with altered let-7-C versions. MiRNA activity was quantified as the fold repression in luciferase levels produced by previously validated "sensors" for each let-7-C miRNA [38]. Individual sensors were co-transfected along with UAS-let-7-C cDNA constructs into Kc-167 cells that do not ordinarily express let-7-C miRNAs [37]. First, confirming the effect of let-7 deletion on miR-100 and miR-125 levels, we found that miR-100 and miR-125 activity reporters were less repressed in cells transfected with a let-7-C cDNA lacking the let-7 hairpin (miR-100: 5.6 ± 0.56 fold repression in Δlet-7 compared to 10.48 ± 1.7 in wild type; miR-125: 8.3 ± 0.66 fold in Δlet-7 compared to 13.8 ± 1.97 fold in wild type). Then, to test whether this effect was due to the absence of mature let-7 or some other cause (e.g. altered RNA conformation of the Δlet-7 primary transcript that reduced miR-100 and miR-125 processing), we generated a chimeric UAS let-7-C cDNA construct in which the Drosophila let-7 hairpin was replaced with the human let-7-a2 hairpin that encoded the same mature let-7 but has a different hairpin structure. While the human let-7-a2 hairpin restored the let-7 mediated repression of its sensor, it did not restore miR-100 and miR-125 mediated repression (Fig 2B,  construct 5). These data indicated that processed let-7 miRNA did not directly regulate the processing of miR-100 or miR-125. Instead, we favor a model where the rate of let-7 processing has an effect on the rate of miR-100 and miR-125 processing, a model consistent with processing of other polycistronic microRNAs [41]. We note that Truscott et al. also recently found evidence for cross-regulatory interaction between let-7-C miRNAs [41], although their results were slightly different-deletion of let-7 and miR-100 but not let-7 alone reduced miR-125 levelsprobably due to technical differences in the constructs used. We also evaluated ΔmiR-100 or ΔmiR-125 let-7-C cDNA constructs in this cell culture assay, but detected no enhancement in miRNA activity (Fig 2B), suggesting that the changes in miRNA levels detected in tissue ( Fig  2A) may not be functionally significant. Taken together, these results indicated that the set of ΔmiR-100, Δlet-7 and ΔmiR-125 strains described above would allow the dissection of the individual contributions of the three miRNAs since neighboring miRNAs continued to be expressed when individual miRNAs were deleted, albeit at altered levels in some cases.
We then used the ΔmiR-100, Δlet-7 and ΔmiR-125 single mutant lines to analyze the consequences of deleting each miRNA on age-associated brain degeneration and behavioral defects. Δlet-7 and ΔmiR-125 single mutant flies displayed significantly reduced longevity compared to control or ΔmiR-100 flies (Fig 2C). In addition, while young Δlet-7 and ΔmiR-125 mutants had normal climbing behavior and brain morphology, a significant decrease in climbing ability as well as a marked increase in vacuole number was observed in both mutants with age ( Fig 2E  and 2F). The vacuoles in both Δlet-7 and ΔmiR-125 mutants appeared to be scattered throughout the central brain region and some enrichment was also seen in the retina (S2 Fig). These data indicated that loss of either let-7 or miR-125 but not miR-100 caused behavioral and morphological changes that were normally seen in much older flies and were indicative of rapid aging of the brain. Given that loss of let-7 and miR-125 triggered physiological processes involved in aging, we tested whether inhibition of individual let-7-C miRNAs enhanced the neurodegeneration of a disease model of fragile X-associated tremor/ataxia syndrome (FXTAS) [42]. FXTAS is a late onset human neurodegenerative disease that is characterized by the presence of ubiquitin positive nuclear inclusions containing RNAs with expanded CGG repeats (rCGG) in neurons and astrocytes [43]. Ectopic expression of transcripts with artificial expansion of these repeats in the fly retina causes a pathology similar to human FXTAS, including photoreceptor degeneration and disorganization of the ommatidia [42]. Using miRNA "sponge" constructs designed to individually inhibit miR-100, let-7, or miR-125 (miR-100SP, let-7SP, or miR-125SP), we tested whether loss of any of these miRNAs' activities enhanced the retinal degeneration in the FXTAS model. We found that driving let-7SP or miR-125SP but not miR-100SP specifically in the eye throughout development and adulthood resulted in significant enhancement of the rCGG phenotype (S3 Fig). This result indicated that, in addition to their role in modulating lifespan, let-7 and miR-125 promoted disease pathogenesis while miR-100 did not.
To pinpoint the specific stage during which let-7 and miR-125 activity were involved in FXTAS disease pathogenesis, we utilized a temperature sensitive allele of Gal80 (tubP-Gal80 ts ). This approach allowed temporal control of both the UAS-rCGG 90 transgene as well as the UAS-miRNA sponges in the eye, since animals at 29°C express UAS transgenes but animals at 18°C do not [44]. We reared strains to control expression in three ways: no expression (18!18), constant expression (29!29), or expression during development but not adulthood (29!18) (Fig 3A-3L). As expected, constant expression of either let-7SP or miR-125SP enhanced the rCGG 90 phenotype whereas no expression did not (Fig 3A-3F). However, let-7SP and miR-125SP behaved differently from one another in the 29!18 regimen: let-7SP enhanced rCGG 90 retinal degeneration while miR-125SP did not (compare Fig 3G-3I). This result, along with the observation that let-7SP animals reared at 29!18 looked no worse than those reared at 29!29, suggested that let-7's main contribution to disease progression occurred during development. Conversely, miR-125SP animals reared at 29!18 looked no worse than those reared at 18!18, suggesting that the phenotypes displayed by those reared at 29!29 was a specific consequence of adult miR-125SP expression. The reciprocal experiment involving adult-only transgene expression was not informative because none of the 18!29 animals displayed a phenotype, even when aged up to 20 days, perhaps because the underlying rCGG 90 phenotype was at least partially of developmental origin. These data indicated that let-7 and mir-125 functioned during distinct temporal periods to effect disease progression, and survival 58d, maximum lifespan 118d, n = 205; chin 1 ; ΔmiR-100: median 74d, maximum survival 120d, n = 117; chin 1 ; Δlet-7: median 38d, maximum survival 66d, n = 253; chin 1 ; ΔmiR-125: median survival 52d, maximum lifespan 100d, n = 157; chinmo RNAi/ΔmiR-125: median survival 56d, maximum lifespan 104d, n = 281. P value of the lifespan curves was calculated by log-rank test. (E) Δlet-7 and ΔmiR-125 mutant males display age associated climbing defects. At 3 days, the climbing ability of Δlet-7 and ΔmiR-125 males was comparable to the control flies expressing the wild type let-7-C transgene. However, at 30 days, only 48.29 ± 9.6% of Δlet-7 flies and 0% of the ΔmiR-125 flies were able to climb as opposed to 91.5 ± 6.7% of control flies. The age related climbing defects of ΔmiR-125 and not Δlet-7 was partially rescued by reducing chinmo levels. When aged for 30 days, 0% of ΔmiR-125 flies, 88.1 ± 9.1% of chin 1 ; ΔmiR-125 flies, and 54.1± 12.5% of chinmo RNAi/ ΔmiR-125 flies display climbing ability upon aging. However, aged chin 1 ; Δlet-7 flies displayed only a very slight increase in climbing ability (51. 6 ± 10.8%) when compared to age-matched Δlet-7 flies (48.29 ± 9.6%). The data were statistically analyzed by an unpaired t-test. The results represented the mean ± S.D of three experiments, n = 15 male flies in each experiment, *** = p value <0.001. (F) ΔmiR-125 and Δlet-7 mutants display late onset brain degeneration that is rescued by reducing chinmo levels. Aged ΔmiR-125 and Δlet-7 males showed an increased number of vacuoles (relative to let-7-C null rescue). Reduced dosage of chinmo in ΔmiR-125 mutants decreased the vacuole count (Two tailed t-test, mean ± S.D n = 5). Genotypes used were the same as those listed for Fig 2A, 2C and 2D in S1 Table. doi:10.1371/journal.pgen.1006247.g002 that the retinal degeneration caused by a decline in let-7 activity was of a developmental origin while that caused by inhibition in miR-125 activity was due to degeneration of adult brains.

miR-125 regulates aging via repression of chinmo
Since miRNAs function by repressing target mRNAs, we investigated whether the age-associated Δlet-7 and ΔmiR-125 phenotypes were due to the elevated expression of chronologically inappropriate morphogenesis (chinmo), the only verified target of both let-7 and miR-125 in flies [38]. Chinmo is a transcription factor that controls neuronal fate in a dosage-sensitive manner. In the mushroom body lineages in the central brain, for example, Chinmo is expressed at high levels early in development to promote early born cell fates and its post-transcriptional downregulation leads to the production of later born fates [45]. The chinmo 3'UTR contains multiple let-7 and miR-125 binding sites and is regulated by let-7-C miRNAs during development [38]. To address whether the adult phenotypes of Δlet-7 and ΔmiR-125 mutants were due to elevated Chinmo, we reduced the dosage of chinmo in these mutants by either removing one copy of chinmo using a null chinmo 1 mutation [45] or by knocking down chinmo using a RNAi transgene that we verified in vivo (S4A- S4L Fig). Lowering chinmo levels dramatically suppressed the premature death (Fig 2D), climbing defects (Fig 2E), and brain necrosis ( Fig 2F  and S4M Fig) of ΔmiR-125 mutants but not Δlet-7 mutants. This result indicated that elevated Chinmo was responsible for ΔmiR-125 phenotypes but that other factors were responsible for Δlet-7 phenotypes. This distinction also indicated that the Δlet-7 phenotypes described above (reduced longevity, climbing and neurodegeneration) were not solely due to the reduction in miR-125 levels observed in these mutants (Fig 2A) and implied the de-repression of other, currently unidentified, mRNAs.
To test whether ectopic Chinmo was sufficient to cause the neurodegenerative phenotypes associated with loss of miR-125, we ectopically expressed a chinmo transgene in adult brains using an inducible neural GAL4 driver ( Fig 4A). This forced expression resulted in a drastic reduction in both lifespan ( Fig 4B) and climbing ability ( Fig 4C) along with a dramatic increase in brain vacuole numbers in 20d aged flies (Fig 4D and 4E). Interestingly over-expression of Chinmo in neurons showed an increased localization of vacuoles in the lamina and central brain regions, indicating that neurons in the lamina region were more sensitive to de-regulation of Chinmo ( Fig 4D and 4E and S2E Fig). These experiments confirmed that deregulated expression of Chinmo in adult neurons results in premature neurodegeneration in adult flies and supported the genetic suppression evidence above that silencing of this protein by miR-125 was critical for maintaining neuronal integrity and viability in adult flies.

Biphasic regulation of chinmo by let-7 and miR-125
Messenger RNAs frequently contain binding sites for multiple miRNAs, which may repress common targets in an additive manner [31]. However, our analysis raised the possibility that the chinmo mRNA might be regulated by miR-125 but not let-7 in adult brains despite containing verified functional binding sites for let-7 [38]. Intriguingly, these experiments were supported by immunostaining of Chinmo in Δlet-7 and ΔmiR-125 adult brains. The degree of de-  : median survival 66d, maximum lifespan 94d, n = 162 males). P value of the lifespan curves was calculated by log-rank test. (C) RU-486-fed males (+RU) displayed age-associated climbing defects. While normal at 10 days, only 18.8 ± 7.9% of RU-486 fed flies (+RU) were able to climb as opposed to 95 ± 5% of control flies (-RU) at 20 days. Mean ± S.D of three experiments, n = 15 male flies, *** = p value <0.001, two tailed t-test. (D-E) Brain sections of 20d RU-486-fed males (+RU) displayed elevated numbers of vacuoles than controls (-RU) (+RU, 86.4 ± 16.9 vacuoles per brain; -RU, 16.2 ± 5.5 vacuoles per brain, p<0.0001, two tailed t-test, mean ± S.D, n = 5). Arrows in D point at vacuoles. The genotype of all flies used in this figure is listed in S1 Table. doi:10.1371/journal.pgen.1006247.g004 38.5 ± 10.3 AUs while ΔmiR-125 mutant brains expressed 72.9 ± 9.2 AUs, indicating that miR-125 was primarily responsible for silencing chinmo in adults ( To support this data, we also quantified the levels of chinmo mRNA in late pupal and adult Δlet-7 and ΔmiR-125 mutant heads, since miRNA regulation is known to cause mRNA destabilization. Consistent with our quantification of Chinmo protein levels, loss of let-7 affected chinmo mRNA levels in late pupae but not adults, while loss of miR-125 affected chinmo mRNA levels in adults but not pupae ( Fig 5J). Together, these data indicated that let-7 and miR-125 predominantly regulated chinmo during development and adulthood, respectively.
Since the derepression of Chinmo during development was higher in Δlet-7 than ΔmiR-125 mutants, we tested whether chinmo-dependent developmental defects were more severe in Δlet-7 than ΔmiR-125 single mutants. To examine this possibility, we investigated the relative roles of let-7 and miR-125 in fate transitions in MB neuronal temporal identity. The MB is composed of four subtypes of neurons that are generated in a sequential manner (γ ! α'/β' ! pioneer α/β ! α/β) by neuroblasts [45][46][47]. High levels of Chinmo specify early born cell fates (γ, α'/β'), while low levels of Chinmo specify later born cell fates (pioneer α/β, α/β). Altered dosages of chinmo lead to changes in the total numbers of these various neuronal classes so that, for example, elevated chinmo is associated with a smaller population of pioneer α/β neurons [38,45]. Therefore, to evaluate the relative contributions of let-7 and miR-125 chinmo regulation during neural development, we counted the number of pioneer α/β neurons in adult brains. As expected, Δlet-7 mutants displayed greater reduction in pioneer α/β neuron number than ΔmiR-125 mutants and, furthermore, Δlet-7, miR-125 double mutants showed a reduction similar to Δlet-7 single mutants ( Fig 5K). Thus, regulation of chinmo was more dependent on let-7 than miR-125 during the larval-to-adult transition but was more dependent on miR-125 in aging adults.
To assess whether let-7 and miR-125 had different strengths of repression that might contribute to their differential activities, we examined the expression of luciferase reporters containing a previously characterized 1.4kb 3'UTR fragment of chinmo that harbors six let-7 and four miR-125 binding sites that are conserved between Drosophila species [38]. Overexpression of the entire let-7-C primary transcript in Kc-167 cells repressed the wild type reporter 17-fold ( Fig 5L). In contrast the fold repression of mutants lacking let-7 binding sites or miR-125 sites or both was reduced 6.9-, 14.0-and 4.8-fold, respectively. While these data indicated that let-7 was a stronger repressor of chinmo than miR-125, it was not clear whether the greater decrease in fold repression was due to a greater number of let-7 sites. To circumvent this issue, we designed and quantified the degree of repression of a luciferase reporter that contained a single verified miRNA binding site for let-7 and a single miR-125 site that had comparable base pairing characteristics (S6 Fig). Ectopic expression of let-7-C miRNAs resulted in a 12-fold repression of the wild type reporter while deletion of the let-7 seed sequence, the miR-125 seed sequence or both sequences resulted in a 2.9-, 4.6-and 0.7-fold repression of luciferase activity respectively ( Fig 5M). Together these data confirmed that both let-7 and miR-125 were capable of silencing the luciferase sensor individually but maximum repression was achieved when both sites were functional and that let-7 was a stronger post-transcriptional repressor than miR-125.
Differential processing and turnover rates of let-7 and miR-125 direct a switch in miRNA targeting activity during the larval-to-adult transition To investigate the basis for the sequential repression of Chinmo by let-7 and miR-125, we first compared the rate of let-7 and miR-125 production in the developing and adult nervous system. To do so, we quantified the ratio of processed miRNA to precursor miRNA for let-7 and miR-125 in staged nervous system samples ( Fig 6A). The let-7/pre-let-7 ratio was significantly higher than the miR-125/pre-miR-125 prior to 72h APF. In contrast, the miR-125/pre-miR-125 ratios were higher than the let-7/pre-let-7 ratios after 72h APF and into adulthood (Fig 6A). The temporal dynamics of let-7 and miR-125 production in the nervous system correlated with their relative roles in chinmo regulation, suggesting that the basis for their sequential activity may involve differential processing and/or turnover.
To determine the basis for this switch in the relative expression of let-7 and miR-125, we first investigated the possibility that let-7 and miR-125 might be differentially processed. To do so, we again took advantage of the Kc-167 embryonic cell line. As mentioned above, the let-7-C locus is not ordinarily transcribed in this cell line, but it is activated in response to the Drosophila steroid hormone 20-hydroxyecdysone (20E) [37]. Kc-167 cells were treated with 20E for 24h to induce primary let-7-C transcript and the levels of let-7 and miR-125 were monitored at different time intervals after washing off the steroid hormone. To measure the relative rates of processed let-7 and miR-125 production, we performed qRT-PCR on 20E-treated Kc-167 samples using a standard curve to extrapolate absolute miRNA levels. Forty-eight hours after the To derive absolute levels of expression, a standard curve was generated using synthetic let-7 and miR-125 RNA oligonucleotides and CT values were extrapolated from the curves. Quantitation of copy number per nanogram of total RNA indicated a significantly higher processing rate of let-7. Values represented as mean of three experiments ± S.D. (C) Primary let-7 transcripts expressing either miR-125 terminal loop (pri-let-7 miR-125L ) or stem base (pri-let-7 miR-125B ) displayed slower kinetics of Drosha processing than the wild type pri-let-7 transcript. In vitro processing of pri-let-7, pri-let-7 miR-125B and pri-let-7 miR-125L with purified Flag tagged Drosha-Pasha complex. The primary transcript (pri), 5' flank (5'F), 3' flank (3'F) and precursor (pre) are indicated on the left. Quantitation of the fraction processed is calculated as precursor/primary +5'F+3'F+precursor. (D) Pre-let-7 is diced more efficiently than pre-miR-125. In vitro processing of radiolabeled pre-let-7 and pre-miR-125 with purified Flag-tagged Dicer 1. Products were analyzed on a 10% polyacrylamide gel. The precursor and the processed miRNA are indicated on the left. Quantitation of fraction diced is indicated at the bottom of the gel. (E) let-7 has a higher turnover rate than miR-125 in the adult central nervous system. Expression levels of let-7 (dark gray) and miR-125 20E treatment, we detected 1024±440 copies of let-7 per nanogram (ng) of total RNA but only 3±1 copies of miR-125 per ng of total RNA, indicating that let-7 was processed more efficiently than miR-125 (Fig 6B). Incubation of cells for 120 hours after the 20E pulse resulted in a 3-4 fold increase in the copy number of let-7 (3872±1365 copies/ng of total RNA) and a~30 fold increase in the copy number of miR-125 (101±24 copies/ng of total RNA). The greater increase in the miR-125 copy number at later time points suggested that processed miR-125 persisted longer than let-7.
In order to identify the key steps in miRNA biogenesis that contributed to the inefficient processing of miR-125, we performed in vitro Drosha and Dicer processing assays (Fig 6C and  6D). We first examined the rate of generation of precursor miRNAs (pre-miRNA) from longer primary miRNA (pri-miRNA) transcripts by the Drosha-Pasha complex. In initial experiments, we found that pri-miR-125 processing was extremely inefficient (S7A Fig). Therefore, we compared the processing of pri-let-7 to the processing of chimeric constructs in which either the pri-let-7 terminal loop or its stem-base were replaced with the pri-miR-125 loop (prilet-7 miR-125L ) or pri-miR-125 stem base (pri-let-7 miR-125B ), respectively (Fig 6B and S7B Fig).
The rates of processing of these three transcripts were examined by incubating with Drosha-Pasha complexes immunoprecipitated from Kc-167 cells (S7B Fig). Substituting either the terminal loop or stem base of pri-miR-125 in pri-let-7 resulted in a dramatic reduction in Drosha processing. While 15% of the unmodified let-7 primary transcript was processed within 30 minutes, only 5% of pri-let-7 miR-125L was cleaved by Drosha. Incubation with Drosha-Pasha complex for 120 minutes increased the percentage of precursor to 30% and 7% for pri-let-7 and pri-let-7 miR-125L , respectively. However, substituting the stem-base of pri-miR-125 in pri-let-7 completely abolished its Drosha processing (Fig 6C). Thus, both the terminal loop and stem base sequence determinants of pri-miR-125 contributed to its inefficient processing by Drosha in vitro, raising the possibility that other post-transcriptional mechanisms may facilitate miR-125's processing in vivo. To evaluate the Dicer-1 processing of pre-let-7 and pre-miR-125, we also performed in vitro processing assays with Flag-tagged Dicer-1 that was, like the Drosha-Pasha complexes described above, purified from Kc-167 cell extracts (S7B Fig). In these assays, pre-miR-125 displayed a significantly lower kinetics of processing than pre-let-7 (Fig 6C). Within 10 minutes of incubation with Dicer-1, 23±3.6% of pre-let-7 and 15±0.6% of pre-miR-125 were processed to their mature forms. After 60 minutes of incubation, the percentage diced was 71±7.2% and 47±0.4% for pre-let-7 and pre-miR-125, respectively (Fig 6D and 6E). Thus, this higher kinetics of processing of let-7 by both Drosha and Dicer likely contributed to its rapid accumulation during metamorphosis.
While differential processing was consistent with the more rapid accumulation of let-7, we hypothesized that the temporal dynamics of miR-125 accumulation (Fig 6A) might also reflect an increased stability. In order to monitor the persistence of let-7 and miR-125 in adult nervous system tissue, we measured the expression of these miRNAs after blocking Dicer-1 activity. Total RNA was extracted from heads of 10d-old adult flies that expressed one of two Dicer-1 shRNA constructs, and the levels of let-7 and miR-125 were quantified by Taqman miRNA assays (Fig 6D and S7C Fig). Knockdown of Dicer-1 resulted in a greater reduction of let-7 relative to miR-125 (Fig 6D): expression of Dicer-1 shRNA1 or Dicer-1 shRNA2 resulted in 0.23±0.1 or 0.31±0.37 fold expression of let-7 relative to control but 0.62±0.18 or 0.58±0.18 fold expression of miR-125 relative to control, respectively. These data indicated that the decay rate of let-7 was significantly higher than that of miR-125 in the adult nervous system.
Finally, to assess whether miR-125 had a longer half-life than let-7, we measured the decay rates and half-lives of let-7 and miR-125 by analyzing Kc-167 cells transfected with synthetic miRNA duplexes. Cells were washed with fresh medium 5 hours after transfection, and samples were collected at the indicated times for total RNA preparation followed by quantitation of let-7 and miR-125 by Taqman assays. Half-lives were inferred from fitted exponential curves (S7D Fig). As expected, the half-life of miR-125 (T 1/2 , 2.7h) was significantly higher than that of let-7 (T 1/2 , 2h). Taken together these experiments suggested a mechanistic basis for the switch in let-7-to-miR-125 activity that occurred during pupal-to-adult transition: while an increased rate of Drosha/Pasha and Dicer processing of let-7 facilitated the attenuation of chinmo in the developing nervous system, the enhanced perdurance of miR-125 ensured that chinmo was silenced in the adult brains (Fig 7).

Discussion
Summary and model let-7 and miR-125 have distinct and non-overlapping functions, despite being co-transcribed and sharing the same target. Loss of either miRNA alone leads to shortened lifespan and premature deterioration of health, as indicated by age-dependent climbing defects and brain degeneration. The aging defects caused specifically by loss of miR-125 are associated with high levels of Chinmo in adult brains, and can be rescued by reducing chinmo levels in the ΔmiR-125 mutant. In contrast, Chinmo is substantially lower in Δlet-7 mutant adult brains and it appears not to contribute to adult Δlet-7 mutant phenotypes: neither adult Δlet-7 mutant climbing defects, brain vacuolization, nor reduced longevity are suppressed by chinmo reduction. Instead, let-7 predominates during development: pupal Chinmo expression is higher and associated defects in neuronal identity are worse in Δlet-7 mutants than ΔmiR-125 mutants. Although deletion of let-7 reduces miR-125 levels, the differences in the Δlet-7 and ΔmiR-125 Biphasic action of two co-transcribed microRNAs. Pre-let-7 and pre-miR-125 are co-transcribed, but mature let-7 accumulates more rapidly than mature miR-125 during development due to its higher rate of processing by Drosha and Dicer. The enhanced stability of miR-125 leads to a switch in the relative abundance of the two miRNAs during adulthood. This differential temporal expression contributes to the distinct functions of the two miRNAs during the life of a neuron: let-7 fine-tunes the gradient of the dosagesensitive transcription factor chinmo to control temporal cell fate determination during neural stem cell (NSC) division (for example, adjusting γ, α'/β', and α/β identities in the mushroom body lineage) while miR-125 ensures the complete silencing of chinmo during adulthood to promote neuron maintenance. Such phasic control may be a general feature of clustered miRNAs. phenotypes indicate that the Δlet-7 phenotypes are not simply due to loss of miR-125. In support of this, distinct temporal periods of let-7 and miR-125 activity were also identified using sponges, an independent method for disrupting miRNA activity in which let-7 interference does not affect miR-125 activity. Based on these results, we conclude that a let-7-to-miR-125 switch during the pupal-to-adult transition ensures chinmo repression in adults, maintaining neuronal integrity and promoting life span.
Our results illuminate a function of miRNA co-targeting that we term "phasic control," which indicates that co-targeting can reflect non-redundant regulation during distinct phases of a cells life, from its birth to its death. Rather than simply reinforcing silencing, such repression at different times may have distinct functions, based not only on the changing status of the cell but also on differences in miRNA::mRNA interactions (e.g. base-pairing characteristics, trans-acting factors, etc). Highlighting such phasic control, we propose a model in which let-7-C miRNAs collectively function as both a rheostat and as a switch but at distinct times ( Fig  7). According to this model, let-7 predominates during nervous system formation, where it shapes the temporal gradient of Chinmo. let-7-dependent attenuation of this dosage-sensitive transcription factor is responsible for the establishment of proper cell fate as neural progenitors divide. Subtle alterations in the rate of let-7 accumulation may adjust the neuronal classes that comprise structures like the mushroom body, whose composition is known to be sensitive to environmental cues [48]. While let-7 adjusts chinmo, miR-125 in contrast switches chinmo off in post-mitotic neurons throughout the adult nervous system. This silencing of a juvenile neuronal marker maintains adult neuronal integrity, since forced Chinmo expression in adults leads to brain deterioration. While our model proposes that miR-125 ensures complete silencing of chinmo, we cannot rule out the possibility that miR-125 repression is alleviated under certain conditions in the adult (e.g. injury-induced repair) so that Chinmo can reprogram neurons to a juvenile state that is needed for certain adult functions. Thus, by independently regulating the same target during two different periods, let-7 and miR-125 miRNAs control cell fate establishment and maintenance, respectively.
Our model is based in part on results that chinmo repression is achieved predominantly by miR-125 in adult brains, even though let-7 is present. What accounts for the muted let-7 activity that, while present, is not responsible for repression of a verified target? Perhaps let-7 has many more targets than miR-125 in the adult brain, since miRNAs with a larger repertoire of target genes have a weaker effect on each individual target [49,50]. In addition, let-7 targets may be highly expressed in the adult brain, thereby titrating away functional let-7 and leading to its reduced effect on all its targets, including those that are co-targeted by both let-7 and miR-125. Such a scenario is supported by increasing evidence that the effectiveness of a particular miRNA is influenced by the cellular concentration of available miRNA binding sites [51,52]. Alternatively, perhaps let-7 silencing requires cofactors that are only expressed during development. Future studies focused on the identification and characterization of the miR-125independent targets of let-7 should provide insight into the networks of let-7 targets in adults.

Significance and scenarios of miRNA co-targeting
While an overarching feature of miRNA regulation is that mRNAs are responsive to multiple miRNAs, our understanding of the biological significance of co-targeting is rudimentary. Supporting the apparently abundant co-targeting identified by miRNA binding site predictions [22-24, 28, 53], there are plenty of examples of multiple miRNAs that can when expressed one-by-one repress the same 3'UTR reporter. Such examples include repression of mtpn 3'UTR by any one of a trio of miRNAs (miR-375, miR-125, and let-7b) and repression of cdkn1A/p21 3'UTR by any one of a staggering 28 different miRNAs [53,54]. Since the effects of combinations of miRNAs are rarely tested, such studies suggest that multiple miRNAs limit the spatial and/or temporal expression of targets and, in cells where they are co-expressed, may function redundantly with one another.
In addition to this simple scenario, there are hints of more complex combinatorial scenarios involving either cooperation or competition between co-targeting miRNAs. In the cooperative scenario, miRNAs with overlapping expression patterns lead to enhanced repression of co-targeted mRNAs in cells where they are expressed together. Examples of this include miR-25 and miR-221/222 co-repression of p57 and miR-148a and the miR-206 co-repression of dmpk [55,56], although it is worth noting that the additive effects of these pairs of miRNAs is small though significant. Supporting such cooperative action, additional studies report that multiple binding sites, especially when they are within 15-35 bp of one another, lead to enhanced reporter repression [29,57,58]. Co-expressed miRNAs can also act competitively, as shown for miR-184 and miR-205 regulation of ship2 [59]. In this scenario, miR-184 does not have repressive activity itself but alleviates the repressive ability that miR-205 exerts via an overlapping binding site, as shown by comparative analysis of miR-205 alone versus in combination with miR-184 as well as analysis of mutated 3'UTR reporters containing intact miR-205 but mutant miR-184 sites. In light of this miR-184 function, systematic assays that have found that many miRNAs when expressed individually have no effect on a 3'UTR reporter do not rule out the possibility that these miRNAs function competitively with others [60,61].
Phasic control expands the repertoire of known co-targeting functions, and emphasizes that co-targeting miRNAs may function at different times from one another and for different purposes. Phasic control may be particularly relevant to clustered miRNAs, since clustered miR-NAs are enriched for co-targeting relationships [22,28]. For example, the vertebrate miR-17~92 cluster, like other miRNA clusters, targets multiple components of related networks and pathways of genes, including the TGF-β pathway [62,63]. As with the let-7-C cluster in flies, members of polycistronic clusters, including the miR-17~92 and miR-1/miR-133 clusters, are differentially processed [64,65]. The resulting differential accumulation of these co-targeting miRNAs, along with differential base pairing and turnover of co-transcribed miRNAs, may lead to the distinct temporal accumulations of processed miRNAs that are indicative of phasic control. Thus, the staggered accumulation of different miRNAs processed from the same polycistronic transcript over time may be an important feature controlling the progression of temporal features of cell and organismal biology.

Conservation of co-targeting by let-7-C miRNAs
The biphasic regulation by let-7-C miRNAs may also be relevant to mRNAs co-targeted by let-7-C orthologues in other animals. These include lin-28 and lin-41, which were originally identified in C. elegans as potential targets of both let-7 and lin-4, the C. elegans miR-125 orthologue [66,67]. These co-targeting relationships are conserved to vertebrates, since mouse lin-28 and lin-41 3'UTRs are responsive to altered levels of both let-7 and miR-125 [68][69][70]. The neurodevelopmental functions of these co-targetings has been extensively investigated and include, for example, let-7/miR-125-mediated repression of lin-28 to control temporal identity during retinal neurogenesis in zebrafish [71,72]. However, the careful dissection of the relative roles of let-7 versus miR-125 as well as their respective post-developmental functions awaits future investigation. While the lin-28/let-7-C relationship does not appear to be conserved to flies [73], the recent identification of a let-7 and miR-125 target with homology to chinmo suggests that regulation of a chinmo orthologue may be conserved [74]. This target, hypermethylated in cancer 2 (hic2), encodes a BTB-zinc finger (BTB-ZF) transcription factor that contains multiple predicted let-7 and miR-125 sites in its 3'UTR. While reciprocal homology searching predicts equally good amino acid similarity between Chinmo and a number of mammalian BTB-ZFs including Hic2, the conservation of let-7 and miR-125 sites in the hic2 3'UTR but not other BTB-ZF 3'UTRs suggests that Hic2 is the mammalian orthologue of Chinmo. Thus, our results predict that mammalian let-7 and miR-125 regulate hic2 in a biphasic manner.

let-7-C miRNAs in aging and neurodegeneration
The persistence and gradual increase of let-7-C miRNAs during adult life may balance the various cellular demands needed for proper tissue and organismal homeostasis over time. Thus, the increasing levels of let-7 that dampen stem cell function in aging tissue, found both in the mouse nervous system and the fly testis [35,75], may be part of a general program that includes the neuronal maintenance function that this study explores. A conserved role for let-7-C miR-NAs in such cell maintenance during adult life is supported by the requirement of lin-4 for proper lifespan in C. elegans [2], since nematodes, like fly brains, exhibit limited cell proliferation during adulthood. In addition, Drosophila let-7 and its target the dp transcription factor promote the maintenance of dopaminergic neurons in the adult brain by regulating the expression of pathogenic Leucine-Rich Repeat Kinase 2 [76]. Interestingly, changes in miR-125b have been linked to Alzheimer's disease and vertebrate cerebellar neurodegeneration, although the molecular mechanisms underlying these changes have not been addressed [77,78]. Taken together, this mounting evidence indicates that let-7 and miR-125 play critical neuroprotective roles in the aging brain so understanding their post-developmental functions in greater detail may be relevant to therapies for human neurodegenerative diseases, including Parkinson's and Alzheimer's diseases. In summary, our work has identified a novel in vivo mechanism by which multiple miRNAs repress a common target during distinct stages. Such differential regulation by subsets of co-expressed miRNAs should be considered for designing therapeutic strategies to treat diseases that are frequently caused by de-regulation of highly targeted mRNAs.

Drosophila husbandry
All flies were cultured on standard cornmeal medium at 25°C under 12 h light, 12 h dark cycles, except for flies analyzed in the temperature sensitive experiment presented in Fig 3. These flies were cultured in one of three regimens: at 18°C, at 29°C, and at 18°C until eclosion and then at 29°C thereafter. For steroid mediated UAS-transgene control using the Gene-Switch driver, flies were fed food containing 200μM RU-486 (Mifepristone, Cayman Chemicals, Ann Arbor MI). Staging of pupae and MARCM clone induction was performed as previously described [37,38]. Unless otherwise noted, adult male flies of indicated ages were used for experiments.

Drosophila genetics and let-7-C mutant strain construction
Detailed genotypes of all strains as well as the sources of the genetic mutations and transgenes used in the study are listed in S1 and S2 Tables, respectively. Transgenesis was performed by Rainbow Transgenic Services (Camarillo, CA) and BestGene, Inc. (Chino Hills, CA). The let-7-C mutant strains analyzed in this study (including let-7-C null , let-7-C hyp , let-7-C hyp rescue, ΔmiR-100, Δlet-7 and ΔmiR-125 strains) were generated by crossing w 1118 ; let-7-C GKI / CyO strains to w 1118 ; let-7-C KO2 , P{neoFRT}40A / CyO strains in which one or both of these strains contained a rescuing transgene inserted on the third chromosome. Since this approach generated trans-heteroyzgous let-7-C GKI / let-7-C KO2 , P{neoFRT}40A animals, it eliminated the effect of any confounding recessive background mutations that might have accumulated on those chromosomes. In addition, because the differing rescuing transgenes were inserted at identical positions on the third chromosome, this approach ensured the pairwise comparison of strains that were otherwise as genetically similar to one another as possible. The detailed genetic scheme for generation of the transgenic samples is described in S1 Fig.

Climbing and lifespan analyses
Climbing assays were performed as described previously [33]. Lifespan analysis was performed as previously reported [1,3,79] using let-7-C mutant flies that were generated as described above. Fifteen male flies (0-1 day old) were transferred to each vial. Flies were transferred to fresh food every 3 days at which time dead flies were counted and removed. The survival curves were plotted using Microsoft Excel. Statistical analysis was performed with the Online Application for the Survival Analysis of lifespan assays (OASIS) [80] and the p values were calculated using the log-rank (Mantel-cox) test. The number of flies used for each experiment are noted in the figure legends, and also included along with the median and maximum lifespans of the tested strains in S3 Table. S4 Table indicates the p values for curves shown in one or more panels. The numbers of flies used for each experiment have been noted in the figure legends. Experiments usually included two independent controls: w 1118 as well as a let-7-C mutant strain containing a fully rescuing transgene. The w 1118 survival curve was generated with flies that had been back crossed five times.

Immunofluorescence
Immunofluorescence was performed as described previously [37,38]. Primary antibodies included rat anti-Chinmo [38] (1:500), chicken anti-GFP (Rockland Immunochemicals, 1: 4000), rabbit anti-Woc [81] (gift from Maurizio Gatti 1:1000), rat anti-Elav (DSHB, 1:250) and mouse anti-Dachshund (DSHB, 1:100). For quantitating Chinmo levels, pixel intensity of 30 individual cells in single confocal sections of 5 independent dissected brains stained with anti-Chinmo and anti-Woc antibodies were quantified using ImageJ software. The expression of Chinmo was normalized to the pixel intensity of Woc and the average pixel intensity of one Δlet-7-C confocal section showing the highest pixel intensity was designated as 100 Arbitrary Units (AU). Samples whose staining was directly compared were prepared and imaged in parallel and under identical conditions. For c708a neuron counts, mushroom bodies were optically sectioned in 0.5 μm increments, and the total number of neurons was determined by manually counting the number of GFPpositive cells section by section, ensuring that cells present on consecutive sections were counted only once. Statistical analysis was performed and histograms generated using Graph-Pad Prism software. P values were calculated using a two-tailed paired t test. Values are presented as mean ± SEM. All images were collected on a Leica SP5 confocal microscope (Light Microscopy Imaging Center, Indiana University, Bloomington IN). Confocal stacks were merged using Leica LSM software.

Northern blot analysis and quantitative real time PCR
Total RNA was extracted with Trizol and treated with DNAse I. The reverse transcription was performed as described previously [82]. For analysis of miRNA copy number in Fig 4A, Kc-167 cells were incubated with 20E (5×10 −6 M) at 25 0 C for 24h before being washed with fresh medium. Reverse transcription (RT) was carried out on 25ng of total RNA using the Reverse Transcription miRNA Taqman assays (Applied Biosystem, Foster City, CA) specific for the miRNA (dme-let-7 and dme-miR-125). Each cDNA sample was diluted 1:25 and real-time quantitative PCR (qPCR) was performed in duplicate using miRNA-specific primers/probe on a StepOnePlus Real Time PCR System (Applied Biosystem, Foster City, CA). For determination of copy number, we generated a standard curve for let-7 and miR-125 using a synthetic let-7 and miR-125 HPLC-purified RNA oligonucleotides synthesized by Integrated DNA Technologies (Coralville, IA) corresponding to the 22 nucleotide miR-125-5p (5'-rUr CrCrCrUrGrArGrArCrCrCrUrArArCrUrUrGrUrGrA-3') and 21 nucleotide let-7-5p (5'-rUr GrArGrGrUrArGrUrArGrGrUrUrGrUrArUrArGrU-3'). For fold change analysis, individual values were normalized to 2S rRNA for Taqman miRNA assays and kinesin levels for Sybr green assays. For qRT-PCR analysis, oligos 2515, 2516, 2599, 2530, 2728, 2729 listed in S5 Table were used. Northern blot analysis was performed as described previously [37].

Plasmid and transgenes
Chinmo transgenes. pP{w+, UAS-chin::SV40} contained the Chinmo open reading frame (ORF) flanked by hsp70 5'UTR and SV40 3'UTR and under the control of UAS sites. It was generated by PCR amplifying the Chinmo ORF with oligos 136 and 137 (see S5 Table for sequences) from reverse transcribed RNA generated from CNS tissue, then sequence verifying and subcloning the resulting fragment into the BglII and NotI restriction enzyme sites of plasmid pUAST. pP{w+, UAS-chinmo RNAi 148 } encoded a short hairpin RNA (TGTGGGCTTTG AATACTACGC) targeting chinmo, designed based on rules described previously [83] and under the control of UAS sites. It was generated by subcloning the annealed oligos 1004 and 1005 (see S4 Table for sequences) into the EcoRI and NheI sites of plasmid pWalium20 (TRiP at Harvard Medical School).
Sponge transgenes. Sponge constructs targeting miR-100, let-7 and miR-125 were designed based on Loya et al. [84] and Bejarano et al. [85]. A silencing cassette for each miRNA was synthesized by BioBasic, Inc that contained twenty miRNA complementary sequences separated by variable four-nucleotide linker sequences (see S6 Table for complete sequences). The entire cassette was subcloned into the NotI and XbaI sites of a modified pVa-lium10 plasmid (TRiP at Harvard Medical School, Boston, MA) and inserted into both attP40 and attP2 sites using phiC31 site-specific genomic integration. Resulting transformants were identified using vermillion as a transformation marker. let-7-C locus transgenes. New transgenes were generated using previously reported P-element based transgenes containing either the full-length let-7-C locus or variants in which miR-100, let-7, and/or miR-125 were specifically deleted [33]. These~18kb fragments were excised using unique restriction sites AvrII and XbaI and subcloned into the XbaI site of a modified pValium10 plasmid (TRiP at Harvard Medical School, Boston, MA)). Resulting plasmids were inserted into attP2 sites using phiC31 site-specific genomic integration and transformants were identified using vermillion as a transformation marker.
Luciferase reporters. The nonmutated chinmo 3'UTR sensor has been described previously [38]. The constructs containing seed deletions of the four miR-125 or six let-7 sites were generated using the QuikChange Lightning Multi Site-Directed Mutagenesis Kit (Agilent Technologies, Santa Clara, CA) and oligonucleotides 978-981 and 972-977 respectively (see S5  Table for oligonucleotide sequences). The synthetic luciferase sensors were generated by annealing oligonucleotides with NotI and XhoI compatible ends (3046/3047, 3048/3049, 3050/ 3051 and 3052/3053). The annealed oligonucleotides (sequences in S5 Table) were cloned into a modified version of pSiCheck2 (Promega Life Science, Madison WI). Fold repression was calculated relative to control samples transfected with empty vector instead of miRNA-encoding plasmids. The luciferase reporters used in Fig 2B were psiCHECK plasmids bearing six perfect sites for either miR-100, let-7 or miR-125 downstream of a Renilla luciferase gene were used and have been described previously [38].
Tagged protein plasmids. Plasmids encoding N-terminal Flag tagged version of Dicer was generated by recombining pENTR-Dicer 1 (kind gift from Mikiko C. Siomi) with the pAFW gateway plasmid (T. Murphy; obtained from the Drosophila Genome Resource Center) using the LR Clonase enzyme (Life Technologies, Carlsbad CA), respectively. Flag tagged Drosha and Pasha constructs have been described previously [82].

Luciferase reporter assays
Drosophila Kc-167 cells were cultured in CCM3 at 23°C. Cells were transfected in 48-well plates with 25 ng of tub-Gal4 plasmid DNA, 25 ng UAS-miRNA plasmid DNA, and 25 ng of 3'UTR-containing sensor plasmid DNA using Effectene (Qiagen). Luciferase assays were performed using the Dual-Luciferase reporter system (Promega Life Science, Madison WI). Transfections were performed in triplicates and resulting luciferase levels were averaged. Fold repression was calculated by dividing the ratio of Renilla luciferase and firefly luciferase in cells transfected with an empty pUAST attB plasmid with the ratio of Renilla luciferase and firefly luciferase in cells transfected with pUAST attB plasmid containing let-7-C cDNAs.

MiRNA duplex transfection and determination of half-life of miRNAs
Si-miRNA duplexes were synthesized as single-stranded RNAs by Integrated DNA Technologies (Coralville, IA) with HPLC purification, and resuspended in duplex buffer (100mM potassium acetate, 30mM HEPES, pH 7.5) to a concentration of 100μM. Annealing was performed by incubating 50 μM complementary single-stranded RNAs at 92°C for 2 min and leaving them for 30 min at room temperature [86].
si miR-125 sense; 5'UCCCUGAGACCCUAACUUGUGAUU si miR-125 antisense; 5'UCACAAGUUAGGGUCUCAGGGACU si let-7 sense; 5'UGAGGUAGUAGGUUGUAUAGUCU si let-7 antisense; 5'ACUAUACAACCUrArCrUrArCrCrUrCrArUrU The miRNA duplexes were transfected using Dharmafect duo according to manufacturer's instructions (GE Life Sciences, Lafayette CO). Briefly, 2.25 μl of mi-siRNA molecules (diluted to 4 μM in duplex buffer) was added such that the final concentration of each siRNA was 5 nM per well (the volumes indicated are for biological triplicate) in a 24 well plate. The cells were incubated at 25 0 C for 5 h before being washed with fresh medium. Quantitative real time PCR was performed to measure the relative levels of the miRNAs in total RNA extracted from transfected cells at different time points. The half lives of let-7 and miR-125 were determined by exponential regression curve fitting using GraphPad Prism version 6 software.

In vitro Drosha and Dicer processing assays
Pri-let-7, and pri-miR-125 were generated by annealing oligos cloned into pLitmus 28i. DNA templates for transcription were generated by PCR with the T7 and 2162 oligo and were transcribed and labelled with 32 UTP (Perkin Elmer, Waltham MA) using the T7 Megashortscript Kit (ThermoFisher, Cambridge MA). The transcript was purified by running the DNAse I treated reaction on a 4% denaturing PAGE gel and the gel piece corresponding to the labeled transcript was excised from the gel and eluted in a Eppendorf Thermomixer set at 400rpm and 37°C in a buffer containing 0.3M Sodium acetate, 0.2% Sodium dodecyl sulphate, and 1mM EDTA. The supernatant was precipitated in Ethanol. The precipitated RNA was refolded by heating at 95°C for 2 minutes followed by 37°C for 1 hour. A typical 25μL reaction contained 15μL of the Flag-Drosha-Pasha beads immunoprecipitate, 6.4mM MgCl 2 , 1 U/μL of Ribonuclease Inhibitor (ThermoFisher, Cambridge MA), and the refolded labeled transcripts (0.5 × 10 5 cpm). The reaction mixture was incubated at 26°C for 30 to 90 min, and RNA was extracted by phenol followed by ethanol precipitation and analyzed on a 10% denaturing polyacrylamide gel.
Supporting Information S1 Fig. Scheme for Generation of Experimental Samples. This study compared flies that were generated using a scheme that ensured that they had similar genetic backgrounds. Flies that were analyzed (F14) were trans-heterozygous for two different let-7-C null alleles (indicated by red and yellow bars), ensuring that phenotypes were not due to recessive mutations on either let-7-C mutant chromosome. In addition, third chromosomes that contained differing rescuing transgenes (indicated by green bar) were derived in parallel from the same population of flies. Finally, all flies had a common X-chromosome (blue bar), derived from an isogenized stock. (S1-1) All rescuing transgenes, including the wildtype rescuing transgene as well as let-7 and miR-125 deleted versions, were injected into embryos from the same population of stock BL#25710 from the Bloomington Drosophila Stock Center. Resulting progeny were backcrossed twice to BL#32261 in order to select and balance vermillion+ transformants (F1 and F2). Single transformants were subsequently backcrossed to an isogenized version of BL#3703 three times (F3-F5) in order to make balanced stocks with isogenized X chromosomes (F6). (S1-2) Stocks with differing rescuing transgenes were crossed to the same population of a stock that contained the let-7-C KO2 chromosome, an isogenized X chromosome, and two 3 rd chromosome balancers. The let-7-C KO2 stock used in F7 was generated in a similar fashion as the rescuing transgenes stocks, by backcrossing three times to an isogenized version of BL#3703. Resulting stocks (F8) had common X (blue), 2 nd (yellow) and 3 rd (green) chromosomes, and were used in F13 to generate the experimental strains. (S1-3) A second let-7-C allele, let-7-C GKI , was prepared by outcrossing twice to an isogenized stock, and then crossed to an isogenized stock containing a T(2:3) Cyo-TM6b compound chromosome. The let-7-C allele was selected based on mini-white, and the T(2:3) Cyo-TM6b balancer was selected based on the dominant Humoral marker. The resulting stock with a fixed second and third chromosome was amplified and used as the source for all virgins in the crosses that yielded the flies for analysis. (S1-4) Flies for analysis were generated by crossing virgins of the stock generated in F12 with males of stocks generated in F8 that harbored differing rescuing transgenes. Histochemistry was performed on brain sections of 40d old ΔmiR-125 mutants and chin 1 , ΔmiR-125 mutants and the number of vacuoles were scored to assess brain morphology. Representative examples of brains sections are shown (vacuoles indicated by blue arrows) and the total vacuole number quantified from such sections of five independent brains is presented in Fig 2D. (TIF) S5 Fig. miR-125 is the predominant miRNA that silences chinmo in adult flies. (A-D) Confocal images of 3d old adult brains immunostained for Chinmo (green). No Chinmo expression was detected in brains of flies harboring either the wild type or the ΔmiR-100 transgene (panels A and B). The level of Chinmo expression in ΔmiR-125 mutants is much higher than in Δlet-7 mutant adult flies (compare panels C and D). Genotypes used are the same as those listed for Sequences and predicted base-pairing of let-7 (green) and miR-125 (red) binding sites in the luciferase sensors. Numbering is relative to the first nucleotide in the 3'UTR. Yellow boxes indicate the sequences that were deleted in the mutant constructs. The sites were designed so that the binding pattern was comparable between the miRNAs and the target sites. The nucleotides 1-9, 11-15, 19-21(let-7) and 19-22(miR-125) formed base-pairing interactions with the 3'UTR. The minimum free energy (mfe) calculated by RNAhybrid is indicated on the right of each binding site.