The MicroRNA mir-71 Inhibits Calcium Signaling by Targeting the TIR-1/Sarm1 Adaptor Protein to Control Stochastic L/R Neuronal Asymmetry in C. elegans

The Caenorhabditis elegans left and right AWC olfactory neurons communicate to establish stochastic asymmetric identities, AWCON and AWCOFF, by inhibiting a calcium-mediated signaling pathway in the future AWCON cell. NSY-4/claudin-like protein and NSY-5/innexin gap junction protein are the two parallel signals that antagonize the calcium signaling pathway to induce the AWCON fate. However, it is not known how the calcium signaling pathway is downregulated by nsy-4 and nsy-5 in the AWCON cell. Here we identify a microRNA, mir-71, that represses the TIR-1/Sarm1 adaptor protein in the calcium signaling pathway to promote the AWCON identity. Similar to tir-1 loss-of-function mutants, overexpression of mir-71 generates two AWCON neurons. tir-1 expression is downregulated through its 3′ UTR in AWCON, in which mir-71 is expressed at a higher level than in AWCOFF. In addition, mir-71 is sufficient to inhibit tir-1 expression in AWC through the mir-71 complementary site in the tir-1 3′ UTR. Our genetic studies suggest that mir-71 acts downstream of nsy-4 and nsy-5 to promote the AWCON identity in a cell autonomous manner. Furthermore, the stability of mature mir-71 is dependent on nsy-4 and nsy-5. Together, these results provide insight into the mechanism by which nsy-4 and nsy-5 inhibit calcium signaling to establish stochastic asymmetric AWC differentiation.


Introduction
Cell fate determination during development requires both the induction of cell type specific genes and the suppression of genes that promote an alternative cell fate [1][2][3][4]. For example, both inductive signaling, mediated by an EGFR-Ras-MAPK pathway, and lateral inhibition, mediated by LIN-12/Notch activity and microRNA (miRNA), are required for six multipotential vulval precursor cells to adopt an invariant pattern of fates in C. elegans [5]. Notch signalingmediated lateral inhibition also plays a crucial role in the neuronal/ glial lineage decisions of neural stem cells; as well as the B/T, alphabeta/gammadelta, and CD4/CD8 lineage choices during lymphocyte development [6,7]. In the Drosophila eye, the kinase Warts and PH-domain containing Melted repress each other's transcription in a bistable feedback loop to regulate the two alternative R8 photoreceptor subtypes expressing Rhodopsin Rh5 or Rh6 [2]. In the C. elegans sensory system, two sets of transcription factors and miRNAs reciprocally repress each other to achieve and stabilize one of the two mutually exclusive ASEL and ASER taste neuronal fates [8][9][10]. Notch signaling acts upstream of the miRNAcontrolled bistable feedback loop to regulate ASE asymmetry through a lineage-based mechanism in early embryos [11].
The C. elegans left and right sides of Amphid Wing Cell C (AWC) olfactory neurons specify asymmetric subtypes through a novel mechanism independent of the Notch pathway in late embryogenesis [12]. Like ASE neurons, the two AWC neurons are morphologically symmetrical but take on asymmetric fates, such that the AWC ON neuron expresses the chemoreceptor gene str-2 and the contralateral AWC OFF neuron does not [12][13][14]. Asymmetric differentiation of AWC neurons allows the worm to discriminate between different odors [15]. In contrast to reproducible ASE asymmetry, AWC asymmetry is stochastic: 50% of animals express str-2 on the left and the other 50% express it on the right. Ablation of either AWC neuron causes the remaining AWC neuron to become AWC OFF , suggesting that AWC OFF is the default state and the induction of AWC ON requires an interaction or competition between the AWC neurons [12]. The axons of the two AWC neurons form chemical synapses with each other; AWC asymmetry is established near the time of AWC synapse formation [16,17]. In addition, axon guidance mutants are defective in inducing the AWC ON state. These results suggest that the synapses could mediate the AWC interaction for asymmetry [12].
nsy-4, encoding a claudin-like tight junction protein, and nsy-5, encoding an innexin gap junction protein, act in parallel to downregulate the calcium-mediated UNC-43 (CaMKII)/TIR-1 (Sarm1)/NSY-1 (MAPKKK) signaling pathway in the future AWC ON cell [18,19]. Both AWCs and non-AWC neurons in the NSY-5 gap junction dependent cell network communicate to participate in signaling that coordinates left-right AWC asymmetry. In addition, non-AWC neurons in the NSY-5 gap junction network are required for the feedback signal that ensures precise AWC asymmetry [18]. Once AWC asymmetry is established in late embryogenesis, both the AWC ON and AWC OFF identities are maintained by cGMP signaling, dauer pheromone signaling, and transcriptional repressors [12,20,21]. unc-43(CaMKII), tir-1 (Sarm1), and nsy-1 (MAPKKK) are also implicated in the maintenance of AWC asymmetry in the first larval (L1) stage [22]. Although multiple genes were identified to be involved in the establishment and the maintenance of AWC asymmetry (for a review, see [23]), it is still unknown how the calcium-regulated signaling pathway is inhibited by nsy-4 and nsy-5 in the AWC ON cell.
The TIR-1/Sarm1 adaptor protein assembles a calciumsignaling complex, UNC-43 (CaMKII)/TIR-1/NSY-1 (ASK1 MAPKKK), at AWC synapses to regulate the default AWC OFF identity [16], thus downregulation of tir-1 expression may represent an efficient mechanism to inhibit calcium signaling in the cell becoming AWC ON . In support of this idea, a prior large scale examination of potential miRNA targets indicated that tir-1 and unc-43 may be downregulated by this class of RNAs [24]. Here, we analyze the function of the miRNA mir-71 in stochastic AWC asymmetry by characterizing its role in downregulation of the calcium signaling pathway in the AWC ON cell. We show that mir-71 acts downstream of nsy-4/claudin and nsy-5/innexin to promote AWC ON in a cell autonomous manner through inhibiting tir-1 expression, in parallel with other processes. We also show that nsy-4 and nsy-5 are required for the stability of mature mir-71. Our results suggest a mechanism for genetic control of AWC asymmetry by nsy-4 and nsy-5 through mir-71-mediated downregulation of calcium signaling.

Identification of miRNAs with predicted target genes in the AWC calcium signaling pathway
The calcium-regulated UNC-43 (CaMKII)/TIR-1 (Sarm1)/ NSY-1 (ASK1 MAPKKK) signaling pathway suppresses expres-sion of the AWC ON gene str-2 in the default AWC OFF cell [12,16,25,26]. To establish AWC asymmetry, the calciummediated signaling pathway is suppressed in the future AWC ON cell. miRNAs are small non-coding RNAs that are robust in mediating post-transcriptional and/or translational downregulation of target genes [27]. In C. elegans, miRNAs are processed from premature form into mature form by alg-1/alg-2 (encoding the Argonaute proteins) and dcr-1 (encoding the ribonuclease III enzyme Dicer) [28]. Gene expression profiling revealed increased levels of unc-43 and tir-1 in dcr-1 mutants [24], suggesting that unc-43 and tir-1 may be downregulated by miRNAs. Thus, we hypothesized that miRNAs may play a role in downregulation of the UNC-43/TIR-1/NSY-1 signaling pathway in the cell becoming AWC ON .
Since most miRNAs are not individually essential and have functional redundancy [37][38][39][40], loss-of-function mutations in a single miRNA may not show a defect in AWC asymmetry. To circumvent potential problems that may be posed by functional redundancy, we took an overexpression approach to determine the role of these six miRNAs in AWC asymmetry. We generated transgenic strains overexpressing individual miRNAs in both AWCs using an odr-3 promoter, expressed strongly in AWC neuron pair and weakly in AWB neuron pair [41]. Wild-type animals have str-2p::GFP (AWC ON marker) expression in only one of the two AWC neurons ( Figure 1A and 1E). Since loss-offunction mutations in the AWC calcium signaling genes (unc-2, unc-36, unc-43, tir-1, nsy-1, and sek-1) led to str-2p::GFP expression in both AWC neurons (2AWC ON phenotype) ( Figure 1B and 1E) [12,16,25,26], we proposed that overexpression of the miRNA downregulating one of these calcium signaling genes would also cause a 2AWC ON phenotype. We found that mir-71(OE) animals overexpressing mir-71, predicted to target tir-1 and nsy-1, had a strong 2AWC ON phenotype ( Figure 1C, 1E, and Figure S1B). This result suggests that mir-71 may downregulate the expression of tir-1 and nsy-1 to control the AWC ON fate and that mir-71 is sufficient to promote AWC ON when overexpressed. However, overexpression of the other five miRNAs individually caused a mixed weak phenotype of 2AWC ON and 2AWC OFF ( Figure S1B). Since the activity of the nsy-1 39 UTR in AWC was independent of mir-71(OE) ( Figure S2B), we focused on the investigation of the potential role of mir-71 in promoting AWC ON through negatively regulating tir-1 expression.

Author Summary
Cell identity determination requires a competition between the induction of cell type-specific genes and the suppression of genes that promote an alternative cell type.
In the nematode C. elegans, a specific sensory neuron pair communicates to establish stochastic asymmetric identities by inhibiting a calcium signaling pathway in the neuron that becomes an induced identity. However, it is not understood how cell-cell communication inhibits the calcium signaling pathway in the induced neuronal identity. In this study, we identify a microRNA that represses the expression of a key molecule in the calcium signaling pathway to promote the induced neuronal identity. Overexpression of the microRNA causes both neurons of the pair to become the induced identity, similar to the mutants that lose function in the calcium signaling pathway. In addition, the stability of the mature microRNA is dependent on a claudin-like protein and a gap junction protein, the two parallel signals that mediate communication of the neuron pair to promote the induced neuronal identity. Our results provide insight into the mechanism by which cell-cell communication inhibits calcium signaling to establish stochastic asymmetric neuronal differentiation. miRNA and Stochastic L/R Neuronal Asymmetry The genetic interaction between mir-71 and tir-1 was characterized by double mutants ( Figure 1E). tir-1(ky648) gain-of-function (gf) mutants had two AWC OFF neurons (2AWC OFF phenotype) ( Figure 1D and 1E) [22]. We found the tir-1(ky648gf) 2AWC OFF phenotype was significantly reduced in the tir-1(ky648gf); mir-71(OE) double mutants (p,0.001) ( Figure 1E). These results support the hypothesis that mir-71 downregulates tir-1 to control the AWC ON fate.
To further determine the requirement of mir-71 in AWC asymmetry, we analyzed str-2p::GFP expression in the mir-71(n4115) deletion null allele [40]. mir-71(n4115) mutants displayed wild-type AWC asymmetry ( Figure 1E), suggesting that mir-71 may function redundantly with other miRNAs or non-miRNA genes to regulate calcium signaling in AWC asymmetry. In addition to mir-71, mir-248 was also predicted to target tir-1 by three programs ( Figure S1A). mir-71 and mir-248 have different predicted target sites in the tir-1 39 UTR. Since mir-248 mutants are not available, we analyzed the effect of mir-248 overexpression on AWC asymmetry. Unlike the highly penetrant 2AWC ON phenotype caused by mir-71 overexpression, mir-248 overexpression generated a mixed weak phenotype of 2AWC ON and 2AWC OFF ( Figure S1B). To test whether mir-71 and mir-248 have a synergistic effect on AWC symmetry, we made transgenic animals overexpressing both mir-71 and mir-248 in AWCs. The 2AWC ON phenotype was not significantly higher in mir-71(OE); mir-248(OE) animals than in mir-71(OE) (data not shown). These results suggest that mir-71 may not act redundantly with mir-248 to regulate tir-1 expression in AWC asymmetry. To knockdown mir-248 expression, we made an anti-mir248 transgene expressing short hairpin RNA (shRNA), consisting of both sense and antisense sequences of mir-248, in AWC. The anti-mir-248 transgene caused an AWC phenotype similar to mir-248(OE) (data not shown), suggesting that the effect of the anti-mir-248 transgene on AWC asymmetry is not through knockdown of mir-248 but mainly due to overexpression of sense mir-248 in the shRNA construct.
The nsy-4 claudin-like gene and the unc-76 axon guidance pathway gene induce the AWC ON state by inhibiting the downstream calcium-signaling pathway. Loss-of-function mutations in nsy-4 and unc-76 cause a partially penetrant 2AWC OFF phenotype ( Figure 1E) [12,19]. mir-71(n4115) mutations significantly enhanced the 2AWC OFF phenotype of nsy-4(ky616) and unc-76(e911) mutants (p,0.001). On the other hand, the 2 AWC ON phenotype of nsy-4(OE) trasnsgenic animals overexpressing nsy-4 in AWCs was significantly suppressed in nsy-4(OE); mir-71(n4115) double mutants (p,0.001; Figure 1E). These results are consistent with a role of mir-71 function in promoting the AWC ON fate, and suggest that mir-71 may act in parallel with other regulatory molecules to antagonize the calcium-regulated signaling pathway to generate the AWC ON identity.

mir-71 inhibits tir-1 expression through its 39 UTR
The predicted mir-71 target site in the tir-1 39 UTR is 96 bp downstream of the stop codon; the prediction is strongly supported by four different programs, including MicroCosm Targets, TargetScan, PicTar, and mirWIP ( Figure S1A). The nucleotides at position 1-8 in the seed region of mir-71 perfectly match the target site of the tir-1 39 UTR; the seed match is conserved between C. elegans and C. briggsae ( Figure 2A).
To determine whether mir-71 acts directly through the predicted binding site in the tir-1 39 UTR, we made GFP sensor constructs with the AWC odr-3 promoter and different 39 UTRs: wild-type tir-1 39 UTR or the tir-1 39 UTRmut with mutated mir-71 target site ( Figure 2B). Transgenic animals expressing each sensor construct were crossed to mir-71(OE) animals. The GFP intensity of each sensor construct in an individual AWC neuron was normalized to the nucleus-localized TagRFP intensity of the transgene odr-3p::2Xnls-TagRFP::unc-54 39 UTR in the same cell. The unc-54 39 UTR does not contain any strongly predicted mir-71 sites. The normalized GFP intensity of each sensor construct was compared between mir-71(OE) animals and their siblings losing the mir-71(OE) transgene in the L1 stage, during which tir-1 is functional for the maintenance of AWC asymmetry [22]. We found that mir-71(OE) animals, compared with wild type, had a significantly reduced normalized expression level of GFP from the tir-1 39 UTR sensor construct (p,0.005; Figure 2B upper panels). However, the normalized expression level of GFP from the tir-1 39 UTRmut was not significantly different between wild-type and mir-71(OE) animals ( Figure 2B bottom panels). These results suggest that mir-71 directly inhibits gene expression through the predicted target site in the tir-1 39 UTR. However, we did not observe a significant difference in the GFP expression level from the tir-1 39 UTR between wild-type animals and mir-71(n4115lf) mutants ( Figure S2A). This result suggests potential functional redundancy of mir-71 in the regulation of tir-1 expression.
is not due to loss of AWC neurons. (E) str-2p::GFP expression phenotypes in wild type, single mutants, and double mutants. nsy-4(OE) animals overexpress the transgene odr-3p::nsy-4 in AWCs. (F) Genetic map of mir-71. mir-71 (blue arrow) is located in an intron of F16A11.3a encoding the ppfr-1 gene. Black bars indicate the location of deletions in ppfr-1(tm2180) and mir-71(n4115) mutants. A schematic of the GFP reporter gene driven by a 2.4 kb region upstream of mir-71 transcript is shown. Arrows, AWC cell body. Scale bar, 10 mm. Statistical analysis was performed using the Z-test for two proportions: *p,0.05; ***p,0.001; ns, not significant. doi:10.1371/journal.pgen.1002864.g001 miRNA and Stochastic L/R Neuronal Asymmetry GFP sensor constructs, driven by the odr-3 promoter, with the tir-1 39 UTR or the tir-1 39 UTR mutated in the predicted mir-71 target site. Middle: Images of GFP expression from GFP sensor constructs and nucleus-localized TagRFP expression from the internal control transgene odr-3p::2Xnls-TagRFP::unc-54 39 UTR in the AWC cell body of wild type and mir-71(OE) animals. All images were taken from animals in the first larval stage. Scale bar, 5 mm. Arrows, AWC cell body. Right: The average normalized GFP intensity of each sensor construct in the AWC cell body. The GFP intensity of an individual cell was normalized to the TagRFP intensity of the same cell. For each sensor construct line, the normalized GFP intensity in wild type was set as 1 arbitrary unit (AU) and the normalized GFP intensity in mir-71(OE) was calibrated to that in wild type. Student's t-test was used for statistical analysis. n = 16-21 for each transgenic line in wild type and mir-71(OE) animals. Error bars, standard error of the mean. ns, not significant. (C) Left: tir-1 overexpression constructs, driven by the odr-3 promoter, with the tir-1 39 UTR or the tir-1 39 UTR mutated in the predicted mir-71 target site. Right: Normalized fold change in tir-1(OE) 2AWC OFF phenotype. The fold change in tir-1(OE) 2AWC OFF phenotype was determined by dividing the 2AWC OFF percentage of tir-1(OE) with the 2AWC OFF percentage of tir-1(OE); mir-71(OE), which was then normalized to the relative tir-1(OE) transgene copy number. Two to three independent lines were analyzed for each tir-1 overexpression construct. Student's t-test was used to calculate statistical significance. Error bars represent standard error of the mean. doi:10.1371/journal.pgen.1002864.g002 Interactions between the 59 and 39 UTRs have been shown to regulate translation in mammalian cells [44], bacteria [45], and RNA viruses [46]. To determine if the tir-1 59 UTR plays a role in regulating the inhibitory effect of mir-71 on the tir-1 39 UTR, we included the tir-1 59 UTR in the GFP sensor constructs ( Figure  S3). Similar to the tir-1 39 UTR sensor constructs without the tir-1 59 UTR ( Figure 2B), the normalized expression level of GFP from the tir-1 59 UTR/tir-1 39 UTR sensor construct was significantly decreased in mir-71(OE) animals compared with wild type (p,0.04; Figure S3A). However, the normalized expression level of GFP from the tir-1 59 UTR/tir-1 39 UTRmut sensor construct was not significantly different between wild-type and mir-71(OE) animals ( Figure S3B). These results suggest that the tir-1 59 UTR does not affect mir-71(OE)-mediated suppression of gene expression through the tir-1 39 UTR.
The nsy-1 39 UTR was also predicted to contain a mir-71 binding site by the four programs used in this study ( Figure S1A), but the GFP expression level from the nsy-1 39 UTR was not significantly different between wild-type and mir-71(OE) animals ( Figure S2B). This result suggests that the predicted mir-71 site in the nsy-1 39 UTR may not be functional in AWC cells, therefore we did not further investigate the regulation of nsy-1 expression by mir-71.
tir-1(OE) animals overexpressing tir-1 in AWC had a 2AWC OFF phenotype [16]. We used the tir-1(OE) 2AWC OFF phenotype as readout to determine if mir-71 acts through the tir-1 39 UTR to suppress the AWC OFF fate. We made tir-1(OE) sensor constructs by replacing GFP in the GFP sensor constructs ( Figure 2B) with tir-1 and crossed transgenic animals expressing each tir-1(OE) sensor construct into mir-71(OE) animals ( Figure 2C). The fold change in tir-1(OE) 2AWC OFF phenotype was determined by dividing the 2AWC OFF percentage of tir-1(OE) animals with the 2AWC OFF percentage of their tir-1(OE); mir-71(OE) siblings, which was then normalized to the relative tir-1(OE) transgene copy number determined by qPCR. The higher normalized fold change in tir-1(OE) 2AWC OFF indicates more suppression of 2AWC OFF phenotype by mir-71(OE) in tir-1(OE); mir-71(OE) animals. The normalized fold change in tir-1(OE) 2AWC OFF of tir-1 39 UTR was significantly higher than that of the tir-1 39 UTRmut (p = 0.03; Figure 2C). These results suggest that mir-71 suppresses the AWC OFF fate by downregulating tir-1 expression through its 39 UTR.

mir-71 is expressed at a higher level in the AWC ON cell than in the AWC OFF cell
To determine if mir-71 is expressed in AWC neurons, we generated transgenic animals expressing YFP under the control of a 2.4 kb promoter upstream of the mir-71 transcript ( Figure 1F). The expression of YFP was detected in several head neurons and the body wall muscle in L1 ( Figure 3A), which is consistent with previously reported expression pattern of mir-71 [47][48][49]. The mir-71p::YFP transgenic animals were crossed into an odr-1p::DsRed strain, expressing DsRed primarily in AWC and AWB neurons ( Figure 3B). YFP was coexpressed with DsRed in AWC and AWB neurons ( Figure 3C), suggesting that mir-71 is expressed in these neurons. We found that 52% of animals had visible mir-71p::YFP in both AWC cells, 28% had visible YFP in only AWC left (AWCL), and 20% had visible YFP in only AWC right (AWCR) ( Figure 3D). These results suggest that the expression of mir-71, when detected in one of the two AWC neurons, does not have a side bias towards AWCL or AWCR, which is consistent with stochastic choice of the AWC ON fate.
We then investigated whether mir-71, when detected in both AWC neurons, has differential expression levels between AWC ON and AWC OFF . Transgenic animals expressing mir-71p::GFP, ceh-36p::myr-TagRFP (myristoylated TagRFP marker of AWC ON and AWC OFF ), and str-2p::2Xnls-TagRFP (nucleus-localized TagRFP marker of AWC ON ) were generated and analyzed in the L1 stage ( Figure 4A, 4A9, 4A0, 4B, 4B9, and 4B0). The ceh-36 promoter is expressed in AWCL, AWCR, ASEL, and ASER [50,51]. mir-71p::GFP expression was significantly higher in the AWC ON cell than in the AWC OFF cell in 71% of the animals (p,0.001; Figure 4C). To confirm this result, we generated transgenic animals expressing mir-71p::NZGFP, odr-3p::CZGFP, and str-2p::2Xnls-TagRFP in which reconstituted GFP (recGFP) [52] expression from two split GFP polypeptides, NZGFP and CZGFP, was restricted mainly in the two AWC cells. Consistent with the mir-71p::GFP result, recGFP expression was significantly higher in the AWC ON cell than in the AWC OFF cell in 81% of the animals (p,0.001; Figure S4). Together, these results suggest that mir-71 is expressed at a higher level in the AWC ON than in the AWC OFF cell. The higher expression of mir-71 in the AWC ON cell is consistent with the role of mir-71 in promoting the AWC ON fate.

tir-1 expression is downregulated through its 39 UTR in the AWC ON cell
The suppression of gene expression by mir-71 through the tir-1 39 UTR (Figure 2B and 2C) and the role of mir-71 in promoting the AWC ON fate ( Figure 1C and 1E) suggest that gene expression Normalized GFP expression was determined by calibrating GFP intensity with 2Xnls-TagRFP intensity of the same cell. All constructs, except for odr-3p::GFP::tir-1 39 UTR, contain the unc-54 39 UTR. All images were taken from first stage larvae. The single focal plane with the brightest fluorescence in each AWC was selected from the acquired image stack and measured for fluorescence intensity. Each animal was categorized into one of three categories: AWC ON = AWC OFF , AWC ON .AWC OFF , and AWC OFF .AWC ON based on the comparison of GFP intensities between AWC ON and AWC OFF cells of the same animal. We did not observe any animals that fell into the ''AWC ON = AWC OFF '' category from our GFP intensity analysis. Total number of animals for each category was tabulated and analyzed as described [86]. p-values were calculated using X 2 test. Error bars represent standard error of proportion. Arrows indicate the AWC cell bodies. Arrowheads represent myr-TagRFP or myr-mCherry signal. Scale bar, 2 mm. doi:10.1371/journal.pgen.1002864.g004 through the tir-1 39 UTR may be downregulated in the AWC ON cell. To investigate this possibility, transgenic animals expressing odr-3p::GFP::tir-1 39 UTR (GFP reporter of the tir-1 39 UTR regulation in both AWCs), odr-3p::2Xnls-TagRFP::unc-54 39 UTR (nucleus-localized TagRFP marker of both AWC ON and AWC-OFF ), and str-2p::myr-mCherry (myristoylated mCherry marker of AWC ON ) were generated and analyzed in the L1 stage ( Figure 4D, 4D9, 4D0, 4E, 4E9, and 4E0). The GFP intensity was normalized to the nucleus-localized TagRFP intensity measured in the same AWC cell to account for variation in focal plane and promoter activity. Normalized GFP intensity was significantly lower in the AWC ON cell than in the AWC OFF cell in more than 85% of the animals (p,0.02; Figure 4F). These results suggest that the expression of tir-1 is downregulated in the AWC ON cell, consistent with a higher expression level of mir-71 in AWC ON and downregulation of tir-1 expression by mir-71.

mir-71 acts cell-autonomously to promote the AWC ON identity
To determine the site of mir-71 action, mosaic animals in which the two AWC neurons have differential mir-71 activity were used to ask whether mir-71 acts in the future AWC ON cell or the future AWC OFF cell. Mosaic animals were generated by random and spontaneous mitotic loss of an unstable transgene expressing the mir-71(OE) construct odr-3p::mir-71 and a mosaic marker odr-1p::DsRed that showed which AWC cells retained the transgene. We specifically looked for the mosaic animals in which only one of the two AWC neurons expressed the mir-71(OE) transgene; this cell was identified by expression of the DsRed marker.
Mosaic analysis was first performed in transgenic lines expressing the mir-71(OE) transgene in a wild-type background. Expression of the mir-71(OE) transgene in both AWC neurons resulted in a 2AWC ON phenotype ( Figure 5A and 5C). When the mir-71(OE) transgene was retained in only one of the two AWC neurons, the mir-71(OE) AWC neuron became AWC ON and wildtype AWC neuron became AWC OFF in the majority of these mosaic animals (p,0.0001; Figure 5B and 5D). This result is consistent with a significant cell-autonomous requirement for mir-71 in the AWC ON cell to regulate its identity, which is opposite to the cell autonomous function of tir-1 in regulation of the AWC OFF identity. This result suggests that the AWC cell with higher mir-71 activity can prevent the contralateral AWC cell from becoming AWC ON and that mir-71 may play a role in a negative-feedback signal sent from pre-AWC ON to pre-AWC OFF . Similar results were obtained from previous mosaic analysis of nsy-4 and nsy-5 [18,19].

Discussion
Stochastic cell fate acquisition in the nervous system is a conserved but poorly understood phenomenon [1]. Here, we report that the miRNA mir-71 is part of the pathway that controls stochastic left-right asymmetric differentiation of the C. elegans AWC olfactory neurons through downregulating the expression of tir-1, encoding the TIR-1/Sarm1 adaptor protein in a calcium signaling pathway. In addition, we have linked NSY-4/claudin-and NSY-5/innexin-dependent stability of mature mir-71 to downregulation of calcium signaling in stochastic AWC neuronal asymmetry. Previous studies have identified the role of miRNAs in reproducible, lineage-based asymmetry of the C. elegans ASE taste neuron pair, in which the miRNA expression pattern is largely fixed along the left-right axis [8,9,55]. This study provides one of the first insights into miRNA function in stochastic left-right asymmetric neuronal differentiation, in which the miRNA expression pattern is not fixed and is likely regulated by the stochastic signaling event driving random asymmetry.
The seed match between mir-71 and the tir-1 39 UTR is conserved between C. elegans and C. briggsae. However, the str-2 promoters share little sequence similarity between C. elegans and C. briggsae. The C. elegans str-2 promoter GFP reporter, when expressed in C. briggsae, does not show detectable GFP expression in AWC neurons in embryos, first stage larvae, or adults (data not shown). This result suggests that the transcriptional regulation of str-2 has diverged in C. briggsae. mir-71 has been implicated in various cell biological and developmental processes including promotion of longevity, resistance to heat and oxidative stress, DNA damage response, control of developmental timing, dauer formation, and recovery from dauer [47,[56][57][58][59][60]. However, it is largely unknown how mir-71 functions to regulate these biological processes. RNA interference (RNAi) of tir-1 did not affect C. elegans longevity [61], suggesting that mir-71 may regulate distinct target genes for different functions.
miRNAs are important post-transcriptional and translational regulators of gene expression during development and disease. Several miRNA target prediction algorithms such as MicroCosm Targets, TargetScan, PicTar, and mirWIP provide useful tools with which to identify potential target genes of miRNAs [62]. However, many miRNAs have redundant functions and therefore give subtle or no phenotypes when mutated [37][38][39][40]. Overexpression approach or phenotypic analysis of miRNA mutants in sensitized genetic backgrounds have been successful in elucidating the role of miRNAs for which null mutants are not available or functional redundancy is a potential problem [5,8,[38][39][40]42,[63][64][65]. Using miRNA target prediction programs, we identified mir-71 and five other miRNAs as potential regulators of the calcium-regulated UNC-43 (CaMKII)/ TIR-1/NSY-1 (MAPKKK) signaling pathway. Through an overexpression approach and functional analysis of mir-71(n4115) mutants in sensitized genetic backgrounds, we revealed the role of mir-71 in genetic control of the AWC ON identity. miRNAs that share the same sequence identity in their seed regions and could be potentially capable of downregulating the same set of target genes are grouped as members of a family [66][67][68][69]. Some miRNA family members have been shown to function redundantly and work together to regulate specific developmental processes [37,38,[70][71][72][73][74]. However, many families of miRNAs did not show synthetic phenotypes, indicating that most miRNA families act redundantly with other miRNAs, miRNA families, or non-miRNA genes [38]. Since there is only one mir-71 family member identified, the absence of an AWC phenotype in mir-71(n4115) single mutants suggests that mir-71 may act redundantly with other miRNA family members or non-miRNA genes to regulate calcium signaling in AWC asymmetry. dcr-1, encoding the ribonuclease III enzyme Dicer, is required for processing of premature miRNAs to mature miRNAs [28]. dcr-1(ok247) null mutants had wild-type AWC asymmetry (data not shown). This result suggests that the dcr-1 mutation may cause simultaneous knockdown of several miRNAs (including mir-71) with opposite functions in AWC asymmetry, thereby masking the role of mir-71 and its redundant miRNAs in AWC asymmetry.
The UNC-76 axon guidance molecule and NSY-4 claudin-like protein act to antagonize the calcium-regulated signaling pathway to generate the AWC ON identity [12,19]. We found that mir-71(n4115) mutants significantly suppressed the 2AWC ON phenotype of nsy-4(OE) and enhanced the 2AWC OFF phenotype of nsy-4(ky627) and unc-76(e911) mutants. These results suggest an alternative mechanism for functional redundancy of mir-71 in AWC asymmetry. mir-71 may act in parallel with other regulatory pathways downstream of unc-76 and nsy-4 to downregulate the calcium signaling pathway in the AWC ON cell. Functional redundancy of miRNAs and other regulatory pathways has been demonstrated by a previous study suggesting that Drosophila miR-7 may act in parallel with a protein-turnover mechanism to downregulate the transcriptional repressor Yan in the fly eye [42].
Our results suggest that mir-71 is regulated at transcriptional and post-transcriptional levels in AWC. At the transcriptional level, mir-71 is expressed at a higher level in the AWC ON cell than in the AWC OFF cell. This transcriptional bias of mir-71 is not dependent on NSY-4 claudin-like protein or NSY-5 innexin gap junction protein. The mechanisms that regulate differential expression of mir-71 in the two AWC cells are yet to be elucidated. At the post-transcriptional level, the stability of mature mir-71 is dependent on nsy-4 and nsy-5. It is possible that nsy-4 and nsy-5 may antagonize the miRNA turnover pathway to increase the level of mature mir-71. The C. elegans 59R39 exoribonuclease XRN-2 has been implicated in degradation of mature miRNAs released from Argonaute [75]. However, xrn-2(RNAi) animals did not show AWC phenotypes (data not shown), suggesting that the stability of mature mir-71 may be independent of xrn-2.
The TIR-1/Sarm1 adaptor protein assembles a calciumregulated signaling complex at synaptic regions to regulate the default AWC OFF identity [16]. Downregulation of the TIR-1 adaptor protein by mir-71 and other parallel pathways may represent an efficient mechanism to inhibit calcium signaling in the cell becoming AWC ON . Calcium signaling is one of the most common and conserved systems that control a wide range of processes including fertilization, embryonic pattern formation, cell proliferation, cell differentiation, learning and memory, and cell death during development and in adult life [76]. In addition, calcium signaling is implicated in left-right patterning in several tissues of different organisms [77]. It has been shown that negative regulation of calcium signaling by miRNAs is important for normal development and health [78][79][80][81]. In summary, our study and the studies from other labs demonstrate that downregulation of calcium signaling by miRNAs is one of the important mechanisms for cellular and developmental processes.
Transgenes maintained as extrachromosomal arrays include kyEx1127

Quantification of fluorescence intensity
Z-stack images of transgenic animals expressing fluorescent markers were acquired using a Zeiss Axio Imager Z1 microscope equipped with a motorized focus drive and a Zeiss AxioCam MRm CCD digital camera. All animals of each set of experiments had the same exposure time for comparison of fluorescence intensity. The single focal plane with the brightest fluorescence in each AWC cell was selected from the acquired image stack and measured for fluorescence intensity. To measure fluorescence intensity, the outline spline tool in the Zeiss AxioVision Rel 4.7 image analysis software was used to draw around the AWC cell body ( Figure 2B; Figure 4A, 4B, 4D, 4E; Figure S4A, S4B; and Figure S6B) or nucleus ( Figure 2B, Figure 4D9 and 4E9) from captured images. To measure fluorescence intensity in dim GFPexpressing cells ( Figure 4B and Figure S4B), the display contrast and brightness were adjusted to visualize and outline the cells. For each category of animals, images from a minimum of 10 animals were collected and analyzed.
qPCR for determining the relative transgene copy number Three adult hermaphrodites from each tir-1(OE) transgene line were collected in 25 ml of worm lysis buffer (50 mM KCl, 0.01% gelatin, 10 mM Tris-HCl pH 8.3, 0.45% Tween 20, 0.45% NP-40, 2.5 mM MgCl 2 , 100 mg/ml Proteinase K). Collected worms were then incubated at 280uC for minimum of one hour, 65uC for one hour, and 95uC for 15 minutes. 5 ml of the worm lysate was used for subsequent qPCR with Fast SYBR Green Master Mix (Invitrogen). qPCR reactions were run in triplicate at 95uC for 3 minutes, followed by 45 cycles of 95uC for 30 seconds, 57uC for 30 seconds, and 72uC for 30 seconds on the CFX96 Real-Time PCR Detection System (Bio-Rad). PCR product was scanned for fluorescent signal at the end of each cycle and the C(T) values were obtained using the CFX Manager Software (Bio-Rad). The relative tir-1(OE) transgene copy number was determined using the 2[2Delta Delta C(T)] method as previously described [85] with the actin-related gene, arx-1, as internal control.

Stem-loop RT-qPCR of premature and mature mir-71
Stem-loop RT-qPCR was performed as described [54] to detect and quantify relative expression levels of premature and mature mir-71. The odr-3p::mir-71 transgenes used in genetic mosaic analysis were crossed into various genetic backgrounds. Total RNA samples were isolated from first stage larvae using RNeasy Mini kit (QIAGEN). Reverse transcription (RT) reactions were performed with 1 mg of total RNA, SuperScript III reverse transcriptase (Invitrogen), and RT primer (oligo d(T) 18 , premature mir-71 stem-loop RT primer, or mature mir-71 stem-loop RT primer). 1 ml of 1:35 diluted reverse transcription product was used as template for subsequent qPCR reactions with Fast SYBR Green Master Mix (Invitrogen). All PCR reactions were run in triplicate at 95uC for 3 minutes, followed by 45 cycles of 95uC for 30 seconds, 51uC for 30 seconds, and 72uC for 30 seconds on the CFX96 Real-Time PCR Detection System (Bio-Rad). PCR product was scanned for fluorescent signal at the end of each cycle and the C(T) values were obtained using the CFX Manager Software (Bio-Rad). The actin-related gene, arx-1, was used as internal control to normalize variation between samples. Relative expression of premature and mature mir-71 was analyzed using the 2[2Delta Delta C(T)] method as previously described [85]. Relative expression was set to one for mir-71(n4115); odr3p::mir-71 and was normalized accordingly for other samples. Student's t-test was used to calculate statistical significance.  Figure S3 The tir-1 59 UTR does not affect mir-71(OE)mediated downregulation of gene expression through the tir-1 39 UTR. (A, B) The average normalized GFP intensity in the AWC cell body of sensor constructs, driven by the odr-3 promoter and the tir-1 59 UTR, with the tir-1 39 UTR (A) or the tir-1 39 UTR mutated in the predicted mir-71 target site (B), in wild type and mir-71(OE) animals. The GFP intensity of an individual cell was normalized to the TagRFP intensity of the internal control transgene odr-3p::2Xnls-TagRFP::unc-54 39 UTR in the same cell in the first larval stage. For each sensor construct, the normalized GFP intensity in wild type was set as 1 arbitrary unit (AU) and the normalized GFP intensity in mir-71(OE) was calibrated to that in wild type. Two independent lines were analyzed for each sensor construct. Student's t-test was used for statistical analysis. Error bars, standard error of the mean. ns, not significant. (TIF) Figure S4 The expression level of mir-71 is higher in the AWC ON cell than in the AWC OFF cell. (A, B) Images of recGFP expressed from mir-71p::NZGFP and odr-3p::CZGFP. (A9, B9) Images of str-2p::2Xnls-TagRFP. AWC ON was identified as str-2p::2Xnls-TagRFP positive (A9). AWC OFF was identified as str-2p::2Xnls-TagRFP negative (B9). (A0) Merge of A and A9 images from the same cell. (B0) Merge of B and B9 images from the same cell. (C) Quantification of recGFP expression in AWC ON and AWC OFF cells. All images were taken from first stage larvae. The single focal plane with the brightest fluorescence in each AWC was selected from the acquired image stack and measured for fluorescence intensity. Each animal was categorized into one of three categories: AWC ON = AWC OFF , AWC ON .AWC OFF , and AWC OFF .AWC ON based on the comparison of recGFP intensities between AWC ON and AWC OFF cells of the same animal. We did not observe any animals that fell into the ''AWC ON = AWC-OFF '' category from our recGFP intensity analysis. Total number of animals for each category was tabulated and analyzed as described [86]. p-values were calculated using X 2 test. Error bars represent standard error of proportion. Scale bar, 2 mm. The actin-related gene arx-1 was used as internal control to normalize the abundance of mature mir-71. All PCR reactions were run in triplicate. p values were calculated using Student's ttest. ns, not significant. Error bars represent standard error of the mean. (TIF) Figure S6 Control experiments to demonstrate that a decreased level of mature mir-71 in nsy-4(ky627) and nsy-5(ky634) mutants is not caused by a reduced transmission rate of the odr-3p::mir-71 extrachromosomal array or reduced activity of the odr-3 promoter.

Supporting Information
(A) Transmission rates of the odr-3p::mir-71 extrachromosomal array in mir-71(n4115), mir-71(n4115);nsy-4(ky627), and mir-71(n4115);nsy-5(ky634) mutants. Error bars represent the standard error of proportion. (B) Top: Representative images of odr-3p::GFP expression in AWC neurons of wild type, nsy-4(ky627), and nsy-5(ky634) mutants at the first larval stage. Bottom: The average intensity of GFP in AWC neurons. Results from two independent odr-3p::GFP transgenic lines are shown. Error bars represent standard error of the mean. Scale bar, 10 mm. (TIF) Figure S7 Differential expression of mir-71 in the two AWC cells is not dependent on nsy-4 or nsy-5. The GFP intensity of mir-71p::GFP was compared between the two AWC cells of the same animal in wild-type, nsy-4(ky627), and nsy-5(ky634) mutants. The percentage difference of mir-71p::GFP expression between the two AWC cells was determined by dividing the higher GFP intensity with the lower GFP intensity. Error bars represent the standard error of proportion. (TIF) Text S1 Supplemental Methods: Quantification of mature mir-71 by stem-loop RT-PCR. (DOCX)