miR-210 controls the evening phase of circadian locomotor rhythms through repression of Fasciclin 2

Circadian clocks control the timing of animal behavioral and physiological rhythms. Fruit flies anticipate daily environmental changes and exhibit two peaks of locomotor activity around dawn and dusk. microRNAs are small non-coding RNAs that play important roles in post-transcriptional regulation. Here we identify Drosophila miR-210 as a critical regulator of circadian rhythms. Under light-dark conditions, flies lacking miR-210 (miR-210KO) exhibit a dramatic 2 hrs phase advance of evening anticipatory behavior. However, circadian rhythms and molecular pacemaker function are intact in miR-210KO flies under constant darkness. Furthermore, we identify that miR-210 determines the evening phase of activity through repression of the cell adhesion molecule Fasciclin 2 (Fas2). Ablation of the miR-210 binding site within the 3’ UTR of Fas2 (Fas2ΔmiR-210) by CRISPR-Cas9 advances the evening phase as in miR-210KO. Indeed, miR-210 genetically interacts with Fas2. Moreover, Fas2 abundance is significantly increased in the optic lobe of miR-210KO. In addition, overexpression of Fas2 in the miR-210 expressing cells recapitulates the phase advance behavior phenotype of miR-210KO. Together, these results reveal a novel mechanism by which miR-210 regulates circadian locomotor behavior.

These neurons can be further divided into 7 clusters based on their cell localization and neurotransmitters they express [4]. There are 3 groups of dorsal neurons (DN1, DN2, and DN3), 2 groups of ventral-lateral neurons (large and small LNv), dorsal lateral neurons (LNd), and lateral posterior neurons (LPN) [4]. The large LNvs (lLNvs) and 4 small LNvs (sLNvs) express the neuropeptide Pigment Dispersing Factor (PDF), while the 5 th sLNv is PDF negative. Under regular light-dark (LD) conditions, flies exhibit a bimodal pattern of locomotor rhythms, peaking around dawn and dusk, which are termed morning peak and evening peak, respectively. According to the dual-oscillator model, separate morning and evening oscillator track these behavior peaks separately [5].
Indeed, studies have shown that the PDF-positive sLNvs are responsible for promoting the morning peak, while the LNds, as well as the fifth sLNv, are mainly responsible for generating the evening activity [6,7]. In addition, the DN1s are important to integrate environmental inputs such as light and temperature, and regulate circadian rhythms and sleep behavior [8,9]. A recent study indicated that DN1s sense acute temperature fluctuations to modulate fly sleep [10].
The PDF positive sLNvs are the master pacemaker neurons in fly brain: they synchronize other circadian neurons to maintain robust circadian rhythms under constant darkness [11]. These sLNvs send axonal projections toward the dorsal brain region, where the DN1s and DN2s are located [12,13]. The dorsal projections of sLNvs exhibit circadian arborization rhythms with higher complexity of axon terminals found in the early day and lower complexity at the night [14]. The physiology relevance and molecular mechanisms underlying this structural plasticity of sLNvs remain unclear.
Recently, the cell adhesion molecule Fasciclin 2 (Fas2) has been found to regulate the arborization rhythms of sLNvs [15]. In addition, two matrix metalloproteinases, MMP1 and MMP2 as well as the pacemaker protein Vrille were shown to be required for the structural remodeling control of sLNv projections [16,17].
Circadian rhythms are generated by an intracellular molecular clock, which is conserved across the animal kingdom [18]. The core of this molecular pacemaker is a negative transcriptional-translational feedback loop [19][20][21]. In Drosophila, CLOCK (CLK) dimerizes with CYCLE (CYC) to activate rhythmic transcription of hundreds of genes through the E-box region [22,23]. Among these clock-controlled genes, PERIOD (PER) and TIMELESS (TIM) are the key transcription repressor: they form heterodimers in the cytoplasm, and then enter into the nucleus to block their own transcription. The abundance of PER and TIM is tightly regulated by a series of post-translational modifications such as phosphorylation, glycosylation and ubiquitination [24][25][26][27]. Recent research has revealed an abundance of post-transcriptional regulation of circadian rhythms by both RNA binding proteins and miRNAs [28][29][30].
miRNAs are small non-coding RNAs, which repress target gene expression through mRNA degradation and/or translation inhibition, thus play crucial roles in posttranscriptional regulation [31]. Recent studies have uncovered functions of the miRNA biogenesis pathway and specific miRNAs in the regulation of different aspects of animal circadian rhythms [30,[32][33][34]. For instance, mouse miR-132/212 modulates the seasonal adaptation and circadian entrainment to day length [35]. miRNAs also target the molecular clock to control period length or rhythmicity of circadian locomotor activity [36]. In Drosophila, the miRNA bantam regulates circadian locomotor period by repressing clk, while miR-276a and let-7 inhibit the clock genes timeless and clockwork orange to modulate circadian rhythms respectively [32,37,38]. Moreover, various circadian output pathways are controlled by miRNAs. The miR959-964 cluster miRNAs regulate the circadian timing of feeding and immune response, while miR-279 modulates circadian locomotor behavior output [39,40]. Recently, we and others have demonstrated that miR-124 specifically controls the phase of circadian locomotor rhythms under constant darkness [41,42].
Previously we identified that depletion of GW182, the key protein for miRNA function, affects PDF-receptor signaling and disrupts circadian rhythms [34]. To further understand the roles of miRNA, we performed a genome-wide screen for miRNAs controlling circadian rhythm phenotypes. Here we show that a conserved miRNA, miR-210, regulates circadian rhythms in Drosophila. miR-210 determines the phase of evening anticipatory behavior through inhibition of Fas2.

A genetic screen of miRNA mutants identifies miR-210 as a regulator of circadian behavior
To identify miRNAs that regulate circadian rhythms, we screened the circadian locomotor rhythms of available miRNA mutants from Bloomington stock center.
Circadian behavior of 46 lines in total was observed both under constant darkness (DD) and light dark (LD) cycles. Most of the miRNA mutants we tested exhibited normal circadian rhythms under DD; though, a few miRNAs had a slightly altered period or rhythmicity (Table S1). We then closely examined for the locomotor behavior under LD.
Interestingly, we found that the miR-210 mutant (miR-210 KO ) clearly advances the phase of the evening anticipatory peak about 2 hours ( Figure 1A, 1B). The miR-210 KO was generated by knocking-in a GAL4 cDNA in replacement of the endogenous miR-210 sequence [43]. To rule out the possibilities of off-target or genetic background effects,  [44]. In miR-210 KO , the overall activity was decreased under TC, however, we still observed an additional 2-hour phase advance compared to the controls ( Figure 1C, 1D). Furthermore, we were able to rescue this behavior defect. These data confirmed that miR-210 is a clear regulator of evening phase during entrainment.
The phase advance of evening anticipation in miR-210 KO is reminiscent of the pdf mutant phenotype [11]. To test the genetic interaction of miR-210 and the PDF pathway, we generated miR-210 KO ; pdf 01 double mutants. Under 12:12 LD conditions, the phase advance of the evening peak in miR-210 KO was indistinguishable from the pdf 01 mutants ( Figure S1A). No additive effect on the phase advance was observed in the miR-210 KO ; pdf 01 double mutants comparing with miR-210 KO or pdf 01 ( Figure S1A, S1B). To observe the phenotype even more clearly, we tested these flies under long photoperiods with 16:8 LD cycle so that we avoid acute effects of the light-off transition on the shape of the evening peak. As observed in pdf 01 mutant, miR-210 KO advanced the phase of evening anticipation by about 4 hours ( Figure S1C, S1D). Again, no additive phase advance was observed in the miR-210 KO ; pdf 01 double mutants ( Figure S1D). These data suggest that miR-210 functions in the same pathway as PDF in regulation of evening phase under entrainment.
To further test the role of miR-210, we first overexpressed it in all circadian tissues using tim-GAL4 [45]. However, most of the flies with overexpression died rapidly after the LD cycles, which indicates that overexpression leads to severe health issues (Table 1). Thus, we then used the pdf-GAL4 to specifically drive miR-210 expression in PDF positive pacemaker neurons. The majority of flies (~63%) with miR-210 overexpression in PDF neurons became arrhythmic under DD. However, for the 37% rhythmic flies, the circadian period was lengthened about 2 hours. Thus, while loss of miR-210 has no effect on the period and amplitude of circadian rhythms in DD, overexpression is detrimental to the proper function of the circadian oscillator in PDF neurons.

The molecular pacemaker is intact in the miR-210 KO flies
Since circadian locomotor rhythms are unaffected in miR-210 KO flies under DD, it is likely that the molecular pacemaker in circadian neurons of miR-210 KO flies is functional. To confirm this, we examined the oscillation of the key pacemaker protein PER in DD. We first focused on the PDF positive sLNvs, which are the master pacemaker neurons under constant conditions. As expected, we found no obvious changes in PER oscillation or abundance in these sLNvs ( Figure S2A, S2B). As an important output molecule for circadian rhythms, the PDF abundance was not affected in miR-210 KO flies ( Figure S2C). We further examined PER levels in another two important groups of circadian neurons: LNds, and DN1s. Similar as in sLNvs, PER oscillation in these neurons were not affected ( Figure S3). Taken together, these data suggest that loss of miR-210 has no effects on the molecular clock.

Loss of miR-210 disrupts the circadian arborization rhythms of sLNv dorsal projections
Although genetic interactions between miR-210 and the PDF pathway was identified ( Figure S1), we observed no significant changes in PDF abundance in the flies missing miR-210 ( Figure S2C). Thus, we examined the projections of the sLNvs.
Circadian arborization rhythms of the dorsal projections of sLNvs have been observed [14,15]. To determine whether miR-210 affects the arborization rhythms of sLNv projections, we used a PDF-specific antibody to examine the termini of sLNv dorsal projections in miR-210 KO flies at early day (Zeitgeber time 2 (ZT2), ZT0 is light on and ZT12 is light off) and early night (ZT14). We found that the wild-type flies had more PDF positive branches of axon terminals at ZT2 than at ZT14 (Figure 2A-2C), as previously observed [15]. Remarkably, this arborization rhythm was abolished in the miR-210 KO flies ( Figure 2A-2C). This is mainly due to the significant decrease of axonal crosses at ZT2 compared to control flies ( Figure 2A). Restoration of miR-210 expression rescued the arborization rhythm in miR-210 KO mutants. Moreover, we did not detect significant changes of PDF abundance in the sLNv soma, indicating that the decrease of axonal branches was not due to the decrease of PDF staining (see Figure S2). Together, these results indicate that miR-210 is required for the circadian arborization rhythms in the sLNv dorsal projections.  , is critical for miRNA function [46], to minimize potential off-target effects, we managed to delete only 7 bps of the matching sequence on the 3'UTR ( Figure 3B). We recovered one line, which is homozygous viable, and named the mutant: Fas2 ΔmiR-210 .
Next we tested the circadian locomotor behavior of the Fas2 ΔmiR-210 . There was no change in period under DD (Table 1) Fas2 is a functional target of miR-210.

miR-210 regulates circadian rhythms through inhibition of Fas2
To identify the cellular mechanism of miR-210 functions, we firstly characterized the miR-210 expression pattern in the fly brain by using the miR-210 knock-in Gal4 to drive GFP as a reporter. We observed GFP signals in the optic lobe, Hofbauer-Buchner eyelet (H-B eyelet), as well as several other brain regions, including the mushroom bodies and antennal lobe ( Figure 4A). To our surprise, no GFP signal was detected in known circadian neurons as labeled by anti-PER antibody staining. Then we examined Fas2 expression using the commercial antibodies (1D4) from DSHB. As previously reported, in wild type flies Fas2 was strongly expressed in the mushroom bodies as well as the ellipsoid bodies ( Figure 4B), [47]. Interestingly, we detected an intensive Fas2 signal in the optic lobe of miR-210 KO mutant flies, which disappeared when miR-210 expression was rescued. In addition, elevated abundance of Fas2 was also observed in the optic lobe of Fas2 ΔmiR-210 ( Figure 4B-4D). Lastly, the increased Fas2 signal clearly overlapped with miR-210 expression in the optic lobe ( Figure 4E).
Although we did not detect miR-210 expression in circadian neurons with the miR-210 knock-in Gal4, the single cell RT-PCR data from Chen and Rosbash indicated that miR-210 has an oscillating expression in the sLNvs [48]. Thus, we first tested the requirement of miR-210 in the sLNvs for normal circadian behavioral rhythms under LD, combining miR-210 GAL4 and Pdf-GAL80. GAL80 is a repressor of GAL4 function, and Pdf-GAL80 has been efficiently used to inhibit gene expression in the PDF neurons [7,49]. Surprisingly, we observed a clear rescue of evening phase when we excluded miR-210 expression in the sLNvs, which suggests that miR-210 is not necessary in PDF neurons for circadian behavior ( Figure 5A). Next, we tested whether overexpression of

Fas2 in the miR-210-expression cells would mimic the behavior phenotype of miR-210 KO
mutants. We observed a weaker (~1 hr) but significant advance of evening peak in Fas2 overexpression ( Figure 5B-5C). Consistent with the miR-210 data, Fas2 is not required in the PDF positive neurons ( Figure 5).

Discussion
Emerging roles of miRNA in the control of different aspects of circadian rhythms have been recently uncovered [33]. Here, we screen Drosophila miRNA mutants and demonstrate that miR-210 is critical for circadian locomotor rhythms. miR-210 determines the proper phase of evening activity peak under entrainment. miR-210 plays its role in circadian rhythms and axonal arborization via repression of Fas2.
Drosophila gradually increases its locomotor behavior and reaches its peak of evening activity at light off. The molecular mechanism underlying the phase control of evening peak is still largely unknown. Loss of miR-210 advanced the evening phase to the same extent of pdf 01 mutants. In fact, our genetic interaction data suggests that miR-210 functions in the same signaling pathway as pdf. However, unlike the pdf 01 mutants, loss of miR-210 has no effect on morning anticipation or circadian rhythmicity. These results indicate that miR-210 may specifically regulate one aspect of pdf signaling functions.
Circadian structural plasticity of sLNv dorsal projections has been recently identified [50,51]. The biological function and molecular mechanisms underlying the structural plasticity however remain elusive. Here we identified that miR-210 KO  First, as far as we know, previous genetic manipulations that cause severe defects on sLNv structural plasticity have no effect on evening phase [50,51]. Second, although overexpression or down regulation of Fas2 in the sLNv abolished the arborization rhythms, it does not affect the evening phase either [15]. So we conclude that these effects of miR-210 on evening phase control and axonal structural plasticity maybe not functional related.
Our evidence that Furthermore, genetic interaction data also indicates that Fas2 and miR-210 function in the same pathway.
In which cells are miR-210 and Fas2 required for the evening phase of activity?
Transcriptome profiling from the Rosbash lab show that expressions of both miR-210 and Fas2 are enriched and oscillating in the PDF positive neurons [15]. Overexpressing miR-210 in the PDF neurons causes flies arrhythmic or rhythmic with a prolonged circadian period (Table 1). However, using miR-210 knock-in GAL4 or Fas2 antibodies, we were not able to detect obvious signals for miR-210 or Fas2 in the PDF neurons. We cannot exclude the possibility that there is weak amount of miR-210 or Fas2 protein expression, which is not detectable due to sensitivity of the techniques we used.
We Here we identify that miR-210 regulates circadian locomotor rhythms and axonal structural plasticity of pacemaker neurons in Drosophila. A similar mechanism may exist in mammals. Interestingly, the glutamatergic synapses on the VIP neurons of mammalian master circadian pacemaker suprachiasmatic nucleus (SCN) also show circadian structural plasticity, which maybe important for light entrainment in mice. miR-210 is a highly conserved miRNA from worms, flies, to humans. In silico miRNA target prediction with targetscan identified a highly conserved binding site of human miR-210 in the 3'UTR of vertebrates brain-derived neurotrophic factor (BDNF). BDNF has wellestablished functions in axonal branching and synaptic plasticity [54,55]. Thus, it is worthwhile to test that whether miR-210 plays conserved functions in structural plasticity in mammals.

Behavioral experiments and analysis
Most of the times, adult male flies (2-5 days old) were used to test locomotor activity rhythms. Since both miR-210 and Fas2 are both on the x chromosome, adult female flies were also used for locomotor rhythms, which were mentioned in the relative figures.
Flies were entrained for 4 days LD cycle at 25 °C and 60% humidity, and released into constant darkness (DD) at 25 °C for at least 5 days. Temperature cycles was performed with 12h:12h 29C:21C cycling for 6d in complete darkness. Trikinetics Drosophila activity monitors were used to record locomotor activity in I36-LL Percival incubators. Activity rhythms were analyzed with Faasx software protocol [56]. Eduction profiles were generated with 3d of LD activity. FAAS-X software was used to analyze behavioral data [56]. Actograms were generated with a signal-processing toolbox for MATLAB. The morning anticipation amplitude was determined by assaying for the locomotor activity as described.

Immunohistochemistry and quantification
Immunohistochemistry was performed with whole Drosophila brains.

Design of CRISPR mediated deletion and screening of Fas2 3'UTR deletions
Fas2