Genetic background-dependent effects of murine micro RNAs on circadian clock function

MicroRNAs (miRs) are important regulators of a wide range of biological processes. Antagomir studies suggest an implication of miR-132 in the functionality of the mammalian circadian clock. miR-212 and miR-132 are tandemly processed from the same transcript and share the same seed region. We found the clock modulator miR-132 and miR-212 to be expressed rhythmically in the central circadian clock. Consequently, mRNAs implicated in circadian functions may likely be targeted by both miRs. To further characterize the circadian role we generated mice with stable deletion of the miR-132/212 locus and compared the circadian behavior of mutant and wild-type control animals on two genetic backgrounds frequently used in chronobiological research, C57BL/6N and 129/Sv. Surprisingly, the wheel-running activity phenotype of miR mutant mice was highly background specific. A prolonged circadian free-running period in constant darkness was found in 129/Sv, but not in C57BL/6N miR-132/212 knockout mice. In contrast, in C57BL/6N, but not in 129/Sv miRNA-132/212 knockout mice a lengthened free-running period was observed in constant light conditions. Furthermore, miR-132/212 knockout mice on 129/Sv background exhibited enhanced photic phase shifts of locomotor activity accompanied by reduced light induction of Period gene transcription in the SCN. This phenotype was absent in miRNA-132/212 knockout mice on a C57BL/6N background. Together, our results reveal a strain and light regimen-specific function of miR-132/212 in the circadian clock machinery suggesting that miR-132 and miR-212 act as background-dependent circadian rhythm modulators.


Introduction
-hour, or circadian, clocks have evolved as an adaptation to daily recurring changes in the environment. In mammals, the central circadian pacemaker is localized in the suprachiasmatic nucleus (SCN) of the ventral hypothalamus [1,2]. Generation of circadian rhythmicity occurs at the single cell level and is based on interlocked positive and negative transcriptional-translational feedback loops (TTL) of clock genes and their protein products [3]. PLOS

Ethical statement
All animal experiments were done with prior ethical assessment by and permission from the Office of Consumer Protection and Food Safety of the State of Lower Saxony and in accordance with the German Law of Animal Welfare.

Study design
Animals were kept for 2 weeks under LD conditions ("lights on", zeitgeber time (ZT) 0; "lights off", ZT12; light intensities of 100 lux) and then released into constant darkness (DD) for at least 14 days (DD 1 ) to compare the free-running period length. To analyze the resetting ability an activity-delaying 15-min light pulse (LP) of 20 lux was given at circadian time (CT) 15 (3 hours after activity onset) on day 15 during DD 1, followed by another period of darkness for 14 days (DD 2 ) and subsequent release into constant light (LL) of 100 lux to compare the freerunning period in LL. The color of the fluorescent light bulbs used during all experiment was around 3k Kelvin. Wheel-running activity including behavioral phase shifts were analyzed using ClockLab software (Actimetrics, Wilmette, IL, USA).

Experimental animals
For all experiments, 2-3 months old in-house bred WT and homozygous miR-132 -/-/212 -/-(KO) littermate mice in either of the background strains (C57BL/6N and 129/Sv) were used The miR-212/132 null mouse line has been generated by genomic targeting of the miR-212/132 locus in ES cells that has been derived from inbred wild-type 129/Sv (full name: 129S2/SvPa-sOrlRj) background. Therefore the mouse line that has been obtained by breeding the founder mice with 129/SV wild-type animals were in pure 129/Sv background, and this line has been maintained at heterozygous state. The founder mice were also bred in parallel with the wildtype C57BL/6N mice and the heterozygote progenies has been subsequently bred with the wildtype C57BL/6N mice for a minimum of 6 generations to obtain miR-212/132 null mouse line in almost pure (>%98, N6) C57BL/6N background. For both mouse lines in 129/Sv and C57BL/6N backgrounds, we have generated the homozygous null mice for analyses via breeding heterozygous animals with each other and have used the wildtype and homozygous mutant littermates in pairs. KO mice were viable in, either, a 129/Sv or a C57BL/6N background and did not show any obvious alterations in behavior, although involvement of both miRs had previously been described for hematopoietic cell development and for endothelial and cardiac functions [30][31][32]. Thus, it was feasible to compare running-wheel performance as a sensitive readout of central circadian clock function in both KO mouse lines. Only male mice were used to avoid additional rhythmic components in females, such as the estrous cycle, which may influence general activity and wheel-running performance [33]. Before experiments mice were bred and housed in small groups of 5 or fewer in the same animal facility under the same LD cycle conditions of 12 hours light, 12 hours dark (LD), with food and water ad libitum.

Behavioral analysis
Handling and activity measurements during experiments were performed as described previously [34]. For constant light conditions and light pulse experiments, animals were singlehoused in running wheel-equipped cages (33 x 15.5 x 13 cm with a 15-cm diameter wheel) with standard bedding. At least 10 days of wheel-running recordings for each condition were used to determine activity period (tau, calculated using a χ 2 periodogram and confirmed by fitting a line to the onsets of activity), phase shifts (difference between regression lines through the onsets of activity before and after the light pulse) and the total amount of activity. Activity profiles have been calculated with the ClockLab software by averaging the data from 10 consecutive days of wheel running activity in each light conditions. Wheel rotations were detected in 6 min intervals and calculated either on a 24 h day or relative to the individual day length of each mouse based on their individual free-running period in DD and LL conditions. The group sizes were as follows: LD profile 129/Sv WT n = 9; KO n = 6; C57BL/6N WT n = 11; KO n = 10; DD profile 129/Sv WT n = 8; KO n = 6; C57BL/6N WT n = 11; KO n = 10; LL profile: 129/Sv WT n = 5; KO n = 6; C57BL/6N WT n = 11; KO n = 10; total activity in DD 129/Sv WT n = 7; KO n = 8; C57BL/6N WT n = 8; KO n = 6; total activity in LL 129/Sv WT n = 5; KO n = 6; C57BL/6N WT n = 10; KO n = 10; period in DD 129/Sv WT n = 15; KO n = 15; C57BL/6N WT n = 12; KO n = 11; period in LL 129/Sv WT n = 6; KO n = 7; C57BL/6N WT n = 10; KO n = 11; phase shift 129/Sv WT n = 8; KO n = 9; C57BL/6N WT n = 11; KO n = 10.

In situ hybridization (ISH)
Circadian profiles of miR-132 and miR-212 expression were determined in the SCN of 129/Sv wildtype mice. Brains were dissected at 6 CTs (CT3, 7,11,15,19,23) during the second day in DD (n = 18 mice; 3 mice per time point) following 2 weeks of LD. Digoxigenin-labelled ISH was performed in 129/Sv and C57BL/6N wildtype mice as previously reported [35] for Per1 and Per2 and signal strength quantified as has been described [36]. For each individual animal, 3-6 sections spanning throughout the SCN were evaluated and a background-subtracted mean value calculated. Mean SCN expression of each miR is represented relative to the averaged expression over the circadian day (set to 1). To measure the induction of gene expression after a light pulse at CT 15 (15 min, 100 lux) (n = 4 per time point and genotype) animals were sacrificed 1 hour later by cervical dislocation (under dim red light during the dark period) and brains were dissected. Tissues were fixed, dehydrated, and paraffin embedded; 8-μm sections were prepared and stored at -80˚C. Sections were hybridized with 35 S-UTP-labelled antisense RNA probes for clock gene transcripts [37]. Relative quantification of expression levels was performed by densitometric analysis of autoradiograph films using Scion Image software (Scion Corp. Frederick, MD, USA).

Statistical analysis
All statistics were performed using GraphPad Prism 5.0 (Prism, La Jolla, CA, USA) software. Diurnal variation of miR-132 and miR-212 expression data was tested by fitting a cosine wave (equation y = B + (A Ã cos (2 Ã π Ã ((x-Ps) / 24)))) to gene expression data, where B is the baseline, A is the amplitude, Ps is the phase shift, with a fixed 24-hour period; significance was determined by F-test. Two-way ANOVA was performed to detect statistical differences of period length, light pulse period length effects, light-induced phase shifts and gene induction followed by Sidak's multiple comparison post-hoc tests. Paired t-test was used to determine period variability between groups and within animals by comparing the period length in DD before and after a light pulse. Correlation analysis of period length and phase shift was performed by fitting a linear regression. The corresponding p-value was determined using the Pearson correlation coefficient, given the correlation value and the sample size. A statistically significant difference was assumed with p-values of less than 0.05. Asterisks indicate significance ( Ã p 0.05; ÃÃ p 0.01; ÃÃÃ p < 0.001).  (Fig 3A, 3C and 3G) and C57BL/6N (Fig 3B, 3D and 3G) background, consistent with previous reports [38]. Strain dependent alterations in activity profiles were apparent (compare Fig 3E and 3F), but total activity in LL was indistinguishable between the cohorts of both background strains and genotypes (Fig 3H). Similar to results obtained in DD conditions, in LL, WT mice showed background strain dependent differences in period length (Fig 3A-3D and 3G). Loss of miR-132/212 did not affect LL period length in 129/Sv mice (Fig 3A, 3C and 3G), but increased the circadian day length in mice on a C57BL/6N background (Fig 3A-3D and 3G). Total activity in LL was altered neither by genetic background nor genotype ( Fig 3H).

miR
Taken together, miR-132/212-deficiency lengthened the circadian period of the SCN pacemaker in a strain-and light regimen-dependent way. Loss of miR-132/212 prolonged the freerunning period in constant darkness in the 129/Sv background, while increasing the free-running period in constant light conditions in the C57BL/6N strain. miR-132/212-deficiency differentially affects photic phase shifts in C57BL/6N and 129/Sv strains Changes in the circadian period length as described above tend to correlate with alterations in the light resetting capability of the SCN pacemaker [39]. To address this in miR-132 -/-/212 -/mouse lines more directly, we measured acute phase-shifting of the wheel-running activity rhythm in response to a delaying 15-min light pulse (20 lux) at CT15 on 129/Sv (Fig 4A and  4C) and C57BL/6N background (Fig 4B and 4D; of note, these were the same cohorts of C57BL/6N animals used for the experiments shown in Fig 2A-2D). Indeed, miR-132/212 lossof-function on a 129/Sv background evoked a marked increase of light pulse-induced behavioral phase delays (Fig 4A, 4C and 4E). In contrast, we found no significant effect on phaseshifting behavior in response to a delaying light pulse in KO mice of C57BL/6N background (Fig 4B, 4D and 4E).
Increased phase shifts in miR-132/212-deficient 129/Sv mice are accompanied by diminished light induction of Per gene transcription in the SCN Photic phase shifting involves acute light induction of the core clock genes Per1 and Per2 in the SCN [6]. It has been reported that, miR-132 transcription itself is also light inducible and miR-132 induction inhibits light induced Per1 and Per2 expression [28]. To test whether lossof-function of miR-132/212 affects light induced Per transactivation in different mouse lines, we examined photic induction of Per1 and Per2 in response to a 15-min light pulse (20 lux) at CT15 in both genotypes and strains by in situ hybridization on SCN-containing sections. In line with the literature [6], Per1 and Per2 mRNA levels were strongly induced by light exposure in the SCN of WT mice of both background strains (Fig 5A, 5C and 5E-5H). Interestingly, miR-132/212-deficiency abolished elevation of Per1 and Per2 mRNA levels after photic    Fig  4A, 4C and4E and Fig 5A-5D, 5E and 5G). In contrast, similar Per1 and Per2 induction was found in C57BL/6N WT and KO mice (Fig 5A-5D, 5F and 5H). This is consistent with a similar trend in behavioral resetting observed in WT and KO mice following a phase-delaying light pulse on this background (Fig 4B, 4D and 4E).

Discussion
Our work highlights strain-specific functions of miR-132/212 in the regulation of the murine SCN pacemaker. Our data confirm previous experiments using miR-132 intra-cerebral antagomir infusion and overexpression in the forebrain showing that miR-132 acts as a negative modulator of light-induced resetting of the SCN pacemaker [28,29]. Moreover, we show that this effect is restricted to the 129/Sv background and absent in C57BL/6N mice. Our results further suggest a strain and light-regimen specific function of miR-132/212 in the clock machinery itself, as revealed by strain-dependent effects on the free-running period length in DD and LL. These results are summarized in Table 1.
We found the clock modulator miR-132 and miR-212 to be expressed rhythmically in the central circadian clock. miR-212 and miR-132 share the same seed region and they can therefore target the same mRNAs [27]. Indeed, double-targeting of both miRs was demonstrated for mRNAs [40] implicated e.g. in circadian functions [41].  Period (DD) longer no effect n.d. no effect Period (LL) no effect longer n.d. n.d.

Phase shift ZT/ CT14
increased no effect decreased increased

Light induction Per1
abolished Per1 induction no effect reduced PER1 induction reduced Per1 induction

Light induction Per2
abolished Per2 induction no effect reduced PER2 induction increased PER2 induction compared to vehicle (but below baseline) n.d. not determined. https://doi.org/10.1371/journal.pone.0176547.t001 Genetic effects on circadian microRNA functions miR-132/212 regulate core clock properties in a background-dependent manner We found that miR-132/212-deficiency affects SCN clock properties as manifested by lengthening of the free-running period length in both inbred strains examined. Interestingly, neither overexpression nor knockdown of miR-132 had an effect on the core clock mechanism (results from different studies are compared in Table 1). The miR-132/212 gene knockout used in our study results in a complete loss-of-function whereas a knockdown approach might not be as efficient. Our double knockout is also efficiently circumventing a potential miR-132 and miR-212 redundancy. miR-132 and miR-212 show overlapping circadian oscillations in the SCN with slightly different phases. Although they are splice variants from a single locus and thus, likely act synergistically, it is possible that genetic interaction of both miRs could account for the period effects missed in previous studies [28,29]. Importantly, depending on light conditions, longer periods were found in KO mice of both genetic backgrounds. Such period changes indicate that miR-132/212-deficiency significantly impairs the properties of the endogenous circadian clockwork in both strains. Recall that Per1 mutant mice show a longer period under constant light exposure [42] and period lengthening in constant light has been reported when degradation of PER2 in the SCN is delayed [43]. Modulation of Per gene activity in the SCN of miR-212/132 -/mice may affect the ability of the central clock to respond to light. Importantly, it has been reported that miR-132 target genes Poly(A) binding protein interacting protein 2 (Paip2a) and B-cell translocation gene 2, anti-proliferative (Btg2) dampen basal Per expression and attenuate PER protein translation, and thereby accelerate the inhibition of Per gene transcription [29,44]. Thus, miR-132/212-deficiency may enhance the free-running period in miR-132 -/-/212 -/mice via this mechanism. Nevertheless, we cannot exclude that some of the observed effects might be mediated by miR effects in SCN-afferent tissues such as the retina where miR-132/212 expression has been found [45] or the bed nucleus of the stria terminalis. Furthermore, baseline differences of clock gene expression in miR-deficient mice between both strains can also influence the circadian phenotype including SCN light responsiveness.

miR-132/212 regulates phase shifts in a background-dependent manner
We tested the prediction that miR-132/212-deficiency influences light sensitivity of the SCN by measuring photic phase shifts of locomotor behavior [46]. Indeed, we found that miR-132/212deficiency in 129/Sv mice modulates circadian light responses, with respect to resetting and Per gene induction. Additionally, a prolonged period length during constant light as observed in KO C57BL/6N mice further suggests an increased light sensitivity of the SCN [39]. Phase shifting of the circadian clock in response to light stimulation is believed to be mediated by Per induction in the SCN which then triggers a resetting of behavioral activity [6,47]. In accordance, miR-132 overexpression in C57BL/6 mice resulted in reduction of light-induced PER1/ 2 levels and photic resetting of behavior [29]. We found that miR-132 -/-/212 -/-C57BL/6N mice did not exhibit a change in Per expression upon light stimulation, which is consistent with their unaffected behavioral resetting. Interestingly, although light responsiveness of Per1 and Per2 expression in the SCN was completely abolished in miR-132 -/-/212 -/-129/Sv mice, they demonstrated enhanced phase-shifting behavior. This is in accordance with previous reports using miR-132 antagomirs in C57BL/6J mice which led to a marked reduction of light induced Per1 gene expression and enhanced phase shifts [28]. Nevertheless, loss-of-function of, both, miR-132 and miR-212 on the 129/Sv background seemed to further enhance the previously described circadian resetting phenotype. Of note, Per1-deficient mice exhibit an increased phase shifting capacity similar to the one documented in miR-132 -/-/212 -/-129/Sv mice [48].
However, other mechanisms than Per regulation may play a role in determining resetting in miR-132 -/-/212 -/mice, e.g. it can be speculated that the weaker adaptation ability of miR-132 -/-/ 212 -/mice, similar to Per1-deficient mice, may be the result of a weakened circadian oscillator, indicated by reduced circadian pacemaker amplitude, as was also suggested for Clock mutant mice [49].
Contradictory results between knockout, knockdown and overexpression studies (summarized in Table 1) might be explained by parallel, but independent effects of miR-132 and miR-212 on behavioral resetting and Per gene expression in the SCN.
Direct targets of miR-212 and miR-132 in the SCN are known to interact with Per genes [29,44]. In particular, Methyl-CpG-binding protein 2 Mecp2, which is involved in chromatin remodeling and a target of miR-132 and miR-212, activates Per transcription while the miR-132 target genes PAIP2A and BTG2 attenuate PER protein translation, both of which may result in a dampening of light induced Per expression or accelerated PER decay [29,44], thus, influencing behavioral resetting. This link between Per gene modulation and phase shift regulation by the miR-132/212 family, however, needs to be further examined.
Circadian phenotypes in miR-132 -/-/212 -/mice were generally more pronounced in the 129/Sv than in the C57BL/6N strain. These data support the possibility that different mouse strains may carry different genetic modifiers influencing the circadian phenotype. Similarly, a Clock Δ19 modulator gene has recently been identified which accounts for the suppression of the Clock Δ19 mutant circadian phenotype in BALB/cJ mice [50]. Interestingly, miRs act as modifiers [19] and impact on circadian clock function [28] and covariation in miR-212 abundance and function was documented in different background strains, including C57BL/6 [17,18,51].
Taken together, we have shown that miR-132/212 are background-dependent modulators of the circadian clock. Future studies are required to identify the miR132/212 targets whose regulation may mediate this effect. Our data further emphasize that, to fully understand the regulatory function of a specific gene or a miR, loss-of-function phenotypes should be analyzed in more than one genetic background strain. The persistence of a phenotype across various strains would underscore a universal function of a specific gene and, thus, suggest a potential significance for humans.