Diurnal Rhythms in Neurexins Transcripts and Inhibitory/Excitatory Synapse Scaffold Proteins in the Biological Clock

The neurexin genes (NRXN1/2/3) encode two families (α and β) of highly polymorphic presynaptic proteins that are involved in excitatory/inhibitory synaptic balance. Recent studies indicate that neuronal activation and memory formation affect NRXN1/2/3α expression and alternative splicing at splice sites 3 and 4 (SS#3/SS#4). Neurons in the biological clock residing in the suprachiasmatic nuclei of the hypothalamus (SCN) act as self-sustained oscillators, generating rhythms in gene expression and electrical activity, to entrain circadian bodily rhythms to the 24 hours day/night cycles. Cell autonomous oscillations in NRXN1/2/3α expression and SS#3/SS#4 exons splicing and their links to rhythms in excitatory/inhibitory synaptic balance in the circadian clock were explored. NRXN1/2/3α expression and SS#3/SS#4 splicing, levels of neurexin-2α and the synaptic scaffolding proteins PSD-95 and gephyrin (representing excitatory and inhibitory synapses, respectively) were studied in mRNA and protein extracts obtained from SCN of C3H/J mice at different times of the 24 hours day/night cycle. Further studies explored the circadian oscillations in these components and causality relationships in immortalized rat SCN2.2 cells. Diurnal rhythms in mNRXN1α and mNRXN2α transcription, SS#3/SS#4 exon-inclusion and PSD-95 gephyrin and neurexin-2α levels were found in the SCN in vivo. No such rhythms were found with mNRXN3α. SCN2.2 cells also exhibited autonomous circadian rhythms in rNRXN1/2 expression SS#3/SS#4 exon inclusion and PSD-95, gephyrin and neurexin-2α levels. rNRXN3α and rNRXN1/2β were not expressed. Causal relationships were demonstrated, by use of specific siRNAs, between rNRXN2α SS#3 exon included transcripts and gephyrin levels in the SCN2.2 cells. These results show for the first time dynamic, cell autonomous, diurnal rhythms in expression and splicing of NRXN1/2 and subsequent effects on the expression of neurexin-2α and postsynaptic scaffolding proteins in SCN across the 24-h cycle. NRXNs gene transcripts may have a role in coupling the circadian clock to diurnal rhythms in excitatory/inhibitory synaptic balance.

We have recently demonstrated in one day-old rat brain neurons, that activation (by depolarization) induced a reversible, calcium-dependent splicing repression of the rNRXN2a SS#3 exon [17]. Further studies indicated that under such conditions SS#3 and SS#4 exons splicing in rNRXN3a but not rNRXN1a are also repressed [18]. Fear conditioning training in rats induced in vivo significant and transient repressions of hippocampal rNRXN1/2/3a, SS#4 splicing, attenuated levels of the excitatory postsynaptic scaffolding protein PSD-95 and the inhibitory postsynaptic scaffolding protein gephyrin in the hippocampus and fear memory formation [18]. Induced neuronal activity may thus affect NRXNs transcripts and subsequently synaptic fine tuning.
The circadian clock residing in the brain's suprachiasmatic nuclei (SCNs), is responsible for the adaptation of physiological and behavioral rhythms to the external 24 h light-dark cycle and attaining proper temporal order within the organism [19]. Neurons in the SCN act as self-sustained oscillators that generate circadian rhythms in gene expression and electrical firing bursts to encode time of day, with high firing rates during the day and lower rates at night [20]. The core molecular clockwork in SCN neurons is based on autoregulatory feedback loops of transcriptional activators (CLOCK/NPAS2 and BMAL1) and inhibitors (PER1-2 and CRY1-2) [21]. The mechanisms by which the core clock produces synchronized rhythms in neural firing and gene expression are postulated to involve intracellular calcium, a second messenger that regulates many cellular processes [22]. In principle, neuronal firing bursts may trigger synaptic remodeling [23], and as such may be pivotal for the communication of the day/night information to the brain and body. Indeed there are several publications showing synaptic plasticity in SCN [24][25][26][27][28][29][30][31]. How the expression of genes encoding proteins that are relevant to synaptic transmission is changing across the circadian cycle is largely unknown.
Because the neurons of the circadian clock display autonomous rhythms in activity we hypothesized that they would also display autonomous oscillations in neurexins SS#3/SS#4 exons splicing and subsequently impinge on excitatory/inhibitory synaptic balance. In the present study we have thus explored circadian patterns in NRXN1/2/3a expression and SS#3/SS#4 exons splicing in the mouse SCN in vivo. The variations in PSD-95 and gephyrin levels were assessed as markers of excitatory/inhibitory synaptic balance.
To further explore the relationship between rhythms in NRXNs expression and SS#3/SS#4 exons splicing and the synaptic scaffold proteins we used immortalized rat SCN2.2 neurons in vitro [32]. These cells retain most biochemical and functional properties of the clock with internal periodicity of ,24 hours [33]. SCN2.2 cells were used to provide homogenous neuronal-derived cell population and facilitate plasmid transfection experiments. We show here that these cells show circadian variations in rNRXN1/2 expression and SS#3/SS#4 exons splicing and in PSD-95 and gephyrin levels and intracellular localization. Finally, we used siRNAs in SCN 2.2 cells to down regulate specific neurexins transcripts for establishing their role of in the circadian rhythms in synaptic scaffold proteins.

Circadian Rhythms in NRXNs Transcripts in the C3H/J Mouse SCN
The mRNA levels of the core clock genes mPer1, mPer2 showed robust rhythms in the mouse SCN ( Fig. 1A and B). In agreement with previous findings in animals entrained to 12:12 hours light-dark cycles [34], two synchronous peaks of mPer1 and mPer2 transcripts were demonstrated at 2 hours before (major peak) and 2 hours after (minor peak) the beginning of the dark phase (ZT10 and ZT14) ( Fig. 1A and B). The levels of total mNRXN1a and mNRXN2a transcripts also varied significantly over the 24-hour cycle ( Fig. 1C and D). NRXN1a transcript levels also displayed 2 peaks (a major peak at ZT6 and a minor at ZT14), whereas NRXN2a displayed a main peak at ZT6 (Fig. 1C and D). There were no significant rhythms in mNRXN3 a transcripts in the SCN (Fig. 1E). Fig. 2 shows the structure of the NRXNs and circadian rhythms in mNRXN1/2/3a SS#3/SS#4 exons splicing (amount of exon included per total transcript of a particular NRXN) in the SCN. Because of 1 exon difference between NRXN1 and NRXN2, SS#3 alternative exons are numbered E12(NRXN1) and E11(NRXN2) and SS#4 alternative exons are numbered E21 (NRXN1) and E20(NRXN2). We observed significant diurnal variations in alternative exon inclusion in mNRXN1a and mNRXN2a SS#3 (E12 and E11, respectively) and mNRXN2a SS#4 (E20) transcripts ( Fig. 2A, B and E). mNRXN1a SS#3 and mNRXN2a SS#4 splicing displayed two synchronous peaks each (ZT2 and ZT10) ( Fig. 2A and E) and mNRXN2a SS#3 exon-included transcript displayed a main peak in the dark phase (ZT22) (Fig. 2B). These peaks did not coincide with the peak expression of the total transcripts of these genes in the SCN (Fig. 1). There were no significant rhythms in mNRXN3a SS#3 and SS#4 or mNRXN1a SS#4 splicing in the SCN (Fig. 2 C, D and F).
Because neurexins impinge on excitatory/inhibitory synaptic balance we looked at diurnal variations in the related scaffold proteins. Figure 3 depicts the levels of gephyrin PSD-95 and neurexin 2a protein levels in the mouse SCN at different time points within a 24 h-cycle. Neurexin-1a levels were not assessed due to non-availability of specific antibodies. Gephyrin levels were significantly elevated during the dark period (ZT14-ZT22) and minimal during the day (Fig. 3A). The levels of PSD-95 protein rose during the day to reach peak levels at ZT14 and declined afterwards in the dark period to minimal values at ZT18 (Fig. 3B). Thus, both gephyrin and PSD-95 were at their peak levels at ZT14 but whereas PSD-95 level was low during the dark hours gephyrin levels were minimal during the light hours. The levels of neurexin 2a protein rose during the day to reach peak levels at ZT12 (namely ,6 hours after the peak in mNRXN2a mRNA; Fig. 1D), and declined afterwards in the dark period to minimal values during the day (Fig. 3C) in a similar manner to gephyrin.

Cells
The role of particular splice variants in the circadian rhythms of PSD-95 and gephyrin was further explored in vitro using SCN2.2 cells. SCN2.2 cells expressed rNRXN1a and rNRXN2a but not rNRXN3a (Fig. 4). Also, rNRXN1b and rNRXN2b were not detected in these cells. The mRNA levels of the core clock gene rPer2 and of rNRXN1/2 a in the SCN2.2 cells at different times after synchronization are presented in Figure 4. There was a significant rhythm in expression of rPer2 in the rat SCN2.2 cells, with peak at 24 hours after synchronization (t = 24). Thus, Per2 levels at t = 24 were significantly higher than at t = 12-18 hours after synchronization (P,0.001 for both). As can be seen in Figure 4, mRNA levels of rNRXN1a and rNRXN2a also varied significantly over the 24 hour cycle and displayed synchronous peaks at t = 24 hours. Thus, rNRXN1a level at t = 24 was significantly higher than at t = 6, 12 and 18 h (p,0.01 for all). rNRXN2a levels at t = 24 were significantly different than at t = 12 and 18 h (p,0.001 for both) and levels at t = 18 also differed from those at t = 6 h (p,0.01). Mesor values of rPer2, rNRXN1a and rNRXN2a transcript Rq levels per GAPDH were 0.96, 0.41 and 0.63, respectively.
Like SCN cells, immortalized cells from the rat suprachiasmatic nucleus express functional melatonin receptors [35]. To evaluate whether the changes in rNRXN1/2 expression and SS#3/SS#4 exons splicing represent clock driven rhythms we performed phase shifting experiments, using melatonin (10 27 M at t = 0 for 6 h and incubation resumed with normal medium until harvest). In the presence of melatonin, rPer2 peak expression was advanced by ,6 hours to t = 18 h (Fig. 4A). Peak rNRXN1a and rNRXN2a transcript levels were also advanced with melatonin from t = 24 h to t = 18. The mesor values (mean 24 h transcript levels) were not affected (0.89, 0.31 and 0.41 for rPer2, rNRXN1a and rNRXN2a transcripts, respectively) and rhythm amplitude, as represented by peak/trough rNRXN1a and rNRXN2a mRNA ratios were maintained or slightly increased under the melatonin conditions (Fig. 4B).
Diurnal rhythms in rNRXN2a SS#3/SS#4 exons splicing (inclusions of E11 and E20 respectively) and rNRXN1a SS#3/ SS#4 exons splicing (corresponding to inclusions of E12 and E21 respectively) in the SCN2.2 cells were also assessed (Fig. 5). Significant rhythms were observed in rNRXN2a SS#3/SS#4 alternative splicing transcripts with synchronous peaks in E11/E20 exons inclusions at t = 18 h after synchronization (Fig. 5). Transcripts levels at t = 18 h were significantly higher than at t = 6 and t = 24 h and did not overlap the phase of peak expression of rNRXN2a (Fig. 4). These rhythms were attenuated with melatonin ( Fig. 5). rNRXN1a SS#3 exon (E12) splicing also demonstrated a significant diurnal oscillation with peak exon included transcript at t = 12. Transcript level (per total rNRXN1a) at t = 12 was significantly higher than at t = 18 and t = 24 h after synchronization. Peak rNRXN1a SS#3 exon included transcript (per total rNRXN1a) was delayed with melatonin to t = 18 h (Fig. 5). rNRXN1a SS#4 exon (E21) included transcripts were not detected in these cells.
The changes in levels of PSD-95 gephyrin and neurexin 2a with the circadian time were also assessed in the SCN2.2 cells (Fig. 6). Western blot analysis of the proteins indicated clear and significant circadian rhythms in the levels of the proteins, with maximal PSD-95 gephyrin levels at t = 12-18 h after synchronization. Minimal levels of both protein were observed at t = 24 h. Neurexin 2a were maximal at t = 6 h after synchronization ( Fig. 6C) namely ,6 hours after the peak in rNRXN2a mRNA (Fig. 4C).
The intracellular localization of PSD-95 and gephyrin in SCN2.2 cells at various times after synchronization, are depicted in Fig. 6C. Immunofluorescence experiments at different hours of the cycle using antibodies for PSD-95 (green) and gephyrin (red) indicated that PSD-95 was mostly associated with dense clusters in the cytoplasm at all time points. Gephyrin was mostly localized in the peri-nuclear region but showed clear membrane staining at t = 18 h and less so at other times. Unlike the Western blots that demonstrated even levels of gephyrin at t = 12-18 ( Fig. 6D), gephyrin staining in the cells consistently demonstrated higher apparent intensity at t = 18 h than at t = 12 h (Fig. 6D).
The effects of melatonin treatment on PSD-95 and gephyrin rhythms are depicted in Figure 7. For both, significant rhythms were still evident following 6 hours melatonin treatment (P = 0.005 for gephyrin and P = 0.02 for PSD-95) without significant changes in mesor values. Following melatonin treatment maximal PSD-95levels shifted to at 6 h (as also at 30 h) after synchronization (Fig. 7A). The phase shifting effects of melatonin treatment on gephyrin was minor, and maximal levels of Gephyrin were seen at t = 18 h after synchronization (Fig. 7B).

Gephyrin and PSD-95 Levels are Linked to NRXN2a SS#3 Exon-including Transcripts
We then used siRNA mediated suppression of rNRXN1a, rNRXN2a and rNRXN2a SS#3 exon-included transcripts to explore causal relationships between rhythms in neurexin genes expression, SS#3/SS#4 alternative exons splicing and the rhythms in PSD-95 and gephyrin. The cells transfected with the respective siRNA were then harvested 12 (corresponding to time of peak levels of PSD-95 and gephyrin), 18 (corresponding to time of peak levels of NRXN2a SS#3 E11 included transcript) and 24 (corresponding to time of peak expression of rPer2 and NRXN2a) hours after synchronization. Expression levels of total rNRXN1a, rNRXN2a, and E11 included rNRXN2a transcripts, and gephyrin and PSD-95 protein levels were assessed (Fig. 8). Scrambled siRNA (SCR) was used as control in these studies. The siRNA directed at the constitutive rNRXN2a exon E8 had no effect on rNRXN1a expression ( Fig 8A) and significantly reduced NRXN2a transcripts at all 3 time points (Fig. 8C). This siRNA also reduced the amount of rNRXN2a E11 including transcripts at t = 18 (a time when this transcript prevails) significantly and to a lesser extent at t = 24 (without reaching statistical significance compared to the scrambled siRNA control) ( Fig 8E). The suppression of rNRXN2a transcripts affected the levels of gephyrin and PSD at t = 18 (i.e. at which rNRXN2 E11 including transcript levels were also suppressed). Thus, a significant decrease in gephyrin and an increase in PSD-95 were noted in the rNRXN2a siRNA treated cells at t = 18 compared to scrambled siRNA control (Fig 8G and I). Also, in the rNRXN2a siRNA treated cells rNRXN2 E11 including transcripts were reduced at t = 24 (not significantly) and gephyrin levels decreased ( Fig 8G) but not at t = 12 at which rNRXN2 E11 including transcripts and gephyrin were both unaffected.
The siRNA directed against rNRXN2 SS#3 exon (E11) including transcripts significantly reduced the level of rNRXN2 E11 included transcripts at all time points ( Fig 8F). As expected, this siRNA also reduced the amount of rNRXN2 a transcripts at t = 18, (at which rNRXN2 SS#3 exon (E11) included transcripts prevail), but not at the other time points (Fig 8D). It also reduced rNRXN1a expression at t = 12 (at which rNRXN1 SS#3 exon (E12) included transcripts prevail, see Fig 5A) but not at the other times ( Fig 8B). The down regulation of rNRXN2 SS#3 exon (E11) included transcripts led to significant diminution of gephyrin levels at all times. In addition rNRXN2 E11 siRNA caused significant upregulation of PSD-95 exclusively at t = 18 (a time at which rNRXN2a transcript levels were also reduced). Down-regulation To further exclude off-target effects of NRXN2a and NRXN2 E11 siRNAs on gephyrin, two additional siRNAs directed at other constitutive rNRXN2a exons (E19, E14) were tested. As expected, only those siRNAs that reduced rNRXN2 E11 transcripts and at the time they did so, also reduced gephyrin levels. Thus, the siRNA directed at the constitutive rNRXN2a exon E19, did not reduce rNRXNa2 levels at 24 h after synchronization but reduced mean(SEM) rNRXN2 SS#3 exon (E11) included transcripts levels to 0.54 (0.21) and gephyrin to 0.7 (0.1) of control values with scrambled siRNA. The siRNAs directed at the constituti-verNRXN2a exon E14 reduced rNRXN2a to 0.79(0.14) but not

Discussion
This study demonstrates diurnal variations in expression and SS#3/SS#4 splicing of mNRXN1a and mNRXN2a (but not NRXN3a) neurexin-2a and postsynaptic scaffolding proteins (PSD-95 and gephyrin) levels in the mouse SCN in vivo. Neurexin-2a protein peak appeared ,6 hours after the mRNA peak. Rhythmic expression and SS#3/SS#4 exons splicing of rNRXN1a and rNRXN2a (NRXN3a is not expressed) was also observed in rat SCN2.2 cells in vitro. Being found in SCN samples from mouse and SCN-derived cells from rat, these rhythms appear to be a general rather than species-specific trait of SCN cells.
In the SCN in vivo, we observed traces of two different oscillations in clock control genes (mPer1 and mPer2) and in mNRXN1/2 genes, by which transcripts reached two peaks, one in the late light phase and another at the beginning of the dark phase. Such split was already demonstrated previously in mice kept in 12:12 light: dark cycles as in our experiment [34]. Some evidence could also be noted in other publications [36,37]. The early peaks in mPer1 and mPer2 were ascribed to the posterior SCN and the later one to the anterior SCN [34] or to existence of separate oscillating cell groups in mouse suprachiasmatic nucleus that couple photoperiodicity to the onset and end of daily activity [34]. A recent paper [38] concluded that in the living SCN tissue individual cells undergo an oscillation in expression of genes and proteins as they participate in complex interactions across the circadian cycle. In this sense they are clearly 'oscillating cells', but their temporal behavior depends critically on their location in the SCN tissue, most probably as a reflection of intercellular communication. The precise timing of PER2 expression within individual neurons is dependent on their location within the nucleus, and that small groups of neurons within the SCN give rise to distinctive and identifiable sub-regions.
The intrinsic nature of the rhythms in rNRXN1/2a expression and SS#3/SS#4 exons splicing is demonstrated in rat immortalized SCN2.2 cells in culture and the ability of melatonin to phase shift them as it does for the core clock components (e.g. Per2) [35]. As SCN2.2 cells act as self sustained pacemakers [32] these results suggest that these rhythms are cell-autonomous. NRXN1a and NRXN2a genes do not include E-box sequences in their promoters and therefore cannot be regarded as primary clockcontrolled genes [39]. The rhythms in expression of NRXN1a and NRXN2a and their alternative SS#3/SS#4 exons splicing thus reveal an additional layer of post-transcriptional regulation of clock controlled genes in mammals as recently found in plants (Arabidopsis) and flies (drosophila) [40].
The rhythm in PSD-95 in the SCN2.2 cells appears to be causally linked to the clock (as it was shifted/modified following melatonin treatment. The rhythm in gephyrin appears to be less dependent on clock phase. It was influenced by NRXN2a SS#3 exon (E11)-included transcript levels and was consistently modified following specific NRXN2a E11 siRNA treatments. With altogether 6 different siRNAs (one scrambled, one targeting rNRXN1, and four targeting rNRXN2 Exons E8, E11, E14 and E19) only those that reduced rNRXN2 E11 transcripts levels and at the times they did so, caused down regulation of gephyrin. Moreover, down regulation of rNRXN2a expression using siRNA directed against a constitutive exon (e.g. E8) resulted in suppression of gephyrin selectively at times when E11 including transcripts comprised most of the NRXN2a mRNA (t = 18 h). The rNRXN2a E11 SiRNA also reduced the levels of NRXN1a mRNA at t = 12 (a time when the SS#3 exon (E12) included transcripts prevail). However, as down regulation of rNRXN1a by specific SiRNA did not affect gephyrin, NRXN1 is probably not involved in the down regulation of gephyrin at t = 12.
Notably, the % of inhibition is calculated from the amount of transcript/protein in the siRNA treated cells to respective values in SCR treated cells at the same time after synchronization. As the amounts of total transcripts of the two neurexins as well as synaptic scaffolds proteins vary with the circadian cycle and at different phases, it is not surprising that the % inhibition at a certain time point may differ for the mRNA transcripts and protein levels. Therefore a high percentage at one time may be related to a small change in actual transcript or protein levels and vice versa. For example, at 12 hours most of the rNRXN2 is in the E11 excluded form and therefore even for an equal reduction in total NRXN2, that reduction is presumably in the E11 excluded form and would not lead to decrease in gephyrin at this hour (Fig. 8E). PSD-95 levels were not affected by changes in rNRXN2a E11 transcripts but upregulated under conditions that total rNRXN2a mRNA transcripts were suppressed. The effects on the synaptic scaffold proteins appear to be specific for NRXN2a as down-regulation of NRXN1a was not associated with significant changes in PSD-95 or gephyrin. Beside diurnal fluctuations in protein levels, gephyrin also exhibited diurnal rhythms in intracellular localization moving between the perinuclear region, the cytoplasm and plasma membrane. Considering the limitations of immunostaining technique as a quantitative method, the results obtained using these different methods are reasonably consistent. However, at t = 12 h after synchronization gephyrin immunostaining intensity appeared to be less than expected according to the western blot data. At this time gephyrin was mostly associated with the cytoplasm. The most obvious explanation for the apparent discrepancy is that the two antibodies (the one used in the immunoblotting and the one used in the immunostaining) differ in their affinities to some unspecified gephyrin isoforms. This remains to be further investigated.
This study suggests a link between the expression of certain NRXN splice variants to excitatory/inhibitory synapse scaffolds in the biological clock. Because there are currently no antibodies capable of detecting the specific variants of native neurexins (comprising ,1000 isoforms), it was only possible to demonstrate circadian rhythms in total endogenous neurexin-2a and show that it followed peak NRXN2a transcripts by ,6 h in both SCN in vivo and in the SCN2.2 cells in vitro. Nevertheless, the data in our study demonstrates that the clock controlled changes in certain endogenous NRXNs transcripts are eventually translated into changes in synaptic proteins. The molecular mechanisms linking the circadian clock to the control of RNA processing is compatible with emerging evidence suggesting that regulation at the RNA level strongly impacts the cellular circadian transcriptome and proteome as well as the oscillator mechanism itself [41].
What our study also shows is that the cross-talk between Nrxn2a E11 included transcripts and the inhibitory synapse scaffold protein Gephyrin are fast enough to be involved in synaptic plasticity over the 24 hours period. Emerging evidence suggests that the entire synaptic structure, including scaffolding proteins and vesicles, is highly dynamic during both development and periods of plasticity. Indeed, SCN oscillations are associated with plastic changes in configuration of the SCN cell network [28,29,31,42,43]. Furthermore, it appears that some of these changes involve inhibitory synapses. [44,45]. Dynamics of GABAergic synaptic components have been studied previously over milliseconds to minutes, revealing mobility of postsynaptic scaffolds and receptors. [46,47,48]. Recent studies in both vertebrate and invertebrate models have shown time-of-day effects on neurophysiology and memory formation, and have revealed a possible role for cycling molecules in memory persistence (reviewed in [49]). Together, these studies suggest that common mechanisms underlie circadian rhythmicity and long-term memory formation. The rhythmic expression and SS#3/SS#4 exons splicing of NRXNs and the effects of NRXNs on the expression of postsynaptic scaffolding proteins in SCN neurons across the lightdark cycle may link the circadian clock to diurnal oscillations in balance of excitatory/inhibitory neurotransmission.

Animals and Sample Acquisition
All animal experiments were conducted in accordance with a protocol approved by the Policy on the Use of Animals in Neuroscience Research, the Policy on Ethics, by the Society for Neuroscience, and were consistent with Federal Guidelines and the European Communities Council Directive (89/609/EEC). Animal and sample acquisition was performed as described previously [50]. Male C3H/J mice (Charles River, Sulzbach, Germany) aged 8-12 weeks were used in all experiments. At least 2 weeks prior to experiments, animals were housed under a standard light-dark (LD) regimen, with 12-h light (250 lux; light on designated as Zeitgeber time [ZT0] and 12-h darkness (dim red light ,10 lux, .680 nm), with constant room temperature, and food and water available ad libitum. Tissue sampling was carried out at ZT2, 6, 8, 10, 12, 14, 18, and 22 (three animals per time point). After decapitation of animals at indicated time points, the brain was removed. The isolation of the SCN essentially followed published procedures [19,51]. Briefly, after brain removal, 500 mm thick slices of the suprachiasmatic region were prepared with a manual tissue chopper. When the desired level of SCN has been reached (the SCN at this point appears as more defined, round or almond shaped structures) sections containing the SCN were cut from the slices under a binocularly microscope using scalpels and surrounding tissue was carefully removed. The isolated SCN tissues were immediately frozen on dry ice. Subsequently, RNA and protein extraction was performed using the NucleoSpinH RNA/Protein kit according to manufacturer's protocol. (Macherey-Nagel, Düren, Germany). Verso TM RT-PCR Kit (Thermo scientific, ABgene, Epsom, UK) was used for cDNA synthesis, from 250 ng of total RNA using Oligo(dT) primer, according to manufacturer's protocol.

SCN2.2 Cell Culture and Sample Acquisition
SCN2.2 Immortalized rat SCN cells, obtained from Dr. David J Earnest [32] were plated onto 6-well plastic tissue culture plates (3610 5 /well) and grown for 48 h to confluence in minimal essential medium (MEM) supplemented with 10% fetal bovine serum (FBS), 4mM L-Glutamine, 4 mg/mL glucose (Merck, Germany), penicillin (100U/mL) and streptomycin (0.1 mg/mL) (all from Biological Industries, Beth Haemek, Israel) at 37 o C in humidified atmosphere with 5% CO 2 . Synchronization was performed essentially as described [52]. Growth medium was replaced with medium containing 5% FBS for at least 8 hr prior to the synchronizing pulse. The medium was then removed, and cells were rinsed with calcium-magnesium free (CMF) Buffer (Dulbecco's PBS (1X) w/o Ca 2+ or Mg + (Sigma)) containing 2 mg/mL glucose, and incubated for 1 hr in serum free medium. The serumfree MEM medium was then replaced with growth medium and cells were kept in this medium until harvested at 6 hours intervals over 30 hours. Samples were subjected to RNA extraction using MasterPure TM Kit (EPICENTRE, Madison, USA), according to manufacturer's protocol. Verso TM RT-PCR Kit (ABgene Nucleic Acid Amplification, Epson, UK) was used for cDNA synthesis, from 1 mg of total RNA using Oligo (dT) primer according to manufacturer's protocol. Synchrony was checked by measuring circadian rhythms in RPer2 mRNA expression using real time PCR. Proteins extraction was performed using Radio Immuno Precipitation Assay buffer (RIPA lysis buffer; 150 mM sodium chloride, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS (sodium dodecyl sulphate) 50 mM Tris, pH 8.0) with Protease Inhibitor Cocktail (Sigma-Aldrich, MO, USA). Cell suspension was maintained for 20min in ice, following centrifugation at 4uC for 10min at maximum speed. Supernatant was removed to a new cold tube, and proteins amount was measured spectrophotometrically with Bradford reagent at 595 nm (Bio-Rad laboratories, CA, USA).

RNA Interference
SCN2.2 cells were plated into 6-well plates, and transfected 24 h later with 40nM of siRNA duplexes specific for rNRXN1 and for rNRXN2 exons E8, E11, E18, E19, and scrambled negative control siRNAs as a mock transfection ( Figure 8. Silencing of NRXN2a and NRXN2a E11 including transcript have differential effects on gephyrin and PSD-95 levels. Total and E11 including rNRXN2a mRNA levels were suppressed using targeted siRNA sequences (siRNA NRXN2 and siRNA NRXN2 E11 respectively), a scrambled sequence (SCR) was used as a control. Cells were synchronized and then incubated with the siRNAs or SCR for 24 hours and harvested 12,18 or 24 hours afterwards. Mean (+SEM) mRNA levels of (A,B) total rNRXN1a, (C,D) total rNRXN2a, (E, F) E11 including rNRXN2a, and protein levels of (G, H) gephyrin and (I, J) PSD-95 are depicted. Significant differences from control values are marked with asterisks (* p,0.05 t-test against SCR; N = 3

Statistical Analyses
The mean (SD) values of experimental and control groups were compared using Student's t tests. Differences were considered significant if P,0.05. Circadian rhythms were evaluated by oneway ANOVA of data from samples collected at different times of the day (P,0.05). One way (time of day) ANOVA was used to establish the presence circadian rhythms [54]. This method assesses whether the variance between time points is significantly greater than the random variation within them. The analysis requires several readings to be taken at each of a group of times within the day; the times of assessment do not have to be evenly spaced. Strength of the ANOVA method is that shape of the rhythm does not affect the statistical outcome [54]. Cosinor best fit method that is sometimes used for the analysis of circadian rhythms was not chosen because for most of the observed rhythms the data span is not sinusoidal in shape.