PER2 mediates CREB-dependent light induction of the clock gene Per1

Light affects many physiological processes in mammals such as entrainment of the circadian clock, regulation of mood, and relaxation of blood vessels. At the molecular level, a stimulus such as light initiates a cascade of kinases that phosphorylate CREB at various sites, including serine 133 (S133). This modification leads CREB to recruit the co-factor CRCT1 and the histone acetyltransferase CBP to stimulate the transcription of genes containing a CRE element in their promoters, such as Period 1 (Per1). However, the details of this pathway are poorly understood. Here we provide evidence that PER2 acts as a co-factor of CREB to facilitate the formation of a transactivation complex on the CRE element of the Per1 gene regulatory region in response to light or forskolin. Using in vitro and in vivo approaches, we show that PER2 modulates the interaction between CREB and its co-regulator CRTC1 to support complex formation only after a light or forskolin stimulus. Furthermore, the absence of PER2 abolished the interaction between the histone acetyltransferase CBP and CREB. This process was accompanied by a reduction of histone H3 acetylation and decreased recruitment of RNA Pol II to the Per1 gene. Collectively, our data show that PER2 supports the stimulus-dependent induction of the Per1 gene via modulation of the CREB/CRTC1/CBP complex.

via an LXXLL consensus sequence, a known motif responsible for the interaction between nuclear receptors and their co-activators 34 . After a light pulse, WC-1 changes conformation and directs NGF1 to the chromatin to acetylate histones. At the same time, it stabilizes the transcriptional factor WC-2 on the promoter of target genes, and target gene expression is promoted 35 .
Structural similarities between the WC-1 and PER2 protein suggest that mammalian PER2 may act in a similar way as WC-1 in N. crassa. Both proteins are similar in size and contain 3 PAS domains. They are both essential regulators of the circadian clock and can act as co-activators via LXXLL domains 33,36 . PER2 accumulates in the SCN between zeitgeber time (ZT) 14 and 16, which corresponds to the time window of light-responsiveness that leads to induction of genes such as Per1 37 . Since WC-1 is involved in light-dependent responses, and due to the resemblance of PER2 with the fungal protein, we wondered whether the mammalian clock factor would modulate the light-mediated CREB signaling pathway to induce genes such as Per1 or cFos in a similar fashion. This resembles the function of WC-2, which is the transcription factor that modulates light responses in N. crassa.
Here, we describe a novel function of PER2 as a mediator of the light-dependent CREB signaling in the early night. Using an in vitro (forskolin-induced cell lines) and an in vivo system (light-stimulated mice), we show that PER2 modulates the light/forskolin mediated assembly of the pSer133-CREB/CRTC1/CBP complex. In particular, we show that PER2 gates the CREB: CRCT1 complex to the early night. Förster resonance energy transfer (FRET) analysis revealed that the complex formation between CREB and CBP depends on the presence of PER2. Consequently, PER2 is important for the histone H3 acetylation at lysine 27 by CBP, and hence induction of Per1 gene expression. We conclude that PER2 is a scaffold facilitating the formation of the CREB: CBP complex after a light pulse or forskolin stimulus.

PER2 modulates light/forskolin-induced Per1 and cFos gene expression in an opposite manner.
A light pulse (LP) applied to mammals in the dark phase of a 12 h light: 12 h dark cycle, where zeitgeber time (ZT) 0 is lights on and ZT12 is light off, elicits phase-shifts in locomotor activity 10,38 paralleled by induction of Per1 gene expression 6,39 . Here we measured the expression of pre-mRNA of Per1 and Per2 genes in the SCN of wild type (wt) mice before and after an LP of 15 min duration at ZT14 (Fig. 1A). The pre-mRNA of Per1 was strongly induced after the LP, whereas the pre-mRNA of Per2 was significantly less responsive (Fig. 1A). To corroborate this observation in vitro, we stimulated NIH3T3 cells with forskolin, which mimics the effect of phase-shifting via activation of the Protein kinase A (PKA) signaling pathway 40 . We observed that Per1, but not Per2, pre-mRNA was induced after 25 and 40 min of forskolin treatment (Fig. 1B). These results suggested that the Per2 gene is less inducible by light or forskolin than Per1. Because Per2 KO mice do not phase delay after a light pulse at ZT14 41 , our general question was whether PER2 was an upstream factor rather than a downstream target of stimulus-mediated clock resetting.
To challenge this hypothesis, we analyzed Per1 pre-mRNA expression in SCN samples obtained from wt and Per2 knock-out (KO) mice 42 collected either in the dark or 15 min after LP at ZT14. We observed that in Per2 KO mice the light inducibility of Per1 is dampened compared to wt (Figs. 1C, S1A upper panel). Although Per2 pre-mRNA was less light-inducible compared to Per1, it was not inducible in the Per2 KO mice (Figs. 1C, S1A lower panel). Subsequently, we analyzed Per1 and Per2 pre-mRNA expression in fibroblast cells derived from wt or Per2 KO animals. After forskolin treatment of the cells we observed induction of Per1 but not Per2 after 25 min in wild-type cells (Figs. 1D, S1B, upper panel). In contrast, Per1 was not induced in Per2 KO cells, but Per2 was slightly increased after 40 min (Figs. 1D, S1B, lower panel). These results are in line with the observation that in the SCN Per1 induction is affected by the loss of Per2. Taken together, our observations suggest that PER2 is important for the regulation of the light-and forskolin-dependent induction of Per1 expression. Of note, other clock genes did not respond in the same way as Per1 after both light and forskolin stimuli, in vivo as well as in vitro ( Fig. S1C-D).
Because cFos is an immediate-early gene (IEG) and can be induced by light in the SCN 23,24 , we wanted to test whether cFos induction may also be affected by the lack of Per2. In the SCN, we observed an induction of c-Fos after an LP in wt animals and surprisingly an even stronger induction in Per2 KO mice (Figs. 1E, S1E). Similarly, forskolin treatment caused stronger cFos induction after forskolin in Per2 KO fibroblasts than in wt cells (Figs. 1F, S1F). In order to corroborate our results, we tested the transcriptional activation of other CREB/CBP targets such as Icer, Nor1 and Nur77. We observed that Icer and Nor1 transcriptional activation after forskolin treatment was comparable to the activation of cFos in wt and Per2 KO cells (Fig. S1G, H). In contrast, transcriptional activation of Nur77 was comparable to the activation of Per1 (Fig. S1I).
These results suggest that PER2 plays an important role in regulating light/forskolin responsive genes, which is different from its role as a circadian regulator. Furthermore, Per2 can modulate light/forskolin responses in a positive (e.g., Per1, Nur77) or negative (e.g., cFos, Icer, Nor1) fashion.
PER2 physically interacts with CREB and modulates its binding to a CRE element in the Per1 regulatory region. Light and forskolin activate signaling pathways that lead to CREB phosphorylation at serine 133 (Ser-133) and subsequently evoke target gene expression such as cFos and Per1 10-12 . The Per1 gene contains a CREB response element (CRE) 14 (Fig. S2A) and is regulated by CREB 14,43 . Since the induction of Per1 gene expression by light or forskolin is affected by the presence or absence of the Per2 gene (Fig. 1C, D), we set out to test whether the PER2 protein affects CREB binding to the Per1 promoter. For this purpose, we studied the CRE-element present in the Per1 regulatory region, including intron 1 (Fig. S2A) that was shown to be functional 14 . We applied forskolin to wt and Per2 KO fibroblast cell lines and collected samples at various time points. Then, we performed chromatin immunoprecipitation (ChIP) using an antibody against CREB, followed by RT-qPCR to measure CREB occupancy at the CRE-element of Per1 and an unrelated element in  www.nature.com/scientificreports/ the Per1 promoter as control (Fig. S2A). We found that forskolin modulated CREB occupancy at the intronic CRE-element of the Per1 gene in a time-dependent fashion with a peak at 25 min after the stimulus. On the other hand, this time-dependent change was absent in the control region ( Fig. 2A, black columns). This transient CREB occupancy increase mirrored the Per1 pre-mRNA profile observed in Fig. 1B. Hence, it appeared that CREB was not recruited in a constant fashion to the Per1 promoter but bound the chromatin upon specific stimulation. Surprisingly, the lack of Per2 led to an increase of CREB binding to the Per1 promoter already after 10 min ( Fig. 2A, grey columns), but did not further increase after 25 min. This indicated that the two profiles of the occupancy of the regulatory region were different between the two genotypes (two-way ANOVA, p < 0.01). Taken together, these results indicate that PER2 modulates CREB recruitment onto the chromatin of Per1 in a stimulus-dependent manner. Next, we asked whether PER2 could affect CREB recruitment to the Per1 gene regulatory region via modulation of CREB phosphorylation at serine 133 (Ser-133), which is associated with light or forskolin-induced phase shifts 44 . Therefore, we performed a ChIP assay using the pSer133-CREB specific antibody as a bait. We observed that this phosphorylated form of CREB was recruited to chromatin in a similar manner as we observed when using the available CREB antibody for both wt (Fig. 2B, black columns) and Per2 KO (Fig. 2B, grey columns) derived chromatin. This result suggests that lack of Per2 did not alter the CREB binding profile to chromatin via changes in phosphorylation of Ser-133.
Above, we described that the lack of the PER2 protein affected the recruitment dynamics of CREB to the CRE-element of the Per1 gene regulatory region. This effect could be of direct or indirect nature. In order to test this, we performed a ChIP assay on the same genomic region using a PER2 specific antibody. Our experiments revealed that the PER2 profile of recruitment to the chromatin (Fig. 2C) was similar to the one for CREB ( Fig. 2A) with a maximum at 25 min after forskolin. Interestingly, the circadian clock component BMAL1 was not recruited to the Per1 CRE element in response to forskolin (Fig. S2B). This observation indicated that the recruitment of PER2 was not due to a general effect on the circadian clock.
Since CREB and PER2 show the same chromatin binding profile upon forskolin stimulation, we wondered whether they might physically interact. We stimulated fibroblast cells with forskolin and collected samples at various time points. Subsequently, we performed a Western blot (WB) and adjusted the protein loading to CREB accumulation. A proper forskolin induction was confirmed by monitoring CREB phosphorylation at serine 133, which was detected with a peak about 25 min after the forskolin stimulus ( Fig. 2D left panel). An immunoprecipitation assay (IP) followed by WB revealed that PER2 binds CREB independently from the forskolin induction and that the interaction is specific (Fig. 2D, right panel, Fig. S2C). Next, we investigated whether we could observe this interaction in vivo in SCN tissue of mice. We immunoprecipitated CREB from SCN protein extracts collected in the dark or 15 min after LP, followed by immunodetection. We observed that PER2 specifically binds CREB (Fig. S2D) and that this was independent of the LP and of CREB phosphorylation (Fig. 2E), as evidenced by the PER2 signal in the absence of the pSer133 signal in the dark. Our in vitro and in vivo data suggest that PER2 physically interacts with CREB and it does so independently from the stimulus (forskolin or light). Therefore, we hypothesize that this interaction influences CREB occupancy in the Per1 regulatory region at the CRE element. PER2 modulates CREB: CRTC1 interaction upon light or forskolin stimulation. CREB occupancy of CRE elements in various promoters has been described to involve a family of factors called CRTCs. It has been demonstrated that in particular CRTC1 localized to the nucleus and dimerized with CREB after exposure to cAMP or calcium 45 . Interestingly, CRTC1 is the main CREB co-activator regulating light responses in the SCN. Upon an LP, CRTC1 was observed to translocate into the nucleus to support CREB activity 46 . These observations prompted us to investigate whether the light or forskolin mediated effect on the induction of the Per1 gene involved CREB: CRTC1 dimerization and whether PER2 modulated this complex.
We applied forskolin to wt and Per2 KO fibroblast cell lines and collected samples at various time points. Then, we performed chromatin immunoprecipitation (ChIP) using an antibody against CRTC1, followed by RT-qPCR to measure CRTC1 occupancy at the CRE-element of Per1 and an unrelated element in the Per1 promoter as control (Fig. 3A). We observed that in wt cells, CRTC1 recruitment to the Per1 intronic chromatin after forskolin treatment displayed a profile comparable to CREB ( Fig. 2A, black columns) with a peak at 25 min after forskolin treatment (Fig. 3A, black columns). Conversely, in Per2 KO cells, the recruitment of CRTC1 to the Per1 promoter after forskolin application was reduced (Fig. 3A, grey column) and comparable to the CREB profile ( Fig. 2A, grey columns). Thus, PER2 appeared to influence CREB and CRTC1 recruitment to the Per1 CRE element in a similar way.
Next, we investigated whether CRTC1 abundance was affected by the lack of PER2. The WB with samples obtained from wt and Per2 KO cell lines collected before and 25 min after forskolin stimulation did not show any difference in CRTC1 levels between the two genotypes. The induction mediated by forskolin increased CRTC1 amounts in both genotypes in a similar manner (Fig. 3B). Interestingly, the interaction between CREB and CRTC1 appeared to be increased by the forskolin stimulus (about 20% of the total input), as shown by immunoprecipitation (IP) using an anti-CREB antibody as bait (Fig. S3A). In the absence of a good knock-out cell line as negative control, we validated the result repeating the IP using either CREB or rabbit IgG (negative control) or TORC-1(CRTC-1, positive reverse CoIP) antibodies as bait. We observed that the interaction between CREB and CRTC-1 is specific, because there was no signal in the IP when the rabbit IgG antibodies were used (Fig. S3B). We wondered whether PER2 might affect CREB: CRTC1 interaction. IP with an anti-CRTC1 antibody as bait revealed that PER2 interacted with CRTC1 independently from the forskolin stimulus (Fig. 3C). Hence, PER2 binds to both CREB (Fig. 2D, E) and CRTC1 (Fig. 3C), although CREB and CRTC1 interact only after a stimulus 45 (Fig. S3A) www.nature.com/scientificreports/ www.nature.com/scientificreports/ www.nature.com/scientificreports/ interacted after this treatment only in wt cells. In contrast, in the Per2 KO cells (Per2 -/-), the interaction was stimulus-independent ( Fig. 3D), as evidenced by the co-IP of both components before forskolin treatment. Next, we investigated whether our observations in cells were also manifested in the SCN of mice. Similar to what we observed in the fibroblast cell line, CREB interacted with CRTC1 only after the light stimulus in SCN samples (Fig. S3C). Immunoprecipitation experiments showed that PER2 interacted with CRTC1 in a specific manner (Fig. S3D). This interaction was independent of an LP, as revealed when sample loading between the two conditions was adjusted to equal amounts of CRTC1 (Fig. 3E). Furthermore, light-induced CRTC1 accumulation was not affected by the lack of PER2 (Fig. 3F), and CREB was still phosphorylated at Ser-133 after the LP, as revealed by WB (Fig. 3F). Although CREB phosphorylation and CRTC1 accumulation were unaffected by PER2, the CREB: CRTC1 interaction was modulated by the clock factor (Fig. 3G) in a comparable manner as observed in cells (Fig. 3D) when a stimulus (light or forskolin) was applied. Additionally, CREB and CRTC1 interacted before the LP in Per2 KO mice (Fig. 3G) which was also observed in cells before the forskolin stimulus (Fig. 3D). Altogether these results define PER2 as a modulator of CREB: CRTC1 interaction. This modulation depends on specific triggers, such as forskolin for cell cultures or light for the SCN cells in animals.
PER2 mediates CREB: CBP interaction upon light or forskolin stimulation. CREB binding protein (CBP) is a histone acetyltransferase that can heterodimerize with CREB when CREB is phosphorylated at Ser-133. This CREB: CBP complex then promotes gene expression 18 . To determine the CBP recruitment profile to the CRE-element of the Per1 regulatory region we performed a ChIP assay using an antibody against CBP, followed by RT-qPCR (Fig. 4A). In wt animals, recruitment of CBP to the CRE-element was very similar (Fig. 4A, left panel, black columns) to the profiles observed for CREB ( Fig. 2A), pSer-133 CREB (Fig. 2B), PER2 (Fig. 2C), and CRTC1 (Fig. 3A). Interestingly, CBP appeared to bind to the control region in the Per1 promoter as well (Fig. 4A, right panel, black columns) possibly due to interaction with BMAL1 47 , since this region contained a BMAL1 binding site (Fig. S2B). In contrast, in Per2 KO cells, CBP did not bind to the CRE-element of Per1 and to a reduced extent to the control element after forskolin treatment compared to background signal at time point 0 (Fig. 4A, grey columns). These results suggested that PER2 affected CBP binding to both Per1 promoter regions.
Since we were interested in the CREB mediated regulation of the Per1 gene, we wondered how PER2 could affect CBP recruitment to the CRE element. We investigated whether PER2 could favor accumulation of CBP in the nucleus. We tested the cells for CBP expression before and after forskolin treatment (Fig. 4B). We saw that CBP accumulation in the nuclei was increased after forskolin treatment, but no distinct difference between the two genotypes was observed (Fig. 4B). We performed immunofluorescence on wt and Per2 KO fibroblasts to evaluate the specificity of our PER2 antibody (Fig. S4A). Next, we tested whether forskolin treatment could induce Ser133 phosphorylation of CREB in our assay using an anti-pSer133 CREB antibody. The phosphorylation of CREB was induced as expected in both wt and Per2 KO cells 25 min after but not before forskolin treatment (Fig. S4B). Since CBP can still accumulate in the nuclei after forskolin induction in the absence of Per2, we investigated whether PER2 could affect the interaction between CBP and CREB. This interaction is essential for the CBP acetyltransferase activity on chromatin. Therefore, we immunoprecipitated CREB at 0 and 25 min after forskolin treatment of cells and performed IP using an antibody against CREB followed by WB (Fig. 4C). We observed that CBP was co-precipitated with CREB 25 min after forskolin treatment.
In contrast, this co-precipitation was not observed in Per2 KO cells (Fig. 4C). Subsequently, we wanted to see whether a Per2-dependent CREB: CBP interaction could also be observed in the SCN of mice after the LP stimulus. Immunohistochemistry on SCN tissue showed that the amount of CBP was increased after LP in both wt and Per2 KO tissue, indicating that lack of Per2 did not affect the increase of CBP levels in response to LP (Fig. 4D, control Fig. S4C). Next, we immunoprecipitated CREB from SCN extracts collected in the dark and after LP at ZT14 of both genotypes (Fig. 4E). Similar to the observation in cells (Fig. 4C), CBP was co-precipitated with CREB in wt extracts but not in extracts of Per2 KO mice. Taken together, our findings indicate that PER2 is necessary for the formation of the CREB: CBP complex after a stimulus such as forskolin or a light pulse.

FRET analysis indicates that PER2 supports the interaction of CREB and CBP domains.
The results presented above lend support to the notion that PER2 serves as a scaffold for CREB: CBP interaction. To further challenge this idea, we performed Förster resonance energy transfer (FRET) experiments, a widely used method to investigate molecular interactions between proteins such as CREB: CBP in living cells 48 . We used a sensor called ICAP (an indicator of CREB activation due to phosphorylation). The sensor is composed of three different elements: 1) the KID domain of CREB containing the Ser-133, which is phosphorylated upon forskolin induction, 2) the KIX domain of CBP, which is responsible for the dimerization with phospho-CREB and 3) a short linker that separates the KID from the KIX domain. KID is flanked by a cyan fluorescent protein (CFP), while KIX is flanked by a yellow fluorescent protein (YFP). When KID is not phosphorylated, the ICAP conformation allows CFP to transfer energy to YFP, producing FRET resulting in yellow light emission. After a stimulus (forskolin), the serine in KID is phosphorylated and binds to KIX. The dimerization separates CFP from YFP, leading to decreased FRET resulting in blue light emission (Fig. 5A upper panel).
We hypothesized that depletion of the endogenous PER2 could affect the KID: KIX dimerization (Fig. 5A  bottom panel). To test this hypothesis, we co-transfected ICAP and either scrambled (scr), or Per2 directed shRNA (shPer2) into NIH3T3 cells. Subsequently, we acquired the FRET signal. 3 min after forskolin treatment, which is the standard latency time for FRET signal detection, we noted that only about 10% of shPer2 transfected cells (Per2 knock-down (KD)) were responsive to the stimulus compared to about 90% of scr transfected cells (Fig. 5B). As evidenced by single-cell traces (Fig. S5A), most shPer2 transfected cells were not responsive, but some with a substantial delay, compared to the scr controls. These results suggested that the knock-down of Per2 www.nature.com/scientificreports/  (Figs. 5C, S5B). 50% of the scr control cells responded to the forskolin treatment within about 3 min, while shPer2 cells needed over 10 min (Fig. 5D). Additionally, www.nature.com/scientificreports/ the relative KID: KIX dimerization was significantly higher in scr control cells compared to shPer2 cells up to around 20 min after the stimulus (Fig. 5E). Altogether these results reinforce our notion that PER2 may support the CREB: CBP dimerization via their respective KID: KIX domains.
PER2 modulates CBP-mediated chromatin acetylation at lysine 27 of histone H3 after forskolin or light stimulation. CBP binds to chromatin by dimerizing with pSer133-CREB to exert its histone acetyltransferase (HAT) activity 49 . Lysine 27 (K27) of histone H3 is the main target of CBP HAT activity to acetylate (Ac) K27 of histone H3 50 . We wondered whether this epigenetic modification (AcH3K27) could be observed on chromatin containing the Per1 CRE-element in cells after forskolin treatment. A ChIP assay, using an anti-AcH3K27 antibody for IP, followed by RT-qPCR revealed that chromatin remodeling at that specific amino acid of histone H3 is forskolin-dependent (Fig. 6A, black columns). We observed highest acetylation of histone H3 in the Per1 CRE-element containing chromatin between 25 and 40 min after stimulus application (Fig. 6A, black columns). Interestingly, acetylation was dramatically reduced in the Per2 KO cells (Fig. 6A, grey  columns). This observation suggested that CBP needed PER2 as a co-factor for assembling a functional complex allowing acetylation of histone H3 on the Per1 CRE-element. This observation is the first evidence indicating that acetylation at K27H3 is a result of PER2-dependent forskolin stimulation. Acetylation of histone H3 leads to the assembly of the RNA polymerase II (RNA Pol II) complex to promote gene transcription 20 . Therefore, we tested whether the formation of such an RNA Pol II containing complex on the Per1 CRE-element was promoted by forskolin treatment of cells and whether PER2 may be involved in this process. We performed a ChIP assay using an antibody against RNA Pol II as bait. We noted that RNA Pol II recruitment to chromatin of wt cells (Fig. 6B, black columns) paralleled the acetylation dynamics of histone H3 on K27 (Fig. 6A, black columns). In contrast, the recruitment of RNA Pol II was lower in cells lacking Per2 (Fig. 6B, grey columns), mirroring the absence of acetylation of H3K27 (Fig. 6A, grey columns). These results provide additional evidence for the importance of PER2 as a facilitating component for assembling the stimulusdependent transcriptional complex on the intronic CRE-element of the Per1 gene.
PER2 can promote the forskolin-induced gene expression of Per1, whereas it can repress that of cFos (Fig. 1). Therefore, we wondered how the chromatin of the CRE element in the cFos promoter was affected after forskolin treatment of cells. Therefore, we performed a ChIP assay using an anti-AcH3K27 antibody for IP, followed by RT-qPCR. Our results showed that the chromatin acetylation profile mirrored the one obtained on the Per1 CRE region (Fig. 6A, C). While we observed in wt samples the highest acetylation of histone H3 in the CRE-element containing chromatin between 25 and 40 min after stimulus application (Fig. 6C, black columns), the same acetylation was profoundly dampened in the Per2 KO samples (Fig. 6C, grey columns). As discussed before histone acetylation leads to the recruitment of RNA Pol II onto the chromatin. Therefore, we tested by ChIP whether forskolin treatment of cells affected the formation of the transcriptional complex including RNA Pol II on the promoter of cFos and whether PER2 could affect this formation. We found that RNA Pol II could be recruited onto the cFos chromatin of wt cells with a profile mirroring the gene expression (Fig. 6D, black columns). Surprisingly, the recruitment of RNA Pol II onto the cFos chromatin in Per2 KO samples was higher compared to wt (Fig. 6D, grey columns). This is consistent with the observation that the cFos pre-mRNA was more abundant in Per2 KO cells compared to wt (Fig. 1F). Together, our results indicate that the forskolin-mediated cFos gene expression is independent of the chromatin acetylation at lysine 27, although no difference was observed between wt and Per2 KO cells in CREB recruitment to the cFos chromatin (Fig. S6A).
A previous study that described light-dependent histone phosphorylation in the SCN of mice 51 prompted us to ask whether acetylation of histone H3 at K27 in the SCN was light-dependent. We applied an LP at ZT14 to mice as described before and subsequently collected SCN tissue. We performed an immunofluorescence assay using an anti AcH3K27 antibody. At ZT14 H3 in the SCN of wt mice displayed a basal level of chromatin acetylation, which was strongly increased after application of an LP (Fig. 6E upper row). In contrast, the immunofluorescence signal was partially dampened in the SCN of Per2 KO mice (Fig. 6E lower row). To confirm this observation, we performed WB on pooled SCN tissue collected in the dark or 15 min after LP from both genotypes (Fig. S6B). Quantification confirmed that AcH3K27 was increased right after the light stimulus in wt SCN tissue, whereas the induction was blunted in SCN extracts obtained from Per2 KO mice (Fig. 6F). Together, these data suggested that the light signaling cascade did regulate chromatin acetylation and that PER2 modulated this process in the SCN.

Discussion
In the present study, we report a role of the clock protein PER2 as a modulator of the light-dependent CREB signaling in the early night. In particular, we observed that PER2 acted as a positive factor in stimulus dependent Per1 and Nur77 gene expression, while it functioned as a negative regulator in cFos, Icer and Nor1 gene induction (Fig. 1C-I). This functional dichotomy indicated that the transcriptional complex involving PER2 is not identical on the Per1 and cFos promoters. That PER2 can act as a positive or negative regulator is consistent with previous observations that this protein has a modulatory influence on transcriptional regulation in both directions in mammals 52,53 . Because deletion of cFos in mice attenuated behavioral responses to light only marginally 54 , we focused on the role of PER2 in the stimulus-dependent activation of the Per1 promoter. Since light activates several signaling pathways in mammals culminating in the induction of the Per1 gene 6,39 , we investigated the role of PER2 as co-factor in CREB mediated transcription, a common downstream target of protein kinase cascades 18 .
ChIP analysis on the intronic CRE-element of the Per1 gene (Fig. S2A) revealed that CREB and its known co-factor CRTC1 are recruited in a time-dependent manner after a forskolin stimulus with a peak around 25 min. However, although CREB and CRTC1 bound to the same region in the Per2 KO cell lines, their binding was less pronounced and occurred earlier, with a peak around 10 min (Figs. 2A, 3A). On the other hand, the histone acetyltransferase CBP and the RNA Pol II were recruited to the same element only in wt, but not in Per2 KO cells www.nature.com/scientificreports/ www.nature.com/scientificreports/ (4A, 6B). Our observations are in agreement with previous findings that describe CREB as an activator of Per transcription 14,55 . Interestingly, PER2 could be recruited to the CRE-element of the Per1 promoter as well, with the same temporal profile as CREB ( Fig. 2A, C). These results suggested that PER2 could be part of the CREBcontaining transcriptional complex. The hypothesis was further corroborated by the observation of a discrete reduction and temporal profile of CREB and CRTC1 binding to the Per1 promoter in Per2 KO cells. Immunoprecipitation experiments revealed that PER2 co-precipitated with CREB (Fig. 2D, E) and with CRTC1 (Fig. 3C, E) in a stimulus-independent manner, suggesting that PER2 could interact with both CREB and CRTC1 before the stimulus occurred (Fig. 7, left). Interestingly, CREB bound to CRTC1 independent of a stimulus when PER2 was absent, however in the presence of PER2 this interaction became stimulus-dependent (Fig. 3D, G). In contrast, interaction between CREB and CBP did not occur in absence of PER2, but appeared in presence of it in a stimulus-dependent fashion (Fig. 4C, E). Hence, a stimulus led to rearrangement of the components CREB, CRTC1 and CBP with PER2 most likely acting as a scaffold to facilitate this rearrangement (Fig. 7,  right). Thus, we conclude that PER2 is the factor that mediates the stimulus-dependent build-up of the complex facilitating CRE element-dependent transcription of genes of the Per1 type family (Per1, Nur77). The limitation of our observation is, that we do not know whether this is a general mechanism or whether this mechanism is restricted to a subset of CRE elements that are flanked by specific uncharacterized regulatory sequences that are present in the Per1 and Nur77 genes. However, the conclusion that PER2 appears to be the stimulus-dependent factor that facilitates gene activation in a subset of CREB dependent target genes seems reasonable. A number of previous observations relate Per gene activation to stimulus-dependent behavioral responses such as entrainment by light 41 , response to drugs [56][57][58][59] , and adaptation to temperature and humidity 60 . The role of PER2 in the CREB complex appears not to be essential (Fig. 1C). However, PER2 increases the transcriptional activation potential of the CREB complex. Since PER2 protein is expressed in a circadian fashion the transcriptional potential of the CREB complex is guided to a particular time window and boosted. Hence, PER2 seems to be important to increase environmental signal transduction to particular times of the day.
Dynamic chromatin remodeling in the SCN has been suggested to occur in response to a physiological stimulus such as light 51 . These findings are in line with our observation that stimulus-dependent activation of Per1 transcription involved the acetylation of histone H3 at lysine 27 and that this acetylation depended on the presence of PER2 (Fig. 6A). The acetylation profile on the Per1 promoter was also mirrored by the recruitment of RNA Pol II and suggested a functional importance of this Per2 dependent H3K27 modification (Fig. 6B). In contrast, recruitment of RNA Pol II on the cFos promoter was independent of Per2 and the H3K27 modification www.nature.com/scientificreports/ (Fig. 6C, D). This suggested that another histone modification is important in the CREB mediated transcriptional activation of promoters of the cFos type family (e.g. Icer, Nor1). The same mechanism as observed in cell cultures on the Per1 promoter is likely to be present in the SCN (Fig. 6E, F). Of note is the strong decrease of the magnitude of H3K27 acetylation in Per2 KO SCN (Fig. 6F). This can't be accounted for by an event happening only at the Per1 genomic locus. It is likely, that PER2 has a more widespread function to promote CBP action in order to acetylate additional genomic loci. Interestingly, light has been described to affect other epigenetic changes in the genome. For example, changes in day length affected the methylation pattern on chromatin in the SCN in a reversible manner 61 . Taken together, it appears that stimuli such as light can affect gene expression via acetylation, methylation, or phosphorylation to modulate transcriptional responses in order to adapt the clock to the current environmental parameters (Fig. 7).
The molecular mechanism that regulates light-dependent responses described here appears to be different from Drosophila, but similar to the one described in Neurospora crassa. In this organism, the transcriptional complex transducing the light signal is composed of WHITE COLLAR 1 (WC-1), WHITE COLLAR-2, and Neurospora GCN5 like-1 (NGF-1) 32,33 . After a light pulse, the large White Collar Complex (WCC) can be rearranged in a similar way as shown in the model we propose here for the mouse (Fig. 7).
Overall our study provides evidence that the PER2 protein can act as a scaffold in the CREB complex. In this role it can influence CREB mediated transcriptional initiation in a positive or negative manner. This most likely depends on additional unknown co-factors necessary for at least two distinct promoter families, one defined by the Per1 promoter and the other by the cFos promoter. Because environmental signals converge in many cases on CREB mediated transcription our model is consistent with the role of PER2 as environmental sensor 41,56,60 . Our observations may be of general nature and will have to be tested on a genome wide scale.

Materials and methods
Animals and housing. Mice were housed with food and water ad libitum in transparent plastic cages  42 were backcrossed over ten generations to obtain Per2 homozygous knock-out mice without Cre recombinase. The MEFs subsequently were serially diluted every three days over 30 passages to yield immortalized cells. All lines were maintained in Dulbecco's modified Eagle's medium (DMEM) containing 4.5 g/l glucose, stable glutamine, sodium pyruvate and 3.7 g/l NaHCO 3 (PAN Biotech), supplemented with 10% fetal calf serum (FCS) and 100 U/mL penicillin-streptomycin at 37 °C in a humidified atmosphere containing 5% CO 2 . Forskolin stimulation (10-100 µM) was used to mimic in vitro the molecular pathway activated by light in mice. Samples were collected at specific time points mentioned in the text.
Plasmids and transfection. The following plasmids were used for the project: RNA extraction from cells. Cells were grown to confluency on 6 cm Petri dishes and induced with 10 µM forskolin (50 mM stock in dimethyl sulfoxide) for the indicated time. Total RNA was extracted using the Nucleospin RNA II kit (Macherey & Nagel) and adjusted to 1 µg/µl with water. An amount of 1 µg was reversetranscribed using Superscript II with random hexamer primers (Thermo Fisher). Real-time PCR was performed using the KAPA probe fast universal master mix and the indicated primers on a Rotorgene 6000 machine. The www.nature.com/scientificreports/ relative expression was calculated compared to the geometric mean of expression of the inert genes Nono, SirT2, Atp5h, and Gsk3b 62 . For a complete list of primers used in the paper, please see Table 1.
RNA extraction from the SCN. RNA from SCN samples was isolated using the Macherey-Nagel RNA Plus kit. Subsequently, 500 ng of purified RNA was used for producing cDNA by reverse transcription (Invitrogen SuperScript II). Real-time PCR was performed using the KAPA probe fast universal master mix and the indicated primers on a Rotorgene 6000 machine.
Chromatin immunoprecipitation. Chromatin immunoprecipitation from cells was performed as described before 63   www.nature.com/scientificreports/ acetic acid) to validate the success of the transfer. The membrane was subsequently washed with TBS 1x/Tween 0.1% and blocked with TBS 1x/Milk 5%/Tween 0.1% for 1 h. After washing, the membrane was stained with the appropriate primary antibodies (Table 2) overnight. The day after, membranes were washed three times with TBS 1x/Tween 0.1% followed by secondary antibody immunoblotting for 1 h at room temperature. The densitometric signal was digitally acquired with the Azure Biosystem.
Immunofluorescence. Mice were cardiovascularly perfused with 0.9% NaCl followed by 4% PFA and brains were collected. They were subsequently cryoprotected by 30% sucrose solution, and sections of SCN were obtained (30 µm, coronal) using a cryostat. Selected sections were washed three times with 1 × TBS (0.1 M Tris/0.15 M NaCl) followed by 2 × SSC (0.3 M NaCl/0.03 M tri-Na-citrate pH 7). Antigen retrieval was performed with 50% formamide/2 × SSC by heating to 65 °C for 50 min followed by washing 2 × SSC and three times in 1 × TBS pH 7.5. Slices were blocked for 1.5 h in 10% fetal bovine serum (Gibco)/0.1% Triton X-100/1 × TBS at RT. After the blocking, the primary antibodies (Table 2) were added to the sections and incubated overnight at 4 °C. The next day, sections were washed with 1 × TBS and incubated with the appropriate fluorescent secondary antibodies diluted 1:500 in 1% FBS/0.1% Triton X-100/1 × TBS for 3 h at RT. (Lot: 132876, Alexa Fluor647-AffiniPure Donkey Anti-Mouse IgG (H + L) no. 715-605-150, Lot: 131725, Alexa Fluor647-AffiniPure Donkey Anti-Rabbit IgG (H + L) no. 711-602-152, Lot: 136317 and all from Jackson Immuno Research). Tissue sections were stained with DAPI (1:5000 in PBS; Roche) for 10 min. Finally, the tissue sections were rewashed twice in 1 × TBS and mounted on glass microscope slides. Fluorescent images were taken using a confocal microscope (Leica TCS SP5), and pictures were taken with a magnification of 10 ×, 20 ×, or 63 ×. Images were processed with the Leica Application Suite Advanced Fluorescence 2.7.3.9723, according to the study by 64 . Live FRET imaging. Transfected NIH3T3 cells were starved for 4 h with 0.5% FBS DMEM. After starvation (0.5% FBS in DMEM), the medium was removed, and cells were washed twice with 1 × HBTS without CaCl 2 and MgCl 2 (25 mM HEPES, 119 mM NaCl, 6gr/L Glucose, 5 mM KCl) gently to avoid cell disturbance. One plate at the time was imaged. Therefore, the medium was changed shortly before the imaging started. NIH3T3 cells were imaged using an inverted epifluorescence microscope (Leica DMI6000B) with an HCX PL Fluotar 5x/0.15 CORR dry objective, a Leica DFC360FX CCD camera (1.4 M pixels, 20 fps), and EL6000 Light Source, and equipped with fast filter wheels for FRET imaging. Excitation filters for CFP and FRET: 427 nm (BP 427/10). Emission filters for CFP: 472 (BP 472/30) and FRET: 542 nm (BP 542/27). Dichroic mirror: RCY 440/520. One frame every 20 s was acquired for at least 90 cycles (0.05 Hz frequency), and the recording lasted at least 30 min. This acquisition rate was ideal to avoid cellular photobleaching and phototoxicity during long recordings. The baseline response in the presence of HBTS only was recorded for 2 min and 40 s. At minute 3:00, 100 µM Forskolin, 2 mM CaCl 2 , and 2 mM MgCl 2 were added to the cells. To avoid buffer perturbations due to the addition of the drug stimulus, the latter was added between frames, so it had approximately 20 s to diffuse and activate a response in the cells.
FRET imaging analysis. The time-lapse recordings were analyzed using LAS X software (Leica), adapting a previously described method 65 . Briefly, two regions of interest (ROI) were selected for each cell, and 50 cells per plate were chosen randomly. A first ROI delimiting the background and a second ROI including the cell nucleus of NIH3T3 cell expressing NLS KIDKIX were used per cell. For each channel, the ROIbackground values were subtracted from the ROIcell values. For baseline normalization, the FRET ratio R was expressed as a ΔR/R, where ΔR is R-R0 and R0 is the average of R over the last 120 s prior stimulus.