CRTC1 mediates preferential transcription at neuronal activity-regulated CRE/TATA promoters

Gene expression mediated by the transcription factor cAMP-responsive element-binding protein (CREB) is essential for a wide range of brain processes. The transcriptional coactivartor CREB-regulated transcription coactivator-1 (CRTC1) is required for efficient induction of CREB target genes during neuronal activity. However, the mechanisms regulating induction of specific CREB/CRTC1-dependent genes during neuronal activity remain largely unclear. Here, we investigated the molecular mechanisms regulating activity-dependent gene transcription upon activation of the CREB/CRTC1 signaling pathway in neurons. Depolarization and cAMP signals induce preferential transcription of activity-dependent genes containing promoters with proximal CRE/TATA sequences, such as c-fos, Dusp1, Nr4a1, Nr4a2 and Ptgs2, but not genes with proximal CRE/TATA-less promoters (e.g. Nr4a3, Presenilin-1 and Presenilin-2). Notably, biochemical and chromatin immunoprecipitation analyses reveal constitutive binding of CREB to target gene promoters in the absence of neuronal activity, whereas recruitment of CRTC1 to proximal CRE/TATA promoters depends on neuronal activity. Neuronal activity induces rapid CRTC1 dephosphorylation, nuclear translocation and binding to endogenous CREB. These results indicate that neuronal activity induces a preferential binding of CRTC1 to the transcriptional complex in CRE/TATA-containing promoters to engage activity-dependent transcription in neurons.

To test whether the proximity of CRE sites to the TSS and/or presence of TATA boxes in the promoter could affect gene expression, we examined the levels of Psen1 and Psen2, which contain promoters with distal consensus CRE/TATA sites (Psen1: −1839 and −1816 bp; Psen2: −3526 and −3464 bp) and proximal CRE/TATA-less sites (Psen1: 244 bp) from the TSS in both mouse and human genes (Table 1 and Supplementary Table S1). Interestingly, treatment did not significantly affect levels of Psen1 or Psen2 at any stimulation time, except for a Figure 1. Neuronal activity induces CREB-dependent gene expression in primary cortical neurons. Quantitative real-time RT-PCR analysis of mRNA levels of CREB target genes in cultured cortical neurons (10 DIV). Levels of c-fos, Dusp1, Nr4a1, Nr4a2 and Ptgs2 mRNAs were differentially increased after KCl, FSK or FSK/KCl treatment in a time-dependent manner. On the contrary, Gapdh mRNA levels were not significantly different between vehicle-(Veh) and FSK/KCl-treated neurons at any time point, while Psen1 and Psen2 mRNA levels were significantly changed only after 8 hours of stimulation. Data represent fold change ± s.e.m relative to vehicle (Veh)-treated neurons from three independent experiments (n = 3). Gene expression levels were normalized to the geometric mean of Ppia, Tbp and Gapdh. N.s: non-significant. *P < 0.05, **P < 0.01, ***P < 0.001, compared to the indicated experimental group as determined by two-way ANOVA followed by Bonferroni test. Activity-dependent CRTC1 activation in neurons. We next studied the molecular mechanisms that regulate activity-dependent CREB transcription in neurons by focusing on CRTC1, a CREB transcriptional coactivator highly expressed in neurons of cortex and hippocampus 8,9 . FSK/KCl induced a significant time-dependent increase of CREB phosphorylation (Ser133) reaching a peak at 15 min (F (8,26) = 2.4, P < 0.05; Fig. 2A). By contrast, chemical-induced neuronal activity induced a rapid (t 1/2 ≈ 1 min) and persistent dephosphorylation of CRTC1 in cultured neurons (F (8,26) = 2.9, P < 0.03), which is associated with translocation of CRTC1 from the cytosol to the nucleus ( Fig. 2A,B).
Activity-dependent recruitment of CRTC1 to target gene promoters. The above results suggest that neuronal activity might differentially activate expression of genes containing CRE sites in their promoters. Gene sequence analyses revealed that the number of proximal CRE sequences (−500 bp to 300 bp of the TSS) containing TATA boxes (CRE/TATA) in the promoter regions differ substantially among the analyzed genes both in human and mouse genomes ( Table 1, Supplementary Table S1). For the following analyzed mouse genes are indicated in parenthesis the number of proximal total CRE and CRE/TATA boxes: c-fos (5, 3), Dusp1 (2,2), Nr4a1 (4,2), Nr4a2 (3,3), Nr4a3 (0,0), Psen1 (1,0), Psen2 (0,0) and Ptgs2 (2,1). We then explored the possibility that CREB and CRTC1 could bind differentially to the proximal promoter gene regions depending on the presence of CRE sites in close proximity to TATA boxes (CRE/TATA sites) by performing quantitative chromatin immunoprecipitation (ChiP-qPCR) analyses using CREB and CRTC1 antibodies and different set of primers amplifying distinct regions of CRE-containing promoters (Fig. 3A). In agreement with a previous report showing recruitment of CRTC1 into CREB target promoters 16 , neuronal activity induced a significant binding of CRTC1 to the proximal CRE/TATA promoter regions of c-fos, Nr4a1 and Nr4a2 (Fig. 3B). Notably, CREB binds constitutively to similar gene proximal promoter regions already in the absence neuronal stimulation (Fig. 3C). Conversely, we detected low binding of CREB and CRTC1 to a CRE/TATA-less site located distally of the TSS (−2698 bp) in the Nr4a3 promoter region (Fig. 3B,C). To examine whether spontaneous neuronal activity was responsible for basal binding of CREB to gene promoters, we performed similar assays in the presence or absence of tetrodotoxin (TTX), a Na + channel blocker that inhibits action potentials. In contrast to reduced binding of CRTC1 to the proximal c-fos promoter in the presence of TTX, basal binding of CREB to this promoter region was largely unaffected by TTX (Fig. 3D). These results suggest a preferential binding of CRTC1 to proximal CRE/TATA-containing target promoters in response to activity resulting in efficient CREB/CRTC1-dependent transcription in neurons.
To evaluate further whether lack of proximal CRE/TATA sequences affects binding of CREB and CRTC1 to target gene promoters, we extended the ChIP-qPCR analysis to the promoter regions of Psen1 and Psen2. CRTC1 does not bind to the promoter regions of Psen1 and Psen2 in basal or stimulated conditions (Fig. 4A,B). By contrast, we found a significant binding of CREB at a proximal CRE/TATA-less containing promoter region of Psen1 (244 bp) in basal but not stimulated conditions (Fig. 4A,C; P < 0.05 Veh vs FSK/KCl). Similarly, CREB occupancy was significantly enriched close to the most distal CRE/TATA-less site of the Psen2 promoter (−4412 bp) in basal but not stimulated conditions (P < 0.05 Veh vs FSK/KCl). Since CREB binds to the proximal promoter region of Psen1 in basal conditions, it is possible that CREB could regulate Psen1 gene expression independently of CRTC1. To test this idea, we developed four different ShRNAs (ShRNA 1-4) targeting the mouse CREB gene (Creb1). Creb1 ShRNA1 and 2 significantly reduced Creb1 protein (F (4,14) = 76.86, P < 0.0001) and mRNA (F (4,14) = 14.59, P < 0.001) in primary neurons (Fig. 5A,B). Notably, Creb1 silencing reduced Nr4a2 (NURR1) protein levels but did not affect PS1 protein levels, as assayed with an antibody that recognize the PS1 C-terminal fragment (CTF), in human HEK-293T cells and cortical neurons (Fig. 5C,D). These results indicate that CREB does not regulate transcription of Psen1, which supports the idea that activity-dependent CRTC1/CREB binding to proximal CRE/ TATA-containing promoter regions is required for optimal gene transcription. dependent manner. Real-time qRT-PCR analysis of mRNA levels normalized to Gapdh in neurons infected with scramble ShRNA (black dots) and Crtc1 ShRNA (red dots). Data represent mean percentage ± s.e.m of three independent experiments performed by triplicate. "#" and "##" represent statistical differences on time and treatment, respectively. *P < 0.05, ***P < 0.001, compared to scramble ShRNA-infected neurons in a specific time point. Statistical analysis was determined by two-way ANOVA followed by Bonferroni test.  suggested that cooperative interactions between the CRTC isoform CRTC2 and CREB/CBP complexes regulate selectively gene transcription 12,20,21 . To test whether neuronal activity induces binding of endogenous CRTC1, the main isoform expressed in neurons, to the CREB transcriptional complex, we next performed coimmunoprecipitation experiments of endogenous CRTC1 and CREB in non-stimulated and stimulated cultured cortical neurons. FSK/KCl efficiently induced CREB phosphorylation and CRTC1 dephosphorylation in primary neurons (Fig. 6A). Immunoblotting confirmed successful immunoprecipitation of CRTC1 and phosphorylated CREB with their respective antibodies. Interestingly, immunoblotting revealed the presence of endogenous CRTC1 in CREB immunoprecipitates and CREB in CRTC1 immunoprecipitates in stimulated conditions (Fig. 6A). Indeed, in FSK/KCl conditions, the percentage of CREB (∼50%) that binds to CRTC1 is similar to the amount of CRTC1 (∼37%) bound to CREB. These results demonstrate that neuronal activity induces CRTC1 assembly into a CREB-containing transcriptional complex.

Discussion
The transcription factor CREB activates the expression of hundred of genes in neurons 2 , but the molecular mechanisms regulating specific transcriptional programs in response to neuronal activity are still largely unclear. The elucidation of these transcriptional mechanisms may be important in cognitive disorders in which gene expression changes parallel synaptic plasticity and memory deficits 22 . This study reveals that whereas CREB binds constitutively to proximal CRE sequences of gene promoters, CRTC1 is preferentially recruited to target proximal promoter regions upon neuronal activity to engage activity-dependent gene transcription. Neuronal activity induces CRTC1/CREB-dependent transcription by a mechanism involving CRTC1 dephosphorylation, nuclear translocation and binding to CREB preferentially into proximal CRE/TATA-containing promoters. These results suggest a transcription model in which neuronal activity promotes cooperative interactions between CRTC1 and CREB/CBP into specific proximal target promoter regions to activate selectively expression of target genes in neurons (Fig. 6B).
The first interesting finding is that neuronal activity induces robust activation of CREB target genes containing proximal CRE/TATA consensus promoter sequences. Notably, whereas CREB binds constitutively to target gene promoters, robust activity-induced transcription is associated with increased recruitment of CRTC1 to the promoter proximal regions containing conserved CRE/TATA sequences. Basal recruitment of CREB to gene promoters is largely independent of neuronal activity since blocking spontaneous activity with TTX does not abolish its binding to the c-fos promoter. These results are consistent with genome-wide studies reporting constitutive binding of CREB to CRE-containing promoters, whereas depolarization increases binding of CREB and CBP at gene enhancers in neurons 18,23 . Indeed, CRTC1 was recently reported to bind to the proximal promoter regions of c-fos and Bdnf IV as well as to the distal (7 kb) half CRE site of a SARE enhancer element of Arc 16,24 . In contrast to binding to proximal CRE/TATA-containing promoters of target genes (e.g c-fos, Nr4a1 and Nr4a2), CRTC1 does not bind to CRE/TATA-less promoter regions of Nr4a3, Psen1 and Psen2. This result is consistent with the requirement of proximal CRE/TATA sequences for efficient CREB-mediated transcription in non-neuronal cells 17 . The absence of CRTC1 recruitment to Psen1 and Psen2 promoters contrasts with basal binding of CREB to the proximal promoter regions of Psen1, a result suggesting the possibility that CREB could regulate Psen1 transcription depending on the physiological conditions. For instance, neuronal activity reduces binding of CREB to the Psen1 promoter while is recruited to other target gene promoters, a result resembling that observed when overexpressing CRTC1 in vivo 24 . However, whereas neuronal activity does not affect Psen1 mRNA levels in cortical neurons, glutamate and BDNF activate Psen1 expression in a CREB-dependent manner in neuroblastoma cells 25 . Notably, multiple consensus transcription binding sites are present in the promoter proximal region (−811 bp to +140 bp) of the human PSEN1 gene including those for CREB, Lyf-1, CdxA, AML-1a, Nkx-2, Pbx-1, E47, MZF1, Sp1 and Elk-1 25 . Our biochemical studies showing that CREB inactivation does not affect Psen1 levels in neurons reinforce the view that CREB is unlikely to be the main transcription factor regulating Psen1 expression. Likewise, we also detected recruitment of CREB to a full CRE site located distally (−4412 bp) from the transcription start site of the Psen2 promoter. It is relevant, however, that presenilins regulate CREB-mediated transcription in the adult brain likely through a mechanism involving the transcriptional coactivators CBP and CRTC1 [26][27][28] . Future experiments will be necessary to elucidate the presenilin-regulated mechanisms of CREB-dependent transcription, and the role of proximal and distal CRE sites as regulatory elements (initiation, enhancers…) of the presenilin genes.
CREB-dependent transcription depends, at the molecular level, on CREB phosphorylation at Ser133 3,29,30 . However, although CREB phosphorylation is important it does not always correlate with activation of gene expression suggesting that phosphorylation kinetics and/or other transcriptional effectors regulate gene transcription 5,31,32 . Our results indicate that Ca 2+ and cAMP signals have differential effects on expression of CREB target genes in neurons, whereas simultaneous Ca 2+ /cAMP signals result in additive or synergistic effects on gene expression. This differential effect on gene expression is unlikely due to differences on CREB phosphorylation.  For instance, we found that Ca 2+ /cAMP signals induce gradual increase of CREB phosphorylation but rapid CRTC1 dephosphorylation in cortical neurons (see also ref. 33 ). It is known that besides inducing similar levels of CREB phosphorylation, cAMP is more effective than stress signals in promoting CREB/CBP complex formation and gene transcription 34 . CBP recruitment to CREB is sufficient for CREB-mediated gene activation, whereas efficient transcriptional induction does not only requires CREB/CBP complex formation but also optimal cAMP signaling and histone acetylation 31,35,36 . Indeed, CRTCs potentiate CREB transcriptional activity through binding to CREB independently of CREB Ser133 phosphorylation 9,21 . Accordingly, our biochemical analysis showed that the Ca 2+ /cAMP-induced potentiation of gene transcription involves recruitment of CRTC1 to CREB/CBP complexes in target promoter regions. This result agrees with previous reports showing that Ca 2+ and cAMP signals act synergistically on gene expression by activating CRTC1 or CRTC2 in neurons and non-neuronal cells, respectively 12,37 . Indeed, neuronal activity regulates CRTC1 nuclear translocation while cAMP modulates its persistence into the nucleus 11 . Conversely, Ca 2+ and cAMP signals do not always cooperate to activate gene expression in response to neuronal activity 38 . The fact that genetic inactivation of CRTC1 blocks Ca 2+ /cAMP-mediated transcription strongly suggests that CRTC1 recruitment to CREB is essential for activity-induced CREB-dependent transcription in neurons. Importantly, neuronal activity and memory training activate preferentially CREB-dependent transcriptional programs 39,40 . In agreement, CRTC1 selectively regulates expression of CREB target genes involved in synaptic plasticity and memory, including among others Bdnf, c-fos, Dusp1, Fgf1, Nr4a1 and Nr4a2 16,24,36,37,41 .
In summary, this study shows that binding of CRTC1 to neuronal gene promoters depends on neuronal activity resulting in induction and/or maintenance of CREB-dependent transcription in neurons. Despite basal binding of CREB to gene promoters, efficient CREB-dependent transcription depends on activity-induced recruitment of CRTC1 to promoters containing proximal CRE/TATA elements. These results are consistent with the idea that the presence of consensus CRE/TATA sequences within the core promoter is required for efficient activation of CREB-dependent transcription mediated by interaction of CREB and the transcriptional complex in non-neuronal cells 9,17 . CRTC1 dephosphorylation and nuclear translocation are key molecular mechanisms triggering activity-dependent CREB target gene transcription in neurons 11,42 . Conversely, deficient dephosphorylation and nuclear translocation of CRTC1 cause transcriptional and memory deficits in experimental models of Alzheimer's disease 24,43 . Future investigations on the mechanisms and effectors involved in CREB-dependent gene transcription may shed light on novel pathways and therapeutics in age-related cognitive disorders.
Briefly, 12 DIV cortical neurons were treated with vehicle or FSK (20 μM) and KCl (30 mM) for 15 min, or pretreated with tetrodotoxin (TTX; 1 μM) for 12 h. Cells were crosslinked with 1% formaldehyde, lysed in ChIP buffer (50 mM Tris-HCl pH 8.1, 100 mM NaCl, 5 mM EDTA, 1% SDS, 0,1% Na deoxycholate and protease/phosphatase inhibitors) and chromatin was sheared between 200 and 500 bp by sonication using a BioruptorPlus (Diagenode, Seraing, Belgium). Fragmented chromatin was analyzed using the High Sensitivity DNA Kit (Agilent Technologies). Immunoprecipitation (2.5 μg DNA) was performed overnight in diluted ChIP buffer (0.1% SDS, 1,1% Triton X-100) with or without anti-CRTC1 or CREB antibodies (Cell Signaling). Input and immunoprecipitated DNA were decrosslinked and amplified by real-time qPCR using specific primers, and the fold enrichment of the amplified target regions was calculated over an irrelevant region in the chromosome 4. Statistical analysis. Statistical analysis was performed using t-test, one-way or two-way analysis of variance (ANOVA) for multiple comparisons followed by Bonferroni or Dunnett's tests using GraphPad Prism software. Differences with P < 0.05 were considered significant.
Ethical experimental statement. This study was performed in accordance with the experimental European Union guidelines and regulations (2010/63/EU). Experimental protocols and experiments involving vertebrate animals were conducted in accordance with the ethical protocol approved by the Animal and Human Ethical Committee (CEEAH) of the Universitat Autònoma de Barcelona (protocol number: CEEAH 2896) and the local Governmental Ethical Committee of the Generalitat de Catalunya (protocol number: DMAH 8787).

Data availability statement.
All data generated during this study are included in this published article or are available from the corresponding author upon reasonable request.