Region-Specific Activation of CRTC1-CREB Signaling Mediates

Neuron Article

Region-Speciﬁc Activation of CRTC1-CREB Signaling Mediates Long-Term Fear Memory

Mio Nonaka,1,4,6 Ryang Kim,1,4 Hotaka Fukushima,2 Kazuki Sasaki,3 Kanzo Suzuki,1,4 Michiko Okamura,1 Yuichiro Ishii,1,4 Takashi Kawashima,1,7 Satoshi Kamijo,1,4 Sayaka Takemoto-Kimura,1,5 Hiroyuki Okuno,1,4,8 Satoshi Kida,2,4 and Haruhiko Bito1,4,* 1Department of Neurochemistry, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan 2Department of Bioscience, Faculty of Applied Bioscience, Tokyo University of Agriculture, 1-1-1 Sakuragaoka, Setagaya-ku, Tokyo 156-8502, Japan 3Department of Molecular Pharmacology, National Cerebral and Cardiovascular Center Research Institute, Suita, Osaka 565-8565, Japan 4CREST-JST, Chiyoda-ku, Tokyo 102-0076, Japan 5PRESTO-JST, Kawaguchi, Saitama 332-0012, Japan 6Present address: Centre for Cognitive and Neural Systems, The University of Edinburgh, Edinburgh EH8 9JZ, UK 7Present address: Howard Hughes Medical Institute, Janelia Farm Research Campus, Ashburn, Virginia 20147, USA 8Present address: Medical Innovation Center, Graduate School of Medicine, Kyoto University, Kyoto 606-8507, Japan *Correspondence: [email protected] http://dx.doi.org/10.1016/j.neuron.2014.08.049

SUMMARY Greenberg, 2008; Josselyn et al., 2001; Kida et al., 2002; Okuno et al., 2012; Silva et al., 1998; Suzuki et al., 2011). The expres- CREB is a pivotal mediator of activity-regulated gene sion of various neuronal immediate early genes (IEGs)—genes transcription that underlies memory formation and that are primarily induced by neuronal activity—such as c-Fos, allocation. The contribution of a key CREB cofactor, brain-derived neurotrophic factor (BDNF), and activity-regulated CREB-regulated transcription coactivator 1 (CRTC1), cytoskeleton-associated protein (Arc, also known as Arg3.1)is has, however, remained elusive. Here we show that predominantly CREB dependent (Bito et al., 1997; Okuno, several constitutive kinase pathways and an ac- 2011). Local manipulation of CREB activation has further revealed strong evidence supporting a critical role of CREB in tivity-regulated phosphatase, calcineurin, converge the formation of memory engrams and allocation of memory to determine the nucleocytoplasmic shuttling of within an active set of neurons (Cowansage et al., 2013; Han CRTC1. This, in turn, triggered an activity-dependent et al., 2007; Zhou et al., 2009). In spite of this significance, our association of CRTC1 with CREB-dependent re- current understanding about the mechanisms by which the pro- gulatory elements found on IEG promoters. Forced moters and enhancers of such IEGs are robustly stimulated by expression of nuclear CRTC1 in hippocampal neu- neuronal activity through convergent effects of different sets of rons activated CREB-dependent transcription, and transcription factors and associated coactivators remains was sufficient to enhance contextual fear memory. limited. Furthermore, we lack insights on how differential coordi- Surprisingly, during contextual fear conditioning, nation of transcription may regulate engrams in distinct brain we found evidence of nuclear recruitment of endog- areas and cooperatively produce a long-term modification of enous CRTC1 only in the basolateral amygdala, and cognitive performance (Lonze and Ginty, 2002; Nonaka, 2009; Nonaka et al., 2014). not in the hippocampus. Consistently, CRTC1 knock- It has been suggested that IEG expression profiles during down in the amygdala, but not in the hippocampus, basal and activated states may differ among brain regions, and significantly attenuated fear memory. Thus, CRTC1 such data were considered as functional evidence in support has a wide impact on CREB-dependent memory pro- for an uneven contribution of distinct brain regions during cogni- cesses, but fine-tunes CREB output in a region-spe- tive behavior (Frankland et al., 2004; Okuno, 2011; Tse et al., cific manner. 2011). In keeping with this, recent work has indicated differential CREB phosphorylation profiles and time courses in different brain regions during reconsolidation and extinction of fear mem- INTRODUCTION ory (Cowansage et al., 2013; Mamiya et al., 2009). However, it remains unsolved whether the distinctive neuronal activities in Ca2+/cAMP-responsive element-binding protein (CREB) is a those brain areas are the sole determinants of such differential ubiquitous transcription factor and is activated by phosphoryla- activity-regulated transcription. An alternative hypothesis that tion in response to external stimuli (Mayr and Montminy, 2001). has not been tested so far is the possibility that cellular consoli- In the nervous system, CREB-dependent gene expression is dation of memory-related events might employ distinct CREB essential for long-term memory and plasticity (Bito and Take- complexes in different brain areas to convert incoming synaptic moto-Kimura, 2003; Bourtchuladze et al., 1994; Greer and inputs to long-term events.

92 Neuron 84, 92–106, October 1, 2014 ª2014 Elsevier Inc. Neuron Region-Speciﬁc Activation of CRTC1-CREB in Memory

CREB-regulated transcriptional coactivator (CRTC), also cific manner that is distinct from the regulation of CREB phos- known as transducer of regulated CREB activity (TORC), has phorylation. Thus, CRTC1 may act as a localized, rather than recently emerged as a unique CREB coactivator that binds to global, tunable switching mechanism for specifying CREB the basic leucine zipper (bZIP) region of CREB (Conkright output and activity-dependent transcription, contributing to opti- et al., 2003). CRTCs are Ser/Thr-rich proteins (Ser/Thr composi- mizing the formation and allocation of memory across brain tion = 18.3%–21.1%) with multiple phosphorylation and glyco- regions. sylation sites. Among the known isoforms of CRTC (CRTC1, 2, and 3), CRTC2 has been best explored. In CRTC2, the cova- RESULTS lent modification of Ser171, by alternative phosphorylation/glyco- sylation, determines its subcellular localization: nuclear in the Nuclear Shuttling of CRTC1 Is Regulated by glycosylated or virgin state, and cytoplasmic in the phosphory- Counteraction of Constitutive Kinase Pathways and lated states. CRTC2 plays key roles in gluconeogenesis, insulin Calcineurin-Dependent Dephosphorylation on Two Key signaling, and cell proliferation in the liver and pancreas (Altare- Serine Residues, Ser151 and Ser245 jos and Montminy, 2011), whereas CRTC3 contributes to the CRTC1 is a Ser/Thr-rich protein (Ser/Thr content = 21.1%). development of obesity by attenuating b-adrenergic receptor To enable high coverage identification for critical phosphory- signaling in adipose tissue (Song et al., 2010). CRTC1-null lated residues, we performed mass spectrometric analysis of mice also develop obesity as a result of dysregulation of the hy- CRTC1 immunoprecipitated from cultured cortical neurons, pothalamic feeding center (Altarejos et al., 2008). Consistently, in and identified 11 Ser or Thr phosphorylation sites (Figure 1A). Caenorhabditis elegans, where only one copy of CRTC gene ex- To investigate activity-dependent regulation of these candidate ists, CRTC1-dependent transcription via CREB is critical for the phosphorylation sites, we established a translocation assay in control of energy homeostasis and longevity (Mair et al., 2011). primary cultures of mouse hippocampal neurons. The neurons Interestingly, CRTC1, an isoform that is most abundantly ex- were initially silenced with 1 mM tetrodotoxin (TTX) in order to pressed in the central nervous system (Watts et al., 2011), trans- restrict CRTC1 in the cytoplasm and then synaptically activated locates to the nucleus upon receipt of dendritic synaptic inputs by washing out TTX and adding a BIC/4AP cocktail (30 mM (Ch’ng et al., 2012). CRTC1 has been implicated in regulation bicuculline, 100 mM 4-aminopyridine, 100 mM glycine, and of the dendrite arborization of developing cortical neurons (Fin- 1 mM strychnine) to induce nuclear translocation of CRTC1 (Fig- sterwald et al., 2010; Li et al., 2009), in the control of neuronal ure 1B). The neurons were fixed after various incubation times, survival after ischemia (Sasaki et al., 2011), and regulation of and CRTC1 immunoreactivity (IR) was measured in the cyto- circadian clock (Jagannath et al., 2013; Sakamoto et al., 2013). plasm and in the nucleus (Figures 1C and 1D). The activity- Despite the critical importance of nuclear CREB-regulated dependent nuclear translocation of CRTC1 was rapid, achieving transcription in cognition, the mechanisms underlying the nucle- a peak and plateau within 2 min after stimulus onset (Figure 1C). ocytoplasmic shuttling of CRTC1, and CRTC1’s impact on mem- Once the culture medium was returned to the TTX-containing ory-related processes, have remained elusive. solution, CRTC1 was exported from the nucleus within 30 min In order to address these questions, we here identified key (Figure 1D). The distribution and redistribution of CRTC1 may amino acid residues underlying nuclear shuttling of neuronal thus serve as a dynamic trace that integrates the incoming syn- CRTC1 via pharmacological and mass spectrometric analyses. aptic activity of the immediate past. Quantitative understanding of CRTC1’s phosphorylation states Diverse kinase and phosphatase pathways have previously enabled rational designing of a constitutively active mutant been implicated in regulation of the nucleocytoplasmic shuttling form of CRTC1 (caCRTC1) whose localization was restricted to of CRTC1–3 (Koo et al., 2005; Screaton et al., 2004). We there- the nucleus of neurons both in vitro and in vivo. Chromatin immu- fore examined the kinases that controlled the phosphorylation noprecipitation (ChIP) assays revealed an activity-dependent states of neuronal CRTC1. Phenformin, a potent activator of association of CRTC1 with CREB-regulated elements on active many kinases including AMP-activated protein kinase (AMPK), IEG promoters. Taking advantage of caCRTC1, we then found prevented the occurrence of both the activity-dependent nuclear that the nuclear accumulation of CRTC1 was sufficient to mark- translocation of CRTC1 (Figure S1A available online) and BIC/ edly stimulate neuronal CREB-dependent gene expression even 4AP-induced CRE-dependent transcription activity (Figure 1E). in the absence of evoked synaptic activity in vitro or in naive mice Conversely, compound C, an inhibitor of AMPK and salt-induc- in vivo. Mice virally expressing caCRTC1 in the dorsal hippocam- ible kinase 2 (SIK2), promoted the nuclear translocation of pus showed enhancement of long-term contextual fear memory, CRTC1 in the presence of spontaneous activity (Figure S1B), which was accompanied by enlargement of dendritic spines. as well as increased the extent of synaptically induced CRE- Surprisingly, during contextual fear conditioning (CFC), nuclear dependent promoter activity (Figure 1F). Consistent with these recruitment of endogenous CRTC1 was detected in the basolat- results, additional experiments showed that kinases that have eral amygdala (BLA), but not in the hippocampus. In keeping with been suggested to function upstream of SIK2, including protein this, CRTC1 knockdown in the amygdala, but not in the hippo- kinase A (PKA) (Figure S1C) as well as a Ca2+/calmodulin-depen- campus, significantly attenuated fear memory. Taken together, dent kinase kinase (CaMKK)-CaMKI/IV cascade (Figure S1D), our study provides evidence for a wide impact of CRTC1 on contributed to the regulation of the nuclear translocation of nuclear CREB-mediated memory processes, and also indicates CRTC1. Furthermore, overexpression of constitutively active or that the phosphorylation states critical for activity-dependent dominant-negative forms of microtubule affinity-regulating ki- nuclear translocation of CRTC1 are controlled in a region-spe- nase 2 (MARK2) shifted CRTC1 localization to the cytoplasm

Neuron 84, 92–106, October 1, 2014 ª2014 Elsevier Inc. 93 Neuron Region-Speciﬁc Activation of CRTC1-CREB in Memory

(legend on next page) 94 Neuron 84, 92–106, October 1, 2014 ª2014 Elsevier Inc. Neuron Region-Speciﬁc Activation of CRTC1-CREB in Memory

or nucleus, respectively (Figure 1G), implicating MARK2 as an in- the basal phosphorylation by a constitutive kinase, MARK2, dependent upstream regulator of neuronal CRTC1. In contrast, and by SIK/AMPK, appeared to critically regulate the calci- we found that a cocktail consisting of inhibitors of a Ca2+/ neurin-triggered nucleocytoplasmic CRTC1 shuttling. calmodulin-dependent protein phosphatase calcineurin (CsA/ FK, cyclosporin A and FK506) attenuated the activity-dependent Forced Nuclear Translocation of CRTC1 Is Sufficient to nuclear translocation of CRTC1 in a concentration-dependent Trigger CRE-Dependent Gene Expression manner (Figure 1H), while also inhibiting activity-induced CRE- We next examined the functional relevance of these two phos- dependent promoter activity (Figure 1F). phorylation/dephosphorylation sites, Ser151 and Ser245,on Together, these findings indicated that in synaptically con- CREB-dependent transcription. The CRTC1 double mutant nected hippocampal neurons, the basal constitutive phos- (S151A/S245A or caCRTC1) localized almost exclusively to the phorylation states of CRTC1 were largely under the control of nucleus even in the presence of TTX (Figure 3A), although muta- SIK2/AMPK and MARK2. In combination, these constitutive ki- tion of either site alone was not sufficient to induce such a de- nases gated the onset of CRTC1 nuclear translocation, which gree of nuclear translocation (Figure S3A; data not shown). A was then turned on by an activity-dependent calcineurin-medi- sizable portion of the CRTC1 mutant (S151A/S245A) remained ated dephosphorylation to facilitate CRE-dependent gene in the nucleus when stimulated in the presence of CsA/FK, indi- expression. cating that dual-alanine mutation of Ser151 and Ser245 conferred We next turned to cultured cortical neurons to determine the resistance to calcineurin inhibitors (Figure 3B). A constitutively key phosphorylation sites underlying CRTC1’s nuclear shuttling. cytosolic mutant (mutCN) was generated by introducing muta- We found marked changes in the electrophoretic mobility of tions in two consensus calcineurin-binding motifs (PxIxIT) of CRTC1 associated with its phosphorylation or dephosphoryla- CRTC1 (Screaton et al., 2004)(Figure 3A). The activities of tion. The abundance of the lowest-mobility (upper) bands were both of these mutant proteins in neurons were quantified by increased on exposure of neurons to phenformin or to CsA/FK, CRE-reporter assays (Figure 3C). First, knockdown of CRTC1 whereas synaptic stimulation of neurons or in vitro treatment with two independent short-hairpin RNAs (shRNAs) (shCRTC1- of cell lysates with calf intestinal phosphatase (CIP) increased #1 and shCRTC1-#2) showed that CRTC1 is critical for ac- the intensity of the highest-mobility (lower) band (Figure 2A). tivity-regulated CRE-dependent transcription (Figure 3C). Alanine-scanning mutagenesis based on the mass spectrometry Expression of an shRNA-resistant form of wild-type CRTC1 data (Figure 1A) revealed that S151A and S245A were mutations (WT-CRTC1res) restored the transcription activity in a concen- that attenuated the phosphorylation-related mobility shift of tration-dependent manner (Figures S3C–S3E). However, FLAG epitope-tagged CRTC1 (Figures 2B, S2A, and S2B). expression of S151A/S245A (caCRTC1res) in this context Consistent with this, increased synaptic activity triggered conferred a high level of CRE-dependent transcription even in dephosphorylation of these Ser residues in neurons (Figures CRTC1-depleted neurons exposed to TTX, and this effect was S2C and S2D). Interestingly, the combined double mutant increased further in the presence of BIC/4AP (Figure 3C). The (S151A/S245A) lost any activity-dependent mobility shift (‘‘ca’’ latter effect of BIC/4AP was blocked by CsA/FK, implying an in Figure 2C). These results suggest that Ser151 and Ser245 might additional calcineurin-sensitive mechanism to further facilitate regulate the surface charges and conformation of CRTC1. This CRE-dependent transcription. In contrast, CRTC1 (mutCNres) was consistent with a prior report of Ser151 phosphorylation by did not evoke transcriptional activity in CRTC1-depleted neu- SIK2/AMPK (Sasaki et al., 2011), but further highlighted the crit- rons exposed to either TTX or BIC/4AP (Figure 3C). Together, ical importance of Ser245 phosphorylation, likely by MARK2, in in contrast to previous studies emphasizing the role of Ser151, cytoplasmic retention of CRTC1 in neurons. Taken together, our results revealed a critical role for both Ser151 and Ser245.

Figure 1. Kinetics and Signaling Mechanisms of Activity-Dependent Nuclear Shuttling of CRTC1 (A) Domain organization of CRTC1 showing the positions of 11 Ser and Thr residues (red arrows) found by mass spectrometric analysis to be phosphorylated in cortical neurons treated with TTX and CsA/FK. CBD, CREB-binding coiled-coil domain; TAD, transactivation domain; NLS, nuclear localization signal; NES, nuclear export signal. The amino acid sequences surrounding each of these 11 Ser or Thr residues are shown on the right. (B) Cultured hippocampal neurons at 7–11 days in vitro (div) treated with TTX or BIC/4AP (1 hr) were stained with an anti-CRTC1 antibody. (C) Hippocampal neurons were silenced with TTX (blue line) and stimulated with BIC/4AP (red line) for the indicated time. The intensity of CRTC1-IR (a.u., arbitrary units) in the nucleus and cytoplasm (top) and the nuclear/cytoplasmic ratio were quantiﬁed (bottom). Data are means ± SD. (D) Nuclear/cytoplasmic ratio of CRTC1-IR in hippocampal neurons treated with BIC/4AP (5 min) then transferred to TTX medium for the indicated time. Data are means ± SD. (E) Luciferase reporter assay for hippocampal neurons transfected with CRE-reporter plasmid and treated with either TTX or BIC/4AP in the absence or presence of phenformin (2 mM) for 4 hr. ***p < 0.001. (F) CRE-luciferase reporter assay as in (E) treated with TTX or BIC/4AP in the presence or absence of compound C (CompC, 5 mM) and CsA/FK as indicated for 4 hr. Note that compound C alone did not evoke CRE activity, but signiﬁcantly enhanced BIC/4AP-induced CRE activity. *p < 0.05. (G) MARK2 is responsible for CRTC1 cytosolic retention. Hippocampal neurons transfected with expression vectors for constitutively active (ca, T208E) or dominant-negative (dn, T208A/S212A) forms of MARK2 and treated with BIC/4AP (2 hr) or TTX, respectively, were stained with anti-CRTC1 antibody. (H) Calcineurin is responsible for nuclear translocation of CRTC1. Hippocampal neurons were treated with TTX or BIC/4AP (1 hr) in the absence or presence of various concentrations of CsA and FK506. The nuclear/cytoplasmic ratio of CRTC1-IR was determined as means ± SD. Representative images for cells stimulated with BIC/4AP in the absence or presence of 5 mM CsA and 6.4 nM FK506 are shown in the upper panels. Reported IC50 for CsA and FK506 for CaN inhibition are 27 nM and 0.23 nM, respectively. Scale bars, 10 mm. See also Figure S1.

Neuron 84, 92–106, October 1, 2014 ª2014 Elsevier Inc. 95 Neuron Region-Speciﬁc Activation of CRTC1-CREB in Memory

Expression of caCRTC1 did not increase the phosphorylation of CREB at Ser133 in neurons exposed to either TTX or BIC/4AP (data not shown), suggesting that the nuclear accumulation of CRTC1 in the absence of CREB phosphorylation is sufﬁcient to induce CRE-dependent transcription. To examine this notion further in an acute system, we constructed caCRTC1-ERT2 by fusing caCRTC1 to a mutant form of the estrogen receptor (ER) that is receptive exclusively to a synthetic ligand of ER (Fig- ure 3D). caCRTC1-ERT2 successfully translocated from cytoplasm to nucleus by 4-hydroxytamoxifen (4-OHT) (Figure 3E). Exposure of the transfected cells to 4-OHT in TTX-containing medium for 4 hr resulted in a marked increase in CRE-dependent transcription in neurons (Figure 3F). This activity was further increased when BIC/4AP was applied with 4-OHT (Figure S3F), suggesting that there is a combinatorial regulation of nuclear CRTC1 and other CRTC1-independent CREB regulatory com- ponents downstream of synaptic activity (see also Figure 3C). Notably, the 4-OHT-induced acute caCRTC1-ERT2 nuclear accumulation was not accompanied by detectable increase in the level of phosphorylated CREB (Figure 3G). We further found that c-Fos is induced in this condition in the absence of synaptic activity (Figure S3G). Together, these results supported the notion that the nuclear translocation of CRTC1 was sufﬁcient to induce CRE-dependent transcription in neurons, and also that nuclear CRTC1 might synergize with other synaptic activity-driven CREB signaling cascades to facilitate CRE-dependent transcription.

CRTC1 Complexes to the IEG Promoter Regions in an Activity-Dependent Manner and Mediates Activity- Dependent IEG Transcription in Neurons To determine the role of nuclear CRTC1 in neuronal CREB- dependent gene expression, we next tested whether CRTC1 directly bound IEG promoters. We prepared cortical neurons in which Myc-tagged CRTC1 was expressed to a minimal level, Figure 2. Identification of Two Critical Ser Residues of CRTC1 that such that no additional increase in transcription activity was Are Dephosphorylated in an Activity-Dependent Manner detectable in sensitive reporter assays (Figure S4A). Quantitative (A) Immunoblot analysis of endogenous CRTC1 in rat-cultured cortical neu- ChIP assays using an anti-Myc antibody revealed that CRTC1 rons incubated in the absence or presence of 2 mM phenformin (Phen) or 5 mM was complexed in an activity-dependent manner with the CsA and 6.4 nM FK506 (CsA/FK) for the indicated time (left). Lysates of neu- promoters of Arc, c-fos, and BDNF, and rather constitutively rons treated with TTX or BIC/4AP (30 min) were also treated (or not) with CIP with the zif-268 promoter (Figure 4A). ChIP assays using anti- before immunoblot analysis (right). (B) Immunoblot analysis of the FLAG-tagged wild-type (WT) and Ser-to-Ala CRTC1 antibodies further confirmed the binding of endogenous + mutants of CRTC1. Cortical neurons transfected with FLAG-CRTC1 WT and CRTC1 to CRE sites of IEG promoters in high K -treated cortical mutants were either treated with TTX (1.5 hr) or BIC/4AP (30 min) at 6 div. Of the neurons (Figure S4B). 11 Ser/Thr residues (Figure 1A), seven serine residues were further analyzed To examine if CRTC1 associates with the promoters via CREB, based on the positional information and the phosphorylation consensus- and we coexpressed A-CREB, a specific inhibitor of CREB function 14-3-3 binding consensus-motif databases (PhosphoMotif Finder, HPRD). that binds to bZIP domain of CREB with high affinity (Ahn Red arrowheads indicate two Ser-to-Ala mutants that show distinct mobility shift both in TTX and BIC/4AP conditions compared to WT. et al., 1998). Expression of A-CREB completely blocked (C) Immunoblot analysis of WT or mutant (ca, S151A/S245A) FLAG-CRTC1 in CRTC1-dependent transcription activity (Figure 4B; see also Fig- cortical neurons treated with TTX or BIC/4AP (30 min). The mobility shift ure S4G). Together, we conclude that a functional CRTC1-CREB caused by activity-induced dephosphorylation was almost completely abol- complex was recruited to the CRE sites in IEG promoters in an 151 245 ished when Ser and Ser were both mutated to Ala. CIP treatment of the activity-regulated manner. TTX sample resulted in further downward mobility shift, indicating the pres- To examine the target specificity of CRTC1 on CRE sites ence of other basal phosphorylation site(s). See also Figure S2. among many regulatory elements in the promoter, we took advantage of the Arc promoter, which contains a single half- Given the ability of S151A/S245A mutant of CRTC1 to confer CRE site in the distal enhancer element termed SARE (Kawa- CRE-dependent promoter activity in neurons even in the pres- shima et al., 2009) that is positioned about 7 kb upstream of the ence of TTX, we hereafter refer to this mutant as caCRTC1. transcription start site of Arc gene (Figure 4C). Knockdown of

96 Neuron 84, 92–106, October 1, 2014 ª2014 Elsevier Inc. Neuron Region-Speciﬁc Activation of CRTC1-CREB in Memory

Figure 3. Nuclear Localization of CRTC1 Is Necessary and Sufficient for Induction of CRE-Dependent Transcriptional Activity in Neurons (A) Schematic representation of WT and mutant CRTC1 forms that localize almost exclusively to the nucleus (caCRTC1, S151A/S245A) or in the cytosol (mutCN) are shown on the upper left. The position of Ser151 and Ser245 are indicated by red arrowheads, and those of the calcineurin-binding motifs mutated in mutCN are indicated by blue arrows. The subcellular localizations of the HA-tagged CRTC1 mutants are shown in the right. (B) The percentage of cells showing a predominantly nuclear localization (Nuc/cyto > 1) of the exogenous proteins after incubation under various conditions for 1 hr was also determined. Note that caCRTC1 is localized in the nucleus in TTX, whereas mutCN is cytoplasmic under any conditions. FSK (5 mM)/IBMX (100 mM) is a protein kinase A activator cocktail. (C) CRE-luciferase reporter assay for hippocampal neurons cotransfected with a CRE-dependent reporter plasmid, shRNA expression vectors for control (shNega) or CRTC1 (shCRTC1-#1), and expression vectors for RNAi-resistant forms (res) of WT or mutant CRTC1. The cells were exposed to TTX or BIC/4AP with or without CsA/FK for 4 hr. Data include five independent experiments. The height of the filled and open bars at the bottom of the graph represents the levels of endogenous CRTC1 and exogenous CRTC1res expression levels, respectively, as determined by ICC in Figure S3C. *p < 0.05, **p < 0.01, ***p < 0.001. (D) Schematic structure of caCRTC1-ERT2. Red arrowheads indicate two Ser residues mutated in caCRTC1. (E) Hippocampal neurons transfected with Myc-caCRTC1-ERT2 were exposed to ethanol (Vehicle) or 4-OHT for 1 hr, and then subjected to ICC with anti-Myc antibody. The percentage of cells showing a predominantly nuclear localization of caCRTC1-ERT2 is shown. (F) CRE-luciferase reporter assay for hippocampal neurons cotransfected with the reporter plasmid and LacZ (Mock) or caCRTC1-ERT2, and treated with either ethanol (Vehicle) or 4-OHT for 4 hr in TTX-containing medium. ***p < 0.001. (G) Acute nuclear translocation of caCRTC1- ERT2 did not induce CREB phosphorylation. Hippocampal neurons expressing Myc-caCRTC1-ERT2 were exposed to BIC/4AP for 5 min, treated with TTX alone, or 4-OHT in TTX-containing medium for the indicated time. Top panels show ICC with antibodies to Myc and to pCREB (Ser133). Middle graph shows the percentage of the neurons with nuclear localization of caCRTC1-ERT2. The pCREB-IRs in nontransfected and transfected neurons were quantified (bottom). Slight pCREB-IR increase at 5 min after 4-OHT onset might reflect a response to the medium change. There are no significant differences between transfected and nontransfected neurons at any time points after 4-OHT treatment. n.s., not significant. Scale bars, 10 mm. See also Figure S3.

Neuron 84, 92–106, October 1, 2014 ª2014 Elsevier Inc. 97 Neuron Region-Speciﬁc Activation of CRTC1-CREB in Memory

Figure 4. CRTC1 Binds to IEG Promoters and Induces CRE-Dependent Transcription via CREB (A) ChIP assay for CRTC1 from Myc-CRTC1-transfected cortical neurons treated with TTX or BIC/4AP. ChIP eluates were subjected to qPCR with primers for the SARE of the Arc promoter, c-fos promoter (proximal), BDNF promoter (pIV), zif268 promoter, or for the unrelated region 7 kb upstream of SARE (Nega) (left). The ratio of the recovery obtained with anti-Myc to that obtained with control IgG is shown as fold enrichment (right). Control ChIP assays performed with un- transfected neurons did not yield substantial recovery with anti-Myc antibody (data not shown). Data are from six independent experiments. *p < 0.05. (B) CRE-reporter assay of neurons transfected with WT-CRTC1 and A-CREB. Note that the transcription activity in this assay is dependent on CRTC1 as shown in Figure 3C, and this component was totally blocked by the expression of A-CREB. (C) Schematic representations of the mouse Arc promoter and corresponding wild-type (7 kb) and mutant (mCRE) promoter luciferase reporter constructs. The CRE is located at À7 kb in the SARE region (yellow box). (D) Luciferase reporter assay of the Arc promoter (7 kb). Rat cortical neurons transfected with the Arc promoter-reporter construct, an shRNA expression vector for control (shNega) or CRTC1 (shCRTC1-#2), and an expression vector for CRTC1res were exposed to TTX or to BIC/4AP for 4 hr. **p < 0.01, ***p < 0.001. (E) Luciferase reporter assay performed as in (D) transfected with an expression vector for caCRTC1 or control GFP. **p < 0.01, ***p < 0.001. (F) Luciferase reporter assay of cortical neurons transfected with SARE-minCMV promoter (top) or the corresponding plasmid with mutations in the CRE, MRE, or SRE (mCRE, mMRE, and mSRE, respectively) and an expression vector for caCRTC1 or control GFP (bottom). ***p < 0.001. See also Figure S4.

CRTC1 significantly attenuated the synaptic activity-dependent CRTC1 (data not shown), similar to the results obtained with the component of the full-length Arc promoter activity, and this effect full-length Arc promoter (Figure 4D). Expression of caCRTC1 was reversed at least in part by expression of WT-CRTC1res (Fig- induced a marked increase in the SARE-minCMV promoter activ- ure 4D). Furthermore, expression of caCRTC1 increased the tran- ity, and this effect was abolished by mutation of the CRE motif scriptional activity of the promoter in the presence of TTX to a (mCRE), but was not influenced by mutation of the MRE, MEF2 level similar to that observed in control neurons stimulated with response element (mMRE), or SRE, the serum response element BIC/4AP (Figure 4E). The combination of caCRTC1 expression (mSRE) (Figure 4F). These results suggested that CRTC1 acts and BIC/4AP treatment induced an additive effect, whereas mu- solely and directly at the half-CRE of SARE without interaction tation of the half-CRE site in the Arc promoter abolished the ef- with MEF2, SRF, or other nonselective factors. fects of caCRTC1 both in TTX and in BIC/4AP (Figure 4E). Next, To determine whether these findings with the Arc promoter are to test if there is any local interaction, we used the SARE-minCMV applicable to other CREB-dependent IEGs, we performed re- promoter (Figure 4F). This promoter was largely dependent on porter assays with the proximal promoter of the c-fos gene and

98 Neuron 84, 92–106, October 1, 2014 ª2014 Elsevier Inc. Neuron Region-Speciﬁc Activation of CRTC1-CREB in Memory

Table 1. Quantification of CRTC1 Knockdown Efficiency in Different Experimental Systems CRTC1 Level (% to shNega) Experiment Target CRTC1 shCRTC1-#1 shCRTC1-#2 Reference Figure In vitro (WB from HEK cell) exogenous 44.8% ± 3.51% 24.4% ± 6.47% Figure S3B In vitro (ICC from neuronal culture) endogenous 27.3% ± 7.35% 25.9% ± 4.24% Figures S3C and S3D In vivo (IHC at amygdala) endogenous 19.8% ± 1.71% 38.3% ± 3.44% Figures 7A, S6A, and S6B In vivo (IHC at hippocampus) endogenous 23.2% ± 2.50% 38.4% ± 2.81% Figures 7E and S6F CRTC1 knockdown efficiency was tested in four different experimental systems shown in the figures (reference figures) and quantified. CRTC1 levels for shCRTC1-#1- and -#2-expressing systems are shown as percentages of those for shNega. In western blotting (WB), the background was defined in the area of 110–130 kDa (the size of mEGFP-CRTC1) within the lane of the mEGFP-CRTC1 nonexpressing sample. In ICC and IHC, the background was defined in the area where there are no cells or tissues, respectively, and subtracted. Please note the shCRTC1-#1 is more effective in suppressing CRTC1 than shCRTC1-#2 in vivo (both in the amygdala and hippocampus) and of equal potency as shCRTC1-#2 in neuronal cultures, while shCRTC1-#2 appears to be more effective than -#1 in HEK293 cells.

promoter IV of the BDNF gene (pIV-BDNF), both of which are c-Fos in the presence of TTX to an extent similar to that induced dependent on CREB for activity (Sassone-Corsi et al., 1988; Tab- by synaptic activity (Figures S5A and S5B). To ascertain these uchi et al., 2000; West et al., 2001). While activity-independent findings in vivo, we constructed adeno-associated viruses phosphoglycerate kinase (PGK) promoter activity was not (AAVs) encoding GFP plus caCRTC1 (AAV-caCRTC1) or GFP affected by CRTC1 (Figure S4C), BIC/4AP-induced activities alone (AAV-GFP) and injected them into the CA regions of the for c-fos promoter or pIV-BDNF were attenuated by CRTC1 mouse hippocampi (Figure 5A). One week after the injection, knockdown, and this effect was reversed in a concentration- caCRTC1 expression was observed in CA1 pyramidal neurons dependent manner by expression of WT-CRTC1res (Figures (Figure 5B). Consistent with our results in cultured neurons, the S4D and S4E; see also Table 1 and Figure S3D). These c-fos expression of Arc, c-Fos, and BDNF increased in the infected and pIV-BDNF promoter activities were markedly increased by CA portions of the hippocampus that were injected with AAV- expression of caCRTC1 even in the TTX condition, but the caCRTC1, while no change was detected when injected with caCRTC1 that lacks the N-terminal CREB-binding domain (ca- AAV-GFP (Figures 5C–5E). We found that IEG expression in vivo DCBD) had no effect (Figure S4F; data not shown). Furthermore, was also increased in some noninfected neurons, though much all of the caCRTC1 effects were completely blocked by coex- less consistently and to a lower extent (Figures 5C–5E, right pression of A-CREB (Figure S4G). panels; Figures S5C–S5E). This could reflect either increased Coimmunoprecipitation experiments in transfected cells re- excitability of caCRTC1-expressing neurons that spread through vealed that CRTC1 is capable of binding the S133A mutant of the hippocampal network, or possibly an effect of released CREB (Figure S4H). In addition, caCRTC1 was coimmunopreci- BDNF toward neighboring neurons, as suggested by increased pitated with an antibody to CBP from lysates of cortical neurons BDNF-IR around some, but not all, infected neurons (e.g., Fig- silenced with TTX when CREB phosphorylation is minimized ures 5E and S5E). Collectively, we conclude that, under our (Figure S4I). These results suggested that caCRTC1 might help experimental conditions, forced nuclear localization of CRTC1 recruit CBP to CRE sites when it is bound to the Ser133 unphos- in hippocampal CA neurons induces significant upregulation of phorylated form of CREB in neurons maintained in the TTX con- neuronal IEG products such as Arc, c-fos, and BDNF throughout dition. In keeping with these findings, our mass spectrometric the infected brain areas in vivo. analysis identified RNA helicase A as an interaction partner of Next, we investigated the consequences of caCRTC1 expres- CRTC1 (data not shown), which was previously shown to sion in vivo. Previous observations in cultured hippocampal mediate the association of CBP with the RNA polymerase II com- slices implicated CRTC1 activity in the regulation of dendritic plex (Nakajima et al., 1997). Together, these results suggested morphology (Li et al., 2009). We used a two-vector lentivirus that following nuclear accumulation of caCRTC1 and its binding system to confer sparse expression of caCRTC1 in hippocam- to CREB, CBP and likely other factors such as TAFII130 or TAF4 pal CA1 neurons (Hioki et al., 2009) for morphological analyses. might be recruited (Conkright et al., 2003; Jeong et al., 2012; We found that caCRTC1 expression induced Arc and c-Fos Ravnskjaer et al., 2007), thereby perhaps even bypassing the expression as early as 1 week after infection. We then acquired need for activity-dependent CREB phosphorylation for CREB/ stacks of images of dendritic spines present on thin secondary CBP-mediated gene expression. dendrites of CA1 pyramidal cells in the stratum radiatum and performed 3D reconstruction to compare various spine param- Dynamics of Nuclear Localization of CRTC1 Play a eters. Expression of caCRTC1 had no detectable effect on Critical Role in Controlling Long-Term Memory spine density (Figures S5F and S5G). In contrast, caCRTC1 Formation expression induced a significant enlargement of spine heads We next examined whether nuclear CRTC1 upregulates the (Figures S5F and S5G). The nuclear accumulation of CRTC1 in expression of endogenous IEGs in neurons. Immunocytochem- hippocampus thus appeared to induce a generalized spread istry (ICC) of cultured neurons transfected with caCRTC1 re- of morphological plasticity of dendritic spines without altering vealed that caCRTC1 induced the expression of both Arc and spine density. The ability of caCRTC1 to upregulate the

Neuron 84, 92–106, October 1, 2014 ª2014 Elsevier Inc. 99 Neuron Region-Speciﬁc Activation of CRTC1-CREB in Memory

Figure 5. Expression of caCRTC1 in Hippocampal CA Regions In Vivo Induces IEG Expressions and Enhances Long-Term Memory Formation (A) Hippocampal sections from a mouse injected with AAV-GFP were counterstained with Hoechst 33342 and monitored for GFP fluorescence. Scale bar, 1 mm. (B) IHC of the mouse hippocampus infected with AAV-caCRTC1 as in (A) using an anti-CRTC1 antibody with Hoechst 33342 counterstaining. caCRTC1 and GFP are coexpressed in CA1 pyramidal neurons of the mice injected with AAV-caCRTC1. Scale bar, 100 mm. (C and D) IHC of Arc (C) and c-Fos (D) in the AAV-GFP- or AAV-caCRTC1-infected CA1 region of the hippocampus at 1 week post infection. White dotted lines indicate pyramidal cell layers (left). Quantifications of the IR intensity in the infected (GFP positive) and noninfected (GFP negative) pyramidal cell soma are represented as means ± SEM (right graphs). *p < 0.05, ***p < 0.001. (E) IHC of BDNF in the AAV-GFP- or AAV-caCRTC1-infected CA3 region of the hippocampus at 4 weeks post infection. Perisomatic BDNF-IR intensity was quantified. ***p < 0.001. Scale bars in (C)–(E), 100 mm. (F) Contextual fear conditioning test (0.2 mA) of mice injected with AAV-GFP or -caCRTC1 (n = 16 each) bilaterally in the hippocampus at 1 week post injection.

Two-way ANOVA, task 3 virus interaction, F(1,30) = 4.64, p < 0.05; post hoc t test, *p < 0.05. (G) 2D plot of fear memory performance and caCRTC1 expression level (CRTC1-IR in CA region) for mice (n = 16) injected with AAV-caCRTC1. Each data point represents one animal, and the bar ends represent the values for caCRTC1 expression level in each hemisphere. Gray shading indicates the area for mean ± SEM of the performance and CRTC1-IR in the AAV-GFP group. Most data points were aligned in the upper-right and lower-left quadrants (dotted lines are medians of CRTC1-IR and freezing score), suggesting that the caCRTC1 expression level correlates with the memory (chi-square test, df = 1, p < 0.05). Behavioral and histological analyses were carried out by an experimenter blinded to the virus type. See also Figure S5.

100 Neuron 84, 92–106, October 1, 2014 ª2014 Elsevier Inc. Neuron Region-Speciﬁc Activation of CRTC1-CREB in Memory

Figure 6. Brain-Region Specific Sensitivity of Nuclear Translocation of CRTC1 in Response to US-CS-Pairing Stimuli (A) IHC using anti-CRTC1 antibody of BLA tissues from mice either in the home cage (Naive), immediate shock (US+/CSÀ) group, conditioned stimulus (CS) alone (USÀ/CS+) group, or CS paired with foot shock (US+/CS+) group (top). Quantifi- cation of cells with nuclear CRTC1 signals out of NeuN-positive neurons (bottom). (B) IHC using anti-CRTC1 antibody of hippocampal CA1 tissue from mice with the same sets of conditions as in (A) (top). Quantification of cells with nuclear CRTC1-positive signals (bottom). For both (A) and (B), images were analyzed by an experimenter who was blinded to the CS/US condition. (C) IHC using an anti-CRTC1 antibody with Hoechst 33342 counterstaining of hippocampus CA1 tissues from mice either in the home cage (Naive) or 90 min after receiving electroconvulsive shocks (ECS) (50 mA, 100 Hz, 2 ms duration 3 3 times with 10 s intervals). Scale bars, 50 mm.

the hippocampus (Figure 6B). A much stronger stimulus, such as an electroconvulsive shock (ECS), triggered nuclear accumulation of CRTC1 in a subset of hippocampal neurons (Figure 6C), suggesting that the gating mechanism to prevent nuclear translocation might be expression of a subset of IEGs in vivo prompted us to test the more preponderant in the hippocampus under physiological impact of caCRTC1 expression on a hippocampus-dependent conditions. behavior, the CFC test. To maximize the infection efficiency, To ascertain the relevance of this region-specific transloca- we switched to AAVs and bilaterally injected AAV-GFP or tion, we set out to examine the requirements of CRTC1 in these -caCRTC1 into mouse hippocampal CA regions. The AAV- areas. We bilaterally injected AAV vectors for two CRTC1 knock- derived expression increased gradually during the initial 4 weeks down constructs and a control AAV-shNega into the BLA (Ta- after injection and reached a plateau thereafter. To minimize any ble 1). After 4 weeks, when CRTC1-IR was greatly weakened cumulative aberration in hippocampal circuit dynamics induced by knockdown (Figures 7A, S6A, and S6B; Table 1), both cued by upregulation of IEGs, we performed the test in 1 week post and contextual fear conditioning test performances were signif- injection. The long-term contextual fear memory was signifi- icantly worsened in the AAV-shCRTC1-#1-injected mice (Fig- cantly enhanced in mice injected with AAV-caCRTC1 (Fig- ure 7B), in keeping with its potency to suppress the task-related ure 5F). After the test, the extent of caCRTC1 expression was induction of both c-Fos and Arc (Figures 7C, 7D, S6C, and S6D). quantified by immunohistochemistry (IHC) (Figure 5B). A chi- The fear memory defects did not reach statistical significance in square test suggested that animals with a higher or lower the AAV-shCRTC1-#2 group (Figure 7B), as a subset of CRTC1- expression of CRTC1 tended to show a higher or lower freezing regulated genes such as Arc were not fully inhibited owing to the score, respectively, in reference to the medians of the respec- lower in vivo knockdown potency of shCRTC1-#2 (Figure S6D; tive parameters (df = 1, p < 0.05) (Figure 5G). Table 1). Forced expression of caCRTC1 in BLA did not enhance long-term fear memory formation (Figure S6E), suggesting that CRTC1 Plays a Critical Region-Specific Role in the BLA CRTC1 signaling is critical, but easily saturates in the amygdala during Fear Memory Formation in accordance with enriched expression of CRTC1 in the amyg- To test whether endogenous CRTC1 is involved in learning and dala (Watts et al., 2011). memory, we first monitored CRTC1’s localization in the BLA neu- In the hippocampus, shCRTC1 virus produced a significant rons during CFC. While CRTC1 remained largely cytoplasmic in reduction in CRTC1 as well (Figures 7E and S6F; Table 1). How- animals in the home cage, pairing the context (conditioned stim- ever, CRTC1 knockdown had little effect on contextual fear ulus [CS]) with the electric foot shock (unconditioned stimulus memory under an experimental condition in which hippocampal [US]) triggered a nuclear accumulation of CRTC1 in some popu- caCRTC1 expression promoted long-term memory (Figure 7F). lations of BLA neurons (Figure 6A). In contrast, however, the CS- This lack of sensitivity to hippocampal CRTC1 knockdown was US pairing did not induce nuclear translocation of CRTC1 in not altered by a change in foot shock strength (Figure S6G).

Neuron 84, 92–106, October 1, 2014 ª2014 Elsevier Inc. 101 Neuron Region-Speciﬁc Activation of CRTC1-CREB in Memory

Figure 7. Differential Contribution of CRTC1 in Long-Term Fear Memory Formation between Amygdala and Hippocampus (A) IHC using an anti-CRTC1 antibody of the mouse injected in BLA with AAV-shNega, -shCRTC1-#1, and -shCRTC1-#2 at 4 weeks post injection. GFP is expressed from one of the cassettes of AAV-shRNA vectors. We observed no significant gross neuronal death in AAV-injected areas, as examined with Hoechst 33342 counterstaining (data not shown). Scale bar, 500 mm. White lines in the CRTC1-IR images are the lines used for quantification in Figure S6A. (B) Cued and contextual fear conditioning test of mice injected with AAV-shNega, -shCRTC1-#1, or -shCRTC1-#2 bilaterally in the BLA at 4 weeks post injection. One-way ANOVA with post hoc Tukey-Kramer test; cued test (tone), ANOVA p = 0.010, **p < 0.01 for shNega versus sh#1; contextual test, ANOVA p = 0.037, **p < 0.05 for shNega versus sh#1 (shNega n = 14, sh#1 n = 15, sh#2 n = 20); note that shCRTC1-#2 is less efficient in knockdown in vivo (Table 1). (C) Quantification of the percentage of c-Fos-positive cells (in the central area of the BLA) after fear conditioning test. Representative IHC images are shown in Figure S6C. (D) Same as (C) but for Arc (IHC images shown in Figure S6D). (E) IHC using an anti-CRTC1 antibody of the mouse hippocampus infected with AAV-shNega, -shCRTC1-#1, and -shCRTC1-#2 at 4 weeks post injection. Scale bar, 200 mm. (F) Contextual fear conditioning test (0.2 mA) of mice injected with AAV-shNega, -shCRTC1-#1, or -shCRTC1-#2 (shNega n = 11, sh#1 n = 7, sh#2 n = 9) bilaterally in the hippocampus at 4 weeks post injection. n.s., not significant by one-way ANOVA. (G) IHC using an anti-c-Fos antibody of the dentate gyrus area of the hippocampus infected with AAV-shNega, -shCRTC1-#1, and -shCRTC1-#2 at 4 weeks post injection with or without enriched environment. Scale bar, 100 mm. Behavioral and histological analyses were carried out by an experimenter blinded to the virus type. See also Figure S6.

102 Neuron 84, 92–106, October 1, 2014 ª2014 Elsevier Inc. Neuron Region-Speciﬁc Activation of CRTC1-CREB in Memory

Furthermore, CRTC1 knockdown had no effect on enriched envi- was controlled through a complex web of both activity-regulated ronment-induced expression of c-Fos and Arc (Figure 7G; data and unregulated pathways. Opposing CRTC1 regulation by ki- not shown). These results are in keeping with our original obser- nases and calcineurin might permit addition of an independent vation that CS-US pairing did not trigger nuclear translocation of tunable switch to CREB-dependent gene expression, perhaps hippocampal CRTC1 (Figure 6B). Taken together, these results to better regulate memory strength specifically during consoli- provide evidence that CRTC1 functions in a region-specific dation (Sekeres et al., 2012). Notwithstanding this interest, manner, with a much higher contribution in the amygdala, to evaluating the impact of endogenous CRTC1 on memory perfor- fine-tune the CREB output during long-term memory formation. mance was previously difficult due to a severe mood dysregulation in CRTC1-null mice (Breuillaud et al., 2012). DISCUSSION Recent evidence suggests that even subtle modulation of CREB signaling significantly affects an amygdala-dependent In this study, we combined biochemical, cell biological, pharma- memory (Cowansage et al., 2013). While CREB phosphorylation cological, virus-mediated genetic engineering, and behavioral is mainly determined via concerted activity of several activity- approaches to decipher key neuronal regulation mechanisms dependent kinases (Bito et al., 1996; Deisseroth and Tsien, of CRTC1 as a CREB coactivator, and revealed an intriguing re- 2002), CRTC1 may be more susceptible to metabolic states of gion-specific requirement for CRTC1 during long-term fear neurons and may also decode neuronal activity that is distinct memory formation. from those governing CREB phosphorylation per se such as The extremely Ser/Thr-rich amino acid composition of CRTC1 long-lasting low-frequency synaptic stimulation that strongly has so far hindered comprehensive investigation of its regulatory stimulates calcineurin (Fujii et al., 2013). Consistent with this, phosphorylation mechanisms. Here we took a highly sensitive we found that nuclear translocation of CRTC1 was triggered by mass spectrometric approach to determine the critical residues fear conditioning only in the amygdala. Region-specific interro- responsible for its activity-regulated nuclear translocation. gation using viral vectors and CRTC1-specific reagents further CRTC1 is multiphosphorylated in the resting state in sharp unveiled that memory consolidation mechanisms downstream contrast to CREB, which is dephosphorylated at a resting state, of CRTC1 might be differentially regulated and contribute constitutively nuclear-localized, and depends on a single critical more to fear memory formation in the amygdala than in the phosphorylation of Ser133. Among the many phosphorylation hippocampus. sites of CRTC1 we identified, two (Ser151 and Ser245) appeared The lack of hippocampal CRTC1 contribution in mice under to critically participate in the activity-dependent dephosphoryla- our experimental conditions (Figure 7) does not necessarily tion process and contribute to the nuclear translocation of rule out a role for hippocampal CRTC1 under circumstances CRTC1. The kinetics of the nuclear shuttling regulated by where CREB-dependent gene expression may become in Ser151 and Ser245 were in keeping with the transport rate of much higher demand. Consistent with this idea, augmented hip- CRTC1 from the dendrites to the nucleus (Ch’ng et al., 2012). pocampal CRTC1 expression was reported to be beneficial in Previous studies in overexpressed cells have suggested that rescuing behavioral and gene expression defects in 6-month- both phosphorylation of CRTC1’s C terminus and dephosphory- old mice that mimicked pathologies underlying Alzheimer’s dis- lation of more N-terminal and central residues might regulate ease, but the same manipulation had little effect in WT mice its nuclear transport. Here we unequivocally determined two (Parra-Damas et al., 2014). classes of kinases, one constitutive (MARK2) and the other Taken together, our findings shed new light on the combinato- metabolically regulated (SIK2/AMPK), for cytoplasmic retention, rial and brain region-specific modulation of molecular cascades and demonstrated that activity-regulated calcineurin unlocked implicated in CRTC1-CREB-dependent gene expression crucial the gating. for formation and allocation of long-term fear memory. Furthermore, a direct transcriptional impact of nuclear CRTC1 transport on CRE/CREB-dependent gene expression was EXPERIMENTAL PROCEDURES shown both in vitro and in vivo. Although a recent report suggested that in nonneuronal cells CRTC1 might augment CREB All animal experiments were performed in accordance with the regulations and Ser133 phosphorylation via PRMT5 (Tsai et al., 2013), we were guidelines of the University of Tokyo, the Tokyo University of Agriculture, and the Japan Neuroscience Society and approved by the institutional review unable to detect such a change in CREB states in neurons committee of the University of Tokyo, Graduate School of Medicine and of (e.g., Figure 3G). The precise mechanism by which CRTC1 the Tokyo University of Agriculture. may facilitate CREB’s transcriptional activity while bypassing CREB phosphorylation, and how CRTC1 cooperates with other Primary Neuron Culture and Transfection activity-regulated mechanisms, should be the focus of future Primary cultures of hippocampal CA neurons from postnatal days 0–1 ICR study. One possible scenario we currently favor is that formation mice (SLC) were prepared as previously reported (Nonaka et al., 2006) and of CRTC1-CREB complex facilitates recruitment of CBP and transfected using Lipofectamine 2000. Cortical neuron cultures from E17–18 rats were prepared, transfected using AMAXA Nucleofector electroporation transcription initiation complex proteins such as TAFII130, as system (Lonza), and seeded onto poly-D-lysine-coated dishes (Ageta-Ishihara has been suggested in nonneuronal cells (Conkright et al., et al., 2009). 2003; Heinrich et al., 2013). Finally, CREB has been widely implicated in long-term mem- Reporter Assays ory formation via regulation of neuronal excitability in various Hippocampal neurons were cotransfected with a promoter-of-interest-driven areas of the brain. Here we uncovered that CRTC1 modification luc2P plasmid, the PGK promoter-driven Renilla plasmid, and either the

Neuron 84, 92–106, October 1, 2014 ª2014 Elsevier Inc. 103 Neuron Region-Speciﬁc Activation of CRTC1-CREB in Memory

CRTC1 knockdown plasmid and/or mutant or WT CRTC1 expression plasmid. sity) for BDNF promoter IV; Y. Ishihama, N. Sugiyama (Kyoto University), and Neurons were treated with TTX for 5 hr and stimulated with BIC/4AP for 4 hr, H. Nakatsumi (Kyushu University) for advice on LC-MS/MS; R. Morris (Univer- then lysed. The luciferase activity of firefly normalized with Renilla was calcu- sity of Edinburgh), L. Monteggia (UTSW), V. Ramirez-Amaya (Autonomous lated (Kawashima et al., 2009). University of Queretaro), and P. Cohen (University of Dundee) for discussion on an early version of the work; all members of the Bito and Kida laboratories Mass Spectrometry for support and discussion; and A. Adachi-Morishima, Y. Kondo, K. Saiki, R. Rat cortical neurons were transfected with FLAG-CRTC1. At 6–8 DIV, the neu- Gyobu, Y. Dobashi, and T. Kinbara for assistance. This work was supported rons were either treated with TTX or BIC/4AP, and FLAG-CRTC1 was immuno- in part by Grants-in-Aid from MEXT and JSPS (to M.N., K. Sasaki, M.O., precipitated from the lysate, digested, and desalted. Mass spectrometry was S.T.-K., H.O., S. Kida, and H.B.), by CREST-JST (to S. Kida and H.B.), by a conducted on an LTQ Orbitrap XL instrument (Thermo Fisher Scientific, San JST-CONACyT SICP grant (to H.O. and H.B.), by awards from the Yamada Sci- Jose). A nanoFrontier HPLC system (Hitachi) was connected to the mass spec- ence Foundation, the Mitsubishi Foundation, the Takeda Science Foundation, trometer for LC-MS/MS. and the Uehara Memorial Foundation (to H.B.), and from The Tokyo Society of Medical Sciences (to H.O.). K. Suzuki, Y.I., S. Kamijo, and T.K. were predoc- ChIP Assay toral fellows of the JSPS. Neuronal nuclear lysate preparation and ChIP were performed using a ChIP-IT Express Kit (Active Motif) as reported previously (Kawashima et al., 2009). Accepted: August 14, 2014 Published: October 1, 2014 In Vivo Virus Injection Lentivirus and AAVs were produced and purified as described in Kawashima REFERENCES et al. (2009) and Kawashima et al. (2013), respectively. The viral injection pro- cedure was essentially as described in Han et al. (2007). Ageta-Ishihara, N., Takemoto-Kimura, S., Nonaka, M., Adachi-Morishima, A., Suzuki, K., Kamijo, S., Fujii, H., Mano, T., Blaeser, F., Chatila, T.A., et al. (2009). Behavioral Analysis Control of cortical axon elongation by a GABA-driven Ca2+/calmodulin- Fear conditioning tests were performed as previously described (Fukushima dependent protein kinase cascade. J. Neurosci. 29, 13720–13729. et al., 2008). Briefly, the mice were placed in a training chamber for 148 s, Ahn, S., Olive, M., Aggarwal, S., Krylov, D., Ginty, D.D., and Vinson, C. (1998). and a foot shock was applied. On day 2, the mice were placed in the same A dominant-negative inhibitor of CREB reveals that it is a general mediator of chamber for 300 s (contextual test) during which the freezing was measured. stimulus-dependent transcription of c-fos. Mol. Cell. Biol. 18, 967–977. Cued and contextual fear conditioning tests for experiments targeting amyg- Altarejos, J.Y., and Montminy, M. (2011). CREB and the CRTC co-activators: dala were performed as follows; the mice were placed in a conditioning cham- sensors for hormonal and metabolic signals. Nat. Rev. Mol. Cell Biol. 12, ber ‘‘A’’ for 120 s, and a tone was applied for 30 s that coterminated with a foot 141–151. shock. On day 2, the mice were placed in chamber ‘‘B,’’ and 180 s later the tone was applied for 180 s (cued test). On day 3, the mice were placed in Altarejos, J.Y., Goebel, N., Conkright, M.D., Inoue, H., Xie, J., Arias, C.M., the chamber ‘‘A’’ for 300 s (contextual test). All experiments were performed Sawchenko, P.E., and Montminy, M. (2008). The Creb1 coactivator Crtc1 is by experimenters blinded to the virus type. required for energy balance and fertility. Nat. Med. 14, 1112–1117. Bito, H., and Takemoto-Kimura, S. (2003). Ca(2+)/CREB/CBP-dependent Coimmunoprecipitation Assay and Western Blotting gene regulation: a shared mechanism critical in long-term synaptic plasticity Coimmunoprecipitation assay was performed as described previously (Okuno and neuronal survival. Cell Calcium 34, 425–430. et al., 2012). HEK293 cells or cortical neurons were lysed in RIPA buffer, and Bito, H., Deisseroth, K., and Tsien, R.W. (1996). CREB phosphorylation and the lysate was mixed with primary antibody, in which protein G Sepharose dephosphorylation: a Ca(2+)- and stimulus duration-dependent switch for hip- (GE Healthcare, Amersham) was added later. The input and eluates were run pocampal gene expression. Cell 87, 1203–1214. on an acrylamide gel, blotted to PVDF membrane, and detected with an Bito, H., Deisseroth, K., and Tsien, R.W. (1997). Ca2+-dependent regulation in antibody. neuronal gene expression. Curr. Opin. Neurobiol. 7, 419–429.

ICC and IHC Bourtchuladze, R., Frenguelli, B., Blendy, J., Cioffi, D., Schutz, G., and Silva, ICC and IHC were performed in general procedures. For ICC, the cells were A.J. (1994). Deficient long-term memory in mice with a targeted mutation of fixed in 4% paraformaldehyde/PBS for 15 min at RT. For IHC, mice were the cAMP-responsive element-binding protein. Cell 79, 59–68. deeply anaesthetized and perfused with cold 2% or 4% paraformaldehyde/ Breuillaud, L., Rossetti, C., Meylan, E.M., Me´ rinat, C., Halfon, O., Magistretti, PB. P.J., and Cardinaux, J.R. (2012). Deletion of CREB-regulated transcription co- All data are shown as the mean ± SEM unless otherwise indicated in the activator 1 induces pathological aggression, depression-related behaviors, figure legend. See Supplemental Information for more detailed procedures. and neuroplasticity genes dysregulation in mice. Biol. Psychiatry 72, 528–536. Ch’ng, T.H., Uzgil, B., Lin, P., Avliyakulov, N.K., O’Dell, T.J., and Martin, K.C. SUPPLEMENTAL INFORMATION (2012). Activity-dependent transport of the transcriptional coactivator CRTC1 from synapse to nucleus. Cell 150, 207–221. Supplemental Information includes six figures and Supplemental Experimental Conkright, M.D., Canettieri, G., Screaton, R., Guzman, E., Miraglia, L., Procedures and can be found with this article online at http://dx.doi.org/10. Hogenesch, J.B., and Montminy, M. (2003). TORCs: transducers of regulated 1016/j.neuron.2014.08.049. CREB activity. Mol. Cell 12, 413–423. Cowansage, K.K., Bush, D.E., Josselyn, S.A., Klann, E., and Ledoux, J.E. ACKNOWLEDGMENTS (2013). Basal variability in CREB phosphorylation predicts trait-like differences in amygdala-dependent memory. Proc. Natl. Acad. Sci. USA 110, 16645– We thank J.-P. Bellier and H. Kimura (Shiga University of Medical Science) as 16650. well as S. Josselyn and P. Frankland (Hospital for Sick Children, Toronto) for invaluable advice during the precursory phase of this study; C. Vinson (NIH) Deisseroth, K., and Tsien, R.W. (2002). Dynamic multiphosphorylation pass- for an A-CREB vector; H. Hioki and T. Kaneko (Kyoto University) for the lenti- words for activity-dependent gene expression. Neuron 34, 179–182. viral vectors; R. Kotin (NIH), B. Davidson (University of Iowa), and I. Bezproz- Finsterwald, C., Fiumelli, H., Cardinaux, J.R., and Martin, J.L. (2010). vanny (UTSW) for access to original AAV vectors; T. Curran (University of Regulation of dendritic development by BDNF requires activation of CRTC1 Pennsylvania) for c-fos promoter; A. Tabuchi and M. Tsuda (Toyama Univer- by glutamate. J. Biol. Chem. 285, 28587–28595.

104 Neuron 84, 92–106, October 1, 2014 ª2014 Elsevier Inc. Neuron Region-Speciﬁc Activation of CRTC1-CREB in Memory

Frankland, P.W., Bontempi, B., Talton, L.E., Kaczmarek, L., and Silva, A.J. Mayr, B., and Montminy, M. (2001). Transcriptional regulation by the phos- (2004). The involvement of the anterior cingulate cortex in remote contextual phorylation-dependent factor CREB. Nat. Rev. Mol. Cell Biol. 2, 599–609. fear memory. Science 304, 881–883. Nakajima, T., Uchida, C., Anderson, S.F., Lee, C.G., Hurwitz, J., Parvin, J.D., Fujii, H., Inoue, M., Okuno, H., Sano, Y., Takemoto-Kimura, S., Kitamura, K., and Montminy, M. (1997). RNA helicase A mediates association of CBP with Kano, M., and Bito, H. (2013). Nonlinear decoding and asymmetric represen- RNA polymerase II. Cell 90, 1107–1112. a tation of neuronal input information by CaMKII and calcineurin. Cell Rep. 3, Nonaka, M. (2009). A Janus-like role of CREB protein: enhancement of syn- 978–987. aptic property in mature neurons and suppression of synaptogenesis and Fukushima, H., Maeda, R., Suzuki, R., Suzuki, A., Nomoto, M., Toyoda, H., Wu, reduced network synchrony in early development. J. Neurosci. 29, 6389– L.J., Xu, H., Zhao, M.G., Ueda, K., et al. (2008). Upregulation of calcium/ 6391. calmodulin-dependent protein kinase IV improves memory formation and Nonaka, M., Doi, T., Fujiyoshi, Y., Takemoto-Kimura, S., and Bito, H. (2006). rescues memory loss with aging. J. Neurosci. 28, 9910–9919. Essential contribution of the ligand-binding beta B/beta C loop of PDZ1 and Greer, P.L., and Greenberg, M.E. (2008). From synapse to nucleus: calcium- PDZ2 in the regulation of postsynaptic clustering, scaffolding, and localization dependent gene transcription in the control of synapse development and func- of postsynaptic density-95. J. Neurosci. 26, 763–774. tion. Neuron 59, 846–860. Nonaka, M., Fujii, H., Kim, R., Kawashima, T., Okuno, H., and Bito, H. (2014). Han, J.H., Kushner, S.A., Yiu, A.P., Cole, C.J., Matynia, A., Brown, R.A., Neve, Untangling the two-way signalling route from synapses to the nucleus, and R.L., Guzowski, J.F., Silva, A.J., and Josselyn, S.A. (2007). Neuronal competi- from the nucleus back to the synapses. Philos. Trans. R. Soc. Lond. B Biol. tion and selection during memory formation. Science 316, 457–460. Sci. 369, 20130150. Heinrich, A., von der Heyde, A.S., Bo¨ er, U., Phu do, T., Tzvetkov, M., and Okuno, H. (2011). Regulation and function of immediate-early genes in the Oetjen, E. (2013). Lithium enhances CRTC oligomer formation and the interac- brain: beyond neuronal activity markers. Neurosci. Res. 69, 175–186. tion between the CREB coactivators CRTC and CBP—implications for CREB- Okuno, H., Akashi, K., Ishii, Y., Yagishita-Kyo, N., Suzuki, K., Nonaka, M., dependent gene transcription. Cell. Signal. 25, 113–125. Kawashima, T., Fujii, H., Takemoto-Kimura, S., Abe, M., et al. (2012). Inverse Hioki, H., Kuramoto, E., Konno, M., Kameda, H., Takahashi, Y., Nakano, T., synaptic tagging of inactive synapses via dynamic interaction of Arc/Arg3.1 Nakamura, K.C., and Kaneko, T. (2009). High-level transgene expression in with CaMKIIb. Cell 149, 886–898. neurons by lentivirus with Tet-Off system. Neurosci. Res. 63, 149–154. Parra-Damas, A., Valero, J., Chen, M., Espan˜ a, J., Martı´n, E., Ferrer, I., Jagannath, A., Butler, R., Godinho, S.I., Couch, Y., Brown, L.A., Vasudevan, Rodrı´guez-Alvarez, J., and Saura, C.A. (2014). Crtc1 activates a transcriptional S.R., Flanagan, K.C., Anthony, D., Churchill, G.C., Wood, M.J., et al. (2013). program deregulated at early Alzheimer’s disease-related stages. J. Neurosci. The CRTC1-SIK1 pathway regulates entrainment of the circadian clock. Cell 34, 5776–5787. 154, 1100–1111. Ravnskjaer, K., Kester, H., Liu, Y., Zhang, X., Lee, D., Yates, J.R., III, and Jeong, H., Cohen, D.E., Cui, L., Supinski, A., Savas, J.N., Mazzulli, J.R., Yates, Montminy, M. (2007). Cooperative interactions between CBP and TORC2 J.R., III, Bordone, L., Guarente, L., and Krainc, D. (2012). Sirt1 mediates neuro- confer selectivity to CREB target gene expression. EMBO J. 26, 2880– protection from mutant huntingtin by activation of the TORC1 and CREB tran- 2889. scriptional pathway. Nat. Med. 18, 159–165. Sakamoto, K., Norona, F.E., Alzate-Correa, D., Scarberry, D., Hoyt, K.R., and Josselyn, S.A., Shi, C., Carlezon, W.A., Jr., Neve, R.L., Nestler, E.J., and Davis, Obrietan, K. (2013). Clock and light regulation of the CREB coactivator CRTC1 M. (2001). Long-term memory is facilitated by cAMP response element-bind- in the suprachiasmatic circadian clock. J. Neurosci. 33, 9021–9027. ing protein overexpression in the amygdala. J. Neurosci. 21, 2404–2412. Sasaki, T., Takemori, H., Yagita, Y., Terasaki, Y., Uebi, T., Horike, N., Kawashima, T., Okuno, H., Nonaka, M., Adachi-Morishima, A., Kyo, N., Takagi, H., Susumu, T., Teraoka, H., Kusano, K., et al. (2011). SIK2 is a Okamura, M., Takemoto-Kimura, S., Worley, P.F., and Bito, H. (2009). key regulator for neuronal survival after ischemia via TORC1-CREB. Synaptic activity-responsive element in the Arc/Arg3.1 promoter essential Neuron 69, 106–119. for synapse-to-nucleus signaling in activated neurons. Proc. Natl. Acad. Sci. Sassone-Corsi, P., Visvader, J., Ferland, L., Mellon, P.L., and Verma, I.M. USA 106, 316–321. (1988). Induction of proto-oncogene fos transcription through the adenylate Kawashima, T., Kitamura, K., Suzuki, K., Nonaka, M., Kamijo, S., Takemoto- cyclase pathway: characterization of a cAMP-responsive element. Genes Kimura, S., Kano, M., Okuno, H., Ohki, K., and Bito, H. (2013). Functional label- Dev. 2, 1529–1538. ing of neurons and their projections using the synthetic activity-dependent Screaton, R.A., Conkright, M.D., Katoh, Y., Best, J.L., Canettieri, G., Jeffries, promoter E-SARE. Nat. Methods 10, 889–895. S., Guzman, E., Niessen, S., Yates, J.R., III, Takemori, H., et al. (2004). The Kida, S., Josselyn, S.A., Pen˜ a de Ortiz, S., Kogan, J.H., Chevere, I., CREB coactivator TORC2 functions as a calcium- and cAMP-sensitive coinci- Masushige, S., and Silva, A.J. (2002). CREB required for the stability of new dence detector. Cell 119, 61–74. and reactivated fear memories. Nat. Neurosci. 5, 348–355. Sekeres, M.J., Mercaldo, V., Richards, B., Sargin, D., Mahadevan, V., Woodin, Koo, S.H., Flechner, L., Qi, L., Zhang, X., Screaton, R.A., Jeffries, S., Hedrick, M.A., Frankland, P.W., and Josselyn, S.A. (2012). Increasing CRTC1 function S., Xu, W., Boussouar, F., Brindle, P., et al. (2005). The CREB coactivator in the dentate gyrus during memory formation or reactivation increases mem- TORC2 is a key regulator of fasting glucose metabolism. Nature 437, 1109– ory strength without compromising memory quality. J. Neurosci. 32, 17857– 1111. 17868. Li, S., Zhang, C., Takemori, H., Zhou, Y., and Xiong, Z.Q. (2009). TORC1 reg- Silva, A.J., Kogan, J.H., Frankland, P.W., and Kida, S. (1998). CREB and mem- ulates activity-dependent CREB-target gene transcription and dendritic ory. Annu. Rev. Neurosci. 21, 127–148. growth of developing cortical neurons. J. Neurosci. 29, 2334–2343. Song, Y., Altarejos, J., Goodarzi, M.O., Inoue, H., Guo, X., Berdeaux, R., Kim, Lonze, B.E., and Ginty, D.D. (2002). Function and regulation of CREB family J.H., Goode, J., Igata, M., Paz, J.C., et al.; CHARGE Consortium; GIANT transcription factors in the nervous system. Neuron 35, 605–623. Consortium (2010). CRTC3 links catecholamine signalling to energy balance. Mair, W., Morantte, I., Rodrigues, A.P., Manning, G., Montminy, M., Shaw, Nature 468, 933–939. R.J., and Dillin, A. (2011). Lifespan extension induced by AMPK and calcineurin Suzuki, A., Fukushima, H., Mukawa, T., Toyoda, H., Wu, L.J., Zhao, M.G., Xu, is mediated by CRTC-1 and CREB. Nature 470, 404–408. H., Shang, Y., Endoh, K., Iwamoto, T., et al. (2011). Upregulation of CREB- Mamiya, N., Fukushima, H., Suzuki, A., Matsuyama, Z., Homma, S., Frankland, mediated transcription enhances both short- and long-term memory. P.W., and Kida, S. (2009). Brain region-speciﬁc gene expression activation J. Neurosci. 31, 8786–8802. required for reconsolidation and extinction of contextual fear memory. Tabuchi, A., Nakaoka, R., Amano, K., Yukimine, M., Andoh, T., Kuraishi, Y., J. Neurosci. 29, 402–413. and Tsuda, M. (2000). Differential activation of brain-derived neurotrophic

Neuron 84, 92–106, October 1, 2014 ª2014 Elsevier Inc. 105 Neuron Region-Speciﬁc Activation of CRTC1-CREB in Memory

factor gene promoters I and III by Ca2+ signals evoked via L-type voltage- lated cAMP responsive element binding protein activity in the rat forebrain. dependent and N-methyl-D-aspartate receptor Ca2+ channels. J. Biol. J. Neuroendocrinol. 23, 754–766. Chem. 275, 17269–17275. West, A.E., Chen, W.G., Dalva, M.B., Dolmetsch, R.E., Kornhauser, J.M., Tsai, W.W., Niessen, S., Goebel, N., Yates, J.R., III, Guccione, E., and Shaywitz, A.J., Takasu, M.A., Tao, X., and Greenberg, M.E. (2001). Calcium Montminy, M. (2013). PRMT5 modulates the metabolic response to fasting regulation of neuronal gene expression. Proc. Natl. Acad. Sci. USA 98, signals. Proc. Natl. Acad. Sci. USA 110, 8870–8875. 11024–11031. Tse, D., Takeuchi, T., Kakeyama, M., Kajii, Y., Okuno, H., Tohyama, C., Bito, H., and Morris, R.G. (2011). Schema-dependent gene activation and memory Zhou, Y., Won, J., Karlsson, M.G., Zhou, M., Rogerson, T., Balaji, J., Neve, R., encoding in neocortex. Science 333, 891–895. Poirazi, P., and Silva, A.J. (2009). CREB regulates excitability and the alloca- Watts, A.G., Sanchez-Watts, G., Liu, Y., and Aguilera, G. (2011). The distribution of memory to subsets of neurons in the amygdala. Nat. Neurosci. 12, tion of messenger RNAs encoding the three isoforms of the transducer of regu- 1438–1443.

106 Neuron 84, 92–106, October 1, 2014 ª2014 Elsevier Inc.