CPEB3 is Associated with Human Episodic Memory

Cytoplasmic polyadenylation element-binding (CPEB) proteins are crucial for synaptic plasticity and memory in model organisms. A highly conserved, mammalian-specific short intronic sequence within CPEB3 has been identified as a ribozyme with self-cleavage properties. In humans, the ribozyme sequence is polymorphic and harbors a single nucleotide polymorphism that influences cleavage activity of the ribozyme. Here we show that this variation is related to performance in an episodic memory task and that the effect of the variation depends on the emotional valence of the presented material. Our data suggest a role for human CPEB3 in human episodic memory.

of the third CPEB3 exon is suffi cient to carry out self-cleavage, forms the core of the ribozyme, and harbors a single nucleotide polymorphism (SNP rs11186856 according to dbSNP; U to C substitution at position 36 of the core sequence; Figure 1). Compared to the ancestral sequence, the rare C containing sequence exhibits more than twofold rate of self-cleavage. A higher rate of selfcleavage can result in an increased number of truncated CPEB3 pre-mRNAs and ultimately in decreased CPEB3 protein expression (Salehi-Ashtiani et al., 2006).
Given the presumed role of CPEB proteins in synaptic plasticity and memory in model organisms and the recent evidence suggesting the existence of a mammalian conserved, functional CPEB3 ribozyme, we aimed at investigating the role of CPEB3 in human memory. The fact that the human CPEB3 ribozyme sequence harbors a SNP affecting the ribozyme functioning enabled us to adopt a behavioral genetics approach by correlating inter-individual variability in memory performance with individual patterns of genetic variability within CPEB3. We have focused on episodic memory as assessed by verbal memory recall tests 5 min and 24 h after learning (de Quervain et al., 2003).

STUDY SUBJECTS AND MEMORY TESTING
Memory testing and genotyping were performed in 333 healthy young Swiss subjects (232 females, 101 males; median age: 22 years, range: 18-45 years). After complete description of the study to the subjects, written informed consent was obtained. The ethics committee of the Canton of Zurich, Switzerland approved the study protocol. Subjects viewed six series of fi ve semantically unrelated German nouns presented at a rate of one word per second with the instruction to learn the words for immediate free

INTRODUCTION
Learning and memory requires biochemical and structural changes at specifi c neuronal synapses and RNA-binding molecules are thought to play an important role in inducing and maintaining such synaptic changes (Klann and Dever, 2004). The cytoplasmic polyadenylation element (CPE) promotes cytoplasmic polyadenylation of dormant mRNAs that have small poly(A) tails. In response to synaptic stimulation the poly(A) tails are elongated and translation can ensue (Huang et al., 2003;Mendez et al., 2000). This mechanism allows for protein dependent cellular and synaptic transformations. Cytoplasmic polyadenylation element-binding (CPEB) proteins constitute a family of regulators of mRNA translation and have been related to synaptic plasticity and memory formation in model organisms (Richter, 2007). In Aplysia, a neuron-specifi c CPEB isoform is upregulated locally at activated synapses and is required for the maintenance of long-term facilitation, possibly through a prion-like mechanism (Si et al., 2003a,b). The Drosophila CPEB2 protein Orb2 is required for long-term courtship memory (Keleman et al., 2007). CPEB1 knockout mice do not show detectable defects in longterm memory formation, however, extinction of hippocampaldependent memories is reduced (Berger-Sweeney et al., 2006). CPEB3, another member of the CPEB family, which has been also related to synaptic plasticity, is present in the human hippocampus and is co-localized with the synaptic marker synaptophysin (Huang et al., 2006).
Recently, a genome-wide search for ribozymes (i.e., catalytic RNAs) in the human genome identifi ed a highly conserved, mammalian-specifi c short intronic sequence within CPEB3 as a ribozyme with self-cleavage properties possibly regulating CPEB3 translation (Salehi-Ashtiani et al., 2006). A 72-nt sequence upstream recall after each series. The words were taken from the collections of Hager and Hasselhorn (1994) and consisted of 10 neutral words such as "angle" [valence: 0.2 ± 0.3 (mean ± SEM), rating on a scale from −10 (most negative valence) to +10 (most positive valence)], 10 positive words such as "joy" (valence: 7.9 ± 0.4) and 10 negative words such as "poverty" (valence: −8.5 ± 0.3). Words were recalled immediately after every series of fi ve words. Five minutes and 24 h after presentation of these words, subjects underwent unexpected delayed free-recall tests for the same words. The design of an unexpected free-recall test was chosen to prevent confounding factors such as unbalanced reiteration of the learned words.

RESULTS
In a population of 333 healthy young individuals who underwent a verbal episodic memory task, 49 homozygous carriers of the rare C allele-containing ribozyme sequence showed signifi cantly poorer delayed memory recall performance (5 min after presentation of 30 semantically unrelated words, see Materials and Methods) than T allele carriers of SNP rs11186856 (7.5 ± 0.4 recalled words 5 min after learning (mean ± SEM) vs. 9.0 ± 0.2, respectively; F = 9.2; df = 1; P = 0.003; Figure 2). Nearby SNPs in the same haplotype block were also signifi cantly associated with episodic memory performance (Figure 2). In accordance with the haplotype structure of this genomic region, signifi cances dropped to chance levels at chromosomal positions 93822581 (rs1935864) and 94052611 (rs11186914). We did not observe an allele-dose effect, i.e., the reduction of episodic memory performance was restricted to homozygous C allele carriers of SNP rs11186856 (CC genotype group, n = 49: 7.5 ± 0.4 words; CT genotype group, n = 140: 9.0 ± 0.2; TT genotype group, n = 144: 9.0 ± 0.2; F = 4.6; df = 2; P = 0.011). Furthermore, there was no genotype-dependent difference in immediate recall performance (P = 0.9). Homozygous carriers of the rare C allele of SNP rs11186856 also showed signifi cantly poorer delayed memory recall performance 24 h after word presentation than T allele carriers (6.7 ± 0.5 recalled words (mean ± SEM) vs. 7.8 ± 0.2, respectively; F = 4.0; df = 1; P = 0.045).
The 30 words presented during the episodic memory task differed in their emotional valence (10 words with positive valence, 10 words with negative valence, 10 words with neutral valence; see Materials and Methods). We therefore ran the genetic association analysis of CPEB3 SNPs in each of the three groups separately. Homozygous carriers of the rare C allele-containing CPEB3 associated with human memory ribozyme sequence had signifi cantly poorer episodic memory performance for words with positive emotional content than T allele carriers of SNP rs11186856 (2.5 ± 0.2 recalled words 5 min after learning vs. 3.3 ± 0.1, respectively; F = 9.3; df = 1; P = 0.002, Figure 3A; 2.4 ± 0.2 recalled words 24 h after learning vs. 2.9 ± 0.1, respectively; F = 4.7; df = 1; P = 0.03). The corresponding signifi cance level for words with negative emotional content was weaker (at 5 min: P = 0.025, Figure 3B; at 24 h: not signifi cant). Interestingly, all of the examined SNPs failed by far to reach signifi cant levels of association for neutral words ( Figure 3C).

DISCUSSION
Behavioral genetic studies have been recently successful in identifying molecules related to human memory functions (de Quervain et al., 2003;Egan et al., 2003;Papassotiropoulos et al., 2006). Signifi cant correlations between genotypic and phenotypic variability point to genes of particular interest and may also identify the precise gene regions related to human memory performance.
Here we show that a human-specifi c polymorphism, a T to C substitution within a highly conserved, small intronic sequence of CPEB3, was associated with decreased episodic memory performance. This intronic sequence has been recently identifi ed as a mammalian conserved, functional CPEB3 ribozyme. Compared to the ancestral sequence, the rare C containing sequence exhibits more than twofold rate of self-cleavage and may therefore result in increased number of truncated CPEB3 pre-mRNAs and ultimately to decreased CPEB3 protein expression (Salehi-Ashtiani et al., 2006). Therefore, the fi nding of decreased delayed verbal memory performance in homozygous C allele carriers suggests a role for human CPEB3 in this memory task. Furthermore, we found that the genotype effects depend on the emotional valence of the presented material, which suggests that CPEB3 might be involved in neural processes and circuits related to emotional memory, such as the amygdala. Given the correlative nature of human behavioral genetic studies the question arises as to the specifi city of the observed genetic association. Importantly, performance levels in the immediate recall task were identical across CPEB3 genotype groups, which indicates that the genotype-dependent differences in delayed episodic memory were not caused by genotype effects on confounding factors such as motivation, attention or working memory. The fact that CPEB3 was associated with episodic memory performance already 5 min after learning does not contradict the presumed role of CPEB proteins in long-term memory formation because human shortterm memory performance is affected by preexisting associative networks stored in long-term memory (Stuart and Hulme, 2000). In addition to the ribozyme SNP rs11186856, adjacent SNPs were also signifi cantly associated with memory performance because they are located within one haplotype block. Given the genomic positions of the adjacent SNPs and the absence of any biological data supporting a functional role of these polymorphisms, we assume that the genetic variation causing the observed association is SNP rs11186856.
Taken together, we show that human CPEB3 is associated with delayed verbal memory recall, especially of emotionally arousing information. Further studies are warranted to explore the precise molecular mechanism of this genetic association and its specifi city for emotional information.