Behavioral tagging is a general mechanism of long-term memory formation Fabricio Ballarinia,1, Diego Moncadaa,1, Maria Cecilia Martineza, Nadia Alena, and Haydee �� Violaa,b,2 aInstituto de Biolog��a Celular y Neurociencias, Facultad de Medicina, Universidad de Buenos Aires-CONICET, Paraguay 2155, 3�� Piso, CP 1121, Buenos Aires, Argentina and bDepartamento de Fisiolog��a, Biolog��a Molecular y Celular, Facultad de Ciencias Exactas y Naturales, Universidad de Buenos Aires, Intendente Guiraldes �� 2160, Ciudad Universitaria, CP 1428, Buenos Aires, Argentina Communicated by Ivan A. Izquierdo, Pontifical Catholic University of Rio Grande do Sul, Porto Alegre, Brazil, June 26, 2009 (received for review March 24, 2009) In daily life, memories are intertwined events. Little is known about the mechanisms involved in their interactions. Using two hippocampus-dependent (spatial object recognition and contex- tual fear conditioning) and one hippocampus-independent (con- ditioned taste aversion) learning tasks, we show that in rats subjected to weak training protocols that induce solely short term memory (STM), long term memory (LTM) is promoted and formed only if training sessions took place in contingence with a novel, but not familiar, experience occurring during a critical time window around training. This process requires newly synthesized proteins induced by novelty and reveals a general mechanism of LTM formation that begins with the setting of a ������learning tag������ estab- lished by a weak training. These findings represent the first comprehensive set of evidences indicating the existence of a behavioral tagging process that in analogy to the synaptic tagging and capture process, need the creation of a transient, protein synthesis-independent, and input specific tag. hippocampus insular cortex memory consolidation protein synthesis novelty Atemporallyofand cquisition new information can be stored into at least two mechanistically different memory types: a short-term (STM) and a long-term memory (LTM). It is well known that in order for LTM to be established, synaptic changes must be stabilized by the action of newly synthesized proteins (1). This process of memory trace consolidation takes place in the brain areas where such forms of learning and memory are likely to reside. Surprisingly, recent evidence demonstrated that the supply of newly synthesized proteins may also derived from another behavioral event occurring in a relatively long-lasting associative time window, helping to promote LTM for a weak learning task that otherwise would only induce STM (2). Current models, based on seminal ideas, propose that memories are stored by stable changes in synaptic weight modifying the activity of specific neuronal circuits (3���5). Therefore, those specific synapses activated by a given learning will require the supply of new plasticity-related proteins (PRPs) for LTM to be formed. As a consequence, there should be a mechanism that restricts the action of PRPs to recently activated synapses but not to others. To address this biological problem, it was suggested that a transient local synaptic tag is established at those recently activated synapses where PRPs are specifically captured. This idea was originally postulated by Frey and Morris (6) and it is now known as the synaptic tagging and capture (STC) hypothesis (7���9). In their seminal work they showed that early-LTP, a transient form of LTP that is induced by a weak stimulus, could be extended to late-LTP, a more persistent form of LTP, if the weak and the strong stimuli were applied in a relatively long-lasting associative time window on different synapses of the same neuron. Frey and colleagues also found that an hippocampal LTP can be reinforced by exposing rats to a novelty or to a holeboard training and this phenomenon is protein synthesis- dependent (10, 11). Considering that LTP and some forms of LTM share a number of properties, such as associativity, durability and protein synthesis dependence (12), it is feasible that a mechanism of STC, also operates in the formation of LTM. The main goal of the present work is to study the process of behavioral tagging using several combinations of behavioral tasks, to investigate whether a behavioral experience can extend the duration of the memory for an independent task by providing the PRPs needed for its consolidation. We found that the behavioral tagging process operates in different hippocampus- and cerebral cortex-dependent learning tasks, suggesting that it represents a general mechanism of LTM formation. Results To study behavioral tagging processes we subjected rats to different hippocampus-dependent or independent learning tasks, and tested whether a weak training gave rise to a LTM by the relatively long-lasting temporal association with a novel experience in a protein synthesis-dependent way. First, rats were trained in a hippocampus-dependent spatial object recognition (SOR) task, consisting of 4 min exploration of two identical objects located in a familiar arena. Fig. 1A shows that time spent exploring both objects in the training session was similar, resulting in an exploration time near 50%. In the test session, performed 30 min after this weak training protocol, animals showed STM expressed by a preferential exploration of the object switched to a new location (P 0.001, Fig. 1A). This weak training was ineffective to yield LTM, because an independent group of rats tested 24 h later did not display a preferential exploration of the objects (P 0.05, Fig. 1A). Next, we asked whether a LTM lasting at least 24 h could be obtained by pairing the SOR with the exploration of a novel environment. Therefore, we exposed rats to a 5 min open field (OF) session at different times before or after the weak SOR training task. OF exploration performed 1 h before the SOR training or between 15 min to 2 h after it, promoted the formation of a SOR-LTM (P 0.01, Fig. 1B). Control animals, that were not exposed to the OF, and those groups that were exposed to the novel OF 4 or 2 h before SOR training did not express SOR-LTM (P 0.05, Fig. 1B). The same applied to the groups that explored the novel OF immediately or 4 h after SOR training (P 0.05, Fig. 1B). Thus, this permissive action of a spatial novelty on memory formation is restricted to a critical time window. To directly address whether the promotion of LTM was due to the novel nature of the OF environment, a group of animals was exposed to a 30 min OF session the previous day to familiarize them with the arena. In contrast to the promoting effect induced by exploration of a novel OF, no SOR-LTM formation was observed when rats explored the familiar OF 1 h after a weak SOR training (P 0.05, Fig. 1C). Author contributions: D.M. and H.V. designed research F.B., D.M., M.C.M., and N.A. performed research F.B., D.M., M.C.M., N.A., and H.V. analyzed data and F.B., D.M., and H.V. wrote the paper. The authors declare no conflict of interest. 1F.B. and D.M. contributed equally to this work. 2To whom correspondence should be addressed. E-mail: hviola@fmed.uba.ar. www.pnas.org cgi doi 10.1073 pnas.0907078106 PNAS August 25, 2009 vol. 106 no. 34 14599���14604 NEUROSCIENCE

By analogy to the STC hypothesis, protein synthesis products derived from one behavioral task and used to stabilize a transient STM into LTM of another learning task reveal the existence of a behavioral tagging process. Thus, it is imperative to demonstrate that OF exploration provides the proteins necessary to consolidate SOR-LTM. For this reason, we infused the translation inhibitor anisomycin in the CA1 region of the dorsal hippocampus, imme- diately after the OF exploration performed 15 min after a weak SOR training. We found that the infusion of anisomycin blocked the promoting effect of OF on SOR-LTM (P 0.01, Fig. 1D). Consistent with previous results obtained in an inhibitory avoidance task (2), de novo protein synthesis elicited by spatial novelty is necessary to promote SOR-LTM. However, the intra-CA1 infusion of anisomycin 15 min before a weak SOR training did not impair the promoting effect of novel OF exposure when given 1 h before training, suggesting that the setting of a ������learning tag������ by weak SOR training does not require protein synthesis (Fig. 1E). Next, we tested if LTM formation of another hippocampus- dependent memory task, the contextual fear conditioning (CFC) paradigm, could be promoted by a novel exposure to an OF. Thus, we trained rats in CFC paradigm employing a weak protocol that only induced STM. An increase in the percentage of freezing was observed when animals were tested in the same context 15 min after the training session (P 0.001, Fig. 2A), but not observed 24 h later (P 0.05, Fig. 2A). To test whether CFC-LTM was promoted by an exploration of a novel arena, we subjected rats to a 5 min OF session 1 h before a weak CFC training. As expected, in test session performed 24 h after training, CFC-LTM was observed only if the OF experience was novel to the animal (P 0.001, Fig. 2B). When rats explored a familiar OF, the promoting effect on CFC-LTM was prevented (P 0.05, Fig. 2B). We next determined whether proteins synthesized after a novel OF exposure were necessary to induce the consolidation of CFC-LTM after a weak training. We infused vehicle or anisomycin into the CA1 immediately after the OF session. One hour later, rats were trained with a weak CFC protocol and LTM tested the following day. Fig. 2C shows the promoting effect of OF exposure on CFC-LTM in animals infused with vehicle solution (P 0.001) again, as happened with the SOR task, this effect was totally abolished by the protein synthesis inhibitor. These experiments, together with others reported recently (2), strongly suggest that a mechanism of behavioral tagging operates in three different hippocampus-dependent memory tasks in rats. Several important questions arise from these results. Is this phe- nomenon selective for hippocampal memories? Could a cerebral cortex-dependent memory be promoted by an appropriate stimu- lus? Is the exploration of a novel OF effective in the promotion of a cortical-dependent memory? To answer these questions, we Fig. 1. Spatial novelty promotes SOR-LTM in a protein synthesis-dependent manner. A���E show the percentage of exploration time as mean SEM. (A) Animals explored the arena for 4 min. Independent groups of rats were tested at 30 min (STM,n 13) or 24 h (LTM, n 15) after training. ***, P 0.001 vs. training, Student���s t test. (B) Control animals (n 20) received 4 min SOR training. Animals in the OF group were exposed to a novel OF 240, 120, or 60 min before (n 15, 17, and 17, respectively) or 0, 15, 60, 120, or 240 min after (n 15, 20, 15, 18, and 15, respectively) SOR training. In the group 0 , animals explore the OF immediately after the SOR training. **, P 0.01 vs. control, Dunnet test after one-way ANOVA. (C) Animals were subjected, or not (Control, n 20), to a novel (n 18) or familiar (n 15) OF 1 h after a 4 min SOR training and LTM was tested 24 h later.**, P 0.01 vs. Control and Fam test sessions, Newman-Keuls after one-way ANOVA. (D) Animals were subjected, or not (Control, n 20), to a novel OF 15 min after a 4 min SOR training. Experimental groups received intra-CA1 infusions of vehicle (Nov Veh, n 17) or anisomycin (Nov Ani, n 15) immediately after OF.**, P 0.01 vs. all groups in test sessions, Newman-Keuls after one-way ANOVA. (E) Rats were tested 24 h after a weak SOR training in the absence (Control, n 8) or in the presence of OF 1 h before training. Experimental groups received intra-CA1 infusions of vehicle (Nov Veh, n 8) or anisomycin (Nov Ani, n 7) 15 min before training. **, P 0.01 vs. Control test session, Newman-Keuls after one-way ANOVA. 14600 www.pnas.org cgi doi 10.1073 pnas.0907078106 Ballarini et al.