Spatial-Memory Formation After Spaced Learning Involves ERKs1/2 Activation Through a Behavioral-Tagging Process

Abstract

The superiority of spaced over massed learning is an established fact in the formation of long-term memories (LTM). Here we addressed the cellular processes and the temporal demands of this phenomenon using a weak spatial object recognition (wSOR) training, which induces short-term memories (STM) but not LTM. We observed SOR-LTM promotion when two identical wSOR training sessions were spaced by an inter-trial interval (ITI) ranging from 15 min to 7 h, consistently with spaced training. The promoting effect was dependent on neural activity, protein synthesis and ERKs1/2 activity in the hippocampus. Based on the "behavioral tagging" hypothesis, which postulates that learning induces a neural tag that requires proteins to induce LTM formation, we propose that retraining will mainly retag the sites initially labeled by the prior training. Thus, when weak, consecutive training sessions are experienced within an appropriate spacing, the intracellular mechanisms triggered by each session would add, thereby reaching the threshold for protein synthesis required for memory consolidation. Our results suggest in addition that ERKs1/2 kinases play a dual role in SOR-LTM formation after spaced learning, both inducing protein synthesis and setting the SOR learning-tag. Overall, our findings bring new light to the mechanisms underlying the promoting effect of spaced trials on LTM formation.

Figure 1

A single weak SOR training session induces SOR-STM but not SOR-LTM; retraining with a second SOR session within a critical time window ranging from 15 min to 7 h is effective to promote LTM. (a,b) show the preference index, expressed as mean ± SEM, registered in a training session (TR) or a test session performed 30 min or 24 h after training. (a) Rats were exposed to a 4-min wSOR training session (TR, n = 12) and independent groups were tested either 30 min (STM, n = 12) or 24 h (LTM, n = 12) after training. Newman-Keuls analysis after one-way ANOVA, F_(2,33) = 21.51; ***p < 0.001 vs. TR and LTM. (b) One-Trial group (n = 17) received a single 4-min wSOR training. Animals in the 2-Trials group were trained with two 4-min SOR sessions spaced by different ITIs spanning from 5 min to 24 h (5 min, n = 10; 15 min, n = 10; 30 min, n = 11; 1 h, n = 15; 4 h, n = 13; 7 h, n = 18; 9 h, n = 13; 24 h, n = 13). Representative first training session (TR, n = 18). Dunnett’s test after one-way ANOVA, F_(9,128) = 5.780; **p < 0.01 vs. 1 Trial and **p < 0.01 vs. TR.

Figure 2

Inhibition of neural activation or protein synthesis in the dorsal hippocampus prevents SOR-LTM formation induced by retraining. (a–c), show the preference index as mean ± SEM in a first training session (TR), which is representative for all groups, and in a test session. (a) Independent animals were submitted to two identical wSOR training sessions spaced by 1 h and received bilateral infusions of either vehicle (2 Trials Veh, n = 10) or muscimol (2 Trials Mus, n = 7) in the dorsal hippocampus, immediately after the second training session. They were tested 24 h later. Animals exposed to a single wSOR training session (1 Trial, n = 8) received a vehicle infusion 65 min after that and were tested 24 h later. Training session (TR, n = 10). Newman–Keuls analysis after one-way ANOVA, F_(3,31) = 7.323; **p < 0.01 vs. all other groups. (b) One-Trial group injected with vehicle (n = 6) and 2-Trials group injected with vehicle (Veh, n = 8) or emetine (Eme, n = 6) immediately after the second training session were tested 24 h later. Training session (TR, n = 8). Newman–Keuls analysis after one-way ANOVA, F_(3,24) = 11.88; ***p < 0.001 vs. all other groups. (c) One-Trial group injected with vehicle (n = 14) and 2-Trials group injected with vehicle (Veh, n = 14) or anisomycin (Ani, n = 10) immediately after the second training session were tested 24 h later. Training session (TR, n = 14). Newman–Keuls analysis after one-way ANOVA, F_(3,48) =6.864; **p < 0.01 vs. all other groups.

Figure 3

Exploration of an open field within a critical time window induces SOR-LTM formation and that is prevented by hippocampal inhibition of ERKs1/2. (Top) The flow chart shows the experimental protocol using wSOR and open field. One-Trial animals (n = 10) received bilateral dorsal hippocampus infusions of vehicle 15 min previous to wSOR training session and were tested the following day. Independent animals received bilateral dorsal hippocampus infusions of either vehicle or U0126 15 min before wSOR training session and were exposed to a novel OF session either 4 h (4 h Veh, n = 9) or 1 h (1 h Veh, n = 10; 1 h U0126, n = 12) after wSOR. Training session (TR, n = 11) is representative for all groups. Data are expressed as mean ± SEM. Newman–Keuls analysis after one-way ANOVA, F_(4,47) =12.05; ***p < 0.001 for all other groups.

Figure 4

Hippocampal inhibition of ERKs1/2 prevents SOR-LTM formation induced by wSOR retraining, acting on learning-tag and protein synthesis processes. (a–d) show the preference index as mean ± SEM registered in the first training session (TR), which is representative for all groups, or in the test session performed 24 h after training. (a) One-Trial group (n = 8) was injected with vehicle 15 min before a single wSOR training session and tested on the following day. Independent animals received intra-dorsal hippocampus infusions of vehicle (n = 11) or U0126 (n = 9) 15 min before being subjected to two identical wSOR training sessions spaced by 4 h; another group was also exposed to an OF session 1 h after both training sessions (n = 6). Training session (TR, n = 12). SOR-LTM was tested 24 h after the second training session. Newman–Keuls analysis after one-way ANOVA, F_(4,41) = 12.18; ***p < 0.001 vs. TR, 1 Trial and 2 Trials U0126. (b) One-Trial group (n = 11) was injected with vehicle 4 h after a single wSOR training session and tested on the next day. Independent animals were subjected to two identical wSOR sessions spaced by 4 h and immediately after that, they received bilaterally infusions of either vehicle (n = 18) or U0126 (n = 15); another group was also exposed to an OF session 1 h after that training (n = 10). Training session (TR, n = 18). SOR-LTM was tested 24 h after the second training session. Newman–Keuls analysis after one-way ANOVA, F_(4,67) = 16.30; ***p < 0.001 vs. all other groups. (c) The experimental protocol is similar to (a), except that the ITI is 1 h. One-Trial group of animals (n = 10), 2-Trials group of rats infused with vehicle (n = 13) or U0126 (n = 11) and the retrained group plus an OF session (n = 7). Training session (TR, n = 13). Newman–Keuls analysis after one-way ANOVA, F_(4,49) = 11.64; **p < 0.01; ***p < 0.001 vs. TR, 1 Trial and 2 Trials U0126. (d) The experimental protocol is similar to (b), except that the ITI is 1 h. One-Trial group of animals was injected with vehicle 1 h after training (n = 6), 2-Trials group of rats infused with vehicle (n = 11) or U0126 (n = 10) and the retrained group plus an OF session (n = 8). Training session (TR, n = 12). Newman–Keuls analysis after one-way ANOVA, F_(4,42) = 13.49; ***p < 0.001 vs. TR and 2 Trials U0126; ^##p < 0.01 vs. 1 Trial.

Figure 5

A model of the effect of different retraining protocols on SOR-LTM formation. The model is proposed on the basis of the present findings and the BT hypothesis, which postulates the requirement of the temporal and spatial convergence of a learning tag (solid line) set by the training, and synthesis of plasticity-related proteins (PRPs; dashed line), in order to promote the formation of LTM 24 h later. Such a convergence is indicated by the intersection of the PRPs and the tag lines, in which case SOR-LTM is observed. For PRP synthesis to occur, we suggest that activated ERKs1/2 levels (p-ERKs1/2, dotted area) induced by each weak SOR (wSOR) training sessions and lasting ca. 7 h should be subjected to an additive or synergistic process (bolder dot area) if separated by an appropriate inter-trial interval (ITI) ranging from 15 min to 7 h. The timeline (top: 0–24 h) indicates the time points of the experimental procedures. The rows show different cases varying from a single wSOR session (upper row) to two wSOR sessions separated by different ITIs (5 min - 9 h). The first row shows that a single wSOR session does not induce LTM on the following day despite inducing tag setting and enhancing ERKs1/2 level as it would be insufficient to trigger the required synthesis of PRPs. Two consecutive and identical wSOR sessions separated by 5 min (second row) or 9 h (fifth row) do not induce LTM as no PRP synthesis would occur in either case. Although each session would tag the same cellular substrates, in the first case, levels of ERKs1/2 induced by the first session would not be further enhanced or extended by the second one to reach the threshold necessary for PRP synthesis. This would be due to the necessity of a minimal ITI for the machinery inducing ERKs1/2 activation by the second session to be operational. In the second case, the tag and ERKs1/2 levels of the first session decay over time during the long ITI and do not reach the second session, thus impeding protein synthesis. The third and fourth rows correspond to ITIs of 1 and 4 h, respectively, in which PRP synthesis would occur. In both cases, ERKs1/2 levels would be enhanced and the same cellular substrates would be retagged by the second session. This convergence would lead to PRP synthesis necessary for LTM formation. However, as ERKs1/2 activity was not quantified under these circumstances, alternative explanations (besides addition or synergy in ERKs1/2 activation level) to explain how ERKs1/2 functionality may relate to wSOR training and PRP synthesis cannot be excluded.