Delayed intrinsic activation of an NMDA-independent CaM-kinase II in a critical time window is necessary for late consolidation of an associative memory

Abstract

Calcium/calmodulin-dependent kinases (CaM-kinases) are central to various forms of long-term memory (LTM) in a number of evolutionarily diverse organisms. However, it is still largely unknown what contributions specific CaM-kinases make to different phases of the same specific type of memory, such as acquisition, or early, intermediate, and late consolidation of associative LTM after classical conditioning. Here, we investigated the involvement of CaM-kinase II (CaMKII) in different phases of associative LTM induced by single-trial reward classical conditioning in Lymnaea, a well established invertebrate experimental system for studying molecular mechanisms of learning and memory. First, by using a general CaM-kinase inhibitor, KN-62, we found that CaM-kinase activation was necessary for acquisition and late consolidation, but not early or intermediate consolidation or retrieval of LTM. Then, we used Western blot-based phosphorylation assays and treatment with CaMKIINtide to identify CaMKII as the main CaM-kinase, the intrinsic activation of which, in a critical time window ( approximately 24 h after learning), is central to late consolidation of LTM. Additionally, using MK-801 and CaMKIINtide we found that acquisition was dependent on both NMDA receptor and CaMKII activation. However, unlike acquisition, CaMKII-dependent late memory consolidation does not require the activation of NMDA receptors. Our new findings support the notion that even apparently stable memory traces may undergo further molecular changes and identify NMDA-independent intrinsic activation of CaMKII as a mechanism underlying this "lingering consolidation." This process may facilitate the preservation of LTM in the face of protein turnover or active molecular processes that underlie forgetting.

Figure 1.

The combinations of training, injection, and memory test regimes used to investigate the role of CaM-kinases in long-term memory after single-trial classical reward conditioning. A, To test for a role of CaM-kinase type enzymes in acquisition, KN-62 or vehicle was injected 30 min before training (CS + US) and memory tested 24 h after training (CS). B, To test for a role in retrieval, KN-62 or vehicle was injected 30 min before presenting the CS at 24 h after training. C, To test for a role in early consolidation, KN-62 or vehicle was injected 15 min after training and memory was tested 24 h after training. D, To test for a role in intermediate and late consolidation, KN-62 or vehicle was injected at various different time points (X = 6, 12, 20, 24, or 30 h) after training and memory was tested 18 h later.

Figure 2.

Comparison of the role of the activation of CaM-kinases in the acquisition and retrieval of long-term memory after single-trial classical conditioning. A, B, Activation of a CaM-kinase type enzyme is required for acquisition, but not for retrieval of memory. KN-62 injection at 30 min before training revealed a significant inhibition of the feeding response to the test stimulus [A; n = 34, 33, and 40 for the KN-62, Vehicle and Naive groups, respectively; test statistics: ANOVA, F_(2,104) = 9.32, p < 0.001; Tukey's test (KN-62 vs Vehicle): p < 0.05]. However, no drug effect was observed when KN-62 was injected at 30 min before memory retrieval [B; n = 23, 21 and 14, respectively; test statistics: ANOVA, F_(2,55) = 11.9, p < 0.001; Tukey's test (KN-62 vs Vehicle]: p > 0.05). C, The feeding response to the US was not affected by injection of KN-62 30 min before testing (n = 18, 21 and 19 for the KN-62, Vehicle and Naive groups, respectively). Test statistics: ANOVA: F_(2,55) = 2.1, p > 0.05 (n.s.). Asterisks in A through C indicate statistically significant pairwise differences (Tukey's test: p < 0.05) compared to the Naive group data. Error bars in this and subsequent figures show ± SEM.

Figure 3.

CaM-kinase activation is involved in late consolidation of memory during a distinct time window, but not in early or intermediate consolidation. A, Injection of KN-62 15 min, 6 or 12 h after training does not lead to impaired long-term memory. Test statistics: ANOVAs: 15 min, F_(2,78) = 33.43, p < 0.001; 6 h, F_(2,103) = 12.70, p < 0.001; 12 h, F_(2,50) = 5.74, p < 0.01. Tukey's tests (all 3 experiments): KN-62 versus Vehicle, p > 0.05 (n.s.). B, Middle panel, Injection of KN-62 24 h after training causes amnesia 18 h later. Test statistics: ANOVA: F_(2,91) = 11.22, p < 0.001; Tukey's test (KN-62 versus Vehicle): p < 0.01. B, Left and right panels, Injection of KN-62 20 or 30 h after training does not lead to impaired long-term memory. Test statistics: ANOVAs: 20 h, F_(2,64) = 4.78, p < 0.05; 30 h, F_(2,85) = 11.58, p < 0.01. Tukey's tests (both experiments): KN-62 versus Vehicle, p > 0.05 (n.s.). Asterisks in A and B indicate statistically significant responses (Tukey's tests, p < 0.05) compared to the Naive group data.

Figure 4.

Amino acid (AA) sequence predicted from the partially cloned Lymnaea αCaMKII cDNA and the corresponding AA sequence of human αCaMKII. The homology of these sequences between the two species is 86%. The cloned region contains the αCaMKII autoregulatory (including autophosphorylation) domain (Lymnaea position 89–101) and part of the catalytic domain (position 1–88). Asterisks indicate identical amino acids in the Lymnaea and human sequences.

Figure 5.

Increased intrinsic phosphorylation of CaMKII 24 h after single-trial classical reward conditioning and late memory consolidation. A, B, Western blot experiments. Three main experimental groups were used, one injected with KN-62 24 h after paired training, one injected with vehicle and one uninjected. Cerebral and buccal ganglia were dissected from all trained animals at 24.5 h after training and also from animals of an unpaired control group and a naive control group. Homogenates from five replicates of five pooled tissue samples per group (see Materials and Methods) were loaded onto gels and the resulting blots probed with either a pCaMKII antibody (example in A, top) or with a total CaMKII antibody (example in A, bottom). The total CaMKII signal was used as an internal standard for densitometric comparisons. Relative integrated density values (pCaMKII/CaMKII) in samples taken from naive animals were assigned a value of 1. All other relative integrated density values were normalized to this naive baseline (dashed line) and used for multiple statistical comparisons (shown in B, significance compared to Unpaired group indicated by asterisks). Test statistics: ANOVA: F_(3,19) = 11.2, p < 0.003; Tukey's tests: No inj. (paired) versus Unpaired, p < 0.05; Vehicle (paired) versus Unpaired, p < 0.01; KN-62 (paired) versus No inj. (paired) and Vehicle (paired), p < 0.05 and 0.01, respectively; KN-62 (paired) versus Unpaired, p > 0.05 (n.s.), No. inj. (paired) versus Vehicle (paired), p > 0.05 (n.s.). C, Memory tests. Three groups of trained (paired stimulus) animals from the experiment described in A and B were kept for behavioral tests performed 18 h after injection. Naive and unpaired control animals from the same experiment were also saved for behavioral tests. Note that for an easier comparison with the densitometric data in B, for the Naive group, only the mean value is shown (dashed line; SE ± 1.6). Test statistics: ANOVA: F_(4,138) = 6.6, p < 0.0001; Tukey's tests: No inj. (paired) versus Unpaired and Naive, p < 0.01 and 0.05, respectively; Vehicle (paired) versus Unpaired and Naive, p < 0.01 and 0.05, respectively; KN-62 (paired) versus No inj. (paired) and Vehicle (paired), p < 0.05; KN-62 (paired) versus Unpaired and Naive, p > 0.05 (n.s.); No. inj. (paired) versus Vehicle (paired), p > 0.05 (n.s.). Significance compared to unpaired group is indicated by asterisks.

Figure 6.

Dissociation of the role of NMDA receptor and CaMKII activity in different phases of memory formation after single-trial classical conditioning. A, B, CaMKII is necessary during both acquisition (injection at 30 min before training) and late consolidation (injection at 24 h after training) of associative memory. Test statistics: A, ANOVA: F_(2,64) = 6.80, p < 0.01; Tukey's test (CaMKIINtide vs Vehicle): p < 0.05; B, ANOVA: F_(2,69) = 8.28, p < 0.001; Tukey's test (CaMKIINtide vs Vehicle): p < 0.05. C, D, Activation of NMDA receptors is only necessary for acquisition but not late consolidation of associative memory. Test statistics: C, ANOVA: F_(2,67) = 5.70, p < 0.01; Tukey's test (MK-801 vs Vehicle): p < 0.05; D, ANOVA: F_(2,66) = 6.4, p < 0.01; Tukey's test (MK-801 vs Vehicle): p > 0.05 (n.s.). Significance compared to naive group is indicated by asterisks.