Retrieval induces reconsolidation of fear extinction memory

Abstract

The nonreinforced expression of long-tem memory may lead to two opposite protein synthesis-dependent processes: extinction and reconsolidation. Extinction weakens consolidated memories, whereas reconsolidation allows incorporation of additional information into them. Knowledge about these two processes has accumulated in recent years, but their possible interaction has not been evaluated yet. Here, we report that inhibition of protein synthesis in the CA1 region of the dorsal hippocampus after retrieval of fear extinction impedes subsequent reactivation of the extinction memory trace without affecting its storage or that of the initial fear memory. Our results suggest that extinction memory is susceptible to a retrieval-induced process similar to reconsolidation in the hippocampus.

Without retrieval, memories would be unusable. Retrieval, however, can weaken long-term memory (LTM). In fact, it is known that reactivation through exposure to training-related stimuli transiently destabilizes the consolidated memory trace, which, to remain behaviorally available, must go through a protein synthesis-dependent process called reconsolidation. Additionally, when retrieval in the absence of appropriate reinforcement is repeated regularly enough, it induces memory extinction, a phenomenon characterized by a decrease in the amplitude and/or frequency of the learned response that also requires protein synthesis in definite areas of the brain.

Reconsolidation and extinction are functionally related; both require acquisition of retrieval-related information connected to previous learning. They are mechanistically different, however; extinction involves additional learning and replaces the expression of the original memory with a newly formed one (1, 2), whereas reconsolidation restabilizes the old memory opened to modification or strengthening by retrieval (3–6).

Given the clinical relevance that reconsolidation blockade and extinction enhancement could have (7, 8), research about the consequences of retrieval on LTM persistence has concentrated on the molecular, neuroanatomical, and electrophysiological requirements of reconsolidation and extinction (9–12) as well as on the elucidation of the behavioral and neurochemical conditions that constrain or facilitate either process (13–15). The potential interaction of extinction and reconsolidation has been comparatively less studied, and when so, the studies produced contradictory results. Thus, it was earlier proposed that extinction can act as a boundary condition preventing fear memories from undergoing reconsolidation (16, 17), but more recent reports indicate that reconsolidation is independent of any influence from fear extinction (18). Indeed, the likelihood of reconsolidation for fear extinction memory has never been evaluated before.

To do that, we analyzed the effect of the intrahippocampal administration of protein synthesis inhibitors on the stability of a consolidated fear extinction memory trace. Our findings strongly support the hypothesis that, as happens for other memory types, fear extinction memory is susceptible to a retrieval-induced process similar to reconsolidation in the hippocampus.

Results

To analyze whether inhibition of protein synthesis after reactivation affects retention of fear extinction, we used a one-trial step-down inhibitory avoidance (IA) task, a form of aversive learning in which stepping down from a platform is paired with a mild foot shock. IA memory is hippocampus-dependent (19) and can be extinguished by allowing the animal to step down to and explore the floor of the training box in the absence of the ensuing foot shock. If this procedure is repeated once a day for 5 consecutive days, it induces clear-cut extinction of the avoidance response that lasts for at least 14 d (Fig. 1A) and requires protein synthesis in dorsal CA1 during a short postretrieval time window to occur (20).

Fig. 1.

Inhibition of hippocampal protein synthesis after reactivation hinders extinction memory. (A) Rats trained in the IA task using a 2-s 0.5-mA foot shock as an unconditioned stimulus were submitted to five extinction sessions 24, 48, 72, 96, and 120 h after training. Retention was evaluated 1, 7, or 14 d later. Note the absence of spontaneous recovery of the avoidance response. (B) Rats trained in the IA task were submitted to five extinction sessions 24, 48, 72, 96, and 120 h posttraining. One day later, extinction memory was reactivated; immediately thereafter, animals were given VEH or ANI in dorsal CA1. Retention was evaluated 3, 24, or 168 h later. (C) Rats were treated as in B except that VEH and ANI were given 6 h after reactivation. (D) Animals were treated as in B except that they received EME or CHX instead of ANI. (E) Rats were treated as in B and C except that reactivation was carried out 7 d after the last extinction training session (n = 10–12 per group). Bars represent the mean (±SEM) of step-down latencies. *P < 0.05; **P < 0.01; ***P < 0.001 using the Student t test or Bonferroni test after ANOVA. The dotted arrows indicate the moment of drug infusion.

Rats trained in the IA task were submitted to the extinction procedure described above (first session, 24 h posttraining), and 24 h after the last extinction session, extinction memory was reactivated. Immediately or 6 h later, the animals received bilateral intra-CA1 infusions of the protein synthesis inhibitor anisomycin (ANI; 160 μg per side) (21–23). Retention was assessed at different postreactivation times. As can be seen in Fig. 1B, when given immediately after reactivation of extinction memory, ANI impaired retention of extinction and, as a consequence, allowed reappearance of the IA response. This effect was observed 24 h and 168 h but not 3 h after reactivation [t₍₁₅₎ = 4.72, P < 0.001 and t₍₁₅₎ = 3.52, P < 0.01 for 24 h and 168 h, respectively]. Conversely, ANI did not affect the retention of extinction at any postinfusion time evaluated when given into dorsal CA1 6 h after reactivation (Fig. 1C), suggesting that the impairment of extinction memory was attributable to a time- and protein synthesis-dependent process and not to an unspecific action of ANI on behavior or hippocampal functionality. Nevertheless, it has been reported that inhibition of protein synthesis with ANI might disrupt other neural functions able to interfere with memory processing (24). Therefore, to discard any ambiguous interpretation of our results, we analyzed the effect of two other widely used protein synthesis inhibitors, emetine (EME) (25) and cycloheximide (CHX) (26), on the persistence of fear extinction. Intra-CA1 infusion of EME (50 μg per side) or CHX (20 μg per side) immediately after reactivation of extinction memory also impaired its retention and allowed recovery of the IA response [F_(2,25) = 15.98, P < 0.001; t_{EME vs. VEH} = 4.61, P < 0.001; t_{CHX vs. VEH} = 5.10, P < 0.001 and F_(2,24) = 7.70, P < 0.01; t_{EME vs. VEH} = 2.60, P < 0.05; t_{CHX vs. VEH} = 3.82, P < 0.01 for 24 h and 168 h, respectively] (Fig. 1D). It has been suggested that the postretrieval effect of some amnesic agents depends on the age of the mnemonic trace. Thus, young memories would be more prone than older ones to pharmacological disruption following reactivation (27, 28). We found, however, that the amnesic effect of ANI, EME, and CHX on the retention of fear extinction memory also occurred when reactivation was carried out 7 d instead of 1 d after the last extinction training session [F_(3,33) = 4.79, P < 0.01; t_{ANI vs. VEH} = 2.72; P < 0.05; t_{EME vs. VEH} = 3.21, P < 0.01; t_{CHX vs. VEH} = 3.04, P < 0.05) (Fig. 1E). Importantly, neither ANI, EME, nor CHX affected retention of fear extinction memory when given 1 or 7 d after the end of the extinction procedure but in the absence of relevant behavioral stimuli (Fig. 2).

Fig. 2.

Inhibition of hippocampal protein synthesis in the absence of relevant behavioral stimuli does not affect retention extinction memory. (A) Rats trained in the IA task were submitted to five extinction sessions at 24, 48, 72, 96, and 120 h after training. One day after the last session, the animals received bilateral infusions of VEH, ANI, EME, or CHX in the CA1 region of the dorsal hippocampus. Retention was evaluated 24 h thereafter. (B) Animals were treated exactly as in A except that the drugs were infused 7 d after the last extinction session (n = 10–12 per group). Bars represent the mean (±SEM) of step-down latencies.

So far, two different scenarios have been proposed to explain the amnesia caused by the postreactivation administration of protein synthesis inhibitors and, hence, the reconsolidation phenomenon. One of these hypotheses postulates that the amnesic effect is attributable to a more or less protracted blockade of the retrieval process (29–31), whereas the other states that the amnesia is consequence of storage impairment (32–34).

Therefore, to study whether the deleterious effect on extinction memory retention and the resulting recovery of the avoidance response induced by hippocampal protein synthesis inhibition was attributable to erasure or, alternatively, to impaired expression of the fear extinction memory trace, animals trained in the IA task were randomly assigned to one of two experimental groups. Animals in group 1 were not submitted to any relevant behavioral experience and received vehicle (VEH) or ANI into dorsal CA1 6 d after IA training. Animals in group 2 were submitted to one daily extinction session for 5 d. Twenty-four hours after the last extinction training session, extinction memory was reactivated; immediately after that, the animals received VEH or ANI into dorsal CA1. Starting 24 h after these procedures, animals in both groups were submitted to one daily extinction training session for 5 consecutive days. We found that when given 6 d after IA training in the absence of any behaviorally relevant stimuli (group 1), ANI did not affect expression of the IA response or the subsequent acquisition of extinction (Fig. 3A). As expected (Fig. 1), when given after reactivation of extinction memory, ANI induced recovery of the avoidance response, and, importantly, extinction of this recovered response was more rapid than that of the original one [t₍₂₂₎ = 5.49, P < 0.001] (Fig. 3B), indicating that there are savings of extinction memory after ANI treatment. Nevertheless, as can be seen in Fig. 3C, extinction of the recovered IA response was also blocked by ANI given into dorsal CA1 immediately after nonreinforced retrieval [t₍₂₄₎ = 3.40, P < 0.01 for ANI/VEH vs. ANI/ANI] suggesting that, as happens for extinction of the original avoidance memory (20), the occurrence of a protein synthesis-dependent process is also required for extinction of the recovered IA trace.

Fig. 3.

Inhibition of hippocampal protein synthesis after reactivation does not erase extinction memory. (A) IA task-trained rats received bilateral intra-CA1 infusions of VEH or ANI 6 d posttraining and, beginning 1 d later, were submitted to one daily extinction session for 5 d. (B) Rats trained in the IA task were submitted to five extinction sessions at 24, 48, 72, 96, and 120 h posttraining. One day later, extinction memory was reactivated; immediately thereafter, animals received VEH or ANI into dorsal CA1. Beginning 1 d after reactivation, animals were submitted to one daily extinction session for 5 d. (C) IA task-trained rats were submitted to five extinction sessions at 24, 48, 72, 96, and 120 h after training. One day later, extinction memory was reactivated; immediately thereafter, animals received VEH or ANI into dorsal CA1. Twenty-four hours later, animals were submitted to a second reactivation session; immediately thereafter, they received VEH or ANI. Retention was reevaluated 24 h later (n = 10–12 per group). Bars represent the mean (±SEM) of step-down latencies. **P < 0.01; ***P < 0.001 using the Student t test. The dotted arrows indicate the moment of drug infusion.

Discussion

Our results show that inhibition of hippocampal protein synthesis during a restricted postreactivation time window hampers retention of consolidated IA extinction memory lastingly, allowing reappearance of the initial fear memory trace. This phenomenon was observed with three different protein synthesis inhibitors, thus ruling out any possible nonspecific pharmacological byproduct. Moreover, the amnesic effect was time-dependent, concomitant with the reactivation of extinction memory and not attributable to spontaneous recovery (35), because it was not seen in animals that received VEH or were given ANI 6 h after reactivation. Indeed, the training protocol used here induces an extinction memory lasting at least 14 d (Fig. 1). Recovery of the IA response cannot be ascribed to renewal or reinstatement either. The former occurs when the conditioned stimulus is presented outside of the extinction context (36), whereas the latter results from the unexpected delivery of the unconditioned stimulus (37). None of these conditions were present in our experimental design. Hence, the most plausible explanation for our data is that fear extinction weakens after reactivation and, to remain behaviorally available, must undergo a protein synthesis-dependent reconsolidation-like process in the hippocampus.

Currently, the prevailing model about the effect of retrieval on memory persistence states that exposing an animal to a learned situation in the absence of proper reinforcement will reactivate the consolidated trace and initiate one of two opposing and competing processes: consolidation of extinction memory and reconsolidation of the original memory (32, 38; a different standpoint is discussed in 39–41). Which one of these two processes will dominate future behavior is thought to depend on environmental, physiological, and behavioral conditions, the so-called “boundary conditions,” at the moment of reactivation (13, 42). Together with quite recent findings (43), our results suggest that the dynamics of memory processing after reactivation are much more complex than previously thought and indicate that the current model must be amended to include the possibility that consolidated extinction memory is also open to updating and modification by retrieval. This possibility certainly has deep implications at the clinical level and could help us to understand, and avoid, relapse following the apparent success of the extinction-based therapies used for the treatment of anxiety disorders (44, 45).

Whether postreactivation administration of protein synthesis inhibitors disintegrates the neurobiological substrates of the consolidated trace or, instead, affects its future expression, either transiently or lastingly, is still matter of debate (5). We believe that these two seemingly antagonist hypotheses are not mutually exclusive, however. Indeed, one could argue that inhibition of protein synthesis after nonreinforced reactivation could hinder future retention not by deleting the trace or impairing its expression but by affecting an active protein synthesis-dependent process initiated at the moment of retrieval and necessary to access or find the information needed to build up a suitable behavioral representation. In this respect, the existence of extinction savings after the recovery of IA memory induced by ANI and the need of protein synthesis for extinction of the recovered avoidance response indicate that reconsolidation blockade does not fully erase extinction memory but, instead, hinders restabilization of, or access to, information necessary for remembering it (34, 46).

Materials and Methods

Animals, Surgery, and Drug Infusion.

Naive male Wistar rats (3 mo old, 300–350 g) raised in our own facilities or bought at the Fundação Estadual de Produção e Pesquisa em Saúde do Rio Grande do Sul were used. The animals were housed five to a cage and kept with free access to food and water under a 12/12-h light/dark cycle, with light onset at 7:00 AM. The temperature of the animal room was maintained at 22–24 °C. To implant them with indwelling cannulas, rats were deeply anesthetized with thiopental (i.p., 30–50 mg/kg) and 27-gauge cannulas were stereotaxically aimed to the CA1 region of the dorsal hippocampus (anterior, −4.2; lateral, ±3.0; ventral, −1.8) in accordance with coordinates taken from the atlas of Paxinos and Watson (47). Animals were allowed to recover from surgery for 4 d before submitting them to any other procedure. At the time of drug delivery, 30-gauge infusion cannulas were tightly fitted into the guides. Infusions (1 μL per side) were carried out over 60 s with an infusion pump, and the cannulas were left in place for an additional 60 s to minimize backflow. The placement of the cannulas was verified postmortem: 2–4 h after the last behavioral test, 1 μL of a 4% (wt/vol) methylene-blue solution was infused as described above, and the extension of the dye 30 min thereafter was taken as an indication of the presumable diffusion of the VEH or drug previously given to each animal. Only data from animals with correct implants were analyzed. All procedures were conducted in accordance with the National Institutes of Health principles of laboratory animal care. Every effort was made to reduce the number of animals used and to minimize their suffering.

IA Learning Task.

Rats were trained in a one-trial step-down IA task during the light phase of the subjective day (between 9:00 and 11:00 AM). The training apparatus was a 50 × 25 × 25-cm Plexiglas box with a 5-cm- high, 8-cm-wide, and 25-cm-long platform on the left end of a series of bronze bars that made up the floor of the box. For training, animals were gently placed on the platform facing the left rear corner of the training box. When they stepped down and placed their four paws on the grid, they received a 2-s 0.5-mA scrambled foot shock and were immediately withdrawn from the training box.

Extinction of IA LTM.

To extinguish the avoidance response, rats were submitted to several nonreinforced IA test sessions 24 h apart. For this purpose, IA-trained animals were put back on the training box platform until they stepped down to the grid. No foot shock was given, and the animals were allowed to explore the training apparatus freely for 30 s after they had stepped down. During this time, the animals stepped up onto the platform and down again several times. To reactivate the extinction memory trace, the animals were put on the training box platform until they stepped down, and right after that, they were removed from the training box. In some experiments (Fig. 3B), animals were submitted to a second extinction protocol after memory reactivation.

Drugs.

ANI, EME, and CHX (Sigma–Aldrich) were dissolved and stored protected from light at −20 °C until use. Right before that, an aliquot was thawed and diluted to a working concentration with 0.1% DMSO in saline (pH 7.2). The doses used were determined based on pilot experiments and previous studies showing the behavioral effects of each compound.

Statistical Analyses.

The Student t test or one-way ANOVA, followed by the Bonferroni test, were used for comparison of two or more than two groups, respectively.