Disrupting reconsolidation is a promising approach to dampen the behavioral expression of fear memory in patients with anxiety disorders or posttraumatic stress disorder. In a series of human fear conditioning studies we showed that disrupting reconsolidation by the noradrenergic β-blocker propranolol is very effective in erasing conditioned fear responding.
The basic design used in our human fear-conditioning studies on disrupting reconsolidation includes testing over different phases across three consecutive days. On day 1 – the fear acquisition phase, healthy participants are exposed to a series of picture presentations. One picture stimulus (CS1+) is repeatedly paired with an aversive electric stimulus (US), resulting in the acquisition of a fear association, whereas another picture stimulus (CS2-) is never followed by an US. On day 2 – the memory reactivation phase, the participants are re-exposed to the conditioned stimulus without the US (CS1-), which typically triggers a conditioned fear response. After the memory reactivation we administer an oral dose of 40 mg of propranolol HCl, a β-adrenergic receptor antagonist that indirectly targets the protein synthesis required for reconsolidation by inhibiting the noradrenaline-stimulated CREB phosphorylation. On day 3 – the test phase, the participants are again exposed to the unreinforced conditioned stimuli (CS1- and CS2-) in order to measure the fear-reducing effect of the manipulation. This retention test is followed by an extinction procedure and the presentation of situational triggers to test for the return of fear. Potentiation of the eye blink startle reflex is measured as an index for conditioned fear responding. Declarative knowledge of the fear association is measured through online US expectancy ratings during each CS presentation. In contrast to extinction learning, disrupting reconsolidation targets the original fear memory thereby preventing the return of fear. Although the clinical applications are still in their infancy, disrupting reconsolidation of fear memory seems to be a promising new technique with the prospect to persistently dampen the expression of fear memory in patients suffering from anxiety disorders and other psychiatric disorders.
Our brains are programmed to learn. Humans are well equipped with the ability to learn the potential dangers in life and perhaps even more important to learn the predictors of danger. Pavlovian aversive conditioning is an excellent tool to study associative fear learning not only in humans but also across a wide range of organisms1,2. This procedure involves presenting an innocuous biologically neutral conditioned stimulus (CS, e.g., a tone or picture) with a noxious or harmful unconditioned stimulus (US), typically a mild electrical shock. If the CS becomes a reliable predictor of the US, the CS will elicit species-typical conditioned behavioral responses (e.g., freezing in rats and potentiated startle reflex in humans), which are conceptualized as expressions of fear (see 3 for critical comments on this terminology). Not only does aversive conditioning research add to a better understanding of the molecular and cellular processes of associative fear learning and memory4 but it also provides a foundation for understanding the etiology and course of anxiety disorders5. Though it bears mentioning that anxiety disorders do not necessarily result from direct conditioning experiences such as traumatic events. They may also result from indirect or vicarious fear learning experiences6. But irrespective of the learning history, associative fear memory lies at the core of anxiety disorders. The Pavlovian conditioning paradigm has not only proven its utility in understanding the origin of anxiety disorders, it is also an excellent translational model to develop and advance treatment for anxiety disorders5.
In the laboratory the two most extensively studied procedures to reduce learned fear are (1) extinction and (2) disruption of reconsolidation. Even though extinction training — i.e., the repeated unreinforced re-exposure to the CS7,8 — is an effective anxiolytic strategy, animal and human fear conditioning studies reliably show that the conditioned fear response can easily return by re-exposure to unsignaled USs (i.e., reinstatement), a context change (i.e., renewal), or the passage of time (i.e., spontaneous recovery)7,9,10. A consensus has been building that extinction learning does not erase the original fear memory but instead reflects the formation of a new inhibitory memory. As a consequence, the fear memory may resurface resulting in a return of fear even after originally successful fear extinction. The various sources of relapse are still a major challenge for clinical practice.
An alternative approach to diminish conditioned fear responding is the more recently (re)discovered procedure of disrupting memory reconsolidation by pharmacological agents. This procedure is very promising as it not only diminishes conditioned fear responding but it even seems to erase associative fear memory, which might eventually solve the problem of relapse.
Memory reconsolidation refers to a two-phase process by which previously consolidated memories transfer to a transient destabilized state at retrieval and require a time-dependent restabilization to persist14-18. Gene transcription, protein and RNA synthesis are necessary for this restabilization and offer a window of opportunity for amnestic agents to affect the fear memory. Amnesia for learned fear has been demonstrated in animals by drugs (e.g., anisomycin) targeting the required protein synthesis14,19,20 or the release of neurotransmitters21,22. Protein synthesis may also be disturbed by the noradrenergic beta-blocker propranolol, which is supposed to inhibit the noradrenaline-stimulated CREB phosphorylation23-25. The fear reducing effects of propranolol have been demonstrated in animals and humans26-32. In a series of discriminative fear-conditioning studies, we consistently demonstrated that propranolol (40 mg) administered prior to or after memory reactivation effectively reduced the conditioned startle fear response and prevented the return of fear in healthy participants.
Proof of Principle
The basic design used in our human fear-conditioning studies on disrupting reconsolidation includes testing over different phases across three consecutive days separated by at least 24 hr in order to (1) support (re)consolidation of the memories and (2) allow the drug to wash out before testing. On day 1 – the fear acquisition phase, healthy participants are exposed to a series of picture presentations. One picture stimulus (CS1+) is repeatedly paired with an aversive electric stimulus (US), resulting in the acquisition of a fear association, whereas another picture stimulus (CS2-) is never followed by an US. On day 2 – the memory reactivation phase, re-exposure to the unreinforced conditioned stimulus (CS1-) typically triggers a conditioned fear response. In this phase, we systemically administer 40 mg propranolol HCl, a β-adrenergic receptor antagonist that indirectly targets the protein synthesis required for reconsolidation25. In view of the peak plasma concentrations of propranolol33, we administered an oral dose of propranolol 90 min prior to the reactivation of the fear memory in our first experiments. However, for a more optimal test of reconsolidation, the amnestic agent should be administered after the memory reactivation. Therefore, in our latter experiments we always administered propranolol after memory reactivation with very similar results. On day 3 – the test phase, the retention of the fear memory is tested 24 hr after the intervention (i.e., first test trial day 3), followed by an extinction procedure and situational triggers to test for the return of fear (i.e., reinstatement, renewal, spontaneous recovery, rapid reacquisition). For determining whether the effect of propranolol requires active retrieval of the fear memory, the drug should be administered to another fear-conditioned group without reactivation of the memory (i.e., propranolol no-reactivation condition)27,28. We used the potentiation of the startle reflex as a measure for conditioned fear responding and online US expectancy ratings as an index for contingency learning. The potentiation of the startle reflex is conceived as a specific and reliable index of fear34, subserved by the amygdala35. The most commonly used startle-eliciting stimulus is the “startle probe”, a loud noise that is presented through headphones during a stimulus or in the interval between two stimulus presentations (i.e., intertrial intervals)36. Stronger startle responses to the loud noise during the fear conditioned stimulus (CS1+) as compared to the control stimulus (CS2-) reflects the fearful state of the participant elicited by the feared stimulus (CS1+).
In a series of 10 consecutive experiments in independent samples 27-32,37 we consistently replicated our original finding where disrupting reconsolidation of fear memory by a beta-adrenergic blocker propranolol HCl effectively reduced the conditioned startle reflex and prevented the return of fear27. The observations that exposure to primary reinforcers (i.e., reinstatement), a change in context (i.e., renewal) or simple the passage of time (i.e., spontaneous recovery) did not lead to the re-emergence of conditioned fear responding as is generally observed after extinction training, support the superiority of disrupting reconsolidation over extinction learning. In addition to these retrieval techniques, reacquisition learning did not reveal any savings of the previously learned fear response. In sum, these findings suggest that disrupting reconsolidation of associative fear memory by propranolol effectively reduced the emotional expression of fear memory.
Furthermore, we tested several boundary and necessary conditions for disrupting reconsolidation: (1) The memory reactivation session seems procedurally similar to extinction training (i.e., unreinforced exposure), but it should involve less unreinforced trials than extinction training because noradrenergic beta-blockers (i.e., propranolol) may also interfere with the formation of extinction memory instead of targeting the original fear memory38,39; (2) Memory retrieval is not sufficient for memory reconsolidation18,31,40,41. Propranolol only reduced the conditioned fear responding when there was something to be learned during the retrieval session31,32. A discrepancy between what has already been learned and what can be learned on a given retrieval session (i.e., prediction error) seems to be necessary for inducing reconsolidation of associative fear memory32. The results indicate that the occurrence of a prediction error is a necessary condition for reconsolidation and provide a useful instrument for developing and optimizing reconsolidation-based treatments for patients suffering from anxiety disorders.
Ethics statement: Our physician-approved protocol meets the requirements of the Ethical Committee of the Department of Psychology at the University of Amsterdam for testing human participants.
1. Attachment of the Fear Potentiated Startle and Shock Electrodes
2. Fear Acquisition – Day 1
Duration about 45 min.
3. Memory Reactivation – Day 2
Duration about 105 min.
4. Test – Day 3
Duration about 45 min.
Manipulation check propranolol: Contrary to pill placebo, both the systolic and diastolic BP as well as the salivary alpha amylase should decrease 90 min after the propranolol intake during memory reactivation, which indicates that the drug manipulation exerted its intended physiological effect. BP and salivary alpha amylase should again return to baseline levels at the test on day 3.
US expectancy ratings: With the current protocol you may not expect any effects of the propranolol treatment on the US expectancy ratings. Irrespective of the pharmacological treatment, the differential US expectancy ratings should increase during fear acquisition and decrease during extinction learning 48 hr later. Moreover the presentation of the unsignaled USs is expected to result in a return of the differential expectancy ratings (Figure 1A-C).
Prerequisite for reconsolidation: In order to trigger the process of reconsolidation, the memory trace should firstly be destabilized as a necessary condition for the restabilization phase. For fear memory destabilization, an expectation of threat at the moment of reactivation (i.e., intact CS1-US expectancy) is necessary31 though not sufficient. Pursuing on the idea that the function of reconsolidation is to update the memory trace to an ever-changing environment, the memory reactivation should also involve a prediction error (PE: a mismatch between what is expected and what actually happens). Given that the US expectancy ratings remain unaffected by the propranolol manipulation, changes in threat expectations from acquisition to test (PE) may serve as an independent indicator of memory destabilization (see 32 for more details). But only when the memory reactivation follows asymptotic learning, these changes in threat expectation may be observed. In case of partially reinforced non-asymptotic learning, a shift in threat expectancies is not necessary for memory destabilization. Hence a retrieval trial may also trigger reconsolidation without a change in expectancies after non-asymptotic learning.
Startle fear responding: Successful fear acquisition on day 1 is demonstrated by an increase in differential startle fear responding (i.e., CS1 vs. CS2) from the first to the last trial of acquisition (Figure 1D-F). Higher startle fear responding to the stimulus that was previously coupled with the shock (CS1) compared to noise alone during memory reactivation on day 2 further indicates that the fear memory is well consolidated.
During the test phase on day 3 you may expect intact startle fear responding in the placebo group: startle responding to the CS1 should be higher compared to the CS2 at the beginning of test. Extinction training subsequently reduces the differential startle responding (i.e., CS1 vs. CS2). Moreover the presentation of the unsignaled USs should result in a return of fear responding to the CS1 compared to the CS2 (Figure 1D).
Contrary to pill placebo, the administration of propranolol HCl is expected to result in an elimination of the differential startle fear responding (i.e., CS1 vs. CS2) at the beginning of the test phase on day 3. Furthermore you may expect that the unsignaled USs will not recover the startle fear responding (Figure 1E).
Note that propranolol should not have any fear-reducing effects when the pill is administered without memory reactivation on day 2 (Figure 1F).
Figure 1. US expectancy ratings to the CS1 and CS2 trials and startle fear responses to the CS1, CS2 and noise alone (NA) trials during acquisition, memory reactivation, extinction and test for the placebo (A,D), propranolol (B,E), and propranolol no- reactivation (C,F) groups. R-CS1– refers to an unreinforced reactivation trial. Please click here to view a larger version of this figure.
In a series of studies we consistently demonstrate that, irrespective of gender, 40 mg of the noradrenergic beta-blocker propranolol either administered prior to or after memory reactivation effectively neutralized the conditioned fear responding (i.e., defensive startle response). None of the four mechanisms of relapse — reinstatement, renewal, spontaneous recovery and rapid reacquisition — were observed after disrupting reconsolidation with propranolol27-32,37. It is noteworthy that the fear-neutralizing effects were only observed for the defensive startle response, but neither for the threat expectancy rating nor for the skin conductance. The data for the electrodermal conditioning are not reported here as the general patterns for the SCR did not significantly deviate from the expectancy ratings. In human fear conditioning research, multiple indices of conditioned responding (e.g., US-expectancy, SCR, startle response, pupil dilation, neural activity) are usually obtained for reasons of cross-validation46. However, there is now convincing evidence that these different response levels do not necessarily act in concert and may even dissociate from each other3,27-32,34,37,47. Note that the startle response is an automatic defensive reflex, which is potentiated in response to a CS that is associated with a US of negative valence and can generally not be observed with USs of neutral or positive valence47,48 (e.g., vibrotactile stimulation or reaction time task). Hence, potentiation of the defensive startle reflex is a reliable and specific index of aversive conditioning34. In contrast, SCR conditioning can occur irrespective of the valence of the US47,48. Given that affective valence of the US does not modify SCR, electrodermal conditioning is a non-specific measure of anticipatory arousal. We therefore believe that the SCR is less suited as behavioral measure in human aversive conditioning research.
If we speculate on translating these findings into clinical practice, several issues and potential limitations are to be considered. First, it may be questioned whether a pharmacological manipulation by the noradrenergic beta-blocker is really necessary to disrupt the process of reconsolidation or whether a behavioral procedure aimed to interfere with reconsolidation would yield a similar neutralizing effect. Even though a one-session treatment of a low dose of propranolol is clearly nontoxic, an entirely behavioral procedure is always preferable over a pharmacological intervention. There is indeed an alternative method where extinction training is presented within the window of reconsolidation49. Several studies failed however to replicate these original findings by Schiller et al.29,50-52 but see 53,54. In addition to these conflicting results, another potential limitation of the extinction within reactivation procedure would be that in clinical practice the fear response is generally not extinguished in a one-session exposure treatment. For instance, imaginary exposure for patients with PTSD traditionally takes ten sessions before the fear subsides55. Thus, even acknowledging the obvious disadvantages of a simple pharmacological treatment in comparison to an entirely behavioral intervention, we believe that the noradrenergic manipulation of memory reconsolidation seems to be more feasible than extinction within the reconsolidation window.
A second issue concerns the optimal conditions to trigger memory reconsolidation. There is growing evidence that the mechanisms mediating the behavioral expression of fear are clearly dissociated from the mechanisms mediating the process of reconsolidation31,32,56-60. For instance, recent animal studies uncovered differential and dissociable receptors in the basolateral amygdala mediating the expression, destabilization and restabilization of previously conditioned fear memories61,62. A behavioral expression of fear memory is not only dissociated from processes mediating memory reactivation (i.e., access to a memory trace), it seems also not being indispensable for reconsolidation to occur61. As such, fear expression during memory reactivation is not informative on whether the memory trace enters a labile phase. Given that memory destabilization is a prerequisite for the noradrenergic beta-blocker to interfere with the restabilization process, an important question is how we can infer successful memory destabilization in clinical practice. A host of findings indicate that a crucial factor in inducing reconsolidation and in demarcating reconsolidation from either memory retrieval or the consolidation of a novel memory38,63 is the degree of prediction error induced during memory retrieval31,32. But given that no objective criterion is available to determine the optimal degree of prediction error in clinical practice, the current laboratory findings cannot easily be translated to treatment protocols.
Another challenge for translating the neuroscience literature into clinical practice concerns the ecological validity of the paradigm. The evidence for disrupting reconsolidation has mainly been shown in animals and humans for relatively new (one day old) and simple fear memories (i.e., tone shock; picture shock). It is not self-evident that disrupting reconsolidation of older, stronger and broader memory networks such as in patients with PTSD is as effective as it has been shown in the laboratory for cued fear conditioning. Also with respect to the dependent variable it is still unclear whether the observations from the animal and human laboratory studies generalize to patients with anxiety disorders. The fear reducing effects are thus far mainly demonstrated for the behavioral expression of aversive conditioning (i.e., freezing behavior in rodents or defensive startle reflex in humans), with only one exception where we demonstrated that also the subjective feelings of distress were significantly neutralized by noradrenergic blockade of memory reconsolidation37. It may be questioned whether these fear-reducing effects in the laboratory are indicative of the typical experiences of fear and avoidance behavior characteristic of patients with anxiety disorders. Future research should investigate whether the current findings indeed generalize to avoidance behaviors, one of the central symptoms of anxiety disorders.
In sum, although the Pavlovian aversive conditioning procedure is an excellent tool to study the basic mechanisms of fear learning and memory, we cannot easily translate the laboratory findings into clinical practice. The insights that we have been acquiring on the optimal, boundary and necessary conditions for memory reconsolidation should only be considered as a starting point for the development of reconsolidation-based treatments. On the other hand, the extensive research on extinction training has resulted in extinction-based exposure interventions, which still belong to the most effective treatments for anxiety disorders and other related disorders. Given that the noradrenergic blockade of memory reconsolidation overshadows the anxiolytic effect of extinction learning, disrupting reconsolidation points to a promising new intervention to effectively reduce excessive and irrational fears.
The authors have nothing to disclose.
This work is supported by a VICI grant (Merel Kindt) from the Netherlands Organization for Scientific Research.
Name of Material – Equipment | Company | Catalog Number | Comments – Description |
2 computers with 4 screens | Dell | Optiplex 9010 | Recording and monitoring physiological responses. Presenting the experimental script. |
Amplifier | Developed by B. Molenkamp – University of Amsterdam | Designed around a Burr Brown INA101 amplifier and ISO103 isolation stage. | |
VSSRP98 | Developed by B. Molenkamp – University of Amsterdam | Physiological registration software. Record electromyography – EMG – activity using a bundled pair of electrodes wires connected to a front-end amplifier with an imput resistacne of 10 MΩ and a bandwidth of DC-1500 Hz. Raw EMG signals are integrated in the amplifier. Integrated EMG signals are sampled at 1000 Hz and used for data analysis. | |
MATLAB | MathWorks | Analyzing data. Peak amplitudes are determined by taking the baseline 50 ms before probe onset to peak differences within 30 – 150 ms following probe onset and are recorded in microvolt. | |
Presentation | NeuroBehavioral Systems Inc. – USA | Stimulus presentation. | |
Constant current stimulator | Digitimer – UK | DS7A | Generates electrical stimulation. |
Shock electrodes | Made by B. Molenkamp – University of Amsterdam | Ag electrodes of 20 mm * 25 mm with fixed inter-electrodes mid-distance of 45 mm. | |
Headphones | Sennheiser Electronic GmbH & CO – Germany | HD 25-1 II | Presentation of startle probse and background noises. |
EMG electrodes | Made by B. Molenkamp – University of Amsterdam | Three 7 mm sintered Ag-AgCl electrodes. | |
Double-sided adhesive collars | MedCaT – the Netherlands | 848125 | 13-mm x 5-mm. For attaching the EMG electrodes to the skin. |
Conductive gel | Signa Gel – Parker Laboratories Inc. – USA | 224.550.011 | Facilitates conduction from the skin to both the EMG and shock electrodes. |
Alcohol swabs | Sanadep 0.5 % – Microtek Medical – the Netherlands | 3053800 | For cleaning the skin of the participant. |
Sphygmomanometer | Omron Healthcare Europe B.V. – the Netherlands | M4-I HEM-752-E | Measuring blood pressures and heart rate. |
Cotton salivettes | Sarstedt – Germany | 511.534 | Obtaining salivary samples. |
Curved tip syringe | Monoject – Covidien – USA | 412 | Applies gel to EMG and shock electrodes. |
Propranolol HCl – 40 mg and placebo pills | Huygens Apotheek – the Netherlands | Pills should be identical in exterior. |