Acutely elevated glucocorticoid levels are known to affect distinct memory functions in opposite ways. Whereas these stress hormones facilitate the long-term consolidation of recent experiences
(1), high levels of glucocorticoids present at the time of retention testing reliably inhibit the retrieval of previously acquired information in animals
(2 –
4) and healthy human subjects
(5 –
8) . Recent evidence from human studies suggests that emotionally arousing information is especially sensitive to the retrieval-impairing effects of glucocorticoids
(7,
8) and that emotional arousal during the test situation is a prerequisite for enabling glucocorticoid effects on memory retrieval
(9) . Studies of rats investigating the neural mechanisms underlying this selectivity indicate that glucocorticoid effects on memory retrieval depend critically on concurrent emotional arousal-induced increases in noradrenergic transmission within the brain
(1,
3) . However, as these animal experiments used highly aversive learning paradigms (e.g., inhibitory avoidance and water maze), which seem difficult to compare with the degree of emotional arousal typically produced by recall of emotionally arousing pictures or words, it is not known whether the impairing effects of glucocorticoids on retrieval of emotionally arousing material in human beings may also require a coactivation of the noradrenergic system.
To investigate this issue, we examined whether a β-adrenoceptor antagonist would block the impairing effects of cortisone on memory retrieval of emotionally arousing words in healthy human volunteers. We used a double-blind, placebo-controlled, within-subject design. We asked volunteers to learn a list of 60 words that had variable emotionality. The following day, we administered cortisone (25 mg), propranolol (40 mg), or both drugs in combination 1 hour before participants performed a free-recall test.
Method
Forty-two healthy volunteers (21 males, 21 females, mean age=24.9 years [SD=3.6]) were recruited from the pool of university students and gave written informed consent to participate in the study, which was approved by the ethics committee of the University of Zürich, Switzerland. Exclusion criteria were the presence of an acute medical condition, a history of psychiatric or neurological illness, a history of oral glucocorticoid use, and exposure to strenuous physical or psychological stress before or during the study.
We used a double-blind, placebo-controlled, within-subject design. On day 1, subjects were presented a list of 60 unrelated German nouns with variable emotionality, ranging from words with a negative valence (e.g., victim, complaint) to words with a positive valence (e.g., palm tree, cookie). We used slightly modified versions of the word lists used in a previous study
(5) . The words were presented (each word for 4 seconds) twice in a different order on a computer screen with the instruction to learn them for delayed recall testing. Twenty-four hours after learning, the study drugs were administered 1 hour before a free-recall test in which participants were asked to write down all the words they could remember. The arousal ratings of the words were assessed with a visual analogue scale ranging from 0 (no arousal) to 8 (maximal arousal). Based on average arousal ratings, words were categorized as high-arousal words (the quartile of words with the highest arousal ratings), low-arousal words (the quartile with the lowest arousal ratings), and medium-arousal words (the middle quartiles with intermediate arousal ratings). Analyses were performed for the entire set of words as well as for the high-arousal, low-arousal, and medium-arousal words separately. After a 2-week washout period, the procedure was repeated with the same individuals with another set of words, this time with participants receiving the treatment they had not received the first time. Treatment order was balanced across subjects. All tests were performed between 1 p.m. and 4 p.m.
Saliva was sampled 1 hour after the administration of the study drugs and immediately preceding recall testing with the Salivette sampling device (Sarstedt Inc., Rommelsdorf, Germany). Free cortisol in saliva was analyzed by use of a commercially available immunoassay (CLIA; IBLHamburg, Germany). The inter- and intra-assay coefficients of variation were below 10%. To reduce the error variance due to interassay imprecision, both samples for each subject were analyzed in the same run.
Results
In a first group of 14 subjects (7 females and 7 males; mean age=26.5 years [SD=4.3]), 25 mg of cortisone or placebo was administered orally 1 hour before the delayed free-recall test, 24 hours after the word-learning exercise. This dose of cortisone, which is rapidly metabolized into the endogenous glucocorticoid hydrocortisone (cortisol), produced salivary cortisol levels at the time of memory retrieval testing comparable to those induced by major psychological or physiological stress (placebo: 6.4 nmol/liter [SD=3.2]; cortisone: 27.6 nmol/liter [SD=30.4]; p<0.05, two-sided paired t test) and impaired delayed free recall when analyzed over all words (placebo: 24.6 words [SD=11.8]; cortisone: 20.6 words [SD=11.2]; p<0.001). When analyzed separately for high- and low-arousal words, cortisone impaired the retrieval of high-arousal words by 42% (p=0.0001,
Figure 1, left panel) but did not impair retrieval of low-arousal words (p=0.3,
Figure 1, left panel) or medium-arousal words (p=0.5). There were no significant interactions between valence (negative or positive) and drug effects or between gender and drug effects (p≥0.05).
In a second group of 14 subjects (7 females and 7 males; mean age=24.2 years [SD=3.0]), 25 mg of cortisone together with 40 mg of propranolol hydrochloride, which readily enters the brain, or placebo was administered orally 1 hour before the delayed free-recall test. The combined administration of cortisone and propranolol elevated salivary cortisol levels to the same extent that cortisone alone did (placebo: 3.2 nmol/liter [SD=1.5]; cortisone plus propranolol: 35.2 nmol/liter [SD=37.8]; p<0.01), but it did not induce memory retrieval impairment when analyzed across all words (placebo: 22.1 words [SD=10.2]; cortisone plus propranolol: 21.8 words [SD=8.4]). Moreover, the combined administration of cortisone and propranolol did not induce retrieval deficits for either high- or low-arousal words (
Figure 1, center panel) or medium-arousal words. A between-groups comparison further revealed that the impairment in recall performance induced by cortisone alone was significantly greater than that induced by the combined administration of cortisone and propranolol (p=0.001, unpaired t test,
Figure 1, left and center panels).
In a third group of 14 subjects (7 females and 7 males; mean age=23.9 years [SD=2.9]), propranolol or placebo was administered 1 hour before the delayed free-recall test. Propranolol administered alone did not affect salivary cortisol levels (placebo: 3.9 nmol/liter [SD=2.1]; propranolol: 4.1 nmol/liter [SD=3.1]) or memory retrieval when analyzed over all words (placebo: 24.4 words [SD=8.8]; propranolol: 23.2 words [SD=5.9]) or separately for high- or low-arousal words (
Figure 1, right panel) or medium-arousal words, excluding general enhancing effects of this drug on memory retrieval. A repeated-measures analysis of variance with treatment condition as within-subject factor and treatment group as between-subjects factor revealed a significant treatment condition by treatment group interaction for high-arousal words (p=0.004) but not for low- or medium-arousal words. In all three experimental groups, we found no evidence of cortisone or propranolol effects, or their combination, on the number of confabulations in the free-recall test or on attention or concentration as measured with the d2 cancellation test
(10) . Demographic parameters did not differ among the three treatment groups.
Discussion
We found that a centrally acting β-adrenoceptor antagonist selectively blocked the impairing effect of cortisone on the retrieval of emotionally arousing verbal material. Propranolol alone did not affect memory retrieval. These findings are consistent with previous evidence from animal experiments
(1,
3) and suggest that emotionally arousing words may be sufficiently arousing to activate the noradrenergic system during memory retrieval. In support of this view, it has been reported that the presentation of emotionally arousing words or pictures is capable of activating the amygdala during encoding to enhance memory consolidation and that a β-adrenoceptor antagonist blocks this activation
(11,
12) . In previous studies we found that although the retrieval-impairing effects of glucocorticoids involve the hippocampus
(4,
6), glucocorticoid effects on memory retrieval require concurrent emotional arousal-induced noradrenergic activation of the amygdala
(3) . Interestingly, and in support of the view that retrieval of emotionally arousing words may increase noradrenergic activity within the amygdala, human imaging studies have indicated that the degree of interaction between the amygdala and the hippocampus is greater during the retrieval of emotionally arousing information
(13,
14) . In an alternative scenario (although these two scenarios are not mutually exclusive), activation of the noradrenergic system may have resulted from an emotionally arousing test situation, as it has been reported that an emotionally nonarousing, familiarized test situation abolishes the impairing effects of glucocorticoids on memory retrieval
(9) .
The present findings may have important clinical implications, as β-adrenoceptor antagonists might prove useful for the prevention of glucocorticoid-induced memory deficits in acute stressful situations
(7), as well as for the treatment of memory deficits in conditions associated with chronically elevated glucocorticoid levels, such as depression and medical conditions requiring glucocorticoid treatment
(15) .