Effects of noradrenergic stimulation on memory in patients with major depressive disorder

Abstract

Major depressive disorder (MDD) has been associated with alterations in the noradrenergic system and impaired memory function. In turn, enhanced memory function has been associated with noradrenergic stimulation. In this study, we examined whether noradrenergic stimulation would differentially improve memory function in patients with MDD and healthy controls. In a placebo-controlled crossover study, 20 patients with MDD and 18 age- and sex-matched healthy controls received either placebo or 5 mg of yohimbine, an alpha-2-adrenoceptor antagonist that causes increased noradrenergic activity, orally before memory testing. A word list paradigm was used to test memory consolidation. Furthermore, the autobiographical memory test assessing memory retrieval and a working memory test were administered. Salivary alpha-amylase and blood pressure were measured. Yohimbine improved memory consolidation (word list learning) across groups (main effect of yohimbine: p = 0.05). This effect was more prominent in depressed patients compared with controls (post hoc t-test: MDD p = 0.01, controls p = 0.77). Memory retrieval (autobiographical memory specificity) and working memory were not affected by yohimbine. Across groups, yohimbine administration resulted in an increase in blood pressure and alpha-amylase. In sum, these results further support the hypothesis that noradrenergic stimulation enhances memory consolidation. The mechanism by which yohimbine leads to stronger memory consolidation in depressed patients compared with healthy controls remains to be elucidated.

Keywords::

• Major depressive disorder
• noradrenergic system
• alpha-2-adrenoceptor antagonist
• yohimbine
• declarative memory
• working memory
• childhood trauma

Introduction

Major depressive disorder (MDD) is associated with marked disturbances of stress regulation systems. Furthermore, memory disturbances are frequent in MDD. One of the most thoroughly investigated cognitive functions is hippocampal-based episodic/declarative memory. Cognitive performance in episodic memory tasks, including paragraph delayed recall, learning, and retrieval of word lists, is impaired in MDD patients (Veiel 1997; Chamberlain and Sahakian 2006; Porter et al. 2007). Additionally, autobiographical memory seems to be less specific in these patients (van Vreeswijk and De Wilde 2004). In healthy humans, there is compelling evidence that stress hormones, e.g. cortisol and norepinephrine (NE), influence cognitive processes (Wolf 2003, 2009; van Stegeren 2008). In major depression, most studies that investigate the association between stress hormones and memory focus on the hypothalamic–pituitary–adrenal (HPA) axis, while the locus coeruleus-noradrenergic (LC-NA) system is often neglected (Wingenfeld and Wolf 2011; Schlosser et al. 2011).

The physiological stress response is mediated by the HPA axis and the LC-NA system with cortisol and NE as their end products (Gold and Chrousos 2002). NE is a monoamine with widespread effects across the brain to regulate arousal and stress responses. Alterations of the noradrenergic system have been extensively investigated in MDD but have yielded inconclusive results (Gold and Chrousos 2002; Nemeroff 2002; Goddard et al. 2010). However, there is some evidence for an increased density of brain alpha-2-adrenoceptors in MDD (Meana and Garcia-Sevilla 1987; Meana et al. 1992; Calogero 1995; Garcia-Sevilla et al. 1999). The alpha-2 receptor contributes to the autoregulation of NE secretion, which is mediated by the LC. Blocking the receptor, e.g. using yohimbine, leads to an increased noradrenergic activity. Interestingly, in MDD elevated agonist binding to alpha-2-adrenoceptor density was found in the LC (Ordway et al. 2003).

The LC-NA system has important modulating effects on memory performance (Southwick et al. 2002; Roozendaal and McGaugh 2011). Preclinical studies have shown that moderate levels of NE enhance working memory, whereas higher levels have impairing effects (Arnsten and Li 2005). Furthermore, it was demonstrated that an alpha-2 antagonist enhances memory consolidation in chickens (Gibbs et al. 2010). In humans, stimulation of the noradrenergic system enhances and noradrenergic blockade reduces memory performance (O'Carroll et al. 1999; Chamberlain et al. 2006). Up to now, most studies have examined the effects of noradrenergic blockade by adrenergic agonists on memory consolidation and working memory (Chamberlain et al. 2006). The majority of those investigations have shown impaired working memory after noradrenergic blockade (Muller et al. 2005; Chamberlain et al. 2006; van Stegeren 2008). Importantly, the studies analyzing noradrenergic stimulation on memory have predominantly used acquisition and consolidation paradigms (O'Carroll et al. 1999; Cahill et al. 2003; van Stegeren 2008; Gibbs et al. 2010). It is also known that noradrenergic activity mediates cortisol effects on memory and learning, e.g. conditioning, by modulating the activation of the amygdala and the hippocampus especially in the context of emotional stimuli (Roozendaal et al. 2003, 2004, 2006a,b, 2008; de Quervain et al. 2007.

Less is known about noradrenergic stimulation on memory retrieval, and research on the effects of noradrenergic stimulation on memory in MDD patients is rare (Harmer et al. 2009). In one study, acute administration of 4 mg of reboxetine, a NE reuptake inhibitor, resulted in an improvement of perception of facial expression, i.e. happiness and surprise in patients and healthy controls (Harmer et al. 2009). Interestingly, a positive effect of reboxetine on memory for positive words was found only in depressed patients. However, consolidation and retrieval were not analyzed separately in this study protocol, and participants were not instructed to learn the words at the time of presentation. Furthermore, a cross-sectional design was used instead of a repeated measure design. Nevertheless, the study provided first evidence that depressed patients are more sensitive to noradrenergic stimulation.

The aim of the current study was (1) to investigate the effects of adrenergic stimulation using a low dose of yohimbine, an alpha-2-adrenoceptor antagonist, on memory by clearly separating consolidation and retrieval and (2) to compare patients with MDD with healthy control participants. We tested different memory domains, i.e. word list learning (consolidation), autobiographical memory specificity (retrieval), and working memory. We hypothesized that yohimbine would enhance memory performance. Because of earlier studies demonstrating an up-regulation of central alpha-2-adrenoceptors (Ordway et al. 2003), we expected this effect to be more pronounced in patients with MDD. To verify the efficacy of yohimbine treatment, salivary alpha amylase; heart rate; and blood pressure were assessed.

Due to the fact that childhood trauma is a risk factor for major depression (Heim et al. 2008) and contributes to differences within depressed patients in terms of physiological disturbances, e.g. higher heart rate in response to stress (Heim et al. 2000), we aimed to analyze whether childhood trauma influences the effects of yohimbine on memory in MDD patients.

Methods

Participants

Twenty patients with MDD and 18 age- and gender-matched healthy controls participated in our study. All patients were inpatients and recruited at the Department of Psychosomatic Medicine and Psychotherapy, University Medical Center Hamburg-Eppendorf & Schön Klinik Hamburg-Eilbek, Germany.

Participants were excluded if they had co-morbidities leading to cognitive impairments, e.g. psychotic symptoms, current or lifetime schizophrenia, alcohol or drug dependence, current anorexia, bipolar disorder, schizoaffective disorder, attention deficit hyperactivity disorder, as assessed by structured interview (see later for details). Further exclusion criteria were contraindications to yohimbine (cardiovascular, renal or hepatic diseases, glaucoma, extrapyramidal disorders, hyperexcitable states, metabolic disease, and pregnancy). Intake of antidepressants did not lead to exclusion, if it was constant in dosage for a week. Eight patients were medication free, 12 took antidepressive medication as described later in detail. Healthy participants were recruited by local advertisement and received financial remuneration for their efforts (100€). Written informed consent was obtained from all participants. The study was approved by the University of Muenster Ethics Committee and the Ethics Committee of the Medical Council of Hamburg and is in accordance with the Declaration of Helsinki.

Clinical assessment

To assess psychiatric diagnoses, participants were interviewed using the Structured Clinical Interview for DSM-IV Axis I (Wittchen et al. 1997). Diagnoses were ascertained by trained interviewers. Depressive mood was measured in all participants using Beck Depression Inventory (BDI) (Beck and Steer 1994). Furthermore, the German version of the Childhood Trauma Questionnaire (CTQ) (Bernstein et al. 2003; Wingenfeld et al. 2010) was used for assessing self-reported childhood maltreatment experiences across five dimensions: emotional, sexual, and physical abuse, as well as emotional and physical neglect. In order to provide a measure for general mental ability, two subtests (LPS 3 and LPS 4) of a well-established German intelligence test, the “Leistungsprüfsystem” (LPS) (Horn 1983) was administered. The LPS is partially based upon Thurstone's (1938) primary mental abilities and correlates highly with other popular psychometric intelligence tests (Horn 1983).

Memory tests

The word list paradigm consisted of 21 words. Participants were asked to memorize as many words as possible in no particular order. Word lists were taken from a study of Kuhlmann et al. (2005) and were reduced from 30 to 21 words (Terfehr et al. 2011a). The two word lists were comparable in emotionality, usage, and length. Each list consists of seven neutral, seven negative, and seven positive words. As in our previous paper, we analyzed the percentage of correctly recalled words relative to the words recalled after the fifth learning trial on the day before (Terfehr et al. 2011a).

A modified version of the autobiographical memory test (AMT; Williams and Broadbent 1986; Buss et al. 2004; Schlosser et al. 2010) was administered. After an initial practice on one cue word, the participants were instructed to write down a specific event from their past in response to six adjectives which were consecutively presented on cards. Participants were instructed to recall events that had happened at least 1 day prior to testing, which had taken place at a certain time and place, and did not last any longer than 1 day. They were also instructed to describe individuals and specific activities involved in the event. The specificity of the answers was evaluated by two trained raters, separately. An answer was considered specific when at least three of the following criteria were met: description of the location, time, and persons involved and activities carried out. Each specific answer was given a score of 1 and non-specific answers a score of 0 (see also Schlosser et al. 2010).

To test the verbal modality of working memory, we administered the self-developed word suppression test (WST) in the style of the Wechsler Memory Scale—Revised (Wechsler 1987; Terfehr et al. 2011b). The WST consisted of two test parts (one with negative and the other with neutral interference words), presented in randomized order. Each test part consisted of 14 recorded trials with a series of alternately presented digits and interference words, which were presented verbally. In the easiest trials, a series consisted of two digits and two words, while in the hardest trials, a series consisted of eight digits and eight words. Two series consisted of the same number of digits and words (e.g. series one and two consisted of two digits and two words, series three and four consisted of three digits and three words…). Digits and words were presented at a rate of one per second. The task was finished when a subject failed to reproduce two trials of the same length. There was no time limit, while the whole test did not take longer than 10 min. Analogous to the Wechsler Memory Scale—Revised (Wechsler 1987), the WST score represented a cumulative raw score, with one point given for each correctly reproduced sequence.

Physiological data

Following American Heart Association guidelines (Perloff et al. 1993), two blood pressure readings were taken at both arms and systolic blood pressure (SBP) and diastolic blood pressure (DBP) were computed as the mean of the readings.

Saliva samples for the assessment of alpha amylase were collected at the same time points. Saliva was collected using salivette collection devices (Sarstedt, Nümbrecht, Germany) and stored at room temperature until completion of the session. Participants were instructed to chew on the cotton roll for 1 min. Salivettes were then stored at − 70°C until biochemical analysis (Biochemical Laboratory, University of Dresden, Institute of Biopsychology, Prof. Dr. C. Kirschbaum). Salivette devices were centrifuged at 3000 rpm for 5 min, resulting in a clear supernatant of low viscosity. Alpha-amylase activity was measured by a quantitative enzyme kinetic method (see also Nater et al. 2007).

Procedure

The study protocol is presented in Table I. Blood pressure was measured at the beginning of the procedure, 1 h after drug administration and at the end of the session.

Table I.  Experimental protocol indicating time of cognitive testing.

Day	Time	Procedure^*
Day 1	1500	Saliva sampling, heart rate, blood pressure
		Word list learning and immediate recall^†
	1515	Administration of yohimbine or placebo
	1615	Saliva sampling, heart rate, blood pressure
		AMT^‡
		Working memory test
	1645	Saliva sampling, heart rate, blood pressure
Day 2	1615	Word list (delayed recall)^†

Notes: ^*One week later the study protocol was repeated using the alternate condition (yohimbine/placebo) and parallel versions of the memory tests (word list, AMT)

^† Test of memory consolidation

^‡ Test of memory retrieval.

A placebo-controlled, double-blind, crossover study was performed. On the first day, all participants learned the word list. About 15 min later, they received either 5 mg yohimbine (Spiegel®, Desma GmbH, Mainz-Kastel, Germany) or placebo orally, respectively. This low dose of yohimbine was chosen to avoid ceiling effect and to be able to detect subtle differences between MDD patients and controls concerning their sensitivity to yohimbine. One hour later, the AMT and the working memory test were performed (test of memory retrieval).

In the afternoon of the second day, participants were asked to recall the words from the word list they had learned the day before (free recall and test of memory consolidation).

Each participant was tested twice with parallel versions of a word list paradigm and an AMT. The two versions of these tests were counterbalanced across the two conditions. The same working memory task was presented two times. All tests took place in the afternoon starting at 15:00. The same procedure with the alternate test condition was repeated after 1 week.

Statistical analysis

Statistical analyses were performed using SPSS Version 18.0. Demographic data were analyzed using Pearson's χ²-test for categorical data and Student's t-test for continuous data. Blood pressure, heart rate, and salivary alpha-amylase were analyzed using a 2 × 3 × 2 [condition × time (three measurement points) × group (depression vs. controls)] analysis of variance (ANOVA) with repeated measurements of the factors condition (placebo vs. yohimbine) and time. Effects of yohimbine on memory performance were analyzed using a 2 × 2 (condition × group) ANOVA with repeated measurements of the factor condition. Multiple regression analysis was used to evaluate the impact of self-reported childhood trauma on the effects of yohimbine on memory consolidation. The differences between memory performance after yohimbine and placebo served as dependent variable.

Results

Demographic and clinical data

Demographic and clinical data as well as the related statistics are presented in Table II. Patients with MDD and healthy controls did not differ in body mass index (BMI), estimated intelligence (LPS), or years of school education. As expected, patients had significantly higher depression scores (BDI). Furthermore, patients with MDD reported higher scores of emotional neglect.

Table II.  Sociodemographic and clinical characteristics.

	MDD	Controls
Characteristics	(n = 20)	(n = 18)	Statistics
Age	35.1 (9.9)	35.3 (10.0)	tdf36 = − 0.06, p = 0.96
Gender (f/m)	16/4	15/3	χ^2 df1 = 0.07, p = 0.79
BMI	23.8 (4.4)	23.3 (4.0)	tdf36 = 0.39, p = 0.69
Years of school education	10.8 (1.5)	10.7 (1.3)	tdf36 = 0.16, p = 0.87
LPS 3 & 4	54.6 (9.4)	52.6 (8.7)	tdf36 = 0.66, p = 0.51
BDI sum^*	22.1 (9.7)	6.2 (7.5)	tdf35 = 5,48, p < 0.001
CTQ^†
Emotional abuse	11.4 (6.4)	9.2 (4.6)	tdf34 = 1.16, p = 0.25
Physical abuse	7.4 (5.4)	6.0 (1.6)	tdf34 = 1.04, p = 0.31
Sexual abuse	6.1 (4.6)	6.5 (4.3)	tdf34 = − 0.27, p = 0.78
Emotional neglect	14.9 (5.6)	9.7 (4.5)	tdf34 = 3.12, p = 0.004
Physical neglect	8.6 (2.2)	8.0 (3.9)	tdf34 = 0.56, p = 0.75
Total score	48.4 (19.2)	39.4 (15.9)	tdf34 = 1.53, p = 0.17

Notes: MDD, major depressive disorder; BMI, body mass index; LPS, “Leistungsprüfsystem” (intelligence test); BDI, Beck Depression Inventory

^* One control participant did not answer the BDI

^† One MDD patient and one control participant did not answer the CTQ.

MDD patients suffered from the following current comorbid diagnosis: anxiety disorder (N = 11) and bulimia nervosa (N = 2). The patients took the following psychiatric medications: selective serotonin reuptake inhibitors (N = 8), serotonin and noradrenaline reuptake inhibitors (N = 1), tricyclic antidepressant (N = 2), tetracyclic antidepressant (N = 1), melatonergic and serotonergic antidepressants (N = 1), and antipsychotics (N = 1). Two of the patients took two different drugs and eight were medication free.

Physiological data

Blood pressure and heart rate: Yohimbine increased SBP (main factor condition: F_1,36 = 11.41, p = 0.02; condition by time interaction effect: F_2,72 = 3.92, p = 0.02). The main effect of group and all other interaction effects were not significant. A similar pattern was seen for DBP (main factor condition: F_1,36 = 13.15, p = 0.001; main factor time: F_2,72 = 16.19, p < 0.001; condition by time interaction effect: F_2,72 = 3. 29, p = 0.04). Again depressed patients and controls did not differ from each other. On the contrary, heart rate was not affected by yohimbine. There was only a significant time effect (F_2,72 = 49.96, p < 0.001), reflecting a decrease over the test session in both conditions. Results are presented in Figure 1.

Figure 1.  SBP, DBP, and heart rate (mean [SE]) before and after administration of yohimbine and placebo in patients with MDD and healthy controls. Yohimbine leads to an increase in blood pressure but did not affect heart rate.

Alpha-amylase (see Figure 2): for alpha-amylase there was only a trend for a condition by time interaction effect (F_2,66 = 2.66, p = 0.08). The main effects of condition (F_1,33 = 1.87, p = 0.18), time (F_2,66 = 2.23, p = 0.16), and group (F_1,33 = 0.02, p = 0.89) were not significant, and neither were any other interaction effects.

Figure 2.  Salivary alpha amylase (mean [SE]) before and after administration of yohimbine and placebo in patients with MDD and healthy controls.

Memory consolidation—Word list paradigm

Conducting a 2 × 2 ANOVA with repeated measures with the main factors group (MDD patients vs. healthy controls) and condition (placebo vs. yohimbine), we found a significant effect of condition (F_1,36 = 4.11, p = 0.05), whereas the group by condition interaction effect did not reach significance (F_1,36 = 2.51, p = 0.12). There was no main effect of group (F_1,36 = 0.58, p = 0.45). The results (mean values) are presented in Figure 3 showing a better memory performance in the yohimbine condition.

Figure 3.  Percentage of words retrieved in the word list paradigm in relation to the last learning list on the previous day (mean [SE]) in patients with MDD and healthy controls after placebo and after administration of 5 mg yohimbine There was a significant effect of drug condition (p = 0.05).

For explorative purposes, we conducted separate t-tests for each group. We found a significant difference between the yohimbine condition and placebo in the MDD group (t_df19 = − 2.75, p = 0.01) but not in the control group (t_df17 = 0.29, p = 0.77). Furthermore, there was a slight, but not significant, difference between patients and controls in the placebo condition (t_df36 = − 1.157, p = 0.09) but not in the yohimbine condition (t_df36 = 0.499, p = 0.62). As shown in Figure 3 in the placebo condition, the percentage of recalled words was reduced in the patients remembered compared with the controls.

In the next step, we calculated the effect sizes for the effect of yohimbine on memory consolidation for the whole group, the controls, and patients, separately. Analyses were carried out using the program G*Power 3 (Faul et al. 2007 2009). The following effect sizes were calculated: whole group d = 0.38, controls only d = 0.08, and patients only d = 0.63.

We also compared patients with (N = 12) and patients without (N = 8) intake of antidepressant medication. Repeated measures ANOVA revealed a significant effect of condition, i.e. a better memory performance in the yohimbine condition (p = 0.014), but no group by condition interaction effect (p = 0.70) or main effect group (p = 0.36) showing no influence of medication intake on the results. Furthermore, we compared patients with and without comorbid anxiety disorder and did not find differences between these groups (p = 0.22).

We also analyzed whether word valence influenced the results. Therefore, a 2 × 2 × 3 ANOVA with repeated measures with the main factors group (MDD patients vs. healthy controls) condition (placebo vs. yohimbine) and word valence (neutral, negative, and positive) was performed. A significant main effect of condition (F_1,36 = 6.377, p = 0.016) could be revealed while there was only a trend for the main effect word valence (F_2,71 = 2.72, p = 0.091). There was no significant group by valence interaction effect (p = 0.13) or treatment by valence interaction effect (p = 0.15).

Order effects: there was no difference between the first test session and the second test session (F_1,37 = 0.356, p = 0.55) or between the two test versions (F_1,36 = 0.004, p = 0.95).

Memory retrieval – AMT

Again, a 2 × 2 ANOVA with repeated measures was conducted with the main factors group and condition. Results (mean values/standard deviations) are presented in Table III. There was no main effect of condition (F_1,36 = 0.18, p = 0.68), no group by condition interaction effect (F_1,36 = 0.05, p = 0.82), and no main effect of group (F_1,36 = 0.96, p = 0.33).

Table III.  Results of the AMT and the working memory test (mean/SD).

	MDD	Controls
	(n = 20)	(n = 18)
AMT—Specificity
Placebo	3.9 (1.7)	3.5 (1.7)
Yohimbine	3.8 (1.6)	3.3 (1.6)
Working memory—neutral
Placebo	8.1 (1.3)	7.8 (2.3)
Yohimbine	8.3 (2.1)	7.9 (2.5)
Working memory—negative
Placebo	8.5 (1.8)	7.6 (2.4)
Yohimbine	8.3 (1.8)	7.7 (2.8)

Notes: MDD, major depressive disorder; AMT, autobiographical memory test.

Working memory

Two 2 × 2 ANOVA with repeated measures were conducted: one analysis for the neutral test part and the other for the negative test part. Mean values and standard deviations are given in Table III. For the neutral test part, there was no main effect of condition (F_1,36 = 0.25, p = 0.62), no group by condition interaction effect (F_1,36 = 0.01, p = 0.932), and no main effect of group (F_1,36 = 0.29, p = 0.59). The same pattern emerged for the negative test part: No main effect of condition (F_1,36 = 0.05, p = 0.83), no group by condition interaction effect (F_1,36 = 0.515, p = 0.48), and no main effect of group (F_1,36 = 1.22, p = 0.28) were revealed.

Association between self-reported childhood trauma and memory consolidation

To analyze whether self-reported childhood trauma contributed to the results, an explorative regression analysis within the group of MDD patients was performed. In the first step, all subscales of the CTQ were included into the regression model. A significant impact of sexual abuse, emotional neglect, and physical neglect on the effects of yohimbine on memory consolidation was revealed (see Table IV). Overall, the model explained 49% of the variance. For explorative purposes, in the second and third steps we also included depression (BDI score) and sociodemographic variables (age, years of education, and estimated intelligence) into the model, which did not change the results.

Table IV.  Multiple regression analyses predicting the effects of yohimbine on memory consolidation in patients with MDDs (N = 19).

Predictor variables	Statistics^*
Step 1	Model
	Fdf5,13 = 4.571
	corr. R ^2 = 0.499
	p = 0.013
CTQ	β	T	p
Emotional abuse	0.233	0.623	0.54
Physical abuse	− 0.454	− 1.444	0.17
Sexual abuse	0.665	3.232	0.007
Emotional neglect	− 0.932	− 3.073	0.009
Physical neglect	0.640	2.948	0.01
Step 2	Model
	Fdf6,13 = 3.837
	corr. R ^2 = 0.489
	p = 0.023
CTQ	β	T	p
Emotional abuse	0.249	0.58	0.52
Physical abuse	− 0.410	1.271	0.29
Sexual abuse	0.561	2.310	0.04
Emotional neglect	− 0.942	− 3.066	0.01
Physical neglect	0.613	2.762	0.02
BDI	0.168	0.836	0.42
Step 3	Model
	Fdf9,9 = 3.340
	corr. R ^2 = 0.540
	p = 0.043
CTQ	β	T	p
Emotional abuse	0.245	0.67	0.52
Physical abuse	− 0.338	− 1.044	0.32
Sexual abuse	0.636	2.709	0.02
Emotional neglect	− 1.061	− 3.370	0.008
Physical neglect	0.646	2.944	0.02
BDI	0.038	0.188	0.85
Age	− 0.271	− 1.438	0.18
Years of education	− 0.240	− 1.155	0.28
LPS 3 & 4	0.151	0.857	0.41

Notes: ^*Dependent variable: memory consolidation after yohimbine minus memory consolidation after placebo.

Using the significant predictor variables, i.e. sexual abuse, emotional neglect, and physical neglect as covariates, we reran our initial ANOVA analyzing memory consolidation. A 2 × 2 ANCOVA with repeated measures was performed with the main factor group (MDD patients vs. healthy controls), the main factor condition (placebo vs. yohimbine), and the covariates such as sexual abuse, emotional neglect, and physical neglect. We found a treatment by group interaction effect (F_1,31 = 5.855, p = 0.02). The treatment by emotional neglect interaction effect (F_1,31 = 3.807, p = 0.06) as well as the treatment by sexual abuse interaction effect failed to reach statistical significance (F_1,31 = 2.356, p = 0.13). All other effects were not significant (all p>0.37). MDD patients showed an improvement in memory consolidation (N = 19, placebo: 47.3%, yohimbine 61.2%). This was not seen in the control group (N = 17, placebo: 54.9%, yohimbine 56.3%). Post hoc ANCOVAs were performed for each group separately. In the control group, there was no significant effect of the covariates alone or any drug by covariate interaction effect. In the MDD patients, the following significant interaction effects could be revealed: drug by sexual abuse (F_1,15 = 8.012, p = 0.013), drug by emotional neglect (F_1,15 = 16.736, p = 0.001), and drug by physical neglect (F_1,15 = 6.745, p = 0.02).

Discussion

This study was designed to compare the effects of noradrenergic stimulation on memory consolidation, memory retrieval, and working memory in depressed patients with the effects observed in healthy control participants. First of all, the administration of 5 mg yohimbine led to an increase in blood pressure, verifying the efficacy of our experimental manipulation. We found a significant improvement of memory consolidation (word list learning) after noradrenergic stimulation across groups. The enhancing effects of noradrenergic stimulation on memory consolidation were more pronounced in patients with MDD compared with healthy controls. Memory retrieval and working memory were not affected by a low dose of yohimbine. Self-reported childhood trauma seems to contribute to the effects of yohimbine on memory consolidation in depressed patients.

Our results in the word list paradigm, i.e. improvement of memory consolidation after yohimbine, are in accordance with several studies finding that increased NE levels improve memory performance, which in turn was impaired after blocking noradrenergic receptors using a beta-adrenoreceptor antagonist (van Stegeren 2008). However, so far the majority of investigations have applied designs with noradrenergic blockade instead of stimulation of the NE system. Our results concerning memory consolidation are in line with the only prior study documenting improvements in memory consolidation following noradrenergic stimulation (O'Carroll et al. 1999). It is remarkable that we found the same effects using a very low dose of yohimbine. Most studies used a higher dosage of yohimbine than that used in this study (Chamberlain et al. 2006). In one study, 0.4 mg per kilogram body weight of yohimbine was used, resulting in higher increases in blood pressure and alpha-amylase relative to the current report (Ehlert et al. 2006).

The main aim of our study was to compare healthy controls with patients suffering from MDD. To be able to detect subtle differences in the sensitivity of brain alpha-2-adrenoceptors in patients versus controls, we decided to use a very low dose to avoid ceiling effects. Indeed, our results suggest that the noradrenergic effects on memory consolidation are more pronounced in the MDD group. This is in line with the study by Harmer and colleagues, who reported a positive effect of acute reboxetine administration on memory for positive words in depressed patients (Harmer et al. 2009). Both studies together further support the hypothesis of an enhanced sensitivity of central alpha-2-adrenoceptor in MDD. This is in line with the finding of increased density of brain alpha-2-adrenoceptor and an elevated agonist binding to alpha-2-adrenoceptors in the LC in depressed patients (Meana and Garcia-Sevilla 1987; Meana et al. 1992; Calogero 1995; Garcia-Sevilla et al. 1999; Ordway et al. 2003).

Although noradrenergic blockade is known to impair working memory, only few studies so far investigated the effects of noradrenergic stimulation on working memory, mostly showing no effect (Chamberlain et al. 2006). Our data did not reveal any evidence for the influence of yohimbine administration on working memory. This is in contrast to one earlier study, which also used yohimbine (20–40 mg) but which suffered from a small sample size and practice effects (Swann et al. 2005). The authors of that study did not find an effect on task performance per se but revealed more commission errors after yohimbine, which seemed to be dose dependent with higher doses being more effective. Accordingly, in our study the very low dose of yohimbine might have made it difficult to find an effect. Thus, more studies are needed to clarify whether there is an effect of noradrenergic stimulation on working memory.

Furthermore, we investigated the effect of noradrenergic activation on autobiographical memory retrieval. As in the working memory test, no effect of yohimbine administration was found. Again, one might argue that the noradrenergic effects on memory retrieval may require a higher dose. Future studies should investigate this topic using higher drug dosages.

Apart from the LC-NA system, hormones of the HPA axis, i.e. cortisol, have profound and well-investigated effects on memory (Wolf 2003). In patients with MDD, there are also well-documented dysregulations of the HPA axis, e.g. glucocorticoid receptor (GR) dysfunction (Hinkelmann et al. 2009; Wingenfeld and Wolf 2011). In a series of studies, we investigated the effects of acute hydrocortisone administration which had no effects on memory in MDD patients, while in healthy controls we found a cortisol-induced memory impairment (Schlosser et al. 2010; Terfehr et al. 2011a,b). This effect was seen in all the tests that we used in this study, i.e. word list learning, autobiographical memory specificity, and working memory. We interpreted these results in the context of reduced central GR functioning, which has been discussed extensively in MDD research (Holsboer 2000; Pariante et al. 2004). Overall, to test cognitive performance after drug stimulation seems to be a promising tool to indirectly investigate central receptor function in MDD patients. As our former studies provided evidence for a reduced GR functioning, the current data further support the hypothesis of an enhanced alpha-2 adrenoceptor sensitivity in these patients.

Childhood trauma is a potent risk factor for psychiatric disorders including MDDs and there is evidence for effects of traumatic experiences on the stress regulation systems (Heim et al. 2000; Orr et al. 2002). It has been shown that humans reporting childhood trauma have a stronger catecholamine response to stress (Otte et al. 2005). Interestingly, in depressed patients a higher autonomic response to stress was only seen in those patients who also suffered from childhood trauma (Heim et al. 2000). This study reveals evidence that childhood trauma effects the impact of adrenergic stimulation on memory consolidation in patients with MDD. Although sexual abuse, emotional neglect, and physical neglect were found to be significant predictors in the regression model, self-reported depression measured with the BDI showed no significance, when simultaneously included in the regression. Thus, self-reported childhood trauma might have stronger influences on the effects of adrenergic stimulation on memory relative to self-reported depression. This is in line with other studies that emphasize that childhood trauma is an important mediator of alterations of physiological stress systems in MDD (Heim et al. 2008).

It has to be mentioned that our MDD patients did not differ significantly in task performance from the controls, which is a major difference to the reported cognitive difficulties in the literature. There was only a slight but not significant difference in the memory consolidation task, i.e. worse performance in the word list learning in the placebo condition in patients compared with controls. There was no evidence for differences between the two groups in working memory or autobiographical memory retrieval. One of the best investigated cognitive functions in depression is the hippocampal-based episodic declarative memory including autobiographical memory, suggesting an impairment in MDD patients (Brand et al. 1992; Burt et al. 1995; Veiel 1997; Williams et al. 2007), although not all studies agree (Purcell et al. 1997; Gualtieri et al. 2006; Alhaj et al. 2007). One might suggest that we investigated only a mild to medium disturbed sample, but all patients were diagnosed with SCID interview and BDI scores are in a range comparable with former studies of our group, in which we saw memory impairments in MDD patients compared with controls (Terfehr et al. 2011a,b).

There are some limitations of the study to mention: first, the sample size is relatively small leading to some statistical limitations, especially with regard to the regression analysis. Thus, the results from the regression analyses are not generalizable outside the sample. Furthermore, the fact that the group by treatment interaction just missed significance is somewhat surprising given the large differences we found in the exploratory analyses. Probably, this is due to a lack of power, and therefore, the results should be replicated in a larger sample. Interestingly, when statistically controlling for self-reported childhood trauma, the group by condition interaction effect became significant. Future studies should further investigate this topic. Another point is that many patients in our study were on antidepressant medication. Therefore, it would be interesting to test the effects of noradrenergic stimulation on memory performance in a sample of medication free MDD patients in the future. However, controlling for medication intake did not alter the results, i.e. the effect of yohimbine on memory was the same in patients with and without antidepressive medication. Though the subsamples of patients with and without medication are small and therefore the statistical power is weak, our results need replication. It would also be of interest to have some basal measurement of noradrenergic activity. Furthermore, it would be useful to compare different doses of yohimbine to investigate dose–response effects. Beside alpha-adrenergic receptors, beta-adrenergic receptors play a well-established role in memory processes. Furthermore, it would be of interest to disentangle pre- and post-synaptic function of alpha-adrenergic receptor system, e.g. by also stimulating alpha-1 adrenoceptors. However, this is one of the first studies in which the effects of adrenergic stimulation on memory are investigated in MDD; thus, the results have to be interpreted in terms of pilot data, due to the preliminary nature of the study.

To conclude, our results further support the enhancing effects of adrenergic stimulation on memory consolidation but not on memory retrieval. Furthermore, there is some evidence for an enhanced alpha-2 adrenoceptor sensitivity in MDD patients, possibly resulting in an enhanced drug responsiveness.

Declaration of interest This study was supported by the Werner-Otto-Stiftung, Hamburg. Dr Otte is on the speaker's board of Astra Zeneca, Lundbeck, and Servier. The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.