E. Huertas, Facultad de Psicología, Campus de Somosaguas, 28223 Madrid, Spain. E-mail: email@example.com
Genetic variants that are related to the dopaminergic system have been frequently found to be associated with various neurological and mental disorders. Here, we studied the relationships between some of these genetic variants and some cognitive and psychophysiological processes that are implicated in such disorders. Two single nucleotide polymorphisms were chosen: one in the dopamine D2 receptor gene (rs6277-C957T) and one in the catechol-O-methyltransferase gene (rs4680-Val158Met), which is involved in the metabolic degradation of dopamine. The performance of participants on two long-term memory tasks was assessed: free recall (declarative memory) and mirror drawing (procedural motor learning). Heart rate (HR) was also monitored during the initial trials of the mirror-drawing task, which is considered to be a laboratory middle-stress generator (moderate stress), and during a rest period (low stress). Data were collected from 213 healthy Caucasian university students. The C957T C homozygous participants showed more rapid learning than the T allele carriers in the procedural motor learning task and smaller differences in HR between the moderate- and the low-stress conditions. These results provide useful information regarding phenotypic variance in both healthy individuals and patients.
Polymorphisms that are related to dopamine, and more specifically, those that are linked to the dopamine D2 receptor gene (DRD2) and the catechol-O-methyltransferase (COMT) gene are frequently associated with various neurological and mental disorders characterized by alterations in cognitive and emotional processes. One of the most clinically relevant single nucleotide polymorphisms (SNPs) of the DRD2 gene is C957T (rs6277), located at the Pro319 codon in the exon 7. The C allele of the C957T SNP has been associated with low striatal DRD2 availability (Hirvonen et al. 2004, 2005) and with high DRD2 binding potential throughout the cortex and thalamus (Hirvonen et al. 2009b).
The Val158Met polymorphism (rs4680), which is located in exon 3 of the COMT gene, affects the activity of the COMT enzyme, which degrades dopamine and other catecholamines (Lachman et al. 1996). The Met variant shows lower dopamine-degrading activity relative to the Val variant, resulting in higher dopamine levels, particularly within the prefrontal cortex and hippocampus (Chen et al. 2004; Dennis et al. 2010; Honea et al. 2009).
The C957T and Val158Met SNPs have also been linked to stress responsiveness. For example, CC genotype carriers of the C957T SNP showed more persistent high-amplitude skin-conductance responses in threatening situations (Huertas et al. 2010). Moreover, an examination of genetically altered mice showed that COMT reduction results in exaggerated stress reactivity (Papaleo et al. 2008). The low-activity COMT variant has also been linked to greater stress sensitivity in humans (van Winkel et al. 2008).
Therefore, the purpose of this study was to investigate whether the C957T and Val158Met SNPs are functionally related to either procedural memory (mirror-drawing learning) or declarative memory (verbal free recall). We also aimed to study possible links between these SNPs and autonomic reactivity [heart rate (HR)] in a moderate-stress situation, such as mirror-drawing learning.
Material and methods
Two hundred and thirteen self-reported Caucasian undergraduate students (149 females) volunteered at the Universidad Complutense de Madrid, Spain. Their ages ranged from 18 to 34 years (mean = 21.12; SD = 4.09). Table 1 shows the gender distribution per genotype. This study was approved by the Ethics Committee of the Facultad de Psicologia, Universidad Complutense de Madrid. The participants signed two informed consent forms, one for the experimental procedure and one for the genetic study.
Table 1. Genotype and gender distributions of the participants
Apparatus and stimuli
An IBM-compatible personal computer and a 15-inch monitor were used for the mirror-drawing task. The software was developed by the Technical Service of the Facultad de Psicologia, Universidad Complutense de Madrid. Subjects used a computer mouse to follow the black outline (11 pixels wide) of a four-point star that was displayed on the computer screen. The star was formed by eight segments that measured 140 pixels each with two points that were vertically aligned and two that were horizontal. The pointer appeared on the screen as a red square that measured five pixels long on each side. Vertical movements of the mouse caused the pointer on the screen to move in the same direction, but horizontal movements caused the mouse to move in the reverse direction, i.e. when the mouse was moved to the left, the pointer moved to the right and vice versa. A green eight-pixel square indicating the starting and termination points was located at the top of the upper vertex of the star. When the pointer passed the green square for the first time, it disappeared and the data recording began. When the pointer passed it for the second time, the recording ceased and the image on the screen disappeared. The horizontal and vertical coordinates of each pixel that was passed over during the task were stored in a data file, and the number of pixels that were traced within the outline of the star, completion time, pixel/second ratio and number of errors (oversteps beyond the outline of the star) were stored in another file.
For the free recall task, a list was used that contained 26 words (two-to-three syllable) with lexical frequencies of between 10 and 14 per million that were selected from the Affective Assessment of 478 Spanish Words (Redondo et al. 2005). Nine of the words were unpleasant (valence: 1.2–1.9 on a scale of 1–9), nine were pleasant (valence: 6.9–8.0) and eight were neutral (valence: 3.8–4.9).
HRs were measured with a Persona O2 Model-11 Portable Pulse Oximeter (Fukuda Sangyo Co., Ltd., Chiba, Japan) and a reusable finger probe that was placed on the index finger of the non-dominant hand. The system achieved ±2% accuracy at 100 beats per minute.
When the participants arrived for their individual laboratory session, the HR sensor was attached to them and they were asked to read and fill out the informed consent for the experimental protocol, the content of which had been explained to them previously. After that, the participants were asked to relax, and there was a 2-min period of inactivity. During this period (when the HR decreased), the minimum HR value was registered. Next, two consecutive mirror-drawing trials were conducted. The participants were instructed to move the pointer along the figure in a clockwise direction as quickly as possible, trying not to go out of the black outline. No time limit was set. Maximum HR values were monitored during the first 20 seconds of each trial. The HR sensor was then detached and the word list for the free recall task was given to the participants, who were asked to read the list aloud twice and told that they would be asked about the words later. Subsequently, they executed two additional mirror-drawing trials and were asked to write down, in any order, the words that they remembered from the free recall list. Finally, two more mirror-drawing trials were performed, the participants filled out the informed consent for the genetic study and saliva samples were collected.
DNA from saliva was collected using the Oragene DNA Self-Collection Kit (DNA Genotek, Ottawa, Ontario, Canada) and purified from 500-µl aliquots using the ethanol precipitation protocol as described by the manufacturer. Purified DNA was dissolved in 100-µl of TE buffer [10 mM Tris-HCl, 1 mM ethylenediaminetetraacetic acid (EDTA), pH 8.0] and stored at −20°C.
For increased accuracy, DNA samples from 133 subjects were genotyped using the TaqMan assay and 80 were genotyped using direct sequencing. The genotyping of 12 samples was repeated using both methods, the results of which were fully consistent. TaqMan genotyping was performed using pre-designed and validated TaqMan SNP genotyping assays for humans from Applied Biosystems (Foster City, CA 94404, USA): C957T (rs6277, C_11339240_10) and Val158Met (A158G, rs4680, C_25746809_50). These genotyping assays were performed using a LightCycler 480-II machine (Roche) with the Endpoint Genotyping method. Colour fluorescence measurements after amplification were analysed by the LightCycler® 480 Endpoint Genotyping Software version 1.5.0. Direct sequencing was performed by Macrogen (Seoul, Korea). The Primer3 software was used to design primers to amplify the two adjacent regions around the two SNPs of interest with an extension of 60–70 base pairs (C957T: left, CACCACCAGCTGACTCTCC, right, CAATCTTGGGGTGGTCTTTG; Val158Met, left, ATCGAGATCAACCCCGACT, right CCCTTTTTCCAGGTCTGACA). These custom primers were synthesized and purchased from Sigma-Genosys (Oakville, Ontario, Canada). The polymerase chain reaction optimization, amplification, product purification and DNA sequencing were performed by Macrogen. The BioEdit Sequence Alignment software was used for the SNP analyses. The genotype and allele distributions of the genetic polymorphisms from the samples are shown in Table 1.
For the mirror-drawing task, the score of each participant in each trial was the total number of pixels that were passed through divided by the time (in seconds) that was required to complete the task. Four participants were excluded from the analysis of this variable. Two of them did not follow the black outline of the star in the two first trials but rather traced the inner contour of the star; one participant executed the two first trials with the right hand and the rest with the left and another participant did not carry out the two last trials. The data were grouped into three blocks, each consisting of two consecutive trials. The measure of performance on the free recall task was the total number of recalled words. One participant was excluded from the analysis of this variable as an outlier (>3 SD from the mean). The mean of the last two HR measurements (taken during the two initial mirror-drawing trials) was used as the HR score under conditions of moderate stress.
Both the mirror-drawing and HR data were analysed using mixed-design analyses of variance (anovas) by means of the Statistical Package for the Social Sciences (SPSS) of International Business Machines (IBM) (IBM SPSS Statistics 19 for Windows, SPSS Inc., Chicago, IL, USA) with an anova for each dependent variable and each SNP. For the mirror-drawing analysis, we used a 3 × 2 design with trial block as a within-subjects factor and genotype as a between-subjects factor. For the HR analysis, we used a 2 × 2 design with stress level as a within-subjects factor and genotype as a between-subjects factor. In both cases, the hypothesized association between the SNP and the dependent variable was indicated by the interaction effect. Because the mean scores in the mirror-drawing task were higher for men than women (F1,207 = 15.403; P < 0.001) and the average HR values were higher for women than men (F1,211 = 6.449; P = 0.012), gender was added to the model as a between-subjects factor in both cases to control for the gender effect. Significant interactions were examined by pairwise post hoc comparisons using univariate or repeated-measures anovas. A Greenhouse–Geisser correction was used when the Mauchly's sphericity test showed significant results.
A one-way anova was used to analyse the free recall data, with the genotype as a between-subjects factor and the total number of recalled words as dependent variable. The hypothesized association between the corresponding SNP and recall was indicated by the genotype effect.
A Bonferroni correction was used for multiple testing. Because there were six hypothesized associations (two SNPs × three dependent variables), the significance level was set at 0.008. Given the unequal size of some of the genetically defined groups (Table 1), additional Mann–Whitney tests were used for comparisons between genotypes. In this case, the dependent variables were the difference between the first and last trial blocks for the mirror-drawing task, the difference between the low-stress level and the moderate-stress level for the HR and the number of words recalled for the recall task. In these analyses, the same comparisons as in the parametric analyses were significant.
The genotype frequencies of the C957T SNP (Table 1) were in Hardy–Weinberg equilibrium (χ2 = 0.189, df = 1, P > 0.65) and were similar to those of other samples of the Spanish healthy population (e.g. Rodriguez-Jimenez et al. 2006). In contrast, the genotype frequencies of the Val158Met SNP differed from the Hardy–Weinberg equilibrium (χ2 = 7.879, df = 1, P < 0.01). This disequilibrium of the Val158Met SNP could be related to the characteristics of the sample. All the participants were undergraduate students and the Val158Met SNP has been associated with neural activity during several cognitive tasks (e.g. Bishop et al. 2008). The gender distribution (Table 1) did not differ significantly between genotypes for the C957T SNP (χ2 = 1.332, df = 2, P = 0.514) or for the Val158Met SNP (χ2 = 3.166, df = 2, P = 0.205).
The anova concerning the two SNPs showed a significant main effect for trial block both in the case of the C957T SNP (F1.8,368 = 195.25; P < 0.001) and in the case of the Val158Met SNP (F1.8,366 = 229.97; P < 0.001), indicating that learning took place during the training.
For the DRD2 C957T SNP (Fig. 1a), the trial block × genotype (CC, CT + TT) interaction was significant (F1.8,368 = 7.46; P = 0.001), showing differences in learning between genotypes. When this interaction effect was parsed into orthogonal linear and quadratic contrasts, only the linear trial block × genotype interaction was significant (F1,205 = 11.074; P = 0.001; partial η2 = 0.053).
With regard to this trial block × genotype interaction, pairwise post hoc analyses showed that the difference between the last and first trial block was significant in both the CC genotype carriers (dif. = 12.94; F1,31 = 88.340; P < 0.001) and the T allele carriers (dif. = 9.34; F1,174 = 436.723; P < 0.001), showing that both groups had learned. The separate analyses of each trial block did not show significant difference between CC carriers and T carriers in the first block (dif. = 0.01; F1,205 = 0.001; P = 0.99), indicating that the performance at the beginning of the learning process was similar for both. However, that difference was significant in the last trial block (dif. = 3.59; F1,205 = 8.497; P = 0.004). In short, both CC carriers and T carriers learned, but CC carriers learned more. The genotype main effect was not significant.
The trial block × genotype × gender interaction was not significant (F1.8,368 = 0.34; P = 0.688). This result justifies an analysis of the data pooled across gender with respect to the trial block × genotype interaction, which was the focus of interest.
For the COMT Val158Met SNP (Fig. 1b), the trial block × genotype (Met/Met, Val/Met + Val/Val) interaction was not significant (F1.8,366 = 3.62; P = 0.032). Therefore, it was not possible to conclude that there were significant learning differences between genotypes. Neither the main effect genotype nor the trial block × genotype × gender interaction was significant.
There were no significant differences in free recall between genotypes for the C957T SNP (M: 5.56 for CC, 5.66 for CT + TT; dif. = 0.10; F1,210 = 0.04; P = 0.839) or the COMT Val158Met SNP (M: 5.49 for Met/Met, 5.70 for Val/Met + Val/Val; dif. = 0.21; F1,210 = 0.29; P = 0.591).
The anovas concerning the two SNPs showed a significant main effect for stress both in the case of the C957T SNP (F1,209 = 120.31; P < 0.001) and in the case of the Val158Met SNP (F1,209 = 276.02; P < 0.001), indicating that HR was higher under the moderate-stress condition than under the low-stress condition.
For the C957T SNP (Fig. 2a), the stress level (low, moderate) × genotype (CC, CT + TT) interaction was significant (F1,209 = 8.31; P = 0.004; partial η2 = 0.040). Pairwise post hoc analyses exploring this interaction showed a significant difference between the two stress levels for both the CC (F1,32 = 34.38; P < 0.001) and T carriers (F1,177 = 272.829; P < 0.001). Separate analyses of each stress level did not show any significant difference between genotypes under either the low-stress condition (F1,209 = 1.756; P = 0.187) or the moderate-stress condition (F1,209 = 0.004; P = 0.947). In short, the HR was higher under the moderate-stress condition than under the low-stress condition in both CC carriers and T carriers, but the difference between conditions was smaller in the CC carriers, as indicated by the stress level × genotype interaction. The main effect genotype was not significant.
The stress level × genotype × gender interaction was not significant (F1,209 = 0.80; P = 0.373). This result justifies an analysis of the data pooled across gender with respect to the stress level × genotype interaction, which was the focus of interest.
For the COMT Val158Met SNP (Fig. 2b), the stress level (low, moderate) × genotype (Met/Met, Val/Met + Val/Val) interaction was not significant (F1,209 = 1.42; P = 0.235). Neither the main effect genotype (F1,209 = 3.47; P = 0.064) nor the stress level × genotype × gender interaction was significant.
Our results show an association between the C957T SNP and the acquisition of a visual–motor skill (the mirror-drawing task), which is considered prototypical of procedural learning. The C homozygous participants showed a similar performance to T allele carriers at the beginning of the training, but they learnt more than the T carriers during the training.
This association between the C957T SNP and the mirror-drawing learning does not seem to be the result of differences in controlled processing (Shiffrin & Schneider 1977), which significantly involves the prefrontal cortex. Skill learning requires, in effect, controlled processing in its initial stage (Anderson 1982). There is an inverted-U relationship between dopamine signalling in the prefrontal cortex and a range of controlled cognitive abilities (see Bolton et al. 2010), and the C allele of the C957T SNP has been associated with high DRD2 binding potential throughout the cortex (Hirvonen et al. 2009b). However, our results show no differences in performance between the CC carriers and the T carriers in the first two learning trials, precisely those that usually require more controlled processing. Thus, our results are not consistent with this explanation.
The differences between genotypes obtained in our results also do not seem to be due to fine motor ability (Hermann et al. 2002) or to other variables that affect performance. The fact that there is no difference between genotypes in the first trial block indicates that the difference observed in the third block does not reflect pre-experimental performance differences but rather a genuine difference in learning.
The C allele of the C957T SNP has also been associated with low striatal DRD2 availability (Hirvonen et al. 2004, 2005), and the striatum plays a fundamental role in procedural learning and, therefore, in skill acquisition (see D’Amours et al. 2011; Seger & Spiering 2011). For instance, Parkinson's disease patients, who have depleted dopamine levels, exhibit less effective skill learning (e.g. Harrington et al. 1990; Heindel et al. 1989; see also Pendt et al. 2011). Therefore, relatively poorer procedural learning would be expected of C homozygous individuals. However, in this study, the C homozygous participants were the most efficient in completing the task. The pattern that is exhibited by CC carriers is, therefore, opposite to that of Parkinson's patients. A similar disparity seems to occur in other tasks. For example, while the Parkinson's patients show enhanced learning to avoid maladaptive choices, CC carriers of the C957T SNP show the poorest performance (Frank et al. 2004, 2007, 2009). Similarly, while Parkinson's patients show impaired suppression of conflicting motor responses in the Simon task (e.g. Wylie et al. 2010), C carriers are the most efficient in inhibiting motor responses to a stop signal, as assessed by the stop-signal procedure (Colzato et al. 2010).
There are several possible explanations for our results. First, the reduced striatal D2 receptor density associated with the CC genotype may be compensated for by a higher affinity. The striatal DRD2 receptors of C homozygous individuals, due to their affinity, may result in optimized G-protein-coupled signalling and in more efficient learning (Hirvonen et al. 2005, 2009a,b; see also Frank & Hutchison 2009).
A second possible explanation is related to the modulatory role of stress. The mirror-drawing task demands high attention, entails frequent errors, increases heart and respiratory rates and has been effectively used to generate moderate stress in previous studies (Homma 2005, 2006; Yoshiuchi et al. 1997). The HR data in this study point in this same direction. Both C homozygous and T carriers show an increased HR during the mirror-drawing task, without differences between them in this situation. This would indicate that the task is moderately stressful for both groups.
It has been proven that stress increases striatal dopamine release (Pruessner et al. 2004). This may compensate for the reduced striatal D2 density of the C homozygous subjects, generating optimal conditions for learning. White et al. (2009) presented a similar rationale to explain why stress leads to an increase in reward-related behavioural impulsivity in individuals with the CC genotype. It would therefore be of great interest to conduct further research to investigate whether the superior performance in procedural learning that is exhibited by the CC carriers in this study is maintained during similar but less stressful and similar but more stressful tasks.
According to our results, neither the C957T SNP nor the Val158Met SNP seems to be associated with performance on the verbal free recall task. In the case of the DRD2 SNP, no other studies have described this association. The evidence for the COMT SNP is contradictory, although a meta-analysis conducted by Barnett et al. (2008) concluded that the experimental data do not support this association. In this study, the retention task was very passive (reading words aloud twice), thus minimizing elaborative codification processes (Craik & Lockhart 1972). This contrasts with the retention tasks used by de Frias et al. (2004), who found an association between episodic memory and COMT. In their study, much more active retention processes were used. Therefore, the association between the Val158Met SNP and performance on episodic memory tests may be due more to the engagement of executive functions than to episodic memory per se. In fact, Schott et al. (2006) found that participants who were homozygous for the Met allele showed stronger functional coupling between the prefrontal cortex and the hippocampus during encoding.
According to our results, participants who were C homozygous for the DRD2 C957T SNP showed a smaller change in HR between the low-stress condition (during the rest period) and the moderate-stress condition (during the first two trials of the mirror-drawing task) compared with the T carriers. However, this association between genotype and HR change is difficult to interpret. The observation of the data seems to indicate that the difference between genotypes does not arise because CC carriers have lower HRs in the moderate-stress condition but rather because they have higher HRs in the low-stress condition. In this sense, CCs would have a higher basal HR level but less of a HR increase when subjected to moderately stressful situations. Nevertheless, the differences in HRs between genotypes were not found to be significant in either the moderate-stress or in the low-stress condition, so this interpretation does not have conclusive empirical support.
Ponce et al. (2008) have found, in a sample of alcoholics, that the C957T SNP and the TaqIA SNP (ANKK1 gene) are epistatically associated with dissocial personality disorder, which has been related with a lower autonomic response to stress (Hansen et al. 2007; Kiehl 2006). Huertas et al. (2010) reported that C homozygotes of the DRD2 C957T SNP showed no decrease in conditioned skin-conductance responses (another index of autonomic reactivity to stress) during conditioning when a moderately aversive unconditioned stimulus was used, whereas T carriers showed a continued decrease. A common conclusion from the present results and the results of Huertas et al. (2010) would be that CC carriers of the C957T SNP show smaller changes in the autonomic response to stress when the stress intensity changes. But this does not necessary imply a lower autonomic response to stress. Nevertheless, this topic needs to be investigated further.
In conclusion, we have explored the associations between two polymorphisms related to several neurological and mental disorders with tests of declarative and procedural memory as well as HR, a measure of autonomic reactivity. With this, we aimed to contribute to the identification of endophenotypes that will help bridge the gap between the patterns of gene expression that underlie these disorders and the corresponding syndromes (Sabb et al. 2009; see also Flint & Munafò 2007). Our results suggest that the C957T C homozygous individuals exhibit more efficient learning of procedural motor tasks and smaller differences in HR between moderate- and low-stress conditions. However, it is not clear whether this pattern of HR response in CC carriers is due to a higher responsiveness under low-stress conditions, lower responsiveness to moderate-stress conditions or both.
This work was supported by Fondo de Investigación Sanitaria (Red de Trastornos Adictivos, RD06/0001/0011), Ministerio de Sanidad y Consumo (Plan Nacional Sobre Drogas, PR61/08-16415), Ministerio de Ciencia e Innovación (SAF2008-03763 and SAF2011-26818), Grupo de Investigación UCM-Banco Santander (Grupo 940157) and Acción Especial UCM AE10/07-15503.