Effect of the COMT Val158Met genotype on lateral prefrontal activations in young children

Abstract Low executive function (EF) during early childhood is a major risk factor for developmental delay, academic failure, and social withdrawal. Susceptible genes may affect the molecular and biological mechanisms underpinning EF. More specifically, genes associated with the regulation of prefrontal dopamine may modulate the response of prefrontal neurons during executive control. Several studies with adults and older children have shown that variants of the catechol‐O‐methyltransferase (COMT) gene are associated with behavioral performance and prefrontal activations in EF tasks. However, the effect of the COMT genotype on prefrontal activations during EF tasks on young children is still unknown. The present study examined whether a common functional polymorphism (Val158Met) in the COMT gene was associated with prefrontal activations and cognitive shifting in 3‐ to 6‐year‐old children. The study revealed that, compared with children with at least one Met allele (Met/Met and Met/Val), children who were Val homozygous (i) were more able to flexibly switch rules in cognitive shifting tasks and (ii) exhibited increased activations in lateral prefrontal regions during these tasks. This is the first evidence that demonstrates the relationship between a gene polymorphism and prefrontal activations in young children. It also indicates that COMT Val homozygosity may be advantageous for cognitive shifting and prefrontal functions, at least during early childhood, and children who possess this variant may have a lower risk of developing future cognitive and social development issues.

ated with the regulation of prefrontal dopamine may modulate the response of prefrontal neurons during executive control. Several studies with adults and older children have shown that variants of the catechol-O-methyltransferase (COMT) gene are associated with behavioral performance and prefrontal activations in EF tasks. However, the effect of the COMT genotype on prefrontal activations during EF tasks on young children is still unknown. The present study examined whether a common functional polymorphism (Val158Met) in the COMT gene was associated with prefrontal activations and cognitive shifting in 3-to 6-year-old children. The study revealed that, compared with children with at least one Met allele (Met/Met and Met/Val), children who were Val homozygous (i) were more able to flexibly switch rules in cognitive shifting tasks and (ii) exhibited increased activations in lateral prefrontal regions during these tasks. This is the first evidence that demonstrates the relationship between a gene polymorphism and prefrontal activations in young children. It also indicates that COMT Val homozygosity may be advantageous for cognitive shifting and prefrontal functions, at least during early childhood, and children who possess this variant may have a lower risk of developing future cognitive and social development issues.

RESEARCH HIGHLIGHTS
• Variants of the catechol-O-methyltransferase (COMT) gene are associated with the regulatory mechanism of prefrontal dopamine that may modulate prefrontal function during executive control.
• We examined whether a common functional polymorphism (Val158Met) in the COMT gene associated with prefrontal activations and cognitive shifting in 3-to 6-year-old Japanese children.
• Older rather than younger children who were Val homozygous appeared to more flexibly switch rules compared to peers with at least one Met allele.
• Children who were Val homozygous exhibited stronger activations in lateral prefrontal regions than children with at least one Met allele during cognitive shifting tasks. and the other (i.e., cognitive shifting) (Garon, Bryson, & Smith, 2008;Miyake & Friedman, 2012). Recently, research has shown that the quality of EF during early childhood is a predictor for several aspects of adolescent and adult life, including academic achievement, peer-topeer relationships, and socioeconomic and health status (Moffitt et al., 2011). Therefore, identifying factors contributing to individual differences in EF during early childhood has become one of the most important research topics in developmental and psychological science.
Behavioral genetic research using a twin-study design suggests that genetic influences contributing to individual differences in EF may be affected by age (Fujisawa, Todo, & Ando, 2017). In more detail, genetic effects on individual differences in EF amongst adults can be stronger compared to younger populations (Friedman, Miyake, Robinson, & Hewitt, 2011;Friedman et al., 2008), despite a strong shared and non-shared environmental influence on individual differences in EF among preschool children (Fujisawa et al., 2017). This latter study suggests that genetic factors begin to contribute to individual differences in EF during early childhood.
It has been shown that lateral prefrontal regions contribute to EF in adult populations (Miller & Cohen, 2001). In the developmental literature, EF development is related to activity in the neural network including prefrontal cortex, parietal cortex and subcortical regions during childhood (Bunge, Dudukovic, Thomason, Vaidya, & Gabrieli, 2002;Moriguchi, 2017;Morton, Bosma, & Ansari, 2009).
In terms of prefrontal cortex, recent neuroimaging studies show that infants and children recruit the lateral prefrontal cortex during EF tasks, including working memory and cognitive shifting (Baird et al., 2002;Moriguchi & Hiraki, 2009, 2011Tsujimoto, Yamamoto, Kawaguchi, Koizumi, & Sawaguchi, 2004). For example, by using functional near-infrared spectroscopy (fNIRS) to measure brain activity in children, Moriguchi and Hiraki (2009) found that the development of cognitive shifting during the Dimensional Change Card Sort (DCCS, Figure 1a) task associated with increased activations within lateral prefrontal cortex. Lateral prefrontal cortex is clearly important for EF, which in turn suggests that there are specific F I G U R E 1 Experimental settings. (a) Experimental sequence. (b) The NIRS probe was attached to the lateral prefrontal area. Each channel consisted of one emitter optode and one detector optode. The regions of interest were located near F3/4, corresponding to channels 2, 4, and 5 and channels 11, 13, and 14, respectively features within this area of the brain that contribute to individual differences in children's EF.
In this regard, dopamine is an important neurotransmitter required for functions within lateral prefrontal cortex. Catechol-Omethyltransferase (COMT) is the major enzyme that contributes to the degradation of released dopamine within this region of the brain (Karoum, Chrapusta, & Egan, 1994;Weinshilboum, Otterness, & Szumlanski, 1999). Several studies have shown that variants of the COMT gene are associated with behavioral performance and prefrontal activation in EF tasks. These include a functional missense mutation at codon 158 that results in a substitution of methionine (Met) for valine (Val). The substitution of Met for Val results in a three-to fourfold reduction in enzyme activity, leading to prolonged dopamine action in the prefrontal cortex (Chen et al., 2004 (Cools, 2006).
The relationship between the COMT gene, EF, and the prefrontal cortex is complex. Adult brain imaging studies repeatedly show that the Met carriers who perform better on working memory tasks show less (i.e., more efficient) lateral prefrontal activation than worse performing Val carriers (Caldú et al., 2007;Egan et al., 2001). Conversely, several other studies examining other tasks, such as inhibitory control, report that increased prefrontal neural activity is found in individuals with the Met allele who also performed these tasks better (Congdon, Constable, Lesch, & Canli, 2009;Winterer et al., 2006). This discrepancy may be due to the association of behavioral performance and activation in the prefrontal regions (Jaspar et al., 2014 Previous research has shown that Val homozygosity is associated with better performance on tasks involving cognitive shifting among the adult population (Colzato et al., 2014;Colzato et al., 2010). In addition, superior performance during the DCCS task is associated with increased activation within lateral prefrontal regions of young children (Moriguchi & Hiraki, 2009). Thus, we predicted that young children who were Val homozygotes would (i) perform better in an EF task involving cognitive shifting (i.e., the DCCS) and (ii) exhibit increased lateral prefrontal activation compared to Met carriers.
All children were recruited from nursery schools in Osaka, Japan and did not have any known developmental abnormalities. An additional 12 children were tested but excluded from the final analyses because of their refusal to wear the NIRS probe (n = 7) and experimental error (n = 3). Moreover, two parents did not report their SES, and these children (n = 2) were also excluded from the analyses. Informed consent was obtained from all parents prior to their child's involvement in the study, which was conducted in accordance with the principles of the Declaration of Helsinki and approved by the local ethics committee.  (Isegawa et al., 2010). Each sample was analyzed twice, and the results of the analyses were in agreement across all samples.

| Socioeconomic status
Maternal education and family income were used as indices of social environment. In terms of maternal education, each parent's education level was assigned a value from 1 to 5 (less than high school 1, high school 2, some college 3, undergraduate degree 4, graduate level 5).

| Behavioral tasks
To assess effects on behavior, children were asked to perform the DCCS task. We used a modified version of the NIH tool box ) that adapted the material and procedures to allow for brain activity acquisition during assessment, as described in previous studies using NIRS (Moriguchi, Sakata, Ishibashi, & Ishikawa, 2015).
Laminated cards (3.5 cm × 7.0 cm) that had two dimensions (shape and color) were used as stimuli (Figure 1a). The task included target cards and test cards: target cards matched test cards in one dimension but did not match in the other dimension (e.g., target cards of a yellow house, or a green car, and test cards of green houses or yellow cars).
The experiments included three different pairs of target and test cards.
At each session, a different tray with a different set of cards was used.
Children performed three consecutive test sessions. One session consisted of a rest phase (15 s), pre-switch phase (20 s), second rest phase (15 s), post-switch phase (20 s), third rest phase (15 s), and mixed phase (20 s). During the pre-switch phase, children were given instructions regarding the first rule (e.g., "This is a shape game. All the cars go here, and all the houses go there"). During the post-switch phase, they were asked to sort the cards according to the second rule (i.e., color).
Finally, during the mixed phase, children performed the task after they were given instructions regarding the rule (e.g., "This is a mixed game. In this game, you will use both shape and color rules."). In each phase, children were given a rule for each trial (e.g., color). The rule order (e.g., color first) during these pre-switch and post-switch phases was held constant across the three sessions for each child, but the rule order was POST. The number of the PRE trials was not the same as the number of the POST trials based on a previous study .
The dependent measures were the percentage of (i) correct responses and (ii) successful switching. We calculated the percentage of successful shifting as a measure to index total performance because the pre-switch and post-switch phases are easy for older children.
Children needed to switch rules between the pre-switch and postswitch phases (one switch). The pass criterion was 90% correct in the pre-switch and post-switch phases (Towse, Redbond, Houston-Price, & Cook, 2000). Moreover, children needed to switch the rules during the mixed phase four times (four switches). Thus, we calculated the percentage of successful switching out of five switches. We measured changes in oxy-Hb and deoxy-Hb in lateral prefrontal areas during the rest phases and each of the task phases. The average changes in oxy-Hb and deoxy-Hb during the rest and the task phases were calculated for each channel (channels 11, 13, and 14 and channels 2, 4, and 5) in each subject.
Recently, it has been shown that the NIRS signal can be contaminated by physiological activities other than cerebral function, such as cardiac pulsation, respiration, and body motion (Huppert, Diamond, Franceschini, & Boas, 2009

| COMT genotyping
The
Post-hoc analyses of this interaction (corrected using the Bonferroni method) revealed that older children performed better than younger children in the post-switch and mixed phases (p < .029), but not in the pre-switch phase (p > .345). No other significant results were observed.
Given that good performance in the pre-switch phase is a prerequisite for being able to conclude that performance in the post-switch phases reflected switching, we examined whether COMT genotype affected the percentage of correct responses in the post-switch and the mixed phases in the DCCS tasks for children who performed more than 90% in the pre-switch phases. We conducted an age ( revealed that the COMT genotype had no effect on successful shifting in younger children (p = .434). However, the Val/Val group were better at switching than the ≥1 Met allele group in older children (p = .020).
No other significant effects were found.
These results are consistent with a previous study suggesting that the effects of the COMT gene were observed in 4-and 5-year-old children, but not in 3-year-old children (Blair et al., 2015).

| NIRS results
Next, we examined whether the lateral prefrontal regions were activated during the DCCS task. The results for the oxy-Hb and deoxy-Hb measurements were consistent after separating the NIRS signal into functional and systematic components, and we therefore report the oxy-Hb results. The significance of the possible difference between changes in oxy-Hb for both the rest phase and task phase was determined by a two-tailed Student's t test for each channel. The behavioral F I G U R E 2 Behavioral results. (a) Percent correct and (b) ratio of successful switching data showed that the percentage of successful shifting was associated with genetic effects, and we calculated brain measures to index successful shifting. Specifically, we combined each rest phase into an aggregated rest phase, and each task phase (pre-switch, post-switch, and mixed phases) into an aggregated task phase. We compared the aggregated task phase to the rest phase using a multiple comparison test and applied a 0.004 (0.05/12) alpha level of significance (two groups and six channels).
We analyzed separately lateral prefrontal activations in the Val/Val and ≥1 Met allele groups. Children in the Val/Val group had significantly activated right (Channel 2, 4, and 5) and left (channel 14) prefrontal regions during the task phases compared to the rest phase (Student's t test, p < .004; Figure 3a). Children in the ≥1 Met allele group exhibited significant activation in both the right (Channel 2 and 4) and left (channel 14) prefrontal regions during the task phases, compared to during the rest phase (Student's t test, p < .004; Figure 3b).
Next, we directly compared activation in the lateral prefrontal regions of the Val/Val and ≥1 Met allele groups. As the prior analyses indicated some differences in the right prefrontal regions (i.e., more channels for the Val/Val group), we therefore focused on the right prefrontal regions. To assess any differences, we conducted a MANCOVA with mean changes in oxy-Hb (channels 2, 4, and 5) as the dependent variable, age (old vs. young) and COMT genotype (

| DISCUSSION
The present study provides the first evidence for a relationship between executive function (EF), gene polymorphism, and the prefrontal cortex in young children. Research on adult subjects suggests that F I G U R E 3 Temporal changes in the oxy-Hb (red) and deoxy-Hb (blue) concentration within the right (channels 2, 4, and 5) and left (channels 11, 13, and 14) lateral prefrontal areas during performance of the DCCS tasks. The brain measures index was calculated as the percentage of successful shifting by combining each task phase (pre-switch, post-switch, and mixed) into an aggregated task phase. Aggregated task phases were compared with the rest phase. Group mean data from children in the ( Although adults of European descent with the Met allele of the COMT gene show better performance in working memory tasks than those who are Val homozygous (Egan et al., 2001), research using cognitive shifting tasks has shown that Val homozygotes outperform individuals homozygous for Met (Colzato et al., 2014;Colzato et al., 2010). Our results are consistent with the latter study, showing superior performance by Val homozygous young Japanese children.
However, the relationship between the COMT gene, EF, and prefrontal activations should be interpreted with caution. Adult Met carriers who perform better in working memory tasks show weaker lateral prefrontal activations than Val homozygote carriers who perform worse Egan et al., 2001). On the other hand, Met carriers who perform better in inhibitory control tasks compared to Val carriers show increased neural activity (Congdon et al., 2009;Winterer et al., 2006). It is possible that efficiency in some EF tasks is associated with increased prefrontal activity (i.e., inhibition), whilst other tasks are associated with decreased prefrontal activity (i.e., working memory) (Jaspar et al., 2014). Regarding young children, superior performance has been associated with increased prefrontal activity during DCCS tasks (Moriguchi & Hiraki, 2009;Moriguchi et al., 2015). Therefore, it seems that Japanese children, at least, who are Val homozygotes show better cognitive shifting and more prefrontal activity than Met carriers, although we have to note that what increased and decreased activities in prefrontal cortex reflect may differ across children and adults, because lateral prefrontal cortex continues to develop during childhood (Gogtay et al., 2004).
We also found that the effects of COMT genotype on behavioral performance were different from those for prefrontal activations. More specifically, the effects on behavioral performance were observed only in older children, whereas the effects of COMT genotype on brain activity were observed regardless of age. These results indicate that the effects associated with COMT genotype are expressed earlier in the brain than in behavior. Such results are not necessarily surprising after acknowledging that activations in prefrontal cortex are a more sensitive (and direct) measure of the effects of dopamine and COMT genotypes. Others, too, have found evidence of effects (advances as well as declines) earlier in the brain than on behavioral measures (Bookheimer et al., 2000;Friederici, Friedrich, & Christophe, 2007 (Moriguchi & Hiraki, 2009, 2011. In addition to NIRS studies, event-related potential (ERP) studies have confirmed that neural activation during the pre-switch and post-switch phases contributed to the performance in the post-switch phases (Espinet, Anderson, & Zelazo, 2012 (Wang et al., 2013;Yeh, Chang, Hu, Yeh, & Lin, 2009). Therefore, future studies should assess whether the observed genetic differences in young children associated with cognitive shifting and prefrontal activations are affected by ethnic grouping.
The results of our study advance understanding of potential individual differences in a child's EF. EF during early childhood is a predictor for academic achievement and social status during adolescence and adulthood (Moffitt et al., 2011), and several studies attempt to identify and understand genetic and environmental effects on individual differences in EF during early childhood (Blair et al., 2015;Diamond et al., 2007;Diamond et al., 2004;Noble et al., 2005). Our study clearly shows that genetic and biological factors play a role. Young children who are homozygous for the Val allele were more likely to show better cognitive shifting than children with at least one Met allele. This, in turn, may decrease the risk of future impairment due to reduced cognitive and social development and this proposal is consistent with Mean Oxy-Hb changes (mM-mm) a previous study (Blair et al., 2015). In contrast, Met carriers may be at risk of developing impaired cognitive shifting during early childhood.
This may be because Met carriers are more sensitive to environmental factors, such as early adverse experiences, when compared to Val homozygotes (Diamond, 2011). Nevertheless, we have assessed only children's cognitive shifting in this study, and different results may be obtained when working memory tasks are examined.
Finally, the COMT gene is almost certainly not the only gene that has alleles that contribute to EF and prefrontal activations in young children. Future research should be undertaken to identify other susceptible genes that may affect the molecular and biological mechanisms that underpin EF. Moreover, prefrontal regions are not the only brain regions that contribute to EF. Indeed, recent brain imaging studies have shown that the brain network including lateral prefrontal cortex, parietal cortex, as well as subcortical regions and their interaction play an important role in the development of EF (Moriguchi, 2017;Morton et al., 2009). Future research should assess how the brain network contributes to the development of EF during childhood.