Evidence from the behavioural and fMRI experiments in this study suggests that the presence of the Met allele at codon 66 of the BDNF gene is associated with a greater probability of using a caudate nucleus-dependent response strategy, at the expense of a hippocampus-dependent spatial strategy, in our virtual navigation memory tasks. This biasing effect was seen in the breakdown of spontaneous strategies on the 4/8VM and in fMRI activation during the CSDLT, two tasks that can be performed with either a hippocampus-based spatial or a caudate-based response strategy. Furthermore, participants using a given strategy in the 4/8VM were more likely to use the same strategy in the CSDLT, showing high reliability of our phenotype across the two tasks. These findings support previous results showing that the Met allele is associated with decreased performance and fMRI activity during tasks that require the use of hippocampus-dependent episodic memory (Egan et al., 2003
; Hariri et al., 2003
; Dempster et al., 2005
; Hashimoto et al., 2008
Whereas the presence of a single Met allele leads to an increased likelihood of using a hippocampus-independent response strategy, the effect of two Met alleles on spontaneous strategies is even more striking (). This supports the findings of Hashimoto et al. (2008)
, who suggested that the effect of the Met allele on hippocampus-dependent memory is dose-dependent. Similarly, Egan et al. (2003)
found that Met/Met homozygotes scored considerably worse on their measures of episodic memory than either their Val/Val or Val/Met counterparts. The pattern that emerges in our data suggests that although the Val/Met genotype is worse for hippocampal function than the Val/Val genotype, the presence of at least one Val allele rescues some function when compared to Met/Met individuals.
The availability of two potential parallel strategies in our tasks is unique relative to the experimental design of previous Val66Met studies (Egan et al., 2003
; Hariri et al., 2003
; Dempster et al., 2005
). The use of the 4/8VM and CSDLT allowed us to study the effects of the Val66Met polymorphism on the interaction between multiple memory systems. The spontaneous quantifiable variability in our phenotype increases the sensitivity to differences in genetic variability. Whereas the use of hippocampus-dependent episodic memory tasks has allowed others to demonstrate impairments in Met carriers, we have shown that the effect of the Met allele on the hippocampus makes a participant less likely to use a spatial strategy when another option is available to them. The fact that 4/8VM and CSDLT performance measures did not differ between genotype groups is critical to exclude differences in performance as a confounding factor in our fMRI study. It also suggests that participants can compensate with the response strategy when spatial memory is at a disadvantage. This is an important strength of our study; the availability of strategies based on different memory systems has greater ecological validity, because it models the strategies available in people’s everyday lives, and demonstrates compensation mechanisms used by Met carriers.
Patients with selective thermo-coagulation lesions to the hippocampus have previously been shown to be impaired on the RAVLT and Rey–Osterreith Complex Figure tasks (Bohbot et al., 1998
). In the present study, healthy young participants with the Met genotype did not show any impairment on these tasks, indicating that they are still able to use their hippocampus despite the decreased functionality of BDNF. Although these neuropsychological tasks are well suited for detecting lesion-dependent impairments in memory, they are not as sensitive as spatial memory tasks. Other studies have shown performance differences between homozygous Val individuals and Met carriers on specific episodic memory-related subtests such as the story recall and logical memory subtests of the Wechsler Memory Scale (Egan et al., 2003
; Dempster et al., 2005
) and episodic memory recognition for scenes (Hariri et al., 2003
The fact that an effect of genotype on spontaneous strategies was demonstrated when no effect was detected on standard neuropsychological tests indicates that the navigational strategies in our study are more sensitive to BDNF genotype differences than standard tests of episodic/declarative memory. The fact that differences were observed in spontaneous strategies but not in any measure of cognitive performance highlights the utility of analysing behavioural processes at a finer level of detail in order to elucidate the often subtle link between genes and behaviour.
Whereas earlier studies demonstrated a direct effect of 4/8VM spontaneous strategies on fMRI activity during task performance (Iaria et al., 2003
; Bohbot et al., 2004
), the present study shows that the Val66Met genotype also predicts fMRI activity in the same brain regions on a similar task. One question that cannot be answered on the basis of the current limited sample is whether the strategy used and genotype have independent effects on fMRI activity. Future studies should investigate whether these effects interact with one another, or whether they independently explain the observed differences in brain activity.
Previous studies have demonstrated that acute and chronic stress can bias strategies towards the caudate nucleus-dependent response strategy in humans (Schwabe et al., 2007
; Schwabe & Wolf, 2009
) and rodents (Kim et al., 2001
), a process that depends on an intact amygdala (Kim et al., 2001
) and is mediated by stress-induced increases in cortisol levels (Schwabe et al., 2010
). An interaction with BDNF expression may represent one potential mechanism by which stress impacts on spontaneous navigational strategy. Chronically stressed rats show a marked decrease in BDNF mRNA expression in the dentate gyrus and hippocampus (Smith et al., 1995
) and concomitant impairment on the hippocampus-dependent Morris water maze task (Song et al., 2006
), just as genetically modified mice with reduced BDNF expression display poor water maze performance (Heldt et al., 2007
). Direct infusion of BDNF into the hippocampus before and during chronic stress rescued water maze performance in rats who had been subjected to repeated immobilization stress (Radecki et al., 2005
It is possible that stress plays an intermediate modulatory role between genotype and behaviour, whereby the impact of BDNF on learning and memory strategies may be influenced by exposure to stress. Having the Val/Val genotype may increase the likelihood of participants using a spatial strategy, even following stress exposure, via increased BDNF expression and long-term potentiation in the hippocampus. On the other hand, participants with the Met allele would be more likely to use response strategies in their everyday lives, following stress exposure. Increased BDNF expression in Val/Val individuals could potentially protect against the demonstrated neurotoxic effects of cortisol on the hippocampus following stress (McEwen, 1998
). This hypothesis is supported by the fact that the Val66Met genotype does not perfectly predict spontaneous learning strategies, suggesting that other factors, such as stress or polymorphisms in other genes, may also play a role.
Whereas genetic factors, modulated by external factors such as stress, play a role in predisposing individuals towards the spontaneous use of one strategy over another, it is probably the reinforcing effect of continually activating one of these memory systems that leads to differences in function and morphology between individuals over time. Previous studies have demonstrated that spatial memory training in taxi drivers can lead to increased grey matter in the hippocampus (Maguire et al., 2000
). Further support for the impact of training on brain morphology comes from studies in adult mice, which have demonstrated that training on either a spatial-dependent or a response-dependent task leads to significant growth in the hippocampus and striatum, respectively (Lerch et al., 2007
). The fact that the hippocampus can grow as a result of spatial memory training is important in the light of evidence linking hippocampal atrophy to psychiatric and neurological brain pathologies, such as Alzheimer’s disease (Bremner et al., 2000
; Apostolova et al., 2006
). Further evidence regarding the impact of genetic and environmental factors on the function and grey matter of the hippocampus will help to determine the risk of neurological and psychiatric illnesses before symptoms occur.
Although the mechanisms by which the polymorphism confers differences in disease risk remain to be worked out, gene association studies have suggested that Met carriers are at increased risk for a number of neurological and psychiatric pathologies known to affect the hippocampus, including Parkinson’s disease (Momose et al., 2002
), bipolar disorder (Neves-Pereira et al., 2002
), eating disorders (Ribases et al., 2003
), and obsessive compulsive disorder (Hall et al., 2003
). A neuroprotective effect of BDNF against Alzheimer’s disease has also been demonstrated in rodents and non-human primates (Tapia-Arancibia et al., 2008
; Nagahara et al., 2009
). The knowledge that spatial memory training can reverse atrophy of the hippocampus, in combination with knowledge of genetic and environmental risk factors for these conditions, may provide us with the tools required to selectively overcome predispositions to certain types of disorder.
The present study demonstrates that the BDNF Val66Met polymorphism plays a role in spontaneous navigational strategy. Met carriers demonstrated a decreased probability of using a hippocampus-dependent spatial strategy as compared with homozygous Val individuals, a phenomenon that could have long-term effects on grey matter and function in this brain structure (Bohbot et al., 2004
). These strategy differences translated into differences in brain activation between the Val and Met groups, with Val individuals preferentially activating their hippocampus and Met carriers exhibiting enhanced caudate nucleus activation with increased use of response strategies over time. Although genetic factors alone do not explain the differences in behaviour between spatial and response learners, the present study demonstrates that the BDNF gene with the Val66Met polymorphism is a novel candidate gene involved in modulating spontaneous strategies during navigation behaviour.