The main finding of this study is that the SNPs rs3766246 and rs2295633 were associated with higher self-reported Arousal in response to amphetamine. Post hoc
analyses of the subcomponents of Arousal revealed that subjects with genotypes C/C at rs3766246 and rs2295633 and also the functional SNP rs324420 showed significantly greater decrease in Fatigue after 10
mg amphetamine administration compared with the other two genotypes. In addition, the CCC haplotype from rs3766246, rs324420, and rs2295633 was significantly related to lower ratings of Fatigue after amphetamine. Haplotype TAT, but not haplotype TCT, was significantly correlated with higher Fatigue scores after 10
mg amphetamine administration. Thus, these three SNPs, and in particular rs324420, showed effects on Fatigue both alone and when combined to form a haplotype. The three investigated FAAH
haplotypes were not significantly associated with levels of Arousal. However, a trend was apparent in line with our findings for Fatigue. Subjects with the CCC haplotype from rs3766246, rs324420, and rs2295633 showed higher ratings of Arousal and subjects with haplotype TAT scored lower on Arousal after 10
mg amphetamine administration (p
When specific doses were examined, only the 10-mg dose showed an effect of genotype groups on Arousal and Fatigue. It is possible that FAAH
genotype effects on response to amphetamine are apparent only at lower, marginally effective doses, and that the genetic differences are overcome by higher doses of the drug. A similar relationship between a SNP in the gene CSNK1E and dose of amphetamine was observed in a previous study (Moreira et al, 2008
) supporting the idea that certain drug–genotype interactions might be more evident at lower doses. FAAH
genotypes were not related to any demographic factors or to POMS-scale baseline mood scores. These findings demonstrate a pharmacogenetic difference, indicating that variations in the FAAH
gene influence acute responses to d-amphetamine.
The only SNP examined in this study with a known functional consequence is rs324420. In vitro
and in vivo
studies have shown reduced cellular activity and expression of the human A/A variant (Chiang et al, 2004
). Thus, the FAAH
C/C variant may result in higher FAAH enzyme activity and consequent lower endocannabinoid levels due to greater degradation of endocannabinoids by FAAH (Doehring et al, 2007
). This might underlie the observed lower levels of fatigue, higher levels of Arousal, and levels of feeling stimulated after 10
mg amphetamine administration. In line with our suggestions, Murillo-Rodriguez (2008)
found activation of the endocannabinoid system via the CB1 receptor to induce sleep and to modulate wakefulness (Murillo-Rodriguez 2008
-knockout mice process higher values of slow wave sleep and more intense episodes of slow wave sleep as compared with wild-type animals (Huitron-Resendiz et al, 2004
). However, in another study oleoylethanolamide, palmitoylethanolamide, and an FAAH antagonist were found to enhance waking (Murillo-Rodríguez et al, 2007
Further, Tyndale et al (2007)
found that subjects (n
=749) with the A/A genotype of rs324420 were at significantly reduced risk for being THC-dependent as compared with those with the C/A or C/C genotype. The C/C variant of rs324420 is also associated with higher craving for marijuana after abstinence (Haughey et al, 2008
) and with lower frequencies of obesity (Sipe et al, 2005
). However, the A/A genotype (not C/C) of rs324420 is more prevalent in problem drug users (Sipe et al, 2002
) and in persons having addictive traits (Flanagan et al, 2006
). Thus, while our results and multiple prior studies have suggested that this SNP is functionally important, neither the C nor the A alleles can be easily identified as the ‘risk allele' for drug abuse and dependence when considering all available data.
Our data suggest that, in addition to the non-synonymous variant rs324420 two other polymorphisms (rs3766246 and rs2295633) located in the same haplotype block are also associated with differences in subjective response to amphetamine. It is conceivable that either of these intronic variants may be associated with differences in mRNA processing, stability, splicing, or changes of transcription rates. Interestingly, in a study of response to cold pain, rs2295633, but not rs324420, was significantly associated with pain sensitivity (Kim et al, 2006
). As these three SNPs form a haplotype block (), it is difficult to determine which (if any) of them are causally related to the observed phenotypes. Thus, some uncertainty remains about whether the coding difference caused by the variant rs324420 or changes by one of the intronic variants is the causal polymorphism at this locus.
In an effort to investigate the possibility that a difference in gene expression, rather than a coding difference, was reasonable for the associations detected in this study, we investigated the possible associations between the SNPs surveyed in this study and gene expression. We used expression data from HapMap CEPH immortalized lymphoblast cell lines (Stranger et al, 2007
; Veyrieras et al, 2008
) for this purpose. We identified a highly significant (log(Bayes factor) >10) association between rs6703669, which is located in intron 1 of the FAAH
gene, and expression of NSUN4 (probe ID hmm8232) (Veyrieras et al, 2008
). A similar observation is reported in the Supplementary materials of Stranger et al (2007)
(Supplementary Table S2, probe ID hmm8232). FAAH
is immediately adjacent to the NSUN4
gene. cDNAs that include exons from NSUN4 and FAAH
have been detected (www.ncbi.nlm.nih.gov/IEB/Research/Acembly
), suggesting that these two genes constitute a single gene complex (Thierry-Mieg and Thierry-Mieg, 2006
). However, rs6703669 was not significantly associated with any of our outcome measures (), and, conversely, none of the three SNPs in and were significantly associated with differential gene expression. In summary, while we did identify an SNP in FAAH
that is associated with differential gene expression, this SNP was not associated with the sensitivity to amphetamine as measured by Arousal or Fatigue in the present study.
The quality and magnitude of subjective responses that individuals experience from their first experience with the drug is related to subsequent abuse or dependence (Di Franza et al, 2004
; Fergusson et al, 2003
; Haertzen et al, 1983
). Thus, our findings may help to predict individual differences in susceptibility to misuse amphetamine. Individuals carrying C alleles of rs3766246, rs2295633, or haplotype CCC from rs3766246, rs324420, and rs2295633, may be possibly more likely to use amphetamine repeatedly as a recreational drug because of their greater sensitivity to its stimulating subjective effects. As we only found FAAH
-genotype-dependent differences in response to the low amphetamine dose, it remains unclear whether our finding has an impact on frequency of higher dosage drug abuse and has to be investigated in future studies. Alternatively, in cases where amphetamine is being clinically used to counteract fatigue, lower doses might be sufficient in individuals with high-sensitivity genotypes. Thus, in the context of clinical use of amphetamine, it may be possible to identify patients who are at risk for amphetamine abuse, or to better calibrate the dose required when using amphetamine for its stimulant properties. In the former case, special precautions may be taken when prescribing this drug to patients at risk for abuse (Shastry, 2006
). However, it should be noted that genetic variation only accounts for a fraction of individual differences, and FAAH
is only one of the many genes involved in the genetics of amphetamine sensitivity (Dlugos et al, 2007
; Lott et al, 2005
; Palmer et al, 2005
; Veenstra-VanderWeele et al, 2006
). Thus, clinically useful predictions will require accounting for multiple genetic loci.
This study has several limitations. Our findings need confirmation by performing larger replication studies. Although this complex pharmacological study used a substantial number of subjects, the inherent level of variability in the outcome measures and the complexity of the gene–environment interactions call for replication to confirm these observations. Further, we only included subjective outcome measures to assess the effects of amphetamine, and functionally intermediate measures such as functional MRI or SPECT (Mattay et al, 2003
; Rohde et al, 2003
) would help to reinforce our observations. Such studies would further elucidate how and in which brain regions the investigated genetic variations might modulate the endocannabinoid system after amphetamine consumption. Another limitation of the study is the fact that the minor-allele frequency of the functional variant for rs324420 was low, (0.225 HapMap CEU sample), requiring us to pool subjects with genotypes A/A and A/C into a single group. Replication in a larger sample may allow separate analyses of all three genotypes.
In summary, our study provides novel evidence that genetic variation in the FAAH gene is associated with specific mood responses after amphetamine administration. These data add to evidence that the endogenous cannabinoid system is related to response to a stimulant drugs in humans, which may lead to improvements in preventing and treating amphetamine abuse disorder.