Although cannabis does not have high abuse liability as other drugs, it is clear that some individuals are more susceptible to becoming dependent on cannabis even at a young age. Decreased negative reward/avoidance behavior conferred by DRD2
individual genetic differences has been hypothesized to account for such vulnerability. While we replicated previous findings showing a role of DRD2
in negative reward learning 
, cannabis-dependence vulnerability did not appear to be driven by this association. Instead, the results suggest for the first time to our knowledge that there are fundamental biological differences driven by genetic impairments in the opioid system in cannabis-dependent individuals compared to those who used the drug but did not become dependent. The study further revealed that Neuroticism, in contrast to other traits characteristic of cannabis-dependent subjects or compared to reward learning and inhibitory control performance, significantly mediates the association between functional SNPs of PENK
(which we now know are associated with mRNA expression levels in regions relevant to negative emotional states) and cannabis dependence. The finding that cannabis dependence is significantly enhanced in a synergistic fashion in individuals with both high Neuroticism and risk genotypes emphasizes the potential important synergistic contribution of negative emotional traits and genetics to vulnerability.
Prevailing theories postulate that addiction vulnerability is linked to impairment of reward sensitivity and/or inhibitory control. Recently, genetic polymorphisms have been used to dissect potential striatal dopamine-receptor involvement in decision-making in relation to positive and negative reinforcement outcomes 
. Our data supports an association of the DRD2
with negative reward performance, but not for other DRD2
SNPs studied. Other investigations 
have also indicated a differential effect of the DRD2
SNPs on behavior that may relate to the role of these SNPs in modulating DRD2
functions (i.e., DRD2
binding, density). The ability to replicate previous studies of the DRD2 rs6277
SNP with avoidance learning emphasizes the validity of our study population. Moreover, the overall lack of association between reinforcement learning and inhibitory control to cannabis dependence either directly or as mediators suggest that although these behavioral traits are important to various aspects of addiction, they do not appear to contribute to cannabis-dependence vulnerability to the same extent as negative emotional traits.
On the other hand, the central role of Neuroticism in the association between PENK
genes and cannabis-use outcomes in our sample has important implications. Both clinical reports and research data have suggested that coping with stress and negative mood states is a common motive for use among heavy abusers 
, which would be consistent with self-medicating subthreshold anxiety and negative affect induced by PENK
dysfunction. Interestingly, previous studies that have addressed the self-medication hypothesis showed that cannabis is more likely to exacerbate mood symptoms than to alleviate them 
. Cannabis exposure and negative affect may thus interact in a complex way within a vicious cycle where cannabis may be used to cope with subthreshold symptoms, but paradoxically further increase them in the long term. That our population consisted of non-depressed subjects underscores the fact that subthreshold symptoms that are not captured by DSM-IV may well have an impact on the emergence and course of a clinically significant disorder such as cannabis dependence.
Another important finding is the role of PENK
SNPs in predicting cannabis dependence. The vulnerability conferred by PENK
(and to a lesser extent DRD2
) SNPs may reflect disturbances of specific neurobiological systems common to these genes. Both DRD2
genes are strongly expressed in the striatum and amygdala, brain regions highly relevant to addiction disorders. A vast literature has emphasized the role of DRD2
in addiction 
; but significant evidence also suggests PENK
involvement in mood/reward regulation and anxiety that are often correlated with alterations of the mesocorticolimbic system 
and the striatopallidal circuitry in aversive behavior 
. The amygdala plays a particularly prominent role in negative mood states and enkephalinergic neurons in the central amygdala are known to be critically involved in anxiety and stress responsivity 
. Interestingly, the association detected in the current study between PENK
SNPs and mRNA
expression was most pronounced in the central amygdala. The rs2576573
A allele, observed to be more frequent in cannabis subjects, was associated with higher PENK mRNA
expression in the central amygdala of the postmortem population in control subjects. This would seem contrary to predictions based on animal studies since upregulation of amygdala PENK mRNA
expression predicts heightened anxiolytic responses and cannabis subjects exhibited more Neuroticism/anxiety. However, it is important to note that it was not possible to study mRNA
expression in the brains of cannabis users, so whether the same relationship exists between the A rs2576573
allele and mRNA expression in cannabis-dependent subjects needs to be investigated. Nevertheless, the current observation does document for the first time that PENK mRNA
expression, particularly in the amygdala, directly associates with these polymorphic noncoding PENK
SNPs. The fact though that the entire PENK
gene is in strong LD suggests that the causative mutation still remains to be identified.
There are a number of limitations that should be considered when evaluating our findings. The lack of association observed regarding some SNPs, inhibitory control and negative-reward sensitivity could be related to the study’s design. The small sample size may have limited our ability to detect genetic effects in subgroups based on dependence diagnoses. For example, few cannabis-dependent individuals carried the A allele of the DRD2 rs6277, so it was not possible to explore behavioral traits in relation to this SNP. Another factor to consider is the cross-sectional approach of this study. The acute neurocognitive effects of cannabis might have affected task performance. While subjects had at least 2 hours of observed abstinence before performing the tasks, we cannot exclude the contribution of such confounding factors and our results should be replicated in settings where chronology of recent use may be better monitored. The fact that our sample was drawn from a population without significant psychiatric comorbidity also opens up the question as to whether these results may be generalizable to other comorbid subgroups. Replication of our results in a larger cohort will also be important to address potential stratification effects. Although not specific to cannabis dependence, results from the Greek Caucasian population suggest that the association between the PENK SNPs and Neuroticism could be replicated and generalized to other populations. An additional issue to address in interpreting the data is that a large percentage of the cannabis group also smoked cigarettes; thus, nicotine withdrawal could be a potential confound to the Neuroticism trait ascribed to cannabis-dependent subjects. However, study participants did not abstain from nicotine on the day of testing and behavioral traits were determined within an hour of the subject’s arrival to reduce potential withdrawal complications. Moreover, cigarette smoking was not associated with PENK SNPs in the large Caucasian population; cigarette smoking is thus unlikely to explain the genetic findings in our subgroup of cannabis-dependent subjects who also smoked cigarettes.
Overall, our results further support the role of DRD2 in negative reward learning, and suggest a central role for Neuroticism as an endophenotype linking PENK polymorphism to cannabis-dependence vulnerability synergistically amplifying the apparent genetic risk. Future studies are needed to correlate neurobiological outcomes with behaviors in animal models and other human populations. Nonetheless, this study suggests that subthreshold mood and anxiety symptoms that do not meet criteria for a DSM-IV disorder may have serious clinical implications. Prevention and early intervention approaches that focus on coping strategies among young individuals genetically prone to Neuroticism may prove to be helpful, and should be specifically examined in such populations.