Our results are in line with those from previous studies demonstrating that the APS and HPS haplotypes, albeit to a different degree and through different mechanisms, both impair COMT function. The met158 allele included in the APS haplotype codes for a ~3-fold reduction in COMT activity, but not amount, that is due to reduced stability of the enzyme at normal physiologic temperature. The HPS haplotype codes for a ≥20-fold reduction in COMT activity that is due to the longer more stable local stem-loop structure in the val158 region that reduces protein translation efficiency.
As an extension of this work, we next tested whether additional SNPs situated in the transcript region that occur at lower frequency in the population contribute to variation in COMT activity. We identified a total of four such minor SNPs that were linked to APS or HPS haplotypes. Interestingly, there was an inverse relationship between the frequency of minor SNPs and major haplotypes in the EST database. Three out of the four minor SNPs were linked to the least frequent HPS haplotype coding for the least COMT activity, while one was linked to the APS haplotype coding for average COMT activity, and none were linked to the most common LPS haplotype coding for the highest COMT activity. The distribution of minor SNPs exclusively within haplotypes coding for reduced COMT activity may suggest different evolutionary models (e.g., enrichment for functional compensatory mutations or positive selection leading to carrying of neutral mutations) depending on the functional contribution of these SNPs.
Compensatory SNPs have been shown to restore fitness by suppressing the deleterious effects of nonsynonymous mutations on receptor folding, DNA-binding, and enzymatic activity 
. Thus, minor SNPs linked to the APS haplotype may compensate for the decreased enzyme thermostability produced by the met158
allele. Compensatory SNPs have also been shown to normalize RNA secondary structure formation altered by synonymous pathogenic mutations 
. Thus, minor SNPs linked to the HPS haplotype may compensate for the loss of Gibbs free energy in the RNA stem-loop structure. In contrast to compensatory SNPs, the existence of neutral SNPs in APS and/or HPS haplotypes would indicate positive selection for haplotypes associated with reduced COMT activity. As COMT regulates neuronal and non-neuronal cellular pathways important for a vast number of basic and complex biologic functions, reduced COMT leading to elevated catecholamine levels could also be beneficial.
To test whether minor SNPs linked to APS or HPS haplotypes rescue COMT activity, further reduce it, or produce neutral effects, a series of molecular modeling and cell culture studies were performed. Inclusion of the 800A (rs769224) minor allele in the APS haplotype or the 417T (rs6267), 422A (740602), or 641T (rs8192488) minor alleles in the HPS haplotype did not alter mRNA folding, RNA transcription, protein translation, or enzymatic activity. The possibility remains that inclusion of minor SNPs with APS and HPS haplotypes produces very subtle effects that were not detected using our methods and might still have effects on individual fitness or that inclusion of minor SNPs produces functional effects on molecular phenotypes not evaluated in the present study.
Therefore, these minor SNPs are likely neutral variations versus functional contributors to COMT activity. However, these data are inconsistent with a simple model of an ancestral selectively advantageous haplotype and two derived deleterious haplotypes. Under a typical population genetic model the ancestral LPS haplotype should have the highest number of neutral variants, even if undergoing strong purifying selection for high activity 
. Likewise, the more recent APS and HPS haplotypes should have a smaller number of neutral variants if undergoing background selection for deleterious low activity 
. Why these haplotypes are more variable than the LPS haplotype is puzzling. What evolutionary processes could cause a departure from a simple model of ancestral selectivity? One possibility is that APS and HPS haplotypes are not unconditionally deleterious, but under balancing selection due to pleiotropic effects on traits other than pain sensitivity. For example, the counterintuitive selection for low activity COMT haplotypes associated with high pain, anxiety, and stress response may be driven by a gain of cognitive function related to higher dopamine and norepinephrine levels. An inverted U-shaped relationship exists between catecholamine levels and cortical function, such that intermediate levels are optimal for cognition 
. Reduced COMT activity would produce a rightward shift in the inverted U-shaped curve, leading to higher levels of catecholamines associated with increased cognitive function advantageous for memory and attention tasks. Thus, both high and low activity haplotypes, each with opposing advantages and disadvantages, may be maintained in the population due to balancing selection 
. While there is no evidence, to date, of greater sequence variation in the COMT locus than that expected under a neutral model 
, the distribution of minor SNPs exclusively within haplotypes associated with reduced COMT activity requires further study. The balancing selection hypothesis would need to be addressed by 1) thorough re-sequencing of the COMT
gene locus followed by haplotype-based and linkage-disequilibrium-based analyses 
, 2) thorough phenotyping of individuals for multiple traits (e.g., pain, cognition, and memory function), and 3) development of more complex population genetic models to account for pleiotropic selection. Whether advantageous or disadvantageous phenotypes associated with low activity COMT haplotypes are ultimately expressed, is probably determined by additional genetic and non-genetic factors.Although the significant functional impact of common APS and HPS haplotypes is independent of minor SNPs situated in the coding region of the gene, it is possible that effects of APS and HPS haplotypes are modified by epistatic interactions occurring at nearby COMT
gene loci not investigated here. A previous report demonstrated that the minor allele of rs2097603 located in the MB-COMT
promoter region associated with schizophrenia 
was found to produce a 1.5-fold reduction in lymphocyte COMT activity independent of the val158
met allele 
. Additionally, a haplotype consisting of two noncoding SNPs, rs737865 in intron 1 and rs165599 in the 3′ untranslated region was associated with Schizophrenia 
and reduced expression of COMT
. These SNPs are not in high LD with the SNPs constituting the three major haplotypes, however may still interact with the coding SNPs to influence the net size and direction of effect. Furthermore, a haplotype consisting of the minor alleles of rs737865 and rs4818 in the HPS haplotype are associated with increased thermal threshold variance, implicating a role for additional unobserved functional polymorphisms 
. It is also possible that the effects of APS and HPS haplotypes are modified by epistatic interactions with mutations located in convergent molecular pathways. For example, hyperhomocysteinemia is an important risk factor for a variety of conditions associated with low COMT activity, including neurodegenerative disorders, cardiovascular disease, and hormonal cancers 
. Homocysteine is a precursor for the biosynthesis of S-adenosyl-L-homocysteine (SAH) which is a strong, noncompetitive inhibitor of COMT. Functional polymorphisms in genes relevant to homocysteine metabolism (e.g., 5,10-methylenetetrahydrofolate reductase; MTHFR and glutamate carboxypepidase II; GCPII 
) that result in accumulation of SAH likely augment the effects of APS and HPS haplotypes. In fact, a recent report demonstrated that individuals homozygous for low enzyme activity alleles of the COMT
genes in combination, but not alone, are at increased risk for schizophrenia 
. Thus, it is possible that minor SNPs situated within APS and HPS haplotypes compensate for low COMT activity not at the level we define, but instead through interactions with other genes or genotypes.
Additionally, the effects of APS and HPS haplotypes may be modified by nongenetic factors, such as environmental events that lead to a sustained elevation in catecholamines (e.g., physical and emotional stress, inflammation, and injury 
) and nutritional deficiencies that contribute to hyperhomocysteinemia (e.g., reduced dietary folate, vitamin B12
, or vitamin B6 
). An example of COMT
gene-environment interaction was recently illustrated by Slade and colleagues, who found that among individuals with the APS or HPS haplotype, TMD incidence was 23% for those with a history of orthodontic treatment and 0% for those with no history of orthodontic treatment 
. In order to truly understand the dynamic etiology of complex catechol-dependent phenotypes, flashlights used to examine the relationship between one SNP and one phenotype need to be traded in for floodlights that will illuminate dynamic relationships between genetic and epigenetic factors. Thus, future studies necessitate careful consideration as to how polymorphisms within and between distinct genetic loci interact with one another as well as with nongenetic factors to place individuals at risk.
The 417T allele was previously associated with clinical effects. This minor SNP, producing an alanine to serine substitution at codon 72, has been associated with aggressive behavior and schizophrenia risk as well as with COMT activity in red blood cells collected from study participants 
. However, we found that the functional impact of common APS and HPS haplotypes is independent of this less frequent second-site mutation situated in the coding region of the human COMT
gene. In light of this, we can revisit the conclusion regarding effects previously associated with 417T. In that study, individuals homozygous for 417G (ala72
) had the highest activity, while those homozygous for 417T (ser72
) had the lowest activity. Individuals exhibiting both the highest and lowest COMT activity were homozygous for the val158
allele, which is present in both the LPS and HPS haplotypes. In correspondence with the APS haplotype, individuals exhibiting average COMT activity were homozygous for the met158
allele. As 417T is carried exclusively with the HPS haplotype and its inclusion in this parent haplotype does not alter enzymatic activity, we can conclude that 417T is a marker of HPS haplotype and that the low activity associated with the nonsynonymous ser72
allele is in fact due to alterations in RNA secondary structure and protein translation efficiency coded for by the parent HPS haplotype.
In summary, we demonstrated that the impact of common COMT haplotypes on enzymatic activity is independent of additional minor SNPs in the coding region of the human COMT gene. Tight linkage of these neutral polymorphisms with more recently acquired low activity APS and HPS haplotypes does not satisfy a simple ancestral model and requires the development of a more complex population genetic model to account for pleiotropic selection. The counterintuitive selection for low activity COMT haplotypes may be driven by a gain of cognitive function, for example, related to increased catecholamine levels. Thus, the expression “no pain – no gain” may have literal meaning when applied to evolution of the human COMT locus.