In the present study, we have found that the valine allele of COMT is associated with diminished NRG1-induced AKT1 phosphorylation in B lymphoblasts from both controls and patients, and showed that COMT overexpression in SH-SY5Y cells led to impaired AKT1 phosphorylation and migration in response to NRG1. These results suggest that the relatively poorer NRG1-induced adhesion and migratory response seen in Val homozygote lymphoblasts 
is due, at least in part, to reduced activation of AKT1. In addition, we have demonstrated a plausible mechanism by which the effect of COMT activity on AKT1 function may be mediated, at least in part. We suggest that consumption of SAM by COMT may affect the ability of cells to regulate PS levels involved in translocation and activation of AKT1 by altering phospholipid methylation, although clearly we cannot rule out the possibility that the competition of COMT for SAM impacts on other potential mechanisms (e.g. DNA methylation) that could affect AKT1 phosphorylation in addition to changes in PS or independent of such changes and could, moreover, be the more important mechanism of the effect of COMT on AKT1 phosphorylation.
As with other chemokine-stimulated migratory responses, activation of AKT1is essential for NRG1-stimulated cell adhesion and migration in B lymphoblasts, evidenced by the severe attenuation of adhesion and migration by PI3K or AKT1 inhibition in these cells 
. Therefore, we suggest that the poor migratory response we previously showed in Val homozygote lymphoblasts might be due at least in part to poor activation of AKT1 in these cells. It is likely that the association between the COMT Val/Met polymorphism and AKT1 activation is mediated via its effects on COMT enzyme activity 
, since COMT overexpression in SH-SY5Y cells significantly decreased NRG1-induced phosphorylation of AKT1. Taken together, these data suggest that high COMT activity (resulting either from overexpression or as the result of Val homozygosity) is inhibitory to the function of AKT1.
Identifying the mechanism of how COMT activity inhibits the function of AKT1, and therefore, NRG-dependent adhesion and migration is challenging. Considering the well characterized function of COMT as an enzyme involved in catechol inactivation, it was conceivable that the effect of COMT on migration and adhesion could be related to a dopamine- or catecholic estrogen-mediated mechanism, since these substrates for COMT can be derived from fetal bovine serum or might be produced by B lymphoblasts in culture. However, this mechanism is unlikely to be responsible, since in B lymphoblasts the production of dopamine is low and the expression of dopamine receptors is scarce (unpublished observation), and because the migration assay was performed in serum-free media. Rather, our earlier results suggested another mechanism, an indirect effect of COMT on methylation of other critical molecules involved in phospholipids to explain the inverse relationship between COMT and AKT1 activation.
It has been suggested that an increase in COMT-mediated methylation decreases the SAM pool and increases SAH, which acts as a feedback inhibitor of SAM-dependent methylation processes 
. Because the Val form of COMT has higher enzyme activity, it would result in relatively higher SAM consumption and SAH generation than the Met form, a hypothesis supported by data showing higher levels of plasma homocysteine, a molecule formed by the hydrolysis of SAH, in Val carriers compared with Met homozygotes 
. Therefore, Val homozygotes would be predicted to have a greater inhibitory effect of COMT on other SAM-dependent methyltransferases, compared with Met homozygotes. Among these methyltransferases, we considered phosphatidylethanolamine N
-methyltransferase (PEMT) to be a good candidate for mediating COMT's association with AKT1 function, as it is involved in the synthesis of PS, which is important for the full activation of AKT1. Activation of AKT1 requires its translocation from the cytoplasm to the plasma membrane, followed by its phosphorylation at residue T308 and then S473 by PDK1 and PDK2, respectively. It has been well demonstrated in many systems that translocation of AKT1 is triggered by PIP3
levels, which are regulated by PI3K and PTEN 
. However, its translocation has been found to not only require membrane PIP3, and but also PS 
. PS serves as an essential attractor for AKT1 by providing a negative charge to the PHD-AKT1. Consistent with this theory, translocation of AKT1 has been shown to be affected dose-dependently by PS content 
After confirming that PEMT, PSS1 and PSS2, which mediate the synthesis of PS 
, are expressed in B lymphoblasts (Supplementary Information, Fig. S5
online), we predicted that high COMT activity might decrease PS in cells via competition with PEMT for SAM, especially in serum-free culture, where exogenous PS, PC and cholines are unavailable. Consistent with this prediction, we found that overnight culture in serum-free media caused significant decreases in PS content in B lymphoblasts and that the extent of the decrease was negatively correlated with COMT activity. Thus, the higher COMT activity the cells possessed, the bigger decreases in PS the cells exhibited. We also showed that NRG1-stimulated translocation of PHD-AKT1 is significantly lower in Val homozygotes than in Met homozygotes while NRG1-stimulated production of PIP3
was not reduced in Val homozygotes compared to Met homozygotes. Therefore, these results were consistent with the hypothesis that COMT's effects on AKT1 are mediated via PS, rather than PIP3
. The impaired NRG1-stimulated PHD-AKT translocation without altering PIP3 generation was observed in COMT-transfected SH-SY5Y cells. Moreover, the inverse relationship between COMT and PS was also supported by COMT transfection studies using SH-SY5Y cells. The transfection studies also demonstrated that the deficit in AKT1 activation induced by COMT overexpression could be rescued by SAM administration, demonstrating the SAM-dependence of COMT's negative regulation of AKT1 function. Thus, our results suggest that high COMT activity affects the cells' ability to maintain or synthesize PS via the PEMT-PSS1 pathway, leading to reduced PS content and, in turn, poorer translocation and activation of AKT1.
If the reduction in PS synthesis and AKT1 activation is related to COMT-mediated SAM consumption, we would expect that these effects of COMT on AKT1 would not be limited only to NRG1-ErbB signaling. Consistent with this prediction, studies using SH-SY5Y and HEK293 cells demonstrated that COMT overexpression inhibited phosphorylation of AKT1 in other relevant signaling pathways such as SDF1-CXCR4, and similar to our findings for NRG1, this inhibition was overcome by SAM augmentation.
Interestingly, in addition to a single effect of COMT genotype on AKT1, we also found evidence of an interaction between COMT Val/Met and a SNP in AKT1 that is associated with variation in AKT1 protein expression. Thus, the minor A allele of rs1130233
, which is associated with lower AKT1 protein levels in lymphoblasts, was associated with poorer NRG1-induced AKT1 phosphorylation in cells that also were in COMT Met homozygotes. This statistical interaction is not based on a simple differential effect of COMT Met and Val alleles on AKT1 phosphorylation in the context of AKT1 genotype, since Val alleles produced maximum inhibition of NRG1-induced AKT1 phosphorylation regardless of AKT1 genotype. Because AKT1 protein levels would be expected to have an effect on AKT1 recruitment and localization for its activation and because of the association of rs1130233
genotype with AKT1protein levels 
, we would expect the effect of the SNP in AKT1 to be more apparent in Met COMT carriers. Consistent with prior data, we were able to replicate the association between AKT1 rs1130233
and protein levels 
, demonstrating that the minor A allele of rs1130233
is associated with low levels of AKT1 protein but not with changes in AKT1 transcript expression (Supplementary Information, Fig. S6
online). Thus we suggest that the low levels of protein in the carriers of the rs1130233
A allele are likely caused by a post-translational modification, i.e., degradation and turnover rate of the protein, perhaps related to its association with scaffold proteins such as 14-3-3 and HSP90 etc 
. According to the finding by Harris et al 
, it seems likely that AKT1 rs1130233
itself is not a functional SNP, since its association with AKT1 protein levels is not found in African Americans, who show very low LD in this genomic region. Therefore they suggested that another relatively common SNP or combination of SNPs, causes the post-translational reduction of the protein. However, our replication of Harris et al's report confirm that rs1130233
can serve as a surrogate marker for protein levels in populations with high LD in this region, such as Caucasians.
Since COMT, AKT1, and ErbB-signaling are each implicated in both cancer and schizophrenia, NRG1-ErbB signaling in B lymphoblasts provides a biologically plausible research tool for elucidating cellular mechanisms relevant to both cancer biology and neurobiology. Using this system, we demonstrated epistatic effects of COMT Val/Met and AKT1 rs1130233
on AKT1 activation. The effects were confirmed and some of the potential mechanisms explored using COMT transfection of cell lines. Effects of COMT Val/Met in cancer biology have almost always been discussed in the context of its role in detoxifying carcinogens and estrogens. Similarly, in neurobiology, the COMT Val/Met polymorphism has most often been linked with cortical dopamine metabolism. The present findings offer an additional facet to COMT's biology that has new, broad and potentially important implications for cancer biology and schizophrenia via its interaction with AKT1 (Supplementary Information, Fig. S7
online). AKT1 signaling mediates a number of critical neuronal signaling molecular pathways that are implicated in schizophrenia and also plays a central role in tumorigenesis in many cancers. If the potential for an AKT1x COMT gene-gene interaction is taken into consideration in future association studies, we would predict that the minor A allele of AKT1 rs1130233
might strengthen associations between COMT Val/Met and the multiple phenotypes with which it has been associated, including cognitive function, schizophrenia and cancer. Consistent with this prediction, a recent report highlighted epistatic interactions of COMT Val/Met and AKT1 rs1130233
on cortical physiology assayed with neuroimaging and on cognitive functions 
. Thus, future studies of epistatic interactions between genes such as COMT and AKT1 might help in elucidating the relationship between schizophrenia and cancer that has been discussed in epidemiological studies