D J Min S Vural
D.-J. Min, S. Vural and J. Krushkal
gene expression may be affected by the genetic background of cancer cells, e.g. by the presence of KRAS mutations which may elevate OCM activity , by the LKB1 and PKCζ MG 132 status and p53 mutation status  or by other sequence variants associated with OCM gene expression .
Our analysis of cell line response to crizotinib suggested that OCM gene expression was associated with response to crizotinib independently from the presence or absence of ge-nomic alterations affecting sensitivity to that agent. This anal-ysis combined the data across different cancer categories, as the number of genomic alterations was low and the cell lines carrying these alterations belonged to different histolo-gies (non-small cell-lung cancer, stomach adenocarcinoma, and breast cancer). Future studies of specific cancer cate-gories with sufficiently large sample sizes may be able to ex-amine the effect of OCM gene expression on drug response in specific cancer types, while accounting for functional genome alterations and expression of target kinase genes.
Although the observed associations were modest, it may be appropriate to investigate whether pretreatment levels of OCM genes could be of potential use for selection of associated agents in Table 1 for cancer treatment, possibly in combination with additional information about cancer-related genomic alterations in non-OCM genes that would further assist in therapy choices. While higher baseline expression levels of many genes presented in Table 1 and in Fig. 3, e.g. GART, TYMS, SHMT2, MTR, BHMT, MAT2B, and MTHFD2, have been associated with increased in vitro sensitivity to multiple agents in our study and in other reports , a number of clinical studies have reported an association of elevated expression of these genes with poorer patient survival [69,88–92]. Some of such clinical associations involved expression changes in OCM genes in response to therapy  as opposed to pretreatment transcriptional levels examined in this study; however, other studies in-volved treatment-naïve patients . Some pretreatment differences in OCM gene expression in the clinical setting may be associated with cancer subtypes, which had been previously demonstrated in ALL patients . There could also be additional explanations for the differences between the associations of OCM gene expression with in vitro drug response and with clinical response. In addition to possible physiological effects of clinical dosing, drug metabolism, toxicity, and patient immune response, there are possible molecular explanations including known OCM-related drug response differences and cellular vulnerabilities between in vitro cancer cell lines and proliferating in vivo tumors that are related to the differences between the nutrient content of the tumor microenvironment and of tissue culture media . As an example of such differences, the tumor microen-vironment has low serine availability. In contrast, cultured cancer cells provide one-carbon units to the OCM cycle from serine but not from glycine, and they also lack homocysteine remethylation, due to the presence of excess methionine and the absence of cobalamine in standard tissue culture media, which also has other alterations in the nutrient content when compared to body fluids [20,120]. If the direction of in vitro associations with drug sensitivity observed in our study and by others  is validated by future studies, additional in vivo studies would be needed to confirm the clinical utility of these correlations. To examine the validity of these molecular associations in vivo, it may be useful to use mouse
PDX models to examine treatment response of tumors with increased or decreased levels of expression of specific OCM genes. It may be appropriate to use such models to examine associations of OCM gene expression levels prior to treatment and during the course of treatment with various measures of in vivo response to agents listed in Table 1, including tumor volume, rates of growth, and other measures of disease progression. If the results of mouse PDX studies are consistent with findings from cancer cell line screens, further clinical analyses of pretreatment expression levels of OCM genes in patients from specific cancer groups and of changes in gene expression during treatment may assist with determining whether any genes reported in this study may be used as potential biomarkers for treatment decisions.
In summary, we found an association of expression lev-els of ten OCM genes with chemosensitivity or chemore-sistance to multiple cancer treatment agents in cancer cell lines. Among these genes, elevated expression of SLC46A1 was consistently associated with chemoresistance to a vari-ety of agents, and that of GART, TYMS, SHMT2, MTR, BHMT, and MAT2B was associated with chemosensitivity to multiple drugs. Although the detected correlations were modest, our findings indicated that pretreatment expression levels of OCM components may be directly or indirectly associated with re-sponse of a number of antitumor agents with diverse mech-anisms of action. If validated in vivo, these associations may need to be taken into account when considering cancer treat-ment regiments and drug combination strategies.