• 2019-10
  • 2019-11
  • 2020-03
  • 2020-07
  • 2020-08
  • 23214-92-8 br Results br Association of OCM


    Association of OCM gene 23214-92-8 with drug response
    We investigated whether cancer drug response was asso-ciated with one-carbon metabolism processes by analyzing Pearson correlation between baseline OCM gene expression and log(IC50) values of cancer cell line response to drug treat-ment. Table 1 provides the list of genes which were associated with sensitivity to antitumor agents (Pearson r ≥ 0.3, FDR-adjusted p < 0.05). Ten genes, GART, TYMS, SHMT2, MTR, ALDH2, BHMT, MAT2B, MTHFD2, NNMT, and SLC46A1, satisfied these criteria. Fig. 1 illustrates several top corre-lations from Table 1. Pretreatment expression levels of 8 of
    Fig. 1 Scatterplots of selected top correlations of gene expression with drug response from Table 1 satisfying r ≥ 0.3 and FDR adjusted p 23214-92-8 < 0.05. Horizontal axis represents log2 of cell line gene expression from the CCLE microarray expression dataset; vertical axis represents log(IC50) of drug response. (A) SHMT2 vs methotrexate; (B) SHMT2 vs XMD14-99; (C) TYMS vs GSK1070916; (D) TYMS vs BX-912; (D) GART vs methotrexate; (E) MAT2B vs crizotinib.
    the 10 genes listed in Table 1, GART, TYMS, SHMT2, MTR, BHMT, MAT2B, NNMT, and SLC46A1, were each correlated with response to multiple drugs. Even though these corre-lations were highly significant (FDR adjusted p < 0.0001 for all associations in Table 1), they were relatively weak ( r ≤ 0.403).
    Methotrexate, an antifolate drug targeting DHFR (dihydro-folate reductase), had the strongest association (r = −0.403 for correlation with SHMT2 expression; Fig. 1A). This result was partially validated in the NCI-60 dataset, in which sensitiv-ity to methotrexate was also strongly associated with SHMT2 expression when using Pearson correlation in all available cell lines, which was statistically significant (r = −0.456, FDR ad-justed p = 0.0017). However, while there was a considerable difference in sensitivity between the six highest and the six lowest SHMT2-expressing cell lines, this difference was not statistically significant (Cohen’s d = −1.0752, FDR adjusted p = 0.2104; Supplementary Table 2).
    In the GDSC–CCLE dataset, sensitivity to methotrexate was also associated (r = −0.362, −0.335, and 0.325) with el-evated expression of GART (the phosphoribosylglycinamide formyltransferase gene), TYMS (encoding thymidylate syn-thase), and NNMT (the nicotinamide N-methyltransferase gene), respectively (Table 1 and Fig. 1E). The log(IC50) val-ues of methotrexate also showed a weak, but statistically sig-nificant (r = −0.245, p = 1.96 × 10−7) correlation with expres-sion of the DHFR gene, which encodes the target of this agent. These results suggest that cell sensitivity to this antifolate agent may be influenced by expression of several compo-nents of the OCM pathway. While association of GART ex-pression with sensitivity to methotrexate was not significant in the NCI-60 panel (Supplementary Table 2), the observed correlation of increased expression of several OCM genes with methotrexate sensitivity in the GDSC-CCLE dataset was consistent with an earlier qualitative analysis of a subset of GDSC tumor cell lines and of patient samples, which had found a statistically significant association of elevated base-line SHMT2, GART, TYMS, and DHFR expression in tu-mor cell lines with methotrexate sensitivity and of SHMT2 expression in acute lymphoblastic leukemia (ALL) patients with response to methotrexate treatment [12]. This associ-ation of increased baseline expression of OCM genes with drug sensitivity is notable and is in contrast to overexpres-sion of DHFR and TYMS that tumors develop in response to treatment, which serves as a mechanism of acquired re-sistance to anti-folate drugs methotrexate, pemetrexed, and 5-fluorouracil [20,69].
    Increased expression of SLC46A1, encoding a proton-coupled folate transporter, was weakly associated with higher levels of resistance to 9 agents including BIX02189, CHIR-99021, crizotinib, cyclopamine, KIN001-260, QL-XII-61, suni-tinib, TL-1-85, and XMD8-85 (r between 0.301 and 0.330; Table 1). The product of this gene participates in the transport of folates and antifolates. Hypermethylation of the SLC46A1 promoter was previously found to be associated with HeLa cell resistance to methotrexate [70]. Agents associated with SLC46A1 expression belong to classes other than antifolate drugs (Table 1). It could be possible that an association of elevated expression of this gene with increased resistance to several agents could be an indirect consequence of an in-creased folate transport to cancer cells.  27
    While the overwhelming majority of the 34 genes exam-ined in our study had unimodal patterns of expression, the distribution of expression levels of NNMT, ALDH2, and CBS in the GDSC-CCLE dataset was bimodal. Fig. 2 shows the bimodal pattern of NNMT expression and unimodal expres-sion of SLC46A1, GART, TYMS, SHMT2, and MTR. Fig. 2C indicates that many cell lines had no or very low levels of NNMT expression, whereas the expression of this gene in other cell lines was at considerable levels (Fig. 2C). The bi-modality of NNMT gene expression and of protein expression and activity of NNMT in human liver has been well estab-lished, and multiple studies have suggested that differences in NNMT expression among patients affect metabolic rates, ther-apeutic response to treatment, and drug toxicity [71–73]. Our analysis using CCLE legacy online portal tools [23] showed that while average NNMT expression values differed among cancer categories, with very low or no expression of that gene observed in most leukemia and multiple myeloma cell lines, the majority of other cancer categories included both high and low NNMT expressing individual cell lines (data not shown).