• 2022-09
  • 2022-08
  • 2022-07
  • 2022-05
  • 2022-04
  • 2021-03
  • 2020-08
  • 2020-07
  • 2020-03
  • 2019-11
  • 2019-10
  • 2019-09
  • 2019-08
  • 2019-07
  • br We evaluated possible joint effects of OCM genes on


    We evaluated possible joint effects of OCM genes on drug sensitivity and a possible co-regulation among OCM genes by examining Pearson correlation among Erastin levels of the 34 OCM genes. In addition, we examined possible molecular mechanisms that could influence the associations of OCM gene expression with drug response by analyzing Pearson correlations between OCM gene expression and expression levels of molecular targets and components of tar-get pathways for the 51 agents listed in Table 1. These gene targets and components of drug target pathways are listed in Supplementary Table 1. Information about gene targets and classification of target pathways for each agent was obtained from the GDSC data download site [25] (Table 1), with additional information about gene targets and components of target pathways collected from the DrugBank [38,39], Erastin Cancer Therapeutics Response Portal v. 2 at the Broad Institute [40,41], Drug SIGnatures DataBase (DSigDB) v. 1.0, [42,43], PharmGKB [44,45], PubChem [46], online resources from drug suppliers and manufacturers including Selleck Chemicals [47], Tocris [48], Aphios [49], and AstraZeneca [50], and original publications [51–66]. The full list of gene targets and target pathway components, which also included several OCM genes, is listed in Supplementary Table 1. Gene expression information was available from the CCLE dataset
    [22] for 532 of these additional target pathway components that were not already a part of the list of 34 candidate OCM genes. To examine possible associations of c-Myc and HIF-1 (hypoxia-inducible factor 1) transcriptional levels and OCM gene expression, we also analyzed Pearson correlations of expression levels of the MYC and HIF1A genes with baseline expression of the 34 OCM genes in untreated cells. FDR adjustment of the p-values derived from the analysis of Pearson correlations of OCM gene expression with other OCM genes and with target pathway components from the top 51 significantly associated agents listed in Table 1 and in Supplementary Table 1 accounted for the comparisons involving 567 genes that included HIF1A, 34 OCM genes, and 532 additional genes (including MYC) that had been reported to be involved in drug target pathways. Because analysis of correlations among log2-transformed RMA-normalized gene expression levels did not involve drug response measures, it was not restricted to the 635 cell lines with matching GDSC drug response data. This analysis was based on a broader dataset of 1036 cell lines with Affymetrix Human Genome U133 Plus 2.0 microarray data, available from CCLE [22].
    Information about chromosomal assignment and genome locations of candidate genes was obtained from GeneCards [67,68].  25
    Regression analysis of association of cell line response to kinase inhibitors with OCM gene expression and mutational status of molecular drug targets
    Among the top agents associated with OCM gene expression (Table 1; Supplementary Table 1), sensitivity and resistance to three kinase inhibitors, erlotinib, lapatinib, and crizotinib, have been previously associated with a number of well-defined ge-nomic alterations and DNA and protein sequence changes in their molecular target genes. We examined which of the genomic alterations known to affect sensitivity or resistance were present in the cell lines with available response data to these three agents. Gene amplification and fusion status was determined using the data from the CCLE web resource of the Broad Institute [22]. Gene level copy number data had been generated by the CCLE Consortium using Affymetrix 6.0 SNP arrays, with segmentation of normalized log2 ratios of the copy number estimates performed using the circular binary segmentation algorithm [22]. We defined gene amplifi-cation status as ≥ 5 copies of a gene (i.e., log2 of normalized values for a diploid genome ≥ 1.322). Data on DNA sequence variants resulting in protein sequence changes were obtained from the whole exome sequencing information downloaded from the GDSC online resource [25].
    Among the three agents, the data for more than one cell line with genomic alterations affecting sensitivity were avail-able only for crizotinib response. We examined four folate genes (MAT2B, MTR, SLC46A1, and SHMT2), expression of which was significantly correlated with log(IC50) of crizotinib (Table 1). We used multiple regression to investigate whether the expression of each of these genes remained a signifi-cant predictor of cancer cell line response to crizotinib af-ter accounting for the presence of genomic alterations with known effects on crizotinib sensitivity and resistance. Expres-sion of each of the four OCM genes, MAT2B, MTR, SLC46A1, and SHMT2, was analyzed separately, and their regression p-values were adjusted for multiple testing using the FDR ap-proach. We used expression of each folate gene, the pres-ence of a genomic alteration with reported effects on crizotinib sensitivity, and the presence of a genomic alteration with re-ported effects on crizotinib resistance as dependent variables in linear regression modeling, whereas log(IC50) of crizotinib was used as the outcome variable.