Search

This file contains Supplementary Figure S1 (Correlation between downregulation of EGFR transcriptional target genes, and sensitivity to erlotinib at the high concentration after 6 hours); Supplementary Figure S2 (EGFR inhibitor and DNA damaging agents induce directionally opposing changes in EGR1 expression that are differentially associated with drug sensitivity); Supplementary Figure S3 (Representative 3D response surface plots for the gemcitabine-erlotinib combination in the indicated cell lines); Supplementary Figure S4 (Representative 3D response surface plots for the vorinostat-talazoparib combination in the indicated cell lines); Supplementary Table S1 (Validation of array-based measurements by qRT-PCR); Supplementary Table S2 (Genes with concerted changes in the same direction [upregulation or downregulation] 24 hours after treatment with each of the 15 drugs); and Supplementary Table S4 (Selected genes potentially mediating drug insensitivity to agents in the NCI TPW dataset).

The intracellular effects and overall efficacies of anticancer therapies can vary significantly by tumor type. To identify patterns of drug-induced gene modulation that occur in different cancer cell types, we measured gene-expression changes across the NCI-60 cell line panel after exposure to 15 anticancer agents. The results were integrated into a combined database and set of interactive analysis tools, designated the NCI Transcriptional Pharmacodynamics Workbench (NCI TPW), that allows exploration of gene-expression modulation by molecular pathway, drug target, and association with drug sensitivity. We identified common transcriptional responses across agents and cell types and uncovered gene-expression changes associated with drug sensitivity. We also demonstrated the value of this tool for investigating clinically relevant molecular hypotheses and identifying candidate biomarkers of drug activity. The NCI TPW, publicly available at https://tpwb.nci.nih.gov, provides a comprehensive resource to facilitate understanding of tumor cell characteristics that define sensitivity to commonly used anticancer drugs.Significance:The NCI Transcriptional Pharmacodynamics Workbench represents the most extensive compilation to date of directly measured longitudinal transcriptional responses to anticancer agents across a thoroughly characterized ensemble of cancer cell lines.

Background Parent of origin-specific allelic expression of imprinted genes is epigenetically controlled. In cancer, imprinted genes undergo both genomic and epigenomic alterations, including frequent copy number changes. We investigated whether copy number loss or gain of imprinted genes in cancer cell lines is associated with response to chemotherapy treatment. Results We analyzed 198 human imprinted genes including protein-coding genes and noncoding RNA genes using data from tumor cell lines from the Cancer Cell Line Encyclopedia and Genomics of Drug Sensitivity in Cancer datasets. We examined whether copy number of the imprinted genes in 35 different genome locations was associated with response to cancer drug treatment. We also analyzed associations of pretreatment expression and DNA methylation of imprinted genes with drug response. Higher copy number of BLCAP, GNAS, NNAT, GNAS-AS1, HM13, MIR296, MIR298, and PSIMCT-1 in the chromosomal region 20q11-q13.32 was associated with resistance to multiple antitumor agents. Increased expression of BLCAP and HM13 was also associated with drug resistance, whereas higher methylation of gene regions of BLCAP, NNAT, SGK2, and GNAS was associated with drug sensitivity. While expression and methylation of imprinted genes in several other chromosomal regions was also associated with drug response and many imprinted genes in different chromosomal locations showed a considerable copy number variation, only imprinted genes at 20q11-q13.32 had a consistent association of their copy number with drug response. Copy number values among the imprinted genes in the 20q11-q13.32 region were strongly correlated. They were also correlated with the copy number of cancer-related non-imprinted genes MYBL2, AURKA, and ZNF217 in that chromosomal region. Expression of genes at 20q11-q13.32 was associated with ex vivo drug response in primary tumor samples from the Beat AML 1.0 acute myeloid leukemia patient cohort. Association of the increased copy number of the 20q11-q13.32 region with drug resistance may be complex and could involve multiple genes. Conclusions Copy number of imprinted and non-imprinted genes in the chromosomal region 20q11-q13.32 was associated with cancer drug resistance. The genes in this chromosomal region may have a modulating effect on tumor response to chemotherapy.

Natural products provide new opportunities for anticancer chemotherapeutics. We examined the associations of molecular features in the NCI‐60 cancer cell lines with response to 1302 plant, marine, and microbial compounds. Expression or mutations in multiple genes were associated with treatment responses, suggesting potential mechanisms of action of natural compounds. This information will assist in future design of new chemotherapy agents., Natural products remain a significant source of anticancer chemotherapeutics. The search for targeted drugs for cancer treatment includes consideration of natural products, which may provide new opportunities for antitumor cytotoxicity as single agents or in combination therapy. We examined the association of molecular genomic features in the well‐characterized NCI‐60 cancer cell line panel with in vitro response to treatment with 1302 small molecules which included natural products, semisynthetic natural product derivatives, and synthetic compounds based on a natural product pharmacophore from the Developmental Therapeutics Program of the US National Cancer Institute's database. These compounds were obtained from a variety of plant, marine, and microbial species. Molecular information utilized for the analysis included expression measures for 23059 annotated transcripts, lncRNAs, and miRNAs, and data on protein‐changing single nucleotide variants in 211 cancer‐related genes. We found associations of expression of multiple genes including SLFN11, CYP2J2, EPHX1, GPC1, ELF3, and MGMT involved in DNA damage repair, NOTCH family members, ABC and SLC transporters, and both mutations in tyrosine kinases and BRAF V600E with NCI‐60 responses to specific categories of natural products. Hierarchical clustering identified groups of natural products, which correlated with a specific mechanism of action. Specifically, several natural product clusters were associated with SLFN11 gene expression, suggesting that potential action of these compounds may involve DNA damage. The associations between gene expression or genome alterations of functionally relevant genes with the response of cancer cells to natural products provide new information about potential mechanisms of action of these identified clusters of compounds with potentially similar biological effects. This information will assist in future drug discovery and in design of new targeted cancer chemotherapy agents.

PURPOSE: Time is a critical factor in drug action. The duration of inhibition of the target or residence time of the drug molecule on the target often guides drug scheduling. METHODS: The effects of time on the concentration-dependent cytotoxicity of approved and investigational agents [300 compounds] were examined in the NCI60 cell line panel in 2D at 2, 3, 7 and in 3D 11 days. RESULTS: There was a moderate positive linear relationship between data from the 2-day NCI60 screen and the 3-, 7- and 11-day and a strong positive linear relationship between 3-, 7- and 11-day luminescence screen IC(50)s by Pearson correlation analysis. Cell growth inhibition by agents selective for a specific cell cycle phase plateaued when susceptible cells were growth inhibited or killed. As time increased the depth of cell growth inhibition increased without change in the IC(50). DNA interactive agents had decreasing IC(50)s with increasing exposure time. Epigenetic agents required longer exposure times; several were only cytotoxic after 11 days’ exposure. For HDAC inhibitors, time had little or no effect on concentration response. There were potency differences amongst the three BET bromodomain inhibitors tested, and an exposure duration effect. The PARP inhibitors, rucaparib, niraparib, and veliparib reached IC(50)s < 10 μM in some cell lines after 11 days. CONCLUSIONS: The results suggest that variations in compound exposure time may reflect either mechanism of action or com-pound chemical half-life. The activity of slow-acting compounds may optimally be assessed in spheroid models that can be monitored over prolonged incubation times.

Background Cancer treatment is increasingly dependent on biomarkers for prognostication and treatment selection. Potential biomarkers are frequently evaluated in prospective-retrospective studies in which biomarkers are measured retrospectively on archived specimens after completion of prospective clinical trials. In light of the high costs of some assays, random sampling designs have been proposed that measure biomarkers for a random sub-sample of subjects selected on the basis of observed outcome and possibly other variables. Compared with a standard design that measures biomarkers on all subjects, a random sampling design can be cost-efficient in the sense of reducing the cost of the study substantially while achieving a reasonable level of precision. Methods For a biomarker that indicates the presence of some molecular alteration (e.g., mutation in a gene), we explore the use of a group testing strategy, which involves physically pooling specimens across subjects and assaying pooled samples for the presence of the molecular alteration of interest, for further improvement in cost-efficiency beyond random sampling. We propose simple and general approaches to estimating the prognostic and predictive values of biomarkers with group testing, and conduct simulation studies to validate the proposed estimation procedures and to assess the cost-efficiency of the group testing design in comparison to the standard and random sampling designs. Results Simulation results show that the proposed estimation procedures perform well in realistic settings and that a group testing design can have considerably higher cost-efficiency than a random sampling design. Conclusions Group testing can be used to improve the cost-efficiency of biomarker studies.

SUMMARY CellMiner-SCLC (https://discover.nci.nih.gov/SclcCellMinerCDB/) integrates drug sensitivity and genomic data, including high-resolution methylome and transcriptome from 118 patient-derived small cell lung cancer (SCLC) cell lines, providing a resource for research into this “recalcitrant cancer.” We demonstrate the reproducibility and stability of data from multiple sources and validate the SCLC consensus nomenclature on the basis of expression of master transcription factors NEUROD1, ASCL1, POU2F3, and YAP1. Our analyses reveal transcription networks linking SCLC subtypes with MYC and its paralogs and the NOTCH and HIPPO pathways. SCLC subsets express specific surface markers, providing potential opportunities for antibody-based targeted therapies. YAP1-driven SCLCs are notable for differential expression of the NOTCH pathway, epithelial-mesenchymal transition (EMT), and antigen-presenting machinery (APM) genes and sensitivity to mTOR and AKT inhibitors. These analyses provide insights into SCLC biology and a framework for future investigations into subtype-specific SCLC vulnerabilities., In Brief Tlemsani et al. provide a unique resource, SCLC-CellMiner, integrating drug sensitivity and multi-omics data from 118 small cell lung cancer (SCLC) cell lines. They demonstrate that SCLCs have differential transcriptional networks driven by lineage-specific transcription factors (NEUROD1, ASCL1, POU2F3, and YAP1). Furthermore, YAP1-driven SCLCs have distinct drug sensitivity profiles., Graphical Abstract

Background Altered DNA methylation patterns play important roles in cancer development and progression. We examined whether expression levels of genes directly or indirectly involved in DNA methylation and demethylation may be associated with response of cancer cell lines to chemotherapy treatment with a variety of antitumor agents. Results We analyzed 72 genes encoding epigenetic factors directly or indirectly involved in DNA methylation and demethylation processes. We examined association of their pretreatment expression levels with methylation beta-values of individual DNA methylation probes, DNA methylation averaged within gene regions, and average epigenome-wide methylation levels. We analyzed data from 645 cancer cell lines and 23 cancer types from the Cancer Cell Line Encyclopedia and Genomics of Drug Sensitivity in Cancer datasets. We observed numerous correlations between expression of genes encoding epigenetic factors and response to chemotherapeutic agents. Expression of genes encoding a variety of epigenetic factors, including KDM2B, DNMT1, EHMT2, SETDB1, EZH2, APOBEC3G, and other genes, was correlated with response to multiple agents. DNA methylation of numerous target probes and gene regions was associated with expression of multiple genes encoding epigenetic factors, underscoring complex regulation of epigenome methylation by multiple intersecting molecular pathways. The genes whose expression was associated with methylation of multiple epigenome targets encode DNA methyltransferases, TET DNA methylcytosine dioxygenases, the methylated DNA-binding protein ZBTB38, KDM2B, SETDB1, and other molecular factors which are involved in diverse epigenetic processes affecting DNA methylation. While baseline DNA methylation of numerous epigenome targets was correlated with cell line response to antitumor agents, the complex relationships between the overlapping effects of each epigenetic factor on methylation of specific targets and the importance of such influences in tumor response to individual agents require further investigation. Conclusions Expression of multiple genes encoding epigenetic factors is associated with drug response and with DNA methylation of numerous epigenome targets that may affect response to therapeutic agents. Our findings suggest complex and interconnected pathways regulating DNA methylation in the epigenome, which may both directly and indirectly affect response to chemotherapy.

Background Small cell lung cancer (SCLC) is an aggressive neuroendocrine lung cancer. SCLC progression and treatment resistance involve epigenetic processes. However, links between SCLC DNA methylation and drug response remain unclear. We performed an epigenome-wide study of 66 human SCLC cell lines using the Illumina Infinium MethylationEPIC BeadChip array. Correlations of SCLC DNA methylation and gene expression with in vitro response to 526 antitumor agents were examined. Results We found multiple significant correlations between DNA methylation and chemosensitivity. A potentially important association was observed for TREX1, which encodes the 3′ exonuclease I that serves as a STING antagonist in the regulation of a cytosolic DNA-sensing pathway. Increased methylation and low expression of TREX1 were associated with the sensitivity to Aurora kinase inhibitors AZD-1152, SCH-1473759, SNS-314, and TAK-901; the CDK inhibitor R-547; the Vertex ATR inhibitor Cpd 45; and the mitotic spindle disruptor vinorelbine. Compared with cell lines of other cancer types, TREX1 had low mRNA expression and increased upstream region methylation in SCLC, suggesting a possible relationship with SCLC sensitivity to Aurora kinase inhibitors. We also identified multiple additional correlations indicative of potential mechanisms of chemosensitivity. Methylation of the 3′UTR of CEP350 and MLPH, involved in centrosome machinery and microtubule tracking, respectively, was associated with response to Aurora kinase inhibitors and other agents. EPAS1 methylation was associated with response to Aurora kinase inhibitors, a PLK-1 inhibitor and a Bcl-2 inhibitor. KDM1A methylation was associated with PLK-1 inhibitors and a KSP inhibitor. Increased promoter methylation of SLFN11 was correlated with resistance to DNA damaging agents, as a result of low or no SLFN11 expression. The 5′ UTR of the epigenetic modifier EZH2 was associated with response to Aurora kinase inhibitors and a FGFR inhibitor. Methylation and expression of YAP1 were correlated with response to an mTOR inhibitor. Among non-neuroendocrine markers, EPHA2 was associated with response to Aurora kinase inhibitors and a PLK-1 inhibitor and CD151 with Bcl-2 inhibitors. Conclusions Multiple associations indicate potential epigenetic mechanisms affecting SCLC response to chemotherapy and suggest targets for combination therapies. While many correlations were not specific to SCLC lineages, several lineage markers were associated with specific agents.

SummaryModel systems are necessary to understand the biology of SCLC and develop new therapies against this recalcitrant disease. Here we provide the first online resource, CellMiner-SCLC (https://discover.nci.nih.gov/SclcCellMinerCDB) incorporating 118 individual SCLC cell lines and extensive omics and drug sensitivity datasets, including high resolution methylome performed for the purpose of the current study. We demonstrate the reproducibility of the cell lines and genomic data across the CCLE, GDSC, CTRP, NCI and UTSW datasets. We validate the SCLC classification based on four master transcription factors: NEUROD1, ASCL1, POU2F3 and YAP1 (NAPY classification) and show transcription networks connecting each them with their downstream and upstream regulators as well as with the NOTCH and HIPPO pathways and the MYC genes (MYC, MYCL1 and MYCN). We find that each of the 4 subsets express specific surface markers for antibody-targeted therapies. The SCLC-Y cell lines differ from the other subsets by expressing the NOTCH pathway and the antigen-presenting machinery (APM), and responding to mTOR and AKT inhibitors. Our analyses suggest the potential value of NOTCH activators, YAP1 inhibitors and immune checkpoint inhibitors in SCLC-Y tumors that can now be independently validated.Graphical AbstractHighlightsSCLC-CellMiner provides the most extensive SCLC resource in terms of number of cell lines (118 cell lines), extensive omics data (exome, microarray, RNA-seq, copy number, methylomes and microRNA) and drug sensitivity testing.We find evidence of distinct epigenetic profile of SCLC cell lines (global hypomethylation and histone gene methylation), which is consistent with their plasticity.Transcriptome analyses demonstrate the coherent transcriptional networks associated with the 4 main genomic subgroups (NEUROD1, ASCL1, POU2F3 & YAP1 = NAPY classification) and their connection with the NOTCH and HIPPO signaling pathways.SCLC-CellMiner provides a conceptual framework for the selection of therapies for SCLC in a personalized fashion allowing putative biomarkers according molecular classifications and molecular characteristics.SCLC-Y cell lines differ from the other cancer cell lines; their transcriptome resemble NSCLC cell lines. YAP1 cell lines while being the most resistant to standard of care treatments (etoposide, cisplatin and topotecan) respond to mTOR and AKT inhibitors and present native immune predisposition suggesting sensitivity to immune checkpoint inhibitors.

The SCLC combination screen examined a 9‐point concentration response of 180 third agents, alone and in combination with etoposide/carboplatin. The predominant effect of adding a third agent to etoposide/carboplatin was additivity. Less than additive effects occurred frequently in SCLC lines sensitive to etoposide/carboplatin. In SCLC lines with little or no response to etoposide/carboplatin, greater than additive SCLC killing occurred over the entire spectrum of SCLC lines but never occurred in all SCLC lines. Exposing SCLC lines to tubulin‐targeted agents (paclitaxel or vinorelbine) simultaneously with etoposide/carboplatin resulted primarily in less than additive cell killing. As single agents, nuclear kinase inhibitors including Aurora kinase inhibitors, Kinesin Spindle Protein/EG5 inhibitors, and Polo‐like kinase‐1 inhibitors were potent cytotoxic agents in SCLC lines; however, simultaneous exposure of the SCLC lines to these agents along with etoposide/carboplatin, generally, resulted in less than additive cell killing. Several classes of agents enhanced the cytotoxicity of etoposide/carboplatin toward the SCLC lines. Exposure of the SCLC lines to the MDM2 inhibitor JNJ‐27291199 produced enhanced killing in 80% of the SCLC lines. Chk‐1 inhibitors such as rabusertib increased the cytotoxicity of etoposide/carboplatin to the SCLC lines in an additive to greater than additive manner. The combination of GSK‐3β inhibitor LY‐2090314 with etoposide/carboplatin increased killing in approximately 40% of the SCLC lines. Exposure to the BET bromodomain inhibitor MK‐8628 increased the SCLC cell killing by etoposide/carboplatin in 20–25% of the SCLC lines. Only 10–15% of the SCLC lines had an increased response to etoposide/carboplatin when simultaneously exposed to the PARP inhibitor talazoparib.

Model systems are necessary to understand the biology of SCLC and develop new therapies against this recalcitrant disease. Here we provide the first online resource, CellMiner-SCLC (https://discover.nci.nih.gov/sclcCellMinerCDB) incorporating 118 individual SCLC cell lines and extensive omics and drug sensitivity datasets, including high resolution methylome performed for the purpose of the current study. We demonstrate the reproducibility of the cell lines and genomic data across the CCLE, GDSC, CTRP, NCI and UTSW datasets. We validate the SCLC classification based on four master transcription factors: NEUROD1, ASCL1, POU2F3 and YAP1 (NAPY classification) and show transcription networks connecting each them with their downstream and upstream regulators as well as with the NOTCH and HIPPO pathways and the MYC genes (MYC, MYCL1 and MYCN). We find that each of the 4 subsets express specific surface markers for antibody-targeted therapies. The SCLC-Y cell lines differ from the other subsets by expressing the NOTCH pathway and the antigen-presenting machinery (APM), and responding to mTOR and AKT inhibitors. Our analyses suggest the potential value of NOTCH activators, YAP1 inhibitors and immune checkpoint inhibitors in SCLC-Y tumors that can now be independently validated.

In recent years, cancer immunotherapy has emerged as a highly promising approach to treat patients with cancer, as the patient's own immune system is harnessed to attack cancer cells. However, the application of these approaches is still limited to a minority of patients with cancer and it is difficult to predict which patients will derive the greatest clinical benefit.One of the challenges faced by the biomedical community in the search of more effective biomarkers is the fact that translational research efforts involve collecting and accessing data at many different levels: from the type of material examined (e.g., cell line, animal models, clinical samples) to multiple data type (e.g., pharmacodynamic markers, genetic sequencing data) to the scale of a study (e.g., small preclinical study, moderate retrospective study on stored specimen sets, clinical trials with large cohorts).This chapter reviews several publicly available bioinformatics tools and data resources for high throughput molecular analyses applied to a range of data types, including those generated from microarray, whole-exome sequencing (WES), RNA-seq, DNA copy number, and DNA methylation assays, that are extensively used for integrative multidimensional data analysis and visualization.

In recent years, cancer immunotherapy has emerged as a highly promising approach to treat patients with cancer, as the patient's own immune system is harnessed to attack cancer cells. However, the application of these approaches is still limited to a minority of patients with cancer and it is difficult to predict which patients will derive the greatest clinical benefit.One of the challenges faced by the biomedical community in the search of more effective biomarkers is the fact that translational research efforts involve collecting and accessing data at many different levels: from the type of material examined (e.g., cell line, animal models, clinical samples) to multiple data type (e.g., pharmacodynamic markers, genetic sequencing data) to the scale of a study (e.g., small preclinical study, moderate retrospective study on stored specimen sets, clinical trials with large cohorts).This chapter reviews several publicly available bioinformatics tools and data resources for high throughput molecular analyses applied to a range of data types, including those generated from microarray, whole-exome sequencing (WES), RNA-seq, DNA copy number, and DNA methylation assays, that are extensively used for integrative multidimensional data analysis and visualization.

The typical low life expectancy and limited therapeutic options for patients with small cell lung cancer (SCLC) caused the National Cancer Institute (NCI) to categorize SCLC as “recalcitrant” cancer. SCLC-CellMiner (https://discover.nci.nih.gov/SclcCellMinerCDB) integrates drug sensitivity and genomic data from 118 patient-derived SCLC cell lines, providing a unique genomic and pharmacological resource. Transcriptomic profiling validates the SCLC consensus nomenclature based on expression of 4 master transcription factors NEUROD1, ASCL1, POU2F3 and YAP1 (NAPY classification) and demonstrate differential transcriptional networks driven by these lineage specific transcription factors. Our analyses reveal transcription networks linking SCLC subtypes with MYC and its paralogs MYCL and MYCN and inactivation of the NOTCH pathway in the neuroendocrine SCLC (N, A & P subgroups). By contrast, YAP1-driven SCLC (SCLC-Y) express the NOTCH pathway and co-express both YAP/TAZ and its negative regulator genes driving the Hippo pathway. SCLC-Y cell lines show the greatest resistance to the standard of care drugs (etoposide, cisplatin and topotecan) while PI3K-AKT-mTOR inhibitors show a higher activity in this subgroup. To explore the immune pathways and the potential value of the transciption factors based classification for selecting SCLC patients likely to respond to immune checkpoint inhibitors, we explored a transcriptome signature based on 18 established native immune response and antigen-presenting genes (APM score). The SCLC-Y cell lines are the only subset expressing innate immune response genes. SCLC-CellMiner is a powerfull tool demonstrating the value of cancer cell line genomic and pharmacological databases. Our analyses suggest the potential genomic molecular classifications to select patients for targeted therapies and immunotherapy, such as patients in the SCLC-Y subgroup who may be most responsive to immune checkpoints modulators. Citation Format: Camille Tlemsani, Lorinc Pongor, Fathi Elloumi, Luc Girard, Kenneth Huffman, Nitin Roper, Sudhir Varma, Augustin Luna, Vinodh Rajapakse, Pascaline Boudou-Rouquette, Robin Sebastian, Kurt Kohn, Julia Krushkal, Mirit Aladjem, Beverly Teicher, Paul Meltzer, William Reinhold, John Minna, Anish Thomas, Yves Pommier. SCLC-CellMiner: An extensive cell line genomic and pharmacology resource identifies a subgroup of small cell lung cancers sensitive to targeted therapies and immunotherapies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 203.

Sarcomas represent a heterogeneous group of cancers with many histological subtypes. Their prognosis remains poor and treatment is mainly based on palliative chemotherapy at metastatic stage. Tumor genome sequencing failed to identify recurrent somatic drivers while several oncogenic gene fusion-translocations have been identified in specific sarcoma subtypes. Because of the rarity and heterogeneity of sarcomas, only few sarcoma patients are enrolled in clinicals trials according their subtype. Increased translational research according subtypes of sarcoma patients is needed to improve patient management. To characterize the differences between subgroups and build translational hypotheses, we built a novel resource, Sarcoma-Cellminer, which integrates drug sensitivity and genomic data from 112 patient-derived sarcoma cell lines. These data will be available from a web-based tool (https://discover.nci.nih.gov/SclcCellMinerCDB/) derived from our CellMiner cross-database web application (https://discover.nci.nih.gov/cellminercdb). Among the 112 cell lines, 65 are bone sarcomas (including 38 Ewing sarcomas and 22 osteosarcomas), 45 are soft tissue sarcomas (including 21 rhabdomyosarcomas). Transcriptome (RNAseq and microarray), copy number, microRNA, genome-wide methylation, and drug sensitivity data are included and made publicly available. We also generated new genomic data including copy number and methylation (850 k) for 79 cell lines from the NCI in addition to the 42 cell lines from Broad Institute (CCLE) and the 40 cell lines from the MGH-Sanger (GDSC). We created the “SCLC-Global” expression set by regrouping all datasets by Z-score normalization, which enables cross-database analyses of gene expression and molecular pathways. Hierarchical clustering based on expression and methylation data identifies subtypes of sarcomas. Histone genes stand out suggesting that epigenetic regulation of canonical histones is a feature of sarcoma genesis. Sarcoma-CellMiner includes drug sensitivity data for over 500 different drugs tested in the NCI, CCLE and GDSC databases. They show two subgroups of Ewing sarcomas: one sensitive to PARP inhibitors and one resistant. Similarly, profile of response to dasatinib is different when comparing alveolar and embryonal rhabdomyosarcomas. Sarcoma-CellMiner is a powerful tool demonstrating the value of patient-derived cancer cell line databases. It provides hypothesis-driven rationale for using omics, especially transcriptome and epigenetic data to better understand sarcoma heterogeneity and select personalized treatments for clinical trials. Citation Format: Camille Tlemsani, Lorinc Pongor, Javed Khan, Fathi Elloumi, Sudhir Varma, Augustin Luna, Vinodh Rajapakse, Kurt Kohn, Julia Krushkal, Mirit Aladjem, Beverly Teicher, Paul Meltzer, William Reinhold, Christine Heske, Yves Pommier. Sarcoma-CellMiner: An extensive resource for patient-derived sarcoma cell line epigenetics, genomics and pharmacology [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 212.

Folate-mediated one-carbon metabolism is essential for growth and survival of cancer cells. We investigated whether the response of cancer cells to antitumor treatment may be partially influenced by variation in expression of one-carbon metabolism genes. We used cancer cell line information from the Cancer Cell Line Encyclopedia and the Genomics of Drug Sensitivity in Cancer resources to examine whether variation in pretreatment expression of one-carbon metabolism-related genes was associated with response to treatment. GART, TYMS, SHMT2, MTR, ALDH2, BHMT, MAT2B, MTHFD2, NNMT, and SLC46A1 showed modest statistically significant correlations with response to a variety of antitumor agents. Higher expression levels of SLC46A1 were associated with resistance to multiple agents, whereas elevated expression of GART, TYMS, SHMT2, MTR, BHMT, and MAT2B was associated with chemosensitivity to multiple drugs. NNMT expression was bimodally distributed and showed different directions of association with various agents. Correlation of increased NNMT expression with sensitivity to dasatinib was validated in the NCI-60 cancer cell line panel. Pretreatment expression levels were correlated among many one-carbon metabolism genes. Expression of several folate genes was strongly associated with expression of multiple components of drug target pathways. Molecular mechanisms underlying associations of one-carbon metabolism gene with drug response require further investigation.

The intracellular effects and overall efficacies of anticancer therapies can vary significantly by tumor type. To identify patterns of drug-induced gene modulation that occur in different cancer cell types, we measured gene-expression changes across the NCI-60 cell line panel after exposure to 15 anticancer agents. The results were integrated into a combined database and set of interactive analysis tools, designated the NCI Transcriptional Pharmacodynamics Workbench (NCI TPW), that allows exploration of gene-expression modulation by molecular pathway, drug target, and association with drug sensitivity. We identified common transcriptional responses across agents and cell types and uncovered gene-expression changes associated with drug sensitivity. We also demonstrated the value of this tool for investigating clinically relevant molecular hypotheses and identifying candidate biomarkers of drug activity. The NCI TPW, publicly available at https://tpwb.nci.nih.gov, provides a comprehensive resource to facilitate understanding of tumor cell characteristics that define sensitivity to commonly used anticancer drugs. Significance: The NCI Transcriptional Pharmacodynamics Workbench represents the most extensive compilation to date of directly measured longitudinal transcriptional responses to anticancer agents across a thoroughly characterized ensemble of cancer cell lines.

Cellular glycosylation processes are vital to cell functioning. In malignant cells, they are profoundly altered. We used time-course gene expression data from the NCI-60 cancer cell lines treated with 11 antitumor agents to analyze expression changes of genes involved in glycosylation pathways, genes encoding glycosylation targets or regulators, and members of cancer pathways affected by glycosylation. We also identified glycosylation genes for which pretreatment expression levels or changes after treatment were correlated with drug sensitivity. Their products are involved in N-glycosylation and O-glycosylation, fucosylation, biosynthesis of poly- N-acetyllactosamine, removal of misfolded proteins, binding to hyaluronic acid and other glycans, and cell adhesion. Tumor cell sensitivity to multiple agents was correlated with transcriptional response of C1GALT1C1, FUCA1, SDC1, MUC1; members of the MGAT, GALNT, B4GALT, B3GNT, MAN, and EDEM families; and other genes. These genes may be considered as potential candidates for drug targeting in combination therapy to enhance treatment response.

Small cell lung cancer (SCLC), an aggressive neuroendocrine type of lung cancer, rapidly acquires resistance to treatment. SCLC progression, lineage differentiation, and resistance to therapy have been suggested to involve epigenetic processes. To date, epigenetic links connecting SCLC DNA methylation patterns to drug response and the ways in which these links are mediated by gene expression remain unclear. In order to understand how DNA methylation may affect SCLC response to chemotherapy, we performed an epigenome-wide association study of 66 SCLC cell lines. We used Illumina Infinium MethylationEPIC BeadChip to measure methylation of 866,091 probes. We examined how methylation of probes and gene regions was associated with SCLC in vitro response to 526 antitumor agents. We also identified associations of epigenetic variation with drug response which may be mediated by regulation of gene expression. A potentially important strong association was observed for TREX1, which encodes the 3’ exonuclease I (DNase III) that is involved in resolution of chromatin bridges and has a potential role in chromothripsis. Increased methylation and low expression of TREX1 were associated with SCLC cell line sensitivity to multiple Aurora kinase inhibitors AZD-1152, SCH-1473759, SNS-314, and TAK-901, as well as to the CDK inhibitor R-547, Vertex ATR inhibitor Cpd 45, and the mitotic spindle disruptor vinorelbine. TREX1 upregulation has been previously associated with resistance of other cancers to DNA damaging agents and with DNA repair or DNA degradation after drug exposure. In our analysis, when compared to other cancer categories, TREX1 in SCLC cell lines had low mRNA expression and increased DNA methylation upstream of its transcription start site, which may provide a possible molecular mechanism for SCLC sensitivity to Aurora kinase inhibitors. CEP350 and MLPH, which are involved in centrosome machinery and microtubule tracking, were associated with several Aurora kinase inhibitors and other agents. Among other examples, EPAS1 (HIF2A) was associated with several Aurora kinase inhibitors, the PLK1 inhibitor GSK-461364, and the Bcl-2 inhibitor ABT-737. Methylation of KDM1A, encoding the histone modifier lysine demethylase 1A (LSD1), was associated with PLK1 inhibitors and the KSP inhibitor SB-743921. IGFBP5, which is expressed in the tuft cell-like SCLC subtype, was associated with the mTOR inhibitor INK-128. Upstream regions of MDM2 and DLL3, a Notch pathway regulator overexpressed in ASCL1-high SCLC tumors, were associated with Bcl-2 inhibitors. Methylation and expression of YAP1, a SCLC lineage driver regulating the Hippo pathway, were correlated with the MTOR inhibitor rapamycin. Among non-neuroendocrine lineage markers, EPHA2 was associated with Aurora kinase inhibitors and a PLK1 inhibitor, and CD151 with Bcl-2 inhibitors. Increased methylation upstream of SLFN11 was correlated with resistance to DNA damaging agents, which is likely mediated by SLFN11 expression. The 5’ UTR region of the epigenetic modifier EZH2 was associated with Aurora kinase inhibitors and the FGFR inhibitor BGJ-398. These and multiple other associations identified in this study provide a novel understanding of epigenetic mechanisms which may modulate SCLC response to chemotherapy, and suggest potential molecular targets for combination therapies. This research was supported in part with federal funds from the National Cancer Institute, NIH, under contract HHSN261200800001E. Citation Format: Julia Krushkal, Thomas Silvers, Dmitriy Sonkin, Suleyman Vural, John Connelly, Sudhir Varma, Paul S. Meltzer, William C. Reinhold, Annamaria Rapisarda, David Evans, Yves Pommier, Beverly A. Teicher. Associations of epigenome-wide DNA methylation patterns with chemosensitivity and chemoresistance of small cell lung cancer cell lines [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference on Molecular Targets and Cancer Therapeutics; 2019 Oct 26-30; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2019;18(12 Suppl):Abstract nr B013. doi:10.1158/1535-7163.TARG-19-B013

Folate-mediated one-carbon metabolism (OCM) is essential for growth and survival of cancer cells. It affects biosynthesis of nucleotides and amino acids, regulation of redox status, methylation of nucleic acids and proteins, and genome maintenance. We investigated whether the response of cancer cells to antitumor therapy treatment may be partially influenced by variation in expression of one-carbon metabolism genes. We used publicly available gene expression data and drug response measures for 251 antitumor agents in 635 cancer cell lines with matching information from the Cancer Cell Line Encyclopedia and the Genomics of Drug Sensitivity in Cancer resources. We examined whether pre-treatment expression levels of OCM-related genes were associated with drug response. Among the 34 OCM genes examined, expression of GART, TYMS, SHMT2, MTR, ALDH2, BHMT, MAT2B, MTHFD2, NNMT, and SLC46A1 showed modest correlations with response to treatment with a variety of antitumor agents. Higher expression levels of the SLC46A1 transporter gene were associated with resistance to multiple drugs, whereas elevated expression of GART, TYMS, SHMT2, MTR, BHMT, and MAT2B was associated with chemosensitivity to multiple antitumor agents. NNMT expression was bimodally distributed and showed different directions of association with various agents. Correlation of increased NNMT expression with sensitivity to dasatinib was validated in the NCI-60 cancer cell line panel. Expression of several OCM genes was strongly associated with expression of multiple components of drug target pathways. For example, expression of both TYMS and GART was strongly positively correlated with BRCA1 expression, NNMT expression was associated with expression of multiple drug target genes including EGFR and ABL2, and the expression of TYMS, DHFR, and SHMT1 was positively correlated with AURKB expression. Pretreatment expression levels of many OCM genes including DHFR, TYMS, ATIC, GART, AHCY, MTHFD1, SLC19A1, and multiple other genes were positively correlated with each other, suggesting their co-regulation. In contrast, NNMT expression was negatively correlated with expression of several other OCM genes. Further studies could investigate whether correlations of expression levels of individual OCM genes with drug response are related to specific metabolic roles of these genes or whether such associations may be due to the general increase in cellular growth and proliferation and tumor progression, which may affect sensitivity to cancer treatment. Citation Format: Julia Krushkal, Suleyman Vural, Dong-Joon Min. Association of transcriptional levels of folate-mediated one-carbon metabolism related genes in cancer cell lines with drug treatment response [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 4246.

DNA methylation has a powerful influence on cellular processes. It plays important roles in cancer initiation and progression. Cancer cells exhibit frequent epigenomic alterations that affect expression of genes involved in apoptosis, cell proliferation, and DNA repair. We used expression information of known methylating and demethylating genes in cancer cell lines to analyze their association with methylation levels of individual probes and median epigenome-wide methylation levels. We utilized RNA-seq gene expression measures and Illumina Infinium HumanMethylation450 BeadChip DNA methylation data of 645 cancer cell lines from 23 cancer types from the Cancer Cell Line Encyclopedia and Genomics of Drug Sensitivity in Cancer resources. We analyzed 73 genes, products of which are involved in methylation and demethylation processes, and examined correlations between expression levels of these genes and methylation status of individual probes and epigenome-wide. In a combined analysis across all cancer types, we observed statistically significant correlations between expression of global DNA methyltransferase genes, DNMT1, DNMT3A, and DNMT3B, and probe methylation levels in 566, 909, and 230 genes, respectively. In particular, DNMT3A expression was associated with methylation status of growth factor and receptor genes FGF1, FGFR4, and EGFR. Expression of hydroxymethylating genes TET1, TET2, and TET3 was correlated with methylation status of probes in 363, 1, and 1055 genes, respectively, including associations of TET3 expression with methylation of probes in the FGF1, KIT, FHIT, IGF1R, IGF2BP2, and PARP3 genes, which may affect growth, proliferation, and DNA repair of tumor cells. APOBEC1, CBX2, UHRF1 and ZBTB38, products of which are involved in various molecular steps of DNA methylation or demethylation, had the largest numbers of associations in different cancer types. For example, in chronic lymphocytic leukemia cell lines we observed a strong negative correlation between ZBTB38 expression, the product of which binds methylated DNA, and methylation levels of probes in 204 genes. We also found a strong correlation between UHRF1 gene expression and methylation levels of probes in 150 genes in chronic lymphocytic leukemia cell lines, in agreement with reported roles of the product of this gene in binding DNA methyltransferases and affecting DNMT1 methylation activity. These results provide a better understanding of molecular mechanisms of epigenetic dysregulation in cancer cells and may provide strategies for future analysis of molecular targets of drug response that may be affected via epigenetic processes. Citation Format: Suleyman Vural, Julia Krushkal. A survey of molecular factors of DNA methylation and demethylation in a broad range of cancer cell lines [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2479.

Multiple studies show that molecular genetic changes and epigenetic modifications affect the risk of cognitive disability or impairment. However, the role of epigenetic variation in cognitive development of neurotypical young children remains largely unknown. Using data from a prospective, community-based study of mother-infant pairs, we investigated the association of DNA methylation patterns in neonatal umbilical cord blood with cognitive and language development at 1 year of age. No CpG loci achieved genome-wide significance, although a small number of weakly suggestive associations with Bayley-III Receptive Communication scales were noted. While umbilical cord blood is a convenient resource for genetic analyses of birth outcomes, our results do not provide conclusive evidence that its use for DNA methylation profiling yields epigenetic markers that are directly related to postnatal neurocognitive outcomes at 1 year of age.

Cardiovascular reactivity to stress and beta-adrenergic receptor (beta-AR) function may contribute to the development of hypertension. As Black Americans have an increased risk of hypertension, we evaluated associations between beta(1)-AR (Arg389Gly) and beta(2)-AR (Arg16Gly, Gln27Glu) gene variants and cardiovascular reactivity in 500 Black youth. Heart rate, preejection period, total peripheral resistance, and blood pressure reactivity were measured during cold and psychological stress. The Arg389Gly polymorphism in the beta(1)-AR was associated with preejection period reactivity in males but not in females. The Arg16Gly polymorphism in the beta(2)-AR was associated with diastolic blood pressure reactivity only during video game stress. An association between the Gln27Glu polymorphism in the beta(2)-AR and vascular reactivity depended on sex. Thus, specific patterns of associations emerged between genetic variations in beta-ARs and cardiovascular reactivity in young Blacks.

Members of the TetR family of bacterial transcriptional regulators affect expression of genes whose products are involved in a variety of important functions, including osmotic stress, catabolic pathways, homeostasis, biosynthesis of antibiotics, expression of efflux pumps, multidrug resistance, and virulence of pathogenic bacteria. We used genome sequence information to carry out phylogenetic classification of 864 TetR family members with a special focus on TetR regulators in Geobacteraceae, an environmentally important family of delta-Proteobacteria. The genome of Geobacter sulfurreducens, a model representative of Geobacteraceae, contains nine genes from the tetR family. Several of these genes are located immediately upstream of operons encoding functionally important c-type cytochromes. Computational analyses identified the presence of conserved promoters and other regulatory binding sites upstream of several G. sulfurreducens tetR genes. This suggests the possibility of an intermediary role of TetR family proteins in Geobacteraceae in regulatory cascades involving a variety of sigma factors. In order to understand the role of the TetR regulatory family in Geobacteraceae, we have inferred phylogenetic relationships among the Geobacteraceae TetR proteins and their homologs in other microbial species.

We used gene expression information and mutational signature analysis of cell lines from a diverse range of cancerous tissues to examine the activity of two members of the APOBEC family of cytidine deaminases, APOBEC3A and APOBEC3B, and three additional genes. APOBEC3B is known to increase the mutation load in many cancers, generating kataegis clusters of closely spaced, single strand-specific DNA substitutions with a characteristic hypermutation signature. Some studies also suggested that APOBEC3A, REV1, UNG, and FHIT may also participate in hypermutation processes associated with APOBEC activity. We investigated how the activities of these five genes may affect the abundance of APOBEC-like signatures and whether they may be associated with sensitivity of cancer cells to treatment in different cancer categories. We analyzed several data sources for 1,408 cell lines from 26 cancer types available from the Cancer Cell Line Encyclopedia (CCLE) and the Genomics of Drug Sensitivity in Cancer (GDSC) resources. These data included whole exome sequencing (WES) data, microarray gene expression information, and drug sensitivity information. We examined correlations of the abundance of APOBEC-associated motifs and WES-wide mutation loads with APOBEC3A, APOBEC3B, REV1, UNG, and FHIT gene expression, as well as their association with cell line chemosensitivity to 255 antitumor drugs in multiple cancer subtypes. We were able to confirm several previously reported expression and chemosensitivity associations and found additional correlations, which may be clinically important. Sensitivity to JQ1, JNK inhibitors AS601245 and VIII, BMS-509744, bicalutamide, and several other agents was correlated with candidate gene expression levels or with abundance of APOBEC-like motif clusters in specific cancer categories such as pancreatic, breast and non-small cell lung cancer cell lines. For example, we observed a strong negative correlation between APOBEC3A expression in pancreatic cell lines and sensitivity to JQ1, a BET inhibitor, which had been reported to inhibit pancreatic cancer cells in vitro and in vivo. In glioma cell lines, APOBEC3B expression was significantly negatively correlated with sensitivity to a CDK9 inhibitor THZ-2-49 and HSP90 inhibitors AUY922 and 17-AAG (tanespimycin). We observed a strong correlation between the combined length of kataegis clusters and chemoresistance to bicalutamide, a nonsteroidal antiandrogen drug, in breast cancer cell lines. Our findings suggest that associations of sensitivity to drug treatment with activities of APOBEC3A/B, REV1, UNG, and FHIT and with the rates of APOBEC-associated mutagenic processes may vary among different cancer categories. Citation Format: Suleyman Vural, Julia Krushkal, Richard Simon. Activity of APOBEC3A, APOBEC3B, REV1, UNG, and FHIT is associated with drug sensitivity in specific cancer subtypes [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 1282.

Isoprenoids are a highly diverse and important group of natural compounds. The enzyme 1-deoxy-D-xylulose-5-phosphate reductoisomerase (DXR) catalyzes a key regulatory step in the non-mevalonate isoprenoid biosynthetic pathway in eubacteria and in plant plastids. For example, in Artemisia annua DXR participates in regulation of the biosynthesis of artemisinin, an important antimalarial drug. We performed phylogenetic analysis using DXR protein sequences from a model prokaryote, Escherichia coli, a picoplanktonic alga, Ostreococcus lucimarinus, and higher plants. The functional domain of DXR was conserved, allowing molecular evolutionary comparisons of both prokaryotic and eukaryotic sequences of DXR. Despite this conservation, for some plant species such as Campthoteca acuminata and Arabidopsis thaliana, phylogenetic relationships of their lineages were consistently violated. Our analysis revealed that plant DXR has an N-terminal transit domain that is likely bipartite, consisting of a chloroplast transit peptide (cTP) and a lumen transit peptide (lTP). Several features observed in the lTP suggest that, while DXR is targeted to the chloroplast, it is localized to the thylakoid lumen. These features include a twin arginine motif, a hydrophobic region, and a proline-rich region. The transit peptide also showed putative motifs for a 14-3-3 binding site with a chaperone phosphorylation site at Thr.

Geobacteraceae are a family of metal reducing bacteria with important applications in bioremediation and electricity generation. G. sulfurreducens is a representative of Geobacteraceae that has been extensively studied with the goal of extending the understanding of this family of organisms for optimizing their practical applications. Here, we have analyzed gene expression data from 10 experiments involving environmental and genetic perturbations and have identified putative transcription factor binding sites (TFBS) involved in regulating key aspects of metabolism. Specifically, we considered data from both a subset of 10 microarray experiments (7 of 10) and all 10 experiments. The expression data from these two sets were independently clustered, and the upstream regions of genes and operons from the clusters in both sets were used to identify TFBS using the AlignACE program. This analysis resulted in the identification of motifs upstream of several genes involved in central metabolism, sulfate assimilation, and energy metabolism, as well as genes potentially encoding acetate permease. Further, similar TFBS were identified from the analysis of both sets, suggesting that these TFBS are significant in the regulation of metabolism in G. sulfurreducens. In addition, we have utilized microarray data to derive condition specific constraints on the capacity of key enzymes in central metabolism. We have incorporated these constraints into the metabolic model of G. sulfurreducens and simulated Fe(II)-limited growth. The resulting prediction was consistent with data, suggesting that regulatory constraints are important for simulating growth phenotypes in nonoptimal environments.

The insulin (INS) and IGF 2 (IGF2) genes are in close proximity to each other and undergo maternal imprinting during fetal growth. We investigated the association between maternal and umbilical cord IGF 2 protein (IGF-II) levels and single nucleotide polymorphisms (SNPs) in the INS and IGF2 genes in 207 healthy African-American mother-newborn pairs. No association was found between maternal IGF-II levels and polymorphism in the INS-IGF2 locus. A significant association was found between newborn IGF-II levels and two SNPs (rs3842738 and rs689) at the 5' end of the INS-IGF2 locus. Analyses of haplotypes inferred from these two SNPs demonstrate a significant relationship between paternally transmitted haplotypes and newborn IGF-II levels, but no association with maternally transmitted haplotypes.

To test the association between small for gestational age and polymorphisms in the insulin gene in newborns and their mothers, as well as the effect of the parental transmission of haplotypes. Pairs of healthy African-American full-term newborns (N=207) and mothers were recruited from Memphis TN and Jackson MS with birth weights ranging from 2210 to 4735 g. Six single nucleotide polymorphisms (SNPs) located in the insulin (INS) and insulin-like growth factor 2 (IGF2) genes were genotyped in mothers and newborns. Haplotypes composed of three SNPs in the 5′ region of the INS–IGF2 locus were computationally inferred. Odds ratios for risk of small for gestational age (SGA) birth were calculated for individual SNPs and inferred haplotypes in the newborns and in the mothers using logistic regression. For 162 mother – newborn pairs the parental transmission of the haplotypes could be inferred, and the risks for SGA birth were calculated for the three common haplotypes in this sample. Three INS SNPs exhibited significant association with risk for SGA birth. The SNP alleles associated with increased risk for SGA were opposite in the maternal and newborn genomes, implying opposing influences on the rate of fetal growth. Consistent with these results, haplotypes composed of complementary nucleotide sequences (CAC at rs3842738, rs689 and rs3842748, respectively, in the newborn versus GTG in the mother) were significantly associated with risk for SGA birth. In analyses of haplotypes according to parental transmission, the same trend in risk for SGA was observed for both maternally and paternally transmitted haplotypes, although a significant difference in risk was observed only for paternally transmitted haplotypes. Polymorphisms near the 5′ end of the INS–IGF2 locus are significantly associated with risk for SGA birth with a major effect due to the paternally transmitted haplotype, which is preferentially expressed due to imprinting.

In a prospective therapeutic trial, features of the hepatitis C quasispecies were investigated as possible markers of therapeutic response. Individuals chronically infected with hepatitis C genotype 1 received antiviral therapy consisting of alpha-interferon plus ribavirin. The study targeted the most rapidly evolving segment of the viral genome, hypervariable region 1 within the envelope-2 gene. Among individuals failing to clear virus in response to therapy, significant differences were observed between quasispecies of African-American and Caucasian subjects. While distance measures for synonymous substitutions were similar between racial subgroups, measures of distance at the amino acid level (nonsynonymous substitutions) varied significantly. Taken together, the observed patterns of variability corresponded to reduced host selection pressure against hypervariable region 1 in African-American nonresponders. Reduced selection pressure was present at baseline and persisted through treatment and follow-up, suggesting population stratification of host factors that influence selection pressure on hepatitis C virus.