Dissertation submitted to the Combined Faculties for the Natural Sciences and for Mathematics of the Ruperto-Carola University of Heidelberg, Germany for the degree of Doctor of Natural Sciences Metabolic Profiling of Cancer Cells and Correlations between Metabolism, Gene Expression and Drug Sensitivity Diplom- Mariela Huichalaf Carbonell Dissertation submitted to the Combined Faculties for the Natural Sciences and for Mathematics of the Ruperto-Carola University of Heidelberg, Germany for the degree of Doctor of Natural Sciences presented by Diplom- Mariela Huichalaf Carbonell born in Santiago, Chile Oral-examination 17 of October 2017 Metabolic Profiling of Cancer Cells and Correlations between Metabolism, Gene Expression and Drug Sensitivity Referees: Prof. Dr. Stefan Wölfl Prof. Dr. Tobias Werner Table of Contents List of Tables 4 List of Figures 6 Abbreviations 7 Summary 10 Zusammenfassung 11 1. Introduction 13 1.1 Cellular Features 13 1.1.1 Glycolysis and Gluconeogenesis 13 1.1.2 The Contribution of Mitochondria to Cellular Respiration 15 1.1.3 Cellular Proliferation 16 1.1.4 Membrane Capacitance 17 1.2 Public Data Repositories 19 1.3 Correlations in Cellular Biology 21 2. Experimental Methods 24 2.1 Cellular Properties Acquisition 24 2.1.1 Cell Cultures 24 2.1.2 Online Monitoring of Membrane Capacitance and Cellular Metabolism 25 2.1.3 End Point Assays to Measure Cellular Metabolism 28 2.2 Hierarchical Clustering 30 2.3 Gene Expression Analyses 31 2.3.1 Selection of Genes Related to Metabolism Pathways with Extreme Expression in the Affymetrix Array 32 1 2.3.2 Selection of Genes in the Total Probe Sets of the Affymetrix Array with Extreme Expression 40 2.3.3 Real-Time Polymerase Chain Reaction (qPCR) 40 2.4 Half Maximal Inhibitory Concentration (IC50) Estimations 41 2.4.1 IC50 Values from the Genomics of Drug Sensitivity in Cancer Project 41 2.4.2 Drug IC50 Estimations 44 2.5 Dependencies between Cellular Properties and Gene Expression or Drug IC50s 44 2.6 Graphics and Statistics 45 3. Results 46 3.1 Cellular Capacitance and Metabolism 48 3.1.1 Cellular Capacitance 48 3.1.2 Glycolytic Activity 50 3.1.3 Respiration Activity 51 3.1.4 Energy Metabolism (ATP Level) 52 3.1.5 Total Mitochondrial Mass 53 3.1.6 Reactive Oxygen Species (ROS) Accumulation 54 3.1.7 Cell Proliferation Rate 55 3.2 Cellular Properties of the Cancer Cell Lines by Tissue Type. Summary Figure and Cluster Dendrograms 58 3.3 Linear Relations between the Cellular Properties of the Studied Cell Lines 63 3.4 Cellular Properties and Affymetrix Gene Expression 65 3.4.1 Selection of Genes Belonging to the Selected Metabolism Pathways 70 3.4.2 Relations between Cell Properties of the Studied Cell Lines and the Extremely Expressed Probe Sets in Affymetrix Arrays 78 2 3.4.3 Relationships between Cell Properties of the Studied Cell Lines and Affymetrix Gene Expression of the Probe Sets with High Significance 85 3.4.4 Affymetrix Gene Expression and qPCR 94 3.5 Cellular Properties and Drug IC50s 97 3.5.1 Relationships between Cell Properties and Drug IC50s of the Studied Cell Lines 97 3.5.2 Drug Sensitivity (IC50) of the Cell Lines 104 4. Discussion 107 4.1 Cellular Properties 107 4.2 Gene Expression and Metabolism 109 4.3 Drug Sensitivity and Cellular Properties of Cancer Cell Lines 114 5. Supplements 118 5.1 R Scripts 118 5.1.1 Bionas Data Collection 118 5.1.2 Robust Multi-Array Average (RMA) 119 5.1.3 Pathways Gene Selection 120 5.1.4 Present and Absence Calls 121 5.1.5 Selection of the Highly and Lowest Express Probes 122 5.1.6 Heatmap and Hierarchical Clustering 124 5.1.7 Pearson Correlation Coefficients Estimation 128 6. References 136 3 List of Tables 1. List of cell lines used in this work 24 2. Cell lines and gene accession numbers 31 3. Entrez gene identifiers of genes belong to the pentose phospate pathway WP134 from wikipathways 32 4. Entrez gene identifiers of genes belonging to the glycolysis- gluconeogenesis pathway WP534 from wikipathways 33 5. Entrez gene identifiers of genes belonging to the tricarboxylic acid cycle and electron transport chain, pathways WP78 and WP111 from wikipathways 34 6. List of the primer sequences used in real-time PCR for estimating the expression of some of the genes involves in the glycolysis- gluconeogenesis pathway 41 7. IC50 values in μM obtained from the Genomics of Drug Sensitivity in Cancer project 42 8. Compounds and concentrations (μM) used for the IC50 calculations 44 9. Mean value of the cell capacitance and metabolism features of cancer cell lines 46 10. r coefficients and p values of the linear dependencies between the electrical and metabolic properties of the studies cell lines 64 11. Affymetric gene expression means of selected pathways by cell line 68 4 12. r coefficients and p-values of the significant dependencies between metabolism features and Affymetrix gene expression of candidate genes of the selected pathway 75 13. r coefficients and p values of the extreme expressed probe sets in Affymetrix array and the cell properties of the studied cell lines 79 14. Affymetrix gene expression of the genes that show the highly significant dependencies with the cellular properties of the studied cell lines 89 15. p-values of the highly significant dependencias between the cellular properties and the total probe sets from Affymetrix array 90 16. r coefficients of the highly significant dependencies between metabolism features and total probe sets from Affymetrix Array 92 17. Database drug IC50s compared with own estimations in our cell lines 105 5 List of Figures 1. Real-time monitoring of cell properties with the Bionas Discovery 2500 instrument and the Bionas Discovery SC1000 chip 26 2. Cellular capacitance per cell lines 49 3. Glycolytic activity per cell lines 50 4. Respiration activity per cell line 52 5. Energy metabolism (ATP level) per cell line 53 6. Total mitochondrial mass per cell line 54 7. Reactive oxygen species (ROS) formation per cell line 55 8. Proliferation rate measure as metabolic activity (MTT assay) per cell line 56 9. Proliferation rate measure as total protein content (SRB assay) per cell line 57 10. Summary of cell capacitance and metabolic features of cell lines by tissue type 60 11. Hierarchical clustering of the cellular capacitance and the metabolic features 62 12. Significant dependencies between the cellular features of the analyzed cell lines 66 13. Relation of cellular capacitance and respiration activity with Affymetrix gene expression level of selected metabolism pathways 69 14. Selection of genes related to glycolysis- gluconeogenesis and PPP 71 6 15. Selection of genes related to TCA cycle and the electron transport chain 74 16. Significant dependencies between mitochondrial mass content and the gene expression of the candidate gene PGLS 77 17. Significant dependencies between glycolytic activity and mitochondrial mass with one of the probe sets of RHBDL2 gene 82 18. Significant dependencies between respiration activity and the two probe sets of IFI16 84 19. Highly significant dependencies between the metabolic features of the analyzed cell lines and the total probes sets in Affymetrix Array 91 20. Real-time polymerase chain reaction (qPCR) in key metabolic genes of the glycolysis- gluconeogenesis pathway 95 21. Affymetrix Gene expression and qPCR comparison 96 22. Drugs IC50 98 23. Elesclomol - significant dependencies between cellular capacitance and the log10 IC50 of elesclomol 99 24. Nutlin 3a and PF 4708671 – significant dependencies between respiration activity and the log10 IC50 of Nutlin 3a (A) and PF 4708671 (B) 100 25. EHT 1864, IPA3, and RDEA119 – significant dependencies between energy metabolism and the log10 IC50 of EHT 1864 (A), IPA3 (B), and RDEA119 (C) 102 26. Methotrexate – significant dependencies between ROS accumulation and the log10 IC50 of methotrexate 104 7 Abbreviations • ATP: Adenosine Triphosphate • COX5B: Cytochrome C Oxidase Subunit 5b • DHE: Dihydroethidium • DMEM: Dulbecco's Modified Eagle Medium • DMSO: Dimethyl sulfoxide • DNA: Deoxyribonucleic Acid • DPBS: Dulbecco’s Phosphate Buffered Saline • F: Farads • GCN1L1: EIF-2-Alpha Kinase Activator GCN1 • GDSC: Genomics of Drug Sensitivity in Cancer Project • GEO: Gene Expression Omnibus • GOT1: Glutamic-Oxaloacetic Transaminase 1 • HK1: Hexokinase 1 • HK2: Hexokinase 2 • HOXA7: Homeobox 7 • IDES sensor: Interdigitated Electrodes • IFI16: Gamma-Interferon-Inducible Protein 16 • INPP5B: Inositol Polyphosphate-5-Phosphatase B • ISFET sensors: Ion-Sensitive Field-Effect Transistors • LDHA: Lactate Dehydrogenase A • LGALS8: Galectin 8 • miRNA: microRNA • MTT: 3-(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide • NADH: Nicotinamide Adenine Dinucleotide • NCBI: National Center for Biotechnology Information • OXPHOS: Oxidative Phosphorylation • PANP: Presence-Absence calls with Negative Probe sets • PFKM: Phosphofructokinase-Muscle • PGLS: 6-phosphogluconolactonase • PPP: Pentose Phosphate Pathway • qPCR: Real-Time Polymerase Chain Reaction • r: Correlation Coefficient 8 • RHBDL2: Rhomboid-Related Protein 2 • ROS: Reactive Oxygen Species • SLC2A1: Solute Carrier Family 2 Member 1 • SRB: Sulforhodamine B • TCA: Tricarboxylic Acid 9 Summary Detailed metabolic characterization of cancer cells provides an important cellular footprint in addition to well-studied genomic features and could play an important role for understanding sensitivity of cancer cells to drug treatment. Thus, detailed understanding of metabolism and gene expression profiles of cancer cells could be an important guideline supporting the steps of drug selection and a determining factor for designing treatment strategies with available anticancer drugs. The aim of the presented study was to analyze the metabolic activities and cell-cell interaction and cell-matrix adhesion properties of a large number of cancer cell lines and correlate these cellular characteristics with drug sensitivity and compare this relation to established knowledge using gene expression and drug sensitivity prediction. For this, correlations among these different set of analytical information were analyzed to obtain cancer cell profiles linking metabolism, gene expression and drug IC50 values, by calculating Pearson correlation coefficients.