In addition to mediating cell metabolism, the metabolic system developed in animals and humans for the chemical conversion of xenobiotics. Over millions of years, a plethora of oxidizing, hydrolyzing, conjugating, and other enzymes were optimized by evolution. Modification, degradation, and/or conjugation, in many cases to polar products, enable a safe elimination from the organism. Whereas many plant products are toxic, there are only rare examples that the metabolic system converts harmless natural substances into toxic entities. The situation changed about two centuries ago, after the advent of synthetic organic compounds: many of them contain structural features that the metabolic system cannot handle in the same manner as natural products, In only a few generations, evolution did not have enough time to optimize the enzymes for this new challenge. Of course, also potential drug candidates offer such a challenge to the metabolic system. The development of many compounds must be discontinued because of severe side effects of some toxic metabolites, most often chemically reactive compounds [1]. Some metabolites, even formed in only minor amounts, may cause idiosyncratic toxicity, rarely observed but with fatal consequences for the individual.
Chemical features that are easily metabolized are responsible for short biological half-life of some potential drug candidates; on the other hand, lack of such moieties might cause a half-life that is too long for safe use of the drug. In addition, such compounds as well as highly lipophilic analogs have a higher risk to form toxic metabolites. Thus, it is most important to understand metabolic pathways and to have tools to predict which compounds might be generated. This necessity applies especially for the common oxidation of xenobiotics by various cytochrome P450s (CYPs). Three approaches are suited to achieve this task: theoretical treatment, by calculating the accessibility and chemical reactivity of the chemical features of the compound; molecular modeling, especially pharma-cophore searches and docking, using 3D structures of the cytochrome binding pockets; and empirical approaches, using the large databases of known metabolic pathways. All these methods have their pros and cons, and none of them seems to be perfect. Especially species selectivity, to conclude from animal results to humans, and the relative amount of certain metabolites are difficult or even impossible to predict.
The introduction of this book provides an overview of the role of metabolism in drug development, followed by a part on software and databases for the study of metabolism. The next part discusses computational approaches for the study of the most important metabolic enzymes, the cytochrome P450 enzymes. 3D structures, substrate recognition and binding, and theoretical and experimental methods for the study of ligand-protein interactions are discussed in this part The chapters of the next part go into more detail with respect to the sites and products of metabolism, using either molecular interaction fields or structure-, reactivity-, and knowledge-based approaches. The important aspect of enzyme inhibition and induction is discussed in the next chapters, using quantitative structure-activity relationships and pharmacophore-based methods; separate chapters discuss the role of P-gp-mediated disposition and the prediction of toxic effects of metabolites. Last but not least, three chapters describe experimental approaches, that is, in vitro models for the study of metabolism and drug—drug interactions and experimental metabolite detection and profiling.
We are very grateful to Johannes Kirchmair for having accepted our invitation to edit this book, which will be of great importance and practical value for all scientists involved in drug research. Our thanks also go to all chapter authors for their valuable contributions, as well as to Frank Weinreich and Heike Nothe at Wiley-VCH for their engagement in this project and in our entire book series "Methods and Principles in Medicinal Chemistry." Dusseldorf Raimund Mannhold, Weisenheim am Sand Hugo Kubinyi, Zurich Gerd Folkers June 2014

List of Contributors XVII

Preface XXI

A Personal Foreword XXIII

Part One Introduction 1

1 Metabolism in Drug Development 3

      1.1 What? An Introduction 3

      1.2 Why? Metabolism in Drug Development 4

      1.2.1 The Pharmacological Context 4

      1.2.2 Consequences of Drug Metabolism on Activity 6

      1.2.3 Adverse Consequences of Drug Metabolism 7

      1.2.4 Impact of Metabolism on Absorption, Distribution, and Excretion 10

      1.3 How? From Experimental Results to Databases to Expert Software Packages 11

      1.3.1 The Many Factors Influencing Drug Metabolism 11

      1.3.2 Acquiring and Interpreting Experimental Results 13

      1.3.3 Expert Software Tools and Their Domains of Applicability 14

      1.3.4 Roads to Progress 16

      1.4 Who? Human Intelligence as a Conclusion 17

      References 19

Part Two Software, Web Servers and Data Resources to Study Metabolism 27

2 Software for Metabolism Prediction 29

      2.1 Introduction 29

      2.2 Ligand-Based and Structure-Based Methods for Predicting Metabolism 30

      2.3 Software for Predicting Sites of Metabolism 38

      2.3.1 Knowledge-Based Systems 38

      2.3.2 Molecular Interaction Fields 39

      2.3.3 Docking 39

      2.3.4 Reactivity Models 40

      2.3.5 Data Mining and Machine Learning Approaches 41

      2.3.6 Shape-Focused Approaches 42

      2.4 Software for Predicting Metabolites 43

      2.4.1 Knowledge-Based Systems 44

      2.4.2 Data Mining and Machine Learning Approaches 46

      2.4.3 Molecular Interaction Fields 46

      2.5 Software for Predicting Interactions of Small Molecules with Metabolizing Enzymes 46

      2.6 Conclusions 48

      References 49

3 Online Databases and Web Servers for Drug Metabolism Research 53

      3.1 Introduction 53

      3.2 Online Drug Metabolism Databases 54

      3.2.1 DrugBank 57

      3.2.2 HMDB 59

      3.2.3 PharmGKB 59

      3.2.4 Wikipedia 60

      3.2.5 PubChem 61

      3.2.6 Synoptic Databases: ChEBI, ChEMBL, KEGG, and BindingDB 61

      3.2.7 Specialized Databases: UM-BBD, SuperCYP, PKKB, and PK/DB 63

      3.2.8 Online Database Summary 64

      3.3 Online Drug Metabolism Prediction Servers 65

      3.3.1 Metabolite Predictors 66

      3.3.2 SoM Predictors 66

      3.3.3 Specialized Predictors 68

      3.3.4 ADMET Predictors 70

      3.3.5 Web Server Summary 71

      References 71

Part Three Computational Approaches to Study Cytochrome P450 Enzymes 75

4 Structure and Dynamics of Human Drug-Metabolizing Cytochrome P450 Enzymes 77

      4.1 Introduction 77

      4.2 Three-Dimensional Structures of Human CYPs 78

      4.3 Structural Features of CYPs 78

      4.3.1 CYP-Electron Transfer Protein Interactions 81

      4.3.2 Substrate Recognition Sites 82

      4.3.3 Structural Variability and Substrate Specificity Profiles 83

      4.3.3.1 CYP1A2 83

      4.3.3.2 CYP2A6 85

      4.3.3.3 CYP2C9 85

      4.3.3.4 CYP2D6 86

      4.3.3.5 CYP2E1 87

      4.3.3.6 CYP3A4 87

      4.4 Dynamics of CYPs 88

      4.4.1 Active Site Flexibility 88

      4.4.2 Active Site Solvation 93

      4.4.3 Active Site Access and Egress Pathways 93

      4.4.4 MD Simulations of CYPs in Lipid Bilayers 96

      4.5 Conclusions 96

      References 97

5 Cytochrome P450 Substrate Recognition and Binding 103

      5.1 Introduction 103

      5.2 Substrate Recognition in the Catalytic Cycle of CYPs 103

      5.3 Substrate Identity in Various Species 104

      5.4 Structural Insight into Substrate Recognition by CYPs 107

      5.4.1 CYP1A1, CYP1A2, and CYP1B1 108

      5.4.2 CYP2A6 108

      5.4.3 CYP2A13 109

      5.4.4 CYP2C8 110

      5.4.5 CYP2C9 112

      5.4.6 CYP2D6 112

      5.4.7 CYP2E1 113

      5.4.8 CYP2R1 113

      5.4.9 CYP3A4 115

      5.4.10 CYP8A1 115

      5.4.11 CYP11A1 116

      5.4.12 CYP11B2 118

      5.4.13 CYP19A1 118

      5.4.14 CYP46A1 119

      5.4.15 General Insights from Protein-Ligand Crystal Structures 119

      5.5 The Challenges of Using Docking for Predicting Kinetic Parameters 120

      5.6 Substrate Properties for Various Human Isoforms 120

      5.6.1 Kinetic Parameters K_m and K_cat and Their Relationship with Substrate and Protein Structure 124

      5.7 Conclusions 128

      References 128

6 QM/MM Studies of Structure and Reactivity of Cytochrome P450 Enzymes: Methodology and Selected Applications 133

      6.1 Introduction 133

      6.2 QM/MM Methods 135

      6.2.1 Methodological Issues in QM/MM Studies 136

      6.2.1.1 QM/MM Partitioning 136

      6.2.1.2 QM Methods 137

      6.2.1.3 MM Methods 138

      6.2.1.4 Subtractive versus Additive QM/MM Schemes 139

      6.2.1.5 Electrostatic QM/MM Interactions 139

      6.2.1.6 QM/MM Boundary Treatments 139

      6.2.1.7 QM/MM Geometry Optimization 140

      6.2.1.8 QM/MM Molecular Dynamics and Free Energy Calculations 140

      6.2.1.9 QM/MM Energy versus Free Energy Calculations 141

      6.2.2 Practical Issues in QM/MM Studies 141

      6.2.2.1 QM/MM Software 141

      6.2.2.2 QM/MM Setup 142

      6.2.2.3 Accuracy of QM/MM Results 143

      6.2.2.4 QM/MM Geometry Optimization 143

      6.2.2.5 Extracting Insights from QM/MM Calculations 144

      6.3 Selected QM/MM Applications to Cytochrome P450 Enzymes 144

      6.3.1 Formation of Cpd I from Cpd 0 146

      6.3.1.1 Conversion of Cpd 0 into Cpd I in the T252X Mutants 148

      6.3.2 Properties of Cpd 1 151

      6.3.2.1 Cpd I Species of Different Cytochrome P450s 154

      6.3.3 The Mechanism of Cytochrome P450 StaP 155

      6.3.4 The Mechanism of Dopamine Formation 160

      6.3.4.1 The Electrostatic Effect is Not Due to Simple Bulk Polarity 163

      6.4 An Overview of Cytochrome P460 Function Requires Reliable MD Calculations 163

      6.5 Conclusions 164

      References 165

7 Computational Free Energy Methods for Ascertaining Ligand Interaction with Metabolizing Enzymes 179

      7.1 Introduction 179

      7.2 Linking Experiment and Simulation: Statistical Mechanics 180

      7.2.1 A Note on Chemical Transformations 182

      7.3 Taxonomy of Free Energy Methods 183

      7.3.1 Pathway Methods 183

      7.3.1.1 Pathway Planning: Using the State Nature of the Free Energy Cycle 184

      7.3.1.2 Free Energy Perturbation 185

      7.3.1.3 Bennett Acceptance Ratio 185

      7.3.1.4 Thermodynamic Integration 186

      7.3.2 Endpoint Methods 186

      7.3.2.1 Molecular Mechanics-Generalized Born Surface Area (MM-GBSA) 186

      7.3.2.2 Linear Interaction Energy 187

      7.3.2.3 QM Endpoint Methods 187

      7.3.3 Summary of Free Energy Methods 187

      7.4 Ligand Parameterization 188

      7.5 Specific Examples 189

      7.5.1 Cytochrome P450 (CYP) 189

      7.5.2 Chorismate Mutase 192

      7.6 Conclusions 192

      References 193

8 Experimental Approaches to Analysis of Reactions of Cytochrome P450 Enzymes 199

      8.1 Introduction 199

      8.2 Structural Data and Substrate Binding 199

      8.3 Systems for Production of Reaction Products and Analysis of Systems 200

      8.3.1 In Vivo Systems 201

      8.3.2 Tissue Microsomal Systems 201

      8.3.3 Purified CYPs in Reconstituted Systems 201

      8.3.4 Membranes from Heterologous Expression Systems 202

      8.3.4.1 Mammalian Cells 202

      8.3.4.2 Insect Cell Systems (Using Baculovirus Infection for Expression) 202

      8.3.4.3 Microbial Membrane Systems 202

      8.4 Methods for Analysis of Products of Drugs 203

      8.4.1 Separation Methods 203

      8.4.1.1 High-Performance Liquid Chromatography 203

      8.4.1.2 Other Separation Methods 204

      8.4.2 Analysis Methods 204

      8.4.2.1 HPLC-UV 204

      8.4.2.2 LC-MS 205

      8.4.2.3 LC-MS/MS 205

      8.4.2.4 LC-HRMS 205

      8.4.2.5 NMR 205

      8.4.2.6 Other Spectroscopy of Metabolites 206

      8.5 Untargeted Searches for CYP Reactions 208

      8.6 Complex CYP Products 208

      8.7 Structure-Activity Relationships Based on Products 210

      8.7.1 SARs Based on Chemical Bond Energy 211

      8.7.2 SARs Based on Docking 211

      8.7.3 Knowledge-Based SAR 212

      8.8 SAR of Reaction Rates 213

      8.9 Other Issues in Predictions 213

      8.10 Conclusions 214

      References 214

Part Four Computational Approaches to Study Sites and Products of Metabolism 221

9 Molecular Interaction Fields for Predicting the Sites and Products of Metabolism 223

      9.1 Introduction 223

      9.2 CYP from a GRID Perspective 224

      9.3 From Lead Optimization to Preclinical Phases: the Challenge of SoM Prediction 226

      9.3.1 MetaSite: Accessibility Function 227

      9.3.2 MetaSite: Reactivity Function 229

      9.3.3 MetaSite: Site of Metabolism Prediction 230

      9.3.4 MetaSite: Validation and Case Studies 231

      9.3.5 MetaSite: Prediction of CYP Inhibition 234

      9.3.6 MassMetaSite: Automated Metabolite Identification 236

      9.4 Conclusions 239

      References 241

10 Structure-Based Methods for Predicting the Sites and Products of Metabolism 243

      10.1 Introduction 243

      10.2 6 A Rule 243

      10.3 Methodological Approaches 245

      10.4 Prediction of Binding Poses 247

      10.5 Protein Flexibility 249

      10.6 Role of Water Molecules 254

      10.7 Effect of Mutations 256

      10.8 Conclusions 258

      References 259

11 Reactivity-Based Approaches and Machine Learning Methods for Predicting the Sites of Cytochrome P450-Mediated Metabolism 265

      11.1 Introduction 265

      11.2 Reactivity Models for CYP Reactions 268

      11.2.1 Hydroxylation of Aliphatic Carbon Atoms 268

      11.2.2 Hydroxyiation and Epoxidation of Aromatic and Double Bonded Carbon Atoms 271

      11.2.3 Combined Carbon Atom Models 273

      11.2.4 Comprehensive Models 273

      11.3 Reactivity-Based Methods Applied to CYP-Mediated Site of Metabolism Prediction 274

      11.3.1 Methods Only Applicable to Carbon Atoms 274

      11.3.2 Comprehensive Methods 276

      11.4 Machine Learning Methods Applied to CYP-Mediated Site of Metabolism Prediction 278

      11.4.1 Atomic Descriptors 278

      11.4.2 Machine Learning Methods and Optimization Criteria 279

      11.5 Applications to SoM Prediction 280

      11.5.1 Isoform-Specific Models 281

      11.5.2 Isoform-Unspecific Models 283

      11.6 Combinations of Structure-Based Models and Reactivity 284

      11.7 Conclusions 285

      References 286

12 Knowledge-Based Approaches for Predicting the Sites and Products of Metabolism 293

      12.1 Introduction 293

      12.2 Building and Maintaining a Knowledge Base 295

      12.3 Encoding Rules in a Knowledge Base 299

      12.4 Ways of Working with Rules 301

      12.5 Using the Logic of Argumentation 303

      12.6 Combining Absolute and Relative Reasoning 307

      12.7 Combining Predictions from Multiple Sources 310

      12.8 Validation and Assessment of Performance 312

      12.9 Conclusions 314

      References 314

Part Five Computational Approaches to Study Enzyme Inhibition and Induction 319

13 Quantitative Structure-Activity Relationship (Q5AR) Methods for the Prediction of Substrates, Inhibitors, and Inducers of Metabolic Enzymes 321

      13.1 Introduction 321

      13.2 In Silico QSAR Methods 322

      13.2.1 Experimental Variability 323

      13.2.2 Data Curation and Manipulation 324

      13.2.3 Molecular Descriptors 324

      13.2.4 Training SAR, QSAR, and Machine Learning Models 325

      13.2.5 Local versus Global QSAR Models 525

      13.2.6 SAR and Classical QSAR Methods 326

      13.2.7 Machine Learning QSAR Methods 327

      13.2.8 Model Assessment and Validation 327

      13.2.8.1 Assessing the Predictive Ability of QSAR Models 327

      13.2.8.2 Applicability Domains of QSAR Models 328

      13.3 QSAR Models for Cytochrome P450 328

      13.3.1 Inhibition QSAR 328

      13.3.1.1 SAR 328

      13.3.1.2 Classical QSAR Models 329

      13.3.1.3 Machine Learning QSAR Models 333

      13.3.1.4 Classification Models 334

      13.3.1.5 3D QSAR Models 335

      13.3.2 Enzyme Induction QSAR 336

      13.4 Conjugative Metabolizing Enzymes 357

      13.4.1 Uridine Diphosphate Glucosyltransferase (UGT) QSAR 338

      13.4.2 Sulfotransferases QSAR 338

      13.5 In Vitro Clearance QSAR 339

      13.6 Conclusions 340

      References 341

14 Pharmacophore-Based Methods for Predicting the Inhibition and Induction of Metabolic Enzymes 351

      14.1 Introduction 351

      14.2 Substrate and Inhibitor Pharmacophore Models 354

      14.2.1 Cytochrome P450 enzymes 354

      14.2.1.1 CYP1A2 354

      14.2.1.2 CYP2B6 355

      14.2.1.3 CYP2C9 356

      14.2.1.4 CYP2C19 357

      14.2.1.5 CYP2D6 358

      14.2.1.6 CYP3A4 359

      14.2.1.7 CYP3A5 and CYP3A7 360

      14.2.2 UDP-Glucuronosyltransferases (UGTs) 361

      14.2.2.1 UGT1A1 361

      14.2.2.2 UGT1A4 361

      14.2.2.3 UGT1A9 361

      14.2.2.4 UGT2B7 362

      14.2.3 Interference with Recently Identified Phase I Metabolic Enzymes 362

      14.3 Inducer Models 363

      14.3.1 Hetero- and Autoactivation 363

      14.3.1.1 CYP2C9 363

      14.3.1.2 CYP3A4 364

      14.3.2 Nuclear Receptors 364

      14.3.2.1 Pregnane X Receptor 364

      14.3.2.2 CAR 366

      14.4 Conclusions 366

      References 368

15 Prediction of Phosphoglycoprotein (P-gp)-Mediated Disposition in Early Drug Discovery 373

      15.1 Introduction 373

      15.2 QSAR Modeling of Compounds Interacting with Transporters 376

      15.2.1 Experimental Data and Assays 376

      15.2.2 Descriptors Used in P-gp Substrate Identification 378

      15.2.3 QSAR Methods Used in P-gp Substrate Identification 380

      15.3 Influence of Compound Structure on P-gp Substrate Identity 380

      15.4 QSAR Models for P-gp Substrates 385

      15.5 Application to Drug Discovery 388

      15.6 Conclusions 391

      References 392

16 Predicting Toxic Effects of Metabolites 397

      16.1 Introduction 397

      16.2 Methods for Predicting Toxic Effects 401

      16.2.1 Predicting Metabolites 401

      16.2.2 Predicting Relative and Absolute Metabolism Likelihoods and Rates 401

      16.2.3 Utilizing Pharmacogenetic Data to Anticipate Dose, Rate, and Time Information in an Individual Patient 402

      16.2.4 Predicting the Effect of the Resulting Metabolites 402

      16.2.4.1 Bioactivity-Based Mechanistic Models 403

      16.2.4.2 Incorporating Pathway Information into Toxicity Models 404

      16.2.4.3 Toxicogenetic and Pharmacogenomic Approaches 406

      16.2.4.4 Knowledge-Based Systems 407

      16.2.4.5 Reactive Metabolites 407

      16.2.5 Current Scientific and Political Developments Regarding Metabolism and Toxicity Prediction 408

      16.3 Conclusions 408

      References 409

Part Six Experimental Approaches to Study Metabolism 413

17 In Vitro Models for Metabolism: Applicability for Research on Food Bioactives 415

      17.1 Introduction 415

      17.1.1 Bioavailability 416

      17.1.2 Intestinal Absorption 416

      17.1.3 First-Pass Metabolism 418

      17.2 Classification of In Vitro Models for Metabolism 418

      17.3 Modifications via Gut (Colon) Microflora 419

      17.3.1 Background Information 419

      17.3.2 In Vitro Models 420

      17.3.2.1 Fecal Slurry 421

      17.3.2.2 Isolated Pure Bacterial Cultures 421

      17.3.2.3 Complex Intestinal Models (TIM-2) 421

      17.4 Intestinal (Gut Wall) Metabolism 421

      17.4.1 Background Information 421

      17.4.2 In Vitro Models 422

      17.4.2.1 Tissue Intact Models 423

      17.4.2.2 Subcellular and Cellular Models 423

      17.5 Hepatic Metabolism 423

      17.5.1 Background Information 423

      17.5.2 In Vitro Models 424

      17.5.2.1 Supersomes: Recombinant Phase I and Phase II Enzymes 424

      17.5.2.2 Microsomes 424

      17.5.2.3 S9 Fractions 426

      17.5.2.4 Hepatocyte Cell Lines 426

      17.5.2.5 Primary Cultures: Cryopreserved Hepatocytes 427

      17.5.2.6 Cryopreserved Hepatocytes versus Microsomes 428

      17.5.2.7 Hepatocytes in Culture 429

      17.6 Pharmacokinetic Data Obtainable from In Vitro Metabolism Models 431

      17.6.1 Pharmacokinetic Analysis 431

      17.6.1.1 Measurement Methodology: Substrate Depletion versus Metabolite Formation 432

      17.6.1.2 Mathematical Models for Metabolism: Well-Stirred, Parallel Tube, and Dispersion Models 432

      17.7 Assay Validation 433

      17.7.1 Selection and Preparation of Reference Compounds 433

      17.7.2 Analytics 434

      17.7.3 Theoretical Steps to Establish an In Vitro Model 434

      17.8 Conclusions 435

      17.8.1 What Can We Summarize from the Literature? 435

      17.8.2 What Questions We Wish to Have Answered Will Determine Which Model We Select 436

      References 438

18 In Vitro Approaches to Study Drug-Drug Interactions 441

      18.1 Introduction 441

      18.1.1 Additional Factors Influencing Drug Metabolism 442

      18.2 Inhibition of Drug Metabolism 444

      18.2.1 In Vitro Models for Predicting Inhibition of Drug Metabolism 444

      18.2.1.1 Human Liver Microsomes 445

      18.2.1.2 S9 and Cytosol 456

      18.2.1.3 Recombinant Enzymes 457

      18.2.1.4 Primary Hepatocytes 458

      18.3 Transcriptional Regulation of Metabolism 460

      18.3.1 Gene Induction Pathways 460

      18.3.2 Gene Repression/Suppression 462

      18.3.3 In Vitro Models for Predicting Induction of Drug Metabolism Enzymes 463

      18.3.3.1 Ligand Binding Assays 463

      18.3.3.2 Gene Reporter Assays 465

      18.3.3.3 Cellular Models for Induction Studies 466

      18.3.3.4 Induction Assays in Cellular Models 468

      18.3.3.5 Treatment with Control and Test Compounds 470

      18.3.3.6 Gene Expression in Cellular Models for Induction 471

      18.3.3.7 Enzymatic Activity in Metabolically Competent Cellular Models of Induction 474

      18.4 Next-Generation Models and Concluding Remarks 474

      References 477

19 Metabolite Detection and Profiling 485

      19.1 Introduction 485

      19.2 Chromatography 486

      19.3 Mass Spectrometry 487

      19.4 Sample Preparation for LC-MS-Based Metabolite Profiling 490

      19.5 Metabolic Profiling by LC-MS 491

      19.5.1 Metabolic Stability and Cytochrome P450 Inhibition Assays 491

      19.5.2 Metabolite Profiling, Detection, and Identification from In Vivo and In Vitro Studies 492

      19.5.3 Reactive Metabolite Detection 496

      19.6 Conclusions 496

      References 497

Index 499

中国医科院医学信息研究所 中科院文献情报中心