Computational modeling of cytochrome P450-mediated drug metabolism ; Citochromų P450 katalizuojamo vaistų metabolizmo kompiuterinis modeliavimas

Justas Dapkunas

Sujets

• Biochemistry
• Drug metabolism
• QSAR

Informations

• Publié par: vilnius_university
• Publié le: 01 janvier 2011
• Nombre de lectures: 14
• Langue: English
• Poids de l'ouvrage: 3 Mo

Extrait

VILNIUS UNIVERSITY
¯Justas Dapkunas
COMPUTATIONAL MODELING OF CYTOCHROME P450-MEDIATED
DRUG METABOLISM
Doctoral dissertation
Physical sciences, biochemistry (04P)
Vilnius, 2011The dissertation work was carried out at Vilnius University from 2007 to 2011 in
cooperation with researchers at VšI ˛ “Aukštieji algoritmai”.
Scientific supervisor:
dr. Remigijus Didžiapetris (VšI˛ “Aukštieji algoritmai”, physical sciences, biochemistry
– 04P)VILNIAUS UNIVERSITETAS
Justas Dapkunas¯
CITOCHROMU ˛ P450 KATALIZUOJAMO VAISTU ˛ METABOLIZMO
KOMPIUTERINIS MODELIAVIMAS
Daktaro disertacija
Fiziniai mokslai, biochemija (04P)
Vilnius, 2011Disertacija rengta 2007 - 2011 metais Vilniaus universitete bendradarbiaujant su
VšI ˛ „Aukštieji algoritmai“.
Mokslinis vadovas:
dr. Remigijus Didžiapetris (VšI˛ „Aukštieji algoritmai“, fiziniai mokslai, biochemija
– 04P)Acknowledgements
I would like to thank my scientific advisor dr. Remigijus Didžiapetris and
the director of VšI˛ “Aukštieji Algoritmai” dr. Pranas Japertas for their help,
patience, and guidance through all the years of working there. I very appreciate
the valuable discussions on development of the models and interpretation of the
obtained results with my colleagues dr. Andrius Sazonovas and Kiril Lanevskij.
Additionally, I acknowledge dr. Laura Steponenien˙ e˙ and Liutauras Juška who
helped me with the analysis of experimental regioselectivity data, and also
˙Dainius Šimelevicius,ˇ dr. Rytis Kubilius, and Tomas Bukenas for development
of the software tools that were used in this work.
I would like to thank the people at the Department of Biochemistry and Bio-
physics of Vilnius University who allowed and encouraged me to do this work.
˙Especially I am grateful to prof. dr. Vida Kirveliene, prof. dr. Dobilas Kirvelis
and dr. Saulius Gražulis for the complicated examinations during which I
learned a lot.
Finally, special thanks go to my family. This thesis would have not been written
without the help of my wife, parents, sister, brother, and, of course, daughters,
who always supported me.
vTable of Contents
Introduction 1
1 Overview of Literature 6
1.1 Cytochrome P450 and Drug Metabolism . . . . . . . . . . . . . . . 6
1.1.1 Reactions Catalyzed by Cytochrome P450 . . . . . . . . . . 6
1.1.2 Human Cytochrome P450 Enzymes . . . . . . . . . . . . . 9
1.1.3 Inhibition of Cytochrome P450 . . . . . . . . . . . . . . . . 13
1.2 Experimental Methods for Estimation of Drug Metabolism . . . . 15
1.2.1 Metabolite Identification . . . . . . . . . . . . . . . . . . . . 15
1.2.2 Cytochrome P450 Reaction Phenotyping . . . . . . . . . . 16
1.2.3ome P450 Inhibition Assays . . . . . . . . . . . . . 17
1.3 QSAR Models of CYP3A4 . . . . . . . . . . . . . . . . . 20
1.4 Prediction of Drug Metabolism Regioselectivity . . . . . . . . . . 23
1.4.1 Biotransformation Rules . . . . . . . . . . . . . . . . . . . . 24
1.4.2 Prediction of Regioselectivity by Quantum Chemistry Me-
thods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.4.3 Regioselectivity Models Using Structures of Enzymes . . . 27
1.4.4 Data Mining Models for Prediction of Metabolism Sites . . 29
2 Data and Modeling Methods 33
2.1 CYP3A4 Inhibition Data . . . . . . . . . . . . . . . . . . . . . . . . 33
2.1.1 Literature Dataset . . . . . . . . . . . . . . . . . . . . . . . . 33
2.1.2 PubChem . . . . . . . . . . . . . . . . . . . . . . . . 34
2.1.3 Summary of CYP3A4 Inhibition Datasets . . . . . . . . . . 35
2.2 Regioselectivity Data . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.2.1 Modeling Dataset . . . . . . . . . . . . . . . . . . . . . . . . 36
2.2.2 External Validation Dataset . . . . . . . . . . . . . . . . . . 37
2.3 Descriptors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.3.1 Fragmental Descriptors . . . . . . . . . . . . . . . . . . . . 38
2.3.2 Atom-centered Fragmental Descriptors . . . . . . . . . . . 38
2.4 Statistical Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.4.1 Global Model . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.4.2 Dynamic Similarity . . . . . . . . . . . . . . . . . . . . . . . 42
2.4.3 Local Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
viiTABLEOFCONTENTS
2.4.4 Estimation of Prediction Reliability . . . . . . . . . . . . . . 44
2.4.5 Training of the GALAS Model . . . . . . . . . . . . . . . . 45
2.5 Development and Validation of Models . . . . . . . . . . . . . . . 46
2.5.1 CYP3A4 Inhibition Model . . . . . . . . . . . . . . . . . . . 46
2.5.2 Regioselectivity Model . . . . . . . . . . . . . . . . . . . . . 47
2.5.3 Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3 Results and Discussion: CYP3A4 Inhibition Modeling 50
3.1 Global Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.2 Local . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.3 Training of the GALAS Model . . . . . . . . . . . . . . . . . . . . . 61
3.3.1 Training with Data from a Similar Assay . . . . . . . . . . 62
3.3.2 T with Data from an Assay with a Different Potency
Threshold . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.3.3 Training with Compounds from a New Structural Class . 65
4 Results and Discussion: Regioselectivity Modeling 68
4.1 Internal Validation of the Model . . . . . . . . . . . . . . . . . . . . 69
4.2 External V of the . . . . . . . . . . . . . . . . . . . 73
4.3 Comparison to Other Models . . . . . . . . . . . . . . . . . . . . . 79
4.4 Adaptation of the Model to Compounds of Novel Classes . . . . . 81
4.5 of the to Cytochrome P450 Phenotyping . . . . 83
Conclusions 86
A Results of External Validation of Regioselectivity Model 88
References 103
List of Publications 121
Curriculum Vitae 123
Summary in Lithuanian (Santrauka) 124
viiiAbbreviations
ADME – Absorption, Distribution, Metabolism, Excretion;
ANN – Artificial Neural Networks;
BC – naive Bayesian Classifier;
BFC – 7-benzyloxy-4-trifluormethylcoumarin;
CYP1A2 – cytochrome P450 1A2;
CYP2C9 –ome P450 2C9;
CYP2C19 – cytochrome P450 2C19;
CYP2D6 –ome P450 2D6;
CYP3A4 – cytochrome P450 3A4;
DFT – Density Functional Theory;
DMCI – Data-Model Consistency Index;
GALAS – Global, Adjusted Locally According to Similarity;
HLM – Human Liver Microsomes;
HPLC – High Performance Liquid Chromatography;
HTS – High-Throughput Screening;
IC – half maximal Inhibitory Concentration;50
IGF-1R – Insulin-like Growth Factor-1 Receptor;
K – inhibition constant;i
K – Michaelism
kNN – k-Nearest Neighbors;
LC – Liquid Chromatography;
LD – median lethal dose (the dose required to kill half of the members of a50
tested population);
LogP – 1-octanol/water partition coefficient;
LR – Logistic Regression;
MLR – Multiple Linear Regression;
MS – Mass Spectrometry;
ixAbbreviations
NCBI – National Center for Biotechnology Information;
NMR – Nuclear Magnetic Resonance;
OECD – Organization for Economic Co-operation and Development;
p – probability;
PLS – Partial Least Squares/Projection to Latent Structures;
QSAR – Quantitative Structure-Activity Relationship;
RI – Reliability Index;
ROC – Receiving Operating Characteristic;
RT – Regression Trees;
SI – Similarity Index;
SOME – Site of Metabolism Estimator;
SPORCalc – Substrate Product Occurrence Ratio Calculator;
SVM – Support Vector Machines;
SyGMa – Systematic Generation of Potential Metabolites.
x