Professional Documents
Culture Documents
Copyright
c Informa Healthcare USA, Inc.
ISSN: 1537-6516 print; 1537-6524 online
DOI: 10.1080/15376510701857189
243
metabolism relationships (QSMRs)/quantitative structure- similarity as a measure of similarity to training set molecules.
activity relationships (QSARs), pharmacophores (de Groot and Structural alerts for likely reactive metabolites (Li 2002; Uetrecht
Ekins 2002), statistical QSAR approaches (Shen et al. 2003; 2003; Williams and Park 2003) are also provided. MetaDrug
Balakin et al. 2004a, 2004b), rule-based approaches, electronic has been tested with 66 molecules and captured approximately
models (Singh et al. 2003; Korzekwa et al. 2004), homol- 79% of first-pass metabolites (Ekins et al. 2006). A more recent
ogy models, and crystal structures with docking approaches application of the underlying metabolism database has been for
(Jolivette and Ekins 2007). Metabolism knowledge bases (Darvas the creation of Kernel-Partial Least Squares (K-PLS) models for
1987; Klopman et al. 1994, 1997; Talafous et al. 1994; Boyer individual metabolic transformations that could be used to rank
and Zamora 2002; Langowski and Long 2002; Borodina et al. suggested metabolites (Embrechts and Ekins 2007).
2003, 2004; Button et al. 2003; Erhardt 2003) combine data In the drug discovery process, it is imperative for medicinal
or rules from many different mammalian species and tend to chemists to quickly find out the major metabolic “soft spots”
predict all the metabolic possibilities for a molecule (Nassar and in the lead molecule structure, so that analog compounds can
Talaat 2004), rather than the likely metabolites, thus limiting be designed and synthesized to address the metabolic stability
their usefulness in drug discovery. problems. The conventional method is to perform in vitro
These techniques when used individually have different and/or in vivo metabolic studies and identify the structures
levels of success and could be combined to improve predictions, of the metabolites using spectroscopic methods such as LC-
perhaps by deploying the most appropriate models based on MS/MS and NMR. This is often a costly and time-consuming
the molecule neighborhood for the test molecule compared to process. With considerable emphasis now on increasing the
training sets. P450–substrate/inhibitor recognition interactions efficiency of drug discovery, there is interest in using the types
have been studied extensively and have generally shown the of predictive computational approaches (as described above) in
importance of hydrophobic, hydrogen bonding and ionizable as many areas as possible, particularly in early drug discovery.
features for both substrates based on Km data and inhibitors Reliable and rapid metabolite prediction using computational
using Ki , IC50 , and percent inhibition data (Ekins et al. approaches, if achievable, could become a valuable tool for drug
2001). Molecular models that account for electronic effects discovery.
of ligands for P450-mediated metabolism have also been The most powerful tool for analysis of in vivo and in vitro
produced (Jones and Korzekwa 1996; Jones et al. 2002) and metabolite samples is mass spectrometry, typically coupled
these have combined aliphatic and aromatic oxidation reactions with liquid chromatographic separation in an online LC/MS
to generate predictions for metabolic regioselectivities. The instrument. Mass spectrometry can provide molecular weight
combination of approaches may also balance the strengths and information with high accuracy and sensitivity, and with
weaknesses of each approach and hence the introduction of such tandem MS/MS instruments, information about molecular
hybrid methods is ongoing. For example, a recent technique structure as well. Confirmation of the likely presence of pre-
called MetaSite (Molecular Discovery, Middlesex, UK) gener- dicted metabolites using LC/MS and LC/MS/MS is generally
ates GRID field descriptors used for determining energetically a straightforward procedure, but requires expert knowledge
favorable binding sites on molecules of known structure using in mass spectral interpretation. This manual interpretation is
crystal structures or homology models for the P450 enzymes and a bottleneck in the metabolite ID process, severely limiting
the interaction energy descriptors for the molecules evaluated throughput. Mass spectrometer manufacturers have responded
as substrates (Berellini et al. 2005). A reactivity component is to this need by providing some tools for metabolite detection,
also used in the MetaSite calculation to produce a probability but to our knowledge none of them explicitly takes the chemical
for an atom to be metabolized. This approach has been applied structure of the metabolite into account. Thus, there is always
with angiotensin II receptor antagonists predicting the site of some question where isomers are concerned, and a manual
metabolism for CYP2C9 and CYP3A4 (Berellini et al. 2005). interpretation is required.
A recent study has expanded the application to CYP2D6, With recent interest in using metabolite prediction software
CYP2C19, and CYP1A2 with 75% to 86% correct predictions with mass spectrometry data analysis (Anari et al. 2004; Nassar
(Cruciani et al. 2005). MetaSite has been compared with docking and Talaat 2004; Shu et al. 2004; Anari and Baillie 2005; Lin
of ligands into crystal structures and homology models of et al. 2005), we have used MetaDrug along with a second
CYP3A4, and MetaSite had 78% overall success compared with new software tool, Apex, which can predict the molecular
57% for docking (Zhou et al. 2006). MetaSite has a limitation of ion spectrum and a fragmentation spectrum that can in turn
not being able to predict the absolute or relative amounts of the be used to determine which predicted molecules are present
major and minor metabolites as well as the rate of metabolite in the analyte. This technology is compatible with all mass
formation for a molecule. A second hybrid method, MetaDrug, spectrometers and hence the combination of MetaDrug (or
includes a manually annotated Oracle database of human another predictive metabolism tool) and Apex enables the
drug metabolism information including xenobiotic reactions, streamlining of metabolite prediction and mass spectrometry
enzyme substrates, and enzyme inhibitors with kinetic data analysis. In silico metabolite prediction typically generates many
(Ekins et al. 2005a; Ekins et al. 2005b; Ekins et al. 2005c). more potential metabolites than are actually observed. This
The MetaDrug database has been used to predict some of presents a considerable problem for manual interpretation of
the major metabolic pathways and identify the involvement the mass spectral data. The analyst must examine each peak
of P450s using multiple QSAR models enabling the prediction detected in the liquid chromatogram and compare the mass
of affinity and rate of metabolism for numerous enzymes (Ekins spectra under that peak with the spectra expected for the
et al. 2005a, 2006). The user may also upload his or her own structures. Quite often, many of the predicted metabolites are
QSAR or QSMR data into the software to offer a further level of isomers, requiring a detailed examination of both MS and
utility. Steps have been undertaken to provide more confidence MS/MS spectra to determine which, if any, are correct. By
in predictions as the QSAR methods also provide Tanimoto automating this tedious process, Apex removes the potential
N-demethylation (desipramine) Yes Yes Best match overall; all MS/MS features
matched
Aromatic hydroxylation (e.g., Yes (4 isomers) Yes Moderate score; isomers are
2-hydroxylation, 10-hydroxylation) indistinguishable by MS and MS/MS
Alkyl hydroxylation Yes (3 isomers) Yes High score, but biased by lack of observed
MS/MS fragments
N-oxidation Yes (2 isomers) Yes Oxidation of dimethylamine most likely
2-hydroxy desipramine No∗ N/A
Didesmethylimipramine No∗ N/A
Iminodibenzyl Yes No
∗ These metabolites would be predicted if MetaDrug was used in the sequential mode.
† Apex cannot determine presence of structures de novo. The only structures that can be determined are those imported from the prediction software.
correlation of a molecule against experimental data. Apex uses the spectra in the data set. (4) MS/MS similarity score: Using the
each of these properties in several ways to create a set of scores. estimated fragmentation spectrum, match the predicted features
Each score is in the range of zero (no match) to one (perfect against those observed and weight according to the intensity of
match). (1) Similarity score: Using predicted mass and isotope the matched features. The weighting takes into account the
cluster, compute the “distance” between the predicted features expectation that the major features in the spectrum should be
and those observed in a spectrum. If the spectrum contains accounted for by a predicted fragment. (5) In addition, external
none of the predicted features, the score is zero; if the spectrum signals (such as from an LC/UV or radiochemical detector) can
is identical to the prediction, the score is one. A partial match be incorporated as “scores,” where the signal is normalized to
to the prediction, either by mass or relative isotopic abundance, a 0–1 range. Each of the structure-based scores is computed
yields a score between zero and one. (2) Purity score: Compute for each spectrum in the data set. If a score is plotted for each
the similarity score, and then weight it by the summed intensity spectrum in time order, the result is a score chromatogram. Peaks
of the matched spectral features divided by the total intensity appear in the score chromatogram where the underlying spectra
of the spectrum. A high score results when the isotope cluster have high score values for that metric. Each individual structure
is both a good match and is the most intense set of features will have a set of score chromatograms, one for each computed
in the spectrum; the spectrum is pure for that species. (3) score.
Molecular ion cluster (MIC) score: Sum the intensities of the When a number of potential metabolite structures are
matched spectral features, and then normalize this over all of presented to Apex, as is the case with MetaDrug, Apex
FIGURE 1 The integrated in silico/in vitro metabolite prediction protocol using MetaDrug and Apex for predicting and verifying
metabolites with mass spectrometry data.