Today, bioanalysts have a range of computational tools to help with method development and data interpretation. These are designed for the user without specialist computational knowledge and are readily available from either commer-cial sources or as freeware. There is software available for determining pKa, logP, solubility, chromatographic retention, sites of metabolism and mass spectral interpretation. But how many of us take time to understand the reasoning behind the packages we are using and, hence, their limitations? All these software packages, whatever the sup -plier or application, have a significant degree of error associated with them. It is this error that makes them predictive rather than definitive. There are two types of computational tool; data modeling and molecular modeling. The majority of software is based on data modeling. Data modeling establishes the relationship between the property of interest and experi-mentally determined ‘descriptors’, which can be statistically extrapolated to other compounds. In molecular modeling, quantum mechanical methods are used to calculate various molecu-lar properties dependent on its conformation, making molecular modeling a 3D technique.Quantitative structure–property relation-ship (QSPR) or quantitative–activity relation-ship (QSAR) approaches establish relationships between molecular structure (expressed as a ‘descriptor’, which is a numerical value represent -ing chemical information

BioanalysisVol. 5, No. 4 EditorialFree Access3D thinking: computational aids for the bioanalystPat Wright, Alexander Alex & Frank PullenPat Wright* Author for correspondenceDepartment of Chemistry, School of Science, University of Greenwich, Medway Campus, Chatham Maritime at Medway, Central Avenue, Kent, ME4 4TB, UK. , Alexander AlexAlexander Alex Consulting, Deal, Kent, UK & Frank PullenDepartment of Chemistry, School of Science, University of Greenwich, Medway Campus, Chatham Maritime at Medway, Central Avenue, Kent, ME4 4TB, UKPublished Online:18 Feb 2013https://doi.org/10.4155/bio.12.332AboutSectionsPDF/EPUB ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareShare onFacebookTwitterLinkedInRedditEmail Keywords: computationaldensity functional theorymass spectral fragmentationmetabolite identificationphysicochemical propertiesToday, bioanalysts have a range of computational tools to help with method development and data interpretation. These are designed for the user without specialist computational knowledge and are readily available from either commercial sources or as freeware. There is software available for determining pKa, logP, solubility, chromatographic retention, sites of metabolism and mass spectral interpretation. But how many of us take time to understand the reasoning behind the packages we are using and, hence, their limitations?All these software packages, whatever the supplier or application, have a significant degree of error associated with them. It is this error that makes them predictive rather than definitive.There are two types of computational tool; data modeling and molecular modeling. The majority of software is based on data modeling. Data modeling establishes the relationship between the property of interest and experimentally determined 'descriptors', which can be statistically extrapolated to other compounds. In molecular modeling, quantum mechanical methods are used to calculate various molecular properties dependent on its conformation, making molecular modeling a 3D technique.Quantitative structure–property relationship (QSPR) or quantitative–activity relationship (QSAR) approaches establish relationships between molecular structure (expressed as a 'descriptor', which is a numerical value representing chemical information [1,2]) and their properties by looking at a 'training' set of molecules for which experimentally determined values are available.QSPR-based software has the advantage that it is computationally economical, with results being generated rapidly. The quality of the prediction of the properties will depend on the structure of the molecule of interest being within the chemical space of the training set. Therefore, the predictions are likely to be less valid for molecules that exhibit novel chemistry. For commercially available software, the nature of the training set is usually unknown to the customer, so it can be difficult to judge if the predictions are likely to be valid for a particular series of compounds.There are many QSPR software packages available: for example ACD/Labs [101], ADMET predictor [102], ChemSilico [103], Pipeline pilot [104] and SPARC [105], which will generate cLogP, aqueous solubility and pKa values. These programs may give different values for the same prediction [3]. Therefore, the bioanalyst may want to try more than one program to find the one that gives suitable results for their specific compounds.Programs are also available for predicting chromatographic retention times. These include DryLab®, ChromGenius and ChromSword®[106–108]. These combine using the physicochemical properties of the molecules with data-based sorbent/eluent/analyte properties to predict chromatographic retention. There is a significant degree of error in these predictions – a recent publication reported ChromGenius predicting the correct elution order for only 68% of 118 compounds [4].Accurately predicting physicochemical properties where a numerical value is generated is a challenge, but stepping beyond this to predicting what happens structurally to a molecule in a complex physiological environment increases the demands on the prediction tool still further.There are several methodologies for the in silico prediction of sites of metabolism. Some of these are QSAR-based, while others are protein–pharmacophore (3D-QSAR) models or predictive databases.Protein–phamacophore modeling predicts from the 3D structure of a known phamacophore and how it interacts with the protein (enzyme). This assumes the mode of protein binding is unique, which of course may not be the case [5]. These models are limited to enzymes where the active site is well modeled, in practice restricting the application to cytochrome P450s and UDP-glucuronosyltransferases.An example of a commercially available metabolite prediction software package is MetaSite. MetaSite considers potential molecular interactions (e.g., electrostatic, hydrogen bonding, hydrophobic and van der Waals and protein/ligand distance) for ligand and binding sites [6]. In addition, MetaSite uses molecular orbital calculations (e.g., Gaussian, B3LYP 6–31G** and AM1) to determine the probability of a metabolic reaction occurring at a certain atom [7].An example of the predictive database approach to metabolite prediction is that taken by the software Meteor (Lhasa). Meteor contains a database of mammalian reactions to which it applies reasoning rules to predict if the molecule is likely to undergo any of the data-based reactions. Data-based predictions have the advantage that they are not limited to enzymes for which the binding site is well characterized.Various commercially available software packages aid with interpretation of MS/MS (collisionally induced dissociation) data. Packages such as Mass Frontier and MS Fragmenter [109,110] are fragment data-based and/or fragmentation rule-based. Fragment iDentificator takes an alternative approach of generating the possible fragments corresponding to the accurate mass of the observed ions [8]; these are then ranked in order of probability on the basis that weak bonds are likely to break first. Bond strength is approximated using literature standard covalent bond energies. Elucidation of product ion connectivity (EPIC) [9] and MetFrag [10] are 'systemic bond dissociation' methods that break all possible bonds in the molecule and compare the accurate masses of the resulting fragments with those of the collisionally induced data products. In addition, MetFrag applies rules to generate neutral loss (i.e., loss of water, ammonia or hydrogen cyanide) driven rearrangements. The fragments are then ranked in order of probability in terms of bond cleavage based on literature-reported bond dissociation energies.Both data- and rule-based approaches have the major limitation that it is assumed the molecule of interest obeys the rules and/or behaves like the molecules in the database. This leads to variation in results generated between programs and deviation from experimentally generated values. In the case of programs that make predictions regarding structural changes, such as metabolite prediction and mass spectral interpretation, it is our experience that the limitations exhibit themselves by overprediction (both in terms of number and type), thus generating many possibilities that the bioanalyst has to manually examine.Is there a way of increasing confidence in in silico predictions for your molecules? Software programs that undertake parameter-based (quantum mechanical) calculations, where no prior knowledge is required, lead to predictions that result entirely from the properties of the molecule itself. This offers a theoretical advantage over the QSPR and database approaches in that the prediction is tailored to the molecule of interest and not based on assumptions that may or may not be valid. One of the most widely applied quantum mechanical tools in chemistry is density functional theory (DFT), which is used to determine the electronic structure of molecules [11]. DFT was first applied to chemistry in the early 1990s, and since then, because of its relative speed and accuracy in prediction, has revolutionized the use of computational tools by chemists [12]. A search of PubMed of the terms 'DFT + Chemistry' produced over 11,000 hits.DFT is ideally applied to molecules in the gas phase. It has been shown to aid mass spectral interpretation by predicting the fragmentation of pharmaceutical molecules [13]. DFT can be used to calculate pKa and lipophilicity (e.g., using software packages such as Jaguar and COSMO-RS [111,112]). It does not always perform well in predicting properties of molecules in solution because solvation is not adequately modeled [14–16]. However, DFT can improve the quality of prediction if used together with other approaches. For example, Szaleniec et al. reported that introducing 3D descriptors (molecular conformation calculated by DFT) into a QSPR method increases the accuracy of the HPLC retention and separation predictions [17].Why are quantum chemistry methods based on 3D conformation not more widely used by bioanalysts? There are several reasons. Often, bioanalysts are unaware of what these programs can do for them and, if they have heard of them, feel that their use is beyond their expertise as quantum mechanics must surely be the realm of a specialist? The word 'quantum' immediately brings images of such scientific giants as Heisenberg and Schrodinger to mind. Yet we do not need to be a physicist to apply quantum mechanics on a daily basis; most of us find operation of a DVD player (the laser being a quantum mechanical device) straightforward because it is designed by experts to be operated by amateurs. Similarly, software programmers and computational chemists are looking to design quantum mechanics packages tailored to specific applications such that they can be easily used by scientists in many fields, the 3D modeling calculations being essentially invisible to the user unless they wish otherwise. The other limitation of quantum methods is that they are computationally 'expensive'. Calculations can take from several minutes to several hours and require a high specification computer (although nothing that cannot be bought at a reasonable cost on the high street). Again work is under way to minimize the type and number of calculations to speed up the time to generate the predictions. Relatively fast packages are already out there; users of MetaSite may not realise they are running quantum (molecular orbital) calculations when they generate their site of metabolism predictions, these predictions taking only minutes.The authors believe that in silico predictions of molecular properties and activities will never completely replace experimental determinations. However, the confidence in predictions for many classes of compounds will be increased using quantum mechanical approaches in two ways: first, by development of software that makes its predictions tailored to the molecule utilizing DFT, these 3D conformational calculations occurring in the background and requiring no user intervention. Second, rule sets describing molecular behaviour with a broader validity will be developed, thanks to a deeper understanding of molecular properties arising from offline molecular modeling using quantum mechanical approaches.Financial & competing interests disclosureThe authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.No writing assistance was utilized in the production of this manuscript.References1 Todeschini R, Consonni V. Handbook of Molecular Descriptors. Wiley-VCH, Hoboken, NJ, USA (2000).Google Scholar2 Waterbeemd H, Gifford E . ADMET in silico modeling: towards prediction paradise? Nat. Rev. Drug Discov.2,192–204 (2003).Crossref, Medline, Google Scholar3 Dearden J, Worth A. In Silico Prediction of Physicochemical Properties. Office for Official Publications of the European Communities, Luxembourg (2007).Google Scholar4 Tyrkkö E, Pelander A, Ojanperä I. Prediction of liquid chromatographic retention for differentiation of structural isomers. Anal. Chim. Acta720,142–148 (2012).Crossref, Medline, Google Scholar5 Smith PA, Sorich MJ, Low LSC, McKinnon RA, Miners JO. Towards integrated ADME prediction: past, present and future directions for modeling metabolism by UDP-glucuronosyltransferases. J. Mol. Graph. Mod.22,507–517 (2004).Crossref, Medline, CAS, Google Scholar6 Cruciani G, Carosati E, De Boeck B, Ethirajulu K, Mackie C. MetaSite: understanding metabolism in human cytochromes from the perspective of the chemist. J. Med. Chem.48(22),6970–6979 (2005).Crossref, Medline, CAS, Google Scholar7 Long A, Walker JD. Quantitative structure–activity relationships for predicting metabolism and modeling cytochrome P450 enzyme activities. Environ. Toxicol. Chem.22(8),1894–1899 (2003).Crossref, Medline, CAS, Google Scholar8 Heinonen M, Rantanen A, Mielikäinen T et al. FiD: a software for ab initio structural identification of product ions from tandem mass spectrometric data. Rapid Commun. Mass Spectrom.22(19),3043–3052 (2008).Crossref, Medline, CAS, Google Scholar9 Hill AW, Mortishire-Smith R-J. Automated assignment of high-resolution collisionally activated dissociation mass spectra using a systematic bond disconnection approach. Rapid Commun. Mass Spectrom.19,3111–3118 (2005).Crossref, CAS, Google Scholar10 Wolf S, Schmidt S, Müller-Hannemann M, Neumann S. In silico fragmentation for computer assisted identification of metabolite mass spectra. BMC Bioinformatics11(1),148–160(2010).Crossref, Medline, Google Scholar11 Burke K. Perspective on density functional theory. J. Chem. Phys.136(50),150901 (2012).Crossref, Medline, Google Scholar12 Alex A. Quantum mechanical calculations. In: Medicinal Chemistry in Comprehensive Medicinal Chemistry II (Volume 4). Jonathan Mason (Ed.). Elsevier, Oxford, UK, 379–418 (2007).Google Scholar13 Wright P, Alex A, Nyaruwata T, Parsons T, Pullen F. Using density functional theory to rationalise the mass spectral fragmentation of maraviroc and its metabolites. Rapid Commun. Mass Spectrom.24,1025–1031 (2010).Crossref, Medline, CAS, Google Scholar14 Chenzhong C, Liao, Marc C, Nicklaus MC. Comparison of nine programs predicting pK(a) values of pharmaceutical substances. J. Chem. Inf. Model49(12),2801–2812 (2009).Crossref, Medline, Google Scholar15 Margabandu R, Subramani K. Comparative study of various quantum mechanical descriptors for prediction of ionization constant (pKa) of substituted anilines. Int. J. ChemTech Res.2(3),1507–1513 (2010).CAS, Google Scholar16 Bryantsev VS, Diallo MS, Goddard WA 3rd. pKa calculations of aliphatic amines, diamines, and aminoamides via density functional theory with a Poisson–Boltzmann continuum solvent model. J. Phys. Chem. A.111(20),4422–4430 (2007).Crossref, Medline, CAS, Google Scholar17 Szaleniec M, Dudzik A, Pawul M, Kozik B. Quantitative structure enantioselective retention relationship for high-performance liquid chromatography chiral separation of 1-phenylethanol derivatives. J. Chromatogr. A1216(34),6224–6235 (2009).Crossref, Medline, CAS, Google Scholar101 ACD/Labs. www.acdlabs.com/homeGoogle Scholar102 Simulations Plus. www.simulations-plus.comGoogle Scholar103 ChemSilico. www.chemsilico.comGoogle Scholar104 Accelrys, Pipeline Pilot. http://accelrys.com/products/pipeline-pilotGoogle Scholar105 ARChem, SPARC. http://archemcalc.com/sparc.phpGoogle Scholar106 Molnar Institute, DryLab®. www.molnar-institut.com/HP/Software/DryLab.phpGoogle Scholar107 ChromSword. www.chromsword.comGoogle Scholar108 ACD Labs, ACD/ChromGenius. www.acdlabs.com/products/com_iden/meth_dev/chromgenGoogle Scholar109 HighChem, Mass Frontier™. www.highchem.com/massfrontier/mass-frontier.htmlGoogle Scholar110 ACD/Labs, ACD/MS Fragmenter. www.acdlabs.com/products/adh/ms/ms_fragGoogle Scholar111 Schrödinger. www.schrodinger.comGoogle Scholar112 COSMOlogic. www.cosmologic.deGoogle ScholarFiguresReferencesRelatedDetailsCited ByDevelopment of a highly sensitive and specific immunoassay for enrofloxacin based on heterologous coating haptensAnalytica Chimica Acta, Vol. 820Understanding collision-induced dissociation of dofetilide: a case study in the application of density functional theory as an aid to mass spectral interpretationThe Analyst, Vol. 138, No. 22 Vol. 5, No. 4 Follow us on social media for the latest updates Metrics History Published online 18 February 2013 Published in print February 2013 Information© Future Science LtdKeywordscomputationaldensity functional theorymass spectral fragmentationmetabolite identificationphysicochemical propertiesFinancial & competing interests disclosureThe authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.No writing assistance was utilized in the production of this manuscript.PDF download

When the Drug Metabolism Discussion Group was instigated in 1971, metabolite identification by mass spectrometry was a slow and laborious process undertaken by mass spectrometrists who seemed to continually disappoint their colleagues by failing to obtain the metabolite spectra. This was usually because not enough material was supplied or the material was impure. Today, accurate metabolite information can be obtained rapidly with little material by utilizing a range of mass spectrometers with complementary properties. This review will discuss how both technology and strategy have evolved over the past forty years to meet the changing demands of metabolism studies within the pharmaceutical industry.

The publication of the US FDA MIST guidance document in 2008 reignited the debate around the most appropriate strategies to underwrite metabolite safety for novel compounds. Whilst some organizations have suggested that the guidelines necessitate a paradigm shift to more thorough metabolite analysis during early development, an evaluation of historical practices shows that the principles of the guidelines have always largely underpinned metabolism studies within the pharmaceutical industry. Therefore, it is argued that existing practices, when coupled to appropriate emerging analytical tools and a case-by-case consideration of the relevance of the generated metabolism data in terms of structure, physicochemisty, abundance and activity, represent a fit-for-purpose approach to metabolite-safety assessments.

HPLC detector technology has advanced dramatically over the past 20 years, with a range of highly sensitive and specific detectors becoming available. What is still missing from the bioanalyst's armoury, however, is a highly sensitive detector that gives an equimolar response independent of the compound. This would allow for quantification of compounds without the requirement for a synthetic standard or a radiolabeled analogue. In particular, such a detector applied to metabolism studies would establish the relative significance of the various metabolic routes. The recently issued US FDA guidelines on metabolites in safety testing (MIST) focus on the relative quantitation of human metabolites being obtained as soon as feasible in the drug-development process. In this article, current detector technology is reviewed with respect to its potential for quantitation without authentic standards or a radiolabel and put in the context of the MIST guidelines. The potential for future developments are explored.

Supervisor's supporting comments Stephen has shown himself to be an exceptional advocate for bioanalysis and an outstanding researcher. He quickly adapted his skills to the requirements of his PhD project, acquiring an understanding of mass spectrometry to which even more experienced practitioners would aspire. Within 18 months, he published his first paper, with a second being recently accepted for publication, and he has presented his work at a number of meetings. In September 2008, he received the Michael Barber award for the best student oral presentation at the British Mass Spectrometry Society conference in York, which attests to his enthusiastic delivery as well as the high standard of his science. His outstanding work and position in his peer group was further recognized when he won the poster prize competition, held at the end of the second year of PhD study at the School of Chemistry, University of Southampton. Stephen has expanded his project to a self-initiated and exciting area that is not only of extreme relevance to metabolite identification, but also increases fundamental knowledge of gas-phase ion chemistry within the collision cell of a mass spectrometer.

The subject of metabolites in safety testing has had much debate in the recent past and has shown itself to be a complex issue with no simple solutions to providing absolute assurance of drug safety. Much of the attention has focused on the ability to identify metabolites and then demonstrate that their risk has been adequately characterized, either through their exposure in toxicology species or, failing this, by direct safety testing. In this review, we summarize our forward operational strategy that combines the principles summarized in the FDA Guidance, together with discussions at scientific meetings and literature opinions. It is a balance between the primary goal of assuring patient safety with one of reasonable investment. A key principle in striking this balance is to build stepwise information on metabolites through the drug discovery and development continuum. This allows assessments to be made from early nonclinical studies onward as to whether or not metabolite safety is underwritten by exposure in toxicology species. This strategy does not require absolute quantitation of the metabolites in early clinical trials but relies upon comparison of relative exposures between animals and humans using the capabilities of modern analytical techniques. Through this strategy, human disproportionate metabolites can be identified to allow a decision regarding the need for absolute quantitation and direct safety testing of the metabolite. Definitive radiolabeled studies would be initiated following proof of pharmacology or efficacy in humans, and nonclinical safety coverage would be adequately assessed prior to large-scale clinical trials. In cases where metabolite safety is not supported through the parent compound toxicology program, approaches for the direct safety testing of metabolites with regard to general and reproductive toxicology, safety pharmacology, and genetic safety have been defined.

1. UK- 343,664 is a novel potent and selective PDE5 inhibitor. Plasma clearances in the male and female rat were high (120 and 54 ml min(-1) kg(-1)), giving rise to short elimination half-lives (0.2 and 0.3 h respectively). Lower clearance in dog (14 ml min(-1) kg(-1)) was the primary factor resulting in a longer elimination half-life (3.7 h). The higher clearance in rat than dog was in agreement with in vitro metabolism rates in hepatic microsomes. 2. The volume of distribution was lower in rat (1.3-2.1 l kg(-1)) compared with dog (4.6 l kg(-1)) probably due to increased plasma protein binding in rat (96 versus 81% in dog). 3. Oral bioavailabilities were 2, 12 and 70% in the male and female rat and dog respectively. T-max less than or equal to 0.5 h in all animals. 4. In multiple oral dose studies, increased systemic exposure was seen with increasing dose up to doses of 200 mg kg(-1) in rat and 150 mg kg(-1) in dog. A marked super-proportional increase in the male rat indicated a capacity-limited clearance at high doses. 5. At the maximal dose of 200 mg kg(-1) in the female rat, no clinical signs were observed after 14 days of treatment. Only minimal signs were recorded in the male rat and dog at the highest dose levels investigated. 6. After single oral or intravenous doses of [C-14]-UK-343,664, the majority of radioactivity was excreted in the faeces of both species. 7. UK- 343,664 was extensively metabolized in both rat and dog. The major primary pathways in dog involved piperazine N-deethylation and loss of a two carbon fragment from the piperazine ring (N,N'-de-ethylation). More extensive metabolism in the rat included additional notable metabolites arising from hydroxylation and lactamization of the piperazine ring, which were only minor metabolites in the dog.

This paper presents a novel MEMS optical mirror based on a proprietary fabrication process. The mirror is fabricated with single crystal silicon and has hexagonal reflective surface 600 /spl mu/m across, with a measured surface roughness is less than 20 angstroms RMS and a radius of curvature of greater than 5 meters. The device has a full 360/spl deg/ of Z rotation at up to 3/spl deg/ (/spl sim/1/spl deg/ controllable) of out of X-Y plane tilt angle depending the design parameters. This mirror has no perforation holes on the reflective surface and no stiction problems during fabrication or operation. The addition of lateral comb drive actuators gives the mirror up to 4 /spl mu/m X and Y in-plane movement. The control of X and Y translation is totally independent and free of movement interference. Due to all electrostatic actuation, the device has lower power consumption, with a designed driving voltage of less than 120 volts. Simulation results, including modal analysis, are included.

1. Pharmacokinetics were studied in mouse, rat, rabbit, dog and man after single intravenous and/or oral doses of sildenafil or [14C]-sildenafil (Viagra). 2. In man, absorption from the gastrointestinal tract was essentially complete. With the exception of male rat, Tmax occurred at approximately 1 h or less. Bioavailability was attenuated by pre-systemic hepatic metabolism in all species. 3. The volume of distribution was similar in rodents and humans (1-2 l/kg) but was greater in dog (5.2 l/kg), due to lower plasma protein binding (84 versus 94-96% respectively). 4. High clearance was the principal determinant of short elimination half-lives in rodents (0.4-1.3 h), whereas moderate clearance in dog and man resulted in longer half-lives (6.1 and 3.7 h respectively). Clearances were in agreement with in vitro metabolism rates by liver microsomes from the various species. 5. After single oral or intravenous doses of [14C]-sildenafil, the majority of radioactivity was excreted in the faeces of all species. No unchanged drug was detected in the excreta of man. 6. Five principal pathways of metabolism in all species were piperazine N-demethylation, pyrazole N-demethylation, loss of a two-carbon fragment from the piperazine ring (N,N'-deethylation), oxidation of the piperazine ring and aliphatic hydroxylation. Additional metabolites arose through combinations of these pathways. 7. Sildenafil was the major component detected in human plasma. Following oral doses, AUC(infinity) for the piperazine N-desmethyl and piperazine N,N'-desethyl metabolites were 55 and 27% that of parent compound respectively.

Hepatic metabolism of melatonin has been investigated. Melatonin was converted in vitro by rat liver microsomes to 6-hydroxymelatonin and to a lesser extent to N-acetylserotonin. Induction with phenobarbitone caused a fourfold increase in the formation of the minor product with little effect on 6-hydroxymelatonin production or melatonin turnover. In contrast, benzpyrene induction caused a dramatic increase in melatonin turnover but the formation of both metabolites, particularly 6-hydroxymelatonin was significantly reduced. This suggests that an alternative inducible pathway is involved in melatonin catabolism.