Elucidation of Drug Metabolite Structural Isomers ... - ACS Publications

Subscriber access provided by United Arab Emirates University | Libraries Deanship

Article

Elucidation of drug metabolite structural isomers using molecular modelling coupled with ion mobility mass spectrometry Eamonn Reading, Jordi Munoz-Muriedas, Andrew D Roberts, Gordon J. Dear, Carol V. Robinson, and Claire Beaumont Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.5b04068 • Publication Date (Web): 11 Jan 2016 Downloaded from http://pubs.acs.org on January 16, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Elucidation of drug metabolite structural isomers using molecular modelling coupled with ion mobility mass spectrometry Eamonn Reading1ǂ†*, Jordi Munoz-Muriedas2†, Andrew D. Roberts3, Gordon J. Dear3, Carol V. Robinson1, Claire Beaumont3* 1

Department of Chemistry, Physical and Theoretical Chemistry Laboratory, University of Oxford,

Oxford, UK 2

Chemical Sciences, Computational Chemistry, GlaxoSmithKline, Stevenage, Hertfordshire, SG1

2NY, UK 3

Drug Metabolism and Pharmacokinetics, GlaxoSmithKline, Ware, Hertfordshire, SG12 0DP, UK

ǂ

Present address: Department of Chemistry, King’s College London, Britannia House, 7 Trinity

Street, London WC2R 2LS, UK *Correspondence: [email protected]; [email protected] †

These authors contributed equally to this work

1 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Abstract

Ion mobility mass spectrometry (IM-MS) in combination with molecular modelling offers the potential for small molecule structural isomer identification by measurement of their gas phase collision cross sections (CCSs). Successful application of this approach to drug metabolite identification would facilitate resource reduction, including animal usage, and may benefit other areas of pharmaceutical structural characterisation including impurity profiling and degradation chemistry. However, the conformational behaviour of drug molecules and their metabolites in the gas phase is poorly understood. Here the gas phase conformational space of drug and drug-like molecules has been investigated, as well as the influence of protonation and adduct formation on the conformations of drug metabolite structural isomers. The use of CCSs, measured from IM-MS and molecular modelling information, for the structural identification of drug metabolites has also been critically assessed. Detection of structural isomers of drug metabolites using IM-MS is demonstrated and, in addition, a molecular modelling approach has been developed offering rapid conformational searching and energy assessment of candidate structures which agree with experimental CCSs. Here it is illustrated that isomers must possess markedly dissimilar CCS values for structural differentiation – the existence and extent of CCS differences being ionization state and molecule dependent. The results present that IM-MS and molecular modelling can inform on the identity of drug metabolites and highlight the limitations of this approach in differentiating structural isomers.


Page 2 of 27

Page 3 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Introduction

In the pharmaceutical industry, it is important to understand the absorption, distribution, metabolism and excretion properties of a drug in order to understand its pharmacology and safety. Drug metabolism is a major determinant for the changes in physiological drug concentration and can determine or alter its pharmacological/toxicological pathway1. It is therefore imperative to ascertain what happens to a drug in man with a view of relating this to the action of the drug or its metabolites2.

Drug metabolites are typically identified using a combination of techniques but primarily by liquid chromatography (LC), mass spectrometry (MS), and nuclear magnetic resonance (NMR)3-5. Although MS can provide many structural clues to a metabolites identity, it is generally not definitive, and can rarely differentiate between structurally isomeric species which possess both the same mass and fragmentation information but that differ in connectivity6. This can be problematic when the exact structure of the metabolite is required to make an assessment of metabolite pharmacological activity or potential reactivity, or for the synthesis of a standard for metabolite quantification. In such cases robust metabolite isolation and subsequent NMR, both of which are time consuming, are typically employed to generate unambiguous structural information7,8. Although NMR facilitates unequivocal structure determination a major drawback of the technique is its limited sensitivity compared to MS. Consequently, an alternative approach is desirable.

Ion mobility (IM) coupled with MS (IM-MS) has been increasingly utilized to gain structural insight into small molecules and proteins9. IM-MS facilitates simultaneous detection of an ion’s mass-tocharge ratio (m/z) and arrival time. The additional temporal dimension afforded by IM has proven to be a powerful tool for separation of complex ion mixtures in the gas phase, aiding mass spectral deconvolution of lipids10, peptides11-13, sugars12,14-16, proteins17 and metabolite species18,19. Additionally, an ions collision cross section (CCS) can be calculated from its arrival time providing a descriptor of the ion gas phase shape – the utilization of molecular modelling and theoretical CCS calculations facilitate the elucidation, or at least rationalization, of ion gas phase structure20-28. The 3 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ability of IM-MS in providing a molecular structural descriptor, coupled with its low sample requirement and high sensitivity, offers advantages in drug metabolite detection and identification when compared to NMR. Potentially definitive structural characterization could therefore be deciphered on up to a thousand times less material, ultimately leading to reduced animal usage in preclinical metabolism studies, as well as allowing samples from lower dosed clinical and pre-clinical studies to be analysed.

Structural identification of moderately flexible biomolecules by IM-MS has proven challenging29. This is due to lack of sufficient, and cost effective, molecular modelling and CCS calculation protocols for the production of accurate theoretical CCS values which are consistent with experimental values for flexible biomolecules.

For a successful protocol to be adopted by the

pharmaceutical industry, a balance must be struck between the cost of performing the calculations and the value (accuracy) of the information gained.

Parity between empirical and theoretical CCS measurements for drugs and drug metabolites has been tested previously6,28,30-32; however the strength of IM-MS in identifying unknowns (with no prior information) has not been rigorously tested. Here the ability of IM-MS information coupled with a novel molecular modelling protocol and established theoretical CCS calculations are investigated to elucidate structural isomers of drug metabolites; contributing to the rapidly developing field of small molecule interrogation and structural identification using IM-MS9.


Page 4 of 27

Page 5 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Experimental

Drugs and drug metabolites

Naringenin-7-O-β-D-glucuronide (naringenin-7-glucuronide) and naringenin-4’-O-β-D-glucuronide (naringenin-4’-glucuronide) were purchased from Cayman Chemicals. β-Estradiol 3-(β-Dglucuronide)

(β-3-estradiol

glucuronide),

β-Estradiol

17-(β-D-glucuronide)

(β-17-estradiol

glucuronide), and all other drug molecules used were purchased from Sigma-Aldrich. For positive ion electrospray IM-MS analysis the samples were prepared at 1 µg/ml in a 50% (v/v) solution of acetonitrile and MilliQ water supplemented with 0.01% formic acid6. For negative ion electrospray IM-MS analysis the samples were prepared at 1 µg/ml in a 50% (v/v) solution of acetonitrile and MilliQ water supplemented with 0.01% glacial acetic acid33.

Ion mobility mass spectrometry Data were collected on both a Synapt G2 HDMS34 and a modified Synapt G1 HDMS (Waters, Manchester, UK); the Synapt G1 HDMS has its travelling wave ion mobility cell replaced by a RF ion confinement linear voltage gradient drift tube35. The data were analysed using MassLynxTM software (Waters).

Molecular modelling and in silico collision cross section calculations

Molecular modelling calculations and conformational searches were performed using the Molecular Operating Environment (MOE) software package36,37 utilizing LowModeMD38. Quantum Mechanical (QM) calculations were carried out using Gaussian 03 software39. All theoretical CCS calculations were carried out with the MOBCAL24,25 or Sigma40 codes at 300 K.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Results and Discussion

IM-MS separates ions based on their gas phase size and shape. To be successful in identifying unknown structural isomers from CCS information the isomers must possess different and detectable gas phase shapes. It is therefore important to investigate and understand the impact of parameters such as ionisation state and adduct type on CCSs, as well as the gas phase conformational space drug-like molecules occupy. Moreover, a successful identification strategy relies on molecular modelling and theoretical CCS measurement reproducibility in providing accurate and representative CCSs of druglike molecules.

Gas phase conformational analysis of drug-like molecules

The ‘global’ gas phase molecular conformational space can be measured by monitoring how CCS varies with mass, or rather mass-to-charge ratio (m/z). The empirical CCSs of forty-five protonated and eighteen deprotonated drug-like molecules were investigated here, using drift tube-IM (DT-IM) MS instrumentation, to comprehend the conformational space of drug-like molecules. To evaluate the conformational ordering of drug-like molecules, their gas phase shape behaviour was also compared to those of other biomolecules (lipids, peptides and carbohydrates) which had previously been inspected by May et al41. Both protonated and deprotonated (Figure S-1) drug-like molecules were found to produce apparently identical gas phase packing efficiencies however they were more compact than those of the ammonium salts and other biomolecules investigated by May et al 41.

The gas phase packing efficiency is dependent on the type of monomer unit (if any) comprising the molecule, the polymeric organization of the molecules (i.e. whether addition of mass leads to branched or linear increase in volume), and the intramolecular forces within the molecule42; the larger packing efficiency expressed by drug-like molecules could therefore be due to their typical nonadditive constituent units. Additionally, it is credible that drug-like molecules express an enhanced rigidity compared to the other polymeric biomolecules, as well as a higher abundance of functional


Page 6 of 27

Page 7 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


groups available for cumulative intramolecular interactions in the gas phase, which could lead to their observed compactness in comparison to other biomolecules.

Influence of ionization state and adduct formation on drug metabolite structural isomer gas phase conformation

The structural isomers of both naringenin and estradiol glucuronide conjugates were used to investigate IM-MS’s capability for drug metabolite isomer distinction (Figure 1). Interestingly, structural isomer CCS and CCS rank order were dependent on both ionization state and adduct formation.

Protonated naringenin glucuronide structural isomers, naringenin-4’-glucuronide (4’) and naringenin7-glucuronide (7), possessed substantial drift-tube CCS (DTCCSHe) differences (∆CCS = 6.6 %), with 4’ being smaller than 7 (Figure 1B). However, the deprotonated species offered no discernible CCS difference (∆CCS = 0.5 %). Furthermore, considerable CCS difference was observed between the protonated and deprotonated forms of naringenin-7-glucuronide (∆CCS = 6.2 %). Interestingly, there was no apparent CCS difference between protonated and deprotonated forms for the 4’ isomer, suggesting that ionisation state influence on the ion’s shape was molecule dependent.

Interestingly, adduct formation was found to impact CCSs. Both ammonium and sodium adducts dramatically changed the DTCCSHe for the naringenin glucuronide isomers compared to their protonated forms, altering the isomer CCS rank order in the process, with 4’ being larger than 7. However, both possessed indiscernible CCS differences between the structural isomers (∆CCS = 0.9 and 1.4 %, respectively) in comparison to the protonated forms. Additionally, all naringenin glucuronide isomers produced similar arrival time distribution (ATD) full width at half maxima (DTFWHM); the width of the Gaussian shaped ATD indicates multiple ion conformations, reflecting a molecules structural dynamics and heterogeneity43.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

These effects are emulated by the estradiol glucuronide structural isomers, β-3-estradiol-glucuronide (3) and β-17-estradiol-glucuronide (17), where ionization state and adduct formation caused different DTCCSHe’s and CCS rank orders (Figure 1C). In this case, however, the deprotonated species presented a different CCS isomer ranking (3 > 17) compared to the protonated and adducted species (3 < 17). The deprotonated species produced the largest CCS difference between isomers (∆CCS = 3.3 %), with the protonated species producing a smaller CCS difference (∆CCS = 2.6 %), and the ammonium and sodium adduct species producing indiscernible CCS differences (∆CCS = 0.4 and 0.6 %, respectively). Interestingly, substantial DTFWHM differences were found between the deprotonated isomers (3 = 18.8 and 17 = 28.0 Å2) suggesting a less dynamic, more homogenous, ion population for the deprotonated β-3-estradiol-glucuronide.

Overall, this demonstrates that both ionization state and adduct formation can alter the CCS, and therefore shape, of drug-like molecules in the gas phase although the impact of these parameters appears to be molecule specific. Additionally, these factors can also influence the drug metabolite gas phase structural dynamics and/or ion heterogeneity as demonstrated by their difference in DTFWHM’s. This phenomenon has also been observed for sugars16,20 and is likely caused by both the availability and strength of intramolecular interactions causing different gas phase molecular behaviour. The existence and strength of these interactions are facilitated by the molecules polarity and through different adduct formation.

It is clear that even though drug-like molecules possess high gas phase packing efficiency, structural isomers can still form subtlety different conformer populations detectable by their CCSs and ATDs, and that IM-MS can successfully discern between them. We observe that ionization state and adduct formation influences the gas phase conformer populations in a manner which is specific to each molecule investigated. However if the structural isomers do not conform to dissimilar shapes in the gas phase or there are only minor differences between their CCSs then they may not be discernible with current IM-MS instrumentation.


Page 8 of 27

Page 9 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Molecular modelling of drug-like molecule conformers

The identification of unknown metabolite molecules from their CCSs requires comparison with theoretically derived CCSs. Previous computational protocols for moderately flexible molecules have been attempted29, although no ‘gold standard’ protocol exists to date. A novel molecular modelling protocol is presented here for generating minimum energy conformers of drug-like molecules for calculation of their theoretical CCSs – the workflow is summarized in Figure S-2.

Firstly, the tridimensional coordinates of all the protonated/deprotonated versions of a molecule of interest are obtained and minimised. The minimised geometries become starting points for a conformational search using the LowModeMD procedure. The PFROSST force-field and semiempirical AM1-BCC partial charges are used for these calculations44.

A comprehensive exploration of the conformational space is critical to the procedure. A typical conformational searching procedure uses simulated annealing molecular dynamics (SAMD) followed by further optimization of low energy conformers using QM theory21,29. However, SAMD is computationally expensive and molecules undergoing SAMD have a higher propensity to become trapped in potential energy wells due to its annealing procedures. LowModeMD38 was used to circumvent some of these issues. LowModeMD avoids generation of high energy and improbable conformers, reducing the time spent exploring the higher energy surfaces of the potential energy surface. Moreover, this process avoids a conformer annealing process. Additionally, LowModeMD has been shown to be very efficient for moderate to large molecules and is capable of efficiently sampling low-strain energy conformations with non-trivial non-bonded interactions38. These interactions are potentially important for drug-like molecules in the gas phase as demonstrated by their high gas phase packing efficiency (Figure S-1).

The conformers generated by a LowModeMD search are then ranked by their energy relative to the minimum energy structure of each protonated/deprotonated form (∆E). At this stage, several 9 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

conformational properties can be assessed and visualized if desired, such as the molecules’ radius of gyration (Figure S-3), to establish the number of potential conformers and their relative sizes. An energy cut-off of 2 kcal mol-1 is then used to filter out conformers lowly populated at the empirical temperature; the temperature of the DT-IM-MS ion mobility cell was monitored at 25-27 °C therefore this cut-off was judged acceptable and has been employed previously6. The geometries of the surviving conformations are then refined at QM level. Finally, theoretical CCSs are calculated for the refined geometries using the projection approximation (PA), exact hard-sphere scattering (EHSS) and/or trajectory (TJ) methods within the MOBCAL code. The Boltzmann-averaged theoretical CCS is then calculated for each method from the ensemble of CCSs.

Assessment of QM theory level

Quantum Mechanics explicitly represents the electrons in a calculation and so are used in theoretical modelling to describe properties that depend on electronic properties. QM methods can be classified as “ab-initio” or semi-empirical. All electrons in the system are considered in detail for ab-initio calculations and the only physical constants required are the speed of light, Planck’s constant and masses of elementary particles. Only electrons in the valence shell are considered in detail in semiempirical methods to simplify the calculations - the core electrons are approximated with empirical parameters. Alternative modelling approaches are those based on density functional theory (DFT), where systems are described as functions based on charge density. Other approximations in QM methods involve the expression of molecular orbitals as linear combinations of a predefined set of one electron functions known as basis functions. The larger the basis set, the better the description of the system but also the higher computational cost.

Given the large number of conformations produced for each system analysed, especially in those with multiple ionization sites and several rotatable bonds, QM refinement may become a very computationally intensive procedure. In order to establish a cost-effective approach, the performance of different levels of QM theory and basis sets in producing conformers with CCSs comparable to 10 ACS Paragon Plus Environment

Page 10 of 27

Page 11 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


empirical CCSs were compared using the estradiol and naringenin glucuronides in their protonated forms. Approaches used included ab initio Hartree-Fock methods (HF/6-31+G** and HF/3-21G), density functional theory (DFT) (B3LYP/6-31+G** and B3LYP/3-21G) and the semi-empirical Hamiltonian AM1 method (Figure 2).

Results indicated that a QM refinement was needed to overcome some of the potential weaknesses of the force-fields applied in the conformation generation stage – possibly in the description of electronic interactions. For example, the PFROSST force-field used within LowModeMD produced conformers with comparatively larger CCSs for naringenin-4’-glucuronide (17.9 % larger). Here, the PFROSST systematically assumed the carboxylic group present within the glucuronide group would interact by means of an internal hydrogen bond with a collateral hydroxyl group (for the lowest energy protomer). However with QM refinement, a contrary interaction was observed; the carboxylic group interacting with hydroxyl groups on the opposite side of the molecule, thus, moving the glucuronide group out of plane with the molecule, forming a more compact molecule and therefore reducing its CCS to 0.8 - 2.5 % parity with experiment (see Figure 4A). Additionally, QM refinement was shown to be critical for both naringenin and estradiol glucuronide isomer pairs - when omitted, the CCS orderings were incorrectly predicted whereas CCS rank order was always predicted correctly for each isomer pair with QM refinement applied.

The use of ab-initio methods, HF or DFT theory, in conformation refinement did not lead to a significant improvement in predictive power compared to semi-empirical AM1 (Figure 2). Therefore, and for the sake of efficiency and cost, AM1 was selected as the QM method of choice in our protocol.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Impact of protonation on structural isomer gas phase conformation

In order to understand the impacts of protonation site on structural isomer conformation, 32 protomers from seven different drug-like molecules were investigated (Figure S-4). All molecules demonstrated differences in energy between protomers.

Metformin and the naringenin glucuronides expressed large theoretical CCS differences between protomers. In both cases, the relatively higher energy protomers provided a larger CCS compared to the lower energy protomers, suggesting that protonation site can drastically alter gas phase shape and that smaller compact protomers are energetically more favourable. In contrast caffeine, atazanavir, and the β-estradiol glucuronide protomers did not differ significantly in theoretical CCS. Interestingly, and usefully, the lowest energy protomers in all cases gave theoretical CCSs closest to their empirical CCSs.

This demonstrates that protonation can alter the molecular shape (and energy) in both a protonation site and molecule specific manner; indicative of certain protonation sites being able to take part in intramolecular interactions which lead to energetically distinctive conformers. In the cases of both metformin and atazanavir, two protomers exist below the 2 kcal mol-1 energy cut-off and both would be predicted to exist under the experiment conditions; as gas phase CCSs are rotationally averaged (when measured within a weak-field limit) they would both contribute to the same arrival time and subsequently to the CCS. In all cases, the two lowest energy protomers gave very similar CCSs but this may not necessarily be the case for all molecules.

Additionally, the CCSs for modelled neutral conformers can differ substantially from their lowest energy protomers (Table S-1). For example, the empirical CCS for atazanavir was 198.5 Å2 and the theoretical CCS for its lowest energy protomers was ~196 Å2, however, the theoretical CCS computed


Page 12 of 27

Page 13 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


for its neutral molecule was 159.7 Å2. Interestingly, the theoretical CCS difference between the lowest energy protomer and the neutral conformer for both naringen-7-glucuronide and naringen-4’glucuronide were > 10 Å2 and < 2 Å2, respectively. However, for all other molecules investigated their neutral molecule gave similar CCS values (< 2 % difference) to their lowest energy protomers.

In summary, protonation site can have a striking but unpredictable effect on gas phase structure highlighting the importance of protonation incorporation during a computational protocol for theoretical molecule CCS generation.

Assessment of theoretical CCS calculation methods

The aptitude of the molecular modelling protocol combined with theoretical CCS calculations (using MOBCAL) for the prediction of empirical CCSs was evaluated. A total of twelve protonated and seven deprotonated rigid drug-like molecules within a range of masses (124-822 Da) and DTCCSHe’s (~60-210 Å2) were used for this assessment (Figure 3 and Table S-2). EHSS and PA theoretical CCSs (CCST) correlate well with DTCCSHe for both protonated and deprotonated drug-like molecule ions. However, although the values correlate well both the EHSS and PA method overestimate their CCSs (EHSS being considerably larger), as reported for other molecules29,45. The correlation (R2 = 0.999) observed reflects that the DTCCSHe can be well-estimated for small molecules. It should be noted however that these correlations are for limited mass and CCS ranges where extrapolation outside their regions may not be linear. This relationship is also specific to drug-like molecules (and their inherent gas phase packing efficiency) and the computational procedure used here.

Additionally, the use of travelling wave IM-MS for the production of relevant CCSs was evaluated (Figures S-5 and S-6, Supporting Discussion), demonstrating good correlation (R2 = 0.99) with DTCCSHe’s.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The most detailed and expensive CCS calculation within MOBCAL, the TJ theory, as well as the Sigma method40 (based on the PA method) were also compared for the estradiol and naringenin glucuronide isomers, enabling a thorough comparison of the CCS calculations available (Table 1). The PA and TJ methods gave similar theoretical CCSs (< 1.5 % difference) even though the TJ uses a forcefield with Lennard-Jones and (optionally) Coulombic potentials. The apparent negligible influence of the more refined TJ method could be due to the parameterisation of the MOBCAL forcefield being sub-optimal for small molecules; attributable to its tuning on limited datasets32.

Glucuronide metabolites of both naringenin and estradiol were used to assess an empirical correction of the PA and EHSS values (CorrPA and CorrEHSS). Improved parity with empirical CCSs was found when the correction was applied (Table 1). For example, the protonated naringenin glucuronide metabolites gave notably poorer parity with the PA method compared to the empirically corrected CCSs.

Overall, these data validate that this modelling protocol can produce conformers for drug and druglike molecules that are comparable to empirical CCSs, as demonstrated by the high correlation between modelled drug-like molecules in Figure 3 (both protonated and deprotonated) and agreement between the protonated metabolite structural isomers in Table 1. However, a case is also presented which shows the failure of the modelling protocol: it failed to produce experimentally comparable deprotonated conformers for the naringenin glucuronide isomers. This failure is likely due to the molecular modelling procedure and not the CCS calculations. Performing the molecular modelling in vacuo may have neglected influential experimental features, such as the presence of buffer gas, preventing the theoretical production of the deprotonated conformations observed experimentally.


Page 14 of 27

Page 15 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Unequivocal identification of metabolite structural isomers from CCS information

It has been shown that structural isomers of drug-like molecules can be measured using IM-MS technology. However, to identify unknown structural isomers using CCS information, assessment criteria for discernment is required and the confidence between assignments must be evaluated or at least challenged. A harmonic penalty scoring function, S = ((DTCCSHe - CCST) / σexp)2, was applied, enabling the agreement (percentage difference) between theoretical (CCST) and empirical (DTCCSHe) CCSs in context of an empirical error (σexp) to be judged (Figure S-7) – this scoring function has previously been used for scoring protein models by their CCSs46.

To test the scoring function naringenin and estradiol glucuronide metabolites are assigned as ‘unknowns’ and the empirical error (σexp) as 2 % of the measured DTCCSHe. It has been stated that CCS accuracy for small molecules using DT-IM-MS technology can range between 0.5-2%; due to errors propagating from the inaccuracy in drift cell temperature and pressure determination, random error (from repeated inter-day measurements), systematic instrumental error (from replicate measurements or scans) and error from the linear fits used for the mobility calculation28,41.

The theoretical CCSs derived from molecular modelling of the potential isomers are then compared and scored against DTCCSHe’s. For ‘unknown 1’ substantially lower scores were achieved with the theoretical CCSs from the modelled naringenin-4’-glucuronide (PA = 1.5, CorrPA = 0.3, and CorrEHSS = 1.7) than for the modelled naringenin-7-glucuronide (PA = 47.0, CorrPA = 24.2, and CorrEHSS = 22.3) (Figure 4B). For ‘unknown 2’ lower scores were achieved for the modelled naringenin-7-glucuronide (CorrPA = 1.7, and CorrEHSS = 1.3) than for the modelled naringenin-4’glucuronide (CorrPA = 14.2, and CorrEHSS = 4.3) for empirically corrected DTCCSHe’s. The scoring method suggests that ‘unknown 1’ corresponds to naringenin-4’-glucuronide and that ‘unknown 2’ corresponds to naringenin-7-glucuronide; this is indeed the case. Interestingly, the score ranking for



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

the naringenin glucuronide isomers was always correct, except for the theoretical PA CCS for unknown 2, conveying the efficacy of the empirical corrections for correct isomer assignment.

In contrast, the scores for the estradiol glucuronide isomers (‘unknowns 3 and 4’) are both small and similar (Figure 4C). Additionally, the score rank order is inconsistent for both ‘unknowns’. The scoring method therefore suggests that estradiol glucuronide isomers, with an empirical error of 2 %, cannot be confidently identified using the CCS measurement alone.

The extent of CCS difference between structural isomers is derived from their respective gas phase shape: the lowest energy structure for naringenin-4’-glucuronide is distinctly compact compared to the naringenin-7-glucuronide, whereas the β-3-estradiol and β-17-estradiol glucuronides possess similar shapes, possibly due to rigidity of the estradiol ring system (Figure 4A).

The cases described here are ‘artificial’ as the CCSs for these molecules have been measured; however this test does present the challenges and limitations inherent with using CCS information for unknown molecule identification. Larger CCS differences between structural isomers will likely aid in their identification due to the penalty of the empirical error; the CCS differences between naringenin glucuronide structural isomers and between estradiol glucuronide structural isomers being ~6.6 % and ~2.6%, respectively, with the naringenin glucuronide isomers providing more discernible scoring. It is therefore likely that CCS differences between structural isomers would necessitate being upward of ~6%, although a larger data set would be required to understand the true benchmark of the current technology.


Page 16 of 27

Page 17 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Conclusion

This work demonstrates that IM-MS technology has the ability to distinguish structural isomers of drug metabolites and rationalize their structures. For structural isomers to be distinguished by IM-MS the isomers themselves must conform to energetically distinct gas phase conformations. The glucuronide drug metabolite structural isomers demonstrate both discernible and indiscernible differences between their CCSs. The adoption of different CCSs likely arises from different gas phase shapes, with their ability to form different intramolecular bonding encouraging distinct structures.

The conformational space analysis of drug-like molecules indicates that intramolecular bonding and gas phase packing will become more prevalent with increasing mass of the molecules. To enforce more flexibility and increase the chance of characteristic intramolecular bonding between structural isomers enhancements could be made, for example through chemical derivitisation30. Interestingly, different ionization state and adduct formation can lead to different gas phase conformers; increasing the conformational space available for investigation. Further molecular modelling and scoring method developments for a variety of adducts (e.g. sodium and ammonium) will increase the arsenal of information available for IM-MS structural isomer identification.

The molecular modelling approach developed performed well for drug and drug-like molecules as well as for structural isomers. Utilizing LowModeMD in combination with semi-empirical QM calculations made the protocol highly efficient and computationally cheaper than other protocols. Comparisons between PA, EHSS and TJ theoretical CCS calculations showed that the extra refinement and computational cost of TJ, as used by Lapthorn et al for the evaluation of small pharmaceutically relevant molecules32, provides similar results to the much cheaper PA method – both overestimating the empirical CCSs. However, the empirically scaled PA and EHSS CCSs, as well as the empirically scaled TJ method derived by Lapthorn et al32, provide enhanced parity between theoretical and empirical CCSs demonstrating its utility.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

In conclusion, IM-MS technology has been shown to offer useful structural information for drug metabolite structural isomers, with the potential to provide an alternative or complementary method to NMR. Prerequisite for identification is the computational modelling which can be used successfully to predict drug-like molecule conformers inexpensively, enabling it to be potentially utilized as an in silico screening strategy for structural differences, and their extents. Therefore, a pre-IM-MS in silico analysis could be performed to rationalize whether structural isomers have the potential to produce different gas phase shapes. Additionally, this protocol may prove most useful in filtering candidate structures when many metabolite structures are plausible, by identifying the isomers that are highly unlikely within the restraint of the measurement. Ultimately, IM-MS proves to be a strong addition to the metabolite identification armoury.


Page 18 of 27

Page 19 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Supporting Information Supporting experimental, discussion, figures, and tables can be found here.

Acknowledgements We acknowledge Erik Marklund for calculating CCSs using the Sigma code. The authors declare no competing financial interests.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

References (1) Iyanagi, T. Int. Rev. Cytol. 2007, 260, 35-112. (2) Caldwell, J.; Gardner, I.; Swales, N. Toxicol. Pathol. 1995, 23, 102-114. (3) Dear, G. J.; Beaumont, C.; Roberts, A.; Squillaci, B.; Thomas, S.; Nash, M.; Fraser, D. Bioanalysis 2011, 3, 197-213. (4) Beaumont, C.; Young, G. C.; Cavalier, T.; Young, M. A. Br. J. Clin. Pharmacol. 2014, 78, 1185-1200 (5) Moco, S.; Vervoort, J.; Moco, S.; Bino, R. J.; De Vos, R. C. H.; Bino, R. TrAC 2007, 26, 855-866. (6) Dear, G. J.; Munoz-Muriedas, J.; Beaumont, C.; Roberts, A.; Kirk, J.; Williams, J. P.; Campuzano, I. Rapid Commun. Mass Spectrom. 2010, 24, 3157-3162. (7) Dear, G. J.; Roberts, A. D.; Beaumont, C.; North, S. E. J. Chromatogr. B 2008, 876, 182-190. (8) Bingol, K.; Bruschweiler-Li, L.; Yu, C.; Somogyi, A.; Zhang, F.; Bruschweiler, R. Anal. Chem. 2015. (9) Lanucara, F.; Holman, S. W.; Gray, C. J.; Eyers, C. E. Nat. Chem. 2014, 6, 281-294. (10) Kliman, M.; May, J. C.; McLean, J. A. Biochim. Biophys. Acta 2011, 1811, 935-945. (11) Harvey, S. R.; Macphee, C. E.; Barran, P. E. Methods 2011, 54, 454-461. (12) Zekavat, B.; Solouki, T. J. Am. Soc. Mass. Spectrom. 2012, 23, 1873-1884. (13) Bush, M. F.; Campuzano, I. D.; Robinson, C. V. Anal. Chem. 2012, 84, 7124-7130. (14) Dwivedi, P.; Schultz, A. J.; Jr, H. H. H. Int. J. Mass. Spectrom. 2010, 298, 78-90. (15) Dwivedi, P.; Bendiak, B.; Clowers, B. H.; Hill, H. H., Jr. J. Am. Soc. Mass Spectrom. 2007, 18, 11631175. (16) Struwe, W. B.; Benesch, J. L.; Harvey, D. J.; Pagel, K. Analyst 2015. (17) Ruotolo, B. T.; Benesch, J. L.; Sandercock, A. M.; Hyung, S. J.; Robinson, C. V. Nat. Protoc. 2008, 3, 1139-1152. (18) Chan, E. C.; New, L. S.; Yap, C. W.; Goh, L. T. Rapid Commun. Mass Spectrom. 2009, 23, 384-394. (19) Paglia, G.; Williams, J. P.; Menikarachchi, L.; Thompson, J. W.; Tyldesley-Worster, R.; Halldórsson, S.; Rolfsson, O.; Moseley, A.; Grant, D.; Langridge, J.; Palsson, B. O.; Astarita, G. Anal. Chem. 2014, 86, 3985-3993.


Page 20 of 27

Page 21 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(20) Pagel, K.; Harvey, D. J. Anal. Chem. 2013, 85, 5138-5145. (21) Williams, J. P.; Grabenauer, M.; Holland, R. J.; Carpenter, C. J.; Wormald, M. R.; Giles, K.; Harvey, D. J.; Bateman, R. H.; Scrivens, J. H.; Bowers, M. T. Int. J. Mass. Spectrom. 2010, 298, 119-127. (22) Kanu, A. B.; Dwivedi, P.; Tam, M.; Matz, L.; Hill, H. H., Jr. J. Mass Spectrom. 2008, 43, 1-22. (23) Mason, E. A.; Schamp, H. W. Ann. Phys. 1958, 4, 233-270. (24) Shvartsburg, A. A.; Jarrold, M. F. Chem. Phys. Lett. 1996, 261, 86-91. (25) Mesleh, M. F.; Hunter, J. M.; Shvartsburg, A. A.; Schatz, G. C.; Jarrold, M. F. J. Phys. Chem. 1996, 100, 16082-16086. (26) Clemmer, D. E.; Jarrold, M. F. J. Mass Spectrom. 1997, 32, 577-592. (27) Clemmer, D. E.; Hudgins, R. R.; Jarrold, M. F. J. Am. Chem. Soc. 1995, 117, 10141-10142. (28) Campuzano, I.; Bush, M. F.; Robinson, C. V.; Beaumont, C.; Richardson, K.; Kim, H.; Kim, H. I. Anal. Chem. 2012, 84, 1026-1033. (29) Zakharova, N. L.; Crawford, C. L.; Hauck, B. C.; Quinton, J. K.; Seims, W. F.; Hill, H. H., Jr.; Clark, A. E. J. Am. Soc. Mass Spectrom. 2012, 23, 792-805. (30) Shimizu, A.; Chiba, M. Drug Metab. Dispos. 2013, 41, 1295-1299. (31) Shimizu, A.; Ohe, T.; Chiba, M. Drug Metab. Dispos. 2012, 40, 1456-1459. (32) Lapthorn, C.; Pullen, F. S.; Chowdhry, B. Z.; Wright, P.; Perkins, G. L.; Heredia, Y. Analyst 2015 (33) Wu, Z.; Gao, W.; Phelps, M. A.; Wu, D.; Miller, D. D.; Dalton, J. T. Anal. Chem. 2004, 76, 839-847. (34) Pringle, S. D.; Giles, K.; Wildgoose, J. L.; Williams, J. P.; Slade, S. E.; Thalassinos, K.; Bateman, R. H.; Bowers, M. T.; Scrivens, J. H. Int. J. Mass. Spectrom. 2007, 261, 1-12. (35) Bush, M. F.; Hall, Z.; Giles, K.; Hoyes, J.; Robinson, C. V.; Ruotolo, B. T. Anal. Chem. 2010, 82, 9557-9565. (36) In Molecular Operating Environment (MOE); Chemical Computing Group Inc.: 101 Sherbrooke St. West, Suite #910, Montreal, QC, Canada, H3A 2R7, 2013. (37) Chen, I. J.; Foloppe, N. J. Chem. Inf. Model. 2008, 48, 1773-1791. (38) Labute, P. J. Chem. Inf. Model. 2010, 50, 792-800. 21 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(39) Gaussian 09, Revision D.01, M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery, Jr., J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, Ö. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski, and D. J. Fox, Gaussian, Inc., Wallingford CT, 2009. (40) Wyttenbach, T.; von Helden, G.; Batka, J.; Carlat, D.; Bowers, M. J. Am. Soc. Mass Spectrom. 1997, 8, 275-282. (41) May, J. C.; Goodwin, C. R.; Lareau, N. M.; Leaptrot, K. L.; Morris, C. B.; Kurulugama, R. T.; Mordehai, A.; Klein, C.; Barry, W.; Darland, E.; Overney, G.; Imatani, K.; Stafford, G. C.; Fjeldsted, J. C.; McLean, J. A. Anal. Chem. 2014, 86, 2107-2116. (42) McLean, J. A. J. Am. Soc. Mass Spectrom. 2009, 20, 1775-1781. (43) Dupuis, N. F.; Wu, C.; Shea, J. E.; Bowers, M. T., J. Am. Chem. Soc. 2009, 131, 18283-18292. (44) Jakalian, A.; Jack, D. B.; Bayly, C. I. J. Comput. Chem. 2002, 23, 1623-1641. (45) Siu, C. K.; Guo, Y.; Saminathan, I. S.; Hopkinson, A. C.; Siu, K. W. J. Phys. Chem. B 2010, 114, 1204-1212. (46) Hall, Z.; Politis, A.; Robinson, C. V. Structure 2012, 20, 1596-1609.


Page 22 of 27

Page 23 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 1. CCSs of drug glucuronide metabolite structural isomers with different adducts and polarity (A) Structures of drug glucuronide metabolite structural isomers. Overlays of representative arrival time distributions (ATDs) at 50V drift voltage for (B) naringenin-4’-glucuronide (4’) and naringenin7-glucuronide (7), MW (Da) = 448.5 and for (C) β-3-estradiol-glucuronide (3) and β-17-estradiolglucuronide (17), MW (Da) = 448.5. ATD full width half maxima (DTFWHM) and DTCCSHe averages are inset. The errors quoted are the standard deviations from at least two inter-day repeated measurements.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 2. Comparison of QM theory level for molecule conformer refinement. All theoretical CCSs shown are calculated using the PA method (CCSPA).


Page 24 of 27

Page 25 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 3. Parity of theoretical CCSs with DTCCSHe values. Both CCSPA and CCSEHSS were calculated using the MOBCAL code and the Boltzmann average values calculated. Protonated ([M+H]+) and deprotonated ([M-H]-) drug-like molecules possessed a mass range of 122-704 Da and 123-607 Da, respectively (Table S-2). A linear regression analysis (y = mx+c), with a fixed intercept of c = 0, was used to establish a scaling factor to bring the theoretical and empirical CCSs into parity (insets). The fit statistics are - [M+H]+ CCSPA: m = 0.966 ± 0.0094, N = 12, R2 = 0.999, and %RMSD = 2.9%. [M+H]+ CCSEHSS: m = 0.885 ± 0.0073, N = 11, R2 = 0.999, and %RMSD = 2.4%. [M-H]- CCSPA: m = 0.998 ± 0.0137, N = 7, R2 = 0.999, and %RMSD = 4.9%. [M-H]- CCSEHSS: m = 0.919 ± 0.0121, N = 7, R2 = 0.999 and %RMSD = 4.7%.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 4. IM-MS for identifying unknown metabolites. (A) Structures of the lowest energy confirmations after molecular modelling and AM1 level QM optimization. Atoms are coloured green, red and white for carbon, oxygen and hydrogen respectively. (B) Comparison of ‘unknown’ CCSs for the naringenin glucuronide structural isomers with computational derived CCSs expressed by their scores for different in silico CCS calculations. (C) Comparison of ‘unknown’ CCSs for the β-estradiol glucuronide structural isomers with computational derived CCSs expressed by their scores for different in silico CCS calculations. Score values < 2 are annotated.


Page 26 of 27

Page 27 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table 1. Summary of CCSs for β-estradiol glucuronide and naringenin glucuronide structural isomers and their percentage differences with theoretical CCSs (CCST), for both protonated and deprotonated ions. All theoretical values are Boltzmann averages of molecular ensembles generated as described in Figure S-2. 2

CCS (Å ) ∆CCST – DTCCSHe (%)

Glucuronide isomer

β-3-estradiol [M+H]+ β-17-estradiol [M+H]+ β-3-estradiol [M-H]

-

β-17-estradiol [M-H]

-

Naringenin-4’ [M+H]+ Naringenin-7 [M+H]+ Naringenin-4’ [M-H] Naringenin-7 [M-H]

-

-

DTCCSHe

CCSPA

CCSEHSS

CCSTJ

CCSSigma

CCSCorrPA

CCSCorrEHSS

136.7 ± 0.57 140.3 ± 0.74 143.9 ± 1.46 139.2 ± 0.21 124.0 ± 0.46 132.7 ± 0.26 124.0 ± 0.37 124.5 ± 0.54

142.7 4.4 143.4 2.2 142.4 -1.0 140.8 1.1 127.0 2.4 141.0 6.3 140.3 13.1 140.8 13.1

154.8 13.2 154.7 10.3 153.4 6.6 151.9 9.1 143.8 16.0 153.4 15.6 151.5 22.2 153.6 23.4

143.8 5.3 142.5 1.6 141.6 -1.6 141.3 1.5

149.0 9.0 150.0 6.9 149.2 3.7 148.2 6.5

-

-

-

-

-

-

-

-

137.8 0.8 138.5 -1.3 142.1 -1.3 140.5 0.9 122.7 -1.0 136.2 2.6 140.0 12.9 140.5 12.9

137.0 0.2 136.9 -2.5 141.0 -2.0 139.6 0.3 127.3 2.6 135.8 2.3 139.2 12.3 141.2 13.4


Elucidation of Drug Metabolite Structural Isomers ... - ACS Publications

Recommend Documents