View Article Online View Journal
Analyst Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: C. Lapthorn, F. Pullen, B. Chowdhry, P. Wright, G. Perkins and Y. Heredia, Analyst, 2015, DOI: 10.1039/C5AN00411J.
This is an Accepted Manuscript, which has been through the Royal Society of Chemistry peer review process and has been accepted for publication. Accepted Manuscripts are published online shortly after acceptance, before technical editing, formatting and proof reading. Using this free service, authors can make their results available to the community, in citable form, before we publish the edited article. We will replace this Accepted Manuscript with the edited and formatted Advance Article as soon as it is available. You can find more information about Accepted Manuscripts in the Information for Authors. Please note that technical editing may introduce minor changes to the text and/or graphics, which may alter content. The journal’s standard Terms & Conditions and the Ethical guidelines still apply. In no event shall the Royal Society of Chemistry be held responsible for any errors or omissions in this Accepted Manuscript or any consequences arising from the use of any information it contains.
www.rsc.org/analyst
Page 1 of 13
Analyst View Article Online
DOI: 10.1039/C5AN00411J
How useful is molecular modelling in combination with ion mobility mass spectrometry for ‘small molecule’ ion mobility collision cross-sections? a
a
a
a
b
Cris Lapthorn* , Frank Pullen , Babur Z. Chowdhry , Patricia Wright ,George Perkins and Yanira c Heredia a Faculty of Engineering & Science, University of Greenwich, Medway Campus, Chatham, Kent ME4 4TB, UK b 149 Hickory Corner Road, Milford, NJ c Department of Hematology. Hospital 12 de Octubre. Instituto de Investigación 12 de Octubre, Madrid, Spain. * Corresponding author: E-mail: [email protected]
(Abstract) Ion mobility mass spectrometry is used to measure the drift-time of an ion. The drift-time of an ion can be used to calculate the collision cross-section (CCS) in travelling wave ion mobility (e.g. Waters Synapt and Vion instruments) or directly determine the experimental CCS (e.g. Agilent 6560 instrument and many drift-tube instruments). A comparison of the experimental CCS and theoretical CCS values obtained from trajectory method He(g) parameterised MOBCAL and N2(g) parameterised MOBCAL software, for a range of 20 ‘small molecules’ is presented. This study utilises density functional theory B3LYP methods and the 6-31G+(d,p) basis set to calculate theoretical CCS values. This study seeks to assess the accuracy of a common procedure using CCS calibration with poly(D/L)-alanine derived from drift-cell measurements and the original release of MOBCAL software and compare it with recent improvements with a drug-like molecule calibration set and a revision of MOBCAL parameterised for N2(g) drift gas. This study represents one of the first quantitative evaluations of the agreement between theoretical CCS and experimental CCS values for a range of small pharmaceutically relevant molecules using travelling wave ion mobility mass spectrometry. Accurate theoretical CCS may allow optimisation of ion mobility separations in silico, provide CCS databases that can confirm structures without the need for alternative analytical tools such as nuclear magnetic resonance spectroscopy (NMR) and assignment of unknowns and positional isomers without requiring reference materials. Keywords: ion mobility mass spectrometry, tandem mass spectrometry, fragmentation mechanisms, protonation, proton affinity, MOBCAL, molecular modelling, ab initio, projection approximation, trajectory method.
Introduction The workflow and computational time required for ‘small molecules’ is such that it is feasible to calculate many theoretical collision cross-sections (tCCSs) in under 24 hours using commonly available computing resources. It is then possible to compare the tCCS and experimental CCS (eCCS) values to gain further information, as shown in Fig. 1. The increase in computing resources available to scientists and consumers enable tCCSs of ions to be calculated relevant for large sets of molecules, in contrast to many large biomolecule challenges where molecular size and complexity often limits throughput.
Analyst Accepted Manuscript
Published on 23 June 2015. Downloaded by University of Michigan Library on 23/06/2015 16:38:39.
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
Analyst
Page 2 of 13 View Article Online
Published on 23 June 2015. Downloaded by University of Michigan Library on 23/06/2015 16:38:39.
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
Fig 1 General workflow for combining molecular modelling with ion mobility mass spectrometry
However, there does not appear to be a consensus or evidence via blind trial to establish whether there is good agreement between eCCSs and tCCSs and which tCCS calculation methods provide the closest matches to eCCS values. Indeed, recent publications1–8 on ‘small molecules’ have utilised various and diverse approaches to calculate tCCSs, a summary of which is shown in Table 1. Table 1. The approaches evaluated in this study are shown in italics, as well as alternative approaches to predicting tCCS values Geometry optimisation Density Functional Theory Molecular dynamics Semi-Empirical Molecular mechanics
Mobility calculation Trajectory Method (T.M.) Exact Hard Sphere Scattering Projection Approximation -
Software MOBCAL Sigma Driftscope -
Partial atomic charge Electrostatic potential Mulliken -
Overall this study seeks to establish a baseline for comparing tCCS and eCCS values based on existing protocols. This study does not seek to optimise and improve on existing protocols, but instead to benchmark them. This may provide evidence for practitioners to reflect on the results and accuracy from previous publications and help understand which factors may exacerbate differences between tCCS and eCCS values. A structurally diverse set of ‘small molecule’ type compounds (see Fig. 2), including pharmaceutically relevant compounds was studied. The number of protonation sites, range of m/z, and sub-structures were varied, as shown in Fig 3. The compounds studied include pregabalin (Lyrica®) and sildenafil (Viagra®) hence this study is especially relevant to drug-like molecules.
Analyst Accepted Manuscript
DOI: 10.1039/C5AN00411J
Page 3 of 13
Analyst View Article Online
DOI: 10.1039/C5AN00411J
3 8 1 7
6
7
CH3
NH
5
3
HO 1
9
4
NH
2
OH
11
5
12
6
8
O 9
15
11
1
4
H3C
9
Phenacetin
8
CH2 13
10
7
13
H3C
19
3
O
N
8
19
16
*
23
CH2 10
3
*OH
4a
N 1
H H
11a 7a
*
17
*NH
18
H3C
13
11
Cl
17
14
3
*
26
1
19
2
CH3OH 18
16
O
*
22
O
Cortisone
24
NH
26
32
33
*
25 27
O
24
19
22
O
30
O 1
1
10
8
CH3 18
15
CH3 23
N
14
5
16
4
7
*
31
25
O
24
22
30
26 27
29
2
7
Doxazosin
29
O
28
32
9
H3C 10 17
*
20 1 14 19
N H
13
CH3
4
H
20a 21
3 7 5
H
4a
H
5a
O
23
CH3 24
22
33
O 6
27
H
28
32
19
36
14
O
O
41
N
*
O
43
CH3
Reserpine
44
N H3C 9
31
15
20
N
28
27
32
26
16
7
O
10
2
4
*
1
S
33
17
11
8
6 5
CH3
O
12
42
6a
O
CH3
18
CH3
37 38
7a O O 29 3 H 25 CH 26
10
O
34
3a 8
12
O
30
H
33
Verapamil
9
13 39 35
2
N
18
H3C
40
9
15 11
CH3
O
8
16
2
25
17
1
O 10
H2N
11
Amlodipine
3
31
31
13
6
H3C
5
6
20
21
CH3 12
11
19 4
N
23
H3C
*NH
8
H3C
23
24
20
O
18
O
H3C
9
O
NH 6
21
2
17
7 2
5
11
N
3
13
16
O
H3C
11
N
28
13
3
22
Trichlormethiazide
CH3
O 12
4 28
12
12
N
N
27
2
O
16
O 14 15
Cl
1
O
22
17
18
S
7
15
20
20
21
S
H2N
9 17 CH3 H 5a
19
6
-
23
15
21 O 26
18
Cl
3
8
O
CH3
20
19
16
4
O
5
14
14 13
5
9
25
OH
5
6
10
4
6
CH3 7
H N
+
N
Clozapine n-oxide
Cl
15
H H
4
*
6
Sulpiride
11
24
3 1
*
8
18
Cl
19
4a 3a
H3C
1
8
6
12
O
2
N
10
O
19
2
N
N H
20
5
9
O
5
7
4
13 7
9
3
14
12
11 10
21
21
13
11
8
N
O
12
14
4
7
Cinchonine
23
16
15
2
12
14
20
H
8
13
S
22
9
H
22
15
NH2 O
16
O*
21
17
21
10
CH3
Trimethoprim
3a
15
11
7
Ondansetron
21
20
*N
N
10 9
4
18
17 16
12
NH2
3
5-(p-methylphenyl)-5-phenylhydantoin 19
15
8
5
4
20
H3C
14 13
1
O
N H
13 18
*O
5
*
7
N
3
2
H2N
14
18
1
4
15
18
19
22
6
2
N
*
N
5
O
H N
15
6
12
17
17
O
20
9
16
H3C
10
1
17
Alprenolol H3C
11
O
2
CH3
H3C
Naproxen
20
11 17
15 16
18
12
3
13a
6
2 11
H3C
14
3
*
NH
14
7
12
H
13
10
9
O
CH3
2
5
14
CH3
8 1
4
6 13
CH3
6
10
16
17
7
12
5
Ephedrine 5
OH 7
*O
O
1
10
16
6
7a
10
4
9
H
9a
Pregabalin 8
*
4
H3C
3
O
H
11
9
5a
Acetaminophen
NH
H3C
CH3
H
N-ethylaniline
1
9
10
8
3
8
7
NH
7 5
4
11
12
*
6
CH3
9
2
CH3 OH
NH2
3
8
5
*
1
2
O
10
6
*
OH
*O
2
2 4
HN
21
22
23
*N
N
25
24
3
Sildenafil
O
29
CH3 30
*
Fig 2 Structures of all compounds studied (* marks lowest energy structures or those within 5 kcal/mol of lowest energy structure, inferring most basic site(s))
Analyst Accepted Manuscript
Published on 23 June 2015. Downloaded by University of Michigan Library on 23/06/2015 16:38:39.
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
Analyst
Page 4 of 13 View Article Online
DOI: 10.1039/C5AN00411J
a)
b)
Fig 3 a) Histogram showing range of m/z values b) Histogram showing spread of number of protonation sites
For calculations using the original He(g) MOBCAL code, it was not possible to calculate tCCS for nine of the compounds without modification to the original MOBCAL code or calibration protocol. Specifically the first 6 tCCSs data-points in Table 2 are omitted because the poly-(D/L)-alanine calibration does not cover that CCS range; so it may be questioned whether an experimental measurement is valid since it will be outside the eCCS calibration range. The remaining 3 tCCSs data-points were omitted because the original He(g) MOBCAL code does not include Cl. It is possible to modify the MOBCAL code before compilation to include Cl (and other atom types if desired), this would require either: 1) assuming that the Cl atom type has identical parameters to Si, which is unsupported by any experimental evidence that we are aware of and does not represent common past practice, or 2) measure and tune the MOBCAL parameters for Cl with representative compounds; which was felt to be outside of the scope of this study and does not represent common past practice. The exclusion of some compounds using the original He(g) code highlights the usefulness of the druglike calibration set and N2(g) MOBCAL code which includes parameterisation of elements including Cl 9 and F and a larger m/z range from 122 to 609 .
Analyst Accepted Manuscript
Published on 23 June 2015. Downloaded by University of Michigan Library on 23/06/2015 16:38:39.
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
Page 5 of 13
Analyst View Article Online
DOI: 10.1039/C5AN00411J
Table 2. Summary of tCCS calculations conducted.
Identity
MW (Da) Molecular formula
He(g) MOBCAL
N2(g) MOBCAL
N-Ethylaniline
122
C8H11N
n
y
Acetaminophen
152
C8H9NO2
n
y
Pregabalin
160
C8H17NO2
n
y
Ephedrine
166
C10H15NO
n
y
Phenacetin
180
C10H13NO2
n
y
Naproxen
230
C14H14O3
n
y
Alprenolol
250
C15H23NO2
y
y
5-(p-Methylphenyl)-5-phenylhydantoin
267
C16H14N2O2
y
y
Trimethoprim
291
C14H18N4O3
y
y
Ondansetron
294
C18H19N3O
y
y
Cinchonine
295
C19H22N2O
y
y
Sulpiride
342
C15H23N3O4S
y
y
Clozapine N-oxide
343
C18H19ClN4O
n
y
Cortisone
361
C21H28O5
y
y
Trichlormethiazide
380
C8H8Cl3N3O4S2
n
y
Amlodipine
409
C20H25ClN2O5
n
y
Doxazosin
452
C23H25N5O5
y
y
Verapamil
455
C27H38N2O4
y
y
Sildenafil
475
C22H30N6O4S
y
y
Reserpine
609
C33H40N2O9
y
y
count =
11
20
% calculated
55%
100%
Results and discussion
Experimental vs theoretical CCS for N2(g) calibration protocol On plotting the tCCS values for the lowest energy structure or the average tCCS of structures within 5 kcal/mol of the lowest energy structure, vs eCCS, there is a linear correlation between the eCCS and tCCS with an R2 value > 0.99 (see Fig 4) showing a good linear correlation. In contrast to previously 9 described results where the slope of the line is very close to 1 (i.e. y = mx + c where m is 1) our results demonstrate a linear correlation that has a defined slope of ~1.25, and a constant, derived from an apparent larger tCCS than eCCS for many ions. Our study includes additional compounds to the N-ethylaniline, acetaminophen, alprenolol, ondansetron, clozapine N-oxide, verapamil and 20 reserpine described in previous results . Using the same level of theory, basis set and parameters to calculate the optimised geometries using Gaussian 09, many of the of the same calibration compounds9 and running the supplied N2(g) MOBCAL script we obtained tCCS values that appear to be well described by a linear regression analysis with an excellent R2 value of >0.99.
Analyst Accepted Manuscript
Published on 23 June 2015. Downloaded by University of Michigan Library on 23/06/2015 16:38:39.
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
Analyst
Page 6 of 13 View Article Online
Published on 23 June 2015. Downloaded by University of Michigan Library on 23/06/2015 16:38:39.
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
Fig 4 Linear regression of eCCS and tCCS for N2(g) MOBCAL
We postulate that there may be two or more contributing factors for differences between tCCS and eCCS: 1) It is conceivable that the parameterisation of interactions in MOBCAL software versions has been sub-optimal as tuning was based on a limited dataset due to computing demands at the time when the software was written (c. 1997)10,11 and on relatively small datasets since9. The slope of ~1.2 obtained herein is also corroborated by the slope of 1.2378 found in the 12 computationally predicted CCS of 125 common metabolites and slopes of up to ~1.2 13 obtained for saxitoxins both using a similar N2(g) MOBCAL T.M. methodology. A subset consisting of molecules from this dataset where the number of rotatable bonds were 3 or less (namely N-ethylaniline, Acetaminophen, Ephedrine, Phenacetin, Naproxen, 5-(pmethylphenyl)-5-phenylhydantoin, Ondansetron, Cinchonine, Clozapine N-oxide, Cortisone and Trichlormethiazide) were selected and their tCCSs plotted against eCCS (see Fig 5). For structures with smaller numbers of rotatable bonds (
Analyst Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: C. Lapthorn, F. Pullen, B. Chowdhry, P. Wright, G. Perkins and Y. Heredia, Analyst, 2015, DOI: 10.1039/C5AN00411J.
This is an Accepted Manuscript, which has been through the Royal Society of Chemistry peer review process and has been accepted for publication. Accepted Manuscripts are published online shortly after acceptance, before technical editing, formatting and proof reading. Using this free service, authors can make their results available to the community, in citable form, before we publish the edited article. We will replace this Accepted Manuscript with the edited and formatted Advance Article as soon as it is available. You can find more information about Accepted Manuscripts in the Information for Authors. Please note that technical editing may introduce minor changes to the text and/or graphics, which may alter content. The journal’s standard Terms & Conditions and the Ethical guidelines still apply. In no event shall the Royal Society of Chemistry be held responsible for any errors or omissions in this Accepted Manuscript or any consequences arising from the use of any information it contains.
www.rsc.org/analyst
Page 1 of 13
Analyst View Article Online
DOI: 10.1039/C5AN00411J
How useful is molecular modelling in combination with ion mobility mass spectrometry for ‘small molecule’ ion mobility collision cross-sections? a
a
a
a
b
Cris Lapthorn* , Frank Pullen , Babur Z. Chowdhry , Patricia Wright ,George Perkins and Yanira c Heredia a Faculty of Engineering & Science, University of Greenwich, Medway Campus, Chatham, Kent ME4 4TB, UK b 149 Hickory Corner Road, Milford, NJ c Department of Hematology. Hospital 12 de Octubre. Instituto de Investigación 12 de Octubre, Madrid, Spain. * Corresponding author: E-mail: [email protected]
(Abstract) Ion mobility mass spectrometry is used to measure the drift-time of an ion. The drift-time of an ion can be used to calculate the collision cross-section (CCS) in travelling wave ion mobility (e.g. Waters Synapt and Vion instruments) or directly determine the experimental CCS (e.g. Agilent 6560 instrument and many drift-tube instruments). A comparison of the experimental CCS and theoretical CCS values obtained from trajectory method He(g) parameterised MOBCAL and N2(g) parameterised MOBCAL software, for a range of 20 ‘small molecules’ is presented. This study utilises density functional theory B3LYP methods and the 6-31G+(d,p) basis set to calculate theoretical CCS values. This study seeks to assess the accuracy of a common procedure using CCS calibration with poly(D/L)-alanine derived from drift-cell measurements and the original release of MOBCAL software and compare it with recent improvements with a drug-like molecule calibration set and a revision of MOBCAL parameterised for N2(g) drift gas. This study represents one of the first quantitative evaluations of the agreement between theoretical CCS and experimental CCS values for a range of small pharmaceutically relevant molecules using travelling wave ion mobility mass spectrometry. Accurate theoretical CCS may allow optimisation of ion mobility separations in silico, provide CCS databases that can confirm structures without the need for alternative analytical tools such as nuclear magnetic resonance spectroscopy (NMR) and assignment of unknowns and positional isomers without requiring reference materials. Keywords: ion mobility mass spectrometry, tandem mass spectrometry, fragmentation mechanisms, protonation, proton affinity, MOBCAL, molecular modelling, ab initio, projection approximation, trajectory method.
Introduction The workflow and computational time required for ‘small molecules’ is such that it is feasible to calculate many theoretical collision cross-sections (tCCSs) in under 24 hours using commonly available computing resources. It is then possible to compare the tCCS and experimental CCS (eCCS) values to gain further information, as shown in Fig. 1. The increase in computing resources available to scientists and consumers enable tCCSs of ions to be calculated relevant for large sets of molecules, in contrast to many large biomolecule challenges where molecular size and complexity often limits throughput.
Analyst Accepted Manuscript
Published on 23 June 2015. Downloaded by University of Michigan Library on 23/06/2015 16:38:39.
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
Analyst
Page 2 of 13 View Article Online
Published on 23 June 2015. Downloaded by University of Michigan Library on 23/06/2015 16:38:39.
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
Fig 1 General workflow for combining molecular modelling with ion mobility mass spectrometry
However, there does not appear to be a consensus or evidence via blind trial to establish whether there is good agreement between eCCSs and tCCSs and which tCCS calculation methods provide the closest matches to eCCS values. Indeed, recent publications1–8 on ‘small molecules’ have utilised various and diverse approaches to calculate tCCSs, a summary of which is shown in Table 1. Table 1. The approaches evaluated in this study are shown in italics, as well as alternative approaches to predicting tCCS values Geometry optimisation Density Functional Theory Molecular dynamics Semi-Empirical Molecular mechanics
Mobility calculation Trajectory Method (T.M.) Exact Hard Sphere Scattering Projection Approximation -
Software MOBCAL Sigma Driftscope -
Partial atomic charge Electrostatic potential Mulliken -
Overall this study seeks to establish a baseline for comparing tCCS and eCCS values based on existing protocols. This study does not seek to optimise and improve on existing protocols, but instead to benchmark them. This may provide evidence for practitioners to reflect on the results and accuracy from previous publications and help understand which factors may exacerbate differences between tCCS and eCCS values. A structurally diverse set of ‘small molecule’ type compounds (see Fig. 2), including pharmaceutically relevant compounds was studied. The number of protonation sites, range of m/z, and sub-structures were varied, as shown in Fig 3. The compounds studied include pregabalin (Lyrica®) and sildenafil (Viagra®) hence this study is especially relevant to drug-like molecules.
Analyst Accepted Manuscript
DOI: 10.1039/C5AN00411J
Page 3 of 13
Analyst View Article Online
DOI: 10.1039/C5AN00411J
3 8 1 7
6
7
CH3
NH
5
3
HO 1
9
4
NH
2
OH
11
5
12
6
8
O 9
15
11
1
4
H3C
9
Phenacetin
8
CH2 13
10
7
13
H3C
19
3
O
N
8
19
16
*
23
CH2 10
3
*OH
4a
N 1
H H
11a 7a
*
17
*NH
18
H3C
13
11
Cl
17
14
3
*
26
1
19
2
CH3OH 18
16
O
*
22
O
Cortisone
24
NH
26
32
33
*
25 27
O
24
19
22
O
30
O 1
1
10
8
CH3 18
15
CH3 23
N
14
5
16
4
7
*
31
25
O
24
22
30
26 27
29
2
7
Doxazosin
29
O
28
32
9
H3C 10 17
*
20 1 14 19
N H
13
CH3
4
H
20a 21
3 7 5
H
4a
H
5a
O
23
CH3 24
22
33
O 6
27
H
28
32
19
36
14
O
O
41
N
*
O
43
CH3
Reserpine
44
N H3C 9
31
15
20
N
28
27
32
26
16
7
O
10
2
4
*
1
S
33
17
11
8
6 5
CH3
O
12
42
6a
O
CH3
18
CH3
37 38
7a O O 29 3 H 25 CH 26
10
O
34
3a 8
12
O
30
H
33
Verapamil
9
13 39 35
2
N
18
H3C
40
9
15 11
CH3
O
8
16
2
25
17
1
O 10
H2N
11
Amlodipine
3
31
31
13
6
H3C
5
6
20
21
CH3 12
11
19 4
N
23
H3C
*NH
8
H3C
23
24
20
O
18
O
H3C
9
O
NH 6
21
2
17
7 2
5
11
N
3
13
16
O
H3C
11
N
28
13
3
22
Trichlormethiazide
CH3
O 12
4 28
12
12
N
N
27
2
O
16
O 14 15
Cl
1
O
22
17
18
S
7
15
20
20
21
S
H2N
9 17 CH3 H 5a
19
6
-
23
15
21 O 26
18
Cl
3
8
O
CH3
20
19
16
4
O
5
14
14 13
5
9
25
OH
5
6
10
4
6
CH3 7
H N
+
N
Clozapine n-oxide
Cl
15
H H
4
*
6
Sulpiride
11
24
3 1
*
8
18
Cl
19
4a 3a
H3C
1
8
6
12
O
2
N
10
O
19
2
N
N H
20
5
9
O
5
7
4
13 7
9
3
14
12
11 10
21
21
13
11
8
N
O
12
14
4
7
Cinchonine
23
16
15
2
12
14
20
H
8
13
S
22
9
H
22
15
NH2 O
16
O*
21
17
21
10
CH3
Trimethoprim
3a
15
11
7
Ondansetron
21
20
*N
N
10 9
4
18
17 16
12
NH2
3
5-(p-methylphenyl)-5-phenylhydantoin 19
15
8
5
4
20
H3C
14 13
1
O
N H
13 18
*O
5
*
7
N
3
2
H2N
14
18
1
4
15
18
19
22
6
2
N
*
N
5
O
H N
15
6
12
17
17
O
20
9
16
H3C
10
1
17
Alprenolol H3C
11
O
2
CH3
H3C
Naproxen
20
11 17
15 16
18
12
3
13a
6
2 11
H3C
14
3
*
NH
14
7
12
H
13
10
9
O
CH3
2
5
14
CH3
8 1
4
6 13
CH3
6
10
16
17
7
12
5
Ephedrine 5
OH 7
*O
O
1
10
16
6
7a
10
4
9
H
9a
Pregabalin 8
*
4
H3C
3
O
H
11
9
5a
Acetaminophen
NH
H3C
CH3
H
N-ethylaniline
1
9
10
8
3
8
7
NH
7 5
4
11
12
*
6
CH3
9
2
CH3 OH
NH2
3
8
5
*
1
2
O
10
6
*
OH
*O
2
2 4
HN
21
22
23
*N
N
25
24
3
Sildenafil
O
29
CH3 30
*
Fig 2 Structures of all compounds studied (* marks lowest energy structures or those within 5 kcal/mol of lowest energy structure, inferring most basic site(s))
Analyst Accepted Manuscript
Published on 23 June 2015. Downloaded by University of Michigan Library on 23/06/2015 16:38:39.
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
Analyst
Page 4 of 13 View Article Online
DOI: 10.1039/C5AN00411J
a)
b)
Fig 3 a) Histogram showing range of m/z values b) Histogram showing spread of number of protonation sites
For calculations using the original He(g) MOBCAL code, it was not possible to calculate tCCS for nine of the compounds without modification to the original MOBCAL code or calibration protocol. Specifically the first 6 tCCSs data-points in Table 2 are omitted because the poly-(D/L)-alanine calibration does not cover that CCS range; so it may be questioned whether an experimental measurement is valid since it will be outside the eCCS calibration range. The remaining 3 tCCSs data-points were omitted because the original He(g) MOBCAL code does not include Cl. It is possible to modify the MOBCAL code before compilation to include Cl (and other atom types if desired), this would require either: 1) assuming that the Cl atom type has identical parameters to Si, which is unsupported by any experimental evidence that we are aware of and does not represent common past practice, or 2) measure and tune the MOBCAL parameters for Cl with representative compounds; which was felt to be outside of the scope of this study and does not represent common past practice. The exclusion of some compounds using the original He(g) code highlights the usefulness of the druglike calibration set and N2(g) MOBCAL code which includes parameterisation of elements including Cl 9 and F and a larger m/z range from 122 to 609 .
Analyst Accepted Manuscript
Published on 23 June 2015. Downloaded by University of Michigan Library on 23/06/2015 16:38:39.
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
Page 5 of 13
Analyst View Article Online
DOI: 10.1039/C5AN00411J
Table 2. Summary of tCCS calculations conducted.
Identity
MW (Da) Molecular formula
He(g) MOBCAL
N2(g) MOBCAL
N-Ethylaniline
122
C8H11N
n
y
Acetaminophen
152
C8H9NO2
n
y
Pregabalin
160
C8H17NO2
n
y
Ephedrine
166
C10H15NO
n
y
Phenacetin
180
C10H13NO2
n
y
Naproxen
230
C14H14O3
n
y
Alprenolol
250
C15H23NO2
y
y
5-(p-Methylphenyl)-5-phenylhydantoin
267
C16H14N2O2
y
y
Trimethoprim
291
C14H18N4O3
y
y
Ondansetron
294
C18H19N3O
y
y
Cinchonine
295
C19H22N2O
y
y
Sulpiride
342
C15H23N3O4S
y
y
Clozapine N-oxide
343
C18H19ClN4O
n
y
Cortisone
361
C21H28O5
y
y
Trichlormethiazide
380
C8H8Cl3N3O4S2
n
y
Amlodipine
409
C20H25ClN2O5
n
y
Doxazosin
452
C23H25N5O5
y
y
Verapamil
455
C27H38N2O4
y
y
Sildenafil
475
C22H30N6O4S
y
y
Reserpine
609
C33H40N2O9
y
y
count =
11
20
% calculated
55%
100%
Results and discussion
Experimental vs theoretical CCS for N2(g) calibration protocol On plotting the tCCS values for the lowest energy structure or the average tCCS of structures within 5 kcal/mol of the lowest energy structure, vs eCCS, there is a linear correlation between the eCCS and tCCS with an R2 value > 0.99 (see Fig 4) showing a good linear correlation. In contrast to previously 9 described results where the slope of the line is very close to 1 (i.e. y = mx + c where m is 1) our results demonstrate a linear correlation that has a defined slope of ~1.25, and a constant, derived from an apparent larger tCCS than eCCS for many ions. Our study includes additional compounds to the N-ethylaniline, acetaminophen, alprenolol, ondansetron, clozapine N-oxide, verapamil and 20 reserpine described in previous results . Using the same level of theory, basis set and parameters to calculate the optimised geometries using Gaussian 09, many of the of the same calibration compounds9 and running the supplied N2(g) MOBCAL script we obtained tCCS values that appear to be well described by a linear regression analysis with an excellent R2 value of >0.99.
Analyst Accepted Manuscript
Published on 23 June 2015. Downloaded by University of Michigan Library on 23/06/2015 16:38:39.
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
Analyst
Page 6 of 13 View Article Online
Published on 23 June 2015. Downloaded by University of Michigan Library on 23/06/2015 16:38:39.
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
Fig 4 Linear regression of eCCS and tCCS for N2(g) MOBCAL
We postulate that there may be two or more contributing factors for differences between tCCS and eCCS: 1) It is conceivable that the parameterisation of interactions in MOBCAL software versions has been sub-optimal as tuning was based on a limited dataset due to computing demands at the time when the software was written (c. 1997)10,11 and on relatively small datasets since9. The slope of ~1.2 obtained herein is also corroborated by the slope of 1.2378 found in the 12 computationally predicted CCS of 125 common metabolites and slopes of up to ~1.2 13 obtained for saxitoxins both using a similar N2(g) MOBCAL T.M. methodology. A subset consisting of molecules from this dataset where the number of rotatable bonds were 3 or less (namely N-ethylaniline, Acetaminophen, Ephedrine, Phenacetin, Naproxen, 5-(pmethylphenyl)-5-phenylhydantoin, Ondansetron, Cinchonine, Clozapine N-oxide, Cortisone and Trichlormethiazide) were selected and their tCCSs plotted against eCCS (see Fig 5). For structures with smaller numbers of rotatable bonds (
