Using ‘small molecule’ CCS from

Ion Mobility Mass Spectrometry

for ID and Prediction

Cris Lapthorn University of Greenwich

Background University of Bath - BSc Chemistry, with industrial year at Pfizer, Sandwich

Pfizer, Sandwich – Team Leader for Open-Access MS, NMR and Separations; >£3M facility

supporting 500,000 samples p.a., ~100 chemists, supervision for 5 FTEs.

Pfizer, Sandwich – Mass spectrometry specialist; supported oligonucleotide, chemical biology &

chemistry projects using high resolution MS techniques, proteomics and chemometrics.

Novartis, Horsham – Analytical Scientist, supported biology and research chemistry.

University of Greenwich - Head of Mass Spectrometry Services; supporting teaching and research

using mass spectrometry and providing consultancy services to external partners.

Thermo Orbitrap XL with FAIMS ion mobility

Waters Synapt G2 with travelling wave ion mobility

Ion mobility adds unique structural

insights to mass spectrometry Mass spectrometry can often quantify using sensitive measurements, and

identify using nominal/accurate mass and elemental composition.

Collision induced dissociation gives some evidence for structure through

product ions.

Ion mobility gives additional separating power on the same timescales as

mass spectrometry with little compromise.

Ion mobility gives rich insights into structure typically investigated by

other techniques including x-ray crystallography or NMR spectroscopy.

Ion mobility spectrometry-mass spectrometry (IMS-MS) of small molecules: Separating and assigning structures to ions

Mass Spectrom. Rev., vol. 32, no. 1, pp. 43, 2013. C. Lapthorn, F. Pullen, and B. Z. Chowdhry

One of the 10 ten most

accessed papers in

Mass Spectrometry Reviews

Overview

1. Evaluation of CCS prediction for small molecules

- initial evaluation, opportunities for improvement

2. MultiMOBCAL – a rapid framework for CCS prediction

- pathways to increased adoption of CCS prediction

3. Evidence from molecular modelling for charge location isomers

- implications for quantitation and utility of IMS

A comparison of theoretical and

experimental CCS for ‘small’ molecules

Geometry optimisation used Gaussian 09 with the

hybrid SCF-DFT B3LYP method and the

6-311++G(d,p) or 6-31G(d,p) basis sets.

Additional keywords pop=(mk,dipole) were used to

generate Merz-Singh-Kollman partial atomic charges

constrained to match the dipole moment.

Collision cross sections were calculated using

Waters Driftscope or Unifi and predicted using

MOBCAL with calculated charge distribution

via projection approximation and trajectory

methods.

How can IMS help in real world problems?

• Build libraries and compare with ‘real’ data

Severine Goscinny - Department of Food,

Medicines and Consumer Safety, Scientific

Institute of Public Health, Belgium,

Mike McCullagh - Waters

How can IMS help in real world problems?

Shimizu, A. & Chiba, M. Ion Mobility Spectrometry–Mass Spectrometry Analysis for the Site of Aromatic Hydroxylation.

Drug Metab. Dispos. 41, 1295–1299 (2013).

CH3CH3

O

NH

N

OH

OH

O-

O

F

OH

CH3

O

OO

OH

OH

M

P

7

O

6

8

Atorvastatin

Warfarin

Parent Compound Potential Metabolites TM-Based Calculated CCS (Å2)

Intact Metabolites N-Methyl Pyridine Derivatives

Atorvastatin O-hydroxy atorvastatin 182.29 200.26

M-hydroxy atorvastatin 188.36 206.61

P-hydroxy atorvastatin 187.73 214.55

Warfarin 6-Hydroxy warfarin 110.42 140.62

7-Hydroxy warfarin 110.23 142.60

8-Hydroxy warfarin 109.84 135.92

ASMS 2015

Investigation of Ion Mobility Mass Spectrometry Analysis of Electrochemically Generated Oxidation Products of Opiates and Comparison with

Theoretical CCS Values

Cris Lapthorn1; Frank Pullen1; Susana da Silva Torres2; Mark R. Taylor2; Russell Mortishire-Smith3; Jayne Kirk3; Andrew Baker4 1University of Greenwich, Chatham Maritime, UK; 2Pfizer, Sandwich, UK; 3Waters Corp, Manchester, UK; 4Waters, Inc., Pleasanton, CA

Predict separation of isomers in impurity analysis, degradation, metabolism, natural products

O P O Cl

Cl

OCH3

OCH3

Dichlorvos

NH

N

S

N

Thiabendazole

CH3

N

N

NH

CH3

Pyrimethanil

CH3 O

NH CH3

CH3

CH3

Ethoxyquin

CH3

CH3

CH3

ClN

N N

OH

Tebuconazole

Cl

N

N

N

S

OF

FF

NH2 Cl

F

F

F

FibrinoprilFlufenoxuron

F

O

NH

O

NH

F

O Cl

F

FF

F

Severine Goscinny - Department of Food,

Medicines and Consumer Safety, Scientific

Institute of Public Health, Belgium,

Mike McCullagh - Waters

How useful is molecular modelling with

ion mobility mass spectrometry? O3

2

OH1

4 5

7

8CH39

CH310

NH26

H5a

pregabalin

OH8

7 1

2 3

4

56

9NH10

CH311

CH312

H7a

H9a

ephedrine

O12

11

NH10

4

3 2

1O7

8CH3965

CH313

phenacetin

CH314

13

6

7

8

9

10

1

2

3

4

5

O11

CH312

15O17

OH16

H13a

naproxen13

12

2

3

15

14

4

5

6

7 8

9

1011

N1

16

17

18

NH19

CH320

desipramine

O18

3NH4

5 O19

NH

12

12

1314

15

1617

6

7

89

10

11

CH320

5-(p-methylphenyl)-5-phenylhydantoin

CH317

O16

12

11

10

15

14

13

O18

CH319

O20CH3

21

9

5

6

N1

2

N3

4

NH28

NH27

trimethoprim

N16

17 22

21

2019

18

13

1415

11

OH12

7

N1

65

4

8

3

9

CH210

2

H11a

H7a

H4a

H3a

cinchonine

O21

S20

O22

NH223

16 17

12

13

O18

CH319

14

15

10

O11

NH9

8

5

N1

6 CH37

2

34

sulpiride

O20

19

25OH26

1

OH18

14

15

34

1312

1110

9O23

8 7

6

CH322

5

17

O24

16

2

CH321

H3a

H4a

H5a

cortisone

NH220

S15

O16

O19 8

9

Cl17

10

5

NH

43

NH2

S1

O11

O12

6

7

13

Cl14

Cl18

trichlormethiazide

Cl27

17

18

19 20

21

16

4

5

6

NH1

2

3

12

O26

O13

14 CH315

7

O8

9

10

NH211

CH325

22

O28

O23CH3

24

amlodipine

O28

23

NH22

21

25

26

18

19

Cl2720

17

16

N13

12

11

N10

3

N2

S1

56

7

89

4

15

14

24

ziprasidone

O12

11

N13

14

15

N1619

N28

27

26

25

O32CH3

33

24

O30

CH331

23

22

21

N20

NH229

17

18

2

O1 6

5

O4

3

10

9

8

7doxazosin CH3

9

N4

3

2

N1

S7

O8 11

12

13

14

O17

18CH319

15 20 N28 27

23 N24

CH330

N25

26

3132

CH333

22NH21

O29

16

O10

6

5

sildenafil

CH311

9

O10

NH8

4

32

1

65

OH7

acetaminophen

CH39

8

NH7

1

23

4

56

n-ethylaniline

CH317

16CH318

NH1514

98

O7

1

65

4

3 2

11

12

CH213

OH10

alprenolol

CH322

N+

4

32

N1

65

7

13

14

15

16

17

12N11

10

9NH818

19

20

21

Cl24

O-

23

clozapine n-oxide

CH313

12

CH321

11

14

15

16

N17

CH318

22

23

24

29

28

27

26

25O30

CH331

O32

CH333

19

N20

5

4

3

2

1

6

O7

CH38

O9

CH310

verapamil

CH331 O

30

18

17

16

15

1419

NH13

12

11

10 9

N1

20

H20a21

4

H4a

3

H3a

2

8

7

H7a

6

H6a

5

H5a22

O32

O23

CH324

O25 CH3

26

O27

28

O29

33

34

3536

37

38O43

CH344

O41

CH342

O39

CH340

reserpine

CH320

2

N3

45

N1

6

7

8

9

10

1413

19 18

17

1612

N11

CH321

15

O22

ondansetron

CH32

N1

CH334

NH8

NH5

6

NH9

NH27

metformin

*

*

n=23

How does experimental CCS fit with

theoretical CCS?

Conclusions • A large ‘small’ molecule dataset comparing

experimental vs theoretical CCS using He(g) and

N2(g) MOBCAL has been evaluated

• There is a very good agreement between

experimental and theoretical CCS, typically within ~

2% CCS or 2 x experimental error

• Provides a key dataset for tuning, learning about

deviations from good agreement.

An automated pipeline for calculating CCS.

Why, how and when?

open source tool open source tool open source tool

collaboration w/ commercial package self written

Rapid CCS prediction for all

MultiMOBCAL

- Runs on multiple PCs, 1 instance is 3-7x faster

- Queues calculations so can run unattended

- Summarises important CCS from verbose MOBCAL OUT files

- Results can be synchronised through shared drives, Dropbox etc

- Predictions can run 10-100 times c.f. manual workflows

Online molecular modelling

Access to molecular modelling typically requires either

dedicated personnel or significant training and infrastructure

(high powered computing and software).

Future accessibility might be enabled by cloud-based

services e.g.

1. Schrodinger recently docked 1.8million compounds on a

600 core CycleCloud Condor cluster.

2. Accelerys Pipeline Pilot is now available on BT’s BT for

Life Sciences cloud computing platform.

3. Chorus now can store, share and visualise data online

using Amazon Web Services.

Can a molecular modelling service provide fee-for-

service molecular modelling?

01MAY12_CL2.raw:1

01MAY12_CL2.raw : 1

LC-MS ion chromatogram [MH]+ for Norfloxacin (m/z 320)

Ion mobilogram [MH]+ for Norfloxacin (m/z 320)

Charge location isomers in

fluoroquinolone antibiotics

Species 1 Species 2

Ion mobilogram [MH]+ for Norfloxacin (m/z 320)

• Fluoroquinolones are a class of antimicrobial

agents administered to livestock to

(a) prevention and control of infections, and

(b) growth promotion.

• Due to the resistant microorganisms in the

human population, the F.D.A. introduced a

ban on the use of enrofloxacin and

ciprofloxacin in livestock production in

September, 2005.

• The use of antibiotic growth promoting

agents (AGPs) in animal husbandry has been

forbidden in the European Union (EU) since

2006, when the final four antibiotics were

banned as growth promoters.

• EU Maximum Residue Levels (MRLs)

currently exist for eight (fluoro)-quinolone

compounds ranging from 10 to 1900 μg kg-1

dependant on the species and tissue type.

F

N N

NH

+

H

O

O

OH

CH3

singly protonated on piperazine

singly protonated on carboxyl

-H2O

F

N N

NH

O

O+

CH3

-CO2

F

N N

NH

+

H

O

CH3

m/z 302

m/z 320m/z 320

1211

109

8

F17

7

1615N

14

13

21

N1

2

3

NH4

5

6

O1819

O23

OH20

CH322

H

F

N N

NH

O

OH

O+

CH3

MSMS of

Species 2

MSMS of

Species 1

Protomer 1 Protomer 2

norflaxin 600 40

m/z200 205 210 215 220 225 230 235 240 245 250 255 260 265 270 275 280 285 290 295 300 305 310 315 320 325 330 335 340

%

0

100

m/z200 205 210 215 220 225 230 235 240 245 250 255 260 265 270 275 280 285 290 295 300 305 310 315 320 325 330 335 340

%

0

100

m/z200 205 210 215 220 225 230 235 240 245 250 255 260 265 270 275 280 285 290 295 300 305 310 315 320 325 330 335 340

%

0

100

01MAY12_CL2_peak3_RTandDTselected 66 (3.510) Cm (66:68) 1: TOF MSMS 0.00ES+ 8.76e4276.1530

233.1090

219.0966256.1427234.1103

320.1400

277.1555321.1489

01MAY12_CL2_peak2_RTandDTselected 57 (3.024) Cm (57:59) 1: TOF MSMS 0.00ES+ 3.27e4302.1311

320.1400

303.1339

321.1411

01MAY12_CL2_peak1_RTandDTselected 55 (2.916) Cm (53:55) 1: TOF MSMS 0.00ES+ 4.50e3276.1530

233.1090

256.1497

302.1311

277.1555 320.1400303.1339

Does molecular modelling predict IM-MS for protomers (I)?

Experimental

Qualitatively observe nr. baseline resolution of m/z 320 species in norfloxacin (R~1.5)

Experimental CCS (eCCS) for different species is ~11Å2 different

Structure tCCS eCCS

Theoretical

Theoretical CCS (tCCS) from projection approximation calculations predicts small

differences in CCS.

Can ion mobility mass spectrometry and density functional theory help elucidate protonation sites in ‘small’ molecules?,

Rapid Commun. Mass Spectrom., vol. 27, no. 21, pp. 2399, 2013. C. Lapthorn, T. J. Dines, B. Z. Chowdhry, G. L. Perkins, and F. S. Pullen

Does molecular modelling predict IM-MS for protomers (II)?

Theoretical

Theoretical CCS (tCCS) from trajectory method calculations correctly predicts

significant differences in CCS

Bordoli Prize

Conclusions

1. Projection approximation calculations

demonstrate for these protomers the IMS

separation does not appear to be based on

physical area presented to buffer gas.

2. Molecular modelling and trajectory

method calculations demonstrate for these

protomers that the charge distribution are

significantly different and the difference in

CCS is correctly predicted, but the absolute

CCSs require improvement

Synapt G2-Si Ionkey Exp.

CCS (Å2) in N2(g) Theo. CCS (Å2) in N2(g)

% Diff (Theo. vs Synapt

G2-Si Exp.)

Norfloxacin

N4 186.0

193.6 4.0%

O18 171.4

180.8 5.4%

Fluoroquinolones N2 @ 20mL/min 200/180V

Time-0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60 2.80 3.00 3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20 5.40 5.60 5.80 6.00 6.20 6.40 6.60 6.80 7.00 7.20 7.40 7.60 7.80

%

0

100

-0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60 2.80 3.00 3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20 5.40 5.60 5.80 6.00 6.20 6.40 6.60 6.80 7.00 7.20 7.40 7.60 7.80

%

0

100

-0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60 2.80 3.00 3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20 5.40 5.60 5.80 6.00 6.20 6.40 6.60 6.80 7.00 7.20 7.40 7.60 7.80

%

0

100

MP102414_008 2: TOF MS ES+ 360.125_360.2

4.22e5

4.20

MP102414_008 2: TOF MS ES+ 332.109_332.13

3.76e5

4.04

3.73

MP102414_008 2: TOF MS ES+ 319.984_320.177

4.76e5

4.00

3.69

Enrofloxacin

Ciprofloxacin

Norfloxacin

N2 – medium polarisability

better separation

He – low polarisability

no discernible separation

Linear Exp. CCS (Å2) in

N2(g) Synapt G2-Si Ionkey

Exp. CCS (Å2) N2(g)

% Diff (G2-Si Ionkey vs

Linear D.T.) Norfloxacin

N4 187.4 186.0 0.8% O18 172.3 171.4 0.5%

12 11 10

9

8

F17

7

16

15N14

13

21

N1

2

3

NH4

5

6

O18

19

O23

OH20

CH322

Norfloxacin• Increased separation of two

major components of

fluoroquinolone precursor m/z

in more polarisizable gases in

the order CO2>N2>He

• Results are consistent in

travelling wave and drift tube

ion mobility regimes

• Theoretical calculations are

consistent with experimental

findings. Charge distribution,

plays major role in

differences in CCS.

Conclusions

Charge location isomers have now been demonstrated

1) Using both positive and negative ionisation modes

2) For a range of (bio)chemical classes inc. simple acids, steroids,

fluoroquinolone antibiotics, pesticides and porphyrins

3) With complementary evidence from molecular modelling, product ion

spectra and action spectroscopy

The existence of charge location isomers has often been uniquely revealed using

ion mobility.

• Where are the rest?

• How can we utilise IMS to improve quantitative performance?

ASMS 2015

The importance of charge isomers in quantitation; ion mobility mass spectrometry of fluoroquinolone antibiotics

Cris Lapthorn1; Mike McCullagh2; Sara Stead2; Martin Palmer2; Kevin Giles2; Keith Richardson2; Jasper Boschmans3; Frank Sobott3; Frank Pullen1; Babur

Chowdhry1; George Perkins4 1University of Greenwich, Chatham Maritime, UK; 2Waters Corp, Manchester, UK;3University of Antwerp, Antwerp, Belgium; 4149 Hickory Corner Road,

Milford, NJ

Acknowledgements Prof Frank Pullen – University of Greenwich

Dr Mike McCullagh - Waters

George Perkins – Sanofi Pasteur, USA

Prof Babur Chowdhry – University of Greenwich

Patricia Wright - University of Greenwich

Prof Trevor Dines – University of Dundee

Dr Jiayun Pang – University of Greenwich

Yanira Ruhe - University of Greenwich

Dr Alex Muck – Waters

Dr Jonathan Williams – Waters

Dr Keith Richardson – Waters

Dr Jeff Brown – Waters

Prof Bela Paizs – University of Bangor

Prof Perdita Barran – University of Manchester

Severine Goscinny - ISP-WIV, Belgium

Scott Rudland – Waters

Alex Hunt – Waters

Dr Grigoriy A. Andrienko – Chemcraft