University of Tartu, Estonia Janssen Pharmaceutica NV, Belgium

Transcription

1 Semiquantitative ESI/MS analysis of metabolites in biological matrices made feasible via prediction of ionization efficiencies Piia Liigand, Jaanus Liigand, Filip Cuyckens, Rob J. Vreeken, Anneli Kruve University of Tartu, Estonia Janssen Pharmaceutica NV, Belgium

2 Methoxychlor HPTE LC/MS Stable isotope labelled internal standards

3 Standard substance free quantitation Aldicarb 4-methoxypyridine ESI+ Aldicarb sulfoxide m/z 4-nitrophenol 2-methoxypyridine m/z m/z Salicylic acid ESI- 2-nitrophenol Benzoic acid m/z m/z 180 m/z

4 Signal Signal Ionization efficiencies C 1 = C 2 Signal m/z Signal 1 < Signal 2 8.E+07 6.E+07 4.E+07 C 1 < C 2 Ionization efficiency 2.E+07 0.E+00 4.E+07 0.E+00 5.E-06 1.E-05 2.E-05 2.E-05 3.E-05 Leito et al Rapid. Commun. Mass Spectrom 22 (2008) E+07 2.E+07 1.E+07 0.E+00 0.E+00 2.E-05 4.E-05 6.E-05 8.E-05 c (M) log IE = log slope 1 slope 2 c (M)

5 Predicted logie Modelling COSMOtherm 5 MS measurements 3 1 Physicochemical parameters log IE predicted = f x 1 x n -3 log IE measured = log slope 1 slope 2-5 Measured logie

6 Oss et al Anal. Chem. 82 (2010) ; Liigand et al Anal. Chem. 89 (2017) Predicted logie ESI /80 0.1% formic acid/acetonitrile 62 compounds: logp = -22 6; pk a (aq) = Mismatch 7 times log IE = α WANS Measured logie R 2 = 0.63 s = 0.87

7 Predicted logie ESI- 20/80 0.1% ammonia/acetonitrile logp = 1 18; pk a (aq) = Model development R 2 = 0.83 mismatch 2.5 times Validation mismatch 3 times log IE = α 0.51 WAPS -4 Measured logie benzoic acid phenol salicylic acid validation set Kruve et al Anal. Chem. 86 (2014)

8 c predicted (M) log IE (Sciex) logie Transferability between instruments 11 setups: 4 vendors, 3 ion sources, 5 mass-analyzers Order of IEs independent of the instrument R 2 = Anchoring is recommended (calibration compounds) Concentration mismatch lower than 2.5 times 1.0E Setup A Setup B Setup C Setup D 4 1.0E E E E E-05 c actual (M) Liigand et al J. Am. Soc. Mass Spectrom 26 (2015) y = 0.80x R² = log IE (Thermo)

9 c predicted (µm) logie predicted Transferability between solvents ESI- 10 solvent compositions 62 compounds Flow injection (0.2 ml/min) logie measured Validation with gradient LC/ESI/MS Mismatch 3.7 times (44 times without model) See poster WP 298 on Wednesday Kruve et al Anal. Chem. 89 (2017) c actual (µm)

10 Could this also work in biological matrices? O HO HO Benzoic acid Salicylic acid 3-CF 3 SO 2 -benzoic acid plasma blood liver tissue AUTOSAMPLER CSF Warfarin Naproxen Sorbic acid Dodecanoic acid Lincomycin Taurocholic acid Fumaric acid brain tissue urine 20/80 0.1% ammonia/acetonitrile Protein precipitation MS 1 ESI- Flow injection Liigand et al Anal. Chim. Acta (available online:

11 logie I. Measuring logie in matrices Warfarin Taurocholic acid 3-CF 3 SO 2 -benzoic acid Salicylic acid Dodecanoic acid Benzoic acid Naproxen Lincomycin Sorbic acid Fumaric acid Urine Plasma Blood CSF Liver Brain Solvent Liigand et al Anal. Chim. Acta (available online:

12 logie I. Measuring logie in matrices Warfarin Taurocholic acid 3-CF 3 SO 2 -benzoic acid Salicylic acid Dodecanoic acid Benzoic acid Naproxen Lincomycin Sorbic acid Fumaric acid Urine Plasma Blood CSF Liver Brain Solvent Liigand et al Anal. Chim. Acta (available online:

13 logie I. Measuring logie in matrices 2 Warfarin 90% signal suppressed 0-4 Warfarin Taurocholic acid 3-CF 3 SO 2 -benzoic acid Salicylic acid Urine Solvent Dodecanoic acid Benzoic acid Naproxen Lincomycin Sorbic acid Fumaric acid Fumaric acid -2 99% signal suppressed Blood Solvent Urine Plasma Blood CSF Liver Brain Solvent Liigand et al Anal. Chim. Acta (available online:

14 II. Correlation with neat solvent Urine Plasma Blood logie urine logie plasma y = 1.80x y = 2.15x 1.74 y = 2.10x R² = R² = R² = 0.71 logie solv logie solv logie solv logie blood CSF Liver Brain logie CSF logie liver y = 1.42x R² = logie solv logie solv y = 1.40x 1.01 R² = logie solv Liigand et al Anal. Chim. Acta (available online: logie brain y = 1.37x 0.80 R² = 0.87

15 III. Model development COSMOtherm log IE = a WAPS + b α + c WAPS pk a α s LOO mismatch... Leave-one-out cross validation MLR Liigand et al Anal. Chim. Acta (available online:

16 Predicted logie III. Model to predict logie log IE = coef WAPS WAPS + coef α α + intercept 2 coef WAPS coef α intercept R 2 mismatch Solvent Urine Plasma Blood Liver Brain CSF Liigand et al Anal. Chim. Acta (available online: Measured logie SOLVENT PLASMA URINE BLOOD LIVER BRAIN CSF -4 R 2 = times mismatch

17 Standard substance free quantitation log IE blood = 0.50 WAPS α Ionization logie (benzoic acid) = Signal (benzoic acid) = logie (salicylic acid) = Signal (salicylic acid) = efficiency c(salicylic acid) c(benzoic acid) = signal (salicylic acid) signal(benzoic acid) IE(salicylic acid) IE(benzoic acid) Salicylic acid c(salicylic acid) c(benzoic acid) = 8x mismatch Benzoic acid m/z

18 Proof-of-concept Approach works! ESI+ and ESI-, different setups, different mobile phases, biological matrices C of calibration compounds 2D structure of compound of interest MS 1 signal Poster WP 298 Wednesday Technical aspects Software (automation) LC, adducts More matrices (environmental, food, botanics) Self-improving How can you use it? IE prediction model Concentration estimate

19 Conclusion Predicting ESI ionization efficiencies Different ESI modes Different mobile phase compositions Transferable between different instruments Biological matrices Semiquantitative analysis prediction mismatch 8 times

20 Poster WP 298 on Wednesday Thank you! Kruvelab.com