V. Chagovets et al., Eur. J. Mass Spectrom. 22, 123–126 (2016) Received: 30 June 2016 ■ Accepted: 3 August 2016 ■ Publication: 19 August 2016 123 EUROPEAN JOURNAL OF MASS SPECTROMETRY Virtual Issue: Papers Presented at the 12th EFTMS Workshop, Matera, Italy Peculiarities of data interpretation upon direct tissue analysis by Fourier transform ion cyclotron resonance mass spectrometry Vitaliy Chagovets,a Aleksey Kononikhin,a,b Nataliia Starodubtseva,a,b Yury Kostyukevich,b,c,d Igor Popov,b,c Vladimir Frankevicha* and Eugene Nikolaevb,c,d a Department of System Biology in Reproduction, Federal State Budget Institution ‘Research Center for Obstetrics, Gynecology and Perinatology’, 4 Oparin Street, Moscow 117997, Russian Federation. E-mail: vfrankevich@gmail.com b Moscow Institute of Physics and Technology, 141700 Dolgoprudnyi, Moscow Region, Russian Federation c Institute for Energy Problems of Chemical Physics, Russian Academy of Sciences, Leninskii pr., 38, bld. 2 Moscow, 119334, Russian Federation d Skolkovo Institute of Science and Technology, 100 Novaya Street, Skolkovo 143025 Russian Federation The importance of high-resolution mass spectrometry for the correct data interpretation of a direct tissue analysis is demonstrated with an example of its clinical application for an endometriosis study. Multivariate analysis of the data discovers lipid species differentially expressed in different tissues under investigation. High-resolution mass spectrometry allows unambiguous separation of peaks with close masses that correspond to proton and sodium adducts of phosphatidylcholines and to phosphatidylcholines differing in double bond number. Keywords: direct tissue analysis, ambient mass spectrometry, Fourier transform ion cyclotron resonance, endometriosis, mass spectrometry clinical application Introduction Investigation of tissue composition is an important stage in biomarker discovery and high-throughput methods are required to obtain the molecular signatures.1 Ambient ionization mass spectrometry is a fast growing branch that allows molecular information to be obtained from tissue samples with minimal sample pretreatment and a set of methods has been proposed.2–7 One of the drawbacks of the ambient analysis is the complexity of the obtained spectra and the lack of additional degrees of separation. This problem can be addressed by ion mobility methods. Another way to overcome biological sample complexity and not to lose important information is to use high-resolution with Fourier transform ion cyclotron resonance (FT-ICR) or Orbitrap mass spectrometers. ISSN: 1469-0667 doi: 10.1255/ejms.1425 The present paper demonstrates profits from the combination of a direct tissue spray ion source with the FT-ICR mass spectrometer with the example of an endometriosis study. Experimental Methanol (HPLC grade) and formic acid were purchased from Sigma-Aldich. Tissue samples were collected from 30 patients with ovarian cysts under laparoscopic surgery in the Operative Gynecology department at the V.I. Kulakov Research Center for Obstetrics, Gynecology and Perinatology. All the patients included in © IM Publications LLP 2016 All rights reserved 124 Peculiarities of Data Interpretation upon Direct Tissue Analysis by FT-ICR-MS Figure 1. Positive-ion direct tissue mass spectra of (a) ovarian cyst endometriosis and (b) the eutopic endometrium. The numbers over the peaks correspond to items in Table 1. Annotated peaks are those giving the highest impact to tissue classification.  – lipid species,  – oxidized lipid species. the study provided written informed consent. All the procedures and study methods were approved by the Commission of Biomedical Ethics at V.I. Kulakov Research Center for Obstetrics, Gynecology and Perinatology. The experimental setup is described elsewhere.7 In brief, a slice of tissue was placed on the tip of medical needle in the front of a LTQ FT Ultra instrument (Thermo Scientific, Bremen, Germany), which includes a FT-ICR analyzer. Methanol with 0.1% formic acid was supplied to the tissue by a fused silica capillary. This solution provides the extraction of compounds from the tissue and is then sprayed on the high voltage application. Acquired mass spectrometric data were analyzed by a partial least squares discriminant analysis (PLS-DA) method realized with the ropls package.8 The lipid nomenclature used throughout the paper is in accordance with LIPID MAPS9 terminology and the shorthand notation is summarized in Liebisch et al.10 Results and discussion All the mass spectra of the tissue samples were acquired in positive ion mode. The most abundant peaks were in the range between m/z 700 and m/z 900. Figure 1 shows the mass spectra characteristics for ovarian cyst endometriosis and a eutopic endometrium of the same patient. PLS-DA multivariate analysis of the mass spectrometric data reveals that molecular information obtained from the direct tissue mass spectra is sufficient for tissues differentiation. The created model for the ovarian vs endometrium describes 80% of the data using the latent variables (R2) and 66% are predicted by the model according to the cross validation, the values showing an accuracy which can be expected to predict new data (Q2). Variable importance in the projection (VIP) values are obtained from PLS-DA models and used to determine compounds with the highest impact to the latent variables (Table 1). According to the accurate mass, most of the selected species are phosphatidylcholines (PC). PCs are registered in mass spectrometry as protonated molecules and adducts with alkali metal ions such as sodium and potassium. Even trace amounts of alkali metal ions can cause comparable intensities of sodiated and protonated molecules. An interference between peaks of sodiated and protonated PCs is observed for PCs with two aliphatic chains with a general formulae PC m:n and PC (m + 2):(n + 3), where m is the total carbon number of fatty acyls and n is the double-bond number. Such an overlapping of peaks hampers the quantitative and V. Chagovets et al., Eur. J. Mass Spectrom. 22, 123–126 (2016) 125 Table 1. List of masses with the highest influence on tissue differentiation according to PLS-DA analysis. # Accurate mass Theoretical mass Mass accuracy (ppm) [Lipid + Na]+ Elemental composition 1 798.5441 798.5619 22 Ox PC 34:1 C 42H82NO9P 2 832.5843 832.5827 2 PC 38:4 C 46H84NO8P 3 796.5257 796.5463 26 Ox PC 34:2 C 42H80NO9P 4 782.5702 782.5670 4 PC 34:1 C 42H82NO8P 5 725.5600 725.5568 4 SM 34:1 C 39H79N2O6P 6 822.5428 822.5619 23 Ox PC 36:3 C 44H82NO9P 7 824.5575 824.5776 24 Ox PC 36:2 C 44H84NO9P 8 806.5679 806.5670 1 PC 36:3 C 44H82NO8P 9 780.5540 780.5514 3 PC 34:2 C 42H80NO8P 10 820.5287 820.5463 21 Ox PC 36:4 C 44H80NO9P 11 808.5850 808.5827 3 PC 36:2 C 44H84NO8P 12 834.5971 834.5983 1 PC 38:3 C 46H86NO8P 13 848.5570 848.5776 24 Ox PC 38:4 C 46H84NO9P 14 772.5276 772.5463 24 Ox PC 32:0 C 40H80NO9P ppm, parts per million. qualitative estimation of the lipid composition. The fastest and most reliable method of interference peaks deconvolution is to resolve the peaks at the instrumentation level. The difference between sodiated and protonated molecules, e.g. for [PC 36:2 + Na]+ and [PC 38:5 + H]+, is 808.5851 – 808.5827 = 0.0024, which requires a resolution of over 3.4 × 105 to separate them. FT-ICR manifests two such cases, shown in Figure 2(a) and (b). Resolution of the presented peaks is up to 8 × 105, which allows the unambiguous identification of the fourth and 11th items in Table 1 as sodium adducts of PC 34:1 and PC 36:2. Another issue for lipid identification is the interference of the isotopic peaks of compounds with a difference of one double bond. In this case, a resolution of about 105 is sufficient. [PC 38:3 + Na]+ has m/z 834.5983, the third isotope of [PC 38:4 + Na]+ has a mass 834.5891 and their difference is 834.5983 – 834.5891 = 0.0092. Resolution of these peaks is shown in Figure 2(c), which gives the 12th item in Table 1 as [PC 38:3 + Na]+. Of note are the groups of peaks marked with the empty circles which also differentiate endometrioid tissues (Figure 1). Their patterns are similar to those of closed circles: 3,1 is similar to 9,4; 6,7 to 8,11; and 13,12. The mass difference between these groups is 16 Da. One can speculate that the open-circle peaks correspond to oxidized products of the respective lipids. Such a possibility has been observed previously with a different ambient ionization method.11 However, this version needs further elaboration because the mass accuracy for these peaks, on the assumption of oxygen attachment to the lipids, is low (Table 1). Conclusion An FT-ICR mass analyzer is essential for direct tissue analysis in order to identify lipid constituents correctly and to avoid some problems connected with features in the mass spectrometric Figure 2. Regions of a high resolution mass spectrum comprising peaks with near-lying m/z. 126 Peculiarities of Data Interpretation upon Direct Tissue Analysis by FT-ICR-MS investigation of lipids. The interference of protonated and sodiated lipid species and of lipids with different double-bond numbers are among these features. Acknowledgments This work was supported by Russian Science Foundation Grant No. 16-14-00029, and N.S. acknowledges MERF Grant No. MK-8484.2016.7 for partial support for the sample collection. References 1. J.E. McDermott, J. Wang, H. Mitchell, B.J. Webb- Robertson, R. Hafen, J. Ramey and K.D. Rodland, “Challenges in biomarker discovery: combining expert insights with statistical analysis of complex omics data”, Exp. Opin. Med. Diagn. 7, 37 (2013). doi: http://dx.doi.org/1 0.1517/17530059.2012.718329 2. D.R. Ifa and L.S. Eberlin, “Ambient Ionization mass spectrometry for cancer diagnosis and surgical margin evaluation”, Clin. Chem. 62, 111 (2016). doi: http://dx.doi. org/10.1373/clinchem.2014.237172 3. J. Laskin and I. Lanekoff, “Ambient mass spectrometry imaging using direct liquid extraction techniques”, Anal. Chem. 88, 52 (2016). doi: http://dx.doi.org/10.1021/acs. analchem.5b04188 4. J. Liu, R.G. Cooks and Z. Ouyang, “Biological tissue diagnostics using needle biopsy and spray ionization mass spectrometry”, Anal. Chem. 83, 9221 (2011). doi: http://dx.doi.org/10.1021/ac202626f | 5. B. Hu, Y.H. Lai, P.K. So, H. Chen and Z.P. Yao, “Direct ionization of biological tissue for mass spectrometric analysis”, Analyst 137, 3613 (2012). doi: http://dx.doi. org/10.1039/c2an16223g 6. K.S. Kerian, A.K. Jarmusch and R.G. Cooks, “Touch spray mass spectrometry for in situ analysis of complex samples”, Analyst 139, 2714 (2014). doi: http://dx.doi. org/10.1039/c4an00548a 7. A. Kononikhin, E. Zhvansky, V. Shurkhay, I. Popov, D. Bormotov, Y. Kostyukevich, S. Karchugina, M. Indeykina, A. Bugrova, N. Starodubtseva, A. Potapov and E. Nikolaev, “A novel direct spray-from-tissue ionization method for mass spectrometric analysis of human brain tumors”, Anal. Bioanal. Chem. 407, 7797 (2015). doi: http://dx.doi.org/10.1007/s00216-015-8947-0 8. S. Wold, M. Sjöströma and L. Eriksson, “PLS-regression: a basic tool of chemometrics”, Chemometr. Intell. Lab. Sys. 58, 109 (2001). doi: http://dx.doi.org/10.1016/S01697439(01)00155-1 9. LipidMAPS, Lipid Metabolites and Pathways Strategy. http://www.lipidmaps.org, 10. G. Liebisch, J.A. Vizcaino, H. Kofeler, M. Trotzmuller, W.J. Griffiths, G. Schmitz, F. Spener and M.J.O. Wakelam, “Shorthand notation for lipid structures derived from mass spectrometry”, J. Lipid Res. 54, 1523 (2013). doi: http://dx.doi.org/10.1194/Jlr.M033506 11. S.P. Pasilis, V. Kertesz and G.J. Van Berkel, “Unexpected analyte oxidation during desorption electrospray ionization-mass spectrometry”, Anal. Chem. 80, 1208 (2008). doi: http://dx.doi.org/10.1021/ac701791w