* Nestec Ltd., Nestle Research Centre, Vers-chez-les-Blanc, P. 0. Box 44, CH-I 000

Transcription

1 TANDEM MASS SPECTROMETRY IN FLAVOUR RESEARCH L.B. Fay,* I. Blank* and C. Cernyt * Nestec Ltd., Nestle Research Centre, Vers-chez-les-Blanc, P. 0. Box 44, CH-I 000 Lausanne 26, Switzerland tnestec Ltd., Nestle R&D Center Kemptthal, CH Kemptthal, Switzerland 1 INTRODUCTION During the last twenty years tandem mass spectrometry has become a major tool for analysis of complex mixtures 2 and is widely employed for forensic chemi~try,~ and environmental4 and bi~logical,~ applications. However, despite its great analytical potential, MS-MS is less frequently used for flavour research, and mass spectrometry coupled to gas chromatography remains the more popular spectroscopic technique for aroma analysis. This can be explained by the longer commercial availability of GC-MS instruments as opposed to MS-MS instruments, by the success of the Gc-MS method itself and more recently of the HPLC- MS c~mbination,~ and by the high cost of the MS-MS apparatus. Nevertheless, because of its sensitivity and selectivity, MS-MS offers features allowing structural identification as well as quantitative determination. In this paper, examples of the use of MS-MS for flavour applications obtained with a quadrupole tandem mass spectrometer will be presented. These examples will highlight the use of MS-MS used alone and in conjunction with GC or HPLC, for identification and quantification of flavour compounds. 2 EXPERIMENTAL All experiments were carried out using a Finnigan TSQ-700 mass spectrometer (Finnigan MAT, Bremen, Germany) working with argon as the collision gas and set to between 1 and 2 mtorr. A collision energy from 5 to 30 ev in the laboratory frame was used. All GC separations were done using an HP-5890 gas chromatograph equipped with a Hp-7673 autosampler. The column was a J&W Sci DB-Wax capillary column 30 m x 0.32 mm internal diameter with a 0.25 pm film thickness. Helium was used as carrier gas at a pressure of 10 psi. Furanone samples were injected in a splitless injector (280 C) whereas pyrazine samples were injected with a cold on-column injector. The ionisation was either electron impact at 70 ev (fhranones) or positive chemical ionisation with ammonia as reagent gas (pyrazines). HPLC separations were made using a Waters 600-MS pump and a Waters 717 autosampler. The column was an HP Lichrospher 100 Rp-18 (125 x 4 mm, 5 pm) operated at room temperature at a flow rate of 1 mi mid. The mobile phase was as follows: solvent A: 50 mm ammonium acetate, ph 5.5; solvent B: methanol. Flow was initially 100% A for

2 272 Flavour Science: Recent Developments 10 min, then going to 100% B over 15 min, and returning to 100% A over 5 min. The HPLC was connected to the MS via a thermospray interface. Fast atom bombardment ionisation was generated with an Ion Tech saddle-field atom gun operated at 10 kv and 0.2 ma using xenon as the gas. All samples were dissolved in thioglycerol used as a matrix. 3 RESULTS AND DISCUSSION 3.1 Application of MS-MS for Qualitative Experiments The use of capillary GC columns with high resolving power allows the analysis of complex flavour profiles. However, it is sometimes still very difficult to get enough resolution to resolve all the constituents of a complex flavour mixture. In such cases, contaminated GC peaks are obtained and the corresponding mass spectra are difficult to interpret. Tandem mass spectrometry offers a third dimension to GC-MS by adding the electronically-based selectivity of the collision induced dissociation (CID) process to the chemically-based selectivity of the chromatographic separation. As an example, Figure 1 A shows the contaminated mass spectrum of a compound tentatively identified as 4-hydroxy- 2,5-dimethyl-3(2H)-firanone. This contaminated spectrum obtained after GC-MS does not allow an unambiguous characterisation of the molecule. On the other hand, the daughter mass spectra of the same compound (Figure 1B) obtained in the same GC conditions, is free of interference At the present time MS-MS spectral libraries are not commercially available. In principle standard CID MS-MS spectra could be collected and stored in reference libraries. Nevertheless, there are several instrument parameters that can cause significant differences in the observed MS-MS spectra for a given molecule. However, a protocol to overcome this problem has recently been prop~sed.~ The Maillard reaction that occurs between sugars and amino acids on heating generates very complex mixtures. GC-MS-MS was employed to identifi the caramel-like smelling compounds rl-hydroxy-2,5-dirnethyl-3(2h)-furanone (HDMF) and 4-hydroxy-2(or 5)-ethyl- 5(or 2)-methyl-3(2H)-furanone (HEW) in Maillard reaction systems based on pentoses." Figure 2 shows the mass spectra of C-labelled HEMF and HDW obtained in the reaction systems D-xylo~e-[3-*~C]-L-alanine and D-xylo~e-[2-'~C]-gIycine respectively. These spectra were obtained by GC-MS-MS and CID of the parent molecular ions at m/z 143 and 129 respectively. In the case of the '3C-HEMF, the daughter ions at m/z 127 [M- CH3]' and at m/z 114 [M-CH2-CH3]' revealed the presence of the C atom in the ethyl group. The incorporation of the C atom in the molecule was also indicated by the fragments at m/z 58, 72, 87 and 100, which correspond to the ions 57, 71, 86 and 99, respectively in the mass spectrum of the unlabelled HEMF. As indicated by the fragment [M-CO-CH3]' at m/z 100 for the labelled HEW and at m/z 99 for the unlabelled HEMF, as well as by the fragment [CO-CH3]' at m/z 43 for both compounds, the methyl group of the labelled HEMF does not bear the C atom. The I3C atom is more difficult to locate in the spectra of the mass spectrum of '3C-HDMF. Indeed, this molecule may occur as a 1:l mixture of two isotopomers and the ion pairs 43/44, 57/58 and 85/86 of nearly equal intensity correspond to its symmetric structure.

3 Novel Methods of Flavour Analysis I I 155 I mass t charge mass t charge Figure 1 Mass spectra of I-hydroxy-2,5-dimethyl-3(2H)-jkranone identified in a Maillard reaction Jystem based on pentose and glycine. A: mass spectrum obtained by GC-MS after electron impact ionisation at 70 ev; B: product ions after collision induced dissociation (10 ev) of the molecular ion at d z 128 formed by electron impact ionisation 3.3 Application of MS-MS for Quantitative Measurements Tandem mass spectrometry has been shown to be useful for quantification of compounds in complex biological matrices because of its selectivity and sensitivity. l1 The two examples that follow highlight the use of this technique for flavour applications. These are the quantification by fast atom bombardment MS-MS (FAB-MS-MS) of the Amadori compound N-( 1 -deoxy-d-fructos-1-y1)-glycine (DFG) and the determination of 8- oxocaffeine in coffee by HPLC-MS-MS.

4 IW a-l H i I00 1 Daughters of ev A 58 I mass f charge 100; 0,,a 73 = o I Daughters of I29 10 ev - B I11 I. mass 1 charge Figure 2 GC-MS-MS spectra of one tautomer 2-( 2- C]ethy~-4-hydroxy-5-methyl- 3(2H)-furanone (A) and of 4-hydro1y-2-[ ~C]methyl-5-methyl-3(2H)--ranone (B) obtained in the reaction systems D-xyIo~e/[3- ~C]-L-alanine and D- xylose/(2- C]-g[ycine respectively (the labelled positions are marked with m) To study the decomposition of DFG in model systems we developed a quantification method based on isotope dilution fast atom bombardment MS-MS after synthesis of the C-labelled DFG used as internal standard. Isotope dilution with FAB-MS is known to show a high dynamic range and therefore to be applicable for quantitative purposes.14 From DFG, FAB-MS produces the commonly observed protonated molecular ion at m/z 238 with little fragmentation. The CID spectrum of the protonated molecular ions leads to spectra showing intense daughter ions at m/z 220 ([M+H-H20] ) and at m/z 202 ([M+H- 2H20] ), and an additional fragment corresponding to the glycine residue at m/z 76. The daughter ion at m/z 220 was selectively recorded (selective reaction monitoring) to quantify DFG. The low mass fragment ion at m/z 76 was of too low an intensity to be used for quantification. Similar daughter ions at m/z 221 and 77 were obtained from C-labelled DFG used as an internal standard. A second order calibration curve was used to quantify DFG in model reactions and the repeatability of the method was found to be below 2%. It should be pointed out that this technique requires neither a purification nor a chromatographic step and is, therefore, very fast and straightforward. However, high salt

5 Novel Methods of Flavour Analysis 275 concentrations or strong buffer solutions should be avoided. Buffers can lead to a complete suppression of all the signals. LCIMSIMS mlz 211 -> 196 I 3:20 6:40 1O:OO :20 16:40 20:OO 23:20 Retention time (min) Figure 3 HPLC-MS (upper trace) and HPLC-A4S-MS (lower trace) detection of& oxocaffne in a solid-phase treated instant cosfee. The protonated molecular ion at mlz 211 was monitored during the HPLC-MS experiment whereas the daughter ion at d z I96 obtained by CID (I 7 ev) from the protonated molecular ion at mlz 21 I was acquired during the HPLC-MS-MS detection The second example of quantification by MS-MS is the determination of 8-oxocaEeine (1,3,7-trimethyluric acid) in coffee subjected to oxidative stress. It was demonstrated that caffeine in coffee is an efficient scavenger of the deleterious hydroxyl radical in situ, forming the major reaction product 8-oxocaffeine. l5 This compound was monitored by HPLC-MS- MS in the coffee extracts obtained by solid phase extraction on CIS cartridges. 8- Oxocaffeine was ionised using a thermospray interface and the protonated molecular ion at m/z 21 1 was submitted to CID, producing fragment ions at m/z 196, 153, 9, 11 1, 83 and 42. The daughter ion at m/z 196 was selectively recorded and the resulting chromatogram is presented in Figure 3 in comparison with the chromatogram of the [M+H]' ion obtained by HPLC-MS only. These data clearly demonstrate the gain in selectivity obtained by HPLC- MS-MS in comparison with HPLC-MS. To quanti@ 8-oxocaffeine, the coffee solutions were spiked with different known amounts of a standard compound and the amount in the non-spiked sample extrapolated from a linear regression equation. For all samples, there was a good linear relationship between oxo-caffeine concentration in the coffee and peak area, giving a correlation coefficient (r) of better than This technique was applied successhlly to instant, roasted and ground and green coffees.

6 276 Flavour Science: Recent Developments 4 CONCLUSION As for many other analytical techniques, the growth of tandem mass spectrometry is linked to instrumental development. The commercial availability of triple quadrupole instruments and, more recently, of quadrupole ion trap mass spectrometers at affordable prices has allowed this tool to enter flavour laboratories. Moreover, improvement in computer control and instrument design has offered users easy-to-operate machines on a routine basis. Flavour analysis deals with complex matrices in which traces of key components are sometimes expected. Therefore, because of its features of structural identification as well as quantitative determinations, we should see an expansion in the use of tandem mass spectrometry (used or not used in conjunction with chromatographic separations) in our research area. ACKNOWLEDGEMENTS The authors are gratehl to R. Fumeaux and S. Marti for expert technical assistance and to Dr. I. Horman for helphl discussions. REFERENCES K.L. Busch, G.L. Glish and S.A. McLuckey, Mass Spectrometry / Mass Spectrometry: Techniques and Applications of Tandem Mass Spectrometry, VCH Publishers Inc., Weinheim, Germany, 1988, p J.V. Johnson and R.A. Yost, Anal. Chem., 1985,57,758A. J. Yinon,MassSpectrom. Rev., 1991,10, 179. D.F. Hunt, J. Shabanowitz, T.M. Harvey and M.L. Coates, Anal. Chem., 1985, 57, 525. D. Favretto and P. Traldi, Mass Spectrom. Rev., 1993, 12, 3. I.A. Blair, Chem. Res. Toxicol., 1993, 6, 741. C. Salles, J.C. Jallageas, F. Fornier, J.C. Tabet and J.C. Crouzet, J. Agric. Food Chem., 1991,39, K.L. Busch and K. Kroha, Characterization and Measurement of Flavor Compounds, A.C.S. Symposium Series 289, Washington, D.C., 1985, p K.R. Mohan, M.G. Barlett, K.L. Busch, A.E. Schoen and N. Gore,.I Am. SOC. Mass Spectrom., 1994, 5, 576. I. Blank and L.B. Fay, J. Agric. Food Chem., 1996,44, 531. G.C. Thorne and S.J. Gaskell, Biomed. MassSpectrom., 1985, 12, 19. A.A. Staempfli, I. Blank, R. Fumeaux and L.B. Fay, Biological Mass Spectrom., 1994, 23,642. C.F. Beckner and R.M. Caprioli, Biomed. Mass Spectrom., 1984, 11, 60. A.A. Staempfli, 0. Ballevre and L.B. Fay, Rapid Comm. Mass Spectrom., 1992, 6, 547. R.H. Stadler and L.B. Fay, J Agric. FoodChem., 1995,43, 32.