Comparison of Volatile Flavour Profiles of Kidney Beans and Soybeans by GC-MS and PTR-MS

Transcription

1 Food Sci. Technol. Res., ++ (+), 0-1*,,**/ Comparison of Volatile Flavour Profiles of Kidney Beans and Soybeans by GC-MS and PTR-MS Saskia M. van RUTH +, Lonneke DINGS +, Eugenio APREA + and Sachiko ODAKE, + Department of Food and Nutritional Sciences, University College Cork, Western Road, Cork, Ireland, Nippon Veterinary and Animal Science University, Food Science and Technology,,,1 / Sakai, Musashino, Tokyo +2* **,,, Japan Received August,-,,**.; Accepted January ++,,**/ The volatile compounds of kidney beans and soybeans were analysed by gas chromatography mass spectrometry (GC-MS) and proton transfer reaction mass spectrometry (PTR-MS). Fingerprint PTR-MS spectra of the kidney beans and soybeans showed similarities but also quantitative di#erences, with soybeans generally presenting higher intensities. Fifty-one compounds were identified by gas chromatography mass spectrometry, -3 in the kidney beans and,3 in the soybeans. Dynamic modelmouth/ptr-ms analysis revealed four masses predominant in both types of beans, m/z --,./, /3 and 1-, which likely originate from methanol, ethanol,,-propanone, and,-butanone, respectively. The two beans di#ered significantly in the headspace concentrations of the masses m/z -- and m/z /3 at various time points, with both higher in the soybeans. The four masses were released at fairly linear rates. The correlation coe$cients of the concentration vs. time profile were in the range *.21. *.33, under mouth conditions. Principal component analysis revealed that high release rates of the masses m/z --, /3 and 1- correlated with the soybeans, whereas mass m/z./ correlated well with the kidney beans. As a result of the quantitative di#erences, the proportions of the four masses di#ered between the two beans. For both types of beans their proportions also changed during the course of release. Keywords : Aroma, flavour, kidney beans, model mouth, PTR-MS, soybeans Introduction Vegetables and legumes form an important part of the human diet. Canned vegetables make up a significant fraction of the total vegetable consumption. For instance, in the UK, at least,/ of vegetables is consumed in a canned form (Competition Commission, +33,). The qauality determining factors are time of planting, seed size, protein quality and digestibility (Forney et al., +33*; Wu et al., +33.; Wuet al., +33/). The texture of kidney beans and the flavour of soybeans are the main sensory characteristics that received attention in scientific studies on these particular type of beans. The important textural attributes of kidney beans are wholeness/toughness of the skin, astringency/flouriness and skin surface (Sanz Calvo, +333). The characteristic beany flavour of soybean products has been attributed to formation of volatile compounds by lipoxygenase activity (Lee Kwang et al.,,***). Apart from their sensory impact, flavour extracts of kidney beans and soybeans have shown antioxidative properties (Takamura et al., +33+). Only a few studies have focused on the volatile flavour of soybeans or fermented soybeans (Lee and Shibamoto,,***; Leejeerajumnean et al.,,**+; Boatright,,**,). To the authors knowledge, we are the only group to have conducted volatile flavour analysis of kidney beans (van Ruth et al.,,**.). The major components of beans that s.vanruth@ucc.ie would potentially a#ect its flavour are lipids, carbohydrates, proteins and phenolic acids (Drumm, +323). The lipid content of kidney beans is ca.,, with a large fraction of highly unsaturated lipids. For soybeans this percentage is higher, ca. 1 (Holland et al., +332). The total mono- and oligo-saccharide content of kidney beans is considerably higher than its lipid content, i.e. ca. +* for kidney beans (Drumm, +323), with sucrose and stachyose predominating. This percentage is ca. / for soybeans (Holland, +332). The protein content of the beans ranges from +/ -*, with a significant fraction of free amino acids. Soaking and canning decrease e#ect on sugar content, and the concentrations of free amino acids and phenolic acids, due to solubilisation and thermal degradation (Drumm, +323). Flavour is an important sensory aspect of food. During the course of eating, the perception of flavour is not a static experience. The overall perception of a food is based on its initial impact, perception during chewing and perception of residual flavour, that is, it is a dynamic process. To measure these dynamics in real-time as during eating, a novel fast analysis technique with high time resolution such as PTR-MS is required (Lindinger et al., +332). In the present study, the volatile profiles of two types of beans were compared using three analytical techniques. Firstly, the volatile profiles were determined by fingerprint PTR-MS, and secondly, the volatile compounds of

2 64 the beans were identified by model mouth/gc-ms analysis. Finally, model mouth analysis combined with PTR-MS was used to elucidate the dynamic release patterns of four major masses from both types of beans under mouth conditions. Materials and Methods Materials Canned red kidney beans (Tesco; Tesco Stores Ltd., Cheshunt, U.K.) were purchased locally in Cork, Ireland. Canned Salad Club soybeans (Nakashimato Co., Ltd, Tokyo, Japan) were purchased in Tokyo, Japan. The beans were removed from the can, drained and placed in a closed glass flask until analysis. Fingerprint PTR-MS analysis Five beans (+* g) were placed in a glass vial (/** ml) and left to equilibrate for /* min at room temperature. The headspace was drawn at +** ml/min, +/ ml/min of which was led into the PTR-MS (Ionicon Analytik, Innsbruck, Austria). The headspace of the samples was analysed by PTR-MS according to the method described by Lindinger et al. (+332), while employing a constant drift voltage of 0** V. Transmission of the ions through the quadrupole was considered according to the specification of the instrument. Background and transmission corrected spectra were averaged over three cycles. GC-MS analysis Six grams of beans were transferred to the flask of the model mouth (van Ruth and Roozen,,***). Four millilitres of water were added as a saliva substitute. No significant di#erences in volatile release were found between addition of water or saliva in preliminary analysis. A mastication rate of /, rpm was employed. The headspace of the beans was flushed with nitrogen gas at a rate of +** ml/min for, min. The volatile compounds in the gas stream were trapped in *.+-g Tenax TA resin (0*/2* mesh; Supelco, Bellefonte, PA) packed in a glass tube. The volatile compounds of the kidney beans were identified by Dr Maarten Posthumus, Wageningen University, the Netherlands, as described previously (van Ruth et al.,,**.). The volatile compounds of the soybeans were identified by GC-MS (QP,*+*; Shimadzu, Kyoto, Japan). The compounds were thermally desorbed from the Tenax resin (TDTS-,*+*; Shimadzu) at,/* for / min. The GC was equipped with a DB+ column of 0* m length, *.-, mm i.d. and a film thickness of +.** mm. An initial oven temperature of.* was used for / min, followed by a rate increase of +* / min to,/*, and a final hold at,/* for. min. Mass spectra were obtained with 1* ev electron impact ionisation while the mass spectrometer was continuously scanning from m/z,3 to //* at a scan speed of *./ s/scan. Two replicates for each type of beans were analysed. Dynamic PTR-MS measurements Six grams of beans were transferred to the flask of the model mouth (van Ruth and Roozen,,***). Four millilitres of water were added as saliva substitute. A mastication rate of /, rpm was employed and three replicates for each type of beans were analysed. The headspace was drawn from the model mouth at +** ml/min by a vacuum pump, +/ ml/ min of which was led through a heated transfer line into S.M. RUTH et al. the PTR-MS for on-line analysis for one minute. Data were collected for a ions that were the predominant masses measured during a scan of all masses in preliminary experiments using the model mouth. They included m/z --,./, /3 and 1-. Dwell time was *., sper mass. Headspace concentrations were calculated as described by Lindinger et al. (+332). Significant di#erences between the samples and between four time points (*.,/, *./*, *.1/ and +.** min) were determined by (multivariate) analysis of variance ([M] ANOVA) and least significant di#erence tests (LSD tests) (O Mahoney, +320). Linearity of the volatile release was evaluated by Pearson product moment correlation coe$cients (O Mahoney, +320). From the release curves, linear release rates were calculated by multiplying the average headspace concentration of a volatile compound in + min by the volume removed from the headspace in + min (+** ml). The linear release rates of the three replications in each type of beans were subjected to principal component analysis (O Mahoney, +320). For the four masses, proportions of the total volatile release at *.,/ and +.** min were calculated. Results and Discussion Fingerprint PTR-MS The spectra of kidney beans and soybeans determined by headspace analysis are presented in Fig. +. The spectra of the two types of beans show similarities. The di#erences are mainly quantitative, with the soybeans showing generally higher intensities for the various masses present than the kidney beans. This technique is very useful to obtain a fast impression (a fingerprint) of the volatile flavour profile of a food product. It also allows discrimination between more, or less, relevant masses in terms of concentration. However, additional measurements are needed to elucidate the identity of the compounds. In the present study this was achieved by GC-MS analysis. Biasioli et al. (,**-) used fingerprint MS to distinguish four types of red orange juices processed by di#erent stabilisation methods. They were able to discriminate/classify the juices based on their volatile profiles using this fast technique, with the further advantage of ascertaining information on the mass of the discriminating compounds. Identification of volatile compounds by GC-MS The volatile compounds of the kidney beans and the soybeans were identified by GC-MS. The results are presented in Table +. Overall, /+ compounds were identified, -3 in the kidney beans and,3 in the soybeans. Among the compounds identified were alcohols, aldehydes, ketones, esters, acids, sulphur compounds and nitrogen compounds. In decreasing order, +-butanol, ethanol, dimethyl trisulfide,,-butanone and heptanal were present at the highest concentrations in the volatile isolate of the kidney beans. For soybeans, the most abundant compounds in decreasing order were,-propanone, ethanol, +- pentanol, octanal and methanol. The present results di#er from Lee and Shibamoto s study (,***). They reported that the major aroma constituents of soybeans were +-octen---ol, maltol, phenylethyl alcohol, hexanal

3 Comparison of Volatile Flavour Profiles of Kidney Beans and Soybeans 65 Fig. +. Fingerprint mass spectra of kidney beans and soybeans determined by headspace-ptr-ms analysis. and g-butyrolactone. Among these volatile compounds, only +-octen---ol was identified in the present study. This may be due to the studies utilizing di#erent isolation techniques. In the present study, the compounds were isolated in a model mouth system in which the headspace was sampled. This type of isolation technique generally results in analysis of the relatively smaller, more volatile compounds. It is thought that this type of extraction more accurately reflects the profile as experienced during eating. This hypothesis was confirmed by studies we performed which showed significant correlation between flavours released from kidney beans measured dynamically in the model mouth and in-nose during consumption (van Ruth et al.,,**.). Other studies on soybean products showed other common compounds. Some of the identified compounds in our study have been determined in fermented soybeans, e.g. dimethyl disulfide and,-pentylfuran (Leejeerajumnean et al.,,**+). Boatright (,**,) reported hexanal, dimethyl trisulfide and,-pentyl furan among the five most powerful odour active compounds in soymilk. Dynamic PTR-MS measurements As the temporal aspect of flavour release is important for flavour perception, the dynamic release of four masses from the kidney beans and soybeans in the model mouth were monitored by PTR-MS (Fig.,). The masses m/z --,./, /3 and 1- were the predominant masses present in the headspace of the beans during a preliminary scan of all masses. These masses were present in high concentrations in the PTR- MS fingerprints, but were not necessarily the highest in quantity in that type of analysis. The conditions are di#erent in model mouth analysis from the headspace conditions used for fingerprint PTR-MS, e.g. addition of water and mastication. Previous studies have shown these conditions impact considerably on volatile flavour release (van Ruth and Buhr,,**-; van Ruth et al.,,**.). PTR-MS is a soft ionisation technique and, therefore, the ionisation results either in ions corresponding to the volatile compound s mass + or in large product ions. Many of these large product ions have been reported (Buhr et al.,,**,; Yeretzian et al.,,**-). These studies have reported that m/z -- is characteristic for methanol. The mass m/z./ can originate from more than one compound, i.e. ethanol, acetaldehyde, --pentanone or methylbutanal.

4 66 S.M. RUTH et al. Table +. Volatile compounds identified by GC-MS a and their expected major ions in PTR-MS analysis b.

5 Comparison of Volatile Flavour Profiles of Kidney Beans and Soybeans 67 Fig.,. Dynamic release of four compounds from kidney beans and soybeans : methanol (m/z --), ethanol (m/z./),,-propanone (m/z /3) and,-butanone (m/z 1-). Propanal, propanone, propanediol and diacetyl can be responsible for mass m/z /3 and,-butanone,,-methylpropanal, butanal and isobutanal for mass m/z 1- (Buhr et al.,,**,; Yeretzian et al.,,**-). To identify the measured masses, the masses were compared with the compounds identified by GC-MS (Table +), taking into account which were most abundant as well as their fragmentation patterns. It is likely that mass m/z -- originated mostly from methanol, m/z./ from ethanol, m/z /3 from,- propanone, and m/z 1- from,-butanone. Mass m/z./ could also originate from acetaldehyde, which may have not been detected by GC-MS because of its high volatility. The four masses showed considerable di#erences in quantitative release, with mass m/z -- (methanol) showing the highest release and mass m/z 1- (,-butanone) the lowest (Table,). Significant di#erences in release patterns between kidney beans and soybeans were observed [F (+,30) ++/.0, P *.***]. In particular, masses m/z -- (methanol) and m/z /3 (,-propanone) di#ered significantly between the two types of beans. The headspace concentration of mass m/z -- (methanol) was significantly higher in the soybeans at *.,/, *./*, *.1/ and +.** min, and mass m/ z /3 (,-propanone) was significantly higher at *.1/ and +.** min for the soybeans (ANOVA, P *.*/) compared to the kidney beans. For all ions and both bean samples, headspace concentrations increased significantly over time [F (-,30),+.*, P *.***]. The increase in headspace concentrations was fairly linear. This is shown by the high Pearson product moment correlation coe$cients of the volatile concentrations in air versus time (Table -). Linear release rates were calculated (Table -). The linearity of the release indicates that release was not limited due to resistance to mass transfer from the sample to the headspace. This is probably due to the relatively high mastication rate employed (/, cycles/min). Furthermore, it also indicates that depletion does not play a role over the time the beans are consumed. Principal component analysis of the release rates revealed that the release rates were separated along the first principal component axis (Fig. -). Mass m/z -- (methanol), mass m/z /3 (,-propanone) and mass m/z 1- (,- butanone) had high positive loadings on the first principal component and mass m/z./ (ethanol) had a high negative loading on the same component. The scores of the two samples were also separated along the first axis. Soybeans coincided with mass m/z --, /3 and 1-, and kidney beans with mass m/z./. This indicates that linear release rates for methanol,,-propanone and,- butanone were relatively high for soybeans, whereas the

6 68 ethanol release rate was higher for the kidney beans. Data in Table - confirm this observation. The di#erences in composition and structure of the beans are likely to have attributed to the di#erences in flavour release rates. As mentioned before, the headspace concentrations of the di#erent masses changed over time. However, the extent of the change varied among the masses and the type of beans (Fig..). For kidney beans, the proportion of m/z -- (methanol) increased over time from *.,/ to +.** S.M. RUTH et al. min, whereas the proportion of m/z /3 (,-propanone) decreased. For the soybeans, however, m/z -- (methanol) decreased over time soybeans and mass m/z /3 (,- propanone) increased. Changes in the proportions of the compounds over time a#ect the balance of the flavour during the eating process. This may have an e#ect on the perception of the compounds and may result in di#erent sensory experiences between people depending on the method and duration of oral processing. In conclusion, methanol,,-propanone and,-butanone Table,. Concentration of masses (m/z -- (methanol),./ (ethanol), /3 (,-propanone) and 1- (,-butanone); ppbv) determined by model mouth/ptr-ms for kidney beans and soybeans at four time points (n -). Table -. Average linear release rates standard deviation and Pearson product moment correlation coe$cients of masses m/z -- (methanol),./ (ethanol), /3 (,-propanone) and 1- (,-butanone) for kidney beans and soybeans.

7 Comparison of Volatile Flavour Profiles of Kidney Beans and Soybeans 69 Fig. -. Plot of the principal component analysis on the linear release rates of the volatile compounds with loadings of the four masses (m/z --,./, /3 and 1-) and scores of the kidney beans (KB) and soybeans (SB). Fig... Proportions of the masses m/z --,./, /3 and 1- released from kidney beans (KB) and soybeans (SB) after *.,/ and +.** min. were released at higher rates from soybeans, and ethanol was released at a higher rate from kidney beans. The release was fairly linear with time. The di#erent rates resulted in di#erent proportions of the four masses between the two bean samples. The proportions of the individual masses for an individual bean type also varied during the course of release. These changes depended on the bean sample. Acknowledgements We thank Dr Maarten Posthumus, of Wageningen University, the Netherlands for GC-MS analysis of the kidney beans. We also gratefully acknowledge the fellowships of Lonneke Dings from Wageningen University, the Netherlands (Erasmus), and Eugenio Aprea from the University of Innsbruck (Marie Curie), Austria, which allowed us to carry out research at University College, Cork, Ireland. Furthermore, we acknowledge a Grant-in-Aid for Scientific Research (C) No. +./2*+0., Ministry of Education, Culture, Sports, Science & Technology, Japan. References Biasioli, F., Gasperi, F., Aprea, A., Colato, L., Boscaini, E. and Maerk, T.D. (,**-). Fingerprinting mass spectrometry by PTR-MS : heat treatment vs. pressure treatment of red orange juice- a case study. Int. J. Mass Spectrom.,,,-,,., -.- -/-. Boatright, W.L. (,**,). E#ect of gallic acid on the aroma constituents of soymilk and soy protein isolates. J. Am. Oil Chem. Soc., 13, -+1 -,-. Buhr, K., van Ruth, S. and Delahunty, C. (,**,). Analysis of volatile flavour compounds by proton transfer reaction-mass spectrometry : Fragmentation patterns and discrimination between isobaric and isomeric compounds. Int. J. Mass Spectrom.,,,+, + 1. Competition Commission. Fruit and vegetable canning in the United Kingdom, London, UK. Available: http : // Drumm, T.D. (+323). Characterization of the major components of dry beans and their potential relationship to flavor. Diss. Abstr. Int.,.3, -/,. -/,/. Forney, A.K., Halseth, D.E. and Kelly, W.C. (+33*). Quality of canned Ruddy kidney beans as influenced by planting date, harvest time, and length of storage before canning. J. Am. Soc. Horticult. Sci., ++/, +*/+ +*/.. Holland, B., Welch, A.A., Unwin, I.D., Buss, D.H., Paul, A.A. and Southgate D.A.T. (+332). The Composition of Foods. The Royal Society of Chemistry and Ministry of Agriculture, Fisheries and Food, London. Kwang Geun, L., Mitchell, A.E. and Shibamoto, T. J. (,***). Determination of antioxidant properties of aroma extracts from various beans. J. Agric. Food Chem.,.2,.2+1.2,*. Lee, KG and Shibamoto, T. (,***). Antioxidant properties of aroma compounds isolated from soybeans and mung beans. J Agric. Food Chem.,.2,.,3*.,3-. Leejeerajumnean, A., Duckham, S.C., Owens, J.D. and Ames J.M. (,**+). Volatile compounds in Bacillus-fermented soybeans. J. Sci. Food Agric., 2+, /,/ /,3. Lindinger, W., Hansel, A. and Jordan, A. (+332). On-line monitoring of volatile organic compounds at pptv levels by means of proton-transfer reaction-mass spectrometry (PTR-MS): Medical application, food control and environmental research. Int. J. Mass Spectrom. Ion Process., +1-, +3+,.+. O Mahoney, M. (+320). Sensory Evaluation of Food. Statistical Methods and Procedures. Marcel Dekker, New York. Sanz Calvo, M., and Atienza del Rey, J. (+333). Sensory analysis of beans (Phaseolus vulgaris L.). Biotechnol. -Agron. -Soc. Envir., -,,*+,*.. Takamura, H., Kitamura, K. and Kito, M. (+33+). Inhibition by lipoxygenase-- of normal hexanal generation in soybeans. FEBS-letters,,3,,.,. van Ruth, S.M. and Roozen, J.P. (,***). Influence of mastication and artificial saliva on aroma release in a model mouth system. Food Chem., 1+, /. van Ruth, S.M. and Buhr, K. (,**-). Influence of saliva on temporal volatile flavour release from rehydrated red bell peppers determined by Proton Transfer Reaction Mass Spectrometry. Eur. Res. Technol.,,+0,,,*,,-. van Ruth, S.M., Dings, L., Buhr, K. and Posthumus, M.A. (,**.). In vitro and in vivo volatile flavour analysis of red kidney beans by proton transfer reaction-mass spectrometry. Food Res. Int., -1, 12/ 13+. Wu, W., Williams, W.P., Kunkel, M.E., Acton, J.C., Huang, Y., Wardlaw, F.B. and Grimes, L.W. (+33/). True protein digestibility and digestibility-corrected amino acid score of red kidney beans (Phaseolus vulgaris L.). J. Agric. Food Chem.,.-, +,3/ +,32. Wu, W., Williams, W.P., Kunkel, M.E., Acton, J.C., Wardlaw, F.B., Huang, Y. and Grimes, L.W. (+33.). Thermal e#ects on in vitro protein quality of red kidney beans (Phaseolus vulgaris L.). J.

8 70 Food Sci., /3, Yeretzian, C., Jordan, A. and Lindinger, W. (,**-). Analysing the S.M. RUTH et al. headspace of co#ee by proton-transfer reaction-mass spectrometry. Int. J. Mass Spectrom.,,,-,,., ++/ +-3.