2016 Volume 22 Issue 1 Pages 111-116
Sake samples having characteristics of 4-vinylguaiacol (4-VG)-like odor were analyzed by GC-MS and organoleptic methods. The quantified levels of 4-VG, 4-vinylphenol, 4-ethylguaiacol, and 4-ethylphenol of 15 sake samples, including 7 samples presenting 4-VG-like odor characteristics by multiple assessors, did not show a clear relation with the perception rate of 4-VG-like odor characteristics. Hence, we screened for other compound(s) responsible for the 4-VG-like odor characteristics, and found that odorous sake samples contain 2-methoxy phenol (guaiacol), 2,6-dimethoxyphenol (syringol), and 3-methyphenol at higher levels than the odor-less sample. The quantified level of guaiacol well matched the perception rate of 4-VG-like odor characteristics. The recognition threshold value of guaiacol in the sake and of 4-VG in the model sake solution was estimated as 14.6 µg/L and 141 µg/L, respectively. The concentration of guaiacol in the most odorous sake samples was higher than the estimated recognition threshold.
Phenolic odor characteristics are sometimes perceived in the sensory evaluation of sake. The odor characteristics are classified as an off-flavor, which notably decreases the sensory evaluation score of sake. 4-Vinylguaiacol (4-VG) is known as a causative compound for the phenolic odor characteristics of sake (i). 4-VG in sake is formed by spoilage bacteria, such as Bacillus spp. and/or Staphylococcus spp., through the decarboxylation of ferulic acid in the sake brewing process (Kaneoke, 2014). Ferulic acid is consequently formed by rice koji enzymes from steamed rice in the sake brewing process (Ito et al., 2014), while the main microbes in sake fermentation, koji mold (Aspergillus oryzae) and sake yeast (Kyokai-yeast; Saccharomyces cerevisiae) do not form 4-VG (Suezawa et al., 1998, Mukai et al., 1998). In wine making, volatile phenolic odor compounds have been extensively studied as off-flavor, namely “phénolé”, components (Chatonnet et al., 1992, 1993). Major components of the odor characteristics in white wine are 4-VG and 4-vinylphenol (4-VP) formed by wine yeasts having substituted cinnamate decarboxylase activity (Chatonnet et al., 1993). Meanwhile, in red wine, 4-ethylphenol (4-EP) and 4-ethylguaiacol (4-EG) are causative components of the odor that is produced by the spoilage yeast Bretanomyces/Dekkera spp. This yeast can reduce vinyl-phenols to ethyl-phenols (Chatonnet et al., 1992, Suarez et al., 2007). Recently, the “phénolé” off-flavor has been identified in some Japanese wines and the spoilage microbe has been investigated intensively (Onda, 2013). 4-VG is also of interest in beer brewing (Coghe et al., 2004) and in Chinese rice wine brewing (Mo and Xu, 2010) as an odorous component. It was also reported that 4-VG may be formed by heating under acidic conditions in Japanese shochu distillation process (Koseki et al., 1996).
Some sake samples entered in the Akita prefectural sake awards of 2013 had disharmonious 4-VG-like odor characteristics and as a result were ranked low. Therefore, we analyzed well known compounds of phenolic odor characteristics in the sake samples using GC-MS. However, the results did not show a clear relation between the levels of phenolic odor compounds and the frequency of 4-VG-like odor characteristics. Thus, we screened for other causative compounds, and found a number of possible candidates. Thereafter, we determined the level of these compounds in the sake samples and elucidated some of their organoleptic properties.
Materials 2-Methoxy phenol (guaiacol) was purchased from Tokyo Chemical Industry Co. (Tokyo, Japan). 2,6-Dimethoxyphenol (syringol), other authentic phenolic compounds and ethyl acetate (infinity pure grade) were obtained from Wako Pure Chemical Industries (Osaka, Japan). Jummai-type sakes entered in the Akita prefectural sake awards of 2013 and 2014 were used as sake samples. Commercial sake samples produced in Akita Prefecture were purchased from a local market.
Preparation of extracts of volatile compounds from sake samples After a 10-mL sake sample was diluted with an equal volume of water, 20 µg of 4-n-butylphenol (4-BP, internal standard; IS) dissolved in 95% ethanol was added. The sample was applied to a Bond Elute C18 LRC 500 MG (Agilent Technologies, Santa Clara, CA, USA) pre-conditioned using 7 mL of methanol and 7 mL of water successively, then the column was washed once with 7 mL of water and trapped compounds were eluted with 2 mL of ethyl acetate. The eluted solution in a glass tube with a PTFE packing cap was washed twice with 1 mL of 0.2 M NaHCO3 aqueous solution, performed by shaking for 1 min and subsequent removal of the aqueous layer after weak centrifugation (2000 rpm, 5 min). The solution was dried by 0.5 g of anhydrous sodium sulfate and concentrated by rotary evaporation (60°C, 50 hPa) to approximately 200 µL in a 1.5 mL glass vial. In the identification experiment, concentrated volatile extracts were prepared from 50-mL sake samples using the same procedure.
Gas chromatography-mass spectrometry An Agilent 7890A GC/5975C inert XL MSD (Agilent Technologies) was used. The column was a 30 m x 0.25 mm i.d. HP-INNOWAX fused silica capillary type with a film thickness of 0.5 µm (Agilent Technologies). The column temperature was held at 55°C for 7 min and then raised from 55°C to 220°C at a rate of 5°C/min and held for 20 min at 220°C. The injector temperature was 220°C, and the flow rate of the helium carrier gas was 1.5 mL/min. An injection volume of 2 µL was applied using the split-less mode. The mass spectrometer was used with an ionization voltage of 70 eV (EI) and an ion source temperature of 230°C. Quantification of phenolic compounds was carried out in the selected ion monitoring mode using m/z 152 for 4-EG, m/z 122 for 4-EP, m/z 150 for 4-VG and 4-BP (IS), m/z 120 for 4-VP, m/z 109 for 2-methoxyphenol (guaiacol), m/z 154 for 2,6-dimethoxyphenol (syringol), and m/z 107 for 3-methylphenol (m-cresol). Calibration curves were constructed by using 17% ethanol 0.1 M succinate-Na buffer solutions (pH 4.3) of the phenolic compounds. The ratio of the peak area of each compound to the peak area of IS was used for calculation. Identification of each component was conducted in the full scan mode.
HPLC analysis of 4-VG Preliminary quantification of 4-VG was conducted using RP-HPLC. Preparation of samples was the same as for GC-MS analysis, with the exception that 3,4-dimethylphenol was used as the IS. A Capcell Pak C18 Type MG column (4.5 mm × 250 mm) (Shiseido, Tokyo, Japan) was used. Solvent A was acetonitrile, and solvent B was 0.1% phosphoric acid / water. A linear gradient was used from A: B=20:80 to A: B=60:40 over 50 min at a flow rate of 1.0 mL/min. The applied sample volume was 20 µL, and absorbance at 280 nm was monitored.
Threshold assessment of guaiacol in the sake The threshold of guaiacol in sake was examined according to the BCOJ sensory analysis method (Brewery Convention of Japan, 2002). This method is founded on the forced-choice ascending concentration series method of limits of ASTM E-679. The base sake used in this experiment was sample No. 1 in Table 2. It is a ginjo-type sake of 15 – 16% ethanol, which does not contain guaiacol. Assessors were comprised of ten undergraduate and graduate students from our school in their twenties (form 21 to 24 years old) who were trained using an authentic compound (guaiacol, 64 µg/L in the sake), and a 42 years old teacher. Each assessor tasted a sample using a traditional tasting cup for sake, Jyanome-choko, by only smelling. We presented 4 sets of the triangle test, although the official method calls for 6 sets of the triangle test, to prevent sensory fatigue of assessors. If the threshold was not determined in the presented test set, another set in the triangle test (comprised of a lower or higher level) was presented to the assessor. We were able to obtain the difference threshold by this method. In addition, we also estimated the recognition threshold by determining the concentration at which the assessor detected the odor.
| Sample No.b | 4-VG | 4-VP | 4-EG | 4-EP | Guaiacol | Syringol | 3-MP |
|---|---|---|---|---|---|---|---|
| Concentration (µg/L)a | |||||||
| 1 | 45 ± 5 | 79 ± 5 | 0.1±0.0 | 0.2±0.0 | 0.0±0.0 | 0.0±0.0 | 0.0±0.0 |
| 2 | 451 ± 7 | 57 ± 1 | 0.2±0.0 | 0.1±0.0 | 0.0±0.0 | 0.3±0.3 | 0.0±0.0 |
| 3 | 683 ± 4 | 70 ± 6 | 0.2±0.0 | 0.1±0.0 | 1.9±0.1 | 0.5±0.6 | 0.0±0.0 |
| 4 | 263 ± 16 | 87 ± 4 | 0.1±0.0 | 0.1±0.1 | 1.2±0.2 | 0.5±0.2 | 0.0±0.0 |
| 5 | 215 ± 8 | 47 ± 0 | 0.1±0.0 | 0.1±0.0 | 8.8±0.1 | 2.0±0.6 | 0.2±0.0 |
Threshold assessment of 4-VG in the model sake solution All sake samples analyzed in this study contained 4-VG at greater than 38 µg/L; therefore, we estimated the threshold of 4-VG in a model sake solution. The base model sake solution contained ethanol, 16.0(v/v)%; n-propylalcohol, 60 mg/L; isobutylalcohol, 40 mg/L; isoamyl alcohol, 120 mg/L; ethylacetate, 50 mg/L; isoamyl acetate, 2 mg/L; ethyl caproate, 5 mg/L; β-phenylethyl alcohol, 80 mg/L; and tyrosol, 90 mg/L. The pH was adjusted to 4.3 using 10 mM sodium succinate. The assessment procedure and the eleven assessors were the same as for the guaiacol threshold assessment. Assessors were trained using an authentic compound (4-VG, 1600 µg/L in the model sake solution) before assessment.
Odor description of phenolic compound spiked sake samples by sake brewing technology specialists We used sake sample No. 1 in Table 2, which was spiked with either guaiacol or 4-VG. Five specialists of sake brewing technology of the Akita Research Institute for Food & Brewing described the odor characteristics of samples. The assessors were informed before the examination that the test samples contained exogenous chemical compounds as potential candidates causing 4-VG-like odor characteristics, but they were not provided the name of the compound. Testing was conducted in a blind manner showing the un-spiked reference sake sample, and a freestyle assessment format was used. Each assessor tasted a sample using the traditional tasting cup for sake, Jyanome-choko, by only smelling.
Quantification of 4-vinyl and 4-ethyl phenols in sake In the analysis of volatile phenolic compounds in wine, the solvent-extracted samples were washed with a weak alkaline aqueous solution to remove phenolic acids, for example ferulic acid (Chatonnet et al., 1992), since phenolic acids produce corresponding vinylphenols through decarboxylation under the high temperature conditions of GC-MS analysis. In this study, we removed ferulic and p-coumaric acids in the ethyl acetate extract by washing twice with 0.2 M NaHCO3 aqueous solution. After washing, we confirmed that phenolic acids were not detected in the extract by HPLC analysis monitored at 320 nm (data not shown). In previous GC-MS analyses of volatile phenolic compounds in wine, 3,4-dimethylphenol was used as an IS (Chatonnet et al., 1993, Campolongo et al., 2010), while we used 4-n-butylphenol as an IS because it is not naturally found in sake samples and the retention time in the GC-MS chromatogram was appropriate. The calibration curve for each phenolic compound showed good linearity (R2 ranged from 0.991 to 1.00).
Fifteen junmai-type sake samples entered in the Akita prefectural sake awards of 2013 were analyzed for the level of 4-vinyl and 4-ethyl phenols, considered as off-flavor components in wine (Table 1). Seven of these contained 4-VG-like odor characteristics and the others contained high 4-VG levels in our preliminary HPLC analysis. The concentration of 4-VG ranged from 38 to 181 µg/L and that of 4-VP ranged from 24 to 145 µg/L. In all samples, 4-EG was less than 0.1 µg/L, and 4-EP was less than 1 µg/L. The observed level of 4-VG was far lower than the recognition threshold value in white wine (440 µg/L) (Chatonnet et al., 1993), and 4-VG and other phenolic compounds did not show clear associations with results of the sensory evaluation.
| Sample No. | 4-VG | 4-VP | 4-EG | 4-EP | Odor indicated No.b |
|---|---|---|---|---|---|
| Concentration (µg/L)a | |||||
| 1 | 65 ± 3 | 44 ± 1 | 0.0±0.0 | 0.3±0.0 | 0 |
| 2 | 73±3 | 77 ± 1 | 0.0±0.0 | 0.3 ±0.0 | 0 |
| 3 | 79 ± 8 | 41 ± 6 | 0.0±0.0 | 0.2 ± 0.1 | 0 |
| 4 | 97 ± 3 | 117 ± 3 | 0y.1±0.0 | 0.4 ±0.0 | 0 |
| 5 | 100 ± 3 | 70 ± 2 | 0.0±0.0 | 0.5 ± 0.1 | 0 |
| 6 | 110 ± 6 | 123 ± 5 | 0.1±0.0 | 0.6±0.0 | 0 |
| 7 | 115 ± 3 | 109 ± 2 | 0.0±0.0 | 0.3 ±0.0 | 0 |
| 8 | 125 ± 5 | 145 ± 10 | 0.1 ± 0.0 | 0.6±0.1 | 0 |
| 9 | 38 ± 2 | 24 ± 0 | 0.0±0.0 | 0.3 ±0.0 | 7.1 |
| 10 | 77 ± 1 | 39 ± 0 | 0.0±0.0 | 0.2 ±0.1 | 7.1 |
| 11 | 88 ± 2 | 41 ± 0 | 0.0±0.0 | 0.2 ±0.0 | 14.3 |
| 12 | 114 ± 6 | 80 ± 1 | 0.1±0.0 | 0.5 ±0.1 | 28.6 |
| 13 | 181 ± 2 | 114 ± 2 | 0.1±0.0 | 0.5 ±0.0 | 42.9 |
| 14 | 61 ± 3 | 41 ± 1 | 0.0±0.0 | 0.5±0.1 | 57.1 |
| 15 | 57 ± 8 | 41 ± 1 | 0.0±0.0 | 0.4 ±0.1 | 64.2 |
Identification of compound(s) responsible for 4-VG-like odor characteristics The four phenolic compounds analyzed could not explain the results of the sensory evaluation; therefore, we expanded our search for candidate phenolic compound(s) in sake. We analyzed two sake samples; one is a junmai-type sake entered in the Akita prefectural sake awards of 2014 having distinct 4-VG-like odor characteristics, and the other is a commercial junmai-daiginjo-type sake sample without 4-VG-like odor characteristics. Knowing that 4-VG has smoky odor characteristics (i), we searched for other smoky flavor compounds (Guillen et al., 2002, Pino, 2014) by comparing full scan GC-MS chromatograms. As a result, we found that guaiacol, syringol, and m-cresol (3-MP) were present at rather high levels in the sake with distinct 4-VG-like odor characteristics. These compounds were identified by comparing GC retention times and mass spectra with authentic compounds. Guaiacol and syringol are well known smoky components (Pino, 2014, Wasserman, 1966). While guaiacol had been identified in the steam distillate of rice bran (Tsugita et al., 1977), its presence in sake samples has not been documented. Another search result, 4-methylguaiacol, a typical smoky compound (Pino, 2014), was not found in the sake sample.
Quantification of guaiacol, syringol, and 3-MP in sake We re-analyzed the sake samples listed in Table 1 and five commercial sake samples (Table 2) with respect to the identified search compounds. The frequency of 4-VG-like odor indication is well matched to the levels of guaiacol and syringol, as shown in Fig. 1. Samples showing high levels of guaiacol and 3-MP; No. 14 and 15 of Fig. 1 and No. 5 of Table 2, are products from the same factory. These samples showed rather low 4-VG and 4-VP levels. In contrast, sample No. 2 in Table 2 showed a rather high 4-VG level, but very low guaiacol and syringol levels. These results suggested that these phenolic compounds might be formed through different pathways.
Concentrations of guaiacol (■), syringol (
), and 3-MP (□) in junmai-type sake samples and 4-VG-like odor perception rate (●). Samples and numbers are the same as in Table 1. Quantification data are means and S.D. of three measurements.
Threshold of guaiacol added to the sake sample The difference and recognition threshold values for guaiacol in the sake sample obtained in this study are shown in Fig. 2. The geometric mean values were 13.7 µg/L and 14.6 µg/L for the difference and recognition thresholds, respectively. The difference between these two threshold values was minimal. The obtained recognition threshold was lower than the level of samples No. 14 and No. 15 in Fig. 1. The distribution of best estimate threshold of each assessor agreed well with the frequency of 4-VG-like odor indication of Fig. 1. These results indicated that guaiacol contributed to the 4-VG-like odor characteristics. It was reported that guaiacol has a very low threshold in apple juice (Eisele and Semon, 2005), while the threshold value obtained in this study was about ten-fold higher. The reason for this observed difference might be due to differences in the background flavor components of the products.
Histogram of estimated thresholds of eleven assessors for guaiacol-spiked sake samples.
G.M. indicates geometric mean. ● or ○ indicates an assessor.
Threshold of 4-VG added to the model sake sample solution The difference and recognition threshold values for 4-VG in the model sake sample obtained in this study are shown in Fig. 3. The geometric mean values were 80 µg/L and 141 µg/L for difference and recognition thresholds, respectively. The obtained difference threshold value is rather higher than the previously reported value, 52 µg/L in sake (URL cited i). The recognition threshold value was higher than that in water or in a model wine solution (Chatonnet et al., 1993), but was lower than for white wine (440 µg/L Chatonnet et al., 1993) and beer (300 µg/L Coghe et al., 2004). The threshold in sake might be higher than reported in the present study data, as Chatonnet (1993) reported a higher recognition threshold in the white wine than in the model solution. It is proposed that the simple flavor components of the model sake solution might enable the recognition of odor characteristics more easily. The large difference between the two thresholds might be due to difficulty in the odor description of 4-VG. The recognition threshold of 4-VG obtained in this study was higher than the concentration of most sake samples listed in Table 1, which suggests that 4-VG contributes little to the 4-VG-like odor characteristics in the Akita prefectural sake award.
Histogram of estimated thresholds of eleven assessors for the 4-VG-spiked model sake sample.
G.M. indicates geometric mean. ● or ○ indicates an assessor.
Odor description of guaiacol or 4-VG spiked sake or model sake sample solution In the threshold assessments, guaiacol spiked to the sake sample was described as medical (5 persons), smoky (3), phenolic (1), burnt (1), and mold-like (1); in contrast, 4-VG was described as old clothes (3), old house (2), bad breath (1), burnt (1), mold-like (1), and unpleasant smell (3). The described characteristics of guaiacol were very clear, indicating that guaiacol in sake is easily recognized. In contrast, the description of 4-VG was diverse and obscure; therefore, it was difficult to describe the odor characteristics clearly, even when the assessor was capable of detecting it. Odor characteristics of spiked phenolic compounds in sake were described by 5 specialists of sake brewing technology (Table 3). Four persons attributed 4-VG-like characteristics to the guaiacol-spiked sample (32 µg/L), whereas two persons described the 4-VG-spiked sample (235 µg/L after spiking) as smoky. These results suggested the possibility that the sake specialists in this study misidentified odor characteristics caused by guaiacol as being caused by 4-VG. A large-scale future investigation is necessary to clarify whether this misidentification was an isolated incident.
| Sample No. | 4-VG | Guaiacol | Odor descriptiona (number of persons) |
|---|---|---|---|
| Concentration (µg/L) | |||
| 1b | 45 | 0 | |
| 2 | 138 | 0 | smoky (1) |
| 3 | 235 | 0 | smoky (1), slightly smoky (1) |
| 4 | 45 | 16 | 4-VG like (1), slightly 4-VG like (2) |
| 5 | 45 | 32 | distinctively 4-VG like (1), 4-VG like (2), smoky (1) |
General considerations Although the sake samples analyzed in this study were the products of a limited area, it was revealed that some sakes from particular factories contain guaiacol at a significant sensory level. Guaiacol is known as an off-flavor component of Alicyclobacillus spp. bacterial spoilage in apple juice. The bacteria produce guaiacol through oxidative degradation of ferulic acid via 4-VG (Siegmund and Pollinger-Zierler, 2006, Goto et al., 2008, Smit et al., 2011). In the case of wine production, guaiacol, 4-methylguaiacol, and syringol were detected in barrel-aged wine (Prida and Chatonnet, 2010) as well as in wine made from bushfire smoke-affected grapes in Australia (Mayer et al., 2014). In beer brewing, guaiacol increases as a result of the use of torrefied malts (Scholtes et al., 2014). In the sake brewing process, however, such high temperature conditions do not exist and sake is not aged in burnt wood containers. In addition, 4-methylguaiacol, a major smoke component, was not detected in sake samples; thus, it was indicated that guaiacol and syringol in sake should be formed by biological conversion from ferulic acid and sinapic acid, respectively, under oxidative conditions. The odor impact of syringol is estimated to be far weaker than that of guaiacol (Wasserman, 1966); however, another report showed that syringol is nearly equal in odor impact to guaiacol. Further study is needed to clarify the contribution of syringol to the phenolic odor characteristics of sake.
Acknowledgements We are grateful to the Akita Prefectural Sake Brewing Society for the kind gift of sake samples, and also to the students of the Food and Brewing Group, Department of Biological Resource Sciences of Akita Prefectural University for their assistance with the sensory analysis.
