Determination of sotolon, sotolon precursors, and minerals in Okinawan awamori

Abstract

Sotolon (3-hydroxy-4,5-dimethyl-2(5H)-furanone) is a volatile aroma compound with various desirable odors, including sweet, honey-like, and caramel-like. In this study, sotolon was quantified in 44 commercial awamori, an Okinawan rice-based distilled alcoholic beverage. Sotolon was detected at concentrations up to 20.5 µg/L, and awamori stored in earthenware jars had a significantly higher sotolon content than that in non-earthenware jars (p < 0.0001). Additionally, sotolon precursors (diacetyl, acetoin, acetaldehyde, and 2-oxobutyric acid) were present at concentrations of 0.13–1.48, 0.24–4.70, 77.50–332.50, and 0.16–0.96 mg/L, respectively, and showed no notable differences between storage vessels or periods. Moreover, Al and Cu were considerably higher in earthenware-jar-stored awamori, whereas Al, Cu, Fe, and Mn were markedly higher in aged awamori. These results suggest the possibility of mineral elution catalyzing the oxidation of sotolon precursors in earthenware jars and provide a foundation for further research on the mechanism of sotolon generation in awamori.

Introduction

Awamori, one of the oldest distilled liquors in Japan, is a rice-based alcoholic beverage that is originated and mainly produced in Okinawa Prefecture, Japan. Awamori is made from rice, which is seeded with black koji mold (Aspergillus luchuensis), treated with water and enzymes to form seed mash (moromi), and is alcohol fermented. The fermented moromi is then distilled in a pot to an alcoholic concentration of approximately 70 % and results in undiluted awamori (genshu). The genshu possesses a pungent odor immediately after production due to the presence of fatty acids, esters, and sulfur compounds, and is therefore stored in vessels. Awamori is traditionally stored in clay pottery (earthenware jars); however, given that the pottery storage method results in high evaporation loss and consumes space and time, commercial awamori is stored in stainless steel or enamelware tanks. Nonetheless, sensory evaluations have shown that the traditional storage method using earthenware jars produces a unique caramel-like aroma.

Sotolon is a volatile compound with an intense odor and has odor thresholds of 2 µg/L in model wine, 8 µg/L in dry white wine (Pons et al., 2008), and 3.1 µg/L in Japanese sake (Boerzhijin et al., 2023; Isogai et al., 2004). The compound is also famous for possessing different aroma characteristics, such as sweet, honey, caramel-like, nutty, walnut, and spicy (Milheiro et al., 2021). Moreover, sotolon is an important volatile compound that adds qualities to Japanese sake (Isogai et al., 2005), sherry, and various types of wines, such as white wine, fortified wine (Pareira et al., 2018), and Chinese Syrah wine (Zhao et al., 2017). Furthermore, Scholtes et al. (2015) suggested diacetyl, acetoin, acetaldehyde, 2-oxobutyric acid, ethanol, ascorbic acid, and hydroxyacetic acid as possible precursors for sotolon production in wine. However, reports on the aroma characteristics of awamori during aging are limited. A study on the formation of volatile aroma compounds in aged awamori suggested a contribution of acetaldehyde evaporation and minerals leaching earthenware jars to ester hydrolyses, fatty acid formations, and oxidation processes (Tamaki et al., 1986).

To the best of our knowledge, no study has investigated sotolon precursors and mineral content in Okinawan awamori. Therefore, the objective of this study was to quantify sotolon, possible sotolon precursors, and nine minerals in awamori stored in various vessels at different ages, and to elucidate their correlation. Sotolon content was determined using liquid chromatography with tandem mass spectrometry (LC-MS/MS) by eliminating nonvolatile substances and unstable products with short half-life periods, such as intermediate products. To evaluate the possible sotolon precursors, gas chromatography–mass spectrometry (GC-MS) was used to analyze diacetyl, acetoin, and acetaldehyde, whereas high-performance liquid chromatography (HPLC) was used to analyze 2-oxobutyric acid. Additionally, inductively coupled plasma mass spectrometry (ICP-MS) was used for mineral analysis. The correlations between these components were then statistically evaluated.

Materials and Methods

Chemicals and reagents Sotolon (> 97 %) was purchased from Sigma-Aldrich (St. Louis, MO, USA). Isovanillic acid (> 98.0 %) was obtained from Tokyo Chemical Industry (Tokyo, Japan). Methanol and formic acid were of LC-MS grade, and nitric acid was of special grade (specific gravity 1.38) and obtained from FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan). The ICP multi-element standard solution IV (23 elements) was purchased from Merck KGaA (Darmstadt, Germany). Acetaldehyde (extra pure grade) and 2-oxobutyric acid (> 97 %) were purchased from FUJIFILM Wako Pure Chemical Corporation. Diacetyl (> 98 %), acetoin, and 3-octanol (> 98.0 %) were obtained from the Tokyo Chemical Industry.

Awamori samples Seventeen types of young awamori (ippan-shu) with maturation periods of less than three years and 27 types of aged awamori (ku-su) with maturation periods longer than three years were obtained from local markets in Okinawa Prefecture, Japan, from April to September 2023 (Table 1). The alcohol content of awamori ranged from 25–44 %, as labeled in the products. Forty and 42 awamori liquor samples were used for quantitative analysis of sotolon and minerals, respectively. Twenty-nine awamori products were used for quantitative measurement of diacetyl, acetoin, acetaldehyde, and 2-oxobutyric acid.

Table 1. Alcohol and sotolon content in awamori.

No.	Sample information	Aging period (year)	Aging container	Alcohol content (%)	Sotolon content(µg/L)
1	-	10	earthenware	35	11.23
2	2 months aged in earthenware, after then 10 months in stainless steel tank	1	earthenware	30	11.09
3	10 years aged (20 %), young (80 %) blended in stainless steel tank	2.8*	non-earthenware	30	4.25
4	-	27	earthenware	40	20.50
5	vacuum distilled young	1	non-earthenware	30	nd#
6	22 years aged (40 %), 21 years (55 %), 10 years (5 %) blended	20.9*	non-earthenware	42	6.39
7	5 years aged (60 %), vacuum distilled young (40 %) blended	3.4*	earthenware	30	2.96
8	-	5	non-earthenware	35	3.06
9	-	3	earthenware	42	7.11
10	aged in stainless steel tank	3	non-earthenware	40	2.73
11	aged in stainless steel tank	3	non-earthenware	30	1.53
12	-	10	earthenware	43	4.34
13	aged in stainless steel tank	15	non-earthenware	43	2.56
14	-	21	earthenware	35	3.33
15	-	1	non-earthenware	30	nd
16	-	1	non-earthenware	43	nd
17	-	1	non-earthenware	30	nd
18	aged in stainless steel tank	3	non-earthenware	35	nd
19	aged in stainless steel tank	5	non-earthenware	40	0.97
20	aged in stainless steel tank	8	non-earthenware	43	0.77
21	direct-heating distilled young	1	non-earthenware	30	nd
22	direct-heating distilled in oak barrel	1	non-earthenware	25	nd
23	-	1	non-earthenware	30	nd
24	aged in oak barrel	3	non-earthenware	25	3.24
25	-	3	earthenware	30	5.75
26	aged (20%), young (80%) blended	1.4*	non-earthenware	30	nd
27	-	1	non-earthenware	30	nd
28	aged in oak barrel	3	non-earthenware	30	2.17
29	-	1	non-earthenware	30	nd
30	-	1	non-earthenware	30	nd
31	-	1	non-earthenware	30	nd
32	aged in stainless steel tank	10	non-earthenware	43	0.98
33	-	10	earthenware	43	nd
34	-	3	earthenware	25	nd
35	3 years aged, young blended in oak barrel	3	non-earthenware	30	1.05
36	3 years aged, young blended in oak barrel	3	non-earthenware	30	3.04
37	-	3	non-earthenware	30	3.68
38	-	5	earthenware	44	15.77
39	5 years aged (30%), 3 years (70%)	3.6*	non-earthenware	35	nd
40	aged in oak barrel	3	non-earthenware	25	nd
41	aged (70%), young (30%) blended in oak barrel	2.4*	non-earthenware	40	not analyzed
42	3 years aged, young blended in oak barrel	2	non-earthenware	30	not analyzed
43	-	10	earthenware	43	not analyzed
44	-	1	non-earthenware	25	not analyzed

*The aging period of blended awamori was determined based on the combination of its composition. For example, No. 3 is composed of 20% of 10 years aged awamori and 80% of young awamori (1 year), thus its aging period is (10 years × 0.2) + (1 year × 0.8) = 2.8 years.

^#nd. not detected.

Sotolon analysis using LC-MS/MS To determine the sotolon analytical conditions, sotolon standard solutions were prepared by adding five sotolon concentrations (0.625, 1.25, 2.5, 5, and 10 µg) to 25 mL of 30 % (v/v) ethanol solution. A sotolon model solution was then prepared in commercial awamori (Kikunotsuyu) with sotolon concentrations ranging from 0.625 to 10 µg. Sotolon was extracted using an Oasis MAX solid-phase extraction (SPE) cartridge containing a mixed-mode ion exchange/reversed-phase sorbent (Waters, Milford, MA, USA). Briefly, an Oasis MAX cartridge (6 cc/150 mg) was conditioned with 3 mL of methanol, 3 mL of H₂O, and 3 mL of 5 % ammonia solution. Subsequently, 25 mL of H₂O and 1 mL of internal standard isovanillic acid (10 µg/mL in methanol) were dissolved in 25 mL of awamori products (the pH of the sample was adjusted to 9.0–9.5 by addition of 50 µL of 5 % ammonia solution) and were then loaded onto the cartridge. The cartridge was subsequently eluted with 2 mL of a 5 % ammonia solution and 2 mL of methanol, followed by 2 mL of methanol containing 2% lactic acid. Dimethyl sulfoxide (10 µL) was then added to this elute to inhibit sotolon evaporation, and the eluate was then concentrated to approximately 0.5 mL using a centrifugal evaporator (EYELA CVE-3110, Tokyo RikaKikai, Tokyo, Japan).

The sotolon content was measured using LC-MS/MS analysis on an Acquity H-Class ultra-performance liquid chromatography platform equipped with a Quattro Micro API triple quadrupole mass spectrometer (Waters). The column used was YMC-Triart C18 (75 mm × 3.0 mm i.d., 1.9 µm, YMC, Kyoto, Japan). Mobile phases A and B were 0.1 % formic acid and methanol, respectively, and mobile phase A:mobile phase B (8:2) was eluted at a flow rate of 0.3 mL/min for 15 min. The column temperature was set to 40 °C and the injection volume was 2.5 µL. The mass spectrometry (MS) ionization conditions were as follows: ESI positive mode; ion source temperature, 120 °C; desolvation temperature, 350 °C; desolvation gas flow, 600 L/h; and cone gas flow, 50 L/h. The cone and collision voltages were set to 22 and 9 V, respectively. Tandem mass spectrometry detection was performed in the multiple reaction monitoring mode. Sotolon was detected as a product ion at m/z 82.9, which resulted from the precursor ion at m/z 128.9, and the internal standard was detected as a product ion at m/z 124.9 from the precursor ion m/z 169.0. Data were processed using MassLynx 4 (Waters). The sotolon content in awamori was calculated using a regression formula obtained from the calibration curve of the model solution. All measurements were performed in triplicate, and quantitative values were converted to 25 % alcohol concentration.

Diacetyl and acetoin analyses using GC-MS Briefly, 10 mL of awamori were mixed with 10 mL of H₂O and 100 µL of internal standard 3-octanol (100 µg/mL in ethanol). Subsequently, the mixture was loaded into acetone and H₂O-preconditioned Sep-Pak AC2 Plus short cartridge (400 mg, 85 µg, Waters), followed by elution with 2 mL of acetone. The eluted extract (1 µL) was injected into a 7890 B GC-5977A mass spectrometer (Agilent Technologies, Santa Clara, CA, USA). The inlet temperature was set at 250 °C, and the split ratio was 1:1. The column used was DB-WAX (30 m × 0.25 mm, 0.25 µm, Agilent Technologies). Helium was used as a carrier gas at a constant flow rate of 1.5 mL/min. The oven was initially held at 40 °C for 1 min, raised to 110 °C at 4 °C/min, and heated to 200 °C at 10°C/min. The MS ionization voltage was set at 70 eV and the scan range was m/z 33–500. Diacetyl and acetoin were detected in selected ion monitoring at m/z 43 and 86, respectively. Diacetyl and acetoin calibration curves were established using awamori solution that was spiked with six standard levels in the range of 5–1 000 µg/L, with three replicates for each level. All measurements were performed in triplicate, and quantitative values were converted to a 25 % alcohol concentration.

Acetaldehyde analysis using GC-MS Briefly, 1 mL of awamori was blended with 100 µL of internal standard 3-octanol (100 µg/mL in ethanol). The mixture was then analyzed using GC-MS under the same aforementioned conditions. Acetaldehyde was detected in selected ion monitoring at m/z 29. Acetaldehyde calibration curves were established using awamori solution that was spiked with six standard levels in the range of 10–2 500 µg/L with three replicates for each level. All measurements were performed in triplicate, and quantitative values were converted to a 25 % alcohol concentration.

Analysis of 2-oxobutyric acid using HPLC The 2-oxobutyric acid was extracted from awamori using SPE and derivatized prior to HPLC analysis, as previously described (Isogai et al., 2004; Singh et al., 1993). Briefly, awamori (0.5 mL) was loaded into a Bond Elut SAX column (500 mg/3 mL; Agilent Technologies), which was preconditioned with methanol and H₂O. The cartridge was eluted with 3 mL of 50 % methanol and the extract was collected using 2 mL of 1 M HCl. Next, 1 mL of the extract was mixed with 1 mL of 1,2-o-phenylenediamine (10 mg/mL) and the mixture was heated to 100 °C for 30 min using a dry block heater. The solution was cooled to room temperature (25 °C), and 2-oxobutyric acid was extracted using 2 mL of ethyl acetate. Subsequently, 1 mL of the extract was dried under N₂ gas flow and the dried extract was dissolved in 1 mL of acetonitrile:methanol (5:8, v/v). The 2-oxobutyric acid was analyzed using HPLC with fluorescence detection (LC-20AD, RF-20AXS, Shimadzu, Kyoto, Japan). The column used was YMC-Triart C18 (150 mm × 4.6 mm i.d., 3 µm, YMC). The injection volume was 2 µL, and the oven was set at 40 °C. Mobile phases A and B were 20 mM phosphate buffer (pH 7.0) and acetonitrile: methanol (5:8, v/v), respectively. Mobile phase A:mobile B (4:6) was eluted at a flow rate of 0.65 mL/min. The excitation and emission wavelengths were programmed at 340 and 389 nm, respectively. The 2-oxobutyric acid calibration curves were established using awamori which was spiked with six standard concentrations ranging from 25–2 500 µg/L, with three replicates for each level. All measurements were performed in triplicate, and quantitative values were converted to a 25 % alcohol concentration.

Mineral analysis using ICP-MS To eliminate ethanol interference in awamori, standard solutions for the calibration curves were diluted with a 1 M nitric acid solution containing 10 % ethanol. Mineral standard solutions (23 elements) were prepared at seven concentrations ranging from 0.01–2.5 mg/L. The awamori sample solution was prepared by diluting 2 mL of awamori with 8 mL of 1 M nitric acid. Aliquots were analyzed using an ICP-MS 7700 Series (Agilent Technologies). The plasma gas (Ar) flow rate was set to 15 L/min and the carrier gas flow rate was 1 L/min. Helium gas at a flow rate of 5.0 mL/min was used as the cell gas for ⁴⁴Ca; no gas was used for the other elements (²³Na, ²⁴Mg, ²⁷Al, ³⁹K, ⁵⁵Mn, ⁵⁶Fe, ⁶³Cu, ⁶⁶Zn). Standard solutions of each element were measured, and the minerals in the awamori were quantified using the regression formula of the calibration curve obtained. The limit of quantification was determined to be 0.001 mg/L, which corresponded to triple the noise count ratio of the blank test using 1 M nitric acid. All measurements were performed in triplicate, and quantitative values were converted to a 25 % alcohol concentration. 

Statistical analysis Statistical differences between groups were analyzed using the Wilcoxon-Mann-Whitney U test on GraphPad Prism Version 9.5.1 (GraphPad Software, Boston, MA, USA). Associations between groups were evaluated using Spearman’s correlation test.

Results

Sotolon content in awamori The LC-MS/MS multiple reaction monitoring chromatograms confirmed the presence of sotolon in the awamori model solution, in which the authentic sotolon compound was added to the awamori sample (Fig. 1). Sotolon and the internal standard isovanillic acid were detected at Rt 4.82 and 10.90 min, respectively. Each sample was analyzed for 15 min, and no carryover was observed in the subsequent analyses. A calibration curve was obtained by adding the authentic compound sotolon to the awamori samples. The curve was obtained with a linear regression of y = 2.2902x + 0.0656 and a good correlation coefficient of r = 0.9968. However, inter-day differences were observed in the LC-MS/MS response, as linearity was conserved. The inter-day variability of the linear inclination was within a range of 2.0–5.1. A standard addition calibration curve was obtained for each analysis day.

Fig. 1. Typical liquid chromatography with tandem mass spectrometry (LC-MS/MS) multiple reaction monitoring chromatograms:

(a) sotolon of model awamori,

(b) isovanillic acid (internal standard) of model awamori,

(c) sotolon of commercial awamori, and

(d) internal standard of commercial awamori.

The sotolon content of 40 commercial awamori samples ranged from 0 to 20.5 µg/L (Table 1). The awamori products with the highest sotolon content was aged inside an earthenware jar for more than 20 years. Conversely, 10 out of the 17 awamori with no quantified sotolon were young awamori samples that were aged for less than three years.

The awamori were compared by dividing them into groups according to their storage vessel (earthenware jars vs. non-earthenware vessels) and storage period (young vs. aged). Products that were kept in jars, even temporarily, such as No. 2, were classified as jar-stored awamori. Sotolon content in awamori stored in earthenware jars was significantly higher than that in awamori stored in non-earthenware vessels (7.01 vs. 1.01 µg/L, respectively; p < 0.0001) (Fig. 2a). Furthermore, when awamori with different storage periods (young vs. aged) were compared, sotolon content averaged 1.22 and 4.01 µg/L, respectively, with a significant difference (p < 0.01) (Fig. 2b). Commercial awamori are often blended by mixing genshu of different storage periods to accommodate desirable flavors. The correlation between sotolon content and relative storage years, which was calculated based on the blending ratio of awamori, was moderately positive (r = 0.617, p < 0.0001) (Supplementary Table S1).

Fig. 2. Sotolon content in awamori according to (a) aging container and (b) aging period. Means of the two groups are significantly different at p < 0.001 (***) or p < 0.0001(****) using Wilcoxon-Mann-Whitney U test.

Sotolon precursors in awamori Diacetyl and acetoin measurements were attempted by directly introducing awamori samples into the GC-MS instrument. However, quantification of the two substances was greatly affected by the ethanol concentration. Thus, SPE was required as a preprocessing procedure, which resulted in high-precision quantitative analysis. Diacetyl content in awamori ranged from 0.13 mg/L by No. 40 to 1.48 mg/L by No. 25, with an average of 0.58 ± 0.32 mg/L; whereas acetoin content varied from 0.24 mg/L by No. 32 to 4.70 mg/L by No. 24, with an average of 1.22 ± 1.09 mg/L (Table 2).

For acetaldehyde measurements, an activated carbon cartridge was used for preprocessing during GC-MS analysis. Acetaldehyde content in awamori ranged from 77.50 mg/L by No. 29 to 332.50 mg/L by No. 2, with an average concentration of 144.47 ± 55.22 mg/L. However, measuring 2-oxobutyric acid using GC-MS is challenging owing to its low volatility. Hence, HPLC was then used to measure 2-oxobutyric acid content in awamori, and it ranged from 0.16 mg/L by both No. 20 and No. 39 to 0.96 mg/L by No. 22, with an average value of 0.53 ± 0.21 mg/L. However, no significant differences in diacetyl, acetoin, and acetaldehyde concentration were observed, despite their aging periods, whereas aged awamori had significantly lower 2-oxobutyric acid levels than that in young awamori (0.45 vs. 0.61 mg/L, respectively; p < 0.05) (Fig. 3). Nevertheless, no correlations were observed between any pair of the four possible sotolon precursors or between sotolon and each of the precursors (Supplementary Table S1).

Fig. 3. Sotolon precursor content in awamori of different aging periods. (a) Diacetyl, (b) acetoin, (c) acetaldehyde, and (d) 2-oxobutyric acid. * Means of the two groups are significantly different at p < 0.05 using Wilcoxon-Mann-Whitney U test.

Table 2. Sotolon precursors content in awamori (mg/L).

No.	Diacetyl	Acetoin	Acetaldehyde	2-Oxobutyric acid
1	0.64	0.38	169.29	0.54
2	0.48	0.36	332.50	0.45
3	0.38	0.97	140.83	0.64
4	0.24	0.91	113.75	0.42
7	0.77	0.48	99.17	0.57
8	1.03	2.86	122.14	0.59
12	0.65	0.28	112.79	0.51
15	0.68	1.40	284.17	0.58
16	0.56	0.65	150.00	0.49
19	0.45	0.33	115.63	0.58
20	0.17	0.67	117.44	0.16
21	0.35	0.83	121.67	0.61
22	0.44	1.11	221.00	0.96
24	1.17	4.70	159.00	0.94
25	1.48	0.99	194.17	0.34
26	0.48	2.59	120.00	0.49
27	0.77	0.98	105.00	0.33
28	0.73	1.06	124.17	0.77
29	1.10	4.47	77.50	0.75
30	0.96	1.32	128.33	0.69
31	0.25	0.53	154.17	0.40
32	0.41	0.24	91.86	0.20
34	0.70	0.56	128.00	0.47
36	0.25	1.23	190.00	0.83
37	0.53	1.17	142.50	0.35
38	0.53	1.61	124.43	0.23
39	0.31	1.17	157.50	0.16
40	0.13	0.66	91.00	0.50
44	0.16	0.90	122.00	0.89
Mean	0.58 ± 0.32	1.22 ± 1.09	144.47 ± 55.22	0.53 ± 0.21

Mineral components of awamori Nine minerals (Al, Ca, Cu, Fe, K, Mg, Mn, Na, and Zn) from awamori were quantified. Nine awamori samples had a total mineral content of less than 1 mg/L, whereas the highest mineral content found was 25.71 mg/L (data not shown). The most predominant mineral in awamori was Na (0–23.82 mg/L, 0–99.81%), followed by K and Mg (0–4.82 and 0–1.68 mg/L, respectively). When the samples were compared according to storage vessels (earthenware jars vs. non-earthenware jars), different average mineral contents were observed (Table 3). When compared according to storage vessels and storage period, the Al content was significantly higher in earthenware jar (p < 0.01) and aged awamori (p < 0.01). Ca was below the limit of detection (0.001 mg/L) in 10 awamori liquors, and there was no significant difference between the storage vessels and storage periods. Moreover, Cu was significantly higher in awamori stored in earthenware jars (p < 0.05) and aged awamori (p < 0.001). Significantly higher Fe content was detected in the earthenware jar (p < 0.05) and aged groups (p < 0.05). Moreover, K, Na, and Mg were not significantly different between storage vessels or storage periods. The Mn concentration was the lowest among the nine minerals, with average concentrations ranging from 0.004–0.007 mg/L; however, it was significantly higher in the aged awamori group (p < 0.05). In terms of Zn content, no significant difference was found in the comparison of storage vessels or storage periods.

The Spearman's rank correlation coefficient was calculated for each mineral pair and between the sotolon and each mineral (Supplementary Table S2). Among the pairs, a strong positive correlation was found between Ca and Mg (r = 0.852; p < 0.001), and moderate positive correlations were observed between K and Mn (r = 0.598, p < 0.0001) and Cu and Mn (r = 0.596; p < 0.001). Additionally, sotolon showed the strongest positive correlation with Al (r = 0.743, p < 0.001), followed by Cu (r = 0.713, p < 0.001). In contrast, the aging period had the strongest positive correlation with Cu (r = 0.484, p < 0.01), followed by Al (r = 0.481, p < 0.01).

Discussion

Fukuda et al. (2016) investigated the effects of storage on volatile aroma content in awamori and found an increase in 28 volatile aroma compounds in ku-su (old-aged awamori). However, sotolon was not found. This may be a result of the difficulty in quantifying sotolon using GC-MS due to its low volatility. Therefore, in this study, sotolon was quantified using LC-MS/MS under mild conditions (Fig. 1, Table 1). As it was difficult to obtain a surrogate substance for sotolon as an internal standard for LC-MS/MS quantification, an alternative substance was identified. Several acidic compounds were tested, and isovanillic acid was selected as the internal standard because of its good column separation and ion strength. To the best of our knowledge, this study is the first to describe and successfully report the sotolon content in awamori. Sotolon is considered a characteristic volatile aroma compound in ku-su due to its sensitive threshold of 2 µg/L in model wine (Pons et al., 2010) and its concentration increase over aging. When compared to the conventional storage method using stainless-steel vessels, the aging process where genshu is stored in earthenware jars possibly contributes to the production of the characteristic volatile aroma compounds in awamori. The sotolon content of Vin jaune and Tokai wines is 120–268 and 80–140 µg/L, respectively, whereas that of old sake ranges from 140–430 µg/L (Guichard et al., 1993). Accordingly, the sotolon content of these brewed liquors is much higher than that of awamori, a distilled liquor. The concentration of sotolon in wine correlated with storage periods but had no relationship with the oak barrel used for wine storage (Cutzach et al., 1999). Similarly, the awamori sotolon content in this study had a moderately positive correlation with the aging period (r = 0.617), which can be attributed to the use of the same jars for decades because of the limited number of earthenware jars (Supplementary S1). Additionally, a distinct awamori flavor property known as koshu-kou (the aged awamori odor produced during the aging process), along with the overall odor remaining in the earthenware jar, is often used to determine the quality of the awamori flavor. Because sotolon content is related to the aging period and container, and the compound emits desirable sweet, honey-, and caramel-like aromas, it is postulated to be one of the important volatile aroma compounds that contribute to koshu-kou and the overall odors that remain in earthenware jars. Therefore, further research on sotolon content and flavor characteristics of aged awamori is necessary.

Table 3. Mean mineral content in awamori (mg/L).

Container/aging type	Al	Ca	Cu	Fe	K	Na	Mg	Mn	Zn
Earthenware (n = 12)	0.016 ± 0.013	0.086 ± 0.093	0.199 ± 0.193	0.020 ± 0.026	0.070 ± 0.134	3.949 ± 4.448	0.313 ± 0.304	0.007 ± 0.012	0.024 ± 0.023
Non-earthenware (n = 30)	0.003 ± 0.004	0.124 ± 0.202	0.068 ± 0.071	0.007 ± 0.011	0.691 ± 1.248	5.227 ± 6.529	0.372 ± 0.471	0.004 ± 0.005	0.013 ± 0.018
*p value*	0.0089**	0.2877	0.0232*	0.0333*	0.6415	0.9017	0.5225	0.1513	0.0730
Young (n = 17)	0.003 ± 0.004	0.171 ± 0.234	0.045 ± 0.086	0.005 ± 0.007	0.657 ± 1.137	6.798 ± 7.781	0.481 ± 0.528	0.004 ± 0.006	0.012 ± 0.017
Aged (n = 25)	0.010 ± 0.011	0.074 ± 0.112	0.146 ± 0.144	0.015 ± 0.022	0.416 ± 1.052	3.545 ± 3.970	0.270 ± 0.323	0.006 ± 0.009	0.019 ± 0.022
*p value*	0.0068**	0.1837	0.0002***	0.0327*	0.1626	0.2852	0.6434	0.0018*	0.1819

The significance value is donated by the asterisk mark as follows: * (p < 0.05), ** (p < 0.01), *** (p < 0.001).

The possible precursors of sotolon in awamori were selected based on studies on sotolon generation in wine. Four compounds (acetaldehyde, acetoin, 2-oxobutyric acid, and diacetyl), which were also determined in awamori, were evaluated and quantified (Table 2). The four compounds were present in the following order of average concentrations: acetaldehyde >> acetoin > diacetyl = 2-oxobutyric acid. Acetaldehyde content decreases with age in Chinese xiaoqu baijiu (Sun et al., 2022) and Chinese fenjiu (Ma et al., 2014). The trend of the sotolon precursors for different aging periods (Fig. 3) is similar to that reported by Fukuda et al. (2016). Moreover, this study is the first to evaluate acetoin and 2-oxobutyric acid content in awamori. Although the acetoin content varied among the samples without changing with aging, 2-oxobutyric acid was significantly decreased by aging (p < 0.05). Sotolon in wine is considered a product of aldol condensation between acetaldehyde and 2-oxobutyric acid, which is oxidatively decomposed from ascorbic acid by the yeast fungus. Therefore, 2-oxobutyric acid is exclusively important in sotolon generation (Pons et al., 2010). Although awamori generally contains lower amounts of 2-oxobutyric acid and acetaldehyde than that in wine (Table 2, Pons et al., 2010), these compounds can be considered as sotolon precursors. Furthermore, the Spearman correlation between each pair of possible sotolon precursors showed the largest r-value of 0.330 between diacetyl and acetoin (Supplementary Table S1). However, no correlations between sotolon and each possible precursor were derived (r = −0.107 to 0.142). This indicates the presence of variable factors other than storage vessels and periods, which might be a result of the different production methods used in commercial Okinawan awamori. Moreover, the thermal degradation of Maillard intermediates, along with the concentration, time, and temperature of dissolved oxygen during awamori storage, could be important factors (Pons et al., 2010). Therefore, further research is required to clarify the involvement of these substances in sotolon generation using an awamori model solution.

Meanwhile, the mineral content of awamori products were measured to investigate the transfer of minerals from jars to awamori (Table 3). Because awamori is a distilled liquor, awamori right after distillation hardly contains minerals. The presence of minerals may be a result of their elution during storage or the addition of tap or groundwater before shipping. Twelve of the 42 awamori products had Na concentrations of 1 mg/L or less and trace amounts of other minerals, therefore ion-exchanged water might be used for these samples. All awamori with K concentrations of 2 mg/L or higher were oak barrel-stored awamori. Moreover, Al, Fe, and Cu were significantly increased in earthenware-jar-stored awamori (p < 0.05, Mann-Whitney U test). This can be explained by the elution of Cu, Zn, Fe, and Mn from the earthenware jars and the addition of Ca and Mg from tap water. Tamaki et al. (1986) also found differences in the Fe, Cu, Ca, Mn, Zn, Mg, K, and Na content between young awamori vs. old awamori stored in earthenware jars. However, neither the number of years the jars had been used in the awamori production nor the correlation between jar age and mineral content are known. Similar results were obtained for Fe³⁺ and Cu²⁺. Furthermore, metal ions (Fe³⁺, Cu²⁺, Mn²⁺, and Zn²⁺) present in distilled spirits act as catalysts for oxidation reactions based on their concentration levels, resulting in the Fenton reaction that produces hydroxyl radicals in the spirits (Scholtes et al., 2015).

Thus, it was inferred that the storage of awamori in earthenware jars is involved in the radical reaction caused by the elution of Fe^3+, Mn²⁺, Zn²⁺, and Cu²⁺ from the jar, which catalyzes oxidation reactions and produces sotolon (Supplementary Table S2). A positive correlation (r = 0.713) was found between Cu and sotolon content; Cu was also found to have the strongest correlation (r = 0.484) between the aging periods among the nine minerals evaluated. Therefore, the elution of Cu ions from earthenware jars during the aging period may be one of the greatest factors in the generation of sotolon. Even though Fe is a well-known mineral that potently catalyzes oxidation, its content in awamori stored in earthenware jars was notably scarce compared with that of Cu (Fe vs. Cu; 0.020 vs. 0.199 mg/L, respectively). Moreover, Fe was less related to sotolon content (r = 0.310) than Cu was. Hence, Cu that was eluted from the earthenware jars was the most important mineral for sotolon generation in awamori. This outcome is in agreement with Scholtes et al. (2015) that described the importance of this mineral in sotolon generation in wine. Additionally, the report indicated that the Fenton reaction might have an indirect role in the production of sotolon. Therefore, the presence of minerals and other key factors in sotolon generation in awamori, such as the concentration of dissolved oxygen involved in the Fenton reaction, should be clarified in future research.

Sotolon was quantified in commercial awamori, and its relationship with aging period and vessel storage was investigated. Sotolon, sotolon precursors, and nine minerals were determined in 44 commercial awamori products of different aging periods and vessels. Sotolon in 40 awamori was detected at a maximum concentration of 20.5 µg/L. This was significantly higher in earthenware-jar-stored awamori and correlated with aging period. The possible precursors of sotolon in distilled awamori products were estimated to be diacetyl, acetoin, acetaldehyde, and 2-oxobutylic acid. The concentrations of diacetyl, acetoin, acetaldehyde, and 2-oxobutylic acid were 0.13–1.48, 0.24–4.70, 77.50–332.50, and 0.16–0.96 mg/L, respectively. No correlations were observed between the four compounds and sotolon. Further studies using modeled awamori are required to clarify the precursors of sotolon in awamori. Al and Cu contents were significantly higher in earthenware-jar-stored awamori than in non-earthenware-jar-stored awamori. The concentrations of Al, Cu, Fe, and Mn were significantly higher in awamori aged for longer periods. These results suggest that sotolon in awamori is a product of the oxidation and polymerization of its precursors present in genshu during aging in earthenware jars, owing to the catalytic effects of Cu, Fe, and Mn eluded from the jars. The study provides the first quantitative determination of sotolon, its precursors, and minerals in awamori, emphasizing the need for further research on modeled awamori and specific minerals to elucidate its generation mechanism.

Acknowledgements This work was supported by the Okinawa Innovation Ecosystem Collaborative Research Promotion Project from Okinawa Prefecture, Japan. LC-MS-MS and ICP-MS analyses were performed at the Center for Research Advancement and Collaboration, University of the Ryukyus.

Conflict of interest There are no conflicts of interest to declare.

Supplementary Table S1. Spearman’s correlation coefficient between sotolon, sotolon precursors, and aging period of awamori

Trait	Acetoin	Acetaldehyde	2-Oxobutyric acid	Sotolon	Aging period
Diacetyl	0.330	0.092	0.171	0.137	-0.086
Acetoin		0.210	0.300	-0.107	-0.255
Acetaldehyde			0.095	0.142	-0.289
2-Oxobutyric acid				-0.101	-0.363
Sotolon					0.617

Supplementary Table S2. Spearman’s correlation coefficient between minerals, sotolon, and aging period of awamori

Trait	Ca	Cu	Fe	K	Na	Mg	Mn	Zn	Sotolon	Aging period
Al	-0.013	0.502	0.154	-0.047	-0.248	0.100	0.429	0.205	0.743	0.481
Ca		-0.107	0.283	0.536	0.538	0.852	0.290	0.311	0.011	-0.130
Cu			0.459	0.087	-0.140	0.058	0.596	0.536	0.713	0.484
Fe				0.235	0.084	0.387	0.473	0.559	0.310	0.335
K					0.332	0.518	0.598	0.146	0.036	-0.282
Na						0.470	0.077	0.359	-0.154	-0.269
Mg							0.348	0.370	0.074	-0.109
Mn								0.332	0.511	0.268
Zn									0.499	0.322

References

• Boerzhijin, S., Isogai, A., and Mukai, N. (2023). Impact of storage conditions on the volatile aroma compounds of sake. J. Food Compos. Anal., 121, 105351. doi:10.1016/j.jfca.2023.105351
• Cutzach, I., Chatonnet, P., and Dubourdieu, D. (1999). Study of the formation mechanism of some volatile compounds during the aging of sweet fortified wines. J. Agric. Food Chem., 47, 2837–2846. doi:10.1021/jf981224s
• Fukuda, H., Han, J., and Yamada, O. (2016). Characteristics of the volatile composition of commercially available awamori. J. Brew. Soc. Japan (Nihon Jozo Kyoukaishi), 111, 261–270. (in Japanese) doi:10.6013/jbrewsocjapan.111.261
• Guichard, E., Pham, T.T., and Etievant, P. (1993). Quantitative determination of sotolon in wines by high-performance liquid chromatography. Chromatographia, 37, 539–542. doi:10.1007/BF02275793
• Isogai, A., Utsunomiya, H., and Iwata, H. (2004). Changes in the concentrations of sotolon and furfural during the maturation of sake. J. Brew. Soc. Japan (Nihon Jozo Kyoukaishi), 99, 374–380. (in Japanese) doi:10.6013/jbrewsocjapan1988.99.374
• Isogai, A., Utsunomiya, H., Kanda, R., and Iwata, H. (2005). Changes in the aroma compounds of sake during aging. J. Agric. Food Chem., 53, 4118–4123. doi:10.1021/jf047933p
• Ma, Y., Qiao, W.W., Chen, T., Du, X., Zhai X., and Zhang, S. (2014). Variations in physicochemical properties of Chinese fenjiu during storage and high-gravity technology of liquor aging. Int. J. Food Prop., 17, 923–936.doi:10.1080/10942912.2012.678536
• Milheiro, J., Vilamarim, R., Filipe-Ribeiro, L., Cosme, F., and Nunes, F.M. (2021). An accurate single step LLE method using keeper solvent for quantification of trace amounts of sotolon in port and white table wines by HPLC-DAD. Food Chem., 350, 129268. doi:10.1016/j.foodchem.2021.129268
• Pareira, V., Leça, J.M., Gaspar, J.M., Pareira, A.C., and Marques, J.C. (2018). Rapid determination of sotolon in fortified wines using a miniaturized liquid-liquid extraction followed by LC-MS/MS analysis. J. Anal. Methods Chem., 2018, 4393040. doi:10.1155/2018/4393040
• Pons, A., Lavigne, V., Landais, Y., Darriet, P., and Dubourdieu, D. (2008). Distribution and organoleptic impact of sotolon enantiomers in dry white wines. J. Agric. Food Chem., 56, 1606–1610. doi:10.1021/jf072337r
• Pons, A., Lavigne, V., Landais, Y., Darriet, P., and Dubourdieu, D. (2010). Identification of a sotolon pathway in dry white wines. J. Agric. Food Chem., 58, 7273–7279. doi:10.1021/jf100150q
• Scholtes, C., Nizet, S., and Collin, S. (2015). How sotolon can impart a madeira off-flavor to aged beers. J. Agric. Food Chem., 63, 2886–2892. doi:10.1021/jf505953u
• Singh, B.K., Szamosi, I., and Shaner, D. (1993). A high-performance liquid chromatography assay for threonine/serine dehydratase. Anal. Biochem., 208, 260–263. doi:10.1006/abio.1993.1043
• Sun, X., Qian, Q., Xiong, Y., Xie, Q., Yue, X., Liu, J., Wei, S., and Yang Q. (2022). Characterization of the key aroma compounds in aged Chinese Xiaoqu Baiju by means of the sensomics approach. Food Chem., 384, 132452. doi: 10.1016/j.foodchem.2022.132452
• Tamaki, T., Takamiya, Y., Miyagi, T., and Nishiya T. (1986). Changes in ester compounds and higher alcohols of awamori during aging. J. Ferment. Technol., 64, 17–24. doi:10.1016/0385-6380(86)90052-X
• Zhao, P., Gao, J., Qian, M., and Li, H. (2017). Characterization of the key aroma compounds in Chinese Syrah wine by gas chromatography-olfactometry-mass spectrometry and aroma reconstitution studies. Molecules, 22, 1045. doi:10.3390/molecules22071045