Multiple roles of lactic acid bacteria microflora in the formation of marker flavour compounds in traditional chinese paocai

Abstract

Traditional paocai brine (PB), which is continuously propagated by back-slopping and contains multiple species of lactic acid bacteria (LAB), is critical for the flavour of paocai. But the flavour-related compounds of PB-paocai, and the relationships between the LAB communities in PB and the marker volatile organic compounds (VOCs) in paocai remain unclear. A metabolomics-based method was utilized to screen the VOCs marker(s) of PB paocai in situ, and the change of these markers in whole fermentation was monitored to reveal the role of LAB communities in marker compound formation in vitro. A total of 13 compounds were screened as discriminant volatile markers for PB-paocai via gas chromatography-mass spectrometry (GC-MS)-based multimarker profiling. The co-culture of LAB from PB significantly increased the amount of acetic acid by producing it directly, and improved the ethyl acetate and ethyl propanoate yields by interacting with other microbes. However, LAB decreased the amount of sulphides, including methanethiol, carbon disulphide, methyl thiolacetate, dimethyl disulphide and dimethyl trisulphide, in PB-paocai due to inhibiting the growth of other microbes. The major flavour features of traditional paocai originate from LAB. Such volatile markers related to LAB communities may serve to promote starter screening and fermentation optimisation to produce paocai-related foods with better sensory qualities. This study could provide a deeper understanding of the role of LAB in marker flavor compound formation and guidance for the industrial production of Chinese paocai.

---

Introduction

Paocai, a traditional Chinese fermented vegetable dish, is extensively consumed in Asia.¹ Chinese paocai has a long history that can be traced back to 1000 B.C., and the traditional paocai fermentation method is still used today. Vegetables are pretreated and soaked in paocai brine (PB), and left at room temperature (20–25 °C) for 6–10 days in pickle jars. PB, constantly propagated by back-slopping, comprises stable lactic acid bacteria (LAB) communities and large flavour compounds, and has been continuously obtained for a long time, even up to a century for some older brands. The flavour quality of traditional paocai depends, to a large extent, on the PB used. Immersing vegetables in salt brine (SB, 4–6% saline) is a basic and widespread method to make fermented vegetables, as well as to make paocai.^1,2 Although PB plays a critical role in producing high-quality paocai on both the domestic and industrial scales, more attention is paid to SB-paocai. Thus, the flavour features of traditional PB-paocai remain unclear.

LAB have been widely used as important bacteria for improving the flavour of fermented food, and are employed as a starter in fermentation to yield abundant volatile organic compounds (VOCs). Lactobacillus casei, as a starter, strongly affects the VOC profiles in sausages, including esters and alcohols.³ Sun and colleagues⁴ revealed through sensory analysis that the global aromatic intensity in May Duck cherry winemaking was enhanced by the introduction of Lactobacillus plantarum SGJ-24. In our previous study, LAB were considered as the main LAB microflora in PB, and these LAB served as a starter to produce the taste marker compounds for the improvement of taste properties in PB-paocai.⁵ Although several species of LAB have been successfully isolated from traditional PB, what effect of LAB on the marker volatile compounds of traditional PB-paocai, which is responsible for the odor properties of traditional PB-paocai, remains unclear. Thus, the effects of LAB from PB on the VOC properties of traditional PB-paocai remain unclear.

Metabolomics approaches are opening up new possibilities to study the volatilomes of fermented food, to screen microorganisms of interest and to profile microbial volatilomes.⁶ A metabolomics-based approach for data processing and analysis has recently been applied to analyse fermented food and bacteria volatilomes to identify the characteristic volatiles.^7,8 The potential of Leuconostoc spp. and Lactobacillus spp. in the production of aroma-giving compounds was evaluated by metabolic fingerprinting of volatiles to select microorganisms that contribute to diversifying the flavour of fermented dairy products.⁶ Although some previous studies have profiled the VOCs of normal salt brine (SB)-paocai, this may be of limited relevance to the effect of PB on traditional paocai.^9,10 Hence, a metabolomics-based approach is used in the present study to clarify the effect of PB on the volatilome of paocai and establish the relationship between the LAB from PB and the volatile markers in PB-paocai.

The major objective of this research was to screen the VOC marker(s) of PB-paocai in situ by a metabolomics-based method and reveal the relationship between these markers and the LAB communities from PB during paocai fermentation in vitro.

Material and methods

Paocai preparation

Four kinds of brine, namely PB, SB, SB plus mixed LAB (LABSB) and LABSB with sterilised vegetables (sterilised LABSB), were used to make paocai. The main LAB communities of paocai (L. plantarum SCYAPCZN11-1, L. buchneri SCYAPCZN11-4 and P. ethanoliduran SCYAPCZN11-8) were previously isolated from paocai⁵ and then co-inoculated in SB to prepare LABSB and sterilised LABSB. The paocai samples fermented with different brines were divided into three groups. The first group was used to identify the marker VOCs (PB-paocai vs. SB-paocai; all paocai samples were collected from Sichuan province). The second group was used to reveal the effect of LAB on those compounds (SB-paocai vs. LABSB-paocai). Finally, the third group was used to clarify the relationship between LAB and volatile markers, and included SB with unsterilised vegetables (SB), LABSB with unsterilised vegetables (LABSB) and LABSB with sterilised vegetables (sterilised LABSB). The LAB inoculum and paocai were prepared following a previous description.⁵ To select the marker flavour compounds, PB- and SB-paocai were sampled at 7 days. To reveal the effect of LAB on the volatile markers and the correlation between LAB and these marker compounds, SB-, LABSB- and sterilised LABSB-paocai were sampled at 0, 1, 2, 3, 5, and 7 days for further analysis. In this section, the radish was mixed and mashed with salt brine as unsterilized medium to make paocai. The sterilized radish medium was treated as following. Firstly, the radish juice and slag was separated with the gauze. And then the juice was sterilized according to the P. Filannino et al.,¹¹ meanwhile the radish slag was heated at 105 °C for 10 min. Finally, sterilized radish juice and slag were homogenized again as sterilized medium to make paocai.

Volatile analysis by headspace SPME-gas chromatography

The evaluation and quantification of the VOCs were performed by headspace solid-phase microextraction (HS-SPME). Five millilitres of paocai juice was kept at 20 °C for 10 min before analysis. Each sample was supplemented with heptanoic acid methyl ester (20 mg mL⁻¹; Sigma-Aldrich Milwaukee, WI) as an internal standard to ensure precise retention times for the volatile component peaks. For automated headspace SPME-GC, a DVB/CAR/PDMS fibre (Supelco, Bellefonte, PA, USA) and an AOC-5000 Plus autosampler (Shimadzu, Japan) were used. After an equilibration time of 30 min at 50 °C using the shaker (250 rpm), the fibre was exposed to the headspace for 30 min at 50 °C. Desorption was performed within 2 min in splitless mode at 240 °C. A GC (GC-2010 plus, Shimadzu, Japan) fitted with a quadrupole MS (GCMS-QP2010 Ultra, Shimadzu, Japan) using an Rtx-wax capillary column (30 m, 0.25 mm ID, 0.25 μm thickness) was used. The injection temperature was 240 °C and the transfer line and ion source were at 220 °C. Helium was used as the carrier gas at a constant linear velocity of 35.0 cm s⁻¹. The temperature program was the following: 40 °C (3 min), from 40 to 130 °C at 5 °C min⁻¹, held for 5 min at 130 °C, from 130 to 155 °C at 25 °C min⁻¹, from 155 to 220 °C at 5 °C min⁻¹ and held for 5 min at 220 °C. All samples were run in triplicate. The NIST 2011 standard mass spectral database was used to identify the VOCs based on the retention time and mass-spectral similarity match (more than 85%). Authentic standards were applied to confirm the marker VOCs of the assignments.^9,10

Spectral data processing

The mass spectra were evaluated using XCMS online following a previous description.⁵ To correct the MS response shift during the run, the spectra were normalised manually by adjusting the peak intensity against the heptanoic acid methyl ester internal standard.

Statistical analysis

Principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) were performed using SIMCA-P 12.0 (Umetrics AB, Umea, Sweden). A student's t-test and Duncan's multiple range test were performed using SPSS 16.0 (SPSS Inc., Chicago, IL) to determine the significance. P values of less than 0.05 were considered to be significant.

Results and discussion

In the preliminary experiment, preference on PB paocai compared with SB paocai was observed (data not shown). To investigate the variation and influence of PB on the flavour of Chinese paocai, metabolomics-based method was utilized to screen the VOCs marker(s) of PB paocai in situ.

Identification of significant volatile flavour compounds for PB-paocai

To obtain detailed information on the compounds contributing to the aroma of PB-paocai, a metabolomics approach was applied. XCMS online analysis was conducted to extract more than 700 ions from each sample. After preprocessing, the multivariate data matrix was subjected to multivariate analysis. According to PCA analysis (Fig. 1), paocai fermented by different brines were clearly distinct, indicating that the brines significantly affected the VOC profile of paocai. Subsequently, a PLS-DA model was established to screen the VOC markers that contributed to the data discrimination. The PLS-DA score plot showed that SB- and PB-paocai were clearly distinguishable, with the model parameters R2Y = 0.884, R2X = 0.344, and Q2 = 0.675 (Fig. 2A). The permutation result (n = 100; intercept of Q2 = −0.263) validated the stability and reliability of this PLS-DA model. Following the score plot, an S plot was constructed to identify the VOCs contributing to the discrimination (Fig. 2B). According to the S plot, 13 compounds with p values of less than 0.05 and VIP values larger than 1.0 were chosen as potential markers,¹² as listed in Table 1. The discriminant compounds were divided into five categories: acids, esters, sulphides, aldehydes and nitriles. Most of those compounds are considered important odorants that contribute to the aromas of fermented food.^13–16

Fig. 1 Principal component analysis (PCA) score plot of the VOC profile of paocai fermented by different brines, based on the volatilome data; triangles (▲), SB-paocai; open triangles (△), PB-paocai.

Fig. 2 Multiple pattern recognition of compounds in SB- and PB-paocai (A) PLS-DA score plot (R2Y = 0.884, R2X = 0.344, and Q2 = 0.675); triangles (▲), SB-paocai; open triangles (△), PB-paocai (B) PLS-DA S-plot. Each triangle in the S-plot represents an ion.

Table 1 Candidates for discriminant VOCs based on the PLS-DA model

Discriminant marker	Group	RT (min)	Trend	Odour descriptor^a	VIP
a Odour: http://pubchem.ncbi.nlm.nih.gov/, http://www.perflavory.com.
Methanethiol	Sulphides	1.70	Down	Vegetable oil, alliaceous, eggy, creamy with savoury nuances	1.72
Carbon disulphide		1.83	Down	Rubber, chokingly repulsive, cabbage	1.08
Methyl thiolacetate		5.13	Down	Sulphurous, eggy, cheese, dairy, vegetable, cabbage	2.94
Dimethyl disulphide		5.69	Down	Sulphur, onion, sweet corn, vegetable, cabbage, tomato, green radish	1.99
Dimethyl trisulphide		13.82	Down	Sulphurous, cooked onion, savoury, meaty	1.54
Ethyl acetate	Esters	2.65	Up	Etherial, fruity, sweet, grape and rum-like	2.02
Ethyl propanoate		3.55	Up	Sweet, etherial, rummy, grape, winey	1.82
Methyl salicylate		23.99	Up	Sweet, salicylate, root beer, wintergreen, aromatic, slightly phenolic and camphoreous	1.02
Acetic acid	Acids	15.47	Up	Sharp pungent sour vinegar	4.14
Propanoic acid		17.75	Up	Pungent, acidic and dairy-like	2.37
2,4-Heptadienal	Aldehydes	16.76	Up	Fatty, green, oily, aldehydic with a vegetative nuance	1.78
Benzaldehyde		17.39	Up	Almond, fruity, powdery, nutty and benzaldehyde-like	1.40
2-Butenenitrile	Nitriles	8.33	Up	Sharp-green	1.33

---

Acids. Acetic acid and propanoic acid were identified as aroma markers for PB-paocai. The concentrations of these acids in PB-paocai were 3.53 and 99.53 times higher than those in SB-paocai, respectively. These acids are considered important odorants that are responsible for odour notes in fermented food, and heavily endow fermented food with distinctive odours, such as pungent, sour, vinegar-like, sweaty, cheesy and rancid.^6,17 Based on their high concentration, these acids may heavily contribute to the distinct aroma and the overall volatile flavours of PB-paocai.

Esters. Three esters (ethyl acetate, ethyl propanoate and methyl salicylate) were selected as the VOC markers for discriminating between the two kinds of paocai. The level of ethyl acetate was increased 17.28 times in PB-paocai compared with SB-paocai. This compound is considered an important olfactory compound in fermented foods with fruity odours.⁶ Ethyl propanoate was another ester marker in PB-paocai, having a concentration 724.10 times higher in PB-paocai than that in SB-paocai. This compound has a distinctive odour redolent of pineapple or bay leaf.⁶ The level of methyl salicylate detected in PB-paocai was approximately 40 times higher than that in SB-paocai at the end of fermentation. Based on its higher concentration, methyl salicylate may partially contribute to the sweet and fragrant odour of PB-paocai.¹⁸ In general, the higher levels of these esters may improve the sensory quality of PB-paocai.

Sulphides. Five volatile sulphide compounds (VSCs), including methanethiol, carbon disulphide, methyl thiolacetate, dimethyl disulphide and dimethyl trisulphide, were selected as markers for discriminating between the two forms of paocai (PB- vs. SB-paocai). The concentrations of these compounds were 6.21, 7.51, 10.43, 1.05 and 1.33 times higher in SB-paocai than those in PB-paocai, respectively. VSCs play important roles in the aroma of foods and beverages, and have highly distinctive olfactory properties due to their super-low odour thresholds.^19,20 Although sulphides are considered important olfactory compounds for fermented food, the variation in these compounds can lead to either consumer acceptance or rejection.¹⁹ Therefore, the higher level of sulphides might result in the lower acceptability of SB-paocai.

Aldehydes. Two kinds of aldehydes were identified as discriminant VOCs on the basis of the PLS-DA model. 2,4-Heptadienal and benzaldehyde were approximately 20 and 17 times more concentrated, respectively, in PB-paocai than in SB-paocai. Their odours have been described as fatty, green, oily, and almond.^14,21

Nitriles. The concentration of 2-butenenitrile was approximately doubled in PB-paocai compared with SB-paocai. This compound can be identified by its sharp-green odour, and is produced by Cruciferaes.⁹

The species of these marker volatile compounds was similar with the previous studies, indicating that these compounds could contribute to the organoleptic properties of fermented vegetable.^10,22 Meanwhile, the results also showed that the content of these compounds was significantly in PB than SB-paocai. The significant differences between the levels of these VOCs may contribute to the overall olfactory properties of PB-paocai compared with SB-paocai. Previous studies have largely focused on identifying the VOCs in SB-paocai.^9,10,22 Rina Wu et al. revealed the changes in flavour during natural fermentation of suan-cai, which is a type of SB-paocai, and esters and aldehydes were in the greatest diversity and abundance, contributing most to the aroma of suan-cai.¹⁰ Hence, unbiased approaches for evaluating the traditional PB-paocai volatilomes in situ remain limited. Herein, the effect of PB on the VOC profile of paocai was investigated, and the target VOCs responsible for the flavour of traditional PB-paocai were screened by a metabolomics approach. Those results indicated that PB affects the VOC profile of paocai, and 13 compounds were identified as responsible for the flavour characteristics of PB-paocai.

Effects of lactic acid bacteria microflora from PB on discriminant VOCs

LAB play important roles in the production of numerous aroma-giving compounds that can improve the flavour of fermented food.^2,23,24 Previously, we isolated three species of LAB from PB and concluded that those strains were the core LAB microflora in PB.⁵ These LAB are used as a starter for traditional paocai fermentation and are considered essential for the quality of traditional paocai. In previous study, Zhao et al. have studied the relationship between LAB and taste marker compounds, but these compounds only included the no-volatile compounds. Although 13 VOCs were identified as markers responsible for the olfactory properties of PB-paocai, the effect of LAB from PB on the production of these compounds was not clear.

To investigate the effects of indigenous LAB communities in PB (L. plantarum, P. ethanoliduran, and L. buchneri) on the markers, two groups of paocai were prepared, including LABSB and SB, and changes in the olfactory markers were monitored during fermentation. On the basis of the PCA score plots acquired from the data sets of these markers (Fig. 3), SB-paocai became distinguishable from LABSB-paocai after 3 days of fermentation and remained so until the end of the fermentation, indicating that the indigenous LAB in PB affected the production and accumulation of markers during paocai fermentation. The inoculum up-regulated the final concentration of ethyl acetate, ethyl propanoate and acetic acid, and down-regulated the final concentration of methanethiol, dimethyl disulphide, carbon disulphide, methyl thiolacetate and dimethyl trisulphide (Fig. 4). The trend in the accumulation of most markers was consistent with the results above (for PB- vs. SB-paocai), indicating that LAB from PB may affect these markers of PB-paocai. However, the concentrations of 2,4-heptadienal and benzaldehyde detected in LABSB-paocai showed no significant difference compared with SB-paocai, suggesting that LABSB may not influence the final content of these compounds. Moreover, three markers (propanoic acid, 2-butenenitrile and methyl salicylate), usually originating from vegetables,^9,25 were not detected in either LABSB- or SB-paocai, indicating that these compounds may originate from PB, which is obtained from vegetable back-slopping. In general, LAB affected the production of some of the volatile markers, and these markers accounted for a large proportion of the VOCs in paocai. Thus, the major differences between PB- and SB-paocai stemmed from the LAB species in the PB.

Fig. 3 Score plots of PCA performed on the marker volatile compounds during paocai fermentation. The directions of the curved arrows indicate the routes of the fermentation time of paocai: (▲) LABSB-paocai and (△) SB-paocai.

Fig. 4 Mean contents of marker volatile compounds in SB- and SB-with-LAB-paocai sampled at the beginning and end of fermentation (0 day and 7 days). The y axis is the peak area normalised by the internal standard. Values are mean ± standard deviation (SD) from triplicate experiments: (*) p < 0.05, (**) p < 0.01, (***) p < 0.001 and (****) p < 0.0001.

Multiple roles of LAB in the formation of the flavour characteristics of PB-paocai

LAB can produce some flavour compounds directly, and can interact with other microbes to yield multiple compounds indirectly.^6,11,26,27 To clarify the detailed relationship between the markers and the LAB inoculum, three groups of paocai (unsterilised SB, unsterilised LABSB and sterilised LABSB) were prepared to monitor the changes during fermentation.

Direct pathway. LAB can directly generate acetic acid to influence the flavour of PB-paocai. The abundance of acetic acid was similar at the early fermentation stage in all of the groups, but increased more rapidly in LABSB-paocai than in the other two groups after 2 days (Fig. 5). Note that the level of acetic acid in LABSB-sterilised paocai was higher than in SB-paocai, but lower than in LABSB-paocai. These results suggest that LAB directly affected the acetic acid level in paocai, while other microbes originating from vegetables might have made additional contributions to the level of this acid.

Fig. 5 Changing levels of acid marker (acetic acid) in paocai fermented by different methods. The y axis is the peak area normalised by the internal standard. Values are the mean ± SD from triplicate experiments: (*) p < 0.05, (**) p < 0.01, (***) p < 0.001 and (****) p < 0.0001.

Indirect pathways. Additionally, while LAB can produce esters directly, they might mainly interact with other microbes to generate further ester markers in PB. The concentrations of ethyl acetate and ethyl propanoate increased rapidly after 3 days, and kept rising until the end of fermentation in all of the groups, except for ethyl propanoate in sterilised LABSB (Fig. 6). The levels of the two esters increased most rapidly in unsterilised LABSB, and were highest at the end of fermentation, but in sterilised LABSB they increased by less than in unsterilised SB. These results suggest that LAB not only contributed to the production of these compounds directly, but may also have interacted with other microbes, which express esterase and are present at the early stage of fermentation, to produce further esters by an indirect pathway (Fig. 7).²⁸

Fig. 6 Changing levels of ester markers in paocai fermented by different methods. The y axis is the peak area normalised by the internal standard. A and B represent ethyl acetate and ethyl propanoate. Values are the mean ± SD values from triplicate experiments: (*) p < 0.05, (**) p < 0.01, (***) p < 0.001 and (****) p < 0.0001.

Fig. 7 Changing levels of sulphide markers in paocai fermented by different methods. The y axis is the peak area normalised by the internal standard. A–E represent methanethiol, carbon disulphide, methyl thiolacetate, dimethyl disulphide and dimethyl trisulphide. Values are the mean ± SD values from triplicate experiments: (*) p < 0.05, (**) p < 0.01, (***) p < 0.001 and (****) p < 0.0001.

Finally, LAB may inhibit the growth of certain microbes to reduce the concentration of sulphides. Sulphides play an important role in flavour, but can negatively affect consumer acceptance.^19,29 Five sulphides were selected as markers to compare the different forms of paocai, as they had lower concentrations in PB-paocai than SB-paocai. The content of methanethiol increased rapidly in all of the groups after 1 day of fermentation, and kept rising slowly after 2 days. Moreover, the level of methyl thiolacetate increased more rapidly after 3 days in unsterilised SB than in the other two groups. Note that methyl thiolacetate was not detected in the LAB-sterilised paocai. Additionally, the concentration of dimethyl disulphide and dimethyl trisulphide increased initially and then continually decreased until the end. These compounds were more concentrated in unsterilised SB than in the other brines throughout the fermentation, while their levels did not significantly change in sterilised LABSB-paocai. Furthermore, the concentration of carbon disulphides decreased in all the groups during the main period of fermentation, and was highest in the unsterilised SB-paocai among the three groups at the end. All of these results indicate that LAB from PB can inhibit the enzyme(s) or microbe(s) that produce these sulphides, which contribute to an unpleasant aroma.

Acids, sulphides and esters contribute strongly to the aroma of fermented vegetables,^9,10 and were also found in large quantities in paocai, regardless of whether from SB or PB, in the current study. Inoculating the indigenous LAB during paocai fermentation was found to either up- or down-regulate most of the markers, and the trends were consistent with preliminary sensory experiment for PB-paocai compared with SB-paocai. Thus, these results indicate that the indigenous LAB strains in the PB contribute significantly to the major olfactory properties of PB-paocai. Moreover, LAB inoculation may affect these markers by either direct and/or indirect pathways. Acetic acid, ethyl acetate and ethyl propanoate, identified as markers, can be produced by LAB directly.⁶ Additionally, LAB may interact with other microorganisms to enhance the production of specific markers. For instance, the level of ethyl acetate was higher in LABSB than in unsterilised SB-paocai at the end of fermentation, because LAB can produce acetic acid and ethanol by heterolactic fermentation, which can then be used as precursors for ethyl acetate production. Some strains of yeast containing abundant esterase may in turn use these compounds to produce more esters. Conversely, LAB may inhibit certain enzymes or microbes that generate compounds with ‘off’ flavours. The sulphide markers were lower in unsterilised LABSB-paocai compared with unsterilised SB-paocai. Organic sulphur compounds can be generated by downstream biological or chemical pathways.²⁹ It has been demonstrated that yeasts can use sulphur-containing pesticides to form sulphur compounds such as carbon disulphides, which give rise to ‘off’ flavours.²⁹ Pseudomonas spp., which exist widely on vegetables, appear to be the major microorganisms responsible for the production of these compounds under aerobic conditions.^30,31 When these LAB were inoculated into SB, the inoculum rapidly occupied the ecological niche and became the dominant strain, providing a low pH and anaerobic environment. The growth of these microbes might be inhibited by the rapid decrease in pH and oxygen, reducing the level of these compounds at the end of fermentation. Overall, the LAB in PB play multiple roles to affect the formation of the major VOC markers in traditional PB-paocai, and the significantly different flavor qualities between SB- and PB-paocai mainly result from these bacteria.

Conclusion

This study investigated the olfactory differences between traditional PB-paocai and regular SB-paocai by screening for the VOC markers responsible for these olfactory features by metabolomics-based analysis. 13 markers were selected according to the metabolomic results, including acids, esters, sulphides, aldehydes and nitriles. More importantly, the indigenous LAB from PB affected the olfactory features through both direct and indirect pathways. These results may facilitate better starter screening and the optimisation of paocai fermentation to achieve higher quality and shorter fermentation times.

Conflicts of interest

The authors declare that there are no conflicts of interest.

Acknowledgements

This work was supported by National Basic Research Program of China (973 Program No. 2012CB720802), the Program for Changjiang Scholars and Innovative Research Team in University (IRT1249), Program of Collaborative innovation center of food safety and quality control in Jiangsu Province, and the 111 Project B07029.

References

1. T. Xiong, F. Peng, Y. Y. Liu, Y. J. Deng, X. Y. Wang and M. Y. Xie, LWT--Food Sci. Technol., 2014, 59, 713–717.
2. T. Xiong, Q. Q. Guan, S. H. Song, M. Y. Hao and M. Y. Xie, Food Control, 2012, 26, 178–181.
3. M. Sidira, P. Kandylis, M. Kanellaki and Y. Kourkoutas, Food Chem., 2015, 178, 201–207.
4. S. Y. Sun, H. S. Gong, W. L. Liu and C. W. Jin, Food Microbiol., 2016, 55, 16–24.
5. N. Zhao, C. Zhang, Q. Yang, Z. Guo, B. Yang, W. Lu, D. Li, F. Tian, X. Liu, H. Zhang and W. Chen, J. Agric. Food Chem., 2016, 64, 2415–2422.
6. T. Pogacic, M. B. Maillard, A. Leclerc, C. Herve, V. Chuat, F. Valence and A. Thierry, Appl. Microbiol. Biotechnol., 2016, 100, 2335–2346.
7. R. Beleggia, C. Platani, R. Papa, A. Di Chio, E. Barros, C. Mashaba, J. Wirth, A. Fammartino, C. Sautter, S. Conner, J. Rauscher, D. Stewart and L. Cattivelli, J. Agric. Food Chem., 2011, 59, 9366–9377.
8. H. J. Kim, K. J. Park and J. H. Lim, J. Agric. Food Chem., 2011, 59, 5707–5713.
9. D. Y. Zhao, J. Tang and X. L. Ding, LWT--Food Sci. Technol., 2007, 40, 439–447.
10. R. N. Wu, M. L. Yu, X. Y. Liu, L. S. Meng, Q. Q. Wang, Y. T. Xue, J. R. Wu and X. Q. Yue, Int. J. Food Microbiol., 2015, 211, 23–31.
11. P. Filannino, G. Cardinali, C. G. Rizzello, S. Buchin, M. De Angelis, M. Gobbetti and R. Di Cagno, Appl. Environ. Microbiol., 2014, 80, 2206–2215.
12. U. Jumhawan, S. P. Putri, Yusianto, E. Marwani, T. Bamba and E. Fukusaki, J. Agric. Food Chem., 2013, 61, 7994–8001.
13. J. M. Wang, D. L. Capone, K. L. Wilkinson and D. W. Jeffery, Food Chem., 2016, 196, 682–693.
14. J. E. Lee, Y. S. Hong and C. H. Lee, J. Agric. Food Chem., 2009, 57, 4810–4817.
15. V. Rojas, J. V. Gil, F. Pinaga and P. Manzanares, Int. J. Food Microbiol., 2003, 86, 181–188.
16. C. Montananri, E. Bargossi, A. Gardini, R. Lanciotti, R. Magnani, F. Gardini and G. Tabanelli, Food Chem., 2016, 192, 736–744.
17. P. Tinchan, Y. Lorjaroenphon, K. R. Cadwallader and S. Chaiseri, J. Food Sci., 2015, 80, C49–C54.
18. Z. H. Qin, X. L. Pang, D. Chen, H. Cheng, X. S. Hu and J. H. Wu, Food Res. Int., 2013, 53, 864–874.
19. M. I. Kinzurik, M. Herbst-Johnstone, R. C. Gardner and B. Fedrizzi, J. Agric. Food Chem., 2015, 63, 8017–8024.
20. B. Fedrizzi, F. Magno, D. Badocco, G. Nicolini and G. Versini, J. Agric. Food Chem., 2007, 55, 10880–10887.
21. G. Mosciano, M. Fasano, J. Cassidy, K. Connelly, P. Mazeiko, A. Montenegro, B. B. Allen, J. Michalski and S. Sadural, Organoleptic Characteristics of Flavor Materials, Allured, Carol Stream, IL, USA, 1993.
22. Y. Z. Yao, S. Y. Pan, G. Fan, L. Dong, J. N. Ren and Y. Zhu, LWT--Food Sci. Technol., 2015, 64, 528–535.
23. V. Plengvidhya, F. Breidt Jr, Z. Lu and H. P. Fleming, Appl. Environ. Microbiol., 2007, 73, 7697–7702.
24. D. T. Nguyen, K. Van Hoorde, M. Cnockaert, E. De Brandt, M. Aerts, L. Binh Thanh and P. Vandamme, Int. J. Food Microbiol., 2013, 163, 19–27.
25. S. Ghosh, J. Thomas, W. Y. Huang, Y. J. Xu and W. G. Jager, J. Phys. Chem. Lett., 2015, 6, 3126–3131.
26. B. E. Wolfe, J. E. Button, M. Santarelli and R. J. Dutton, Cell, 2014, 158, 422–433.
27. O. Erkus, V. C. de Jager, M. Spus, I. J. van Alen-Boerrigter, I. M. van Rijswijck, L. Hazelwood, P. W. Janssen, S. A. van Hijum, M. Kleerebezem and E. J. Smid, ISME J., 2013, 7, 2126–2136.
28. J. B. Hu, S. Gunathilake, Y. C. Chen and P. L. Urban, RSC Adv., 2014, 4, 28865–28870.
29. M. Mestres, O. Busto and J. Guasch, J. Chromatogr. A, 2000, 881, 569–581.
30. J. Zwielehner, M. Handschur, A. Michaelsen, S. Irez, M. Demel, E. B. Denner and A. G. Haslberger, Mol. Nutr. Food Res., 2008, 52, 614–623.
31. F. B. Whitfield, Int. J. Food Sci. Technol., 1998, 33, 31–51.

---