Volatile Organic Compounds (VOCs) Produced by Levilactobacillus brevis WLP672 Fermentation in Defined Media Supplemented with Different Amino Acids

Fermentation by lactic acid bacteria (LAB) is a promising approach to meet the increasing demand for meat or dairy plant-based analogues with realistic flavours. However, a detailed understanding of the impact of the substrate, fermentation conditions, and bacterial strains on the volatile organic compounds (VOCs) produced during fermentation is lacking. As a first step, the current study used a defined medium (DM) supplemented with the amino acids L-leucine (Leu), L-isoleucine (Ile), L-phenylalanine (Phe), L-threonine (Thr), L-methionine (Met), or L-glutamic acid (Glu) separately or combined to determine their impact on the VOCs produced by Levilactobacillus brevis WLP672 (LB672). VOCs were measured using headspace solid-phase microextraction (HS-SPME) gas chromatography–mass spectrometry (GC-MS). VOCs associated with the specific amino acids added included: benzaldehyde, phenylethyl alcohol, and benzyl alcohol with added Phe; methanethiol, methional, and dimethyl disulphide with added Met; 3-methyl butanol with added Leu; and 2-methyl butanol with added Ile. This research demonstrated that fermentation by LB672 of a DM supplemented with different amino acids separately or combined resulted in the formation of a range of dairy- and meat-related VOCs and provides information on how plant-based fermentations could be manipulated to generate desirable flavours.


Introduction
Plant-based foods have gained popularity as consumers choose to reduce their meat and dairy intake due to concerns about their health, the environment, and/or animal welfare [1][2][3][4].This desire has resulted in an increase in sales of meat or dairy analogues worldwide as consumers seek out alternative forms and flavours of products they are familiar with [5][6][7][8][9][10][11][12].A challenge with producing plant based-analogues is obtaining realistic meat-or dairy-like flavours.
Flavour is a complex sensory modality that encompasses volatile organic compounds (VOCs) sensed in the nose at the olfactory epithelium retronasally, non-volatile organic compounds sensed on the tongue (taste attributes: sweet, salt, sour, bitter, and umami), and chemesthetic responses (hot, spicy, and pungent) sensed in the oral cavity.Generally, studies on flavour focus on the analysis of VOCs owing to the importance of aroma/odour in overall flavour perception [13][14][15].VOCs are a low-molecular-weight (<400 Da) compound with a relatively high vapour pressure at room temperature, which means that they can be easily transferred into the gaseous phase [16] and subsequently to olfactory receptors [13,17].
In meat or dairy analogues, the addition of flavour compounds extracted from meat or dairy products is generally not acceptable.In addition, the chemical generation of flavour compounds can be environmentally unfriendly owing to the requirement to use solvents, and such processes lack selectivity, which leads to the formation of unwanted compounds, reduced process efficiency, and increased downstream costs [18].Although desirable flavours can be extracted directly from plants, this is only economically feasible for a small number of VOCs, as plants contain complex mixtures of VOCs [19].A promising route for the generation of flavour VOCs that imitate meat or dairy flavour compounds is via the microbial biosynthesis or fermentation of plant material [20].For example, lactic acid bacteria (LAB) can use plant substrates for energy and nutrition and produce a range of volatile secondary metabolites.In fermented foods, these volatile secondary metabolites are responsible for the production of flavour compounds or flavour precursors [21,22].The composition of the plant substrates and the LAB strains used have been reported to have the greatest impacts on the resulting fermentation flavours produced [23,24].However, as the substrate composition of plants can vary widely, understanding how best to generate targeted flavour VOCs through plant-based fermentation is challenging.
LAB are a fastidious microorganism that require a rich cultivation medium for growth, as the majority of them are auxotrophic for a wide range of amino acids and vitamins [25].Carbohydrates (simple sugars), protein (peptone, yeast extract, beef extract, or whey protein), minerals, vitamins, and buffering agents are common ingredients in LAB cultivation media [26].A rich cultivation medium, however, is not suitable for determining the role of substrates on VOC formation by LAB fermentation owing to the difficulty in determining which substrates and metabolic pathways are responsible for the VOCs detected.To reduce the complexity in the system, a defined medium with only sufficient nutrients to support LAB growth can be used [27][28][29][30][31][32].
Amino acids are important not only for the growth of LAB but also for the production of flavour compounds.Most amino acids do not have a direct impact on flavour, but they do contribute indirectly because they are precursors to key flavour compounds [33].Enzymes found in the LAB, such as deaminases, decarboxylases, transaminases (aminotransferases), and lyases, can convert amino acids in different ways.Amino acid transamination is the primary initiator of the conversion of amino acids to flavour compounds.Transamination of amino acids results in α-keto acid formation, which can subsequently be decarboxylated into aldehydes, which in turn can be dehydrogenated into alcohols or carboxylic acids by alcohol dehydrogenases and aldehyde dehydrogenases, respectively [34][35][36].
For LAB strains to be used in commercial production of plant-based flavours, the bacteria need to be readily available, food grade, and not very fastidious in terms of nutrient requirements or growth temperatures.An initial trial with Levilactobacillus brevis WLP672 (Lev.brevis WLP672) and Lactobacillus delbrueckii WLP677 strains identified the LAB brewing strain Lev.brevis WLP672 as a good candidate for use in fermentation trials.Lev.brevis WLP672 is an obligatory heterofermentative LAB that is used in the production of a wide range of fermented products worldwide.The bacterium uses the phosphoketolase pathway (PK) to ferment hexoses to produce a mixture of lactic acid, ethanol, acetic acid, CO 2 , and an array of volatile secondary metabolites [37].
To date, most studies on the amino acid-derived VOCs produced by yeasts [38][39][40][41][42], fungi [43], and LAB [44,45] have used complex media (natural or synthetic), with only a few studies using LAB in the defined medium [33,46].This is the first study to use headspace solid-phase microextraction (HS-SPME) gas chromatography-mass spectrometry (GC-MS) to provide a comprehensive analysis of VOCs produced by Lev.brevis WLP672 in response to the addition of single or combined amino acids in a defined medium.

Development of Defined Medium
Based on past literature [27,[47][48][49][50][51][52] and a series of trials, a defined medium (DM) was developed (Table 1), which supported the growth of Lev.brevis WLP672 (thereafter referred to as LB672).In addition to the other components in the DM, growth of LB672 did not occur in the absence of sodium acetate, as previously reported [48].Sodium acetate is postulated to stimulate the growth of several LAB species, owing to it being both a buffering agent and an energy source [47,48].Sodium acetate concentrations of 0.1 and 1.2% were the lowest and highest concentrations previously used in nutrient-rich media designed to support the growth of LAB [47,49].In the current study, sodium acetate stimulated the growth of LB672 in the DM when glucose was present, but it did not enable growth (function as an energy source) when glucose was absent.LB672 also did not grow in the DM in the absence of peptone (enzymatic protein digest, Bacto peptone), even in the presence of individual amino acid.Lev.brevis strains have previously been reported to lack genes for amino acid biosynthesis [53,54].Bacto peptone was used as its composition was defined [55,56].Further, LB672 did not grow in the absence of the added vitamins or minerals.
To investigate the role of amino acid supplementation on the VOCs produced by LB672 during fermentation, the DM was supplemented with amino acids, either separately or in combination (Table 1).Note that amino acid supplementation in the DM resulted in an amino acid concentration 4 to 30 times higher than the amino acid concentration in the DM (from the peptone) [55,56].Almost all LAB require glutamic acid (Glu) because of their inability to synthesise its precursor α-ketoglutarate de novo, owing to their lack of a complete TCA cycle [25].This means aminotransferases, which initiate transamination reactions of amino acid, utilise Glu as the donor substrate of the amino groups and convert them into α-ketoglutarate [36].To ensure that sufficient Glu was present to enable transamination reactions, Glu was added to all media except the original DM (DM0.1 and DM1.2, Table 1).

Physiochemical Properties
The pH of the medium impacts on the growth of bacteria, and during LAB fermentation, the pH of the medium can decrease owing to the production of lactic acid [57].In the current study, pH values decreased over the 16 days of fermentation (Table 2) from 5.2 to 4.6 in media containing 1.2% acetate and from 6.6 to 6.1 in media containing 0.1% acetate.During fermentation, the turbidity (OD 600 ) of the media increased over time (Table 2) by about 0.5 to 0.7 units in media containing 1.2% acetate and in the range of 0.2 to 0.4 in media containing 0.1% acetate.It is obvious that there were differences obtained in OD 600 between the two acetate concentrations used in the DM.However, due to the variation in the initial pH of the DM, it is not possible to confirm the impact of acetate on the growth.Further, after 16 days of fermentation in either DM0.1 or DM1.2 (Table 2), OD 600 values were higher in DM containing 0.1 and 1.2% acetate compared to in DM supplemented with amino acids.The reason for the decreased LB672 growth in the presence of amino acids is unclear.However, as the addition of amino acids decreased the initial pH of the medium, it is speculated that this could be one of the reasons for the decreased growth (lower final OD 600 values).1).

Volatile Organic Compounds (VOCs) after Fermentation
A total of 49 VOCs were detected after 16 days of LB672 fermentation, which were attributed to supplementation of the DM with different amino acids separately or combined.The VOCs detected included alcohols (19 compounds), acids (9), esters (6), sulphur compounds (5), ketones (4), aldehydes (1), and unknown compounds (5) (Table 3).In order to distinguish differences across different medium compositions in the relative abundance of the VOCs detected, hierarchical clustering analysis and heatmap visualisation were carried out (Figures 1 and 2).In the dendrogram, VOCs on nearby branches tend to show positive correlations, whereas distant branches tend to show negative correlations.S1).The VOCs represented in the heatmap are numbered according to the peak numbers (Table S1).S2).S2).
Since the heat map analysis highlighted differences between the VOC profile across different medium compositions, the specific amino acid-derived VOCs produced by LB672 across different medium compositions are discussed in the following sections separately.

Phe-Derived VOCs
Phe, an aromatic amino acid, can be converted by LAB into phenyl pyruvate via the action of aromatic aminotransferases (ArAT) in the presence of α-ketoglutarate.The ArAT is active on all aromatic amino acids (tryptophan (Try), Phe, and tyrosine (Tyr)) and also on Leu and Met [36].Further, phenyl pyruvate can subsequently be converted into benzeneacetaldehyde (phenylacetaldehyde), phenyl acetic acid, phenylethyl alcohol (phenyl ethanol), benzyl alcohol (phenyl methanol), and benzaldehyde by LAB via enzymatic, non-enzymatic, and unknown mechanisms [36,58].
Benzaldehyde, which has characteristic almond and burnt sugar notes [57], is produced chemically (non-enzymatic reactions) from phenyl pyruvate in the presence of oxygen and manganese [34,58].DM with added Phe and 0.1% acetate had the highest benzaldehyde peak area (Figure 3A, medium DMP).Benzaldehyde was detected in all media possibly due to the presence of peptone, which is speculated to be the reason why many of the compounds discussed in the following sections were also detected in many of the ferments at low levels.
duced chemically (non-enzymatic reactions) from phenyl pyruvate in the presence of oxygen and manganese [34,58].DM with added Phe and 0.1% acetate had the highest benzaldehyde peak area (Figure 3A, medium DMP).Benzaldehyde was detected in all media possibly due to the presence of peptone, which is speculated to be the reason why many of the compounds discussed in the following sections were also detected in many of the ferments at low levels.1) at 0.1% and 1.2% acetate.Values are presented as mean ± standard error (n = 6).Different letters represent significant difference between the different medium compositions according to Tukey's test at p < 0.05.
Phenylethyl alcohol, which has a characteristic floral note [59], can be produced from phenyl pyruvate, which is generated by transamination of Phe.Phenylpyruvate is then decarboxylated to phenylacetaldehyde (benzeneacetaldehyde) via the action of decarboxylase.Further, it can be dehydrogenated to phenylethyl alcohol via the action of the enzyme alcohol dehydrogenase (AlcDH) [36].The highest peak area for phenylethyl alcohol was obtained in media with added Phe, where a significantly (p < 0.05) higher peak area Figure 3. Normalized peak area of benzaldehyde (3A), phenylethyl alcohol (3B), and benzyl alcohol (3C) across different medium compositions (see Table 1) at 0.1% duced chemically (non-enzymatic reactions) from phenyl pyruvate in the presence of oxygen and manganese [34,58].DM with added Phe and 0.1% acetate had the highest benzaldehyde peak area (Figure 3A, medium DMP).Benzaldehyde was detected in all media possibly due to the presence of peptone, which is speculated to be the reason why many of the compounds discussed in the following sections were also detected in many of the ferments at low levels.1) at 0.1% and 1.2% acetate.Values are presented as mean ± standard error (n = 6).Different letters represent significant difference between the different medium compositions according to Tukey's test at p < 0.05.
Phenylethyl alcohol, which has a characteristic floral note [59], can be produced from phenyl pyruvate, which is generated by transamination of Phe.Phenylpyruvate is then decarboxylated to phenylacetaldehyde (benzeneacetaldehyde) via the action of decarboxylase.Further, it can be dehydrogenated to phenylethyl alcohol via the action of the enzyme alcohol dehydrogenase (AlcDH) [36].The highest peak area for phenylethyl alcohol was obtained in media with added Phe, where a significantly (p < 0.05) higher peak area and 1.2% duced chemically (non-enzymatic reactions) from phenyl pyruvate in the presence of oxygen and manganese [34,58].DM with added Phe and 0.1% acetate had the highest benzaldehyde peak area (Figure 3A, medium DMP).Benzaldehyde was detected in all media possibly due to the presence of peptone, which is speculated to be the reason why many of the compounds discussed in the following sections were also detected in many of the ferments at low levels.1) at 0.1% and 1.2% acetate.Values are presented as mean ± standard error (n = 6).Different letters represent significant difference between the different medium compositions according to Tukey's test at p < 0.05.
Phenylethyl alcohol, which has a characteristic floral note [59], can be produced from phenyl pyruvate, which is generated by transamination of Phe.Phenylpyruvate is then decarboxylated to phenylacetaldehyde (benzeneacetaldehyde) via the action of decarboxylase.Further, it can be dehydrogenated to phenylethyl alcohol via the action of the enzyme alcohol dehydrogenase (AlcDH) [36].The highest peak area for phenylethyl alcohol was obtained in media with added Phe, where a significantly (p < 0.05) higher peak area acetate.Values are presented as mean ± standard error (n = 6).Different letters represent significant difference between the different medium compositions according to Tukey's test at p < 0.05.
Phenylethyl alcohol, which has a characteristic floral note [59], can be produced from phenyl pyruvate, which is generated by transamination of Phe.Phenylpyruvate is then decarboxylated to phenylacetaldehyde (benzeneacetaldehyde) via the action of decarboxylase.Further, it can be dehydrogenated to phenylethyl alcohol via the action of the enzyme alcohol dehydrogenase (AlcDH) [36].The highest peak area for phenylethyl alcohol was obtained in media with added Phe, where a significantly (p < 0.05) higher peak area was obtained with 1.2% acetate compared to 0.1% acetate (Figure 3B, medium DMP).Phe addition has previously been reported to increase phenylethyl alcohol production during the fermentation of lychee wine using Saccharomyces cerevisiae var.cerevisiae MERIT.ferm[38], in synthetic grape must fermentation by commercial wine yeast strains [39], in papaya juice fermented by Williopsis saturnus var.mrakii [41], and in synthetic grape juice fermented by Saccharomyces cerevisiae var.bayanus [42].The addition of an amino acid mixture (valine (Val), Leu, Ile, and Phe) to soy (tofu) whey has also been reported to increase the amount of phenylethyl alcohol following Torulaspora delbrueckii Biodiva fermentation [40].
Benzyl alcohol, which imparts floral notes, can be produced from benzaldehyde via oxidative and non-oxidative pathways [60].It was detected at the highest peak area in media supplemented with Phe (Figure 3C, medium DMP), where a significantly (p < 0.05) higher amount was obtained with 0.1% acetate compared to 1.2% acetate.
A genomic study by Liu et al. [53] reported that Lev.Brevis ATCC 367 lacked the gene araT encoding ArAT (putative) enzyme, which initiates the transamination reaction that converts Phe into phenyl pyruvate.However, in the current study, LB672 produced benzaldehyde, phenylethyl alcohol, and benzyl alcohol presumably via the intermediate phenyl pyruvate.Yvon and Rijnen [61] reported that the presence of aspartate aminotransferase (AspAT) in Brevibacterium linens is responsible for aspartate (Asp) transamination and is also active on aromatic amino acids.The gene aspAT encoding for AspAT enzyme has been detected in several Lev.Brevis strains, including Lev.Brevis ATCC 367 [53] and Lev.Brevis CGMCC 1306 [62].Hence, it seems more likely that, in the current study, AspAT in LB672 carried out the transamination reaction of Phe rather than the ArAT enzyme.
In the current study, the highest peak area for methanethiol was in DM at 0.1% acetate supplemented with the Aamix (Figure 4A, medium DMAa), which contained Met along with other amino acids.The highest peak area for methional was in medium supplemented with Met (Figure 4B, medium DMM), where a significantly (p < 0.05) higher peak area was obtained for 1.2% acetate compared to 0.1% acetate.For dimethyl disulphide, the highest peak area was observed in the medium with Aamix added (Figure 4C, medium DMAa) at 1.2% acetate, suggesting that it was converted from methanethiol.
According to a study by Liu et al. [53], the genes cblA and cglA, which encode CBL and CGL lyases, are present in Lev.Brevis ATCC 367, and catalyse the demethiolation reaction of Met.Lyases, play an important role in the biosynthesis of sulphur-containing amino acids but not a major role in Met catabolism by LAB [36].Therefore, aminotransferases ArAT or BcAT, which might be present in LB672 and involved in the transamination reaction, would appear to be the major source of the degradation products.As the peak area for methanethiol was high and the peak area for methional was low, the results suggest that either methional was produced through transamination (either by ArAT or by BcAT) and converted into methanethiol or that methanethiol was produced directly from Met via the activity of CBL/CGL.
Benzaldehyde, which has characteristic almond and burnt sugar notes [57], is produced chemically (non-enzymatic reactions) from phenyl pyruvate in the presence of oxygen and manganese [34,58].DM with added Phe and 0.1% acetate had the highest benzaldehyde peak area (Figure 3A, medium DMP).Benzaldehyde was detected in all media possibly due to the presence of peptone, which is speculated to be the reason why many of the compounds discussed in the following sections were also detected in many of the ferments at low levels.1) at 0.1% and 1.2% acetate.Values are presented as mean ± standard error (n = 6).Different letters represent significant difference between the different medium compositions according to Tukey's test at p < 0.05.
Phenylethyl alcohol, which has a characteristic floral note [59], can be produced from phenyl pyruvate, which is generated by transamination of Phe.Phenylpyruvate is then decarboxylated to phenylacetaldehyde (benzeneacetaldehyde) via the action of decarboxylase.Further, it can be dehydrogenated to phenylethyl alcohol via the action of the enzyme alcohol dehydrogenase (AlcDH) [36].The highest peak area for phenylethyl alcohol was obtained in media with added Phe, where a significantly (p < 0.05) higher peak area and 1.2% zeneacetaldehyde (phenylacetaldehyde), phenyl acetic acid, phenylethyl alcohol (phenyl ethanol), benzyl alcohol (phenyl methanol), and benzaldehyde by LAB via enzymatic, non-enzymatic, and unknown mechanisms [36,58].
Benzaldehyde, which has characteristic almond and burnt sugar notes [57], is produced chemically (non-enzymatic reactions) from phenyl pyruvate in the presence of oxygen and manganese [34,58].DM with added Phe and 0.1% acetate had the highest benzaldehyde peak area (Figure 3A, medium DMP).Benzaldehyde was detected in all media possibly due to the presence of peptone, which is speculated to be the reason why many of the compounds discussed in the following sections were also detected in many of the ferments at low levels.1) at 0.1% and 1.2% acetate.Values are presented as mean ± standard error (n = 6).Different letters represent significant difference between the different medium compositions according to Tukey's test at p < 0.05.
Phenylethyl alcohol, which has a characteristic floral note [59], can be produced from phenyl pyruvate, which is generated by transamination of Phe.Phenylpyruvate is then decarboxylated to phenylacetaldehyde (benzeneacetaldehyde) via the action of decarboxylase.Further, it can be dehydrogenated to phenylethyl alcohol via the action of the enzyme alcohol dehydrogenase (AlcDH) [36].The highest peak area for phenylethyl alcohol was obtained in media with added Phe, where a significantly (p < 0.05) higher peak area acetate.Values are presented as mean ± standard error (n = 6).Different letters represent significant difference between the different medium compositions according to Tukey's test at p < 0.05.
In the current study, the highest peak area for 2-methyl butanol was observed in DM supplemented with Ile (Figure 5A, DMI) at 0.1% acetate.Similarly, the highest peak area for 3-methyl butanol was obtained in DM supplemented with Leu (Figure 5B, DML) at 0.1% acetate.The addition of Leu has previously been reported to increase 3-methyl butanol production: in synthetic grape must fermented by commercial wine yeast strains [39], in synthetic cassava medium fermented by Ceratocystis fimbriata [43], and in papaya juice fermented by Williopsis saturnus var.mrakii [41].The addition of an amino acid mixture (Val, Leu, Ile, and Phe) to soy (tofu) whey has also been reported to increase the peak area of 3-methyl butanol fermented by Torulaspora delbrueckii Biodiva [40].Further, the addition of Ile has also been reported to increase 2-methyl butanol production during the fermentation of papaya juice by Williopsis saturnus var.mrakii [41].
The gene bcaT encodes the BcAT enzyme which initiates the transamination reaction on branched-chain amino acids, and the gene araT encodes the ArAT (putative) enzyme that initiates the transamination reaction on Leu, were both absent in a genome study of Lev.Brevis ATCC 367 [53].In the current study, LB672 fermentation resulted in the production of 2-and 3-methyl butanol from Ile and Leu, respectively.This result suggests that LB672 possesses BcAT or ArAT enzymes, which catalysed the transamination of Ile and Leu because 2-and 3-methyl butanol could not have been produced without a transamination step.
in synthetic cassava medium fermented by Ceratocystis fimbriata [43], and in papaya juice fermented by Williopsis saturnus var.mrakii [41].The addition of an amino acid mixture (Val, Leu, Ile, and Phe) to soy (tofu) whey has also been reported to increase the peak area of 3-methyl butanol fermented by Torulaspora delbrueckii Biodiva [40].Further, the addition of Ile has also been reported to increase 2-methyl butanol production during the fermentation of papaya juice by Williopsis saturnus var.mrakii [41].
The gene bcaT encodes the BcAT enzyme which initiates the transamination reaction on branched-chain amino acids, and the gene araT encodes the ArAT (putative) enzyme that initiates the transamination reaction on Leu, were both absent in a genome study of Lev.brevis ATCC 367 [53].In the current study, LB672 fermentation resulted in the production of 2-and 3-methyl butanol from Ile and Leu, respectively.This result suggests that LB672 possesses BcAT or ArAT enzymes, which catalysed the transamination of Ile and Leu because 2-and 3-methyl butanol could not have been produced without a transamination step.
Ethanol, which is a marker compound in fermentation studies, can be produced from sugars via an intermediate acetaldehyde using the PK pathway [21], from the degradation of Thr [64], or from acetate through acetyl-CoA [65].Overall, the ethanol peak area was higher in DM containing 1.2% acetate compared to 0.1% (Figure 6A).This may have been a result of the acetate being converted into ethanol.Interestingly, in DM supplemented with either Leu, Ile, or Phe at 0.1% acetate, the peak area for ethanol was significantly (p < 0.05) lower than in other media with either 0.1 or 1.2% acetate.This finding suggests that LB672 in the presence of added Leu, Ile, or Phe at the lower acetate concentrations is using a pathway other than an ethanol-producing pathway.
Acetic acid was detected in all media, with its concentration being significantly (p < 0.05) higher at 1.2% acetate-containing DM compared to 0.1% acetate-containing DM (Figure 6B).Though Lev. brevis strains have been shown to produce acetic acid in addition to lactic acid [66], owing to the fact that acetate was present in the DM, it is not possible to confirm that LB672 produced additional acetate.Acetic acid, however, is the building block for fatty acids consisting of an even number of carbons [32,67].In the current study, Figure 5. Normalized peak area of 2-methyl butanol (5A) and 3-methyl butanol (5B) across different medium compositions (see Table 1) at 0.1% Figure 3. Normalized peak area of benzaldehyde (3A), phenylethyl alcohol (3B), and benzyl alcohol (3C) across different medium compositions (see Table 1) at 0.1% and 1.2% acetate.Values are presented as mean ± standard error (n = 6).Different letters represent significant difference between the different medium compositions according to Tukey's test at p < 0.05.
Phenylethyl alcohol, which has a characteristic floral note [59], can be produced from phenyl pyruvate, which is generated by transamination of Phe.Phenylpyruvate is then decarboxylated to phenylacetaldehyde (benzeneacetaldehyde) via the action of decarboxylase.Further, it can be dehydrogenated to phenylethyl alcohol via the action of the enzyme alcohol dehydrogenase (AlcDH) [36].The highest peak area for phenylethyl alcohol was obtained in media with added Phe, where a significantly (p < 0.05) higher peak area and 1.2% Figure 3. Normalized peak area of benzaldehyde (3A), phenylethyl alcohol (3B), and benzyl alcohol (3C) across different medium compositions (see Table 1) at 0.1% and 1.2% acetate.Values are presented as mean ± standard error (n = 6).Different letters represent significant difference between the different medium compositions according to Tukey's test at p < 0.05.
Phenylethyl alcohol, which has a characteristic floral note [59], can be produced from phenyl pyruvate, which is generated by transamination of Phe.Phenylpyruvate is then decarboxylated to phenylacetaldehyde (benzeneacetaldehyde) via the action of decarboxylase.Further, it can be dehydrogenated to phenylethyl alcohol via the action of the enzyme alcohol dehydrogenase (AlcDH) [36].The highest peak area for phenylethyl alcohol was obtained in media with added Phe, where a significantly (p < 0.05) higher peak area acetate.Values are presented as mean ± standard error (n = 6).Different letters represent significant difference between different medium compositions according to Tukey's test at p < 0.05.
Ethanol, which is a marker compound in fermentation studies, can be produced from sugars via an intermediate acetaldehyde using the PK pathway [21], from the degradation of Thr [64], or from acetate through acetyl-CoA [65].Overall, the ethanol peak area was higher in DM containing 1.2% acetate compared to 0.1% (Figure 6A).This may have been a result of the acetate being converted into ethanol.Interestingly, in DM supplemented with either Leu, Ile, or Phe at 0.1% acetate, the peak area for ethanol was significantly (p < 0.05) lower than in other media with either 0.1 or 1.2% acetate.This finding suggests that LB672 in the presence of added Leu, Ile, or Phe at the lower acetate concentrations is using a pathway other than an ethanol-producing pathway.
butanoic, hexanoic, octanoic, and decanoic acids were present in a higher abundance in media containing 1.2% acetate compared to 0.1%.This result demonstrates in a concentration-dependent manner that acetate could be converted into fatty acids with even carbon chain-length numbers.Figure 6.Normalized peak area of ethanol (6A) and acetic acid (6B) across different medium compositions (see Table 1) at 0.1% and 1.2% acetate.Values are presented as mean ± standard error (n = 6).Different letters represent significant difference between the different medium compositions according to Tukey's test at p < 0.05.
In the current study, the addition of some individual amino acids (Phe, Met, Leu, and Ile) or the addition of the amino acid mixture (Aamix) had an impact on the relative proportions of VOCs formed by LB672 fermentation.In contrast, no particular VOCs were detected in the medium supplemented with Thr.Thr catabolism by LAB is initiated by threonine aldolase (TA) or serine hydroxymethyltransferase (SHMT), which converts Thr to acetaldehyde and glycine [35,36].Acetaldehyde was not detected after LB672 fermentation in the current study.As acetaldehyde is an intermediate compound in the ethanol production pathway [64], it was considered that it was most likely not detected rather than not being present.The inability to detect acetaldehyde was due to the fact that the HS-SPME method used was not optimised for its detection-rather, it was designed to detect a broad range of compounds.
The addition of different concentrations of acetate (0.1, or 1.2%) in the DM significantly influenced the relative abundance of VOCs generated after LB672 fermentation.The mechanism(s) underpinning this observation were, however, unclear, as acetate's role

Phe-Derived VOCs
Phe, an aromatic amino acid, can be converted by LAB into phenyl pyruvate via the action of aromatic aminotransferases (ArAT) in the presence of α-ketoglutarate.The ArAT is active on all aromatic amino acids (tryptophan (Try), Phe, and tyrosine (Tyr)) and also on Leu and Met [36].Further, phenyl pyruvate can subsequently be converted into benzeneacetaldehyde (phenylacetaldehyde), phenyl acetic acid, phenylethyl alcohol (phenyl ethanol), benzyl alcohol (phenyl methanol), and benzaldehyde by LAB via enzymatic, non-enzymatic, and unknown mechanisms [36,58].
Benzaldehyde, which has characteristic almond and burnt sugar notes [57], is produced chemically (non-enzymatic reactions) from phenyl pyruvate in the presence of oxygen and manganese [34,58].DM with added Phe and 0.1% acetate had the highest benzaldehyde peak area (Figure 3A, medium DMP).Benzaldehyde was detected in all media possibly due to the presence of peptone, which is speculated to be the reason why many of the compounds discussed in the following sections were also detected in many of the ferments at low levels.1) at 0.1% and 1.2% acetate.Values are presented as mean ± standard error (n = 6).Different letters represent significant difference between the different medium compositions according to Tukey's test at p < 0.05.
Phenylethyl alcohol, which has a characteristic floral note [59], can be produced from phenyl pyruvate, which is generated by transamination of Phe.Phenylpyruvate is then decarboxylated to phenylacetaldehyde (benzeneacetaldehyde) via the action of decarboxylase.Further, it can be dehydrogenated to phenylethyl alcohol via the action of the enzyme alcohol dehydrogenase (AlcDH) [36].The highest peak area for phenylethyl alcohol was obtained in media with added Phe, where a significantly (p < 0.05) higher peak area

Phe-Derived VOCs
Phe, an aromatic amino acid, can be converted by LAB into phenyl pyruvate via the action of aromatic aminotransferases (ArAT) in the presence of α-ketoglutarate.The ArAT is active on all aromatic amino acids (tryptophan (Try), Phe, and tyrosine (Tyr)) and also on Leu and Met [36].Further, phenyl pyruvate can subsequently be converted into benzeneacetaldehyde (phenylacetaldehyde), phenyl acetic acid, phenylethyl alcohol (phenyl ethanol), benzyl alcohol (phenyl methanol), and benzaldehyde by LAB via enzymatic, non-enzymatic, and unknown mechanisms [36,58].
Benzaldehyde, which has characteristic almond and burnt sugar notes [57], is produced chemically (non-enzymatic reactions) from phenyl pyruvate in the presence of oxygen and manganese [34,58].DM with added Phe and 0.1% acetate had the highest benzaldehyde peak area (Figure 3A, medium DMP).Benzaldehyde was detected in all media possibly due to the presence of peptone, which is speculated to be the reason why many of the compounds discussed in the following sections were also detected in many of the ferments at low levels.1) at 0.1% and 1.2% acetate.Values are presented as mean ± standard error (n = 6).Different letters represent significant difference between the different medium compositions according to Tukey's test at p < 0.05.
Phenylethyl alcohol, which has a characteristic floral note [59], can be produced from phenyl pyruvate, which is generated by transamination of Phe.Phenylpyruvate is then decarboxylated to phenylacetaldehyde (benzeneacetaldehyde) via the action of decarboxylase.Further, it can be dehydrogenated to phenylethyl alcohol via the action of the enzyme alcohol dehydrogenase (AlcDH) [36].The highest peak area for phenylethyl alcohol was obtained in media with added Phe, where a significantly (p < 0.05) higher peak area acetate.Values are presented as mean ± standard error (n = 6).Different letters represent significant difference between the different medium compositions according to Tukey's test at p < 0.05.
Acetic acid was detected in all media, with its concentration being significantly (p < 0.05) higher at 1.2% acetate-containing DM compared to 0.1% acetate-containing DM (Figure 6B).Though Lev.Brevis strains have been shown to produce acetic acid in addition to lactic acid [66], owing to the fact that acetate was present in the DM, it is not possible to confirm that LB672 produced additional acetate.Acetic acid, however, is the building block for fatty acids consisting of an even number of carbons [32,67].In the current study, butanoic, hexanoic, octanoic, and decanoic acids were present in a higher abundance in media containing 1.2% acetate compared to 0.1%.This result demonstrates in a concentration-dependent manner that acetate could be converted into fatty acids with even carbon chain-length numbers.
In the current study, the addition of some individual amino acids (Phe, Met, Leu, and Ile) or the addition of the amino acid mixture (Aamix) had an impact on the relative proportions of VOCs formed by LB672 fermentation.In contrast, no particular VOCs were detected in the medium supplemented with Thr.Thr catabolism by LAB is initiated by threonine aldolase (TA) or serine hydroxymethyltransferase (SHMT), which converts Thr to acetaldehyde and glycine [35,36].Acetaldehyde was not detected after LB672 fermentation in the current study.As acetaldehyde is an intermediate compound in the ethanol production pathway [64], it was considered that it was most likely not detected rather than not being present.The inability to detect acetaldehyde was due to the fact that the HS-SPME method used was not optimised for its detection-rather, it was designed to detect a broad range of compounds.
The addition of different concentrations of acetate (0.1, or 1.2%) in the DM significantly influenced the relative abundance of VOCs generated after LB672 fermentation.The mechanism(s) underpinning this observation were, however, unclear, as acetate's role in LAB growth and VOC generation is complicated.In addition to serving as a buffering agent, acetate has also been reported to demonstrate growth-stimulating activities for several LAB [47], and omitting acetate from the media can result in diminished growth [48].Acetate can also serve as a precursor to a number of lipids/fatty acids [32,68] and ethanol [65].
The presence of aminotransferases such as BcAT, ArAT, and AspAT and other enzymes involved in amino acid catabolism, including KdcA, AlcDH, AldDH, and CBL/CGL, in LB672 is likely in the current study because of the production of VOCs derived from branched-chain, aromatic, or sulphur (Met) amino acids.This could be confirmed by applying a genomic analysis of LB672.
In the current study, the VOCs produced after 16 days of fermentation were analysed using GC-MS.Although GC-MS is a commonly used reference method for VOC analysis, it cannot be used to track changes in VOCs in real time.However, over time, the concentration of VOC produced can fluctuate dramatically.It is therefore possible that compounds were produced but were not detected in the analysis performed at the end of the fermentation (16 days).Proton transfer reaction-time of flight-mass spectrometry (PTR-ToF-MS) is a quick, direct, non-invasive, and highly sensitive (parts per trillion by volume) online method that can be used to monitor the production of VOCs [69].Such an approach would be useful in determining how VOCs change over time in the DM supplemented with different amino acids.

LAB Strain
The LB672 culture was obtained from White Labs, USA.A stock culture was maintained at 4 • C until use.For working cultures, LB672 was initially cultivated in de Man, Rogosa, and Sharpe (MRS) broth at 25 • C for 3 days in sealed containers using anaero-bic packs (Mitsubishi Gas Chemical (MGC) Company, Tokyo, Japan).An aliquot of the resulting culture was streaked onto MRS agar plates, which were incubated at 25 • C in sealed containers containing MGC anaerobic packs for 3 days or until large colonies were visible on the plates.An inoculating suspension was created by picking colonies from the streak plate and adding them to 10 mL of MRS broth, which was incubated at 25 • C for 3 days in sealed containers using MGC anaerobic packs.To prevent carryover of medium components to fermentation experiments, the cells were added to an Eppendorf tube (1 mL) and centrifugated (5000× g for 5 min at 20 • C) using a microcentrifuge (IEC Micromax, Milford, MA, USA).Cells were then washed twice with sterilised phosphate-buffered saline (PBS) (100 mL; 0.8 g NaCl, 0.02 g KCl, 0.144 g Na 2 HPO 4 , and 0.0245 g KH 2 PO 4 , pH of 7.4) and then resuspended to a final concentration of 1 ×10 9 CFU/mL.The resulting cell suspension was used for the fermentation studies.

Fermentation
To confirm their sterility, prepared media were incubated at 25 • C for at least three days prior and checked for an absence of turbidity prior to inoculation.Each medium was inoculated with 100 µL of the LB672 cell suspension (1 × 10 9 CFU/mL) and incubated at 25 • C for 16 days under anaerobic conditions, as mentioned previously.Uninoculated sterile medium was used as control.All the fermentations were carried out in duplicate.At the end of the fermentation, growth was confirmed by measuring the pH and optical density (OD 600 , Ultrospec 3300 pro, Amersham Biosciences, Amersham, UK) of an aseptically removed sub-sample.

Determination of Volatile Organic Compounds
The volatile organic compounds (VOCs) present at the end of fermentation were measured using headspace-solid phase microextraction (HS-SPME) gas chromatographymass spectrometry (GC-MS) (Agilent 6890N GC system, Beijing, China; coupled to an Agilent 5975B VL mass spectrometer with triple axis detector, Wilmington, DE, USA).Samples were prepared by centrifuging (2000 rpm, 4 • C for 10 min, Sorvall RT 6000, Wilmington, DE, USA) a sub-sample (35 mL) of the ferment and adding the resulting supernatant (10 mL) to a 20 mL headspace vial (Phenomenex, Auckland, New Zealand) containing 3 g of NaCl along with 20 µL of the internal standard (3-heptanone (Sigma-Aldrich, Saint Louis, MO, USA) (0.008 mg/mL, prepared with methanol)).The vials were quickly capped with a Teflon-faced septa and stored at 4 • C in a refrigeration block (Peltier cooling tray) in the autosampler until analysis.Three analytical replicates were prepared from each supernatant to provide a total of six replicates for each fermentation medium (biological replicates: 2 and analytical replicates: 3).Samples were analysed in a randomised order, blocked by replicates.At the beginning of each replicate and at the end of each replicate, two blank samples consisting solely of RO water were analysed.
Prior to GC-MS analysis, each headspace vial was incubated at 40 • C for 5 min, after which the SPME fibre (divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS), 50/30 µm Stableflex, Supelco, Bellefonte, PA, USA) was inserted through the septum of the vial and exposed to the headspace for 40 min at 40 • C. Upon completion of the extraction procedure, for thermal desorption, the fibre was inserted into the front injector port of the GC (set at 230 • C) for 2 min in splitless mode followed by 3 min in split mode (purge flow: 50 mL/min).Between samples, the SPME fibre was conditioned in the back inlet at 270 • C with a purge flow of 50 mL/min for 2 min.VOCs were separated through a Zebron ZB-Wax capillary column (60 m × 0.32 mm inner diameter × 0.5 µm film thickness: Phenomenex, Torrance, CA, USA) using helium as the carrier gas at a constant flow rate of 1 mL/min.The oven temperature programme was set at 50 • C for 5 min initially, followed by heating at a rate of 5 • C/min to 210 • C, followed by 10 • C/min until 240 • C, and finally held for 5 min, for a total run time of 45 min.Mass ions were measured between 29 and 300 m/z via electron ionisation (EI mode 70 eV).The ion source temperatures of mass ion traps were sustained at 230 • C with a quadrupole temperature of 150 • C.

Data Analysis 3.5.1. GC-MS Data Extraction
The GC-MS raw data files were exported in CDF format and imported into PARADISe (Version V6.0.0.14) software.PARADISe is based on PARAFAC2 modelling, which allowed simultaneous deconvolution of pure mass spectra of peaks and integration of areas of deconvoluted peaks for all samples.Resolved peaks were identified by matching their deconvoluted pure mass spectra and retention index (RI) [73,74] against the NIST spectral library (NIST14, version 2.2, National Institute of Standards and Technology).The compounds' RI (NIST spectral library) was compared using the calculated RI.A cubic spline interpolation using the n-alkane series (C7-C25), which was run under the same conditions in GC-MS, was used to calculate the RI of each compound [75].The compounds were further selected based on their R match greater than 800.On a few occasions, compounds with R matches below 800 were chosen because they co-eluted with background or air peaks.If a compound only had a calculated RI or if its identification was not confident, VOCs were considered "unknown".Finally, the data matrix of the peak area for each compound (including not-identified peaks (unknown)) for each sample replicate was exported from PARADISe.

Statistical Analysis
The data exported from PARADISe were examined and compounds that had a significant difference in peak area between controls and treatments were identified using t-tests.Significant differences across treatments were assessed via one-way analysis of variance (ANOVA) with a generalised linear model (significance level at p < 0.05) using SPSS (IBM SPSS statistics, version 29.0.0.0 (241)).
Statistical analysis for selected compounds was carried out in R (version R 4.2.1)(R Foundation for Statistical Computing, Vienna, Austria).Heatmaps were developed using "cluster", "gplots", "factoextra", and "tidyverse" external packages as a non-targeted approach to encompass variations between medium compositions and selected VOCs.The peak area

Figure 1 .
Figure 1.Heatmap visualisation and hierarchical clustering analysis of VOCs produced by LB672 based on the log 2 transformed average peak area of each VOC.Fermentation was carried out in the defined medium (DM) with 0.1% acetate added and supplemented with different amino acids separately or in combination (DMP: Phe added, DML: Leu added, DMI: Ile added, DM: defined medium, DMAa: Aamix added, DMM: Met added, DMT: Thr added, and DMG: Glu added).The green colour represents a higher abundance, whereas the red colour indicates a lower abundance.The VOCs represented in the heatmap are numbered according to the peak numbers (TableS1).

Figure 1 .
Figure 1.Heatmap visualisation and hierarchical clustering analysis of VOCs produced by LB672 based on the log 2 transformed average peak area of each VOC.Fermentation was carried out in the defined medium (DM) with 0.1% acetate added and supplemented with different amino acids separately or in combination (DMP: Phe added, DML: Leu added, DMI: Ile added, DM: defined medium, DMAa: Aamix added, DMM: Met added, DMT: Thr added, and DMG: Glu added).The green colour represents a higher abundance, whereas the red colour indicates a lower abundance.The VOCs represented in the heatmap are numbered according to the peak numbers (TableS1).

Figure 2 .
Figure 2. Heatmap visualisation and hierarchical clustering analysis of VOCs produced by LB672 based on the log 2 transformed average peak area of each VOC.Fermentation was carried out in the defined medium (DM) with 1.2% acetate added and supplemented with different amino acids separately or in combination (DMAa: Aamix added, DMI: Ile added, DM: defined medium, DMM: Met added, DMP: Phe added, DML: Leu added, DMT: Thr added, and DMG: Glu added).The green colour represents a higher abundance, whereas the red colour indicates a lower abundance.The VOCs represented in the heatmap are numbered according to the peak numbers (TableS2).

Figure 2 .
Figure 2. Heatmap visualisation and hierarchical clustering analysis of VOCs produced by LB672 based on the log 2 transformed average peak area of each VOC.Fermentation was carried out in the defined medium (DM) with 1.2% acetate added and supplemented with different amino acids separately or in combination (DMAa: Aamix added, DMI: Ile added, DM: defined medium, DMM: Met added, DMP: Phe added, DML: Leu added, DMT: Thr added, and DMG: Glu added).The green colour represents a higher abundance, whereas the red colour indicates a lower abundance.The VOCs represented in the heatmap are numbered according to the peak numbers (TableS2).

Molecules 2024 ,
29,  x FOR PEER REVIEW 11 of 20 by BcAT) and converted into methanethiol or that methanethiol was produced directly from Met via the activity of CBL/CGL.

Table 1 .
Overview of the composition of the different media used.

Table 2 .
The pH and OD 600 of samples after 16 days of fermentation by LB672 in different medium compositions.
bcResults are the mean value of duplicate samples.Values with different superscript lowercase letters ( a-e ) in