The Effect of Different Medium Compositions and LAB Strains on Fermentation Volatile Organic Compounds (VOCs) Analysed by Proton Transfer Reaction-Time of Flight-Mass Spectrometry (PTR-ToF-MS)

: Lactic acid bacteria (LAB) fermentation is a viable approach for producing plant-based flavour compounds; however, little is understood about the impact of different LAB strains and medium compositions on the production of volatile organic compounds (VOCs). This study investigated the impact of the addition of individual amino acids (AAs) (L-leucine, L-isoleucine, L-phenylalanine, L-glutamic acid, L-aspartic acid, L-threonine, or L-methionine) to a defined medium (DM) on the generation of VOCs (after 0, 7, and 14 days) by one of three LAB strains ( Levilactobacillus brevis WLP672 (LB672), Lactiplantibacillus plantarum LP100 (LP100), and Pediococcus pentosaceus PP100 (PP100)), using proton transfer reaction-time of flight-mass spectrometry (PTR-ToF-MS). The concentration of m / z 45.031 (t.i. acetaldehyde) was significantly ( p < 0.05) higher after 7 days of fermentation by LP100 in the DM supplemented with threonine compared to all other media fermented by all three strains. The concentrations of m / z 49.012 (t.i. methanethiol) and m / z 95.000 (t.i. dimethyl disulfide) were significantly ( p < 0.05) higher after 7 days of fermentation by either LP100, PP100, or LB672 in the DM supplemented with methionine compared to all other media. Information on the role of individual AAs on VOCs generation by different LAB strains will help to guide flavour development from the fermentation of plant-based substrates.


Introduction
Plant-based diets have become popular as a means of reducing the environmental impact associated with animal-based diets while also improving human health and animal welfare [1,2].There is a wide range of meat and dairy analogues available in the market [3]; however, such products frequently lack the overall flavour of their traditional counterparts [4][5][6][7].
Fermentation via lactic acid bacteria (LAB) is a promising approach to generate specific meat and dairy VOCs from plant substrates [12][13][14][15][16].However, it is difficult to relate the influence of plant substrate, fermentation conditions, and LAB strain on VOCs produced due to the complexity of the compounds present in plant-based systems.In addition, LAB are a fastidious group of bacteria, and the majority of them are auxotrophic to vitamins and amino acids, requiring rich, complex cultivation media for normal growth [17,18].However, using a rich, poorly defined cultivation medium makes it difficult to determine how changes in medium composition impact the production of specific VOC.To better understand the production of VOCs in accordance with the compositions of each medium, a defined medium (DM) is therefore required [19].The growth of LAB in DM has been thoroughly investigated over many years [20][21][22][23][24].However, only a few studies have examined the impact of LAB on the VOCs produced during growth in DM [25][26][27], and no studies have reported on the VOC production by different LAB strains in response to changes in DM composition.
LAB cultivation media commonly contain carbohydrates (simple sugars), proteins/amino acids, minerals, vitamins, and buffering agents [17].Amino acids (AAs) serve as the building blocks for important flavour compounds in addition to being a substrate for growth.The first stage of AA catabolism is transamination, where LAB uses the aminotransferase enzyme to convert AAs into α-ketoacids with the presence of αketoglutarate.Produced α-ketoacids can then be decarboxylated into aldehydes, which can then be dehydrogenated into alcohols or carboxylic acids by alcohol dehydrogenase and aldehyde dehydrogenase, respectively [28][29][30].The current study focuses on the impact of adding individual AA to a DM on the generation of specific fermentation VOCs.
The generation of specific VOCs by LAB fermentation is highly strain-dependent as it is dependent on different metabolic pathways carried out by LAB strains using various enzymes [31,32].Thus, this study also aims to investigate how commercial LAB strains grown in a DM individually produce specific VOCs.
Proton transfer reaction-mass spectrometry (PTR-MS) enables the rapid, direct, and non-invasive real-time monitoring of VOC with extremely high sensitivity (parts per trillion (ppt) by volume) [33].The fundamental principles of PTR-MS have been well covered in past reviews [34,35].The main challenge with PTR-MS applications is that identification is based on the molecular formula without the capability to separate isomers.Thus, the parallel use of GC-MS and/or fastGC-PTR-MS analysis is usually required to support compound identification [36][37][38].

LAB Strains
Three commercial LAB strains were used in this study.The Levilactobacillus brevis WLP672 (hereafter referred to as LB672) was obtained from White labs, San Diego, CA, USA.Lactiplantibacillus plantarum LP100 (LP100) and Pediococcus pentosaceus PP100 (PP100) were obtained from BIOAGRO, Thiene, Italy.The strains were maintained at 4 • C prior to use.For activation, 1 mL of either a liquid stock culture or 1 g of lyophilized powder was added to 10 mL of de Man, Rogosa, and Sharpe (MRS) broth, which was incubated at 25 • C for 3 days under anaerobic conditions (Anaerobic packs, Mitsubishi Gas Chemical (MGC) Company, Tokyo, Japan).An aliquot of the resulting culture was inoculated onto MRS agar medium using the streak plate method to obtain single colonies and incubated at 25 • C for 3 days using MGC anaerobic packs.An inoculating suspension was prepared by adding three individual colonies from the streak plate to 10 mL of the MRS broth, which was incubated at 25 • C for 3 days using the MGC anaerobic packs.The cells were pelleted by centrifugation (5000× g for 5 min at 20 • C) (PK 121R/ALC International, Cologno Monzese, Italy) and washed twice with sterilized 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 suspension was used as the inoculum in the fermentation trials.

Fermentation
Prior to fermentation, the prepared media were held for 3 days at 25 • C to ensure sterility.Then, 4 mL aliquots of the media were transferred into sterile headspace vials (20 mL) capped with PTFE/silicone septa (Agilent, Cernusco sul Naviglio, Italy).A 0.05 mL aliquot of each LAB cell suspension (1 × 10 9 CFU/mL) was inoculated to each headspace vial, which was flushed with N 2 at a rate of 10 mL/min for 20 min to establish an anaerobic environment.The vials were placed in sample trays in a randomized order in an autosampler (MPS Multi-Purpose Sampler, Gerstel, Mülheim an der Ruhr, Germany) and held at 25 • C for 14 days.Three replicates were prepared from each sample.Controls were the uninoculated media.At the end of the fermentation (after 14 days), growth was confirmed by measuring the pH (inoLab Level 1/WTW, Weilheim, Germany) and optical density (BioPhotometer/Eppendorf, Hamburg, Germany) of a sub-sample of the fermented culture.

PTR-ToF-MS
The VOCs produced during fermentation were measured at three time points (0, 7, and 14 days of fermentation) using a PTR-ToF-MS 8000 (Ionicon Analytik GmbH, Innsbruck, Austria).The drift tube conditions were as follows: 110 • C drift tube temperature, 2.8 mbar drift pressure, and 628 V drift voltage.This led to an E/N ratio of about 140 Townsend (Td), with E corresponding to the electric field strength and N to the gas number density (1 Td = 10 −17 V cm 2 ).The sampling time per channel of ToF acquisition was 0.1 ns, amounting to 350,000 channels for a mass spectrum ranging up to mass peak (m/z) = 340, which resulted in the acquisition rate of 1 spectrum/s.Each measurement was conducted automatically using an autosampler with 60 s between each measurement to prevent any memory effects/carry over.The sample headspace was withdrawn with a 2.5 mL syringe (CTC Analytics AG, Zwingen, Switzerland) and injected into the static headspace (SHS) module (Ionicon Analytik GmbH, Innsbruck, Austria).The flow of zero air inside the static headspace module was 90 sccm, and the syringe injection time was 25 s/sample at a rate of 100 µL/s, resulting in a 16-fold dilution of the sample [43].Pure N 2 was flushed through the syringe immediately before withdrawal to prevent measurement contamination.PTR-MS performance was verified with certified calibration mixtures.Sensitivity was better than 10 cps/ppbv, and the limit of detection (LOD) was lower than 100 pptv at an acquisition rate of 1 spectrum/s.The mass resolution was at least 4000 M/∆M.Deadtime correction, internal calibration of mass spectral data, and peak extraction were performed according to previously described procedures [44,45].The peak intensity in ppb/v (parts per billion per volume) was estimated using the formula described in the literature.The formula uses a constant value for the reaction rate coefficient (k = 2•10 −9 cm 3 s −1 ) [46].

HS-SPME-GC-MS
HS-SPME-GC-MS measurements were included to support the identification of compounds detected by PTR-ToF-MS.At the end of the fermentation (after 14 days), the samples were removed from the PTR-ToF-MS autosampler sample tray and transferred to a GC-MS autosampler sample tray held at 25 • C.An SPME fibre (DVB/CAR/PDMS, 2 cm, 50/30 µm thickness) was exposed to the headspace of the samples for 40 min at 25 • C. VOCs were desorbed from the SPME fibre at 250 • C for 5 min in the injector of the GC in splitless mode, and helium was used as the carrier gas at a flow rate of 2 mL/min for 5 min.The VOCs were separated using a capillary column (InnoWax 30 m/0.32 mm/0.5 µm).The oven temperature program was set at 40 • C held for 1 min, and then increased to 250 • C at 5 • C/min and held for 2 min.MS was performed with an ion source temperature of 200 • C and an electron ionization energy of 70 eV over the mass range of m/z 33-350.
Retention indices (RI) for each VOC were calculated from the retention time (RT) of an n-alkane series (C6-C30) obtained under the same HS-SPME-GC-MS conditions.By comparing the calculated RI and from the NIST library (NIST14, version 2.2, National Institute of Standards and Technology), the VOCs were identified.

FastGC-PTR-ToF-MS
To assist with attributing each signal (m/z) to the correct compound and determining the number of compounds contributing to each m/z (isomers), fastGC-PTR-ToF-MS was carried out on all the samples at each time point after performing SHS-PTR-ToF-MS measurements.The drift tube conditions were the same as described in Section 2.4.1.The polar capillary column (MXT ® -WAX (Siltek ® -treated stainless steel), 6 m) was maintained under pure helium with a constant flow rate of 2.5 sccm.Pure N 2 was used as a make-up gas with a flow of 50 sccm.Sample headspace air was injected through the purge tool (Gerstel, Germany) into a fastGC sampling loop for 15 s, guaranteeing its total filling.The chromatographic measurement was registered for 250 s with the thermal ramp from 40 to 200 • C and data acquisition was set to 5 spectra/s [47].The following pure standards, ethyl acetate, ethyl butanoate, ethyl hexanoate, ethyl octanoate, ethyl decanoate, ethanol, 2-methyl propanol, 3-methyl butanol, phenylethyl alcohol, 2-butanone, 2-hexanone, 2-heptanone, 2-nonanone, and benzaldehyde were prepared individually and diluted to a final concentration of 10 ppm through serial dilutions.Acetic acid was diluted to a final concentration of 50 ppm through serial dilutions.The 15 pure standards were analysed as well in fastGC-PTR-ToF-MS to assist the identification of m/z.TofDAQViewer was used to visually inspect fastGC-PTR-ToF-MS data of the standards and samples after they were saved as h5-files.From the 15 standards, a table was prepared consisting of RT, the literature RI, and fragmentation pattern, which in combination with the literature fragmentation patterns and GC-MS results, was used to assist in assigning compound identities to each m/z (Table 2).

Statistical Analysis
To determine which m/z were significantly (p < 0.05) higher in the samples than in the blanks, an analysis of variance (ANOVA) was run between each sample type and the corresponding blank.
Principal component analysis (PCA) was performed on all samples using related m/z concentrations and coded to highlight the sample differentiation based on DM compositions (DM and DM supplemented with individual AAs), LAB strains, and fermentation time.To ensure that the PCA was focusing on variation in the data from different DM compositions, LAB strains, and fermentation time, data from the control (uninoculated) treatments were excluded from the PCA.The analysis was performed in R (version 4.2.1, R Foundation for Statistical Computing, Vienna, Austria) by using "factoextra", "ggplot2", "reshape", "ggpubr", and "dplyr" packages [48].Data were normalized by autoscaling (mean-centred and divided by the standard deviation) using the "prcomp" function of the "factoextra" package.
To identify m/z that were significantly influenced by experimental conditions, all sample-related m/z were subjected to three-way ANOVA using a general linear model (significance level at p < 0.05), where the main effects were DM compositions, LAB strains, and fermentation time, and all interactions were investigated.The mean separations for each m/z were calculated using Tukey's HSD test at p < 0.05.Analysis was carried out using SPSS (IBM SPSS statistics, version 29.0.0.0 (241), Chicago, IL, USA).
The selected VOCs (m/z) were plotted in bar graphs separating data based upon main experimental factors of DM compositions, and LAB strains as well as the interactions between DM compositions and LAB strains at 7 days of fermentation using "ggplot2", "dplyr", "ggpubr", "reshape", "ggthemes", "multcompView", "readr", and "scales" pack-ages in R. The mean separations for each m/z were calculated using Tukey's HSD test at p < 0.05.

Physicochemical Properties after Fermentation
All the LAB strains grew well either in the DM or in the DM supplemented with individual AAs, as indicated by a decrease in pH (due to acid production) [49], and an increase in optical density (OD 600 ) values (resulting from cell growth) (Table 3).The OD 600 values of the inoculated samples at day 0 ranged between 0.01 and 0.03.There were significant (p < 0.05) differences in the pH and OD 600 values between LAB strains across different medium compositions after 14 days of fermentation.In general, the pH of medium compositions fermented by LB672 was highest except for DMA, DML, and DMI, where there were no significant (p > 0.05) differences in the pH between LAB strains.

VOCs Produced during Fermentation
Fermentation by three LAB strains either in the DM or in the DM supplemented with different AAs, resulted in a total of 184 m/z being measured in the raw PTR-ToF-MS data.After the removal of isotopologues and m/z that were not significantly (p > 0.05) different from the baseline, this number decreased to 88 (Table S1).m/z were tentatively identified (t.i.) based on their exact mass, supported by HS-SPME-GC-MS (22 out of 88 m/z) (Tables 4, 5 and S1), fastGC-PTR-ToF-MS identification (Tables 2 and 5), and/or literature data.    1 : m/z that HS-SPME-GC-MS identified. 2 : m/z identified by fastGC-PTR-ToF-MS and/or the injection of pure standard. 3: 2-methyl-2-butenal and 3-methyl-2-butenal both present, but mainly 2-methyl-2-butenal due to elevated treatment effect based on HS-SPME-GC-MS. 4 : Ethyl acetate + butyric acid dominant in LB672 fermented samples, and acetoin + butyric acid dominant in LP100 and PP100 fermented samples based on fastGC-PTR-ToF-MS.
The concentrations of the 88 m/z were higher at 7 days.At 14 days, the concentrations were lower, which was assumed to be due to a decrease in LAB growth/metabolism as substrates became depleted, which resulted in a decrease in the concentration of the volatiles in the headspace as those present at day 7 were removed during the flushing of the headspace.Hence, to better determine the effects of different LAB strains on the VOCs produced across different medium compositions, only the data on day 0 and after 7 days of fermentation were considered.
To understand the differences between m/z produced by three LAB strains across different medium compositions, PCA was performed for the 88 m/z detected by PTR-ToF-MS.As shown in Figure 1a, the PCA score plot of LB672-, LP100-, and PP100-fermented DM and DM supplemented with individual AAs samples explained 52.7% of the total variance, comprising 38.4% from the first principal component (Dim1) and 14.3% from the second principal component (Dim2).The explained variance was mainly attributed to the separation of LB672-, LP100-, and PP100-fermented DM and DM supplemented with individual AA samples at 7 days from the 0 days samples and LB672-fermented DM samples from LP100-and PP100-fermented DM and DM supplemented with AA samples, and LB672-fermented DM supplemented with AA samples at 7 days (Figure 1a).The separation along Dim 1 was attributed to the presence of higher concentrations of m/z 60.021, 126.968, 128.058, 89.058 (t.i.ethyl acetate, acetoin, and butyric acid), 117.078, 54.006, 57.039, 43.017 (t.i.common fragment), 61.028 (t.i.acetic acid), 49.012 (t.i.methanethiol), and 44.022 associated with mainly LB672-fermented DM supplemented with AA samples at 7 days from all samples at 0 days (Figure 1b).Variation along Dim 2 was attributed to the separation of 7 days of LB672-fermented DM samples, in particular DMA (negative Dim 2), from LP100-and PP100-fermented DM and DM supplemented with AA samples at 7 days.The separation along Dim 2 was mainly attributed to ethanol-related m/z (47.045, 29.04, 65.057, and 75.08) along with m/z 26.016 (common fragment), 27.025, 31.018(t.i.formaldehyde), 63.01 (t.i.carbon dioxide/water cluster), 83.066, and 93.087 (t.i.toluene) which had negative loadings, and 71.050 and 75.036 with positive loadings (Figure 1b).
Three-way ANOVA was used to further investigate differences in VOC emissions from the different media and LAB strain combinations.The analysis found that of the 88 m/z signals attributed to the samples, 46, 55, 81, 32, 44, 48, and 33 m/z were significantly (p < 0.05) differentiated based upon LAB strains, DM compositions, time (at 0 and 7 days), DM compositions*LAB strains interactions, time*LAB strains interactions, DM compositions*time interactions, and DM compositions*time*LAB strains interactions, respectively (Table S1).Finally, 43 m/z were identified (Table 5) that demonstrated a significant (p < 0.05) increase in their concentration during fermentation (time) and significant (p < 0.05) differences in either DM compositions, or LAB strains or interaction effects.
The specific fermentation VOCs (m/z) produced by different LAB strains in DM supplemented with different AAs are discussed in the following sections.and PP100-fermented DM and DM supplemented with AA samples at 7 days.The separation along Dim 2 was mainly attributed to ethanol-related m/z (47.045, 29.04, 65.057, and 75.08) along with m/z 26.016 (common fragment), 27.025, 31.018(t.i.formaldehyde), 63.01 (t.i.carbon dioxide/water cluster), 83.066, and 93.087 (t.i.toluene) which had negative loadings, and 71.050 and 75.036 with positive loadings (Figure 1b).

Ethanol
Ethanol is an end product of sugar fermentation by LAB.Heterofermentative LAB (Lev.brevis) utilise hexose sugars (glucose) via the phosphoketolase (PK) pathway and produces ethanol as the end product via an intermediate of acetaldehyde using an alcohol dehydrogenase (AlcDH) enzyme.In contrast, homofermentative LAB (P.pentosaceus) ferment hexose sugars solely into lactic acid through the Embden/Meyerhof/Parnas (EMP) pathway [16].However, at slow growth and low glycolytic flux rates, homofermentative LAB can shift to a mixed acid fermentation with ethanol as one of the end products [50].Notably, facultative heterofermentative LAB (Lpb.plantarum) utilise hexose sugars to produce lactic acid through the EMP pathway and pentose sugars by the PK pathway [51] and under certain conditions, these LAB have been reported to ferment hexoses through the PK pathway [52].Furthermore, ethanol can also be produced by LAB through the degradation of the AA Thr [29,30,53].
In the present study, the concentration of m/z 47.045 (t.i.ethanol) was significantly (p < 0.05) higher after 7 days of fermentation by LB672 (Figure 2) across the different medium compositions, where ethanol was detected at trace amounts after either LP100 or PP100 fermentation.In the LB672 fermentation, ethanol was higher in the DM compared to DM supplemented with individual AAs.Given the similarities in pH after LB672 fermentation across different medium compositions (Table 3) after 14 days, differences in ethanol in LB672 after 7 days are likely due to differences in fermentation rate.spectively (Table S1).Finally, 43 m/z were identified (Table 5) that demonstrated a significant (p < 0.05) increase in their concentration during fermentation (time) and significant (p < 0.05) differences in either DM compositions, or LAB strains or interaction effects.The specific fermentation VOCs (m/z) produced by different LAB strains in DM supplemented with different AAs are discussed in the following sections.

Ethanol
Ethanol is an end product of sugar fermentation by LAB.Heterofermentative LAB (Lev.brevis) utilise hexose sugars (glucose) via the phosphoketolase (PK) pathway and produces ethanol as the end product via an intermediate of acetaldehyde using an alcohol dehydrogenase (AlcDH) enzyme.In contrast, homofermentative LAB (P.pentosaceus) ferment hexose sugars solely into lactic acid through the Embden/Meyerhof/Parnas (EMP) pathway [16].However, at slow growth and low glycolytic flux rates, homofermentative LAB can shift to a mixed acid fermentation with ethanol as one of the end products [50].Notably, facultative heterofermentative LAB (Lpb.plantarum) utilise hexose sugars to produce lactic acid through the EMP pathway and pentose sugars by the PK pathway [51] and under certain conditions, these LAB have been reported to ferment hexoses through the PK pathway [52].Furthermore, ethanol can also be produced by LAB through the degradation of the AA Thr [29,30,53].
In the present study, the concentration of m/z 47.045 (t.i.ethanol) was significantly (p < 0.05) higher after 7 days of fermentation by LB672 (Figure 2) across the different medium compositions, where ethanol was detected at trace amounts after either LP100 or PP100 fermentation.In the LB672 fermentation, ethanol was higher in the DM compared to DM supplemented with individual AAs.Given the similarities in pH after LB672 fermentation across different medium compositions (Table 3) after 14 days, differences in ethanol in LB672 after 7 days are likely due to differences in fermentation rate.

Thr-Derived VOCs
Acetaldehyde, which is a main flavour com either from the catabolism of Thr using eithe droxymethyltransferase (SHMT) enzymes [29,3 (intermediate in the ethanol production pathwa tration of m/z 45.031 (t.i.acetaldehyde) was sig fermentation by LP100 in the DM with added Th in the medium DMT compared to all other me LB672 (Figure 3).This suggests that the presen likely/higher in LP100 and PP100 strains.Howe ence or absence of TA/SHMT enzymes in LB67 was detected in LB672-fermented media (Figure dehyde was converted into ethanol by the AlcD ative and homofermentative LAB strains, LP100 etaldehyde, but not ethanol after fermentation

Thr-Derived VOCs
Acetaldehyde, which is a main flavour comp either from the catabolism of Thr using either droxymethyltransferase (SHMT) enzymes [29,30 (intermediate in the ethanol production pathway tration of m/z 45.031 (t.i.acetaldehyde) was signi fermentation by LP100 in the DM with added Thr in the medium DMT compared to all other med LB672 (Figure 3).This suggests that the presence likely/higher in LP100 and PP100 strains.Howev ence or absence of TA/SHMT enzymes in LB672 was detected in LB672-fermented media (Figure dehyde was converted into ethanol by the AlcDH ative and homofermentative LAB strains, LP100 a etaldehyde, but not ethanol after fermentation in

Thr-Derived VOCs
Acetaldehyde, which is a main flavour compou either from the catabolism of Thr using either t droxymethyltransferase (SHMT) enzymes [29,30,53 (intermediate in the ethanol production pathway) tration of m/z 45.031 (t.i.acetaldehyde) was signific fermentation by LP100 in the DM with added Thr (D in the medium DMT compared to all other media LB672 (Figure 3).This suggests that the presence/a likely/higher in LP100 and PP100 strains.However ence or absence of TA/SHMT enzymes in LB672 a was detected in LB672-fermented media (Figure 2), dehyde was converted into ethanol by the AlcDH e ative and homofermentative LAB strains, LP100 an etaldehyde, but not ethanol after fermentation in a

Thr-Derived VOCs
Acetaldehyde, which is a main flavour compound in yogurt [54], is produced by LAB either from the catabolism of Thr using either threonine aldolase (TA) or serine hydroxymethyltransferase (SHMT) enzymes [29,30,53] or from sugars via the PK pathway (intermediate in the ethanol production pathway) [16].In the current study, the concentration of m/z 45.031 (t.i.acetaldehyde) was significantly (p < 0.05) higher after 7 days of fermentation by LP100 in the DM with added Thr (DMT), followed by PP100 fermentation in the medium DMT compared to all other media fermented by either LP100, PP100, or LB672 (Figure 3).This suggests that the presence/activity of TA/SHMT enzymes is more likely/higher in LP100 and PP100 strains.However, it is not possible to confirm the presence or absence of TA/SHMT enzymes in LB672 as the highest concentration of ethanol was detected in LB672-fermented media (Figure 2), which suggests that produced acetaldehyde was converted into ethanol by the AlcDH enzyme.The facultative heterofermentative and homofermentative LAB strains, LP100 and PP100, respectively, did produce acetaldehyde, but not ethanol after fermentation in all media studied, which suggests that the AlcDH enzyme is most likely absent/not active in these strains.Notably, acetaldehyde was detected in all media possibly due to the presence of glucose, peptone, and an AA mixture in all DM compositions.tration of m/z 45.031 (t.i.acetaldehyde) was significantly (p < 0.05) higher after 7 days of fermentation by LP100 in the DM with added Thr (DMT), followed by PP100 fermentation in the medium DMT compared to all other media fermented by either LP100, PP100, or LB672 (Figure 3).This suggests that the presence/activity of TA/SHMT enzymes is more likely/higher in LP100 and PP100 strains.However, it is not possible to confirm the presence or absence of TA/SHMT enzymes in LB672 as the highest concentration of ethanol was detected in LB672-fermented media (Figure 2), which suggests that produced acetaldehyde was converted into ethanol by the AlcDH enzyme.The facultative heterofermentative and homofermentative LAB strains, LP100 and PP100, respectively, did produce acetaldehyde, but not ethanol after fermentation in all media studied, which suggests that the AlcDH enzyme is most likely absent/not active in these strains.Notably, acetaldehyde was detected in all media possibly due to the presence of glucose, peptone, and an AA mixture in all DM compositions.Different superscript lowercase letters represent significant differences between different medium compositions fermented by different LAB strains according to Tukey's test at p<0.05.

Met-Derived VOCs
Met is a sulphur-containing AA whose catabolism is initiated by a transamination step, in which the aromatic aminotransferase (ArAT,) or the branched-chain

Thr-Derived VOCs
Acetaldehyde, which is a main flavour co either from the catabolism of Thr using eit droxymethyltransferase (SHMT) enzymes [29 (intermediate in the ethanol production pathw tration of m/z 45.031 (t.i.acetaldehyde) was si fermentation by LP100 in the DM with added T in the medium DMT compared to all other m LB672 (Figure 3).This suggests that the prese likely/higher in LP100 and PP100 strains.How ence or absence of TA/SHMT enzymes in LB was detected in LB672-fermented media (Figu dehyde was converted into ethanol by the Alc ative and homofermentative LAB strains, LP1 etaldehyde, but not ethanol after fermentation the AlcDH enzyme is most likely absent/not ac was detected in all media possibly due to the mixture in all DM compositions.

Thr-Derived VOCs
Acetaldehyde, which is a main flavour com either from the catabolism of Thr using eithe droxymethyltransferase (SHMT) enzymes [29,3 (intermediate in the ethanol production pathwa tration of m/z 45.031 (t.i.acetaldehyde) was sig fermentation by LP100 in the DM with added Th in the medium DMT compared to all other me LB672 (Figure 3).This suggests that the presen likely/higher in LP100 and PP100 strains.Howe ence or absence of TA/SHMT enzymes in LB67 was detected in LB672-fermented media (Figure dehyde was converted into ethanol by the AlcD ative and homofermentative LAB strains, LP100 etaldehyde, but not ethanol after fermentation the AlcDH enzyme is most likely absent/not acti was detected in all media possibly due to the p mixture in all DM compositions.

Thr-Derived VOCs
Acetaldehyde, which is a main flavour compo either from the catabolism of Thr using either droxymethyltransferase (SHMT) enzymes [29,30,5 (intermediate in the ethanol production pathway) tration of m/z 45.031 (t.i.acetaldehyde) was signifi fermentation by LP100 in the DM with added Thr ( in the medium DMT compared to all other media LB672 (Figure 3).This suggests that the presence/ likely/higher in LP100 and PP100 strains.Howeve ence or absence of TA/SHMT enzymes in LB672 a was detected in LB672-fermented media (Figure 2) dehyde was converted into ethanol by the AlcDH e ative and homofermentative LAB strains, LP100 an etaldehyde, but not ethanol after fermentation in the AlcDH enzyme is most likely absent/not active was detected in all media possibly due to the pre mixture in all DM compositions.
) at 25 • C. Values are presented as mean ± standard error (n = 3).Different superscript lowercase letters represent significant differences between different medium compositions fermented by different LAB strains according to Tukey's test at p < 0.05.
In the current study, the concentration of m/z 49.012 (t.i.methanethiol) was significantly (p < 0.05) higher after 7 days of fermentation by either LP100, PP100, or LB672 in the DM supplemented with Met (DMM) compared to all other media fermented by all three strains (Figure 4a).Methanethiol is a characteristic flavour compound in meat [57] and cheese [58,59].Similarly, the concentration of m/z 95.000 (t.i.dimethyl disulfide) was significantly (p < 0.05) higher after 7 days of fermentation by either LP100, PP100, or LB672 in the medium DMM, compared to all other media fermented by all of these strains (Figure 4b).Notably, methanethiol and dimethyl disulfide were detected in all media possibly due to the presence of peptone and an AA mixture in all DM compositions, which contained Met.
the DM supplemented with Met (DMM) compared to all other media fermented by all three strains (Figure 4a).Methanethiol is a characteristic flavour compound in meat [57] and cheese [58,59].Similarly, the concentration of m/z 95.000 (t.i.dimethyl disulfide) was significantly (p < 0.05) higher after 7 days of fermentation by either LP100, PP100, or LB672 in the medium DMM, compared to all other media fermented by all of these strains (Figure 4b).Notably, methanethiol and dimethyl disulfide were detected in all media possibly due to the presence of peptone and an AA mixture in all DM compositions, which contained Met.
According to a study by Liu et al. [60], the genes cblA and cglA encoding CBL and CGL lyases are present in Lev.brevis ATCC 367 and Lpb.plantarum WCFS1, which catalyses the demethiolation reaction of Met.Whereas aminotransferases such as ArAT are present in Lpb.plantarum WCFS1 and P. pentosaceus ATCC 25745 and BcAT is only present in Lpb.plantarum WCFS1 and is involved in the transamination reaction.It is obvious that without these enzymes, methanethiol production is not possible.Therefore, the LAB strains used in the current study LP100, PP100, or LB672 likely contain either CBL and CGL lyases and/or ArAT/BcAT transferases.A genomic study is therefore required to confirm the presence of CGL, CBL, ArAT, and BcAT enzymes in these strains.Different superscript lowercase letters represent significant differences between different medium compositions fermented by different LAB strains according to Tukey's test at p<0.05.

Thr-Derived VOCs
Acetaldehyde, which is a main flavour compound in yogurt [54], is produced by LAB either from the catabolism of Thr using either threonine aldolase (TA) or serine hydroxymethyltransferase (SHMT) enzymes [29,30,53] or from sugars via the PK pathway (intermediate in the ethanol production pathway) [16].In the current study, the concentration of m/z 45.031 (t.i.acetaldehyde) was significantly (p < 0.05) higher after 7 days of fermentation by LP100 in the DM with added Thr (DMT), followed by PP100 fermentation in the medium DMT compared to all other media fermented by either LP100, PP100, or LB672 (Figure 3).This suggests that the presence/activity of TA/SHMT enzymes is more likely/higher in LP100 and PP100 strains.However, it is not possible to confirm the presence or absence of TA/SHMT enzymes in LB672 as the highest concentration of ethanol was detected in LB672-fermented media (Figure 2), which suggests that produced acetaldehyde was converted into ethanol by the AlcDH enzyme.The facultative heterofermentative and homofermentative LAB strains, LP100 and PP100, respectively, did produce acetaldehyde, but not ethanol after fermentation in all media studied, which suggests that the AlcDH enzyme is most likely absent/not active in these strains.Notably, acetaldehyde was detected in all media possibly due to the presence of glucose, peptone, and an AA mixture in all DM compositions.Different superscript lowercase letters represent significant differences between different medium compositions fermented by different LAB strains according to Tukey's test at p<0.05.

Thr-Derived VOCs
Acetaldehyde, which is a main flavour compound in yogurt [54], is produced by LAB either from the catabolism of Thr using either threonine aldolase (TA) or serine hydroxymethyltransferase (SHMT) enzymes [29,30,53] or from sugars via the PK pathway (intermediate in the ethanol production pathway) [16].In the current study, the concentration of m/z 45.031 (t.i.acetaldehyde) was significantly (p < 0.05) higher after 7 days of fermentation by LP100 in the DM with added Thr (DMT), followed by PP100 fermentation in the medium DMT compared to all other media fermented by either LP100, PP100, or LB672 (Figure 3).This suggests that the presence/activity of TA/SHMT enzymes is more likely/higher in LP100 and PP100 strains.However, it is not possible to confirm the presence or absence of TA/SHMT enzymes in LB672 as the highest concentration of ethanol was detected in LB672-fermented media (Figure 2), which suggests that produced acetaldehyde was converted into ethanol by the AlcDH enzyme.The facultative heterofermentative and homofermentative LAB strains, LP100 and PP100, respectively, did produce acetaldehyde, but not ethanol after fermentation in all media studied, which suggests that the AlcDH enzyme is most likely absent/not active in these strains.Notably, acetaldehyde was detected in all media possibly due to the presence of glucose, peptone, and an AA mixture in all DM compositions.Different superscript lowercase letters represent significant differences between different medium compositions fermented by different LAB strains according to Tukey's test at p<0.05.

Thr-Derived VOCs
Acetaldehyde, which is a main flavour compound in yogurt [54], is produced by LAB either from the catabolism of Thr using either threonine aldolase (TA) or serine hydroxymethyltransferase (SHMT) enzymes [29,30,53] or from sugars via the PK pathway (intermediate in the ethanol production pathway) [16].In the current study, the concentration of m/z 45.031 (t.i.acetaldehyde) was significantly (p < 0.05) higher after 7 days of fermentation by LP100 in the DM with added Thr (DMT), followed by PP100 fermentation in the medium DMT compared to all other media fermented by either LP100, PP100, or LB672 (Figure 3).This suggests that the presence/activity of TA/SHMT enzymes is more likely/higher in LP100 and PP100 strains.However, it is not possible to confirm the presence or absence of TA/SHMT enzymes in LB672 as the highest concentration of ethanol was detected in LB672-fermented media (Figure 2), which suggests that produced acetaldehyde was converted into ethanol by the AlcDH enzyme.The facultative heterofermentative and homofermentative LAB strains, LP100 and PP100, respectively, did produce acetaldehyde, but not ethanol after fermentation in all media studied, which suggests that the AlcDH enzyme is most likely absent/not active in these strains.Notably, acetaldehyde was detected in all media possibly due to the presence of glucose, peptone, and an AA mixture in all DM compositions.According to a study by Liu et al. [60], the genes cblA and cglA encoding CBL and CGL lyases are present in Lev.brevis ATCC 367 and Lpb.plantarum WCFS1, which catalyses the demethiolation reaction of Met.Whereas aminotransferases such as ArAT are present in Lpb.plantarum WCFS1 and P. pentosaceus ATCC 25745 and BcAT is only present in Lpb.plantarum WCFS1 and is involved in the transamination reaction.It is obvious that without these enzymes, methanethiol production is not possible.Therefore, the LAB strains used in the current study LP100, PP100, or LB672 likely contain either CBL and CGL lyases and/or ArAT/BcAT transferases.A genomic study is therefore required to confirm the presence of CGL, CBL, ArAT, and BcAT enzymes in these strains.

Other VOCs
The concentration of m/z 89.058 (t.i.acetoin, ethyl acetate, and butyric acid) was significantly (p < 0.05) higher after 7 days of fermentation by LB672 across different medium compositions compared to either LP100 or PP100 fermentation across all media studied (Figure 6).Based on the fastGC-PTR-ToF-MS results, the m/z 89.058 detected by PTR-ToF-MS was considered to be comprised of contributions from ethyl acetate, acetoin, and butyric acid (and a small signal was observed for ethyl butanoate).However, there were differences between LAB strains in the contribution of ethyl acetate, acetoin, and butyric acid to this m/z: acetoin and butyric acid were dominant in LP100 and PP100 ferments, whereas ethyl acetate and butyric acid were dominant in LB672 ferments.fermentation by LP100 i in the medium DMT co LB672 (Figure 3).This s likely/higher in LP100 a ence or absence of TA/ was detected in LB672-f dehyde was converted i ative and homoferment etaldehyde, but not eth the AlcDH enzyme is m was detected in all med mixture in all DM comp superscript lowercase letters represent significant differences between different medium compositions by different LAB strains according to Tukey's test at p<0.05.

Thr-Derived VOCs
Acetaldehyde, which is a main flavour compound in yogurt [54], is produced by LAB either from the catabolism of Thr using either threonine aldolase (TA) or serine hydroxymethyltransferase (SHMT) enzymes [29,30,53] or from sugars via the PK pathway (intermediate in the ethanol production pathway) [16].In the current study, the concentration of m/z 45.031 (t.i.acetaldehyde) was significantly (p < 0.05) higher after 7 days of fermentation by LP100 in the DM with added Thr (DMT), followed by PP100 fermentation in the medium DMT compared to all other media fermented by either LP100, PP100, or LB672 (Figure 3).This suggests that the presence/activity of TA/SHMT enzymes is more likely/higher in LP100 and PP100 strains.However, it is not possible to confirm the presence or absence of TA/SHMT enzymes in LB672 as the highest concentration of ethanol was detected in LB672-fermented media (Figure 2), which suggests that produced acetaldehyde was converted into ethanol by the AlcDH enzyme.The facultative heterofermentative and homofermentative LAB strains, LP100 and PP100, respectively, did produce acetaldehyde, but not ethanol after fermentation in all media studied, which suggests that the AlcDH enzyme is most likely absent/not active in these strains.Notably, acetaldehyde was detected in all media possibly due to the presence of glucose, peptone, and an AA mixture in all DM compositions.
) at 25 • C. Values are presented as mean ± standard error (n = 3).Different l superscript lowercase letters represent significant differences between different medium compositions fermented by different LAB strains according to Tukey's test at p < 0.05.

Other VOCs
The concentration of m/z 89.058 (t.i.acetoin, ethyl acetate, and butyric acid) was significantly (p < 0.05) higher after 7 days of fermentation by LB672 across different medium compositions compared to either LP100 or PP100 fermentation across all media studied (Figure 6).Based on the fastGC-PTR-ToF-MS results, the m/z 89.058 detected by PTR-ToF-MS was considered to be comprised of contributions from ethyl acetate, acetoin, and butyric acid (and a small signal was observed for ethyl butanoate).However, there were differences between LAB strains in the contribution of ethyl acetate, acetoin, and butyric acid to this m/z: acetoin and butyric acid were dominant in LP100 and PP100 ferments, whereas ethyl acetate and butyric acid were dominant in LB672 ferments.
Based on previous experiments, to obtain good growth of LAB in the DM, the DM was supplemented with an AA mixture in addition to peptone.While the AA mixture addition made determining the impact of individual AAs on VOC production a little more challenging, the obvious advantage was that the cultures all grew.It is important to note that the concentration of each individual AA added to the DM was at least five times higher than its concentration in the DM from the AA mixture.A closer look at the VOCs derived from individual AAs reveals that Thr, Met, and Ile had the highest impact on specific VOCs production.In contrast, the addition of either Leu, Glu, Asp, or Phe had no significant effect on the production of specific VOCs.This might be because the addition of a higher concentration of a single AA may result in an increased requirement for other AA, which will influence the VOCs produced [62].
This study investigated how the relative concentrations of VOCs in the DM and DM supplemented with AAs were influenced by only three commercial LAB strains.To better understand the impact of strain on particular VOCs, further study using different commercial LAB strains is required.Further, temperature controls the growth rate of LAB, resulting in impact on VOC generation.As only 25 • C was used in the present study, different fermentation temperatures need to be investigated to determine their effect on the VOCs production.Based on previous experiments, to obtain good growth of LAB in the DM, the DM was supplemented with an AA mixture in addition to peptone.While the AA mixture addition made determining the impact of individual AAs on VOC production a little more challenging, the obvious advantage was that the cultures all grew.It is important to note that the concentration of each individual AA added to the DM was at least five times higher than its concentration in the DM from the AA mixture.A closer look at the VOCs derived from individual AAs reveals that Thr, Met, and Ile had the highest impact on specific VOCs production.In contrast, the addition of either Leu, Glu, Asp, or Phe had no significant effect on the production of specific VOCs.This might be because the addition of a higher concentration of a single AA may result in an increased requirement for other AA, which will influence the VOCs produced [62].
This study investigated how the relative concentrations of VOCs in the DM and DM supplemented with AAs were influenced by only three commercial LAB strains.To better understand the impact of strain on particular VOCs, further study using different commercial LAB strains is required.Further, temperature controls the growth rate of LAB, resulting in impact on VOC generation.As only 25 °C was used in the present study, different fermentation temperatures need to be investigated to determine their effect on the VOCs production.

Conclusions
The present study demonstrated that VOCs produced during fermentation analysed by PTR-ToF-MS, HS-SPME-GC-MS, and fastGC-ToF-MS were influenced by DM compositions (AA addition), and LAB strains.The addition of Thr, Met, and Ile AAs to the DM, noticeably impacted the relative concentrations of VOCs after LB672, LP100, and PP100 fermentation, suggesting the presence of specific AA catabolic enzymes in these strains.Understanding how specific VOC generation responds to varying medium compositions and LAB strains will facilitate the production of the VOC required to improve the flavour profiles of plant-based fermentation products or analogue products.Therefore, further investigations utilising different commercial LAB strains are needed to more broadly understand how AAs or medium compositions impact the generation of fermentation-   fermentation by LP100 i in the medium DMT co LB672 (Figure 3).This s likely/higher in LP100 a ence or absence of TA/ was detected in LB672-f dehyde was converted i ative and homoferment etaldehyde, but not eth the AlcDH enzyme is m was detected in all med mixture in all DM comp superscript lowercase letters represent significant differences between different medium compositions by different LAB strains according to Tukey's test at p<0.05.

Thr-Derived VOCs
Acetaldehyde, which is a main flavour compound in yogurt [54], is produced by LAB either from the catabolism of Thr using either threonine aldolase (TA) or serine hydroxymethyltransferase (SHMT) enzymes [29,30,53] or from sugars via the PK pathway (intermediate in the ethanol production pathway) [16].In the current study, the concentration of m/z 45.031 (t.i.acetaldehyde) was significantly (p < 0.05) higher after 7 days of fermentation by LP100 in the DM with added Thr (DMT), followed by PP100 fermentation in the medium DMT compared to all other media fermented by either LP100, PP100, or LB672 (Figure 3).This suggests that the presence/activity of TA/SHMT enzymes is more likely/higher in LP100 and PP100 strains.However, it is not possible to confirm the presence or absence of TA/SHMT enzymes in LB672 as the highest concentration of ethanol was detected in LB672-fermented media (Figure 2), which suggests that produced acetaldehyde was converted into ethanol by the AlcDH enzyme.The facultative heterofermentative and homofermentative LAB strains, LP100 and PP100, respectively, did produce acetaldehyde, but not ethanol after fermentation in all media studied, which suggests that the AlcDH enzyme is most likely absent/not active in these strains.Notably, acetaldehyde was detected in all media possibly due to the presence of glucose, peptone, and an AA mixture in all DM compositions.
) at 25 • C. Values are presented as mean ± standard error (n = 3).Different superscript lowercase letters represent significant differences between different medium compositions fermented by different LAB strains according to Tukey's test at p < 0.05.

Conclusions
The present study demonstrated that VOCs produced during fermentation analysed by PTR-ToF-MS, HS-SPME-GC-MS, and fastGC-ToF-MS were influenced by DM compositions (AA addition), and LAB strains.The addition of Thr, Met, and Ile AAs to the DM, noticeably impacted the relative concentrations of VOCs after LB672, LP100, and PP100 fermentation, suggesting the presence of specific AA catabolic enzymes in these strains.Understanding how specific VOC generation responds to varying medium compositions and LAB strains will facilitate the production of the VOC required to improve the flavour profiles of plantbased fermentation products or analogue products.Therefore, further investigations utilising different commercial LAB strains are needed to more broadly understand how AAs or medium compositions impact the generation of fermentation-related VOCs.This would provide more information about which LAB strain and which AA combination, or medium compositions work best together to produce target VOCs.

Figure 1 :
Figure 1: Score plot (a) and loadings plot (b) of the principal components (PC) of VOCs produced by three LAB strains (LB672 ( ), LP100 ( ), and PP100 ( )) across different medium compositions (DM , DMA , DMG , DMI , DML , DMM , DMP , and DMT ) at 0 ( ) and 7 ( ) days of fermentation at 25 o C based on the concentrations (ppbV) of finalized 88 m/z from PTR-ToF-MS.The first and second components represented for 38.4% and 14.3% of total variance.

Figure 1 .Figure 1 :
Figure 1.Score plot (a) and loadings plot (b) of the principal components (PC) of VOCs produced by three LAB strains (LB672 (

Figure 2 .
Figure 2. Mean concentration (ppbV) of m/z 47.045 (t.i.ethanol) across different medium compositions after 7 days of fermentation by LB672 (

Figure 2 :
Figure 2: Mean concentration (ppbV) of m/z 47.045 (t.i.ethanol) across differ fermentation by LB672 ( ), LP100 ( ), and PP100 ( ) at 25 o C. Values are Different superscript lowercase letters represent significant differences b fermented by different LAB strains according to Tukey's test at p<0.05.

) at 25 •
C. Values are presented as mean ± standard error (n = 3).Different superscript lowercase letters represent significant differences between different medium compositions fermented by different LAB strains according to Tukey's test at p < 0.05.

Figure 2 :
Figure 2: Mean concentration (ppbV) of m/z 47.045 (t.i.ethanol) across differ fermentation by LB672 ( ), LP100 ( ), and PP100 ( ) at 25 o C. Values are Different superscript lowercase letters represent significant differences b fermented by different LAB strains according to Tukey's test at p<0.05.

) at 25 •
C. Values are presented as mean ± standard error (n = 3).Different superscript lowercase letters represent significant differences between different medium compositions fermented by different LAB strains according to Tukey's test at p < 0.05.

Figure 6 :
Figure 6: Mean concentration (ppbV) of m/z 89.058 (t.i.acetoin, ethyl acetate, and butyric acid) across different medium compositions after 7 days of fermentation by LB672 ( ), LP100 ( ), and PP100 ( ) at 25 o C. Values are presented as mean ± standard error (n=3).Different superscript lowercase letters represent significant differences between different medium compositions fermented by different LAB strains according to Tukey's test at p<0.05.

Funding:
This work was supported by 1. the Accelerating Higher Education Expansion and Development (AHEAD) operation (AHEAD/PhD/R3/Agri/394), a world bank funded project, Ministry of Education, Sri Lanka, 2. the University of Otago doctoral scholarship, 3. the University of Otago postgraduate publishing bursary, 4. Catalyst: Seeding funding was provided by the New Zealand Ministry of Business, Innovation and Employment and administered by the Royal Society Te Apārangi.5.This study was partly carried out within the ON Foods-Research and innovation network on food and nutrition Sustainability, Safety and Security-Working ON Foods and received

Table 1 .
Overview of medium compositions used.

Table 3 .
The mean pH and OD 600 of the samples after 14 days of fermentation by three LAB strains across different medium compositions at 25 • C. Values are the means ± standard error of 2 replicates.Values with different superscript lowercase letters ( a-c ) in each row (either pH or OD 600 column wise) are significantly different according to Tukey's test at p < 0.05.

Table 4 .
VOCs detected after fermentation by three LAB strains in either DM or DM supplemented with individual AAs using HS-SPME-GC-MS at 25 • C.
√: VOCs detected at given LAB-fermented medium.×: VOCs not detected at given LAB-fermented medium.

Table 5 .
VOCs (m/z) detected by PTR-ToF-MS after fermentation by different LAB strains in DM and DM supplemented with individual AAs at 25 • C that significantly (p < 0.05) distinguished between different medium compositions (M), LAB strains (S), and fermentation time (0 and 7 days) (T) and their interaction effects.