Phenyllactic Acid Produced by Geotrichum candidum Reduces Fusarium sporotrichioides and F. langsethiae Growth and T-2 Toxin Concentration.

Fusarium sporotrichioides and F. langsethiae are present in barley crops. Their toxic metabolites, mainly T-2 toxin, affect the quality and safety of raw material and final products such as beer. Therefore, it is crucial to reduce Fusarium spp. proliferation and T-2 toxin contamination during the brewing process. The addition of Geotrichum candidum has been previously demonstrated to reduce the proliferation of Fusarium spp. and the production of toxic metabolites, but the mechanism of action is still not known. Thus, this study focuses on the elucidation of the interaction mechanism between G. candidum and Fusarium spp. in order to improve this bioprocess. First, over a period of 168 h, the co-culture kinetics showed an almost 90% reduction in T-2 toxin concentration, starting at 24 h. Second, sequential cultures lead to a reduction in Fusarium growth and T-2 toxin concentration. Simultaneously, it was demonstrated that G. candidum produces phenyllactic acid (PLA) at the early stages of growth, which could potentially be responsible for the reduction in Fusarium growth and T-2 toxin concentration. To prove the PLA effect, F. sporotrichioides and F. langsethiae were cultivated in PLA supplemented medium. The expected results were achieved with 0.3 g/L of PLA. These promising results contribute to a better understanding of the bioprocess, allowing its optimization at an up-scaled industrial level.


Introduction
Beer is the most consumed alcoholic beverage worldwide and the third most popular drink overall after water and tea. In 2018, beer production in the European Union was estimated to be nearly 406,050 10 8 L and its consumption was calculated to be around 370,092 10 8 L [1]. Barley is the main ingredient in the brewing process and its quality directly influences the characteristics of the final product. However, barley crops can be contaminated by several fungal species belonging to Aspergillus, Penicillium and Fusarium genera [2]. The latter is the most prevalent genus all over the world and the main genus in Europe [3]. Fusarium species are responsible for the production of toxic metabolites called mycotoxins, which are of increasing concern at both health and economic levels [4]. Indeed, recent surveys carried out in Europe have demonstrated that barley crops are frequently contaminated

Effect of Co-Culture between G. candidum and Fusarium Strains on Fungal Growth and T-2 Toxin Concentration
The co-culture experiment consisted of simultaneously inoculating G. candidum and Fusarium strains into Ym medium for different incubation times (ranging from 24 to 168 h) at 22 • C, 150 rpm. For each incubation time, microbial dry weight, T-2 toxin and PLA concentrations were analyzed in control cultures (G. candidum, F. langsethiae 2297 and F. sporotrichioides 186 alone) and in co-cultures. Two co-culture experiments were conducted: G. candidum with F. langsethiae 2297 (Gc/Fl) and G. candidum with F. sporotrichioides 186 (Gc/Fs).
In control cultures, G. candidum dry weight increased during the first 3 days of incubation reaching 3.9 g/L and then slightly decreased to stagnate at 2.3 g/L during the last hours of the experiments. For Fusarium control cultures, fungal biomass increased throughout the whole experimental duration; F. langsethiae 2297 attained a maximum of 3.8 g/L whereas F. sporotrichioides 186 almost reached 3 g/L. In co-culture conditions, where microorganisms were simultaneously inoculated, for the two co-culture experiments (Gc/Fl and Gc/Fs), the total biomass increased during the first 3 days of incubation and then stabilized until the end of the experiment. However, in both co-culture experiments, for each incubation time, the total microbial dry weight was not the sum of dry weights obtained separately in control culture. Thus, co-culture leads to microbial growth reduction without distinguishing the growth of G. candidum from Fusarium species (Figure 1). Figure 2, Panel A, shows that in control culture (F. langsethiae 2297 alone), T-2 toxin was detected from 48 h and the concentration was 99.65 µg/L (±7. 28), and reached 332.7 µg/L (±29.42) after an incubation time of 168 h. In co-culture (G. candidum with F. langsethiae 2297), T-2 toxin was detected from 72 h (20.22 µg/L ± 4.32) and attained 116.44 µg/L (±10.89) after incubation for 168 h. The percentage of T-2 toxin reduction was 100%, 94%, 84% and 65% at 48 h, 72 h, 120 h and 168 h, respectively. These results were similar to those previously obtained for the F. langsethiae 033 strain [40]. The same phenomenon was observed in the second co-culture experiment using F. sporotrichioides 186 strain with slightly different degrees of reduction ( Figure 2, Panel B). In control culture (F. sporotrichioides 186 alone), T-2 toxin was detected from 48 h and the concentration was 82.3 µg/L (±6.1), reaching 294.65 µg/L (±4.74) after incubation for 168 h. As for the first co-culture experiment, in the co-culture G. candidum with F. sporotrichioides 186, T-2 toxin was detected from 72 h (18.8 µg/L ± 6.12) and reached 106.25 µg/L (±3.04) after incubation for 168 h. The percentage of T-2 toxin reduction was 100%, 92%, Toxins 2020, 12, 209 4 of 16 74% and 64% at 48 h, 72 h, 120 h and 168 h, respectively. To ensure that T-2 toxin was not degraded, HT-2 toxin was also monitored and was not detected. experiments. For Fusarium control cultures, fungal biomass increased throughout the whole experimental duration; F. langsethiae 2297 attained a maximum of 3.8 g/L whereas F. sporotrichioides 186 almost reached 3 g/L. In co-culture conditions, where microorganisms were simultaneously inoculated, for the two co-culture experiments (Gc/Fl and Gc/Fs), the total biomass increased during the first 3 days of incubation and then stabilized until the end of the experiment. However, in both co-culture experiments, for each incubation time, the total microbial dry weight was not the sum of dry weights obtained separately in control culture. Thus, co-culture leads to microbial growth reduction without distinguishing the growth of G. candidum from Fusarium species (Figure 1).    Figure 2, Panel A, shows that in control culture (F. langsethiae 2297 alone), T-2 toxin was detected from 48 h and the concentration was 99.65 µ g/L (±7.28), and reached 332.7 µ g/L (±29.42) after an incubation time of 168 h. In co-culture (G. candidum with F. langsethiae 2297), T-2 toxin was detected from 72 h (20.22 µ g/L ± 4.32) and attained 116.44 µ g/L (±10.89) after incubation for 168 h. The percentage of T-2 toxin reduction was 100%, 94%, 84% and 65% at 48 h, 72 h, 120 h and 168 h, respectively. These results were similar to those previously obtained for the F. langsethiae 033 strain [40]. The same phenomenon was observed in the second co-culture experiment using F. sporotrichioides 186 strain with slightly different degrees of reduction ( Figure 2, Panel B). In control culture (F. sporotrichioides 186 alone), T-2 toxin was detected from 48 h and the concentration was 82.3 µ g/L (±6.1), reaching 294.65 µ g/L (±4.74) after incubation for 168 h. As for the first co-culture experiment, in the co-culture G. candidum with F. sporotrichioides 186, T-2 toxin was detected from 72 h (18.8 µ g/L ± 6.12) and reached 106.25 µ g/L (±3.04) after incubation for 168 h. The percentage of T-2 toxin reduction was 100%, 92%, 74% and 64% at 48 h, 72 h, 120 h and 168 h, respectively. To ensure that T-2 toxin was not degraded, HT-2 toxin was also monitored and was not detected.
In both co-culture experiments, the PLA concentration increased rapidly during the first two days of incubation. In the co-culture with F. langsethiae 2297, PLA concentration attained 0.25 g/L (±0.05) at 24 h and 0.46 g/L (±0.06) at 48 h. Afterward, it radically decreased starting at 72 h (0.14 g/L± 0.02) to reach a null value at the end of the incubation time. The same profile was observed in the coculture with F. sporotrichioides 186: PLA concentration attained 0.26 g/L (±0.02) at 24 h and 0.36 g/L (±0.04) at 48 h. PLA concentration in co-culture conditions was inversely proportionate to the T-2 toxin concentration. Indeed, the increase in T-2 toxin concentration was correlated with the reduction of PLA concentration in the medium. When PLA was at its highest level (0.46 g/L in Gc/Fl and 0.36 g/L in Gc/Fs), T-2 toxin was not detected. In both co-culture experiments, the PLA concentration increased rapidly during the first two days of incubation. In the co-culture with F. langsethiae 2297, PLA concentration attained 0.25 g/L (±0.05) at 24 h and 0.46 g/L (±0.06) at 48 h. Afterward, it radically decreased starting at 72 h (0.14 g/L± 0.02) to reach a null value at the end of the incubation time. The same profile was observed in the co-culture with F. sporotrichioides 186: PLA concentration attained 0.26 g/L (±0.02) at 24 h and 0.36 g/L (±0.04) at 48 h. PLA concentration in co-culture conditions was inversely proportionate to the T-2 toxin concentration. Indeed, the increase in T-2 toxin concentration was correlated with the reduction of PLA concentration in the medium. When PLA was at its highest level (0.46 g/L in Gc/Fl and 0.36 g/L in Gc/Fs), T-2 toxin was not detected.

G. candidum Growth and PLA Production Kinetics
The G. candidum strain selected by the IFBM and used in this study produces PLA during the brewing process. To study the growth of this filamentous yeast, Ym medium was initially inoculated with 0.2 g/L of a G. candidum starter culture and incubated at 22 • C, 200 rpm for 5 days. Samples were withdrawn at the starting point and after 6 h, 12 h 24 h, 48 h, 72 h, 96 h, and 120 h of fermentation time and the PLA concentrations were measured.
After 48 h of incubation, the concentration of PLA reached a maximal concentration of 0.41 g/L for 2.25 g/L of yeast dry weight. After 72 h of culture, both G. candidum dry weight and PLA concentration started decreasing, growth went from a maximum of 3.43 g/L (±0.51) to 2.46 g/L (±0.46) and PLA concentration drastically decreased from a maximum of 0.41 g/L (±0.03) to 0.03 g/L (±0.01) (almost 17 times less) (Figure 3, Panel A). Figure 3, Panel B demonstrates the specific production of PLA relative to G. candidum biomass through the fermentation time. It clearly shows that the PLA is highly accumulated in the medium at the early stages of G. candidum growth between 12 and 48 h and then drastically disappears.

G. candidum Growth and PLA Production Kinetics
The G. candidum strain selected by the IFBM and used in this study produces PLA during the brewing process. To study the growth of this filamentous yeast, Ym medium was initially inoculated with 0.  Figure 3, Panel B demonstrates the specific production of PLA relative to G. candidum biomass through the fermentation time. It clearly shows that the PLA is highly accumulated in the medium at the early stages of G. candidum growth between 12 and 48 h and then drastically disappears.

Sequential Cultures
This experiment studied the indirect interactions between G. candidum and the two Fusarium strains. Therefore, the same Ym medium used in Section 2.2 to grow G. candidum was filtrated after 6 h, 12 h, 24 h, 48 h, 72 h, 96 h and 120 h of fermentation time into sterilized Erlenmeyer flasks. Henceforth, the obtained filtrate will be called the "pre-fermented medium" which contains all the metabolites secreted by G. candidum.
In the sequential culture experiment with F. langsethiae 2297, the dry fungal weight was gradually reduced on pre-fermented media up to 48 h (Figure 4, Panel A). The most significant reduction in the F. langsethiae 2297 dry weight occurred in the flasks pre-fermented for 12 h, 24 h and 48 h with a 62%, 72% and 66% reduction percentage, respectively. Beyond 24 h of pre-fermentation, it appeared that F. langsethiae 2297 growth increased slowly. In Ym medium pre-fermented for 120 h, the fungal strain was able to proliferate naturally (3.1 g/L of fungal biomass compared to 3.4 g/L in a non-fermented Ym medium). These results demonstrated that the fungal growth inhibition was more efficient in Ym medium pre-fermented for two days by G. candidum. Previous results showed that the PLA was produced during the early growth phase of the yeast reaching its peak (between 0.25 and 0.41 g/L of PLA) at around 24-48 h of fermentation time. This suggests that the PLA was involved in the reduction of fungal biomass at a rate of 72% (going from 3.4 g/L in a non-fermented medium to 0.95 g/L in 24 h pre-fermented medium). A significant reduction in T-2 toxin concentration was observed for fungal cultures performed in Ym medium pre-fermented from 6 h to 72 h ( Figure 4-Panel B). To ensure that T-2 toxin was not degraded, HT-2 toxin was also monitored and was not detected. Figure 3. PLA concentration (g/L) and G. candidum biomass (g/L) in Ym medium (Panel A) and PLA specific production (g PLA/g dry weight) in Ym medium (Panel B).

Sequential Cultures
This experiment studied the indirect interactions between G. candidum and the two Fusarium strains. Therefore, the same Ym medium used in Section 2.2 to grow G. candidum was filtrated after 6 h, 12 h, 24 h, 48 h, 72 h, 96 h and 120 h of fermentation time into sterilized Erlenmeyer flasks. Henceforth, the obtained filtrate will be called the "pre-fermented medium" which contains all the metabolites secreted by G. candidum.
In the sequential culture experiment with F. langsethiae 2297, the dry fungal weight was gradually reduced on pre-fermented media up to 48 h ( Figure 4, Panel A). The most significant reduction in the F. langsethiae 2297 dry weight occurred in the flasks pre-fermented for 12 h, 24 h and 48 h with a 62%, 72% and 66% reduction percentage, respectively. Beyond 24 h of pre-fermentation, it appeared that F. langsethiae 2297 growth increased slowly. In Ym medium pre-fermented for 120 h, the fungal strain was able to proliferate naturally (3.1 g/L of fungal biomass compared to 3.4 g/L in a non-fermented Ym medium). These results demonstrated that the fungal growth inhibition was more efficient in Ym medium pre-fermented for two days by G. candidum. Previous results showed that the PLA was produced during the early growth phase of the yeast reaching its peak (between 0.25 and 0.41 g/L of PLA) at around 24-48 h of fermentation time. This suggests that the PLA was involved in the reduction of fungal biomass at a rate of 72% (going from 3.4 g/L in a non-fermented medium to 0.95 g/L in 24 h pre-fermented medium). A significant reduction in T-2 toxin concentration was observed for fungal cultures performed in Ym medium pre-fermented from 6 h to 72 h ( Figure 4-Panel B). To ensure that T-2 toxin was not degraded, HT-2 toxin was also monitored and was not detected.  The most significant reduction in T-2 toxin concentration occurred in the flasks pre-fermented for 24 h and 48 h, with a 70% and 56% reduction, respectively. These percentages correlated perfectly with the biomass reduction rate (72% and 66%, respectively). This suggested that the reduction in fungal biomass in the medium is responsible for the reduction in T-2 toxin concentrations. Indeed, specific productions were calculated and demonstrated that the T-2 toxin reduction is correlated to fungal biomass reduction (data not shown).
The same experiment was conducted using F. sporotrichioides 186 ( Figure 5). Fungal growth was drastically reduced in medium pre-fermented for 6 h, 12 h, 24 h and 48 h at almost the same rate of 70% in correlation with the increase of PLA in the medium. As expected, the Ym medium prefermented for 6 h, 12 h, 24 h and 48 h showed an important reduction in T-2 toxin of up to 78%. The equivalence between the growth reduction and the toxin reduction percentages also suggests that it is due to the cessation of fungal growth. One-way ANOVA, Dunnett multiple comparisons post-hoc test, * p-value < 0.05; ** p-value < 0.01; *** p-value < 0.001; ns = not significant).
The most significant reduction in T-2 toxin concentration occurred in the flasks pre-fermented for 24 h and 48 h, with a 70% and 56% reduction, respectively. These percentages correlated perfectly with the biomass reduction rate (72% and 66%, respectively). This suggested that the reduction in fungal biomass in the medium is responsible for the reduction in T-2 toxin concentrations. Indeed, specific productions were calculated and demonstrated that the T-2 toxin reduction is correlated to fungal biomass reduction (data not shown).
The same experiment was conducted using F. sporotrichioides 186 ( Figure 5). Fungal growth was drastically reduced in medium pre-fermented for 6 h, 12 h, 24 h and 48 h at almost the same rate of 70% in correlation with the increase of PLA in the medium. As expected, the Ym medium pre-fermented for 6 h, 12 h, 24 h and 48 h showed an important reduction in T-2 toxin of up to 78%. The equivalence between the growth reduction and the toxin reduction percentages also suggests that it is due to the cessation of fungal growth.
These experiments demonstrated that the interaction between G. candidum and Fusarium strains occurs through a compound released by G. candidum in the medium. As previous results showed, it is highly probable that the PLA, present in G. candidum filtrate is the metabolite responsible for the reduction of fungal dry weight and the subsequent reduction in T-2 toxin concentration. To validate this hypothesis, further experiments using pure PLA compound were required.

Effect of Pure PLA on Fungal Growth and T-2 Toxin Concentration
D-(+)-3Phenyllactic acid was purchased as a pure compound and several concentrations were tested. To validate the results presented in previous sections, PLA solution was prepared at concentrations found at different fermentation times: 0.5 g/L, 0.4 g/L, 0.3 g/L and 0.2 g/L. Lower concentrations of PLA were also tested to determine the minimal inhibitory concentration (MIC): 0.05 g/L and 0.1 g/L. The effect of this pure compound on Fusarium strains growth and its ability to produce T-2 toxin was evaluated.
Both F. langsethiae 2297 growth and T-2 toxin production were highly affected by the addition of D-PLA in Ym medium ( Figure 6). As the concentration of D-PLA increased in the medium, lower fungal mass and lower toxin concentration were quantified. In the control condition (without PLA) F. langsethiae 2297 dry weight was 3.2 g/L (±0.26) and the T-2 toxin concentration was 148 µg/L (±7.8), whereas in the presence of 0.3 g/L of PLA, both dry weight and T-2 concentration were reduced to a rate of 71%, reaching 0.75 g/L (±0.7) and 43.4 µg/L (±1.2), respectively. These experiments demonstrated that the interaction between G. candidum and Fusarium strains occurs through a compound released by G. candidum in the medium. As previous results showed, it is highly probable that the PLA, present in G. candidum filtrate is the metabolite responsible for the reduction of fungal dry weight and the subsequent reduction in T-2 toxin concentration. To validate this hypothesis, further experiments using pure PLA compound were required.

Effect of Pure PLA on Fungal Growth and T-2 Toxin Concentration
D-(+)-3Phenyllactic acid was purchased as a pure compound and several concentrations were tested. To validate the results presented in previous sections, PLA solution was prepared at concentrations found at different fermentation times: 0.5 g/L, 0.4 g/L, 0.3 g/L and 0.2 g/L. Lower concentrations of PLA were also tested to determine the minimal inhibitory concentration (MIC): 0.05 g/L and 0.1 g/L. The effect of this pure compound on Fusarium strains growth and its ability to produce T-2 toxin was evaluated.
Both F. langsethiae 2297 growth and T-2 toxin production were highly affected by the addition of D-PLA in Ym medium ( Figure 6). As the concentration of D-PLA increased in the medium, lower fungal mass and lower toxin concentration were quantified. In the control condition (without PLA) F. langsethiae 2297 dry weight was 3. The same PLA concentrations were tested on F. sporotrichioides 186 and similar results were obtained (Figure 7). The most important reduction occurred in Ym medium supplemented with 0.3 g/L of D-PLA. The same PLA concentrations were tested on F. sporotrichioides 186 and similar results were obtained (Figure 7). The most important reduction occurred in Ym medium supplemented with 0.3 g/L of D-PLA. The same PLA concentrations were tested on F. sporotrichioides 186 and similar results were obtained (Figure 7). The most important reduction occurred in Ym medium supplemented with 0.3 g/L of D-PLA. In both cases, it seems clear that the reduction of T-2 toxin concentration in the medium is directly related to the reduction in fungal growth. Specific production was calculated for each PLA concentration and demonstrated that the T-2 toxin reduction is correlated to the fungal biomass reduction ( Figure 8).  In both cases, it seems clear that the reduction of T-2 toxin concentration in the medium is directly related to the reduction in fungal growth. Specific production was calculated for each PLA concentration and demonstrated that the T-2 toxin reduction is correlated to the fungal biomass reduction (Figure 8). The same PLA concentrations were tested on F. sporotrichioides 186 and similar results were obtained (Figure 7). The most important reduction occurred in Ym medium supplemented with 0.3 g/L of D-PLA. In both cases, it seems clear that the reduction of T-2 toxin concentration in the medium is directly related to the reduction in fungal growth. Specific production was calculated for each PLA concentration and demonstrated that the T-2 toxin reduction is correlated to the fungal biomass reduction ( Figure 8).

Discussion
The contamination of food raw material by fungal species has many consequences. In addition to the alteration of commodities, the loss of nutritional qualities, and the strong reduction in yield, fungal development can lead to the accumulation of toxic compounds such as mycotoxins. In France, the occurrence of several Fusarium species in barley crops intended for brewing has become a source of concern over the past ten years. In barley crops, the introduction of F. sporotrichioides and F. langsethiae has been recently observed, progressively replacing F. poae [41][42][43]. The risk associated with these Fusarium species is the production of T-2 toxin, the most toxic compound in the type-A trichothecenes family. During the brewing process, the malting step provides the best conditions (22 • C and high humidity) for Fusarium development and T-2 toxin production [30]. Currently, G. candidum is used during the brewing process to reduce T-2 toxin contamination. However, its efficiency is variable and the mechanisms of interaction between G. candidum and Fusarium species are still unknown. Previously, Gastélum-Martinez et al. (2012) used the co-culture method between those two microorganisms and demonstrated that the direct interaction between G. candidum and F. langsethiae 033 led to a drastic T-2 toxin concentration reduction (93% in comparison to the control culture) [40]. To decipher the mechanism of interaction that lead to this reduction, in the presented study, the two microorganisms were also cultivated sequentially. First, G. candidum was cultivated, and removed from the medium before Fusarium strain inoculation. Several incubation times for G. candidum culture were tested (0 to 146 h) and Fusarium incubation was always 7 days. Results obtained in these sequential cultures show a reduction in the T-2 toxin concentration linked to a reduction in fungal growth. In addition, the reduction in T-2 toxin concentration varies according to the medium pre-fermentation time by G. candidum. In sequential cultures, T-2 toxin concentrations are inversely correlated with the production of PLA by G. candidum, demonstrating that the mechanism leading to T-2 toxin reduction, was linked directly to the PLA concentrations in the medium. Indeed, while the PLA concentration was at its highest level after 48 h of pre-fermentation time, the T-2 toxin concentration was at its lowest. The correlation between G. candidum growth evolution and PLA concentration in the medium suggests that PLA is a primary metabolite as it is secreted during the growth phase (from 0 h to 48 h, the PLA concentration varied from 0 to 0.41 g/L) and then gradually disappeared from the culture media. PLA biosynthesis is not yet described in G. candidum but well described in lactic acid bacteria (LAB) strains, which can produce large amounts of PLA. In fact, in lactic acid bacteria, the PLA results from amino acid metabolism of phenylalanine and α-ketoglutarate. In a glucose, citric acid or fructose enriched medium, the phenylalanine amino acid group is transferred to α-ketoglutarate under the action of aromatic amino acid transferase (AAT), leading to the formation of phenylpyruvic acid (PPA), an intermediate to PLA. Depending on the type of lactate dehydrogenases (L-LDH or D-LDH) present in lactic acid bacteria, PPA is converted to either L-PLA or D-PLA [44][45][46]. A potential PLA synthesis pathway is explicitly demonstrated by Chaudhari and Gokhale (2016) and simplified in Figure 9 [47]. Studies have shown that the D form of PLA is more effective as an antimicrobial compound than the L form [37].
°C and high humidity) for Fusarium development and T-2 toxin production [30]. Currently, G. candidum is used during the brewing process to reduce T-2 toxin contamination. However, its efficiency is variable and the mechanisms of interaction between G. candidum and Fusarium species are still unknown. Previously, Gastélum-Martinez et al. (2012) used the co-culture method between those two microorganisms and demonstrated that the direct interaction between G. candidum and F. langsethiae 033 led to a drastic T-2 toxin concentration reduction (93% in comparison to the control culture) [40]. To decipher the mechanism of interaction that lead to this reduction, in the presented study, the two microorganisms were also cultivated sequentially. First, G. candidum was cultivated, and removed from the medium before Fusarium strain inoculation. Several incubation times for G. candidum culture were tested (0 to 146 h) and Fusarium incubation was always 7 days. Results obtained in these sequential cultures show a reduction in the T-2 toxin concentration linked to a reduction in fungal growth. In addition, the reduction in T-2 toxin concentration varies according to the medium pre-fermentation time by G. candidum. In sequential cultures, T-2 toxin concentrations are inversely correlated with the production of PLA by G. candidum, demonstrating that the mechanism leading to T-2 toxin reduction, was linked directly to the PLA concentrations in the medium. Indeed, while the PLA concentration was at its highest level after 48 h of pre-fermentation time, the T-2 toxin concentration was at its lowest. The correlation between G. candidum growth evolution and PLA concentration in the medium suggests that PLA is a primary metabolite as it is secreted during the growth phase (from 0 h to 48 h, the PLA concentration varied from 0 to 0.41 g/L) and then gradually disappeared from the culture media. PLA biosynthesis is not yet described in G. candidum but well described in lactic acid bacteria (LAB) strains, which can produce large amounts of PLA. In fact, in lactic acid bacteria, the PLA results from amino acid metabolism of phenylalanine and α-ketoglutarate. In a glucose, citric acid or fructose enriched medium, the phenylalanine amino acid group is transferred to α-ketoglutarate under the action of aromatic amino acid transferase (AAT), leading to the formation of phenylpyruvic acid (PPA), an intermediate to PLA. Depending on the type of lactate dehydrogenases (L-LDH or D-LDH) present in lactic acid bacteria, PPA is converted to either L-PLA or D-PLA [44][45][46]. A potential PLA synthesis pathway is explicitly demonstrated by Chaudhari and Gokhale (2016) and simplified in Figure 9 [47]. Studies have shown that the D form of PLA is more effective as an antimicrobial compound than the L form [37]. Several studies have been conducted on LAB and more precisely, on the Lactobacillus genus, which is frequently involved in their antifungal activity [48][49][50]. Lactobacillus strains and L. plantarum in particular, have been found to produce PLA in sourdough bread. The use of these strains is a means of natural food preservation. Indeed, studies have shown that they improve the shelf life of bread and bakery products by decreasing and/or inhibiting fungal activities. PLA is considered one Several studies have been conducted on LAB and more precisely, on the Lactobacillus genus, which is frequently involved in their antifungal activity [48][49][50]. Lactobacillus strains and L. plantarum in particular, have been found to produce PLA in sourdough bread. The use of these strains is a means of natural food preservation. Indeed, studies have shown that they improve the shelf life of bread and bakery products by decreasing and/or inhibiting fungal activities. PLA is considered one of the responsible inhibitory compounds along with lactic acid and acetic acid [51]. To our knowledge, no studies have been carried out on PLA metabolism and its toxicity effect in the human body. In 2002, Lavermicocca et al. (2003) studied the fungicidal activity of PLA on 23 fungal strains belonging to Aspergillus, Penicillium and Fusarium genera. Among these strains, 90% showed at least a 50% growth inhibition at PLA concentrations lower than 7.5 g/L. Other strains presented a growth delay of at least three days [52]. Dieuleveux et al. have proved that PLA produced by G. candidum strains at 20 g/L also has antibacterial activity against L. monocytogenes, S. aureus, E. coli and A. hydrophila [37,38,53].
Fusarium strains used in this study were more susceptible to PLA that those tested by Lavermicocca et al. (2003). Indeed, F. langsethiae 2297 growth was drastically reduced (72%) when it was exposed to 0.2 g/L of PLA, whereas F. sporotrichioides 186 growth was slightly reduced (47%) when it was exposed to 0.3 g/L of PLA. Thus, there is relevant variability in susceptibility among fungal species. Although the antimicrobial action mechanism is still not elucidated, some suggest that the PLA causes the bacteria to form aggregates with the secretion of polysaccharides described as a "response to the attack". Indeed, as the concentration of PLA increased, a larger amount of polysaccharides were found in the medium and alteration in cell wall rigidity after only 27 h of incubation was observed leading to cell death [38,47]. In this study, as indicated by the specific production of T-2 toxin obtained for each fungal strain, the reduction in the concentration of T-2 toxin is correlated with the reduction in fungal growth. However, in some cases, the inhibition of fungal growth by sub-lethal concentrations of fungicide or some natural products enhances mycotoxin production [54][55][56]. This must be taken into account in the development of biocontrol strategies.
In this study, to provide an explanation for the phenomenon of T-2 toxin concentration control during the malting process previously observed by the IFBM, in vitro experiments were carried out under environmental conditions close to those of the brewing process. Currently, the filamentous yeast is added in a freeze-dried form (100 g per 25 tons of barley) directly into the barley steeping water for at least 10 h. Then, the water is discarded and the steeped barley remains at rest for 3 to 5 days at 16-20 • C. This stage is the most critical step in the brewing process because the operating conditions favor Fusarium growth and T-2 toxin production. Results demonstrate that the reduction in Fusarium contamination and T-2 toxin during the malting process is due to the PLA produced by G. candidum. Based on the results of this study, in order to develop an effective biocontrol method to use G. candidum, preparation of the strain seems essential to activate the PLA production metabolism. Mu et al. developed a medium favorable to PLA production by Lactobacillus sp. strains, highly enriched with glucose, phenylpyruvic acid (phenylalanine intermediate in the PLA biosynthesis pathway) and yeast extract [46]. This medium significantly enhanced Lactobacillus sp. proliferation, and thus PLA yield. However, the use of such broth on an industrial level does not seem to be applicable for several reasons. On one hand, it may alter the organoleptic characteristics of the final product. On the other, using these components in large amounts would have a considerable economic impact on the industry. Consequently, it seems important to combine optimized growth factors (G. candidum activation medium and initial concentration, fermentation duration, temperature, water activity, rotation speed, oxygenation levels, etc.) to enhance PLA production naturally, and to develop an ecofriendly, toxin-free beer product. Moreover, the presence of PLA during the malting step not only helps to reduce Fusarium flora and consequently, to reduce T-2 toxin concentration, but it also improves the organoleptic properties of the final beer product [36,57,58]. This study demonstrates for the first time, the role of PLA as a biocontrol agent in reducing T-2 toxin concentration.

Reagents and Chemicals
T-2 toxin and phenyllactic acid (PLA) were purchased from Sigma-Aldrich (Saint-Quentin-Fallavier, France). Stock solutions were prepared in dimethylsulfoxyd (DMSO) and acetonitrile-water (30:70 v/v) mixture, respectively, and stored at −18 • C until use. Solvents used for T-2 toxin extraction and high-performance liquid chromatography (HPLC) were analytical grade quality and purchased from Thermo-Fisher Scientific (Illkirch, France). Ultrapure water used for HPLC was purified at 0.22 µm by an ELGA purification system (ELGA LabWater, High Wycombe, United Kingdom).

Strains, Media and Culture Conditions
In this study, two Fusarium strains were used: F. sporotrichioides 186 and F. langsethiae 2297. Both strains were previously isolated from contaminated barley kernels and were kindly provided by the French Institute of Brewing and Malting (IFBM). The filamentous yeast Geotrichum candidum is already used as a biocontrol agent during the malting process (IFBM Malting Yeast ®, DMS food specialties, La Ferté sous Jouarre, France) and was purchased from DSM Food Specialties.
Fusarium pre-cultures were performed on potato dextrose agar medium (PDA 39 g/L) and incubated at 22 • C for 7 days. Cultures were then used to induce sporulation or conserved at 4 • C. Fusarium strains sporulation was induced in carboxymethylcellulose (CMC) liquid medium (CMC: carboxymethylcellulose 15 g/L; yeast extract 1 g/L; MgSO 4 7H 2 O 0.5 g/L; NH 4 NO 3 1g/L; KH 2 PO 4 1 g/L). Briefly, at least 15 plugs of each Fusarium strain from a seven-day-old solid pre-culture were inoculated in 150 mL of CMC medium and incubated in an orbital shaker set at 22 • C at 150 rpm for 15 days in the dark. At the end of the incubation time, the solution was filtrated using sterilized Mira cloth. Spores were counted on Thoma cell counting chamber and ultimately used to inoculate culture during further experiments or conserved in 40% glycerol at −80 • C.
G. candidum strain was supplied in freeze-dried form, thus a pre-culture was essential to revivify it prior to experimental use. A 24 g/L culture was prepared in 250 mL of yeast and malt (Ym) liquid medium (Ym: glucose 5 g/L; yeast extract 1.5 g/L; malt extract 1.5 g/L; peptone salt 2.5 g/L pH 7) and incubated in an orbital shaker set at 22 • C at 150 rpm for 24 h. At the end of the incubation time, this culture was used as a starter culture.
Ym liquid medium was used during all experiments (co-cultures and sequential cultures) to elucidate the interaction mechanisms between G. candidum and Fusarium strains.

Kinetic of PLA Production by G. candidum
In an Erlenmeyer flask, 150 mL of Ym medium was inoculated with G. candidum starter culture with a final concentration adjusted at 0.2 g/L and then incubated in an orbital shaker set at 22 • C at 150 rpm for different fermentation times ranging from 6 h to 120 h. At the end of the fermentation time, the medium was aseptically divided into two volumes. First, 50 mL were used to evaluate G. candidum growth by measuring the dry weight and PLA concentration by HPLC-DAD at each sampling time. The remaining 100 mL was aseptically filtered to eliminate G. candidum cells, leaving only its excreted metabolites in the medium. The medium nutrients were then adjusted according to the volume and the pH was adjusted at 7. These volumes were used during the sequential cultures experiments and are henceforth referred to as pre-fermented medium. Experiments were conducted four times in triplicate.

Co-Culture of Fusarium Strains and G.candidum
Erlenmeyer flasks containing 150 mL of Ym medium were inoculated with G. candidum starter culture at the final concentration of 0.2 g/L. Then, F. langsethiae 2297 or F. sporotrichioides 186 was inoculated at a final concentration of 10 6 spores/mL in their respective flasks. For control conditions, each microorganism was inoculated alone at the same concentrations. Cultures were incubated in an orbital shaker set at 22 • C at 150 rpm. Several incubation times were tested: 24 h, 48 h, 72 h, 120 h and 168 h. At the end of all sampling times for all culture conditions the total dry weight, PLA and T-2 toxin concentration were evaluated. All experiments were conducted twice in duplicate.

Sequential Cultures of Fusarium Strains and G.candidum
For sequential cultures, 100 mL of pre-fermented Ym medium at different fermentation times ranging from 6 h to 120 h (used in Section 4.3) were inoculated with F. langsethiae 2297 or F. sporotrichioides 186 at the final concentration of 10 6 spores/mL in their respective flasks. Cultures were incubated in an orbital shaker set at 22 • C at 150 rpm for 7 days. For the control condition, F. sporotrichioides 186 or F. langsethiae 2297 were inoculated in a non-fermented Ym liquid medium at the same concentrations. At the end of the incubation time, fungal growth was evaluated by measuring the dry weight and T-2 toxin concentration by HPLC-DAD. All experiments were conducted twice in duplicates.

Phenyllactic Acid Effect on F. sporotrichioides 186 and F. langsethiae 2297 Growth and T-2 Toxin Concentration
To confirm that PLA is the metabolite produced by G. candidum, which is involved in Fusarium growth reduction and T-2 toxin concentration reduction, fungal cultures were conducted in Ym liquid medium supplemented with PLA. PLA standard stock solution was prepared at 40 mg/mL in a mixture of acetonitrile/water (30/70, v/v) and appropriate volumes of PLA stock solution were added in order to obtain several different concentrations: 0.05 g/L; 0.1 g/L; 0.2 g/L; 0.3 g/L; 0.4 g/L and 0.5 g/L in Erlenmeyer flasks containing 100 mL of Ym liquid medium. Then, F. langsethiae 2297 or F. sporotrichioides 186 was inoculated at the final concentration of 10 6 spores/mL in their respective flasks. Cultures were incubated in an orbital shaker set at 22 • C at 150 rpm for 7 days. At the end of the incubation time, Fusarium strains' growth was evaluated by measuring the dry weight and T-2 toxin concentration by HPLC-DAD. PLA dilutions were prepared to add only 75 µL of acetonitrile in the culture medium, this concentration having been identified as a no-effect dose on both fungal growth and T-2 toxin concentration. Control cultures were performed by adding only 75 µL of acetonitrile to the medium.

G. candidum, F. sporotrichioides 186 and F. langsethiae 2297 Biomass Evaluation
To estimate microorganism growth during the incubation period, vacuum filtration was performed to determine the dry weight (g/L). First, cellulose nitrate filters (pore size 0.45 µm, Sartorius Stedim Biotech, Goettingen, Germany) were left to dry overnight in an oven set at 105 • C. Afterward, 10 mL of culture medium were vacuum-filtered at each sampling time and filters were then incubated at 105 • C for 24 h. The microorganism dry weight refers to the difference between filters post-filtration and pre-filtration.

PLA Quantification
At each sampling time, 1 mL of culture media was withdrawn and filtrated through 0.45 µm PTFE syringe filters (Thermo Scientific Fisher, Villebon-Sur-Yvette, France) to eliminate microorganisms from the supernatant prior to injection into HPLC apparatus. Analyses of PLA were performed using a Luna C18(2) column (5 µm, 250 × 4.6 mm) and a pre-column with the same characteristics (Phenomenex, Torrance, CA, USA). The detection of PLA was performed using a Dionex Ultimate 3000 UHPLC system coupled with a diode-array detector (DAD) set at 210 nm (Thermo Fisher Scientific, Illkirch, France). The analysis was performed in a gradient mode using acidified water (0.2% of acetic acid glacial) as solvent A and pure HPLC grade acetonitrile as solvent B. Flow was set at 1.2 mL/min with A/B ratios of 90:10, 50:50, 50:50, 0:100 and 90:10, with run times of 0.0, 4.0, 9.0, 10.0 and 15.0 min, respectively. Injection volume was set at 50 µL. PLA quantification was calculated according to a standard calibration curve with concentrations ranging between 10 and 1000 mg/L.

T-2 Toxin Extraction and Quantification
After the incubation period, cultures were filtrated with Nalgene™ Rapid-Flow™ Filters of 0.45 µm pore size (Thermofischer Scientific, Waltham, MA, USA) to remove microorganisms. Filtrates were then extracted with 70 mL of ethyl acetate and shaken on a Universal Shaker SM 30 B Control Edmund Bühler ® (Thermofischer Scientific, Waltham, MA, USA) set at 150 rpm overnight. The organic phase was recovered and evaporated until dry under a rotavapor set at 60 • C. Samples were resuspended with 2 mL of acetonitrile/water (30/70, v/v) mixture and filtered through 0.45 µm PTFE syringe filters (Sigma Aldrich, St. Quentin Fallavier, France). Samples were conserved at 4 • C until further analysis. T-2 toxin was analyzed by Gemini C18 columns, 150 mm × 4.6 mm, 3 µm and a pre-column with the same characteristics (Phenomenex). As for PLA, T-2 toxin was detected and quantified using HPLC-DAD (Dionex, Sunnyvale, CA, USA) according to the methodology described by Medina et al. [59]. T-2 toxin quantification was calculated according to a standard calibration curve with concentrations ranging between 0.2 and 50 µg/mL.

Statistical Analysis
First, the normal distribution of data was tested by the Shapiro-Wilk test. Then, one-way analysis of variance (ANOVA) followed by Dunnett's multiple comparisons test was used to analyze the effect of PLA on F. langsethiae 2297 and F. sporotrichioides 186 growth and their T-2 production. One-way ANOVA followed by a Tukey's multiple comparisons test was used to analyze the differences between control and co-culture or sequential culture conditions. The statistical analysis of data was carried out with GraphPad Prism 8 software (GraphPad Software, La Jolla, CA, USA). Differences were considered to be statistically significant when the p-value was lower than 0.05. Graphical values are represented by mean ± standard deviation (SD).