Correspondence: Jean-Paul Vernoux, Research Unit Aliments Bioprocédés Toxicologie Environnements (URABTE) E.A. 4651, IFR146 ICORE, Université de Caen Basse-Normandie, Esplanade de la paix, 14032 CAEN cedex, France. Tel.: +332 31 56 56 21; fax: +332 31 56 61 79; e-mail: email@example.com
Geotrichum candidum ATCC 204307 was previously found to generate phenyllactic acid (PLA) and indoleacetic acid (ILA) in complex culture media. In this study, a relationship between concentrations of PLA, ILA, and hydroxy PLA (OH-PLA) and initial concentrations of phenylalanine, tryptophan, and tyrosine, added respectively as unique sources of nitrogen in synthetic medium, was established. Phenylpyruvic acid (PPA), an intermediate compound of PLA metabolism, was able to induce not only PLA but also phenylethyl alcohol (PEA) production when used separately as initial substrate. Under pH, temperature, and salt concentrations used for cheese-making, phenylalanine was found to be the most efficient substrate for antimicrobial metabolite production. In excess of substrate, different yeast strains of Geotrichum candidum, Yarrowia lipolytica, Candida natalensis, and Candida catenulata were shown here to produce 1.6 ± 0.5–5.0 ± 0.2 mM of PLA from phenylalanine, 5.0 ± 0.1–10.9 ± 0.3 mM of ILA from tryptophan, and 1.3 ± 0.3–7.0 ± 0.02 of PLA and 0.1 ± 0.0–2.22 ± 0.09 mM of PEA from PPA. Geotrichum candidum ATCC 204307 was the highest producer. This is the first time these antimicrobial metabolites PLA, OH-PLA, ILA, and PEA are being reported as the reaction products of aromatic amino acids catabolism in G. candidum.
Geotrichum candidum has been used for a long time in the dairy industry as a starter or adjunct culture (Boutrou et al., 2006). Various enzymatic activities of G. candidum contribute in soft cheese maturation, texture, flavor, aroma, and safety of soft cheeses by producing different aromatic and bioactive metabolites during the processes of ripening (Spinnler et al., 2001; Collins et al., 2003; Sacristán et al., 2012).
Two antimicrobial compounds produced by G. candidum are phenyllactic acid (PLA) and indoleacetic acid (ILA). PLA and ILA have been reported to specifically inhibit the growth of Listeria monocytogenes in complex media TSBYE by inducing behavioral and structural changes in this food-borne pathogen (Dieuleveux & Guéguen, 1998; Dieuleveux et al., 1998a). The inhibitory potential of the commercially produced PLA has been demonstrated against both Gram-positive and Gram-negative bacteria, such as Staphylococcus aureus, Bacillus cereus, and Escherichia coli O157 : H7 (Dieuleveux et al., 1998a, b; Ohhira et al., 2004). PLA and a related compound hydroxy PLA (OH-PLA) can exhibit a broad range of inhibitory activity against a variety of fungal species including some mycotoxigenic species, namely Aspergillus ochraceus, Penicillium roqueforti, and Penicillium citrinum (Lavermicocca et al., 2000). A fourth metabolite, phenyl ethyl alcohol (PEA), is responsible for the aromatic ‘rose’ character of soft cheeses (Lee & Richard, 1984) and can also induce an autoantibiotic effect in Candida albicans (Lingappa et al., 1969). Additionally, the bactericidal action of PEA in Gram-positive and Gram-negative species, for example S. aureus and E. coli, has been associated with cytological changes, including membrane damage, and inhibition of RNA and protein synthesis (Lucchini et al., 1990).
Whereas knowledge of the antimicrobial properties of these compounds already exists, their production and/or metabolic pathways have not been described so far in G. candidum. To evaluate the potential of G. candidum as a bioprotective microorganism involved in the production of antimicrobial compounds, the influence of the three aromatic amino acids, phenylalanine, tyrosine, and tryptophan and of phenylpyruvic acid in synthetic medium under varied experimental conditions was determined in this study.
Materials and methods
D-phenylalanine (Phe), D (+)-3-phenyllactic acid (PLA), phenylpyruvic acid in the chemical form of sodium phenylpyruvate (PPA), β-phenylethylalcohol (PEA), L-tyrosine (Tyr), DL-hydroxy PLA (OH-PLA), DL-tryptophan (Trp), Indole-3-pyruvic acid (IPA), DL-3-indoleacetic acid (ILA), L-glycine (Gly), HPLC gradient grade methanol ChromaSOLV®, and trifluoroacetic acid (HPLC grade) were purchased from Sigma-Aldrich. The glacial acetic acid was obtained from Carlo Erba Reagents, France.
Media and culture conditions
Nine yeast strains were used in these experiments. D1/D2 domains and ITS rRNA fragments were sequenced for all UCMA strains and are available in the GenBank database. Geotrichum candidum ATCC 204307, UCMA 4574 (GenBank: KC542308 and KC542316), 3627 (GenBank: KC542305 and KC542313), 4057 (GenBank: KC542307 and KC542315), 482 (GenBank: KC542304 and KC542312), and 281 (GenBank: KC542303 and KC542311) were isolated from Pont l'Evêque cheese (Dieuleveux et al., 1997). Three other yeast species, Yarrowia lipolytica UCMA3681 (GenBank: KC542309 and KC542317), Candida natalensis UCMA3722 (GenBank: KC542306 and KC542314), and Candida catenulata UCMA3690 (GenBank: KC542310 and KC542318), were also tested.
The synthetic media (SM) were a modified version of mineral media used by Adour et al. (2002) and were composed of 15 g L−1 of glucose, 3.4 g L−1 of potassium phosphate (monobasic), 3.45 g L−1 of sodium phosphate (monobasic monohydrate), 0.51 g L−1 of magnesium sulfate (anhydrous), and 10 mg L−1 of iron sulfate (7H2O).
Yeast strains obtained from agar slope stocks were pregrown in 50 mL of Malt extract broth (Malt extract 20 g L−1, glucose 20 g L−1, and peptone 1 g L−1) pH 5.7 ± 0.2 at 25 °C for 48 h under shaking (Novotron, VWR and 120 rpm). The pellet was obtained by centrifuging the preculture for 20 min at 20 000 g, washed with the same volume of saline (0.9% NaCl), and suspended in 5 mL of tryptone salt broth (tryptone 1 g L−1, sodium chloride 8 g L−1, pH 7.2) to obtain 108 cfu mL−1. The cell suspension was used to inoculate 6-well microplates (test plates 15 mL, TPP) containing synthetic media (6 mL per well) at one percent dilution to have 106 cfu mL−1. The development of yeast strains was effective in these microplates when kept under agitation at 25 °C on a microplate shaker (Wesbart) at 120 rpm min−1. Under all experimental conditions except incubation at 5 and 15 °C, a 1-log increase in growth from 106 to 107 cfu mL−1 was observed following an incubation of 72 h.
To assess the influence of different substrates on PLA or ILA synthesis, 0.1 to 5.0 g L−1 of substrates were added to SM in five separate experiment sets including phenylalanine (Phe-SM), phenylpyruvic acid (PPA-Gly-SM), tyrosine (Tyr-SM), tryptophan (Trp-SM), or indole pyruvic acid (IPA-Gly-SM) (corresponding to 0.6 to 30.2 mM for Phe or 30.4 PPA mM, 27.5 mM for tyrosine, and 0.4 to 24.4 mM for Trp or IPA). Glycine (1 g L−1) was used as the nitrogen source of the PPA-Gly-SM and/or IPA-Gly-SM and in negative control medium (Gly-SM). The initial pH of the synthetic medium was always set to 5.7 ± 0.2, and cultures were incubated at 25 °C unless specified. Samples were taken at 0, 12, 24, 48, and 72 h.
To assess the effects of temperature, pH, and salt concentration, media were prepared with 0.5 g L−1 of the respective amino acid (corresponding to 3.0 mM of Phe or PPA and 2.4 mM of Trp). For temperature study, cultures were subjected to shaking in thermostatic rooms at 5 and 25 °C or in an incubator (Weiss, WK1) at 15 °C. Initial pH of the media (pH range tested: 4–9) was adjusted with 10% NaOH or 10% HCl. For salt experiments, 1, 3, 7, and 10 g L−1 of NaCl were added to the medium. All the experiments were conducted thrice, with two replicas per sample each time.
Preparation of samples
Culture supernatant was used to analyze the extracellular production of the metabolites. 500 μL of culture of each well was microfiltered using the centrifugal filters (VWR, 0.2 μm) at 12 000 g for 5 min at 25 °C. 450 μL of the filtrate was then diluted with equal volume of SM (without any added amino acid) at pH 2.0.
A modified version of the method reported by Valerio et al. (2004) was employed. Briefly, 20 μL of each sample was injected on an HPLC-UV system (Waters e2695) using a Waters Sunfire® C18 Column (150 mm, 4.6 mm, 3.5 μm) connected to a single-channel Photodiode array detector (Waters 2998). The solvent system for Phe, PPA, PLA, OH-PLA, and PEA was solvent A: methanol-0.05% TFA and solvent B: water-0.05% TFA. The mobile phase was a gradient from 10% to 55% A for 25 min with a flow rate of 0.6 mL min−1. The samples were monitored using a single UV detector at 210 nm. The solvent system for Trp, IPA, and ILA was 72% solvent A (1% acetic acid) and 28% solvent B (100% methanol) with a flow rate of 0.8 mL min−1 for 40 min (Chung et al., 2003). The peaks were detected at 280 nm. The calibration curves were linear (R2 > 0.98), and the limit of detection (LOD) was close to 50 μg L−1.
Experimental data were analyzed using Matlab 6.5. The different metabolite productions were compared using anova test to obtain P-values.
Results and discussion
PLA production in synthetic media containing phenylalanine (Phe-SM)
Geotrichum candidum ATCC204307 started producing PLA within first 12-h incubation, and by the end of 72 h, a minimum of 0.50 ± 0.01 mM and a maximum of 5.2 ± 0.7 mM of PLA were produced in the medium containing increasing phenylalanine concentrations (Fig. 1). For smaller substrate concentrations (0.6–3.0 mM), 67.4 ± 3.8 to 76.6 ± 1.4% of the consumed Phe was transformed into PLA (Fig. 4). A substrate saturation phenomenon was observed as the amino acid concentration increased, and only 17.6 ± 2.4% of Phe (30.2 mM) was converted to PLA.
In addition to PLA, small quantities of PPA (from 0.03 ± 0.01 to 0.10 ± 0.02 mM) in the presence of 0.6–30 mM of Phe were also found to be present in the culture media hinting at PPA being present as the intermediate reaction product. It is reminiscent of similar observations made in Saccharomyces cerevisiae and lactic acid bacteria (LAB) (Yvon & Rijnen, 2001) where phenylalanine is changed to phenylpyruvate through a transamination reaction, which is then reduced to phenyllactate (Dickinson et al., 2003). No PEA production from Phe was observed, which is in accordance with results previously reported for G. candidum (Lee & Richard, 1984).
The linear relationship found between the lower concentrations of Phe and PLA (Fig. 1 insert) and the fact that no PPA or PLA was produced in the absence of Phe in negative control samples (Gly-SM) indicate that these compounds are essentially the result of G. candidum's enzymatic degradation of phenylalanine. It has been previously reported also for different Lactobacilli and S. cerevisiae (Yvon & Rijnen, 2001; Dickinson et al., 2003; Lavermicocca et al., 2003; Curtin & McSweeney, 2004; Li et al., 2007). The metabolite production in G. candidum ATCC204307 is far higher compared with only 0.55 mM PLA in the case of Lactobacillus sp. SK007 (Li et al., 2007), 0.57 mM reported in the case of Leuconostoc mesenteroides subsp. mesenteroides ITMY30 (Valerio et al., 2004), or 0.65 ± 0.05 mM in Pediococcus acidilactici DSM20284 (Mu et al., 2012). The substrate consumption ratio was also higher as Phe was readily consumed by the yeast cells, while 96% of the Phe remained unutilized in the case of Lactobacillus sp. SK007 even after 96 h and in the presence of smaller substrate concentrations in MRS medium (Li et al., 2007). However, certain strains of Lactobacillus plantarum (1073, 778, and 1081) are capable of producing PLA in higher amounts (2.6 ± 0.2, 4.1 ± 0.4, and 5.2 ± 0.5 mM) in MRS medium (Gerez et al., 2010), comparable with that produced here by G. candidum ATCC204307.
PLA production in synthetic medium containing phenylpyruvate (PPA-Gly-SM)
Increasing quantities of PLA (0.4 ± 0.3 to a maximum of 7.3 ± 0.7 mM) were produced when the medium was supplemented with various concentrations (0.6–30.4 mM) of PPA as the initial substrate (Fig. 2). While the substrate–product relationship was linear at lower concentrations (Fig. 2 insert), a substrate saturation phenomenon was observed for higher concentrations. However, large percentages of unconsumed substrate (PPA) remained in the medium, even for smaller concentrations (Fig. 4). PLA was not detected in the absence of PPA in negative control (Gly-SM). The partial degradation of PPA could be explained by the probability either that the enzymes responsible for the conversion are not fully functional under our culture conditions as was the case for Lactococci (Gao et al., 1997), or that G. candidum ATCC204307 has difficulty to properly internalize the substrate (PPA) in the minimal medium used here ('Effect of different physical parameters').
This confirms that phenylpyruvic acid (PPA) can act as an intermediate compound of PLA metabolism in G. candidum ATCC204307. In bacteria, PPA has been shown to be implicated in PLA formation from Phe in Lactobacillus sanfranciscensis DSM20451T, L. plantarum TMW1.468 (Vermeulen et al., 2006), lactic acid bacteria (Valerio et al., 2004), Lactobacillus sp. SK007 (Li et al., 2007), and P. acidilactici DSM20284 (Mu et al., 2012) where PLA formation had increased some 14-fold from PPA. However, in the present work, it has been shown that under the given experimental conditions, PPA metabolism in G. candidum ATCC204307 is slower.
OH-PLA production in synthetic media containing tyrosine (Tyr-SM)
In the presence of increasing concentrations of tyrosine in SM, G. candidum ATCC 204307 produced increasing concentrations of hydroxy PLA (OH-PLA) up to a maximum of 2.3 ± 0.1 mM for substrate excess of 5 g L−1 (data not shown). This substance was not previously detected in G. candidum but reported in lactic acid bacteria (Valerio et al., 2004). OH-PLA having a similar antimicrobial spectrum to PLA and being produced at lower yield, no further investigations were carried out.
ILA production in synthetic media containing tryptophan (Trp-SM)
Trp has also been confirmed here to be the amino acid responsible for the production of the second important antilisterial metabolite, ILA. Trp was readily taken up by the yeast cells, and its conversion to ILA began within first 12 h of incubation yielding from 0.60 ± 0.02 to a maximum of 12.3 ± 1.4 mM of ILA. The amino acid was completely utilized within 24 h of incubation for smaller substrate concentrations (Fig. 3). A maximum of 60.9 ± 6.6% of the tryptophan remained unconsumed (Fig. 4), and substrate inhibition effect was observed at 24.4 mM. The relationship between Trp and ILA concentrations was linear at lower concentrations (Fig. 3 insert), and no ILA was produced in negative control samples (Gly-SM).
The results are in agreement with the findings in Lactobacillus casei and Lactobacillus helveticus that produced ILA from Trp through a reversible reaction (Gummalla & Broadbent, 1999). However, contrary to Phe metabolic reaction, no intermediate product was detected at any stage in the culture of G. candidum ATCC204307. Gao et al. (1997) have shown that Lactococcus lactis catabolized Trp by aminotransferase to indole pyruvic acid (IPA) under conditions found in ripening cheese (pH 5.2, 4% NaCl). IPA has also been reported as the sole intermediate in the primary route for Trp catabolism by Lactobacillus casei to ILA via successive transamination and dehydrogenation reactions (Gummalla & Broadbent, 1999). However, IPA is unstable and spontaneously degrades to different metabolites (Gao et al., 1997), and using IPA as the substrate in dedicated assays here failed to yield expected results.
Comparative analysis of PLA, PEA and ILA production by different yeast strains
Six different strains of G. candidum were able to produce from 1.6 ± 0.5–5.0 ± 0.2 mM of PLA in Phe-SM and 5.0 ± 0.1–10.9 ± 0.3 mM of ILA in Trp-SM (Fig. 5). G. candidum ATCC 204307 was the most efficient strain, while G. candidum UCMA 4057 was the least efficient. The three other yeast strains tested are the principal yeast species identified on the surface of soft smear-ripened cheeses. Among these, C. natalensis UCMA 3722 produced the highest amounts of metabolites 4.07 ± 0.52 mM of PLA from Phe-SM, 4.13 ± 0.23 mM of PLA from PPA-Gly-SM, and 9.51 ± 0.57 mM of ILA from Trp-SM.
Geotrichum candidum exhibited a particular metabolic pattern because it produced PEA in addition to PLA only in the presence of PPA (PPA-Gly-SM) but not in the presence of Phe in our experimental conditions. In the presence of 30.4 mM of PPA, the yeast strains produced PEA in the range of 0.11 ± 0.07 mM to 2.20 ± 0.09 mM (Fig. 5), G. candidum ATCC204307 being the highest producer. PPA has been previously reported as an intermediate of PEA production from Phe in different yeast strains but not in G. candidum (Lee & Richard, 1984). This could be due to the defined high quantity of PPA, which was added here and could stimulate another metabolic pathway. This concentration, however, is far lower than reported minimum inhibitory concentration values of 60–240 mM required for exerting bactericidal effects by PEA (Corre et al., 1990; Lucchini et al., 1990).
Effect of different physical parameters
At 5 °C (temperature of cheese storage and transport) and 15 °C (temperature of cheese ripening), the metabolite production was either absent (for PPA-Gly-SM samples) or reduced (for Phe-SM and Trp-SM samples) compared with the optimal 25 °C (Table 1). Geotrichum candidum's initial 106 cfu mL−1 count was maintained at 5 and 15 °C in all three media during the incubation period of 72 h. Thus, lowering of temperature had an overall negative impact on the metabolite production. As smear cheeses are ripened at 15 °C, this factor could explain the reduced antilisterial activity reported in situ in the cheese microcosm (Dieuleveux & Guéguen, 1998; Imran et al., 2010).
Table 1. Production of the metabolites by Geotrichum candidum ATCC204307 following an incubation of 72 h in synthetic media containing 0.5 g L−1 of the respective substrate (corresponding to 3.0 mM for Phe/PPA or 2.4 mM for Trp) under standard culture conditions (pH 5.7 ± 0.2, 25 °C, no NaCl) unless when specified
The present results show that under acidic pH, significantly higher quantities of PLA from Phe were detected at pH 4 and 6.5 than at pH 7, 8, or 9 (P-value < 0.01). On the other hand, ILA production was enhanced significantly under alkaline conditions (pH 7–9 in comparison with pH 4 and 5.6 (P < 0.05) (Table 1). However, PPA does not appear to be a suitable substrate for in situ production of PLA under higher pH conditions. The antimicrobial activity of organic acids such as PLA (pKa 3.46) has been shown to be pH dependent, with a maximum activity at low pH that favors the undissociated active form of the acid molecule (Piper et al., 2001; Gerez et al., 2010). In cheese microenvironment, microbial interactions can increase the pH from 4 to 7 (Imran et al., 2010), and as a result of this change, under fermentation conditions, PLA activity can be negatively affected.
In smear cheese environment, salt is provided at concentrations higher than 7 g L−1. In Table 1, results showed that PLA production from Phe remained unaffected under these conditions. However, an increase in PLA production from PPA was observed (P-value < 0.01) when sodium chloride concentrations were increased in the medium from 1 to 10 g L−1. The possible explanation for this increase could have been resulted from an enhancement of PPA transportation in the presence of Cl− as it was described for pyruvate transport mechanisms across erythrocyte cells, which implicate activation and use of a nonspecific anion transporter in the presence of high concentration of Cl− (Halestrap, 1976). However, Trp's transformation to ILA was significantly lowered (P-value < 0.01) by salt increase from 1 to 3 g L−1.
These results demonstrate that Phe would make the most suitable substrate for the production of the antilisterial metabolite PLA in cheese-ripening environment as its metabolic pathway was not affected drastically by salt, pH, or temperature as compared to PPA.
Conclusion and perspectives
To conclude, G. candidum transforms phenylalanine, tyrosine, and tryptophan into phenylpyruvic acid/phenyllactic acid, hydroxy phenyllactic acid, and indoleacetic acid, respectively. PPA, when used as substrate, subsequently produced PLA and phenyl ethyl alcohol. Previous knowledge of G. candidum's antipathogenic spectrum is here reinforced by the discovery of PEA and OH-PLA. The antimicrobial properties of these liposoluble compounds can generate particular interest for enhancing the safety of soft cheeses containing G. candidum as a ripening agent. Further studies on the enzymes and genes regulating the metabolic reactions involved in the production of these metabolites are in progress to understand how bioprotective effect of G. candidum in soft cheeses can be optimized.
This work was made possible by a PhD scholarship from Higher Education Commission (HEC) of Pakistan.