Identification and characterization of a new antifungal peptide in fermented milk product containing bioprotective Lactobacillus cultures

ABSTRACT

Mold and yeast contamination constitutes a major problem in food commodities, including dairy products, hence new natural preventive measures are in high demand. The aim of the current study is to identify and characterize novel antifungal peptides produced by lactic acid bacteria (LAB) in sour cream. By the use of a newly developed image-based 96-well plate fungal growth inhibition assay targeting Debaryomyces hansenii, combined with a range of analytical tools comprising HPLC-high-resolution mass spectrometry, ultrahigh-performance liquid chromatography-Triple Quadrupole MS and nuclear magnetic resonance spectroscopy, we successfully identified a new antifungal peptide (DMPIQAFLLY; 1211 Da) in sour cream enriched with two bioprotective LAB strains. This peptide represents a fragment of casein, the most abundant protein in milk. Presumably, the proteolytic activity of these bioprotective strains results in the observed 4-fold higher concentration of the peptide during storage. Both bioprotective strains are able to generate this peptide in concentrations up to 0.4 μM, independently of the sour cream starter culture employed. The peptide attenuates the growth rate of D. hansenii at concentrations ≥35 μM, and results in smaller cells and more compact colonies. Hence, the peptide is likely contributing to the overall preserving effect of the investigated bioprotective LAB strains.

lactic acid bacteria, antifungal peptide, food spoilage, bioprotection

INTRODUCTION

Lactic acid bacteria (LAB) are used as bioprotectives for a range of foods, including bread, fresh fruit and vegetables, animal feed and dairy products. The use of LAB is one of the oldest methods for preserving food, and it is estimated to date back more than 3000 years. The most important and best characterized antimicrobials produced by LAB are lactic acid and acetic acid (Arena et al.2016). Other metabolic products from LAB contribute to the overall antimicrobial capacity and preservative potential of LAB, although the contributions from the individual constitutents are difficult to quantify (Lindgren and Dobrogosz 1990). Studies describing antibacterial proteinaceous compounds, e.g. large bacteriocins, are abundant in comparison to those focusing on proteins or peptides with antifungal properties. Nevertheless, during the last decade various LAB-derived proteinaceous compounds with anti-yeast and anti-mold activities have been identified (Coda et al.2008; Rizzello et al.2011; Crowley, Mahony and van Sinderen 2013). Initially, such peptides were identified by the loss of antifungal activity upon treatment with proteolytic enzymes, e.g. Magnusson and Schnürer (2001) identified antifungal activity that disappeared upon addition of proteinase K. Furthermore, the activity was strictly pH-dependent and was lost above pH 6, albeit the activity could be recovered after acidification of the medium (Magnusson and Schnürer 2001). Other studies of fermentation or enzymatic degradation of proteins have identified peptides exhibiting antifungal activity (Table 1). Coda et al. (2008) identified five peptides with moderate antifungal activity, exhibiting MICs in the range 3.5–8.2 mg/mL. Among all possible combinations of these, the lowest minimal inhibitory concentration (MIC; ∼0.95 mg/mL) was found for a mixture of the peptides DPVAPLQRSGPEIP and PHAVAAVPPVLR (Coda et al.2008). In a similar study, peptides produced by Lactobacillus plantarum 1A7 were identified—and they showed MICs within the range 2.5–11.6 mg/mL (Coda et al.2011). None of the identified peptides were de novo synthesized by LAB, but resulted from their proteolytic activity mediating degradation of proteins from wild rice and wheat. In another study, four additional bioactive peptides derived from rice proteins were found to possess antifungal properties with similar MIC values (Rizzello et al.2011).

Table 1.

Antifungal peptides derived from proteolysis by LAB or enzyme from LAB.

Peptide	Derived from	Reference
DPVAPLQRSGPEIP	Rye protein	(Coda et al.2008)
PHAVAAVPPVLR	Rice protein	(Coda et al.2008)
LLGWGHKGSSIID	Rice protein	(Coda et al.2008)
PILQSLIRFDGGACSSF	Rice protein	(Coda et al.2008)
RSQIKREQYTPQDVEMLFSSF	Rice protein	(Coda et al.2008)
SAFEFADEHKGAYS	Rice protein	(Coda et al.2011)
AAIIFGSIFWNVGMKR	Rice protein	(Coda et al.2011)
AEGEVILEDVQPSSVQS	Rice protein	(Coda et al.2011)
PPDVLTKLTAVPAAQQLDEADGHPR	Rice protein	(Coda et al.2011)
VLHEPLF	Rice proteins	(Rizzello et al.2011)
ALKAAPSPA	Rice proteins	(Rizzello et al.2011)
AILIIVMLFGR	Rice proteins	(Rizzello et al.2011)
AAAAVFLSLLAVGHCAAADFNATDADADFA-GNGVDFNSSDAAVYWGPWTKAR	Rice proteins	(Rizzello et al.2011)
SSSEESII	Casein αs2	(Bougherra et al.2017)

Bioactive peptides arising from proteolytic degradation of milk, both with or without the presence of LAB, have previously been reported. Digestion of different milk proteins by e.g. pepsin and chymosin resulted in formation of >30 antimicrobial peptides (Sah et al.2016). A recent study identified an antibacterial peptide derived from the subunit α_s2 of casein. This peptide was obtained by hydrolyzing bovine casein with an extracellular serine metalloprotease originated from L. lactis subsp. lactis BR16. This peptide showed activity against Gram-positive and Gram-negative bacteria with MICs ranging from 0.2 to 1.1 mM (Bougherra et al.2017).

To our knowledge, no further studies have identified bioactive peptides resulting from proteolytic activity of LAB, and more specifically, no antifungal peptides have been identified in dairy products. We therefore wanted to investigate whether LAB, used as starter culture and/or bioprotective strains, produce bioactive peptides in sour cream.

MATERIALS AND METHODS

Milk ferments

Sour cream was produced from pasteurized (90°C for 20 min) 9% (w/v) fat cow milk inoculated with 0.01% (w/v) starter culture (Chr. Hansen A/S, Denmark) followed by fermentation at 26°C for 13 h. Sour creams were prepared with (+BioP (bioprotective culture)) or without (–BioP) the addition of two bioprotective strains (Lactobacillus paracasei (strain 1: CH127) and Lactobacillus rhamnosus (strain 2: CH126), Chr. Hansen A/S, Denmark). BioP strain inoculation levels were 5 × 10⁶ colony-forming units (CFU)/mL. Sour creams were stirred for a few minutes at 25°C before storage at 4°C–7°C. Additional milk ferments were prepared by fermenting milk at 30°C for 24 h only in the presence of one of the two or both BioP strains (no starter culture) and these ferments are referred to as Strain 1, Strain 2 and Strain 1 + 2, respectively.

Image-based 96-well plate fungal growth inhibition assay

The image-based 96-well plate fungal growth inhibition assay measured the growth rate of a reference Debaryomyces hansenii yeast strain based on images acquired using an oCelloScope (Phillips BioCell, Allerød, Denmark). Test solutions containing reference and test compounds were prepared in saline peptone (pepsal, Oxoid, Roskilde, Denmark) with pH adjusted to 4 using 1 M HCl and containing 0.025%–3% (v/v) dimethyl sulfoxide (DMSO). Reference compounds included potassium sorbate and benzoic acid (Sigma-Aldrich, Copenhagen, Denmark) and test compounds were peptides synthesized in-house. Aqueous sour cream samples, vide infra, were prepared for the growth inhibition assay by mixing these samples 50:950 with DMSO-water solutions, resulting in final concentrations of 0.025%–3.0% (v/v) DMSO. A 48-h yeast culture was prepared at 25°C by inoculating yeast extract glucose chloramphenicol medium (1 g/L yeast extract (Merck, Darmstedt, Germany), 20 g/L D-glucose (Merck, Darmstedt, Germany) and 0.1 g/L chloramphenicol (Sigma-Aldrich, Copenhagen, Denmark)) with D. hansenii (stored in glycerol at −80°C). The 48-h yeast culture was diluted in pepsal to obtain a yeast solution of approximately 2 × 10⁵ CFU/mL. Test and yeast solutions were mixed (1000:20), and 100 μL of the resulting solution were applied to wells of a transparent 96-well microplate in the required number of replicates. The n-values are denoted as ‘n = x(y)’ and designate x biological experiments each with y technical replicates. The microplates were covered with a lid and the yeast growth monitored at room temperature for up to 48 h by using an oCelloScope (Phillips BioCell, Allerød, Denmark), recording an image of each well at a preset frequency and duration. Recorded images were processed by the oCelloScope software applying Segmentation and Extraction of Surface Areas or Background Corrected Absorption algorithms depending on concentration. The output of the algorithms is log to the number of pixels covered with yeast and is thereby an estimate of yeast growth. Yeast growth rates were quantified as the slope of the exponential part of the growth curves (as determined by visual inspection in each case) using the ‘linear regression’ function in GraphPad Prism® version 7.00 (GraphPad Software, La Jolla, CA, USA).

Preparation of samples from milk ferments

For image-based 96-well plate growth inhibition assays

Sour cream was exposed to 40 kHz ultra-sonication using a Branson 3510E-DTH for 60 min (Branson Ultrasonics, Danbury, CT, USA) followed by centrifugation (at 14 000 rpm) using a Sigma 3-18K centrifuge (SciQuip, Shrewsbury, UK) for 1 h and filtration of the resulting liquid phase through 0.22 μm Sartolab-P20 cellulose acetate filters (Sartorius Stedim Biotech, Göttingen, Germany). These aqueous sour cream samples were used directly for the image-based 96-well plate growth inhibition assay.

For HPLC-high-resolution mass spectrometry and HPLC-HRMS²

Concentrated aqueous sour cream samples for HPLC-high-resolution mass spectrometry (HRMS) analyses, vide infra, were prepared from aqueous sour cream samples by removal of very polar constituents using reversed-phase vacuum liquid chromatography (VLC). Eluents comprised HPLC-grade methanol (VWR, Radnor, PA, USA), milliQ water (Merck-Millipore, Burlington, MA, USA) and acetonitrile (VWR, Radnor, PA, USA), all acidified with 0.1% (v/v) trifluoroacetic acid (Iris Biotech, Marktredwitz, Germany). A 40 × 40 mm RP C₁₈ VLC column (Merck, Darmstedt, Germany) was preconditioned with 50 mL methanol, then 100 mL acetonitrile and finally conditioned with 100 mL water. A sample of 100 mL aqueous sour cream was added to the column, and very polar constituents were removed by elution with 200 mL water. The less polar constituents were collected by elution with 100 mL acetonitrile and the sample volume was reduced in vacuo.

For ultrahigh-performance liquid chromatography

Acetonitrile extracts of sour cream and milk ferments were prepared for quantitative ultrahigh-performance liquid chromatography (UHPLC)-triple quadrupole mass spectrometer (TQMS) analyses by mixing these with acetonitrile (VWR, Radnor, PA, USA) at a 1:9 ratio, if not otherwise stated, then exposing the mixture to 40 kHz ultra-sonication, centrifugation at 14 000 rpm and finally, collection of the supernatant.

HPLC-HRMS analysis

Concentrated aqueous sour cream samples (MeCN fraction from VLC, vide supra) were separated on an Agilent 1260 series instrument (Santa Clara, CA, USA) consisting of a G1329B auto sampler, a G1311B quaternary pump with integrated degasser, a G1330B thermostatted column compartment and a G1315D photodiode-array detector. Separations were performed at 40°C using a Phenomenex C₁₈(2) Luna column, 150 × 4.6 mm i.d., 3 μm particles, 100 Å pore size (Phenomenex, Torrance, CA, USA) with a flow rate of 0.5 mL/min. The eluate was directed to a Bruker micrOTOF-Q II mass spectrometer equipped with an electrospray ionization source (Bruker Daltonik, Bremen, Germany) operated in positive mode with a corona potential of 4 kV, and dry gas with a flow of 7 L/min at 200°C. All operations were controlled by Bruker Hystar ver. 3.2 software, and data analysis was performed using Compass Data Analysis ver. 4.0 (Bruker Daltonik, Bremen, Germany) and MatLab ver. R2017a software (MathWorks, Natick, MA, USA). Spectra were matched with possible fragment masses using mMass for structure verification (Strohalm et al.2008; Strohalm et al.2010; Niedermeyer and Strohalm 2012).

Nuclear magnetic resonance spectroscopy and MS structural elucidation

All nuclear magnetic resonance spectroscopy (NMR) experiments were performed on a 600 MHz Bruker Avance III instrument (operating frequency of 600.13 MHz) equipped with a cryogenically cooled 1.7-mm TCI probe head and a Bruker SampleJet sample changer (Bruker Biospin, Karlsruhe, Germany). All experiments were acquired in automation (temperature equilibration to 300K, optimization of lock parameters, gradient shimming and setting of receiver gain). ¹H NMR spectra were acquired with 30°-pulses and 64k data points. 2D homo- and heteronuclear experiments were acquired with 2048 data points in the direct dimension and 128 (HMBC) or 512 (DQF-COSY) or 256 (multiplicity edited HSQC and NOESY) data points in the indirect dimension. IconNMR ver. 4.2 (Bruker Biospin, Karlsruhe, Germany) was used for controlling automated sample change and acquisition of NMR data, whereas Topspin ver. 3.5 (Bruker Biospin, Karlsruhe, Germany) was used for acquisition and processing of NMR data.

Quantitative analysis of free amino acids and peptides

The content of free amino acids was determined in milk ferments by gas chromatography-mass spectrometry (GC-MS) after derivatization as described previously (Villas-Bôas et al.2011). The content of peptide GLPQEVL and DMPIQAFLLY in acetonitrile extracts of milk ferments were determined using a Bruker Advance UHPLC with an EVOQ TQMS. The column used was a Kinetex XB-C18 (Phenomenex, 50 × 2.1 mm, 1.7 μm, 100 Å) maintained at 40°C. The aqueous eluent (A) consisted of water/acetonitrile (95:5, v/v), and the organic eluent (B) consisted of acetonitrile/water (95:5, v/v), both acidified with 0.1% formic acid. The elution profile: 0 min, 300 μL/min, 30% B; 2.5 min, 300 μL/min, 58% B, followed by 2.5 min wash out at 500 μL/min 100% B and 2.5 min of equilibration at 500 μL/min 30% B. The injection volume was 5 μL and standard peptide solutions were prepared in 90% (v/v) acetonitrile-water.

Protein sequence matching

Protein sequence matching was performed using BLAST searches in the annotated protein sequence database SWISS-PROT distributed by The EMBL Outstation—The European Bioinformatics Institute, UK (The UniProt Consortium 2017).

Peptide synthesis

Fmoc-protected amino acid building blocks, 2-chlorotrityl polystyrene resin (loading: 1.60 mmol/g), piperidine and trifluoroacetic acid were obtained from IRIS Biotech (Marktredwitz, Germany). Automated SPPS was performed on a LIBERTY system (CEM, Matthews, NC, USA) using a DISCOVER microwave unit (CEM, Matthews, NC, USA). Water used for analytical and preparative-scale HPLC was filtered through a 0.22 μm membrane filter. Analytical-scale HPLC was performed on a Shimadzu Nexera UPLC System (Shimadzu, Kyoto, Japan) comprising two LC-30AD pumps, an SIL-30AC auto-sampler, CTO-20AC oven, a SPD-M20A PDA-Detector and a DGU-20A5R Controller. Analytical UHPLC was performed on a Phenomenex Luna 2.5 μm C18(2) HST column, 100 × 3.0 mm i.d., 2.5 μm particle size (Phenomenex, Torrance, CA, USA). Preparative-scale HPLC was performed on a Shimadzu Prominence system (Shimadzu, Kyoto, Japan) comprising a CBM-20A controller, two LC-20AP pumps, an SIL-20AHT, an SPD-M20A diode array detector, by using a Phenomenex Luna C18(2) column, 250 × 30 mm i.d., 5 μm particle size (Phenomenex, Torrance, CA, USA). For both systems, binary mixtures of eluent A (5% MeCN in H₂O with 0.1% trifluoracetic acid (TFA) added) and eluent B (95% MeCN in H₂O with 0.1% TFA added) were used. Peak identity was confirmed by matrix-assisted laser desorption ionization-time-of-flight-mass spectrometry (MALDI-TOF-MS) using a Bruker Microflex LRF (Bruker Daltonik, Bremen, Germany).

GLPQEVL-OH

To obtain an Fmoc-Leu-preloaded 2-chlorotrityl chloride resin, a sample of resin (0.1 mmol) was treated with 10% N, N-diisopropylethylamine (DIPEA) in dry CH₂Cl₂ (3 mL) for 3 min followed by washing with dry CH₂Cl₂ (5 mL; 2 × 2 min), and then Fmoc-Leu-OH (2 equiv; 0.2 mmol) and DIPEA (0.4 mmol; 67 μL) in dry CH₂Cl₂ (3 mL) were added to the resin, which subsequently was shaken for 2 h, and then washed successively with N, N-dimethylformamide (DMF), MeOH and CH₂Cl₂ (5 mL, each for 3 × 3 min). Next, the preloaded resin was employed for the assembly of the remaining part of the sequences on a Liberty synthesizer (CEM, Matthews, NC USA). Finally, the peptide was cleaved from the resin by treatment with TFA (2 × 2 mL) for 1 h. Purification of the obtained crude product was performed by preparative-scale HPLC using a 0%–35% B linear gradient over 20 min to yield the desired peptide (t_R = 17.3 min) for which purity was assessed to be >97% by analytical UHPLC using a 0% to 40% B linear gradient (t_R = 7.8 min).

DMPIQAFLLY-OH

To obtain an Fmoc-Tyr(tBu)-preloaded 2-chlorotrityl chloride resin, a sample of resin (0.1 mmol) was treated with 10% DIPEA in dry CH₂Cl₂ (3 mL) for 3 min followed by washing with dry CH₂Cl₂ (5 mL; 2 × 2 min), and then Fmoc-Tyr(tBu)-OH (2 equiv; 0.2 mmol) and DIPEA (0.4 mmol; 67 μL) in dry CH₂Cl₂ (3 mL) were added to the resin, which subsequently was shaken for 2 h, and then washed successively with DMF, MeOH and CH₂Cl₂ (5 mL, each for 3 × 3 min). Next, the preloaded resin was employed for the assembly of the remaining part of the sequences on a Liberty synthesizer. Finally, the peptide was cleaved from the resin by treatment with TFA (2 × 2 mL) for 1 h. Purification of the obtained crude product was performed by preparative-scale HPLC using a 0%–55% B linear gradient over 20 min to yield the desired peptide (t_R = 16.6 min) for which purity was assessed to be >97% by analytical UHPLC using a 0%–60% B linear gradient (t_R = 8.4 min).

Statistical analysis

Student's unpaired t-test, 1-way ANOVA followed by Tukey's multiple comparisons test or multiple t-tests corrected for multiple comparisons using the Holm-Sidak method were used to assess statistically significant differences between two or more groups as appropriate. Significant outliers were identified by Grubbs’ test (α = 0.05). The significance level was set at P < 0.05 and is indicated with a single asterisk in figures. For linear regression analysis r² > 0.90 is accepted. All tests were conducted using GraphPad Prism® version 7.00 (GraphPad Software, La Jolla, CA, USA).

RESULTS

Antifungal activity in aqueous sour cream samples

Prolonged shelf life of a fermented milk product is defined as increased time until visible growth of yeast and/or mold is detected during storage. Sour cream containing bioprotective (BioP) Lactobaccillus paracasei and Lactobaccillus rhamnosus exhibits prolonged shelf life as compared to sour cream without these. However, a microplate-based assay for fast and reliable quantitative measurement of fungal growth inhibition in sour cream samples appears to be lacking. Therefore, we developed an image-based 96-well plate fungal growth inhibition assay capable of assessing the antifungal activity in aqueous sour cream samples using an oCelloScope. Debaromyces hansenii is a very common spoilage organism, and was chosen as test strain in this assay because it is very sensitive towards +BioP, making it a good candidate to elucidate the mechanisms behind the BioP. The oCelloScope fungal growth measurements are based on image capture, and the sample preparation and growth inhibition assay were both optimized for a transparent aqueous sample of sour cream retaining the antifungal activity observed for the crude BioP sour cream (Fig. 1A). In short, sour cream samples were subjected to 40 kHz ultrasonication, centrifugation at 14 000 rpm and filtration through 22-μm cellulose acetate filters. Preliminary experiments showed higher sensitivity toward detection of yeast growth using an oCelloScope as opposed to OD₆₀₀ spectrophotometric measurements, and thus allowed for a shorter experiment time, as previously described for detection of mold growth (Aunsbjerg, Andersen and Knøchel 2015). In particular, in this assay, exponential yeast growth can be detected within 24 h (Fig. 1B).

Figure 1.

Image-based 96-well plate fungal growth inhibition assay. Aqueous samples of sour cream produced by fermentation of milk and starter culture with the addition of bioprotective strains (+BioP) compared to those without (–BioP). The growth was compared to growth in pure saline peptone (Pepsal) and a control in saline peptone containing 33 mM (5 mg/mL) sorbate. All test solutions contained 0.025% (v/v) DMSO. (A) Yeast growth rates expressed as percentage relative to growth in saline peptone (pH 4) set to 100%. All growth rates measured during exponential growth and results shown as mean ± standard error of the mean (SEM), n = 3(7). (B) Representative growth curves.

Identification of peptides present in increased concentrations in BioP-containing sour cream

To identify peptides present in increased concentrations in BioP-containing sour cream, 2D-HPLC-HRMS plots of concentrated aqueous samples of sour cream with and without BioP were acquired. Visual inspection of mass-to-charge-ratio (m/z) ranges for which peptides are expected revealed two peaks with higher intensity in the BioP-containing sour cream, i.e. a peak corresponding to a peptide with m/z 605 eluting at 44.5 min and a peak corresponding to a peptide with m/z 378 eluting at 27.5 min (Fig. 2).

Figure 2.

Representative 2D HPLC-HRMS plots from sour cream prepared with (+BioP) or without (–BioP) addition of bioprotective strains. Blue to red color scale: low to high MS signal; m/z: mass-to-charge ratio. Characteristic differences in the sour creams with and without BioP strains (n = 6) are indicated by arrows, i.e. at m/z = 605 and RT = 44.5 min for DMPIQAFLLY (A) and at m/z = 378 and RT = 27.5 min for GLPQEVL (B).

The structures of these two peptides were elucidated by using HRMS² of the precursor ions and by analysis of the derived b- and y-ion series. The peptide with m/z = 605 was identified as DMPIQAFLLY (Fig. S1, Supporting Information) and the peptide with m/z = 378 was identified as GLPQEVL (Fig. S2, Supporting Information). The isoleucine/leucine ambiguities for GLPQEVL were resolved by using NMR spectroscopy (Figs S3 and S4 and Tables S1 and S2, Supporting Information), revealing both residues to be leucine. Both peptides were also subjected to a BLAST search in the SwissProt database revealing that the GLPQEVL sequence corresponds to S1-α-casein residues 25–31, while the DMPIQAFLLY sequence matches residue 199–208 in β-casein. The two peptides were furthermore synthesized allowing their structures to be unambiguously identified as DMPIQAFLLY and GLPQEVL by comparison of HPLC-HRMS, HRMS² and NMR data (not shown).

Fungal growth inhibition of identified peptides

The two identified peptides, GLPQEVL and DMPIQAFLLY, were synthesized as TFA salts to determine their antifungal properties. The fungal growth inhibitory activity of the two peptides was assessed separately in pepsal at pH 4 using the image-based 96-well plate growth inhibition assay. The effect on the yeast growth rate of the peptide counter ion TFA was examined and found non-significant (data not shown), and hence, the peptides were tested as TFA salts. GLPQEVL (755 Dalton (Da)) in concentrations up to 142 μM (112 μg/mL) exhibited no significant effect on the yeast growth rate during exponential growth (Fig. 3A). In contrast, 35 μM (42 μg/mL) DMPIQAFLLY (MW 1211 Da) significantly reduced the yeast growth rate during exponential growth (Fig. 3A). The peptide was very insoluble in water and no higher concentrations than 46 μM (56 μg/mL) DMPIQAFLLY could be tested before a precipitate was visible. The plateau of the dose-response curve for DMPIQAFLLY (Fig. 3A) indicates a maximum inhibitory activity of approximately 75% in saline peptone compared to the untreated control. This is a similar growth rate reduction as seen for the BioP-containing sour cream compared to sour cream samples without BioP (Fig. 1A). No additive inhibitory activity was observed by adding 9–70 μM of GLPQEVL to samples containing 23 or 46 μM DMPIQAFLLY (Fig. 3B).

Figure 3.

Analysis of antifungal properties of the identified peptides. (A) The peptide DMPIQAFLLY (purple triangles; n = 6(3–4)), but not GLPQEVL (black circles; n = 4(3–6)), exhibited antifungal activity in the image-based 96-well plate fungal growth inhibition assay. (B) The antifungal activity of DMPIQAFLLY at 23 μM (green squares) and 46 μM (gray diamonds) was not affected by the addition of 9–70 μM GLPQEVL; n = 2(3–4). Test solutions were prepared in pepsal (pH 4) containing 3% (v/v) DMSO. Yeast growth rates expressed as percentage relative to control growth (test solution with no peptide) set to 100%. All growth rates measured during exponential growth and results shown as mean ± SEM.

In order to evaluate the antifungal potency of DMPIQAFLLY, the bioactivity of two well-known antifungal compounds benzoic acid and potassium sorbate was assessed in the image-based 96-well plate fungal growth inhibition assay. The IC₅₀ and approximate MIC values were 390 μM and 131 μM, n = 1(3), for benzoic acid, and 330 ± 10 μM and 166 μM, n = 2(7), for potassium sorbate (data not shown). Compounds were tested in pepsal (pH 4) containing 0.25% (v/v) DMSO. Comparison of these MIC values to the inhibitory concentration of DMPIQAFLLY indicates that the peptide exhibits higher antifungal potency than potassium sorbate and benzoic acid.

Concentration of the identified peptides in sour cream

After testing the antifungal activity of the two peptides and verifying the bioactivity of the peptide DMPIQAFLLY, the amount of peptides was quantified in sour cream with and without BioP. Sour cream samples were extracted with acetonitrile and quantified using UHPLC-TQMS (Fig. 4). This showed that the concentration of DMPIQAFLLY in BioP-containing sour cream was 2-fold higher (128 ± 13 nM) than in sour cream without BioP (55 ± 9 nM; Fig. 4A). However, the concentration of DMPIQAFLLY found in BioP-containing sour cream was close to 300-fold lower than the concentration at which the peptide exhibited significant antifungal activity in the image-based 96-well plate fungal growth inhibition assay (≥35 μM). Similarly, the concentration of GLPQEVL found in sour cream containing BioP (68 ± 5 nM; Fig. 4B) was much lower than the maximal concentrations of this peptide tested in the image-based 96-well plate fungal growth inhibition assay (up to 142 μM).

Figure 4.

Peptide content in sour cream with and without BioP strains. Sour cream was produced in triplicate by fermentation of milk and starter culture with (+BioP) and without (–BioP) addition of bioprotective strains, then frozen immediately at –20^○C until analyzed. Acetonitrile extracts of sour creams were prepared and analyzed for the content of the two peptides DMPIQAFLLY (A) and GLPQEVL (B) using UHPLC-TQMS. The peptide content in milk used for the production of sour creams was similarly assessed. Results are shown as mean of a biological triplicate ± SEM, asterisks indicate P < 0.05 compared to milk.

Change in peptide concentrations during storage

As a next step, it was investigated whether the presence of the sour cream starter culture is required for the BioP strains to produce the bioactive peptide DMPIQAFLLY, and whether the peptide is produced during the original fermentation process and/or during subsequent storage in the refrigerator. Milk before fermentation, milk fermented by the individual BioP strains without the starter culture present and milk with starter culture ( = sour cream) with and without BioP strains were stored at 5°C, and samples were frozen at the indicated days for subsequent analysis of the peptide and free amino acid concentrations. The content of DMPIQAFLLY in sour cream containing BioP increased significantly (R² = 0.98, P = 0.0009) during storage (4-fold), whereas its concentration in sour cream without BioP was nearly constant (Fig. 5A). Without the starter culture, a different pattern is observed. During the first 14 days of storage, an increase in the concentration of DMPIQAFLLY was observed in milk ferments prepared with only bioprotective strain 1 (strain 1), only bioprotective strain 2 (strain 2) or both strains (strain 1 + 2; Fig. 5B). However, a decrease in the concentration of DMPIQAFLLY was observed in strain 2 and strain 1 + 2 samples on storage day 28, while strain1 samples exhibited even higher concentrations at day 28. These data suggest that DMPIQAFLLY is produced by the starter culture during fermentation and by the BioP strains during storage, while it is also degraded by strain 2 during long-term storage. To verify, that the individual BioP strains are able to proteolyse milk proteins, we determined the free amino acid concentrations in milk fermented for 24 h at 30°C by one or both of the BioP strains as well as in sour cream +/– BioP (Fig. 5C). Ferments prepared with strain 1 showed lower free amino acid concentrations as compared to ferments prepared with strain 2 or strain 1 + 2, indicating lower proteolytic activity of strain 1. The free amino acid content was similar for ferments prepared with strain 2, strain 1 + 2 and regular sour cream containing BioP (prepared with starter culture).

Figure 5.

Concentration of antifungal peptide DMPIQAFLLY measured during storage of milk with or without starter culture and with or without bioprotective LAB strains. The fermented products were stored at 5^○C and samples were collected for analysis at the given time point. (A) Samples with starter culture (= sour cream) were prepared for UHPLC-TQMS analysis using a 1:1 ratio between acetonitrile and sour cream for the extraction:—BioP samples (red squares) and +BioP (blue circles). (B) Samples without starter culture were prepared in a 9:1 ratio between acetonitrile and fermented milk: milk (gray diamonds), ferments prepared with bioprotective strain 1 (green circles), bioprotective strain 2 (magenta squares) and both strains (black triangles). (C) Relative free amino acid concentrations after one day of storage determined by GC-MS: milk (gray), milk + strain1 (green), milk + strain2 (magenta) and milk + strain1 + 2 (black) as well as sour cream (= milk with starter culture) without BioP (red) or with BioP (blue).

Morphological changes observed for spoilage yeast exposed to bioactive peptide

Using the image-based 96-well plate fungal growth inhibition assay, it was shown that the growth rate of the spoilage yeast D. hansenii is reduced in the presence of the peptide DMPIQAFLLY, and furthermore the oCelloScope-based method revealed morphological changes of D. hansenii colonies when cultivated in the presence of this peptide. As seen in Fig. 6, the colonies are smaller and denser in the presence of DMPIQAFLLY.

Figure 6.

Debaromyces hansenii grown in pepsal containing DMPIQAFLLY exhibits altered morphology. Representative images from the image-based 96-well plate fungal growth inhibition assay. Pepsal (pH 4, 3% (v/v) DMSO) with and without 35 μM DMPIQAFLLY was inoculated with D. hansenii and morphological changes were detected by following the yeast growth with measurements in the oCelloScope.

DISCUSSION

In this study it was shown that bioprotective LAB strains contribute to the production of a milk-derived bioactive peptide, DMPIQAFLLY, which shows growth-inhibitory activity against Debaromyces hansenii. This peptide is also present in milk and milk fermented traditionally by a starter culture, but in lower concentrations as compared to fermentations including BioP strains. Cultivation of BioP strains alone, without the starter culture, also provides the bioactive peptide. In particular, no increase in the concentration of the bioactive peptide was detected directly after fermentation; however, following 7 days of storage, fermentations including one or both BioP strains exhibited a significant increased content of DMPIQAFLLY. Interestingly, the peptide concentration decreased in fermentations prepared with strain 2 or strain 1 + 2 upon storage for beyond 14 days, while fermented milk containing strain 1 and traditionally made BioP-containing sour creams (including a starter culture) continued to increase in peptide concentration during the time measured. These data suggest that strain 1 and strain 2 are capable of producing the bioactive peptide during the fermentation and subsequent storage when the starter culture is present, but the BioP strains require storage time to produce significant amounts of the peptide in fermented milk not containing the starter culture. Moreover, it appears that strain 2 may be counteracting the accumulation of the bioactive peptide. This strain produced a high concentration of free amino acids during fermentation, indicating a high proteolytic activity of this strain (Fig. 5C). In contrast, strain 1 might contain more specific proteases and not those that are cleaving peptides into free amino acids, resulting in an increase in bioactive peptide and no or decreased degradation of this during storage.

The DMPIQAFLLY peptide inhibited the growth of D. hansenii in the image-based 96-well plate fungal growth inhibition assay at a concentration of 35 μM (42 μg/mL; Fig. 3), which is at least a 10-fold lower concentration than the MIC values reported previously for protein-derived antifungal peptides (Rizzello et al.2011). The maximal yeast growth inhibition by DMPIQAFLLY was about 25% as compared to growth of the untreated control in saline peptone. Higher concentrations could not be tested due to the low solubility of the peptide in aqueous media. Interestingly, the yeast growth rate in aqueous samples of sour cream containing BioP was also only reduced by ∼50% as compared to the sample without BioP. The concentration of the peptide in sour cream is at least 100-fold lower than concentrations found to be inhibitory in the antifungal assay. LAB strains are known for producing several antifungal compounds (Yang and Chang 2010; Pawlowska et al.2012) working synergistically in the combat of spoilage fungi (Cabo, Braber and Koenraad 2002; Ndagano et al.2011). Therefore, we believe that this peptide is not the sole source of the enhanced shelf life observed for sour cream containing BioP strains, but we propose that it contributes to the overall bioactivity.

The bioactive peptide identified in this study corresponds to a fragment of a peptide discovered previously by enzymatic proteolysis of casein. This peptide, DMPIQAFLLYQQPVLGPVR, has been patented as an antibacterial peptide, showing a MIC of 2.4 μM against oral bacterial pathogens (Reynolds, Dashper and O’Brien-Simpson 1998). Hence, it is possible that DMPIQAFLLY likewise exhibits inhibitory activity against bacteria. Furthermore, they speculate that these casein-derived peptides exert their antimicrobial effect by virtue of their amphipathic nature enabling incorporation into cell membranes of other microorganisms where it may form aggregates that lead to membrane distortion and growth inhibition (Reynolds, Dashper and O’Brien-Simpson 1998). DMPIQAFLLY has a single charge, and hence most likely does not retain the amphipathic nature of DMPIQAFLLYQQPVLGPVR. Nevertheless, the proposed mode of action for casein-derived peptides could still be valid for DMPIQAFLLY, since its lipophilic nature could likewise allow for interactions with cell membranes. In line with this, we show that DMPIQAFLLY exposure results in a more condensed yeast phenotype, likely caused by effects on the cell membrane (Fig. 6). A similar mode of action has been proposed for certain antibacterial peptides, typical strongly cationic, and is termed the Shai–Matsuzaki–Huang model (Baltzer and Brown 2011).

Bovine milk contains a complex mixture of several hundreds of proteins with the major part being caseins (Gagnaire et al.2009). It is well known that when milk is subjected to fermentation with LAB, caseins are partly hydrolyzed to oligomers with a length of 4–18 amino acids by a range of cell envelope proteases (Honoré, Thorsen and Skov 2013). Hence, it is not surprising that DMPIQAFLLY and GLPQEVL are produced from these caseins by the added BioP LAB strains during the fermentation process and during storage. We showed a significant increase in the concentration of DMPIQAFLLY in sour cream containing BioP strains during storage, supporting that the BioP strains mediate the formation of this peptide from casein. Future studies aim to identify the proteases involved in DMPIQAFLLY peptide generation.

SUPPLEMENTARY DATA

Supplementary data are available at FEMSYR online.

Acknowledgements

We acknowledge the excellent help and assistance from laboratory technician Maiken Lund Jensen and the help from Tina Hornbaek, Arife Önder, Katrine Juhl Krydsfelt and the Analytics Department from Chr. Hansen A/S.

Author contribution statement

CG, DS, ARN, SS and JMV conceived this project. LKFM, JMV, AMH, SS performed all of the experiments. ARN, CG, SS, LKFM, JMV, KJ, HF and DS analyzed the data. SS, LKFM, DS and ARN wrote the manuscript with contributions from all authors.

FUNDING

These studies were sponsored by Chr. Hansen A/S (Denmark).

Conflict of interest. SS, ARN and CG work for Chr Hansen A/S who sponsored a postdoc grant to JMV and LKFM and who develops and sells BioP.