In situ production of human β defensin-3 in lager yeasts provides bactericidal activity against beer-spoiling bacteria under fermentation conditions




To examine the use of a natural antimicrobial peptide, human β-defensin-3 (HBD3), as a means of preventing spoilage from bacterial contamination in brewery fermentations and in bottled beer.

Methods and Results

A chemically synthesised HBD3 peptide was tested for bactericidal activity against common Gram-positive and Gram-negative beer-spoiling bacteria, including species of Lactobacillus, Pediococcus and Pectinatus. The peptide was effective at the μmol l−1 range in vitro, reducing bacterial counts by 95%. A gene construct encoding a secretable form of HBD3 was integrated into the genome of the lager yeast Saccharomyces pastorianus strain CMBS-33. The integrated gene was expressed under fermentation conditions and was secreted from the cell into the medium, but a significant amount remains associated with yeast cell surface. We demonstrate that under pilot-scale fermentation conditions, secreted HBD3 possesses bactericidal activity against beer-spoiling bacteria. Furthermore, when added to bottled beer, a synthetic form of HBD3 reduces the growth of beer-spoiling bacteria.


Defensins provide prophylactic protection against beer-spoiling bacteria under brewing conditions and also in bottled beer.

Significance and Impact of the study

The results have direct application to the brewing industry where beer spoilage due to bacterial contamination continues to be a major problem in breweries around the world.


Bacterial contamination of yeast fermentations is an ongoing problem in many breweries around the world and can occur at many stages of the fermentation process. Several bacterial species can tolerate the inhospitable conditions of the brew environment, the most common being anaerobic Gram-positives of the Lactobacillus and Pediococcus genera, including species such as Lactobacillus brevis, Lactobacillus brevismilis, Lactobacillus lindneri, Lactobacillus malefermentans, Lactobacillus dextrinicus and Pediococcus damnosus. (Jespersen and Jakobsen 1996; Weber et al. 2008). Additionally, Gram-negative bacteria such as Pectinatus frisingensis and Pectinatus cerevisiiphilus have also been identified in spoilt beer. The ability of beer-spoiling bacteria to persist and replicate under the harsh environmental conditions experienced in the brewery, ensures that once contaminated, it is difficult to control or to rectify the problem.

The use of antimicrobial compounds in the fermentation process to reduce bacterial contamination has been investigated. The addition of synthetic antibiotics to fermentation vessels is considered uneconomical and anathema to consumer sentiment and therefore has not been pursued. Hops (Humulus lupulus) cones or their extracts, which are added to beer and lager fermentations for flavour, can act as natural antimicrobial agents against many Gram-positive bacteria; however, resistance to hops compounds has been observed lately in bacterial strains isolated from breweries (Iijima et al. 2006). Other postbrew processes such as refrigeration, pasteurization, filtration and chemical preservatives further help to reduce beer spoilage.

Antimicrobial peptides (AMPs) are small (<50 amino acids), generally positively charged hydrophobic molecules that exhibit antimicrobial activity against a broad range of micro-organisms. AMPs are produced by a wide range of organisms, including bacteria, fungi, plants and animals. Bacterial AMPs, referred to as bacteriocins, are active against other, often closely related bacteria, thus offering a competitive growth advantage (Hassan et al. 2012). The use of bacteriocins to control bacterial contaminations in industrial settings has been examined. Nisin, a bacteriocin produced by Lact. lactis M30 strain, has been shown to inhibit the growth of Pediococcus sp. and other wine bacteria, such as Oenococcus oeni, in wine and fruit brandies (Radler 1990; Diez et al. 2012), although the effect was more bacteriostatic rather than bactericidal. Nisin also increased microbial stability when added to unpasteurized beer; however, it exerts a deleterious effect on the viability of brewery yeast (Chihib et al. 1999; Galvagno et al. 2007). The addition of Nisin to beer has been approved in Australia and New Zealand. A different bacteriocin, also isolated from Lact. lactis strain M30, displayed bactericidal activity when added to lagers postfermentation but not during fermentations (Vaughan et al. 2004). Bacteriocins from Enterococcus faecium strain L50 displayed bactericidal activity against Lact. brevis in hopped and unhopped wort; however, the bactericidal activity has not been examined under fermentation conditions (Basanta et al. 2008).

Several studies have reported the expression of genes encoding bacteriocins mainly in the baker's yeast, Saccharomyces cerevisiae. Recombinant yeast strains expressing the bacteriocins, Pediocin PA-1, from Pediococcus acidilactici (Schoeman et al. 1999) or Plantaricin 423 from Lactobacillus plantarum (Van Reenen et al. 2003), demonstrated antimicrobial activity against Lact. monocytogenes, while heterologous production of both enterocin L50A and L50B from Enterococcus faecium demonstrated antimicrobial activity against Ped. damnosus (Basanta et al. 2009). These studies confirmed the proof-of-principle that AMPs can be produced in yeasts and retain activity when secreted into the medium, although the antimicrobial activity of these peptides during fermentations has not been examined.

In higher eukaryotes, AMPs form part of the innate immune response, acting as a first line of defence against any invading bacteria (Higgs et al. 2005; Wiesner and Vilcinskas 2010; Bos and Salzman 2011; Nakatsuji and Gallo 2012; Jarczak et al. 2013). In such cases, the cells that come into contact with bacteria (e.g. the skin, oral cavity or digestive track of animals) produce AMPs. Eukaryotic AMPs have a broad susceptibility range; they are effective against many Gram-positive and Gram-negative bacteria, certain viruses and fungi (Hoover et al. 2003; Hoffmann 2007). Just over 2000 unique peptides with antimicrobial activity have been identified to date ( (Wang et al. 2009). The most prominent class of antimicrobial agents produced by human cells is the evolutionarily conserved small, cationic peptides called defensins. Defensins are grouped into three classes (α, β and θ) based on properties such as molecular weight, disulfide bridge patterns between conserved cysteine residues present in the peptides and/or their macrocyclic nature (Dhople et al. 2006; Bos and Salzman 2011). Six α-defensin (HNP1-6) and four β-defensin (HBD1-4) peptides have been characterized in humans while over 30 HBD-like peptide-encoding cDNAs have been identified in the human genome ( (Dhople et al. 2006; Wiesner and Vilcinskas 2010; Jarczak et al. 2013). The θ-defensins have only been identified in rhesus monkeys. Defensins differ in their antimicrobial specificities, for example HNP1, HNP5, HBD1 and HBD3 display broad antimicrobial activity against bacteria, some viruses and fungi, while HBD2 specificity is restricted to Gram-negative bacteria and fungi such as Candida albicans. Several models, mainly based on the net positive charge and amphipathic nature of the peptides, have been proposed to explain the mode of action of AMPs against bacteria. Most models predict that AMPs are adsorbed onto the bacterial cell surface through electrostatic interactions with acidic phospholipids and then integrate into the phospholipid bilayer leading to localized membrane thinning and membrane disruption through the formation of transmembrane pores, leading to cell death (Wiesner and Vilcinskas 2010; Teixeira et al. 2012).

Considering the broad antimicrobial activity of defensins, the aim of this study was to examine the use of these natural peptides in controlling bacterial spoilage in yeast fermentations. We demonstrate that HBD3 displays bactericidal activity against common beer-spoiling bacteria in vitro. By integrating a synthetic copy of the gene, coding for a secretable form of HBD3 peptide, into the genome of an industrial lager yeast, we show that HBD3 is secreted from the cells and effectively reduces the growth of bacteria seeded into pilot-scale fermentations. Furthermore, when added to bottled beer, HBD3 is an effective bactericidal agent during long-term storage.

Methods and materials

Strains and media

All bacterial species were obtained from the Deutsche Sammlung von Mikroorganismen and Zelkulturen, Leibniz Institute, Braunschweig, Germany, with the exception of strain Lact. brevis (strain DMS20054), which was obtained from the UCC strain collection. Bacteria were cultured at 30–37°C in MRS medium (Deman et al. 1960) containing 0·1% Tween-80 (Sigma-Aldrich, Poole, UK). Pectinatus frisingensis and Pediococcus damnosus were cultured under anaerobic conditions, while Lact. brevis was cultured under microaerophilic conditions.

Saccharomyces cerevisiae strain S150-2B (Mata, leu2-3, 112 ura3-52, trp1-289, his3) is a haploid laboratory yeast (Canavan and Bond 2007). The polyploid lager yeast, Saccharomyces pastorianus strain, CMBS-33, (Centre for Malting and Brewing Science, Leuven, Belgium) has previously been described (Bond et al. 2004; James et al. 2008). For culturing and propagation, yeast cells were grown in YEP (1% yeast extract, 2% bacto peptone) with 2% dextrose (YEPD) or maltose (YEPM).

Wort was prepared from a mixture of three different malted barley varieties as follows: 93% Pale malt, 0·5% Pale Caramalt (colour 25 ± 5 EBC), 0·2% Acid malt (with lactic acid bacteria acidified barley malt to reduce the pH of the mash). All malt samples came from Mälzerei Gebr. Steinbach GmbH (Zirndorf, Germany). Four different cultivars of hop pellets type 90 were used. Two bitter hops, Hallertau Perle (Crop 2011, 9·5% α-acid acc. to EBC 7·5) and Hallertau Magnum (Crop 2009, 15·5% α-acid acc. to EBC 7·5), were added at the start of wort boiling. Two aroma hops, Hallertau Hallertauer Tradition (Crop 2011, Type 90, 7·4% α-acid acc. to EBC 7·5) and Hallertau Spalter Select (Crop 2011, Type pellet 90, 5·8% α-acid acc. to EBC 7·5), were added at the end of wort boiling. All hops were supplied by Simon H. Steiner, Hopfen, GmbH (Mainburg, Germany).

Pilot-scale fermentations

Fermentations were carried out in c. 28 l of wort with an average extract of 11·4% w/w. The prepared wort was added to cylindro-conical fermenters, and yeast cells were pitched at a cell density of approx. 5 × 107 cells ml−1. The wort and beer were analysed according to standard methods specified by Mitteleuropäische Brautechnische Analysenkommission (MEBAK) or the European Brewery Convention (EBC). Extract [% w/w], original gravity [°P], apparent degree of fermentation [%] and alcohol [% v/v] were determined using an Alcolyzer Beer ME in combination with a DMA 4500 mol l−1 (Anton Paar GmbH, Graz, Austria). pH levels were measured with a InLab Expert Pro sensor (Mettler-Toledo GmbH, Greifensee, Switzerland). Cell counts were performed with a Thoma chamber cell. After fermentation, the beer was matured for 12 days at room temperature, then filtered, carbonated (9°C, 1·8 bar, 48 h) and bottled. Filtration was carried out using a plate filter with standard depth filter sheets (K200, pore size c. 3–7 μm; Pall SeitzSchenk Filtersystems GmbH, Bad Kreuznach, Germany). Half the beer was then pasteurized in the bottles.

Bactericidal activity of HBD3

A synthetic form of HBD3 (sHBD3; Fig. 1a, amino acids 19–66) was chemically synthesised (GL Biochem Ltd, Shanghai, China) and dissolved in sterile distilled water (SDW). Bacterial cultures, grown for 12–16 h, were diluted in SDW to give a final cell density of 104 CFU ml−1. sHBD3 at varying concentrations was added to the bacterial suspension (final volume 20 μl), and the cells were incubated at 37°C for 2–2·5 h. Serial dilutions of bacteria were then plated onto MRS agar and incubated under microaerophilic or anaerobic conditions depending on the bacterial strain. The same experimental approach was used to determine the bactericidal activity in bottled beer with the exception that the bacteria were seeded into the beer.

Figure 1.

Structure and expression of human β-defensin-3 (HBD3) gene construct. (a) Amino acid sequence of human defensin-3 including the α-factor secretory signal (underlined). (b) Schematic of the gene construct of HBD3. The promoter is indicated by a rectangle containing the name of the gene from which the promoter was derived (PGK). The α-factor secretory signal is shown as a black rectangle. The 3′ untranslated regions from the ADH gene is shown as is the KanMx cassette used during integration into the chromosome. CYC refers to legacy termination sequences left over from pGreg vector. The primers used for generation of gene constructs are labelled and referenced to Table 1 and Data S1. The open rectangles on primers A and F represent regions of sequence homology at the site of integration at YPR159C-A on chromosome XVI. (c) Cultures of the strain CM-INT-51 were grown YEPD, and RNA was extracted from cells harvested at the times indicated above the lanes (h). HBD3 mRNA levels were monitored by semiquantitative RT-PCR. Lane M, molecular weight markers, the sizes of the major bands (nts) are shown to the left of the figure.

Pilot-scale fermentations were seeded with c. 300 CFU ml−1 of Lact. brevis (strain DMS20054) at the start of the fermentation. Samples were taken from the fermentations at varying intervals thereafter. The surviving bacteria were enumerated by plating on MRS agar containing 80 μg ml−1 cycloheximide.

Cloning and chromosomal integration of the HBD-3 gene cassette into CMBS-33

The amino acid sequence of human β-defensin-3 (HBD-3) was obtained from the Protein Database (GenBank Accession Number >gi|17372441). The protein sequence was reverse translated into the corresponding nucleotide sequence, with codons optimized for expression in yeast. A DNA fragment, consisting of the sequences encoding the S. cerevisiae α-factor secretory signal peptide (Kjeldsen 2000) in-frame with the mature HBD-3 sequence and alcohol dehydrogenase (ADH) termination sequence downstream, was chemically synthesized (GenScript Inc., Piscataway, NJ, USA) and cloned into pBlueScript SK (Stratagene Inc., Santa Clara, CA, USA) at the EcoR1 site. The HBD3 gene cassette was PCR amplified, from this plasmid using primers Rec1 and Rec2 (Table 1), which contain 30 nt homology to sequences either upstream or downstream of the insertion site in the yeast shuttle vector, pGREG505 (Jansen et al. 2005). The PCR products were cloned into the plasmid by in vivo homologous recombination in the S. cerevisiae strain S150-2B as previously described (Jansen et al. 2005) to generate the plasmid pGREGHBD3-505, placing the HBD-3 gene construct under the control of the GAL promoter and upstream of the plasmid-encoded 3′ CYC termination sequences.

Table 1. Oligonucleotides
PrimerDNA sequence 5′–3′Location (nts)
  1. R, Reverse; F, Forward primers.

  2. a

    Contains 30 nt homology to pGREG505 vector.








The pGREGHBD3-505 was used as a template for amplification of the HBD3 gene cassette for integration into the brewery yeast CMBS-33 at YPR159C-A on the right arm of the chromosome XVI. As part of the integration step, the GAL promoter was replaced with the constitutive phosphoglycerate kinase (PGK) promoter from Saccharomyces bayanus. The KanMX cassette from the pGREGHBD3-505, which confers resistance to the aminoglycoside genecitin (G418), was inserted into the genome downstream of the HBD3 gene cassette (Fig. 1b). Procedural details for the gene integration are included in Data S1. The HBD3 gene cassette in the context of the promoter and terminator sequences was verified by DNA sequencing GATC Biotech (Konstanz, Germany), and the sequences are shown in Fig. S1.

RNA extraction and RT-PCR

Saccharomyces pastorianus strain CMBS-33, expressing the HBD3 gene, were cultured at 30°C in YEPD-G418. Total RNA was extracted from the cells as previously described (Campbell et al. 2002). The RNA samples were treated with DNase I (RNase-free; RQ1; Promega Co., Southampton, UK) and reverse transcribed into cDNA as previously described (Usher and Bond 2009). The oligonucleotides used for cDNA synthesis (G) and PCR amplification (G and H; reverse and forward respectively) are listed in Table 1. Semiquantitative PCR amplification was carried out as follows: 1 μl of cDNA was amplified in a reaction volume of 25 μl using 0·04 U of Taq Polymerase, 0·25 mm dNTPS, 10 mm Tris-HCl, pH 8·8, 50 mmol l−1 KCl, 1·5 mmol l−1 MgSO4, 0·1% Triton X-100, 1·5 mmol l−1 MgCl2, 400 nmol l−1 forward and reverse primers. The amplification cycle was 95°C for 5 min, followed by 20 cycles of 95°C for 30 s, 55°C for 45 s, 72°C for 1 min and a final elongation step of 72°C for 10 min.

Purification, detection and quantification of rHBD3

Yeast cell pellets (2·5-ml packed cell volume) from pilot-scale fermentations (CM-INT-51 and CMBS-33) were washed twice in 20 volumes of phosphate-buffered saline (PBS). Cell-associated HBD3 was released by resuspending the cell pellet in 10 volumes of freshly prepared 5% acetic acid. Following incubation at room temperature (20°C) for 10 min, the yeast cells were pelleted by centrifugation (4800 g), and the extraction was repeated a second time. The two supernatants were pooled and lyophilized. The lyophilisate was resuspended in 0·5–1·0 ml of PBS. An aliquot of the lyophilisate was resuspended in 0·1% triflouro acetic acid, and rHBD3 (recombinant HBD3) was purified on a Sep-pak C18 cartridge as described (Conlon 2007). Total cell lysates were prepared using CelLytic-Y (Yeast Cell Lysis/Extraction Reagent; Sigma-Aldrich). Yeast cells (0·1 ml packed cell volume) from pilot fermentations were washed in PBS as described earlier and resuspended in CelLytic-Y (0·4 ml). Approximately 0·1-ml packed volume of acid-washed glass beads (0·5 μm) was added and vortexed for 10 min. The cell lysis (routinely c. 80%) was checked by microscopy. The supernatant was collected following the removal of the glass beads and cell debris by centrifugation.

rHBD3 levels in the medium, cell associated and cell lysates were quantified by capture enzyme-linked immunosorbent assay (capture ELISA) kit (Peprotech Co., Rocky Hill, NJ, USA) using the supplier's detailed instructions. Briefly, Nunc Maxisorb (96-well) plates (Thermofisher, Loughborough, UK) were coated overnight with polyclonal anti-HBD3 capture antibody in PBS (3 μg ml−1). The wells were washed twice with PBS containing 0·05% Tween-20 (PBS-Tween). The wells were blocked for 2 h. with 1% bovine serum albumin (BSA) in PBS-Tween (0·3 ml well−1). The wells were incubated (2–12 h) with various dilutions of either the cell medium, acid-washed fraction or cell lysates of CMBS-33, CM-INT-51 or synthetic HBD3. Following washing with PBS-Tween, the wells were incubated for 2 h with biotinylated polyclonal anti-HBD3 antibody (0·25 μg ml−1) and washed as before. The wells were incubated for 30 min with streptavidin-HRP and then washed with PBS-Tween. Antibody binding was detected with 3,3′,5,5′-Tetramethylbenzidine (TMB, T8665; Sigma-Aldrich). Following incubation at 30°C for 20–30 min., absorbance at 630 nm and at 450 nm (following acidification with 0·1 mol l−1 Sulfuric Acid) was measured. A standard curve was generated using serially diluted synthetic HBD3.

Immunofluorescence of formaldehyde-fixed yeast cells

The cell-associated HBD3 was visualized by immunofluorescence microscopy using the same principles of the HBD3 capture ELISA. Aliquots of cells from the pilot-scale fermentations of CM-INT-51, CMBS-33 were resuspended in 5 ml PBS containing 4% formaldehyde and incubated for 60 min at room temperature. The cells were pelleted by centrifugation at 4800 g for 5 min. Following extensive washing in PBS, the cells were incubated in 1% BSA in PBS for 60 min. 200-μl aliquots were pelleted and resuspended in 100 μl PBS, 0·1% BSA (PBS-BSA-DB) containing biotinylated anti-HBD3 detection antibody (Peprotech) at 1 μg ml−1 and incubated for 1 h. Following three washes in PBS-Tween, the cells were incubated with Extravidin-Cy3 (E4142, SigmaAldrich) in 100 μl PBS-BSA-DB for 30 min. The cells were washed four times in PBS-Tween and twice in PBS. The cells were recovered and resuspended in 100 μl PBS, and 5 μl was mounted on polylysine-treated microscope slides in Vectashield mounting medium (Vector Labs, Burlingame, CA, USA). The slides were observed and photographed using a Nikon Eclipse Fluorescent microscope equipped with a rhodamine (TRITC) filter.

Western blotting

Eluates from acid wash fractions or Sep-pak columns were lyophilized, dissolved in SDS-sample buffer and alkylated. The samples along with 0·25 μg synthetic HBD3 were separated on 15% polyacrylamide-SDS gels (Sambrook and Russell 2001; Siebke et al. 2012), and the proteins were electrotransferred to membranes and processed as previously described (Siebke et al. 2012), except that polyvinylidene difluoride (PVDF, Immobilon-P Millipore, Billerica, MA, USA) membrane was used according to manufacturer's recommendations. The membranes were blocked in 1% BSA in PBS-Tween (2–12 h) and then incubated (2–12 h) with biotinylated anti-HBD3 polyclonal antibody (detection antibody, 0·25 μg ml−1 Peprotech) in PBS-BSA-DB. The membrane was washed three times in excess PBS-Tween and then incubated with HRP-conjugated streptavidin (Peprotech), at a 1 : 1000 dilution. Streptavidin binding to biotin-anti-HBD3 was detected using a luminol-based chemiluminescent peroxidase substrate (Santa Cruz Biotechnology, Dallas, Texas, USA) and captured on HyperFilm ECL (GE Healthcare, Amersham, UK).


Synthetic HBD3 peptide is effective against beer-spoiling bacteria

To test whether human β-defensin 3 peptide (HBD3) is bactericidal for beer-spoiling bacteria, cultures of Lact. brevis, Ped. frisingensis and Ped. damnosus were incubated with increasing concentrations of a chemically synthesized mature form of HBD3 peptide (sHBD3; Fig. 1a). As shown in Fig. 2, sHBD3 at 500 ng ml−1 was sufficient to reduce the bacterial load by >95% in bacterial cultures seeded at a cell density of c. 104 ml−1. Other beer-spoiling bacteria, including Lact. brevisimilis and Lact. malefermentans, displayed similar sensitivities to sHBD3 (data not shown). The strict anaerobes Ped. frisingensis and Ped. damnosus appear to be more sensitive to sHBD3, displaying a higher killing rate at lower concentrations of peptide (P = 0·02). At lower cell densities (102–103 CFU ml−1), 90% of bacteria can be eliminated with sHBD3 at concentrations as low as 50 ng ml−1 (data not shown). Lager yeasts on the other hand showed a much lower sensitivity to sHBD3: Saccharomyces pastorianus cultures at concentrations used for pitching in lager fermentations (107 cells ml−1) were unaffected by sHBD3 at concentrations up to 100 μg ml−1 (data not shown).

Figure 2.

Bactericidal activity of human β-defensin-3 (HBD3). Cultures of Lactobacillus brevis, Pediococcus frisingensis or Pediococcus damnosus (c. 104 cells ml−1) were mixed with increasing concentrations of synthetic HBD3 and incubated at 37°C for 2·5 h. Cell survival was enumerated by plating on MRS agar. Error bars represent the standard deviation from four independent experiments. (image) Lact. brevis; (image) Ped. frisingensis and (image) Peddamnosus

Cloning and expression of HBD3 in the lager yeast CMBS-33

Having demonstrated that sHBD3 is an effective bactericidal agent against beer-spoiling bacteria in vitro, we next set out to express the peptide in lager yeast to test whether in situ production of HBD3 could provide prophylactic protection against bacterial infection during fermentation. We first used a gene cassette in which a secretable form of HBD3 was expressed from an inducible GAL1 promoter in the pGreg505 yeast shuttle vector. DNA sequences encoding the secretory signal sequence for the yeast mating-type factor (α) was engineered upstream of the HBD3 gene (Fig. 1a, underlined sequence) to direct the secretion of HBD3 into the growth medium. To ensure proper transcript processing, termination sequences from the yeast alcohol dehydrogenase (ADH) was placed downstream of the HBD3 stop codon but upstream of the legacy termination signal sequences of yeast CYC gene in pGreg505. (Fig. 1b). Subsequently, the GAL1 promoter in the cassette was replaced with a S. bayanus PGK promoter to drive the constitutive expression of HBD3 (Fig. 1b). The HBD3 gene cassette driven by the PGK promoter, along with the termination sequences, was integrated into the genome of the lager yeast CMBS-33 between YPR159 and YPR160W disrupting the dubious ORF YPR159C-A on chromosome 16. The derived strain was called CM-INT-51. The target location was chosen based on our previous DNA sequence analysis, transcript mapping and copy number variation analysis of this region (Usher and Bond 2009), and the fact that the region appears to contain genes that are either non-essential or redundant. The integration of the HBD3 gene cassette at YPR159CA was confirmed through PCR amplification of upstream and downstream fragments, anchoring one of the primers in the gene cassette and the other either downstream of YPR159 or upstream of YPR160W. These fragments were sequenced to identify the precise location of the integrant.

To determine whether the integrated gene is expressed, reverse transcription-PCR (RT-PCR) was conducted with RNA isolated from the yeast strain CM-INT-51 cultured as described in the Methods section. As shown in Fig. 1c, constitutive expression of recombinant HBD3 (rHBD3) mRNA was observed in the CM-INT-51 strain with maximum expression observed after 8 h.

Detection of secreted rHBD3

To determine whether recombinant HBD3 (rHBD3) was secreted into the media, strain CM-INT-51 was cultured in YEPM or wort and the spent media assayed for the presence of HBD3 by ELISA as described in the Methods section. rHBD3 was detected in the media from YEPM cultures at a concentration of 3–10 ng ml−1 (average 5 ng ml ± 3·4) in multiple independent experiments. Comparable levels of rHBD3 were detected in the media from wort fermentations (Fig. 3a).

Figure 3.

(a) Detection of rHBD3. CM-51 cells were inoculated into either YEPM or wort. The cultures were grown aerobically for the times indicated. rHBD3 was detected by ELISA. The mean of triplicate values and standard deviation are shown. (□) YEPM and (image_n/jam12382-gra-0004.png) WORT. (b) Western blotting. Postbrew cell pellets (108 cells) of CM-INT-51 or its parent CMBS-33 were extracted with acetic acid and processed as described in Materials and Methods. The extract was applied to a Sep-pak C-18 column and bound proteins eluted in 80% acetonitrile (ACN). Samples were dissolved in SDS-sample buffer and alkylated. 8–10 μg total protein from the eluates was separated on denaturing 15% acrylamide SDS gels. Chemiluminescence image of anti- human β-defensin-3 (HBD3) bound to HBD3 is shown. Lane 1, acetic acid extract from CMBS-33; lane 3, acetic acid extract from CM-INT-51; lane 4, 80% ACN eluate from CM-INT-51, lane 6, 250 ng of sHBD3. Lanes 2 and 5 were left empty. The apparent slightly faster migration of rHBD3 in lanes 3 and 4 (compared with sHBD3 in lane 6) is due to the high protein content in the extracts compared with the amount of sHBD3 loaded. (c) Immunofluorescence staining with anti-HBD3. Postfermentation CMBS-33 or CM-INT-51 yeast cells were fixed in formaldehyde and blocked with 1% BSA and incubated with biotinylated rabbit anti-HBD3 antibody. The cell bound antibodies were detected by Extravidin-Cy3 using a Nikon Eclipse fluorescent microscope at 800× magnification. Panels A and C are phase-contrast while B and D are Cy3 fluorescence views of CMBS-33 and CM-INT-51 cells, respectively.

As the levels of rHBD3 detected in the medium were relatively low, we wondered whether the peptide remained associated with the cell either due to defective secretion or due to natural affinity of defensins for cell membranes. Others have exploited the latter feature to purify defensins from media by first binding the peptide to glutaraldehyde-fixed Staphylococcus aureus cells and then eluting the bound peptides with dilute acids (Harder et al. 2001). To test this possibility, the CMBS-33 and CM-INT-51 cell pellets from pilot-scale fermentation were washed with acetic acid to extract any cell-associated rHBD3 (see methods for details). We observed that a significant quantity of rHBD3 remained bound to the yeast cells, with approximately fivefold more rHBD3 remaining cell associated than free rHBD3 in the medium, (21 and 4·2 ng ml−1, respectively). Western blot analysis of protein extracts from acid-washed cell pellets or following a further purification on a C-18 Sep-pak column identified a band migrating at same molecular weight as the sHBD3 in strain CM-INT-51 (Fig. 3b), while no bands were detected in the parental strain CMBS-33. The presence of rHBD3 on the external surface of the cells was also confirmed by immunofluorescence using an anti-HBD3 antibody (Fig. 3c). rHBD3 were also detected in cell lysates of yeast strains carrying the HBD3 gene cassette (see below), especially during the early stages of fermentation, suggesting concomitant synthesis and secretion of rHBD3.

Expression and In situ bactericidal activity of rHBD3 in pilot-scale fermentations

To determine whether rHBD3 is produced under industrial fermentation conditions and if so could provide prophylactic protection from beer spoilage bacteria, pilot-scale fermentations were carried using strain CM-INT-51 or the parental strain CMBS-33 in the presence or absence of the beer-spoiling lactic acid bacterium Lact. brevis (+LAB). As shown in Fig. 4a,b, strain CM-INT-51 fermented faster than the parental strain CMBS-33. This trend was replicated in two separate fermentations. The presence of Lact. brevis did not affect the fermentation rate of either strain. Strain CM-INT-51 had an apparent degree of fermentation (ADF) of 80%, and a final percentage alcohol of 5·16% while strain CMBS-33 reached on ADF of 60% and a final percentage alcohol of 4·10% (Fig. 4b). The increased fermentation rate in strain CM-INT-51 appears to result from the slightly faster growth rate of this strain during fermentation (Fig. 4c,d). Cell counts for both strains were similar up to day 3 when thereafter, growth of strain CM-INT-51 was significantly faster than strain CMBS-33.

Figure 4.

Fermentation profiles of CM–INT–51 and CMBS–33. Pilot–scale fermentations were carried out at 13°C in the presence (+LAB) or absence (−LAB) of Lactobacillus brevis seeded at the start of the fermentation. Samples were extracted at regular intervals during the fermentation. (a) Specific gravity, Plato, (b) % Alcohol (v/v), (image_n/jam12382-gra-0003.png) CM–INT–51(+LAB), (image_n/jam12382-gra-0006.png) CM–INT–51(−LAB), (image_n/jam12382-gra-0005.png) CMBS–33(+LAB), (image_n/jam12382-gra-0007.png) CMBS–33(−LAB). Data for CM–INT–51(+LAB) and CMBS–33(+LAB) represent the mean of two independent fermentations (fermentations 1 and 2), while data for CM–INT–51(−LAB) and CMBS–33(−LAB) are from a single fermentation. Error bars represent the standard error from the mean. (c and d) Yeast cell counts from two independent fermentations (Fermentation 1 and 2), (image_n/jam12382-gra-0003.png) CM–INT–51(+LAB), (image_n/jam12382-gra-0001.png) CMBS–33(+LAB). The difference between growth rate of CM–INT–51 and CMBS–33 was significant at = 0·02 and 0·001 for Fermentations 1 and 2, respectively.

Samples were drawn from the fermentation at regular intervals and levels of rHBD3 in the medium, acid-washed or in the cell lysate fractions were quantified. As shown in Fig. 5, rHBD3 was detected in the cell lysate fraction on days 3–5. The peptide was secreted from the cell and was detected in the cell-associated fraction on days 2–7, with levels decreasing over the course of the fermentation. Free rHBD3 was also detected in the medium in lower amounts (data not shown), reflecting what was previously observed in laboratory-scale fermentations.

Figure 5.

Production and secretion of human β-defensin-3 (HBD3) during fermentation. Samples (wort plus yeast cells) were taken at regular intervals during the course of the fermentation. Cells were pelleted and HBD3 levels in the cell lysate and cell-associated membrane were detected by ELISA. Levels of HBD3 are shown. Error bars represent the standard error from 3–4 replicates. (■) Lysate and (image_n/jam12382-gra-0008.png) membrane.

To test the bactericidal activity of rHBD3 produced in situ, fermentations were seeded with Lact. brevis at the start of the fermentation. In the absence of rHBD3 (CMBS-33), the bacterial count increased by c. 2·5 logs over the course of the fermentation, while the presence of rHBD3 in CM-INT-51 significantly reduced the level of contaminating bacteria throughout the fermentation (Fig. 6). Thus, we conclude that beta defensin provides protection from infecting bacteria during the course of the fermentation.

Figure 6.

Bactericidal activity of rHBD3 during fermentation. Pilot-scale fermentations were seeded with Lactobacillus brevis c. 45 min prior to yeast pitching. Samples were taken from the fermentation at regular intervals, and bacterial counts enumerated by plating on MRS agar supplemented with 80 μg ml−1 cycloheximide. Error bars represent the standard error from the mean of triplicate samples. (image_n/jam12382-gra-0003.png) CM-INT-51 and (image_n/jam12382-gra-0005.png) CMBS 33.

Bacterial analysis of both pasteurized and unpasteurized beer revealed that very few bacteria that were seeded into the fermentations survived the maturation, filtering and bottling process, even in the absence of pasteurization. Bacterial counts in randomly sampled bottles ranged between 10 and 70 CFU ml−1 in beer from both CM-INT-51 and CMBS33 fermentations.

As bacterial contamination of beer can occur at the bottling stage, we asked whether the addition of synthetic HBD3 could prophylactically prevent bacterial contamination in bottled beer and thereby extend post-production shelf life. To test this, sHBD3 was added to samples of the bottled beer reseeded with either Ped. damnosus or Lact. brevis. The addition of HBD3 at a concentration of 500 ng ml−1 to bottled beer effectively eliminated the beer-spoiling bacteria in pasteurized beer (Fig. 7a,b). As previously observed (Fig. 2), the beer spoiler Ped. damnosus appears to be more sensitive to HBD3 than Lact. brevis. Surprisingly, the pasteurization process appears to enhance the bactericidal activity of HBD3 as bacterial elimination was less effective in nonpasteurized beer (Fig. 7, columns labelled NP). This was most noticeable with Lact. brevis, which persisted in the beer sample even in the presence of 500 ng ml−1 of sHBD3 (Fig. 7b), indicating that pasteurization may enhance the activity of sHBD3.

Figure 7.

Bactericidal activity of sHBD3 in bottled beer. Beer produced from pilot-scale fermentations of the yeast CM-INT-51 or CMBS33, conducted in the presence (+B) of added Lactobacillus brevis, was bottled and pasteurized (P) or left unpasteurized (NP). The bottles were then infected with Pediococcus damnosus (a) or Lact. brevis (b) at 1 × 104 CFU ml−1 and synthetic human β-defensin-3 (HBD3) (sHBD3) was added at the indicated concentration. The bottles were incubated as described in the methods section. Bacteria were enumerated by plating on MRS agar. Error bars represent the standard error from the mean of three replicates. (■) 0 ng ml−1; (image_n/jam12382-gra-0008.png) 25 ng ml−1 and (image_n/jam12382-gra-0009.png) 500 ng ml−1.


In this study, we have explored the use of human defensins as an antimicrobial agents against common beer-spoiling bacteria in industrial lager fermentations. Our data indicate that a synthetic form of human β-defensin-3 (sHBD3) displays antimicrobial activity against common beer-spoiling bacteria in the μmol l−1 range in vitro. This is similar to the bactericidal activity range of the bacteriocin, enterocin L50 when produced in S. cerevisiae (Basanta et al. 2009). The effectiveness of sHBD3 is dependent on the bacterial load. The lager strain of yeast, S. pastorianus CMBS-33, used in this study was at least 100 times less sensitive to sHBD3 than any of the bacterial species tested, and in fact, under fermentation conditions, little or no loss of viability was observed.

A gene cassette encoding for a secretable form of HBD3 was constructed and integrated into XVI chromosome of the yeast strain CMBS-33. Competitive genome hybridization (Bond et al. 2004) analysis of DNA from the parent CMBS-33 and strain CM-INT-51, indicated that the gene was not inserted into any other region of the genome nor were there any other gross genetic changes detected in the genome of the integrated strain (data not shown). Transcription of the integrated copy of HBD3 from the PGK promoter resulted in constitutive expression of rHBD3, with levels peaking at 8–12 h. This pattern of expression is consistent with what has previously been shown for PGK expression under fermentation conditions (James et al. 2003).

Analysis of the fermentation profiles of the parent CMBS-33 and strain CM-INT-51 revealed that the latter displayed a faster fermentation rate. Strain CM-INT-51 also displayed faster growth in wort. We currently do not understand the reason for this positive difference in fermentation rate. It is possible that the insertion of a gene sequence upstream of the gene YPR160W alters its expression. Being allopolyploid, there is more than one copy of YPR160W in the lager strain, and we also know that at least one copy is a hybrid pseudogene, arising from a recombination event between two of the homeologous copies (Usher and Bond 2009). YPR160W encodes for the non-essential gene, glycogen phosphorylase, which is required for the mobilization of glycogen in yeast cells through sequential phosphorolysis of alpha-1,4-linked glucose units in glycogen. Alteration in expression levels of YPR160W may influence the fermentation rate, for example, by increasing the intracellular concentration of glucose. It is also possible that the expression of HBD3 may positively affect fermentation rate, although there is no intuitive rationale for such an effect.

Our results demonstrate for the first time that a secretable form of rHBD3 can be generated in lager yeasts. Using an ELISA assay, rHBD3 was detected in the spent medium at concentrations in the μmol l−1 range. While the levels are lower than that required for the bactericidal effects of synthetic HBD3 in vitro, we subsequently discovered that a substantial quantity of rHBD3 remains bound to the external surface of the cells and also within the cell. The peptide was actively bound to the exterior surface of the cell as incubation with acetic acid released most of the bound rHBD3. Similar dissociation kinetics were also observed in the release of HBD3 bound to glutaraldehyde-fixed Staphylococcus aureus (Harder et al. 2001). The cell-associated fraction of rHBD3 was detected by Western blotting as a band migrating the same molecular weight marker as synthetic HBD3 using an anti-HBD3 antibody. Previous studies have shown that the cationic defensins have an inherent ability to interact with the anionic surfaces of eukaryotic and bacterial cells through electrostatic interactions (Zhang et al. 2010). rHBD3 was detected on the external surface of the cells on Day 2 of the fermentation with levels decreasing thereafter. Levels of HBD3 within the cell lysate remain high up to Day 5 of the fermentation. Based on fermentation performance, such binding does not appear to be harmful to the vitality of the yeast cell.

Our results demonstrate that rHBD3 significantly reduced the growth of Lact. brevis in industrial fermentations. Interestingly, the amount of rHBD3 produced and secreted from the yeast cells during fermentation was substantially less than the concentration required to kill 95% of bacteria as estimated from the in vitro bactericidal assay. The differences in the bactericidal activity of HBD3 in vitro and in vivo may be accounted for by differences in the assays. The bactericidal activity of sHBD3 was dependent on the bacterial load in vitro: the in vitro assay was conducted using a higher initial bacterial load (104 CFU ml−1), while the starting inoculum for the in vivo assay was thirty times lower. The commercial sHBD3 peptide has a purity of 70%, but has not been specifically folded to achieve the correct CYS-CYS linkages. The rHBD3 produced in situ may have proper folding and correct secondary structure. It is also possible that the synthetic peptide may be partially oxidized, which may reduce the formation of di-sulfide bonds within the peptide, thus reducing its activity. The requirement for a reduced level of rHBD3 for bactericidal activity in vivo may also reflect enhanced stabilization of the peptide activity under fermentation conditions. Other components in the wort or cellular metabolites produced during fermentation may also contribute to the bactericidal activity of rHBD3 in vivo. For example, the presence of ethanol in the medium as fermentation proceeds may increase the susceptibility of the bacteria to the actions of rHBD3 through altering the composition of the bacterial membrane. Alternatively, ethanol or other metabolites of fermentation could act directly as bactericidal agents, thus contributing to the overall level of bacterial killing (Koshiro and Oie 1984). The latter is less likely as we did not observe a decrease in bacterial load in CMBS-33 cultures lacking the rHBD3 gene cassette.

As bacterial contamination can occur at the bottling stage of beer production, we also tested the efficacy of sHBD3 when added to bottled beer. Our finding demonstrates that sHBD3 can reduce the bacterial load in pasteurized bottled beer by 95 and 80% for Ped. damnosus and Lact. brevis (added postpasteurization), respectively. The increased sensitivity of Ped. damnosus to sHBD3 in beer is consistent with our finding in the in vitro bactericidal assay (Fig. 2), suggesting that strict anaerobes may be more sensitive to HBD3 than microaerophilic bacteria. Interestingly, we observed that sHBD3 was less effective in nonpasteurized beer samples. The process of pasteurization has been shown to improve the oxidative stability of beer while unpasteurized beer displays a higher degree of free radical formation. Furthermore, long-term storage of unpasteurized beer has been shown to lead to the accumulation of significantly more degraded proteins (Lund et al. 2012). Any or all of these events may influence the observed bactericidal activity of HBD3 in pasteurized beer. It is also possible that residual bacteria present at the end of fermentation may have escaped the filtration and lagering processes and may have been reactivated during the in vitro bactericidal assay. This is unlikely as <30–70 CFU ml−1 was detected in beer samples before the challenge test. This level of residual bacteria could not account for the apparent increase in survival of seeded bacteria in unpasteurized beer.

Taken together, the data presented here indicate that human β-defensin-3 is an effective antimicrobial agent against both Gram-positive and Gram-negative beer-spoiling bacteria and is functional in situ in industrial fermentations. Our data indicate that HBD3 and possibly other classes of defensin peptides from different plant or animal sources may have the potential to be developed as natural biopreservatives for beers and other beverages.


The research reported in this article was supported by grants from the Department of Agriculture, Food and Fisheries (FIRM 06RDTCD411 and 10FPlusNATDEF) to UB.

Conflict of interest

No conflict of interest declared.