key: cord-0042991-7bz25ddz
authors: Pham, T.; Wimalasena, T.; Box, W. G.; Koivuranta, K.; Storgårds, E; Smart, K. A.; Gibson, B. R.
title: Evaluation of ITS PCR and RFLP for Differentiation and Identification of Brewing Yeast and Brewery ‘Wild’ Yeast Contaminants
date: 2012-05-16
journal: J
DOI: 10.1002/j.2050-0416.2011.tb00504.x
sha: 51e8e117cf3caa0aea971a3b81646f93e620bbd6
doc_id: 42991
cord_uid: 7bz25ddz

A reference library of ITS PCR/RFLP profiles was collated and augmented to evaluate its potential for routine identification of domestic brewing yeast and known ‘wild’ yeast contaminants associated with wort, beer and brewing processes. This library contains information on band sizes generated by restriction digestion of the ribosomal RNA‐encoding DNA (rDNA) internal transcribed spacer (ITS) region consisting of the 5.8 rRNA gene and two flanking regions (ITS1 and ITS2) with the endonucleases CfoI, HaeIII, HinfI and includes strains from 39 non‐Saccharomyces yeast species as well as for brewing and non‐brewing strains of Saccharomyces. The efficacy of the technique was assessed by isolation of 59 wild yeasts from industrial fermentation vessels and conditioning tanks and by matching their ITS amplicon sizes and RFLP profiles with those of the constructed library. Five separate, non‐introduced yeast taxa were putatively identified. These included Pichia species, which were associated with conditioning tanks and Saccharomyces species isolated from fermentation vessels. Strains of the lager yeast S. pastorianus could be reliably identified as belonging to either the Saaz or Frohberg hybrid group by restriction digestion of the ITS amplicon with the enzyme HaeIII. Frohberg group strains could be further sub‐grouped depending on restriction profiles generated with HinfI.

The use of pure, uncontaminated cultures of specific strains of brewing yeast for fermentation has been a central tenet of the brewing industry for over a century 4 . Use of pure cultures minimizes the risk of microbial contamination and ensures consistent fermentation performance and product quality. The closed systems employed in modern breweries reduce the risk of contamination by 'wild' yeast, which may be considered to be any yeast occurring other than the specified production yeast. Wild yeasts can therefore be non-Saccharomyces yeast, Saccharomyces yeast species other than brewing production yeast, or even production yeast strains other than those intended for a specific fermentation. In other words, yeast not deliberately used and under full control 20 . Contamination, albeit at a low concentration and rarely influencing the fermentation process or product, is however normal 55 and is used a quality control indicator within modern breweries.

Beer spoilage due to the presence of wild yeast contaminants can take a number of forms and is influenced by the contaminant taxa 10, 31 . Reported problems include the production of off-flavours, particularly phenolic offflavours, which are formed by the carboxylation of ferulic and cinnamic acids 49 , as well as the production of other off flavours such as acetic acid or esters. Competition for nutrients may also occur if wild yeast growth rates are sufficiently fast. An extreme case of competition involves the presence of so-called killer yeast, which can completely replace a brewing yeast through the production of killer factors or zymocins 61 . The presence of small or nonflocculating wild yeast cells can, in addition, result in unacceptably high levels of turbidity and difficulties with clarification 10 .

The wild yeast isolated from raw materials, brewery facilities, malt extract (wort) in fermentation vessels or beer in collection vessels, kegs or bottles are rarely identified to species level and little published data is available regarding the dynamics of wild yeast contamination in the brewery, i.e. the relative abundance of different species at different stages in the production of beer. An improved understanding of wild yeast and their occurrence is necessary to inform decisions regarding the optimization of processes to eliminate unwanted yeast contaminants and thus prevent unnecessary beer spoilage.

Conventional methods used to detect and isolate wild yeast contaminants have mainly been based on morphological traits and especially their physiological abilities 5, 30 . The most popular method for differentiation is cultivation of unknown yeasts on a wide range of selective media, e.g., lysine medium 57 , actidione medium 23 , crystal violet medium 1 , MYGP+copper medium 2, 46 and CLEN media 46 . Incubation at 37°C for detection of wild yeast in lager breweries has also been used 56 . Different organisms have different growth requirements with regard to nutrients, carbon sources and oxygen availability, thus selective media can aid in detection of contaminants. However, no single medium is capable of detecting or differentiating all wild yeast strains and in many cases, e.g. selection of non-brewing Saccharomyces yeast based on Cu-sensitivity, the mechanistic reasons for the differentiation are obscure. Breweries typically focus on detection and differentiation of brewing and non-brewing yeast rather than identification of contaminant yeast taxa. Traditional identification of wild yeasts to species level is a complex, laborious and time-consuming process which requires approximately 50-100 biochemical tests 5, 30, 34 . Final results are usually available several days or even weeks after beer packaging; therefore, these methods are often used as a retrospective view of quality rather than proactive process control 51 .

Though not routinely used in breweries, molecular techniques for the identification of wild yeast contaminants have the advantage over traditional techniques in that they can often be used to accurately identify yeast isolates to species level. One such technique is ITS-PCR involving the amplification of the ribosomal RNA-encoding DNA (rDNA) internal transcribed spacer (ITS) region consisting of the 5.8S rRNA gene and two variable flanking regions ITS1 and ITS2. Interspecific polymorphism in this region allows for differentiation of species 59 . Restriction fragment length polymorphism (RFLP) can be used in conjunction with ITS-PCR to differentiate closely-related species such as those of the Saccharomyces sensustricto complex 18 .

PCR amplification of rDNA and restriction digestion of the amplicon has not been used extensively for the identification of yeasts isolated during brewing or from beer 27, 53 , but has seen wide application in other fermentation industries, particularly for the identification of yeasts occurring during oenological fermentations 13, 14, 17, 18, 21, 22, 36, 43, 44, 48, 62 . The technique has also been used to identify yeasts occurring as contaminants or natural, spontaneous fermenters of cider 15, 39 and rice wine 28 as well as yeast associated with non-alcoholic citrus juice beverages 3, 32 . Despite its accuracy and relative ease of use, this technique has not been widely adopted for routine identification of isolated wild yeast in breweries. One possible reason for this is that there is currently no comprehensive reference library of ITS-PCR fragment sizes and RFLP band sizes for yeast associated with brewery fermentation and beer.

The aims of this study were, firstly, the collation and augmentation of rDNA-PCR (ITS1, 5.8S and ITS2) data and gel-detectable RFLP profiles for known brewery wild yeast contaminants, as well as common production yeast, for use as a reference tool and, secondly, to critically evaluate the potential of this reference library for identification of wild yeast contaminants isolated from industrial fermentation vessels and beer conditioning tanks.

Fifty-nine wild yeast isolates were obtained from wort and beer sampled from five different fermentation vessels and three different conditioning tanks used for the production of two different lager beers. All samples were ob-tained from one brewery and at the same time. Known yeast strains, including several type strains were also obtained from the National Collection of Yeast cultures (NCYC; Norwich, UK), Dipartimento Biologia Vegetale Perugia, Yeasts Industrial Collection (DBVPG; Perugia, Italy), European Saccharomyces cerevisiae Archive for Functional Analysis (EUROSCARF; University of Frankfurt, Frankfurt, Germany) and the VTT Culture Collection (VTT; Technical Research Centre of Finland, Espoo, Finland). Also included was the production lager strain CB11. Yeast strains were maintained on YPD agar (1% yeast extract, 2% neutralized bactopeptone, 2% dextrose and 1.2% agar) at 4°C.

Multiple wort samples (10 mL) and beer samples (100 mL) were filtered through Millipore ® membrane (pore size of 0.45 µm) filters, which were transferred to MYGP+copper plates (0.3% malt extract, 0.3% yeast extract, 0.5% bactopeptone, 1% glucose, 1% agar technical and 100 ppm Cu added as copper sulphate). The high copper concentrations prevented growth of the production lager yeast strains and allowed single colony wild yeast strains to be isolated. These wild yeasts were taken from the original copper plates and subcultured on 100 and 200 ppm copper-supplemented Bacto TM agar (Becton, Dickinson and Company) to ensure that the production yeast strain was not present. All plates were incubated at 25°C for 48 h before subsequent subculturing on plates without additional copper.

Rapid DNA extraction was carried out by transferring a fresh yeast colony (approx. 2 days old) to a 50 µL yeast DNA extraction buffer (2 × 10 -3 M NaOH and 0.001% sarcosine) and boiling for 10 min at 100°C. Sarcosine was included to permeabilise cell membranes. The supernatant containing DNA crude extract was collected by brief centrifugation and stored at -20°C. The PCR reactions were performed as described by White et al. 59 Briefly, 50 µL PCR reaction mixture contained 0.5 µM primer ITS1 (5 TCCGTAGGTGAACCTGCGG 3), 0.5 μM primer ITS4 (5 TCCTCCGCTTATTGATATGC 3), 25 µL Master mix (Quick Load Taq 2× from New England Biolabs), 3 µL DNA crude extract and water. The PCR conditions were as follows: 95°C for 15 min (initial denaturing), 35 cycles of 95°C for 1 min (denaturing), 55°C for 2 min (annealing), 72°C for 2 min (elongation) followed by a final elongation step at 72°C for 10 min. A quantity of 6 µL of product from each reaction was separated on a 1.4% (w/v) agarose gel in TAE buffer and 8 μL of the PCR product was digested with CfoI (Promega), HaeIII or HinfI endonucleases (New England Biolabs) and separated on a 4% agarose gel. Gels were stained with ethidium bromide, visualized under UV light and photographed. Band sizes were calculated with reference to a 100 bp ladder (New England Biolabs). Tentative identification of unknown brewery contaminant yeast was carried out by first comparing ITS amplicon sizes with those of known species, including type strains, to produce a short-list of candidate species and secondly by comparing restriction profiles produced using the enzymes CfoI, HaeIII and HinfI.

Where available, ITS sequence data was obtained from databases such as MycoBank (http://www.mycobank.org/), Genbank (http://www.ncbi.nlm.nih.gov/genbank/) and CBS-KNAW Fungal Biodiversity Centre database (http://www. cbs.knaw.nl/). Virtual restriction profiles of the full length sequences (ITS sequences containing sequences of both ITS1 and ITS4 oligonucleotides) were determined using DNAMAN v4.1 (Lynnon BioSoft) and Clone Manager 5 software (Scientific & Educational Software).

While numerous methods are available for the extraction of DNA from yeast cultures, the one used here was found to be particularly simple and rapid. Better results were obtained than with methods involving extraction with SDS or colony PCR (data not shown). However, any method that produces relatively pure DNA for PCR is sufficient.

Collation of the available data and grouping of yeasts according to ITS amplicon size in ascending order revealed that the many species and indeed genera can be separated according to amplicon size ( Table I) . Species of the genus Candida, with amplicon sizes ranging from 400-700 bp were found in all amplicon size categories. Candida rugosa, C. pararugosa, C. intermedia, C. sake and C. stellata, however, all had amplicon sizes of 500 bp or less. A characteristic of these species was the lack of restriction sites with the HaeIII enzyme. Restriction with HinfI resulted in production of double bands, with restriction sites in each case being close to the centre of the amplified region. The similar amplicon sizes and restriction profiles of species within the group prevent definitive identification of individual species based on ITS/RFLP, though smaller amplicon sizes within this group may suggest C. rugosa or C. intermedia rather than, for example, C. stellata. Predicted bands sizes, based on available sequence data, were similar to observed gel-detectable band sizes, with differences in restricted band sizes typically less than 20 bp (Table I) when different strains are compared. Other Candida species, including C. tropicalis, C. norvegica and C. parapsilosis had larger amplicon sizes (>500 bp) and while the data suggests that C. tropicalis and C. parapsilosis are not differentiable based on ITS amplicon size and restriction pattern, C. norvegica due to its larger size (580 bp) and unique banding pattern may be differentiated using this technique. Banding patterns for C. tropicalis and parapsilosis corresponded well to predicted band sizes (Table I) . Candida species with amplicon sizes above 600 bp included C. mesenterica, C. santamariae (basionym C. beechii) and C. tenuis. The latter species were observed to have similar amplicon sizes and restriction profiles. The exceptions here were the profiles generated with HaeIII and HinfI for C. tenuis by Guillamon et al. 22 , which deviated significantly from the predicted profile. Candida mesenterica, on the other hand could be distinguished due to its larger amplicon size (approx. 650 bp) and lack of restriction with CfoI. Amplicon size and restriction profile with HaeIII were similar to those of Scheffersomyces stipitis (basionym Pichia stipitis), emphasising the requirement for more than one restriction profile to be generated to reliably identify a species or genus. Candida boidinii can be distinguished from other members of the genus by its larger amplicon size (≥700 bp). Care should be taken however to avoid confusion of this strain with Hanseniaspora uvarum when basing identification on visible fragment sizes.

The Dekkera species, D. bruxellensis and D. anomala, while producing similar restriction profiles may be distinguished by the larger size of the D. anomala amplicon (>500 bp) compared to the D. bruxellensis amplicon (≤500 bp) and the tendency of D. anomala to generate a double band with the HinfI enzyme. Dekkera bruxellensis can likewise be distinguished from Pichia fermentans based on the CfoI restriction profile. Of note here is the discrepancy in the D. anomala results obtained by Granchi et al. 21 and the authors of this study who found amplicon sizes of 514 and 540 bp, respectively, compared with an amplicon size of 800 bp obtained by Esteve-Zarzoso et al. 17 and Morrissey et al. 39 Of the Pichia species, those with the smallest ITS amplicon sizes were P. fermentans, P. membranifaciens and P. kudriavzevii (basionym Issatchenkia orientalis), with P. fermentans strains normally having a smaller size (~450 bp) than the other two species. Results suggest that P. fermentans and P. membranifaciens cannot be reliably differentiated based purely on observation of gel-detectable fragment sizes with the enzymes used here. However, it is possible to differentiate these species with XhoII and MaeII restriction enzymes according to in silico data: P. fermentans ITS amplicon is digested into 322 and 123 bp fragments with XhoII and undigested with MaeII, whereas P. membranifaciens ITS amplicon is digested into 175, 170, 105 and 29 bp fragments with XhoII and 260, 125 and 94 bp fragments with MaeII (data not shown). These two species could, in the majority of cases, be differentiated from P. kudriavzevii by the restriction profiles obtained with CfoI and HaeIII.

The two Kazachstania species included here 31 , K. exigua and K. unispora could be distinguished from each other by the profiles generated with HaeIII and HinfI and could furthermore be differentiated from other species due to their large amplicon size and restriction profile with HaeIII. In the same amplicon size class were the Kluyveromyces species, which could not be distinguished from one another based on amplicon size or restriction profile, but could be distinguished from non-Kluyveromyces species as long as more than one restriction profile was available. The profile generated with CfoI was, for example, similar to that obtained from Zygosaccharomyces rouxii. Torulaspora delbrueckii was another species with a large ITS amplicon (800 bp) and which generated numerous bands with CfoI. A distinctive characteristic of this species was the lack of restriction sites for HaeIII. This unrestricted DNA could be distinguished from that of H. uvarum (also without restriction sites) by size. According to in silico data, it is possible to use restriction enzymes other than those employed here to differentiate Kluyveromyces species. Namely, AvaI, HindIII and MslI restriction enzymes do not cleave the K. lactis ITS amplicon, whereas the K. marxianus ITS amplicon is cleaved into two fragments, based on a partial ITS amplicon sequence available, having sizes of 458 and over 240 bp (AvaI), 573 and over 125 bp (HindIII) and 494 and over 204 bp (MslI).

Of the three Zygosaccharomyces species included in this study, all were in the largest amplicon size class with sizes ranging from 725-790 bp. Zygosaccharomyces rouxii could be distinguished from Z. bailii and Z. bisporus based on its smaller amplicon size, but as indicated previously care must be taken to avoid misidentification of this species as Kluyveromyces. The available data suggests that Z. bailii and Z. bisporus are not readily distinguishable using this technique and, furthermore, that there is the possibility of this species being confused with Hanseniaspora uvarum based on amplicon sizes and restriction profiles.

For the remaining non-Saccharomyces species within this library, amplicon sizes ranged from 600-700 bp and these species included Millerozyma farinosa (synonym P. farinosa), Meyerozyma guilliermondii (basionym P. guilliermondii), S. stipitis, Wickerhamomyces anomalus (synonym P. anomala), and Wickerhamomyces subpelliculosus (basionym P. subpelliculosa). (Table I) . Of these, the most difficult to differentiate are the Wickerhamomyces species.

Yeasts belonging to the Saccharomyces sensu stricto group have ITS sizes of 840-880 bp (Table II) , a feature which distinguishes them immediately from the known non-Saccharomyces contaminants. In many cases, restriction profiles were identical in different species. Restriction with HaeIII however reveals that these species fall into one of two groups. The first group yields 4 distinct bands (approx. 310, 230, 170, 130 bp) which correspond in size to the predicted ITS fragments that would be generated with HaeIII digestion of the S. cerevisiae S288c ITS region (Table II) . All S. cerevisiae strains, including the investigated ale strains, belong to this group. Other species that fall into this group (based on observed and predicted restriction profiles) include S. bayanus var. uvarum and S. paradoxus. The second HaeIII restriction group generates three fragments of approx. 480, 230 and 130 bp and includes S. kudriavzevii and S. mikatae 7 .

Interestingly, HaeIII restriction can separate S. pastorianus strains into either restriction group 1 (with 4 bands) or restriction group 2 (with 3 bands); this differentiation is consistent with the study of hybrid groups of S. pastorianus 16 . The majority of S. pastorianus restriction group 1 strains were indistinguishable from S. cerevisiae ale strains as well as the sister laboratory strains BY4741 and S288c. Those belonging to restriction group 2 had profiles similar to those observed with several strains of S. bayanus including the S. bayanus var. bayanus type strain. Previously categorized hybrid group 1 (Saaz) strains included in this study showed the same restriction profiles with individual enzymes. Restriction group 1 lager strains could however be further sub-grouped depending on profiles generated with HinfI restriction of amplicons. This profile was found with three strains belonging to restriction group 1 lager yeasts. Otherwise, HinfI could not differentiate different Saccharomyces species or strains investigated (fragment sizes were approx 360, 350 and 120, with the two larger bands frequently observed as a single band on gels). Strain W34 and the production strain (CB11) used in this brewery had RFLP patterns consistent with Frohberg strains. Some differences were observed when comparing restriction profiles with CfoI, though these differences were as often observed when comparing strains of one species as comparing different species and are therefore not applicable for differentiation of species within the Saccharomyces sensu stricto group.

Wild yeast contaminants, 59 in total, were isolated from samples collected from conditioning tanks and fermentation vessels. Results showed that ITS-PCR/RFLP can be used, to some extent, to differentiate a number of wild yeast taxa present in the brewing process. Wild yeast species putatively identified mainly belonged to the genera Candida, Pichia, Dekkera, Rhodotorula and Saccharomyces. Putative identifications are listed in Table III according to the constructed 5.8S ITS library. All isolates with PCR products of 880 bp belonged to Saccharomyces restriction group 1 (Table III) , which includes S. cerevisiae, S. paradoxus, S. pastorianus hybrid group 2 (Frohberg) and S. bayanus var. uvarum. It is therefore not possible to differentiate the isolated Saccharomyces species. Pichia species were the most commonly isolated yeasts from conditioning tanks but were not detected in fermentation vessel samples. Saccharomyces spp. made up the largest proportion of yeast found in fermentation tanks. Strains belonging to Saccharomyces restriction group 1 were the most abundant contaminants found in fermentation vessels (Table III) . They made up around 82% of the wild yeasts found in these vessels and 10% in conditioning tanks. Approximately 8-10% of Saccharomyces isolates found in both conditioning tanks and fermentation vessels were putatively identified as S. paradoxus based on a distinctive double band generated with HinfI digestion. Dekkera bruxellensis made up 16% of wild yeast isolates in conditioning tanks. Candida species occurred at a low frequency of 3% in conditioning tanks and fermentation vessels. Composition of the wild yeast population varied depending on vessel and beer type; two fermentation vessels were only contaminated by Saccharomyces restriction group 1 species while the others had additional S. paradoxus, C. intermedia/C. pararugosa/ C. rugosa, D. bruxellensis and R. mucilaginosa as contaminants. One lager beer type was contaminated with Pichia spp. only while the other had a mixture of Saccharomyces, Pichia, Candida and Rhodotorula (data not shown).

Restriction analysis of the rDNA region spanning the 5.8S rRNA gene and flanking internal transcribed spacers (ITS1 and ITS2) has previously been shown to be an effective, rapid and simple method to identify a variety of yeasts isolated from alcoholic fermentations. Compared to traditional methods based on selective media and incubation conditions, this method has potential advantages in terms of speed, efficiency and reduced work load 6, 25, 55 . Although the method has been applied for detection of wild yeast contaminants of food and wine, there are few reports of its application for differentiation of wild yeast contaminants in breweries and their occurrence at different stages of brewing process 26, 27, 47, 55, 60 . Yeast contaminants in the brewing industry are rarely identified to species level, possibly due to a lack of appropriate reference strains, the continued use of traditional, non-specific means of identification and the occurrence of misleading information in brewing literature, for example, the name Candida mycoderma has been used to describe a mixture of different yeasts, moulds and bacteria, which contribute to surface films 24,25 rather than the actual species. Additionally, nomenclature for asexual/sexual stages of yeasts cause confusion, e.g., Pichia kudriavzevii (basionym Issatchenkia orientalis) the teleomorph of Candida acidothermophilum (synonym Candida krusei) 54 . For these reasons, we have constructed a reference library for known wild yeast contaminants found in breweries and tested this library using unknown yeast contaminants isolated from industrial fermentation vessels and conditioning tanks. A number of production and non-production yeast strains were included to determine if these can be differentiated using the PCR-RFLP technique. Strains included laboratory strains, ale strains and lager strains from both hybrid groups identified by Dunn and Sherlock 16 .

A comparison of PCR fragment sizes and restriction profiles indicated that the data compiled in Table I could be used to identify a number of non-Saccharomyces wild yeast contaminants. Species with unique restriction profiles that have been observed in two or more independent investigations and with multiple strains include D. bruxellensis, H. valbyensis, K. exigua, K. unispora, M. farinosa, M. guilliermondii, P. kudriavzevii, R. mucilaginosa, S. stipitis, T. delbrueckii and Z. rouxii. Other species may potentially provide reliable restriction profiles for identifi-cation but these first require verification, in most cases because only one strain has been included in one or more studies (C. mesenterica, C. norvegica, Filobasidium capsuligenum, Schwanniomyces occidentalis, W. subpelliculosus, Z. bisporus). In other cases, species within a genus have displayed similar restriction profiles (C. santamariae and C. tenuis or C. intermedia, C. rugosa, C. sake and C. stellata or C. tropicalis and C. parapsilosis or P. fermentans and P. membranifaciens as well as Z. bisporus and Z. bailii), meaning that the profiles generated cannot be considered to be species-specific. In other cases similar profiles are generated with species of different genera (Kluyveromyces lactis and W. anomalus or C. boidinii and H. uvarum) and such profiles can therefore only be considered indicative of a species and not reliable for definitive identification.

In some cases, there are discrepancies in restriction profiles among strains of a given species and, while the majority of these differences are observed with only one of the three restriction enzymes, there are cases in which profiles bear no similarity to each other. One notable example occurs with Dekkera anomala, in which a distinct profile was observed in two independent studies, while a different profile was obtained by two other independent studies. The strongly conflicting results in this case may suggest that mis-identification or mis-labelling of strains may have occurred, mostly likely as a consequence of the rapidly changing taxonomic status of yeasts. Otherwise, small differences in fragment sizes may be related to sequence differences between strains of a given species or due to the way in which the DNA was amplified and restricted or the bands sizes determined. It may be expected that differences in recorded fragment sizes as great as 20 bp are possible simply due to the manner in which band size was calculated and this would account for many of the small size variations reported by different researchers for the same strain of a species. Previous studies suggested that the restriction enzyme HaeIII could divide four siblings of the Saccharomyces sensu stricto group into two categories: S. bayanus/S. pastorianus which has a three-band pattern and S. cerevisiae/S. paradoxus which has a four-band pattern 18, 27, 38 . Interestingly, we found that HaeIII digestion of S. pastorianus amplicons could also yield either a three or four band pattern, depending on the strain involved. Studies in recent years have shown that the S. pastorianus lager strains belong to at least two discrete groups 16, 35 . The hybridization event 41, 58 which led to the creation of S. pastorianus, believed now to be a cross between an S. cerevisiae ale strain and the recently discovered S. eubayanus 33 , and the subsequent isolation of the hybrid, is believed to have occurred on at least two separate occasions. These hybridizations led to the formation of the Saaz type (group 1) strains used originally in Denmark and Bohemia 16 and the Frohberg type (group 2) used mainly in the Netherlands and Germany. It is probable that the success and proliferation of these strains in the brewing industry was brought about by improved cryotolerance due to the presence of the S. (eu)bayanus subgenome, as S. bayanus is known to be considerably more cryotolerant than S. cerevisiae 50 and S. eubayanus is found naturally in cold environments 33 . Increased tolerance to stress is typical of hybrid strains and appears to bestow an advantage to strains, particularly those used in the brewing and winemaking industries 47 . The increase in genome stability associated with the polyploid hybrid condition 47 may in fact encourage adaptive evolution of yeasts to their environment. Despite both lager hybrid groups sharing phenotypic characteristics allowing them to ferment wort efficiently at colder temperatures, they are genomically distinct and representatives of both groups were included in this study to determine the potential for ITS-PCR/RFLP to differentiate the two groups. This is the first study to directly compare ITS regions of strains from both of the known hybrids. Results were consistent with those reported by Dunn and Sherlock 16 who used cross genome hybridization to differentiate the two lager strain types. In this study, the HaeIII enzyme was found also to distinguish the two hybrid groups, with one generating profiles typical of S. cerevisiae (hybrid group 2) and one generating profiles typical of S. bayanus var. bayanus (hybrid group 1). Dunn and Sherlock 16 and Nakao et al. 41 have established that, of the two hybrid groups, hybrid group 2 has retained more of the S. cerevisiae genome than the other, with 16 S. cerevisiae-type chromosomes, 12 S. bayanustype chromosomes and eight chimeric chromosomes 16, 41 . Hybrid group 2 strains appear to have also inherited their ITS region from S. cerevisiae.

Results of an investigation by Fernández-Espinar et al. 18 showed that S. bayanus and S. pastorianus strains could not be distinguished by ITS-PCR/RFLP using oligonucleotide primers and restriction enzymes identical to those used in this study 18 . Similarly, Manzano et al. 37 were unable to differentiate those yeasts using PCR-DGGE 37 . It was supposed that this result was due to the similarity between the ITS regions of both S. bayanus and S. pastorianus, leading to the difficulty in designing specific primers to distinguish them 29 . Those studies, however, only included S. pastorianus strains belonging to hybrid group 1 (Saaz type) and results are therefore in accordance with those of the current study. Barszczewski and Robak 6 found that RFLP patterns generated by a strain of S. pastorianus could not be distinguished from those of several S. cerevisiae strains. This result may have been influenced by the oligonucleotide primers and restriction enzymes chosen in that study or may have been due to the S. pastorianus strain belonging to hybrid group 2 and therefore having an ITS region similar to that of S. cerevisiae. Likewise, Tornai-Lehoczki and Dlauchy 53 found that four lager yeast strains could not be differentiated from three type strains of S. cerevisiae (and 12 ale strains) but were different to three type strains of S. pastorianus. Again, this result may be influenced by the hybrid group to which the strains belonged. The reference strains of S. pastorianus used in that study all belonged to hybrid group 1, as determined by Dunn and Sherlock 16 . The four lager strains used in the study may conceivably belong to hybrid group 2 (three of the strains originated in Germany and could be closely related).

Restriction endonucleases used in this study, as well as distinguishing the S. pastorianus strains as either Saaz or Frohberg type, also placed the Frohberg strains into one of two groups depending on the RFLP profile generated with HinfI. The reason for this separation is not known but suggests a different post-hybridization history for these lager strains.

We could not distinguish brewing and non-brewing Saccharomyces cerevisiae strains from each other based on the technology described probably due to a low level of phylogenetic separation of industrial and non-industrial strains of the species. Based on 5.8S rDNA restriction patterns, they all independently belong to the S. cerevisiae species. However, one study by Yamagishi et al. 60 employed a combination of specific PCR of flocculation gene FLO1 and amplification of rDNA followed by RFLP to distinguish brewing and non-brewing strains suggesting that differentiation of these may be possible with the right combination of amplicon and restriction enzyme. Similarly, Barszczewski and Robak 6 (2004) applied ITS-PCR RFLP and RAPD (randomly amplified polymorphic DNA) to discriminate between brewing and wild yeast isolates.

We have attempted to identify unknown contaminants found in conditioning tanks and fermentation vessels using the constructed ITS library. With reference to current taxonomy, 90% of isolates in fermentation vessels were found to be Saccharomyces spp. while Pichia species appeared to be dominant in conditioning tanks with 61% of the isolates further identified either as P. fermentans or P. membranifaciens. The genus Pichia is considered the most common of the non-fermentative spoilage yeasts which can cause turbidity and estery off-flavour of beer 8, 45 . The predominance of Pichia in conditioning tanks, but not in fermentation vessels, highlighted the possible effect of oxygen in beer post-fermentation. Air, if inadvertently introduced, can be the source of yeast or bacterial contamination, but even if not, can allow the growth of aerobic yeasts such as those from the Debaryomyces, Dekkera, or Candida genera 9 . A more detailed study with frequent sampling of all stages in the brewing process is, however, necessary to fully appreciate the wild yeast population changes that occur in response to changing environmental conditions.

The ITS-PCR RFLP technique has a sufficient level of resolution to identify a number of yeasts associated with industrial breweries. Its advantages in terms of reproducibility and ease of use may overcome limitations of classic identification methods. It should however be noted that certain profiles were only genus specific, while others profiles were not specific to either species or genus. Of note was the fact that all profiles generated from the 59 unknown brewery-isolated yeasts corresponded to a profile in the constructed library. The library therefore may not, in all instances, be reliably used to identify an isolate to species level, but may still be used to at least produce a list of possible candidate species which can be verified by other means.

Identification of wild yeasts, even if not to species level, may help brewers to detect if specific problems are occurring at specific steps of the brewing process. Because different species and genera occur under different conditions, the presence of particular yeast types may be a good indication of which process parameters need to be reviewed. For example, Candida, Kluyveromyces, Pichia and Torulaspora are opportunistic contaminants and usually occur during the aerobic phase of fermentation. Although the majority of these yeasts are obligate aerobes, some Candida and Torulaspora species show some growth in anaerobic conditions. Dekkera species can ferment sugar to ethanol when oxygen is available and their presence is associated with accumulation of high concentrations of acetic acid 12 .

Individual Saccharomyces sensu stricto species cannot be identified using this technique. Saccharomyces species can, however, be separated into two different groups based on the restriction profiles generated with HaeIII. This technique appears to be a simple and effective way to identify known lager yeast strains as belonging to either the traditional Saaz or Frohberg hybrid groups.

In this study, we generated and tested a 5.8S ITS RFLP library for brewery wild yeast contaminants as well as ale and lager brewing yeast strains based on findings in the literature. However, there are some yeasts reported as being brewery contaminants which have not been included here. Further studies should be conducted in order to strengthen and complete the database. ITS/ RFLP profiles have yet to be generated for Candida solani, C. humilis, C. oleophila, C. versatilis, Debaryomyces marama, Kodamaea ohmeri (synonym Pichia ohmeri) and Wickerhamomyces onychis (basionym P. onychis) 5, 25 .

European Brewing Convention. Analytica Microbiologica: Part II. Institute of Brewing: London

Copper media for wild yeast detection

Yeast species associated with orange juice: evaluation of different identification methods

A history of research on yeasts 3: Emil Fischer, Eduard Buchner and their contemporaries

Yeasts, Characteristics and Identification

Differentiation of contaminating yeasts in brewery by PCR-based techniques

Chimeric genomes of natural hybrids of Saccharomyces cerevisiae and Saccharomyces kudriavzevii

Brewing: Science and Practice

Wild yeasts in brewing and distilling

Microbiological Methods in Brewing Analysis in Brewing Microbiology

Role of oxygen on acetic acid production by Brettanomyces/ Dekkera in winemaking

Dynamics of indigenous yeast populations during spontaneous fermentation of wines from Mendoza

Molecular analysis of yeast population dynamics: Effect of sulphur dioxide and inoculum on must fermentation

Yeast ecology in French cider and black olive natural fermentations

Reconstruction of the genome origins and evolution of the hybrid lager yeast Saccharomyces pastorianus

Identification of yeasts by RFLP analysis of the 5.8S rRNA gene and the two ribosomal internal trancribed spacers

RFLP analysis of the ribosomal internal trancribed spacers and the 5.8S rRNA gene region of the genus Saccharomyces: a fast method for species identification and the differentiation of flor yeasts

Genbank

Yeast classification

Rapid detetion and quantification of yeast species during spontaneous wine fermentation by PCR-RFLP analysis of rDNA ITS region

Rapid identification of wine yeast species based on RFLP analysis of the ribosomal internal transcribed spacer (ITS) region

The use of control levels of actidione for brewing and non-brewing strain differentiation

The use and understanding of media used in brewing microbiology. Media for wild yeast

Specific spoilage organisms in breweries and laboratory media for their detection

Flow cytometric detection of wild yeast in lager breweries

Phenotypic and genetic diversity of Sacchaomyces contaminants isolates from lager breweries and their phylogenic relationship with brewing yeasts

Molecular identification of yeast species associated with 'Hamei'-a traditional starter used for rice wine production in Manipur

PCR differentiation of Saccharomyces cerevisiae from Saccharomyces bayanus/Saccharomyces pastorianus using specific primers

Classification of yeasts

Phylogenetic circumscription of Saccharomyces, Kluyveromyces and other members of the Saccharomycetaceae, and the proposal of the new genera Lachancea, Nakaseomyces, Naumovia,Vanderwaltozyma and Zygotorulaspora

Identification of yeast species from orange fruit and juice by RFLP and sequence analysis of the 5.8S rRNA gene and the two internal transcribed spacers

Microbe domestication and the identification of the wild genetic stock of lagerbrewing yeast

Conventional and rapid methods for yeast indentification

Sequence diversity, reproductive isolation and species concepts in Saccharomyces

Molecular profiling of yeasts isolated during spontaneous fermentations of Austrian wines

PCR-DGGE differentiation of strains of Saccharomyces sensu stricto

Intergenic transcribed spacer PCR ripotyping for differentation of Saccharomyces species and interspecies hybrids

The role of indigenous yeasts in traditional Irish cider fermentations

Mycobank

Genome sequence of the lager brewing yeast, an Interspecies Hybrid

Molecular genetic study of introgression between Saccharomyces bayanus and S. cerevisiae

Yeast populations residing on healthy or Botrytis

The effect of Debina grapevine indigenous yeast strains of Metschnikowia and Saccharomyces on wine flavour

Biotechnology, physiology and genetics of the yeast Pichia anomala

A comparison of new and traditional techniques for the detection of brewery contaminants

The complex and dynamic genomes of industrial yeasts

Application of whole genome amplification and quantitative PCR for detection and quantification of spoilage yeasts in orange juice

Phenolic off-flavour problems caused by Saccharomyces wild yeast

Effect of growth media and strains on structural stability in small chromosomes (Chromosome I, VI and III) of bottom-fermenting yeast

Wild yeasts in brewing: a review. The Inst. of Brewing & Distilling Africa Sect

Microbial attachment and biofilm formation in brewery bottling plants

Delimination of brewing yeast strains using different molecular techniques

A method for the detection of Issatchenkia orientalis in a baker's yeast factory

Detection and identification of wild yeasts in lager breweries

Growth of Saccharomyces cerevisiae and Saccharomyce uvarum in a temperature gradient incubator

Utilisation of lysine by yeasts

Saccharomyces arboricolus sp. nov., a yeast species from tree bark

Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics

Differentiation between brewing and non-brewing yeasts using a combination of PCR and RFLP

The Yeasts

Dynamics and diversity of non-Saccharomyces yeasts during the early stages in winemaking

Molson Coors, Burton-on-Trent, UK, are thanked for their technical assistance as well as their permission to publish this work. The authors gratefully acknowledge the gift of the Group 1 and 2 lager yeast strains from Gianni Liti