Reduction of Brettanomyces bruxellensis Populations from Oak Barrel Staves Using Steam

Reduction of Brettanomyces bruxellensis Populations from Oak Barrel Staves Using Steam Zachary M. Cartwright, 1 Dean A. Glawe, 2 and Charles G. Edwards 1 * Abstract: Brettanomyces bruxellensis, a spoilage yeast associated with red wines, can be difficult to remove from oak barrels. New (16 L) and used (225 L) barrels representing French or American oaks with different toasting levels were obtained. A commercially prepared Cabernet Sauvignon wine was transferred into the new barrels and inoculated with B. bruxellensis for 6 to 7 mos of aging. Upon disassembly of all barrels, yeast penetration into the wood was evaluated by collecting shavings from 2.5 cm diameter holes or by sawing staves into 3 10 cm blocks and then into 4 mm-thick cross sections. Additional blocks from the center of staves were exposed to steam for different times before being sawn into cross sections. Shavings and cross sections were transferred to either enhancement (shavings) or wine recovery (cross sections) media to detect culturable cells. In general, staves contained populations of 10 3 cfu/mm 3 at depths up to 4 mm, with lower populations ( 10 2 cfu/mm 3 ) occasionally detected between 5 to 8 mm, depending on the oak species or barrel age. During steaming, 2 to 3 min was required to achieve 55 C at a depth of 4.5 mm, but slightly longer times were needed for 9.5 mm (~4 min). Yeast present in 0 to 4 mm cross sections required a total steaming time of 9 min for inactivation, whereas 12 min was necessary for a depth of 5 to 9 mm. Because it is not possible for wineries to routinely determine yeast penetration depths in individual barrels, a minimum steaming time of 12 min can be used to limit future infections by B. bruxellensis in contaminated barrels. Key words: Brettanomyces, heat, microbial populations, staves, wine barrels Brettanomyces bruxellensis is generally regarded as detrimental to wine quality during the aging of wines in oak barrels (Fugelsang and Edwards 2007). This yeast is known to produce several undesirable compounds, including volatile phenols, specifically, 4-ethylphenol (4-EP) and 4-ethylguaiacol (4-EG). Whereas 4-EP has an aroma described as smoky, horse sweat, or cow manure, 4-EG has been described as being medicinal, spicy, or clove (Chatonnet et al. 1992, Licker et al. 1998, Romano et al. 2009). Even after various rinsing, cleaning, or sanitizing protocols are applied, viable cells of B. bruxellensis tend to persist in barrels, in part because of the structure of the wood. The porous network of xylem vessels allows cells to penetrate staves up to 8 mm, a depth that may correspond to that of wine pigments (Malfeito-Ferreira et al. 2004, Barata et al. 2013). Most certainly, yeast recovery from any given stave would be influenced by factors known to affect wood porosity such as oak species, forest location, and coopering methods such as toasting (Chatonnet and Dubourdieu 1998, del Alamo- 1 School of Food Science and 2 formerly with the Department of Plant Pathology, Washington State University, Pullman, WA 99164-6376. *Corresponding author (edwardsc@wsu.edu; tel: 509-335-6612; fax: 509-335-4815) Acknowledgments: The authors gratefully acknowledge the Washington State Grape and Wine Research Program, the Ivory Tower Scholarship Fund, regional wineries, Mr. Gordon Steel (Rio Grande Vineyards and Winery, Las Cruces, NM), and Washington State University (viticulture/enology program, Franceschi Microscopy and Imaging Center, and School of Food Science) for financial and material support. Manuscript submitted Feb 2018, revised May 2018, accepted May 2018 Copyright 2018 by the American Society for Enology and Viticulture. All rights reserved. doi: 10.5344/ajev.2018.18024 Sanza and Nevares 2017). In agreement, González-Arenzana et al. (2013) analyzed shavings collected from contaminated commercial barrels and reported that the yeast was present in higher populations in French (~10 4 cfu/g) than in American (~10 3 cfu/g) oak staves. As an infection of only 6 cells/ml can result in the production of 1 mg/l 4-EP (Chatonnet et al. 1997), any effective sanitizing method must be able to reduce yeast populations not only on the surface of staves, but also those present at depths of up to at least 8 mm. To reduce microbial populations in barrels, both thermal and nonthermal methods have been studied. Traditionally, barrels are filled with 50 to 70 C water or are steamed with a wand-type system for a specified amount of time. In support of this practice, B. bruxellensis is sensitive to heat, with inactivation occurring at 50 C or at lower temperatures if ethanol is present (Couto et al. 2005). D-values, the time at a certain temperature to kill 90% of exposed microorganisms, range between 12 to 80 sec for 55 C when a matrix of water or buffer is used (Couto et al. 2005, Fabrizio et al. 2015). More recently, a number of nonthermal methods were studied, but with varied success. For example, a microwave treatment (3000 W for 3 min) of barrels filled with water reduced populations only by 35% (French oak) to 67% (American oak) up to a stave depth of 8 mm (González-Arenzana et al. 2013). Guzzon et al. (2013) noted that use of 1 mg/l ozone for 30 min inactivated B. bruxellensis, although the sanitizer can be hindered by slow diffusion into staves (Palacios et al. 2012) and loses effectiveness due to the presence of organic matter (Guzzon et al. 2011). Finally, culturable cells could not be recovered after application of high-power ultrasonic sound waves (17 W/L for 3 min) to barrels containing 60 C water, but this technique exhibits limited stave penetrations of up to 4 mm (Schmid et al. 2011). 400

Reduction of B. bruxellensis Populations 401 Because of limitations associated with nonthermal methods, hot water and/or steam continue to be the best options for reducing microbial populations from barrels. However, an establishment of specific temperature and time relationships that would eliminate B. bruxellensis present in oak staves has been lacking. Although hot-water washing (60 C) reduces yeast populations, Schmid et al. (2011) concluded that the method does not satisfactorily sanitize the surface of wine barrels, in agreement with Barata et al. (2013), who compared various barrel-cleaning/sanitizing protocols, including steaming. However, Schmid et al. (2011) relied upon oak pieces soaked in inoculated nutrient broths as opposed to staves inoculated by storing infected wine under commercial conditions. Although Barata et al. (2013) concluded that steaming 4-year old commercial barrels would not eradicate B. bruxellensis, treatment times were not optimized, that is, barrels were steamed for only 10 min regardless of initial populations or degree of yeast penetration. In contrast, Fabrizio et al. (2015) concluded that immersion of barrels in water at 60 C provided an estimated 7-log reduction in population, thereby practically eliminating the yeast from barrels, depending on the initial population. Although Fabrizio et al. (2015) represented one of the few reports that have examined heat penetration into barrel oak staves, these authors only studied a penetration depth of 8 mm, and neither oak species nor toasting level was described. The objective of this research therefore was to determine the impacts of oak species, barrel-toasting levels, stave location within barrels, and barrel age on the recovery of B. bruxellensis from staves before and after application of steam. Materials and Methods Wine recovery medium. A commercially produced, blended red wine (ph 3.5, 11.0% [v/v] ethanol) was obtained from a commercial winery. Total sulfur dioxide (SO 2 ) was reduced to <3 mg/l by adding hydrogen peroxide, as confirmed by the aeration-oxidation method (Buechsenstein and Ough 1978), and the ph was adjusted to 3.75 with 5 M NaOH. To encourage yeast growth in the wine, 0.5 g/l glucose, 0.5 g/l fructose, 0.1 g/l yeast extract, and 20 mg/l p-coumaric acid (Sigma-Aldrich Chemical Co.) were added. Growth of molds and/or bacteria was inhibited by adding biphenyl (0.02% [w/v]) or chloramphenicol (0.005% [w/v]), respectively. Prior to inoculation, the wine was sterilized using 0.22 µm absolute filters (Millipore). Strains and starter cultures. B. bruxellensis strains I1a and E1 were obtained from the Washington State University culture collection (Jensen et al. 2009). Starter cultures were prepared by transferring a single colony into 10 ml Yeast Mold (YM) broth (ph 4.5, Difco) and incubated at 27 C for 10 days. Once populations reached 10 6 cfu/ml, 100 µl was inoculated into 100 ml YM broth (ph 4.5, 5.0% [v/v] ethanol) with growth monitored spectrophotometrically at 660 nm. Cells were harvested by centrifugation (2000g for 20 min), washed twice with 0.2 M Na 2 HPO 4 (ph 7.0) buffer, and suspended in 25 ml wine recovery medium (WRM) prior to inoculation into wines. Oak barrels. Small (16 L) and large (225 L) barrels were used as part of this research. New barrels (16 L) made from American (Quercus alba) or French (Quercus petraea) oaks were obtained from Tonelería Cordobesa S.L. (Montilla) and produced with either light or heavy toasting levels (eight of each type). Each barrel was constructed of ~24 staves having 20-mm-thick midpoints. In addition, three-year-old 225 L barrels made from American oak with medium-toheavy toasting (Canton) or French oak with medium-to-light toasting (Artisan Barrels & Tanks Inc.) were received from a commercial winery (two of each type). All four barrels were identified by the winery as being infected with an unidentified strain(s) of B. bruxellensis and received a hot-water wash and a cold-water rinse, and were gassed with SO 2 prior to transport. These barrels were constructed of ~30 staves having 25-mm-thick midpoints. As noted by Boulton et al. (1996), individual barrels represent a fairly random wood sample given that the staves are typically sawn from different portions of different trees. Inoculation of 16 L barrels. The barrels were filled to minimal headspace with a commercially prepared Cabernet Sauvignon wine (ph 3.39, 6.09 g/l titratable acidity as tartaric acid, 13.5% [v/v] ethanol, and 18 mg/l free SO 2 ) that had been filtered through 0.45 µm absolute filters (3M Company). To encourage yeast growth, 0.5 g/l fructose, 0.5 g/l glucose, and 0.1 g/l yeast extract were dissolved in 200 ml wine, filtered through 0.22 µm, and added to each barrel under aseptic conditions. A high-performance food-grade silicon (American Sealants Inc.) was used as a sealant against any leakage. One barrel of each oak type (American or French) and toasting level (light or heavy) was inoculated with B. bruxellensis strain I1a or E1 at ~10 2 cfu/ml (eight infected barrels total). For each infected barrel, a noninoculated barrel served as an experimental control (eight control barrels total). All barrels were stored at 13 C and 70% relative humidity. Every other week, barrels were topped off with sterile-filtered wine and stirred for 30 sec using a sterilized stainless-steel barrel wand (Napa Fermentation). After 6 to 7 mos, the barrels were drained and sequentially rinsed with (1) tap water (<4 mg/l chlorine), (2) water at 46 C, (3) 100 mg/l total SO 2 in distilled water, and (4) water at 46 C (following guidelines set by regional wineries) prior to air drying. Barrel disassembly. Staves from 16 L and 225 L barrels were numbered clockwise from the bunghole, with those between 1100 and 1300 hr and 1700 and 1900 hr positions classified as top and bottom staves, respectively (approximately five staves per classification). All other staves were considered to be from the sides of barrels. Following barrel disassembly, staves were placed in sterile sampling bags (Labplas) and stored at 13 C. Yeast penetration. Penetration of B. bruxellensis into staves during wine aging was determined with two different methods. In the first method, four top or bottom staves were randomly selected from each barrel size and type. Staves were initially sawn with a table saw into 3 10 cm blocks at midpoints and then with a band saw into four horizontal cross sections or layers (0 to 4, 5 to 9, 10 to 14, and 15 mm).

402 Cartwright et al. Cross sections were individually placed into sterile 100 ml sampling bags (Nasco) and stored overnight at 13 C prior to addition of 50 ml WRM. All bags were incubated at 20 C for 30 days and rotated every three days to ensure wine contact with all surfaces. A second method relied on using a 2.5 cm Forstner drill bit to bore holes (36) into the midpoints of 18 bottom or lateral/side staves at 2 mm depth increments. The collected shavings were transferred into 25 ml EBB Brettanomyces enhancement medium (Renouf and Lonvaud- Funel 2007), which was then incubated at 20 C for 12 hr with shaking (100 rpm). EBB was used on oak shavings instead of WRM because of its ability to effectively remove yeasts from surfaces and to provide a better estimate of resident Brettanomyces populations. For both methods, saw blades, drill bits, and surfaces were regularly wiped with 70% (v/v) ethanol between the sawing of staves previously subjected to different heat treatments to prevent cross-contamination among shavings, blocks, or cross sections. Steam treatment of staves. Lateral/side staves (eight per barrel) were sawn into 3 10 cm blocks as previously described. The blocks were randomly placed onto a custommade, stainless-steel plate (34 cm diam, 1.0 cm thickness) with 3.25 10.25 cm rectangular apertures. Metal strips were welded to the underside so that the oak samples would be flush to the metal surface, and any gaps between apertures and blocks were filled with weather-stripping tape. Once the blocks were inserted, the steel plate was placed over a boiling steam kettle containing enough distilled water for the plate surface to be 30 cm above the liquid. A plywood cover was overlaid onto the steel plate during heating. At designated times, the blocks were removed from the steel plate and placed into sterile bags before being sawn into horizontal cross sections. The cross sections were individually placed into sterile 100 ml sampling bags containing 50 ml WRM for incubation at 27 C for at least 60 days. Heat penetration into 3 10 cm blocks was evaluated with thermocouples (Omega Engineering) installed into 1.0 mm diam holes drilled 1.5 cm into the sides of blocks. Analyses. Yeast culturability was determined with the Wallerstein differential medium (Difco) via the spread plate method either manually or with an Autoplate 4000 spiral plater (Spiral Biotech). All plates were incubated at 27 C for six days prior to enumeration. Representative yeast colonies were initially identified based on colony and cell morphologies (Edwards 2005) and later confirmed by genetic internal transcribed spacer sequencing (MIDI Labs) of three randomly selected colonies per barrel. Chemical analyses included ph and titratable acidities assessed with standardized methods (Edwards and Watson 2013), 4-EP and 4-EG by gas chromatography-mass spectroscopy (Jensen et al. 2009), and ethanol with an ebulliometer (Alla). Scanning electronic microscopy (SEM) was performed on stave midpoints obtained with a 9.5 mm plug cutter. Plugs were horizontally sectioned into 2 mm cross sections and desiccated overnight before fixation on metal stubs (Flegler et al. 1993). A mini-plasma sputter coater (Anatech) applied a uniform 20 nm gold coat on samples prior to visualization with an S-570 SEM (Nissei Sangyo America Ltd.). Analysis of variance and Tukey s honest significant difference for mean separation were determined at p 0.05 with XLSTAT software (Addinsoft). Results Yeast penetration into staves. With an initial population of 10 2 cfu/ml in wines incubated in 16 L barrels, a minimum of 14 wks was required for B. bruxellensis strain I1a (Figure 1A) or E1 (Figure 1B) to achieve populations of >10 6 cfu/ml. Both strains consistently entered logarithmic growth slightly faster in the heavily toasted French oak barrels than in staves from the lightly toasted French oak or any of the American oaks. At the same time, B. bruxellensis was never detected in wines present in the eight uninoculated barrels (results not shown). After 6 to 7 mos of incubation with the infected wines, the 16 L barrels were disassembled, and staves were observed with SEM. The collected images revealed possible structural differences among the oak species and confirmed the presence of yeast cells in surface and subsurface oak. Generally, xylem vessels appeared larger in diameter and less dense/ compact in French oak than in American oak, but a lack of sufficient imaging precluded accurate measurements of differences between the oak species. Despite potential physical differences, the xylem vessels and ray parenchyma provided Figure 1 Culturabilities of Brettanomyces bruxellensis strain I1a (A) or E1 (B) in Cabernet Sauvignon wine aged in light- or heavy- toasted barrels made from American or French oaks. Data represent means ± standard errors (n = 3).

Reduction of B. bruxellensis Populations 403 a porous network where the majority of suspected B. bruxellensis could be found. Although large yeast populations were observed within the first 4 mm of American oak samples, cells were not found at greater depths, in contrast to French oak where cells were commonly observed at depths 6 mm (Figure 2A). In addition to individual cells, pseudohyphae structures were also observed, most apparently between mm in American oak and mm in French oak (Figure 2B). The estimated lengths of these structures ranged between 10 and 15 µm, with several reaching 20 µm. Although the pseudohyphae could not be positively identified as being from B. bruxellensis, these structures were often surrounded by cells exhibiting shapes and sizes characteristic of this yeast. Depths of penetration by B. bruxellensis in 16 L barrels were determined with two separate methods. First, center portions of top and bottom staves were sawn into cross sections that were then incubated in WRM to detect viable yeasts. For top staves, penetration was limited to the first 4 mm from inside-to-outside the barrel, with slight differences among oak types and/or toasting levels (Figure 3A and 3C). Notably, populations of 10 3 to 10 4 cfu/ml were present in the WRM Figure 2 Scanning electron micrographs from 16 L French barrel staves infected with Brettanomyces bruxellensis, illustrating yeast cells located at 6 to 8 mm stave depths (A) and presence of pseudohyphae structures, as indicated by arrows (B). Figure 3 Culturabilities of Brettanomyces bruxellensis strain I1a (A, B) or E1 (C, D) from top (A, C) or bottom (B, D) staves from light- or heavytoasted 16 L American or French oak barrels. Dashed lines refer to presumed populations below the limit of detection. Data represent means ± standard errors (n = 4).

404 Cartwright et al. after only one day of incubation. For all 0 to 4 mm cross sections, populations eventually achieved >10 6 cfu/ml. In contrast, populations of I1a or E1 present in bottom staves from French oak barrels consistently entered logarithmic growth faster than those from staves made from American oak (Figure 3B and 3D), possibly because of higher initial populations (p 0.05). Furthermore, French oaks (light and heavy toasts) contained viable yeast cells present in the 5 to 9 mm cross section, unlike the American oaks in which no growth was detected after incubation for 30 days (Figure 3B and 3D). Depths of penetrations were also determined in side/lateral and bottom staves by collecting stave shavings produced by using Forstner drill bits and incubating them in EBB recovery medium. Here, results similarly supported the notion that viable populations were present at greater depths in French oak than in American oak (Table 1). In fact, populations were recovered from up to 6 mm (light toast) or 8 mm (heavy toast) of French oak. Additionally, heavy toasted staves exhibited slightly larger B. bruxellensis populations than the light toasted staves. Some similarities and some differences were observed between the 225 L and 16 L barrels. By incubating stave cross sections in WRM, B. bruxellensis could only be recovered from top staves from the 225 L barrels within the 0 to 4 mm cross sections (Figure 4A). Consistent with the 16 L French oak barrel analyses, viable populations were also obtained from the 5 to 9 mm cross sections in 225 L French oak bottom staves (Figure 4B). Unlike the 16 L American oak barrel staves, B. bruxellensis was also recovered from the 5 to 9 mm cross sections from the 225 L American oak staves. Consistently, longer incubation in WRM was required for cross sections from 225 L barrels than from the 16 L barrels. Whereas B. bruxellensis from 16 L barrels could be detected after one day of incubation (Figure 3), 4 to 10 days of incubation in WRM was required to detect viable cells from all 225 L barrel samples (Figure 4). Regardless of incubation requirements, analysis of 225 L barrel shavings produced by a Forstner drill bit (Table 1) was in agreement with the recovery of Brettanomyces from different oak layers in 225 L bottom staves (Figure 4B). Most notably, populations of 10 1 or 10 2 cfu/mm 3 were recovered from mm (American) and 6 to 8 mm (French) cross sections, respectively, after removing samples following 12 hr incubation (Table 1). Furthermore, yeast cells were never recovered from the 225 L barrels beyond 8 mm, even after 60 days of incubation in WRM. Besides indicating yeast penetration, Table 1 also illustrates that the depth of polymeric pigments and pigmented tannins into staves from 16 L or 225 L barrels was typically correlated with yeast recovery data. Accordingly, pigments moved the farthest through 225 L French barrels (5.17 mm) and the shortest for 16 L American barrels ( 3.8 mm). In individual samples, yeasts were occasionally detected beyond Table 1 Penetration of Brettanomyces bruxellensis and red wine pigments into oak lateral/side staves from different barrel sizes, types, and toasting levels (n = 36). a Barrel size and age 16 L 1-year-old 16 L 1-year-old 16 L 1-year-old 16 L 1-year-old 225 L 3-year-old 225 L 3-year-old Oak type and toast level American Light American Heavy French Light French Heavy American Mediumheavy French Mediumlight Pigment penetration (mm) Stave layer (mm) 3.46 a 0 to 2 3.78 a 0 to 2 4.56 c 0 to 2 6 to 8 4.65 c 0 to 2 6 to 8 8 to 10 4.44 c 0 to 2 6 to 8 5.17 d 0 to 2 6 to 8 8 to 10 Population (log 10 cfu/ mm 3 ) 3.62 b 2.73 cd * b 3.65 ab 2.84 c * 3.75 a 2.95 c 2.09 f * 3.81 a 3.02 c 2.13 ef 1 * 1.75 g 2.29 e 1.60 g * 1.85 g 2.64 d 2.30 e 2.08 f * a Mean values within a column followed by different letters are significant at p 0.05. b Asterisks indicate 30 cfu/ml for 60 days in the EBB recovery medium. Figure 4 Culturabilities of unknown strain(s) of Brettanomyces bruxellensis from top (A) or bottom (B) staves from medium-heavy toasted American or medium-light toasted French 225 L oak barrels. Dashed lines refer to presumed populations below the limit of detection. Data represent means ± standard errors (n = 4).

Reduction of B. bruxellensis Populations 405 the pigments, but only the 225 L French oak barrels routinely exhibited this trend. Statistical analysis of both 16 L and 225 L barrels revealed several significant main and interactive effects that impacted populations of B. bruxellensis in untreated barrel staves (Table 2). For the 16 L barrels, oak species (O) and yeast penetration depth in 2 mm increments (D) had the greatest influence on the size of the yeast populations (p 0.001), with higher populations observed at the inner surfaces. The location of staves within the barrel (L) and toasting level (T) also significantly affected population size (p 0.05 and p 0.01, respectively), with bottom and heavy-toasted staves having larger populations. Of the many potential interactive effects, only O D and D T were significant at p 0.001, Table 2 Calculated F-values for main and interactive (statistically significant values only) effects of five factors on culturable populations of Brettanomyces bruxellensis in untreated barrel staves. a Source of variation 16 L barrels 225 L barrels Main effects Oak species (O) 67.60*** n/a b Location of stave (L) 2.66* 2.30* Oak depth (D) 22.60*** 130*** Toasting level (T) 11.60** n/a Interactive effects O D 56.40*** n/a O T 4.40* n/a L D 2.34* 2.11* D T 7.78*** n/a a Significance denoted as p 0.05 (*), p 0.01 (**), or p 0.001 (***). b n/a: Not applicable to 225 L barrels, since specific yeast strain(s) were not known, and toasting levels were different between French and American oak barrels. whereas O T and L D were significant at p 0.05. Similar trends were observed for the 225 L barrels (Table 2), except that oak species and toasting level were not evaluated because each oak had a different toasting level (Table 1). As observed for the 16 L barrels, stave location (L) and yeast penetration depth (D) for 225 L barrels had significant impacts on culturability (p 0.05 and p 0.001, respectively), and an interaction between L D was observed (p 0.05). Interactions of more than two main effects were not statistically significant for any of the barrels. Yeast recovery from steamed staves. Heat penetration of 3 10 cm stave blocks depended primarily on depths, and not on barrel size (16 or 225 L), oak species, toasting level, or stave thickness (Figure 5). In general, innermost surfaces (0 mm) quickly exceeded 95 C within 2 min, while external surfaces farthest away from the steam (20 or 25 mm) never surpassed 65 C within 12 min. The maximum depth where B. bruxellensis was previously recovered ( 9 mm) required a minimum of 3 to 4 min to reach temperatures between 50 to 60 C, and eventually attained 75 C after 8 to 9 min. The duration of steaming affected recovery of B. bruxellensis, depending on the depth of yeast penetration into the staves. For American oak lateral/side staves (16 L barrels), B. bruxellensis strain I1a was recovered only from cross sections located 4 mm (Figure 6), in agreement with previous analyses of top- and bottom-located staves (Figure 2 and Table 1). After 6 min of steaming, the 0 to 4 mm cross section required 30 days of incubation in WRM to yield culturable populations. Steaming either light (Figure 6A) or heavy (Figure 6B) toasted American oak blocks for a longer period of time (9 min) resulted in undetectable populations, even after incubation for 60 days in WRM. In contrast, B. bruxellensis was recovered from 5 to 9 mm cross sections from 16 L barrel French oak lateral/side staves (Figure 7), similar to the results for French Figure 5 Heat penetration into staves from 16 L (A, B) or 225 L (C, D) barrels made from light or medium-heavy toasted American oak (A, C), and heavy or medium-light toasted French oak (B, D). Data represent means ± standard errors (n = 8)

406 Cartwright et al. oak bottom staves (Figure 2 and Table 1). Because the yeast penetrated farther into French oak, B. bruxellensis could be recovered from 5 mm, but not 4 mm after 9 min of steaming. However, to be detectable, viable cells present in 5 to 9 mm cross sections after 9 min of steaming required 30 days of incubation in WRM. Culturable cells were not recovered after steaming for 12 min, even after Figure 6 Culturabilities of Brettanomyces bruxellensis strain I1a recovered from light- (A) or heavy- (B) toasted 16 L American oak barrels. Data represent means ± standard errors (n = 8). WRM, wine recovery medium. Figure 7 Culturabilities of Brettanomyces bruxellensis strain I1a recovered from light- (A) or heavy- (B) toasted 16 L French oak barrels. Data represent means ± standard errors (n = 8). WRM, wine recovery medium. 60 days of incubation in WRM. Significant differences in recoveries were not observed between B. bruxellensis strains I1a or E1 (data not shown). Lateral/side staves obtained from the commercial 225 L barrels behaved similarly as those from 16 L barrels. Generally, the biggest difference was that the unidentified strain(s) of B. bruxellensis required additional incubation in WRM to yield detectable populations as the time of steaming increased (Figure 8). For instance, 7 or 15 days of incubation were needed to detect viable cells from the unheated French or American oak staves, respectively. After 6 min steaming, 30 or 60 days of incubation were required for B. bruxellensis identification between 5 to 9 mm and 0 to 4 mm depths, respectively. Lastly, no viable populations were detected between 0 to 4 mm after 9 min steaming, or 5 to 9 mm depths after 12 min steaming, in agreement with the results for the 16 L French oak staves (Figure 7). B. bruxellensis recovered in the WRM after steaming exhibited production of 4-EP and 4-EG. After 60 days of incubation, when populations had exceeded 10 6 cfu/ml, concentrations were >2000 µg/l (4-EP) and 300 µg/l (4-EG). Furthermore, concentrations of ethylphenols were significantly greater for strains I1a and E1 (p 0.05) than for the unidentified strain(s) originating from the 225 L barrels. Neither 4-EP or 4-EG was detected in WRM if B. bruxellensis populations were <30 cfu/ml. Discussion Yeast penetration into staves. B. bruxellensis was found to penetrate staves to different depths depending on several factors. For instance, populations in staves from 16 L or 225 L oak barrels approached 10 3 cfu/ mm 3, up to a depth of 2 mm, in agreement with reports by Yap et al. (2008) and Schmid et al. (2011) who had contaminated staves by soaking wood in yeast-inoculated broths. In the present study, culturable populations as high as 10 2 cfu/mm 3 were recovered from increased depths, higher than those reported by Barata et al. (2013). In fact, B. bruxellensis was occasionally located beyond perceptible red wine pigments because of unknown mechanisms, in contrast to previous reports (Malfeito-Ferreira et al. 2004, Barata et al. 2013). Similar to the findings of González-Arenzana et al. (2013), new French oak contained higher yeast populations at greater penetrations (up to 8 mm), possibly due to less blockage of xylem vessels by thyloses than in new American oak (Chatonnet and Dubourdieu 1998, del Alamo-Sanza and Nevares 2017). The commercial 225 L barrels usually had smaller populations at depths of 2 mm than at 4 mm, probably due to cleaning protocols (periodic hot-water washes and SO 2 gassings) used by the winery that reduced surface populations. Within a barrel, viable yeast cells were recovered up to 8 mm from staves located at the bottom or side of the barrels. However, populations in top staves were

Reduction of B. bruxellensis Populations 407 Figure 8 Culturabilities of an unknown strain(s) of Brettanomyces bruxellensis recovered from medium-heavy toasted American (A) or medium-light toasted French (B) 225 L oak barrels. Data represent means ± standard errors (n = 8). WRM, wine recovery medium. limited to depths 4 mm, a finding in contrast to those of Barata et al. (2013) and Leaute and Gilboulot (2013), who observed large populations near bungholes or top-inner surfaces. Although yeast growth is stimulated in semiaerobic environments (Ciani and Ferraro 1997, Aguilar-Uscanga et al. 2003), discrepancies could also be due to variations in barrel specifications or experimental protocols including sample sizes. For instance, sampling oak staves by using chisels (Barata et al. 2013) or electric planers (González-Arenzana et al. 2013) may have yielded small amounts of material with potentially undetectable populations or difficulty in accurately reproducing specific stave depths compared with sawing 4 mm-thick cross sections. However, concentrations of dissolved oxygen in wine are higher closer to the bungholes when barrels are not hermetically sealed (del Alamo-Sanza and Nevares 2017). Staves located at the top of barrels may also have lower SO 2 levels through exposure of dissolved oxygen, thereby providing favorable conditions for yeast growth (Agnolucci et al. 2017). Besides oak species and stave location, toasting levels also influenced the recovery of viable yeast. After incubation with infected wine, heavily toasted oak contained higher levels of B. bruxellensis, as evidenced by rapid growth upon inoculation into WRM. Intensive toasting creates transverse cracks in oak wood (Hale et al. 1999), which would provide more opportunities for the yeast to penetrate deeper into the wood. In addition, the yeast would also benefit from increased availability of cellobiose (Blondin et al. 1982), a sugar released during heating and bending of staves (Matthews et al. 2011, Crauwels et al. 2015). Presence of pseudohyphae. Besides noting normal yeast cell morphology (Edwards 2005), pseudohyphae were occasionally observed in stave cross sections, commonly near the maximum depths of yeast penetration. Similar cell structures have been observed under laboratory conditions (Aguilar-Uscanga et al. 2000, Kurtzman and Fell 2011), but this is the first report of pseudohyphae formation in oak barrels. Metabolically, these structures increase cell-surface-to-volume ratios to more efficiently take up nutrients and to physically extend to nutrient-rich areas (Bartles 1998), an important mechanism for other yeasts in nutrient-limited environments (Scherr and Weaver 1953, Brown and Hough 1965, Gimeno et al. 1992). Heat penetration and cell inactivation. During steaming, thermocouple data illustrated lag times regarding temperature increases of staves depending on depth. Generally, ~4 min was required for 9.5 mm depths to reach at least 55 C, a depth slightly greater than the maximum penetration achieved by the yeast. At a temperature of 55 C, B. bruxellensis exhibits a D-value of 12 to 80 sec (Couto et al. 2005, Fabrizio et al. 2015). Little variation in heat penetration was noted among staves from different oak species or toasting levels. As expected, eradication of B. bruxellensis from staves depended on the depth of wood penetration. If the extent of yeast penetration was 4 mm, increased steaming times of >3 min reduced culturable populations until a time of 9 min was reached when the yeast could no longer be recovered. If the yeast was present up to a depth of 9 mm, as in French oak (new or used barrels) or in American oak (used barrels only), culturable cells were not recovered from 0 to 4 mm cross sections, but remained viable in the deeper layers. In these situations, additional heating was necessary, with a total steaming time of 12 min being required to no longer recover the yeast up to 9 mm. A lack of growth in these heated cross sections was confirmed by the absence of 4-EP or 4-EG in WRM, even after 60 days of incubation (data not shown). In contrast, staves incubated in WRM with culturable yeast present frequently contained concentrations of 4-EP and 4-EG in excess of 2000 and 300 µg/l, respectively (data not shown). Several studies have presented conflicting observations regarding the influence of heat treatments on B. bruxellensis. On the one hand, Guzzon et al. (2011) reported reductions of 3.5 log in yeast populations after steaming barrels for 30 min. Because these authors did not disassemble the barrels, the effectiveness of the steam treatments was determined by monitoring microbial populations in water placed in the treated barrels. Similarly, Barata et al. (2013) noted that culturable cells still remained after steaming barrels for 10 min. On the other hand, Fabrizio et al. (2015) noted a 7-log reduction in populations of B. bruxellensis after a hot-water soak (60 C) treatment, in conflict with the findings of Schmid et al. (2011). A potential concern with prolonged steaming of barrels is degradation of oak-derived volatile compounds, which are extracted into wines during barrel aging and are essential to wine flavor and aroma profiles (Towey and Waterhouse 1996a, 1996b). Here, potential losses

408 Cartwright et al. of guaiacol, 4-methylguiaicol, furfural, 5-methylfurfural, cis- and trans-oak lactones, eugenol, and vanillin are of concern to winemakers using barrel-cleaning/sanitizing protocols (Schmid et al. 2011). However, temperatures greatly exceeding those achieved during steaming are necessary to considerably alter these compounds. Whereas temperatures 120 C must be maintained for 1 to 6 hrs for measurable differences, degradation of many of these desired compounds requires >185 C (Sarni et al. 1990). Since application of a 12 min steaming regimen resulted in maximum temperatures of 75 to 100 C reached at critical depths for yeast penetration, significant changes in oak-derived volatile compounds would not be expected. Conclusion This study quantified populations of B. bruxellensis at different depths and present in a wide range of infected barrels and evaluated the efficacy of steam treatments for removal of this yeast. Populations tended to be higher and at greater depths for staves having longer contact time with wine, that is, bottom and lateral/side barrel staves and those from older barrels. Steaming was found to eradicate the yeast from staves if present at depths of 0 to 4 mm (9 min) or 5 to 9 mm (12 min), as evidenced by a lack of culturable cells even after incubation of stave cross sections in WRM for 60 days. Future research using entire barrels instead of stave center points should be conducted to confirm the effectiveness of these steaming times for removing B. bruxellensis from highly infected barrels. Literature Cited Agnolucci M, Tirelli A, Cocolin L and Toffanin A. 2017. Brettanomyces bruxellensis yeasts: Impact on wine and winemaking. World J Microbiol Biotechnol 33:180. Aguilar-Uscanga MG, Delia ML and Strehaiano P. 2000. Nutritional requirements of Brettanomyces bruxellensis: Growth and physiology in batch and chemostat cultures. Can J Microbiol 46:1046-1050. Aguilar-Uscanga MG, Delia ML and Strehaiano P. 2003. Brettanomyces bruxellensis: Effect of oxygen on growth and acetic acid production. Appl Microbiol Biotechnol 61:157-162. Barata A, Laureano P, D Antuono I, Martorell P, Stender H, Malfeito- Ferreira M, Querol A and Loureiro V. 2013. Enumeration and identification of 4-ethylphenol producing yeasts recovered from the wood of wine ageing barriques after different sanitation treatments. J Food Res 2:140-149. Bartles JR. 1998. Advances in Molecular and Cell Biology. JR Bartles (ed.). Elsevier Science, Greenwich, CT. Blondin B, Ratomahenina R, Arnaud A and Galzy P. 1982. A study of cellobiose fermentation by a Dekkera strain. Biotechnol Bioeng 24:2031-2037. Boulton RB, Singleton VL, Bisson LF and Kunkee RE. 1996. Principles and Practices of Winemaking. Chapman and Hall, New York, NY. Brown CM and Hough JS. 1965. Elongation of yeast cells in continuous culture. Nature 206:676-678. Buechsenstein JW and Ough CS. 1978. SO 2 determination by aerationoxidation: A comparison with Ripper. Am J Enol Vitic 29:161-164. Chatonnet P and Dubourdieu D. 1998. Comparative study of the characteristics of American white oak (Quercus alba) and European oak (Quercus petraea and Q. robur) for production of barrels used in barrel aging of wines. Am J Enol Vitic 49:79-85. Chatonnet P, Dubourdieu D, Boidron JN and Pons M. 1992. The origin of ethylphenols in wines. J Sci Food Agric 60:165-178. Chatonnet P, Viala C and Dubourdieu D. 1997. Influence of polyphenolic components of red wines on the microbial synthesis of volatile phenols. Am J Enol Vitic 48:443-448. Ciani M and Ferraro L. 1997. Role of oxygen on acetic acid production by Brettanomyces/Dekkera in winemaking. J Sci Food Agric 75:489-495. Couto JA, Neves F, Campos F and Hogg T. 2005. Thermal inactivation of the wine spoilage yeasts Dekkera/Brettanomyces. Int J Food Microbiol 104:337-344. Crauwels S et al. 2015. Comparative phenomics and targeted use of genomics reveals variation in carbon and nitrogen assimilation among different Brettanomyces bruxellensis strains. Appl Microbiol Biotechnol 99:9123-9134. del Alamo-Sanza M and Nevares I. 2017. Oak wine barrel as an active vessel: A critical review of past and current knowledge. Crit Rev Food Sci Nutr, doi: 10.1080/10408398.2017.1330250. Edwards CG. 2005. Illustrated Guide to Microbes and Sediments in Wine, Beer, and Juice. WineBugs LLC, Pullman, WA. Edwards CG and Watson BA. 2013. Basic Microbiological and Chemical Analyses for Wine. Washington State University Cooperative Extension EM047, Pullman, WA. Fabrizio V, Vigentini I, Parisi N, Picozzi C, Compagno C and Foschino R. 2015. Heat inactivation of wine spoilage yeast Dekkera bruxellensis by hot water treatment. Lett Appl Microbiol 61:186-191. Flegler SL, Heckman JW and Klomparens KL. 1993. Scanning and Transmission Electron Microscopy: An Introduction. WH Freeman, New York, NY. Fugelsang KC and Edwards CG. 2007. Wine Microbiology: Practical Applications and Procedures. Springer, New York, NY. Gimeno CJ, Ljungdahl PO, Styles CA and Fink GR. 1992. Unipolar cell divisions in the yeast S. cerevisiae lead to filamentous growth: Regulation by starvation and RAS. Cell 68:1077-1090. González-Arenzana L, Santamaría P, López R, Garijo P, Gutiérrez AR, Garde-Cerdán T and López-Alfaro I. 2013. Microwave technology as a new tool to improve microbiological control of oak barrels: A preliminary study. Food Control 30:536-539. Guzzon R, Widmann G, Malacarne M, Nardin T, Nicolini G and Larcher R. 2011. Survey of the yeast population inside wine barrels and the effects of certain techniques in preventing microbiological spoilage. Eur Food Res Technol 233:285-291. Guzzon R, Nardin T, Micheletti O, Nicolini G and Larcher R. 2013. Antimicrobial activity of ozone. Effectiveness against the main wine spoilage microorganisms and evaluation of impact on simple phenols in wine. Aust J Grape Wine Res 19:180-188. Hale MD, McCafferty K, Larmie E, Newton J and Swan JS. 1999. The influence of oak seasoning and toasting parameters on the composition and quality of wine. Am J Enol Vitic 50:495-502. Jensen SL, Umiker NL, Arneborg N and Edwards CG. 2009. Identification and characterization of Dekkera bruxellensis, Candida pararugosa, and Pichia guilliermondii isolated from commercial red wines. Food Microbiol 26:915-921. Kurtzman C, Fell JW and Boekhout T. 2011. The Yeasts: A Taxonomic Study. Elsevier, Amsterdam. Leaute B and Giboulot M. 2013. Study of the distribution of the contamination of the inner surface of a new wine barrel by Brettanomyces bruxellensis. Revue des Œnologues 40:29-30.

Reduction of B. bruxellensis Populations 409 Licker J, Acree T and Henick-Kling T. 1998. What is Brett (Brettanomyces) Flavor? A Preliminary Investigation. In Chemistry of Wine Flavor. Waterhouse AL and Ebeler SE (eds.), pp. 96-115. American Chemical Society, Washington, DC. Malfeito-Ferreira M, Laureano P, Barata A, D Antuono I, Stender H and Loureiro V. 2004. Effect of different barrique sanitation procedures on yeasts isolated from the inner layers of wood. Am J Enol Vitic 55:304A. Matthews JF, Bergenstråhle M, Beckham GT, Himmel ME, Nimlos MR, Brady JW and Crowley MF. 2011. High-temperature behavior of cellulose I. J Phys Chem B 115:2155-2166. Palacios A, Borinaga I and Carrillo D. 2012. Comparative study on wine barrel disinfection systems and their use as alternatives to sulfur treatment. Enólogos 14:46-54. Renouf V and Lonvaud-Funel A. 2007. Development of an enrichment medium to detect Dekkera/Brettanomyces bruxellensis, a spoilage wine yeast, on the surface of grape berries. Microbiol Res 162:154-167. Romano A, Perello MC, Lonvaud-Funel A, Sicard G and de Revel G. 2009. Sensory and analytical re-evaluation of Brett character. Food Chem 114:15-19. Sarni F, Moutounet M, Puech J-L and Rabier P. 1990. Effect of heat treatment of oak wood extractable compounds. Holzforschung 44:461-466. Scherr GH and Weaver R. 1953. The dimorphism phenomenon in yeasts. Bacteriol Rev 17:51-92. Schmid F, Grbin P, Yap A and Jiranek V. 2011. Relative efficacy of high-pressure hot water and high-power ultrasonics for wine oak barrel sanitization. Am J Enol Vitic 62:519-526. Towey JP and Waterhouse AL 1996a. Barrel-to-barrel variation of volatile oak extractives in barrel-fermented Chardonnay. Am J Enol Vitic 47:17-20. Towey JP and Waterhouse AL 1996b. The extraction of volatile compounds from French and American oak barrels in Chardonnay during three successive vintages. Am J Enol Vitic 47:163-172. Yap A, Schmid F, Jiranek V, Grbin P and Bates D. 2008. Inactivation of Brettanomyces/Dekkera in wine barrels by high power ultrasound. Aust NZ Wine Ind J 23:32-40.