Scholarly article on topic 'Bioflavoring by non-conventional yeasts in sequential beer fermentations'

Bioflavoring by non-conventional yeasts in sequential beer fermentations Academic research paper on "Chemical sciences"

Share paper
Academic journal
Food Microbiology
OECD Field of science

Abstract of research paper on Chemical sciences, author of scientific article — Sylvester Holt, Vaskar Mukherjee, Bart Lievens, Kevin J. Verstrepen, Johan M. Thevelein

Abstract Non-conventional yeast species have great capacity for producing diverse flavor profiles in production of alcoholic beverages, but their potential for beer brewing, in particular in consecutive fermentations with Saccharomyces cerevisiae, has only poorly been explored. We have screened 17 non-conventional yeast species for production of an appealing profile of flavor esters and phenolics in the first phase of alcoholic fermentation, followed by inoculation with S. cerevisiae to complete the fermentation. For measurement of phenolic compounds and their precursors we developed an improved and highly sensitive methodology. The results show that non-conventional yeast species possess promising potential for enhancement of desirable flavors in beer production. Notable examples are increasing isoamyl acetate (fruity, banana flavor) by application of P. kluyverii, augmenting ethyl phenolic compounds (spicy notes) with Brettanomyces species and enhancing 4-vinyl guaiacol (clove-like aroma) with T. delbrueckii. All Pichia strains also produced high levels of ethyl acetate (solvent-like flavor). This might be selectively counteracted by selection of an appropriate S. cerevisiae strain for the second fermentation phase, which lowers total ester profile. Hence, optimization of the process conditions and/or proper strain selection in sequentially inoculated fermentations are required to unlock the full potential for aroma improvement by the non-conventional yeast species.

Academic research paper on topic "Bioflavoring by non-conventional yeasts in sequential beer fermentations"

Accepted Manuscript

Bioflavoring by non-conventional yeasts in sequential beer fermentations

Sylvester Holt, Vaskar Mukherjee, Bart Lievens, Kevin J. Verstrepen, Johan M. Thevelein

PII: S0740-0020(17)30376-3

DOI: 10.1016/

Reference: YFMIC 2901

To appear in: Food Microbiology

Received Date: 21 April 2017 Revised Date: 8 October 2017 Accepted Date: 15 November 2017

Please cite this article as: Holt, S., Mukherjee, V., Lievens, B., Verstrepen, K.J., Thevelein, J.M., Bioflavoring by non-conventional yeasts in sequential beer fermentations, Food Microbiology (2017), doi: 10.1016/

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Bioflavoring by non-conventional yeasts in sequential beer fermentations

Sylvester Holt1, Vaskar Mukherjee1,2 Bart Lievens2, Kevin J. Verstrepen3 and Johan M. Thevelein1*

1 Laboratory of Molecular Cell Biology, Institute of Botany and Microbiology, VIB Center for Microbiology, KU Leuven, Kasteelpark Arenberg 31, B-3001 Leuven, Belgium

2 Laboratory for Process Microbial Ecology and Bioinspirational Management, Department of Microbial and Molecular Systems (M2S), Cluster for Bioengineering Technology (CBeT), KU Leuven, Campus De Nayer, Fortsesteenweg 30A, 2860 Sint-Katelijne-Waver, Belgium

3 Laboratory for Genetics and Genomics & VIB Laboratory for Systems Biology, Centre of Microbial and Plant Genetics (CMPG), Department of Microbial and Molecular Systems (M2S), KU Leuven, Gaston Geenslaan 1, B-3001 Leuven, Belgium

# Current address: University of Gothenburg, Department of Chemistry and Molecular Biology, Gothenburg, Sweden

Corresponding author: Highlights

• Non-Saccharomyces yeasts have been screened for their bioflavoring potential in beer

• Pichia kluyverii is unique for its strong production of fruity esters

• Ethyl and vinyl phenols and their precursors can be determined simultaneously with HPLC

• Brettanomyces sequential inoculation can be positive or negative for flavor profile

• Aroma contribution is not additive in sequential beer fermentations Abstract

Non-conventional yeast species have great capacity for producing diverse flavor profiles in production of alcoholic beverages, but their potential for beer brewing, in particular in consecutive fermentations with Saccharomyces cerevisiae, has only poorly been explored. We have screened 17 non-conventional yeast species for production of an appealing profile of flavor esters and phenolics in the first phase of alcoholic fermentation, followed by inoculation with S. cerevisiae to complete the fermentation. For measurement of phenolic compounds and their precursors we developed an improved and highly sensitive methodology. The results show that non-conventional yeast species possess promising potential for enhancement of desirable flavors in beer production. Notable examples are increasing isoamyl acetate (fruity, banana flavor) by application of P. kluyverii, augmenting ethyl phenolic compounds (spicy notes) with Brettanomyces species and enhancing 4-vinyl guaiacol (clove-like aroma) with T. delbrueckii. All Pichia strains also produced high levels of ethyl acetate (solvent-like flavor). This might be selectively counteracted by selection of an

38 appropriate S. cerevisiae strain for the second fermentation phase, which lowers total ester profile.

39 Hence, optimization of the process conditions and/or proper strain selection in sequentially

40 inoculated fermentations are required to unlock the full potential for aroma improvement by the

41 non-conventional yeast species.

1. Introduction

In spite of their great potential for flavor production, it is surprising that only few non-conventional yeast species have been evaluated in beer fermentations, which is reflected in recent reviews (Basso et al., 2016; Michel, Meier-Dörnberg, et al., 2016). Most examples are from beers with an outspoken phenolic profile, influenced by 4-vinyl guaiacol (4VG) with "clove-like" descriptor, 4-vinyl phenol (4VP) with "phenolic" descriptor, 4-ethyl guaiacol (4EG) with "spicy", "clove-like" and "vanilla" descriptors, and 4-ethyl phenol (4EP) with "spicy", "smokey" and "horsy" descriptors (Callemien and Collin, 2009; Mukai et al., 2010). The 4VG derived clove-like aroma is essential for the flavor of some wheat and blond beer styles, which enjoy considerable commercial success. Saccharomyces cerevisiae brewing yeasts that are classified as POF+ (phenolic off-flavor positive), contain active hydroxycinnamate decarboxylase enzymes, encoded by FDC1 and PAD1, that can transform trans-ferulic acid into 4VG and p-coumaric acid into 4VP (Mukai et al., 2010). Conventional POF+ S. cerevisiae brewing yeasts are used in the fermentation of beers in which a clove-like aroma is desired, except for some Hefeweizen beers which are presumably fermented with Torulaspora delbrueckii (Tataridis et al., 2013).

The most commonly used non-conventional brewing yeast is Brettanomyces, which is inoculated in the fermentation of some Trappist and sour beers. It is also found in spontaneous fermentations of Gueuze and Lambic beers, in which it contributes to the typical sourness and unique complex flavor profile (Bokulich and Bamforth, 2013; Schifferdecker et al., 2014; Steensels et al., 2015). Brettanomyces yeasts contain vinyl phenol reductases that catalyze the transformation of vinyl phenols and vinyl guaiacols into the ethyl forms 4EP and 4EG (Granato et al., 2014). In contrast to wine, where high levels of 4EP are mostly associated with "medicinal" and "band aid" off-flavors, the spicy "Brett" character in beer is usually dominated by higher levels of 4EG and is considered positive for some beer styles, especially spontaneously fermented Gueuze and Lambic beers. The concentrations of 4EG and 4EP in the beer are determined by the hydroxycinnamic acid precursor levels in the fermentation media and Brettanomyces yeast's ability to transform the compound (Steensels et al., 2015). Some Brettanomyces yeasts also have high esterase activity, which is unwanted because it destroys the acetate esters which are essential for the fruity flavors of beer (Steensels et al., 2015; Verstrepen, Derdelinckx, et al., 2003). For application of candidate Brettanomyces strains in beer brewing, it is therefore important to evaluate the levels produced of both phenolics and acetate esters.

Non-conventional yeasts have already shown a great potential for enhancement of flavor-active esters in wine making, both in mixed or sequentially inoculated with S. cerevisiae [table 1; for

76 reviews see (Ciani and Comitini, 2010; Ciani et al., 2010; Jolly et al., 2014)]. However, depending on

77 the composition of the grape must, the application of non-conventional yeast in the wine

78 fermentation can modulate the flavor profile in an unpredictable manner. For example, application

79 of Candida stellata, Torulaspora delbrueckii or Pichia anomala has been shown to either increase or

80 decrease the acetate ester and higher alcohol levels (table 1). Here, we have assessed their potential

81 for bioflavoring of beer by performing sequential inoculations with non-Saccharomyces and the

82 Saccharomyces MauriBrew Ale 514 yeast. This approach was shown to be promising also in beer

83 production with sequential inoculation of Pichia kluyverii and S. cerevisiae using commercially

84 available FrootZen P. kluyverii yeast, which is frequently used for enhancing tropical fruit aromas in

85 Sauvignon blanc wines (Saerens and Swiegers, 2013). Previously, beer fermentations with a single

86 and sequential inoculum of T. delbrueckii and S. cerevisiae have been evaluated under standard

87 gravity (12°P) conditions, but the level of the important ester isoamyl acetate was negligible (< 0.01

88 mg/L) (Canonico et al., 2016; Michel, Kopecka, et al., 2016). Saccharomycodes ludwigii and

89 Zygosaccharomyces rouxii have also been evaluated for production of low alcohol beer, but they did

90 not produce a commercially-acceptable flavor profile in single strain fermentations (De Francesco et

91 al., 2015).

92 In this study, we evaluated 17 non-conventional yeast species for production of flavor esters and

93 phenolics, in order to possibly serve as starter cultures in sequential beer fermentations with S.

94 cerevisiae. Initial observations showed that P. kluyverii was able to produce very high amounts of

95 isoamyl acetate and ethyl acetate, while P. anomala and P. kudriavzevii produced extremely high

96 levels of ethyl acetate, but low levels of isoamyl acetate. This shows that P. kluyverii may have

97 unique potential for enhancing fruitiness in beer. In sequential fermentations with P. kluyverii (2

98 days) followed by S. cerevisiae to complete the high-gravity beer fermentation, we indeed observed

99 a higher isoamyl acetate/isoamyl alcohol ratio although the total level of isoamyl acetate was much

100 lower than expected. Further optimization of the process conditions and/or selection of appropriate

101 S. cerevisiae strains for the second inoculation can allow to fully profit from the unusual capacity of

102 P. kluyverii as starter culture to produce a highly attractive flavor profile.

2. Materials and methods

2.1 Chemicals. All chemicals were from Sigma Aldrich. Acetonitrile and formic acid was of UV-grade, and the ethanol used for gas chromatography was GC-grade.

2.2 Preparation of malt extract medium.

Malt extract medium consisting of 220 g/L of malt extract (Brewferm spraymalt 8 EBC, Brouwland, Belgium) supplemented with 0.5 mg/L ZnSO4 [as commonly applied in high gravity brewing (Gibson, 2011)], was autoclaved at 110°C for 15 min. After autoclaving, the malt extract medium was cold settled overnight and filtered through a nylon filter (GE Healthcare) to remove insoluble precipitates. The final gravity of the malt extract medium was 19.2-20.5 °P. The malt extract medium contained 98.9 g/L maltose, 17.3 g/L glucose, 6.2 mg/L ferulic acid and 2.0 mg/L coumaric acid, as well as 0-0.3 mg/L 4VG, likely formed by degradation of ferulic acid during autoclaving.

2.3 Selection of strains from non-conventional yeast species

A collection of yeasts mainly isolated from high sugar niches such as flower nectar, cacao fermentations, grape must, sugar juice, and syrup, was pre-screened for ability to grow in high sugar concentrations, in view of their possible application in high-gravity beer fermentations. The pre-screening results and methodology have been described previously (Mukherjee et al., 2017). From the 160 strains able to grow in 40% glucose and in 40% fructose, 22 strains belonging to 17 species were selected based on previous usage in diverse alcoholic fermentations (Table 1). The selected strains included seven strains from the Pichia genus known for high ester production in wine fermentations (Clemente-Jimenez et al., 2005; Domizio et al., 2011; Kurita, 2008; Moreira et al., 2008; Moreira et al., 2005), including the commercialized Pichia kluyverii strain FrootZen (Saerens and Swiegers, 2013), and four from the Brettanomyces genus known for its vinyl reductase activity, including the well characterized and fully sequenced wine spoilage strain B. bruxellensis AWRI1499 (Curtin et al., 2012). Apart from Pichia and Brettanomyces strains, several strains from other genera previously isolated from wine and spontaneous beer fermentations, or evaluated in co- and sequential wine fermentations, were included (table 2). Strains used in this study were either obtained from the CBS-KNAW or NRRL fungal collections, Chr. Hansen A/S, The Australian Wine Research Institute (AWRI), or investigated in-house by PCR amplification and sequencing of the D1/D2 domain of large ribosomal subunit. Briefly, genomic DNA was extracted after treatment of the cells with zymolyase (purchased from Seikagaku Biobusiness, Tokyo, Japan). A single colony was suspended into 50 ^L of lysis solution (3 mg zymolyase/mL ultrapure water). The solution was heated at 37°C for 60 min, followed by 10 min at 98°C. The variable D1/D2 domain of the large-subunit (26S) rDNA gene was amplified using primers NL-1 and NL-4 (Kurtzman and Robnett, 1998).

The PCR product was purified and single read sequencing was performed by the VIB Genetic Service Facility (Antwerp, Belgium) using Applied Biosystems 3730XL DNA Analyzer (White et al., 1990). Identification was performed by a BLASTN (version 2.2.29) search against the non-redundant nucleotide collection in GenBank.

2.4 Cell cultures and inoculation by dry weight.

Strains were recovered from -80°C glycerol stocks and plated on YPD agar (10 g/L yeast extract, 20 g/L bacteriological peptone, 15 g/L agar, and 20 g/L D-glucose) and allowed to grow for 3-5 days at 25°C. Next, strains were pre-cultured overnight in 3 mL YPD (containing 50 mg/L ampicillin and 25 mg/L kanamycin) and then 100 mL of overnight pre-culture was re-inoculated into 5 mL of YPD and cells were allowed to grow for 3 days at 25°C. After the growth, cell cultures were maintained on ice while the dry cell weight was measured. An aliquot of 0.5 mL was filtered through a Whatman glass microfiber filter, and dried at 65°C for a minimum of 30 min, and the dry cell weight per volume was calculated as the net weight gain of the filter. The cells (maintained on ice) were washed with malt extract media, and inoculated at a final cell density of 100 mg dry weight/L. In sequential fermentations, a second inoculation was performed after 48 h with 100 mg dry weight/L of S. cerevisiae brewing yeast cell counting per mg dry weight was done with the Scepter cell counter (Millipore). One milligram of dried S. cerevisiae MauriBrew Ale 514 yeast cells contained approximately 1.0x108 cells, corresponding to an inoculation rate of 1.0x107 cells/mL.

2.5 High-gravity beer fermentations.

Small-scale high-gravity fermentations were performed at 20°C in a total volume of 60 mL in 100 mL tall tubes [validated against 2L fermentation tubes of the European Brewery Convention (EBC)] equipped with water locks and under continuous stirring at 130 rpm. Samples (10 mL) were taken after 48 and 72 h, yeast cells and residuals were removed by centrifugation, and the samples were kept at 4°C until analysis. Sequential fermentations, inoculated after 48h with S. cerevisiae brewing yeast, were carried out in a total volume of 250 mL in 370 mL (EBC validated) tall tubes. After 72 h of stirring at 130 rpm all fermentations were left unstirred for the remaining part of the fermentation.

2.6 HPLC-UV/Fluorescence detection of phenolic compounds.

A reverse phase Macherey-Nagel NUCLEOSIL 100-10 C18 column with 4 mm internal diameter was used for separation on a Shimatzu LC-20AT HPLC equipped with a SIL-HT autosampler. We have first performed a two-step linear gradient of 0.1% formic acid and acetonitrile as eluent as described for separation of polyphenolic compounds (Aznar et al., 2011). This resulted in good separation of the ferulic acid and p-coumaric acid, but poor separation of 4VP and 4VG and also suffered from a long

elution time for the 4EG. We therefore tested isocratic elution with higher levels of acetonitrile from 23-50% to enhance the separation and reduce the run time. The final method used an isocratic eluent containing 0.1% formic acid and 25% acetonitrile. This enabled good separation of trans-ferulic acid, p-coumaric acid, 4-vinyl phenol (4VP), 4-vinyl guaiacol (4VG), 4-ethyl phenol (4EP), and 4-ethyl guaiacol (4EG) with elution performed for 35 min. Subsequent conditioning of the column was done with 95% acetonitrile and 5% acetonitrile for 5 min each. The HPLC was coupled in parallel to an SPD-20A UV detector and an RF-10AXL fluorescence detector. The fluorescence detector gave higher signals for all phenolic compounds than the UV detector, with the strongest signal for 4VG only measured by UV detection, being out of range with fluorescence detection. On the other hand, the ferulic acid and p-coumaric acid showed poor emission between 305 to 360 nm with excitation wavelengths between 210 to 260 nm, but good UV absorbance at 280 nm. P-coumaric acid, ferulic acid and 4VG were therefore measured by the UV detector at 280 nm, while the fluorescence detector was used for 4EG and 4EP with excitation and emission wavelengths of 210 and 340nm, respectively. Standards were prepared in 100% methanol (kept at -20°C) and spiked (1:20) into malt extract medium to a final concentration of 0.3-10 mg/L. The injection volume was 100 ml. All compounds were well separated with retention times as follows: p-Coumaric acid at 8.7 min; Ferulic acid at 9.5 min; 4VP at 21.7 min; 4VG at 24.0 min; 4EP at 25.5 min and 4EG at 28.4 min. The detection of the standards spiked in malt extract medium was linear with the concentration for all compounds (R2> 0.99). For 4VP, however, the observed values in the final fermentations were unexpectedly higher than the range of the standards, which was possibly due to interference with an unknown compound eluting at the same time from the column. Therefore, we did not include the values for 4VP in the final data.

2.7 GC-FID of esters and alcohols. Fermentation samples with a volume of 2 mL were analyzed in 15 mL vials on a Thermoscience Trace GC, equipped with a ResTek Stabilowax polyethylene glycol column with 0.25 mm diameter. Split-injection (1:25), with a resting flow of 2 mL/min He, was performed after 10 min equilibration at 60°C with a Thermoscience TriPlus RSH autosampler. The oven was kept at 40°C for 2 min, heated to 240°C with a ramping of 15°C/min, and kept at 240°C for 2 min. The detection was carried out with a flame ionization detector (FID), using 20 mL/min of N2, 350 mL/min high grade compressed air, and 30 mL/min H2 provided from a VWR H 2 generator. Concentrated standards were kept at -20°C in GC grade absolute ethanol and brought to the final concentration in 5% ethanol in volumetric dilutions. Fermentation samples were kept at 4°C prior to analysis.

2.8 HPLC-RID of maltose and ethanol

Maltose and ethanol concentrations in sequential fermentations were measured on a Waters HPLC using an Ion Exclusion Column, 7 ^m, 7.8 x 300 mm. Column temperature was 75°C, 5 mM H2SO4 was used as eluent with a flow rate of 1 mL/min and a Waters refractive index detector (RID) was used for detection. We have estimated the alcohol by volume content [ABV (v/v)] by the approximated formula ABV = ABW /(0.789+0.211*ABW). It uses the density of alcohol 0.789 g/l at 20°C in the fraction of alcohol/water and the alcohol by weight [ABW (w/v)] determined by the HPLC measurements.

3. Results

3.1 A novel method for simultaneous detection of ferulic and coumaric acid, 4VG, 4EP and 4EG

Previous detection methods for beer phenolic compounds (Vanbeneden et al., 2006; Zhu and Cui, 2013) suffered from very long elution times (> 90 min) of 4EG, incomplete separation of 4VG and 4EP, and from the inherent instability in sensitivity of electrochemical detectors. We therefore developed a novel method, which allows simultaneous detection of all four phenolic compounds with a sensitive fluorescence detector and the acid precursors with a UV detector. The method uses acetonitrile and formic acid as eluent (Aznar et al., 2011; Shinohara et al., 2000). The best separation of the phenolic flavor compounds and their precursor acids was obtained with a 35 min isocratic flow at a concentration of 25% acetonitrile and 0.1% formic acid (Fig. 1A and 1B). The sensitivity was high, with the lowest concentration of 0.3 mg/L tested still being in the linear range of the standards (Fig. 1C). The fluorescence detection of 4VG was so sensitive that the signal was already saturated at 2 mg/L, but also with UV absorbance the sensitivity was high (Fig. 1B, 1C, and other data not shown). The linearity of the measurement as a function of the concentration was confirmed (R2 > 0.99) by spiking standards into brewing malt extract medium (Fig. 1C). For 4VP, however, the observed values in the fermentations were unexpectedly high, possibly due to interference with an unknown compound eluting at the same position. Therefore, we eliminated the measurements of 4VP from the final data.

3.2 Evaluation of flavor compound production in fermentations

To assess flavor compound production, we carried out small-scale high-gravity beer fermentations (see materials and methods 2.5) with the 23 selected non-Saccharomyces strains, the standard brewing yeast Mauribrew Ale 514 and the Zymaflore VL3 wine yeast (used for production of fruity white wines). The fermentations were sampled after 48 and 72 h and the flavor metabolite profiles of phenolic compounds, esters and higher alcohols were determined and analyzed by phenotypic clustering (Fig. 2A and 2B). As expected, the S. cerevisiae Ale 514 brewing yeast consumed the highest amount of maltose (60 g/L) and produced 6.0% ABV after three days of fermentation. On the

other hand, Zygotorulaspora florentina (Zfl) and Lachancea thermotolerans (Lth) were the most efficient non-conventional yeast strains for ethanol production in malt extract media with a production of 4.0 and 2.6% ABV, respectively. The Zfl strain even produced a slightly higher level of ethanol than the commercial S. cerevisiae Zymoflore VL3 wine yeast (3.6% ABV). It was the only non-conventional yeast species that consumed a significant amount of maltose (approximately 30 g/L) after three days of fermentation. The average ethanol production in the case of non-Saccharomyces yeasts was 1.4% ABV, with low consumption of maltose. Interestingly, the Brettanomyces (Bbrl, Bbr2 and Bna) and the P. kluyverii (Pkl2 and Pkl3) strains, which showed to be phenotypic outliers for production of flavor compounds, produced very low levels of ethanol (0.2-0.4% ABV), with residual glucose in the fermentations, even after three days.

3.2.1 Acetate esters

Production of ethyl acetate, generally considered as an off-flavor due to its "solvent-like" aroma, remained for most species and strains clearly below the sensory threshold of 21-30 mg/L in beer (Meilgaard, 1982). Exceptions were T. delbrueckii and B. naardenensis (71.5 and 24.3 mg/L after 72 h, respectively) and especially the Pichia strains, which produced very high levels in the range of 41171 mg/L and 88-924 mg/L after 48 and 72 h, respectively (Fig. 2A and B). P. anomala and P. kudriavzevii strains differed from the other species in producing high levels only of ethyl acetate (110-249 mg/L after 72h) without clear increases in the other acetate esters. The Pichia species were actually standing out in producing high levels of all acetate esters (Fig. 2 and Supplementary Fig. 1). P. kluyverii was unique among the Pichia species in producing also high levels of the beneficial flavor compound isoamyl acetate ("banana flavor") (9.7-11.6 mg/L isoamyl acetate, after 48h). The commercial Pkl3 P. kluyverii (FrootZen) outperformed all other strains in this respect with 42.6 mg/L isoamyl acetate produced after 72h (Fig. 2B). The isoamyl and ethyl acetate (924 mg/L) levels produced by the commercial Pkl3 P. kluyverii (FrootZen) strain are 19 and 31 times higher than those of the S. cerevisiae brewing yeast (2.2 mg/L isoamyl acetate and 30.1 mg/L ethyl acetate). They are in the same range as those obtained with the commercial CMBS33 S. pastorianus lager strain with constitutive overexpression of ATF1 (49 mg/L isoamyl acetate and 273 mg/L ethyl acetate) (Verstrepen, Van Laere, et al., 2003).

3.2.2 Fusel alcohols

The only yeasts that were able to form isoamyl alcohol above its sensory threshold of 70 mg/L (Meilgaard, 1975) were Z. florentina (Zfl) and P. kudriavzevii Pku1. All the other strains displayed a range between 1.5 and 59.6 mg/L after three days (Supplementary Data 1 and Fig. 2). The three P. kluyverii yeasts showed low to intermediate levels (11.6-33.1 mg/L), indicating

that the high production of isoamyl acetate in P. kluyverii is not directly linked to the level of its precursor alcohol, and that the naturally occurring AATase(s) in P. kluyverii are highly active and/or expressed at high levels compared to other yeast species.

3.2.3 Ethyl esters

None of the non-conventional yeast strains were able to produce ethyl esters above the minimum threshold of sensory perception (Fig. 2 and Supplementary Data 1). Only the S. cerevisiae ale yeast was able to produce ethyl hexanoate (fruity "apple" flavor) at a level (0.60 mg/L), significantly exceeding the sensory threshold of 0.21 - 0.23 mg/L (Meilgaard, 1975). The production of ethyl esters was correlated to ethanol production (Supplementary Fig. 1), suggesting that the ability to form ethyl esters is somehow linked to alcoholic fermentation capacity.

3.2.4 4VG

For all non-conventional yeast strains, production of the phenolic compounds was correlated to te capacity to transform hydroxycinnamic acids present in the malt extract medium (Fig. 2A and 2B). The Mauribrew Ale 514 strain was POF-, while T. delbrueckii and the wine yeast VL3 were POF+, transforming almost all ferulic acid into 4VG in 72 h. C. stellata, Z. florentina, and the two P. anomala strains were also POF+, but with intermediate levels of 4VG (Cst, 1.77 mg/L; Zfl, 1.78 mg/L; Pan1, 1.89 mg/L; Pan2, 2.89 mg/L). The P. kluyverii and P. kudriavzevii strains were POF- and showed no production of 4VG even after 72 h of fermentation.

3.2.5 4EP and 4EG

Significant transformation of coumaric and ferulic acid into 4EP and 4EG were observed in fermentations with Brettanomyces strains (0.12-0.22 mg/L 4EP and 0.79-1.22 mg/L 4EG after 48 h, and 0.38-0.54 mg/L 4EP and 1.50-2.75 mg/L 4EG after 72 h), except for the B. naardenensis Bna strain which seemed to be POF-. The fermentations with Brettanomyces strains (Bbr1, Bbr2 and Ban) did not contain any detectable levels of 4VG after 48 h of fermentation, indicating that all 4VG was transformed into 4EG. The levels of ferulic and coumaric acids decreased further after 72 h, with an increase in 4VG for Ban (0.39 mg/L) and Bbr1 (0.00 and 1.65 mg/L in the fermentation replicates, respectively), while the level of 4EG strongly increased up to 2.75 mg/L. The concentrations of 4EP and 4EG in fermentations with Bbr1, Bbr2 and Ban, are all above the sensory thresholds of 0.10 mg/L for 4EP and 0.13 mg/L for 4EG, respectively [Supplementary Data 1; (Callemien and Collin, 2009)]. The appearance of 4VG after 72 h suggests a two-step process, in which the bottlenecks are the decarboxylation of ferulic acid into 4VG and the subsequent reduction of 4VG into 4EG. The Ban and Bbr1 strains showed efficient conversion of ferulic acid into 4EG compared to the B. bruxellensis

Bbr2 wine spoilage yeast, making these good candidate strains for beers in which spicy phenolic notes are desirable, while the Bna strain appears as a candidate Brettanomyces strain for beers in which phenolic notes are not desirable.

3.3 Sequentially inoculated beer fermentations

Next, we performed trial beer fermentations first inoculated with a non-conventional yeast species, and after 48 h of flavor development, sequentially inoculated with the Ale 514 brewing yeast. The strains evaluated in sequential fermentations were selected for high production of specific aromatic compounds and were phenotypic outliers in comparison with S. cerevisiae in single strain fermentations (Fig. 2 and Supplementary Data 1). This included primarily strains from the Pichia and Brettanomyces genera, as well as Z. florentina and T. delbrueckii. The P. kluyverii strains differed in their ability to produce acetate ester, P. anomala strains showed high ethyl acetate production, T. delbrueckii produced the highest level of 4VG (POF+), and Z. florentina showed the highest production of fusel alcohols. Only Brettanomyces strains managed to produce 4EG and 4EP, which is probably because of the unique vinyl reductase activity of these species. Fermentations were semi-static with stirring in the first 72 h to allow proliferation of both yeast species in the small tube fermentations. The formation of an aerobic cap on top of the fermentations with film-forming Pichia yeasts was taken as an indirect confirmation of biomass production and flavor development (Fugelsang and Edwards, 2010). The sequential combinations of non-conventional yeast species and S. cerevisiae mostly showed similar fermentation progress using weight loss as an indirect measure of ethanol production, except for the Z. florentina strain which showed intermediate fermentation capacity [Fig. 3; weight loss of 1.22% (w/v) (=1.6% ABV) on day 2 and 1.91% (w/v) (=2.6% ABV) on day 3] and the Bbr2 B. bruxellensis strain which caused stuck fermentation by S. cerevisiae resulting in residual maltose levels and lower ethanol yields (Fig. 3 and Supplementary Fig. 2 and Supplementary Data 1 for all quantified metabolites). Alcoholic fermentation with Z. florentina were also slower in the sequential fermentation compared to the corresponding single strain fermentation after three days (see section 3.2), which is likely due to inhibition on this species by S. cerevisiae after inoculation at 48 h of fermentation. The Bbr2 (AWRI 1499) yeast has not been investigated before in sequential fermentations such as performed in this study. Brettanomyces yeasts are not known to produce killer toxins and the B. bruxellensis Bbr2 yeast is, on the contrary, sensitive to killer toxins produced by Klyveromyces wickerhamii (Mehlomakulu et al., 2014). In wine sequential fermentations with T. delbrueckii and S.cerevisiae (inoculation after 48 h), sluggish fermentations were observed to be linked to low levels of assimilable nitrogen (Taillandier et al., 2014). The sluggish Bbr2-Ale 514 mixed fermentation could therefore either be due to exhaustion of essential nutrients, such as vitamins, trace metals ions (e.g. Zn2+), nitrogen and lipids by Bbr2 in the

malt extract medium, or due to inhibition of S. cerevisiae metabolism (Bisson, 1999; Ciani et al., 2016).

The flavor metabolite profiles of the sequential fermentations showed significant differences compared to single fermentation with S. cerevisiae, seemingly disclosing a species-dependent relationship. All P. kluyverii yeasts were clearly distinguishable from the other yeasts, with more resemblance to S. cerevisiae (Fig. 4).

The sequential fermentation with Bbr2 revealed that it was also a phenotypic outlier for aroma production (Fig. 4), leading to high levels of 4EG, 4EP, ethyl octanoate [1.01 mg/L, well above the sensory threshold of 0.9 mg/L (Meilgaard, 1975)], ethyl decanoate, acetic acid, ethyl acetate and glycerol (Supplementary Fig. 2). On the other hand, we observed lower levels of ethyl hexanoate, isobutanol, isoamyl alcohol, and an absence of isoamyl and isobutyl acetate (Supplementary Fig. 2A). The latter is consistent with previous reports that Brettanomyces yeast may display acetate esterase activity (Steensels et al., 2015). However, the sequential fermentations with the Bbr1 B. bruxellensis yeast did not blot out the isoamyl and isobutyl acetate production, while there was still production of 4EG and 4EP. These results reveal a clear difference between spoilage and beneficial Brettanomyces strains. In addition, the Bbr1 strain also produced the highest levels of acetaldehyde in sequential fermentations (Supplementary Fig. 2A and 4). Interestingly, sequential fermentations with P. kluyverii yielded lower levels of ethyl acetate than the corresponding single strain fermentations (39.4, 38.5, and 53.4 mg/L vs. 41.8, 61.1, and 83.6 mg/L after 48 h for Pkl1, 2 and 3, respectively). This is similar to the ethyl acetate yield of S. cerevisiae in the control fermentation (47.0 mg/L). It was also observed that P. kluyverii Pkl3 retained isoamyl acetate yield similar to that of the S. cerevisiae control fermentation (4.4 vs. 4.7 mg/L, Supplementary Fig. 2A) despite the fact that production of isoamyl acetate was inhibited in all sequential fermentations with non-conventional yeasts. On the other hand, the level of the precursor isoamyl alcohol was much lower in the P. kluyverii sequential fermentation than that produced in the single strain S. cerevisiae control fermentation (Supplementary Fig. 2A). This resulted in a higher isoamyl acetate/isoamyl alcohol conversion ratio (5.0 vs 6.7% for S. cerevisiae single strain control fermentation and P. kluyverii/S. cerevisiae sequential fermentation, respectively). Except for these differences mentioned above, the overall flavor metabolite profiles of the P. kluyverii sequential fermentations were similar to the complete fermentations inoculated only with S. cerevisiae (Fig. 4).

The production of ethyl acetate in sequential fermentations with P. anomala Pan1 was similar to the control single strain fermentation with S. cerevisiae Ale 514 (49.8 mg/L compared to 47.0 mg/L, respectively), whereas P. anomala strain Pan2 produced the highest ethyl acetate yield in all of the

367 sequential fermentations (129.1 mg/L) (Supplementary Fig. 2A and 4). These results suggest lower

368 performance for ethyl acetate production or higher esterase activity occurring in mixed

369 fermentations with the Panl P. anomala strain compared to Pan2, and appears to be a strain-

370 dependent trait.

371 The sequential fermentations using Brettanomyces confirmed the unique ability of Brettanomyces

372 strains to produce ethyl phenols, as 4EP and 4EG were absent with any of the other non-

373 conventional yeasts (Supplementary Fig. 2B and Supplementary Data 1).

374 The sequential fermentations also showed that the P. anomala Pan1 and Pan2 strains produced

375 slightly lower levels of 4VG than in their corresponding single strain fermentations (1.12 and 2.46

376 mg/L vs. 1.89 and 2.89 mg/L after 72 h, respectively), whereas T. delbrueckii and Z. florentina

377 produced the highest levels of 4VG in all the sequential fermentations with 3.04 and 3.40 mg/L,

378 respectively. The similarity between Z. florentina and T. delbrueckii was striking, given that T.

379 delbrueckii performed much better for 4VG production in single strain fermentations (5.77 vs. 1.78

380 mg/L after 72 h, respectively). This revealed that the Z. florentina strain may be a candidate for

381 production of beers with a phenolic profile, although it was unable to produce fruity acetate and

382 ethyl esters, except for ethyl decanoate (Supplementary Fig. 2A and Fig. 2B).

4. Discussion and conclusions

In this work we have first developed a novel, more powerful and convenient methodology for the determination of phenolic compounds simultaneously with their hydroxycinnamic acid precursors in alcoholic fermentations. Previously, methanol has generally been used as a solvent together with electrochemical detection. Most attention has been on 4VG while the ethyl phenolics were not properly measurable. The new methodology presented in this work, which uses an acetonitrile and formic acid mixture as eluent and fluorescence detection, has a better resolution and is more sensitive, allowing accurate measurement of 4VG, 4EP and 4EG, as well as ferulic acid and p-coumaric acid. This facilitates determination of the relevant phenolics compounds more precisely in brewing and winemaking processes with non-conventional yeasts and allowed us to evaluate their use in mixed brewing fermentations.

Non-conventional yeast species generally have a lower performance for ethanol production compared to S. cerevisiae. As a result, they have only been rarely used as pure starter cultures for production of alcoholic beverages. However, they are generally encountered in spontaneous alcoholic fermentations, in which they impart and contribute to unique flavor profiles. However, spontaneous alcoholic fermentations generally suffer from great variability and unpredictability. These observations have raised interest in evaluating non-conventional yeast species for use in co-or sequential fermentations with S. cerevisiae. This has been explored previously for wine fermentations (Ciani and Comitini, 2010; Ciani et al., 2010; Jolly et al., 2014) and it has now been extended in the present work to beer fermentations.

Among the different non-conventional yeast species evaluated for flavor production in the present work, P. kluyverii stood out in producing exceptional levels of isoamyl acetate and ethyl acetate. Isoamyl acetate is a highly desirable flavor compound in beer fermentations, imparting a fruity and banana-like flavor, whereas ethyl acetate gives a solvent-like off-flavor. At the extreme level of more than 900 mg/L ethyl acetate produced in the single fermentation at day 3, there will be a solvent-like off-flavor. However, if the fermentation with P. kluyverii is arrested before this excessive production (after 2 days: 83.6 mg/L ethyl acetate and 11.3 mg/L isoamyl acetate), the production of isoamyl acetate is so high that it might override the negative flavor effect of ethyl acetate.

Although the production of the two esters in single culture fermentations with P. kluyverii was highly pronounced, in sequential fermentations with S. cerevisiae, it was strongly reduced to a similar level as in S. cerevisiae single strain fermentations. There are multiple possible explanations that can be evaluated. First of all, it should be determined whether the removal of P. kluyverii, e.g. by centrifugation and subsequent separation from the medium, after the first fermentation prevents

the disappearance of the huge levels of the two acetate esters in the next fermentation with S. cerevisiae. If this would be the case, it would point to a specific synergistic interaction between the two yeast species, triggering an increase in esterase activity much higher than the sum of that of the two yeast species separately. This could be accomplished for instance by enhanced secretion of esterases (Kurita, 2008). On the other hand, if the removal of P. kluyverii after the first fermentation does not prevent the disappearance of the large amounts of acetate esters accumulated, S. cerevisiae should in some way be responsible for it. This could be caused by intracellular esterase activity in S. cerevisiae, for instance encoded by IAH1 (Fukuda et al., 2000), since acetate esters easily diffuse through cellular membranes. Another way in which P. kluyverii could affect S. cerevisiae esterase activity or its own endogenous acetate ester production in the second fermentation, is through depletion of nutrient levels, especially nitrogen sources [for review see e.g. Verstrepen, Derdelinckx, et al. (2003)], mainly because of biomass formation in the first phase of the fermentation.

Depending on the cause of the ester disappearance, several approaches can be taken to prevent or reduce it. When the disappearance is due to P. kluyverii (secreted) esterase activity, many strains could be screened to identify a strain with low (secreted) esterase activity. Alternatively, putative esterase encoding genes in P. kluyverii could be inactivated by targeted genetic modification, e.g. by CRISPR/Cas9 (Wagner and Alper, 2016), eliminating or reducing esterase activity. This also opens up the possibility that one or more esterases would have higher activity towards isoamyl acetate or ethyl acetate, and that the ratio between the two esters could thus be improved in this way. If S. cerevisiae would be responsible for the dramatic decrease in acetate ester levels, Saccharomyces sp. strain collections could be screened to identify strains with lower esterase activity and compatibility with P. kluyverii flavor production. This might also create the possibility that isoamyl acetate esterase activity would be reduced less than ethyl acetate esterase activity, resulting in an improved ratio of the two acetate esters.

A previous study in wheat beer with lower gravity (10 °P), showed a much more pronounced enhancement of isoamyl acetate in sequential fermentations with P. kluyverii FrootZen and S. cerevisiae (Saerens and Swiegers, 2013). This could be due to the low-gravity conditions and/or the specific P. kluyverii strain or S. cerevisiae wheat strain used. The levels of isoamyl acetate and ethyl acetate reported were significantly lower than the levels measured in our work in the single fermentations with the same P. kluyverii FrootZen yeast. This indicates that careful evaluation of the fermentation conditions and proper strain selection are crucial to unlock the full potential of aroma production by the non-conventional yeast species.

451 P. kluyverii has also been reported to convert precursors present in grapes into polyfunctional thiols

452 for production of Sauvignon Blanc wine (Anfang et al., 2009). The polyfunctional thiols provide

453 conspicuous tropical fruit flavors to the wine. These precursors are also present in aroma hops and a

454 fraction is released by Saccharomyces yeasts during beer fermentation (Cibaka et al., 2016; Cibaka et

455 al., 2015). Sequential or co-fermentation with P. kluyverii might increase the amount of

456 polyfunctional thiols that is released from the precursors in the hop, improving beer aroma. Hence,

457 more non-conventional yeast species could be screened for their capacity to release polyfunctional

458 thiols from hop precursors.

459 Brettanomyces species also show potential for use as starter culture in sequential beer

460 fermentations with S. cerevisiae. They were the only yeasts producing conspicuous levels of 4EG. It

461 imparts spicy notes to the aroma, which is attractive in certain specialty beers. Brettanomyces yeasts

462 have been identified in spontaneous beer fermentations and have also been used already as starter

463 cultures in combination with S. cerevisiae (Steensels et al., 2015). Brettanomyces species secrete

464 glucosidases that hydrolyze glycosylated terpenoid compounds also present in grapes and other

465 fruits. They have been isolated from spontaneous beer fermentations for production of cherry

466 lambic beer and likely contribute in this way to the conspicuous flavor profile of these beers. Also in

467 this case notable differences in glycoside hydrolase activity between species and strains have been

468 observed (Daenen, Saison, et al., 2008; Daenen, Sterckx, et al., 2008). Finally, King and Richard

469 (2000) showed that T. delbrueckii, Lachancea thermotolerans as well as S. cerevisiae brewing yeast

470 could transform hop monoterpene alcohols into other derivatives.

471 The implementation of non-conventional yeast species will introduce new challenges for breweries.

472 For example, it will be problematic to perform serial re-pitching of the inoculum (as it will contain an

473 unknown proportion of both Saccharomyces and non-Saccharomyces yeast), and questions also

474 remain regarding bottling, foam stability and haze formation. The aerobic yeasts, such as P. kluyverii,

475 which showed the greatest potential for production of fruity esters, require oxygen to propagate

476 and form a layer on top of the fermentation medium. For craft brewing of ales, which is performed

477 at 20-25°C, the current sequential fermentation method may be implemented directly in open

478 fermentors. On the other hand, for large-scale high-gravity beer fermentations, which is performed

479 in closed tanks, the wort is typically oxygenated with sterile air to allow the growth of brewing yeast

480 (Gibson et al., 2007). The sequential fermentation procedure ensures optimal flavor production by

481 the non-conventional yeast in the first part of the fermentation, whereas inhibitory conditions such

as low temperature (8-12°C), CO2 pressure, and an anaerobic environment would occur only later in the alcoholic fermentation when the brewing yeast is active.

In conclusion, we have developed a new HPLC-UV/fluorescence methodology that facilitates determination of phenolic compounds for alcoholic beverages, and is particularly useful for determination of ethyl phenolic compounds produced by Brettanomyces species. We have screened 17 non-conventional yeast species for their potential usage as natural flavor enhancers and evaluated the best candidate strains in sequential fermentations with S. cerevisiae. Application of P. kluyverii in the brewing process is a very promising method to enhance fruity acetate esters, but gave surprisingly low yields in sequential fermentations. Future studies should investigate optimization with regards to development of strains with less esterase activity and using optimal fermentation conditions, especially oxygen levels, to unlock the full potential for aroma improvement by the non-conventional yeast species.


The authors gratefully acknowledge the kind borrowing of HPLC equipment from Wim Van den Ende and the Sugar Metabolism Group in the Molecular Plant Biology laboratory of KU Leuven. We acknowledge Tom Struyf and Jimmy Soares for practical advice and fruitful discussions regarding development of the novel fluorescence/UV method. We also thank Sofie Saerens and Hentie Swiegers from Chr. Hansen A/S for the kind gift of the P. kluyverii FrootZen yeast, and the Australian Wine Research Institute for providing the B. bruxellensis AWRI1499 strain. We thank the two anonymous reviewers for careful reading and useful feedback for improvement of the manuscript.


This work was supported by an SBO grant (IWT 140044) from IWT-Flanders, KU Leuven Industrial Research Fund (IOF Knowledge platform) and the EC 7th Framework program (CORNUCOPIA project).


Andorra, I., Berradre, M., Rozes, N., Mas, A., Guillamon, J.M., Esteve-Zarzoso, B., 2010. Effect of pure and mixed cultures of the main wine yeast species on grape must fermentations. Eur Food Res Technol 231, 215-224.

514 Anfang, N., Brajkovich, M., Goddard, M.R., 2009. Co-fermentation with Pichia kluyverii increases

515 varietal thiol concentrations in Sauvignon Blanc. Aust J Grape Wine Res 15, 1-8.

516 Aznar, O., Checa, A., Oliver, R., Hernandez-Cassou, S., Saurina, J., 2011. Determination of

517 polyphenols in wines by liquid chromatography with UV spectrophotometry detection. J Sep Sci 34,

518 527-535.

519 Azzolini, M., Tosi, E., Lorenzini, M., Finato, F., Zapparoli, G., 2015. Contribution to the aroma of white

520 wines by controlled Torulaspora delbrueckii cultures in association with Saccharomyces cerevisiae.

521 World J Microbiol Biotechnol 31, 277-293.

522 Basso, R.F., Alcarde, A.R., Portugal, C.B., 2016. Could non-Saccharomyces yeasts contribute on

523 innovative brewing fermentations? Food Res Int 86, 112-120.

524 Bisson, L.F., 1999. Stuck and sluggish fermentations. Am J Enol Vitic 50, 107-119.

525 Bokulich, N.A., Bamforth, C.W., 2013. The microbiology of malting and brewing. Microbiol Mol Biol

526 Rev 77, 157-172.

527 Callemien, D., Collin, S., 2009. Structure, organoleptic properties, quantification methods, and

528 stability of phenolic compounds in beer—a review. Food Rev Int 26, 1-84.

529 Canonico, L., Agarbati, A., Comitini, F., Ciani, M., 2016. Torulaspora delbrueckii in the brewing

530 process: A new approach to enhance bioflavour and to reduce ethanol content. Food Microbiol 56,

531 45-51.

532 Ciani, M., Capece, A., Comitini, F., Canonico, L., Siesto, G., Romano, P., 2016. Yeast interactions in

533 inoculated wine fermentation. Front Microbiol 7, 555.

534 Ciani, M., Comitini, F., 2010. Non-Saccharomyces wine yeasts have a promising role in

535 biotechnological approaches to winemaking. Ann Microbiol 61, 25-32.

536 Ciani, M., Comitini, F., Mannazzu, I., Domizio, P., 2010. Controlled mixed culture fermentation: a new

537 perspective on the use of non-Saccharomyces yeasts in winemaking. FEMS Yeast Res 10, 123-133.

538 Ciani, M., Ferraro, L., 1998. Combined use of immobilized Candida stellata cells and Saccharomyces

539 cerevisiae to improve the quality of wines. Journal of Applied Microbiology 85, 247-254.

540 Cibaka, M.L., Decourriere, L., Lorenzo-Alonso, C.J., Bodart, E., Robiette, R., Collin, S., 2016. 3541 Sulfanyl-4-methylpentan-1-ol in Dry-Hopped Beers: First Evidence of Glutathione S-Conjugates in

542 Hop (Humulus lupulus L.). J Agric Food Chem 64, 8572-8582.

543 Cibaka, M.L., Gros, J., Nizet, S., Collin, S., 2015. Quantitation of selected terpenoids and mercaptans

544 in the dual-purpose hop varieties Amarillo, Citra, Hallertau Blanc, Mosaic, and Sorachi Ace. J Agric

545 Food Chem 63, 3022-3030.

546 Clemente-Jimenez, J.M., Mingorance-Cazorla, L., Martinez-Rodriguez, S., Las Heras-Vazquez, F.J.,

547 Rodriguez-Vico, F., 2005. Influence of sequential yeast mixtures on wine fermentation. Int J Food

548 Microbiol 98, 301-308.

549 Comitini, F., Gobbi, M., Domizio, P., Romani, C., Lencioni, L., Mannazzu, I., Ciani, M., 2011. Selected

550 non-Saccharomyces wine yeasts in controlled multistarter fermentations with Saccharomyces

551 cerevisiae. Food Microbiol 28, 873-882.

552 Curtin, C.D., Borneman, A.R., Chambers, P.J., Pretorius, I.S., 2012. De-novo assembly and analysis of

553 the heterozygous triploid genome of the wine spoilage yeast Dekkera bruxellensis AWRI1499. PLoS

554 One 7, e33840.

555 Daenen, L., Saison, D., Sterckx, F., Delvaux, F.R., Verachtert, H., Derdelinckx, G., 2008. Screening and

556 evaluation of the glucoside hydrolase activity in Saccharomyces and Brettanomyces brewing yeasts. J

557 Appl Microbiol 104, 478-488.

558 Daenen, L., Sterckx, F., Delvaux, F.R., Verachtert, H., Derdelinckx, G., 2008. Evaluation of the

559 glycoside hydrolase activity of a Brettanomyces strain on glycosides from sour cherry (Prunus

560 cerasus L.) used in the production of special fruit beers. FEMS Yeast Res 8, 1103-1114.

561 De Francesco, G., Turchetti, B., Sileoni, V., Marconi, O., Perretti, G., 2015. Screening of new strains of

562 Saccharomycodes ludwigii and Zygosaccharomyces rouxii to produce low-alcohol beer. J Inst Brew

563 121, 113-121.

564 Domizio, P., Romani, C., Lencioni, L., Comitini, F., Gobbi, M., Mannazzu, I., Ciani, M., 2011. Outlining

565 a future for non-Saccharomyces yeasts: selection of putative spoilage wine strains to be used in

566 association with Saccharomyces cerevisiae for grape juice fermentation. Int J Food Microbiol 147,

567 170-180.

568 Fugelsang, K.C., Edwards, C.G., 2010. Chaptor 11 - Wine spoilage, in: Fugelsang, K.C., Edwards, C.G.

569 (Eds.), Wine Microbiology. Springer Science and Business Media, New York, pp. 166-167.

570 Fukuda, K., Kiyokawa, Y., Yanagiuchi, T., Wakai, Y., Kitamoto, K., Inoue, Y., Kimura, A., 2000.

571 Purification and characterization of isoamyl acetate-hydrolyzing esterase encoded by the IAH1 gene

572 of Saccharomyces cerevisiae from a recombinant Escherichia coli. Appl Microbiol Biotechnol 53, 596573 600.

574 Gibson, B.R., 2011. 125th Anniversary Review: Improvement of Higher Gravity Brewery

575 Fermentation via Wort Enrichment and Supplementation. Journal of the Institute of Brewing 117,

576 268-284.

577 Gibson, B.R., Lawrence, S.J., Leclaire, J.P., Powell, C.D., Smart, K.A., 2007. Yeast responses to stresses

578 associated with industrial brewery handling. FEMS Microbiol Rev 31, 535-569.

579 Granato, T.M., Romano, D., Vigentini, I., Foschino, R.C., Monti, D., Mamone, G., Ferranti, P., Nitride,

580 C., Iametti, S., Bonomi, F., Molinari, F., 2014. New insights on the features of the vinyl phenol

581 reductase from the wine-spoilage yeast Dekkera/Brettanomyces bruxellensis. Ann Microbiol 65, 321582 329.

583 Izquierdo Canas, P.M., Palacios Garcia, A.T., Garcia Romero, E., 2011. Enhancement of flavour

584 properties in wines using sequential inoculations of non-Saccharomyces (Hansenula and

585 Torulaspora) and Saccharomyces yeast starter. Vitis 50, 177-182.

586 Jolly, N.P., Varela, C., Pretorius, I.S., 2014. Not your ordinary yeast: non-Saccharomyces yeasts in

587 wine production uncovered. FEMS Yeast Res 14, 215-237.

588 Kim, D.H., Hong, Y.A., Park, H.D., 2008. Co-fermentation of grape must by Issatchenkia orientalis and

589 Saccharomyces cerevisiae reduces the malic acid content in wine. Biotechnol Lett 30, 1633-1638.

590 King, A., Richard, D.J., 2000. Biotransformation of monoterpene alcohols by Saccharomyces

591 cerevisiae,Torulaspora delbrueckii and Kluyveromyces lactis. Yeast 16, 499-506.

592 Kurita, O., 2008. Increase of acetate ester-hydrolysing esterase activity in mixed cultures of

593 Saccharomyces cerevisiae and Pichia anomala. J Appl Microbiol 104, 1051-1058.

594 Kurtzman, C.P., Robnett, C.J., 1998. Identification and phylogeny of ascomycetous yeasts from

595 analysis of nuclear large subunit (26S) ribosomal DNA partial sequences. Antonie Van Leeuwenhoek

596 73, 331-371.

597 Mehlomakulu, N.N., Setati, M.E., Divol, B., 2014. Characterization of novel killer toxins secreted by

598 wine-related non-Saccharomyces yeasts and their action on Brettanomyces spp. Int J Food Microbiol

599 188, 83-91.

600 Meilgaard, M.C., 1975. Flavor chemistry of beer: part II: flavour and threshold of 239 aroma

601 volatiles. M BAA TQ 12, 151-168.

602 Meilgaard, M.C., 1982. Prediction of flavor differences between beers from their chemical

603 composition. J Agric Food Chem 30, 1009-1017.

604 Michel, M., Kopecka, J., Meier-Dornberg, T., Zarnkow, M., Jacob, F., Hutzler, M., 2016. Screening for

605 new brewing yeasts in the non-Saccharomyces sector with Torulaspora delbrueckii as model. Yeast

606 33, 129-144.

607 Michel, M., Meier-Dörnberg, T., Jacob, F., Methner, F., Wagner, R.S., Hutzler, M., 2016. Review: Pure

608 non-Saccharomyces starter cultures for beer fermentation with a focus on secondary metabolites

609 and practical applications. J Inst Brew 122, 569-587.

610 Moreira, N., Mendes, F., Guedes de Pinho, P., Hogg, T., Vasconcelos, I., 2008. Heavy sulphur

611 compounds, higher alcohols and esters production profile of Hanseniaspora uvarum and

612 Hanseniaspora guilliermondii grown as pure and mixed cultures in grape must. Int J Food Microbiol

613 124, 231-238.

Moreira, N., Mendes, F., Hogg, T., Vasconcelos, I., 2005. Alcohols, esters and heavy sulphur compounds production by pure and mixed cultures of apiculate wine yeasts. Int J Food Microbiol 103, 285-294.

Mukai, N., Masaki, K., Fujii, T., Kawamukai, M., Iefuji, H., 2010. PAD1 and FDC1 are essential for the decarboxylation of phenylacrylic acids in Saccharomyces cerevisiae. J Biosci Bioeng 109, 564-569. Mukherjee, V., Radecka, D., Aerts, G., Verstrepen, K.J., Lievens, B., Thevelein, J.M., 2017. Phenotypic landscape of non-conventional yeast species for different stress tolerance traits desirable in bioethanol fermentation. Biotechnol Biofuels 10, 216.

Rodriguez, M.E., Lopes, C.A., Barbagelata, R.J., Barda, N.B., Caballero, A.C., 2010. Influence of Candida pulcherrima Patagonian strain on alcoholic fermentation behaviour and wine aroma. Int J Food Microbiol 138, 19-25.

Sadoudi, M., Tourdot-Marechal, R., Rousseaux, S., Steyer, D., Gallardo-Chacon, J.J., Ballester, J., Vichi, S., Guerin-Schneider, R., Caixach, J., Alexandre, H., 2012. Yeast-yeast interactions revealed by aromatic profile analysis of Sauvignon Blanc wine fermented by single or co-culture of non-Saccharomyces and Saccharomyces yeasts. Food Microbiol 32, 243-253.

Saerens, S., Swiegers, J.H., 2013. W02013030398: Enhancement of beer flavor by a combination of pichia yeast and different hop varieties.

Schifferdecker, A.J., Dashko, S., Ishchuk, O.P., Piskur, J., 2014. The wine and beer yeast Dekkera bruxellensis. Yeast 31, 323-332.

Shinohara, T., Kubodera, S., Yanagida, F., 2000. Distribution of phenolic yeasts and production of phenolic off-flavors in wine fermentation. J Biosci Bioeng 90, 90-97.

Steensels, J., Daenen, L., Malcorps, P., Derdelinckx, G., Verachtert, H., Verstrepen, K.J., 2015. Brettanomyces yeasts--From spoilage organisms to valuable contributors to industrial fermentations. Int J Food Microbiol 206, 24-38.

Taillandier, P., Lai, Q.P., Julien-Ortiz, A., Brandam, C., 2014. Interactions between Torulaspora delbrueckii and Saccharomyces cerevisiae in wine fermentation: influence of inoculation and nitrogen content. World J Microbiol Biotechnol 30, 1959-1967.

Tataridis, P., Kanelis, A., Logotetis, S., Nerancis, E., 2013. Use of non-saccharomyces Torulaspora delbrueckii yeast strains in winemaking and brewing. Zb Mat Srp Prir Nauk, 415-426. Vanbeneden, N., Delvaux, F., Delvaux, F.R., 2006. Determination of hydroxycinnamic acids and volatile phenols in wort and beer by isocratic high-performance liquid chromatography using electrochemical detection. J Chromatogr A 1136, 237-242.

Verstrepen, K.J., Derdelinckx, G., Dufour, J.P., Winderickx, J., Thevelein, J.M., Pretorius, I.S., Delvaux, F.R., 2003. Flavor-active esters: adding fruitiness to beer. J Biosci Bioeng 96, 110-118. Verstrepen, K.J., Van Laere, S.D., Vanderhaegen, B.M., Derdelinckx, G., Dufour, J.P., Pretorius, I.S., Winderickx, J., Thevelein, J.M., Delvaux, F.R., 2003. Expression levels of the yeast alcohol acetyltransferase genes ATF1, Lg-ATF1, and ATF2 control the formation of a broad range of volatile esters. Appl Environ Microbiol 69, 5228-5237.

Viana, F., Gil, J.V., Valles, S., Manzanares, P., 2009. Increasing the levels of 2-phenylethyl acetate in wine through the use of a mixed culture of Hanseniaspora osmophila and Saccharomyces cerevisiae. Int J Food Microbiol 135, 68-74.

Wagner, J.M., Alper, H.S., 2016. Synthetic biology and molecular genetics in non-conventional yeasts: Current tools and future advances. Fungal Genet Biol 89, 126-136.

White, T.J., Bruns, T., Lee, S., Taylor, J., 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics, in: Innis, M.A., Gelfand, D.H., Sninsky, J.J., White, T.J. (Eds.), PCR protocols: a guide to methods and applications, San Diego, Calif: Academic Press, Inc, pp. 315322.

Zhu, M., Cui, Y., 2013. Determination of 4-vinylgaiacol and 4-vinylphenol in Top-fermented Wheat Beers by Isocratic High Performance Liquid Chromatography with Ultraviolet Detector. Braz Arch Biol Technol 56, 1018-1023.

Figure legends

Figure 1. UV and fluorescence HPLC chromatograms of phenolic compounds and hydroxycinnamic acid precursor standards spiked in malt extract medium. The standard shown contains 2.38 mg/L of each phenolic compound and precursor acids spiked into malt extract medium. A) UV chromatogram recorded at 280 nm B) Fluorescence chromatogram with excitation/emission at 210/340 nm. C) Standard curves showing linearity of phenolic compounds measurement with UV and fluorescence in spiked malt extract medium. The standards included 0.28, 0.57, 1.16, 2.38, 4.87 and 10 mg/L. The x-scale indicates the phenolic compound in mg/L and the y-scale indicates the peak height of the fluorescence intensity or UV absorbance in ^V. The malt extract medium contains ferulic and p-coumaric acid, a low amount of 4VG and the calibration curve therefore does not pass through zero.

Figure 2. Heatmap indicating secondary quantified metabolites from fermentations of single non-conventional yeast species. The fermentations were sampled after 48 h (A) and 72 h (B). The fermentations were performed in EBC-validated tubes with 60 mL 20°P malt extract media. Values of each of the metabolites (X) were normalized with the average and the standard deviation (a), to calculate Z-scores = (X- ^/a, which is a measure of the deviation from the mean. Scores for all flavor compounds ranged from -5 (red), through 0 (white) indicating average measurements, to 5 (green) standard deviations. Euclidean distances calculated from the Z-scores were used to make a neighbor joining phenotypic tree with BioNumerics 7.6. Numbers on the nodes indicate the goodness of the clustering in % (cophenetic correlation coefficients). Branch lengths are the relative intercluster NJ-distances.

Figure 3. Profile of fermentations sequentially inoculated with non-conventional yeast species and S. cerevisiae. The weight loss is shown as percentage of the starting weight of the malt extract medium. After day 2 (48 h) the fermentations were inoculated with S. cerevisiae Ale 514 brewing yeast to finish the alcoholic fermentation. A single inoculation of the S. cerevisiae Ale 514 strain was included as a reference (closed circles). Sequential inoculation with Bbr1 (closed downward triangles), Bbr2 (open downward triangles), Pan1 (closed upward triangles), Pan2 (open upward triangles), Pkl1 (closed squares), Pkl2 (open squares), Pkl3 (open squares with cross), Tde (closed diamonds), Zfl (open diamonds). After completion of the alcoholic fermentation samples were kept at 4°C until analysis.

Figure 4. Metabolites produced in sequential fermentations with S. cerevisiae MauriBrew Ale 514 analyzed with principle component analysis (PCA). The fermentations were first inoculated with non-Saccharomyces yeast, and after 48 h with the MauriBrew Ale 514 yeast to complete the

698 alcoholic fermentation. Shown here are the first two components PC1 and PC2 that together explain

699 72.1% of the variance in the data. Supplementary Fig. 3 shows a scree plot of the total variances and

700 Supplementary Fig. 4 PC3 and PC4, which together explain 19.3% of the variance. The variation in

701 PC3 and PC4 is mainly due to differences in ethyl acetate and acetaldehyde. The data was

702 normalized and analyzed with the R package factoextra.

703 Tables

704 Table 1. Comparison of fermentation studies performed up till now with non-Saccharomyces yeast

705 co- or sequentially inoculated with Saccharomyces to complete the alcoholic fermentation. The

706 studies are primarily done in wine fermentations for which the grape cultivar is indicated. CD

707 positive, cinnamic acid decarboxylase positive; + increase in the production; - decrease in the

708 production; = the levels were similar; n.a., information not available. A slash represents two

709 fermentation trails in the same study.

Strain Acetate esters Fusel alcohols Ethyl esters (C6-C10) CD positive Reference Grape must (cultivar), media, or beer Inoculation type

Candida stellata + Sadoudi et al. (2012) Sauvignon Blanc Co

stellata + + Ciani and Ferraro (1998) Pinot grigio Co,Seq

stellata + + + Andorra et al. (2010) Macabeo Co

Hanseniaspora osmophilia + Viana et al. (2009) Bobal Co

uvarum + + + Andorra et al. (2010) Macabeo Co

uvarum + Moreira et al. (2008) Cultivar not specified Co

Lachancea thermotolerans + Comitini et al. (2011) Airen Seq

Metschnikowia pulcherrima + + + Rodriguez et al. (2010) Muscat d'Alexandrie Co, Seq

pulcherrima + + + Sadoudi et al. (2012) Sauvignon Blanc Co

Pichia anomal a + Kurita (2008) Synthetic YN medium Co

anomala + + + Domizio et al. (2011) Macabeo Co

anomala +/- =/= Izquierdo Canas et al. (2011) Airen Seq

fermentans + + + Clemente-Jimenez Macabeo Co

et al. (2005)

guilliermondii + Moreira et al. (2005) Synthetic YM medium Co

guilliermondii + Moreira et al. (2008) Cultivar not specified Co

kluyverii + = (C6) + (C8-C10) Saerens and Swiegers (2013) Wheat beer Seq

kudriavzevii Kim et al. (2008) Campbell's Early Co

Saccharomycodes ludwigii +(EA) Isobutanol n.a. Domizio et al. (2011) Macabeo Co

Torulaspora delbrueckii + Low Azzolini et al. (2015) Soave, Chardonnay Co

delbrueckii +/- -/+ =/- =/+ Izquierdo Canas et al. (2011) Airen Seq

Zygotorulaspora florentina + + + (C6) - (C8-C10) Domizio et al. (2011) Macabeo Co

711 Table 2. Strains used in this study. Genera are indicated with bold. n.a., information not available

Strain number Species Origin Geo Collection ref.

Ale 514 Saccharomyces cerevisiae Commercial wine strain. Ab Mauri, Australia n.a.

Zymaflore VL3 cerevisiae Commercial wine strain. Laffort, France France

Bmo Blastobotrys mokoenaii Soil South Africa x9113

Ban Brettanomyces anomalus n.a. n.a. X9073

Bbr1 bruxellensis Beer UK CBS 3025

Bbr2 bruxellensis Wine spoilage. Isolated by AWRI, Australia Australia AWRI1499

Bna naardenensis Dr. Pepper soda USA NRRL Y-5740

Cst Candida stellata Wine grapes Germany X9023

Cma Citeromyces matritensis Beet sugar thick Juice Belgium ST1312/081

Dha Debaryomyces hansenii Cacao fermentation Ghana x38

Koh Kodamaea ohmeri Cacao fermentation Ghana x22

Lth Lachancea thermotolerans Mirabelle plum conserve Russia x9005

Mre Metschnikowia reukaufii Nectar (Pulmonaria officinalis) Belgium Y6.3K/FT11 B

Pan1 Pichia anomala Maple syrup n.a. x9015

Pan2 anomala Cacao fermentation Ivory Coast x10

Pkl1 kluyverii Cacao fermentation Ghana x21

Pkl2 kluyverii Cacao fermentation Ghana x36

Pkl3 kluyverii Commercial wine strain. Chr. Hansen, Denmark New Zealand Frootzen

Pku1 kudriavzevii Cacao fermentation Ivory Coast x12

Pku2 kudriavzevii Ginger beer West Africa X9035

Sba Starmerella bacillaris Wine Italy X9029

Sbo bombicola Nectar (Helleborus foetidus) Belgium V10.2Y A1

Tde Torulaspora delbrueckii Beet sugar thick Juice Belgium ST1312/167

Zfl Zygotorulaspora florentin a Sulphited grape must Italy X9022

713 Supplementary Data 1. Excel sheets containing quantified values of metabolites produced in single

714 fermentations on day 2 (A.) and day 3 (B.) and sequentially inoculated fermentations (C.).

715 Supplementary Fig. 1. Correlations in the production of flavor compounds. The phenotypic

716 dendrograms show correlations in the levels of the quantified aroma compounds and primary

717 metabolites after 48h and 72h in single strain inoculum fermentations. Numbers on the nodes

718 indicate the goodness of the clustering in % (cophenetic correlation coefficients). Branch lengths are

719 the relative intercluster NJ-distances.

720 Supplementary Fig. 2. Metabolite production of sequential fermentations with non-

721 Saccharomyces yeast and the MauriBrew Ale 514 yeast. The fermentations were first inoculated

722 with non-Saccharomyces yeast, and after 48 h with the MauriBrew Ale 514 yeast to complete the

723 alcoholic fermentation. Bar graphs indicate duplicate fermentations. In case of no values, it is below

724 detection threshold of the assay. All glucose was consumed after the sequential fermentations.

725 Supplementary Fig. 3. Scree plot of the PCA analysis. The scree plot illustrates the number of

726 dimensions for the transformed data and the percentage of the explained variances.

727 Supplementary Fig. 4. PCA component 3 and 4 of the metabolites produced in sequential

728 fermentations. The fermentations were first inoculated with non-Saccharomyces yeast, and after 48

729 h with the MauriBrew Ale 514 yeast to complete the alcoholic fermentation. Shown here are the

730 two components PC3 and PC4, which together explain 19.3% of the variance. The variation here is

731 mainly due to differences in ethyl acetate and acetaldehyde.

fig1.pdf 1 20/04/2017 19:45

> 150000

CD O c CO .Q



Coumaric acid Ferulic acid


10 15 20 25 30

Retention time (min)

^ 200000] (D



g 20,000 CL



g> 200,000-

"<D .C

® 100,000

y = 4,341x + 827

R2 = 0.9999

Retention time (min) 1,500,000

g> 1,000,000

® 500,000H


Ferulic acid

y = 15,455x + 89,602

R2 = 0.9962


y = 28,470x + 52,741 R2 = 0.9987

2.5 5.0 7.5 10.0 Concentration (mg/l)

800,000 600,000 400,000 200,000 0


g> 200,000 g 100,000

y = 95,456x

R2 = 0.9998

y = 66,894x

R2 = 0.9994

y = 23,847x + 22,624

R2 = 0.996

0.0 2.5 5.0 7.5 Concentration (mg/l)



■ ■

-5 0 5




O — — (D O (D TO p .C O TD

E 3 ^ ^


Ale 514

Zymaflore VL3


-s Z-score О s




И ñ £

3 — — —

Ш " (Л

■■ ■ ■

■ ■ ■

■ ■

■ ■ ■

■ ■ ■

Ale 514

Zymaflore VL3

^ 5-■S

w 3-\ O

01 23456789 10 11 | Fermentation time (days)

S. cerevisiae inoculated Non-conventional yeast inoculated

PC1 (54.3%)

1 Highlights

2 • Non-Saccharomyces yeasts have been screened for their bioflavoring potential in beer

3 • Pichia kluyverii is unique for its strong production of fruity esters

4 • Ethyl and vinyl phenols and their precursors can be determined simultaneously with HPLC

5 • Brettanomyces sequential inoculation can be positive or negative for flavor profile

6 • Aroma contribution is not additive in sequential beer fermentations