Applied Microbiology and Biotechnology

, Volume 82, Issue 6, pp 1143–1156

The role of oxygen in yeast metabolism during high cell density brewery fermentations


    • Centre for Malting and Brewing Science, Faculty of Bioscience EngineeringKatholieke Universiteit Leuven
  • S. M. G. Saerens
    • Centre for Malting and Brewing Science, Faculty of Bioscience EngineeringKatholieke Universiteit Leuven
    • Laboratory of Molecular Cell Biology (VIB), Department of Molecular Microbiology, Institute of Botany and MicrobiologyKatholieke Universiteit Leuven
  • S. E. Van Mulders
    • Centre for Malting and Brewing Science, Faculty of Bioscience EngineeringKatholieke Universiteit Leuven
  • F. Delvaux
    • Centre for Malting and Brewing Science, Faculty of Bioscience EngineeringKatholieke Universiteit Leuven
  • F. R. Delvaux
    • Centre for Malting and Brewing Science, Faculty of Bioscience EngineeringKatholieke Universiteit Leuven
Applied Microbial and Cell Physiology

DOI: 10.1007/s00253-009-1909-8

Cite this article as:
Verbelen, P.J., Saerens, S.M.G., Van Mulders, S.E. et al. Appl Microbiol Biotechnol (2009) 82: 1143. doi:10.1007/s00253-009-1909-8


The volumetric productivity of the beer fermentation process can be increased by using a higher pitching rate (i.e., higher inoculum size). However, the decreased yeast net growth observed in these high cell density fermentations can have a negative impact on the physiological stability throughout subsequent yeast generations. The use of different oxygen conditions (wort aeration, wort oxygenation, yeast preoxygenation) was investigated to improve the growth yield during high cell density fermentations and yeast metabolic and physiological parameters were assessed systematically. Together with a higher extent of growth (dependent on the applied oxygen conditions), the fermentation power and the formation of unsaturated fatty acids were also affected. Wort oxygenation had a significant decreasing effect on the formation of esters, which was caused by a decreased expression of the alcohol acetyl transferase gene ATF1, compared with the other conditions. Lower glycogen and trehalose levels at the end of fermentation were observed in case of the high cell density fermentations with oxygenated wort and the reference fermentation. The expression levels of BAP2 (encoding the branched chain amino acid permease), ERG1 (encoding squalene epoxidase), and the stress responsive gene HSP12 were predominantly influenced by the high cell concentrations, while OLE1 (encoding the fatty acid desaturase) and the oxidative stress responsive genes SOD1 and CTT1 were mainly affected by the oxygen availability per cell. These results demonstrate that optimisation of high cell density fermentations could be achieved by improving the oxygen conditions, without drastically affecting the physiological condition of the yeast and beer quality.


FermentationBrewer’s yeastYeast physiologyStress responseOxygenFlavour compounds


In the traditional production of lager beer, the fermentation process is the most time-consuming step, which takes about 1–2 weeks before entering the maturation period. Therefore, an important objective of modern fermentation science and technology is to reduce the fermentation time while producing an end product of similar quality and thus allowing great time and money savings. To improve the volumetric productivity of the beer fermentation process, a promising strategy may be the enhancement of the amount of suspended yeast cells in a batch fermentor (i.e., increasing “the pitching rate”; Okabe et al. 1992; Verbelen et al. 2008).

Previous research on high cell density (HCD) fermentations revealed that the fermentation time could be significantly reduced (Okabe et al. 1992; Edelen et al. 1996; Erten et al. 2007; Verbelen et al. 2008). Despite the observation of a slightly higher stress response during fermentations pitched with higher yeast concentrations, Verbelen et al. (2009a) concluded that the yeast physiological condition was only moderately affected by high cell densities.

However, to ensure the reuse and quality of the resulting yeast slurry after fermentation, adequate yeast growth must be ensured. Higher cell concentrations in the initial phase of the fermentations result in a lower build-up of unsaturated fatty acids per cell, which causes a decreased yeast net growth (Verbelen et al. 2009a). It is hypothesised that oxygen availability is decreased when high concentrations of yeast cells are present (Edelen et al. 1996). Therefore, an optimisation of the oxygen supply could improve the sustainability of HCD fermentations.

The major role of oxygen in brewery fermentations is to promote the biosynthesis of unsaturated fatty acids (UFA) and sterol, which are required for adequate yeast growth during fermentation (Boulton and Quain 2001). In traditional brewing, yeast is cropped after fermentation and stored for reuse in subsequent fermentations. Cropped yeast cells contain insufficient sterols and UFA and therefore oxygen supplementation is needed to ensure optimal fermentation performance (Moonjai et al. 2002). Sterols and UFA regulate the membrane fluidity and permeability. They also influence the activity of membrane-bound enzymes and affect the response to stress (Swan and Watson 1998; Chatterjee et al. 2000; Jones et al. 2005). If oxygen becomes limiting, the sterol content decreases through yeast growth and below a threshold of 0.1% (w/w) no further growth is possible (Aries and Kirsop 1977). Regarding critical UFA levels, Casey et al. (1984) and Ohno and Takahashi (1986) found UFA levels between 4.8 and 7.6 mg/g at the end of fermentation.

Normally, oxygen is provided through wort aeration or oxygenation prior to pitching. This technique, however, has several drawbacks, such as over-aeration or under-aeration, leading to inconsistent fermentations. To avoid these shortcomings, yeast can be oxygenated in a controlled process before fermentation, thereby removing the requirement for subsequent wort oxygenation (Masschelein et al. 1995; Boulton et al. 2000; Depraetere et al. 2003).

Paradoxical to the need of oxygen for lipid biosynthesis, oxygen can cause yeast degeneration through the formation of highly reactive oxygen species (ROS; O2·, H2O2, OH·). ROS can attack cellular compounds (DNA, lipids, sugar, and proteins) and can lead to cell death. To protect against oxidative damage, cells possess defense mechanisms, including enzymes, such as peroxidases, catalases, and superoxide dismutases, and antioxidants such as glutathione and vitamins. Genes such as SOD1, SOD2, CTT1, CTA1, GSH1, GLR1, GRX1, and GRX2 are involved in the oxidative stress response (Gibson et al. 2008). Different transcription factors have been involved in the regulation of gene expression under oxidative stress. Msn2/4 are transcription factors of the general stress response, which bind with the STRE (CCCCT) sequence (Martinez-Pastor et al. 1996). Msn2/4 are negatively regulated by the cyclic adenosine monophosphate–protein kinase A (cAMP-PKA) pathway (Smith et al. 1998). STRE sequences have been identified in promoter sections of many stress-induced genes, such as heat shock protein (HSP) genes (e.g., HSP12 and HSP104), CTT1, and genes that contribute to the synthesis (TPS1, TPS2, TPS3, and TSL1) and degradation (NTH1, NTH2) of trehalose (Winderickx et al. 1996; Zähringer et al. 1997). Furthermore, Yap1 is a specific transcription factor, which can bind to the antioxidant response element (TGACTCA) sequence (Estruch 2000).

In this study, different oxygenation conditions were applied to HCD fermentations to investigate the role of oxygen on yeast physiology and metabolism in these fermentation types. Because it is believed that the oxygen availability is limited in high cell density worts, the application of preoxygenated yeast was also investigated. Specific gene expression and physiological parameters were monitored predominantly in the first hours of fermentation because this period determines the further course of fermentation. These results are crucial in elucidating the physiological behavior of yeast in high cell density populations and for the application of HCD fermentations in the brewing industry.

Materials and methods

Yeast strain and medium

The study was carried out with an industrial lager brewing strain of Saccharomyces cerevisiae (pastorianus; CMBSPV09; Katholieke Universiteit Leuven, Centre for Malting and Brewing Science, Heverlee, Belgium), obtained from a brewery yeast storage vessel.

Sterile all-malt hopped wort with an extract content of 15°P (gram extract per 100 g wort; 53% maltose, 33% maltotriose, 9% glucose, 4% sucrose) and with a free amino nitrogen (FAN) content of 291 ppm was made in a pilot brewery. After centrifugation (3,000 rpm, 3 min), the wort was decantated. This wort had a reduced fatty acid content (0.1 ppm C16:0, 2.6 ppm C16:1, 0.6 ppm C18:0, 2.9 ppm C18:1, 1.6 ppm C18:2, and 0.2 ppm C18:3) and was used throughout the study.

Fermentation conditions and sampling

Yeast preoxygenation was performed at 20°C in a membrane loop reactor, according to Depraetere et al. (2003). Yeast slurries (3 l; 75–80% moisture) were circulated at 750 ml/min during 5 h. Oxygen was delivered to the slurry via the membrane sparger to obtain an oxygen concentration of 8 mg/l in the slurry. Maltose (3%) was added to the yeast solution prior to preoxygenation.

Five different experimental conditions were examined in this study (Table 1). The worts were sparged with air, oxygen or nitrogen, for 10 min before pitching. Four different oxygenation conditions were applied to HCD fermentations (pitching rate 80 × 106 cells/ml): (1) non-preoxygenated yeast with aerated wort (A/NP), (2) non-preoxygenated yeast with oxygenated wort (O/NP), (3) preoxygenated yeast with oxygen-depleted wort (N/PR), (4) preoxygenated yeast with aerated wort (A/PR). In addition, a reference fermentation (REF; pitching rate 20 × 106 cells/ml, non-preoxygenated yeast with wort aeration) was taken along. The dissolved oxygen concentration of each medium was measured with an oxygen sensor (Oxi 340/SET, WTW, Weinheim, Germany). The concentration and viability of the yeast slurries were determined by flow cytometry (YeastCyte, BioDETECT AS, Oslo, Norway) before the required amount was pitched in the wort. All fermentations were carried out in duplicate, in tall tubes (75 cm tall, 8 cm internal diameter), containing 1.8 l sterile 15°P wort medium. The fermentations were performed at 15°C and were monitored frequently by withdrawing samples through a narrow sampling tube (15 cm from the bottom) with the aid of N2 overpressure. Samples were cooled directly on ice and the yeast and fermenting wort were separated by centrifugation (2,800 rpm, 3 min, 2°C). Fermentations were stopped at around 80% apparent degree of fermentation (ADF) and the tubes were cooled down at 2°C for 24 h to sediment the yeast. The supernatant of the tube (beer) was collected and the remaining slurry was resuspended in 1 l cold sterile water, after which samples were taken to characterise the cropped yeast.
Table 1

Properties of the five different experimental conditions

Test condition

Sparging gas

Oxygen concentration in the wort (ppm)

Pitching yeast

Pitching rate (cells per milliliter)





80 × 106





80 × 106





80 × 106





80 × 106





20 × 106

Fermentation analysis

Before centrifugation, the number and the viability of suspended yeast cells was counted by flow cytometry (YeastCyte, BioDETECT AS, Oslo, Norway). Dead cells were stained in-line with 1% (v/v) propidium iodide in phosphate-buffered saline (70355, Sigma).

The specific gravity of the fermenting medium was measured with a handheld density meter (DMA 35N, Anton Paar, Graz, Austria), but the final extract and alcohol content were measured with the DMA 4500 density analyzer and Alcolyser Plus (Anton Paar, Graz, Austria).

FAN was determined by a ninhydrin-based method, according to the standard method as defined by the European Brewery Convention (EBC-Analytica 1998).

Volatile compound concentrations were determined by headspace gas chromatography. The vials were analysed with a calibrated Autosystem XL gas chromatograph with a headspace autosampler (HS40; Perkin Elmer, Wellesley, MA, USA), equipped with a Chrompack-Wax 52 CB column (length 50 m; 0.32-mm internal diameter; layer thickness 1.2 μm; Varian, Palo Alto, CA, USA) and a flame ionization detector (FID) and an electron capture detector. Analyses were carried out in duplicate and the results were analysed with Perkin Elmer Turbochrom Navigator software and were recalculated to 5% (v/v) ethanol.

Yeast physiology

Fatty acid analysis

Total fatty acids [palmitic acid (C16:0), palmitoleic acid (C16:1), stearic acid (C18:0), oleic acid (C18:1), and linoleic acid (C18:2)] of frozen yeast pellets were extracted following the procedure of Moonjai et al. (2002). Gas chromatography was done with a calibrated Varian 300 analyzer (Varian, Palo Alto, Ca, USA) equipped with a 30-m length, 0.32-mm internal diameter, 0.25-μm film thickness, Alltech Heliflex AT-225 capillary column (Alltech Associated, Inc., Deerfield, IL, USA) and a FID (Verbelen et al. 2009a). Fatty acid concentrations were calculated as mg/g cell dry weight (CDW). The fatty acid analysis of each fermentation sample was done in duplicate.

Glycogen and trehalose analysis

The quantification of glycogen and trehalose was performed simultaneously following the method of Neves et al. (1991) with some changes. Yeast samples were washed three times with cold distilled water. Afterwards, the yeast suspension was filtered over a membrane filter (pore size 0.45 μm) and 25–50 mg (wet weight) of cell pellet was weighed and immediately stored at −80°C. The extraction of glycogen and trehalose, the enzymatic breakdown with α-amyloglucosidase (500 units/mg; Roche Diagnostics, Indianapolis, IN, USA) and trehalase (2,100 units/ml; E-TREH, Megazyme, Wicklow, Ireland), and measurement of final glucose concentrations were performed as described elsewhere (Verbelen et al. 2009a). The results are expressed as percent of wet weight (ww). Glycogen and trehalose analysis of each sample was done in triplicate.

Quantitative PCR

The expression levels of specific genes (Table 2) were determined using quantitative polymerase chain reaction (qPCR). Samples were collected from the tall tubes after 1 and 4.5 h and at an ADF of 35% and 50% (exponential phase), cooled on ice, and centrifuged at 3,000 rpm for 3 min. The pellets were washed two times with cold sterile RNase-free water and 150 × 106 pelleted cells were then frozen at −80°C. RNA extraction, reverse transcription, and making the reaction mix was performed as described before (Verbelen et al. 2009a). The PCR program used on the ABI Prism 7500 instrument (Applied Biosystems) consisted of an initial denaturation of 10 min at 95°C, amplification by 40 cycles of 15 s at 95°C, and 1 min at 58°C. The primers were designed to anneal close to the 3’-end of each gene. The PCR primers were all designed with the Primer Express software (Applied Biosystems, Cheshire, UK) according to the Applied Biosystems guidelines. Primer sequences used for qPCR analysis are given in Table 2. The specificity of the primers was tested using conventional PCR and the melting curves of the amplified product. The gene for 18S rRNA (RDN18) was used as the reference gene. The expression levels were determined using the ABI Prism 7500 System Gene Quantification Software (Applied Biosystems) through quantification of Sybr Green fluorescence. A standard curve of each gene was constructed with genomic DNA of CMBSPV09. The expression levels of the different genes were normalised with respect to 18S expression levels and are means of two independent fermentation samples, each consisting of three replicates of the PCR reaction.
Table 2

Genes, their products, and primer sequences of those used for qPCR analysis (from 5′ to 3′)



Forward primer sequence

Reverse primer sequence


18S ribosomal RNA




12-kDa heat shock protein




Cytosolic catalase T




Cu/Zn superoxide dismutase




Squalene epoxidase




Δ9 fatty acid desaturase




Alcohol acetyltransferase




Branched-chain amino acid permease




Acetohydroxy acid synthase




Acetohydroxy acid reductoisomerase




The impact of oxygen on yeast metabolism in HCD fermentations was evaluated, using a normal-fermenting moderately flocculating industrial lager brewery yeast strain (CMBSPV09) selected from a previous study (Verbelen et al. 2008).

Fermentation characteristics

The progress of the fermentations with different pitching rates is depicted in Fig. 1. The time required to reach an ADF of 80% was approximately 270 h for the REF with an inoculum size of 20 × 106 viable cells/ml and between 50 and 70 h for the HCD fermentations. The A/NP condition had the longest HCD fermentation time of approximately 70 h, followed by the conditions with preoxygenated yeast (N/PR 58 h; A/PR 52 h). The fastest fermentation rate was observed in the case of the O/NP condition (46 h). In general, the higher pitching rate had a larger impact on the fermentation rate than the oxygen conditions.
Fig. 1

Decrease of sugar density (n = 2; expressed as °P) as a function of time (filled circles = non-preoxygenated yeast, aerated wort (A/NP); empty circles = non-preoxygenated yeast, oxygenated wort (O/NP); filled triangles = preoxygenated yeast, deaerated wort (N/PR); empty triangles = preoxygenated yeast, aerated wort (A/PR); filled squares = reference (REF)). In the box, the decrease of density of the different HCD fermentations is enlarged to emphasize the subtle differences between them

In Fig. 2a, the net growth pattern (maximal cell count minus the initial inoculum size) of the different conditions are shown. The lowest net growth was obtained in the case of the A/NP condition, which was 28% lower than N/PR, 40% lower than REF, 45% lower than A/PR, and 60% lower than the O/NP condition. Hence, the oxygen conditions had a significant impact on the extent of yeast growth in the HCD fermentations. These results were confirmed when the total biomass (which was the sum of the yeast in suspension and the flocculated yeast formed during fermentation) was measured at the end of fermentation (results not shown).
Fig. 2

a Net growth (maximal cell count minus the initial pitching rate) patterns (in 106 cells per milliliter; n = 2), b cell viability (in percent living cells; n = 2) of the total yeast populations after fermentation, and c FAN uptake (in parts per million; n = 2) of the different fermentations

The viability of the yeast population was monitored during fermentation (results not shown) and remained similar for the different fermentations. When the total biomass was characterised at the end of the fermentation, the cell viability was also measured (Fig. 2b). The HCD fermentations showed similar viabilities at the end of fermentation, but the REF condition showed a slightly decreased percentage of living cells. This indicates that the long duration of fermentation had a negative influence on the yeast viability but that yeast viability remained at an acceptable level in all fermentations.

The total FAN uptake levels were 68% enhanced when higher than normal pitching rates were used (Fig. 2c). The highest FAN uptake was found in the O/NP fermentation condition and was 15% higher than the other HCD fermentations, which were not significantly different.

Table 3 shows the results of the beer analysis for pH and aroma compounds in the fermentations performed. Despite the large differences in net growth, the observed yields of ethanol were identical \(\left( {{{Y_{{{\text{p}} \mathord{\left/ {\vphantom {{\text{p}} {\text{s}}}} \right. \kern-\nulldelimiterspace} {\text{s}}}} = {\text{0}}{\text{.565}} \pm {\text{0}}{\text{.008\% }}\left( {{v \mathord{\left/ {\vphantom {v v}} \right. \kern-\nulldelimiterspace} v}} \right)} \mathord{\left/ {\vphantom {{Y_{{{\text{p}} \mathord{\left/ {\vphantom {{\text{p}} {\text{s}}}} \right. \kern-\nulldelimiterspace} {\text{s}}}} = {\text{0}}{\text{.565}} \pm {\text{0}}{\text{.008\% }}\left( {{v \mathord{\left/ {\vphantom {v v}} \right. \kern-\nulldelimiterspace} v}} \right)} {{\text{ $ ^\circ $ P}}}}} \right. \kern-\nulldelimiterspace} {{\text{ $ ^\circ $ P}}}}} \right)\). The pH of the beer was inversely correlated with the extent of net growth and FAN uptake in the HCD fermentations. The highest final pH was found in the REF fermentation condition.
Table 3

Final alcohol, pH, and aroma concentrations in the different fermentations performed


Oxygen conditions






Alcohol (% v/v)

6.62 ± 0.02

6.72 ± 0.06

6.46 ± 0.03

6.72 ± 0.01

6.73 ± 0.01


4.28 ± 0.01

4.16 ± 0.01

4.25 ± 0.01

4.22 ± 0.01

4.51 ± 0.03

Flavour profile

 Acetaldehyde (ppm)

4.96 ± 0.02

2.57 ± 0.24

7.00 ± 0.80

4.25 ± 0.19

7.48 ± 1.46

 DMS (ppb)

15.0 ± 0.4

11.7 ± 0.5

11.4 ± 0.7

12.6 ± 0.3

16.8 ± 0.4

 Higher alcohols (ppm)


11.4 ± 0.2

13.0 ± 0.2

12.1 ± 0.2

11.8 ± 1.0

11.1 ± 0.1


8.59 ± 0.12

8.64 ± 0.22

8.27 ± 0.03

7.80 ± 0.67

6.66 ± 0.10

  Isoamyl alcohol

54.9 ± 0.9

58.9 ± 0.3

57.9 ± 1.0

56.6 ± 2.5

44.5 ± 0.6

  Total higher alcohols

74.9 ± 1.2

80.6 ± 0.7

78.3 ± 1.2

76.2 ± 4.3

62.3 ± 0.8

 Esters (ppm)

  Ethyl acetate

24.9 ± 0.5

19.2 ± 0.5

22.8 ± 0.1

21.6 ± 0.6

27.9 ± 0.5

  Isoamyl acetate

1.30 ± 0.03

1.12 ± 0.06

1.44 ± 0.02

1.38 ± 0.04

1.19 ± 0.06

  Ethyl caproate

0.17 ± 0.01

0.12 ± 0.01

0.18 ± 0.01

0.15 ± 0.00

0.16 ± 0.01

  Total esters

26.4 ± 0.6

20.5 ± 0.6

24.4 ± 0.1

23.2 ± 0.6

29.3 ± 0.6

 Diacetyl (ppb)

528 ± 2

334 ± 7

626 ± 6

538 ± 10

60.9 ± 16.8

Values of all the flavour compounds have been recalculated based on the normal ethanol percentage in lager beers (=5% v/v)

aThese different conditions are described in Table 1.

With respect to the flavour compounds (Table 3), the concentrations of higher alcohols are positively influenced by the pitching rate (Edelen et al. 1996; Verbelen et al. 2008). However, no large differences were found between the different oxygen conditions applied to the HCD fermentations.

The lowest concentration of the ethyl acetate ester, which has a solvent-like flavour and a threshold of 30 ppm, was found in the O/NP condition, which was 31% lower than the highest concentrations found in the reference fermentation. Maximum concentrations of isoamyl acetate (banana-like flavour, threshold 1.2 ppm) were found in N/PR and A/PR. Less isoamyl acetate (30%) was found in O/NP, compared with the highest level found (N/PR). As can be noticed, isoamyl acetate was inversely correlated with the oxygen availability per cell. Concerning ethyl caproate (apple-like flavour, threshold 0.2 ppm), the lowest concentration was also measured in the O/NP condition. In contrast with ethyl acetate, isoamyl acetate and ethyl caproate were not negatively affected by the applied high pitching rate.

Slightly higher DMS concentrations were observed in the REF condition, which is in correspondence with earlier results (Verbelen et al. 2008). The acetaldehyde concentrations at the end of fermentation showed no direct trend with the applied conditions.

The total diacetyl (diacetyl + α-acetolactate) levels in HCD fermentations were drastically increased compared to the reference fermentation, which was also observed in earlier studies (Okabe et al. 1992; Verbelen et al. 2009a). Diacetyl can cause a buttery off-flavour above its threshold of 80 ppb. Surprisingly, the total diacetyl level of the O/NP condition was significantly lower during (results not shown) and after fermentation than the other HCD fermentations. It was expected that the higher uptake levels of FAN in this fermentation condition (Fig. 2c) should enhance the formation of α-acetolactate, thereby resulting in higher levels of total diacetyl.

Yeast physiological parameters

Fatty acid profiles

In Fig. 3, the fatty acid profiles of the yeast populations are shown for the five conditions in the course of the first part of the fermentation. Since the wort was centrifuged, the levels of exogenous lipids were reduced compared with a normal wort. Hence, the increase in UFA content of the yeast in the beginning of the fermentation was due to biosynthesis and not to assimilation. The increase of the unsaturated fatty acid C16:1 and C18:1 content was strongly influenced by the pitching rate and the oxygen conditions. In case of the A/NP condition, the sum of the concentrations of C16:1 and C18:1 increased from 5.03 mg/g CDW to 8.33 mg/g CDW in 4.5 h. With respect to the O/NP fermentation, the UFA content increased from 6.33 to 13.36 mg/g CDW in 4.5 h. During preoxygenation, the levels of UFA increased by 53% (results not shown). Without wort aeration (N/PR), no build-up of C16:1 and C18:1 was observed during fermentation; the level of these UFAs was at its maximum (8.06 mg/g CDW) in the beginning of fermentation. With wort aeration (A/PR), a further build-up of UFA during the aerobic phase of the fermentation was observed (from 8.16 mg/g CDW to 9.49 mg/g CDW in the first 2.5 h). Although the same oxygen levels were supplemented to the wort, the reference fermentation was characterised by a much higher formation of C16:1 and C18:1 (from 5.68 mg/g CDW to 14.39 mg/g CDW in 2.5 h) in comparison with A/NP and A/PR. In the following anaerobic phase of the fermentation, the newly synthesised membrane lipids were distributed over mother and daughter cells until the UFA content dropped to a growth limiting concentration (Casey et al. 1984). Indeed, a higher maximum level of UFA in the HCD fermentations was consistent with a higher yeast net growth (Figs. 2a and 3). The final UFA levels towards the end of fermentation in case of O/NP were significantly higher than in the other conditions (Fig. 3b), suggesting that in this condition other nutrients (e.g., fermentable carbohydrates) were limiting at that point.
Fig. 3

Fatty acid profiles (in mg/g cell dry weight (CDW)) as a function of time (n = 2). a Non-preoxygenated yeast, aerated wort (A/NP); b Non-preoxygenated yeast, oxygenated wort (O/NP); c Preoxygenated yeast, deaerated wort (N/PR); d Preoxygenated yeast, aerated wort (A/PR); e Reference (REF). Fatty acids: filled circles = C16:0, empty circles = C16:1, filled triangles = C18:0, empty triangles = C18:1, and filled squares = C18:2

Glycogen and trehalose content

The general trend of glycogen during brewery fermentation consists of a first intense initial period of breakdown, followed by accumulation in the exponential phase (Boulton 2000). The REF fermentation followed this trend (Fig. 4a). When the pitching rate increased, this typical profile was more leveled out: the initial breakdown after 4.5-h fermentation was less intense, compared with the reference fermentation (Fig. 4a, white bars) and no significant differences were found between the HCD fermentations with different oxygen conditions. After fermentation, the glycogen content was measured again (Fig. 4a, grey bars). As can be seen, the lowest glycogen levels were found in case of the REF fermentation and this level was 21% lower than O/NP, 38% lower than N/PR, 41% lower than A/PR, and 45% lower than A/NP. Despite a similar glycogen level reached during exponential phase (results not shown), the low final concentrations in the reference fermentation were caused by a prolonged stationary phase.
Fig. 4

Glycogen (a) and trehalose contents (b) (in % of wet weight) of the yeast suspensions 4.5 h after pitching (white bars) and of the total yeast populations after the different conducted fermentations (grey bars; n = 3)

Initially (4.5 h after pitching), trehalose levels were minimal and no significant differences were found between the various fermentations (Fig. 4b, white bars). Afterwards, the disaccharide accumulated towards the end of fermentation, which corresponds with the literature (Majara et al. 1996). In Fig. 4b (grey bars), the trehalose level of the total yeast population after fermentation is depicted. A/NP reached the highest trehalose level (1.84% ww). N/PR and A/PR had similar final trehalose values (average 1.55% ww) and were significantly higher (24%) than O/NP and REF (average 1.25% ww).

Gene expression analysis

At four time points in the fermentation (1 h, 4.5 h, and time points corresponding to an ADF of 35% and 50%), yeast samples were taken and expression of HSP12, CTT1, SOD1, OLE1, ERG1, ATF1, BAP2, ILV2, and ILV5 was monitored by qPCR (Fig. 5).
Fig. 5

Expression profiles (n = 4) of the genes HSP12 (a), CTT1 (b), SOD1 (c), OLE1 (d), ERG1 (e), ATF1 (f), and BAP2 (g) and the ratio of ILV2 and ILV5 expression at four fermentation points: (1) 1 h after pitching, (2) 4.5 h after pitching, (3) when 35% ADF, and (4) 50% ADF (exponential phase) was reached. (White bars) Non-preoxygenated yeast, aerated wort (A/NP); (light grey bars) non-preoxygenated yeast, oxygenated wort (O/NP); (dark grey bars) preoxygenated yeast, deaerated wort (N/PR); (black bars) preoxygenated yeast, aerated wort (A/PR); (striped bars) reference (REF). The expression levels are expressed relative to that of the housekeeping gene RDN18

As can be seen from Fig. 5a, the expression level of HSP12 was low in the initial lag phase but increased during exponential phase. Higher expression levels of HSP12 were observed in the yeast cells from the HCD conditions, although no significant differences in HSP12 expression are observed between the different applied oxygen conditions in the HCD fermentations. HSP12 has been evaluated for the use as a molecular marker for stress resistance (Ivorra et al. 1999).

In Fig. 5b, CTT1 expression is depicted. CTT1 encodes the cytosolic catalase T, which catalyses the breakdown of hydrogen peroxide to oxygen and water. CTT1 is associated with the oxidative stress response, through the transcription factors Yap1 and Skn7, and the general stress response (STRE dependent; Estruch 2000). In general, the CTT1 expression peaked in the lag phase. The highest expression levels were found in case of the O/NP condition 1 h after pitching, followed by the REF fermentations 4.5 h after pitching. The lowest expression level in the lag phase was found in case of N/PR. The A/NP fermentation had a similar peak value.

Superoxide dismutase, encoded by SOD1 and SOD2, catalyses the conversion of the superoxide anion to oxygen and hydrogen peroxide (Van Loon et al. 1986) and are both associated with the oxidative stress response (Herrero et al. 2008). In the O/NP condition, SOD1 gene expression increased drastically in the first hour after pitching (Fig. 5c). The REF fermentation possessed the second highest peak value after 4.5 h. This suggests that the response of oxygen in case of the reference fermentation came a bit later than in the HCD fermentations. In case of the A/NP, A/PR, and N/PR fermentations, the expression of SOD1 remained low.

The expression of OLE1, which encodes the oxygen dependent fatty acid desaturase, was very high throughout the whole experiment (Fig. 5d), although maximum levels were found after 4.5 h. The highest peak was observed in N/PR, which is in accordance with the fact that OLE1 is known as a hypoxic gene (Zitomer et al. 1997; Kwast et al. 1998). The second highest level of expression was found in the REF condition, which was similar to the A/NP and A/PR fermentations. A low relative level of expression was found in case of the O/NP fermentation condition. No direct correlation could therefore be found between the formation of UFA and expression of OLE1 (Fig. 3 vs. 5c).

ERG1 (squalene epoxidase) catalyses the oxygen-dependent epoxidation of squalene to 2,3-oxidosqualene and plays an essential role in the ergosterol biosynthesis pathway. Surprisingly, expression of ERG1, which is induced with increased oxygen availability and upon sterol limitation (Jahnke and Klein 1983; M’Baya and Karst 1987), was drastically increased after a few hours after pitching in the REF condition (Fig. 5e). However, after 4.5 h, the medium was already depleted of oxygen in the REF fermentation (results not shown). On the other hand, only small differences were observed in the different HCD fermentations. Therefore, no direct connection existed between ERG1 and oxygen availability in this study.

ATF1 encodes the alcohol acetyl transferase, an important enzyme in the synthesis of ethyl acetate and isoamyl acetate. ATF1 transcription is a limiting factor in the synthesis of these esters (Verstrepen et al. 2003b; Saerens et al. 2008). In this study, the expression profile of this gene was similar with that of OLE1 (Fig. 5d vs. f). However, the highest peak value was found in the REF fermentation 4.5 h after pitching and not in N/PR, as was the case with OLE1. The lowest expression peak level was found in the O/NP condition. The ethyl acetate and isoamyl acetate concentrations were indeed lower in this fermentation condition. The high ATF1 expression in the REF fermentation was accompanied with significant higher concentrations ethyl acetate but not isoamyl acetate, confirming that other factors (e.g., substrate availability) than ATF1 expression also play an important role in the synthesis of esters (Verstrepen et al. 2003a). In the exponential phase, the expression levels of ATF1 decreased to low levels (Fig. 5f), which confirms the outcome of other studies (Shen et al. 2003; Rautio et al. 2007; Saerens et al. 2008).

In addition, the expression of BAP2 was monitored because the branched chain amino acid permease plays an important role in the assimilation of important amino acids (Grauslund et al. 1995) and in the production of higher alcohols (Kodama et al. 2001). Surprisingly, BAP2 was only highly expressed at 1 h after pitching (Fig. 5g). The expression levels in the yeast cells of the REF fermentation and of the O/NP fermentation were remarkably higher than in the other fermentation conditions. The higher expression of BAP2 in O/NP was accompanied by a higher FAN uptake from the medium (Fig. 2c) but did not result in a significant higher total production of isobutanol and isoamyl alcohol (Table 3).

To investigate the role of ILV2 (encoding α-acetohydroxy acid synthase) and ILV5 (encoding α-acetohydroxy acid reductoisomerase) in the metabolism of diacetyl, their expression levels were monitored (Fig. 5h). α-Acetolactate is an overflow product between the reactions mediated by Ilv2 and Ilv5 (Villanueba et al. 1990). Therefore, the ratio ILV2/ILV5 was calculated from the expression data and is depicted in Fig. 5h. This ratio was high in the beginning of fermentation and decreased to a minimum. However, no significant differences were found between the various fermentation conditions. Therefore, no correlation was found between the observed differences in the amount of total diacetyl (Table 3) and the ILV2/ILV5 ratio (Fig. 5h).


Higher initial cell concentrations increase the fermentation speed, which creates a big economical advantage for the breweries. However, to ensure the physiological stability of the yeast populations, an adequate yeast growth is needed, which could be accomplished by optimisation of the oxygen conditions. Therefore, the purpose of this study was to determine the impact of HCD fermentations with different oxygen conditions on yeast fermentation performance and beer quality.

Impact of HCD fermentations with different oxygen conditions on fermentation performance and beer characteristics

Adequate yeast growth is of critical importance for the quality of the resulting beer and the stability of the yeast slurry that often is reused in subsequent fermentations. It can be assumed that low growth rates will eventually lead to older yeast populations, compared with normal pitched wort, in which the yeast population increases two to three times. Aged yeast cells are often associated with decreased fermentation capacity, changed morphology, altered flocculation patterns, and extended generation times (Barker and Smart 1996; Powell et al. 2003). Improving the oxygen conditions can result in a significant enhancement of yeast growth during HCD fermentations for the production of beer. From this study, it can be concluded that higher oxygen levels are needed in the wort, pitched with high yeast concentrations, to obtain a similar net growth compared with a normal pitched fermentation. This suggests that oxygen is a critical growth limiting factor in HCD fermentations. To circumvent the problems associated with wort aeration, the use of yeast preoxygenation was also investigated in this study. During this process, the yeast population is treated with oxygen in a low-nutrient environment, stimulating the build-up of essential membrane lipids. Therefore, the need of wort oxygenation or aeration is abolished (Boulton et al. 2000). However, our results show that yeast preoxygenation alone cannot stimulate the same growth rate in HCD fermentations compared with the reference fermentation. On the other hand, yeast preoxygenation together with wort aeration prior to HCD fermentation gives satisfying results. The fact that high oxygen levels in HCD fermentation do not lead to similar relative yeast growth (ratio of maximum cell count and applied pitching rate) compared with the reference conditions suggests that still other growth limiting factors exist in HCD fermentations.

Acetate ester levels have previously been reported to be negatively influenced by higher pitching rates (Suihko et al. 1993; Edelen et al. 1996; Verbelen et al. 2008) and by higher levels of oxygen in the wort (Verstrepen et al. 2003a). In this study, it is confirmed that the production of higher alcohols is stimulated and the ester ethyl acetate is repressed by higher pitching rates (Verbelen et al. 2008). Moreover, adding high levels of oxygen to the wort has a negative impact on the synthesis of ethyl acetate, isoamyl acetate, and ethyl caproate, which make up the consumer-appreciated fruitiness of a lager beer. From that point of view, preoxygenation seems to have a benefit against the other conditions, for having the highest isoamyl acetate levels, which has the highest flavour impact (ratio of concentration and threshold).

The total diacetyl content is drastically higher in the HCD fermentations, but wort oxygenation seems to have a reductive effect on total diacetyl. As the chemical decarboxylation is the rate-limiting step in the metabolism of diacetyl (Yamauchi et al. 1995), the higher residual amounts in many accelerated fermentation systems can be explained by the short residence times (Okabe et al. 1992; Willaert and Nedovic 2006; Verbelen et al. 2008). Therefore, efficient strategies are needed to decrease the diacetyl content in those beers (Willaert 2007).

Impact of HCD fermentations with different oxygen conditions on yeast physiology

Varying the oxygen conditions of HCD fermentations has a distinct effect on the resulting glycogen and trehalose concentrations. Adding high amounts of oxygen to wort, prior to HCD fermentations, decreases glycogen and to a lesser extent also trehalose in the resulting yeast slurry. Glycogen is an important physiological parameter because it fuels the lipid synthesis in the early events of fermentation (Quain 1988). However, in the HCD fermentation with wort oxygenation, similar peak levels of UFA are produced, compared with the reference fermentation (Fig. 3b, e), while a higher breakdown of glycogen occurs in the aerobic phase of the reference fermentation. As exogenous carbohydrates are already being taken up in the early hours of HCD fermentations, a reason for the higher breakdown of glycogen in the reference fermentation could be that both exogenous and endogenous carbon sources are used for the build-up of essential lipids in HCD fermentation. It can be hypothesised that HCD fermentations can cope better with their glycogen pool than a normal fermentation and that the lower build-up of glycogen seen in the exponential phase of HCD fermentations with wort oxygenation is caused by the high growth rate, compared with the other HCD fermentations.

The response to stressful conditions during fermentation in yeast depends on the activation of signal transduction pathways, which result in transcriptional change and synthesis of protective molecules. The expression levels of the stress gene HSP12 and the accumulation of trehalose in this study confirm the earlier reports by Verbelen et al. (2009a) that the HCD yeast populations trigger a stress response. However, between the different HCD fermentations, no significant differences were found in HSP12 expression, suggesting that this gene is not triggered by high oxygen levels. The observed accumulation of the stress protectant trehalose during the fermentations is most likely a response to nutrient limitation and/or ethanol toxicity (Boulton 2000). The trehalose level was only different in the case of the HCD fermentation condition with wort oxygenation. Higher trehalose levels are believed to ensure yeast viability during the starvation period after fermentation is completed and also during the fermentation itself, which may lead to a reduction in fermentation time (Guldfeldt and Arneborg 1998). As a consequence, lower trehalose concentrations in the cropped slurry, could make the yeast vulnerable to stresses associated with a new fermentation.

Monitoring the expression of oxidative stress related genes SOD1 and CTT1 shows that SOD1 is predominantly expressed in the early phase of the fermentations. Both SOD1 and CTT1 are highly expressed when high oxygen concentrations are added to the wort, which suggests that the yeast populations respond to the high oxygen levels. Other studies also report an induction of genes involved in oxidative stress response during the period of aeration (Higgins et al. 2003; Pérez-Torrado et al. 2009). Higher expression levels of SOD1 and CTT1 do not necessarily imply that yeast cells are exposed to oxidative stress. It is the balance between ROS production and cell defenses that determines the degree of oxidative stress (Jamieson 1998). In this study, high expression levels of SOD1 and CTT1 do not result in lower viabilities in the HCD yeast populations with oxygenated wort and therefore we conclude that the balance is in favor of the defense mechanisms. However, ROS have been described as an important factor in aging of yeast cells (Powell et al. 2000), influencing the life span of yeast cells (Van Zandycke et al. 2002), and prolonged exposure to these ROS could eventually result in an inability to prevent cellular damage and cell death due to an increase in oxidant production, a decline in antioxidant defense mechanisms, or a decline in the efficiency of repair mechanisms (Sohal and Weindruch 1996). Serial repitching with excessive oxygen levels could have a negative influence on the yeast and eventually lead to an accelerated loss of viability in comparison with normal wort. Therefore, optimum oxygen levels for high cell density fermentations are therefore necessary, through the application of preoxygenation or moderate concentrations of oxygen in the wort. In a recent study, performed by Verbelen et al. (2009b), expression of stress responsive genes was monitored during preoxygenation. These authors suggest that yeast cells acquire a stress response during preoxygenation, making them more resistant against the stressful conditions during the subsequent fermentation. The observation that expression levels of CTT1 and SOD1 and build-up of UFA in the reference fermentation are higher compared with the HCD fermentations with aerated wort indicates that oxygen availability to the yeast cell is better in these fermentations and that oxygen addition to yeast cells in HCD fermentations has to be improved.

Besides UFA, sterols are also formed in the beginning of fermentation, when oxygen is available, through the action of Erg1. We found only high expression of ERG1 in the reference fermentation, when the medium is already depleted of oxygen, suggesting that ERG1 expression is not a determining factor in build-up of sterols.

OLE1 and ATF1 gene expression shows similar profiles and similar oxygen dependence. This is in accordance with the literature (Fuji et al. 1997; Fujiwara et al. 1998), in which ATF1 transcription is found to be co-regulated by the same mechanism as the OLE1 gene and are both repressed by oxygen and unsaturated fatty acids, though by a different regulation pathway. High levels of expression are found in case of preoxygenated yeast in anaerobic wort and low levels in case of oxygenated wort, in accordance with the fact that these are co-regulated hypoxic genes (Zitomer et al. 1997; Fujiwara et al. 1998). From the results of this study, it also became clear that OLE1 gene expression is not correlated with the amount of UFA formed. Several reasons could be mentioned. First of all, OLE1 gene expression was very high during the whole experiment even in the case of oxygenated wort. Therefore, transcript levels of OLE1 are not the rate limiting step in the formation of UFA, but the need for molecular oxygen in the desaturation reaction could be the limiting factor. Kajiwara et al. (2000) observed only an increase of 7% of C18:1 when using an OLE1-overexpressing strain. Secondly, OLE1 gene expression has been shown to be dependent on several regulatory sequences, such as STRE, O2R, fatty acid response element, and low oxygen response element (Nakagawa et al. 2001; Vasconcelles et al. 2001; Martin et al. 2006). The expression of OLE1 is therefore difficult to interpret. In contrast, ATF1 transcription levels correlated well with the resulting concentrations of ethyl acetate, but other factors, such as growth and thus acyl-CoA availability, probably played an important role as well.

High expression levels of BAP2 in the reference fermentation and in the HCD fermentations with wort oxygenation suggest that uptake of branched chain amino acids was enhanced in these conditions. However, we found no correlation with the total uptake of FAN in the reference fermentation (probably because the yeast concentration was lower) nor in a higher production of higher alcohols in both conditions.

In the initial phase of the fermentations, a relative change in the expression of ILV2 and ILV5 occurred that might promote accumulation of α-acetolactate, the precursor for diacetyl. However, no differences in ILV2/ILV5 expression ratio are observed between the different fermentations and thus no correlation with the final total diacetyl content exists. It can be suggested that other factors, such as yeast growth, enzymatic activity, and substrate composition, which are known to play a role in the diacetyl metabolism, determine the extent of total diacetyl in this experiment.

Taken together, this work provides evidence that optimisation of the oxygen conditions prior to HCD fermentation can efficiently increase the performance and stability of these fermentations. Both wort oxygenation and yeast preoxygenation with wort aeration give a high net growth, ensuring the sustainability of HCD fermentations. From a yeast physiological (higher glycogen and trehalose content, less response to oxidative stress) and beer quality (less repression of ester synthesis) point of view, yeast preoxygenation in combination with aeration seems to be more advantageous, provided that efficient strategies are adopted to decrease the diacetyl content. However, more experimental work is needed to prove the physiological stability of HCD populations in subsequent generations and the technological feasibility at commercial scale.


The authors wish to thank Stefan Van Loy for the excellent help with the experiments. Financial support from the Institute for the Promotion of Innovation through Science and Technology in Flanders (IWT-Vlaanderen) is acknowledged.

Copyright information

© Springer-Verlag 2009