To test the impact of airtight storage of cereal grain on ethanol production from the grain, storage was simulated on a laboratory scale. Moist wheat (30% water content) was stored in airtight test tubes with a simulated air leakage [4]. A sub-set of these test tubes was inoculated with the biocontrol yeast, P. anomala J121. The other tubes were not inoculated to estimate an eventual impact of the biocontrol yeast on the subsequent enzymatic pre-treatment and fermentation. After 4 weeks incubation, 50% of the tubes without yeast inoculation showed substantial mould growth. These tubes were not included in further analyses. No mould growth was observed in tubes inoculated with P. anomala. These results showed that it was necessary to include the biocontrol yeast for grain conservation. Farm grain with 18% water content was used as a control.

The material was stored for 4 weeks. After this time, the grain was ground and pre-treated, as described in the Methods section.

The resulting material was used as a substrate in test fermentations in shake flask cultures using S. cerevisiae as the fermentation yeast. Samples were taken and the sugar and ethanol concentrations were monitored. Fermentations were run until a plateau concentration of ethanol had been reached, which usually occurred within 30 hours of the start of the fermentation. After 30 hours, the yields of gram ethanol per gram added wheat grains were determined using the maximum ethanol concentration determined for the according fermentation for the calculations. The ethanol yield from dry grains was 0.42 g ethanol/g grain (± 0.01 standard deviation (SD), N = 6). Unexpectedly, the yield from stored moist wheat grains was significantly enhanced (P < 0.001), to 0.48 g ethanol/g added grain in samples without biocontrol yeast (± 0.01 SD, N = 3), and 0.47 g ethanol/g added grain in samples with added biocontrol yeast (± 0.01 SD, N = 5). The difference between yields obtained from the inoculated and non-inoculated grain was not significant. These results imply that the biocontrol yeast does not negatively influence the ethanol yield, and that moist storage of the grain increases ethanol yield by more than 10%.

To test starch degradation in the samples after the pre-treatment, mash samples were incubated with an I₂-KI solution and the colour development was monitored. Mash samples obtained from dry grain were visibly darker than those from moist grain (Figure 1). This indicates a higher accessibility of the starch in moist grain to the enzymatic pre-treatment compared with the dry grain. Obviously a substantial amount of starch had not been degraded during the pre-treatments of the farm grain.

The more efficient starch degradation in the stored moist grain was also illustrated by the higher initial glucose concentrations in the test fermentation broth. The initial glucose concentrations in fermentations of stored moist grain, with or without biocontrol yeast, were around 80 g/L, compared with about 60 g/L in the fermentation medium obtained from farm grain (Figure 2). In the fermentations of the mash from farm grain, the sugar concentration increased during the first hours of cultivation, indicating a degradation of residual starch by glucoamylase, which was obviously still active in the fermentation. This increase in glucose concentration also indicated that there was starch left after the pre-treatment of stored dry grain. In fermentations of moist grains, with or without added biocontrol yeast, no increase or only a small initial increase in the glucose concentration was observed (Figure 2). The concentrations of free glucose dropped to 0 g/L between 12 and 15 hours of test fermentation in mashes from stored farm grain, and between 15 and 18 hours in those from stored moist grain. Some release of glucose from residual starch can be assumed even after the measured apparent glucose concentrations dropped to zero, as the plateaux in the ethanol concentration were reached somewhat later (Figure 2).

Due to the high proportion of non-soluble particles in the substrate suspension, it was not possible to determine yeast biomass by gravimetric methods or optical density measurements. Therefore, yeast growth was monitored by plating dilutions of the fermentation broth on selective YPD medium. As shown in Figure 2, the number of yeast colony forming units (CFU) increased during fermentation, indicating that sugar consumption and ethanol production were connected to yeast growth. Cell numbers reached their maximum value some time before the plateau in the ethanol concentration was reached. In contrast to ethanol formation, no significant difference was found in the numbers of CFU in the different fermentations.

We also tested whether the use of biocontrol yeast had an impact on the composition of the yeast population in the test fermentations. DNA was isolated from 20 randomly selected colonies and a polymerase chain reaction (PCR)-fingerprint was generated and compared with the fingerprint of P. anomala J121 and that of the inoculated fermentation yeast (S. cerevisiae). The fingerprints of all tested isolates were identical to that of S. cerevisiae (results not shown). P. anomala was most probably already inactivated during the pre-treatment of the substrate. The substrate was not handled sterilely throughout the experiments and contaminating bacteria might thus influence the fermentations, so we also tested bacterial growth during the fermentation. Bacteria were present in the fermentation, but in low numbers compared with the yeast population (10⁴-10⁵ bacterial CFU/ml of the fermentation liquid, that is, 10⁵-10⁶/g (dry weight) grain) (Figure 2). Moreover, their numbers did not increase during the fermentation, indicating that these bacteria were not actively growing. Probably the bacterial CFU were due to spores present on the grain and survived the pre-treatment. These results indicate that bacteria from the grain did not influence the ethanol production, independently of whether the grain was stored moist or dry.

We also tested whether a treatment with a mixture of cellulases, hemicellulases and pectinases (CHP) could improve the ethanol yield. However, this treatment did not improve the ethanol yield, neither for dry nor moist stored grain. The high-performance liquid chromatography (HPLC) profiles of the fermentation broths from these treatments were almost identical to those obtained from material treated with Stargen 001 only. However, for the CHP-treated samples we found a peak in the HPLC chromatogram that did not disappear during the course of the fermentation. In contrast, in the samples without the additional enzyme treatment, a peak at the same retention time vanished towards the end of the fermentation. According to the standard curve, this peak represented maltose, which can be fermented to ethanol by several S. cerevisiae strains [13].

We further investigated the fermentation medium using a high-performance anion exchange chromatography coupled with pulsed amperometric detection (HPAE-PAD) analysis with a gradient method. This analysis showed that the peak was generated by both maltose and cellobiose. HPAE-PAD analysis showed that maltose and cellobiose were also present after fermentation in samples that were only treated with starch-degrading enzymes, however, the concentrations were below the lowest concentration of the standard (10 μM). In samples additionally treated with CHP, the amounts of maltose and cellobiose were higher compared with the fermentations of materials not treated with these enzymes. The concentrations of both disaccharides decreased during fermentation, indicating an assimilation of the maltose and/or an enzymatic degradation of both sugars during fermentation. Both GC200 and Multifect^® Pectinase FE have been shown to contain considerable β-glucosidase activity [14,15]. On the other hand, it has also been shown that an amendment of additional β-glucosidase to a GC200/Multifect^® Pectinase FE mixture can improve the degradation of cellulosic biomass [14]. However, in our experiments the maximum concentrations of cellobiose were below 0.3 g/L (results not shown), thus it does not seem to be relevant for increasing the ethanol yield from wheat grain. HPAE-PAD analysis also revealed the presence of small amounts of xylose, arabinose and galactose in the fermentation broths. Here concentrations were almost twice as high in CHP-treated samples compared with those treated only with Stargen 001. The amount of these sugars did not decrease during the fermentations in accordance with the stated inability of S. cerevisiae to ferment these sugars [13]. However, in all fermentations the concentrations of these sugars were always less than 1 g/L (results not shown), thus the potential impact of fermenting these sugars on the total ethanol yield from wheat grain seems to be insignificant.

Saccharomyces cerevisiae AEF1, an isolate from a Swedish ethanol production plant was kindly provided by the Agroetanol company (Norrköping, Sweden). Pichia anomala J121 (CBS 100487) was used as the biopreservative yeast [4].

Yeasts were grown on YPD medium (20 g/L glucose, 20 g/L peptone and 10 g/L yeast extract, 16 g/L agar for solid medium) at 30°C. Selective YPD medium additionally contained 0.1 g/L chloramphenicol (Sigma-Aldrich Inc., St Louis, MO, USA) to suppress the growth of bacteria.

To study the presence of bacteria in the fermentations, a solid medium selective for bacteria was used, LB medium (10 g/L trypton, 5 g/L NaCl, 5 g/L yeast extract, 16 g/L agar and 0.1 g/L Delvocid [active compound natamycin, Gist-Brocades, Delft, The Netherlands]).

For starch degradation, Stargen™ 001 (Danisco US Inc., Rochester, NY, USA), a mixture of alpha amylase and glucoamylase for industrial ethanol fermentations was used. The activity of the Stargen 001 enzyme mixture has been determined to be at least 456 granular starch hydrolysing units/g protein. GC 220 (Danisco US Inc., Rochester, NY, USA) was used as a cellulose and hemicellulose-degrading enzyme mixture, and Multifect^® Pectinase FE (Danisco US Inc., Rochester, NY, USA) was used for pectin degradation. The cellulase activity of GC 220 has been determined to be at least 6200 carboxymethylcellulose activity units (IU)/g protein. One IU unit is defined as the activity needed to liberate 1 μmol of reducing sugars in 1 minute. The pectinase activity of the Multifect^® Pectinase has been determined to be at least 145 IU pectinase/g protein. Enzyme activities were determined by the provider. All enzymes were a kind gift from Danisco US Inc., Genencor Division (Rochester, NY, USA).

The wheat grain was kindly provided by a local farmer (Anders Eriksson, Uppsala, Sweden). The farm grain (18% water content) was stored in a covered plastic barrel and used as a control. Moist wheat grain (30% water content) was obtained by re-moistening the dry grain and stored as 18-g portions in airtight reaction tubes with a simulated air leakage as previously described [4]. Biocontrol yeast was added to some of the tubes at a concentration of 10⁵ cells/g wheat grain [21]. Moist grain was stored for 4 weeks before preparation of the fermentation mash. All storage occurred at room temperature.

Aliquots of 10 g of wheat grains (dry weight) were milled with a mixer (Braun kitchen machine) until the milled material could pass through a 1-mm screen. The resulting flour was mixed with water and the pH was adjusted to 5 with sulphuric acid. The final volume was 40 ml, which was poured into a 100-ml glass bottle. The suspension was gelatinised in a 100°C water bath for 25 minutes. Water was added to a volume of approximately 80 ml and the pH was readjusted to 5 if required. Subsequently, 25 μl of an amylase mixture (Stargen 001) was added and the bottles were incubated on a rotary shaker at 37°C and 100 rpm for 24 hours. A mixture of CHP-enzymes was added to the suspension of some of the samples before amylase incubation. This mixture contained GC 220 (3 mg protein/g grain dry weight) and Multifect^® Pectinase FE (0.1 mg protein/g grain dry weight). These amounts were calculated according to the supplier's recommendation to give a final enzyme concentration of 25 mg/g cellulose and 10 mg/g pectin, using the grain composition reported by Åman [20]. The samples with added CHP-enzymes were incubated at 37°C and 100 rpm for 72 hours; subsequently the samples were treated with amylases as described above.

After enzymatic pre-treatment, the bottles were filled to a volume of exactly 100 ml with water, the pH was adjusted to 5 and 1-ml sample was taken for sugar concentration determination.

S. cerevisiae was inoculated in a YPD medium and grown for 15 hours at 30°C and 150 rpm. Then 1 ml of this culture was added to 100 ml wheat grain mash in a 100-ml serum flask sealed by a rubber cork with a syringe needle through it and incubated at 30°C and 150 rpm. The test fermentations were run for 30 hours. Concentration of free glucose was directly tested during the fermentations using test stripes for glucose (Keto-Diabur-Test 5000, Roche, Mannheim, Germany) from 15 hours of cultivation until fermentation was finished. This was to make sure that no residual glucose was present when the fermentation was stopped.

Samples from the test fermentations were diluted in steps of 10-fold dilutions and 10 μl of the dilutions were dropped on selective plates for yeast and bacteria, and incubated at room temperature for 24 hours. Colonies in the drops were counted and the numbers of yeast and bacteria per millilitre of fermentation broth was calculated.

Yeast colonies from the fermentations were picked randomly from the quantification plates. Yeast DNA was isolated according to Liberal et al. [26] and PCR-fingerprints were generated as described earlier [12,27]. As controls, PCR-fingerprints were produced using genomic DNA of S. cerevisiae AEF1 and P. anomala J121 as templates.

To test if there were significant differences between the investigated storage methods, Student's t-test was performed with a significance level of 5% using Microsoft^® Excel 2000.