Validation of a Rapid Method for Measuring β-Glucosidase Activity in Fermenting Muscat Grape Musts

Method validation.

The absorbances of both p-nitrophenol (PNP) alone and the possible chromophore complex were tested to determine whether the absorbance at 405 nm of PNP increased via the addition of α-cyclodextrin. Cuvettes (2 mL) were prepared containing 2 mM of PNP and 6 mM of α-cyclodextrin (Sigma-Aldrich Canada, Oakville, ON) in 0.1 M monobasic sodium phosphate (NaH₂PO₄) adjusted to pH 5.0, 6.0, 7.0, and 8.0 with concentrated hydrochloric acid. Absorbances of PNP alone and the possible α-cyclodextrin-based chromophore complex were measured with an Ultraspec 1000E spectrophotometer (Pharmacia Biotech, Cambridge, UK) at 405 nm. The absorbances with and without α-cyclodextrin at the various pH levels allowed calculation of the molar absorption coefficients (ε) via the Beer–Lambert equation [A = ε·b·c, where A = absorbance, b = path length (cm), and c = concentration (moles/L)], to determine if the addition of α-cyclodextrin increased ε, and to determine the effect of pH on the formation of the chromophore. Determinations were done in triplicate.

Assay validation.

The effect of different concentrations (2 to 12 mM) of the substrate p-nitrophenyl β-O-d-glucopyranoside (PNPG) on the rate of its hydrolysis by 0.75 units of β-glucosidase (where 1 mg of pure β-glucosidase is 3.4 units) was examined. When PNPG is subjected to the activity of β-glucosidase, PNP is released by hydrolysis and can then bind with α-cyclodextrin (which was present in solution at 2 mM) and form the chromophore. The solution was buffered at pH 6.5 with 0.1 M sodium phosphate. The complex was then read spectrophotometrically at 405 nm. The reaction was followed for 5 min and the increase in absorbance from the complex formation was recorded each minute. From these data, enzyme kinetics were calculated, and Michaelis–Menten and Lineweaver–Burk plots were drawn. Determinations were done in triplicate.

Testing the assay on grape must.

Chardonnay Musqué must (~25 Brix) previously stored at −25°C was thawed at room temperature over 24 hr. Treatment of the must was necessary to reduce color and protein-complexing compounds that might interfere with the measurement of the β-glucosidase. Ultimately, to correctly measure chromophore formation through this modified assay, the wine/must was first treated to a high degree of centrifugation to remove extracellular material. This began with centrifugation of 25 mL of must at 8500 g for 30 min in an IEC Centra CL-2 centrifuge to remove cell debris. The pellet, with ≤1% of total glycosidase activity, was discarded (Aryan et al. 1987). Ammonium sulfate was then added to the supernatant to ~80% saturation (0.56 g/mL) as a protein-binding salt such that the β-glucosidase was bound up. This slurry was stirred for one hour and the precipitate was collected via centrifugation at 8500 g for 30 min to isolate the bound proteins, after which the pellet was resuspended in a sodium phosphate buffer containing 5 mM of sodium-ethylenediamine tetraacetate (Na₂EDTA). This resulted in a colorless solution containing the majority of the proteins. This solution was then passed through a Sephadex G-25 column to liberate the proteins from their salt-bound state. Elution of the solution from the column was performed in fractions and measured at 260 nm, where absorbance was related to the amount of protein present. Thereafter, the higher protein fractions were then isolated, and the β-glucosidase activity was tested using the above spectrophotometric method. Since the levels of enzymes in must were anticipated to be much lower than those of pure enzyme, measurements were required over 48 hr. Observations over such a lengthy time did not preclude the possibility of some spontaneous hydrolysis of PNPG in addition to actual enzyme-induced hydrolysis.

β-Glucosidase activity in wine treated with AR2000.

Approximately 200 kg of Chardonnay Musqué grapes harvested at Cave Spring Vineyards, Beamsville, Ontario, in September 2001 was destemmed and crushed in a pilot-scale destemmer-crusher and allowed 48-hr skin contact. The crushed grapes were then pressed at 2 bar in a pilot-scale basket press containing a water bladder. The unsettled must (~25 Brix) was then divided into six 20-L glass carboys and inoculated with Saccharomyces bayanus strain EC1118 (Lallemand Corp., Montreal, QB). The wine was fermented to dryness and AR2000 (Gist-Brocades, Servian, France) was added to three of the six carboys at a rate of 3 g/hL and thereafter incubated at 10°C for 5 weeks according to manufacturer’s specifications. The remaining three carboys served as a control. Wine samples were taken at the end of the 5-week period, and the previously mentioned extraction method and subsequent spectrophotometric assay were performed on both control and AR2000 samples, on each of 10 different protein fractions from each treatment replicate, at three different reaction pH levels (3.0, 4.0, and 6.5), and at two different incubation temperatures (10°C and 20°C) over a 48-hr sample reaction incubation period. Absorbance readings were taken at 0, 24, and 48 hr after addition of PNPG and α-cyclodextrin.

β-Glucosidase activity in fermentations by S. cerevisiae strain VL1.

Approximately 10 kg each of muscat grapes (Muscat Hamburg, Muscat Ottonel, Morio-Muskat, and Chardonnay Musqué), harvested from the University of Guelph Vineland Research Vineyard in September 2001 were stored at −25°C. These were thawed at room temperature, destemmed and crushed, and pressed as described previously, and the must (~20 Brix) was then fermented in duplicate 12-L glass carboys by S. cerevisiae strain VL1 (Lallemand) that was inoculated at ≈2 X 10⁶ cells/mL. Fermentation to dryness (<0.5 Brix) lasted 7 days. Over the course of the fermentation the previously mentioned extraction method and spectrophotometric assay were used to determine the daily β-glucosidase activity.

Comparison of enzyme kinetics.

Three 4-L replicates of muscat must (~20 Brix) reconstituted from a concentrate kit (Vineco International Ltd., St. Catharines, ON) were each inoculated with ≈2 x 10⁶ cells/mL of S. cerevisiae strain VL1. Samples were drawn at 24 hr, which had been determined to be the peak of β-glucosidase activity in the fermentation. These samples were then subjected to six concentrations of the substrate PNPG (2, 4, 6, 8, 10, and 12 mM), plus exactly 6 mM of α-cyclodextrin. From this the enzyme kinetics were followed and Michaelis–Menten and Lineweaver–Burk plots were drawn. These were compared to kinetics derived from pure almond β-glucosidase.

β-Glucosidase activity of yeast strains.

Six yeast strains, VL1, VL3, BA11, EC1118, D47, and CY3079 (Lallemand), were each inoculated into three 4-L replicates of reconstituted (~20 Brix) muscat juice (VineCo International). Fermentations were conducted over a 7-day period. Must samples were drawn daily and subjected to the extraction method and spectrophotometric assay described previously. Following an initial 5-min sample incubation at 20°C, absorbance was measured after 24 and 48 hr of incubation at 4°C.

Sensory evaluation.

Once fermentations had concluded, wines were settled, racked, cold stabilized at −2°C, racked again, and stored at 4°C in anticipation of sensory analysis. Descriptive analysis was performed on the wines from the yeast treatments using a 100-mm unstructured line scale to test for melon, pear, lemon, floral, muscat, and earthy aromas, and flavors. Nine judges tasted the wines in duplicate with the aid of CompuSense software (Compusense Inc., Guelph, ON). Standards for judge training (Table 1⇓) were agreed upon by consensus as representative of their descriptors by the sensory panel. The sensory panel had already received rigorous training for descriptive analysis of aromatic white wines; therefore, familiarization with the scorecard, standards, and a subset of the sample wines was deemed sufficient. Samples were randomly presented in duplicate sets of three wines.

Statistical analysis.

All data were analyzed using the SAS statistical package (SAS Institute, Cary, NC). Analysis of variance was performed using the general linear models procedure (proc glm), while correlation analysis was carried out using proc corr.

PNP-α-cyclodextrin chromophore complex.

A preliminary step was to assess the molar absorbance coefficient (ε) for both PNP and the PNP-α-cyclodextrin complex to determine whether the addition of α-cyclodextrin aided in spectrophotometric detection. Addition of 6 mM α-cyclodextrin to 2 mM PNP to form the chromophore complex dramatically increased both the absorbance and ε (Table 2⇓). The increase in pH increased both absorbance and εconsiderably, regardless of α-cyclodextrin addition (Table 2⇓). As absorbance increased so did the perceivable yellow color in the cuvette. This initial portion of this research therefore validated the chemical basis for the assay developed by Arnaldos et al. (1999) for its eventual modification for the determination of β-glucosidase activity in grape must and wine. The chromophore complex formation was enhanced by increased pH and was indeed substantially higher in absorbance compared to PNP alone. The higher ε of the complex is thought to be due to the decrease in the pKa of PNP caused by α-cyclodextrin (Connors and Lipari 1976).

Assay with almond β-glucosidase and PNPG.

To test for the validity of the chromophore production in the presence of β-glucosidase, hydrolysis of different concentrations of the substrate PNPG by 0.75 units of pure almond β-glucosidase in the presence of 6 mM of α-cyclodextrin was analyzed spectrophotometrically over 5 min. When increasing doses of the substrate were present with the same concentration of α-cyclodextrin and β-glycosidase and at the same pH, the reaction kinetics were of a standard Michaelis–Menten type (Figure 1A⇓), from which the Km and Vmax of the reaction were calculated (Karp 2004). The Michaelis–Menten and Lineweaver–Burk plots (Figure 1A,B⇓) indicated that an increase in substrate increased β-glucosidase activity until the enzymes began to reach a saturation point of ~6 to 8 mM of PNPG, which determined the level of its future use in the assay. The formulated Vmax was a 5.82 x 10⁻³ change in absorbance per second, and the Km was 7.15 mM of PNPG. The control, which contained the same chemical additions with no enzyme, exhibited no chromophore formation. Thus, PNP can be formed from the substrate PNPG when β-glucosidase cleaves it, after which the PNP is free to form the complex with α-cyclodextrin. The reaction was measurable at a range of pH values, but it was evident that glycosidase activity within wine pH (3 to 4) was not high.

β-Glucosidase activity in wine treated with AR2000.

The assay was modified to ensure that color and extracellular material from a natural must or wine source of β-glucosidase would not interfere with the spectrophotometric assay. Use of centrifugation, a column (Sephadex G-25), and protective agents (EDTA) allowed the original assay (Arnaldos et al. 1999) to be used to measure the enzyme in must and wine. The 10 protein fractions that could possibly contain β-glucosidase were isolated from Chardonnay Musqué wines with or without (control) AR2000 treatment. Six subfractions of each protein fraction were taken to simultaneously test the effects of three different pH levels (3.0, 4.0, and 6.5) combined with two incubation temperatures (10°C and 20°C) for each pH via the modified assay. Absorbance increased with incubation time in AR2000 treatments only (Figure 2A⇓). As buffer pH increased, so did chromophore formation (in both control and AR2000 treatments); absorbance increased slightly as pH increased from 3.0 to 4.0 and increased substantially between 4.0 and 6.5 (Figure 2B⇓). Apparent differences in enzyme activity between the control and AR2000 were noticeable at pH 6.5 only. Increased incubation temperature also increased absorbance in both AR2000 and the control, and treatment differences were apparent at both incubation temperatures (Figure 2C⇓). Temperature effects on absorbance were equal in magnitude for all pH and incubation times (data not shown); in other words, there were no two-way or three-way interactions among time, pH, and temperature. In all cases, AR2000 absorbances substantially exceeded those of the control wines. A control containing no protein isolate from wine but all the chemical additions exhibited no chromophore formation.

Overall, AR2000 wines exhibited a greater amount of β-glucosidase activity than the controls. Absorbance readings in the controls were likely attributable to spontaneous hydrolysis of PNPG in addition to possibly yeast-derived glycosidases. Since wines were dry, it is unlikely that the presence of glucose played a role in the assay. All three variables impacted the magnitude of the reaction on the AR2000 wines, but pH and temperature also showed some differences in the control, presumably because of nonenzymatic hydrolysis. In general, enzyme activity was optimized by pH values close to neutral. Longer incubation times (up to 48 hr) were necessary to enhance the colorimetric reaction, and 20°C incubation temperatures increased the colorimetric reaction over 10°C treatments. The effect of wine pH in reducing β-glucosidase activity agrees with result of other experiments (Gallifuoco et al. 1997, Günata et al. 1997), as does the effect of lowering the incubation temperature (Günata et al. 1990b). Yet, examining the effect of incubation time on the reaction kinetics is unique to these experiments, and it was a significant factor in allowing the hydrolysis of the substrate, forming the chromophore, and enhancing the absorbance.

β-Glucosidase activity in fermentation by S. cerevisiae strain VL1.

The modified assay was used to test β-glucosidase activity of S. cerevisiae strain VL1 on a daily basis throughout two replicate muscat fermentations. Five protein fractions were eluted from each sample. Only the second and fifth fractions were found to be different from each other at 260 nm, and these showed the highest and lowest absorbances, respectively (data not shown). The largest increase in the hydrolysis of PNPG occurred after 24 hr into the fermentation, after which absorbance decreased dramatically by 48 hr and plateaued throughout the following 120 hr (Figure 3⇓). There were no differences in absorbance with respect to time other than those between the samples collected after 24 hr and all the others (Figure 3⇓). A control containing the same chemical additions and no protein isolate from the fermenting must exhibited no chromophore formation.

Fermentation of a muscat must by S. cerevisiae VL1 thus allowed use of the modified assay on a daily basis to determine how yeast growth and fermentation rate affected the production of β-glucosidase. When individual days were compared with respect to mean β-glucosidase, day 1 activity was substantially higher than all others. Immediately after inoculation, the enzyme activity was moderate; it then rose with exponential growth of the yeast, then dropped quickly, falling below initial inoculation values. This is similar to an experiment that endeavored to assess β-glucosidase through a different assay, where a sharp peak in enzyme activity corresponded to the exponential growth of the yeast population (~24 hr after fermentation) (Delcroix et al. 1994).

Enzyme kinetics of β-glucosidase.

A fermentation using S. cerevisiae strain VL1 on muscat must was conducted to test whether its β-glucosidase enzyme followed similar Michaelis–Menten kinetics to that of the pure almond β-glucosidase. The Michaelis–Menten (Figure 4A⇓) and Lineweaver–Burk (Figure 4B⇓) plots indicated that an increase in substrate increased β-glucosidase activity until the enzymes reach their saturation point, at ~6 to 8 mM of PNPG. The formulated Vmax was a 4.59 x 10⁻⁷ change in absorbance per second (405 nm), and the Km was 6.28 mM of PNPG. A control containing the same chemical additions, but no protein isolate from the must, exhibited no chromophore formation.

Fermentation of muscat must by S. cerevisiae strain VL1 was performed with the same methodology as with the almond β-glucosidase so that the kinetics of the enzyme at peak production could be examined and compared. The yeast-excreted enzyme had a saturation point of ~6 to 8 mM of PNPG, a formulated Vmax of 4.59 x 10⁻⁷ change in absorbance per second, and a Km of 6.27 mM of PNPG (compared to a 5.82 x 10⁻³ change in absorbance per second, and a Km of 7.15 for pure almond glucosidase). Hence, although Vmax values differed between the enzymes by several orders of magnitude, the Km values were similar for both the pure almond glucosidase and the must enzyme isolate; values were 7.15 and 6.28 mM PNPG, respectively. The Km for β-glucosidase has been variously reported as being between 0.63 and 3.3 mM PNPG (Gallifuoco et al. 1997, Günata et al. 1997, Spagna et al. 1998). The Vmax of an enzymatic reaction can change according to the concentration of substrate, but the Km should be constant for a given enzyme (Karp 2004). The parameters of the modified assay described here, particularly the use of the PNP/α-cyclodextrin chromophore formation, heighten the accuracy in measuring the hydrolysis of PNPG and may explain the apparently higher Km in this study compared to those previously reported.

β-Glucosidase activity of six yeast strains.

The five S. cerevisiae strains (VL1, BA11, CY3079, D47, and VL3) and one S. bayanus strain (EC1118) were each used to ferment three 4-L replicates of muscat must. The β-glucosidase activity for each yeast strain varied throughout the fermentation period (Figure 5⇓). The individual fractions separated via protein concentration (260 nm) were each measured, and data in Figure 5⇓ represent mean β-glucosidase activity across all fractions for each day. The six yeast strain fermentations showed significant differences in β-glucosidase activity. Initially (15 min after inoculation), VL1 showed high of β-glucosidase activity compared with the other yeast strains, noteworthy because VL1 has been marketed for having high β-glucosidase activity. This elevated activity was observed initially and for slightly over 48 hr, but it then dropped off. All yeast strains except EC1118 peaked in activity after 24 hr, with D47, VL1, and CY3079 exhibiting the highest activity. This general pattern of activity was observed for D47, VL3, BA11, and CY3079, but lower levels of activity were measured. CY3079 seemed to have a second smaller peak in activity later in the fermentation. The EC1118 fermentations were slower and peaked by day 3, exhibiting the highest sustained activity from day 3 to 5, inclusive (Figure 5⇓). BA11 tended to show the lowest activity throughout. Controls containing all the chemical components but no must protein showed no observable chromophore formation.

Least squares (LS) means were calculated in order to express mean overall enzyme activity throughout the 7-day fermentation. Strains VL1, CY 3079, and EC1118 were high in overall glucosidase activity, BA11 and VL3 were relatively low, and D47 was moderate (Table 3⇓).

EC1118 therefore showed the overall highest activity based on LS means, but only peaked by the third day and then slowly dropped off, perhaps because of the slower fermentation of the must for all three EC1118 fermentations compared with the other yeast strains. The patterns of β-glucosidase activity found here are similar to those of three unidentified strains, which peaked after the first day and then declined (Delcroix et al. 1994). However, this work is unique in that significant differences were found in activity among the yeast strains. Other research that has attempted to measure β-glucosidase activity from yeast strains has also found little to no differences (Rosi et al. 1994). Overall, LS means of β-glucosidase activity were highest for EC1118 followed by VL1, with the lowest activity found for VL3 followed by BA11. A caution must be attached to these data because glucosidases may be inhibited by glucose present in the fermentation medium. The assay performed on the fermenting musts did not ultimately contain added glucose, thus the activities reported herein may be higher than those found within actual fermentations. However, the relative activities measured by this assay should reflect those expected under normal fermentation conditions.

Time of incubation.

One of the major aspects of enumerating the β-glucosidase activity of the samples that had gone through the enzyme isolation process was measuring the hydrolysis activity over time (e.g., VL1, Figure 6⇓). The sample absorbance represented the enzyme’s ability to continually hydrolyze substrates after being released into a medium, and although sustained hydrolysis has been shown to occur for concentrated β-glucosidase (Martino et al. 1995a,b, Mateo and Di Stefano 1997), it has not been observed on β-glucosidase isolated from a fermenting must. The fractions were incubated with the substrate and chromophore-forming compound for a total of 48 hr and were measured at 24- and 48-hr intervals (Figure 6⇓). A substantial increase of hydrolysis was observed after 24 and 48 hr (depending on day and therefore β-glucosidase present). During the 48-hr incubation, the absorbances of the chromophore complex increased in some of the fractions, suggesting that the activity of the β-glucosidase in the isolate continued with time. It is also likely that some spontaneous hydrolysis of PNPG occurred during this timeframe.

Sensory evaluation of wines.

Two replicate wines of each yeast treatment were randomly presented to a sensory panel of nine judges and assessed for six aroma descriptors. Although there were no significant aroma differences, trends in the data were nonetheless apparent (Figure 7⇓). Melon and muscat aromas were highest in VL1 and EC1118 treatments, whereas the earthy aroma was lowest among these fermentations. EC1118 was also high in pear and lemon aromas. VL3 had more floral and earthy aromas and was low in muscat, pear, and lemon aromas. D47, CY3079, and BA11 had moderate intensities of all aroma descriptors, with the exception of high earthy aroma for D47. Intensities of melon and lemon flavors were not impacted by yeast strain. However, muscat flavor differed among the strains; D47 wines had lowest muscat flavor, whereas all others produced roughly equal intensities. Both sweetness and bitterness were also impacted by the yeast strains, with VL1 producing the least bitter and least sweet wines and D47 producing the most bitterness and sweetness. Lemon and muscat flavor were positively correlated with most other aroma and flavor descriptors (Table 4⇓). Anomalous data included a lack of correlation between muscat aroma and flavor. Hence, although no differences were observed among the yeasts for the various aroma descriptors, muscat flavor, sweetness, and bitterness were affected by yeast strain. Data also suggested that β-glucosidase activity of the yeast strains was related to increased aromas and flavors, particularly muscat and melon aroma. Other aroma descriptors such as pear, lemon, and floral were also related to increased β-glucosidase activity. Although we cannot state definitively that the sensory differences between the wines were due to hydrolysis of terpene glycosides by the various yeast strains, the yeast strains did have a sensory impact that might be glycosidase related. The differences between the wines may also be partly due to esters or other aroma-active compounds synthesized by the yeasts.