Comparative Investigations on Different β -Glucosidase Surrogate Substrates

: β -glucosidases are hydrolyzing enzymes which can release many aroma-active compounds from their glycoside form. Several yeasts produce these enzymes and thus are applied during the wine production process. To be able to test speciﬁc organisms for the presence of β -glucosidases and to investigate this enzyme activity, four main surrogate substrates have been described. The properties and applicability of these compounds, named arbutin (hydroquinone- β -D-glucopyranoside), esculin (6-O-(-D-glucosyl)aesculetin), 4-nitrophenyl- β -D-glucopyranoside (pNPG) and 4-methylumbelliferyl-β -D-glucopyranoside (4-MUG), are discussed after comparing their advantages and disadvantages. Although all four substrates were found suitable for photometric assays, 4-MUG has proven to be most appropriate due to high sensitivity, high robustness and simple processing. Furthermore, the investigation of β -glucosidase product accumulation is described, which could be used to give indications about β -glucosidase localization. 325 assay microplates used. Arbutin, esculin and 4-MU were analyzed in black microplates. The ampliﬁcation of the ﬂuorescence after the from into electricity (Gain) was set to 60 for all ﬂuorescence


Introduction
A high proportion of the aromatic substances that are responsible for the flavor and taste of fruits occur as glycosides and thus do not contribute to the aroma profile [1]. Studies with grape juices or vine grapes showed that aroma glycoside concentrations are two to ten times higher than the concentrations of the free aromatic molecules [2,3]. The presence of aroma glycosides is also confirmed for many other industrially processed fruits like apricot, peach, yellow plum, quince, sour cherry, passion fruit, kiwi, papaya, pineapple, mango, lulo, raspberry and strawberry [3]. The most probable reason for the frequent binding of aromatic substances to several glucose molecules or other sugars is lower toxicity and better solubility of the glycosides, which is associated with a better storage capability [4,5].
For the industrial processing of fruits, glycosides represent a great opportunity to enhance the aroma intensity of the products. A natural possibility to release aroma compounds from the glycosylated forms is the application of β-glucosidases, which are present in many non-Saccharomyces yeasts such as Aureobasidium pullulas [6][7][8] and Issatchenkia terricola [9]. Moreover, β-glucosidases are not only interesting with regard to the release of aroma-active substances, but for example also play an important role in the degradation of lignocellulosic biomass through the enzymatic route [10,11].
In this study, four β-glucosidase surrogate substrates are compared and characterized with regard to their practical application possibilities. Since the first description of 4-nitrophenyl-β-D-glucopyranoside (pNPG) for the assessment of β-glucosidase activities [12], this substrate has been widely used for the kinetic characterization of enzymes. Arbutin (hydroquinone-β-D-glucopyranoside) was used previously as a surrogate substrate for the screening of β-glucosidase activities in yeasts of oenological origin [13]. 6-O-(-D-Glucosyl)aesculetin (esculin) has traditionally been used to identify bacteria [14], Glucosyl)aesculetin (esculin) has traditionally been used to identify bacteria [14], and has recently been used in an esculin-glycerol solid medium for rapid screening of larger collections of native wine yeast strains [15]. Beside the frequent use of 4-methylumbelliferylβ-D-glucopyranoside  in the context of human β-glucosidase research [16,17], the use of agar plates containing this surrogate substrate for microbiological screening purposes was described [18]. In 2005, a modified fast 4-MUG-based assay method in microplates with the possibility of β-glucosidase activity quantification was developed [19].
The structures of the aforementioned compounds are very similar (Figure 1). Hydrolysis of the β-glycosidic bond can be photometrically detected due to the absorption or fluorescence properties of either the glycoside or the aglycon forms. However, these surrogate substrates differ in their spectroscopic properties and in their application possibilities for β-glucosidase screening. A direct comparison of these substances is not yet available but will be helpful for the selection of the most suitable compound for screening approaches or other applications.

Agar Plate Assays
If not otherwise stated, all chemicals and enzymes were purchased from Sigma Aldrich (St. Louis, MO, USA). For esculin agar plates, 0.3 g L −1 ferric chloride, 1 g L −1 casein hydrolysate, 25 g L −1 yeast extract, 8 mL L −1 glycerol, 20 g L −1 agar and 1 g L −1 esculin were autoclaved and poured into petri dishes [15]. For arbutin plates, an equal amount of arbutin instead of esculin was used (respective data is not shown in the manuscript). For liquid assays with arbutin or esculin, the medium was prepared without agar. After incubation of the commercially available yeasts at 30 °C for 48 h, the formation of brown colorations was examined.

Photometric Measurements
A multimode microplate reader (Tecan Spark, Tecan AG, Männedorf, Switzerland) was used for photometric measurements. The below mentioned wavelengths for the measurements were previously identified for each surrogate substrate by performing excitation and emission scans between 300 and 600 nm of 100 µL samples with 5 mM surrogate concentration in 100 mM citrate-phosphate buffer pH 7. The absorption of p-nitrophenol (pNP) was measured at 400 nm. Fluorescence emission of 4-Methylumbelliferone (4-MU) was measured at 450 nm after excitation at 330 nm, whereas fluorescence emission of esculin was measured at 454 nm after excitation at 367 nm. Fluorescence measurements of arbutin took place at 325 nm after excitation at 260 nm. For the assay with pNP, transparent microplates were used. Arbutin, esculin and 4-MU were analyzed in black microplates. The amplification of the fluorescence signal after the transformation from light into electricity (Gain) was set to 60 AU for all fluorescence measurements. However, these surrogate substrates differ in their spectroscopic properties and in their application possibilities for β-glucosidase screening. A direct comparison of these substances is not yet available but will be helpful for the selection of the most suitable compound for screening approaches or other applications.

Agar Plate Assays
If not otherwise stated, all chemicals and enzymes were purchased from Sigma Aldrich (St. Louis, MO, USA). For esculin agar plates, 0.3 g L −1 ferric chloride, 1 g L −1 casein hydrolysate, 25 g L −1 yeast extract, 8 mL L −1 glycerol, 20 g L −1 agar and 1 g L −1 esculin were autoclaved and poured into petri dishes [15]. For arbutin plates, an equal amount of arbutin instead of esculin was used (respective data is not shown in the manuscript). For liquid assays with arbutin or esculin, the medium was prepared without agar. After incubation of the commercially available yeasts at 30 • C for 48 h, the formation of brown colorations was examined.

Photometric Measurements
A multimode microplate reader (Tecan Spark, Tecan AG, Männedorf, Switzerland) was used for photometric measurements. The below mentioned wavelengths for the measurements were previously identified for each surrogate substrate by performing excitation and emission scans between 300 and 600 nm of 100 µL samples with 5 mM surrogate concentration in 100 mM citrate-phosphate buffer pH 7. The absorption of p-nitrophenol (pNP) was measured at 400 nm. Fluorescence emission of 4-Methylumbelliferone (4-MU) was measured at 450 nm after excitation at 330 nm, whereas fluorescence emission of esculin was measured at 454 nm after excitation at 367 nm. Fluorescence measurements of arbutin took place at 325 nm after excitation at 260 nm. For the assay with pNP, transparent microplates were used. Arbutin, esculin and 4-MU were analyzed in black microplates. The amplification of the fluorescence signal after the transformation from light into electricity (Gain) was set to 60 AU for all fluorescence measurements.

β-Glucosidase Substrate Hydrolysis in Solution
Enzyme kinetics were measured in 100 mM citrate-phosphate buffer pH 7 at room temperature after the addition of 10 µL of a 2.5 U mL −1 solution of β-glucosidase from almonds (Sigma Aldrich) to the samples. Arbutin, esculin or 4-MUG were added using 100 µL of a 5 mmol L −1 solution. Fluorescence spectra were measured in the same samples following the kinetic measurements. Accordingly, 50 µL of a 10 mmol L −1 solution of pNPG were used for the experiments. Before measurement, 50 µL of 0.1 mol L −1 Na 2 CO 3 were added to the sample. The absorption spectra of pNP were determined 20 min after enzyme addition. The total assay volume was 110 µL.

Comparison of Assay Sensitivity and Linearity
Fifty microliters of a 10 mmol L −1 4-MUG or pNPG solution in 100 mM citratephosphate buffer pH 7 were used and the assay was started by the addition of 10 µL of different almond β-glucosidase sample dilutions. Afterwards, the increase in absorption or fluorescence was measured for five minutes. Samples of pNPG were mixed with 40 µL of a 0.1 mol L −1 Na 2 CO 3 solution, samples of 4-MUG with 40 µL H 2 O before measurement. The signal emission slope per minute was calculated based on the measured values. For the pNPG assays we used enzyme amounts which resulted in constant slopes over an assay time of five minutes.

Determination of pH Robustness of 4-MU Quantification
Fluorescence spectra were measured in 50 µL of a 10 mmol L −1 4-MU solution which was combined with 50 µL of a 100 mM citrate-phosphate buffer of the mentioned pH value. For the estimation of glucose and ethanol influences, the applied buffer was supplemented with the respective component. The total volume of the reaction mixture was 100 µL.

Fluorescence Microscopy
Prior to imaging, 50 µL of cultures of Aureobasidium pullulans (CBS 100280) and Issatchenkia terricola (originated from the strain collection of the Hochschule Geisenheim University) were combined with 50 µL of a 0.76 mmol L −1 4-MUG solution [19] in 100 mM citrate-phosphate buffer pH 5.6. The used cultures were cultivated for 48 h at 30 • C in deepwell plates containing 600 µL Yeast Extract-Peptone-Dextrose Medium. After 10 min incubation at room temperature, images were acquired with an Axio Imager Z1 m fluorescence microscope (Carl Zeiss AG, Oberkochen, Germany) using a DAPI filter. The exposure time was set to 15 s.

Necessity of Photometric Assays for Reliable β-Glucosidase Activity Determination
Arbutin and esculin were often used in the context of agar plates to which iron compounds like ferric ammonium citrate [20] or ferric chloride [15] are simultaneously added. The complexation of iron by the released aglycons leads to the visible formation of elemental iron which indicates β-glucosidase activity. Initial tests with both compounds showed their suitability for respective assays in liquid culture. However, the outcome of respective tests with yeast suspension cultures of several commercially available wine yeasts was clearly different to the corresponding agar plate assay result in many cases ( Figure 2). Only two of the commercial wine yeast strains caused a dark staining in the agar plate assay and in the liquid culture assay with esculin Kitzinger Aromatic was able to improve flavor diversity during fermentation, while Bioferm Malic improved the degradation of malic acid. One strain (Bioferm Rouge) caused no staining in both assays, but the other eight strains led to contrary results. Whereas the Kitzinger strains Champagner, Portwein, Universal, Bordeaux and Steinberg showed positive signals only in the liquid assays, the Bioferm strains Blanc, Doux and Killer caused iron complexation only in the agar plate assays. assays, the Bioferm strains Blanc, Doux and Killer caused iron complexation only in the agar plate assays. The described differences between yeast suspension cultures and agar plate assays confirm the necessity to adapt the assay conditions to the subsequent application in liquid conditions. Photometric measurements of liquid assays additionally enable precise activity quantification and are feasible in a high-throughput format using 96-well plates.
Because previous experiments revealed the possibility of direct fluorescence measurement of arbutin and esculin (data not shown), these compounds were measured in the following experiments without iron addition.

Comparison of the Surrogate Substrates for Photometric Assays
After investigation of the absorption or fluorescence properties of all compounds in form of the glycoside or the aglycon in scanning experiments ( Figure S1), their hydrolysis was followed photometrically in experiments with 100 µL of a 5 mmol L −1 glycoside solution to which almond β-glucosidase was added to a final concentration of 0.23 U mL −1 (Figure 3). The corresponding fluorescence emission and absorption spectra are presented in Figure S1.
Investigation of the arbutin solution showed fluorescence signals with an emission maximum at 325 nm under the selected conditions ( Figure 3A). After the addition of almond β-glucosidase, the signal intensity slowly decreased within 12 h of measurement. Esculin showed higher fluorescence signals with an emission maximum at 455 nm, which rapidly decreased over 5 min after the addition of β-glucosidase ( Figure 3B). Although these two surrogate substrates were mainly used for agar plate applications during screening procedures, they can also be used to quantify β-glucosidase activities with fluorescence measurements in solutions, which has been recently stated also by two other studies [21,22]. In comparison, β-glucosidase-catalyzed pNPG hydrolysis led to increasing absorption values with a maximum at 400 nm ( Figure 3C). 4-MUG hydrolysis led to increasing fluorescence signals at 450 nm up to 50,000 units within 10 min after addition of the enzyme ( Figure 3D).
Hydrolysis of arbutin, esculin, pNPG or 4-MUG by an almond β-glucosidase clearly resulted in changes of the detectable photometric signals. While the detected compounds of 4-MUG and pNPG derive from products of the enzymatic reaction (4-MU and pNP), the hydrolysis of arbutin and esculin can be followed photometrically via the reduction of the substrate's fluorescence. Although the detection of fluorescence intensity decrease during β-glucosidase-mediated esculin hydrolysis represents a suitable assay, the detection of increasing absorption or fluorescence signal intensities during conversion of pNPG or 4-MUG is less error-prone. This is due to the fact that many parameters can lead to The described differences between yeast suspension cultures and agar plate assays confirm the necessity to adapt the assay conditions to the subsequent application in liquid conditions. Photometric measurements of liquid assays additionally enable precise activity quantification and are feasible in a high-throughput format using 96-well plates.
Because previous experiments revealed the possibility of direct fluorescence measurement of arbutin and esculin (data not shown), these compounds were measured in the following experiments without iron addition.

Comparison of the Surrogate Substrates for Photometric Assays
After investigation of the absorption or fluorescence properties of all compounds in form of the glycoside or the aglycon in scanning experiments ( Figure S1), their hydrolysis was followed photometrically in experiments with 100 µL of a 5 mmol L −1 glycoside solution to which almond β-glucosidase was added to a final concentration of 0.23 U mL −1 (Figure 3). The corresponding fluorescence emission and absorption spectra are presented in Figure S1.
Investigation of the arbutin solution showed fluorescence signals with an emission maximum at 325 nm under the selected conditions ( Figure 3A). After the addition of almond β-glucosidase, the signal intensity slowly decreased within 12 h of measurement. Esculin showed higher fluorescence signals with an emission maximum at 455 nm, which rapidly decreased over 5 min after the addition of β-glucosidase ( Figure 3B). Although these two surrogate substrates were mainly used for agar plate applications during screening procedures, they can also be used to quantify β-glucosidase activities with fluorescence measurements in solutions, which has been recently stated also by two other studies [21,22]. In comparison, β-glucosidase-catalyzed pNPG hydrolysis led to increasing absorption values with a maximum at 400 nm ( Figure 3C). 4-MUG hydrolysis led to increasing fluorescence signals at 450 nm up to 50,000 units within 10 min after addition of the enzyme ( Figure 3D).
Hydrolysis of arbutin, esculin, pNPG or 4-MUG by an almond β-glucosidase clearly resulted in changes of the detectable photometric signals. While the detected compounds of 4-MUG and pNPG derive from products of the enzymatic reaction (4-MU and pNP), the hydrolysis of arbutin and esculin can be followed photometrically via the reduction of the substrate's fluorescence. Although the detection of fluorescence intensity decrease during β-glucosidase-mediated esculin hydrolysis represents a suitable assay, the detection of increasing absorption or fluorescence signal intensities during conversion of pNPG or 4-MUG is less error-prone. This is due to the fact that many parameters can lead to decrease of absorption or fluorescence signals, e.g., precipitation phenomena or pH changes, whereas an increase of signal intensity is rarely caused by artifacts. decrease of absorption or fluorescence signals, e.g., precipitation phenomena or pH changes, whereas an increase of signal intensity is rarely caused by artifacts.

Assessment of Assay Limitations and Linearity
Based on the previously performed experiments, 4-MUG and pNPG were identified as the most promising surrogate substrates. However, for the detailed characterization of β-glucosidases, a reliable quantification of the activity is of high importance. Due to the different measurement methods of the two surrogate substrates, differences in the quantification limits are very likely. To further investigate these differences, varying amounts of almond β-glucosidase were added to the surrogate substrates and the resulting signal increase was compared (Figure 4).
While a fluorescence signal intensity increase of 30,763 AU per minute could be detected with 100 µU mL −1 of the enzyme in the reaction with 4-MUG as substrate, the signal strength was reduced from 14,706 AU per minute (50 µU mL −1 β-glucosidase), over 1470 AU (5 µU mL −1 β-glucosidase) to 64 AU (0.5 µU mL −1 β-glucosidase). In comparison, the absorption increase resulting from pNP formation by equal amounts of almond β-glucosidase declined unequally especially with high enzyme concentrations. The absorption signal increase of 0.1735 AU per minute determined with 100 µU mL −1 of enzyme dropped to an absorption increase of 0.0016 AU after the addition of 5 µU ml −1 of almond β-glucosidase.
Furthermore, the fluorescence signal increase of 4-MU is proportional to concentrations of almond β-glucosidase from 100 µU mL −1 to 1 µU mL −1 . Only the lowest tested enzyme amount (0.5 µU mL −1 ) was outside of the linear range. The observed changes in the absorption signal of pNP, in contrast, did only show a linear correlation for enzyme concentrations up to 10 or 50 µU mL −1 . At least the value measured with 100 µU mL −1 seems to be outside the linear range.

Assessment of Assay Limitations and Linearity
Based on the previously performed experiments, 4-MUG and pNPG were identified as the most promising surrogate substrates. However, for the detailed characterization of β-glucosidases, a reliable quantification of the activity is of high importance. Due to the different measurement methods of the two surrogate substrates, differences in the quantification limits are very likely. To further investigate these differences, varying amounts of almond β-glucosidase were added to the surrogate substrates and the resulting signal increase was compared ( Figure 4).

Investigation of the pH Robustness of 4-MUG Assays
Due to the strong pH-dependence of pNP absorption [23], the samples have to be mixed with a Na2CO3 solution before the measurement. This procedure is time consuming, increases measurement inaccuracies and complicates kinetic measurements. To identify potential changes in the spectral properties of 4-MU, its emission spectra were ana- While a fluorescence signal intensity increase of 30,763 AU per minute could be detected with 100 µU mL −1 of the enzyme in the reaction with 4-MUG as substrate, the signal strength was reduced from 14,706 AU per minute (50 µU mL −1 β-glucosidase), over 1470 AU (5 µU mL −1 β-glucosidase) to 64 AU (0.5 µU mL −1 β-glucosidase). In comparison, the absorption increase resulting from pNP formation by equal amounts of almond β-glucosidase declined unequally especially with high enzyme concentrations. The absorption signal increase of 0.1735 AU per minute determined with 100 µU mL −1 of enzyme dropped to an absorption increase of 0.0016 AU after the addition of 5 µU ml −1 of almond β-glucosidase.
Furthermore, the fluorescence signal increase of 4-MU is proportional to concentrations of almond β-glucosidase from 100 µU mL −1 to 1 µU mL −1 . Only the lowest tested enzyme amount (0.5 µU mL −1 ) was outside of the linear range. The observed changes in the absorption signal of pNP, in contrast, did only show a linear correlation for enzyme concentrations up to 10 or 50 µU mL −1 . At least the value measured with 100 µU mL −1 seems to be outside the linear range.

Investigation of the pH Robustness of 4-MUG Assays
Due to the strong pH-dependence of pNP absorption [23], the samples have to be mixed with a Na 2 CO 3 solution before the measurement. This procedure is time consuming, increases measurement inaccuracies and complicates kinetic measurements. To identify potential changes in the spectral properties of 4-MU, its emission spectra were analyzed at different pH levels ( Figure 5A). Furthermore, the influence of different concentrations of glucose ( Figure 5B) or ethanol ( Figure 5C) was investigated.

Investigation of the pH Robustness of 4-MUG Assays
Due to the strong pH-dependence of pNP absorption [23], the samples have to be mixed with a Na2CO3 solution before the measurement. This procedure is time consuming, increases measurement inaccuracies and complicates kinetic measurements. To identify potential changes in the spectral properties of 4-MU, its emission spectra were analyzed at different pH levels ( Figure 5A). Furthermore, the influence of different concentrations of glucose ( Figure 5B) or ethanol ( Figure 5C) was investigated.

Microscopic Investigation of 4-MU Localization in Cultures with β-Glucosidase Activity
In order to test if 4-MUG can furthermore be used to investigate the localization of βglucosidase enzymes in different strains, fluorescence microscopy analyses of cell cultures before and after the addition of 4-MUG ( Figure 6) were performed.
After addition of the surrogate substrate, A. pullulans cells clearly accumulated 4-MU in the cytoplasm( Figure 6B) which indicates a cytoplasmic localization of high amounts of the responsible β-glucosidase enzymes. In the case of I. terricola, 4-MU fluorescence could be detected in cells and in the culture medium ( Figure 6D). This observation is

Microscopic Investigation of 4-MU Localization in Cultures with β-Glucosidase Activity
In order to test if 4-MUG can furthermore be used to investigate the localization of β-glucosidase enzymes in different strains, fluorescence microscopy analyses of cell cultures before and after the addition of 4-MUG ( Figure 6) were performed. consistent with the previously described secretion of significant amounts of β-glucosidase by this strain [9].

Discussion
In this comparative study of the most commonly used surrogate substrates, 4-MUG turned out to be most suitable for the photometric identification of β-glucosidase activities.
The described differences in solid and liquid assay conditions can be caused, for example, by the different availabilities of oxygen. Since the possible application of such strains takes place in a liquid environment (e.g., in a wine fermentation) and the liquid culture approach simplifies activity quantification, liquid assay conditions are most appropriate for screening.
Although arbutin and esculin were so far mainly used for the qualitative determination of β-glycosidic activities on agar plates, photometric determination and quantification is also possible under liquid assay conditions. Whereas the slow hydrolysis of arbutin is clearly a drawback for its potential use in a screening, esculin revealed to be suitable for this purpose. However, photometric detection of the substrate is more amenable to artifacts if compared to the detection of hydrolysis products in case of pNPG or 4-MUG.
A comparison of these two surrogate substrates revealed that the fluorescence-based assay is much more sensitive and apparently does not show saturation effects with the used enzyme concentrations. The possibility of adjusting the strength of the fluorescence signal amplification after the transformation from light into electricity (gain setting) in the case of more strongly deviating enzyme activities also leads to a considerable expansion of the measurable enzyme activities. In addition, the measurement of a fluorescent molecule is less influenced by cells and other sample ingredients compared to the absorptionbased assay. Although pNPG has been commonly used as substrate for β-glucosidase characterization in previous studies [24][25][26][27], 4-MUG represents a highly suitable alternative surrogate substrate for screenings and kinetic investigations of β-glucosidase activities.
The detected changes in the spectral properties of 4-MU due to variation of pH or the addition of ethanol should be taken into account, but do not prevent the use of 4-MUG as a surrogate substrate for β-glucosidase assays. To determine the amount of surrogate converted, separate calibration curves can be established for each tested condition. Therefore, After addition of the surrogate substrate, A. pullulans cells clearly accumulated 4-MU in the cytoplasm ( Figure 6B) which indicates a cytoplasmic localization of high amounts of the responsible β-glucosidase enzymes. In the case of I. terricola, 4-MU fluorescence could be detected in cells and in the culture medium ( Figure 6D). This observation is consistent with the previously described secretion of significant amounts of β-glucosidase by this strain [9].

Discussion
In this comparative study of the most commonly used surrogate substrates, 4-MUG turned out to be most suitable for the photometric identification of β-glucosidase activities.
The described differences in solid and liquid assay conditions can be caused, for example, by the different availabilities of oxygen. Since the possible application of such strains takes place in a liquid environment (e.g., in a wine fermentation) and the liquid culture approach simplifies activity quantification, liquid assay conditions are most appropriate for screening.
Although arbutin and esculin were so far mainly used for the qualitative determination of β-glycosidic activities on agar plates, photometric determination and quantification is also possible under liquid assay conditions. Whereas the slow hydrolysis of arbutin is clearly a drawback for its potential use in a screening, esculin revealed to be suitable for this purpose. However, photometric detection of the substrate is more amenable to artifacts if compared to the detection of hydrolysis products in case of pNPG or 4-MUG.
A comparison of these two surrogate substrates revealed that the fluorescence-based assay is much more sensitive and apparently does not show saturation effects with the used enzyme concentrations. The possibility of adjusting the strength of the fluorescence signal amplification after the transformation from light into electricity (gain setting) in the case of more strongly deviating enzyme activities also leads to a considerable expansion of the measurable enzyme activities. In addition, the measurement of a fluorescent molecule is less influenced by cells and other sample ingredients compared to the absorption-based assay. Although pNPG has been commonly used as substrate for β-glucosidase characterization in previous studies [24][25][26][27], 4-MUG represents a highly suitable alternative surrogate substrate for screenings and kinetic investigations of β-glucosidase activities.
The detected changes in the spectral properties of 4-MU due to variation of pH or the addition of ethanol should be taken into account, but do not prevent the use of