Abstract
Conventional approaches to measuring sulfur dioxide (SO2) in wine, such as aeration-oxidation, iodometric titration, and flow-injection analysis, are known to overestimate molecular SO2, particularly in red wine because of the dissolution of weak anthocyanin-bisulfite complexes during analysis. Methods for determining molecular SO2 without perturbing anthocyanin-bisulfite complexes and other weak adducts exist, e.g., headspace gas-detection tube measurements. However, it is unclear whether SO2 values achieved through conventional methods (“Molecular SO2”) or nonperturbing methods (molecular SO2) provide a better measure of antimicrobial activity. In our work, white and pseudo-red wines were prepared with varying additions of SO2; the red wine was produced by spiking the white wine with an anthocyanin extract. “Molecular SO2” and molecular SO2 concentration in white wines were well correlated, but “Molecular SO2” was significantly higher in red wines. Wines were inoculated with Saccharomyces cerevisiae strain EC1118 (Lallemand), and viability and culturability were evaluated at regular intervals. Both culturable and viable cell counts decreased significantly for treatments with 0.5 to 2.0 mg/L molecular SO2 in the white and red wines, and for 0.5 to 2.0 mg/L “Molecular SO2” in the white wine, but concentrations >2.0 mg/L “Molecular SO2” were necessary to decrease cell counts in the red wine. These results indicate that anthocyanin-bisulfite complexes have negligible antimicrobial activity, and that conventional approaches to measuring “Molecular SO2” are poorly suited to predicting the microbial stability of red wines.
Sulfur dioxide (SO2) has been used since at least the end of the 18th century as a wine preservative because of its antioxidant and antimicrobial effects (McGovern 2003). Although trace amounts of SO2 are produced by yeast during fermentation, the majority of SO2 in wine is intentionally added by winemakers. Because the first logarithmic acid dissociation constant of SO2 (pKa = 1.81 in H2O at 20°C) is low compared with wine pH, the major SO2 species at wine pH is bisulfite (HSO3−), and less than 5% of SO2 typically exists in its neutral, so-called molecular form (Waterhouse et al. 2016b). The sum of HSO3− and molecular SO2 is referred to as free SO2 (Waterhouse et al. 2016b). A portion of HSO3− in wine will also exist in the form of covalent adducts with wine nucleophiles, e.g., acetaldehyde. These bound SO2 forms have diminished preservative activity, but are counted along with free SO2 as part of total SO2. The antimicrobial activity of SO2 is primarily due to its molecular form, putatively because this neutral species can readily diffuse across the cell membrane (Divol et al. 2012).
Because the molecular SO2 species is volatile and perceived as “burning” or “irritating” at high concentrations (sensory threshold = 2 mg/L) (Waterhouse et al. 2016b), and because total SO2 concentrations are regulated in most winemaking countries, defining appropriate minimum molecular SO2 concentrations is of importance to winemakers. Specific recommendations vary with challenge study conditions and microorganism, and have been summarized elsewhere (Boulton et al. 1999). For example, for Saccharomyces (a major spoilage risk for sweet wines), decreasing cell counts from 5 × 104 cfu/mL to 1 or fewer cfu/mL after 24 hr at room temperature requires 1.41 to 1.74 mg/L molecular SO2 across Saccharomyces cerevisiae strains, whereas decimal (10-fold) reduction requires 0.24 to 0.32 mg/L molecular SO2, respectively (King et al. 1981). This previous work was performed in growth media, and other recommendations include 0.825 mg/L molecular SO2 for S. cerevisiae control, according to studies in model wine (10% ethanol) (Beech et al. 1979), or up to 1.55 mg/L molecular SO2 to reduce Saccharomyces bayanus to undetectable levels in white wines supplemented with 36 g/L sugars and adjusted to different pH levels and ethanol concentrations (Sudraud and Chauvet 1985).
Because of the range of different literature values, wine production textbooks may give a range of values, for example, 0.5 to 0.8 mg/L (Margalit and Crum 2004), with higher values recommended for wines at greater risk for spoilage, such as sweet wines. A complication to determining appropriate molecular SO2 recommendations is that molecular SO2 is not directly measured. Instead, free SO2 and pH are measured, and molecular SO2 is calculated using a modified Henderson-Hasselbalch equation (Waterhouse et al. 2016b). One problem with this approach is that the pKa value of SO2 depends on ethanol concentration, temperature, and ionic strength of the wine. These factors can have considerable consequences, as typical table wines with 11 to 14% alcohol by volume will have a pKa = 1.9 to 2.1 at 20°C, compared with 1.81 in pure water (Usseglio Tomasset and Bosia 1984).
A potentially larger issue to establishing appropriate SO2 targets is that conventional analytical approaches to free SO2 measurement, such as flow-injection analysis, aeration-oxidation (A–O), and iodometric titration, all involve an initial acidification and dilution step. These approaches can result in disruption of weakly bound bisulfite adducts, resulting in artifactually high concentrations of free SO2, and there-fore molecular SO2. This effect is particularly noticeable in red wines because of dissociation of anthocyanin-bisulfite complexes (Burroughs 1975). Although formation of adducts between the cationic form of anthocyanins (flavylium ion) and HSO3− is strongly favored (with a low dissociation constant of Kd = 1 × 10−5 for malvidin-3-glucoside) (Brouillard and El Hage Chahine 1980), the half-life for dissociation is on the order of minutes (Brouillard and El Hage Chahine 1980), comparable to the time necessary for conventional SO2 analyses (Coelho et al. 2015). In a recent study, Waterhouse et al. (2016a) referred to the sum of free SO2 and weakly bound SO2 measured by conventional analyses as “Free SO2”, in quotes, to distinguish it from the true amount of free SO2. We will adopt this convention in this paper, along with the analogous concept of “Molecular SO2” as compared with molecular SO2.
The measurement of molecular SO2 instead of “Molecular SO2” can be achieved by using nonperturbing methods that avoid changes to wine composition. For example, gas-detection tubes (GDT) can be used to measure headspace SO2 concentrations in an equilibrated sample, and molecular SO2 can then be calculated using Henry’s law (Coelho et al. 2015). Molecular SO2 concentrations measured by headspace GDT (HS-GDT) in red wines were, on average, only 32% that of molecular SO2 measured by A–O. The discrepancy between the two analytical methods could be modeled as a function of anthocyanin concentration and pH. By comparison, GDT measurements of white and rosé wines yielded molecular SO2 concentration that averaged 86% of A–O values (Coelho et al. 2015). Similar conclusions have been reached using other nonperturbing methods such as headspace gas chromatography (Davis et al. 1983).
“Free SO2” measurements of red wines are still fit for purpose if the weakly bound SO2 has antimicrobial activity comparable to that of free SO2. This assumption has justification, since strongly bound acetaldehyde-bisulfite complexes are reported to have activity against lactic acid bacteria (Wells and Osborne 2012), either because the adducts will release free HSO3− following metabolism of acetaldehyde, or because the acetaldehyde-bisulfite complex is itself inhibitory. However, similar antimicrobial effects of strongly bound SO2 on yeast are less pronounced; for example, acetaldehyde-bisulfite (160 mg/L as SO2) showed no effect on Brettanomyces bruxellensis viability (measured by direct epifluorescence filter technique) and survival (by plate counts) in a red wine over a two day period (Du Toit et al. 2005).
To our knowledge, an evaluation of the antimicrobial activity of anthocyanin-bisulfite adducts has been reported only once in the literature. Usseglio-Tomasset and colleagues reported that the presence of anthocyanin-bisulfite complexes (30 mg/L as SO2) delay fermentation by S. cerevisiae, S. bayanus, Saccharomyces uvarum, and Saccharomycodes ludwigii compared with an unsulfited control, but this delay is less pronounced than that observed with additions of free SO2; the authors concluded that anthocyanin-bisulfite adducts retain some antiseptic activity (Usseglio-Tomasset et al. 1982). However, survival was not evaluated, for example, by growth on selective media. Moreover, molecular (or “Molecular”) SO2 concentrations were not measured during the fermentation, so it is unclear whether initial differences in SO2 among treatments persisted.
In summary, current literature reports are ambiguous as to whether anthocyanin-bisulfite adducts possess antimicrobial activity and, thus, whether inadvertent measurement of these adducts in conventional “Free SO2” measurements is justified. In this study, we compared the validity of conventional (A–O) and nonperturbing (HS-GDT) measurements of molecular SO2 for predicting S. cerevisiae viability in challenge studies using wines with varying SO2 additions, and in the presence or absence of anthocyanins.
Materials and Methods
Chemicals
Potassium metabisulfite (97% [w/w]) and ethanol (95% [v/v]) were obtained from Acros Organics. Potassium bitartrate (99% [w/v]), hydrogen peroxide (30% [w/v]), sodium hydroxide (0.01 N), and o-phosphoric acid (85% [w/w]) were obtained from Fisher Scientific. A nominally 25% phosphoric acid solution was prepared as a 2.38:1 dilution of 294 mL phosphoric acid (85%) with 700 mL deionized water. Hydrochloric acid (36.5% [w/w]) was obtained from BDH Merck.
SO2 working standards
SO2 stock solutions at nominal concentrations of 10 g/L as SO2 were prepared weekly by dissolution of potassium metabisulfite in a solution of 10% (v/v) ethanol in water to avoid SO2 autooxidation. Analysis by A–O was used to confirm the concentration in the stock and working solutions (Iland 2004).
SO2 measurements by A–O
During the experiments, conventional analyses of “Free SO2” and total SO2 were performed by A–O, as described elsewhere (Iland 2004). To compare “Molecular SO2” with the HS-GDT molecular SO2 value, “Free SO2” for each wine was converted to “Molecular SO2” using a pKa value (2.09) calculated on the basis of the alcohol content (11.8%), HS-GDT measurement temperature (23°C), and an assumed wine ionic strength of 0.05 M, using equations described elsewhere (Usseglio Tomasset and Bosia 1984). The reported detection limit for “Free SO2” by A–O is 2 mg/L (Iland 2004), which equates to a “Molecular SO2” detection limit of 0.06 mg/L at the pKa value of the tested wines.
SO2 measurements by HS-GDT
HS-GDT measurements of headspace SO2 concentrations were performed using Gastec 5Lb GDT tubes (Gastec Corporation), and free SO2 concentrations were calculated, based on a protocol described elsewhere (Coelho et al. 2015). Two hundred mL of headspace was sampled for each analysis. According to the minimum detectable length of stain detectable on the GDT (0.3 mm), the limit of detection was estimated to be 0.03 mg/L molecular SO2, which would equate to a free SO2 detection limit of 1 mg/L for the pKa value (2.09) of the tested wines.
Wines
A 2013 commercial, “no sulfites added” white wine (45% Riesling, 24% Müller Thurgau, 17% Muscat Canelli, and 14% Chenin blanc; all grapes were Vitis vinifera, Columbia Valley AVA, WA) was used in all trials. Wines with identical lot numbers and bottling dates were used within each biological replicate. To produce a “red wine” with a basic composition similar to that of white wine, 2 g/L of grape anthocyanin powder (21.2% anthocyanin by weight) (Polyphenolics) was added to the control white wine. Basic wine compositional parameters are reported in Table 1 and were measured at ETS Labs (St. Helena, CA) by ISO 17025–accredited methods. Analyses of monomeric anthocyanins, polymeric anthocyanins, and tannins were performed by high-performance liquid chromatography (HPLC) using methods described elsewhere (Waterhouse et al. 1999).
Experiment A: challenge experiment with varying SO2 additions
An overview of this experiment is depicted in Figure 1. Prior to SO2 adjustments, both white and “red” wines were sterile-filtered by 0.2 μm polyethersulfone Nalgene Rapid-Flow disposable filter units (Thermo-Fisher). Total SO2 was then adjusted by addition of the 10 g/L stock SO2 solution. For both wines, SO2 was added to 1 L samples to yield the following concentrations: 0 (control), 22, 32, 44, 56, and 67 mg/L. For “red wines”, an additional five 1 L treatments were prepared at the following SO2 levels: 78, 92, 104, 114, and 128 mg/L. An additional 67 mg/L sample was prepared for both wines for use as an uninoculated control. “Free” and “Molecular SO2” were determined in each wine by A–O, and free and molecular SO2 were determined by HS-GDT, as described above.
Yeast/mold media
Yeast/mold (YM) broth media (Difco-BD) were prepared from 3 g/L yeast extract, 3 g/L malt extract, 5 g/L peptone, and 10 g/L dextrose; YM agar was prepared identically to the YM broth, except that 21 g/L agar (Difco-BD) was also added, and heat was used to facilitate dissolution. Both broth and agar media were sterilized by autoclaving for 15 min.
SO2 challenge experiments
Ethanol-conditioned yeasts were prepared twice, that is, before each set of replicate challenge experiments. One gram of S. cerevisiae strain EC1118 (Scott Laboratories) was added to 10 mL sterile water at 40°C for 30 min, with gentle agitation by hand after 15 min. The rehydrated sample was added to 125 mL sterilized 100% strength YM broth and agitated by platform rotary shaker at 20 to 25°C for 4 to 12 hr. This protocol was repeated three times, using broth with increasing alcohol concentrations (4, 8, and 12% [v/v]) and decreasing YM concentrations (76, 38, and 19% of full-strength YM broth). Yeast were collected by centrifugation for 10 min at 6000 rpm, washed twice with 0.1% peptone, consolidated into one tube, and refrigerated at an approximate concentration of 108 cfu/mL under peptone until use.
Ethanol-conditioned yeasts were then inoculated into each 1 L wine treatment at target rates of 106 cells/mL and mixed by gentle agitation and inversion; flasks were loosely capped. The flasks were stored in the dark except during sampling. The challenge experiment took place at room temperature (~22°C). Aliquots of each treatment were sampled prior to inoculation (T0) and at 10 time points (T1 to T10) after inoculation throughout the experiment for microbial analysis, and in some cases SO2 analyses, out to a maximum of 14,400 min (10 days) (Table 2). The same bottles were used and repeatedly sampled throughout the experiment. At each of the sampling times (T0-T10), 62 mL was drawn, for a total of 682 mL removed. The total liquid volume was 1 L at the start of fermentation, with ~100 mL headspace at T0, and the final headspace volume was ~782 mL after the T10 sampling. As discussed below, the increase in headspace may have resulted in nonenzymatic loss of SO2 by the last time point (T10). After sampling, SO2 was removed or inactivated, as previously described (Johnston et al. 2002). One mL of sample was added to 9 mL of 0.1% peptone (resulting in pH = 3.9), and SO2 activity was quenched by the combined effects of dilution and pH shift.
Survival assessed by plating on YM agar
Treatments were mixed prior to sampling by vigorous shaking. For the first biological replicate, a 4 mL sample was transferred directly into 16 mL of 0.1% peptone for a 5-fold dilution and serially repeated for 200- and 2000-fold dilutions. Samples were vortexed for 5 sec between dilutions or prior to plating. Plating was done using the drop-plate method (Herigstad et al. 2001), with half a dilution series (100 and 101, or 102 and 103) duplicated once per plate. Samples were incubated at 25°C and read between 48 and 72 hr. Counts between 10 and 200 cfu per 100 μL were used for calculating cell concentrations in the original treatment.
Viability assessed by flow cytometry
Flow cytometer staining buffers were prepared with 0.2% Pluronic F68 (BASF Corporation) and 1 mmol/L EDTA (Sigma) in phosphate-buffered saline (8 g/L NaCl, 0.2 g/L KCl, 1.44 g/L Na2HPO4, and 0.24 g/L KH2PO4) tablets (Sigma-Aldrich), adjusted to pH 7.4 with HCl, and filtered on Nalgene Rapid-Flow sterile, disposable filter units through a 0.2 μm polyethersulfone membrane (Thermo-Fisher). Thiazole orange (TO, “live/dead” stain) at 42 μmol/L in dimethyl sulfoxide and propidium iodide (PI, “dead” stain) at 4.3 mmol/L in water were obtained from Becton, Dickinson and Company. One hundred μL of sample was added to 400 μL of staining buffer, with 5 μL of each TO and PI (for a final concentration of 420 nmol/L TO and 42 μmol/L PI), mixed, and allowed to develop for 10 min in the dark.
Samples from the unsulfured white control wines were used as yeast controls. Dead yeast controls were produced through two methods. Heat-killed samples were made using 1 mL of live culture subjected to 10 min in a 90°C water bath. Alcohol-killed samples were produced by adding 1 mL of culture to 3.0 mL of 95% ethanol (final concentration 71%) for 10 min. Additionally, high level SO2-treated samples were made by adding 1 mL of live culture to 1 mL of 1 g/L SO2 solution (for a final concentration of 500 mg/L) for 10 min.
Cell counts and cell viability were then measured on an Accuri C6 flow cytometer (BD Biosciences) with a 488 nm laser and equipped with a 24 sample plate. A forward scatter threshold was set at fewer than 80,000 arbitrary units. Ultrapure HPLC-grade water (Alpha Aeser) with bacteriostatic solution (BD Biosciences) was used as sheath fluid. CFlow Plus software (BD Biosciences) was used for collection and evaluation. Samples were gated on yeast from the unsulfured white wine sample, using forward and side scatter, and viability was evaluated using FL1 (530 BP) and FL3 (670 LP) after staining with TO (emission 530 nm) and PI (emission at 617 nm), using killed yeast samples to confirm dead gates. No color compensation was used. Sample size varied from 25 to 100 μL, depending on culture concentration. Viability was determined in replicate at each time point.
Experiment B: confirmation experiment
To confirm the results of experiment A, a follow-up challenge experiment with a limited number of SO2 treatments was performed to generate a range of “Molecular” and molecular SO2 concentrations. Sterile-filtered wines were prepared as described above, and SO2 was added at the following levels: 0 mg/L (white, control), 74 (white), 0 (red), 74 (red), and 142 (red). “Free” and “Molecular SO2” were determined in each wine by A–O, and free and molecular SO2 were determined by HS-GDT, as described above. S. cerevisiae EC1118 inocula were prepared as described above, with the exception that the ethanol-acclimated yeast was stored in 0.2 μm-filtered, sulfite-free control white wine until inoculation. After inoculation, the wines were stored at room temperature for 960 min (equivalent to T8 in experiment A). To quench the effect of SO2 prior to plating analyses, 50 mL of each sample was centrifuged for 10 min at 6000 rpm, the supernatant wine sample carefully removed, and the pellet washed and resuspended in 50 mL of 0.2 μm-filtered control white wine (Johnston et al. 2002). Survival was then evaluated by duplicate plating, as described above, with the following exception of the dilution step size. In this experiment, 1 mL of sample was added to 9 mL of 0.1% peptone for a 10-fold dilution, and serially repeated for 102 and 103 dilutions.
Calculating log reductions and death curves
Viability counts in cfu/mL for plating (Vp) and in events/mL for flow cytometry (Vfc) at each time point (T) were converted to log values. The log values for replicates of both wine controls (no SO2 added) were averaged at each time point, and this value was subtracted from the sample value to generate a log reduction value for each sample (Equation 1).
Eq. 1
Statistical analyses
Minitab v. 16 (Minitab Inc) was used for statistical analyses. Paired Student’s t tests were used to evaluate whether differences existed in survival or viability between the molecular SO2 and “Molecular SO2” measurements for each wine type. Student’s t tests were also used to evaluate whether different levels of molecular SO2 or “Molecular SO2” (low, medium, or high) resulted in significant decreases in survival compared with the unsulfited control. Statistical significance was defined as p < 0.05.
Results
An initial goal of this study was to generate two wines for challenge studies with near-identical compositions and molecular SO2, but differing in “Molecular SO2”, i.e., the apparent molecular SO2 based on conventional SO2 measurements that include some weakly bound SO2 forms. To create control and treated wines differing in weak SO2 binders, we selected a base white wine with negligible “Free SO2” and total SO2 (<2 and <5 mg/L, respectively, Table 1) and spiked it with a commercial anthocyanin extract. Both wines were then analyzed for monomeric anthocyanin, polymeric anthocyanin, and tannin by an HPLC method. Monomeric anthocyanins were not detectable in the white wine, but were present in the “red wine” at 458 mg/L as malvidin-3-glucoside equivalents. The monomeric anthocyanin concentration in our “red wine” was higher than in typical commercial red wines, which would have undergone aging and lost monomeric anthocyanins through formation of polymeric pigment and other reactions, but were within the ranges observed in newly fermented wines (Monagas et al. 2006). Tannins were detectable only in the “red wine” (299 mg/L catechin equivalents). The presence of tannins in this wine presumably resulted from impurities in the anthocyanin extract, but the tannin concentration was still at the low end of a typical commercial red wine (Harbertson et al. 2008). “Free SO2” was still undetectable after the anthocyanin addition, but total SO2 (7 mg/L) exceeded the detection limit, possibly because of trace amounts of bound SO2 in the anthocyanin extract.
SO2 was then added at varying concentrations to each wine to generate wines with “Molecular SO2” (determined with A–O) and molecular SO2 (determined with HS-GDT) ranging from undetectable to at least 1.0 mg/L. The resulting “red” and white wines had similar total SO2 concentrations for a given SO2 addition (Supplemental Figure 1A). Determination of “Molecular SO2” involved initial measurement of “Free SO2”, and subsequent calculation of “Molecular SO2”. Our calculations accounted for effects of temperature, ionic strength, and ethanol on the acid dissociation constant, resulting in pKa values of 2.09, compared with the more common value of 1.81, based on the pKa in water at 20°C.
Challenge studies were performed in which a commercial wine yeast (S. cerevisiae strain EC1118) was inoculated into each of the SO2-adjusted wines and the control. Samples were taken at 10 time points over 10 days, with microbial analyses performed for all time points and SO2 analyses performed for some time points (Table 2). No significant change in total, “Free”, or free SO2 (Supplemental Figure 1A to 1E) occurred over the first 32 hrs (up to T9) for any treatment level, but total SO2 for a given SO2 addition level was significantly lower at 10 days for “red wine” with ≥32 mg/L SO2 (Tukey’s test, p < 0.05). Because the decrease in total SO2 was not significantly different between the 67 mg/L challenge treatment and the uninoculated control, this decrease was attributed to chemical oxidation, likely resulting from repeated opening of bottles for sampling. To avoid confounding effects of the loss of SO2 during the experimental course, data on yeast viability and survival at t = 10 days (T10) were not used.
“Molecular SO2” and molecular SO2 for the T0 to T9 time points in the white wine as a function of SO2 addition are shown in Figure 2 (left plot). The proportional increase in “Molecular SO2” and “Free SO2” for the lowest addition level (22 mg/L) was smaller than the increase observed with subsequent additions, presumably because of formation of adducts with SO2 binders present in the original wine, e.g., acetaldehyde or other carbonyls (Waterhouse et al. 2016b). “Molecular SO2” and molecular SO2 did not significantly differ for SO2 additions <50 mg/L, but molecular SO2 was significantly greater for the highest addition levels (56 and 67 mg/L). For the white wine, molecular and “Molecular” SO2 >0.8 mg/L could be achieved through addition of 67 mg/L SO2 (Figure 2), with similar results observed for “Molecular SO2” in the “red wine”. However, greater additions of SO2 (114 mg/L or more) were necessary to achieve >0.8 mg/L molecular SO2 in the “red wine” at T0 (Figure 2). The largest absolute difference between molecular and “Molecular” SO2 in the “red wine” was observed for an addition of 92 mg/L SO2, which resulted in molecular SO2 = 0.38 mg/L and “Molecular SO2” = 1.78 mg/L. Similarly, SO2 additions of 44 mg/L to the “red wine” were necessary to have >0.6 mg/L “Molecular SO2” (a typical winemaking target), compared with an SO2 addition of 104 mg/L to have >0.6 mg/L molecular SO2. “Molecular SO2” was also slightly higher (average = 0.2 mg/L “Molecular SO2”, or ~5 mg/L “Free SO2”) in the “red wine” than in the white wine for SO2 additions of 22 to 67 mg/L (paired t test, p < 0.05). The reason for this was unclear, but this possibly occurred because the added anthocyanins partially reacted with other SO2-binding nucleophiles in the time between addition and the SO2 measurement.
An analogous discrepancy was observed for “Free SO2” (Supplemental Figure 1C) and free SO2 (Supplemental Figure 1E), in that higher SO2 additions were necessary for the “red wine” than the white wine to achieve similar free SO2 levels, but no differences were observed among wine “Free SO2” values for the same SO2 addition. Total SO2 did not significantly differ between red and white wines at a given SO2 addition level (Supplemental Figure 1A).
Yeast survival at T0 to T9 (0 to 1920 sec) during the challenge study were determined by serial dilution and plating onto YM media. Changes in yeast counts (log cfu/mL) with respect to the 0 mg/L SO2 control were then calculated, and these changes were plotted as a function of time for each SO2 addition level (Figure 3). Data from T0 to T4 (0 to 63 sec) highly varied because of difficulties in quenching the SO2 quickly, so these data were averaged prior to plotting. SO2 additions were less effective in decreasing survival in “red wines” then in white wines. For example, in the white wine (Figure 3, left), decimal decreases (1 log or greater) were observed for all time points at or after 240 sec (T6) for SO2 additions ≥56 mg/L, and for ≥44 mg/L at the final time point (1920 sec, T9). However, in the “red wine” (Figure 3, right), SO2 additions of at least 104 mg/L were necessary to observe a decimal reduction for any time point. We observed an increase in viability for the 104 and 114 mg/L (but not 128 mg/L) “red wine” treatments between 480 and 1920 min, presumably because the yeast were initially rendered nonviable, but then recovered and grew at these intermediate SO2 concentrations.
Using the data in Figure 2, we identified red and white treatments with similar molecular SO2 and “Molecular SO2” (Table 3). For example, the 44 mg/L SO2 addition to white wine and 92 mg/L SO2 addition to “red wine” both yielded ~0.37 mg/L molecular SO2 and were treated as one pair, and the 44 mg/L SO2 addition to white wine and 32 mg/L SO2 addition to “red wine” yielded ~0.43 mg/L “Molecular SO2” as a different pair.
The maximum decrease in yeast survival occurred at the final time points, T6 to T9 (240 to 1920 sec, Figure 3). To evaluate whether differences in SO2 addition effects in red and white wines were statistically different, average log reduction at T6 to T9 was calculated for each pair listed in Table 3, and plotted as a function of either molecular SO2 or “Molecular SO2” (Figure 4). Pairs with similar molecular SO2 (Figure 4, left) did not differ significantly in survival between red and white wines, but pairs with similar “Molecular SO2” did differ significantly (paired t tests, p > 0.05). For example, a 2-log reduction in survival was observed for the 0.67 mg/L molecular SO2 red and white wine pair. The 0.67 mg/L “Molecular SO2” white wine also had a more than 2-log reduction in survival, but the “red wine” with 0.69 mg/L “Molecular SO2” had no significant decrease in survival.
SO2 treatments were then divided into categories of low (<0.5 mg/L molecular SO2 or “Molecular SO2”), medium (0.5 to 2.0 mg/L), or high (>2.0 mg/L). For both low molecular and “Molecular SO2”, we observed no significant decrease in the average survival at T6 to T9 for either wine type. For the medium SO2 levels, we observed a significant decrease (p < 0.05, t test) in survival compared with that in the control for the white wine based on either molecular or “Molecular SO2”, and for the “red wine” based on molecular SO2, but no significant decrease based on “Molecular SO2”. None of the treatments had molecular SO2 in the high category, but “red wines” with high “Molecular SO2” gave significantly lower survival than the control.
SO2-treated wines were also characterized by flow cytometry to selectively measure viable cells (as opposed to surviving cells measured by plating). Results for matched pairs of red and white wines were similar to those seen with survival, in that pairs with similar molecular SO2 gave similar decreases in viable cells compared with the untreated control, but pairs with similar “Molecular SO2” did differ significantly in viability (data not shown). SO2 treatments were again divided into categories (low, medium, and high), and average decreases at T6 to T9 in viable cells are shown in Figure 4 (right). Medium levels of molecular SO2 levels significantly decreased viability (p < 0.05, t test) in both red and white wines, but no significant decrease was observed for “red wines” with medium “Molecular SO2”.
Finally, we performed a confirmation challenge experiment with selected SO2 treatments (experiment B). The SO2 treatments were chosen to create the following wines:
Medium “Molecular SO2”, medium molecular SO2 white wine (74 mg/L SO2 added, 1.20 mg/L “Molecular”, 1.19 mg/L molecular).
High “Molecular SO2”, medium molecular SO2 “red wine” (142 mg/L SO2 added, 2.56 mg/L “Molecular”, 1.59 mg/L molecular).
Medium “Molecular SO2”, low molecular SO2 “red wine” (74 mg/L SO2 added, 1.33 mg/L “Molecular”, 0.27 mg/L molecular).
A control “red wine” and a control white, with no added SO2.
The challenge experiment lasted 960 min, and yeast survival was again determined by plating of serial dilutions. The effects of the treatments were determined by comparing changes in surviving cells (log cfu/mL) in a treatment with the unsulfited white control. The results were similar to those of the original experiment (data not shown): molecular SO2 >0.5 mg/L in either wine type, or “Molecular SO2” in the white wine >0.5 mg/L was sufficient to decrease survival compared with the control, but survival at molecular SO2 <0.5 mg/L, i.e., the “red wine” with 74 mg/L added SO2, was not significantly different from that in the control.
Discussion
We have recently reported that “Molecular SO2” of wines measured by conventional analytical approaches (e.g., A–O) that rely on initial acidification and dilution of the sample are often much higher than molecular SO2 values measured by headspace methods, e.g., HS-GDT, that do not perturb the sample (Coelho et al. 2015). This discrepancy also has been reported by other authors (Burroughs 1975, Davis et al. 1983), and is particularly severe for red wines. For example, we previously observed that molecular SO2 averaged only 32% of “Molecular SO2” values across a range of red wines. The different values measured between methodologies are likely due to dissociation of weakly bound anthocyanin-bisulfite adducts over the time course of a conventional SO2 analysis (Coelho et al. 2015). The key question investigated by our group was whether this weakly bound anthocyanin-bisulfite fraction has antimicrobial activity and, hence, whether conventional “Molecular SO2” analyses are more fit for purpose than nonperturbing molecular SO2 analyses. In principle, it could be possible to evaluate the efficacy of SO2 in a set of commercial white and red wines, but such a study could be confounded by differences in other parameters (e.g., pH, acetic acid, residual sugar).
To control for these differences, a “red wine” was prepared by adding a commercial anthocyanin extract to the white wine. Although other weak SO2 binders exist in wine, our recent work has indicated that the discrepancy between “Molecular” and molecular SO2 analyses can be largely explained by differences in monomeric anthocyanin concentration (Coelho et al. 2015). The resulting “red wine” had elevated monomeric anthocyanins at concentrations representative of a young red wine, and had low levels of tannins, but otherwise resembled the white wine (Table 1). SO2 addition resulted in nearly identical increases in “Molecular SO2” in both wine types, but in much lower levels of molecular SO2 in the “red wine” than in the white wine (Figure 2). Because the wines were of similar composition, besides higher anthocyanins and tannins, and because anthocyanins are well known to bind SO2, this difference in SO2 activity was presumed to result from binding of SO2 by anthocyanins. In support of this hypothesis, 78 mg/L SO2 needed to be added to the “red wine” to significantly increase molecular SO2 levels, compared with 32 mg/L to significantly increase “Molecular SO2”, a difference of 46 mg/L total SO2 (0.72 mM). This concentration represents 78% of the monomeric anthocyanin concentration (458 mg/L as malvidin-3-glucoside, or 0.92 mM), which is squarely within the range of monomeric anthocyanins reported to be bound in typical red wines (70 to 85%) (Usseglio-Tomasset et al. 1982).
Interestingly, the SO2 addition (56 mg/L) that led to a typical winemaking target “Free SO2” (~30 mg/L, Supplemental Figure 1C) resulted in a “red wine” with molecular SO2 below the analytical detection limit (<0.05 mg/L, Supplemental Figure 1D); thus, these concentrations were far below any typical winemaking recommendation. Although not tested in the current report, we would expect similar “Free SO2” for other methods that perturb equilibrium, e.g., the Ripper iodometric titration. An earlier study reported that free SO2 measured by capillary electrophoresis were up to an order of magnitude lower than “Free SO2” measured by the Ripper method or A–O (Bogren 1996). The use of a modified Ripper method to account for other nonreducing substances (Zoecklein et al. 1999) is not expected to improve agreement with free SO2 methods, since the initial acidification step in Ripper would still perturb equilibrium.
S. cerevisiae was used for the challenge studies in sweet wines because of the well-known importance of using SO2 to prevent refermentation, and because there are multiple literature reports on recommended SO2 levels for S. cerevisiae control in both wine and defined media (Beech et al. 1979, King et al. 1981, Sudraud and Chauvet 1985). The EC1118 strain was selected because of its widespread use in international winemaking (Hornsey 2007). The results of our challenge studies with S. cerevisiae on the “red” and white wines demonstrate that although yeast survival decreases with increasing total SO2 level, the efficacy of SO2 is lower in “red wine” than in white wine for the same total SO2 addition (Figure 3). Using conventional SO2 analysis approaches (A–O), we could not easily explain these differences in survival, since apparent “Molecular SO2” at a given total SO2 was similar in both the red and white wines (Figure 2). However, the molecular SO2 concentration determined by HS-GDT was much lower in the “red wine” for a given total SO2 addition (Figure 2). We used two statistical approaches to evaluate whether molecular SO2 was a better predictor of antimicrobial activity than “Molecular SO2”. First, we identified pairs of “red” and white wines similar in either molecular SO2 or “Molecular SO2”. No significant difference was observed in survival between molecular SO2 pairs, but a significant difference was observed for “Molecular SO2” pairs (Figure 4). Similar results were observed for viability measured by flow cytometry.
Second, we classified the SO2 treatments according to the measured molecular or “Molecular” SO2, and compared antimicrobial activity among these treatments. These categories were based on the broad range of “target” molecular SO2 levels suggested in the literature. One report states that ~1.5 mg/L molecular SO2 was required for 4-log reductions in S. cerevisiae in media after 24 hr (King et al. 1981), another paper recommends at least 0.9 mg/L to prevent yeast growth in sweet wines (Sudraud and Chauvet 1985), and yet another recommends a range of 0.5 to 0.9 mg/L for controlling yeast spoilage (Beech et al. 1979). On the basis of these reports, we categorized wines with 0.5 to 2.0 mg/L molecular or “Molecular” SO2 as having “medium” levels that covered the broad range suggested in the literature, with wine treatments having <0.5 mg/L or >2.0 mg/L SO2 classified as “low” or “high”, respectively. We observed that molecular SO2 >0.5 mg/L results in a significant and greater than 2-log reduction in both viable and culturable cells in “red” and white wines (Figure 5). This result was in good agreement with the minimal concentrations suggested in the previous reports. By comparison, to yield an antimicrobial effect, high concentrations of “Molecular SO2” were necessary in the “red wine”. A follow-up confirmation challenge study (experiment B) gave results similar to those in experiment A: a red wine with “Molecular SO2” >0.5 mg/L, but molecular SO2 <0.5 mg/L, showed no decrease in yeast survival, but the red and white wines with both “Molecular SO2” and molecular SO2 >0.5 mg/L had a 2-log decrease in survival.
Taken together, the data from experiments A and B suggest that anthocyanin-bisulfite complexes measured by conventional SO2 analyses like A–O have little (if any) antimicrobial effect. To our knowledge, this is the first demonstration that the anthocyanin-bisulfite adducts have negligible activity in wine. Our results contradict the observations of Usseglio-Tomasset and colleagues, who reported that anthocyanin-bisulfite complexes (30 mg/L as SO2) delayed fermentation by S. cerevisiae and other yeasts almost as much as free SO2 (Usseglio-Tomasset et al. 1982). Because molecular (or “Molecular”) SO2 concentrations were not measured after addition or during fermentation in this earlier work, it is unclear whether the SO2 stayed in the anthocyanin-bisulfite complexes, or whether molecular SO2 increased because of reactions that would consume anthocyanins.
Reports that have attempted to establish appropriate molecular SO2 levels for control of S. cerevisiae spoilage using white wines, model wines, or media would not have observed any effect due to anthocyanin binding of SO2. However, several studies on SO2 inhibition of B. bruxellensis in red wine are reported in the literature. Brettanomyces control is of considerable interest to the wine industry because of its high ethanol tolerance and low nutrient requirements, and because it can convert vinylphenols to more potent and malodorous ethylphenols (yielding “Band-Aid”, “horse”, or “spicy” odor) (Smith and Divol 2016). Brettanomyces can be inhibited by molecular SO2 (Smith and Divol 2016), but requisite concentrations appear to vary not only by strain, but also between media and red wine. For example, challenge studies in red wines evaluating multiple Brettanomyces strains have reported that most of these strains are culturable at 0.4 mg/L “Molecular SO2”, and sometimes up to 1.0 mg/L (Barata et al. 2008, Zuehlke and Edwards 2013). However, a large survey of SO2 tolerance among Brettanomyces strains grown in media revealed that most strains could grow only at below 0.2 mg/L “Molecular SO2” (Vigentini et al. 2013). A separate did not survive (>4-log reduction) at 0.4 mg/L “Molecular SO2” (Curtin et al. 2012). Another report using model wines indicated growth inhibition at 0.3 mg/L “Molecular SO2 and 10% ethanol for all but one of 18 Brettanomyces strains (Duckitt 2012).
Although direct comparisons among results are complicated by varying methodologies for evaluating SO2 antimicrobial activity, the higher levels of “Molecular SO2” necessary for Brettanomyces control in red wines as compared to model wine or media reported in the literature could be explained by the fact that conventional “Molecular SO2” analyses overestimate the true molecular SO2 fraction. In a recent survey of commercial wines, we observed that “Molecular SO2” ranged from 0.2 to 1.2 mg/L (Coelho et al. 2015). In the same work, using HS-GDT, we observed that only two of nine red wines had molecular SO2 >0.2 mg/L. On the basis of the slope of the regression line for HS-GDT versus A–O data (0.32) (Coelho et al. 2015), we can surmise that on average, the molecular SO2 value was only 32% of the “Molecular SO2” value in red wines. By comparison, 17 of 18 white and blush wines in this previous study had molecular SO2 >0.2 mg/L, and the molecular SO2 was only 82% (slope of regression line = 0.82) of the “Molecular SO2” value. As demonstrated in our current work, molecular SO2 is a much better predictor of antimicrobial activity against Saccharomyces than “Molecular SO2”. Brettanomyces spoilage is much more common in barrel-aged red wines than barrel-aged white wines, a phenomenon often attributed to the lower pH of white wines favoring the molecular SO2 form (Oelofse et al. 2008). However, our work suggests that an additional issue is that conventional “Molecular SO2” measurements overestimate the amount of active molecular SO2 in red wines. This hypothesis was not evaluated in our current work, but should be considered in future studies.
As a caveat, previous reports that used red wines in microbial challenge studies did not report the monomeric anthocyanin concentration of the wines used. Monomeric anthocyanin concentrations in red wines decrease to near undetectable concentrations within the first few years after fermentation because anthocyanins react with other wine components, leading to formation of a range of pigmented and nonpigmented products, e.g., pyranoanthocyanins and tannin-anthocyanin adducts, that are less able to bind SO2 (Waterhouse et al. 2016a). Thus, in older red wines, as opposed to young red wines or “red wines” produced from white wines spiked with anthocyanins, we expect that “Molecular” and molecular SO2 concentrations should be closer to each other, and a reasonable prediction of SO2 antimicrobial activity can be achieved based on conventional “Molecular SO2” analyses.
Our current study evaluated the antimicrobial activity of molecular SO2 compared with that of “Molecular SO2”. However, in winemaking, free SO2 in the form of HSO3− also serves as an antioxidant, capable of reacting with major oxidation products (acetaldehyde and related carbonyls, quinones, or H2O2) (Waterhouse et al. 2016b). Potentially, “Free SO2” measurements may also overestimate the antioxidant activity of wines, analogous to the situation with “Molecular SO2”. Waterhouse et al. (2016a) recently observed that lees-aged wines exposed to large amounts of air during bottle storage develop oxidized aromas despite having ~10 mg/L “Free SO2” (measured by iodometric titration) at the end of the storage period. By comparison, wines aged without lees had lower “Free SO2”, but did not have oxidized aromas. The authors speculated that this may have been due to the presence of lees-derived weak SO2 binders that were detected as “Free SO2”, but did not prevent formation of oxidized off-aromas. Based on our current work, this is a reasonable hypothesis to investigate.
Conclusion
Using a “red wine” produced by spiking a white wine with anthocyanin extract, we have demonstrated that anthocyanin-bisulfite complexes have negligible antimicrobial activity. These complexes are measured as part of conventional “Free” and “Molecular SO2” analyses, such as A–O and flow-injection analysis. Our results indicate that conventional approaches to measuring “Molecular SO2” are ill-suited for predicting the microbial stability of red wines, particularly of younger red wines that are rich in monomeric anthocyanins. Future work is necessary to determine whether conventional SO2 analyses also overestimate protection against other spoilage organisms (e.g., Brettanomyces) and oxidation in red wines.
Acknowledgments
The authors gratefully thank ETS Laboratories, St. Helena, CA, for collaboration and analytical support. Additional thanks to Scott Laboratories (Petaluma, CA) for the donation of yeast and bacteria and to Polyphenolics (Madera, CA) for the donation of anthocyanin powder. Funding was from the Peter and Tacie Saltonstall Endowment and the New York Wine & Grape Foundation.
Footnotes
The authors have no conflicts of interest.
Supplemental data is freely available with the online version of this article at www.ajevonline.org.
- Received March 2017.
- Revision received October 2017.
- Revision received January 2018.
- Accepted February 2018.
- Published online June 2018
- ©2018 by the American Society for Enology and Viticulture