Abstract
Studies have yet to evaluate how bentonite properties may affect the protein profile, polyphenol content, metal concentration, and heat stability of a white wine at different pH values. Therefore, this work assessed the proteins, polyphenols, metals, and haze forming tendency when heating white wine samples before and after a fining treatment with four activated sodium bentonites in a typical wine pH range (3.00 to 3.60). Soluble wine proteins were separated by sodium dodecyl sulfate - polyacrylamide gel electrophoresis, and gel images were compared using the Quantity One software package (Bio-Rad Laboratories, Inc., Hercules, CA). The wine haze forming tendency, metals, and polyphenols were measured using heat tests and International Organisation of Vine and Wine (OIV) methods. Low molecular mass proteins were efficiently removed by all of the bentonites, regardless of the pH. High and medium molecular mass proteins were less likely to be removed and the efficiency, which depended on the pH, was variable. Reductions of vacuolar invertase (GIN1) and VVTL1 fractions of the thaumatin-like proteins were induced by bentonites with pH values less than 10. These bentonites were affected to a lesser extent by the negative effect of acidic pH. The reduction in haze forming tendency of the unfined Erbaluce wine was particularly noticeable in bentonite fined samples heated at 50 to 60°C, 60 to 80°C, and 70 to 80°C at pH 3.17, pH 3.30, and pH 3.60, respectively. The poor removal of glycoproteins (YGP1 and Hmp1) at higher pH values contributed to an increased thermal stability. The exchange of cationic species, notably sodium and potassium, between the bentonites and the wine was related more to the wine pH than to the clay type. Finally, the extent of polyphenol removal correlated with the amount of protein removed. When protein removal did not occur, the reduction of polyphenols was driven by the specific surface area and the surface charge density of the bentonite.
During bottle storage, occasional temperature extremes may lead to protein aggregation and flocculation, resulting in turbidity of wine. A haze or deposit in bottled wine can reduce or invalidate its commercial value, and winemakers typically perform fining treatments to avoid wine turbidity. The major contributors to natural white wine haze that have been identified include β-glucanases (Esteruelas et al. 2009), class IV chitinases, and thaumatin-like proteins (TLP; Waters et al. 1996). Wine instability may also be influenced by non-protein factors, including the wine pH, ionic strength, ethanol content, and the concentrations of polysaccharides, polyphenols, sulfates, and organic acids (Van Sluyter et al. 2015). In a recent experiment (Lambri et al. 2013a), even small variations in wine pH may have increased the solubilization of glycoproteins, particularly highly glycosylated yeast proteins, after the dissolution of the protein–tannin interactions. These yeast glycoproteins, which have been described in the past as “haze-protective” (Waters et al. 2005), clearly modify the haze forming tendency of the wine (Fusi et al. 2010), and their release induces modifications in wine heat stability (Lambri et al. 2013a).
Although the range of innovative methods to avoid protein-driven visual changes in white wines increases every year, including enzymatic, mineral, and physical approaches (Van Sluyter et al. 2015), the use of bentonite (a 2:1 smectite that is primarily montmorillonite) is still considered the best method for the prevention of white wine protein haze in the wine industry (Chagas et al. 2012). The increasing consumer interest in less processed wines (IFOAM 2003) moves the technicians toward the use of fining agents that do not represent any potential risk for consumer health. In this regard, the use of a mineral-based adjuvant such as bentonite is of potential interest for stabilizing the colloidal state both of the white and the red wines (Dordoni et al. 2015). Previous studies have demonstrated (Lambri et al. 2012a) how different bentonite labels can selectively remove specific proteins that are responsible for turbidity upon the heating of white wine.
To improve protein removal, bentonites are activated by the partial substitution of Ca2+ by Na+ (i.e., activation that improves the swelling properties of the clay; Catarino et al. 2008). The negative net charge of the clay is partially balanced by exchangeable cations, mainly Ca2+, K+, and Mg2+. These ions are localized within the interlayer space and on the external surface of the clay particles, and can be exchanged with the medium (i.e., wine). The limits of the exchangeable substances that can be released in wine by bentonite have been set by Resolution Oeno 11/2003 (OIV 2003). Additionally, several countries require that for every new bentonite used, the possible increase in the concentrations of calcium, magnesium, iron, arsenic, and lead in wine should be analyzed (Castiñeira Gómez et al. 2004). Despite the importance of evaluating elemental exchanges between bentonite and wine for safety and stability purposes, only a few reports studied the influence of bentonite on the elemental composition of wine in recent years (Castiñeira Gómez et al. 2004, Nicolini et al. 2004). The side effects of bentonite treatment include the depletion of phenolic (Gonzáles-Neves et al. 2014, Salazar et al. 2006) and aroma compounds (Lira et al. 2015) that are coupled with protein removal (Lambri et al. 2010), and through adsorption phenomena (Lambri et al. 2013b). The removal of phenolic compounds, the exchange of metals between wine and bentonite, and the modifications of the protein profile and haze forming tendency during fining with bentonites with different characteristics in a typical wine pH range may have a substantial effect on enological practices. An understanding of the bentonite characteristics that can improve wine stability, minimize polyphenol removal, and minimize metal perturbations would be useful for the wine industry. To assess important attributes of bentonite, this paper estimates the proteins, polyphenols, metals, and haze forming tendency when heating white wine samples before and after a fining treatment. The fining treatment employed 1 g/L of each of four activated sodium bentonites in a typical wine pH range (3.00 to 3.60). The work also evaluated whether a release of metals from the bentonite or a removal of metals from the wine occurred with respect to the maximum contents of extractable iron, sodium, and combined calcium and magnesium expressed by Resolution Oeno 11/2003 (OIV 2003) and with the legal limits reported in the Compendium of International Methods of Analysis (OIV 2011).
Materials and Methods
Chemicals.
All chemicals were of high purity and were commercially available. The products used for wine pH adjustments and analyses were provided by Carlo Erba (Carlo Erba Reagenti S.p.A., Arese, Italy). All reagents and equipment used for electrophoresis were purchased from Bio-Rad (Hercules, CA) and were all of proteomic grade, unless otherwise stated.
Bentonite samples and analyses.
Four samples of activated sodium bentonite were purchased from Dal Cin Gildo S.p.A. (Concorezzo, Italy): Granular W (GW), Top Gran (TG), Powder N (PN), and Powder W (PW). The bentonite samples were analyzed in triplicate. Elemental analyses of the inorganic content were conducted with an energy dispersive X-ray detector (EDS-EDAX Genesis; EDAX, Inc., Mahwah, NJ) coupled to a scanning electron microscope (XL30 Esem; Philips, Eindhoven, Netherlands). The surface charge densities were measured using the procedure previously described by Ferrarini et al. (1996). The specific surface area (SSA) was determined according to COEI-1-Benton:2011 (OIV 2013). The pH and the swell index were measured using the methods described by Resolution Oeno 11/2003 (OIV 2003). The charge density per surface unit was calculated as the ratio between the surface charge density (meq/100 g) and the SSA (m2/100 g).
Wine samples and pH adjustments.
The Erbaluce wine was characterized as reported (Lambri et al. 2013a). The wine was produced from Vitis vinifera L. Erbaluce grapes harvested at commercial maturity, following winemaking practices standardized by the Erbaluce di Caluso Controlled and Guaranteed Designation of Origin specifications. The resulting wine was not aged in oak, and lees contact and/or any additional fining treatments before the tests were avoided. The wine was divided into eight aliquots. Two specimens were unmodified (and their pH was determined), whereas the other aliquots were subjected in duplicate to pH adjustments to 3.00, 3.30, and 3.60 using 1 mol/L HCl or 1 mol/L NaOH.
Wine fining with bentonite labels.
Laboratory-scale trials were conducted. For each bentonite label, two sets of samples were prepared as described (Lambri et al. 2012a). The bentonite slurries were prepared in deionized water at a concentration of 10% (w/w). After 90 min rehydration, the gels were stirred; then they were added at a single dose of 1 g/L to 4 L of wine and thoroughly mixed. Four liters of untreated sample was maintained in duplicate as a control. The bentonite dose of 1 g/L was selected by taking into account different issues. First, the COEI-1-BENTON:2003 (OIV 2003) requires the use of bentonite doses from 0.1 to 0.8 g/L for the protein adsorption test trial. Moreover, the use of higher or of lower doses involves or produces excessive and undesired bentonite side effects (Van Sluyter et al. 2015), or does not allow sufficient protein and heat stability of the wine (Hsu and Heatherbell 1987, Pocock and Waters 2006, Lira et al. 2015). The dose of 1 g/L has been reported to result in the removal of 85% of the protein content (Sauvage et al. 2010) without severely affecting the volatile profile of the wine (Lambri et al. 2010, 2012b). After the addition of the bentonite slurry, all the samples and controls were transferred to glass demi-johns of identical shapes and dimensions. They were maintained for five days in a room at 16 to 18°C and 60% relative humidity. The clear liquid phases were then separated and filtered under nitrogen pressure through folded 595½ filters (Whatman; GE Healthcare Europe GmbH, Milan, Italy). The control samples were treated and filtered under the identical conditions as the fined samples, but without the addition of the bentonite.
Protein analyses.
Proteins were precipitated as reported in Lambri et al. (2012a) by the addition of 400 mL of absolute ethanol to 100 mL of wine, and the samples were left for 72 hr at 4°C. The samples were then centrifuged for 20 min at 4470 × g. The protein pellets were suspended in MilliQ water and then dialyzed overnight at 20°C in 3500 Da cut off tubes (Membrane Filtration Products, San Antonio, TX). Dialysis against water eliminates amino acids and small peptides from the sample (Moreno-Arribas et al. 2002). The dialyzed samples were lyophilized and then resuspended in 10 mL of water before being quantified by the Schacterle and Pollack method (1973) with bovine serum albumin as a standard. The assay procedure, a modification of the Lowry method, entailed mixing the sample with alkaline copper and phenol regents; the solution was then heated at 55°C, cooled, and absorbance determined at 650 nm. This assay exhibits the best accuracy with regard to absolute protein concentrations due to the chemical reaction with polypeptides and the quantitation of oligopeptides (Moreno-Arribas et al. 2002).
Electrophoresis was done as previously described (Lambri et al. 2012a). Eight μL from each resuspended protein pellet sample was mixed with 10 μL of 2X Laemmli sample buffer (Laemmli 1970) with dithiothreitol as the reducing agent. The samples were then loaded onto precast 10-well 8 to 16% gradient polyacrylamide gels. The Low Molecular Weight SDS Marker Kit (GE Healthcare Europe GmbH, Milan, Italy) (3 μL) was used as the molecular weight standard. Sodium dodecyl sulfate - polyacrylamide gel electrophoresis (SDS-PAGE; Laemmli 1970) was run using a Mini Protean 3 Cell at 200 V, 120 mA, and 25 W until the dye front reached the bottom of the gel (~70 min). The gels were stained with a colloidal solution of Coomassie Brilliant Blue (CBB) G-250 following Neuhoff’s method, with a modification (0.12% CBB G-250; Neuhoff et al. 1988). Each gel was run in duplicate. Images of the CBB-stained gels were digitized using a VersaDoc device and analyzed using the Quantity One 4.6 software package (Bio-Rad Laboratories, Inc., Hercules, CA). The relative intensity of each protein band in each lane was calculated as the symmetrized percent change (SPC) of the band intensity with respect to the unfined wine at the identical pH (Berry and Ayers 2006).
Polyphenol and heat stability analyses.
Wine polyphenols were measured as absorbance at 280 nm and quantitatively determined using a calibration curve with catechin as a standard, as reported in Ribéreau-Gayon et al. (2006). The heat stability of the wine before and after bentonite treatment was determined following the procedure of Pocock and Rankine (1973) as applied in Lambri et al. (2012b). All measurements were taken in triplicate, and the samples (5 mL each) were saturated with nitrogen and sealed in test tubes with screw caps. The tubes were heated to 30, 40, 50, 60, 70 or 80°C in a water bath for six hr. Then, the tubes were kept at 4°C for 16 hr and allowed to warm to room temperature. The amount of haze formed was estimated visually by comparison to an unheated control sample. When the level of haze is visible, it is possible to set it as an absorbance limit over which the sample is considered heat-unstable (Pocock and Waters 2006). The absorbance was detected by a spectrophotometer (Shimadzu UV-1601 spectrophotometer; Shimadzu Europe, Duisburg, Germany) at 540 nm and 25°C in 1-mL plastic cuvettes.
Metal analyses.
An Analyst 100/300 Perkin Elmer Atomic Absorption Spectrometer (AAS) with continuum source background corrector and FIAS AS-800 auto sampler was used. Atomization was achieved electrothermally using an HGA-800 graphite furnace (ETAAS). AAS was employed with air/acetylene (10/1.5) flames. The wine samples were treated with hot HNO3–H2O2 to decompose the organic matrix. For each sample, 25 mL of wine was transferred to a Kjeldahl flask. Then, 5 mL of the certified HNO3 (63%, d = 1.43 g/mL) and 5 mL of H2O2 were added to the flask, and the mixture was boiled for ~30 min or until colorless. This solution was then transferred to a 50 mL flask and diluted to 50 mL with deionized water at 0.55 mS/cm conductivity before the samples were injected into the AAS. Calibration curves (r2 ≥ 0.999) were prepared before analysis with standard commercial stock solutions (Merck, Darmstadt, Germany). During analysis, 5 to 10% of the samples were randomly selected and analyzed six times to examine the reproducibility of the results.
Statistical analysis.
Protein measurements were performed in two analytical replicates on two technical replicates, for a total of four replicate measurements. The heat stability test, polyphenol tests, and metal analyses were performed using analytical triplicates on each technical duplicate for a total of six replicates. The data were processed by a factorial ANOVA at p ≤ 0.05. Significant differences were detected using a post-hoc Tukey’s test. The statistics were performed in Kyplot 2.0 statistical software (Kyens Lab Inc., Tokyo, Japan) and SPSS Statistics 21 (IBM Corp., Armonk, NY).
Results and Discussion
Characteristics of bentonite samples.
In this work, four natural calcium bentonites were tested. The nature of these bentonite samples was changed through a commercial activation process, which enriched the natural calcium-dominant clay with sodium. This treatment can lead to differences (Lambri et al. 2010) or similarities (Lambri et al. 2012b) in the resulting elemental composition, depending on the types of raw clay and the modification conditions (Catarino et al. 2008). The elemental composition of the tested bentonites is shown in Table 1. No significant differences in the levels of sodium, magnesium, or potassium among the bentonite samples were observed. The amount of iron was significantly higher in PN; PN also had the greatest surface charge density, the largest SSA and swell index, and the highest pH. Conversely, the PW sample showed the lowest SSA value and a significantly higher charge density per surface unit area. Finally, the TG and GW samples demonstrated similar SSAs and surface charge densities and, therefore, similar charge densities per surface unit area. The ratios of sodium to calcium were similar among the bentonites TG (1.14), PN (1.20), and PW (1.03), whereas bentonite GW, because of its lowest content of calcium, showed a relatively high ratio of sodium to calcium (2.41). The PN bentonite tested in this study differed from the others in most of the physicochemical characteristics investigated: it accommodated more polyferric iron and showed a significantly higher swell index value than the other clays (Table 1), but these characteristic features did not positively influence protein adsorption (Table 2). As shown in Table 1, all pH values we observed were basic and over the isoelectric point (pI), which is usually assumed to be around pH 7 (Benna et al. 1999). Specifically, the highest pH value was observed for PN, a slight but significantly lower pH was observed for PW, and finally, pHs below 10 were observed for TG and GW (Table 1). In bentonite samples having a pH > pI, the likelihood of electrostatic attraction between layers in the house of cards-like structure progressively decreases as the pH increases (Benna et al. 1999). As a consequence, in PN and PW samples showing pH > 10 (Table 1), the interactions among the particles can occur only by electrostatic repulsion, which indicates edge-to-edge, edge-to-face, and face-to-face repulsion.
Metal exchange between bentonites and wine.
Metals in wine occur at mg/L concentrations or less. Although not directly related to the taste of the final product, their content should be determined because excess metals are undesirable and, in some cases, restricted by legislation to guarantee the protection of consumer health (Ribéreau-Gayon et al. 2006). Five elements were determined in wines at different pH values before and after bentonite treatment (Table 3). The concentrations detected fall within the range typical of wines (OIV 2013) and none were above the limits established by the Office International de la Vigne et du Vin (OIV; OIV 2011). The iron in wine was significantly augmented after bentonite treatment with GW, TG, and PN, irrespective of the pH of the wine. Fining with PW bentonite did not increase the iron content in the wine. The magnesium content in wine was slightly affected by bentonite treatment irrespective of the pH. The clays GW and TG produced slight, but significant increases in the magnesium in wine at pH 3.00 and 3.17. The magnesium concentration significantly diminished only after treatment with TG at pH 3.60. The calcium concentration increased after treatment with all the bentonites at acidic pH (3.00 and 3.17). In the fined wines using PN and PW at pH 3.30, significantly lower amounts of calcium were recorded than in the unfined wine. At pH 3.60, the calcium amounts were significantly lower after fining, notably with PW. The level of potassium diminished after fining according to the increase in pH, except for the treatment with TG at pH 3.30. The only increase in potassium concentration was observed after GW treatment at pH 3.00. Finally, with respect to the original wine, the content of sodium was increased after each bentonite treatment at pH 3.30 and 3.60.
The quantities of iron, magnesium, calcium, sodium, and potassium released in wine were determined from the data in Table 3 to consider possible excessive releases by bentonites. These data were then compared with the limits determined by the Resolution Oeno 11/2003 (OIV 2003) and with the maximum acceptable limits of elements contained in wine reported in the Compendium of International Methods of Analysis (OIV 2011). Notably when GW was used in the pH range 3.00 to 3.30, the iron enrichment in the wine was above the limit of 0.6 mg released by 1 g of bentonite (Table 3). This result agreed with Enkelmann (1988) who observed the release of considerable amounts of iron into tartaric acid by eight out of 10 commercial bentonites. The release of iron observed for GW, TG, and PN was not strongly influenced by the pH in the range tested but depended mainly on the clay label. This finding was difficult to explain by evaluating the physicochemical parameters of the clays (Table 1) and were not in accordance with a previous investigation performed on bentonites with a high smectite content (Catarino et al. 2008). These authors reported that pH 3.32 promoted iron release in wine from bentonites having the highest content of both constitutive and non-crystalline iron. The release of constitutive metals such as iron is affected by a series of factors not yet fully understood; therefore, other authors stated that iron content and that of other elements did not seem to be altered by the addition of bentonite (Castiñeira Gómez et al. 2004). Limits of 1 meq/1 g for the release of calcium and magnesium (OIV 2003) were exceeded by GW, PN, and PW at pH 3.00. At that pH, 1.63 meq/g of GW, 1.17 meq/g of PN, and 1.60 meq/g of PW were dissolved in wine (data calculated from the results in Table 3). GW, PN, and PW increased the wine magnesium concentration by 1.13, 0.52, and 0.69 meq/L, respectively. Although the magnesium clay concentrations were not significantly different, these labels had a high magnesium content (Table 1). Regarding calcium, wine concentration increased by 0.50, 0.65, and 0.90 meq/L for GW, PN, and PW, respectively (data calculated from the results in Table 3), but compositional data (Table 1) did not explain this result. Both calcium and magnesium were exchanged at a lower rate by bentonites as pH increased (Table 3). This finding contrasted with the findings of Catarino et al. (2008), who observed a non-significant effect of pH on the wine magnesium content and a higher concentration of calcium when pH was 3.58 with respect to pH values of 2.94 and 3.32. By contrast to our results, Catarino et al. (2008) found a positive correlation between the magnesium and calcium content of the clays and the increased magnesium and calcium concentration of the wine. The sodium released was below the accepted OIV limit (OIV 2003) for all clays irrespective of the pH (data calculated from the results in Table 3). The more basic pH values (3.30 and 3.60) induced higher release of sodium, producing significant increases in the wine concentration of sodium at pH 3.60 after each bentonite treatment. Catarino et al. (2008) did not report a correlation between the sodium released in the wine and the pH or sodium clay content. Potassium decreased in wines after bentonite treatment (Table 3). This decrease was enhanced as the pH increased. By contrast, Catarino et al. (2008) reported the lowest potassium concentration at the lowest pH. Similar to our work, they did not discover a relationship between the sodium content of the clay and the behavior of the clay in wine.
To describe the real ion balances and fluxes that occurred in the fined wine samples, the mmol/L concentration of each element was calculated (Table 4) from the data presented in Table 3. The exchange of cationic species between the clays and the wine seemed to be related both to the pH and the clay label. Considering the total amounts of sodium and potassium ions involved at different pH values, acidic pH reduced the exchange between wines and bentonites, whereas higher pH (in this case, pH 3.60) exchanged higher quantities of potassium and sodium between the wine and clay. At lower pH levels, the release of sodium could depend mainly on the concentration, because at pH 3.00 the removal of potassium was less than at higher pH. The substitution of sodium with potassium reduces the hydration energy and enhances the van der Waals energy because potassium can fit into the ditrigonal cavity of the silicate layer (Bergaya et al. 2013). Catarino et al. (2008) suggested that the chemical composition of the type of clay could be important in determining the extent of ionic exchange. In addition to the selectivity of the smectites for larger over smaller cations (i.e., for potassium over sodium), other factors that influence the ion exchange are the cation hydration, electrostatic interactions, entropic effects, and oxidation of the structural iron. The effect of the pH is associated with the concentration of protons in the medium and with the relative ionic strength generated. Additionally, the affinity of protons for silicate structures is higher than that of potassium (Bergaya et al. 2013). The reduced competition of protons at pH 3.17, 3.30, and 3.60 allowed more potassium to be adsorbed, confirming the results of Nicolini et al. (2004). With the affinity of potassium for aluminum silicate structures and the competition with protons (Catarino et al. 2008), a strong concentration gradient of ‘wine versus bentonite’ seemed to be responsible for the observed removal of potassium. Decreasing quantities of potassium in wine improve the tartrate stability of the product because smaller amounts of potassium raise the solubility of potassium hydrogen tartrate (Ribéreau-Gayon et al. 2006). In addition, the release of sodium in the medium allows the formation of soluble sodium tartrate and sodium hydrogen tartrate that further improves the stability and minimizes the tartrate casse. Finally, considering that the pH influences the concentration of hydrogen tartrate in wine (Lambri et al. 2014), the removal of the hydrogen tartrate anion by ionic interactions with the positively charged edges of the clay could be important. Notwithstanding these hypothetical possibilities, further studies are required for a complete assessment of the risk, and influence of, bentonite on tartrate casse.
Relationships between protein removal, wine heat stability, and bentonite characteristics.
When the bentonite slurry is added to the wine (an acidic medium), the sedimentation of aggregates (which are likely partially disrupted) are favored. This sedimentation is optimal for enological purposes aimed at obtaining a clear supernatant in a few hours. As a rule, the acidic conditions tested in this study represent a limitation for the adsorption performance of a clay-water suspension because under acidic pH, the structure of the clay is likely disrupted and the house of cards-like structure gives an aggregated structure as a result of attractive interparticle forces (Benna et al. 1999). The change in pH from the natural water-clay suspension to that of the wine samples changes the ionic strength. This change in ionic strength contributes to the compression of the double layers and reductions in the dominance of the edge-to-face attraction, leading to the breakdown of the house of cards-like structure (Benna et al. 1999). This structural change could be particularly noticeable for PN, which has a large SSA and a natural strong basic pH (Table 1). However, this structural change may be reasonably considered to occur for all of the bentonites considered here. Moreover, because bentonite shows a strong preference for hydrogen, a reduced competition between hydrogen ions and proteins occurs in wines with less acidic pH, conditions under which the proteins are less cationic. Indeed, different pH values of wines modified the efficacy of bentonite deproteinization (Lambri et al. 2010) and the specific protein removal (mg protein/g bentonite) was higher in wines with higher pH and lower in wines under more acidic conditions (Lambri et al. 2012a).
Specifically, at a more acidic pH (3.17 and 3.00), none of the bentonites significantly differed from the others in protein removal behavior (Table 2). At the slightly higher pH (3.60), TG and GW, which have similar SSAs and natural pH values <10 (Table 1), behaved similarly and showed a significantly higher protein removal capacity than PN and PW (Table 2). Generally, protein removal is driven by adsorption phenomena in which the surface charges play a fundamental role. At a more acidic pH (3.00), the wine proteins are protonated and organized in small micelles and small aggregates with tannins (Lambri et al. 2014), i.e., conditions under which electrostatic interactions between the proteins and the bentonite result in protein removal. Furthermore, at strong acidic pH, the house of cards-like structure of the bentonite is negatively affected and the residual negative surface charge is reduced, resulting in better protein removal by those clays with a minor collapse into double layers that carry a certain density of negative charges (Table 2). A similar situation was observed at pH 3.17 where a decrease in the net positive charge of proteins resulted in a less effective removal of proteins. At higher pH (3.30 and 3.60), the double layers of the bentonite are covered with a greater number of negative charges, and the charge interaction with less cationic proteins is therefore less favored. The two bentonites that were effective are those that were more likely to retain a residual house of cards-like structure with a minor collapse into double layers.
The protein profiles of the Erbaluce wine samples treated with the four bentonites at different pH values are reported in Figure 1. The protein profile of unfined wine at the original pH (3.17) is also shown, and is enlarged to indicate the band numbers as defined in the image analysis. The identification of the proteins in each band was described in a previous report (Lambri et al. 2012a) and is summarized in Table 5. A statistical analysis of the SPC of the band intensities (Table 6) highlighted significant differences among the bentonite-treated and the unfined wine samples, and different behaviors among the different bentonite labels at varying pH. Taking into account the differences (Moreno-Arribas et al. 2002) between the Lowry-Folin method modified by Schacterle and Pollack (1973) used to detect the absolute protein concentration (Table 2) and the CBB SDS-PAGE of protein fractions with molecular weight ranging from 14 to 97 kDa (Figure 1, Tables 5 and 6), it seems that bentonite fining decreased the absolute protein content (~30% in the most drastic decrease), but it had completely removed most of the protein bands. The quantitative nature of the determination of total proteins, even if carried out on purified samples after dialysis with 3 kDa cut off membranes, also allows measurement of oligopeptides and polypeptides, which leads to overestimates of the protein content as compared with the CBB analysis (Esteruelas et al. 2009). On the other hand, extremely faint bands seen by SDS-PAGE were observed after adsorption on bentonite of the proteins contained in wine (Hsu and Heatherbell 1987, Sauvage et al. 2010). Moreover, when bands were quantified separately, the standard deviation may be dependent upon the concentration of each protein in the original wine: a range from 10% for the band corresponding to the most concentrated protein fraction to 25% for the bands corresponding to the less concentrated protein fractions (Sauvage et al. 2010). The results in Table 6 are reported as a qualitative determination of band intensity calculated as the SPC of the band intensity with respect to the unfined wine at the identical pH (Berry and Ayers 2006).
In the Erbaluce wine protein profile, no chitinases were detected as reported for Chardonnay (Okuda et al. 2006) and Sauvignon wine (Esteruelas et al. 2009). Additionally, similar to results obtained by Yokotsuka and Singleton (1997), invertase GIN1 was contained in several protein bands and glycosylated proteins were identified (Figure 1 and Table 5). Exo-β-1,3-glucanase and VVTL1 from TLP were reported to be found in naturally occurring haze that had formed in Sauvignon wine (Esteruelas et al. 2009). The high and medium molecular weight protein bands were removed with variable efficiency by the different bentonite labels and there appears to be a differential behavior in response to differences in the wine pH (Table 6) as found for bands 1, 2, 3, 4, and 6, which included the grape vacuolar invertase (GIN1), yeast cell wall proteins (YGP1), yeast glycosidase (Crh1p), and yeast mitochondrial helicase (Hmi1p). Bands containing yeast exo-β-1,3-glucanase and a fraction of GIN1 and TLP (Table 6) were previously reported to be glycosylated and to have increased abundance at pH 3.30 and 3.60 (Lambri et al. 2013a). A diffuse band between 26 and 18 kDa containing glycoproteins was reported in Sauvignon wine (Esteruelas et al. 2009). Since glycosylation reduced the positive net charge of the proteins at acidic pH, the presence of more abundant glycosylated proteins at higher pH makes protein removal more difficult (Table 2) as demonstrated by Vanrell et al. (2007) in base and sparkling white wines. These circumstances corroborate the more effective action of TG and GW labels at pH 3.30 and 3.60, which were affected to a lesser extent by the negative effect of wine pH on the structure and surface charge of the clay double layers. Finally, the higher efficacy of removal of the low molecular weight proteins by the different bentonite labels (Table 6) confirmed a previous report by our group on Chardonnay and Sauvignon wines (Lambri et al. 2012a).
In the study of Sauvage et al. (2010) 1 g/L of bentonite removed 85% of the protein content of a Chardonnay wine at its natural pH. Moreover, it was also reported that ~15% of residual proteins remained non-adsorbed even with a bentonite concentration as high as 1.5 g/L. Similar percentages of residual proteins that belonged to the thaumatin-like family were reported in our previous research (Lambri et al. 2012a) and were observed here for the band 13, which contained VVTL1 (Table 6). The other bands containing VVTL1 were differentially removed: bands 11 and 12 were removed by 50 to 77% with variable efficiency depending on the labels and pH, while bands 14 and 15 were depleted by ~70%, irrespective of the label and pH. Sauvage et al. (2010) observed a similar value of 70% adsorption for TLP, but with bentonite doses ranging from 0.6 to 0.8 g/L. In their study, the depletion of 14 kDa protein and invertase was achieved with bentonite doses of 0.8 g/L and 1 g/L, respectively. By contrast, SPC of band 19 (14 kDa) intensity (Table 6) highlighted protein depletion ranging from 62% (at more acidic pH) to 77% (at higher pH), and a significantly higher and variable percentage of residual invertase (bands 4 to 8, 10, and 16) in wines fined with the different labels under the various pHs. According to Sauvage et al. (2010), invertases were shown to be the most resistant to removal by either bentonite or to heating. The same patterns of results for lower (11.2 to 25 kDa), intermediate (32 to 45 kDa), and higher molecular weight (60 to 65 kDa) fractions was observed for Gewürztraminer, Riesling, and Sauvignon blanc wines by Hsu and Heatherbell (1987). These authors tested bentonite doses from 0.2 to 0.8 g/L and found that heat stability was achieved by only the highest dose applied.
The responses of the unfined and bentonite-treated Erbaluce white wine samples to the heat stability test at 30 to 80°C are shown in Table 7. Haze formation was estimated visually by comparison to an unheated control sample and an absorbance >0.100 AU was used in this study as an absorbance limit over which the sample was considered heat-unstable (Pocock and Waters 2006). Significant differences were observed between the unfined wines and bentonite-treated wines at all pH values. The absence of chitinase (Table 5), which plays an important role in bentonite rate determination (Hung et al. 2014), resulted in a marked decrease in haze formation of fined wines (Table 7) that showed no clouding formation, although significant differences that depended on the pH were observed among the fined wines. The reduction in haze forming tendency of the unfined Erbaluce wine was particularly noticeable in bentonite fined samples heated at 50 to 60°C, 60 to 80°C, and 70 to 80°C at pH 3.17, pH 3.30, and pH 3.60, respectively (Table 7). The fining performance of GW clay in the pH range under investigation and of TG label at higher pH values (3.30 and 3.60; Table 3) enhanced the thermal stability of wine samples, particularly at 70°C (Table 7). Furthermore, GW removed glycoproteins (YGP1 and Hmp1) with a lower efficiency at higher pH (Table 6), perhaps because of the hydrophilic nature of the sugar moiety. This persistence contributed to an increased thermal stability of the residual proteins and a lower haze forming tendency (Table 7) as reported by others (Waters et al. 2005, Fusi et al. 2010, Van Sluyter et al. 2015). At a more acidic pH (3.00 and 3.17), PN displayed a less efficient removal of protein bands that contained GIN1, YGP1, and a fraction of TLP (Table 6). These proteins are known to be more resistant to bentonite fining (Vanrell et al. 2007, Sauvage et al. 2010). However, this lower removal caused no significant difference in the heat stability of the wine, except at 70°C (Table 7). At pH 3.17, the large reduction of GIN1 fractions contained in band 5 (Table 6) may have affected the higher thermal stability of fined samples at temperatures above 70°C (Table 7), because the vacuolar invertase has been shown to unfold at this temperature (Falconer et al. 2010). At 70°C, an important reduction of the invertase band was observed by Sauvage et al. (2010) and stable wines were obtained by the reduction of protein fractions containing invertase after the addition of bentonite both to must and wine in a recent study by Lira et al. (2015).
Moreover, the depletion of bands 12 and 13 containing the VVTL1 fraction from TLP (Table 6) highlighted significant differences among the bentonite labels, which may be conceivable with the differentiated clouding observed after heating of the fined samples at 50 to 60°C (Table 7). Marangon et al. (2014) recently discovered the existence of different isoforms of TLP with an unfolding temperature at 56°C, which form haze. Results demonstrated that progressive removal until completion (Table 5) of TLP and invertases (respectively, over 80 and 70% reduction of bands 12, 13, and 5) was achieved at 1.0 g/L bentonite close to heat stability (Table 7). Even with bentonite addition, some of the TLP still remain in stabilized wines similar to PR-proteins (Table 6), and as already reported in other studies (Okuda et al. 2006, Sauvage et al. 2010, Lambri et al. 2012a, Lira et al. 2015). At higher wine pH, GW seemed to be more effective at all temperatures than the other bentonites tested (Table 7). However, GW displayed a lower removal of protein bands containing yeast exo-β-1,3-glucanase and a fraction of GIN1 and VVTL1 (Table 6) that were previously reported to be glycosylated in Erbaluce wine and to change their abundance when the pH was modified (Lambri et al. 2013a). The removal of glycosylated proteins at higher pHs varies dependent upon the type of bentonite preparation used. This variation will impact the heat-stability of the wine. Thus, the higher presence of glycosylated proteins at higher pHs can affect white wine quality.
These results demonstrated that varying the pH range in a wine affects the tendency of a bentonite to increase the wine thermal stability because of the effect of the pH on the structure and surface charge density of double layers of the clay and the modifications in the wine protein profile.
Polyphenol removal.
The amount of polyphenols in the fined and unfined wines is reported in Table 8. Irrespective of the pH, the lowest level of polyphenols was observed after the treatment with GW. Additionally, at pH 3.00 and 3.17, the polyphenol concentration was significantly decreased after each bentonite treatment. At pH 3.00, the polyphenol concentrations of the wines treated with GW and PN were the most affected. The results of each bentonite treatment was more similar when pH 3.30 and pH 3.60 were compared than when more acidic pH values were considered. At pH 3.30 and 3.60, the largest residual concentration of polyphenols was observed after fining with TG and PW. The extent of polyphenol removal from wine was correlated to the amount of wine proteins that were removed by the tested bentonites (Figure 2). This could be expected since substantial evidence exists to suggest interactions of phenolic compounds and proteins (Salazar et al. 2006). Researchers (Siebert et al. 1996, Lambri et al. 2014) have suggested that the aggregations between wine proteins and tannins are driven by hydrophobic interactions occurring on protein hydrophobic tannin-binding sites, the exposure of which depends on both protein heating and reduction (Marangon et al. 2010). The correlation between the removal of polyphenols and the removal of protein was stronger at pH 3.00 and 3.17 than it was at pH 3.30 and 3.60 (Figure 2). The removal of proteins by bentonite is influenced by the pH as reported by several authors (Lambri et al. 2012a, Shamsuddin et al. 2014). The capacity of the clays to remove organic molecules such as phenolic compounds is low (Jiang et al. 2002) because of the hydrophilic nature of the clay surface (Schoonheydt and Johnston 2013). Modifications of the chemical affinity of the clay can increase its effectiveness in removing organic molecules. The deposition of proteins on the clay particles should change the surface properties by covering the clay surface. In these conditions, the removal of hydrophobic molecules from wine is mediated by proteins, resulting in multilayer adsorption (Shamsuddin et al. 2014).
At pH 3.00, all bentonites removed the same amount of protein (Table 2), but TG and PW were significantly less efficient in polyphenol removal (Table 8). At more basic pH (3.30 and 3.60), PN did not reduce the protein content (Table 2) but at the same pH, PN and GW caused the highest removal of polyphenols (Table 8). These results suggest that direct removal of polyphenols may be obtained using PN at higher pH values. By contrast to literature reports investigating phenolic compounds (Zaghouane-Boudiaf and Boutahala 2011), the removal of wine phenolics appears to be related to clay properties (Dordoni et al. 2015). As observed for the favorable adsorption isotherm of the β-phenylethyl acetate on a clay having the largest SSA (Lambri et al. 2013a), the bentonite PN with the highest SSA and the more negative surface charge density (Table 1) appeared to be the most effective at removing phenolic molecules even when protein removal was not involved (Figure 2 and Table 2). This removal was likely because of strong chemical interactions (Lambri et al. 2013a). These results demonstrated that the effect of the type of clay on wine polyphenol compounds depends on the balance between protein-mediated polyphenol removal and the direct adsorption of polyphenol onto the clay surface.
Conclusions
The fining performance of bentonite labels with different properties were distinguished better at higher wine pH than at more acidic pH values because of lower perturbations in the bentonite structure and enhanced changes in the wine protein profile. Low molecular mass proteins were removed at a high efficiency by all of the bentonites, regardless of the pH. High and medium molecular weight proteins were less removed by bentonite than low molecular weight proteins, and were highly affected by clay characteristics and pH value. Vacuolar invertase (GIN1) and some protein bands containing VVTL1 of the TLP were the fractions more affected by the bentonites having a natural pH lower than 10, thus inducing an increased thermal stability of the fined wines. The clay with the lowest calcium content was the least effective in removing bands containing GIN1, VVTL1 proteins, and a glycoprotein, YGP1. Over the range of pH values tested, the removal of phenolic compounds was mediated by the removal of proteins, and this removal was affected by pH-induced variations. As a result of the physical properties of the clay, a small removal of polyphenols was noted without a significant depletion of proteins. The greatest modification induced by pH upon the composition of wine was the extent of sodium-potassium exchange between the clays and the wine. Acidic pH values resulted in less release of sodium and less removal of potassium from wine, whereas wine with a more basic pH displayed the largest replacement of potassium with sodium. In view of protective actions against the risk of tartrate precipitation, this finding could enhance the wine tartrate-holding capacity. Although further studies are needed on the definition of a safe, allergen-free, and effective adjuvant for colloidal stabilization targeted to wine type, this work explains the differential action of clays under the most common wine pH values used in the industry.
Acknowledgments
Acknowledgments: The authors disclose any conflict of interest. All authors have participated in both research and article preparation. All authors have approved the final article. The present study received no financial support.
- Received January 2015.
- Revision received June 2015.
- Accepted August 2015.
- Published online October 2015
- ©2015 by the American Society for Enology and Viticulture