Abstract
Analysis of protein precipitable wine tannins has become more commonplace due to the simplicity of the method and the strong association between tannin content and perception of astringency. The traditional protein precipitation method for wine measures tannins and polymeric pigments using 5% w/v sodium dodecyl sulfate (SDS) and the alkaline buffer triethanolamine (TEA) pH 9.4 to dissolve the tannin-protein precipitate and support the colorimetric reaction with ferric chloride. However, this method results in a high background absorbance presumably due to the alkaline pH of the resuspension buffer, which has been shown to oxidize phenolics. Experimentation with several buffer formulations showed that pairing TEA with urea instead of SDS allowed formulation of buffers at lower pH. Urea-TEA buffers at pH 7 and 8 showed significantly lower background absorbance and drift over time as well as a significantly greater amount of tannin recovered. These trends in tannin content and iron reactive phenolics were maintained across a wide range of diluted samples (100–1200 mg/L catechin equivalents). Using a TEA-urea buffer at pH 7 or 8 may improve on previous methods of protein tannin precipitation by increasing yield of tannins from the wine and decreasing the background absorbance and drift.
Measurement of tannins has become more commonplace in the wine industry. Numerous methodologies have been developed to measure tannins, and their use in research and winery settings are reviewed here (Thorngate 2006, Harbertson and Spayd 2006). Recently, spectroscopy-based predictive models have been developed for tannins and other phenolic compounds that use spectra gathered from the UV-visible (Skogerson et al. 2007, Dambergs 2012) and Fourier transformed mid-infra red (Fernandez and Agosin 2007) to obtain data rapidly. However, the strength of the models depends on the methodology on which they are based. Thus, improving on the methods used to develop these models is of utmost importance.
The present work focused on the protein tannin precipitation method originally developed by Hagerman and Butler (1978) and later adapted to wine (Harbertson et al. 2002). Protein tannin precipitation has also been combined with traditional sulfur dioxide bleaching to measure polymeric pigments (Harbertson et al. 2003). The advantages of this method are its simplicity and its strong correlation to perceived astringency (Kennedy et al. 2006, Mercurio and Smith 2008). However, the protein tannin precipitation assay requires a minimum amount of tannins to form a protein-containing precipitate (Jensen et al. 2008) and the use of an alkaline buffer to resuspend the tannin-protein precipitate and support its reaction with iron. Additionally, previous work suggests that alkaline conditions may cause inherent errors due to the oxidation of phenolic compounds (Singleton and Rossi 1965).
Protein-tannin complexes can be disassociated using detergents (Hagerman and Butler 1978), caffeine (Cai et al. 1990), or protein denaturants such as phenol (Hagerman and Butler 1980). Urea, a known chaotropic agent, disrupts protein and tannin-protein complex formation (Rowe et al. 2010) by disrupting the hydrogen bonds that maintain the structure (Almarza et al. 2009) and unstacking anthocyanin self-aggregates (Hoshino et al. 1982). Urea also has a high solubility in water (107.9 g/100 mL at 20°C) and thus may be a better alternative to sodium dodecyl sulfate (SDS), which tends to form bubbles that increase the risk of volumetric errors while pipetting and limit the automation of the methodology. Here we evaluated alternative buffer formulations to determine if we could improve tannin analyses. We hypothesized that using urea at a neutral pH would be superior to SDS in terms of improving tannin yield and reducing background absorbance and oxidation of phenolic compounds, which would improve the accuracy of protein tannin precipitation methods across multiple tannin concentrations and types of wine.
Materials and Methods
Reagents.
Bovine serum albumin (BSA, Fraction V powder), SDS, triethanolamine (TEA), sodium chloride (NaCl), ferric chloride hexahydrate (FeCl3*6H2O), (+) catechin, urea, and chemicals used in buffer preparation were purchased from Sigma-Aldrich (St. Louis, MO).
Tannins and iron reactive phenolics quantification.
Tannins were measured using the method by Hagerman and Butler (1978). Volumes were adapted for the measurement of tannins in grape skin, seed extracts, and wines using a micro scale as described here (Harbertson et al. 2002). Iron reactive phenolics in the wine were analyzed as previously described (Heredia et al. 2006).
Experimental resuspension buffers.
The original method (Hagerman and Butler 1978) used 1% (w/v) SDS and 5% (v/v) TEA pH 9.4 as a buffer to dissolve the tannin-protein precipitate and to support the eventual colorimetric reaction with ferric chloride (FeCl3). In the modified method by Harbertson et al. (2002), the percentage of SDS was increased to 5% to improve the dissolution of tannin-protein precipitates for grape skins and seeds. In our experiments, we modified the composition of the resuspension buffer. SDS was replaced with urea (8.3 M), and three buffers with a pH of 7, 8, and 9.4, respectively, were created. We were still able to dissolve the precipitate and maintain a colorimetric reaction with FeCl3. We compared these TEA-urea buffers with two 5% (v/v) TEA-SDS buffers (adjusted to pH 8 and 9.4) in the incubation experiment.
Evaluation of resuspension incubation.
A 2011 Cabernet Sauvignon red wine from the Columbia Valley in Washington State was analyzed in triplicate for this experiment. Following dissolution of the protein-tannin precipitate, tannins and background were measured after incubating for 10, 20, 30, or 60 min in the resuspension buffers.
Spectral evaluation.
Absorbance spectra (400–700 nm) were measured in 1-nm increments using an Agilent diode array spectrophotometer (Model Number 8453; Agilent Technologies, Santa Clara, CA). Tannins and total iron reactive phenolics were assayed in triplicate in samples of red wine as described earlier. Briefly, the diluted wine or tannin protein precipitate was dissolved in resuspension buffer and incubated for 10 min before the spectrum of each sample was collected. Spectral data were measured again after adding FeCl3 and incubating for 10 min.
Dilution evaluation.
Fifteen wine samples were analyzed for tannins in triplicate using different assay dilutions (1:9, 1:4, 1:2.333, 1:1.5, 1:0.667, 1:0.25, and no dilution) with the TEA-SDS pH 9.4 and the TEA-urea pH 7 and pH 8 experimental buffers. The percentage of tannin precipitated and the difference in absorbance between the background and FeCl3 reaction spectra was calculated for each dilution series. Samples in which the final absorbance measurements (after FeCl3 was added) were outside of the instrument’s linear range were omitted.
Multiple sample evaluation.
Forty-four vintage red wine samples (22 Cabernet Sauvignon, 22 Syrah) from Yakima and Columbia Valley American Viticulture Areas were analyzed in triplicate for tannins and iron reactive phenolics using the control and experimental buffers.
Results and Discussion
Standardization and spectra.
Standard curves of catechin (50–300 mg/L) with the control and experimental buffers were carried out in triplicate. A one-way ANOVA showed that neither the slopes (0.0060 ± 0.0002) nor the intercepts (0.03 ± 0.007) were significantly different, indicating that neither the pH nor the buffer composition affected the reaction between iron reagent and the phenolic compounds.
Incubation experiments.
A two-way ANOVA was performed to determine the effects of the alkaline buffer formulation (pH 8 and pH 9.4) and duration of incubation with the protein-tannin precipitate (10, 20, 30, and 60 min) on background absorbance (BKGD A510nm), total absorbance after addition of FeCl3 (Total A510nm), change in absorbance after addition of FeCl3 (Δ A510nm FeCl3), and tannin concentration in a sample of Cabernet Sauvignon wine (Table 1). Buffer pH and incubation time had significant effects on BKGD A510nm, Δ A510nm FeCl3, and tannin concentration; however, only pH had a significant effect on total A510nm.
All of the variables were significantly greater at pH 8 than at pH 9.4, suggesting that although the background is greater at the lower pH, the greater change in absorbance at the lower pH indicated that a greater amount of tannin was recovered. Furthermore, the slope of the standard curve was similar between the two pHs, suggesting that more tannins are recovered at pH 8 than at pH 9.4. This is consistent with previous work that suggests a more neutral pH is better for phenolic analysis because it reduces oxidation of the compounds (Singleton and Rossi 1965). However, we observed a significant increase in background absorbance (A510nm) at 30 and 60 min and concomitant decrease in Δ A510nm FeCl3, resulting in a significant decrease in tannin concentration at 20 min and the subsequent time points. Previous work has suggested that alkaline conditions may oxidize phenolic compounds and thus make them unavailable to react with the color reagent (Singleton and Rossi 1965). We performed incubations without the metal color reagent, ruling out oxidation of the metal or metal phenolic complex as a source of variation. We found a significant increase in background absorbance without any change in total absorbance after the metal color reagent had been added, suggesting that the change was due to loss of tannins through oxidation as well as an increase in background absorbance.
A two-way ANOVA was used to evaluate the impact of buffer pH and incubation period on the experimental TEA-urea buffer formulation (Table 2). The background absorbance during the incubation period significantly increased only at 60 min for both the pH 8 and pH 9.4 buffers, whereas background absorbance did not change significantly during the 60-min incubation for the pH 7 buffer. As expected, total A510nm, Δ A510nm FeCl3, and tannins were highest at pH 7, followed by pH 8 and pH 9.4, respectively (all p < 0.05). Overall, our results suggested that buffer formulations with urea and TEA were superior to those formulated with SDS and TEA. Buffers containing urea and TEA had significantly lower background absorbance at all pHs, and none of the reaction absorbances was different at pH 7. Similar to our earlier results, using the pH 7 buffer increased the recovery of tannin in the protein-tannin precipitates.
The effects of pH on the background absorbance and FeCl3 tannin reaction spectra for the same wine sample used in the incubation experiment are shown in Figure 1 and 2, respectively. Similar to our findings in the incubation experiment, altering buffer pH or buffer formulation (TEA-SDS or TEA-urea) did not change the overall shape of the background spectra or iron reactive tannins, strengthening the argument for pairing TEA with urea instead of SDS.
Protein-tannin complexes are primarily formed through hydrogen bonding and hydrophobic interactions (Hagerman et al. 1998). The principle behind dissolving the tannin-protein complex in the original method by Hagerman and Butler (1978) was that tannins will not interact with proteins at a pH high enough to be in their charged phenolate form (Hagerman and Butler 1978). In our study, we showed that a high concentration of urea with the original TEA buffer dissolved the protein-tannin complex, even at a neutral pH in which the tannins would not be in the charged form. Condensed tannin protein complexes are primarily driven by hydrogen bonding (Hagerman et al. 1998), which suggests that the urea dissolves the protein-tannin complex by acting as a hydrogen bond disruptor.
Dilution evaluation.
Protein-tannin precipitation analysis requires excess protein to ensure precipitation. As a result, dilution of wine or grape samples is necessary and may be a source of variation in the analysis (Jensen et al. 2008). Additionally, changes in background absorbance and the reaction of FeCl3 must be taken into account. Because the background absorbance was significantly lower with the TEA-urea buffers, we diluted 15 wines with the TEA-SDS pH 9.4, the TEA-urea pH 7 or 8 buffer to concentrations ranging from 100–1200 mg/L catechin equivalents (CE). Supplementary Figure 1A–C shows the difference in absorbance in relation to the percentage of tannin precipitated in the dilution series. Similar to previous findings by Jensen et al. (2008), we observed a valid range of absorbance change (≥95% tannin recovery), with a range of 0.7 to 1.5 for the TEA-SDS pH 9.4 buffer and a range of 0.6 to 1.6 for the TEA-urea pH 7 and 8 experimental buffers. Absorbance changes between 0.4 and 0.5 deviated from the remaining data in both the pH 7 and 8 buffers (0.4 to 0.5) and showed a high percentage of tannin precipitation. This range was found primarily for low tannin wines that were near the lower limit for precipitation (100 mg/L CE). Our results were ultimately compatible with the previous recommendations; however, the absorbance range observed was slightly larger than previously reported by Jensen et al. (2008), likely due to the slightly smaller range of tannin concentrations in the previous work (which ranged from 300 to 700 mg/L CE).
Multiple sample evaluation.
Figure 3A and B shows the analysis of the background absorbance and tannins (A) and total iron reactive phenolics (B) of multiple samples of Cabernet Sauvignon (n = 22) and Syrah (n = 22) wines dissolved in the original SDS-TEA buffer at pH 9.4 or one of the three experimental TEA-urea buffers (pH 7, 8, or 9.4). A one-way ANOVA with post hoc analysis showed that the background absorbance and the phenolic content (of both iron reactive phenolics and tannins) varied significantly as a result of the different buffers. The earlier trend was also observed here, with the TEA-urea buffers containing significantly higher tannin concentrations and significantly lower background values. However, content of total iron reactive phenolics exhibited a different pattern, with the TEA-urea pH 8 containing the highest content, followed by TEA-SDS pH 9.4, TEA-urea pH 7, and TEA-urea pH 9.4, respectively. As previously discussed, tannin recovery, but not iron reactive phenolics, was greater with the TEA-urea pH 7 buffer than with the higher pH buffers, suggesting the wine contained one or more compounds that interfered with analysis of total iron reactive phenolics at pH 7. Tannins are ubiquitous within the plant kingdom, and the TEA-urea pH 7 buffer may improve the characterization and evaluation of protein-tannin complexes found in other foods and feeds because of the more stable pH conditions at pH 7 (Hagerman 1992).
Conclusion
Our results showed that substituting urea for SDS in the TEA-based resuspension buffer may improve tannin protein precipitation assays by lowering the background absorbance and background drift that may occur if practitioners do not follow the original method (Harbertson et al. 2002). Lowering the pH of the TEA-urea buffer used to dissolve the tannin protein precipitate significantly increases the amount of tannin recovered, possibly due to less oxidative destruction of the phenolic compounds. Although the oxidative destruction is likely independent of buffer composition, only the TEA-urea buffer dissolved the protein-tannin precipitate at pH 7. The range of absorbance change required to precipitate ≥95% of the tannins was slightly larger for the pH 7 and pH 8 TEA-urea buffers than for the control buffer, but ultimately comparable to the values reported for the original buffer. An analysis of 44 wines combined with the control and TEA-urea buffers confirmed our earlier results in terms of content of tannin precipitation and total iron reactive phenolics. Although the tannin content was higher in the TEA-urea pH 7 buffer than in the higher pH buffers, the content of total iron reactive phenolics was not higher in the pH 7 buffer, suggesting that one or more compounds in the wine may have interfered with analysis at neutral pH. We recommend the use of pH 7 and pH 8 TEA-urea buffers to improve tannin recovery and reduce background absorbance shift and account for the 10–15% reduction in total iron reactive phenolics if the pH 7 buffer is used.
Acknowledgments
Acknowledgments: The authors would like to thank the Wine Advisory Committee, the Washington Wine Commission, and the WSU Agricultural Research Center for their support of this project. Richard Larsen is thanked for his careful editing and Douglas Adams and Roger Boulton are thanked for their helpful comments on this manuscript.
Footnotes
-
Supplemental data is freely available with the online version of this article at www.ajevonline.org.
- Received July 2014.
- Revision received October 2014.
- Accepted October 2014.
- Published online December 1969
- ©2015 by the American Society for Enology and Viticulture