Editor's note: The analysis of wine tannin has traditionally meant that an enologist will use one of the relatively popular precipitation-based techniques, most of which are derived from Hagerman and Butler's procedure published in the Journal of Agricultural Food Chemistry (1978). Assaying tannin by traditional methods involves several steps and is fairly slow. Furthermore, the accuracy and precision of these protocols have been repeatedly questioned over the years. This has led to several refinements to Hagerman and Butler's assay, most notably by Martin and Martin, and most recently by Harbertson and Adams. But it appears that a truly definitive tannin assay has remained somewhat elusive. In the following article, Silber and Fellman suggest that a bioassay using a stable protein-dye marker should prove somewhat faster, less expensive and more accurate than the current protein precipitation-based protocols.
Red wine consumption continues to flourish in the U.S. Over the past 25 years consumption of better quality table wines has risen dramatically while consumption of lower quality ("jug") wines has decreased. Other studies conducted throughout the world confirm the health benefit of red wine, linking this effect to the antioxidant activity of flavonoid phenols abundant in red wine. Principal phenolic compounds found in red wines include condensed tannins (proanthocyanidins) formed from the polymerization of flavonoid phenols during grape berry development, which are prominent components that contribute taste and quality to the wine.
During vinification, tannic balance in red wine arises from the relationship between seed and skin tannins. Skin tannins contribute to the wine's organoleptic properties, "fullness, roundness and color" while "structure and body" properties are attributed to seed tannins.
Perceptions of bitterness and astringency in red wine are elicited mainly by condensed tannin. Monomeric flavonoid phenols are primarily bitter, but upon polymerization astringency increases more rapidly than bitterness. An abundance of seed tannins can make wine excessively astringent while high skin tannin levels may cause a herbaceous character and bitterness. These roles for tannin in the sensory quality of wine justify the avid interest of winemakers in developing a method to measure grape tannin content during growth and wine manufacture, as well as provide a tool to investigate cause and effect relationships throughout the winemaking process.
Historically, studies directed toward optimizing the production of fine wine have been pursued from two interactive perspectives. A considerable amount of viticultural research has identified various cultural practices used to optimize tannin content in grapes at harvest and in finished wine, including irrigation, canopy management and cropping yields. On the other hand, tannin analysis has been, and is presently, used to monitor wine quality development in vineyards and cellars.
Numerous biochemical methods for measurement based on tannin structure or chemical properties have been studied. However, the chemical complexity and diversity of tannins restrict use to those based on the protein precipitation technique, which serves best to both quantify tannins and assess their biological activity.
Among the precipitation-based procedures, the method of Martin and Martin that also exploits the precipitation-based protocol of Hagerman and Butler is widely popular and is used in ecological and nutritional studies. Although precision, accuracy and thoroughness of this method have been repeatedly questioned, concerns have recently been raised as some the apparent equilibrium constant of binding between the TA and BSA, soluble tannin-protein complexes are increasingly formed. It was postulated that for the reasons of analytical accuracy a protein precipitation tannin assay should include, along with the assessment of the protein-precipitating capacity, a simultaneous determination of the PEP of any plant extract. Scientific evidence supports the rationale for the PEP determination to improve sensitivity, accuracy and consistency in tannin bioassays.
Shortly after Calderon et al. identified soluble tannin-protein complexes, other reports indicated a critical point existed for the tannin-BSA precipitation from a solution. The point was defined by Hagerman and Klucher as the PEP or optimum precipitation point (OPP) and was shown to be strongly associated with the tannin-BSA ratio in the reaction solution. Dissolving of the tannin-BSA complex usually occurred beyond this characteristic point, at the BSA-tannin ratio > 1.0. It has been demonstrated in theoretical and experimental studies that high accuracy and consistency patterns of tannin bioassays imply precipitation of BSA to varying degrees while avoiding dissolving of the tannin-BSA complex.
A practical solution involves determination of the PEP prior to or simultaneously with the assessment of the protein-precipitating capacity of a plant extract. To make the PEP determination more easily available in large-scale bio-screening studies, we proposed a hands-on modification of our core tannin bioassay by substituting the optical reading of the protein precipitation results by a visual one.
Operating a tannin bioassay using a known PEP value makes it easier and faster to find the optimum protein concentration range and the proper tannin to protein ratio when setting up a tannin bioassay. It also minimizes undesirable guesswork and related error in assessing the protein concentration range and the optimum tannin-protein ratio in the reaction mixture, thus practically eliminating inaccurate results. The assessment of the PEP value before each tannin assay is critical to maintain accuracy and consistency, and to minimize guesswork when estimating the range of protein concentration needed in the tannin assay.
Tannin bioassay by a protein precipitation technique serves best to both quantify tannin, and assess its biological activity when both the precipitate (the insoluble tannin bound protein) and the supernatant (the soluble tannin bound protein) are used for tannin assessment. This protocol provides reliable information regarding tannin content in wine because it inherently includes determination of the PEP value for each individual wine sample.
Here we report application of a new tannin bioassay protocol (patent pending) using a highly stable hydrophobically associated protein-dye marker (alkaline CBB-BSA) that allows hands-on, rapid, accurate and low-cost tannin monitoring for grape growing and winemaking.
Materials and Methods
Dye-protein marker: The manufacture of alkaline CBB-BSA dye-protein marker is described elsewhere. Each new batch of the alkaline CBB-BSA reagent is standardized by evaluation of the optimum number of binding sites for CBB on BSA, the lmax shifts in solvents of different polarity and the l602nm absorbance spectra. In addition, each new preparation of the alkaline CBB-BSA is randomly standardized by its molar absorptivity (El602nm), net absorbance, detergent interference, stability under various pH and storage conditions, and tannin-binding capacity with standard samples of tannic acid (TA) or quebraco tannin (QT).
Sample treatment: Grape berry and/or wine samples were obtained from experimental and commercial sources in Washington state.
Grape berry extraction: One hundred grape berries selected from cluster samples are taken from each treatment/ replication combination. Berry samples are frozen and stored for later analysis. Seeds and skins are separated from pulp, blotted dry, weighed, freeze dried, ground and stored at -20°C until extraction for tannin. Extracts of tannin in 50 percent aqueous methanol are prepared as follows.
Briefly, aliquots of 400mg samples are weighed into 50ml Pyrex-glass extraction vials. Ten ml 50 percent aqueous methanol is added, and large-diameter glass marbles placed on top of each tube. Initial extraction is performed by placing the tubes in a room-temperature, ultrasonic water bath for 20 minutes, then in a dry-heating block at 92-95°C for 15 minutes. After cooling to room-temperature, the samples are transferred to a refrigerated centrifuge and spun in an angle rotor at 9130 × g for 20 minutes. The supernatants from each sample are combined into 50ml volumetric flasks while the precipitates are resuspended in another 10ml of 50 percent aqueous methanol, and two subsequent extraction-centrifugation cycles are repeated. The final precipitates are discarded. All supernatants from the same samples are combined in one volumetric flask, and the final uniform volume of the supernatants is adjusted with aqueous methanol (up to 25ml) before tannin analysis.
Wine sample preparation: The original wine samples are diluted twice in deionized water before tannin analysis. Aliquots of 0.5 to 200µL (approximate equivalents to tannin concentration range of 10 to 60mg/ml) are then transferred to a set of tubes containing tannin reaction mixture to determine PEP or tannin values.
Tannin assay with established methods: Protocols of established methods for tannin quantification in grapes and wine are listed here with pertinent reagents and expression differences.
Tannin assay with alkCBB-BSA: Ordinarily, individual sets of four to seven test tubes containing incrementally increasing (by 200µL) amounts of grape berry aqueous methanol extracts, or diluted wine samples, are arranged. The range of the extract solution concentrations for each individual sample is preliminarily assessed in a PEP optical or visual test. A constant amount of alkCBB-BSA reagent (50µL) is added, and the final volume of the reaction mixtures is adjusted to 2.76ml with 0.2M acetate buffer, pH 5.0. The test tubes are mechanically agitated (Vortex mixer) and refrigerated to optimize the precipitation process. After hue appears (approximately 15 min utes), tubes are centrifuged in a refrigerated (5°C) angle rotor at 9,130 × g for 20 minutes.
In the tubes with the range of the sample concentration close to the PEP value, the color of the supernatants decreases as a simultaneous increase in blue precipitate occurs. The supernatants from each test tube are separately combined and diluted with 2ml of 0.2M acetate buffer while the precipitates are dissolved in 2ml of a detergent mixture, 1 percent SDS-5 percent TEA plus 2ml of .2M acetate buffer. The absorption at 602nm of both the supernatants and dissolved precipitates is measured. A series of double-Y absorption graphs is plotted against increasing concentrations of wine/grape samples, and the location where the supernatant curve intersects with the precipitate curve on the graph, the intersection point (IP), is determined. This value is used in Equation 1 (see below) to quantify tannin content.
Assessment of the precipitation equivalence point: PEP assessment always precedes tannin analysis. A set of four to seven test tubes is usually sufficient. An extended set of seven to 15 test tubes might be required for unknown plant material or when low tannin content is expected. Wine samples are added in increments of increasing concentration so that CBB-BSA is precipitated to varying degrees, from minimum to maximum. The final volume of the reaction mixture is then adjusted to 2.76ml. The PEP is determined either with a colorimeter or more rapidly by direct visual reading to identify the tube in which the blue color of the supernatant has disappeared after centrifugation, or the tube with the largest blue-colored precipitate. In most cases, both readings are complementary.
Standard tannic acid bioassay: A standard precipitation curve is plotted with varying concentrations of standard tannic acid (TA) in 50 percent aqueous methanol. A set of five test tubes is made with incrementally increased amounts of standard TA, from 0.05mg to 1.0 mg. A constant amount (50µL) of alkaline CBB-BSA reagent is added to each tube, and the final volume is adjusted to 2.76ml with 0.2M acetate buffer, pH 5.0. The standard samples are then treated as previously described for test samples.
Double-Y graph plot: The absorbency at 602nm is measured in both the precipitate and the supernatant from the same sample. The intersection point (IP) on the double-Y graph plot is identified. The values of the IP for both the standard and test samples are then treated in Equation 1 below:
where TAE is the tannic acid equivalent; A equals the product (Xs × Ys), where Xs is the X-axis projection value of the sample IP, and Ys is the Y-axis projection value of the sample IP; B - equals the product (Xst × Yst) where Xst is the X-axis projection value of the standard IP, and Yst is the Y-axis projection value of the standard IP.
Tannin assay validation
We paired the alkCBB-BSA tannin assay that includes PEP evaluation plus tannin assessment in both the precipitates and supernatants with three other established methods for tannins, which protocols do not include PEP evaluation but rather determining tannin only in the supernatant or in the precipitate, plus different analytical techniques for tannin detection and quantification at the final steps of the assay. Thus, the Martin and Martin assay deals with quantifying tannin through unprecipitated protein with Bradford reagent. The two other methods have adapted the classic tannin-protein precipitation method of Hagerman and Butler for use with grapes and wine. In these tannin assays the resulting precipitates are treated either with Ferric chloride (FeCl3) or a vanillin analogue Dimethylaminocynnamaldehyde (DMCA) for differential determination of tannin/pigments or polymeric (in grapes) and total (in wine) flavan-3-ols, respectively.
Martin and Martin Tannin Assay: A set of 8 to 12 test tubes contained incrementally increased quantities of 50 percent aqueous methanol extracts of each sample so that the BSA would be precipitated to varying degrees for the purpose of subsequent regression analysis of the unprecipitated proteins with Bradford technique. The range of the incremental concentrations of plant extracts is from 1.6mg to 6.4mg (0.2ml to 0.8ml) per test tube, and the final volume in each test tube is brought to 0.8ml with 50 percent aqueous methanol. BSA stock solution is added equally, 1.39mg (dissolved in 1.39ml of 0.2M acetate buffer solution, pH 5.0, concentration 1mg/ml) to each probe. A similar set of solutions is made up with blanks instead of sample extracts. The probes and controls are stirred and kept refrigerated overnight. The next day they are filtered through Sephadex G-25 to isolate the unprecipitated proteins and to remove unreacted (excess) tannins and other compounds that might interfere with the Bradford colorimetric protein assay.
Isolation of the unprecipitated proteins with Sephadex G-25: Prepacked Sephadex G-25M columns (PD-10, Pharmacia), bed volume 9ml are originally equilibrated with 0.9 percent NaCl solution containing 0.05 percent merthiolate and finally rinsed three times with 0.2M acetate buffer, the filtrates being discarded. The refrigerated samples are centrifuged at 9,130 × g for 15 minutes and the obtained supernatants poured onto the columns. The precipitated pellets are washed with 0.51ml acetate buffer, stirred and re-centrifuged. The second supernatants are added to the columns. Finally, 3.5ml of methanol buffer is added to the columns with eluants being collected at all steps.
Bradford Protein Assay: The original Bradford micro-method for colorimetric protein assay is used. Two colorimetric cuvettes filled with 2.5ml of acidic Coomassie Blue (BioRad) solution are used for each extract solution. An amount of 0.1ml sample in methanol buffer is added into the cuvettes. Five minutes later the absorbance is read at 595nm.
Plotting regression curves: Regression curves for unbound BSA after the centrifugation are plotted against constant BSA concentration (1mg/ml). The amount of protein left in each supernatant after precipitation reaction is determined by substituting the log percent of the sample transmittance into the regression equation. The amount of BSA precipitated by tannins in the sample is determined as the difference between the amount of BSA (1.39mg) added to the reaction mixture minus the BSA left in the sample (unbound BSA detected in the supernatant). Tannin binding capacity is reported as mg BSA precipitated per g plant tissue (expressed on a dry matter basis). Hence, if 0.4g of plant tissue is initially treated with 50ml of extraction solution, the concentration of the plant sample is 8mg/ml. If 0.5ml of the extract is used in the tannin bioassay and 0.6g BSA is precipitated, the tannin binding capacity would be reported as 0.15mg protein precipitated/g plant material: 0.6 / (0.5 × 8 mg/ml) = 0.15.
Tannin Analysis With FeCl3: The method of Hagerman and Butler is scaled down, and volumes are adjusted to adapt the method to the amount of tannin found in grape extract and wines. A protein solution for tannin precipitation is prepared by dissolving BSA in 0.2M acetate buffer, pH 4.9, to give a final protein concentration of 1mg/ml. The sample assay is started with the addition of grape extract or wine containing tannin to a BSA solution. A 1.0ml aliquot of the protein solution is dispensed into a 1.5ml microcentrifuge tube, then 500µL of the diluted extract or wine is added, and the mixture is allowed to incubate at room temperature for 15 minutes with slow agitation. The sample is centrifuged for 15 minutes at 13,500 × g to pellet the tannin-protein precipitate. The resulting pellet is then resuspended in a basic detergent buffer, 5 percent TEA (v/v) - 10 percent SDS (w/v). The tube is incubated at room temperature for 10 minutes then vortexed. An aliquot of 0.125µL of a ferric chloride reagent (10mM FeCl3 in 0.01 N HCl) is immediately added to the tube. After 10 minutes, the absorbance at 510nm is determined. A standard curve is plotted with (+)-Catechin (50 to 300mg/L), and tannin content is expressed as mg/L CE (Catechin equivalents).
Tannin Analysis With DMCA: The protocol for this method is similar to that described for FeCl3 except that FeCl3 is substituted for a different chromogen, DMCA, that forms a reaction product with a lambda max at 640nm.
Performance Characteristics: It appears that the performance efficiency of the Martin and Martin method relies primarily on performance characteristics (sensitivity, accuracy and reproducibility) for detection and measurement of micro quantities of unprecipitated proteins in the supernatants after centrifugation. For this purpose, an acidic Coomassie blue-dye from the Bio-Rad Life Science Group is used. As the performance efficiency of this technology is fully described in the Bio-Rad Life Science Group publications, here we refer to this source.
alkCBB-BSA assay in comparison to Martins' method: Both methods are compared for data frequency distribution and sensitivity at the threshold of lower limit detection (LLD), higher limit detection (HLD), close-running thresholds for the minimum and maximum data values, and available concentration range. Our method provides no zero or negligible results for tannin in all analyzed white and red wine samples. It has both higher sensitivity and accuracy for tannin estimated by data frequency occurrence and characteristic histogram profile, a 35.9 percent lower LLD, twice higher HLD, and a twice as broad concentration range, compared to the paired method. Importantly, the paired method provides a high frequency of negligible (zero) results (35.9 percent) in white and diluted wines. It is concluded that conducting tannin bioassay based on calculated PEP value and tannin assessment in both the precipitates and supernatants significantly increases the sensitivity and accuracy of the analysis (Table 1).
Clear difference is observed upon comparison of results obtained from each of the paired assay. While the alkaline CBB method p during berry cell division, followed by weekly irrigation to replenish vine water usage (ED);
3. the standard practice until veraison followed by a period of water deficit (VD).
The crop level treatments consisted of 3 (Low) and 6 (High) tons per acre. Plot sizes were designed for pilot plant winemaking (one-ton lots). A split-plot design was used with the main-plot consisting of the three irrigation treatments, replicated four times. Each main-plot was divided into the two crop-level subplots.
Figure 3 shows the irrigation effect and interaction with crop load on tannin contents of wine made from the experimental vineyard. Apparently the deficit applied after veraison in high crop load vines increases the tannin levels in the finished wine. A deficit applied early in grape cluster development elevated wine tannin levels in low crop load vines when compared to high crop loads subjected to the same treatments.
Use of the tannin assay may assist winegrape growers in pre-harvest assessment of grape quality to determine ultimate winemaking styles while adjusting for field-derived quality parameters.
The principal effect of light levels appears to occur in seed tannin content as shown in Figure 4. Since the sensory properties of astringency as well as "structure and body" are attributable to seed tannins, it is possible to measure grape tannin levels for potential blending before maceration according to the winemaker's preference.
Tannin Measurements in Relation to Red Wine Quality as a Blending Guide: Recently some Washington vintners have used tannin assays to quantify wine quality. Measurements of phenolic content (anthocyanin, polymeric pigment, total color and tannin) in all red wine lots after primary fermentation, after two months of aging, prior to blending and one year after bottling, are used in conjunction with sensory information data to make primary blending decisions. Without this information to guide the process, it is difficult to ascertain tannin extraction efficiency and determine which wine lots to blend together for "synergy." Instead, the winemaker is able to move the wine quality in a desired direction based on hard data while providing a meaningful historical record. Quantifying tannin on a routine basis provides an advantage of vintage evaluation earlier than other producers and to measure comparative vineyard quality or experimental treatments. Data presented in Table 2 and Figure 2 clearly demonstrate the advantage of our method.
Upon comparison, differences between absolute tannin values as measured by the three different methods were apparent. Use of the FeCl3 reagent to estimate precipitated tannin concentrations showed very little difference between samples, but greater differences between samples were noted using the other two assay protocols. Regardless of the absolute values, the DMCA and alkCBB-BSA methods showed similar comparative quantities.
Apparently the alkCBB-BSA precipitation method is more sensitive as evidenced by the greater quantitative differences observed between individual wine samples. Using the same samples, taste panel estimates of astringency were highly correlated to values obtained using the alkCBB-BSA method (R2=0.858) and the ferric chloride method (R2=0.905). The lowest correlation (R2=0.675) existed between perception of astringency and the DMCA method. Of course, these figures may r
Moris L. Silber and John K. Fellman
John K. Fellman is professor of Postharvest Physiology in the Department of Horticulture at Washington State University, where he has been a resident faculty member since 1995. His principal research interests include phytochemistry, chemistry and biochemistry of fruits and their products, and postharvest plant physiology. Moris L. Silber, M.D. Ph.D. is currently assistant research professor in the Department of Natural Resource Sciences at Washington State University, Pullman, where he has been on the research faculty since 1990. His research interests revolve around phytochemistry, especially of tannins and other secondary plant metabolites as applied to the study of human nutrition and medicine. For more information, contact John K. Fellman (fellman@wsu.edu)
