|Home | About | Journals | Submit | Contact Us | Français|
We demonstrate proof-of-concept for the use of electrophoretically mediated microanalysis (EMMA) as a new approach to the determination of total antioxidant capacity (TAC). EMMA is a low volume, high efficiency capillary electrophoretic technique that has to date been underutilized for small molecule reactions. Here, nanoliter volumes of 2,6 – dichlorophenolindophenol (DCIP) reagent solution are mixed with an antioxidant containing sample within the confines of a narrow bore capillary tube. The mixing is accomplished by exploiting differential migration rates of the reagents when a voltage field applied across the length of the capillary tube. The ensuing electron transfer reaction between DCIP and the antioxidant(s) is then used as a quantitative measure of the TAC of the sample. Linear calibration using either redox form of DCIP is accomplished with standard solutions of ascorbic acid. Several commercial beverage samples are analyzed, and the TAC values obtained with the reported methodology are compared to results obtained with the widely used ferric reducing antioxidant power (FRAP) spectroscopic method. For the analysis of real samples of unknown ionic strength, the method of standard additions is shown to be superior to the use of external calibration. This easily automated EMMA method may represent a useful new approach to TAC determination.
Antioxidants help prevent oxidative damage to biomolecules caused by free radicals and are thought to aid in the prevention of many health problems, such as heart disease, cancer, and inflammatory diseases (1). Thus, the measurement of antioxidants in consumer products is of interest. Separations-based analytical approaches to identifying and quantifying the individual antioxidants found in a food and beverage sample is a challenging task, but a few approaches have been reported (2–5). Due to the chemical diversity of antioxidants, and the complex mixtures of antioxidants often found in many real samples, the use successful use of any single separation technique has proven difficult. In addition, the collective effect of a complex mixture of antioxidants is often different than the sum of the contributions of the individual antioxidants in the mixture (6). These factors, along with commercial pressure for a simple numerical representation of the antioxidant content in a sample, have lead to the useful concept of Total Antioxidant Capacity (TAC). Instead of being concerned with the individual responses of each antioxidant, TAC considers the aggregate strength of all antioxidant compounds present in a given sample (7–14).
There have been numerous laboratory methods for TAC determination reported in the literature (8–15), and these methods vary widely in both the chemical approach and the resulting numerical values obtained for TAC. The chemistry behind several of these methods was the topic of a very nice recent review article (15). The chemistry employed in TAC analyses generally falls into two categories: methods involving electron transfer (ET) reactions, and methods involving hydrogen atom transfer (HAT) reactions. Regardless, the analyses generally involve mixing a sample solution with one or more standard reagent solutions while monitoring changes in either absorbance or fluorescence. The most widely used methods are the ferric reducing antioxidant power (FRAP) assay (8) and trolox equivalent antioxidant capacity assay (TEAC) (9), and the oxygen radical absorbance capacity (ORAC) assay (10). While each of these methods is designed to measure the aggregate antioxidant capability of a given sample, the methods are performed under widely differing conditions, so it is perhaps not surprising that these methods often produce differing numerical results (11–14), making attempts to quantitatively compare data obtained with different procedures ill-advised. Nonetheless, TAC has become an increasingly popular metric by which antioxidant properties can be described. Indeed, TAC values are now found on the labels of many commercial beverages in the USA.
Within the past two decades, capillary electrophoresis (CE) has become a viable analytical technique (16–21). CE gained a great deal of credibility as the technique that allowed the human genome to be solved several years ahead of schedule (17). The general advantages of CE-based analyses include low sample volume requirements (and correspondingly low amounts of chemical waste), rapid analysis times, and highly efficiency. These compelling aspects of CE-based methods, in capillary or microfluidic chip format, increasingly attractive alternative to traditional method of analysis. Here we consider the utility of electrophoretically mediated microanalysis (EMMA), a CE-based approach for performing low volume in-line chemical reaction(s), to determine TAC. EMMA, which was first reported by Bao and Regnier in 1992 (22), is an excellent analytical tool for in-line mixing of nanoliter volumes of reagent(s) and a sample for chemical analysis. EMMA is performed in a fused silica capillary and the mixing of chemical reagents within the capillary is accomplished by applying a voltage field, exploiting the differences in the electrophoretic mobilities of the reactants. The EMMA technique retains all the attractive features of capillary electrophoresis (CE) methods, such as small sample volumes, minimal waste, use of a single on-line flow-through detector, and rapid analysis time. Indeed, these advantages have lead to significant interest in EMMA methodology; however, most of the interest has been focused toward the development of enzymatic analyses (23–25) where enzyme turnover can compensate for the low concentration sensitivity that is inherent to capillary electrophoretic systems with UV absorbance detection. Still, an increasing number of reports of EMMA with small molecule reactions have recently appeared (26–28), and these have been primarily in areas where high micromolar to millimolar concentrations of analytes are possible. Because TAC values are often in the millimolar range, CE analyses with in-line UV detection should have more than adequate sensitivity. Consequently, the determination of TAC with EMMA becomes an intriguing possibility.
In this work, we investigate the utility of EMMA to determine TAC values of aqueous samples containing antioxidants, and demonstrate proof-of-concept for this approach using a limited set of commercial beverages. We use 2,6 – dichlorophenolindophenol (DCIP), a common redox indicator dye molecule, to act as the oxidizing agent in a rapid in-line electron transfer reaction with antioxidants. DCIP turns out to be an attractive choice of oxidizing agent as it (i) is easily detected by UV/VIS absorbance in either redox form (ii) has redox state0sensitive spectral properties, (iii) exhibits fast electron transfer kinetics and (iv) has reasonable mobility so that the individual redox forms of DCIP can be easily separated with CE distances and voltages without significant electrodispersion. The reported in-line method is applied to assess the TAC to a variety of commercial juice/beverage samples, and the results are compared to those obtained with the accepted FRAP method. The effect of sample conductivity on the EMMA results is also investigated.
Sodium phosphate monobasic, L-ascorbic acid, and 2,4,6-tris(2-pyridyl)-s-triazine (TPTZ) were purchased from Sigma-Aldrich Company (St. Louis, MO). 2,6 – dichlorophenolindophenol sodium salt (DCIP) was purchased from Aldrich Chemical Company (Milwaukee, WI). Ferric chloride was purchased from J.T. Baker Chemical Company (Philipsburg, PA), glacial acetic acid from EM Science (Gibbstown, NJ), and sodium hydroxide and hydrochloric acid from Fisher Scientific (Fairlawn, NJ).
Phosphate buffer (50.0 mM) was prepared by dissolving monobasic sodium phosphate in 18 MΩ·cm water. The pH was adjusted to 7.20 using 1 M NaOH and an Accument pH meter (Fisher Scientific, Fairlawn, NJ). The acetate buffer used in the FRAP experiments was prepared by diluting glacial acetic acid with 18 MΩ·cm water and adjusting the pH to 3.60 with aqueous sodium hydroxide. All buffer solutions were filtered through 0.45 µm PTFE syringe filters into 1-mL polypropylene vials (Agilent Technologies, Santa Clara, CA) before use. Buffer solutions were prepared weekly and stored at 4°C when not in use.
Standard and analysis solutions of the antioxidant ascorbic acid were prepared daily in degassed 18 MΩ·cm water with, in some cases, 0 to 100 mM NaCl added to adjust the sample conductivity. All solutions were filtered through a 0.45 µm PTFE syringe filter (JSI Scientific, Freehold, NJ) prior to use, and stored at 4°C when not in use.
Seven commercial beverage samples were obtained at a local retail outlet. Samples were filtered through a 0.45 µm PTFE syringe filter (JSI Scientific, Freehold, NJ) and diluted with either 18 MΩ·cm water or an aqueous NaCl solution prior to use. The beverage samples included Minute Maid apple juice (FEB0909 AM65 2126 CT809), Ocean Spray 100% cranberry juice (CT981 0507 PCJ), Welch's 100% grape juice (NE08104 20:33 P) and four different Glaceau Vitamin Water products: XXX acai-blueberry-pomegranate (B8206 14:02 A CT795), Formula 50 grape (B8164 10:10 B CT735), Rescue green tea (B8148 19:39 B CT735) and multi-v lemonade (B8140 22:37 C CT735). All analyses of beverage samples were carried out on the same day that the samples were first opened as many samples suffered from significant air oxidation upon overnight storage after being opened.
An Agilent 3D capillary electrophoresis system (Agilent Technologies, Santa Clara, CA) with photodiode array UV/Visible detection and Agilent Chemstations software was used for all separations. An HP 8452A diode array spectrophotometer (Hewlett-Packard, Palo Alto, CA) controlled by HP 89530A UV/Vis operating software on a PC were used for FRAP analysis.
A 50.0 µm i.d. unmodified fused silica capillary (Polymicro Technologies, Phoenix, AZ) 33.0 cm in length (24.5 cm effective) was used for all EMMA analyses. The running buffer was 50.0 mM phosphate at pH 7.2, and the capillary was maintained at a constant temperature of 25.0 °C. Absorbance detection was performed at 590 nm, 265 nm, and 210 nm, each with an 8 nm bandwidth. A 2.00 minute flush with run buffer was performed between runs. Each day, capillaries were reconditioned with 2.00 minute flushes of 1.0 M NaOH, 18 MΩ·cm water, and the running buffer, sequentially.
The EMMA procedure involved the use of the oxidizing agent 2,6 – dichlorophenolindophenol (DCIP) as outlined in Figure 1. A 150 mbar·s injection of sample is sandwiched between two 150 mbar·s injections of 0.60 mM DCIP. A 300 mbar·s plug of buffer is injected behind the analytes to guard against loss of reagent to the inlet vial. Following the injection sequence, an alternating potential of ±15.0 kV (2.7 sec at each polarity, 10 iterations), referred to as rapid polarity switching (RPS) (29), was applied to mix the antioxidants with DCIP. After the RPS step, a 10.0 kV separation potential is applied, to separate the remaining DCIP and the reduced DCIP product via capillary zone electrophoresis (CZE). L-Ascorbic acid was chosen as the calibration standard. For the analysis of commercial beverages, 400 µL of 5-fold diluted beverage was mixed with "x" µL of 1.00 mM ascorbic acid standard and (400-x) µL of 18 MΩ·cm water. All samples were vortex mixed in polypropylene samples vials (Agilent Technologies, Palo Alto, CA, USA) prior to being placed in the sample tray.
The ferric reducing antioxidant power (FRAP) assay was performed using a modified version of the procedure detailed by Benzie and Strain (8). The FRAP reagent consisted of a 10:1:1 ratio of solutions of acetate buffer, ferric chloride, and 2,4,6-tris(2-pyridyl)-s-triazine (TPTZ). The acetate buffer system was 100 mM at pH 3.6. The ferric chloride solution was 20.0 mM, prepared in 18 MΩ·cm water. The TPTZ solution was 10.0 mM, prepared in 40.0 mM HCl. FRAP calibration was accomplished via analysis of solutions consisting of 900 µL of FRAP reagent and a varied amount ("x" µL) of 0.500 mM ascorbic acid, with an appropriate volume (100-x µL) of 18 MΩ·cm water. For the analysis of commercial beverages, 100 µL of 10-fold diluted beverage was added to 900 µL of the FRAP reagent solution. The FRAP procedure was carried out at room temperature in a 0.50 cm path length disposable cuvette, and the absorbance of the product at 594 nm was recorded every 10 seconds for a period of 10 minutes. The final absorbance reading was used for quantification.
A schematic representation of the use of electrophoretically mediated microanalysis (EMMA) for the determination of total antioxidant capacity (TAC) is shown in Figure 1. The EMMA method employed here involves a sandwich assay in which a plug of approximately 9 nL of the antioxidant containing sample is injected between plugs of a reagent solution containing an excess of 2,6 – dichlorophenolindophenol (DCIP). When mixed, an electron transfer (ET) reaction between DCIP and the antioxidant(s) in the sample causes the formation of dichlorodihydroxydiphenylamine, the reduced form of DCIP. The system can be calibrated using ascorbic acid, a commonly used antioxidant standard for aqueous systems that is known to react rapidly with DCIP with 1:1 stoichiometry (30, 31). At neutral pH, DCIP carries a negative charge (32), allowing its two REDOX forms to be easily separated from one another and from other anions in a sample via electrophoresis. DCIP also has a strong chromophore that is conveniently sensitive to oxidation state, making it an attractive choice for wavelength-selective detection of the individual redox forms. Sandwiching the sample between plugs of DCIP establishes conditions under which sample antioxidants, regardless of charge, will mix with the DCIP reagent when the RPS mixing potential is initiated. The dual plug arrangement also allows for bidirectional diffusional mixing of any neutral antioxidants that may be present in the sample.
Typical electopherograms obtained via the in-capillary TAC analysis of an aqueous standard of L-ascorbic acid are given in Figure 2. A broad neutral peak marking the flow of the sample matrix past the detector is observed at about 2.3 minutes while the analytically useful DCIP peaks are observed at later times. The anionic nature of DCIP allows both the original reagent and its reduced product to be electrophoretically separated from other sample components. The excess DCIP reactant, which under the conditions employed migrates upstream with an electrophoretic mobility of ca. −2.1 × 10−4 cm2V−1s−1, absorbs strongly at 590 nm (peak 3 in Figure 2A) where many potential interfering ions will not absorb. The reduced DCIP product (peak 2, μep = −1.5 ×10−4 cm2V−1s−1) has a local absorbance maximum at 265 nm, and can be detected with greater sensitivity at this wavelength (Figure 2B). Both forms can be sensitively detected at very low wavelength (210 nm, Figure 2C) but at this wavelength there is a greater risk of co-migrating sample ions interfering with quantification (vide infra, Figure 6).
To explore the linearity of response, standard aqueous solutions of L-ascorbic acid concentration standards were analyzed with the EMMA methodology. As seen in Figure 3, increasing the concentration of ascorbic acid (bottom to top) in the sample plug leads to decreasing peaks areas for DCIP and correspondingly increasing peak areas for the reduced product. The calibration data for the decreasing DCIP peak area at 590 nm, its λmax, were described by y = (−370±10)x+ (642±6) with r2 = 0.997. Calibration data using the appearance of the reduced DCIP product peak areas at 265 nm were y = (132 ± 2)x− (1.3±1.2) with r2 = 0.9991. Thus, the in-line reaction scheme can be calibrated via the disappearance of the DCIP, as well as by the appearance of the reduced DCIP product; however, the RSDs at 590 (≤ 2.5 %) were uniformly superior to those at 265, which were as high as 10% with the lowest concentration samples. Still, both 590 nm and 265 nm data were monitored with real samples (i) to ensure that DCIP was not completely depleted and (ii) as a control for the possibility of bias from an anion from the sample co-migrating with DCIP. We did not encounter the latter case with any of the samples analyzed.
The analytical response of the reported method to the polyphenolic antioxidant catechin was also examined. The method gave linear and reproducible results for this large antioxidant molecule, but the calibration slope with catechin was considerably lower than with ascorbic acid, indicating differing response factors for these two antioxidants. With the EMMA method presented, the calibration data for catechin, using the signal for DCIP at 590 nm is described by y = (−96±3)x + (650±2) (r2 = 0.996). That is, the sensitivity of this method for catechin was only about 26% of that seen with ascorbic acid. The lower response factor for catechin is most likely due to a combination of less favorable thermodynamic factors (lower E° value for the redox half reaction) and slower electron transfer kinetics. Differing response factors is one of the necessary compromises one makes when measuring TAC, regardless of the methodology employed, and the reported EMMA methodology presented here is also subject to this limitation. Indeed, significant and sometimes drastically different values for TAC have been reported for analyses of the same samples with published methods for TAC determination (11–14). Nonetheless, TAC remains a useful measure of the aggregate ability of a given sample to adjust to oxidative assault, and this method reported here is shown to respond linearly, albeit with different slope, to two very different antioxidants.
One of the important aspects to consider with any electromigration method is the salt content of the sample. Because the conductivity of the sample can affect the local voltage field within the sample plug, and therefore the initial migration velocity of ions in that environment, varied ionic content between standard and sample matrixes can be problematic. With conventional CE separations, samples are often prepared in low ionic strength solutions so ionic samples will concentrate or "stack" at the sharp boundary in the field strength that develops where sample matrix and the background buffer meet. Conversely, a deconcentrating effect can occur when very conductive samples are injected. The effect of sample salt on performance with the in-line EMMA system does not follow this trend. The reaction between DCIP and ascorbate was investigated by analyzing a series of ascorbic acid standards that also contained 100 mM NaCl (Figure 4). The calibration data remained linear (y=(137 ± 2)x − (6.0±0.9), r2 = 0.9997), and reproducibility was not compromised. However, the DCIP response is no longer detected as a single peak, but rather as two overlapping peaks corresponding to the discrete plugs of this reagent that were injected but did not merge into a single zone by the time they reached the detection point. That is, the field within the sample was too low to allow the DCIP to fully traverse the sample and merge with the second plug of DCIP prior to detection. In addition, the reduced DCIP peak widened and also became somewhat bimodal. The summed peak areas from the two peaks (or single bimodal peak) still indicate good linearity, and the absence of an ascorbic acid peak indicates that the reaction between DCIP and ascorbic acid appears to go to completion with or without salt in the sample. However, the initial migration dynamics of DCIP are clearly altered. Without salt in the sample, the applied separation potential creates a strong voltage field within the resistive sample plug and the anionic DCIP migrates rapidly through the resistive sample matrix and merges with the DCIP anions on the other side of the sample plug. When NaCl is added to the standards, the samples have higher conductivity which slows the migration velocity of DCIP within the sample plug, and two zones of DCIP are evident at the detection point.
One must be careful to address the potentially troublesome effect of varied sample ionic strength on quantifications of TAC. One should be cautioned against external calibration with aqueous (no salt) ascorbic acid standards; this approach gave rise to apparent percent recoveries of ascorbic acid that varied from 65 to 110%. To probe the range and magnitude of the effect of sample conductivity, varied amounts of salt were added to an apple juice sample and gradually lower DCIP peak areas were obtained with increased salt content. The decrease in area amounted to only about a 10% difference between no salt and 100 mM salt added to the sample (Figure 5). Higher amounts of salt had little additional effect, presumably because the conductivity of the sample had reached that of the background buffer (at 100 mM NaCl). However, a modest increase in reproducibility was achieved by adding salt to the samples.
While the problems associated with varied sample conductivity could be minimized by adding a large amount of salt to all samples and standards, the method of standard additions is perhaps the best approach to quantification. Several commercial juice samples were analyzed with the EMMA approach using the method of multiple standard additions to obtain quantitative data. Some minor baseline disruptions were observed in the region of interest with some samples at 210 nm (Figure 6), and these were avoided by performing detection at 265 nm. Typical raw data obtained at 265 nm for a juice sample spiked with various amounts of ascorbic acid is given in Figure 7. For all samples analyzed, the product peak area at 265 nm exhibited linearity with the volume (0, 80, 160 and 240 µL) of 1.00 mM ascorbic acid addition to the 800 µL total sample volume, but the slope of resultant data varied from 0.070 to 0.206, depending on the sample. The resulting quantitative determinations of TAC, expressed as millimolar equivalents of ascorbic acid, are presented in Table 1. The same beverage samples were also analyzed with the commonly used FRAP assay and despite the differing redox chemistry and solution pH between the two methods, the TAC values are surprisingly comparable. It is perhaps not too surprising that the samples most likely to have polyphenolic compounds (grape juice and green tea) show the largest deviations between the two techniques. It appears from this limited, but diverse set of samples that the rapid and easily automatable EMMA methodology, used in conjunction with the method of standard additions, may be a viable alternative to more cumbersome methods to determine TAC.
The use of the oxidant DCIP as an in-capillary reactant appears to be a reasonable alternative to published methodologies for assessing the total antioxidant capacity (TAC). The EMMA methodology requires minimal amounts of sample and generates negligible waste while providing quantitative data that are well within the range of variance typically seen when comparing TAC methods. Owing to differences in ionic strength in commercial samples, a spiking approach to quantification appears to be advisable. Overall, this new in-line methodology shows reasonable promise as a rapid and low-volume alternative to the current TAC methodology.
This work was supported by the National Institutes of Health (R15 EB003854-02). The authors wish to thank Dr. John Stahl of Geneva College for the suggestion to use DCIP, and for valuable comments regarding this work.