Quantification of taurine in energy drinks using 1H NMR
Abstract
The consumption of so called energy drinks is increasing, especially among adolescents. These bever- ages commonly contain considerable amounts of the amino sulfonic acid taurine, which is related to a magnitude of various physiological effects. The customary method to control the legal limit of taurine in energy drinks is LC–UV/vis with postcolumn derivatization using ninhydrin. In this paper we describe the quantification of taurine in energy drinks by 1 H NMR as an alternative to existing methods of quantifica- tion. Variation of pH values revealed the separation of a distinct taurine signal in 1 H NMR spectra, which was applied for integration and quantification. Quantification was performed using external calibration (R2 > 0.9999; linearity verified by Mandel’s fitting test with a 95% confidence level) and PULCON. Tau- rine concentrations in 20 different energy drinks were analyzed by both using 1 H NMR and LC–UV/vis. The deviation between 1 H NMR and LC–UV/vis results was always below the expanded measurement uncertainty of 12.2% for the LC–UV/vis method (95% confidence level) and at worst 10.4%. Due to the high accordance to LC–UV/vis data and adequate recovery rates (ranging between 97.1% and 108.2%), 1 H NMR measurement presents a suitable method to quantify taurine in energy drinks.
1. Introduction
According to the Directorate General for Health and Consumers of the European Commission, the name energy drink describes a category of beverages comprising various combinations of sub- stances such as carbohydrates, vitamins, minerals and commonly one or more of the substances caffeine, taurine, glucuronolac- tone [1] and inositol. The label energy drink (ED) is a commercial designation that aims to implicate that EDs provide energy in a nutritional and physiological sense. These beverages are increas- ingly offered on the market [2] and corresponding to a survey in 2012 especially adolescent people (aged between 10 and 18 years) show a high inclination to the consumption of EDs in Europe [3].
Taurine is an amino sulfonic acid that can be naturally found in mammalian muscles [4] and presents the most abundant free amino acid in animal tissues [5]. A variety of physiological effects are connected to the exposure of taurine [5,6], e.g. higher per- formances with respect to hormonal responses [7], and protection functions for heart [8], vessels [9] and kidney [10] have been described. However, health effects connected to ED consumption are frequently discussed, especially as previous studies on chronic toxicity or carcinogenicity were inadequate and the available data were insufficient to assign an upper safety level for the daily intake of taurine [1]. In Germany, taurine concentrations up to 300 mg/kg are legally granted for flavoring [11] and as caffeine-bearing bev- erages, EDs can contain concentrations up to 4000 mg/l taurine [12]. As these legal limits are generally exhausted [2], the taurine concentrations in EDs account either for 300 mg/l or 4000 mg/l. Hence, the consumption of EDs increases the daily intake of taurine from omnivore diets, varying between 40 mg/day and 400 mg/day [13–15], substantially.
In order to control the legal limits of taurine in EDs, the most frequently applied precise method is HPLC [2]. Besides, fourier transform infrared (FTIR) spectroscopy can be used [2], but requires prior calibration by means of an appropriate amount of samples with known taurine content. A feasible option for quantification purposes is presented by 1H NMR, considering concomitant bene- fits like short measuring times, non-destructive analysis, no need for prior isolation of analytes in mixtures and the possibility of measuring several analytes in mixtures simultaneously [16]. Fur- thermore NMR spectroscopy can be regarded as a primary method [17] and thus, is very suitable for quantification purposes [18]. The application of 1H NMR to quantify taurine has already been described for cells and tissue extracts [19] and urine [20]. This paper describes the quantification of taurine in EDs by 1H NMR, based on improved conditions of sample preparation.
2. Materials and methods
2.1. NMR measurement
2.1.1. Chemicals
TSPd4 (3-(trimethylsilyl)propionic acid-D4 sodium salt, 98 atom% D), D2O (99.9 atom% D), NaN3 and KH2PO4 were purchased from Merck (Darmstadt, Germany). NaOH pellets were purchased from VWR (Leuven, Belgium), HCl (37%) from Sigma Aldrich (Saint Louis, USA) and taurine from Fluka (Buchs, Switzerland).
2.1.2. Sample preparation
20 different EDs were purchased from supermarkets in Germany. Each sample preparation was carried out directly before measurement. After degassing the samples by ultrasonication for 10 min, 800 µl of EDs were mixed with 100 µl of TSPd4 solution (containing 7 mM TSPd4 and 2 mM NaN3 in D2O) and 100 µl of 1 M NaOH. Calibration standards were prepared mixing 800 µl with 100 µl of TSPd4 solution and 100 µl of 1 M KH2PO4-buffer pH 11 (13.61 g KH2PO4 were filled up to a volume of 100 ml with deminer- alized water prior to pH-adjustment with NaOH pellets for rough and 1 N NaOH for final adjustment). For spiked recoveries 700 µl ED sample were mixed with 100 µl TSPd4 solution, 100 µl 1 M NaOH and subsequently either 100 µl of 100 mM aqueous taurine solution or 100 µl demineralized water were added.
The pH of all samples was adjusted to pH 11, using 1 M NaOH or 1 M HCl. Thereby the amount of NaOH or HCl for pH adjust- ment was noted down to consider the resulting dilution factors for examination. Finally 600 µl of the pH adjusted solutions were filled into 5 mm NMR-tubes (PP-507, purchased from Rototec-Spintec, Griesheim, Germany) for NMR-measurement.
2.1.3. 1H NMR parameters
Hardware and software equipment used for NMR measurement was completely purchased from Bruker (Bruker BioSpin, Rheinstet- ten, Germany) comprising a Bruker Avance 400 spectrometer with a 5 mm SEI probe with Z-gradient coils, an automatic Sample Changer (B-ACS-60), a BCU 05
temperature unit and TopSpin 3.0 software.
All samples were measured at 301.8 ± 0.1 K, without rotation and using 4 dummy scans prior to 128 scans. Acquisition parameters have been set as follows: size of fid = 64 k, spectral width = 20.5524, receiver gain = 16, acquisition time = 3.98 s, relax- ation delay = 10 s, FID resolution = 0.25 Hz. Data acquisition was achieved using an experiment with a NOESY-presaturation pulse sequence (Bruker 1D noesygppr1d) with water suppression via irradiation of the water frequency during the recycle and mixing time delays. Received spectra were automatically phased, baseline- corrected and calibrated to the TSP signal at 0.0 ppm.
2.1.4. Quantification
The taurine quantification was achieved by using both exter- nal calibration and the principle of PULCON (pulse length based concentration determination). To perform integration, a constant integration range of 35 Hz (each 17.5 Hz left- and right-hand from signal top) was used. A calibration curve was achieved by mea- suring calibration standards with concentrations ranging from 125.1 mg/l (1 mM) to 625.7 mg/l (5 mM) and from 1251.4 mg/l (10 mM) to 6257.0 mg/l (50 mM). To perform PULCON, a sample containing 4000.0 mg/l (32 mM) taurine was directly prepared and measured with each test series.
2.2. LC–UV/vis measurement
2.2.1. Chemicals
HCl (37%) was purchased from J.T. Baker (Phillipsburg, USA), sodium acetate trihydrate from BDH laboratory supplies (Poole, England), methanol, norleucine and taurine from Fluka (Buchs, Switzerland), citric acid from Grüssing (Filsum, Germany), di- sodium EDTA, ethanol, LiOH monohydrate, ninhydrin, phenol and potassium acetate from Merck (Darmstadt, Germany), acetic acid (100%), boric acid and LiCl from Roth (Karlruhe, Germany), caprylic acid and isopropyl alcohol from Sigma Aldrich (Steinheim, Germany) and reducing agent (order N◦ 6005007) from Sykam (Eresing, Germany).
Solvents for LC–UV/vis analysis were prepared as follows: for buffer A (0.12 M, pH 2.9): 5.04 g LiOH monohydrate, 15 g citric acid, 50 ml ethanol, 8 ml HCl (37%) and 0.1 ml caprylic acid, for buffer B (0.3 M pH 4.2): 4.2 g LiCl, 8.4 g LiOH monohydrate, 15 g citric acid, 9 ml HCl (37%) and 0.1 ml caprylic acid and for buffer C (0.3 M pH 10.5): 4.2 g LiCl, 8.4 g LiOH monohydrate, 10 g citric acid, 2.5 g boric acid, 9 ml HCl (37%) and 0.1 ml caprylic acid were each filled up to a volume of 1 l with demineralized water and subsequently pH adjusted with 37% HCl. Regeneration solution (0.5 M) consisted of an aqueous solution of 21 g/l LiOH monohydrate and 0.4 g/l di-sodium EDTA. Ninhydrin solution was prepared mixing 20 g nin- hydrin, 2 g phenol, 1 ml of reducing agent and 600 ml methanol with 400 ml acetate buffer (2 M sodium acetate, 2 M potassium acetate, pH adjusted to pH 5.51 with acetic acid (100%)).
2.2.2. Sample preparation
Degassing of EDs was performed by shaking and filtering sam- ples three times through a folded filter. Samples were diluted with demineralized water to taurine contents ranging between 15 mg/l (0.12 mM) and 20 mg/l (0.16 mM) and 1 M aqueous norleucine ref- erence was added, so that the final concentration in the diluted sample was 0.1 M norleucine.
2.2.3. LC–UV/vis parameters
For chromatographic analysis of taurine an amino acid ana- lyzer (Sykam, Eresing, Germany) with a cooling part for reagents (S 7130), an autosampler (S5200), a quaternary pump-system, a module for amino-acid derivatization (S4300) and Chromeleon 6.6 software (ThermoFisher Scientific, Waltham, Massachusetts, USA) were used. Chromatographic separation was carried out with an ammonium-filter-column (LCA K05/Li, Peek 4.6 × 100 mm) prior to a cation-column (LCA K07/Li, PEEK 4.6 × 150 mm, filled with CK 10F). The mobile phase consisted of buffer A, B, C and regeneration solution (D) using the following gradient elution with a flow rate of 0.45 ml/min: A:B:C:D = 100:0:0:0 (v/v/v/v) from start to 13 min, 79:21:0:0 from 14 to 33 min, 62:38:0:0 at 47 min, 0:100:0:0 at 66 min, 0:0:100:0 from 71 to 81 min, 0:0:86:14 from 84 to 86 min, 0:0:80:20 at 86.1 min, 0:0:78:22 at 102 min, 0:0:0:100 from 102.1 to 106.1 min and 100:0:0:0 from 106.2 to 127 min. From start to 104 min ninhydrin solution was added to the reaction coil (tem- perature of 130 ◦C) and subsequently water, each with a flow rate of 0.25 ml/min. The column temperature gradient program started with 37 ◦C for 40 min, 60 ◦C from 58 to 77 min, 74 ◦C from 95 to 106 min and 37 ◦C from 116 min on. As injection volume 100 µl were used and UV/vis detection was achieved at λ = 570 nm.
2.2.4. Quantification
Classification of analytes in the spectra was based on charac- teristic retention times and signals were integrated automatically. The quantification was carried out by referring the signal inten- sity of taurine to the signal intensity of the internal standard norleucine.
Fig. 1. Overlaid spectra of an ED (spotted line) and a taurine reference (continuous line) at different pH values (left) with an extended view on spectra at pH 11 (right).
3. Results and discussion
3.1. Sample preparation for NMR spectroscopy and signal separation
Research on the quantification of taurine in EDs via one dimen- sional 1H NMR has already been accomplished, but proved to be difficult due to overlapping signals [21]. The existence of separated signals for integration and quantification purposes is an important prerequisite. In order to achieve completely separated signals, the influence of different solvents, pH values and ion concentrations on the chemical shift can be utilized [18]. Thus, the applicability of different pH values for quantitative NMR measurement (qNMR) of taurine in EDs was tested. Fig. 1 shows the spectra of taurine and an ED, adjusted to different pH values ranging between pH 3 and pH 11. It distinctly illustrates the improved signal separation of a taurine multiplet at pH 11, which is formed by two overlapping taurine multiplets and was used for quantification. The drift of the taurine signal as a function of different pH values indicates a titra- tion curve, suggesting a pKa value of the amino function between 8 and 10. This is consistent with reported pKa values of 9.1 or 9.0 [22,23].
As pH adjustment to pH 11 constitutes a strong modification compared to the original pH value of pH 3–4, the stability of tau- rine at pH 11 was analyzed. Therefore, an ED (pH adjusted to pH 11) was measured at different points in time for a period of 20 h. With regard to interday deviations, this experiment was repeated on three consecutive days. As Fig. 2 demonstrates, the variation of the signal intensity is marginal and even more influenced by inter- day variations than by different points in time after pH adjustment.After 20 h the percentage of signal intensity compared to the ini- tial signal intensity still amounted to 99.14 ± 1.27%. Consequently, it can be assumed that the stability of taurine at pH 11 is sufficient for quantification.
3.2. Validation of the NMR and LC/UV–vis methods
Precision was determined by preparing and measuring one ED sample repeatedly (n = 5), yielding a relative standard deviation of 1.85% for LC–UV/vis and 0.79% for NMR analysis. The limit of detec- tion (LOD) and the limit of quantification (LOQ) were determined according to DIN 38402 [24] (significance level 5%; uncertainty of result 33%). LOD was 0.6 mg/l for LC–UV/vis and 2.6 mg/l for NMR analysis, LOQ was 2.4 mg/l for LC–UV/vis and 10.4 mg/l for NMR analysis (measuring 5 equidistant concentration levels between 5 and 25 mg/l for LC–UV/vis and 125.1 and 625.7 mg/l for NMR anal- ysis).
The expanded measurement uncertainty of LC–UV/vis anal- ysis amounted to 12.2%, calculated as the combined standard uncertainty (uc) multiplied by k = 2 (95% confidence level). For the calculation of uc the random element of uncertainty (repro- ducibility within the laboratory, determined by relative standard deviations of results and result ranges measuring an ED sample with 5 repetitions) and bias in measurement (determined by the recovery rate of 11 spiking experiments) were used. To analyze the recovery rate, EDs were spiked with aqueous taurine solution as fol- lows: ED:taurin = 7:1 (v/v) with c(taurine) = 1251.4 mg/l (100 mM) and undiluted ED for 1H NMR (5 different samples; 3 repetitions for each sample), ED:taurine = 1:1 (v/v) with c(taurine) = 20 mg/l and diluted ED (1:200, v/v) for LC–UV/vis (2 different samples; 5 repe- titions for each sample). Results of recovery rates are discussed in Section 3.3.4.
3.3. Quantification of taurine
In order to analyze a wide range of different samples, 20 EDs varying in brand and ingredients were purchased. These EDs cov- ered 11 different brands, hereof 3 EDs with sweeteners instead of sugar and 1 ED with reduced sugar content in combination with sweeteners. Moreover 4 EDs contained fruit concentrates and 1 ED tea. To ensure similar integration terms despite small variations of the chemical shift of taurine, a permanent integration range and adjusted bounds subject to the chemical shift were used.
Fig. 2. Stability of taurine at pH 11 as a function of time, painted as percentage of initial taurine signal intensity.
3.3.1. Calibration curve
As mentioned in the introduction, EDs contain either around 300 mg/l or 4000 mg/l taurine. Given the large interval between these concentrations, two separate calibration graphs with concentrations ranging from 125.1 mg/l (1 mM) to 625.7 (5 mM) and from 1251.4 mg/l (10 mM) to 6257 mg/l (50 mM) were elab- orated. High correlation coefficients (R2 > 0.9999) and linearity (verified by Mandel’s fitting test; F-values were lower than reported values in F-tables with a confidence level of 95%) ensure optimal quantification conditions for both calibration graphs.
3.3.2. PULCON
In general qNMR can be performed by using an internal or an external reference. The choice of an adequate internal reference can be cumbersome with regard to the magnitude of requirements that need to be fulfilled: e.g. the reference signal has to be distinct from the signals in the sample, shall not interact with sample molecules, has to be soluble in the sample and besides that the relaxation prop- erties should be ideally similar to the respective molecule to be quantified [25]. Hence, the usage of external references provides a suitable solution and is one basis for PULCON (pulse length based concentration determination). This method derives from the fact that the NMR signal strength is inversely proportional to the 90◦ pulse length [26,27]. Combined with the direct proportionality of signal intensity and the number of nuclei evoking the signal [18] the respective concentration of the sample can be calculated from a reference with known concentration [25].
To perform PULCON, a sample containing 4000 mg/l (32 mM) was measured with each test series and defined as quantification reference (using the ERETIC 2 quantification tool offered by Bruker TopSpin 3.0 software). Quantification of EDs was performed by
comparing the taurine signal intensity to the signal intensity of the quantification reference, considering the respective 90◦ pulse
length.
3.3.3. Comparison of qNMR and LC–UV/vis results
The results of the NMR quantification of the 20 EDs using the cal- ibration curve and the PULCON method as well as LC–UV/vis results are displayed in Table 1. The percentage of the deviation between 1H NMR results from LC–UV/vis data ranged between −10.4% and 8.7% for 1H NMR using calibration and −12.0% and 11.0% for 1H NMR using PULCON. Hence, for both versions of qNMR techniques the deviations from LC–UV/vis data are always smaller than the expanded measurement uncertainty of LC–UV/vis measurement of 12.2% (computation shown in part 3.2 validation) itself and thus, offer a suitable alternative to LC–UV/vis analysis. Comparing the qNMR quantification techniques among themselves, the results achieved by PULCON are slightly inferior, as the absolute value of deviation from LC–UV/vis data averaged 5.5 ± 3.7%, which is marginally higher than 4.5 ± 3.0% for the calibration curve method. However, the average interday range (each sample was prepared and measured twice on different days) amounts to 1.1 ± 1.5% for PULCON and 1.9 ± 1.2% when using the calibration curve. Further- more the quantification based on PULCON provides a rapid and comfortable option for routinely sample measurement, as one ref- erence with known taurine content replaces several solutions with various concentration levels to calculate a calibration curve.
3.3.4. Spiked recoveries
Besides, it has to be kept in mind that LC–UV/vis data were used as reference for examination. This leads to the fact that possible quantification errors of LC–UV/vis data are not considered when validating qNMR on the basis of the deviation between LC–UV/vis and qNMR results. Thus, proper qNMR results are possibly down- graded due to errors of LC–UV/vis data. Otherwise a simultaneous drift of error for qNMR and LC–UV/vis causes minor deviations between the two methods themselves despite potentially high deviations from factual values. One considerable source of error for LC–Uv/Vis analysis is the requirement for derivatization: incom- plete reaction with ninhydrin provides misleadingly lower taurine concentrations.
In order to investigate the quality of qNMR for taurine quantification independently from LC–UV/vis results, spiked recoveries of 5 different ED samples were analyzed on three consecutive days by means of external calibration. The mean recovery rates ranged between 97.1% and 108.2% yielding an average value of 104.2 ± 4.9% (see Table 2). In comparison to this, recovery rates for LC–UV/vis amounted to 95.2 ± 2.1%. Thus, the previously shown suitability of qNMR for taurine quantification is confirmed by results of recovery experiments.
4. Conclusion
Summing up the results achieved, qNMR provides a feasible option to quantify taurine in EDs, as it is both convenient and suf- ficiently accurate. Due to facile sample preparation and short time of sample measurement (NMR measurement took 31 min instead of 127 min for LC–UV/vis analysis) a huge amount of samples can be measured quickly, replacing elaborate conventional methods.
A common argument against NMR measurement is high cost. However, LC-analysis involves the requirement for high priced columns that need to be renewed at regular intervals. Considering the negligible costs once that hardware and software are purchased, NMR is able to be price-competitive in the long term.