Sensitive Quantitation of Polyamines in Plant Foods by Ultrasound

Apr 28, 2014 - polyamine analysis, enabling great sensitivity and selectivity to be achieved. ..... presented in Figure 3 in the form of a Pareto char...
1 downloads 0 Views 3MB Size
Article pubs.acs.org/JAFC

Sensitive Quantitation of Polyamines in Plant Foods by UltrasoundAssisted Benzoylation and Dispersive Liquid−Liquid Microextraction with the Aid of Experimental Designs Edgar Pinto, Armindo Melo, and Isabel M. P. L. V. O. Ferreira* REQUIMTE/Departamento de Ciências Químicas, Laboratório de Bromatologia e Hidrologia, Faculdade de Farmácia, Universidade do Porto. Rua de Jorge Viterbo Ferreira 228, 4050-313 Porto, Portugal S Supporting Information *

ABSTRACT: A new method involving ultrasound-assisted benzoylation and dispersive liquid−liquid microextraction was optimized with the aid of chemometrics for the extraction, cleanup, and determination of polyamines in plant foods. Putrescine, cadaverine, spermidine, and spermine were derivatized with 3,5-dinitrobenzoyl chloride and extracted by dispersive liquid−liquid microextraction using acetonitrile and carbon tetrachloride as dispersive and extraction solvents, respectively. Two-level full factorial design and central composite design were applied to select the most appropriate derivatization and extraction conditions. The developed method was linear in the 0.5−10.0 mg/L range, with a R2 ≥ 0.9989. Intra- and interday precisions ranged from 0.8 to 6.9% and from 3.0 to 10.3%, respectively, and the limit of detection ranged between 0.018 and 0.042 μg/g of fresh weight. This method was applied to the analyses of six different types of plant foods, presenting recoveries between 81.7 and 114.2%. The method is inexpensive, versatile, simple, and sensitive. KEYWORDS: plant foods, putrescine, cadaverine, spermidine, spermine, response surface methodology, HPLC−DAD



INTRODUCTION The polyamines putrescine, cadaverine, spermidine, and spermine are low-molecular-weight organic amines with aliphatic structure, which have been implicated in several biological processes, such as plant growth and development.1 These compounds are naturally found in many plant foods and are involved in the regulation of a diverse range of vital cellular processes in eukaryotic cells. In plants, the most common polyamines are putrescine, spermidine, and spermine and are usually localized in the cytoplasm and organelles (e.g., vacuoles, mitochondria, and chloroplasts) of plant cells.2,3 The dietary intake of polyamines influences human health. Their biological role in humans includes participation in cell growth and proliferation, pain sensitization, stabilization of DNA, gene transcription, apoptosis, regulation of the immune response, maintenance of intestinal function, and protein synthesis.4,5 It has been suggested that increasing polyamine intake has several health benefits, such as decreasing ageassociated pathologies and increasing longevity.6 However, a polyamine-deficient diet has also been proposed as an efficient nutritional strategy for several human conditions.7 Plant foods have very low levels of polyamines, ranging from a few milligrams to less than a 100 mg/kg. Thus, methods to perform polyamine quantitation in plant foods frequently report results as non-detectable or below the limit of detection (LOD)/limit of quantitation (LOQ) because of the high sensitivity required.8−10 Plant foods are a very complex and variable matrix;2,3 thus, efficient cleanup and pre-concentration techniques are needed to obtain sensitive and accurate quantitation of polyamines. Because plant foods are solid matrixes, an extraction/deproteinization is also required. The deproteinization process is usually performed with diluted © 2014 American Chemical Society

strong acids, such as perchloric, hydrochloric, and trichloroacetic acids.11−13 Chemical derivatization has become widely accepted in polyamine analysis, enabling great sensitivity and selectivity to be achieved. Nowadays, there are several derivatization reagents that can be used for ultraviolet (UV) and/or fluorescence detection, including o-phthaldialdehyde, dansyl chloride, 6aminoquinolyl-N-hydroxysuccinimidyl carbamate, dabsyl chloride, 3,5-dinitrobenzoyl chloride, 1,2-naphthoquinone-4-sulfonate, benzoyl chloride, 4-chloro-3,5-dinitrobenzotrifluoride, diethyl ethoxymethylenemalonate, and others.14,15 Among them, 3,5-dinitrobenzoyl chloride shows several advantages, such as its low price, commercial availability, and fast reaction times, at room temperature.16 Many sample preparation techniques have been developed for cleanup purposes. Liquid−liquid extraction (LLE) and solid-phase extraction (SPE) are some of the most widely used techniques; however, they require large quantities of organic solvents and are time-consuming.12,13 Solid-phase microextraction (SPME) and single-drop microextraction (SDME) have been increasingly used in the last few years because of their eco-friendly nature.17,18 However, these procedures also had some drawbacks, such as high price, short durability of the extraction fibers, and sample carryover.19 Dispersive liquid− liquid microextraction (DLLME) is a novel sample preparation technique based on a ternary component mixture consisting of an aqueous sample, a dispersive solvent, and an extraction Received: Revised: Accepted: Published: 4276

February April 28, April 28, April 28,

24, 2014 2014 2014 2014

dx.doi.org/10.1021/jf500959g | J. Agric. Food Chem. 2014, 62, 4276−4284

Journal of Agricultural and Food Chemistry

Article

MD-2010 Plus multiwavelength detector, and a column oven model 7981 (Jones Chromatography, Hengoed, U.K.). The column used was a 250 × 4.6 mm inner diameter, 5 μm, Ultracarb 30-ODS RP-18 (Phenomenex, Milford, MA). The Borwin PDA controller software (JMBS Developments, Le Fontanil, France) was used for system control and data acquisition. The column temperature was set at 35 °C, and the chromatographic separation was carried out by gradient elution with two solvent mixtures at a 1 mL/min flow rate. Solvent A consisted of 10 mM aqueous sodium formate/formic acid buffer at pH 3, and solvent B consisted of acetonitrile. The linear gradient program was as follow: 0−5 min, 40% B; 5−11 min, 58% B; 11−15 min, 58% B; 15−20 min, 65% B; 20−23 min, 65% B; 23−28 min, 95% B; 28−30 min, 95% B; and 30−40 min, column rinse and re-equilibration. The total run time, including polyamine separation, column rinse, and reequilibration, was of 40 min. Diode array detection was carried out at the wavelength of 250 nm. Sample Preparation. Plant food samples (lettuce, carrot, onion, cabbage, tomato, and apple) were purchased at local supermarkets. Samples were washed with ultrapure water, frozen at −80 °C, and freeze-dried. The lyophilized samples were homogenized by grinding in a blender and sieved through a nylon sieve of 150 μm mesh size. Free polyamines were determined according to Zhou and Yu,11 with minor modifications. Lyophilized samples (100 mg) were homogenized in 2.5 mL of 5% (v/v) cold HClO4 and incubated at 4 °C for 1 h. After centrifugation at 17000g for 30 min, the supernatant was collected and used as a sample extract to determine the content of polyamines. Derivatization, Cleanup, and Pre-concentration. A 2 mL volume of standard solution or sample extract was placed in a 10 mL tube, and 0.5 M borate buffer (pH 10.0) and 2 M NaOH were added to adjust pH to 10. After that, 680 μL of 3,5-dinitrobenzoyl chloride in acetonitrile (56 mM) was added to the previous solution, which was subjected to vortexing. Afterward, the mixture was ultrasonicated for 15 min at room temperature. Benzoylated polyamines were extracted by the DLLME method by adding 500 mg of NaCl and 205 μL of carbon tetrachloride (extraction solvent) into the sample solution to form a cloudy solution. After vortexing for 1 min, the mixture was centrifuged for 5 min at 5000 rpm, and the organic phase was aspirated into a microsyringe and collected into a 2 mL glass vial. The organic phase was evaporated in a gentle stream of N2 and reconstituted in 100 μL of acetonitrile. Finally, 20 μL was injected into the HPLC system for analysis. Experimental Design. Several trials were conducted to optimize the benzoylation and DLLME conditions for the quantitative analyses of polyamines in plant foods using RSM. A two-level (2k) factorial design was created to screen the main variables affecting the derivatization and extraction procedures. This technique is a powerful tool for screening the key variables in a multivariable system. Two different full factorial designs were created for trials at low (−1) and high (+1) concentration levels with five central points. The experiments were performed to evaluate the significance of seven variables (3,5-dinitrobenzoyl chloride concentration, 3,5-dinitrobenzoyl chloride volume, derivatization time, derivatization temperature, dispersive solvent volume, extraction solvent volume, and amount of NaCl) under study. In the first full factorial design, the evaluated variables were 3,5-dinitrobenzoyl chloride concentration (35−65 mM), 3,5-dinitrobenzoyl chloride volume (400−800 μL), derivatization time (15−40 min), and derivatization temperature (20−50 °C). The screening of the main variables affecting the derivatization procedure was performed without performing the extraction procedure to exclude potential effects arising from it. In the second full factorial design, the variables dispersive solvent volume (400−800 μL), extraction solvent volume (50−250 μL), and amount of NaCl (500−800 mg) were evaluated. The DLLME procedure was carried out after fixing the most suitable derivatization conditions. Afterward, the significant variables were optimized using a CCD. Experiments with three independent variables, 3,5-dinitrobenzoyl chloride concentration (X1), dispersive solvent volume (X2), and extraction solvent volume (X3), were conducted by applying a CCD consisting of a complete 20-factorial design with six center points and

solvent. It presents several advantages, such as simplicity, high recovery, quickness, and low cost.20 Several methods have been developed for the analysis of polyamines, including micellar electrokinetic chromatography,21 gas chromatography (GC),17 thin-layer chromatography,22 ion chromatography,23 and liquid chromatography (LC).24 Overall, LC is the most widely used technique in the determination of polyamines in different kinds of food because of its high selectivity, sensitivity, and applicability. Growing interest arises in coupling derivatization and DLLME, followed by chromatographic separation (i.e., LC or GC) for the analysis of different kinds of compounds.25,26 The potentiality of these methods to simplify experimental procedures, decrease sample loss, increase method sensitivity, and remove interferences is seen as a great improvement in the analytical chemistry field.27 The development of a derivatization and DLLME method requires selection of several variables (concentration of derivatization reagent, reaction time, reaction temperature, dispersive solvent volume, extraction solvent volume, and ionic strength) to establish optimum conditions. The optimization of a multivariable system through a onevariable-at-a-time procedure can be time-consuming and neglect potential interactions between variables. The two-level full factorial design is a useful tool for preliminary screening of experimental factors that needed to be optimized, whereas response surface methodology (RSM) aids to understand interactions among the parameters that should be optimized to determine the optimum operational conditions for the system or to determine a specific region that satisfies the operating specifications.28 Consequently, these chemometric methods, when applied to derivatization and miniaturized extraction procedures, improve the quality of analytical data. However, until now, few studies apply these procedures. A new method is presented for the detection and quantitation of putrescine, cadaverine, spermidine, and spermine using ultrasound-assisted benzoylation followed by DLLME and high-performance liquid chromatography−diode array detection (HPLC−DAD). The benzoylation procedure and extraction conditions were optimized with the aid of twolevel full factorial designs and central composite design (CCD). The developed method was applied to several plant food samples.



MATERIALS AND METHODS

Chemicals and Standard Solutions. Putrescine dihydrochloride, cadaverine dihydrochloride, spermidine trihydrochloride, 1,7-diaminoheptane [internal standard (IS)], spermine tetrahydrochloride, hydrochloric acid, acetonitrile, methanol, acetone, and 3,5-dinitrobenzoyl chloride were purchased from Sigma-Aldrich (St. Louis, MO). Boric acid, sodium hydroxide, potassium chloride, and sodium chloride were from VWR (Radnor, PA). Perchloric acid (HClO4), carbon tetrachloride, chlorobenzene, chloroform, dichloromethane, tetrachloroethylene, and trichloroethylene were obtained from Merck (Darmstadt, Germany). Standard stock solutions were prepared by dissolving each compound in 0.1 M HCl at a final concentration of 1000 mg/L. Working standard solutions were prepared daily from the stock solutions by adequate dilution with 0.1 M HCl. Sodium borate buffer was prepared from dilution of H3BO3 and KCl in water (0.5 M each), followed by titration with NaOH (0.5 M) to pH 10. 3,5Dinitrobenzoyl chloride (56 mM) was prepared by dissolving a proper amount of the solid compound in acetonitrile. Instrumentation and Chromatographic Conditions. Chromatographic analysis was performed with an analytical HPLC unit (Jasco, Tokyo, Japan) equipped with Jasco PU-2080 HPLC pumps, Jasco 4277

dx.doi.org/10.1021/jf500959g | J. Agric. Food Chem. 2014, 62, 4276−4284

Journal of Agricultural and Food Chemistry

Article

two axial points on the axis of each design variable at a distance of α = 1.682 from the design center. The response was the sum of all polyamine peak areas. Experiments were carried out as follows: 3,5dinitrobenzoyl chloride concentration (35−65 mM), dispersive solvent volume (400−800 μL), and extraction solvent volume (100−250 μL). The complete experimental design (see the Supporting Information) was performed using the Design Expert Trial Version 7 (Stat-Ease, Inc., Minneapolis, MN) software. Method Validation. Mixed standard solutions (ranging from 0.5 to 10 mg/L) containing putrescine, cadaverine, spermidine, and spermine were subjected to the derivatization, cleanup, and preconcentration procedures described above. Calibration curves were constructed by the least-squares linear regression model, using the ratio between the peak area of each polyamine and the peak area of IS. The deviation from linearity was evaluated by applying the lack-of-fit test to the calibration curves of each polyamine. The intraday precision was calculated from the results of five replicate analyses in a single day of standards prepared at three concentration levels (0.5, 2.5, and 10 mg/L), whereas the interday precision was calculated from results obtained in five consecutive days. The LOD and LOQ were calculated on the basis of the calibration curve parameters according to Miller and Miller.29 Accuracy, on the basis of the recovery percentage {R, % = [(concentration of the spiked sample − concentration of the unspiked sample)/added concentration] × 100}, was estimated from triplicate experiments performed with lettuce, carrot, onion, cabbage, tomato, and apple samples (ca. 100 mg) spiked with 0.1 mL volume of a mixed standard solution (2.5 mg/L of each polyamine) to obtain a content of 2.5 μg/g in each plant food. Results of polyamine contents in plant foods were expressed as micrograms per gram of fresh weight. The stability of the extracts was tested by analyzing the extract of 5 mg/L standard solution at 0, 1, 2, 6, and 24 h after the derivatization and extraction procedure. During this time period, extracts were stored at 4 °C.

Choosing the most suitable dispersive and extraction solvents is a crucial step to obtain the maximum extraction efficiency. The dispersive solvent should be miscible with both aqueous and organic phases, while the extraction solvent must have higher density than water, low water solubility, and high extraction efficiency for the target compounds.20 Three dispersive solvents (methanol, acetonitrile, and acetone) and six extraction solvents (carbon tetrachloride, chloroform, chlorobenzene, dichloromethane, trichloroethylene, and tetrachloroethylene) were evaluated. The 5 mg/L standard solution was diluted in 5% (v/v) HClO4 solution, derivatized, and subjected to the DLLME procedure. Triplicate measurements were performed for each tested solvent. The use of acetonitrile as a dispersive solvent led to higher peak areas for all of the tested polyamines (Figure 2A), indicating that higher sensitivity



RESULTS AND DISCUSSION Experimental Parameter Selection. Working standard solutions containing putrescine, cadaverine, spermidine, and spermine (see chemical structures in Figure 1) at a fixed

Figure 2. Selection of (A) dispersive and (B) extraction solvents. Figure 1. Chemical structure of (1) putrescine, (2) cadaverine, (3) spermidine, (4) IS, and (5) spermine.

could be obtained with this solvent. Carbon tetrachloride presented higher extraction efficiency for all of the target compounds (Figure 2B); thus, it was selected as an extraction solvent. Two-Level Full Factorial Design. The main effects of the seven studied variables in the screening experiments are presented in Figure 3 in the form of a Pareto chart. The same standard solution used to perform the selection of the dispersive and extraction solvents was used in the two-step experimental design (two-level full factorial design for screening and CCD for optimization). The magnitude of the effects is highlighted in an ordered bar chart, where the bar length of the vertical axis is proportional to the significance of the variables.

concentration (5 mg/L) were used to study the buffer capacity of the derivatization procedure and the performance of different dispersive and extraction solvents on the DLLME procedure. To evaluate the buffer capacity, standard solutions were diluted in 5% (v/v) HClO4 solution to replicate the extraction conditions used in real plant food sample extraction. The obtained extracts presented a pH value around 1. Because the benzoylation reaction requires pH around 10, borate buffer (0.5 M) with 2 M NaOH provided a pH of 10 and was used in further experiments. 4278

dx.doi.org/10.1021/jf500959g | J. Agric. Food Chem. 2014, 62, 4276−4284

Journal of Agricultural and Food Chemistry

Article

extraction solvent volume, and amount of NaCl was assessed. The variables dispersive solvent volume and extraction solvent volume as well as the interaction between them were statistically significant (p < 0.05) and, thus, were considered for optimization. Because the variables “3,5-dinitrobenzoyl chloride volume” and “dispersive solvent volume” tested in the first and second full factorial designs, respectively, rely on the use of acetonitrile, the two variables were designated throughout the optimization procedure as just “dispersive solvent volume”. The variables derivatization time, derivatization temperature, and amount of NaCl showed no significant effect on the peak area of polyamines and, thus, were fixed for further experiments at 15 min, room temperature, and 500 mg, respectively. Faster reaction times, without the need for heating and with a lower amount of reagents involved, were achieved without compromising method sensitivity. CCD. For CCD, 20 experiments were carried out with randomly selected different combinations of the 3,5-dinitrobenzoyl chloride concentration (X1), dispersive solvent volume (X2), and extraction solvent volume (X3) parameters using statistically designed experiments. The sum of all polyamine peak areas as a function of X1, X2, and X3 was used as a response for the CCD design. Models were validated by two diagnostic residuals, the R2 and Q2 values. Typical values indicating good models are R2 > 0.75 and Q2 > 0.60.30,31 In our study, the values of R2 predicted and R2 adjusted were 0.92 and 0.97, respectively, and the Q2 value was 0.92, indicating the goodness of the model. Adequate precision, measured as a signal-to-noise ratio, was 26.54 (greater than 4 as desirable), which also indicates the adequacy of the present model. As shown in Table 1, the F value was 66.1, which indicates that the regression model is statistically significant (p < 0.01) at the 99% confidence level. Values of probability > F for X1, X2, X3, X1X2, X2X3, X12, X22, and X32 were lower than 0.05, which indicates that they are significant model terms. Despite its nonsignificance, the interaction X1X3 was considered in the model. Thus, a hierarchical model was applied. The lack-of-fit was 1.29, non-significant as desired, and the quadratic model was valid.32 Three-dimensional surface and contour plots were drawn to investigate the interactive effect of two factors on the sum of polyamine peak areas (Figure 4). Figure 4A shows the combined effect of the 3,5-dinitrobenzoyl chloride concentration and dispersive solvent volume. Figure 4B shows the

Figure 3. Main effect Pareto chart for the two-level full factorial design of the screening experiments of (A) derivatization and (B) extraction.

With regard to the derivatization procedure, the significance of the variables 3,5-dinitrobenzoyl chloride concentration, 3,5dinitrobenzoyl chloride volume, derivatization time, and derivatization temperature was evaluated. The variables 3,5dinitrobenzoyl chloride concentration and 3,5-dinitrobenzoyl chloride volume as well as the interaction between them were statistically significant (p < 0.05), with a positive effect on the response. Thus, these variables were selected for further assessment in the CCD. For the extraction procedure, the significance of the variables dispersive solvent volume,

Table 1. Analysis of Variance (ANOVA) for the Response Surface Model for the Sum of Polyamine Peak Areas source model intercept X1 X2 X3 X1X2 X1X3 X2X3 X12 X22 X32 residual lack-of-fit pure error total

coefficient estimate

standard error

sum of squares 2.54 × 10

15

3.62 4.41 1.08 3.06 −3.56 9.77 −3.42 −2.09 −4.52 −1.70

× × × × × × × × × ×

107 106 107 106 106 104 106 106 106 106

8.43 5.59 5.59 5.59 7.31 7.31 7.31 5.45 5.45 5.45

× × × × × × × × × ×

105 105 105 105 105 105 105 105 105 105

2.66 1.60 1.28 1.02 7.64 9.36 6.28 2.94 4.18 4.27 2.40 1.87 2.59

× × × × × × × × × × × × ×

1014 1015 1014 1014 1010 1013 1013 1014 1013 1013 1013 1013 1015

degrees of freedom 9 1 1 1 1 1 1 1 1 1 1 10 5 5 19 4279

mean square 2.83 × 10

14

2.66 1.60 1.28 1.02 7.64 9.36 6.28 2.94 4.18 4.27 4.81 3.74

× × × × × × × × × × × ×

1014 1015 1014 1014 1010 1013 1013 1014 1013 1013 1012 1012

F value

probability > F

66.1