High-performance liquid chromatographic determination of biogenic amines in wines with an experimental design optimization procedure

Analytica Chimica Acta 575 (2006) 97–105

High-performance liquid chromatographic determination of biogenic amines in wines with an experimental design optimization procedure Natividad Garc´ıa-Villar ∗ , Javier Saurina, Santiago Hern´andez-Cassou Department of Analytical Chemistry, University of Barcelona, Diagonal 647, 08028 Barcelona, Spain Received 31 January 2006; received in revised form 12 May 2006; accepted 18 May 2006 Available online 27 May 2006

Abstract A novel and sensitive HPLC method for determining biogenic amines in wine samples is described. It involves pre-column labeling of the analytes with 1,2-naphthoquinone-4-sulfonate (NQS) and liquid–liquid extraction of derivatives with chloroform for analyte preconcentration and sample clean-up. A linear gradient elution consisting of a mixture of 2% of acetic acid aqueous solution and methanol is used to separate the amine derivatives in a C18 column. The eluted compounds are detected spectrophotometrically at 270 nm. The optimization of both derivatization and separation conditions is accomplished by means of factorial and central composite designs and multicriteria decision functions. The analytical parameters of the method are established using red wine samples. Detection limits range from 0.006 to 0.315 mg L−1 . The run-to-run repeatabilities of retention times and peak areas are around 0.6 and 5.6%, respectively. Recoveries ranging from 91.9 to 105% prove the accuracy of the method for determining histamine, putrescine, cadaverine, tryptamine, phenylethylamine and serotonin in red wines. The proposed method has been applied to the analysis of commercial wines from different Spanish regions. © 2006 Elsevier B.V. All rights reserved. Keywords: Biogenic amines; Derivatization; High-performance liquid chromatography (HPLC); Experimental design; Wines

1. Introduction New trends in food industry are directed to elaborate healthier products and to improve food quality. As a result, novel food analysis methods searching for trace compounds that can affect human health are receiving increasingly more attention. Among other bioactive compounds, biogenic amines are specially relevant as they are usually present in fermented and spoiled foods and may induce several health disorders in sensitive humans [1,2]. Amine contents are specially controlled in wines and related products as they are mostly generated during the winemaking from the decarboxylation of the corresponding amino acids by microorganisms [3,4]. The main factors influencing on the formation of biogenic amines during wine production are the amino acidic composition of the wine must and the presence of yeasts and lactic acid bacteria [5–7]. Certain chemical factors like pH and SO2 have a remarkable effect on the microoganisms

∗

Corresponding author. Tel.: +34 93 403 44 45; fax: +34 93 402 12 33. E-mail address: [email protected] (N. Garc´ıa-Villar).

0003-2670/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2006.05.074

activity and, thus, on the formation of biogenic amines. High concentrations of SO2 have a negative influence on the bacteriological activity, whereas high pH gives rise to more complex bacterial microflora [3,7]. To date, there is no legislation dealing with the biogenic amine content in wines, although some countries recommend the limitation of histamine to a few milligrams per liter (2–10 mg L−1 ). In the near future amine levels will be an issue to be regulated in most countries. For this reason, the establishment of accurate methods to determine such substances is of particular interest. Biogenic amines are usually determined by separation techniques like HPLC [8–14], capillary electrophoresis (CE) [15–21] and gas chromatography (GC) [22,23]. These methods mainly involve spectrophotometric or fluorometric detection and/or hyphenation with mass spectrometry (MS) [24–27]. The development of rapid, straightforward and miniaturized analytical methods involving immunoassay, enzymatic and polymerase chain reaction processes is deserving also attention. Enzymelinked immunosorbent assays have proved to be specific enough to perform accurate determination of histamine in wine and cheese samples [28,29]. An enzymatic test to determine histamine in wines [30] and a biosensor array using an artificial

98

N. Garc´ıa-Villar et al. / Analytica Chimica Acta 575 (2006) 97–105

neural network to determine histamine, tyramine and putrescine in food samples [31] have been also developed. Recently, a multiplex polymerase chain reaction (PCR) method allowed determination of histamine, tyramine and putrescine producing acid lactic bacteria in foods [32]. However, to date, these rapid methods cannot be applied to determine a large number of biogenic amines and the detection limits are much poorer than those obtained with HPLC and CE methods, which are still required for analytical purposes. In the above-mentioned separation methods, a pre- or postcolumn derivatization step is necessary, in many cases, to attain a highly sensitive photometric or fluorimetric detection. For such a purpose, o-phthaldialdehyde, fluorenylmethylchloroformate and dansyl chloride are the most utilized labeling reagents. Regarding to sample clean-up and pre-concentration, solid phase extraction (SPE) is widely exploited. Recent developments in this field deal with the on-line hyphenation of derivatization and SPE to the separation system [33,34]. Despite the efforts to automate the SPE, clean-up of biogenic amines by using this technique is often not very efficient, and losses of amines are another common drawback. Liquid–liquid extraction of the analytes with an organic solvent, like diethylether or chloroform before derivatization results in another possibility to tackle the sample preparation [35–37], although the reported recoveries of amines are often low. In the present study, a highly sensitive method to determine biogenic amines in wine samples by pre-column labeling with 1,2-naphthoquinone-4-sulfonate (NQS) is described. This reagent reacts with primary and secondary amino groups and aromatic primary amino groups (see Fig. 1). Recent uses of this reagent deal with pre-column derivatization and HPLC determination of histamine [38] and in-capillary CE determination of several biogenic amines in wine samples [39]. In this new approach, derivatives are extracted with chloroform after the pre-column reaction. This procedure allows preconcentration of biogenic amines while removing efficiently potential interferences such as amino acid derivatives and other polar compounds, which usually hinder the chromatographic separation of biogenic amines. The combination of both derivatization and extraction procedures prior to the HPLC separation, results in an effective clean-up of samples and affords a sensitive determination of biogenic amines with limits of detection in the vicinity of picograms, which are similar to those obtained with fluorimetric detection. As another goal of this work, the feasibility of the experimental design methodology, which is used for the optimization of the different stages of the whole method, is also proved.

2. Experimental 2.1. Standards and reagents Unless specified all reagents used were of analytical grade. Milli-Q water (Millipore, Milford, MA, USA) was used to prepare all solutions. Labeling reagent (NQS) was obtained from Carlo Erba, Milan, Italy. Hydrochloric acid (37%, w/w) was from Panreac, Barcelona, Spain. The buffer solution for the derivatization medium was prepared with sodium tetraborate (Carlo Erba) and sodium hydroxide (Merck, Darmstadt, Germany). Acetic acid (96%, w/w) and HPLC grade methanol and chloroform were purchased from Merck. Cadaverine hydrochloride, putrescine hydrochloride, histamine dihydrochloride, tryptamine hydrochloride, tyramine hydrochloride, phenylethylamine hydrochloride, ethanolamine and cycloheptylamine, as internal standard, were purchased from Fluka (Buchs, Switzerland), and serotonin sulfate was from Sigma–Aldrich (Steinheim, Germany). Stock solutions of each amine were prepared and stored in the dark at 4 ◦ C. 2.2. Apparatus The chromatographic systems consisted of an Agilent 1100 Series HPLC instrument equipped with a G1311A quaternary pump, a G1379A degasser, a G1315B diode-array detector furnished with a 13-␮L flow cell and an Agilent Chemstation for data acquisition and analysis (Rev. A 10.02), all of them from Agilent Technologies (Waldbronn, Germany). Samples were injected with a Rheodyne 7725(i) (Cotati, CA, USA) injection valve equipped with a 20 ␮L sample loop. A block heater model Stuart SBH200D from Bibby Sterilin LTD (Stone, Staffordshire, UK) was used to perform the derivatization reaction. An evaporator Mini-Vap from Supelco (Bellefonte, PA, USA) was used to dry sample extracts. A Cyberscan 2500 pH meter from Eutech Instruments (Singapore) with a Hamilton pH electrode (Bonaduz, Switzerland) was used for pH measurement of samples and buffers. 2.3. Samples Commercial red wines from the following Spanish wine producing regions were purchased in retail stores: Campo de Borja, Jumilla, La Mancha, Rioja, Ribera del Duero, Cari˜nena and Valdepe˜nas. Wines analyzed can be classified according to their vintages, ranging between 2000 and 2004, and their aging period after fermentation in barrels or bottles (12 months for reserva wines, 6 months for crianza wines and no aging period for young wines). Internal standard (cycloheptylamine) was added to wine samples, which were filtered through a 0.45 ␮m nylon membrane (Cameo Nylon, Scharlab, Barcelona, Spain) prior to their analysis. 2.4. Derivatization and liquid–liquid extraction procedure

Fig. 1. Derivatization reaction between NQS and biogenic amines.

A 250 ␮L aliquot of biogenic amine standard solution or wine sample was mixed in a reaction vial with 250 ␮L of reagent solu-

N. Garc´ıa-Villar et al. / Analytica Chimica Acta 575 (2006) 97–105

tion (0.07 M NQS + 0.1 M HCl) and 250 ␮L of buffer solution (0.125 M Na2 B4 O7 + 0.1 M NaOH). The reaction was developed (pH 9.2) for 5 min at 65 ◦ C. Derivatized samples were extracted with 1250 ␮L of chloroform. After 1.5 min of shaking, 1000 ␮L of the organic phase extract were recovered and the solvent was evaporated to dryness through a nitrogen flow. Dried residues were reconstituted in 130 ␮L of a mixture 2% acetic acid aqueous solution and methanol (15:85, v/v). Finally, 20 ␮L of the resulting solution were injected into the chromatographic system. 2.5. Chromatographic analysis Derivatized biogenic amines were separated on a Synergi Hydro-RP C18 column (150 mm × 4.6 mm i.d., particle size ˚ equipped with a guard column (4 mm × 3 mm i.d.), 4 ␮m, 80 A) both from Phenomenex (Torrance, CA, USA). Derivatives were separated using a binary gradient based on a 2% (v/v) acetic acid aqueous solution (channel A) and methanol as organic modifier (channel B). The total flow-rate was 1 mL min−1 . The initial eluent contained a 15% of methanol and was linearly increased up to 55% in 5 min. After maintaining these conditions for 6 min the content of methanol was linearly increased from 55 to 95% in 5 min. Derivatives were detected spectrophotometrically at 305 and 270 nm.

99

The simultaneous optimization of time and temperature was accomplished with a central composite design. The temperature was varied from 30 to 90 ◦ C and the reaction time from 2 to 20 min. The remaining variables, pH, buffer and NQS concentration were kept constant at 9.8, 125 and 45 mM, respectively. Since the maximum responses for most of amines were not attained under the same conditions, a Derringer desirability function was defined to find an optimum compromise for all of them: 1/8

D = (d12 × d2 × d3 × d4 × d5 × d6 × d7 )

where d1 , d2 , d3 , d4 , d5 , d6 , d7 correspond to histamine, putrescine, cadaverine, ethanolamine, tryptamine, phenylethylamine, and tyramine peak areas, respectively (see Table 2 for optimization criteria and desirability limits). As shown in Fig. 2a the maximum of the response was attained at temperatures between 65 and 75 ◦ C and at derivatization periods from 5 to

3. Results and discussion 3.1. Optimization of derivatization conditions Derivatization conditions were thoroughly optimized by using screening and central composite designs [40]. Initially, the main factors and interactions influencing the derivatization were statistically evaluated on standard samples by using two-level factorial designs. The highest order interactions were taken as a reference of random variations. The following variables involved in the labeling reaction were studied: pH, buffer concentration, temperature, reaction time and reagent concentration. In order to decrease the number of experiments required, two preliminary systems were checked, comprising reaction time, temperature and NQS concentration (system A) and pH, buffer concentration and NQS concentration (system B). See Table 1 for details. Results from system A, revealed that time and temperature were interrelated. From system B, pH and reagent concentration were significant, while the influence of buffer concentration on the derivatization was negligible. Table 1 Factorial analysis design corresponding to the optimization of the derivatization conditions Factor

Temperature (◦ C) Reaction time (min) NQS concentration (mM) pH Buffer concentration (mM)

Screening system A A A, B B B

Level High (+) 80 15 80 11.5 125

Low (−) 40 3 10 9.8 12.5

Fig. 2. Multicriteria desirability surface responses obtained for the optimization of derivatization time and temperature. (a) Standard samples containing 50 ␮M each amine. (b) Red wine samples spiked with 50 ␮M each amine. Other experimental conditions are specified in Section 3.1.

100

N. Garc´ıa-Villar et al. / Analytica Chimica Acta 575 (2006) 97–105

Table 2 Multicriteria decision making criteria and desirability limits for the simultaneous optimization of interacting variables through overall desirability response Variables optimized and Derringer desirability function

Individual responses

Optimization criteria/desirability limits


d1 = histamine peak area (AHis )

Time and temperature of the derivatization 1/8 D = (d12 × d2 × d3 × d4 × d5 × d6 × d7 )

d1 =


d2 = putrescine peak area (APut )

d2 =


d3 = cadaverine peak area (ACad )

d3 =


d4 = ethanolamine peak area (AEta )

d4 =

d5 = tryptamine peak area (ATrp )

d5 =





d6 = phenylethylamine peak area (APhe )

d6 =

d7 = tyramine peak area (ATyr )

d7 =





Slope and time of elution gradient D = (d1 × d2 × d3 × d4 )1/4

d1 = resolution tryptamine–phenylethylamine (RsTrp-Phe )

d1 =

d2 = resolution cycloheptylamine–serotonin (RsChp-Ser )

d2 =

d3 = resolution serotonin–tyramine (RsSer-Tyr )

d3 =

d4 = time of analysis (ta )

d4 =

12 min. In order to determine the most appropriate conditions for wine samples, a second composite design was performed. The studied ranges of temperature and time were narrowed from 50 to 80 ◦ C and from 3 to 15 min, respectively, since the previous study revealed that the optimum combination of these two variables should be located inside this zone. The responses obtained for each amine (see Fig. 2b) were similar to those obtained in the former studies, so that the Derringer desirability function previously defined for standard samples was again used for commercial wine samples. Finally, the optimum conditions of time and temperature of derivatization were attained at 5 min and 65 ◦ C. Optimization of the non-interacting derivatization parameters was independently performed. The influence of the reagent concentration on the derivatization was studied in the range from 5 to 100 mM NQS and it was assayed on spiked red wine samples (50 ␮M each amine). From this study, a reagent concentration of 70 mM was selected for further studies as a compromise between sensitivity for all the amines and the absence of interfering peaks in the chromatogram. The effect of pH of the reaction mixture






(AHis − 0.6AHis(max) )/AHis(max) (0.9 − 0.6) 0 if AHis < 0.6AHis(max) 1 if AHis > 0.9AHis(max) (APut − 0.1APut(max) )/APut(max) (0.9 − 0.1) 0 if APut < 0.1APut(max) 1 if APut > 0.9APut(max) (ACad − 0.5ACad(max) )/ATrp(max) (0.9 − 0.5) 0 if ACad < 0.5ACad(max) 1 if ACad > 0.9ACad(max) (AEta − 0.3AEta(max) )/AEta(max) (0.9 − 0.3) 0 if AEta < 0.3AEta(max) 1 if AEta > 0.9AEta(max) (ATrp − 0.5ATrp(max) )/ATrp(max) (0.9 − 0.5) 0 if ATrp < 0.5ATrp(max) 1 if ATrp > 0.9ATrp(max) (APhe − 0.5APhe(max) )/APhe(max) (0.9 − 0.5) 0 if APhe < 0.5APhe(max) 1 if APhe > 0.9APhe(max) (ATyr − 0.4ATyr(max) )/ATyr(max) (0.9 − 0.4) 0 if ATyr < 0.4ATyr(max) 1 if ATyr > 0.9ATyr(max) (RsTrp-Phe − 0.5RsTrp-Phe(max) )/RsTrp-Phe(max) (0.9 − 0.5) 0 if RsTrp-Phe < 0.5RsTrp-Phe(max) 1 if RsTrp-Phe > 0.9RsTrp-Phe(max) (RsChp-Ser − 0.3RsChp-Ser(max) )/RsChp-Ser(max) (0.9 − 0.3) 0 if RsChp-Ser < 0.3RsChp-Ser(max) 1 if RsChp-Ser > 0.9RsChp-Ser(max) (RsSer-Tyr − 0.5RsSer-Tyr(max) )/RsSer-Tyr(max) (0.9 − 0.5) 0 if RsSer-Tyr < 0.5RsSer-Tyr(max) 1 if RsSer-Tyr > 0.9RsSer-Tyr(max) (1ta(max) − ta )/ta(max) (1 − 0.6) 0 if ta < 0.6ta(max) 1 if ta = 1ta(max)

was evaluated in range from 8.0 to 11.0. In general, the response of most amines decreased by increasing the pH. However, in the case of histamine, tryptamine and ethanolamine, the sensitivity increased with pH until a maximum around pH 9.0–9.5. From this study, a pH value of 9.2 was selected as it provided a high sensitivity for all the amines. Although, the buffer concentration was not a significant factor in the derivatization process, the large amounts of tartaric and malic acids present in wines can modify the final pH of the reaction mixture, thus, affecting the yield of the reaction. Here, tetraborate solutions with concentrations varying from 6.25 to 125 mM were tested. Finally, a 125 mM tetraborate solution was selected as it was robust enough to maintain the pH of derivatization at 9.2 for a wide variety of wines. 3.2. Optimization of the liquid–liquid extraction procedure A liquid–liquid extraction step with an organic solvent was expected to be a suitable sample clean-up treatment, in

N. Garc´ıa-Villar et al. / Analytica Chimica Acta 575 (2006) 97–105

order to remove amino acid derivatives and other interfering polar compounds. Additionally, this procedure may be useful to improve the sensitivity by concentrating biogenic amines derivatives. First, ethyl ether, methylene chloride, chloroform, ethyl acetate and methyl ethyl ketone as well as various binary mixtures were assayed for a selective extraction of derivatives. Among these possibilities, chloroform provided the best recoveries for the amine derivatives and signals increased up to 6000fold. At the same time, impurities remained at the aqueous phase, so that clearer chromatograms were obtained. As a conclusion, chloroform was finally selected as an extraction solvent giving the best compromise between enhancement of sensitivity and effective clean-up of side products and sample matrix components. The possibility of interaction between the variables concerning the extraction procedure was investigated by means of a two-level factorial design. These variables were the extraction pH, the solvent volume and the extraction time. See Table 3 for details of the experimental design. Results from the factorial screening showed that there was no interaction between any of the factors for any of the amines and, thus, optimization of each variable could be performed independently. It was also noted that extraction pH had the highest influence on the extraction yield, whereas the influence of time of extraction was negligible. After derivatization, the pH of samples was adjusted between 3.0 and 12.0 by adding appropriate amounts of the corresponding buffers. In all cases, the maximum extractions as well as the clearest chromatograms were achieved at pH values between 8.5 and 10.0 (see Fig. 3). Finally, an extraction pH of 9.2 was chosen for further studies. Note that the optimal derivatization and extraction pHs are the same, so that the derivatives can directly be extracted with no further experimental changes. The chloroform/aqueous solution ratios and the extraction time were also investigated. From the corresponding results, a chloroform volume of 1.25 mL and an extraction time of 1.5 min were chosen. After evaporation, samples were reconstituted in a mixture of 2% acetic acid aqueous solution and methanol (15:85, v:v).

101

Fig. 3. Influence of the pH on the extraction of the biogenic amine derivatives. See Section 2.4 for experimental conditions. Amine concentration: 500 ␮M histamine; 10 ␮M tryptamine, phenylethylamine and tyramine; 50 ␮M putrescine, ethanolamine, cadaverine and serotonin.

3.3. Optimization of chromatographic conditions Amine derivatives were separated in an octadecylsilane column using a mobile phase consisting on a 2% of acetic acid aqueous solution and methanol. Four linear gradient profiles were investigated, in which the organic phase varied from 15 to 95% in 5, 10, 16 and 22.5 min. The steepest gradient leaded to co-elution of most of the amine derivatives. Intermediate gradients resulted in co-elution of some compounds. The shallowest gradient enabled a complete separation of the analytes, although the time of analysis was considerably longer. This information was used to establish a three-step elution gradient as detailed as follows. The initial gradient step consisted of a variation of methanol content from 15 to 55% in 5 min. This range allowed a complete separation of histamine, putrescine and cadaverine derivatives from interferences. The resolution of the remaining amine derivatives was optimized on the second gradient step from the study of slope and time as the most relevant variables influencing on the separation. Their simultaneous optimization was accomplished by means of a three level factorial design (see Table 4 for experimental design details) and further construction of response surfaces. A total of nine experiments were

Table 3 Estimation of the effects in the two-level factorial design for the optimization of the liquid–liquid extraction conditions Factor/interactiona

pH of extraction Volume of organic solvent Time of extraction pH/volume Volume/time pH/time pH/volume/time E criticalb

Effect Histamine

Putrescine

Phenylethylamine

Tyramine

90.83 17.20 −0.45 13.15 −1.60 0.3 −1.60 20.33

246.98 38.54 −38.62 27.64 −10.82 18.99 −10.82 137.42

4230.87 652.16 −149.56 86.81 39.78 192.53 39.78 505.48

1170.65 178.95 −68.25 18.75 80.15 20.25 80.15 1018.40

Four amines chosen as models are presented. a pH levels: high (+) = 9.2, low (−) = 3.0; volume of organic solvent levels (␮L): high (+) = 1500, low (−) = 250; time of extraction levels (min): high (+) = 5, low (−) = 1. b E critical estimated from the highest order interaction at a significance level of α = 0.05. All effects larger in absolute value than E critical are significant.

102

N. Garc´ıa-Villar et al. / Analytica Chimica Acta 575 (2006) 97–105

Table 4 Three level full-factorial design used for the optimization of the separation conditions on the second gradient step Experiment

Time range (min)

MeOH range (%)

Slope (%MeOH min−1 )

Number of resolved peaksa

Tyramine retention timeb (min)

1 2 3 4 5 6 7 8 9

5–7 5–7 5–7 5–11 5–11 5–11 5–15 5–15 5–15

55–55 55–60 55–63 55–55 55–70 55–79 55–55 55–80 55–95

0 2.5 4.0 0 2.5 4.0 0 2.5 4.0

7 6 5 9 6 5 9 7 6

12.27 11.52 10.97 15.90 12.02 10.84 20.02 13.95 12.03

a Nine amines were used in the optimization experiments: histamine, putrescine, cadaverine, ethanolamine, tryptamine, phenylethylamine, cycloheptylamine, serotonin and tyramine. b Tyramine is the last amine eluted from the column.

assayed by varying the slope and time, all of them starting at 5 min with a methanol percentage of 55%. After this second step, the methanol percentage was linearly increased up to 95% with a slope of 8% MeOH min−1 , for a complete elution of the amine derivatives and effective column washing. Responses considered to be relevant in the optimization were combined in the following Derringer desirability function: D = (d1 × d2 × d3 × d4 )1/4 where d1 , d2 and d3 corresponded to resolution between tryptamine and phenylethylamine, cycloheptylamine and serotonin, and serotonin and tyramine, and d4 was the analysis time. Optimization criteria and desirability limits are summarized in Table 2. The global desirability response was maximum for a slope of 0% MeOH min−1 for 6 min. (See Fig. 4). This finding involves that after increasing methanol from 15 to 55% for 5 min, the composition of methanol was held at 55% for 6 min before increasing the solvent percentage up to 95% in 5 min (see Fig. 5). This gradient was also tested with wine samples, and an excellent separation performance was also obtained. No co-elution among the possible interferences of the wine matrix compounds with the peaks of interest was found.

Fig. 4. Multicriteria surface response obtained for the optimization of the second step elution gradient by variation of time and increase of methanol composition per minute. See Section 2.5 for experimental conditions.

3.4. Method validation Analytical parameters of the method were evaluated at 270 nm using red wine samples. Results are summarized in Table 5.

Table 5 Analytical features of the proposed method established in wines Amine

Histamine Putrescine Cadaverine Ethanolamine Tryptamine Phenylethylamine Serotonin Tyramine a b c d

Retention time (min) ± S.D.a (n=6) 4.50 6.75 8.55 9.50 11.00 11.90 15.30 15.75

± ± ± ± ± ± ± ±

0.03 0.03 0.06 0.08 0.09 0.09 0.04 0.03

Peak area repeatability (R.S.D., %, n=6) 25 (␮M)b

50 (␮M)b

75 (␮M)b

3.1 6.4 7.8 6.7 5.7 6.1 7.9 1.7

3.9 7.0 5.9 8.3 3.6 7.9 7.3 6.3

3.7 4.5 3.9 10.0 4.0 1.2 3.8 5.5

Dynamic range (␮M)

Correlation coefficientc,d

Detection limit (mg L−1 )

Limit of quantification (mg L−1 )

Up to 1000c Up to 1000c Up to 250 Up to 1000c Up to 1000c Up to 1000c Up to 1000c Up to 750

0.9959 0.9987 0.9983 0.9994 0.9997 0.9997 0.9996 0.9990

0.217 0.073 0.024 0.315 0.008 0.006 0.067 0.036

0.724 0.245 0.080 1.05 0.025 0.019 0.224 0.121

Standard deviation. Concentration range: 25, 50 and 75 ␮M. Maximum checked. Concentration ranges covered from 0.75 to 250 ␮M for cadaverine, from 0.75 to 750 ␮M for tyramine and from 0.75 to 1000 ␮M for the rest of the amines.

N. Garc´ıa-Villar et al. / Analytica Chimica Acta 575 (2006) 97–105

103

Fig. 5. Chromatogram of a biogenic amine standard mixture using the optimized conditions. Composition of the sample = 350 ␮M His, Cad and Put; 105 ␮M Trp, Phe, Ser and Tyr; 50 ␮M Chp (internal standard). Peak assignment: His = histamine; Put = putrescine; Cad = cadaverine; Eta = ethanolamine; Trp = tryptamine; Phe = phenylethylamine; Chp (IS) = cycloheptylamine (internal standard); Ser = serotonin; Tyr = tyramine.

A series of studies were carried to detect possible matrix effects (changes in the sensitivity due to the sample matrix). These experiments were performed by adding increasing known amounts of a standard containing the biogenic amines to three red wines of different vintage. Matrix effects were evaluated statistically by performing t-tests, where the slopes of the corresponding calibration graphs were compared with those obtained using a biogenic amine pure standard. For most of the amine derivatives, the calibration slopes were statistically similar to the pure standard curve. As an exception, ethanolamine and tyramine, were affected by the matrix composition as their slopes were statistically lower than those obtained for the water standards at a level of 5% significance. When the analytical interest is focused on the analysis of tyramine and ethanolamine, the standard addition method seems to be recommendable for a more accurate quantification. Finally, external pure standards can be used for calibration, either in the validation of the method or in the determination of biogenic amines in wine samples. Standard mixtures of biogenic amines, ranging from 0.75 to 1000 ␮M of each amine (with 50 ␮M of cycloheptylamine as an internal standard (I.S.)) were used for the evaluation of the linearity of the method. Calibration graphs were calculated according to I.S. method using linear regression. Good correla-

tion coefficients were obtained in all cases, with values ranging from 0.9959 for histamine to 0.9997 for phenylethylamine and tryptamine. The detection limits of biogenic amine derivatives, defined at a signal-to-noise ratio of 3, ranged from 0.006 mg L−1 for phenylethylamine to 0.315 mg L−1 for ethanolamine. The repeatability of the method was tested at three different concentrations levels of biogenic amines (25, 50 and 75 ␮M of each amine), which were added to a wine matrix. Six replicates of each concentration level were assayed. The relative standard deviation of the retention time was around 0.6% and the peak area repeatability was around 5.6%. The accuracy was evaluated by means of a spiking and recovery study on red wines. The spiked levels were 25, 50 and 75 ␮M for each amine. Four determinations were made at each level of concentration. Table 6 shows excellent results for histamine, cadaverine, putrescine, phenlylethylamine, tryptamine and serotonin, with recoveries ranging from 91.9% for serotonin to 105% for cadaverine. Therefore, all these amines can be quantified with the proposed method under the selected conditions. As expected from the preliminary calibration studies, ethanolamine and tyramine quantification resulted in a more complex issue due to matrix effects on the derivatization of these amines (probably

Table 6 Recoveries and R.S.D. values obtained for some amines from the spiking and recovery study performed on red wine samples Amine

Histamine Putrescine Cadaverine Tryptamine Phenylethylamine Serotonin

Addition level I (25 ␮M)

Addition level II (50 ␮M)

Addition level III (75 ␮M)

Recovery (%)

R.S.D. (%) n = 4

Recovery (%)

R.S.D. (%) n = 4

Recovery (%)

R.S.D. (%) n = 4

103.5 98.9 103.2 94.6 102.9 99.2

3.3 2.8 6.3 0.8 6.2 4.8

99.7 101.6 105.0 95.3 104.4 94.3

7.2 6.1 2.9 1.2 2.5 2.8

100.8 99.1 100.1 101.3 99.3 91.9

6.6 0.2 2.7 5.8 5.3 8.4

104

N. Garc´ıa-Villar et al. / Analytica Chimica Acta 575 (2006) 97–105

Fig. 6. Chromatogram of a commercial red wine sample (spiked with 50 ␮M of internal standard) analyzed under the optimized experimental conditions. Peak assignment as in Fig. 5.

Table 7 Quantitation of biogenic aminesa in commercial red wines using the proposed method (n = 4) Wine region

Vintage

Aging period

pH

Histamine (mg L−1 )

Campo de Borja Jumilla La Mancha Valencia Table wineb Rioja 1 Rioja 2 Ribera del Duero Cari˜nena La Mancha Valdepe˜nas

2004 2004 2004 2003 2004 2001 2001 2001 2000 2000 2000

Young Young Young Young Young Crianza Crianza Crianza Reserva Reserva Reserva

3.42 3.39 3.09 3.42 3.50 3.23 3.39 3.65 3.26 3.34 3.34

5.0 5.4 0.8 2.9 12.2 4.8 11.6 9.2 9.3 4.5 6.1

a b

± ± ± ± ± ± ± ± ± ± ±

0.2 0.1 0.1 0.1 0.2 0.5 0.1 0.5 0.1 0.4 0.6

Putrescine (mg L−1 ) 24.8 20.4 21.1 10.5 15.5 33.9 28.6 22.6 29.9 11.6 13.7

± ± ± ± ± ± ± ± ± ± ±

0.1 0.9 0.6 0.9 0.9 0.2 0.8 0.4 0.6 0.9 0.3

Cadaverine (mg L−1 )

Tryptamine (mg L−1 )

Phenylethylamine (mg L−1 )

± ± ± ± ± ± ± ± ± ± ±

<0.025 0.29 ± 0.02 0.62 ± 0.03 0.31 ± 0.01 <0.025 0.29 ± 0.02 0.57 ± 0.03 0.66 ± 0.01 0.37 ± .01 0.12 ± .01 0.20 ± 0.09

2.40 0.11 0.19 0.06 0.50 2.63 1.12 0.83 5.15 0.47 0.20

2.3 3.1 9.1 1.9 6.2 11.2 14.1 10.3 4.8 9.7 5.0

0.3 0.3 0.7 0.1 0.5 0.1 0.9 0.9 0.1 0.3 0.2

± ± ± ± ± ± ± ± ± ± ±

0.09 0.01 0.02 0.01 0.09 0.07 0.07 0.05 0.07 0.07 0.02

Serotonin (mg L−1 ) <0.224 <0.224 0.32 ± 0.01 <0.224 <0.224 <0.224 0.50 ± 0.09 <0.224 <0.224 0.33 ± 0.05 <0.224

Mean value ± standard deviation. Region not specified.

related to the oxidation of their derivatives to other side products). 3.5. Determination of biogenic amines in wine samples Red wines of different Spanish regions were analyzed under the selected experimental conditions. Four replicates for each determination were performed. As an example, a chromatogram of one of the red wines analyzed is shown in Fig. 6. Results obtained are given in Table 7. Additional information about pH, vintage and type of wine according to their post-fermentative aging process is also included. Putrescine, cadaverine and histamine are the most abundant amines in the samples analyzed. It is observed that both wine aging and acidity are variables influencing the content of biogenic amines in wines. Higher global amounts of biogenic amines are generally found in older wines such as crianzas and reservas, which have aged for longer periods in barrels or in bottles after winemaking. Conversely, young wines (from 2003 and 2004 vintages) contain lower amounts of

these compounds as they were directly bottled after winemaking and have not undergone further maturation processes. In parallel, less acid wines gave rise to higher histamine contents. This can be associated with the development of microorganisms with decarboxylase activity, which is enhanced at higher pH, and thus, histidine decarboxylation is enhanced. None of these wines surpass the toxic levels reported in the literature (100–500 mg L−1 ) [41] and, thus, these Spanish wines analyzed do not represent any toxicological risk for human health. 4. Conclusions In this work the combination of derivatization and liquid– liquid extraction procedures, ensures a quantitative elimination of potential interferences like amino acids. Furthermore, a significant pre-concentration factor is accomplished by the extraction step with a noticeable enhancement of the sensitivity of the detection for most amines. Detection limits verging on the picogram range are good enough to determine typical

N. Garc´ıa-Villar et al. / Analytica Chimica Acta 575 (2006) 97–105

amine levels present in wines. As the validation studies demonstrated, good linearities, precisions and recoveries are achieved. The experimental design techniques, applied to the optimization of the derivatization, the liquid–liquid extraction and the chromatographic separation conditions, helped to attain the most suitable analytical conditions from a reduced number of experiments. As a result, the separation is accomplished in a short analysis time with an excellent resolution for all the biogenic amines peaks of interest.

[16] [17] [18] [19] [20] [21] [22] [23] [24]

Acknowledgements [25]

This work was supported by Ministerio de Ciencia y Tecnolog´ıa project BQU2003-03103. NGV thanks the Generalitat de Catalunya for a FI grant. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15]

S. Bard´ocz, Trends Food Sci. Technol. 6 (1995) 341. M.H. Silla-Santos, Int. J. Food Microbiol. 29 (1996) 213. A. Lonvaud-Funel, FEMS Microbiol. Lett. 199 (2001) 9. P. Lehtonen, Am. J. Enol. Vitic. 47 (1996) 127. M. Caruso, C. Fiore, M. Contursi, G. Salzano, A. Paparella, P. Romano, W. J. Microbiol. Biotechnol. 18 (2002) 159. S.-Q. Liu, J. Appl. Microbiol. 92 (2002) 589. J.M. Landete, S. Ferrer, L. Polo, I. Pardo, J. Agric. Food Chem. 53 (2005) 1119. I. Moln´ar-Perl, J. Chromatogr. A 987 (2003) 291. M.C. Vidal-Carou, F. Lahoz-Portol´es, S. Bover-Cid, A. Marin´e-Font, J. Chromatogr. A 998 (2003) 235. J. Kirschbaum, K. Rebscher, H. Br¨uckner, J. Chromatogr. A 881 (2000) 517. R. Romero, D. G´azquez, M.G. Bagur, M. S´anchez-Vi˜nas, J. Chromatogr. A 871 (2000) 75. O. Busto, M. Miracle, J. Guasch, F. Borrull, J. Chromatogr. A 757 (1997) 311. Z. Loukou, A. Zotou, J. Chromatogr. A 996 (2003) 103. V. Lozanov, S. Petrov, V. Mitev, J. Chromatogr. A 1025 (2004) 201. ´ Kov´acs, L. Simon-Sarkadi, K. Ganzler, J. Chromatogr. A 836 (1999) A. 305.

[26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40]

[41]

105

X. Liu, L.X. Yang, Y.T. Lu, J. Chromatogr. A 998 (2003) 213. M. Molina, M. Silva, Electrophoresis 23 (2002) 2333. S. Oguri, J. Chromatogr. B 747 (2000) 1. N.P. Beard, A.J. de Mello, Electrophoresis 23 (2002) 1722. L.Y. Zhang, X.C. Tang, M.X. Sun, J. Chromatogr. B 820 (2005) 211. L.W. Cao, H. Wang, H.S. Zhang, Electrophoresis 26 (2005) 1954. F.R. Antoine, C.I. Wei, W.S. Otwell, C.A. Sims, R.C. Littell, A.D. Hogle, M.R. Marshall, J. Agric. Food Chem. 50 (2002) 4754. Y.Y. Zhao, L.S. Cai, Z.Z. Jing, H. Wang, H.X. Yu, H.S. Zhang, J. Chromatogr. A 1021 (2003) 175. F. Calbiani, M. Careri, L. Elviri, A. Mangia, L. Pistarpa, I. Zagnoni, J. Agric. Food Chem. 53 (2005) 3779. Y. Song, Z. Quan, J.L. Evans, E.A. Byrd, Y.M. Liu, Rapid Commun. Mass Spectrom 18 (2004) 989. B. Santos, B.M. Simonet, A. Rios, M. Valc´arcel, Electrophoresis 25 (2004) 3427. K.K. Ngim, S.E. Ebeler, M.E. Lew, D.G. Crosby, J.W. Wong, J. Agric. Food Chem. 48 (2000) 3311. ´ Marcobal, M.C. Polo, P.J. Mart´ın-Alvarez, ´ A. M.V. Moreno-Arribas, Food Res. Int. 38 (2005) 387. O. Ayg¨un, E. Schneider, R. Scheuer, E. Usleber, M. Gareis, E. M¨artlbauer, J. Agric. Food Chem. 47 (1999) 1961. J.M. Landete, S. Ferrer, I. Pardo, Food Addit. Contam. 12 (2004) 1149. J. Lange, C. Whittmann, Anal. Bioanal. Chem. 372 (2002) 276. ´ Marcobal, R. Mu˜noz, FEMS Microbiol. Lett. 244 B. De las Rivas, A. (2005) 367. E.K. Paleologos, M.G. Kontominas, Anal. Chem. 76 (2004) 1289. T. Hy¨otyl¨ainen, N. Savola, P. Lehtonen, M.L. Riekkola, Analyst 126 (2001) 2124. J.O. Fernandes, M.A. Ferreira, J. Chromatogr. A 886 (2000) 183. S. Moret, L.S. Conte, J. Chromatogr. A 729 (1996) 363–369. N.M. Tamim, Ll.W. Bennett, T.A. Shellem, J.A. Doerr, J. Agric. Food Chem. 50 (2002) 5012. N. Garc´ıa-Villar, J. Saurina, S. Hern´andez-Cassou, Analyst 130 (2005) 1286. N. Garc´ıa-Villar, J. Saurina, S. Hern´andez-Cassou, Electrophoresis 27 (2006) 474. D.L. Massart, B.G.M. Vandeginste, L.M.C. Buydens, S. de Jong, P.J. Lewi, J. Smeyers-Verbeke, Handbook of Chemometrics and Qualimetrics, Elsevier, Amsterdam, 1997, pp. 659, 701, 783. S.L. Taylor, Crit. Rev. Toxicol. 17 (1986) 91.