Liquid Chromatography–Single-Quadrupole Mass Spectrometry as a Responsive Tool for Determination of Biogenic Amines in Ready-to-Eat Baby Foods

Liquid Chromatography–Single-Quadrupole Mass Spectrometry as a Responsive Tool for... Baby food has never been the object of biogenic amine profiling. The aim of this study was to develop a highly sensitive method for analysis of biogenic amines in ready-to-eat baby foods. The principle of the developed method involves high- performance liquid chromatography coupled to single-quadrupole mass spectrometry (HPLC–APCI–MS) of dansyl deriva- tives, presented also in comparison with common diode array and fluorescence detection systems. The confirmation of correct identification of derivatives was performed by in-source fragmentation of the product ion at 170  m/z, performed only in one MS analyzer. The method was used to identify the amine profile and quantify the putrescine, cadaverine, histamine, tyramine, spermidine, and spermine content in 68 ready-to-eat baby foods. The limits of detection and quantification were − 1 in the range of 0.07–1.67 and 0.2–5.0 ng mL . The method enabled quantification of amines at ng/g level in almost all ana - lyzed samples, without any preconcentration step. Amine recoveries of 86.0–105.2% were obtained with RSD ≤ 9.7%. The developed method could be used for quantification of the most frequently occurring BAs in foods including vegetables, fish, meat, or fruit at previously undetectable concentration levels, making the method multimatrix applicable and highly-sensitive. Keywords Baby food · Biogenic amines · Dansyl chloride · HPLC–APCI–MS · Ion-source fragmentation Introduction particularly in those with high protein content [3]. Since BAs are thermo-stable compounds, they are present even in It has been acknowledged for a long time that formation of the heat-treated foods [4]. certain chemicals may pose a risk to human health. Such Potential adverse reactions among infants and young chemicals are biogenic amines (BAs)—nitrogenous com- children could appear after consumption of food containing pounds mainly formed by decarboxylation of the corre- toxic BAs, but the available data’s on digestive disorders sponding amino acids by spoilage and other microorgan- in children are limited [5]. Complementary foods intended isms, with the exception of polyamines, which can be formed for infants and young children under the age of 3 years old in vivo by the amination and transamination of aldehydes including ready-to-eat products are not currently screened or ketones [1, 2]. BAs can be found in all food products, for BAs. As a result, no data on the profiles and concentra- tions of individual BAs in baby foods are available. Such challenging analysis requires a new methodology, with lower Electronic supplementary material The online version of this LODs and greater chromatographic separation to be appli- article (https ://doi.org/10.1007/s1033 7-018-3527-z) contains cable for routine analysis. supplementary material, which is available to authorized users. The most common technique used for analysis of BAs, due to its sensitivity and selectivity, is HPLC coupled to * Anna Czajkowska-Mysłek anna.czajkowskamyslek@gmail.com various detection systems. The methods involve pre-/post- column derivatization, and UV or FL detection [6], evapo- Mass Spectrometry Laboratory, Wroclaw Research Centre rative light-scattering detector (ELSD) [7], or more often EIT+, 147 Stabłowicka, 54-066 Wroclaw, Poland are performed with MS detectors [8]. Separation of BAs Faculty of Biotechnology and Food Sciences, Lodz is generally performed on columns with alkyl chain also University of Technology, 4/10 Stefanowskiego, 90-924 Lodz, Poland Vol.:(0123456789) 1 3 902 A. Czajkowska-Mysłek, J. Leszczyńska conducted in UPLC systems [9, 10], or on HILIC (Hydro- (Merck, Darmstadt, Germany). Glacial acetic acid, per- philic Interaction Liquid Chromatography) columns [11]. chloric acid 60%, l -proline for biochemistry, trichloro- Each method consists of two basic steps: extraction from acetic acid, diethyl ether, sodium carbonate anhydrous, food matrices, with optional clean-up with SPE [12, 13] or and acetone for LC were acquired from Merck (Darm- µSPE [14], matrix solid-phase dispersion (MSPD) [11, 15], stadt, Germany). Ammonium formate, formic acid, dansyl dispersive liquid–liquid microextraction (DLLME) [16, 17] chloride, and 1,7-diaminoheptane were purchased from and derivatization to appropriate compounds for the detec- Sigma-Aldrich (St. Louis, MO, USA). All certified materi- tion technique used. The derivatization step may be per- als, including histamine dihydrochloride, tyramine hydro- formed using many reagents, such as o-phthaldialdehyde chloride, cadaverine dihydrochloride, putrescine dihy- (OPA), dansyl chloride (Dns-Cl), benzoyl chloride, dabsyl drochloride, spermidine trihydrochloride, and spermine chloride, 4-chloro-3,5-dinitrobenzotrifluoride, 1,2-naph- tetrahydrochloride, were purchased from Dr. Ehrenstorfer thoquinone-4sulfonate, 6-aminoquinolyl-N-hydroxysuccin- GmbH (Augsburg, Germany). Other chemicals, including imidyl carbamate, or N-hydroxy-succinimide ester [8]. Co- sodium hydroxide micropills, were obtained from POCH extractives like free amino acids might compete with BAs (Gliwice, Poland). Ammonium acetate for HPLC was sup- in the analytical process what might result in poor recovery plied by J.T. Baker (Deventer, The Netherlands). rates. Therefore, it is necessary to use different extraction solvents (perchloric acid, trichloroacetic acid, hydrochloric acid) and perform LLE to remove the amino acids from the Samples sample matrix. BAs profiling is certainly a challenge, primarily due to Sixty-eight samples of commercial ready-to-eat baby food the complexity of food matrices, the presence of free amino preserves intended for infants (4–12 months) and young acids and compounds that could interfere with the analytes, children (1–3 years old), produced by ten (coded from A the low concentration of BAs and in some cases significant to J) leading manufacturers available in Poland were pur- differences in the concentrations of individual amines in the chased from local shops. The products, available as din- amine profile. Limits of detection reported for the real sam- ners, soups, or desserts were pasteurized by manufactur- ples for the majority of the available analytical methods are ers and packed in glass jars (125–250 g) or plastic boxes/ − 1 − 1 generally in µg g (µg mL ) range with a few exceptions pouches (50–250 g). The samples contained 8–12% of fish [18]. LC–MS is the most precise and sensitive method, but is (vegetable-based with fish, sample Nos. 1–23, 23 prod- still not widely employed, particularly for routine analysis of ucts), 10% of meat (vegetable-based with meat, sample food [8]. Using tandem MS/MS detectors, the derivatization Nos. 24–38, 15 products), vegetables (sample Nos. 39–53, step could be excluded, but analysis needs to be performed 15 products), and fruit (sample Nos. 54–68, 15 products). in HILIC mode. The aim of this work was to develop an HPLC–APCI–MS method to evaluate, for the first time, the BAs content in Standards/Samples Preparation Procedure commercial ready-to-eat baby foods. A methodological requirement was to achieve much lower (1–2 orders of Standard solutions of selected amines and ISTD magnitude) LOQs in comparison with current methods for (1,7-diaminoheptane) were diluted with 0.4  M HClO . quantification of the most frequently occurring BAs in foods Stock solutions were stored at 4 °C for 3 months and cali- (vegetables, fish, meat or fruit), without additional sample bration solutions prepared daily before analysis. The con- − 1 preconcentration, making the method also multimatrix appli- centration of ISTD was maintained at 50 ng mL . The cable. The HPLC–APCI–MS method could be used as a calibration curves range was based on the concentration responsive analytical tool for the identification of possible levels of individual BAs determined in the analyzed sam- food constituents, which might have the most allergy-like ples. Dansyl chloride solution was prepared in acetone potential and elicit food adverse reactions among infants and just prior to use. young children. In the case of baby foods, two combined jars/boxes with diverse date codes were homogenized (if not homogene- ous) using a laboratory mixer. The sample preparation Materials and Methods procedure consists of three steps: an acid extraction, deri- vatization, and LLE extraction (Fig. 1). Chemicals The sample acid extracts could be stored at − 18  °C for approximately 6 months prior to further analysis. All LC–MS grade Acetonitrile (VWR, Radnor, PA, USA) was samples were analyzed in triplicate. used, and water purified using a MilliQ Direct 8 system 1 3 Liquid Chromatography–Single-Quadrupole Mass Spectrometry as a Responsive Tool for… 903 Fig. 1 Sample preparation procedure flowchart program: 0.01–16.00 min 60–90% B; 16.01–24.00 90% B; RP–HPLC–APCI–MS Conditions and 24.01–30.00 60% B (re-equilibration). HPLC analyses using different detection systems: an Analyses were performed using HPLC–APCI–MS with SPD-M20A Shimadzu diode array detector (connected selected ion monitoring (SIM) in positive ion mode. The in-line before MS) and a Shimadzu fluorescence detector HPLC system consisted of a Shimadzu Prominence UFLC RF-20A (offline MS) were also compared to establish the liquid chromatography binary system equipped with two response of the detectors to the obtained dansyl derivatives. LC-20AD pumps, a DGU-20A3 degasser unit, an SIL- The conditions of HPLC–FLD analysis were as follows: 20ACHT autosampler, and a CTO-10ASVP thermostated mobile phase A (water)/ B (ACN), gradient 0–19.0  min column oven coupled to an LCMS-2020 detector, with an 60–90% B, 19.0–20.0 min 90% B, 20.01 min 60% B, re- APCI interface, all supervised via CMB-20A controller. − 1 equilibration 27.0 min, flow rate 0.8 mL min , column tem- Data analysis was performed using the LabSolution soft- perature 25 °C, 20 µL injection, λ /λ 352/515 nm. ware (ver. 5.72 Shimadzu, Kyoto, Japan). MS acquisition ex em was performed under the following conditions: APCI tem- − 1 Method Validation and QA/QC Procedure perature 450 °C, nebulizing gas (N ) flow rate 4 L min , − 1 drying gas (N ) flow rate 10 L min , heat block temperature After selecting the optimum conditions for the sample prepa- 300 °C, and desolvation line temperature 200 °C. The BAs ration and HPLC–APCI–MS separation, method validation were separated on a Gemini-NX C column (150 × 4.6 mm, was performed. The method was validated on the most com- 3 µm particle size, Phenomenex, Torrence, CA, USA), with plex baby food matrix—vegetable with fish. The quantifica - a pre-column (4 × 3 mm) containing the same stationary − 1 tion of BAs was based on internal standard calibration. The phase, operated at 25 °C with a flow rate of 0.8 mL min . regression equations were calculated as six-point calibration The mobile phase consisted of 10 mM ammonium formate curves with weighting factor 1/x, based on quantification (A) and acetonitrile (B) with the following gradient elution 1 3 904 A. Czajkowska-Mysłek, J. Leszczyńska of the ratio of the amine peak area to the peak area of the reproducibility (three consecutive days, n = 9), in triplicate ISTD vs. the concentration of BA. The calculation of LODs analyses of sample spiked with approx. 25, 125, and 250 ng − 1 and LOQs was estimated for the LC–MS technique and the mL of PUT, SPD, and SPM and approx. 5, 25, and 50 ng − 1 method. LOQs were measured as the lowest concentrations mL of CAD, HIS, and TYR, respectively. The accuracy at which the analyte could be detected reliably, and defined of the method was evaluated by quantifying the recovery as concentrations resulting in a relative standard deviation of standard solutions. The recovery test was performed on (RSD) below 20%. They were thus a measure of the assay’s laboratory made vegetable baby food with fish using the precision at low analyte levels. The LOQ values obtained method of standard addition. The sample was spiked with were set as the lowest concentration levels on calibration high, intermediate, and low levels of standard BAs solution. curves. LODs were calculated from LOQ as LOD = LOQ/3. The method was validated with regard to linearity, matrix effect, precision, and accuracy. To evaluate the linearity of Results and Discussion the method, the standard solutions were prepared by diluting a specific volume of the stock standard to achieve several BA Profile of Ready‑to‑Eat Baby Foods concentrations. To estimate the matrix effect, five concen- trations of the standard were added (in triplicate) to a baby The BA profile of commercial ready-to-eat baby foods con- food containing endogenous BAs. The slopes of the calibra- sists mainly of ten amines (Fig. 2). The identified amines tion curves were compared with those obtained for stand- (HPLC–APCI–MS in full scan mode) were methylamine + + ard solutions. The matrix effect was calculated as ratios of (METH, [M+H] = 265), ethylamine (ETH, [M+H] = 279), the slope of matrix-matched calibration curves (5-points) putrescine (PUT, [M+H] = 555), cadaver ine (CAD, + + and the slope of the calibration curves in solvent (6-points) [M+H] = 569), histamine (HIS, [M+H] = 578), agmatine + + multiplied by 100 [19]. The quality assurance/quality con- (AGM, [M+H] = 597), tyramine (TYR, [M+H] = 604), trol (QA/QC) samples were inserted into each batch (blank serotonin (SER, [M+H] = 643), sper midine (SPD, + + solvent, duplicates of the sample from the beginning of the [M+H] = 845), and spermine (SPM, [M+H] = 1135). In batch at the end of batch). In addition, the stock solutions analyzed samples, other psychoactive amines commonly were used as the QC samples. The QC samples were pre- found in food products including phenylethylamine (PEA, + + pared daily at the low, medium, and high concentrations [M+H] = 355), tryptamine (TRP, [M+H] = 394) were not of standard BAs solution in fish-based baby food extract, detected. Apart from the above-mentioned BAs, other psy- and were measured at the beginning or end of each batch. choactive amines including dopamine (DOP, [M+H] = 853) The RSD for the peak area was determined as a measure of and norepinephrine (NE, [M+H] = 869) have been precision. The precision of the method was assessed based detected, but only in fruit-based products. To evaluate the on intra-day repeatability (one day, n = 3) and inter-day toxicity potential of ready-to-baby foods, only six amines Fig. 2 HPLC–APCI–MS (SIM+) chromatogram obtained in scan amine), PUT (putrescine), CAD (cadaverine), HIS (histamine), AGM mode (m/z 100–1300) presenting the BA profile of baby food sam- (agmatine), SER (serotonin), TYR (tyramine), SPD (spermidine), ple (No. 3). Peak identification: METH (methylamine), ETH (ethyl- SPM (spermine) 1 3 Liquid Chromatography–Single-Quadrupole Mass Spectrometry as a Responsive Tool for… 905 were selected from BAs identified in their profile. These The most promising results were obtained using the FL were selected due to their direct toxicity (HIS, TYR), their detection (Fig. 3a, b), but the low intensity of HIS and TYR potential to enhance HIS toxicity (PUT, CAD) or their car- fluorescence exclude its application with sufficient sensitiv - cinogenic properties (SPD, SPM) [2]. ity for analysis of baby food. Other amine derivatization agents such as o-phthaldialdehyde/2-mercaptoethanol (only RP–HPLC–APCI–MS Method Development primary amines), or benzoyl chloride used in the pre-column mode with HPLC–DAD/FLD system (data not shown) could Determining the BAs content in food matrices containing not be exploited for rewarding sensitivity level. low concentrations of HIS, TYR and CAD in comparison Using only a single-quadrupole MS detector in SIM with PUT, SPD, and SPM, which occur in food products mode, we could overcome both, the issue of differences in for children at much higher concentrations, presents cer- amine concentration in samples and the interferences caused tain difficulties, especially for classic detection systems by side-products of dansylation and co-eluting derivatives such as UV or FL (Fig. 3a–d). First, side-products of the (Fig. 3e–f). Our method is fully adequate for the challenging derivatization agent, dansyl chloride, show up on the chro- task of determining BAs in baby foods, but requires care- matograms, and second, the significant differences in the ful optimization of the chromatographic system, ion source, concentrations of amines present in the samples impede and MS detector parameters to obtain desired sensitivity. To their quantification. The side-products of Dns-amine reac- optimize the ionization of each BA in the source, various tion: dansyl acid (Dns-OH, [M+H] = 252), dansylamide mobile phases were tested. In general, only acetonitrile was (Dns-NH , [M+H] = 251), dansyl hydrazine (Dns-N H , used as an organic phase modifier, in combination with dif- 2 2 3 [M+H] = 266), N-dansyl ethylamine (Dns-NHC H , ferent volatile salts with acid additives, to obtain the appro- 2 5 [M+H] = 279) (Fig. 3f) observed with UV and FL detection priate value of pH for ionization. All amines (gas-phase − 1 systems essentially elute before BAs, but some compounds basicity > 200  kcal mol ) gave more intensive signals in present in the real samples also partially co-elute with PUT. mobile phase consisting of ACN with ammonium formate Fig. 3 HPLC chromatograms of dansylated amines obtained for for a standard solution of BAs: PUT = 510, CAD = 97, HIS = 98, − 1 different detection systems: a, b FLD, c, d DAD and e, f APCI- ISTD = 50, TYR = 99, SPD = 495, SPM = 303 ng mL and b, d, f for MS SIM(+) operated in-line. Chromatograms a, c, e were obtained a vegetable sample with fish intended for infants 1 3 906 A. Czajkowska-Mysłek, J. Leszczyńska − 1 (gas-phase basicity 179  kcal mol ) than in ammonium All experiments were conducted in triplicate. To quantify − 1 acetate (gas-phase basicity 191 kcal mol ), and a higher the BAs with appropriate sensitivity, the SIM(+) mode was signal was obtained with 10 mM salt concentration than with used in four segments (Table 1). 15 or 20 mM. SPM, a polyamine similar to SPD but which Using Dns-Cl as the pre-column derivatization agent elutes in most current MS methods [1, 10] as an irregular resulted in the formation of stable derivatives with rela- and low-intensity peak, had the highest intensity in ammo- tively high molecular masses, producing strong signal ions nium formate mobile phase (pH ~ 5), due to its higher Δ in positive mode. Furthermore, the high masses of the Dns- energy. In this instance, no acid modifier (formic or acetic BAs precursor ions obtained enabled fragmentation with acid) was needed. The mobile phase gradient was optimized formation of the product ion, specific fragment of m /z 170 for the shortest analysis time with appropriate resolution of (5-(N,N-dimethylamino)naphthyl ion) [1]. amines, which allowed for separation of all BAs in 24 min In-source fragmentation of BAs to the m/z 170 ion in (30 min re-equilibration). single-quadrupole MS was used to identify dansylated ESI is commonly used in analyses of BAs and amino amines. In-source fragmentation was achieved by increas- acids for the ionization of highly polar compounds. To ing the desolvation line (DL) and Qarray DC voltages (DL/ achieve proper ionization efficiency of the analyzed com- DC) for segment 1: 40/70 V, segment 2: 60/90 V, segment pounds, the ESI and APCI sources results were compared. 3–4: 60/100 V. In-source fragmentation was used to confirm The APCI source could be operated at higher mobile phase identification of all dansylated amines in this study, yielding − 1 flow rates (up to 2 mL min ) and enabled a higher ioni- in fragment m/z 170 in a range from 1% for SPM, ISTD, 2% zation temperature (up to 500 °C) to be used, resulting in for TYR, CAD, 6–7% for SPD and HIS, to 10% for PUT of minimum suppression of ionization and good ionization effi- precursor ion intensity in mass spectra. Greater fragmenta- ciency for BAs, without their needing to be pre-ionizable by tion of HIS occurred during in-source fragmentation than of acid additives in the mobile phase. other amines. This approach also confirmed the identifica- The APCI source parameters [nebulizing gas flow (NG) tion of peak which eluted after PUT in real samples (peak and drying gas flow (DG), temperature of APCI, heat block present in UV chromatograms (DAD) and in fluorescence (HB) temperature, and desolvation line (DL) temperature] detection (FLD) as a tailing peak) as dansyl derivative. were optimized to achieve the best analytical conditions. The optimizations were performed by triple injection of the Optimization of Sample Preparation Procedure standard solutions of all amines, taking into account the changes in the composition of the mobile phase during the The application of a derivatization step, as well as the LLE process in specified segments (Figs. S1–S5). The optimal clean-up procedure, was essential for quantification of BAs source parameters were temperature for APCI = 450  °C in almost all analyzed baby food samples. (maximum setting value 500  °C was not used), for The optimization of the sample preparation procedure HB = 300 °C and for DL = 200 °C, and the optimal nitrogen for baby foods included selection of the extraction solvent − 1 flows for NG = 4.0 L min (maximum setting value 4.4 L (5% TCA, 0.4  M HClO performed in 1 or 2 extraction − 1 min was not used due to the occurrence of a whistling steps—45 mL or 25 + 20 mL) (Fig. S6a), adjusting the pH − 1 sound) and for DG = 15 L min . The APCI voltage was set (9.0, 10.0, 11.0, 11.5, and 12.0) (Fig. S6b) and modifying at 4.5 kV, and the detector at 1.3 kV. The lowest ionization the temperature (40, 50, 60, and 70 °C) together with the efficiency in the optimized method was obtained for SPM. reaction time (30, 45, 60, and 75 min) for higher dansyla- Therefore, for this amine, the interface and detector voltages tion efficiency (Fig.  4) and the conditions of LLE extraction were increased to 5.0 and 1.5 kV, respectively. Optimiza- (2 × 1 mL, 3 × 1 mL, 4 × 1 mL of diethyl ether) (Fig. S6c). tion of the ion optic parameters for standard solutions of the The best conditions for BAs isolation from samples were amines using the LabSolution software was also performed. using 0.4 M HClO as an acid extraction solvent added in a Table 1 SIM(+) conditions for + Segment Time (min) Voltage (kV) DL Volt, Qarray DC, [M+H] the APCI-MS method Qarray RF (V) (m/z) APCI Detector 1 0.0–11.45 4.5 1.3 40, 30, 78 555.3 2 11.45–17.2 4.5 1.5 40, 30, 78 569.3, 578.3, 597.3, 604.2 3 17.2–19.0 4.5 1.3 40, 30, 104 845.3 4 19.0–24.0 5.0 1.5 40, 30, 130 1135.5 1 3 Liquid Chromatography–Single-Quadrupole Mass Spectrometry as a Responsive Tool for… 907 Fig. 4 Desirable surface response as relative MS response for 7 BAs obtained for the optimization of dansylation time and temperature in a amine standard solution, b baby food sample (n = 3) − 1 single volume of the process and the dansylation reaction, estimated LODs ranged from 0.56  ng mL for TYR to − 1 which was performed in pH 11.0 at 50 °C for 60 min. Most 15.3 ng mL for SPM using HPLC–APCI–MS [21]. For − 1 − 1 current methods involving the dansylation reaction are per- fish, LODs ranged from 20 ng g for SPD to 250 ng g for formed in pH = 9 [1, 20] or in generally alkaline medium SPM using HPLC–ESI–MS/MS [13]. [21]. The estimated results of pH optimization show that For all the studied BAs, good linearity was obtained with in pH = 11.0 amines such as HIS and CAD provide more R ranging from 0.9990 to 0.9999 (Table 2). Linearity was intensive signals. Dns-Cl was dissolved in acetone because established over 2–3 orders of magnitude. of difficulties related to the evaporation of acetonitrile fre- The relative standard deviation ranges of the compo- quently-used in other methods described in the literature. nents were 1.7–7.5% for intra-day analysis and 1.4–9.7% Certain results for these optimal dansylation parameters are for inter-day analysis, indicating a good standard of preci- in-line with those presented in the literature [1, 21]. The sion (Table 2). obtained derivatives were extracted with 3 × 1 mL of diethyl The developed method was reproducible with good recov- ether. With this method, there is no need to use a large vol- ery in the range of 86.0% for SPM and 105.2% for SPD with ume of extraction solvent (only 3 mL per sample) and the RSD ≤ 17.2% (Table 2). emulsification which occurs during the LLE extraction step Signal suppression as %SSE (signal suppression/enhance- is overcome through centrifugation, resulting in high BA ment) was observed for all BAs (100% SSE means that recovery values. This methodology enables the use of low there is no matrix effect) (Table S1). Very low suppression, sample dilution factor (the lowest equals 10), without pre- observed as a reduction in signal response, ranged from an concentration (LLE in a ratio of 1:1), because most matrix estimated − 2 for CAD (98% SSE) to − 7 for SPM (93% components, amino acids, and reagents remain in the aque- SSE) was obtained. Mild signal suppression, over 10%, was ous phase. estimated for TYR − 14 (86% SSE). Nevertheless, the SSE values obtained are relatively low and the estimated matrix- Method Validation Results matched calibration curves did not differ significantly. The calibration data, LODs, and LOQs are presented in Application of HPLC–APCI–MS Method to Baby Food Table 2. The sensitivity of the method is reflected by the Samples − 1 LODs values ranged from 0.07 ng mL (HIS, CAD, and − 1 TYR) to 1.67 ng mL (SPD and SPM), which are lower The optimized method was applied to determine the content than those reported in the literature, even for LC–MS/MS of 6 BAs in 68 samples of baby foods. HPLC–APCI–MS − 1 methods. The obtained 2.0 ng g LOQs for HIS, CAD, and method was fully adequate for quantification of almost TYR are out of range for any currently existing method of all selected amines present in the samples within a suit- BAs analysis in food matrices. For soybean meal, LODs able concentration range (4/408 results were < LOQ). − 1 − 1 ranged from 14.9 ng g for TYR to 19.1 ng g for HIS The summary amine levels in analyzed baby food prod- − 1 using UHPLC–ESI–MS/MS [10]. For wines estimated, ucts were found in a wide range of 1283–101423 ng g − 1 − 1 LODs ranged from 30.8 ng mL for CAD to 441 ng mL (Fig. 5). The particular amines were quantified as being in − 1 − 1 for TYR using HPLC–APCI–MS [22]. In donkey milk, the range of 704–53416 ng g for PUT, 2–1263 ng g 1 3 908 A. Czajkowska-Mysłek, J. Leszczyńska 1 3 Table 2 Analytical characteristics of the HPLC–APCI–MS method (n = 3) a 2 b b c − 1 Dansylamide Retention time (min) Linear Calibration equation R LOD LOQ LOQ (ng g ) Amount Intra-day, n = 3 Inter-day, n = 9 Recovery ± SD (%) − 1 range (ng added (ng (ng mL ) (RSD, %) − 1 − 1 mL ) mL ) (Dns) -PUT 11.01 ± 0.02 2.5–500 y = 0.351474x − 0.00249921 0.9996 0.80 2.5 25.0 25 2.2 2.5 100.6 ± 15.0 127 1.7 3.9 102.5 ± 7.9 255 1.8 2.7 103.6 ± 4.2 (Dns) -CAD 11.95 ± 0.02 0.2–100 y = 1.13648x + 0.000709481 0.9999 0.07 0.2 2.0 5 3.6 4.1 100.0 ± 7.2 24 2.9 4.1 98.8 ± 4.7 48 3.4 3.1 100.1 ± 3.4 (Dns) -HIS 12.45 ± 0.02 0.2–100 y = 0.977505x + 0.00704862 0.9999 0.07 0.2 2.0 5 4.9 3.2 102.1 ± 12.7 25 6.3 9.7 90.6 ± 14.6 49 7.3 8.0 91.2 ± 9.6 (Dns) -TYR 16.59 ± 0.01 0.2–100 y = 0.908411x + 0.00548783 0.9998 0.07 0.2 2.0 5 1.8 1.4 97.9 ± 11.0 25 6.5 7.5 91.4 ± 17.1 50 3.8 5.3 95.7 ± 5.8 (Dns) -SPD 17.79 ± 0.01 5.0–500 y = 0.346042x − 0.0195579 0.9995 1.67 5.0 50.0 25 2.6 2.7 105.2 ± 17.2 124 7.5 4.9 97.0 ± 9.8 248 4.4 3.9 95.7 ± 5.8 (Dns) -SPM 22.48 ± 0.02 5.0–300 y = 0.528549x − 0.0219078 0.9990 1.67 5.0 50.0 24 1.8 3.8 100.8 ± 11.4 121 2.6 5.1 86.4 ± 6.5 243 3.9 5.8 86.0 ± 6.1 Weighted least-squares linear regression (weighting factor 1/x) For LC–MS technique For RP–HPLC–APCI–MS with Dns-Cl method in baby foods Liquid Chromatography–Single-Quadrupole Mass Spectrometry as a Responsive Tool for… 909 cost spectrometer). The presented method could be easily extended to other sample matrix with low amine levels, with the opportunity of high sample preconcentration during the LLE procedure, e.g., water or ice. The HPLC–APCI–MS method could be a milestone achievement, providing an essential analytical tool for the specific identification of components in baby foods with the most potential to provoke allergy-like responses and other adverse reactions. Acknowledgements The first author would like to thank SHIM-POL (A. M. Borzymowski, Poland) for providing the APCI-MS detector used in the present work, and especially Dr. Paweł Stalica for his valu- able support. Dr. Elżbieta Piątkowska and Dr. Paweł Pasikowski (EIT+) for their valuable editing assistance of this manuscript. Special thanks also to Prof. Arkadiusz Szterk for kindly providing the dansyl chloride used in this study. Funding This research was financially supported by the Polish Minis- try of Science and Higher Education for the maintenance of research potential of Wacław Dąbrowski Institute of Agricultural and Food Bio- Fig. 5 Profile of total amine content obtained for analyzed baby foods technology (Warsaw, Poland) in 2014/2015 (500-01-ZJ-01). (n = 68) Compliance with ethical standards − 1 − 1 for CAD, 2–2375 ng g for HIS, 2–1668 ng g for TYR, − 1 − 1 Conflict of interest The authors declare that they have no conflict of 408–46680 ng g for SPD and 34–5619 ng g for SPM interest. [23]. The application of highly-sensitive HPLC–APCI–MS method in amine analysis of baby foods allowed identifica- Open Access This article is distributed under the terms of the Crea- tion of food ingredients which may be necessary to remove tive Commons Attribution 4.0 International License (http://creat iveco mmons.or g/licenses/b y/4.0/), which permits unrestricted use, distribu- (tuna, possibly spinach), reduce the amount added (spin- tion, and reproduction in any medium, provided you give appropriate ach, green peas), either reduce its using by infants under 12 credit to the original author(s) and the source, provide a link to the months of age (beef), or control the consumption (fruit baby Creative Commons license, and indicate if changes were made. products with banana) [23]. The obtained results enabled, for the first time, the assessment of a potential acute refer - ence dose (ARfD), and the BAI (biogenic amine index) for baby foods [23]. References 1. Lee S, Yoo M, Shin D (2015) The identification and quantifica- tion of biogenic amines in Korean turbid rice wine, Makgeolli by Conclusion HPLC with mass spectrometry detection. LWT–Food Sci Technol 62:350–356 To the best of our knowledge, there has been no research to 2. Santos MHS (1996) Biogenic amines: their importance in foods. Int J Food Microbiol 29:213–231 date to determine the biogenic amine profile of baby foods, 3. Suzzi G, Gardini F (2003) Biogenic amines in dry fermented sau- due to extremely low limit of detection to be obtained. Using sages: a review. Int J Food Microbiol 88:41–54 HPLC–APCI–MS and pre-column derivatization with dan- 4. Lehane L, Olley J (2000) Histamine fish poisoning revisited. syl chloride, we identified the BAs profile and quantified Intern J Food Microbiol 58:1–37 5. Reese I, Ballmer-Weber B, Beyer K, Fuchs T, Kleine-Tebbe J, the most toxic ones (HIS, TYR) in commercial ready-to-eat − 1 Klimek L, Lepp U, Niggemann B, Saloga J, Schäfer Ch, Werfel baby foods at concentration levels as low as 2 ng g , with- T, Zuberbier T, Worm M (2017) German guideline for the man- out requiring the use of SPE for clean-up and preconcentra- agement of adverse reactions to ingested histamine. Allergo J Int. tion with minimum use of organic solvents for the sample https ://doi.org/10.1007/s4062 9-017-0011-5 6. Tahmouzi S, Khaksar R, Ghasemlou M (2011) Development and preparation. The optimized method is suitable for simulta- validation of an HPLC-FLD method for rapid determination of neous analyses of PUT, CAD, HIS, TYR, SPD and SPM histamine in skipjack tuna s fi h ( Katsuwonus pelamis). Food Chem at low LODs. It provides significantly improved sensitivity 126:756–761 when compared to MS and MS/MS methods described in 7. Önal A, Tekkeli SEK, Önal C (2013) A review of the liquid chro- matographic methods for the determination of biogenic amines in the literature, particularly for CAD, HIS and TYR analy- foods. Food Chem 138:509–515 sis, and enables in-source fragmentation to the product ion without using tandem mass spectrometry (use of low 1 3 910 A. Czajkowska-Mysłek, J. Leszczyńska 8. Restuccia D, Spizzirri UG, Puoci F, Cirillo G, Curcio M, Parisi 16. Huang K-J, Jin C-X, Song S-L, Wie C-Y, Zhang J-F, Liu Y-M, Li OI, Iemma F, Picci N (2011) A new method for the determination J (2011) Development of an ionic liquid-based ultrasonic-assisted of biogenic amines in cheese by LC with evaporative light scat- liquid-liquid microextraction method for sensitive determination tering detector. Talanta 85:363–369 of biogenic amines: application to the analysis of octopamine, 9. Dadáková E, Křižek M, Pelikánová T (2009) Determination of tyramine and phenethylamine in beer samples. J Chromatogr B biogenic amines in foods using ultra-performance liquid chroma- 879:579–584 tography (UPLC). Food Chem 116:365–370 17. Płotka-Wasylka J, Simeonov V, Namieśnik J (2016) An in situ 10. Lee S, Eom H, Yoo M, Cho Y, Shin D (2011) Determination derivatization—dispersive liquid-liquid microextraction combines of biogenic amines in Cheonggukjang using ultra high pressure with gas-chromatography—mass spectrometry for determining liquid chromatography coupled with mass spectrometry. Food Sci biogenic amines in home-made fermented alcoholic drinks. J Biotechnol 20:123–129 Chromatogr A 1453:10–18 11. Self RL, Wu WH, Marks HS (2011) Simultaneous quantification 18. Erim FB (2013) Recent analytical approaches in the analysis of of eight biogenic amine compounds in tuna by matrix solid-phase biogenic amines in food samples. Trends Anal Chem 52:239–247 dispersion followed by HPLC-orbitrap mass spectrometry. J Agric 19. Sulyok M, Berthiller F, Krska R, Schuhmacher R (2006) Devel- Food Chem 59:5906–5913 opment and validation of a liquid chromatography/tandem mass 12. De Mey E, Drabik-Markiewicz G, De Maere H, Peeters M-C, spectrometric method for the determination of 39 mycotoxins in Derdelickx G, Paelinck H, Kowalska T (2012) Dabsyl derivatisa- wheat and maize. Rapid Commun Mass Spectrom 20:2649–2659 tion as an alternative for dansylation in the detection of biogenic 20. Szterk A, Roszko M (2014) Simultaneous determination of free amines in fermented meat products by reversed phase high per- amino acids, l -carnosine, purine, pyrimidine, and nucleosides in formance liquid chromatography. Food Chem 130:1017–1023 meat by liquid chromatography/single quadrupole mass spectrom- 13. Sagratini G, Fernández-Franzón M, De Berardinis F, Font G, etry. J Liq Chromatogr R T 37:664–680 Vittori S, Mañes J (2012) Simultaneous determination of eight 21. La Torre GL, Saitta M, Potortí AG, Di Bella G, Dugo G (2010) underivatised biogenic amines in fish by solid phase extraction High performance liquid chromatography coupled with atmos- and liquid chromatography-tandem mass spectrometry. Food pheric pressure chemical ionization mass spectrometry for sensi- Chem 132:537–543 tive determination of bioactive amines in donkey milk. J Chroma- 14. Basheer C, Wong W, Makahleh A, Tameem A, Salhin A, Saad B, togr A 1217:5215–5224 Lee HK (2011) Hydrazone-based ligands for micro-solid phase 22. García-Villar N, Hernández-Cassou S, Saurina J (2009) Determi- extraction-high performance liquid chromatographic deter- nation of biogenic amines in wines by pre-column derivatization mination of biogenic amines in orange juice. J Chromatogr A and high-performance liquid chromatography coupled to mass 1218:4332–4339 spectrometry. J Chromatogr A 1216:6387–6393 15. Calbiani F, Careri M, Elviri L, Mangia A, Pistarà L, Zagnoni 23. Czajkowska-Mysłek A, Leszczyńska J (2017) Risk assessment I (2005) Rapid Assay for analyzing biogenic amines in cheese: related to biogenic amines occurrence in ready-to-eat baby food. matrix solid-phase dispersion followed by liquid chromatogra- Food Chem Toxicol 105:82–92 phy-electrospray-tandem mass spectrometry. J Agric Food Chem 53:3779–3783 1 3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Chromatographia Springer Journals

Liquid Chromatography–Single-Quadrupole Mass Spectrometry as a Responsive Tool for Determination of Biogenic Amines in Ready-to-Eat Baby Foods

Free
10 pages

Loading next page...
 
/lp/springer_journal/liquid-chromatography-single-quadrupole-mass-spectrometry-as-a-pMI3TekkiU
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s)
Subject
Chemistry; Chromatography; Proteomics; Pharmacy; Laboratory Medicine; Analytical Chemistry
ISSN
0009-5893
eISSN
1612-1112
D.O.I.
10.1007/s10337-018-3527-z
Publisher site
See Article on Publisher Site

Abstract

Baby food has never been the object of biogenic amine profiling. The aim of this study was to develop a highly sensitive method for analysis of biogenic amines in ready-to-eat baby foods. The principle of the developed method involves high- performance liquid chromatography coupled to single-quadrupole mass spectrometry (HPLC–APCI–MS) of dansyl deriva- tives, presented also in comparison with common diode array and fluorescence detection systems. The confirmation of correct identification of derivatives was performed by in-source fragmentation of the product ion at 170  m/z, performed only in one MS analyzer. The method was used to identify the amine profile and quantify the putrescine, cadaverine, histamine, tyramine, spermidine, and spermine content in 68 ready-to-eat baby foods. The limits of detection and quantification were − 1 in the range of 0.07–1.67 and 0.2–5.0 ng mL . The method enabled quantification of amines at ng/g level in almost all ana - lyzed samples, without any preconcentration step. Amine recoveries of 86.0–105.2% were obtained with RSD ≤ 9.7%. The developed method could be used for quantification of the most frequently occurring BAs in foods including vegetables, fish, meat, or fruit at previously undetectable concentration levels, making the method multimatrix applicable and highly-sensitive. Keywords Baby food · Biogenic amines · Dansyl chloride · HPLC–APCI–MS · Ion-source fragmentation Introduction particularly in those with high protein content [3]. Since BAs are thermo-stable compounds, they are present even in It has been acknowledged for a long time that formation of the heat-treated foods [4]. certain chemicals may pose a risk to human health. Such Potential adverse reactions among infants and young chemicals are biogenic amines (BAs)—nitrogenous com- children could appear after consumption of food containing pounds mainly formed by decarboxylation of the corre- toxic BAs, but the available data’s on digestive disorders sponding amino acids by spoilage and other microorgan- in children are limited [5]. Complementary foods intended isms, with the exception of polyamines, which can be formed for infants and young children under the age of 3 years old in vivo by the amination and transamination of aldehydes including ready-to-eat products are not currently screened or ketones [1, 2]. BAs can be found in all food products, for BAs. As a result, no data on the profiles and concentra- tions of individual BAs in baby foods are available. Such challenging analysis requires a new methodology, with lower Electronic supplementary material The online version of this LODs and greater chromatographic separation to be appli- article (https ://doi.org/10.1007/s1033 7-018-3527-z) contains cable for routine analysis. supplementary material, which is available to authorized users. The most common technique used for analysis of BAs, due to its sensitivity and selectivity, is HPLC coupled to * Anna Czajkowska-Mysłek anna.czajkowskamyslek@gmail.com various detection systems. The methods involve pre-/post- column derivatization, and UV or FL detection [6], evapo- Mass Spectrometry Laboratory, Wroclaw Research Centre rative light-scattering detector (ELSD) [7], or more often EIT+, 147 Stabłowicka, 54-066 Wroclaw, Poland are performed with MS detectors [8]. Separation of BAs Faculty of Biotechnology and Food Sciences, Lodz is generally performed on columns with alkyl chain also University of Technology, 4/10 Stefanowskiego, 90-924 Lodz, Poland Vol.:(0123456789) 1 3 902 A. Czajkowska-Mysłek, J. Leszczyńska conducted in UPLC systems [9, 10], or on HILIC (Hydro- (Merck, Darmstadt, Germany). Glacial acetic acid, per- philic Interaction Liquid Chromatography) columns [11]. chloric acid 60%, l -proline for biochemistry, trichloro- Each method consists of two basic steps: extraction from acetic acid, diethyl ether, sodium carbonate anhydrous, food matrices, with optional clean-up with SPE [12, 13] or and acetone for LC were acquired from Merck (Darm- µSPE [14], matrix solid-phase dispersion (MSPD) [11, 15], stadt, Germany). Ammonium formate, formic acid, dansyl dispersive liquid–liquid microextraction (DLLME) [16, 17] chloride, and 1,7-diaminoheptane were purchased from and derivatization to appropriate compounds for the detec- Sigma-Aldrich (St. Louis, MO, USA). All certified materi- tion technique used. The derivatization step may be per- als, including histamine dihydrochloride, tyramine hydro- formed using many reagents, such as o-phthaldialdehyde chloride, cadaverine dihydrochloride, putrescine dihy- (OPA), dansyl chloride (Dns-Cl), benzoyl chloride, dabsyl drochloride, spermidine trihydrochloride, and spermine chloride, 4-chloro-3,5-dinitrobenzotrifluoride, 1,2-naph- tetrahydrochloride, were purchased from Dr. Ehrenstorfer thoquinone-4sulfonate, 6-aminoquinolyl-N-hydroxysuccin- GmbH (Augsburg, Germany). Other chemicals, including imidyl carbamate, or N-hydroxy-succinimide ester [8]. Co- sodium hydroxide micropills, were obtained from POCH extractives like free amino acids might compete with BAs (Gliwice, Poland). Ammonium acetate for HPLC was sup- in the analytical process what might result in poor recovery plied by J.T. Baker (Deventer, The Netherlands). rates. Therefore, it is necessary to use different extraction solvents (perchloric acid, trichloroacetic acid, hydrochloric acid) and perform LLE to remove the amino acids from the Samples sample matrix. BAs profiling is certainly a challenge, primarily due to Sixty-eight samples of commercial ready-to-eat baby food the complexity of food matrices, the presence of free amino preserves intended for infants (4–12 months) and young acids and compounds that could interfere with the analytes, children (1–3 years old), produced by ten (coded from A the low concentration of BAs and in some cases significant to J) leading manufacturers available in Poland were pur- differences in the concentrations of individual amines in the chased from local shops. The products, available as din- amine profile. Limits of detection reported for the real sam- ners, soups, or desserts were pasteurized by manufactur- ples for the majority of the available analytical methods are ers and packed in glass jars (125–250 g) or plastic boxes/ − 1 − 1 generally in µg g (µg mL ) range with a few exceptions pouches (50–250 g). The samples contained 8–12% of fish [18]. LC–MS is the most precise and sensitive method, but is (vegetable-based with fish, sample Nos. 1–23, 23 prod- still not widely employed, particularly for routine analysis of ucts), 10% of meat (vegetable-based with meat, sample food [8]. Using tandem MS/MS detectors, the derivatization Nos. 24–38, 15 products), vegetables (sample Nos. 39–53, step could be excluded, but analysis needs to be performed 15 products), and fruit (sample Nos. 54–68, 15 products). in HILIC mode. The aim of this work was to develop an HPLC–APCI–MS method to evaluate, for the first time, the BAs content in Standards/Samples Preparation Procedure commercial ready-to-eat baby foods. A methodological requirement was to achieve much lower (1–2 orders of Standard solutions of selected amines and ISTD magnitude) LOQs in comparison with current methods for (1,7-diaminoheptane) were diluted with 0.4  M HClO . quantification of the most frequently occurring BAs in foods Stock solutions were stored at 4 °C for 3 months and cali- (vegetables, fish, meat or fruit), without additional sample bration solutions prepared daily before analysis. The con- − 1 preconcentration, making the method also multimatrix appli- centration of ISTD was maintained at 50 ng mL . The cable. The HPLC–APCI–MS method could be used as a calibration curves range was based on the concentration responsive analytical tool for the identification of possible levels of individual BAs determined in the analyzed sam- food constituents, which might have the most allergy-like ples. Dansyl chloride solution was prepared in acetone potential and elicit food adverse reactions among infants and just prior to use. young children. In the case of baby foods, two combined jars/boxes with diverse date codes were homogenized (if not homogene- ous) using a laboratory mixer. The sample preparation Materials and Methods procedure consists of three steps: an acid extraction, deri- vatization, and LLE extraction (Fig. 1). Chemicals The sample acid extracts could be stored at − 18  °C for approximately 6 months prior to further analysis. All LC–MS grade Acetonitrile (VWR, Radnor, PA, USA) was samples were analyzed in triplicate. used, and water purified using a MilliQ Direct 8 system 1 3 Liquid Chromatography–Single-Quadrupole Mass Spectrometry as a Responsive Tool for… 903 Fig. 1 Sample preparation procedure flowchart program: 0.01–16.00 min 60–90% B; 16.01–24.00 90% B; RP–HPLC–APCI–MS Conditions and 24.01–30.00 60% B (re-equilibration). HPLC analyses using different detection systems: an Analyses were performed using HPLC–APCI–MS with SPD-M20A Shimadzu diode array detector (connected selected ion monitoring (SIM) in positive ion mode. The in-line before MS) and a Shimadzu fluorescence detector HPLC system consisted of a Shimadzu Prominence UFLC RF-20A (offline MS) were also compared to establish the liquid chromatography binary system equipped with two response of the detectors to the obtained dansyl derivatives. LC-20AD pumps, a DGU-20A3 degasser unit, an SIL- The conditions of HPLC–FLD analysis were as follows: 20ACHT autosampler, and a CTO-10ASVP thermostated mobile phase A (water)/ B (ACN), gradient 0–19.0  min column oven coupled to an LCMS-2020 detector, with an 60–90% B, 19.0–20.0 min 90% B, 20.01 min 60% B, re- APCI interface, all supervised via CMB-20A controller. − 1 equilibration 27.0 min, flow rate 0.8 mL min , column tem- Data analysis was performed using the LabSolution soft- perature 25 °C, 20 µL injection, λ /λ 352/515 nm. ware (ver. 5.72 Shimadzu, Kyoto, Japan). MS acquisition ex em was performed under the following conditions: APCI tem- − 1 Method Validation and QA/QC Procedure perature 450 °C, nebulizing gas (N ) flow rate 4 L min , − 1 drying gas (N ) flow rate 10 L min , heat block temperature After selecting the optimum conditions for the sample prepa- 300 °C, and desolvation line temperature 200 °C. The BAs ration and HPLC–APCI–MS separation, method validation were separated on a Gemini-NX C column (150 × 4.6 mm, was performed. The method was validated on the most com- 3 µm particle size, Phenomenex, Torrence, CA, USA), with plex baby food matrix—vegetable with fish. The quantifica - a pre-column (4 × 3 mm) containing the same stationary − 1 tion of BAs was based on internal standard calibration. The phase, operated at 25 °C with a flow rate of 0.8 mL min . regression equations were calculated as six-point calibration The mobile phase consisted of 10 mM ammonium formate curves with weighting factor 1/x, based on quantification (A) and acetonitrile (B) with the following gradient elution 1 3 904 A. Czajkowska-Mysłek, J. Leszczyńska of the ratio of the amine peak area to the peak area of the reproducibility (three consecutive days, n = 9), in triplicate ISTD vs. the concentration of BA. The calculation of LODs analyses of sample spiked with approx. 25, 125, and 250 ng − 1 and LOQs was estimated for the LC–MS technique and the mL of PUT, SPD, and SPM and approx. 5, 25, and 50 ng − 1 method. LOQs were measured as the lowest concentrations mL of CAD, HIS, and TYR, respectively. The accuracy at which the analyte could be detected reliably, and defined of the method was evaluated by quantifying the recovery as concentrations resulting in a relative standard deviation of standard solutions. The recovery test was performed on (RSD) below 20%. They were thus a measure of the assay’s laboratory made vegetable baby food with fish using the precision at low analyte levels. The LOQ values obtained method of standard addition. The sample was spiked with were set as the lowest concentration levels on calibration high, intermediate, and low levels of standard BAs solution. curves. LODs were calculated from LOQ as LOD = LOQ/3. The method was validated with regard to linearity, matrix effect, precision, and accuracy. To evaluate the linearity of Results and Discussion the method, the standard solutions were prepared by diluting a specific volume of the stock standard to achieve several BA Profile of Ready‑to‑Eat Baby Foods concentrations. To estimate the matrix effect, five concen- trations of the standard were added (in triplicate) to a baby The BA profile of commercial ready-to-eat baby foods con- food containing endogenous BAs. The slopes of the calibra- sists mainly of ten amines (Fig. 2). The identified amines tion curves were compared with those obtained for stand- (HPLC–APCI–MS in full scan mode) were methylamine + + ard solutions. The matrix effect was calculated as ratios of (METH, [M+H] = 265), ethylamine (ETH, [M+H] = 279), the slope of matrix-matched calibration curves (5-points) putrescine (PUT, [M+H] = 555), cadaver ine (CAD, + + and the slope of the calibration curves in solvent (6-points) [M+H] = 569), histamine (HIS, [M+H] = 578), agmatine + + multiplied by 100 [19]. The quality assurance/quality con- (AGM, [M+H] = 597), tyramine (TYR, [M+H] = 604), trol (QA/QC) samples were inserted into each batch (blank serotonin (SER, [M+H] = 643), sper midine (SPD, + + solvent, duplicates of the sample from the beginning of the [M+H] = 845), and spermine (SPM, [M+H] = 1135). In batch at the end of batch). In addition, the stock solutions analyzed samples, other psychoactive amines commonly were used as the QC samples. The QC samples were pre- found in food products including phenylethylamine (PEA, + + pared daily at the low, medium, and high concentrations [M+H] = 355), tryptamine (TRP, [M+H] = 394) were not of standard BAs solution in fish-based baby food extract, detected. Apart from the above-mentioned BAs, other psy- and were measured at the beginning or end of each batch. choactive amines including dopamine (DOP, [M+H] = 853) The RSD for the peak area was determined as a measure of and norepinephrine (NE, [M+H] = 869) have been precision. The precision of the method was assessed based detected, but only in fruit-based products. To evaluate the on intra-day repeatability (one day, n = 3) and inter-day toxicity potential of ready-to-baby foods, only six amines Fig. 2 HPLC–APCI–MS (SIM+) chromatogram obtained in scan amine), PUT (putrescine), CAD (cadaverine), HIS (histamine), AGM mode (m/z 100–1300) presenting the BA profile of baby food sam- (agmatine), SER (serotonin), TYR (tyramine), SPD (spermidine), ple (No. 3). Peak identification: METH (methylamine), ETH (ethyl- SPM (spermine) 1 3 Liquid Chromatography–Single-Quadrupole Mass Spectrometry as a Responsive Tool for… 905 were selected from BAs identified in their profile. These The most promising results were obtained using the FL were selected due to their direct toxicity (HIS, TYR), their detection (Fig. 3a, b), but the low intensity of HIS and TYR potential to enhance HIS toxicity (PUT, CAD) or their car- fluorescence exclude its application with sufficient sensitiv - cinogenic properties (SPD, SPM) [2]. ity for analysis of baby food. Other amine derivatization agents such as o-phthaldialdehyde/2-mercaptoethanol (only RP–HPLC–APCI–MS Method Development primary amines), or benzoyl chloride used in the pre-column mode with HPLC–DAD/FLD system (data not shown) could Determining the BAs content in food matrices containing not be exploited for rewarding sensitivity level. low concentrations of HIS, TYR and CAD in comparison Using only a single-quadrupole MS detector in SIM with PUT, SPD, and SPM, which occur in food products mode, we could overcome both, the issue of differences in for children at much higher concentrations, presents cer- amine concentration in samples and the interferences caused tain difficulties, especially for classic detection systems by side-products of dansylation and co-eluting derivatives such as UV or FL (Fig. 3a–d). First, side-products of the (Fig. 3e–f). Our method is fully adequate for the challenging derivatization agent, dansyl chloride, show up on the chro- task of determining BAs in baby foods, but requires care- matograms, and second, the significant differences in the ful optimization of the chromatographic system, ion source, concentrations of amines present in the samples impede and MS detector parameters to obtain desired sensitivity. To their quantification. The side-products of Dns-amine reac- optimize the ionization of each BA in the source, various tion: dansyl acid (Dns-OH, [M+H] = 252), dansylamide mobile phases were tested. In general, only acetonitrile was (Dns-NH , [M+H] = 251), dansyl hydrazine (Dns-N H , used as an organic phase modifier, in combination with dif- 2 2 3 [M+H] = 266), N-dansyl ethylamine (Dns-NHC H , ferent volatile salts with acid additives, to obtain the appro- 2 5 [M+H] = 279) (Fig. 3f) observed with UV and FL detection priate value of pH for ionization. All amines (gas-phase − 1 systems essentially elute before BAs, but some compounds basicity > 200  kcal mol ) gave more intensive signals in present in the real samples also partially co-elute with PUT. mobile phase consisting of ACN with ammonium formate Fig. 3 HPLC chromatograms of dansylated amines obtained for for a standard solution of BAs: PUT = 510, CAD = 97, HIS = 98, − 1 different detection systems: a, b FLD, c, d DAD and e, f APCI- ISTD = 50, TYR = 99, SPD = 495, SPM = 303 ng mL and b, d, f for MS SIM(+) operated in-line. Chromatograms a, c, e were obtained a vegetable sample with fish intended for infants 1 3 906 A. Czajkowska-Mysłek, J. Leszczyńska − 1 (gas-phase basicity 179  kcal mol ) than in ammonium All experiments were conducted in triplicate. To quantify − 1 acetate (gas-phase basicity 191 kcal mol ), and a higher the BAs with appropriate sensitivity, the SIM(+) mode was signal was obtained with 10 mM salt concentration than with used in four segments (Table 1). 15 or 20 mM. SPM, a polyamine similar to SPD but which Using Dns-Cl as the pre-column derivatization agent elutes in most current MS methods [1, 10] as an irregular resulted in the formation of stable derivatives with rela- and low-intensity peak, had the highest intensity in ammo- tively high molecular masses, producing strong signal ions nium formate mobile phase (pH ~ 5), due to its higher Δ in positive mode. Furthermore, the high masses of the Dns- energy. In this instance, no acid modifier (formic or acetic BAs precursor ions obtained enabled fragmentation with acid) was needed. The mobile phase gradient was optimized formation of the product ion, specific fragment of m /z 170 for the shortest analysis time with appropriate resolution of (5-(N,N-dimethylamino)naphthyl ion) [1]. amines, which allowed for separation of all BAs in 24 min In-source fragmentation of BAs to the m/z 170 ion in (30 min re-equilibration). single-quadrupole MS was used to identify dansylated ESI is commonly used in analyses of BAs and amino amines. In-source fragmentation was achieved by increas- acids for the ionization of highly polar compounds. To ing the desolvation line (DL) and Qarray DC voltages (DL/ achieve proper ionization efficiency of the analyzed com- DC) for segment 1: 40/70 V, segment 2: 60/90 V, segment pounds, the ESI and APCI sources results were compared. 3–4: 60/100 V. In-source fragmentation was used to confirm The APCI source could be operated at higher mobile phase identification of all dansylated amines in this study, yielding − 1 flow rates (up to 2 mL min ) and enabled a higher ioni- in fragment m/z 170 in a range from 1% for SPM, ISTD, 2% zation temperature (up to 500 °C) to be used, resulting in for TYR, CAD, 6–7% for SPD and HIS, to 10% for PUT of minimum suppression of ionization and good ionization effi- precursor ion intensity in mass spectra. Greater fragmenta- ciency for BAs, without their needing to be pre-ionizable by tion of HIS occurred during in-source fragmentation than of acid additives in the mobile phase. other amines. This approach also confirmed the identifica- The APCI source parameters [nebulizing gas flow (NG) tion of peak which eluted after PUT in real samples (peak and drying gas flow (DG), temperature of APCI, heat block present in UV chromatograms (DAD) and in fluorescence (HB) temperature, and desolvation line (DL) temperature] detection (FLD) as a tailing peak) as dansyl derivative. were optimized to achieve the best analytical conditions. The optimizations were performed by triple injection of the Optimization of Sample Preparation Procedure standard solutions of all amines, taking into account the changes in the composition of the mobile phase during the The application of a derivatization step, as well as the LLE process in specified segments (Figs. S1–S5). The optimal clean-up procedure, was essential for quantification of BAs source parameters were temperature for APCI = 450  °C in almost all analyzed baby food samples. (maximum setting value 500  °C was not used), for The optimization of the sample preparation procedure HB = 300 °C and for DL = 200 °C, and the optimal nitrogen for baby foods included selection of the extraction solvent − 1 flows for NG = 4.0 L min (maximum setting value 4.4 L (5% TCA, 0.4  M HClO performed in 1 or 2 extraction − 1 min was not used due to the occurrence of a whistling steps—45 mL or 25 + 20 mL) (Fig. S6a), adjusting the pH − 1 sound) and for DG = 15 L min . The APCI voltage was set (9.0, 10.0, 11.0, 11.5, and 12.0) (Fig. S6b) and modifying at 4.5 kV, and the detector at 1.3 kV. The lowest ionization the temperature (40, 50, 60, and 70 °C) together with the efficiency in the optimized method was obtained for SPM. reaction time (30, 45, 60, and 75 min) for higher dansyla- Therefore, for this amine, the interface and detector voltages tion efficiency (Fig.  4) and the conditions of LLE extraction were increased to 5.0 and 1.5 kV, respectively. Optimiza- (2 × 1 mL, 3 × 1 mL, 4 × 1 mL of diethyl ether) (Fig. S6c). tion of the ion optic parameters for standard solutions of the The best conditions for BAs isolation from samples were amines using the LabSolution software was also performed. using 0.4 M HClO as an acid extraction solvent added in a Table 1 SIM(+) conditions for + Segment Time (min) Voltage (kV) DL Volt, Qarray DC, [M+H] the APCI-MS method Qarray RF (V) (m/z) APCI Detector 1 0.0–11.45 4.5 1.3 40, 30, 78 555.3 2 11.45–17.2 4.5 1.5 40, 30, 78 569.3, 578.3, 597.3, 604.2 3 17.2–19.0 4.5 1.3 40, 30, 104 845.3 4 19.0–24.0 5.0 1.5 40, 30, 130 1135.5 1 3 Liquid Chromatography–Single-Quadrupole Mass Spectrometry as a Responsive Tool for… 907 Fig. 4 Desirable surface response as relative MS response for 7 BAs obtained for the optimization of dansylation time and temperature in a amine standard solution, b baby food sample (n = 3) − 1 single volume of the process and the dansylation reaction, estimated LODs ranged from 0.56  ng mL for TYR to − 1 which was performed in pH 11.0 at 50 °C for 60 min. Most 15.3 ng mL for SPM using HPLC–APCI–MS [21]. For − 1 − 1 current methods involving the dansylation reaction are per- fish, LODs ranged from 20 ng g for SPD to 250 ng g for formed in pH = 9 [1, 20] or in generally alkaline medium SPM using HPLC–ESI–MS/MS [13]. [21]. The estimated results of pH optimization show that For all the studied BAs, good linearity was obtained with in pH = 11.0 amines such as HIS and CAD provide more R ranging from 0.9990 to 0.9999 (Table 2). Linearity was intensive signals. Dns-Cl was dissolved in acetone because established over 2–3 orders of magnitude. of difficulties related to the evaporation of acetonitrile fre- The relative standard deviation ranges of the compo- quently-used in other methods described in the literature. nents were 1.7–7.5% for intra-day analysis and 1.4–9.7% Certain results for these optimal dansylation parameters are for inter-day analysis, indicating a good standard of preci- in-line with those presented in the literature [1, 21]. The sion (Table 2). obtained derivatives were extracted with 3 × 1 mL of diethyl The developed method was reproducible with good recov- ether. With this method, there is no need to use a large vol- ery in the range of 86.0% for SPM and 105.2% for SPD with ume of extraction solvent (only 3 mL per sample) and the RSD ≤ 17.2% (Table 2). emulsification which occurs during the LLE extraction step Signal suppression as %SSE (signal suppression/enhance- is overcome through centrifugation, resulting in high BA ment) was observed for all BAs (100% SSE means that recovery values. This methodology enables the use of low there is no matrix effect) (Table S1). Very low suppression, sample dilution factor (the lowest equals 10), without pre- observed as a reduction in signal response, ranged from an concentration (LLE in a ratio of 1:1), because most matrix estimated − 2 for CAD (98% SSE) to − 7 for SPM (93% components, amino acids, and reagents remain in the aque- SSE) was obtained. Mild signal suppression, over 10%, was ous phase. estimated for TYR − 14 (86% SSE). Nevertheless, the SSE values obtained are relatively low and the estimated matrix- Method Validation Results matched calibration curves did not differ significantly. The calibration data, LODs, and LOQs are presented in Application of HPLC–APCI–MS Method to Baby Food Table 2. The sensitivity of the method is reflected by the Samples − 1 LODs values ranged from 0.07 ng mL (HIS, CAD, and − 1 TYR) to 1.67 ng mL (SPD and SPM), which are lower The optimized method was applied to determine the content than those reported in the literature, even for LC–MS/MS of 6 BAs in 68 samples of baby foods. HPLC–APCI–MS − 1 methods. The obtained 2.0 ng g LOQs for HIS, CAD, and method was fully adequate for quantification of almost TYR are out of range for any currently existing method of all selected amines present in the samples within a suit- BAs analysis in food matrices. For soybean meal, LODs able concentration range (4/408 results were < LOQ). − 1 − 1 ranged from 14.9 ng g for TYR to 19.1 ng g for HIS The summary amine levels in analyzed baby food prod- − 1 using UHPLC–ESI–MS/MS [10]. For wines estimated, ucts were found in a wide range of 1283–101423 ng g − 1 − 1 LODs ranged from 30.8 ng mL for CAD to 441 ng mL (Fig. 5). The particular amines were quantified as being in − 1 − 1 for TYR using HPLC–APCI–MS [22]. In donkey milk, the range of 704–53416 ng g for PUT, 2–1263 ng g 1 3 908 A. Czajkowska-Mysłek, J. Leszczyńska 1 3 Table 2 Analytical characteristics of the HPLC–APCI–MS method (n = 3) a 2 b b c − 1 Dansylamide Retention time (min) Linear Calibration equation R LOD LOQ LOQ (ng g ) Amount Intra-day, n = 3 Inter-day, n = 9 Recovery ± SD (%) − 1 range (ng added (ng (ng mL ) (RSD, %) − 1 − 1 mL ) mL ) (Dns) -PUT 11.01 ± 0.02 2.5–500 y = 0.351474x − 0.00249921 0.9996 0.80 2.5 25.0 25 2.2 2.5 100.6 ± 15.0 127 1.7 3.9 102.5 ± 7.9 255 1.8 2.7 103.6 ± 4.2 (Dns) -CAD 11.95 ± 0.02 0.2–100 y = 1.13648x + 0.000709481 0.9999 0.07 0.2 2.0 5 3.6 4.1 100.0 ± 7.2 24 2.9 4.1 98.8 ± 4.7 48 3.4 3.1 100.1 ± 3.4 (Dns) -HIS 12.45 ± 0.02 0.2–100 y = 0.977505x + 0.00704862 0.9999 0.07 0.2 2.0 5 4.9 3.2 102.1 ± 12.7 25 6.3 9.7 90.6 ± 14.6 49 7.3 8.0 91.2 ± 9.6 (Dns) -TYR 16.59 ± 0.01 0.2–100 y = 0.908411x + 0.00548783 0.9998 0.07 0.2 2.0 5 1.8 1.4 97.9 ± 11.0 25 6.5 7.5 91.4 ± 17.1 50 3.8 5.3 95.7 ± 5.8 (Dns) -SPD 17.79 ± 0.01 5.0–500 y = 0.346042x − 0.0195579 0.9995 1.67 5.0 50.0 25 2.6 2.7 105.2 ± 17.2 124 7.5 4.9 97.0 ± 9.8 248 4.4 3.9 95.7 ± 5.8 (Dns) -SPM 22.48 ± 0.02 5.0–300 y = 0.528549x − 0.0219078 0.9990 1.67 5.0 50.0 24 1.8 3.8 100.8 ± 11.4 121 2.6 5.1 86.4 ± 6.5 243 3.9 5.8 86.0 ± 6.1 Weighted least-squares linear regression (weighting factor 1/x) For LC–MS technique For RP–HPLC–APCI–MS with Dns-Cl method in baby foods Liquid Chromatography–Single-Quadrupole Mass Spectrometry as a Responsive Tool for… 909 cost spectrometer). The presented method could be easily extended to other sample matrix with low amine levels, with the opportunity of high sample preconcentration during the LLE procedure, e.g., water or ice. The HPLC–APCI–MS method could be a milestone achievement, providing an essential analytical tool for the specific identification of components in baby foods with the most potential to provoke allergy-like responses and other adverse reactions. Acknowledgements The first author would like to thank SHIM-POL (A. M. Borzymowski, Poland) for providing the APCI-MS detector used in the present work, and especially Dr. Paweł Stalica for his valu- able support. Dr. Elżbieta Piątkowska and Dr. Paweł Pasikowski (EIT+) for their valuable editing assistance of this manuscript. Special thanks also to Prof. Arkadiusz Szterk for kindly providing the dansyl chloride used in this study. Funding This research was financially supported by the Polish Minis- try of Science and Higher Education for the maintenance of research potential of Wacław Dąbrowski Institute of Agricultural and Food Bio- Fig. 5 Profile of total amine content obtained for analyzed baby foods technology (Warsaw, Poland) in 2014/2015 (500-01-ZJ-01). (n = 68) Compliance with ethical standards − 1 − 1 for CAD, 2–2375 ng g for HIS, 2–1668 ng g for TYR, − 1 − 1 Conflict of interest The authors declare that they have no conflict of 408–46680 ng g for SPD and 34–5619 ng g for SPM interest. [23]. The application of highly-sensitive HPLC–APCI–MS method in amine analysis of baby foods allowed identifica- Open Access This article is distributed under the terms of the Crea- tion of food ingredients which may be necessary to remove tive Commons Attribution 4.0 International License (http://creat iveco mmons.or g/licenses/b y/4.0/), which permits unrestricted use, distribu- (tuna, possibly spinach), reduce the amount added (spin- tion, and reproduction in any medium, provided you give appropriate ach, green peas), either reduce its using by infants under 12 credit to the original author(s) and the source, provide a link to the months of age (beef), or control the consumption (fruit baby Creative Commons license, and indicate if changes were made. products with banana) [23]. The obtained results enabled, for the first time, the assessment of a potential acute refer - ence dose (ARfD), and the BAI (biogenic amine index) for baby foods [23]. References 1. Lee S, Yoo M, Shin D (2015) The identification and quantifica- tion of biogenic amines in Korean turbid rice wine, Makgeolli by Conclusion HPLC with mass spectrometry detection. LWT–Food Sci Technol 62:350–356 To the best of our knowledge, there has been no research to 2. Santos MHS (1996) Biogenic amines: their importance in foods. Int J Food Microbiol 29:213–231 date to determine the biogenic amine profile of baby foods, 3. Suzzi G, Gardini F (2003) Biogenic amines in dry fermented sau- due to extremely low limit of detection to be obtained. Using sages: a review. Int J Food Microbiol 88:41–54 HPLC–APCI–MS and pre-column derivatization with dan- 4. Lehane L, Olley J (2000) Histamine fish poisoning revisited. syl chloride, we identified the BAs profile and quantified Intern J Food Microbiol 58:1–37 5. Reese I, Ballmer-Weber B, Beyer K, Fuchs T, Kleine-Tebbe J, the most toxic ones (HIS, TYR) in commercial ready-to-eat − 1 Klimek L, Lepp U, Niggemann B, Saloga J, Schäfer Ch, Werfel baby foods at concentration levels as low as 2 ng g , with- T, Zuberbier T, Worm M (2017) German guideline for the man- out requiring the use of SPE for clean-up and preconcentra- agement of adverse reactions to ingested histamine. Allergo J Int. tion with minimum use of organic solvents for the sample https ://doi.org/10.1007/s4062 9-017-0011-5 6. Tahmouzi S, Khaksar R, Ghasemlou M (2011) Development and preparation. The optimized method is suitable for simulta- validation of an HPLC-FLD method for rapid determination of neous analyses of PUT, CAD, HIS, TYR, SPD and SPM histamine in skipjack tuna s fi h ( Katsuwonus pelamis). Food Chem at low LODs. It provides significantly improved sensitivity 126:756–761 when compared to MS and MS/MS methods described in 7. Önal A, Tekkeli SEK, Önal C (2013) A review of the liquid chro- matographic methods for the determination of biogenic amines in the literature, particularly for CAD, HIS and TYR analy- foods. Food Chem 138:509–515 sis, and enables in-source fragmentation to the product ion without using tandem mass spectrometry (use of low 1 3 910 A. Czajkowska-Mysłek, J. Leszczyńska 8. Restuccia D, Spizzirri UG, Puoci F, Cirillo G, Curcio M, Parisi 16. Huang K-J, Jin C-X, Song S-L, Wie C-Y, Zhang J-F, Liu Y-M, Li OI, Iemma F, Picci N (2011) A new method for the determination J (2011) Development of an ionic liquid-based ultrasonic-assisted of biogenic amines in cheese by LC with evaporative light scat- liquid-liquid microextraction method for sensitive determination tering detector. Talanta 85:363–369 of biogenic amines: application to the analysis of octopamine, 9. Dadáková E, Křižek M, Pelikánová T (2009) Determination of tyramine and phenethylamine in beer samples. J Chromatogr B biogenic amines in foods using ultra-performance liquid chroma- 879:579–584 tography (UPLC). Food Chem 116:365–370 17. Płotka-Wasylka J, Simeonov V, Namieśnik J (2016) An in situ 10. Lee S, Eom H, Yoo M, Cho Y, Shin D (2011) Determination derivatization—dispersive liquid-liquid microextraction combines of biogenic amines in Cheonggukjang using ultra high pressure with gas-chromatography—mass spectrometry for determining liquid chromatography coupled with mass spectrometry. Food Sci biogenic amines in home-made fermented alcoholic drinks. J Biotechnol 20:123–129 Chromatogr A 1453:10–18 11. Self RL, Wu WH, Marks HS (2011) Simultaneous quantification 18. Erim FB (2013) Recent analytical approaches in the analysis of of eight biogenic amine compounds in tuna by matrix solid-phase biogenic amines in food samples. Trends Anal Chem 52:239–247 dispersion followed by HPLC-orbitrap mass spectrometry. J Agric 19. Sulyok M, Berthiller F, Krska R, Schuhmacher R (2006) Devel- Food Chem 59:5906–5913 opment and validation of a liquid chromatography/tandem mass 12. De Mey E, Drabik-Markiewicz G, De Maere H, Peeters M-C, spectrometric method for the determination of 39 mycotoxins in Derdelickx G, Paelinck H, Kowalska T (2012) Dabsyl derivatisa- wheat and maize. Rapid Commun Mass Spectrom 20:2649–2659 tion as an alternative for dansylation in the detection of biogenic 20. Szterk A, Roszko M (2014) Simultaneous determination of free amines in fermented meat products by reversed phase high per- amino acids, l -carnosine, purine, pyrimidine, and nucleosides in formance liquid chromatography. Food Chem 130:1017–1023 meat by liquid chromatography/single quadrupole mass spectrom- 13. Sagratini G, Fernández-Franzón M, De Berardinis F, Font G, etry. J Liq Chromatogr R T 37:664–680 Vittori S, Mañes J (2012) Simultaneous determination of eight 21. La Torre GL, Saitta M, Potortí AG, Di Bella G, Dugo G (2010) underivatised biogenic amines in fish by solid phase extraction High performance liquid chromatography coupled with atmos- and liquid chromatography-tandem mass spectrometry. Food pheric pressure chemical ionization mass spectrometry for sensi- Chem 132:537–543 tive determination of bioactive amines in donkey milk. J Chroma- 14. Basheer C, Wong W, Makahleh A, Tameem A, Salhin A, Saad B, togr A 1217:5215–5224 Lee HK (2011) Hydrazone-based ligands for micro-solid phase 22. García-Villar N, Hernández-Cassou S, Saurina J (2009) Determi- extraction-high performance liquid chromatographic deter- nation of biogenic amines in wines by pre-column derivatization mination of biogenic amines in orange juice. J Chromatogr A and high-performance liquid chromatography coupled to mass 1218:4332–4339 spectrometry. J Chromatogr A 1216:6387–6393 15. Calbiani F, Careri M, Elviri L, Mangia A, Pistarà L, Zagnoni 23. Czajkowska-Mysłek A, Leszczyńska J (2017) Risk assessment I (2005) Rapid Assay for analyzing biogenic amines in cheese: related to biogenic amines occurrence in ready-to-eat baby food. matrix solid-phase dispersion followed by liquid chromatogra- Food Chem Toxicol 105:82–92 phy-electrospray-tandem mass spectrometry. J Agric Food Chem 53:3779–3783 1 3

Journal

ChromatographiaSpringer Journals

Published: May 2, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off