Sensitive profiling of trace neurotoxin domoic acid by pressurized capillary electrochromatography with laser-induced fluorescence detection

A stable, sensitive and low-reagent consumption method for the quantification of trace neurotoxin domoic acid (DA) was presented by pressurized capillary electrochromatography (pCEC) with laser-induced fluorescence (LIF) detection. Migration behaviours of DA were studied under different separation modes such as pCEC and capillary HPLC. Separation parameters including the mobile phase, applied voltage, supplementary pressure and sample loading rate were investigated. Double modes including electrophoretic migration and chromatographic retention were employed and an efficient separation of DA was achieved with a flow of 1.2 mL min 1 in a capillary column. Sensitive analysis of DA was achieved with a detection limit of DA in shellfish tissue extracts as low as 10 ng mL 1 (equivalent to 15 ng DA per g, wet weight). The stability and repeatability of pCEC-LIF was evaluated and a good result was gained with RSDs (n 1⁄4 3) for the retention time and peak area of DA less than 0.5% and 2.0%, respectively. Applied to shellfish samples, the recoveries were satisfactory and the variation of the quantification of DA was acceptable when compared with the HPLC-MS/MS method.


Introduction
Domoic acid (DA), the principal causative toxin of amnesic shellsh poisoning (ASP), is produced by the algae pseudonitzschia and could accumulate in the edible tissue of shell-sh. 1,2 Consuming the contaminated shellsh seafoods could cause severe neurological symptoms such as memory loss, headaches, loss of balance, nausea, etc. 3 Marine animals and even human beings will exhibit acute intoxication aer consumption of these contaminated foods. Fabricating a facile and sensitive method to monitor trace DA is important and is attracting more and more attention.
Commonly, HPLC with UV detection has been widely utilized for quantifying DA in shellsh. 4 By means of uorescence detection (FLD) with pre-column or post-column derivatization, the HPLC-FLD method is oen used with a low detection limit for trace analysis of DA. [5][6][7] HPLC with MS detection including LC-MS/ MS, LC-laser ablation ESI-MS and SPE-LC-MS/MS, have also been developed for the detection of DA. [8][9][10] These methods are stable, efficient and sensitive, however, a considerable amount of organic elution solution is consumed and consequently these methodologies might be unfavourable for economical and environment-friendly applications to a certain degree.
As the effective microuid separation technology, capillary electrophoresis (CE) and capillary electrochromatography (CEC) have been proposed. Fast analysis of DA using microchip electrophoresis with laser-induced uorescence (LIF) detection has been proposed. 11 CEC coupling a high voltage to induce electroosmotic ow (EOF) with a capillary LC column has been formed for the analysis of ASP toxins. 12,13a Leaó Martins et al. rstly developed a CEC method for the analysis of DA. 12 A. Gago-Martínez et al. reported the detection of DA with the combination of CEC and photodiode array detection (PAD). 13 CE and CEC have been studied and applied to algae toxins. However, the primary problem in CE or CEC was associated with bubble formation produced by Joule heating as a result of the electric current passing through the column. 13 The bubbles generate baseline noise and make it difficult to obtain stable ow conditions. Besides, by using electrokinetic injection, quantitative sample introduction was achieved in CE or CEC for the DA analysis, but it might be inuenced by the conductivity of sample matrix. The main drawback associated with CE or CEC is the lack of adequate robustness and reproducibility. 13,14 To address these problems mentioned above, pressurized capillary electrochromatography (pCEC) has been proposed by coupling an HPLC pump onto the inlet end of the capillary to provide auxiliary pressure to ush capillary column with solvent and inhibit bubble formation. 13a,b Quantitative sample injection could be facilely achieved through a reliable rotary-type injector in pCEC system. In pCEC, both high efficiency of CEC and high selectivity of HPLC were successfully combined. [15][16][17] The mobile phase is driven by both hydraulic pressure and electroosmotic ow (EOF), and resulted in an efficient elution and the questions such as bubble formation and unstable quantitation could be effectively solved. [18][19][20] By superimposing a hydraulic pressure onto the electrically generated ow in capillary column, double modes that are electronic interaction and chromatographic retention could be employed for the efficient separation with a high resolution. 18,19 Furthermore, the mobile solution as low as nano-litre ow level in capillary column could be carried out facilely in pCEC. It is desirable and become an attractive alternative to common HPLC and pure CEC for the robust and effective analysis.
To date, with the popular UV detection, pCEC-UV method has been used to rapid quantify of DA in shellsh seafood with a determination limit equivalent to 2.0 mg DA per g. 21 It always suffers from low detection sensitivity due to a rather short optical path length in capillary column for on-line photometric detection. To gain a high sensitivity, laser induced uorescence detection (LIF) that is well suitable for capillary microseparation with a strong laser to overcome the limit of optical path length in capillary column could be utilized. Ultratrace levels of detection limit (10 À11 mol L À1 levels for uorescein) were achieved with LIF and presented an obvious advantage over common UV detection. 22 Since introduced rstly by Yan et al. in 1995, 23 capillary electrochromatography with LIF methods have attracted close attentions in trace detection of polycyclic aromatic hydrocarbons, 23 aatoxins, 24 avins in plasma 25 or proteins. 26 So far, pCEC-LIF method for the trace detection of non-protein amino acids toxin DA has still not reported and its utilization will provide an attractive way for the low-reagent consumption and sensitive analysis of ASP.
Here, we present the development and application of the pCEC-LIF method for the quantication of trace DA in shellsh tissue extracts. Migration behaviours of DA were studied under different separation modes including pCEC and capillary HPLC. A series of parameters including mobile phase, applied voltage, hydraulic pressure and sample loading rate were investigated for optimum electrochromatography behaviour. The stability and reproducibility of pCEC-LIF method were evaluated, as well as the detection limit of DA. The recoveries of shellshes samples spiked with DA were analyzed, and the method validation in detection results was nally evaluated by comparing with the authoritative HPLC-MS/MS method.

Instrumentation
pCEC was performed on pCEC system (Trisep™ 2100 GV, Unimicro Technologies, USA), which consisted of a solvent delivery gradient module, a micro-uid manipulation module (including a 0.8 mL six-port injection valve, a four-port split valve, and a back-pressure regulator), a high voltage power supply (0-AE30 kV), a data acquisition module, and a LD-DPSS LIF detector (10 mW, l ex ¼ 473 nm; l em ¼ 530 nm). Samples are injected to sample loop and then carried to four-port split valve by mobile phase. Under a constant pressure, the total ow from four-port valve was split and a nano-ow was driven into capillary column with a split device. Applied voltage was imposed at the outlet end of column, and the inlet end was connected to the four-port split valve and grounded. HPLC-MS/ MS system and the program were shown in ESI. †

Sample preparation
The procedure for handling shellsh tissue was based on the previous ref. 6 and 21. Approximately 50 g of shellsh tissue was homogenized with the blender. A 4.0 g homogenated tissue was used and extracted with 15.0 mL extraction solvent (1 : 1 methanol-water) and homogenized again for 3 min at 13 000 rpm. The resulting slurry was then centrifuged at 4000 rpm for 10 min, and the centrifuged sample was then ltered through a 0.22 mm membrane and was puried further using the following SPE procedure according to the work reported by M. Maroulis et al. 6 A 5.0 mL fraction of the ltered crude extract was loaded onto the cartridge in a ow rate of about one drop per second. The cartridge was washed at about one drop per second with 5 mL wash solution (acetonitrile-water ¼ 10 : 90, v/v). Then the effluent was discarded and 0.5 mL of 0.1 mol L À1 formic buffer eluent (pH 2.92 AE 0.01, adjusted with sodium hydroxide) was added. The effluent was discarded again. The formic buffer (0.1 mol L À1 ) were loaded onto the cartridge and allowed to ow at a rate of one drop per second. Exactly 2.0 mL of the nal eluted DA extract were collected for further analysis.

Derivatization of domoic acid with NBD-F
According the ref. 8, NBD-F in acetonitrile (1.0 mg mL À1 , 50 mL) and 0.10 mol L À1 sodium borate (50 mL) were mixed with sample extract or standard solution (50 mL) in an amber vial (2 mL), at 20 C and the reaction was terminated aer 3 min by adding 1.0 mol L À1 HCl (50 mL) to the mixture. An aliquot of this derivatised mixture was analysed directly by pCEC-LIF.

pCEC-LIF procedure
Mobile phase was composed of phosphate buffer solution and ACN with the appropriate concentration and desired pH value.
The pH values and salt concentrations mentioned in this work were that of phosphate buffer solution.
By using acetonitrile-phosphate buffer solution (60 : 40, v/v) as mobile phase, isocratic pressurized electrochromatography was performed with a octadecyl silica (ODS) packed capillary column mentioned above. A supplementary pressure (7.2 MPa) was applied to the column inlet and the nano-ow was driven by the pump with a split ratio of 1/15. Before conducting pCEC experiments, capillary column was conditioned with mobile phase for 30 min and the column was kept at 25 C. The applied voltage was ramped from 0 to +8 kV at rst and then maintained at +6 kV. Fluorimetric detection was used (l ex ¼ 473 nm; l em ¼ 530 nm).

Method validation
The method was evaluated and validated through linearity, limit of detection (LOD), fortied recoveries and precision, and the program was shown in ESI. † Precision of the method was evaluated by the intra-day and inter-day RSD. The intra-day RSD was obtained from analysing ve replicates of the sample spiked with 0.25 mg mL À1 of DA on the same day, and the interday RSDs were determined by analysing the sample based on ve replicates on three consecutive days. The comparison with the authoritative HPLC-MS/MS method was carried out to further evaluate the feasibility of this method.

pCEC mode for DA detection
In pCEC, an HPLC pump to superimpose a hydraulic pressure at the inlet of capillary column onto an electrically generated ow. Mobile phase is driven by both EOF and pressurized ow, and the linear velocity of a charged solute in mobile phase (u m ) is the sum of the velocities contributed from electrophoretic migration velocity (u em ), pressurized ow (u p ), and EOF (u eo ). As shown in Fig. 1, double driving forces including a hydraulic pressure and an electroosmotic ow (EOF) could be employed and double modes including electrophoretic migration (EM) between the electrode and charged solutes and capillary micro-liquid chromatographic (m-HPLC) retention of packed stationary phase could be employed for the separation of DA derivatization products.
To evaluate the chromatographic behaviour of DA derivatization product DA-NBD, DA standard solutions and shellsh tissue extracts spiked with DA were used and the results were shown in Fig. 2 and 3. In Fig. 2, the migration time of DA-NBD was shortened to 12 min in pCEC with a positive high voltage of +6 kV, while prolonged to more than 16 min when a negative high voltage À6 kV was applied. It indicated that DA-NBD was of negatively charge and the electrophoresis action could enhance the migration rate and the uorescence response of DA-NBD remarkably, which was in favour of the rapid and sensitive analysis of trace DA. By using a positive high voltage, the migration rate of DA-NBD was accelerated obviously benetting from electrophoretic migration with the same direction of hydraulic pressure though the cathode EOF was in the oppose direction. In pCEC, the electrophoretic migration and electroosmotic ow were of importance for the migration and effective separation of DA.
Compared with the m-HPLC method which achieved under the same experimental conditions as that of pCEC method except applied voltage, a higher sensitivity and a rapid elution of DA were obtained (shown in Fig. 3). Applied to shellsh tissue extracts, the pCEC-LIF method was suitable for DA detection. The separation of DA-NBD and co-eluted matrixes were successfully achieved with a high sensitivity.

Optimization of separation parameters
3.2.1 Effect of pH of buffer solution. The pH value mentioned in this case was that of phosphate buffer solution. DA has three carboxyl groups and a secondary amine, and the   pH of mobile phase has an obvious inuence on its chromatographic retention and electronic interactions. In this case, the migration of neutral NBD-F was drove by hydraulic pressure and electroosmotic ow (EOF), and then was studied to evaluate the change of EOF. As shown in Fig. S3, † a little change of migration time occurred at pH 5.50-3.50 while a notable decrease was observed with pH increasing from 3.50 to 2.50, which indicated that the EOF decreased at pH 2.50. Interestingly, in Fig. 4, the retention and uorescence response of DA derivation product (DA-NBD) increase with the pH decreasing from 5.50 to 3.50, which was owing to the comprehensive action of the stable EOF in capillary column and the decreasing electrophoretic mobility (with an opposite direction to EOF). At pH 2.50, the deprotonation of three carboxyl groups and protonation of secondary amine in DA was in the highest level. So the net negative charge of DA-NBD was inhibited and thus electrophoretic migration of DA-NBD rapidly reduced. The vector sum of electrophoresis and EOF migration was in the opposite direction of hydraulic pressure, which was helpful to enhance the retention and band compression of DA-NBD in the column. As seen from Fig. 4, both migration time and uorescence intensity of DA-NBD reached the maximum at pH 2.50, which was favourable to seek a high sensitivity.
3.2.2 Effect of acetonitrile. The content of acetonitrile in mobile phase is a fundamental factor that could affect both the velocity of EOF and chromatographic retention. A high content of acetonitrile could cause an increase in electroosmotic mobility and a stronger elution in the reverse phase chromatography. As seen from Fig. S4, † an obvious decrease of the retention of DA-NBD and NBD-F hydrolysis products could be observed while the ACN content increased from 55% to 70%. With the ACN content of 70%, DA-NBD was eluted too fast and mixed with the co-existed substances together. While the ACN content got down to 55%, the retention of DA-NBD was intense and the migration was slow, which resulted in an overlap of DA-NBD and another co-existed substances. If the ACN content was of 60%, a good separation of DA-NBD and co-eluted compounds could be well achieved with an appropriate retention time.

Effect of buffer concentration.
The inuence of ionic strength in mobile phase on the separation of DA-NBD was also studied. Generally, the increase of buffer concentration will reduce the EOF of separation column. In Fig. S5, † the retention time of charged DA-NBD decreased slightly when buffer concentration increased from 2.50 mmol L À1 to 7.50 mmol L À1 . The increased phosphate buffer concentration in mobile phase   may contribute to the reduction of the EOF on the surface of stationary phase. When the buffer solution with a high concentration of 7.50 mmol L À1 was used, the migration of DA-NBD could be accelerated, while the migration of neutral NBD-F hydrolysis product was slowed with peak broadening obviously. As a result, the concentration of 5.0 mmol L À1 of PBS buffer solution was selected for the high resolution of DA.
3.2.4 Effect of applied voltage. In Fig. 5, the m-HPLC method was used for the separation without the application of voltage. In such a way, DA-NBD could be recognized from the matrix but the sensitivity was lower and retention time was longer than that of pCEC, which was unfavourable to sensitive monitoring of trace shellsh toxin. In pCEC, the increase of electric eld could enhance the EOF and alters the electrophoretic migration of the charged substances. In this case, the electrophoretic velocity of DA was faster than the EOF in the opposite direction. With the applied voltage increasing, the migration rate of DA-NBD was enhanced as well as uorescence response, which attributed to the increase of band compression of DA-NBD controlled by the reverse migration of EOF. For neutral NBD-F hydrolysis products, the migration was prolonged with applied voltage increasing, owing to the obvious increase of the reverse migration of EOF. At high voltage of +6 kV, a favourable resolution of DA-NBD was achieved with a high uorescence response. If the applied voltage increased further and got up to +8 kV, the higher value of applied voltage could lead to a faster migration rate and cause a peak overlap of DA-NBD and matrix solutes, which resulted in a serious decrease of the resolution of DA.
3.2.5 Effect of pump ow. In pCEC, the valid sample volume injected into capillary column is controlled by splitting ratio of the four port split valve. If keeping hydraulic pressure and the ow-rate constant, the spit-ow was constant. But if the ow rate decreased, the injection samples could stay in the spit valve for a longer time. With the ow rate of pump decreasing, the split-ow of injection solution to waste solution decrease, and the injection volume of samples into capillary column would increase. As seen in Fig. 6, when the applied voltage and hydraulic pressure keep constant, decreasing the ow rate in the capillary column from 3.0 mL min À1 to 1.2 mL min À1 resulted in a remarkable increase in the response of DA-NBD with an acceptable resolution. If the ow rate decreased further and reached 0.6 mL min À1 , the repeatability of pump ow was poor due to the limitation of HPLC pump. And so a low ow rate of 1.2 mL min À1 was selected for the stable and sensitive analysis of DA.

Analytical performance
3.3.1 Calibration curve, detection limit and repeatability. Based on the investigations mentioned above, the optimal conditions were achieved (mobile phase, 60% v/v ACN, 40% v/v of 5 mmol L À1 PBS buffer solution (pH 2.5), supplementary pressure: 1000 psi, applied voltage: +6 kV). And then a baseline separation of DA was gained and the co-existed components in the mussel extracts had no obvious inuence (Fig. 7). The linearity of the method was determined by constructing a calibration curve with a series of concentrations of DAs in shellsh tissue extracts. As shown in Table 1, the linear range, regression equation, correlation coefficient, and the limit of detection (LOD) are listed. Owing to the high uorescence yields and stability, the relative standard deviations including inter-day RSD (n ¼ 5) and inter-day RSD (n ¼ 15) for retention time and the peak area respectively was acceptable. The LOD of DA was of 10 ng mL À1 (S/N ¼ 3) (equivalent to 15 ng DA per g, w.w. of  shellsh tissue), which was 130 times more sensitive than that of pCEC-UV method (LOD, 2.0 mg DA per g). The sensitivity of this method would be better than that of CEC-PAD (the LOD was not provided in the reference, and it could be deduced as a level of mg DA per g like that of UV detector), better than that of CE-MS (70 ng DA per mL), 28 and kept in a similar level to normal HPLC methods for DA analysis (LOD, 6-25 ng DA per g). 7,8 Especially, in this pCEC method, a small amount of mobile phase was used with the ow in capillary column of 1.2 mL min À1 , which was only 1/800 of that used in normal HPLC. The LOD of pCEC-LIF is far less than the guideline value in mussels (20 mg DA per g, w.w. of shellsh tissue), and it is stable, low-reagent consumption and sufficiently sensitive to quantify DA in food safety monitoring.

Sample analysis.
This analytical method was also applied to three kinds of shellsh samples (oyster, Mytilus edulis and ruditapes). The shellsh samples were prepared through spiking standard DA solutions into blank shellsh samples. Oyster, Mytilus edulis and ruditapes were extracted and subjected to SPE as described above, produced a pCEC chromatogram in which the DA-NBD derivative product was well resolved. The mean recoveries of certied shellsh samples spiked with standard DA at three concentrations ranged from 90.9% to 110% with an acceptable RSD (RSD < 4.4%) (as shown in Table 2). Compared with the accepted authoritative HPLC-MS/MS method, the detection results were similar and acceptable, which shows the developed pCEC-LIF method is reliable and suitable to quantify trace DA toxin.

Conclusions
A stable and sensitive pCEC-LIF hyphenated method for the analysis of trace DA in shellshes was formed with low-reagent consumption, and the quantication for trace DA in shellsh seafoods was successfully achieved. Under the optimal conditions, a baseline separation with a high resolution of DA from shellsh matrices was gained within 12 min. The pCEC-LIF method was robust and sensitive with the LOD of DA as low as 10 ng mL À1 (S/N ¼ 3) (equivalent to 15 ng DA per g), which was more sensitive than most method fabricated with CEC-PAD and pCEC-UV. An extremely little amount of mobile phase was consumed economically with a ow of 1.2 mL min À1 in capillary column, which was only 1/800 of that used in normal HPLC. A good repeatability was also obtained and RSD values (intraday, n ¼ 3) for retention time and peak area were less than 0.5% and 2.0%, respectively. Applied to shellsh seafoods, the recoveries were obtained with satisfactory values in the range of 90.9-110%. Compared with the accepted authoritative HPLC-MS/MS method, the detection results were similar and in a good level. The results indicated that pCEC-LIF hyphenated technology was robust and sensitive enough for the quantitative analysis of trace DA toxin in shellsh seafoods and hold promise for the application to food safety monitoring.

Conflicts of interest
There are no conicts to declare.