Establishment of a Method for Determination of Arsenic Species in Seafood by LC-ICP-MS

14 An analytical method for determination of arsenic species (inorganic arsenic (iAs), 15 methylarsonic acid (MA), dimethylarsinic acid (DMA), arsenobetaine (AB), 16 trimethylarsine oxide (TMAO) and arsenocholine (AC)) in Brazilian and Spanish seafood 17 samples is reported. This study was focused on extraction and quantification of inorganic 18 arsenic (iAs), the most toxic form. Arsenic speciation was carried out via LC with both 19 anionic and cationic exchange with ICP-MS detection (LC-ICP-MS). The detection limits 20 (LODs), quantification limits (LOQs), precision and accuracy for each arsenic species were 21 established. The proposed method was evaluated using eight reference materials (RMs). 22 Arsenobetaine was the main species found in all samples. The total and iAs concentration 23 in 22 seafood samples and RMs ranged between 0.27 – 35.2 and 0.02 – 0.71 mg As kg -1 , 24 respectively. Recoveries of between 100% and 106% for iAs, based on spikes, were 25 achieved. The present results provide reliable iAs data for future risk assessment analysis.


32
The rapid expansion in trade of seafood products makes this an important market 33 worldwide (De Silva & Bjondal, 2013). The increase in global consumption of seafood is 34 associated with several benefits such as a reduction in risk of several diseases (Innis, 2007;35 Zmozinski, Passos, Damin, Espirito Santo, Vale, & Silva, 2013). On the other hand, 36 concerns about human health have arisen since several arsenic species have been detected 37 in seafood (Leufroy, Noël, Dufailly, Beauchemin, & Guérin, 2011). The toxicity of As is 38 dependent on its chemical species, with inorganic species (iAs) such as arsenite (As(III)) 39 and arsenate (As(V)) being the most toxic (Geng,Komine,Ohta,Nakajima,Takanashi,& 40 Ohki, 2009). Other arsenic species such as monomethylarsonic acid (MA) and 41 dimethylarsenic acid (DMA) are less toxic to humans, with asenobetaine (AB) being quality control, the As concentration in in-house prepared As speciation standards was   155 The following certified reference materials (CRM) were used for method bivalve samples were purchased from local supermarkets in Barcelona, Spain, during 2013. 173 All these samples were analyzed in a raw state (wet weight) without lyophilization or other 174 pretreatments. Only edible parts of each fish and seafood were used for the analysis. 175 Samples were washed with Milli-Q water, cut, and homogenized using a blender (non-176 contaminating kitchen mixer; Multiquick 5 Hand Processor, Braun, Barcelona, Spain).

177
After homogenization, samples were stored in the refrigerator at 4-10 °C until analysis 178 (before 2 days). The moisture of fresh samples was determined in triplicate by drying 0.5 g aliquots 182 in an oven at 102 ± 3°C until constant weight. Moisture ranged from 45% to 94%, and all 183 results are expressed as dry mass.  185 The total arsenic content in seafood and CRM samples was determined by ICP-MS 186 following microwave digestion. Initially, 0.5 g and 2 g aliquots of lyophilized and fresh 187 samples, respectively, were weighed in digestion vessels, after which 8 mL of concentrated 188 nitric acid and 2 mL of hydrogen peroxide were added. The microwave digestion procedure 189 was carried out according to the following programme: 10 min from room temperature to  The extraction of As species was based on our previous study 206 Calderón, Centrich, Rubio, & López-Sánchez, 2014). For this, 0.2 g and 1.0 g aliquots of 207 10 lyophilized and fresh samples, respectively, were weighed in digestion vessels and 10 mL 208 of a solution containing 0.2% (w/v) of nitric acid and 1% (w/v) of hydrogen peroxide were 209 added to perform a microwave assisted extraction (MAE) at temperature of 95 °C. Samples 210 were cooled to room temperature and centrifuged at 3500 rpm for 25 min. The supernatant 211 was filtered through PET filters (Chromafil, Macherey-Nagel, pore size 0.45 µm).

212
Triplicate analyses were performed for each sample. This extraction method completely 213 oxidizes As(III) into As(V), without conversion of the other organoarsenic species into 214 inorganic arsenic (iAs). The iAs was identified and quantified as As(V) in the extracts by 215 comparing the chromatographic peak for the samples with the peak of As(V) standard 216 solution. Total arsenic in the extracts was determined by ICP-MS (as described previously).

217
Arsenic speciation was carried out in the extracts by LC-ICP-MS. Two chromatographic 218 separation methods were used for separation of the arsenic species. As(III), As(V), DMA  To evaluate the accuracy of the applied procedure, several CRMs were analysed.

240
Seafood CRMs (TORT-2, DOLT-4, SRM 2976, SRM 1566b, BCR-627, ERM-BC211 and 241 ERM-CE278) and one material reference (9 Th ) were analysed during the study. The 242 concurrent analyses of the CRMs listed above were used to measure the accuracy of the 243 determination of total As (Table 1). For quality control of acid digestion, a CRM was 244 analysed in every batch of samples measurements (total As concentration). The comparison 245 between each obtained value of total As with its corresponding certified value (Table 1) 246 showed no significant difference at a 95% confidence level when Student's t-test was 247 applied. The repeatability (six times within a day, n=6) was assessed for the results 248 obtained by analysis of different replicates of CRMs (Table 1). The RSD (%) values were: 249 4.9% for TORT-2 and 1.2% for DOLT-4. The detection (LOD) and quantification limits 250 (LOQ) were calculated as three times the standard deviation (3σ) and ten times the standard 251 deviation signal (10σ) of ten digestion blanks, respectively (Llorente-Mirandes et al., 252 2014). The results obtained were as follows: 0.006 mg As kg -1 dry weight basis for method 253 detection limit and 0.021 mg As kg -1 dry weight basis for method quantification limit. The extraction efficiency was evaluated by calculating the ratio between total 258 arsenic present in the samples, given by the acid digestion, and the total arsenic present in 259 the extracts. The extraction efficiencies are presented in Table 1 for the CRMs and Table 2 260 for the real samples. The efficiency obtained in this work varied between 73% and 104% 261 with an average of 89%, which is consistent with the literature (Amayo, Petursdottir,  Column recovery is expressed as the ratio of total As (sum of all arsenic species) 282 eluted from the chromatographic column to the total As in the extract injected into the 283 chromatographic column. Measurement of column recovery is essential to provide a control 284 of chromatographic separation and to evaluate the quantification of the As species. The 285 column recovery values ranged from 58% to 99% for CRMs (Table 1) and 70% to 104% 286 for all samples (Table 2). These values are in agreement with those reported by Zheng & 287 Hintelmann (2004), which found values from 85 to 110% using HPLC-ICP-SFMS and 288 methanol/water as extracting agent.

290
Recovery of inorganic arsenic 291 Standards of As(III) and As(V) were spiked in solid samples of red porgy, tuna-1, 292 clam-1, mussel and CRM TORT-2 and then homogenized. Samples were taken for 293 extraction 30 minutes after spiking. Quantitative oxidation of As(III) to As(V) was 294 achieved since only As(V) was found as iAs in the spiked samples. Thus, anion LC-ICP-

295
MS was used to quantify the As(V) as iAs in the samples. The recoveries found for red 296 porgy, tuna-1, clam-1, mussel and TORT-2 were 102 ± 2, 100 ± 5, 100 ± 4, 101 ± 2 and 297 106 ± 2 (mean % ± standard deviation, n=3), respectively. These recovery values were and red porgy extracts, respectively. The clam-1 was fortified with 0.200 mg As kg -1 of 301 As(III) and As(V); the red porgy with 0.250 mg As kg -1 of As(III) and As(V). As can be 302 seen, iAs was recovered successfully as As(V) from the two samples. In order to verify the accuracy of the proposed speciation method, two CRMs were 306 analysed and evaluated: BCR-627 (Tuna fish) and ERM-BC211 (Rice). The CRM BCR-307 627 has a certified value of 3.9 ± 0.22 mg As kg -1 for AB and 0.15 ± 0.02 mg As kg -1 for 308 DMA. To assess the accuracy of the inorganic arsenic results, the ERM-BC211 rice 309 material was analysed because there is no CRM for measurement of inorganic arsenic in 310 seafood. The ERM-BC211 has a certified value of 0.124 ± 0.011 mg As kg -1 for iAs and 311 0.119 ± 0.013 mg As kg -1 for DMA. The values found for the ERM-BC211 and CRM BCR-312 627 are shown in Table 1 and did not differ significantly from certified values at a 95% 313 confidence level.  RMs, however the concentrations found in this work are given in Table 1.  The reported values ranged between 0.010-0.036 mg kg -1 and 0.315-0.823 mg kg -1 for 348 DOLT-4 and TORT-2, respectively (Table 3). This fact illustrates that solvent plays a role

365
In summary, the concentrations of iAs found in this work (Table 1)   66% for oyster and mussel, and 95% for fish, Table 2), which is considered non-toxic. In 400 contrast, Zheng & Hintelmann (2004) found lower levels of AB in samples collected from 401 the Moira Lake (less than 16% of total arsenic). Those data demonstrate the need to carry 402 out speciation in seafood samples as the total amount of As does not provide enough 403 information about the toxicity of the analysed sample.  Table 2.  dimethylarsinoylsugarphosphate, which were identified in fish and molluscs (Nischwitz & 436 Pergantis, 2005). Due to the lack of appropriate standards, this attribution was not checked.

437
The inorganic arsenic was extracted, identified and quantified as As(V), and 438 selectively separated from other arsenic compounds. It was found in 36% of all samples 439 being always below 3.3% of the total arsenic. For fish samples, the inorganic arsenic 440 content is in all cases below the limit of detection. (n=14). This is illustrated in Figure 2a, 441 which shows that inorganic arsenic was not detected in red porgy extracts (continuous line), 442 and also shows that the all the spiked iAs was successfully recovered as As(V) (dotted 443 line). The extraction method not converted the other organoarsenic species into inorganic 444 arsenic (iAs). Figure 2b shows that the major arsenic compound in red porgy extracts was 1a, continuous line), which was fortified with As(III) and As(V), and as can be seen, iAs 456 was recovered successfully as As(V) (Fig. 1a, dotted line). The lowest concentration of iAs 457 (0.033 ± 0.003 mg kg 1 ) was found in shrimp, as previously observed (Baeyens et al., 2009;458 Leufroy et al., 2011;Sirot et al., 2009;Sloth, Larsen & Julshamn, 2005).

459
The present results showed a wide variability in the arsenic species found in seafood 460 samples, highlighting the need to carry out speciation to discern the toxic from the non- column and it could co-elute with other cationic species usually found in seafood (specially 471 AB). Therefore, the oxidation of As(III) to As(V) allows the determination of iAs as As(V) 472 which is well separated from other As species. Also it is remarkable that is not necessary to 473 quantify two peaks to determine iAs, so errors are minimized. Thus, the present method 474 allows an accurate quantification of iAs and could be a valuable tool for food control 475 laboratories which assessing the iAs in seafood samples.

476
To assess the applicability of the method, total arsenic and arsenic species in 477 different seafood samples, including fish, crustaceans and bivalves, were determined. AB 478 was the predominant arsenic species in all samples. Inorganic arsenic content was below 479 the detection limit in all fish samples, whereas it was found in all bivalves and crustacean 480 samples ranged from 0.02 to 0.71 mg As kg -1 of iAs.

481
For an accurate assessment of food safety more efforts will be needed such as   Table 1. Total arsenic and arsenic species in reference materials; concentrations are expressed as mg As kg -1 dry mass (mean ± SD, n = 3 and *n=6).    Table 4. Total arsenic in seafood samples, concentrations are expressed as mg As kg -1 dry mass (mean ± SD, n = 3).