Establishing a method of HPLC involving precolumn derivatization by 2,2′‐dithiobis (5‐nitropyridine) to determine the sulfites in shrimps in comparison with ion chromatography

Abstract Although sulfites are widely used in shrimp processing, the contents of residual sulfite need to be strictly controlled due to their potential toxicity. In this paper, a novel method was developed for determination of the free and total sulfites in shrimps. Major procedures of the method includes separation of free and total sulfites with ultrasound‐assisted extraction and pH adjustment for 20 min, then a precolumn derivatization was conducted by 2,2′‐Dithiobis (5‐nitropyridine) and verified by LC‐MS, and finally HPLC coupled with an ultraviolet (UV) detector was carried out. Results indicated that the UV absorption wavelength shifted from 213 (sulfites) to 320 nm (new disulfide compounds), significantly reducing the interference of natural occurring compounds and solvents in the matrix. The standard curves exhibited a good linear range of 3.2–51.2 mg/L (R 2 = 0.9996). The limit of detection (LOD) and limit of quantification (LOQ) were 0.3 and 1.0 mg/L, respectively. The contents of free and total sulfite in frozen shrimps were 26.58 ± 0.48 and 31.44 ± 0.83 mg/kg calculated by SO2, respectively. These were similar (p > 0.05) to the data obtained by the method of ion chromatography. In conclusion, the new developed method has been proved to be a reliable and economic method for effective determination of free and total sulfites in the shrimps, and the method could be expanded in determination of the sulfites in other food products.

reversibly bound sulfites could be released by extraction with an appropriate pH (McLeod & Davey, 2007).
Sulfites are widely used in food processing and hard to be replaced, because they are economic and easy for application (Zhang et al., 2014). Despite of these remarkable advantages, sulfites should be applied with strict limitation due to their potential toxicity (da Costa Machado Matos Carvalho et al., 2011). People may suffer from asthmatic and allergic reactions if they ingested foods containing large amounts of sulfites, especially in free (or reversibly bound) sulfite form (Moseholm, Taudorf, & Frøsig, 2010;Reno, Brooks, & Ameredes, 2015). Besides, long-term exposure to sulfites can cause neurotoxicity and even damage to the reproductive system (Kencebay et al., 2013;Marshall, Reist, Jenner, & Halliwell, 1999). Irreversibly bound sulfites are hard to be detected by common analytical techniques, because of their extraordinary stable forms.
High-performance liquid chromatography (HPLC) is widely used nowadays. HPLC coupled with multi-wavelength spectrophotometer was used to analyze the free and total sulfites in foodstuffs (Lim et al., 2014;Pizzoferrato, Di, & Quattrucci, 1998). HPLC instruments, sometimes combined with preseparation treatments, were used to analyze the sulfites in fresh sausages, shrimps, grapes, dehydrated fruits and vegetables, and the results were similar to those obtained by traditional Monier-Williams distillation method (Ni et al., 2015;Pizzoferrato, Quattrucci, & Di, 1990).
Three main pretreatment methods were available for the determination of sulfites in foods by HPLC: (a) direct determination of the sulfite concentration of the sample solution; (b) oxidation of sulfite to sulfate, and (c) sulfite derivatization (Jackowetz & Orduña, 2013;McFeeters & Barish, 2003). The ultraviolet absorption wavelength of sulfites is 213 nm, at which a large number of natural compounds (such as protein and amino acids) and organic solvents can absorb and cause disturbances, resulting in decreased accuracy. Compared with other methods, the derivatization method possessed obvious superiority with lower detection limits and less interference.
Sulfite ions can react with many organic disulfides to displace thiol anions and form organic thiosulfates, commonly referred as "Bunte" salts (Field, 1977). The reaction is according to Equation 1.
The disulfide (RSSR) reacts quantitatively with sulfite ions (SO 3 2− ) to generate new sulfite-containing dithio compounds (RSSO 3 − ) and thiol (RS − ). In almost all cases, the disulfides can be fully reacted with sulfite ions in a ratio of 1:1 stoichiometry, and this principle was often used for the detection of disulfides (Li & Zhao, 2006). In addition, nitro-containing disulfides were more sensitive to this type of reaction, and traces amounts of sulfite ions can react quantitatively with nitro-containing aromatic disulfides, making it detectable by UV spectrophotometer (Humphrey, Ward, & Hinze, 1970). Certain disulfides, particularly of 2,2′-Dithiobis (5-nitropyridine) (DTNP), were commonly used to determine the thiol in various biochemical samples. The reaction process of DTNP and SO 3 2− was shown in Figure 1.
Herein, we aimed to separately extract the free and total sulfites from shrimps and then quantify by precolumn DTNP derivatization HPLC (PD-HPLC). Both detection limit and recovery rate were evaluated, and this proposed method has been also verified and compared with ion chromatography.

| Preparation of samples
Frozen shrimps were delivered to the laboratory by an ice incubator in two hours with the temperature kept below −10°C. Once delivered, it was stored in a refrigerator at −18°C for further use.
Prior to the analysis, the shrimps were homogenized using a meat grinder (JYS-A960 Joyoung), then weighed accurately (10,0000 g), put into a 250-ml Erlenmeyer flasks, and sealed. 50 ml of propyl alcohol (1%, v/v) solution was added to stabilize sulfite and prevented oxidation.
The method of separation of sulfites was based on the previous method (Wang & Xu, 2014) with slight modifications. The extraction solution was adjusted to pH 8.0 for free sulfite extraction and pH 12.0 for total sulfite extraction (Rita, Maria, & Eduardo, 2009), using 0.5 mol/L of sodium hydroxide, then extracted with the assistance of ultrasonic leaching for 15 min, and centrifuged (Bechman Allegra model) at 7104 g for 15 min at 4°C. The supernatants were transferred to a volumetric flask and then diluted with ultrapure water to be 100 ml for HPLC measurements.

| Derivatization process
8.0 ml of ultrapure water, 0.5 ml of 0.2 mol/L acetic acid-sodium acetate buffer, 1.0 ml of the above prepared standard or sample solution, and 0.5 ml of 5.0 mmol/L DTNP solution (dissolved in acetonitrile) were added in turn to a 20 ml test tube and mixed thoroughly for 30 s at 25°C. The mix was filtered through a 0.22 μm membrane (Millipore) for HPLC analysis.

| HPLC conditions
High-performance liquid chromatography separation was performed using a Waters 1525 chromatography systems (Waters Inc.), with a Waters 2487 dual-channel UV-visible detector at wavelength of 320 nm. The injection (20 μl) was automatic using a W2707 autosampler, and a C 18 column (5 μm; 4.6 × 250 mm internal diameter) from Waters was used. Column temperature was kept at 30°C in a column oven, and samples were held in the autosampler tray at 20°C prior to the injection. The mobile phase were as follows: 0.05 mol/L acetic acid-sodium acetate solution (A) and pure acetonitrile (B).
Intraday precision determination included the following: the retention time and the RSD value of the peak area of seven continuous injections; interday precision: the retention time and the RSD value of five injections in 3 days.

LC-MS analysis was carried out on a Waters 2695 coupled with
MS spectrometry (Thermo Fisher LCQ™ Deca XP plus) and using a Waters C 18 column (5 μm; 4.6 × 250 mm). The mobile phase was 5 mmol/L ammonium acetate (A) and pure acetonitrile (B), and the elution procedure was the same as HPLC. Negative electron spray ionization mass spectrometry (ESI-MS) was used to analyze the chemical structure of sulfite derivatives.

| The PD-HPLC method compared with the IC method
ICS-1100 ion chromatography system equipped with a conductivity detector (Thermo Fisher Scientific Inc) and a Dionex IonPac AS9-HC Column (4 mm × 50 mm) were used to determine sulfites.

| Data analysis
All analyses were triplicated, and the results were expressed as mean ± SD. SPSS version 19.0 (IBM, 2018) was used to conduct all statistical analyses. A t test was used for comparison between two means, and a one-way analysis of variance (ANOVA) was used for comparison of more than two means; a p value of less than 0.05 was assumed to be statistically significant.

| UV absorption spectrum
The absorption spectra of SO were consistent with the reports of similar compounds (Humphrey et al., 1970;McFeeters & Barish, 2003). As the reaction solution containing excess of sulfite ion, and the UV absorption spectra showed a reduced absorption peak around 320 nm compared with original DTNP, it was speculated that there might be some new reaction products, such as the disulfide compounds containing sulfites. This speculation will be verified by LC-MS. Figure 2 displayed that the absorption wavelengths of the sulfites and DTNP derivative products were shifted from 213 to 320 nm, significantly increasing the selective detection of sulfites. This was also verified in Figure 3 that peak 2 at 9.76 min had significantly higher response than the peak 1 at 9.13 min. Within a certain concentration range, the area of peak 2 was changed accordingly with different sulfite concentration. Figure   confirmed that SO 3 2− and DTNP were reacted at a molar ratio of 1:1, as shown in Figure 1. Figure 3 also compared the HPLC chromatographs of standard sulfite solution, shrimp sample without added sulfites, and shrimp sample added with sulfites. The results revealed that no other compound in the sample could interfere with the derivatization process. concentration of SO 2 , mg/L), with an excellent correlation coefficient (R 2 = 0.9996). It could be seen in Table 1, the intraday precision was better than the daytime precision. The intraday precision and interday precision of the total sulfites under alkaline extraction conditions were slightly lower than the free sulfites, which might result from the effect of higher pH on the stability of the derivative product. However,

| Calibration curves, precision, detection limits, and repeatabilities
the RSD values were all within 4%, meeting the testing requirements.
Moreover, the limit of detection (LOD) and limit of quantitation (LOQ) of sulfites were 0.3 and 1.0 mg/L, calculated as SO 2 , based on a signalto-noise ratio of 3 (S/N = 3) and 10 (S/N = 10), respectively.

| Sample analysis and comparison
Three different levels of standard sulfites (5.0, 20, and 50 mg/kg, in content of SO 2 ) were added into the frozen shrimp homogenates.
The precision and recovery rates of the PD-HPLC method were carried out instantly, and the parallels were also conducted and compared with an IC method. As described previously, the free and total sulfites were selectively extracted from the matrix at pH 8.0 for free sulfites and pH 12.0 for total sulfites. Both of them were determined by PD-HPLC and IC methods, and all the results were listed in

| CON CLUS IONS
In summary, a HPLC involving precolumn derivatization method for the analysis of sulfites in shrimps has been developed. The derivatization was based on the reaction of SO 3 2− with DNTP and verified by LC-MS. UV and visible absorption spectrum showed that all the derivatized products had absorption at 320 nm, and they were well separated by gradient elution in HPLC. The linearity was good in a range of 3.2-51.2 mg/L (R 2 = 0.9996). The proposed method could be successfully used to detect sulfites in shrimp samples with high recoveries (89.2%-94.8%) and reasonable relative standard deviations (RSD <2.4%). The detection limits of the PD-HPLC method were slightly lower than the IC method. Accordingly, this new developed method can be used to determine the residue of sulfites in shrimp and other foodstuffs.

ACK N OWLED G M ENT
This work was funded by the project of Zhoushan Science and Technology Bureau (No. 2016C21081).

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

AUTH O R CO NTR I B UTI O N S
Authors K. Yang led the relevant project, designed the study, interpreted the results, and revised the paper. C. Zhou and Z. Yang designed and carried out the experiments, performed the data analyses, and wrote the manuscript draft. L. Yu provided the shrimps and collected the test data. C. Wu, M. Cai, and P. Sun helped with data analysis, polished the language, and edited the manuscript.

E TH I C A L S TATEM ENT
This study does not involve any human nor animal testing.