Quantitative detection of organic mercury in whole blood using derivatization and gas chromatography-negative chemical ionization-mass spectrometry

Reagents

Standard solutions of organic mercury [MeHg and ethyl Hg (EtHg) in benzene, 10 μg/mL] and inorganic mercury (Hg²⁺, 10 μg/mL) were obtained from Wako Pure Chemicals (Osaka, Japan). Sodium tetraphenylborate, sodium tetrakis [3,5-bis(trifluoromethyl)phenyl] borate (BArF₂₄) and lithium tetrakis (pentafluorophenyl) borate ethyl ether complex (BArF₂₀) were purchased from Tokyo Chemical Industries (Tokyo, Japan). A certified reference material (CRM) for MeHg (cod fish tissue, NMIJ CRM 7402-a No.250) was obtained from the National Institute of Advanced Industrial Science and Technology (Tsukuba, Japan). 2,4,6-Trichloroanisole-d₃ (TCA-d₃) standard solution (Wako Pure Chemicals) was used as an internal standard (1 μg/mL toluene solution) because labeled MeHg was not readily available.

Analysis of standard solutions

Standard mercury solutions (MeHg and EtHg in benzene, 10 μg/mL) were treated with derivatizing reagents (tetraarylborates) and then analyzed using GC-EI-MS and GC-NCI-MS. First, 0.1 M acetate buffer (pH 5) and 0.1 mL of 2% BArF₂₄, BArF₂₀, or tetraphenylborate and 0.5 mL of toluene were combined. The standard solution was shaken for 30 s and centrifuged (830 g, 5 min). The water phase was removed, and the solution was washed with 0.5 mL of fresh water. The toluene phase was divided into a GC vial. Mercury halides were extracted using halide salts and also analyzed using EI and NCI. The standard solution was mixed with HBr (5 M, 0.1 mL), CuCl₂ (2 M, 0.05 mL), and toluene (0.5 mL) in sequence. The mixture was shaken for 30 s and centrifuged at 830 g for 5 min. The toluene phase (0.5 mL) was transferred to a GC vial for GC-MS analysis.

Sample preparation

Purification and extraction of inorganic and organic mercury were performed following the method of a previous study^[12]. To extract mercury compounds from whole blood samples (0.5 mL for unknown concentration samples and 0.1 mL for known high concentration samples), 0.1% cysteine solution (0.1 mL) was added to 2-mL polypropylene tubes. Then, the solution was made up to 0.5 mL with distilled water. Methyl isobutyl ketone (MIBK, 0.5 mL) was added, and the mixture was shaken for 30 s and centrifuged at 830 g for 5 min to remove the fat contents. The upper MIBK layer was removed, and the remaining solubilized water solution was again washed with 0.5 mL of hexane. The hexane layer was discarded.

To extract organic mercury, solubilized water solutions were mixed with HBr (5 M, 0.1 mL), CuCl₂(2 M, 0.05 mL), 10 μL of the internal standard solution (10 ng per sample), and 0.5 mL of toluene in sequence. The mixture was shaken for 30 s and centrifuged at 830 g for 5 min. The toluene phase (0.5 mL) was transferred to a GC vial for GC-MS analysis.

To derivatize inorganic mercury in the remaining solubilized water solution, 0.1 mL of 0.1 M acetate buffer (pH 5), 0.1 mL of 2% BArF₂₄, and 0.5 mL of toluene were combined. The solution was shaken for 30 s and centrifuged (830 g , 5 min). The water phase was removed, and the solution was washed with 0.5 mL of fresh water. The toluene phase was divided into a GC vial.

Whole blood samples

Whole blood samples from a specimen bank were used to analyze organic mercury by GC-ECD^[13]. Samples were chosen according to their MeHg concentration, from high to low. They were originally collected from Japanese adults in Kyoto Prefecture. Written informed consent was obtained from individuals before blood sampling. This analysis was approved by the Kyoto University Ethics Committee (approved no. R1478).

Instrumental conditions

Agilent 6890GC equipped with a DB-5MS (15 m long, 0.25 mm i.d., 0.10 μm film thickness) capillary column was used (Agilent Technologies, Santa Clara, CA). The injection volume was 1 μL, and the inlet temperature was maintained at 280 °C. Pulsed splitless injection (30 psi for 2 min) was used, and the vent line was opened at 2 min. Helium gas was used as the carrier gas at a flow rate of 1 mL/min. The oven temperature was maintained at 40 °C for 2 min and ramped at 5 °C/min to 130 °C, 2 °C/min to 150 °C, and 30 °C/min to 240 °C. The transfer line temperature was 280 °C. Mass spectra were obtained using Agilent 5973MSD. In the EI mode, the ion source temperature was 230 °C, the emission current was 35 μA, and the electron energy was 70 eV. In the NCI mode, the ion source temperature was 150 °C, the emission current was 50 μA, and the electron energy was 120 eV. Methane gas was used as the reagent gas at a flow rate of 2 mL/min.

Full-scan mass spectra from m/z 50 to 700 were used to analyze standard solutions. Selected ion monitoring (SIM) was chosen for quantitative and qualitative detection using NCI-MS. Specific ions of organic mercury bromides were monitored for quantification (m/z 281) and identification (m/z 283 and 279) in the NCI mode. Calibration curves of target analytes were made by spiking the standard mixture at 7 points (0.01 ng per sample to 10 ng per sample) with the internal standard (10 ng of TCA-d₃ per sample). The amount ratio was calculated from the response of organic mercury to that of the internal standard (10 ng TCA-d₃). Linear calibration curves were fitted with a weight of the inverse of concentration. Instrument detection limits were calculated from the relative standard deviation of 8 replicated extracts from standard solution multiplied by 3 at 0.03 ng/mL standard solutions. Blank experiments were conducted before and after sample analyses (n = 6), while no signals of organic mercury were detected. To evaluate the method detection limits in whole blood samples, whole blood of unexposed Wistar rat were used because human blood contains residual levels of methyl mercury more than the limits. Methyl and ethyl mercury concentrations in the rat blood were determined by the method of standard addition and were 0.1 ng/mL and < 0.12 ng/mL, respectively. The rat blood was spiked to 0.15 ng/mL MeHg and 0.15 ng/mL EtHg. Method detection limits were calculated from the relative standard deviation of 8 replicated samples and extracts from spiked rat blood samples multiplied by 3.

Derivatization of standard solutions

Tetraaryl borate salts were employed in this study because the carbon attached to borate is part of a nucleophilic group containing a polarized boron-carbon bond. In contrast, divalent mercury compounds are electrophilic and susceptible to transmetalation^[14] [Figure 1].

Figure 1. Schemes of the derivatization of mercury compounds with fluorinated tetraaryl borate and tetraphenyl borate salts.

For phenylmercury compounds, fragment ions were observed in the EI mode, whereas no fragment ions were observed in the NCI mode (data not shown) [Figure 2].

Figure 2. Extracted ion chromatograms of phenylated and bistrifluoromethyl phenylated organic and inorganic mercury. Ph: phenyl group; BTFMPh: 3,5-bis(trifluoromethyl) phenyl group. Mercury compounds in standard solutions were derivatized using sodium tetraphenylborate and sodium tetrakis [3,5-bis(trifluoromethyl) phenyl] borate separately and combined for this analysis. Ions were monitored in the scan mode of electron ionization.

For bistrifluoromethyl phenyl (BTFMPh) mercury compounds with a trifluoromethyl group detectable in the NCI mode, molecular ions were observed (m/z 406 for BTFMPhMeHg), whereas methyl and ethyl mercury fragments were undetectable [Figure 3]. In contrast, derivatization of inorganic mercury enabled the detection of a specific fragment ion with m/z 630, the molecular weight of BTFMPh₂Hg.

Figure 3. (A): Full-scan mass spectra of BTFMPhMeHg and (B): BTFMPhHg obtained using negative chemical ionization. Both analytes had the same concentration.

No derivative products were observed after the pentafluorophenylation reaction (data not shown). It is considered that the pentafluorophenyl ring is not reactive owing to the electron deficiency by substituted fluorine atoms^[15].

For mercury chloride, a fragment ion with m/z 237 derived from [HgCl]^- was observed, and the sensitivity was high [Figure 4]. MeHgBr also gave a fragment ion with m/z 281 derived from [HgBr]^-. However, MeHgI gave a base peak with m/z 127 from iodine. Among these three halides, MeHgBr is considered applicable to blood samples because it has the highest m/z and is moderately volatile. In contrast, MeHgCl is volatile and thus loss of the analyte during sample preparation can occur.

Figure 4. (A): Chromatogram of different halide salts of organic mercury and (B): mass spectra of MeHgCl, (C): MeHgBr, and (D): MeHgI. Organic mercuries in standard solutions were extracted using aqueous solutions containing chloride, bromide, and iodide ions. Spectra were obtained using negative chemical ionization.

The analytical procedure is divided into two steps: derivatization of inorganic mercury by BArF₂₄ and extraction of organic mercury as a bromide complex.

Stability of derivatized compounds

The stability of derivatized compounds during sample preparation was examined. In the extraction of mercury compounds, cysteine is often used as the reservoir of organic mercury in the water phase owing to strong coordination binding between the sulfhydryl group and mercury^[16]. To evaluate the effect of cysteine on the derivatization of inorganic mercury, a standard solution of inorganic mercury (10 μL of 10 μg/mL standard solution) was reacted with BArF₂₄ with or without cysteine. The cysteine-containing solution did not contain derivatives (data not shown). In addition, to confirm whether the derivatization was inhibited or whether the reaction with cysteine occurred, a cysteine solution was added to the derivatized standard solution without cysteine. The result showed the absence of derivatized mercury, and cysteine reacted with derivatized mercury because of the nucleophilicity of the sulfhydryl group. Removal of cysteine is required for the analysis of inorganic mercury, although the process has not been established. Analysis of inorganic mercury was not further conducted in this study. Finally, organic mercury was analyzed as a mercury halide complex.

Calibration curves and detection limits

Calibration curves were plotted for MeHgBr and EtHgBr. The linearity (R²) of the response ratio vs. concentration curves was 0.987 for MeHgBr and 0.990 for EtHgBr [Figure 5]. Although a higher response was obtained with a higher concentration, the linearity of the calibration curves between 0.01 ng and 1 ng per sample was 0.978 for MeHgBr and 0.979 for EtHgBr. The instrument detection limits were 0.05 ng and 0.06 ng per sample for MeHgBr and EtHgBr, respectively. The method detection limits in blood matrix were 0.06 ng MeHg per sample and 0.07 ng EtHg per sample. These values correspond to detection limits of 0.12 ng/mL (MeHgBr) and 0.14 ng/mL (EtHgBr) per 0.5 mL of blood samples and final extracts.

Figure 5. (A): Calibration curves for MeHgBr and (B): EtHgBr. Concentration of analytes and response ratios are based on 0.5 mL samples spiked with the internal standard (10 ng of TCA-d₃). Linear portions were fitted with a weight of the inverse of concentration. Linearity (R²) was 0.987 for MeHgBr and 0.990 for EtHgBr.

Sample analysis and validation against GC-ECD analysis

Recovery of organic mercury from whole blood samples with low mercury contamination was evaluated. The blood sample with the lowest mercury concentration (0.66 ng/mL MeHg) was used for this test. Whole blood was spiked at 20 ng/mL with the standard solution (MeHg and EtHg in benzene, 10 μg/mL). The spiked samples were analyzed using the same method for recovery evaluation. The sample signal was subtracted from the spiked sample signal. The recovery of MeHg was 67.1%, while the recovery of EtHg was 49.3% [Table 1]. This difference might be caused by the lipophilicity of the chemicals, and EtHg could be lost during the degreasing process.

The concentrations of MeHg in whole blood samples determined using GC-ECD were used for cross-checking [Figure 6]. The concentrations measured using GC-ECD were in the range of 3.4-32 ng/mL, and the concentrations obtained using GC-NCI-MS were in the range of 1.5-44 ng/mL. The results of the two methods had a correlation coefficient (r) of 0.923. The slope of the best-fit line was 0.80 ± 0.07. Therefore, GC-NCI-MS is applicable for the analysis of MeHg in whole blood samples.

Figure 6. Cross-check of the concentrations of methylmercury in whole blood samples determined using GC-NCI-MS with those determined using the GC-ECD method. n = 13, r = 0.923. Concentrations determined using GC-NCI-MS were corrected for average recovery rates.

Analysis of a CRM (cod fish tissue)

The concentration of MeHg in a CRM, namely cod fish tissue, was analyzed using GC-NCI-MS. The recovery of MeHg from fish samples was 59% (n = 3). The mean concentration of MeHg determined using our method was 0.64 μg/g, which agreed with the certified value of 0.58 μg/g [Table 2].

Comparison with existing studies

We aimed to develop a GC-NCI-MS method for the quantitative detection of mercury in blood samples. Zachariadis et al. quantified inorganic mercury and MeHg in human urine, saliva, and serum using GC-MS in the EI mode^[17]. They applied in situ derivatization in aqueous solutions, followed by headspace solid-phase microextraction. This method gave a recovery ranging from 85% to 98% for inorganic mercury and MeHg, and the detection limits for inorganic mercury and MeHg were 10 and 15 ng/L in urine, 54 and 60 ng/L in saliva, and 61 and 81 ng/L in serum, respectively. Watanabe et al. evaluated the performance of GC-EI-MS in the detection of MeHg in fish samples^[18]. They used sodium tetraphenylborate for derivatization and co-injected polyethylene glycol 200 to suppress the absorption of methyl phenyl mercury within the GC-MS instrument. This method was validated using CRMs and spiked samples, although the detection limit was not determined. Cavalheiro et al. compared GC-EI-MS and GC-ICP-MS for monitoring inorganic mercury and MeHg, as well as tin compounds in biological tissues^[19]. They used the isotope dilution method and sodium tetrapropylborate for derivatization. The detection limits using GC-EI-MS and GC-ICP-MS were, respectively, 0.079 pg and 0.005 pg for MeHg and 0.041 pg and 0.013 pg for inorganic mercury. Although GC-ICP-MS performs better than GC-MS, the former requires expensive instrumentation and maintenance. Hippler et al. compared a coupled GC-EI-MS/ICP-MS system and a cryo-focusing headspace-GC-MS system for the analysis of MeHg in whole blood samples^[20]. The limit of quantification was 0.3 μg/L for GC-EI-MS/ICP-MS and 0.5 μg/L for headspace-GC-MS, and the results obtained using the two methods were in good agreement. In general, GC-MS can offer excellent selectivity and good sensitivity with moderate cost and skill requirements. However, some of the extraction and purification methods provided by the above-mentioned studies are relatively complicated and time-consuming^[17,20]. In comparison, our study provides a simpler and lower-cost option for the determination of mercury in biological samples, although further optimization and investigation are needed considering some limitations described below.

Limitations

In this study, we analyzed only blood samples. Analysis of a CRM (cod fish tissue) was conducted, and the recovery from the CRM was slightly lower than that from blood. Although the obtained concentration of MeHg agreed with the certified value, our method needs to be improved to obtain a higher recovery rate, thus enhancing the accuracy and sensitivity. Most of the analytical methods used to determine the levels of mercury species achieve recoveries ranging from 80% to 120%, including the above-mentioned studies^[17-18,20]. For other sample matrices, further investigations are needed because the effect of different sample matrices may affect the analysis result^[17].

In addition, optimization of the sample preparation steps is required. This method could not be applied to the analysis of inorganic mercury, and other derivatizing reagents or purification methods to remove cysteine and similar interferents are required. In our extraction procedure, MIBK was used, a Group 2 substance in the Ordinance on Prevention of Hazards Due to Specified Chemical Substances. Safe organic solvents will be examined instead of MIBK in future studies.