MIT (purity >95%), DIT (purity >95%), rT3 (purity >95%), T4 (purity >95%), ¹³C₆-MIT (purity >98%), ¹³C₆-T3 (purity >98%), and ¹³C₆-T4 (purity >98%) were obtained from Toronto Research Chemicals (Downsview, ON, Canada). T3 (purity >97%) was obtained from Sigma Aldrich (Oakville, ON, Canada). LC-MS grade methanol and ethyl acetate (EtOAc) were obtained from Merck (Darmstadt, Germany). Ultrapure water was produced through the Milli-Q purification system (Millipore, Billerica, United States) at conductivity of 18.2 Ω·cm⁻¹. Sodium hydroxide and hydrochloric acid were purchased from Xilong Scientific Co., Ltd. (Chengdu, China).

Thirty pairs of sediment and water samples were collected in October 2019 from Taihu Lake of China. The sampling sites are shown in Figure 2. Sediment samples (upper 20–30 cm) were collected at the lake bottom with a stainless-steel grab sampler, and water samples (upper 50 cm) were collected with a stainless-steel vertical sampler. All water samples were collected in precleaned 4-L amber glass bottles. The samples were sealed without a headspace and transported to the laboratory within 6 h of collection. Water samples were filtered and extracted immediately, and sediment samples were stored at −20°C until further analyses.

Glass fiber pads (GF/F, 0.7 μm, Whatman, Maidstone, United Kingdom) were used for filtering water samples to remove suspended materials. The target thyroid hormones were extracted from water samples with Waters Oasis HLB extraction cartridges (500 mg, 6 cc). Prior to extraction, isotope-labelled standards (¹³C₆-MIT, ¹³C₆-T3, and ¹³C₆-T4) were added into water samples at concentrations of 100 ng/L as surrogates to compensate for losses and matrix effects, and then water samples were acidified to pH∼2.5. The HLB cartridges were preconditioned with 6 ml of EtOAc, 6 ml of methanol, and 6 ml of distilled water (pH 2.5), and then water samples were loaded at a flow rate of 5–10 ml/min. After rinsed with 10 ml of acidified water, the cartridges were dried and eluted with 6 ml of EtOAc. The extracts were evaporated under a gentle stream of nitrogen to dryness and reconstituted with 200 μl of methanol for UPLC-ESI-MS/MS analysis.

For sediment samples, they were freeze-dried and grinded to pass through 100 mesh sieve. Prior to extraction, sediment samples were spiked with 20 ng of each surrogates (¹³C₆-MIT, ¹³C₆-T3, and ¹³C₆-T4) and were ultrasonically extracted with 20 ml of mixture of methanol and EtOAc (1:1, v:v). The eluent was evaporated to dryness and reconstituted with 200 ml of distilled water. The subsequent purified steps followed the same procedure as for the water samples described above.

The LC apparatus was an ACQUITY Ultra Performance LC system (Waters, Milford, MA, United States). Separation was carried out on a Waters BEH C18 column (100 mm × 2.1 mm, 1.7 μm particle size). The column was maintained at 35°C, and the flow rate and injection volume were 0.3 ml/min and 2 μl, respectively. Aqueous (0.1%) acetic acid (A) and 0.1% acetic acid in methanol (B) were used as the mobile phases. The initial gradient of 10% B was held for 3 min, and then increased to 50% at 6 min, 90% at 10 min, and then to 100% at 10.5 min. After that the column was washed with 100% B for 2 min, then the column was re-equilibrated with the initial mobile phase composition for 4 min before the next injection.

Mass spectrometry was performed using a Waters Xevo TQ-XS Triple Quadrupole Mass Spectrometer equipped with an electrospray ionization source in positive ion mode. A flow injection of a standard solution of each compound was used to find the optimum conditions in the ESI source and the average parameters selected were: source temperature 150°C; desolvation temperature 500°C; capillary voltage 3.2 kV; desolvation gas flow 1000 L/h and cone gas flow 150 L/h. Quantitative analysis of the target thyroid hormones was performed in multi-selected reaction monitoring (MRM). In the MRM transitions, the dwell times were automatically selected. Data acquisition was performed with MassLynx 4.2 (Micromass, Manchester, United Kingdom). The MRM transitions, cone and collision energy of target thyroid hormones were listed in Table 1.

The five thyroid hormones were simultaneously analyzed by MS/MS in the MRM mode. The two abundant MRM transitions, cone voltage and collision energies, were optimized for each analyte by infusing the standard solutions into the mass spectrometer (Table 1). All of the precursor ions were protonated molecular ions [(M + H)⁺]. Under MS/MS conditions, the protonated molecular ions lost water and carbon monoxide, corresponding to a mass loss of 46 Da, yielding [M + H-H₂O-CO]⁺ ions for all target thyroid hormones. At high collision energy, T4, T3 and rT3 showed losses of 127 Da, corresponding to the loss of HI. No loss of HI was observed in the spectrum of MIT and DIT, suggesting that the losses of HI in T4, T3 and rT3 spectrums were more likely to occur on the phenolic rings. For MIT, the other abundant product ion was the result of loss of water from the protonated molecule at low collision energy. In the case of DIT, the other abundant ion at m/z 289.9 was produced by the cleavage of benzene ring. Chromatographic separation was performed on a BEH C18 column. Methanol and the addition of acetic acid in the mobile phases increased the retention of target thyroid hormones to the column, improving the chromatographic behavior of the analytes (Figure 3).

This study used the isotope dilution method for quantifying the target analytes. The use of such method allowed to correct the losses of analytes during sample pretreatment, and to compensate for matrix effects from different samples and variations in instrumental signals between injections. The recovery experiments were conducted using pooled water and sediment samples from Taihu Lake. Analyte addition was made with the criterion of at least three times the original concentration that was determined prior to the fortification experiment. As shown in Table 2, the recoveries of all target analytes were 84–102% in sediment and 87–101% in water. The relative standard deviation (RSD) values were less than 19% in all of the spiked samples. The extent of signal suppression and enhancement in the UPLC-ESI/MS/MS detection was evaluated by spiking extracts of the pooled water and sediment samples with a standard solution. The matrix effect observed with each analyte was calculated using the percentage signal intensity in the sample matrix versus the signal for the same concentration in methanol. The response ratios for all target analytes in pooled water and sediment samples were 25–30% and 40–51%, respectively (Table 2). Based on a signal-to-noise ratio of 10, the limit of quantifications (LOQs) for target thyroid hormones in water and sediment samples were 0.08–0.1 ng/L and 0.008–0.01 ng/g dw, respectively. To our knowledge, only one study has previously reported the detection of T4 and T3 in surface water, and the LOQs of T4 (0.6 ng/L), T3 (0.9 ng/L) and rT3 (3.2 ng/L) (Svanfelt et al., 2010) are lower than the present study.

As shown in Figure 2 and Table 3, MIT and DIT were detected in all sediment and water samples. MIT was detected with 0.01–1.1 ng/g dw in sediment and 1.9–9.9 ng/L in water, and DIT was 0.06–0.33 ng/g dw and 4.1–9.6 ng/L, respectively. T4 was only observed in sediment, with the measured concentrations and frequency of <LOQ-0.07 ng/L and 87%, respectively. T3 and rT3 were not found in this study. To our knowledge, the present study is the first one that reports such a contamination of surface water by MIT and DIT which are important for sustaining thyroid hormone synthesis in organisms and humans. One study reported 0.8–1.7 ng/ml of urinary MIT and DIT in euthyroid people, but much higher of MIT (316 ng/ml) and DIT (129 ng/ml) were measured in DEHAL1 deficient humans. It has been reported that some algae species can form MIT and DIT which are probably bound to proteins (Shah et al., 2005). Thus, in addition to human/animal excretion, algae may be a potential source for MIT and DIT contamination. It requires further research on whether algae species from Taihu Lake are able to metabolize MIT and DIT. Besides, MIT and DIT are the characteristic hydrolysis products of cascoiodine (Wang et al., 2008), which is widely used in aquaculture and livestock and poultry breeding to regulate the growth and development of animals (Yu, 2004), and thus the addition of cascoiodine in the feed maybe another source of MIT and DIT found in Taihu Lake. In the present study, T4 was found in sediment but not water. This is on one hand due to the different sediment/water partition [the log Kow of T4: 4.12 (Li et al., 2014)]. On the other hand, combination of T4 with protein compounds would probably make it prone to sorption onto sediment. It has been reported that the mean daily urinary excretion of free T4 and T3 by healthy humans are 1.4 and 0.63 μg (Faber et al., 1981), respectively, while that for fecal T4 (probably in the sorbed form to proteins) amounts to 20 μg (Svanfelt et al., 2010). Protein compounds not only exist in organisms but also are ubiquitous in aquatic environment. Xia et al. (2013) reported that both bovine albumin from animal and soy peptone from plant could combine with perfluoroalkyl substances and affect their bioaccumulation by Daphnia magna in water. Thus, the protein compounds in sediment from Taihu Lake possibly help T4 sorb onto sediment, which needs further researches. T3 is naturally less produced compared with T4 (Faber et al., 1981). T4 is generated by coupling two DIT residues while T3 is generated by coupling one DIT with one MIT in organisms. But thyroid peroxidase is much more efficient at combining of two DIT residues and thus generation of T4 occurs much more readily, explaining why T4 excretion is much higher than that of T3. Also, T4 is a widely used medicine with more than 400 different formulations worldwide, but synthetic T3 is relatively less applied (Brown and Wong, 2017). These can partly explain the “non-detection” of T3 in this study.

The concentration ratios of MIT/DIT in water and sediment samples from Taihu Lake are given in Figure 4. It is interesting to find that the MIT/DIT ratios in sediment (2.4 ± 0.92) is significantly higher than those in water (0.84 ± 0.18) (p < 0.001), and the different sorption behavior between them should be the major reason. The pH, clay content, and organic C content in the environment can affect their sorption onto suspended particles and sediment. In this study, MIT and DIT would be in neutral or anionic forms in these samples at the measured pH values of 7.59–9.26, indicating that sorption to organic C could be the dominant sorption mechanism. Similar phenomenon has been reported for the sorption of hydrophilic sulfonamide antibiotics in freshwater-sediment microcosms (Carstens et al., 2013). Furthermore, as one of important components of organic C, protein compounds readily combined with DIT compared to MIT, which possibly increased sorption and binding of DIT residues to sediment by producing nonextractable sediment-bound residue or transformation products (such as T4). Figure 5 showed a significant correlation between the MIT/DIT ratios and DOC values in water samples (R ² = 0.3684, p < 0.001), also indicating that sorption to organic C could be of important mechanism for their sorption to the solid phases.

As shown in Figure 2, the total concentrations of detected thyroid hormones were 6.2–18 ng/L in water and 0.15–1.5 ng/g dw in sediment. The total concentrations of detected thyroid hormones in the eastern area (sites 22, 23 and 24) were relatively higher than those in other areas of Taihu Lake. This was probably due to the usage of cascoiodine in the feedstuff in the aquaculture and livestock and poultry breeding in the eastern area which used to have lots of crab, fish, livestock and poultry farms. Cascoiodine is widely used as synthetic thyroxine additives to regulate the growth and development of animals (Yu, 2004), and it can be hydrolyzed into MIT, DIT, and T4 under alkaline conditions (Wang et al., 2007). In comparison, relatively low concentrations of thyroid hormones were found in the northwestern area of Taihu Lake (including Zhushan Bay, Meiliang Bay, and Gonghu Bay) where relatively heavy contamination has been reported (Zhao et al., 2019). Thus, MIT and DIT may be a possible indicator of wastewater input from aquaculture and livestock and poultry breeding, and deserve more future study.