Reference standards of emodin and puerarin (internal standard, IS) were purchased from National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China). For quantitative analysis, stock solutions of emodin and IS were prepared in methanol at a concentration of 88.4 and 66.5 μg/mL, respectively. For the bioconcentration study, a stock solution of emodin was prepared in DMSO at a concentration of 2 mg/mL. All stock solutions were stored at 4°C and diluted daily to working solutions for experiments. HPLC grade methanol was purchased from SK Chemicals (Ulsan, Korea). Deionized water with a resistivity of 18.2 MΩ was produced by a Milli-Q water purification unit (Barnstead International Inc., Dubuque, Iowa, USA).

HPLC was performed on a Shimadzu LC-20AD series HPLC system (Kyoto, Japan) consisting of two LC-20AD pumps, a CTO-20A column oven (set at 35°C) and a SIL-20A autosampler. Analyte separation was achieved on a Phenomenex Luna C₁₈ column (150 × 2.00 mm, 5 μm) preceded by a Phenomenex C₁₈ security guard column, with an isocratic mobile phase consisting of 0.1% formic acid and methanol (20:80, v/v) at a flow rate of 0.3 mL/min.

MS detection was performed on a Thermo Finnigan TSQ quantum mass spectrometer (San Jose, CA, USA). Ultra-high pure nitrogen was used as the sheath and auxiliary gas. The ESI source was operated in negative ion mode under the following conditions: spray voltage, 3.5 kV; capillary temperature, 400°C; sheath gas, 35 arbitrary units; auxiliary gas, 15 arbitrary units. For quantification in multiple-reaction monitoring (MRM) cone voltage, collision energy and precursor to production transition m/z for emodin and IS were as shown in Table 1. For metabolism study of emodin, three phase II metabolites of emodin were monitored simultaneously using MRM mode (Teng et al., 2007).

Our method provided recoveries of 87.0–90.5% range for eleutheroembryos suspension and 97.7–101.4% for medium, respectively. No matrix effect was observed, which were assessed by comparing the responses achieved from post-extraction blank sample and mobile phase spiked with the analyte. The inaccuracies ranged from −3.4–2.9% to −2.3–4.6% for eleutheroembryos suspension and medium, respectively. The precisions represented as RSD were not more than 8.6 and 8.8% for eleutheroembryos suspension and medium, respectively. Calibration curves were plotted from 5 to 7 concentration points (0.5–150 ng/mL for eleutheroembryos suspension and 0.25–50 ng/mL for medium). The instrumental limits of detection (LODs), calculated as a signal to noise ratio of three, were 0.2 ng/mL for eleutheroembryos suspension and 0.1 ng/mL for the exposure medium.

Zebrafish eleutheroembryo is a powerful tool applied to various aspects of toxicity studies such as acute toxicity, teratogenicity, toxicokinetecs, toxicodynamics, adverse effects, and long-term effects (Scholz et al., 2008). A number of studies have provided evidences that eleutheroembryo as an alternative to adult fish can greatly be applied to bioconcentration study (Wiegand et al., 1999; Schreiber et al., 2009; El-Amrani et al., 2012; Tu et al., 2014). Compared with adult fish, one of the major characters of eleutheroembryos is that eleutheroembryos develop depending on its own yolk consumption instead of external feeding (Halder et al., 2010). Generally, it is considered that the eleutheroembryos cannot feed freely until 139.5 hpf at 26°C breeding temperature (Straehle et al., 2012). According to the definition, the process of the bioconcentration comprises breath and direct contact but the uptake in the diet is not included (Arnot and Gobas, 2006); i.e., only the passive diffusion should be taken into account. Considering the character of eleutheroembryos, the use of eleutheroembryos can avoid the feeding-induced extra intake of test chemicals in long-term exposure test. Moreover, the poor digestive system of eleutheroembryos ensure that the excretion slightly affects the depuration of test chemicals. Therefore, the bioconcentration in eleutheroembryos can be completely considered as a passive process, which greatly conforms the definition of bioconcentration. Our study also confirmed the applicability of zebrafish eleutheroembryos on bioconcentration study.

In our study, the results of bioconcentration process of emodin in zebrafish eleutheroembryos are shown in Figure 1 for the standard test, and Figure 2 for the modified test. In both tests, emodin obviously accumulated in the bodies of the eleutheroembryos.

In the standard test, the toxicokinetics of different concentration groups were similar. The concentrations of emodin in exposure media (hereinafter referred to as C_w) were considered to be constant (the variations of concentrations were below 10%) with the values (mean ± SD) of 14.2 ± 0.5 and 8.13 ± 0.70 ng/mL for high and low concentration groups respectively. A slight decrease of the analyte concentration was observed, which was likely due to the metabolism and possibly some degree of adherence on the walls of the tanks. The accumulation profiles of emodin in zebrafish eleutheroembryos are shown in Figures 1A,C. The concentrations of emodin in zebrafish eleutheroembryos (hereinafter referred to as C_f) increased gradually with the exposure time and then reached plateau at 24 h, which suggested that the bioconcentration process reached steady state after only 24 h of exposure. The concentrations at plateau in the high concentration group were nearly double of that in the low concentration group. In the depuration phase, the zebrafish lavae were transferred to blank medium to eliminate the residual emodin, whose concentrations were decreased by about 50% within 24 h.

In the modified test, the toxicokinetic process was remarkably different from the standard test. As the exposure media were not renewed, C_w were not constant and decreased by over 90% within 72 h, with the initial concentrations of 14.2 ± 1.42 and 7.66 ± 0.92 ng/mL for two concentration groups, respectively. No obvious variations of concentration were observed in control groups, which showed that emodin was stable under the test environment. In other words, the physical factors, chemical factors and the biodegradation slightly affected the C_w and could be ignored. On the other hand, since the mass of emodin in the eleutheroembryos' bodies was below 3% of total mass of emodin in the tanks, we thought that the influence of the bioconcentration on C_w could be ignored. Meanwhile, we concluded that the decrease of C_w was mainly due to the metabolism. In the bodies of eleutheroembryos, C_f increased gradually at first with peak concentrations obtained at 12 h of exposure, and then decreased continuously with the decline of C_w. The real-time BCFs, calculated by the ratios of C_f and C_w, increased with time and reached plateau after 24 h of exposure, suggesting that the distribution had reached steady state.

The toxicokinetic parameters are shown in Table 2. The symbols k₁ and k₂ designate the uptake and depuration rate constants, respectively, obtained from Equations (2), (7). k₁ and k_2(bioc) were determined in exposure process, and k_2(dep) was from depuration phase in the standard test. The dissipation rate constants (λ) of emodin in media were obtained from Equation (8). BCF values were obtained from Equations (3), (9). BCF_ss values were the experimental values at steady state, which were calculated as the ratio between C_f and C_w at the maximum time of exposure time (48 h for the standard test and 72 h for the modified test). BCF_k values were obtained from parameters k₁, k₂ and λ, which showed that BCF values in the standard test were lower than those in the modified test, due to the apparent metabolism of emodin.

In previous studies, several quantitative structure-activity relationship (QSAR) models have been established to calculate the BCF (cBCF) of organic compounds. These QSAR models showed that the BCF values of the chemicals can be correlated with n-octanal/water partition coefficients (K_ow) of the chemicals. A classical equation has been established for the BCF_L of lipophilic substances in zebrafish eleutheroembryos (Petersen and Kristensen, 1998): (12)log BCFL=0.64+0.92 log Kow Herein, for ionizable chemicals, apparent oil-water partition coefficients (D_ow) rather than K_ow are used (Ding et al., 2016). The D_ow of chemicals change with pH, which also called pH-corrected K_ow. In this study, the pH of medium was 7 and did not change with the concentration of emodin. To compare with the estimated values, the BCF values obtained in this study were standardized to wet-weight BCF (sBCF), by transferring the dry-weight concentration to wet-weight concentration. sBCF represents the real accumulation level of chemicals in alive organism rather than in dry organism. Generally, for zebrafish eleutheroembryos, sBCF is one fifth of dry-weight BCF.

The metabolites of emodin were monitored by MRM mode. The previous study has proved that phase II metabolism is the major way of elimination of emodin in vivo. In our work, three classical phase II metabolites, mono-sulfate (M1), mono-glucosylate (M2), and mono-glucuronide (M3) were chosen for this study. Both M1 and M2 are common metabolites of emodin in human (Liu et al., 2012), while M3 is a specific metabolite existing in zebrafish eleutheroembryos (Hu et al., 2012). In zebrafish eleutheroembryos, the metabolites of emodin accumulated with increased exposure time in all tests (Figure 3). In the uptake phase of the standard test, the concentrations of metabolites increased by a linear trend, which indicated that the formation rates of the metabolites could be considered as approximately constant. In the depuration phase of the standard test, the metabolites concentration decreased with time but more than half of the metabolites remained in the body after 24-h depuration. Under the microscope, it was observed that emodin and its metabolites were accumulated mainly in the intestines of eleutheroembryos (Supplementary Material 1). At 72 h of test time, the color of the intestinal contents was still deep, while it became light after an extra 24-h depuration. In the modified test, the accumulation rates of metabolites were fast at first and turned slow subsequently. These results showed that emodin could be highly metabolized in vivo, which is in agreement with the previous studies performed in mammals (Teng et al., 2007; Liu et al., 2012). No metabolite was detected in the media probably due to their low concentrations that were below the low limit of detection. More interestingly, di-conjugates were also detected in zebrafish eleutheroembryos, which maybe contributed by the fact that two metabolic reactions (any two of the glucuronidation, glucosylation, and sulfation processes) can be induced by the active phenolic hydroxyl groups in emodin (Hu et al., 2012). These results further implicated that emodin can be highly metabolized. Since the mono-conjugate is the main form of metabolites of emodin in the human body (Liu et al., 2012), no further study on di-conjugates was performed in our work.

The bioconcentration processes of the two tests are illustrated in Figure 4. These processes comprised uptake (k_in), elimination (k_out) and metabolism (k_m), where k_in is equal to k₁, and the sum of k_out and k_m is equal to k₂.

The standard test is a bioconcentration experiment preformed at constant exposure concentration, which can be regarded as a long-term exposure model or a repeated exposure model. Therefore, the process of the standard test is simple and we just need to focus on the emodin concentration in zebrafish eleutheroembryos. In this model, k₂ is considered as the k_out with the k_m assumed as zero. For the calculation of BCF of emodin, since the high metabolism was not included, k_out was overestimated, leading the BCF value being lower than the actual value.

The modified test is a bioconcentration experiment preformed at a certain initial exposure concentration, which can be considered as a single exposure model. The relations among k₁, k₂, k_in, k_out, and k_m are the same as those in the standard test. In this study, during 0–12 h of the exposure time, the concentration process was the same as the standard test, thus the concentration-time profiles for zebrafish eleutheroembryos were similar to those of the standard test. Only at 12 h, C_f reached a peak value, and the uptake rate was equal to the sum of elimination rate and metabolism rate (Equation 13). (13)Vin=Vout + Vm → kintCw=koutCf + kmCf, only at 12 h Since k_m was not zero, the uptake rate (V_in) was higher than the elimination rate (V_out). In the other word, for the exposure medium, the amount of the lost emodin was more than that of the reuptake of emodin, leading the C_w decreasing continually. The decrease of C_w caused the uptake rate (k_intC_w) to decline, further causing the decrease of C_f at 12–72 h of exposure time. Within 24–72 h of exposure time, a steady state was reached with the BCF tending to be constant (Figures 2A4,B4). During this period, the natural logarithmic concentration-time curve of C_f of the modified test was approximately linear (Figures 2A1,B1), which indicated that the internal concentration decreased with an unchanged rate constant. Here, we defined this period as “depuration period” of the modified test, with a depuration rate constant k₃ (1/h). The k₃ could be obtained by linear fitting the Equation (14), and its values was the slope of this equation. (14)LnCf=-k3t+b During the depuration period, the process of C_f followed Equation (15), derived from Equations (4), (5): (15)dCfdt=-k3Cf=kintCw-koutCf-kmCf From Equations (4), (9), the following equation could be obtained: (16)BCF=kintkout+km-λ=CfCw From Equations (15), (16), Equation (17) could be obtained: (17)k3=λ Equation (17) further supports our claim that the decrease of C_f was directly attributed to the decrease of the exposure concentration. However, the observed values of k₃ at 0.0251 and 0.0284 in the high and low concentration groups respectively were obviously lower than the values of λ, which were probably due to the intestinal accumulation of emodin. When the decrease of C_w contributed to a decline in C_f, the intestinal emodin was released into the whole body, providing a buffer for slowing down the decrease rate of C_f. This hypothesis can be confirmed by the fact that the values of k₂ in depuration phase was lower than that in uptake phase in the standard test (Table 2). In this case, the bioconcentration process of emodin may be fitted better with the two-compartments model.

The decrease of C_w was from uptake and elimination, expressed as Equation (18). It is important to note that as the uptake rate and elimination rate are the variables for C_f, a conversion factor R_f−w (g/L), the mass of fish per unit volume of water, should be introduced into the equation. In our work, the average of the dry weight of one fish was 0.0609 ± 0.0076 mg. With the fish-to-water ratio kept at 1:1 (fish: mL), the R_f−w was calculated to be 0.0609 g/L. (18)-λCw=Rf-w(koutCf-kinCw) From Equations (15–18), Equation (19) could be derived. It is shown that the value of k_m can be obtained from λ, R_f−w and BCF. In other words, to calculate the k_m, it was crucial to keep the R_f−w (or the fish-to-water ratio) constant in the whole test. Since k_m, like k_in and k_out, is a characteristic constant only related to the chemical reactivity of the substance and the animal species, the ratio of k_m–k₂ should be a fixed value. From Equation (19), values of k_m were calculated with the values of 0.0455 and 0.0570, which was 29.5 and 40.7% of the values of k₂, for high and low concentration group respectively. Finally, it is concluded that the metabolism constituted 29.5–40.7% of the depuration process of emodin in zebrafish eleutheroembryos. (19)km=λ·(1+1Rf-w·BCF) In the standard test the underestimated BCF values can be corrected by subtracting the k_m from k₂ and the corrected BCF values were 169.6 and 168.6 (with the sBCF of 33.9 and 33.7) for high and low exposure level, respectively. In this study, k_m was approximately equal to λ because of very small value of 1Rf-w·BCF. Based on Equation (9), the BCF in the modified test could be approximately considered as the corrected BCF. Nevertheless, with the intestinal accumulation, the values of BCF in the modified test were high than the corrected BCF.

About all, the modified test can be used for the calculation for the km. However, it is a special process only fitted to the chemicals which are easily biotransformed. For those depurated mainly with the unchanged form, only the standard method is suitable to study their BCF. Moreover, the form of the Equation (19) is variable and may be changed if the test chemicals are unstable in the medium. In these cases, more factors should be concerned (including physical, chemical and biological factors), but the method of derivation is similar as above.

In the previous study, it was showed that emodin has very poor oral bioavailability owing to its extensive glucuronidation in intestine and liver. In the gut, emodin is absorbed in the intestinal cells by passive diffusion and metabolized to phase II metabolites (Liu et al., 2012). It seems that the accumulation and the biotransformation of emodin in intestinal cells is similar to that in zebrafish eleutheroembryos. Although the half-life of emodin is not long (5.4 h in rats, Huang et al., 2012), emodin still can be locally accumulated in the intestinal cells by repeated intake, as the intestinal wall directly contacts to the ingested food. Since emodin is highly metabolized, its single dose cannot cause accumulation, as the process of the modified test. However, when the emodin used for a long time (for weight loss or chronic constipation), the intracellular concentration of emodin could be accumulate to a high level continually (dozens of times higher than the concentration in lumen), as the process of the standard test. Although, the emodin could be accumulated in the whole gut with it passing the intestine, the accumulation most likely occurred in the colon instead of other parts of the digestive tract, as the in-take food stays in colon for most of the time. Accordingly, as the alimentary canal toxicity of anthraquinones only occurred in the colon, it could be inferred the potential accumulation of anthraquinones may be a risk factor concerned with MC process. On the other hand, as the lipid content of zebrafish eleutheroembryos and intestinal cells are similar (Petersen and Kristensen, 1998; Trapp and Horobin, 2005), the accumulation level of substances in these two organisms are similar. In other words, the extent of accumulation of chemicals in intestinal cells can be estimated by the data from zebrafish eleutheroembryos. Since the estimated value with QSAR model was similar with the observed values in our study, it can be considered that this QSAR model can be applied to other anthraquinones. With the value of pH of colon lumen was about 6–7 (Nugent et al., 2001), the estimated BCF of typical anthraquinones are listed in Table 3, which shows that some kinds of anthraquinones have high BCF values. These illustrated that the emodin as well as other anthraquinones could highly accumulate in the colonic cells.

Moreover, the effect of intestinal flora should not be ignored. It was reported that the metabolites of emodin and combined emodin can be microbiologically converted to free emodin, and only the free emodin can be absorbed by intestinal cells (Moreau et al., 1985). When taking herbal laxative, the combined anthraquinones cannot be absorbed, owing to their high molecular weight and polarity, and finally reach to the colon (van Gorkom et al., 1999). In the colon, their aglycones can be released by beta-glucoside enzyme from resident microflora, and subsequently the free anthraquinones can be absorbed and function on the colon, which works synergistically with the bioconcentration effect, leading to a further increase of the intracellular concentration of anthraquinones.

It is considered that the apoptosis of colonic epithelial cells is the main mechanism of the anthraquinone-induced MC (Chen et al., 2011). As the targets of apoptosis activity of emodin are inside the cells (Dong et al., 2016), its toxicity is affected by the intracellular concentration rather than the exposure concentration. In the standard test, the test media were renewed every 12 h, which was similar as the repeated administration, while the modified test was similar as the single administration. Comparing the results of these two tests, it could be inferred that the long-term use of anthraquinones could make their concentrations in the intestinal cells remain at high level, while short-term use could not make it. The accumulation highly increases the risk of long-term use of anthraquinones, which may be a possible reason why only long-term use (instead of short-term use or a single high dose) of anthraquinones can cause MC. In conclusion, the accumulation of emodin as well as anthraquinones is a non-negligible risk factor of the MC process.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.