Introduction

Drug safety screening plays an important role in early drug discovery. The screening strategies involve multiple disciplines including toxicogenomics, toxicoproteomics, and metabonomics in both in vitro and in vivo systems [1, 2]. Liver is the major metabolizing organ in the body, and reactive intermediates may be generated upon metabolism of drug candidates [3]. Therefore, screening compounds with their hepatotoxicity profiles are most relevant in triggering the safety evaluation in the early phase of drug discovery.

Reduced glutathione (GSH, γ-l-glutamyl-l-cysteinylglycine) contains a thiol group which can function as a free radical scavenger [4]. By reacting with reactive oxygen species such as free radicals or peroxides, GSH serves as an electron donor and is converted to its oxidized form, GSSG. GSSG can be reduced back to GSH by glutathione reductase in the presence of reduced nicotinamide adenine dinucleotide phosphate (NADPH) [5–7]. The GSH and GSSG pair is one of the major intracellular redox regulating couples. The quantitative determination of the GSSG/GSH ratios and the individual concentrations of GSH and GSSG have been used to indicate oxidative stress in cells and tissues [8–12]. Yet, there is a knowledge gap to make the GSSG/GSH pair a practical biliary index for screening hepatic oxidative stress in early drug discovery. For example, the kinetics and magnitudes of biliary GSSG or GSH changes upon toxin/drug-induced hepatotoxicity are not clear. This paper is intended to provide such in vivo data with selected compounds and dosages that are known to cause hepatotoxicity.

Since GSH and GSSG lack strong chromophores and fluorophores, analytical methods for determining GSH and GSSG typically involve extensive sample preparation such as derivatization of GSH or GSSG to facilitate their detection [13–16]. The auto-oxidation of GSH to GSSG adds to the complexity of the analysis of GSH in bile; for an example, GSH was reported to be rapidly auto-oxidized to GSSG upon exposure to air (t 1/2 = 5 min) [17]. Derivatization of GSH has therefore also been used to prevent cysteine oxidation in GSH [15, 18, 19]. An alternative strategy to stabilize biliary GSH and prevent its conversion to GSSG is to acidify bile samples upon sample collection [17, 20–22].

In this study, we adopted the method of acidifying bile samples to stabilize biliary GSH upon collection. There were no derivatization steps, such that quantification errors due to sample manipulation were minimized. Bile samples from bile-duct-cannulated Sprague Dawley rats after administration of acetaminophen or diquat dibromide monohydrate were quantitated for GSH and GSSG. Bile samples from gallbladders of the “unmodified mice” (mice which had not been surgically modified) after administration of paraquat or tert-butyl hydroperoxide (TBHP) were also profiled based on their biliary GSH, GSSG, and the GSSG/GSH ratios. The time profiles of GSSG/GSH ratios post-dose were obtained as a potential biomarker to estimate hepatic oxidative stress with drug-induced oxidative stress in the rat and mouse.

In addition, the high sensitivity of the LC–MS/MS (Liquid chromatography coupled to tandem mass spectrometry) detection allowed for direct measurement of both biliary GSH and GSSG in a small volume of individual mouse samples. Although mice are routinely used for pharmacological studies, only a very small amount of bile can be obtained from individual gallbladders. On the other hand, the complexity of the bile duct catheterization surgery for mice is likely the reason for the limited work on assays of biliary GSH and GSSG in mice. This study provides examples of direct measurement of biliary GSH and GSSG as potential indexes to the drug-induced hepatic oxidative stress in mice without bile duct catheterization surgeries, which may greatly benefit drug discovery programs.

Methods

Chemicals and reagents

Acetaminophen (≥99.0 % pure), diquat dibromide monohydrate (PESTANAL®), paraquat (PESTANAL®), TBHP (over a 4-Å molecular sieve ), a dipeptide of Glu-Glu, potassium chloride (≥99.0 % pure), and taurocholic acid (≥95.0 % pure) were purchased from Sigma-Aldrich (St. Louis, MO). All solvents used for HPLC and LC/MS were of chromatographic grade (Burdick and Jackson, Muskegon, MI). Fresh bile from male Sprague Dawley rats and fresh gallbladders from male CD-1 mice were obtained within the facility. The acidification solution contained 0.9 μM Glu-Glu in H2O with 0.1 % formic acid.

Biliary excretion studies in rats

All animal procedures were conducted under protocols approved by the Amgen (Cambridge) Institutional Animal Care and Use Committee. Male Sprague Dawley rats weighing 300–350 g (Charles River Laboratories, Wilmington, MA) had silicon catheters implanted in the bile duct and proximal duodenum using aseptic surgical techniques; the exteriorized biliary and duodenal catheters were connected to permit recirculation of bile. On the first day post-surgery, the catheters were disconnected, and the rats received an infusion of artificial bile (0.05 % potassium chloride, 1.35 % taurocholic acid in 0.9 % saline, pH = 8) at 1 mL/h via the duodenal catheter that continued throughout the study. On the second day post-surgery, the rats were administered a single dose of diquat at 31 mg/kg in 0.9 % saline intraperitoneal (IP) or acetaminophen at 1,500 mg/kg in Ora-Plus® oral (PO) or a sham dose of 50 % DMSO in saline intravenous (IV). Bile samples were collected for 30 min at specified time points which started at 30 min before dose, 0 min, 30 min, 1 h, 2 h, 4 h, 8 h, and 24 h post-dose into pre-weighed bottles containing 9 mL of the acidification solution (“Chemicals and reagents”)

Biliary excretion studies in mice

Male CD-1 mice weighing 38 g on average were administrated a single dose of paraquat at 130 mg/kg in 0.9 % saline (subcutaneously) or TBHP at 135 mg/kg in Ora-Plus® (IP). For each compound, three mice were sacrificed to collect gallbladders at 30 min before dose (−30 min), 15 min, 30 min, 45 min (TBHP only), 1 h, 2 h, and 4 h post-dose. The collected gallbladders were placed in 100 μL of the acidification solution and frozen at −80 °C. The mice were not given anesthesia or any surgery; thus, they are referred to as “unmodified mice” in this paper.

Analytical methods

Preparation of rat bile samples

The volume of bile was determined gravimetrically, assuming a density of 1 g/mL. Each bile sample was diluted with 9 mL of the acidification solution upon collection and was further adjusted by dilution with acidified water to exactly 1,000-fold of the original concentration, followed with centrifugation at 3,750 rpm for 5 min and analysis of LC/MS/MS. The biliary GSH and GSSG levels in rats were normalized by both bile flow rates and animal weights and expressed as nanomole per minute per kilogram to allow for direct comparison of previously published values.

Preparation of mouse bile samples

Gallbladders were collected from male CD-1 mice and placed into pre-weighed vials filled with 100 μL of the acidification solution. The filled vials were measured gravimetrically again after the collection of gallbladders. The weights of the whole gallbladders including bile were calculated by subtracting the pre-weights from the post-weights. Additional acidification solution was added according to the resulting weights to generate exact 1,000-fold dilution of the original bile concentrations (10,000-fold dilution was applied for some samples due to concentrations above the quantitation limit of the standard curve). The gallbladders were then carefully minced with needles, to fully release the content into the buffer, followed by centrifugation. The supernatant was taken out for LC/MS/MS analysis.

Preparation of the standard curves

The standards and quality controls for the rat samples were prepared by serial dilution of the standard GSH or GSSG in the artificial bile and diluted with the acidification solution. The artificial bile consisted of 0.05 % potassium chloride and 1.35 % taurocholic acid in 0.9 % saline (pH = 8) to mimic high concentrations of biliary salts. The dipeptide Glu-Glu was used as the internal standard throughout. For the mouse samples, the standard curves and quality controls were prepared in artificial bile which was diluted tenfold with the acidification solution. Duplicate intra-day runs for both standards and quality controls were performed. The standards were accepted within ±20 % from the nominal values, and the quality controls were observed within ±20 % from the nominal values.

Chromatography and mass spectrometry conditions

The method utilized a ultraperformance liquid chromatography (UPLC) (Acquity™, Waters, Milford, MA) coupled to a triple quadrupole mass spectrometer with a turbo spray ion source (Sciex APCI 4000™, Applied Biosystems, Foster City, CA) for efficient separation of GSH and GSSG. Detection of GSH, GSSG, and the internal standard Glu-Glu was performed by multiple reaction monitoring (MRM) with transition(s), namely m/z 308.2 → 233.06, m/z 308.2 → 179.1, and m/z 308.2 → 162.08 for GSH; m/z 613.2 → 484.1 and m/z 613.2 → 355.3 for GSSG; and m/z 277.1 → 259.0 for Glu-Glu, respectively.

The separation was conducted on a UPLC reversed-phase HSS T3 analytical column (1.7 μm, 2.1 × 50 mm) with water containing 0.1 % formic acid as the mobile phase A and acetonitrile containing 0.1 % formic acid as the mobile phase B. A multistep gradient was applied with a flow rate of 0.4 mL/min, which started at 1 % mobile phase B for 1 min, increased to 5 % mobile phase B at 1.5 min, and to 95 % B at 2.5 min, followed by a 0.5-min holding at 95 % mobile phase B, then decreased to 1 % mobile phase B at 3.5 min which was followed by another 1.5 min holding at 1 % B.

Statistical analysis

The statistics were performed using SigmaPlot 11.0 (San Jose, CA, USA). For the rat studies of acetaminophen and diquat, respectively, the concentration changes from Cmax to baseline at observed Tmax were tested between the dosed groups and the control groups using the nonparametric Wilcoxon two-sample rank-sum test (two groups). For the mouse studies of paraquat and TBHP, respectively, the statistical tests using Wilcoxon two-sample rank-sum test were performed to compare the concentrations post-dose to the pre-dose control baseline at −0.5 h.

Results

GSH and GSSG stability with sample acidification

Bile samples that were collected directly into the acidification solution showed no conversion of GSH to GSSG. To assess the stability of biliary GSH and GSSG in the acidified bile samples, three bile samples from three individual rats, respectively, were diluted 1,000-fold with the acidification solution and analyzed over a 3-day period at 4 °C. The mean area ratios of GSH for each of the three bile samples across the 3-day period were found to stay virtually constant, at 0.51 ± 0.02, 0.45 ± 0.02, and 0.34 ± 0.01, respectively, and at 1.14 ± 0.04, 0.85 ± 0.04, and 0.45 ± 0.02, respectively, for GSSG (Fig. 1a, b). The coefficients of variation were within 5 % for each subject for up to 94 h. The acidification method used in this assay was thus sufficient to stabilize GSH and GSSG in the bile samples.

Fig. 1
figure 1

Stability of a GSH and b GSSG in the bile samples from three individual rats prepared with the acidification method. Three bile samples from rat A (solid circle), rat B (open circle), and rat C (solid triangle), respectively, were acidified upon collection with formic acid solution and further diluted to 1,000-fold of the original concentrations. The acidified samples were analyzed separately over a 3-day period at 4 °C. The coefficients of variation were within 5 % for each subject across the time period up to 94 h

Full size image

Using artificial bile to prepare the calibration standards

Due to the high salt content in bile, matrix effects of the calibration standards prepared in bile vs. standards prepared in water was a major issue for quantitation of biliary GSH and GSSG using mass spectrometric detection. Endogenous GSH and GSSG in bile can additionally affect the quantitation with the calibration curves prepared in bile, especially in the low calibration range. In this study, the absolute concentration values of biliary GSH and GSSG in rat were quantitated against calibration curves made with artificial bile. The artificial bile contains high concentrations of biliary salts but does not have the endogenous GSH or GSSG. It was tested as a substitute for bile or water to prepare standards and quality controls. The calibration ranges for GSH with the artificial bile was 0.40 to 813.80 μM \( \left( {y=0.0075x+0.007;{R^2}=0.9952} \right) \). The calibration range for GSSG is 0.80 to 102.09 μM \( \left( {y=0.0003x;{R^2}=0.9905} \right) \). The calibration curves prepared in the artificial bile of GSH or GSSG were roughly parallel to that prepared in fresh bile with the ordinate shift by the value of the endogenous biliary GSH or GSSG in fresh bile (not shown). The assay accuracy and precision were within ±20 % for both GSH and GSSG.

Assay of biliary GSH and GSSG in rats after administration of diquat or acetaminophen

Diquat is known to cause oxidative stress and to increase the GSSG level in rats. The two individual rats dosed with diquat at 31 mg/kg (IP) showed an excretion profile of biliary GSH peaking at 13 and 32 nmol/min/kg, respectively, at the bile collection interval of 2.0–2.5 h post-dose (Fig. 2a). The corresponding biliary GSSG excretion for the two tested rats peaked at 28 and 88 nmol/min/kg, respectively, at the collection interval of 1–1.5 h post-dose (Fig. 2b). The GSSG/GSH ratios of those two treated rats both peaked at 7 around 0.5–1.5 h post-dose, compared with that of the two control rats (whose ratios were 1 and 3). The increased biliary GSSG/GSH ratios indicated that oxidative stress was measurable after the administration of diquat (Fig. 2c).

Fig. 2
figure 2

Time profiles of GSH, GSSG, and the ratio of GSSG/GSH in rat bile after an administration of diquat. With a single dose of diquat at 31 mg/kg in 0.9 % saline (IP), biliary a GSH, b GSSG, and c the ratio of GSSG/GSH were measured, respectively. Control rat 1 (closed circle) and control rat 2 (open circle) were intravenously dosed with a sham dose of saline containing 50 % DMSO. Rat 6 (closed star) and rat 7 (open star) were dosed with diquat. GSH peaked at 13 and 32 nmol/min/kg for rat 6 and rat 7, respectively, at the bile collection interval of 2.0–2.5 h post-dose. The corresponding biliary GSSG excretion for the two tested rats peaked at 28 and 88 nmol/min/kg, respectively, at the collection interval of 1–1.5 h post-dose. The GSSG/GSH ratios of those two treated rats both peaked at 7 around 0.5–1.5 h post-dose, compared with that of the two control rats (whose ratios were 1 and 3)

Full size image

Acetaminophen, when overdosed, is metabolized to a toxic electrophilic metabolite NAPQI (N-acetyl-p-benzoquinone imine) in a large amount which can deplete the GSH reserves. A single PO dose of acetaminophen at 1,500 mg/kg in Ora-Plus® depleted biliary GSH causing an initial increased secretion of biliary GSH and GSSG. The biliary GSH excretion peaked at 21, 30, 16 nmol/min/kg for three individual rats at the bile collection interval of 1–1.5 h post-dose (Fig. 3a). The GSSG excretion levels for the same animals reached 88, 51, and 28 nmol/min/kg, respectively, during the collection interval of 0–0.5 h post-dose (Fig. 3b). There was an initial increase of the GSSG/GSH ratios at 6, 3, and 4, respectively, for the three tested rats upon an administration of acetaminophen at 0–0.5 h post-dose, compared to the GSSG/GSH ratios of 1 and 3 in the two control rats. The ratios surged up at the later time points again, at 6–24 h post-dose, reaching 5, 6, and 5 for the three rats (Fig. 3c) and indicating possible hepatic oxidative stress.

Fig. 3
figure 3

Time profiles of GSH, GSSG, and the ratio of GSSG/GSH in rat bile after an administration of acetaminophen. With a single dose of acetaminophen at 1,500 mg/kg (PO), biliary a GSH, b GSSG, and c the ratio of GSSG/GSH were measured, respectively. Control rat 1 (closed circle) and control rat 2 (open circle) were the same subjects in the diquat study (Fig. 2). Rat 3 (closed triangle), rat 4 (open triangle), and rat 5 (open square) were dosed with acetaminophen. The biliary GSH excretion peaked at 21, 30, and 16 nmol/min/kg for rat 3, rat 4, and rat 5, respectively, at the bile collection interval of 1–1.5 h post-dose. The corresponding GSSG excretion levels reached 88, 51, and 28 nmol/min/kg, respectively, at the collection interval of 0–0.5 h post-dose. There was an initial increase of the GSSG/GSH ratios at 6, 3, and 4, respectively, for the three tested rats upon an administration of acetaminophen at 0–0.5 h post-dose, compared to the GSSG/GSH ratios of 1 and 3 in the two control rats. The ratios surged up at the later time points again at 6–24 h post-dose, reaching 5, 6, and 5 for the three rats

Full size image

Rat 4 had a peak of the GSSH/GSH ratio at 6–6.5 h post-dose (Fig. 3c). A close examination of the experimental records on rat 4 indicated that the bile flow rate of this rat was reduced to its one third at the collection interval of 6–6.5 h post-dose. While the decreased bile flow rate was not supposed to change the GSSG/GSH ratio, it indicated that the animal underwent pharmacological changes, possibly some hepatic oxidative stress.

Of the two rats used as control in this study, control rat 1 had low biliary GSH and GSSG excretion, both at 2 nmol/min/kg, which were comparable to the levels previously reported as “non-responder” [23]. The GSSG/GSH ratio for this “non-responder” control rat was 1. Control rat 2 had a biliary excretion of GSH and GSSG at 10 and 34 nmol/min/kg, respectively. Its GSSG/GSH ratio stayed at 3 for up to 4.5 h after an IV administration of a sham dose of 50 % DMSO in saline, and then increased to 4 at 8.5 h after the sham dose.

Detection of biliary GSH and GSSG in mice after administration of paraquat or TBHP

After a single dose of paraquat at 130 mg/kg in 0.9 % saline (subcutaneously), the mean GSSG in mouse bile increased between 0.5 and 4 h post-dose which was the last collection time point (Fig. 4b). The mean GSSG/GSH ratio increased from 1.1 (n = 3, or 0.69, n = 6) at pre-dose to 4 at 2 h post-dose (Fig. 4c). After a single dose of TBHP at 135 mg/kg (IP), the biliary GSSG levels had a continuous increase to 4 h post-dose which was the last collection time point (Fig. 5b), while the biliary GSH level continuously decreased. The mean GSSG/GSH ratio increased from 0.28 (n = 3) or 0.69 (n = 6) at pre-dose to 4.5 at 4 h post-dose of TBHP (Fig. 5c). The increase in the GSSG/GSH ratio after a dose of paraquat or TBHP is potentially indicative of possible hepatic oxidative stress caused by the drug treatments.

Fig. 4
figure 4

Time profiles of GSH, GSSG, and the ratio of GSSG/GSH in mouse bile after an administration of paraquat. Three mice served as controls and were sacrificed at 30 min before dose. Another 15 mice were subcutaneously dosed with paraquat at 130 mg/kg in 0.9 % saline. At each collection time point, three mice were sacrificed to collect gallbladders. The mean GSSG in mouse bile increased between 0.5 and 4 h post-dose which was the last collection time point. The mean GSSG/GSH ratio increased from 1.1 (n = 3, or 0.69, n = 6) at pre-dose to 4 at 2 h post-dose. *p < 0.1 (Wilcoxon two-sample test)

Full size image
Fig. 5
figure 5

Time profiles of GSH, GSSG, and the ratio of GSSG/GSH in mouse bile after an administration of TBHP. Three mice served as controls and were sacrificed at 30 min before dose. Another 18 mice were dosed with TBHP at 135 mg/kg in Ora-Plus (IP). At each collection time point, three mice were sacrificed to collect gallbladders. The biliary GSSG levels had a continuous increase to 4 h post-dose which was the last collection time point, while the biliary GSH level continuously decreased. The mean GSSG/GSH ratio increased from 0.28 (n = 3, or 0.69, n = 6) at pre-dose to 4.5 at 4 h post-dose of TBHP. *p < 0.1 (Wilcoxon two-sample test)

Full size image

Discussion

The assay was proven to be a straightforward way to detect changes of biliary GSSG/GSH ratios induced by the four tested compounds in rats (with two compounds) and mice (with two compounds) to facilitate assessing drug-induced oxidative stress. This study provides examples of assessing the GSSG/GSH ratio as a biomarker for drug-induced oxidative stress. Further investigation will be needed to address the ranges of the GSSG/GSH ratios and the criteria associated with the induced oxidative stress.

The method we described in this study for measuring GSH and GSSG in rat and mouse bile employed an acidification collection step but no derivatization, thus minimizing quantification errors due to sample manipulation. In addition, this method utilizes a UPLC–MS/MS detection which has high sensitivity and allows for directly measuring both biliary GSH and GSSG in a small volume of samples, for example, with individual mouse bile samples. This creates the opportunity that the individual hepatic redox status could be estimated with a small amount of bile directly collected from a mouse gallbladder without extra pre-study surgery.

It should be noted that the animal models used in our studies were different from the animal models used in some previously published work. Our rats underwent bile duct cannulation surgery followed by a 24-h infusion of artificial bile which could alter the levels of the biliary GSH and GSSG, and affect the baseline GSSG/GSH ratio in the control rats. The mean baseline of the biliary GSH excretion in control rat 2 in our study was observed as 10 nmol/min/kg across a period of 24.5 h (nine time points) which was15-fold less than the published 14.8 ± 1.1 nmol/min/100 g [24]. In addition, the biliary GSSG/GSH ratios in the two control rats in this study were observed as 1 and 3 compared to less than 1 in the previous work [24]. The other difference was that the rats in this study were not anesthetized for bile sample collection. It is well known that anesthetized animals can have altered pharmacology and changes in hepatic blood flow rates which could affect the GSH elimination rates and possibly its total concentrations [25]. Unmodified mice used in this study (which were not given anesthesia or any surgery) were also of a different animal model from the anesthetized and bile-duct-cannulated mice [23].

The amount of bile collected from a single mouse gallbladder is only 5–10 μL. This small volume would have required a large number of mice to be sacrificed to collect enough mouse bile as a blank matrix for preparation of calibration standard curves. In this study, we prepared the calibration curves for quantitation of biliary GSH and GSSG in the mouse in 10-fold-diluted artificial bile. As no assessment of matrix effect for mass spectrometric detection between the calibration curves prepared in the 10-fold-diluted artificial bile and the ones prepared in the blank mouse bile were conducted in this study, using 10-fold-diluted artificial bile could affect the obtaining of the absolute values of GSH or GSSG in mouse bile using LC/MS. However, since it is more applicable to assess the relative changes of biliary GSSG and GSH for assessing induced oxidative stress than the absolute quantitation values, the nominal concentrations of biliary GSH or GSSG in the mouse, which might be affected by matrix effect, were reported here.

The compounds tested were known to cause hepatic toxicity, yet statistical significance of p < 0.05 was not reached. The reasons for not being statistically significant in the rat studies or having less desired statistical significance (p < 0.1) in the mouse studies are twofold. First, the inter-subject variability of biliary GSSG or GSH levels was high in both rat and mouse studies (Figs. 2, 3, 4, and 5). This observation was especially true in the rat studies with the existing “non-responder” rat as one of the controls which was probably due to animal surgical manipulation. Secondly, the selected probe drugs at their dosages in this manuscript are known to cause hepatotoxicity. Thus, the studies were proof-of-concept studies with small sample sizes (n = 2 or 3) and were not designed to achieve statistical significance. Nevertheless, the observed high inter-subject variability suggested bigger sample sizes for future studies. And most importantly, even though statistical significance was not achieved, the individual kinetic profiles contain important information. For an example, the GSSG/GSH ratios and the biliary Cmax of GSSG and GSH were clearly elevated above each rat's baseline.

Future studies should test the concept with investigational candidates, with a secondary confirmation to verify induced hepatotoxicity. Analysis of hepatic tissues or detection of oxidative stress on blood chemistry panel with terminal blood can be planned to verify acute oxidative stress or hepatic injury. Liver histology may be needed in some cases where there are hepatic injuries caused by a long-term exposure of drugs of interest.

Last but not the least, since the GSSG/GSH ratios were dynamic and returned to physiological values after the initial increase, the bile collection time is critical to detect potential changes of the GSSG/GSH ratios. The extent and duration of the peak of the GSSG/GSH ratio may be very relevant to the total drug-induced oxidative stress. More understanding of how changes in the kinetics of the biliary GSSG/GSH ratios relate to drug-induced oxidative stress and further relate to toxicology and pathology would greatly help in development of an exposure response index for screening drug toxicity.