Characterizing the metabolites of the tyrosine-kinase inhibitor pexidartinib in mouse feces, urine, plasma, and liver

Abstract

Pexidartinib (PEX, TURALIO^®), a tyrosine kinase inhibitor, is approved for treating tenosynovial giant cell tumor in adults. However, its potential to cause fatal liver injury has prompted the U.S. FDA to issue a black box warning, and the mechanisms underlying its hepatotoxicity remain largely unknown. As biotransformation may contribute to PEX-induced hepatotoxicity, understanding its metabolism is essential. Our previous research indicated that PEX forms reactive metabolites in human and mouse liver microsomes and in human hepatocytes. We investigated PEX metabolism and liver distribution in mice with a focus on metabolite characterization. Our data shows that PEX is mainly excreted into mouse feces as an unchanged drug, in line with findings in humans. Thirty phase I metabolites reported in our previous in vitro studies were detected in mouse feces, urine, plasma, and/or liver; these include the products of unusual carbon-carbon bond cleavages. Twenty-eight phase II PEX metabolites were tentatively identified, including 12 glucuronides, 6 sulfates, 1 glucose conjugate, 2 glutathione, 1 cysteinyl-glycine, and 6 N-acetylcysteine adducts; 24 of these have not previously been reported. The detection of glutathione-PEX adducts and their degradation products indicates that reactive PEX metabolites are generated in mice, consistent with our previous findings in liver microsomes. Since glutathione-PEX adducts are also generated in human primary hepatocytes, the discovery of these new metabolites may help others to clarify the previously unknown metabolic fates of some PEX in human studies and provide starting points for investigations into PEX toxicity by further assessing the safety of its metabolites.

1. Introduction

Pexidartinib (PEX, also known as PLX3397, TURALIO^®) is a tyrosine kinase inhibitor that selectively targets the colony-stimulating factor 1 receptor (CSF1R) and the c-KIT receptor [1]. PEX is the first drug approved by the U.S. Food and Drug Administration (FDA) to treat tenosynovial giant cell tumor in adults who are not likely to benefit from surgery [1-5]. Additionally, PEX is also currently under the evaluation in several clinical trials for other diseases such as refractory leukemias and advanced solid tumors [6-9]. Clinically reported serious and even fatal liver injuries have led the FDA to issue a black box warning for PEX hepatotoxicity [4,10,11]. The mechanism underlying PEX-induced liver injury remains unknown. However, it is well appreciated that drug metabolism plays an important role in drug toxicity [12,13].

A previous report indicated that PEX in human subjects is primarily excreted in feces as its unchanged form [14]. The major metabolite is PEX-N-glucuronide, mainly produced by UDP-glucuronosyltransferase 1A4 (UGT1A4), which is much less potent as a CSF1R inhibitor compared to PEX [10]. The systemic exposure of PEX-N-glucuronide is approximately 10 % higher than that of PEX [14]. Other reported metabolites include mono- or di-hydroxylated products, as well as the glucuronide of dealkylated PEX [15,16]. Pharmacokinetic studies using ¹⁴C-labelled PEX in humans indicated that the overall recovery of administered radioactivity was 92.2 %, with 64.8 % recovered in the feces and 27.4 % in urine [15]. However, 28.4 % of the recovered radioactivity was not chemically identified (11.7 % in urine and 16.7 % in feces). These data suggest that unidentified PEX metabolites are present in humans. In our previous studies, we investigated the metabolism of PEX in human and mouse liver microsomes (HLM/MLM) and in primary human hepatocytes; we found that PEX produces reactive metabolites that can be trapped using reduced glutathione (GSH) or methoxyamine [17]. These reactive metabolites may have the potential to react with macromolecules in vivo, such as proteins and DNA [18,19], potentially causing adverse effects and accounting for unrecovered radioactivity in PK studies. We also discovered that CYP3A4/3A5 mediate the formation of uncommon metabolites via carbon-carbon bond (C-C) cleavage reactions [20]. Because these products are chemically unusual, they might have been overlooked in prior studies in humans.

In this study, we focused on comprehensive investigation of PEX biotransformation in mice by characterizing and semi-quantifying PEX metabolites in mouse urine, feces, plasma, and liver. We tentatively identified 28 phase II metabolites including GSH- and N-acetyl-cysteine (NAc)-PEX adducts using a liquid chromatography–mass spectrometry (LC-MS) based metabolomic approach. This approach has unique advantages in discovering the uncommon and unpredictable metabolites, especially those formed from molecular breakdown that leads to the loss of radiolabeled atoms are frequently missed in human subjects. For example, we have successfully applied this strategy in unveiling the metabolites from C-C cleavage of PEX [20] and reactive adducts [17] in liver microsomes. In this study, the phase II conjugates of the uncommon metabolites produced by C-C cleavage and the terminal fate of the reactive adducts (e.g., N-acetylcysteine adducts) are discovered for the first time. Meanwhile, 30 phase I metabolites reported in our previous in vitro study [20] were also observed and semi-quantified in mice. This extensive examination of PEX metabolism in mice may provide a basis for exploring PEX metabolic profiles in humans and provide valuable insights for future studying the roles of PEX metabolites in its toxicity using animal models (e.g., CYP3A-knockout mice).

2. Materials and methods

2.1. Materials

PEX, (5-[(5-chloro-1H-pyrrolo-[2,3-b]pyridin-3-yl)methyl]-N-[[6-(trifluoromethyl)pyridin-3-yl]-methyl]pyridin-2-amine), was purchased from Cayman Chemical (Ann Arbor, MI, USA). Agomelatine, methyl cellulose and dimethyl sulfoxide (DMSO) were obtained from Sigma-Aldrich (St. Louis, MO, USA). All solvents used for LC-MS were of the highest grade commercially available (Thermo Fisher Scientific, San Jose, CA, USA).

2.2. Animal experiments

Male C57BL/6NJ mice (8–12 weeks old) were purchased from the Center for Comparative Medicine at Baylor College of Medicine and maintained under a standard 12-hour dark/light cycle with free access to water and chow. All the protocols for animal experiments were approved by the Institutional Animal Care and Use Committee at Baylor College of Medicine (Protocol No. AN-6589). PEX was dissolved in 0.5 % methyl cellulose (w/v) containing 10 % DMSO (v/v) at the concentration of 8 mg/mL. The dose of PEX in mice was converted from the clinical dose at 500 mg/daily [10]. The equivalent dose for mice was calculated by multiplying the human dose (approximately 7 mg/kg for a 70 kg individual) by a conversion factor of 12.3, based on the body surface area differences between the two species [21]. Mice (n = 4) were administered orally with PEX at the dose of 80 mg/kg or vehicle (10 μL/g) and kept in individual metabolic cages. The urine and feces were collected continually for 16 h. On the 2nd day, the mice of control group were administered orally again with PEX (80 mg/kg). After 1.5 h, the mice were euthanized and whole blood samples were collected via cardiac puncture, and liver samples were collected in cryovials (Corning, Corning, NY, USA) and snap-frozen in liquid N₂. The whole blood samples were centrifuged at 2000 rcf for 5 min to separate the plasma. The plasma and liver samples were stored at −80 °C until analysis.

2.3. Sample preparation and LC-MS analysis

The sample preparation methods of feces, urine, liver, and plasma were modified from those used in our previous work [22]. Briefly, plasma samples were prepared by mixing 10 μL of plasma with 60 μL of ice-cold methanol containing an internal standard (0.1 μM agomelatine) and 20 μL of urine samples were added into 180 μL of methanol with internal standard. The sample mixtures were vortexed, centrifuged at 15,000 rcf for 15 min. Liver and feces samples were weighed and homogenized in 50 % methanol with the ratios of 1:6 and 1:10 (w/v), respectively. To 50 μL of the liver or fecal homogenate, 200 μL of ice-cold methanol with internal standard was added, and the resulting mixtures were vortexed, centrifuged at 15,000 rcf for 15 min. The resulting supernatants were transferred to a new Eppendorf tube and subjected to a second centrifugation at 15,000 rcf for 15 min. Three microliters of each supernatant were injected into LC-MS system for analysis.

All samples were analyzed on a Vanquish Ultra-high-performance LC coupled with a Q Exactive Orbitrap MS (UHPLC-Q Exactive MS) (Thermo Fisher Scientific) using the method described in our previous work [17]. Q Exactive MS was operated in positive mode with electrospray ionization. For full MS, data ranging from m/z 80–1200 Da in profile mode were acquired. The ion at m/z 371.1012 was used as a reference ion for positive mode during acquisition. For targeted MS/MS, the precursor ions from potential metabolites were fragmented under the High Collision Dissociation mode. The product ions for each precursor were the mixture obtained with normalized collision energies of 10, 35, and 50 % (arbitrary unit). The mass resolution for full MS was 140,000 and the resolution for MS/MS was 15,000.

2.4. Data analysis

The LC-MS data were acquired with Xcalibur software (Thermo Fisher Scientific). The raw data were first imported to Compound Discoverer software (version 3.1, Thermo Fisher Scientific), and then processed using the untargeted metabolomics workflow. The mass tolerance for peak alignment was set at 5 ppm and the retention time shift at 0.1 min. Multivariate data matrix was extracted from Compound Discoverer in Excel format, and then exported into SIMCA-14 (Umetrics, Kinnelon, NJ, USA) for multivariate data analysis [17,22]. Orthogonal projection to latent structures-discriminant analysis (OPLS-DA) was performed on Pareto-scaled data [23]. For chemometric analysis, the matrix data from m/z 100–800 were processed. The percentage of each metabolite in the sample was calculated by normalizing its peak area with the sum of the peak area of all metabolites in this sample. All the data related to metabolite abundances are semi-quantitative, as they are based on ion relative abundance and not calibrated standard curves. The bar graphs were plotted in GraphPad Prism 10.4. All the data are presented as mean ±S.E.M.

3. Results and discussion

3.1. Profiling PEX metabolites in mouse feces and urine using a metabolomic approach

The current work extensively investigated PEX metabolism in mice using an LC-MS-based metabolomic approach, which has previously been employed in our drug metabolism research and mechanistic studies of drug-induced liver toxicity [18,22,24,25]. Results of the chemometric analysis on the ions generated from the LC-Q Exactive MS analysis of fecal and urine samples from the vehicle and PEX-treated groups are presented in Fig. 1. Principal component analysis revealed two clusters corresponding to the vehicle and PEX-treated groups in fecal (inlaid in Fig. 1A) and urine samples (inlaid in Fig. 1B). The S-plots generated from OPLS-DA display the ion contribution to group separation. The top-ranking ions contributing to cluster separation are PEX and its metabolites, which are marked in the S-plots. PEX and its metabolites were present in both mouse urine and fecal samples.

Fig. 1.

Metabolomic analysis of feces and urine samples from mice treated with vehicle and PEX. Male mice were orally administered with either vehicle (Control) or 80 mg/kg PEX (PEX) (n = 4). Fecal and urine samples were collected continually for 16 h and analyzed by LC-MS in the positive ion mode. Ions within the m/z range of 100–800 were subjected to chemometric analysis. (A) The S-plots generated from OPLS-DA analysis of feces samples. (B) The S-plots generated from OPLS-DA analysis of urine samples. The X-axis is a measure of the relative abundance of ions and the Y-axis is a measure of the correlation of each ion to the model. The score plots were individually inlaid in the S-plots showing well-separated clusters in both fecal and urine samples. The t[1] and to[1] values represent the score of each sample in principal component 1 and 2, respectively. The top-ranking ions that contribute to the group separation in feces and urine are labeled in the corresponding S-plots. The identity of ions (metabolite identification) was accordant with those in Table 1.

In contrast to conventional LC-MS methods or approaches using radiolabeled compounds, the metabolomic strategy offers significant advantages in identifying the unexpected and unusual important metabolites, especially those that lose the radiolabeled atoms. For example, our prior studies of the phase I metabolism of PEX in HLM, MLM and in primary human hepatocytes revealed 31 metabolites, including uncommon metabolites formed via C-C bond cleavages and reactive metabolites captured by glutathione or methoxamine [17,20]. In the current study, a similar approach is adopted to further study the fate of PEX in mice including phase I and phase II metabolism. Phase I and Phase II metabolism are two primary processes of drug and xenobiotic detoxification in vivo, mainly occurring in the liver. Phase I reactions, typically mediated by enzymes such as P450s, introduce polar functional groups (e.g., through oxidation), to increase water solubility. Phase II reactions follow by conjugating these modified molecules like glucuronidation and sulfation, further enhancing water solubility and promoting excretion of metabolites [26,27]. Interestingly, 30 of the 31 phase I metabolites detected in HLM [20] were observed in mouse samples (M1-M26 and M28-M31) (Fig. 2A-2D, Supplemental Figs. 1A-1D, and Supplemental Table 1), highlighting a high degree of similarity between humans and mice in phase I metabolism of PEX.

Fig. 2.

Percentages of major PEX metabolites in mouse feces, urine, plasma, and liver. The urine and feces samples were collected as described in Fig. 1. The liver and plasma samples were collected 1.5 h after PEX treatment. (A) The relative abundances of major metabolites in feces. (B), The relative abundances of major metabolites in urine. (C) The relative abundances of major metabolites in plasma. (D) The relative abundances of major metabolites in liver. Phase I and phase II metabolites were extracted using their exact mass (error < 5 ppm). The peak area of each metabolite was normalized by the sum of the peak area of all the metabolites detected in the matrix to give a percentage. Metabolites with larger than 1 % are listed as major metabolites. (E) Relative levels of PEX and its metabolites in feces and urine. The peak areas of PEX, sum of phase I metabolites, and sum of phase II metabolites are normalized to per milligram of feces or per microliter of urine. The abundances of metabolites shown in the figure are semi-quantitative based on their peak areas. Since the ion response of each metabolite in MS may vary, their absolute levels may not be directly comparable. Semi-quantification is an approach that estimates the quantity of a metabolite in a sample based on the peak area, rather than providing an absolute quantification.

We tentatively identified 28 phase II metabolites (M33-M60) in the fecal, urine, plasma, and liver samples from mice treated with PEX, including 12 glucuronides, 6 sulfates, 1 glucose conjugate, 2 GSH adducts, one cysteinyl-glycine adduct, and 6 NAc adducts of PEX or its metabolites (Table 1). Alcohol M27 and a GSH adduct (M32), derived from the C-C bond cleavage reactions of PEX (Fig. 6) [20] were not detected in any sample species, but its degradation product M27-NAc (M60) was observed in all samples (Table 1). As PEX phase II metabolism and bioactivation have not been well-studied in humans, our findings provide a basis for identifying as-yet unknown human metabolites of PEX that may contribute to its hepatoxicity.

Table 1.

Summary of phase II metabolites of PEX in mouse feces, urine, plasma and liver.

RT (min)	Observed m/z [M+H]+	Calculated m/z [M+H]+	Mass error (ppm)	Predicted formula	Identification	Metabolite ID	Source
10.13	594.1367	594.1362	0.84	C26H23ClF3N5O6	PEX+Gluc	M33	F, U, L, P
11.55	594.1356	594.1362	−1.01	C26H23ClF3N5O6	PEX+Gluc	M34	F, L, P
10.11	580.1571	580.1569	0.34	C26H25ClF3N5O5	PEX+Glu	M35	F, U, L, P
8.78	610.1317	610.1311	0.98	C26H23ClF3N5O7	PEX+O+Gluc	M36	F, U, L, P
10.16	610.1314	610.1311	0.49	C26H23ClF3N5O7	PEX+O+Gluc	M37	F, U, L, P
10.57	610.1315	610.1311	0.66	C26H23ClF3N5O7	PEX+O+Gluc	M38	U, P
11.17	608.1143	608.1154	−1.81	C26H21ClF3N5O7	PEX+O-2H+Gluc	M39	F, U, L, P
9.17	498.0614	498.0609	1.00	C20H15ClF3N5O3S	PEX+SO3H	M40	F, U
10.03	498.0609	498.0609	0.00	C20H15ClF3N5O3S	PEX+SO3H	M41	F, U, P
11.03	498.0608	498.0609	−0.20	C20H15ClF3N5O3S	PEX+SO3H	M42	F, L, P
12.08	498.0607	498.0609	−0.41	C20H15ClF3N5O3S	PEX+SO3H	M43	F, U, L, P
9.17	516.0714	516.0715	−0.19	C20H17ClF3N5O4S	PEX+O+2H+SO3H	M44	F, U, L, P
6.65	435.1069	435.1066	0.69	C19H19ClN4O6	Aminopyridine+Gluc	M45	U, L, P
4.82	451.1014	451.1015	−0.22	C19H19ClN4O7	Aminopyridine+O+Gluc	M46	U, P
5.97	451.1015	451.1015	0.00	C19H19ClN4O7	Aminopyridine+O+Gluc	M47	U, P
8.21	355.0261	355.0262	−0.28	C13H11ClN4O4S	Aminopyridine+O+SO3H	M48	F, U, L, P
7.56	474.1121	474.1119	0.42	C19H18F3N3O8	M26 +Gluc	M49	U, P
4.85	446.1168	446.1170	−0.45	C18H18F3N3O7	M28 +Gluc	M50	U, P
7.00	373.0429	373.0433	−1.07	C14H13ClN2O8	M30 +Gluc	M51	U, P
9.04	723.1722	723.1722	0.00	C30H30ClF3N8O6S	PEX+GSH	M52	L, P
9.16	723.1717	723.1722	−0.69	C30H30ClF3N8O6S	PEX+GSH	M53	L, P
8.66	594.1298	594.1296	0.34	C25H24ClF3N7O3S	PEX+Cys+Gly	M54	F, L, P
10.30	579.1177	579.1187	−1.73	C25H22ClF3N6O3S	PEX+NAc	M55	F, U, L, P
10.78	579.1189	579.1187	0.35	C25H22ClF3N6O3S	PEX+NAc	M56	F, L
6.69	420.0890	420.0892	−0.48	C18H18ClN5O3S	Aminopyridine+NAc	M57	F, U
7.71	420.0891	420.0892	−0.24	C18H18ClN5O3S	Aminopyridine+NAc	M58	F, U, L
8.02	415.1046	415.1046	0.00	C17H17F3N4O3S	Cleaved right+NAc	M59	U, L, P
9.80	328.0515	328.0517	−0.61	C13H14ClN3O3S	Cleaved left+NAc	M60	F, U, L, P

PEX, pexidartinib; +O, monohydroxylation; + 2 O, dihydroxylation; +2O+2H, monohydroxylation + hydrogenation; Gluc, glucuronic acid; Glu, glucose; Cys, cysteine; Gly, glycine; NAc, N-acetylcysteine. U, urine; F, feces; P, plasma; L, liver. The metabolite IDs (M33–M60) in this table are labelled as an extension of the ID numbers (M1–M32) from [20].

Fig. 6.

Chemical structures of PEX and its metabolites. Putative structures of phase I and phase II metabolites were identified based on their exact masses and MS/MS fragments. The structures of M1–M32 were previously reported in our work [20], while M52–M54 were described in [17] (labeled as M1/M2 and M11). The ID numbers in blue are novel metabolites. GSH refers to reduced glutathione. The metabolite IDs in this figure extend the ID numbering in [20]. GSH, reduced glutathione.

3.2. PEX and its metabolites in mouse feces, urine, plasma, and liver

In mouse feces, a total of 47 metabolites were detected (Table 1 & Supplemental Table 1, Fig. 2A and Supplemental Fig. 1A), including 29 phase I metabolites (M1-M26 and M28-M30) and 18 phase II metabolites (M33-M37, M39-M44, M48, M54-M58 and M60). Among these, M1 (O+PEX) accounts for 10.3 %, M4 (O+PEX) for 14.0 %, M7 (O+PEX-2H) for 11.7 %, M8 (2 O+PEX) for 11.0 %, M12 (aminopyridine) for 11.3 %, and M44 (PEX+O+2H+SO₃H) for 19.6 % of the total peak area (Fig. 2A). In mouse urine, 50 metabolites were resolved, comprising 28 phase I metabolites (M1-M4, M6-M26, M28, M30 and M31) and 22 phase II metabolites (M33, M35-M41, M43-M51, M55 and M57-M60) (Table 1 & Supplemental Table 1, Fig. 2B, and Supplemental Fig. 1B). Among these metabolites, dihydroxylated aminopyridine M17 (2 O+M12) accounts for 8.7 %, acid M23 for 42.3 %, and acid M26 for 12.6 % of the total peak area (Fig. 2B). In plasma, 52 circulating metabolites were observed, consisting of 29 phase I metabolites (M1-M26 and M28-M30) and 23 phase II metabolites (M33-M37, M39, M41-M55, M59 and M60) (Table 1 & Supplemental Table 1, Fig. 2C and Supplemental Fig. 1 C). Metabolites accounting for more than 10 % of total peak area of metabolites in plasma included M2 (O+PEX, 38.7 %), ketone M7 (12.5 %), and aminopyridine M12 (19.5 %) (Fig. 2C). In mouse liver, a total of 41 metabolites were detected, including 22 phase I metabolites (M1-M5, M7-M8, M11-M18, M22-M26, M28 and M30) and 19 phase II metabolites (M33-M37, M39, M42-M45, M48, M52-M56, and M58-M60) (Table 1 & Supplemental Table 1, Fig. 2D and Supplemental Fig. 1D). Among them, M12 had the highest relative abundance at 37.1 %, followed by M7 (17.3 %) and M8 (11.9 %) (Fig. 2D). Percentages in Fig. 2A-2D were calculated by dividing the peak area of each metabolite by the total peak area of all detected metabolites in the sample, excluding the peak area of the parent PEX.

Except the proportions of major and minor PEX metabolites in Fig. 2 and supplemental Fig. 1, their abundances based on the peak area were also provided in Supplemental Fig. 2.

In Fig. 2E, peak areas are normalized by the sum of the peak areas for PEX plus all its metabolites. The components were divided into unchanged PEX, phase I metabolites and phase II metabolites, and the peak area sum (PEX + all its metabolites) was regarded as 100 %. Unchanged PEX was mainly excreted into feces (Fig. 2E). The peak area of PEX per milligram of feces was 1574-fold greater than that per microliter of urine (5.97e+11 vs 3.79e+8, Fig. 2E). In humans, 65 % of PEX ¹⁴C radioactivity is reported to be excreted into feces, with 44 % as unchanged PEX, and small amounts as a dihydrodiol (M11, Fig. 6), tri-oxidized, or dioxidized forms. Around 27 % of PEX administered was excreted into urine as metabolites, including over 10 % as the PEX N-glucuronide (M33, Fig. 6) [15]. This strategy relies on the fact that the novelly identified metabolites often have no commercially available standards, and is well utilized to provide a preliminary classification of major and minor metabolites in different systems [20,22]. However, due to potential variability in MS response factors among different metabolites, these percentage estimates are considered semi-quantitative.

In mouse feces, unchanged PEX accounts for 84.5 % of the peak area in the sum of the parent drug and detected metabolites, while it contributed to only 1.7 % in urine. In contrast, phase I metabolites were the most abundant class in urine (81.8 %), with phase II metabolites also representing a significant proportion (16.5 %) (Fig. 2E). In short, PEX was largely excreted into feces as its unchanged form in mice, with only a limited amount into the urine (Fig. 2E), similar to what was seen in a human study [14].

3.3. Identification of PEX-related glucuronides in mice

M33 and M34 were tentatively identified as the direct conjugates of glucuronide acid (Gluc) to different positions of PEX. Both were detected in the feces, plasma, and liver, but M34 was also present in urine samples. M33 has the protonated m/z of 594.1367 and retention time (RT) of 10.13 min (Table 1, Fig. 3A & 3B). The major MS/MS fragments of M33 were m/z 165, 258, 266, and 418, as illustrated in Fig. 3B. M34 has a protonated m/z of 594.1356 with a retention time of 11.55 min (Table 1, Fig. 3A & 3C). The major MS/MS fragments of M34 were 165, 258, 271, 283, 418 and 460, and are interpreted in Fig. 3C. The putative structures of M33 and M34 were proposed as N-glucuronides as shown in Figs. 3B and 3C. M34 was proposed as an amide based on its retention time and MS/MS fragments.

Fig. 3.

Identification of major glucuronide metabolites associated with PEX. The chromatograms were extracted from a fecal or urine sample according to their relative abundances with the mass error of 5 ppm. The MS/MS spectra were obtained with a mixed normalized collision energy of 10 %, 35 % and 50 % (arbitrary units). (A) Chromatograms of M33 and M34. (B) MS/MS of M33. (C) MS/MS of M34. (D) Chromatograms of M36, M37, M38 and M50. (E) MS/MS of M36. (F) MS/MS of M37. (G) MS/MS of M38. (H) MS/MS of M50.

M35 eluted at 10.11 min and has a protonated m/z of 580.1571, with major fragment ions of 160, 165, 258, 266 and 418. It was tentatively identified as a PEX-glucose conjugate and was present in all the samples examined (Table 1, and Supplemental Figs. 4A & 4B). The protonated m/z value of M36-M38 was 610.1311, 16 Da larger than those of M33 and M34. Thus, these three metabolites were tentatively identified as glucuronides of monohydroxylated PEX (O+PEX) (Table 1, Fig. 3D-3G). M36-M38 eluted at 8.78, 10.16 and 10.57 min, respectively (Fig. 3D). The major MS/MS fragments of M36 were 181, 266 and 434, and are interpreted in the inlaid structural diagram of Fig. 3E, suggesting that the 5-chloro-7-azaindole ring was monohydroxylated. M37 was identified as a glucuronide of M3 (Supplemental Table 1 and Fig. 6) based on its MS/MS fragments. The major MS/MS fragments of M37 were m/z 181, 254, 282 and 434, and are interpreted in the inlaid structural diagram in Fig. 3F. M38 was less abundant than M36 and M37 exhibited a similar MS/MS fragmentation pattern to that of M36. Glucuronidation of M38 also occurred on the 5-chloro-7-azaindole ring, but at a different position. The major fragments of M38 are interpreted in the inlaid structural diagram of Fig. 3G. M39 eluted at 11.17 min (Table 1 and Supplemental Figs. 3 A & 3B) and was inferred to be the glucuronide of ketone M7 (Supplemental Table 1 and Fig. 6) [20] based on its protonated m/z of 608.1143 and major MS/MS fragments of 160, 280 and 432. The fragments are interpreted in Supplemental Fig. 3B, and the proposed structure is shown in the inlaid diagram. M36, M37 and M39 were detected in all the samples, while M38 only was observed in the urine and plasma samples (Table 1).

M45-M47 were tentatively identified as glucuronides derived from aminopyridine M12 (Supplemental Table 1 and Fig. 6) [20]. M45 eluted at the retention time of 6.65 min, had a protonated m/z of 435.1069 (Supplemental Figs. 4A & 4C), and is proposed to be the N-glucuronide of M12. The major MS/MS fragments of M45 were m/z 107, 178, 259 and 301 as illustrated in Supplemental Fig. 4C. M46 and M47 eluted at 4.82 and 5.97 min respectively; each has a protonated m/z of 451.1015, 16 Da larger than that of M45, and are proposed to be the glucuronides of monohydroxylated aminopyridine M12 (Supplemental Table 1 and Supplemental Figs. 3A, 3C & 3D). The major MS/MS fragments of M46 were m/z 95, 181 and 275, while those of M47 were m/z 123, 165 and 275, and these are interpreted in Supplemental Figs. 3C and 3D, respectively. M45-M47 were detected in urine and plasma samples; M45 was also observed in the liver (Table 1).

Our previous study demonstrated that C-C bond cleavages of PEX resulted in several unusual metabolites (M24-M31) [20] (Supplemental Table 1 and Fig. 6). We observed these metabolites undergoing phase II metabolism in mice. M49 eluted at 7.56 min, has a protonated m/z of 474.1121, and is proposed as a glucuronide of acid M26 (Table 1 & Supplemental Table 1 and Supplemental Figs. 4D & 4E). The major MS/MS fragments of m/z 121, 160, 254, 280 and 298 are shown in inlaid structural diagrams of Supplemental Fig. 4E. M50 eluted at 4.85 min, has a protonated m/z of 446.1168, and is proposed to be a glucuronide of phenol M28 (Table 1 & Supplemental Table 1 and Fig. 3D & 3H) [20]. The major MS/MS fragments of 110, 160 and 270 are shown in Fig. 3H and interpreted in the inlaid structural diagram. Acid M30 is another metabolite derived from C-C bond cleavage of PEX (Supplemental Table 1 and Fig. 6) [20]. M51 eluted at 7.00 min and has a protonated m/z of 373.0429, and is proposed to be a glucuronide of acid M30 (Table 1 and Supplemental Figs. 4D & 4 F). The major MS/MS fragments of m/z 135, 153, 179 and 197 are interpreted in the inlaid structural diagram of Supplemental Fig. 4F. M49-M51 were observed in urine and plasma.

The detection of M26 and M28 (and of their phase II metabolites) indicates that the unusual C-C bond cleavage reactions we detected in microsomes also occur in mice. In urine, acids M23 and M26 were abundant, but phenol M28 was less abundant (Fig. 2B and Fig. 6). M50 is a relatively abundant glucuronide in mouse urine (Fig. 2B) that is derived from M28, which is a major metabolite in HLM [20]. Thus, M28 may be a substrate for UGTs in humans. The 12 % of radioactivity that was unidentified in human urine could potentially be attributed to these uncommon PEX metabolites generated from C-C bond cleavages, which may not have been detected using conventional methods, leading to the ignorance of some important metabolites [15].

3.4. Identification of PEX-sulfate adducts in mice

PEX and its phase I metabolites also underwent sulfation in mice. Four sulfate adducts of PEX (M40-M43) were revealed. M40-M43 have the identical theoretical protonated m/z of 498.0609, and they eluted at 9.17, 10.03, 11.03 and 12.08 min, respectively (Table 1 and Fig. 4A). Their exact masses indicated that M40-M43 were formed by directly conjugating a sulfate group to different nitrogen atoms on PEX. M40-M43 exhibited similar patterns of MS/MS fragments, and their major MS/MS fragments were interpreted in the inlaid structural diagrams of Fig. 4B-4E. M44 eluted at 9.17 min, has a protonated m/z of 516.0714 (Fig. 4F), and is tentatively identified as a sulfate of M6 (O+PEX+2H) (Figs. 4A and 6, Table 1 & Supplemental Table 1) [20]. The MS/MS fragments of m/z 107, 160, 258, 266, 418, and 434 are shown in Fig. 4F and interpreted in the inlaid structural diagram. M48 eluted at 8.21 min, has a protonated m/z of 355.0261 (Table 1, Supplemental Figs. 3E & 3 F), and is determined to be a sulfate of monohydroxylated aminopyridine (Table 1). The major MS/MS fragments of m/z 107, 181, and 275 are interpreted in the inlaid structural diagram of Supplemental Fig. 3F. M43, M44, and M48 were detected in all the samples, while M40-M42 were observed in some samples (Table 1). Among these sulfate adducts, M44 is the most abundant (Figs. 2A and 4A).

Fig. 4.

Identification of major sulfate metabolites associated with PEX. The chromatograms were extracted from a fecal sample with the mass error of 5 ppm. The MS/MS spectra were obtained with a mixed normalized collision energy of 10 %, 35 % and 50 % (arbitrary units). (A) Chromatograms of M40-M44. (B) MS/MS of M40. (C) MS/MS of M41. (D) MS/MS of M42. (E) MS/MS of M43. (F) MS/MS of M44.

As discussed above, PEX N-glucuronidation (0.33 % of total metabolite peak areas) is a minor pathway in mouse urine compared to humans. On the contrary, five sulfate adducts were observed in mouse feces, with the species PEX+O+2H+SO₃H (M44, Fig. 6) accounting for 19.6 % of all the metabolite ion counts in mouse feces (Fig. 2A). In human studies, M6 (PEX+O+2H, Supplemental Table 1 and Fig. 6) was reported to account for 2.4 % of the total radioactivity in feces, while another 16.7 % of the radioactivity remained uncharacterized [15]. M6 was found in mouse feces and at low concentrations in plasma but was undetectable in the liver (Supplemental Figs. 1 C and 1D). M44, the M6 sulfate adduct, is present at high levels in mouse feces, and is detected in mouse liver and plasma. We infer that M6 readily undergoes sulfation in mice, and is excreted as M44. It is plausible that some of the unidentified metabolites in human feces may be sulfate adducts of PEX; further investigation is warranted.

So far, it is unknown whether PEX sulfate adducts form in humans. A previous pharmacokinetic study in human subjects [28] reported that UGT inhibitors increase the C_max of PEX less efficiently than CYP3A inhibitors, suggesting that inhibition of UGTs may redirect PEX metabolism toward sulfation. Thus, re-evaluation of human metabolism studies might uncover sulfates and uncommon C-C cleavage products that would improve our knowledge of PEX metabolic profiles in humans.

3.5. Identification of PEX-related GSH or NAc adducts in mice

Our previous studies in liver microsomes indicated that PEX can be activated to react with GSH [17]. In the current study in mice, both GSH and NAc adducts were detected, indicating that reactive PEX metabolites are generated in vivo. M52 and M53 are GSH adducts of PEX and M54 is a cysteinylglycine adduct of PEX (Table 1 and Fig. 6), which have been elucidated in our previous study [17]. In mice, M52, M53 and M54 were observed in plasma and liver, while M54 was also detected in feces (Table 1). In urine and feces, the NAc adducts M55 and M56 (PEX+NAc) were observed, as GSH moieties are typically further metabolized into NAc [29]. M55 and M56 each had a protonated m/z of 579.1187 and eluted at 10.30 and 10.78 min, respectively (Table 1 and Fig. 5A-5C). The MS/MS of M55 produced a major fragment ion of 416, which is interpreted in the inlaid structural diagram (Fig. 5B). The MS/MS of M56 produced fragments at 130, 197, 266 and 450, which are interpreted in the inlaid structural diagram (Fig. 5C).

Fig. 5.

Identification of N-acetylcysteine adducts of PEX. The chromatograms were extracted from a fecal or urine sample according to their relative abundance with the mass error of 5 ppm. The MS/MS spectra were obtained with a mixed normalized collision energy of 10 %, 35 % and 50 % (arbitrary units). (A) Chromatograms of M55 and M56. (B) MS/MS of M55. (C) MS/MS of M56. (D) Chromatograms of M59 and M60. (E) MS/MS of M59. (F) MS/MS of M60. (J) The proposed mechanisms of M59 and M60 formation.

M57 and M58 each had a protonated m/z of 420.0892, eluted at 6.69 min and 7.71 min respectively (Table 1 and Supplemental Figs. 5A-5C), and are proposed as M12 +NAc (M12 is the major N-dealkylated metabolite of PEX) [20]. The MS/MS of M57 produced a major fragment ion of 257, which is interpreted in the inlaid structural diagram (Supplemental Fig. 5B), while the MS/MS of M58 produced fragments at 95, 107, 197 and 291, which are interpreted in the inlaid structural diagram (Supplemental Fig. 5C). Two GSH adducts derived from the metabolites formed by C-C bond cleavages (Fig. 5G) [17,20] were found in vitro, but not found in mice; however their corresponding NAc adducts (M59 and M60) were observed (Fig. 5D and Fig. 6). M59 eluted at 8.02 min and had a protonated ion of m/z 415.1046 (Fig. 5D & 5E). The MS/MS of M59 yielded major fragment ions of m/z 126, 160, 254, 284 and 373, which are interpreted in the inlaid structural diagram (Fig. 5E). M60 eluted at 9.80 min and had a protonated ion of m/z 328.0515 (Fig. 5D & 5F). The major MS/MS fragments of M60 are interpreted in the inlaid structural diagram (Fig. 5F). The proposed mechanism(s) of M59 and M60 formation are presented in Fig. 5G. M55 and M60 were observed across all sample types. M60 was more abundant in urine and plasma compared to M55, while the opposite was seen in feces and liver. M56-M59 were detected only in specific sample types (Table 1). The abundance of GSH adducts to PEX and its metabolites, and of degradation products of these adducts, shows that the reactivity of either activated PEX or its phase I metabolites could contribute to PEX hepatotoxicity.

It is well appreciated that reactive metabolites can modify biological macromolecules [30]. A trace of amount of adduct M54 (PEX+Cys+Gly, Fig. 6) [17] was also present in mouse liver (Table 1 and Supplemental Fig. 1D), indicating that M54 may be an intermediate in the formation of M55 and M56. Adducts M57 and M58 (M12-NAc) were detected in the mouse feces and urine (less than 1 %), likely resulting from the dealkylation of M55 and M56, as M12-GSH adducts were not observed in mouse liver and plasma. Interestingly, NAc adducts M59 and M60, which derive from the products of unusual C-C bond cleavages [17], were also observed in mouse liver and plasma. These adducts are likely formed by the degradation of their corresponding GSH adduct precursors, which we reported in previous in vitro studies [17] (Table 1 & Supplemental Table 1, Fig. 5G). Generally, GSH adducts in vivo are converted to NAc adducts in several steps, sequentially catalyzed by gamma-glutamyltransferase, dipeptidases and N-acetyltransferases [24,29,31]. The formation of M60 has been interpreted as occurring by an ipso-addition of oxygen mechanism, while the formation of M59 remains elusive. The proposed mechanisms for M59 and M60 formation are depicted in Fig. 5G. Further studies are needed to determine the exact role of PEX reactive metabolites in its liver toxicity.

In mouse plasma and liver, metabolites M2 and M3 (PEX+O), ketone M7, M8 (diol), and aminopyridine M12 were relatively abundant (Figs. 2C and 2D). Although these metabolites were not reported in human plasma [15], a conjugate corresponding to M45 (M12 +Gluc) was detected in human urine, and M8 was observed in human feces, indicating that both M8 and M12 form in human subjects [15]. Aminopyridine M12 was abundant in mouse liver, as were its metabolites M12-Gluc (M45) and M12 +O-sulfate (M48) (Fig. 2D and Supplemental Fig. 1D). Aminopyridine moieties are known to be hepatotoxic [32]; however, the role of M12 in PEX toxicity remains unclear and requires further evaluation. Our previous study demonstrated that metabolite M3 is not stable, with its hydroxyl group capable of interchanging with a methoxyl group in methanol [20]. Additionally, incubating M3 in water at 37°C in the presence of NAc could form a PEX-NAc adduct (Supplemental Fig. 6). Metabolite M3 may pose the risk to liver toxicity by reacting with hepatic proteins. Given the high abundance and special chemical properties of metabolites M3 and M12, further research is needed to evaluate their potential contribution to PEX-induced hepatic toxicity. The identified in vivo fate of PEX in plasma, feces, and urine may also provide some valuable insights into understanding the PEX liver toxicity as well. Additionally, we cannot rule out the possibility that other metabolites may also contribute to PEX-induced hepatotoxicity.

4. Summary and conclusion

Our study extensively examined the Phase I and Phase II metabolism of PEX in mice, characterizing and semi-quantifying its metabolites in feces, urine, plasma, and liver. Current studies comprehensively investigate phase I and phase II metabolites in mice, in which the conjugates of the uncommon metabolites produced by carbon-carbon cleavage, sulfate conjugates of PEX, and terminal fate of the reactive adducts in urine and feces were reported for the first time [14,15]. In human subject, there are plenty of unrecovered radioactivity, indicating that some significant metabolites were not unidentified. These results could potentially account for unrecovered radioactivity in human PK studies and may be valuable for studying the PEX hepatoxicity. The detection of GSH- and NAc- adducts in mice indicates that reactive metabolites of PEX previously observed in primary human hepatocytes (Supplemental Fig. 7) and liver microsomes also are produced in mouse liver and strongly suggests that these will be made in human liver. The major phase I metabolites (e.g., M12, a pyridine anime) are quite similar in human and mice. Future investigations might focus on how the relatively abundant phase I metabolites M3 and M12 (a major metabolite in human subject), and any reactive metabolites, may contribute to PEX liver toxicity. Overall, this study provided a comprehensive PEX metabolism in mice and offered insights for the understanding of the PEX metabolism in humans, and for studying PEX toxicity from a metabolic perspective.

Supplementary Material

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.jpba.2025.117034.

Abbreviations:

PEX

pexidartinib

CSF1R

colony-stimulating factor 1 receptor

UGT

UDP-glucuronosyltransferase

GSH

reduced glutathione

LC-MS

liquid chromatography –mass spectrometry

UHPLC

ultra-high performance liquid chromatography

Q Exactive MS

Q Exactive Hybrid Quadrupole-Orbitrap mass spectrometer

OPLS-DA

orthogonal projection to latent structures-discriminant analysis

HLM

human liver microsomes

NAc

N-acetylcysteine

Gluc

glucuronic acid

Glu

glucose

Data Availability

The authors declare that all the data supporting the findings of this study are available within the paper and its Supplemental Data.