Identification and characterization of metabolites of ASP015K, a novel oral Janus kinase inhibitor, in rats, chimeric mice with humanized liver, and humans

Abstract

1. Here, we elucidated the structure of metabolites of novel oral Janus kinase inhibitor ASP015K in rats and humans and evaluated the predictability of human metabolites using chimeric mice with humanized liver (PXB mice).
2. Rat biological samples collected after oral dosing of 14C-labelled ASP015K were examined using a liquid chromatography–radiometric detector and mass spectrometer (LC–RAD/MS). The molecular weight of metabolites in human and the liver chimeric mouse biological samples collected after oral dosing of non-labelled ASP015K was also investigated via LC–MS. Metabolites were also isolated from rat bile samples and analyzed using nuclear magnetic resonance.
3. Metabolic pathways of ASP015K in rats and humans were found to be glucuronide conjugation, methyl conjugation, sulfate conjugation, glutathione conjugation, hydroxyl- ation of the adamantane ring and N-oxidation of the 1H-pyrrolo[2,3-b]pyridine ring. The main metabolite of ASP015K in rats was the glucuronide conjugate, while the main metabolite in humans was the sulfate conjugate. Given that human metabolites were produced by human hepatocytes in chimeric mice with humanized liver, this human model mouse was believed to be useful in predicting the human metabolic profile of various drug candidates.

Keywords : ASP015K, chimeric mice, humanized, metabolite

Introduction

Members of the Janus kinase (JAK) family are protein tyrosine kinases associated with the membrane proximal region which transduce cytokine-mediated signals (Ihle, 1994; Johnston et al., 1994; Linnekin et al., 1997), and consist of JAK1, JAK2, JAK3 and tyrosine kinase 2 (TYK2). JAK1 is essential for the signalling pathways of cytokines involved in inflammatory responses in psoriasis and rheumatoid arthritis (RA), such as interleukin (IL)-6 and interferon (IFN)-g (Imada & Leonard, 2000; Leonard, 2001). JAK3 is associated with the signalling pathways of cytokines that play a crucial role in T-cell differentiation, proliferation and survival, such as IL-2, IL-4, IL-7, IL-9, IL-15 and IL-21 (Imada & Leonard, 2000; Leonard, 2001). We recently developed ASP015K, an orally bioavailable JAK inhibitor with moderate selectivity for JAK1 and JAK3 which inhibits those kinases more potently than JAK2 (by 1.3 and 7.0 times, respectively) (Higashi et al., 2012). This drug demonstrated potent efficacy in both a rat model of adjuvant-induced arthritis and RA patients at once- daily doses of 100 and 150 mg (Papp et al., 2012). However, while previous studies have clarified the kinetic profile of unchanged ASP015K in rats and humans, the structures of its metabolites remain unknown.

Regarding metabolites of drugs, the United States Food and Drug Administration (FDA) and the International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) have issued, ‘‘Metabolites in Safety Testing’’, and ‘‘M3(R2) Nonclinical Safety Studies for the Conduct of Human Clinical Trials and Marketing Authorization for Pharmaceuticals’’, respectively (FDA, 2008; ICH, 2009). These guidance recommend that drugs for which the metabolites ratio exceeds 10% of total drug exposure should be qualified in at least one of the animals used in toxicity studies. Drug metabolites frequently retain the pharmacological and toxicological activities of their parent compounds (Kalasz et al., 2013; Klimas & Mikus, 2014; Ram et al., 2012; Smith & Dalvie, 2012; Testa et al., 2012), and reactive metabolites generated by metabolic activation of drugs may cause idiosyncratic toxicity in humans (Hussaini & Farrington, 2014). Therefore, determining the structures of metabolites for evaluation of safety assessment and clarification of the pharmacological mechanism is of the utmost importance to ensuring compound safety.

Interspecies differences in drug metabolism are often observed during drug development. Although in vitro systems using human materials have been applied to predict in vivo human metabolites, Dalvie et al. (2009) reported that the predictability of phase I metabolites was only 50–69% in primary human hepatocytes, with that of circulated secondary metabolites even less (547%). This poor predictability is believed to be due to the preparation, storage and experi- mental treatment methods for hepatocytes altering the func- tion of metabolizing enzymes (Wang et al., 2005). Precise prediction of human metabolites therefore remains a challenge.

Chimeric mice with humanized liver (PXB mice; PhoenixBio Co., Ltd., Hiroshima, Japan) have been generated from urokinase-type plasminogen activator/severe combined immunodeficiency (SCID) mice transplanted with human hepatocytes (Tateno et al., 2004), and approximately 80% of hepatocytes are humanized in these animals. The expression levels and metabolic activities of cytochrome P450 (CYP) and non-CYP enzymes in livers of the liver chimeric mice are similar to values in humans (Katoh et al., 2004, 2005), and human-specific metabolites have been successfully identified in this model (De Serres et al., 2011; Inoue et al., 2009; Kamimura et al., 2010; Yamazaki et al., 2010).

Here, the purpose of this current study was to elucidate the structures of metabolites of ASP015K to understand the metabolism and pharmacokinetics of this drug. Additionally, the metabolite profiling of the drug in the chimeric mice with humanized liver was conducted to assess the utility of the humanized mice in drug discovery and development.

Materials and methods

Chemicals and regents

Test articles of 14C-labelled ASP015K (Figure 1) synthesized by PerkinElmer (Waltham, MA) and non-labelled ASP015K, as well as an authentic sample of metabolite M2, ASP015K- 50-O-sulfate, synthesized by Astellas Pharma Inc. (Osaka, Japan) were used. Commercially available ammonium acetate (Nacalai Tesque, Kyoto, Japan), ammonia solution (Nacalai Tesque), methanol (Nacalai Tesque) and acetonitrile (Wako Pure Chemical, Osaka, Japan) were used to prepare the high-performance liquid chromatography (HPLC) mobile phase. Methylcellulose (Metolose, SM-400; Shin-Etsu Chemical, Tokyo, Japan) was used for the preparation of the dosing suspension. The deuterated solvent for nuclear magnetic resonance (NMR) analysis, dimethylsulfoxide-d6 and methanol-d4, were purchased from Acros Organics (Geel, Belgium). Scintillation cocktails, Pico-fluor-40 (PerkinElmer) and Ultima-Flo M (PerkinElmer) were used for radiometric detection via a liquid scintillation counter and flow scintil- lation analyzer coupled with HPLC, respectively.

Equipment

Agilent 1100 or 1200 (Agilent Technologies, Inc., Palo Alto, CA) coupled with LTQ XL (Thermo Fisher Scientific, Waltham, MA) or LCQ Deca XP Plus (Thermo Fisher Scientific) were used to perform liquid chromatography–mass spectrometry (LC–MS) or liquid chromatography–radiomet- ric detection and mass spectrometry (LC–RAD/MS), respect- ively. A liquid scintillation counter (Tri-carb 3100 TR; PerkinElmer) and a scintillation analyzer as the detector of LC–RAD/MS (Radiomatic 625 TR; PerkinElmer) were used for the radiometric detection of 14C-labelled ASP015K. An Agilent 1200 coupled with a single quadrupole mass analyzer (G6130; Agilent Technologies, Inc.) was used as the mass- based fractionation system for isolation of ASP015K metabolites. A fast atom bombardment mass spectrometer (FAB-MS; JMS-GCmate II; JEOL, Ltd., Tokyo, Japan), an electrospray ionization mass spectrometer (ESI-MS; TSQ Quantum; Thermo Fisher Scientific) and an NMR setup (Varian Inova 600; Agilent Technologies, Inc.) were used for structure elucidation of ASP015K metabolites.

Animals

Six- to seven-week-old male Sprague Dawley rats were purchased from CLEA Japan, Inc. (Tokyo, Japan). The rats were acclimated for 7–10 days with free access to pellet diet (CE-2; Oriental Yeast Co., Ltd., Tokyo, Japan) and water before being used. Chimeric mice with humanized liver and non-transplanted uPA+/+/SCID genotype mice were supplied by PhoenixBio Co., Ltd. The mice were given free access to solid food CRF1 (Oriental Yeast Co., Ltd.) and water. Human hepatocytes, Lot No. BD87, 2-year-old of male Caucasian had been purchased from BD Biosciences (Discovery Labware, BD GentestTM Products and Services, Franklin Lakes, NJ) and chimeric mice with humanized liver were produced as 14C-labelled ASP015K was orally administered to rats at 3 mg/kg (4.64 MBq/kg) in a previous study (Oda et al., 2014), and plasma samples collected at the time to reach maximum plasma concentration (tmax) were used for meta- bolic profiling. The radioactivity concentration in rat plasma collected at 0.25 h after dosing was 0.46 mg eq. of ASP015K/ mL. Urine and bile samples collected at the intervals of 0–6 and 6–24 h after dosing were used for metabolic profiling. The radioactivity excretion values for 0–6 and 0–24 h post- dosing were 5.3% and 8.2% in rat urine, and 36.7% and 43.7% in rat bile, respectively (Oda et al., 2014). Individual plasma, urine and bile samples (n 4 each) were pooled, and these pooled samples were applied to a solid phase extraction cartridge (Oasis HLB 3 cc/60 mg; Waters, Milford, MA). The cartridge was washed with water and eluted with methanol, and the methanol eluate was concentrated and reconstituted in 20% methanol for LC–RAD/MS analysis.

To obtain the sample for isolation of metabolites, non-labelled ASP015K was orally administered to 30 rats at 150 mg/kg, and bile samples were collected from their bile ducts for 6 h after dosing. Non-labelled ASP015K was also orally administered to the liver chimeric and control mice twice at 40 mg/kg each, as follows: ASP015K was suspended in 0.5% methylcellulose to prepare an 8 mg/mL dosing suspension directly prior to administration. All doses were calculated based on the body weight of each animal prior to dosing. Urine samples were collected for 24 h after the first dosing using metabolic cages. Blood and bile samples were collected at 1 h after the second dosing from the heart and the gallbladders of the mice, respectively. Animal test conditions, dose, urine sampling interval and plasma and bile sampling points were determined based on findings from the prelim- inary test. All of the blood samples were centrifuged immediately after collection at 1000g, 4 ◦C for 5 min to obtain plasma. Plasma, urine and bile samples (n 2 each) were pooled by dosing group for analysis. Mouse biological samples were deproteinized using acetonitrile, and the supernatant was concentrated and reconstituted in 20% methanol for LC–MS analysis.

Human plasma and urine samples

For the metabolic profiling of ASP015K in humans, leftover plasma and urine samples collected in a previous study (Zhu, 2012) were used. The human plasma samples collected at 0.25, 0.5, 1, 1.5, 2, 3, 4, 6, 8, 10, 12, 24, 36, 48 and 72 h after oral administration of ASP015K were pooled in accordance with the plasma pooling method of Hop et al. (1998) by dosing group to prepare a single sample with a concentration proportional to the pharmacokinetic AUC72 (AUC from time 0–72 h). Equivalent volumes of human urine samples (n 6) collected for 24 h after dosing were also pooled by dosing group. An aliquot of the samples was deproteinized using acetonitrile, and the supernatant was concentrated and reconstituted in 20% methanol for LC–MS analysis. The remaining urine samples (approximately 2.5 L) were pooled for metabolite isolation.

LC–RAD/MS and LC–MS analytical conditions

The mobile phase consisted of 100 mmol/L ammonium acetate (pH 8)/acetonitrile/water at 10/5/85 (A) and 100 mmol/L ammonium acetate (pH 8)/acetonitrile/water at 1/8/1 (B) volumetric ratio on a Develosil C30-UG-5 (5 mm, 4.6 250 mm; Nomura Chemical; Aichi, Japan) at 40 ◦C. Chromatographic separation was performed via gradient elution. The proportion of solvent B was increased linearly from 0% to 20% over 50 min, at this point the proportion was then increased linearly to 100% over 10 min and held there for 10 min. The sample eluents were monitored by UV absorbance at 295 nm and MS at m/z 150–1000 in positive ion mode. To obtain the mass fragmentation data, the metabolites were also subjected to MS2 analysis. For LC–RAD/MS analysis, the rat urine and bile samples were analyzed using the flow scintil- lation analyzer with 3 mL/min of Ultima Flo M scintillation cocktail. HPLC eluate of the plasma sample was fractionated using a fraction collector, and samples were mixed with 3 mL of Pico-Flour-40 scintillation cocktail (PerkinElmer) for indi- vidual analysis using the liquid scintillation counter.

Isolation and structure elucidation of ASP015K metabolites

ASP015K metabolites were isolated from rat bile and human urine samples via column chromatography followed by mass-based purification with an Agilent G6130 (Agilent Technologies). Structures of metabolites were elucidated using a fast atom bombardment mass spectrometry (FAB- MS) or LC–MS and NMR. As the matrix for FAB-MS, 50% 3-nitrobenzylalchol in dimethyl sulfoxide was used. Metabolites M1 and M3 through M12 were dissolved in dimethylsulfoxide-d6 for NMR analysis. ASP015K was also analyzed in the same manner to compare the MS and NMR spectra.

One metabolite detected in rat bile (M13) was unable to be isolated due to instability; therefore, the structure of the metabolite was estimated by mass fragmentation using LC– RAD/MS. The isolated metabolite and authentic samples of M2 were analyzed using LC–MS or LC–RAD/MS to identify the metabolite in biological samples of liver chimeric mice, the control mice, rats and human.

Results

Structure elucidation of ASP015K metabolites

Thirteen ASP015K metabolites, M1–M13, were detected in rats and humans. Ten of these metabolites (M1, M3, M4 and M6–M12) were isolated from rat bile samples, and one metabolite (M5) was isolated from pooled human urine samples. M2 was identified as ASP015K-50-O-sulfate using its authentic samples. In positive ion mode, the monoisotopic ion masses of the protonated forms were detected as follows: M1, m/z 421; M2, m/z 407; M3, m/z 343; M4, m/z 341; M5, m/z 343: M6, m/z 503; M7, m/z 519; M8, m/z 632; M9, m/z 343; M10, m/z 632; M11, m/z 519; M12, m/z 519 and M13, m/z 650. The structures of ASP015K, M1 and M3–M12 were unambiguously assigned by the acquisition and rationalization of their NMR spectra (Table 1).

Since some of the metabolites were suspected to be methyl, glucuronide and glutathione conjugates, the connect- ing positions were determined by analyses of hetero nuclear multiple bond connectivity (HMBC) and rotating frame Overhauser enhancement spectroscopy (ROESY). M1 and M4 were found to be the 7-N-methyl conjugates of ASP015K with shifting bonding patterns of r electrons on the pyrrolopyridine ring. M6, M7, M11 and M12 were found to be glucuronide conjugates, with M7, M11 and M12 also identified as the hydroxy metabolites in the hydroxyadaman- tyl group. M8 and M10 were found to be glutathione conjugates in the 3- and 2-position of the pyrrolopyridine ring, respectively. M3, M5 and M9 were elucidated to be oxidized metabolites. M13 was suspected to be a dihydrohy- droxy ASP015K–glutathione conjugate according to the mass shift. The structures of the metabolites determined in our investigation were as follows: M1, 7-N-methyl ASP015K-50- O-sulfate; M3, 60-hydroxy ASP015K; M4, 7-N-methyl ASP015K; M5, ASP015K-7-N-Oxide; M6, ASP015K-50-O-glucuronide; M7, (40S)-40-hydroxy-ASP015K-40-O-glucuro- nide; M8, ASP015K-3-glutathione conjugate; M9, 70-hydroxy ASP015K; M10, ASP015K-2-glutathione conjugate; M11, 70-hydroxy ASP015K-7-N-glucuronide and M12, 60-hydroxy ASP015K-50-O-glucuronide (Table 2).

Identification and characterization of ASP015K meta- bolites in rats using LC–RAD/MS

In the HPLC-radiochromatogram of the rat plasma, the unchanged compound and two minor metabolites (M6 and M9) were detected (Figure 2). In urine, ASP015K was the most abundant peak, with multiple minor metabolites, including M3, M4, M6, M9 and M11, also detected (Figure 3). In contrast, ASP015K was the minor component in rat bile, and M6 was the largest peak, with M3, M4, M7, M8, M9, M10, M12 and M13 detected as minor metabolites (Figure 4).
In bile samples, M1 and M11 were detected only by MS as trace amount of metabolites.

In vivo metabolite profiling of ASP015K in chimeric mice with humanized liver and control mice

No metabolites of ASP015K were observed in the control mouse plasma samples, although, three metabolites observed in humans, M1, M2 and M4, were detected in the UV chromatograms of the liver chimeric mouse plasma samples (Figure 7). In the UV chromatograms of urine and bile samples from the liver chimeric mice, several metabolites,and M2 was the most abundant component. In urine, ASP015K and M1–M5 were detected (Figure 6).

Discussion

The metabolic pathways of ASP015K are glucuronide con- jugation, methyl conjugation, sulfate conjugation, glutathione conjugation, hydroxylation of the adamantane ring and N-oxidation of 1H-pyrrolo[2,3-b]pyridine ring of ASP015K (Figure 10). The main metabolite of ASP015K was the glucuronide conjugate M6 in rats, and the sulfate conjugate M2 in humans. The biliary excretion of the glucuronide conjugate is considered to be the main elimination pathway of ASP015K in rats, and M2 also exists as the circulated metabolite in humans. Taken together, these results suggest that species differences are indeed present in the major metabolic pathways of ASP015K between rats and humans.

These results will be important finding to clarify the pharmacokinetics of ASP015K in the species. However, human faeces or bile were not examined in the current study, and so further investigation is still required.ASP015K is detected in the plasma of the control mice, liver chimeric mice and humans, with no metabolites present in the plasma of the control mice but both M1 and M2 detected in the liver chimeric mouse and human plasma. In addition, M1–M4, which are detected in human urine, are also detected in both urine and bile of the liver chimeric mice. In contrast, M3, M4 and unidentified metabolites are only observed in the urine and bile samples of the control mice. These results indicate that M1 and M2 are produced by the human hepatocytes in the liver chimeric mice. Thus, the metabolite profile of ASP015K in the liver chimeric mice is very close to that in humans, and underscoring the utility of this humanized model as reported in the past studies (Bateman et al., 2014; Kamimura et al., 2010). In addition, this model can likely be used to predict human metabolite profiles of other drug candidates as well. Further, given that the invasive nature of the bile obtaining procedure hampers sampling in humans, this model may also be useful for estimating metabolite profiles in human bile, however, this assumption should be verified by further investigations.

Of note, one human metabolite (M5) was not detected in any biological samples from the liver chimeric mice. Previous studies have also reported that metabolite profiles for certain compounds differ between the liver chimeric mice and humans. We believe these issues to be due to the influence of enzymes derived from the remaining mouse hepatocytes (Bateman et al., 2014; Kamimura et al., 2010). In the case of M5, further metabolism of the metabolite by these enzymes may have affected its detection. The liver chimeric mice should therefore be used with caution when examining the kinetic and metabolic features of drugs in mice and humans. Overall, the metabolite profile of ASP015K in the liver chimeric mice was quite informative in the development of this drug since this experiment,Peficitinib using the humanized model mice, was conducted in advance of the clinical studies.