Profiling of in vivo, in vitro and reactive zorifertinib metabolites using liquid chromatography ion trap mass spectrometry

Zorifertinib (AZD-3759; ZFB) is a potent, novel, oral, small molecule used for the treatment of non-small cell lung cancer (NSCLC). ZFB is Epidermal Growth Factor Receptor (EGFR) inhibitor that is characterized by good permeability of the blood–brain barrier for (NSCLC) patients with EGFR mutations. The present research reports the profiling of in vitro, in vivo and reactive metabolites of ZFB. Prediction of vulnerable metabolic sites and reactivity pathways (cyanide and GSH) of ZFB were performed by WhichP450™ module (StarDrop software package) and XenoSite reactivity model (XenoSite Web Predictor-Home), respectively. ZFB in vitro metabolites were done by incubation with isolated perfused rat liver hepatocytes and rat liver microsomes (RLMs). Extraction of ZFB and its related metabolites from the incubation matrix was done by protein precipitation. In vivo metabolism was performed by giving ZFB (10 mg kg−1) through oral gavage to Sprague Dawley rats that were housed in metabolic cages. Urine was collected at specific time intervals (0, 6, 12, 18, 24, 48, 72, 96 and 120 h) from ZFB dosing. The collected urine samples were filtered then stored at −70 °C. N-Methyl piperazine ring of ZFB undergoes phase I metabolism forming iminium intermediates that were stabilized using potassium cyanide as a trapping agent. Incubation of ZFB with RLMs were performed in the presence of 1.0 mM KCN and 1.0 mM glutathione to check reactive intermediates as it is may be responsible for toxicities associated with ZFB usage. For in vitro metabolites there were six in vitro phase I metabolites, three in vitro phase II metabolites, seven reactive intermediates (four GSH conjugates and three cyano adducts) of ZFB were detected by LC-IT-MS. For in vivo metabolites there were six in vivo phase I and three in vivo phase II metabolites of ZFB were detected by LC-IT-MS. In vitro and in vivo phase I metabolic pathways were N-demethylation, O-demethylation, hydroxylation, reduction, defluorination and dechlorination. In vivo phase II metabolic reaction was direct sulphate and glucuronic acid conjugation with ZFB.


Introduction
Non-small-cell lung carcinoma (NSCLC) is any type of epithelial lung cancer other than small cell lung carcinoma (SCLC). NSCLC accounts for about 85% of all lung cancers. 1 NSCLC treatment had a breakthrough in the past 15 years with tyrosine kinase inhibitors, specically with the rst generation getinib and erlotinib and, currently, with the second generation afatinib. These agents are referred to as "targeted" therapy since they target epidermal growth factor (EGF) mutations in lung cancer adenocarcinoma patients. [2][3][4] EGF is a single polypeptide of ythree amino acid residues which is contributed in the cell proliferation regulation. EGF and EGFR play a crucial role in wound healing through stimulating dermal and epidermal regeneration.
The new therapies development for improving wound healing has involved the usage of EGF. In addition, EGFR inhibitors have become a way for the cancer treatment. Thus, therapies targeting EGF/EGFR are suitable for the treatment of cancer and cutaneous wounds. 5 The rst agent developed as a target cancer therapy was EGFR inhibitor. There are two classes of EGFR inhibitors are in current use: the monoclonal antibodies (cetuximab, panitumumab, and matuzumab) that target the extracellular ligandbinding domain and small-molecule tyrosine kinase inhibitors such as ZFB which target intracellular domain. 6,7 ZFB is a novel small molecule used for NSCLC treatment through inhibition of EGFR and characterized by good permeability of the blood-brain barrier (BBB) for (NSCLC) patients with EGFR mutations. 8 ZFB can penetrate the BBB and was conrmed to be effective in both clinical and preclinical settings, and showed promising activity in patients with central nervous system (CNS) metastasis as monotherapy. 8,9 ZFB is currently at a phase II/III clinical trials. 10 Side effects of ZFB was skin and gastrointestinal disorders of occurred in 92% and 76% patients respectively. Other disorders included hepatobiliary and renal disorders (13%), asthenia (7%), infections and infestations (7%), and metabolism and nutrition disorders at 4% and 7%, respectively. 11 ZFB chemical name is 4-((3-chloro-2-uorophenyl)amino)-7methoxyquinazolin-6-yl (R)-2,4-dimethylpiperazine-1carboxylate. ZFB chemical structure contains N-methyl piperazine and phenyl amine rings that could undergo metabolic bioactivation generating iminium and iminoquinone reactive intermediate, respectively that can be captured by nucleophile such as potassium cyanide (KCN) and GSH, respectively forming cyano adducts and GSH conjugates, respectively that are stable and could be identied by liquid chromatography ion trap mass spectrometry (LC-IT-MS). Extraction, identication, separation and characterization of these stabilized adducts and conjugates were performed using LC-IT-MS. [12][13][14] These reactive metabolites helped us to predict toxicities associated with ZFB.
Our research team previously established LC-MS/MS analytical method for ZFB quantication with application to assess its metabolic stability which revealed the moderate extraction ratio and the good bioavailability of ZFB. 15,16 This research will focus on the proling of ZFB metabolites either in vitro or in vivo using LC-IT-MS. Metabolic bioactivation of a drug to reactive intermediates that can covalently modify proteins leading to druginduced organ toxicities. Hence, we will check the reactive intermediates that will aid in predicting the reasons for toxicities associated with these drugs. Reactive intermediates cannot be checked in vivo as a result of as once they are formed, they will bind to endogenous materials as proteins or DNA that prevents the detection by LC-IT-MS. 17

Chemicals and instruments
Zorifertinib was purchased from Med Chem Express LLC company (Princeton, NJ, USA). Reference powders (analytical grade) and solvents (HPLC grade) were used. Sprague Dawley rats were obtained from Health Research Center of Prince Naif bin Abdul Aziz at King Saud University (Riyadh, KSA). Rat liver microsomes (RLMs) were prepared in our laboratory following a published protocol using Sprague Dawley rats. 18,19 Acetonitrile (ACN), nicotinamide adenine dinucleotide phosphate (NADPH), formic acid (HCOOH), potassium cyanide (KCN) and glutathione reductase (GSH) were purchased from Sigma-Aldrich company (St. Louis, MO, USA). Water (HPLC grade) was supplied by Milli-Q plus purication system (Billerica, MA, USA) that is available in-house. Tween 80 were obtained from Eurostar Scientic Ltd. (Liverpool, UK). Ethical approval for the Animal experiments of were obtained from the Animal Ethics Committee at King Saud University (No. KSU-SE-19-52). The separation of ZFB and its related metabolites and adducts was done on an Agilent HPLC 1200 series that consists of G1367B HIP-ALS auto sampler, G1322A degasser, G1316 thermostated column compartment and G1311A binary pump that was connected using an electrospray ionization (ESI) to an Agilent 6320 Ion Trap (Agilent Technologies, Palo Alto, CA, USA). Mass Hunter soware (Agilent Technologies, Palo Alto, CA, USA) was used to regulate the data acquisition and instruments.

Chromatographic conditions
The analytical parameters that were used for separation of ZFB and its related metabolites are listed are mentioned in Table 1. Agilent eclipse plus C18 analytical column (150 mm Â 2.1 mm, 3.5 mm particle size) (Agilent Technologies, Palo Alto, CA, USA). Column temperature was kept constant at 23 AE 2 C. The most appropriate chromatographic conditions were attained at a ow rate of 0.25 mL min À1 with a gradient system for 65 min. Mobile phase composed of 0.1% HCOOH (pH: 3.2) and ACN. The post time was 15 min. Sample injection volume was 10 mL. Mass parameters were optimized for ZFB. Dissociation for ZFB and related metabolites was done using collision-induced dissociation (CID) in the collision cell. Our target is to recognize the susceptibility of metabolism key sites, as shown by the low value of the site lability that was revealed by the composite site lability (CSL) and also the proposed major isoforms regioselectivity that was supposed to be responsible for metabolism. The results from the WhichP450™ soware, displayed by the pie chart used for indication of most probably cyp450 isoform that has the major role in ZFB metabolism. These predictions were used as a director during proling of ZFB in vivo and in vitro metabolites. 20,21 2.4. In silico of ZFB reactive intermediates and toxicity prediction using XenoSite reactivity model and DEREK soware XenoSite reactivity model is freely available online at https:// swami.wustl.edu/xenosite that was used to detect the labile sites that could produce reactive intermediates that could form complex with GSH, cyanide, DNA and protein stable adducts. [22][23][24] This in silico model is based on neural networks of around 680 molecules. 25 This free online soware characterized by short running time depending on the molecule size. 21 SMILES format (CC(C)Oc1ccc(cc1)NC(]O) N2CCN(CC2)c3c4cc(c(cc4ncn3)OCCCN5CCCCC5)OC) of ZFB was uploaded to XenoSite reactivity model that is an online website. The outcomes were used in searching and identifying reactive metabolites of ZFB. DEREK module (StarDrop soware package) was utilized for screening for ZFB and its related metabolites structural alerts and to verify our bioactivation pathway proposal. DEREK soware can be used to propose structural modication at the bioactive sites that may stop the initiation of toxicity sequence. The chemical structure of ZFB and its related metabolites (SMILES format) was uploaded on the program soware for the toxicity prediction.

2.5.
In vitro metabolism 2.5.1. RLMs incubations. RLMs were incubated with ZFB following a previously published procedure. 26,27 Thirty mM ZFB in DMSO was added to 0.08 M phosphate buffer (pH 7.4) with RLMs (1 mg mL À1 ). The incubation mixture was rst kept in a shaking water bath for 5 min at 37 C. Incubation was initiated by NADPH (1 mM) solution. The nal volume of incubation mixture is 1 mL and the total time of incubation is 90 min. Controls were prepared either by replacing RLMs or NAPDH with buffer. The metabolic reaction was stopped adding 2 mL ice-cold acetonitrile then centrifugation at 9000g for 10 min for protein precipitation. The supernatant was removed to clean vials for evaporation using a stream of nitrogen as inert gas. The residue was solubilized in a mixture of 0.1% HCOOH and ACN in a ratio similar to mobile phase and transferred to HPLC vial for analysis. To trap reactive intermediates, the same  Similarly, negative controls were done using the above steps. Phenytoin (2 mM) was used to evaluate the activity of RLMs as the positive control.

Liver hepatocytes incubation.
Rat liver hepatocytes (RLHs) will be done following the established steps. 28 RLHs incubation experiment with ZFB (30 mM) was performed in a rotary evaporation instrument tted with a custom made 4way adaptor to permit the connections of four 100 mL round bottom asks. The RLHs incubations were performed at 37 C for 2 hours. One mL aliquots (at 0, 15, 30, 60, 90, and 120 minutes were taken from each ask, mixed with 2 mL of ACN and immediately frozen at À70 C refrigerator). Cell viability was retested using trypan blue at the end of the experiment. Stored frozen samples were thawed and then centrifuged at 14 000 rpm. Supernatants were transferred to new vials and evaporated under a stream of inert gas nitrogen. The residue was reconstituted in 300 ml of the mobile phase in HPLC vials for LC-MS analysis. Negative control experiments were done following the above steps.

In vivo metabolism
Four male one month Sprague-Dawley rats of average weight (400 g) were housed individually in a metabolism cage that was kept in animal care facility in a 12 h light/dark cycle (7:00-19:00). Rats had free access to standard animal food and water. Rats were kept in metabolism cages for 72 h before the start of the experiment. ZFB was solubilized in a special solution (4% DMSO, 5% Tween 80 and 30% PEG 300, in HPLC H 2 O) to allow ZFB dispersion and to maintain the drug solubility. The ZFB

In silico prediction of ZFB metabolites
The composite site lability for ZFB is 0.9979 that indicates the high lability of ZFB to metabolism. The metabolic Landscape indicates that atoms of C32 and N4 of N-methyl piperazine ring and C27 of phenyl group are the most labile atoms to metabolism, while C3 and C5 (the a carbons) of the N-methyl piperazine ring and C18 of methoxy group are moderate labile atoms to metabolism that matched with the current experiments ( Fig. 1). Cyp3A4 had the main role in ZFB metabolism. Consecutively, depending on the literature knowledge and in silico postulations, a list of predicted reactive intermediates and metabolites was prepared and used as a guide during practical experiments.

In silico ZFB bioactivity and toxicity prediction
The ZFB atomic sites were proposed for generation of electrophiles that could be trapped by cyanide    Fig. 2A.
In silico toxicity assessment of ZFB metabolites was done using DEREK module (StarDrop soware package) (Fig. 2B).
ZFB and its metabolites show equivocal nephrotoxicity and equivocal mutagenicity due to halogenated benzene and substituted 4-anilinoquinazoline, respectively (Fig. 2B). M4 shows plausible HERG channel inhibition due to 2-aminophenol derivative. Table 2 shows a complete list of in vitro and in    vivo ZFB metabolites with DEREK outcomes for supposed toxicity prole.

MS/MS fragmentation behavior of ZFB
The ZFB precursor ion peak (PIP) appeared at 29.9 min in the total ion chromatogram (TIC). Dissociation of precursor ion (PI) at m/z 460 resulted in three characteristic fragment ions at m/z 346, m/z 320 and m/z 141 (Fig. 3A). The dissociation of ZFB targets the weak carboxylate link between the piperazine and quinazoline moieties inside its chemical structure. Fragment ions at m/z 320 and m/z 141 represent the two building blocks of ZFB. Fragment ion at m/z 141 was used for conrmation of the metabolic reaction at piperazine ring while m/z 320 for the remaining chemical structure (Fig. 3B). 30

Identication of ZFB in vitro metabolites
Puried extracts recovered from RLMs incubates were injected into LC-IT-MS. Six phase I metabolic reactions produced by Ndemethylation, O-demethylation, hydroxylation, reduction, aoxidation and dechlorination. One sulphate conjugate and two glucuronic acid conjugates were characterized aer incubation of ZFB with isolated perfused RLHs as phase II metabolite (Table 3). In vitro phase I metabolites (M2, M3, M4, M5 and M6) gures and schemes have been moved to a ESI (Fig. S1-S5 †). In vitro phase II metabolites (M9) gures and schemes have been moved to a ESI (Fig. S6 †). 3.4.1. Identication of the M1 metabolite of ZFB. The M1 PIP appeared at 30.3 min in the TIC. Dissociation of PI at m/z 446 resulted in two characteristic fragment ions at m/z 320 and m/z 100 (Fig. 4A). Comparing to the dissociation of ZFB, fragment ion at m/z 320 indicates that no metabolic reaction in part B while part A shows a 14 m/z less that indicates the N-demethylation of piperazine ring (Fig. 4B). Radical cation (m/z 100) was formed in small intensity due to collision induced dissociation (CID) in LC-IT-MS. 31 3.4.2. Identication of the M7 metabolite of ZFB. The M7 PIP appeared at 36.5 min (m/z 526) in the TIC. PIP exhibits 80 m/ z higher than that of the M1 (precursor metabolite) with m/z 446 that was predicted to be sulfate conjugation of M1. Dissociation of PI at m/z 526 resulted in two characteristic fragment ions at m/z 508 and m/z 446 (Fig. 5A). Comparing to the dissociation of ZFB, 32 fragment ion at m/z 446 indicates that the metabolic reaction in part B (14 m/z less) while part A shows no metabolic reaction that indicates the O-demethylation of methoxy group attached to quinazoline moiety. Fragment ion at m/z 508 (18 m/z less) that indicates water loss and conrm the O-demethylation metabolic reaction. M7 was proposed to be a direct conjugation of M1 with sulfate ion (Fig. 5B). Another LC-IT-MS screening for sulfate adduct was done by neutral loss scan of ions that lose 80 Da (Fig. 5C). 33 3.4.3. Identication of the M8 metabolite of ZFB. The M8 PPI appeared at 26.7 min in the TIC. Dissociation of PI at m/z 622 resulted in three characteristic fragment ions at m/z 605, m/ z 446 and m/z 430 (Fig. 6A). It showed 176 m/z higher than that of PI with m/z 446 (M2), which proposed to be a glucuronic acid conjugate of ZFB metabolite. Fragment ion at m/z 446 that showed neutral loss of glucuronic acid (loss of 176 m/z units) that matched with the other fragment ions at m/z 430 and m/z 605 ( Fig. 6B and C). Constant neutral loss scan for M8 metabolite conjugates are more specic (due to the characteristic loss of 176 mass units), and provide much cleaner mass spectral data. 32 3.5. Identication of ZFB in vivo metabolites PIP mass spectra comparison between urine extracts with control urine samples as well as PIP comparison of ZFB proposed metabolites (Table 4) allowed the proling of six in vivo phase I and three in vivo phase II metabolites. Metabolic pathways for in vivo phase I metabolites were supposed to be Ndemethylation, hydroxylation, reduction, dechlorination and a oxidation, while for phase II metabolites were the result of two glucuronic acid conjugation and one sulphate (Table 4). M1 metabolites is previously mentioned in in vitro ZFB phase I metabolism. M7 and M9 metabolites are previously mentioned in in vivo ZFB phase II metabolism. In vivo phase I metabolites (M10, M11, M12, M13 and M14) gures and schemes have been moved to a ESI (Fig. S7-S10 †). The same metabolic incubation (ZFB with RLMs) was done in the presence of 1.0 mM KCN and 1.0 mM GSH to trap iminium and electro decient conjugated system intermediates, respectively. Four GSH conjugates and three cyanide adducts were identied ( Table 5), revealing that the N-phenyl amine group and methyl piperazine ring in ZFB can become bioactivated and then trapped by the nucleophile GSH and cyanide ion. 34 ZFB cyano adducts (AZDCN515, AZDCN501) and ZFB GSH conjugates (AZDGSH761, AZDGSH763, AZDGSH717) gures and schemes have been moved to a ESI (Fig. S11-S15 †).
3.6.1. Identication of AZDCN485 cyano adduct. The AZDCN485 PIP appeared at 46.6 min in the total ion chromatogram (TIC). Dissociation of PI at m/z 485 resulted in two characteristic fragment ions at m/z 458 and m/z 168 (Fig. 8A). Fragment ion at m/z 458 (27 m/z less) that indicates HCN loss and conrms the cyano adduct formation. Comparing to the dissociation of ZFB, fragment ion at m/z 168 indicates that cyano nucleophile trapped the bioactivated piperazine ring (part A) forming cyano adduct (Fig. 8B).
3.6.2. Identication of AZDGSH715 GSH conjugates. The AZDGSH715 PIP appeared at 42.4 min in the TIC. Dissociation of PI at m/z 715 resulted in two characteristic fragment ions at m/z 640 and m/z 586 (Fig. 9A). Fragment ion at m/z 586 (129 m/z less) that indicates anhydroglutamic acid moiety and conrms the formation of GSH conjugate. Constant neutral loss scan for screening for GSH adduct was done by monitoring of ions that lose 129 Da (Fig. 9B). 32 AZDGSH715 formation indicated the phenyl ring bioactivation in in vitro metabolism of ZFB. The metabolic reactions that occurred in AZDGSH715 were proposed dechlorination, N-demethylation, deuorination then hydroxylation and GSH conjugation at bioactivated phenyl ring (iminoquinone derivative) (Fig. 9C).

Proposed pathways of bioactivation of ZFB
Bioactivation pathways of ZFB are proposed in Fig. 10. AZDCN485, AZDCN515 and AZDCN501 cyanide adducts revealed the metabolic formation of reactive iminium intermediates at N-methyl piperazine ring in ZFB metabolism. The bioactivation pathway was predicted as hydroxylation metabolic reaction at N-methyl piperazine ring in ZFB followed by loss of H 2 O (dehydration) that resulted in iminium ions intermediates (reactive hard nucleophile) that were trapped using cyanide electrophile (KCN) forming stable cyanide adducts that were characterized and identied using LC-IT-MS. This bioactivation pathway that involves iminium intermediates formation is previously reported with other drugs containing cyclic tertiary amine ring. 35,36 The formation of AZDGSH715, AZDGSH761, AZDGSH763 and AZDGSH717 supposed the formation of iminoquinone reactive intermediates in the metabolism of ZFB that were formed by dechlorination, deuorination and hydroxylation of phenyl group and then oxidation 34 generating reactive intermediates that were trapped using GSH leading to the formation of stable conjugates that were characterized and identied using LC-IT-MS (Fig. 10). 37

Conclusions
In silico soware (WhichP450™ module, DEREK and XenoSite reactivity model) has proposed the most labile atoms in the ZFB structure that revealed to be the same as seen in the detected metabolites in phase I metabolism and the same groups were bioactivated as expected. Three in vitro phase II metabolites (sulphate and glucuronic acid conjugates) and six in vitro phase I (N-demethylation, O-demethylation, hydroxylation, reduction, and oxidation) and were found for ZFB. Three in vivo phase II metabolites (one sulphate and two glucuronic acid conjugates) and six in vivo phase I metabolites (N-demethylation, a-oxidation, hydroxylation, reduction, and dechlorination) of ZFB were identied (Fig. 11). Seven reactive intermediates (four GSH conjugates and three cyano adducts) were characterized and the mechanisms of their formation were proposed (Fig. 11) that may illuminate the way for guring out the causes behind ZFB toxic side effects.

Ethics approval
Ethical approval for the Animal experiments of were obtained from the Animal Ethics Committee of King Saud University (No. KSU-SE-19-52).

Conflicts of interest
The authors declare no conicts of interest for the current work.