Liquid chromatography–tandem mass spectrometry metabolic profiling of nazartinib reveals the formation of unexpected reactive metabolites

Nazartinib (EGF816, NZB) is a promising third-generation human epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor. This novel irreversible mutant-selective EGFR inhibitor targets EGFR containing both the resistance mutation (T790M) and the activating mutations (L858R and Del19), while it does not affect wild-type EGFR. However, the metabolic pathway and bioactivation mechanisms of NZB are still unexplored. Thus, using liquid chromatography–tandem mass spectrometry, we screened for products of NZB metabolism formed in vitro by human liver microsomal preparations and investigated the formation of reactive intermediates using potassium cyanide as a nucleophile trap. Unexpectedly, the azepane ring was not bioactivated. Instead, the carbon atom between the aliphatic linear tertiary amine and electron-withdrawing system (butenoyl amide group) was bioactivated, generating iminium intermediates as reactive species. Six NZB phase I metabolites, formed by hydroxylation, oxidation and N-demethylation, were characterized. Moreover, two reactive iminium ions were characterized and their corresponding bioactivation mechanisms were proposed. Based on our results, we speculate that bioactivation of NZB can be blocked by small sterically hindering groups, isosteric replacement or a spacer. This approach might reduce the toxicity of NZB by avoiding the generation of reactive species.


Introduction
Non-small-cell lung cancer (NSCLC) encompasses a heterogeneous group of lung cancer subtypes [1][2][3][4][5], which affects 90% of patients with lung cancer [6]. This class of lung cancer is associated with several mutations, such as those in human epidermal growth factor receptor (EGFR). Tyrosine kinase inhibitors (TKIs) regulate the activity of human EGFR and have become the standard treatment for patients suffering from advanced EGFR-mutant NSCLC. The first-generation EGFR TKIs (e.g. gefitinib and erlotinib) bind reversibly and competitively to the ATP-binding site of the EGFR tyrosine kinase (TK) domain, which improves the outcome of NSCLC patients bearing EGFR-activating mutations (L858R and Del19) [7,8]. However, after satisfactory responses for a period, patients' tumours acquired resistance to first-generation TKIs because of the development of a T790M mutation, which affects the ATP-binding site of the human EGFR [9][10][11][12].
Thus, second-generation EGFR TKIs (e.g. avitinib and dacomitinib) were designed to target tumours with T790M mutation and EGFR-activating mutations. These compounds showed promising anti-T790M activity in laboratory experiments. However, their clinical activity towards T790M-associated NSCLC was limited because of their inhibitory effects on wild-type EGFR, which resulted in toxicity and a narrow therapeutic index [13][14][15]. More recently, third-generation EGFR TKIs (e.g. osimertinib and nazartinib (NZB)) were developed. They irreversibly and selectively target EGFR with T790M and other mutations, whereas they have little effect on wildtype EGFR activity [13,14]. Third-generation EGFR TKIs were developed to overcome EGFR T790M-mediated resistance to first-and second-generation EGFR TKIs with minor toxicity. Thirdgeneration EGFR TKIs combine effectiveness against NSCLC that is resistant to both first-and second-generation EGFR TKIs [16,17]. Osimertinib, for example, is approved by both the American and European regulatory agencies for the management of patients with metastatic EGFR T790M NSCLC [18]. Pre-clinical data show that NZB, another third-generation EGFR TKI [19], does not affect wild-type EGFR activity and presents selectivity against mutated EGFR, similar to other third-generation EGFR TKIs. Nevertheless, it presents some side effects, such as diarrhoea, pruritus and rash [20].
In addition to the drug itself, by-products of detoxification pathways may be responsible for such adverse effects in patients. Detoxification involves metabolic reactions that transform endogenous compounds and xenobiotics, increasing their polarity to be excreted from the human body. Although metabolites usually exhibit less toxicity than their parents, in some cases, bioactivation may generate reactive intermediates that are more toxic than the unmetabolized molecules [21][22][23]. Reactive intermediates are unstable and can modify DNA and proteins by the formation of covalent bonds, which is considered the initial step in drug-induced organ toxicity [24,25]. Thus, the identification of generated reactive metabolites is crucial for understanding drug-induced toxicity. However, reactive metabolites are usually generated by phase I metabolic pathways and their identification is hindered by their transient nature. To overcome this limitation, a nucleophile can be used to capture reactive intermediates, and the resulting adducts can be characterized and identified by mass spectrometry technique [26,27].
The chemical structure of NZB (N- figure 1) contains two tertiary nitrogen atoms (an azepane ring and a terminal dimethylamino group) that can be bioactivated, generating iminium ion intermediates [28][29][30][31]. The formation of unstable intermediates reveals side effects of NZB as was approved with similar drugs. Cyclic tertiary amine rings can perform bioactivation by iminium ion generation [28][29][30][31]. These intermediates react poorly with glutathione; however, they can be trapped using potassium cyanide [21,28,29]. The obtained reactive iminium intermediates trapped efficiently using cyanide to form cyano conjugates can be characterized by mass spectrometry [26][27][28]32,33]. Moreover, although the azepane ring was expected to undergo bioactivation during NZB metabolism, this does not occur. Instead, the carbon between the aliphatic linear tertiary amine and the unsaturated conjugated system are bioactivated.
It is hypothesized that these reactive metabolites might be responsible for the side effects of NZB. However, there are no reports on specific metabolic pathways associated with the bioactivation mechanism of NZB. Thus, the aim of this work was to use in vitro experiments to characterize the bioactivation pathways of NZB that form reactive intermediates. To do so, we used a scavenging molecule ( potassium cyanide) to trap reactive intermediates of NZB metabolism. This approach was used because when reactive metabolites form in vivo, they bind to DNA and proteins via covalent bonds and hence cannot be detected [24,27,32].

Chromatographic conditions
Resolution and identification of in vitro NZB metabolites and its related cyano adducts from the HLM incubation mixtures was performed on an Agilent Triple Quadrupole system comprising an Agilent rapid resolution liquid chromatography (RRLC) 1200 as an HPLC system and an Agilent 6410 triple quadrupole (QqQ) as a mass detector (Agilent Technologies, Palo Alto, CA, USA) with an electrospray ionization (ESI) source. Chromatographic resolution of the metabolic mixtures components was done on a C 18 column (length, 150 mm; internal diameter, 2.1 mm; and particle size, 3.5 µm). The column temperature was fixed at 22 ± 1°C, and we used a gradient mobile phase at a flow rate of 0.2 ml min −1 and consisting of 10 mM ammonium formate (solvent A; pH 4.2) and acetonitrile (solvent B). The gradients steps involved solvent B (5%; 0-5 min), solvent B (5-50%; 5-35 min), solvent B (50-90%; 35-50 min) and solvent B (90-5%; 50-60 min), with a post time of 15 min. The sample injection volume was 10 µl. The run time was 60 min, with the chromatographic and mass parameters preoptimized for NZB. The generation of daughter ions (DIs) of NZB metabolites and cyano adducts was done in the collision cell by collision-induced dissociation (CID). Mass analysis was performed on a mass detector using positive ESI source. Nitrogen (N 2 ) was used as drying gas at a flow rate of 11 l min −1 , and as collision gas at a pressure of 55 psi. Capillary voltage, source temperature, fragmentor voltage and collision energy were set to 4000 V, 350°C, 140 V and 18 eV, respectively. Agilent Mass Hunter software was used for controlling instrument and data acquisition.

Human liver microsomes incubation
We first exposed HLMs to several NZB concentrations (2-30 µM) and found that the composition of metabolites did not vary within this range. However, the concentration of metabolites increased as the concentration of NZB increased. Thus, 30 µM was used in all experiments to increase the yield of metabolites and make their characterization easier. The screening of NZB metabolites was performed in vitro by incubating NZB (30 µM) with HLMs (1.0 mg ml −1 ) in phosphate buffer (50 mM at pH 7.4) and MgCl 2 (3.3 mM) for 120 min at 37°C in a shaking water bath. The in vitro metabolization of NZB was stimulated by the addition of NADPH (1.0 mM) and terminated by the addition of ice-cold acetonitrile [34,35]. The same HLM incubation experiment was repeated in the presence of potassium cyanide to capture the reactive intermediates. All reactions were performed in triplicate to verify the  [36][37][38]. Controls were prepared following the same steps except the addition of the drug or NADPH.

Identification of NZB reactive intermediates
Full mass spectrometry scans and extracted ion chromatograms of the detected mass to charge ratio (m/z) peaks were used to identify the in vitro metabolites in the incubation mixtures. Molecular ions were used as parent ions (PIs) for fragmentation into daughter ions (DIs). The fragmentation behaviour was used to characterize the reactive metabolites formed during NZB metabolism by HLMs in vitro.

Reactive metabolites
In addition to the metabolites described above, two cyano adducts were characterized, indicating the generation of reactive intermediates in NZB metabolism by HLMs.

Bioactivation mechanism of NZB
The characterization of the NZB506 and NZB520 cyano adducts revealed the generation of reactive iminium intermediates in NZB metabolism. The hydroxylation of the bioactivated carbon in NZB followed by dehydration resulted in the generation of reactive iminium electrophiles that were captured by a cyanide nucleophile to form a stable cyano adduct (scheme 10). The bioactivation pathway for the formation of reactive intermediates has been previously studied using drugs containing cyclic tertiary amines. However, herein, the reactive intermediates were generated by bioactivation of an aliphatic noncyclic carbon attached to a tertiary amine rather than by azepane bioactivation [39][40][41][42][43][44].

Conclusion
The current study provided experimental evidence to support further work on NZB toxicity. Six in vitro NZB phase I metabolites and two cyano adducts were identified (figure 8) and bioactivation mechanisms were proposed. The knowledge on bioactivation mechanisms is crucial for determining the chemical groups involved in bioactivation. This information may be used for the development of new molecules containing small sterically hindering groups, isosteric replacement or a spacer to prevent NZB bioactivation; inhibiting the generation of reactive species in this way would result in reduced toxicity. The data obtained in this study will contribute towards the development of new drugs with enhanced safety profiles.
Ethics. The study's design (in vitro assays using commercially available liver microsomes) exempts it from the approval by Ethics Committees.
Data accessibility. The data supporting the results in this article can be accessed at the Dryad Digital Repository: https:// doi.org/10.5061/dryad.j5m8h10 [45].