Introduction

Furmonertinib (Alflutinib, AST2818), a newly developed third-generation epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor, is designed to treat non-small-cell lung carcinoma with confirmed EGFR T790M mutation. This inhibitor has a desirable antitumor efficacy with a relatively wide therapeutic range (40–240 mg) [1]. Furmonertinib is mainly metabolized by CYP3A4 in human liver, forming the main N-demethylated metabolite AST5902 (Fig. 1a) [2]. However, the clearance of furmonertinib in multiple-dose administrations is dose and time dependent. Liu et al. suggested that this condition may be caused by CYP3A4 self-induction. They observed that furmonertinib has an induction potential comparable with that of rifampin, exhibiting a similar EC50 (0.25 µM) of the mRNA expression of induced CYP3A4; by contrast, AST5902 has a relatively weaker effect (Figs. 1b, c) [2]. In the past decades, rifampin has been used as a potent positive model drug to evaluate CYP3A4 induction potential. Based on Liu’s research, furmonertinib can also be a potential positive model drug. However, robust evidence is needed to prove furmonertinib as a model positive CYP3A4 inducer.

Fig. 1: mRNA expression changes of CYP3A4 induced by furmonertinib and its main metabolite AST5902.
figure 1

a Chemical structures of furmonertinib and AST5902; b dose-induction curves of the furmonertinib-induced mRNA expression levels of CYP3A4 in three different donors; and c dose-induction curves of the mRNA expression levels of CYP3A4 induced by AST5902 in three different donors [2].

Full size image

According to the 2020 final US Food and Drug Administration (FDA) guide for industry entitled In Vitro Drug Interaction Studies—Cytochrome P450 Enzyme- and Transporter-Mediated Drug Interactions (https://www.fda.gov/media/134582/download), changes in mRNA and enzyme activity levels should be evaluated in in vitro studies to determine whether a drug is a latent inducer of specific drug-metabolizing enzymes. Commonly, xenobiotics activate orphan nuclear receptors, such as pregnane X receptor (PXR) or constitutive androstane receptor, which can bind to the elements of the promoter region of cytochrome P450 enzymes and induce enzyme expression transcriptionally [3, 4]. Thus, changes in mRNA expression of targeted proteins are usually required. The metabolism of probe drugs is commonly measured to estimate specific enzyme activities. Testosterone, whose metabolite is 6β-hydroxytestosterone, is a typical probe drug of CYP3A4 and often quantified for the evaluation of the enzyme’s activity. Changes in the protein expression of drug-metabolizing enzyme and transporter (DMET) have been widely explored because of several considerations. For instance, the increase in mRNA levels has a poor correlation with enzyme activities. In addition, enzyme activity can be influenced by multiple factors. Certain chemicals can inhibit enzymes but also act as drug inducers [5]. The conventional way of protein quantification, Western blotting, is a semi-quantification method with a low throughput. Furthermore, the high sequence homology among DMET proteins and lack of specific antibodies result in their limited use in research and industrial laboratories.

Although proteomic quantification has been widely used in inter-individual studies with the development of liquid chromatography-mass spectrometry (LC-MS) [6,7,8,9,10], most of them involve the isolation of human liver microsomes directly from liver tissues or a considerable number of cells [5]. In in vitro studies, plated primary human hepatocytes (PHHs) are often used to evaluate drug induction potential. However, extracting microsomes from plated cells is challenging because of the limited number of cells in each well, especially when several concentrations of the test drug or different drugs should be evaluated simultaneously. Another significant challenge is the measurement of enzyme activity and protein expression in the same well in 96- or 48-well cell culture plates to obtain a distinct correlation between protein expression and enzyme activity.

This study explored the induction mechanism and potential of furmonertinib in plated PHHs in terms of several aspects. Savaryn et al. [11] developed an enrichment-free method to quantify cytochrome P450 proteins in 96-well-plated cells. With modifications of this method, we performed simultaneously the absolute quantification of CYP3A4 proteins and measured the activity of the enzyme with cells from the same well. Using this evaluation model, we tested furmonertinib. We also compared species differences in terms of CYP3A4 induction by furmonertinib between plated PHHs and rat/mouse primary hepatocytes.

Materials and methods

Materials

A dual luciferase® reporter assay system and sequencing-grade trypsin were purchased from Promega (Madison, WI, USA). The following reagents or assay kits were purchased from Thermo Fisher Scientific (Waltham, MA, USA): Lipofectamine™ 3000 transfection reagent, type IV collagenase, rat tail collagen I, Dulbecco’s modified Eagle’s medium (DMEM), William’s E medium, Gibco Opti-Minimum Essential Medium-reduced serum medium, and fetal bovine serum. A Matrigel matrix was procured from Corning (NY, USA). A gelatine solution (0.1%), testosterone, and furmonertinib were obtained from Sigma-Aldrich (St. Louis, MO, USA). AST5902 was a kind gift from Prof. Da-fang Zhong’s laboratory (Shanghai Institute of Materia Medica, Chinese Academy of Sciences). Recombinant human cytochrome P450 enzymes were purchased from Cypex (Scotland, UK). TRIzol reagent was bought from Life Technology (CA, USA). PrimeScript RT Master Mix was obtained from Takara (Tokyo, Japan). Hieff qPCR SYBR Green Master Mix and sodium dodecyl sulfate (SDS) loading buffer were from Yeasen Biotech (Shanghai, China). The internal standard d7-6β-hydroxytestosterone was procured from Corning. High-performance LC-grade formic acid, dithiothreitol, and ammonium bicarbonate were supplied by Sinopharm (Shanghai, China). Iodine acetamide was purchased from Meilun Biotechnology (Dalian, China). Surrogate peptides and isotopic internal standards were synthesized by ChinaPeptides (Shanghai, China). Cryopreserved PHHs (lot #GKJ, Caucasian, male, 52 years old; lot #QBU, Caucasian, male, 50 years old; and lot #HJK, Hispanic, female, 30 years old) were purchased from BioIVT (Baltimore, MD, USA). Deionized water was prepared with a Synergy UV-R system (Millipore, Molsheim, France) in our laboratory.

PXR luciferase assay

The luciferase reporter expression plasmids pcDNA3.1-RXRα, pcDNA3.1-PXRα, pGL3-CYP3A4-XREM-Luc, and plasmid pRL-SV40 containing the Renilla luciferase gene were kind gifts from Prof. Li-kun Gong’s laboratory [12] (Shanghai Institute of Materia Medica, Chinese Academy of Sciences). HEK-293T cells were cultured in 48-well plates (plated with 0.1% gelatine in advance) for luciferase assay experiments. They were cultured overnight and transfected with Lipofectamine 3000 transfection reagent in accordance with the protocol provided. After 12 h of transfection, the medium was replaced with DMEM containing different inducers. Cell extracts were collected 24 h after induction, and luciferase activity was measured with the dual luciferase reporter assay system by using the Renilla luciferase activity as a normalization standard.

Isolation, culture, and induction of primary rat/mouse hepatocytes

Primary rat/mouse hepatocytes were isolated from Sprague–Dawley rats or C57BL/6 mice by using a two-step collagenase digestion method with some modifications [13]. Once isolated, 5 × 104 cells were seeded into each well of a 96-well cell culture plate (covered with type I collagen in advance). For real-time PCR and Western blot, the cells were seeded into 24- and 6-well cell culture plates, respectively. After 4 h of attachment, the medium was replaced with ice-cold medium containing 0.2 mg/mL Matrigel and cultured overnight to allow the cells to stabilize and form a sandwich culture model. In the next 3 days, the cells were treated with the test compounds or a vehicle (0.1% dimethyl sulfoxide [DMSO]) and incubated at 37 °C and 5% CO2. The medium was changed and replenished every 24 h. All the samples were prepared in triplicate. After the treatment, total RNA was extracted, and cDNA was synthesized. Protein samples were collected with RIPA lysis buffer containing 1% protease inhibitor cocktail. Real-time PCR was conducted to measure the changes in mRNA expression. The forward and reverse primers of mouse CYP3A11 were 5ʹ-AACTGCAGGATGAGATCGATGAG-3ʹ and 5ʹ-TTCATTAAGCACCATATCCAGGTATT-3ʹ, respectively. Mouse β-actin was used as an internal standard. Western blot was carried out to quantify the furmonertinib-induced changes in CYP3A11 in mouse hepatocytes at different concentrations. Samples were prepared in accordance with previously described protocols [14]. They were electrophoresed on 10% SDS-PAGE gels and then transferred to polyvinylidene difluoride membranes. After they were probed with anti-CYP3A11 and anti-GAPDH, images were captured with a CLINX ChemiScope 3300 mini system. Western blot experiments were repeated three times.

Culture of primary human hepatocytes

Cryopreserved PHHs were thawed in accordance with the provided protocol. Cell viability was determined with trypan blue. Only the cells with a viability of >85% were used. A total of 1.4 × 105 viable hepatocytes were plated in each well of 48-well cell culture plates coated with type I rat tail collagen and incubated at 37 °C under 5% CO2. After 4 h, the cells were overlaid with 0.2 mg/mL Matrigel in an ice-cold medium to form a sandwich configuration and cultured overnight to stabilize. The following procedures were similar to those used for primary rat/mouse hepatocytes, as described above.

Cytochrome P450 enzyme activity assay via LC-MS/MS

Metabolite formation was measured via LC-MS/MS in a multiple-reaction monitoring (MRM) mode to compare the enzyme activities among different inducers in primary human, mouse, and rat hepatocytes. Testosterone was used as a probe drug to determine the CYP3A4 activity in human hepatocytes, CYP3A11 activity in mouse hepatocytes, and CYP3A1 activity in rat hepatocytes. 6β-Hydroxytestosterone, one of the main metabolites of testosterone, was also quantified. Primary human, mouse, and rat hepatocytes were incubated with William’s E medium containing 200 µM testosterone for 1 h in duplicate. An accurate volume of supernatant was collected and mixed with the quench solution (methanol:acetonitrile = 1:1, containing the internal standard). All the samples were vortexed and centrifuged at 15,000 × g for 10 min before LC-MS/MS analysis. An Agilent 6495 triple quadrupole mass spectrometer coupled with a 1290 Infinity ultra-high-performance liquid chromatography (UPLC) system was used to quantify the metabolite concentrations. Chromatography was performed with a gradient elution by using an Acquity UPLC HSS T3 column (50 mm × 2.1 mm, 1.8 µm, Waters, CA, USA). Tables 1 and 2 summarize the details of the gradient and mass transitions, respectively.

Table 1 LC-MS/MS method details.
Full size table
Table 2 MRM transitions for metabolite quantification and protein quantification.
Full size table

Absolute protein quantification of CYP3A4 in human hepatocytes treated with different inducers

The methods described by Savaryn et al. [11] were optimized to absolutely quantify CYP3A4 protein in plated PHHs. In brief, the medium was used to measure enzyme activities, and 50 mM ammonium bicarbonate coupled with 100 mM iodoacetamide and 100 mM dithiothreitol was added to the plates. Cell culture plates were covered with adhesive polymerase chain reaction plate seals, heated at 90 °C in a water bath for 20 min, stored in the dark, and cooled to room temperature. The mixture of each well was collected into 1.5 mL tubes, and the total protein concentration was determined with a Take3 micro-volume plate (BioTek, Winooski, VT, USA). The total protein concentration was adjusted to 1–2 mg/mL, and sequencing-grade trypsin was added to each sample at an enzyme-to-protein ratio of 1:45 (w:w). The mixture was incubated at 37 °C for 4 h. The reaction was stopped by adding the quench solution containing stable isotope-labeled peptides as an internal standard to a final concentration of 10% acetonitrile and 0.1% formic acid. The surrogate peptide for CYP3A4 is EVTNFLR, which has been verified by several laboratories [8, 15]. Chromatography was performed with a gradient elution by using an Acquity UPLC HSS T3 column (50 mm × 2.1 mm, 1.8 µm). Water with 0.1% formic acid (solvent A) and acetonitrile with 0.1% formic acid (solvent B) were used as mobile phases at a flow rate of 0.5 mL/min. The injection volume was 18 µL. Table 1 lists the elution gradients, and Table 2 summarizes two to three mass transitions that were used to identify each peptide.

Calibration standards and quality control samples

Calibration standards and quality control (QC) samples were prepared simultaneously with hepatocyte samples. The following serial concentrations of standard peptides were prepared in 50 mM ammonium bicarbonate buffer: 0.5, 1, 2, 5, 10, 15, 25, and 40 ng/mL. The concentrations of the QC samples were 1.5, 4, and 20 ng/mL. All the QC samples were prepared simultaneously and added with the quench solution containing the internal standards. The least square regression algorithm was used to fit the standard curve with a weight of 1/x2.

Measurement of enzyme digestion efficiency

Different protein–enzyme ratios and incubation times were examined in terms of the digestion of recombinant CYP3A4 protein to determine the best and stable incubation conditions. Enzyme–protein ratios of 1:15, 1:30, and 1:45 (w:w) and incubation periods of 3, 4, and 6 h were tested (Table 3). The other steps were the same as mentioned above.

Table 3 Results of enzyme digestion efficiency.
Full size table

Statistical analysis

Data were analyzed with GraphPad Prism 7.0. Unpaired Student’s t test was used for the comparison between two groups. P < 0.05(*) was used to indicate significant differences between the two groups. P < 0.01(**) and P < 0.001(***) were also presented as needed. The data in Table 4 are presented as mean ± standard deviation (SD), which was also calculated with GraphPad Prism 7.0.

Table 4 Concentration of CYP3A4, CYP2C9 and fold change of CYP3A4 in enzyme activity with different inducers (mean ± SD)
Full size table

Results

PXR activity

Luciferase activity was measured after 293T cells were treated with different concentrations of furmonertinib plus the positive drug (10 µM rifampin) and the vehicle (0.1% DMSO). As shown in Fig. 2, at very low concentrations of furmonertinib (0.03 and 0.1 µM), PXR was strongly activated compared with the vehicle (0.1% DMSO; P < 0.01). However, its activity decreased as the concentration of furmonertinib increased. Cell Counting Kit-8 assay was used to determine the cell viability, but no significant difference was found among the groups. As an anti-cancer drug, furmonertinib can influence several pathways in 293T cell lines at high concentrations. Hence, this drug caused the transcription efficiency to decrease. However, furmonertinib has a potent PXR activation potential under 0.5 µM.

Fig. 2: PXR activation by different furmonertinib concentrations, with 10 µM rifampin as a positive control and 0.1% DMSO as a vehicle.
figure 2

*P < 0.05 and **P < 0.01.

Full size image

Quantification of the protein levels of CYP3A4 in plated primary human hepatocytes

Enzyme digestion efficiency was determined before induction incubation by using recombinant CYP3A4 enzymes. The optimal conditions (enzyme: protein = 1:45, digestion 4 h) were selected for the following experiments (Table 3). The matrix effects were tested, and no evident interference was found. The CYP3A4 protein expression was distinctly increased by rifampin and furmonertinib, and it increased as the concentrations of furmonertinib increased. The CYP3A4 protein expression in the cells treated with 10 µM rifampin was 20.16 ± 5.78 pmol/mg total protein, whereas the CYP3A4 protein expression in the cells induced by 0.5 µM furmonertinib was 4.8 ± 0.66 pmol/mg protein compared with that of the vehicle (0.1% DMSO), which expressed 0.65 ± 0.45 pmol/mg protein. The cells treated with 10 µM rifampin and 0.5 µM furmonertinib had fold-increase values of 31.21 ± 8.95 and 7.43 ± 1.03, respectively (Fig. 3a). The data, which were obtained using the samples in all three batches, are shown in Table 4 as mean ± SD.

Fig. 3: Changes of CYP3A4 protein expression and enzyme activity in SCHH induced by rifampin and furmonertinib.
figure 3

a Relative changes in the protein expression of CYP3A4 compared with that of the vehicle in three different donors (lot #GKJ, lot #QBU, and lot #HJK), and each batch of rifampin or furmonertinib was compared with the vehicle; *P < 0.05 and **P < 0.01. b Relative changes in the enzyme activity of CYP3A4 (evaluated in terms of the formation of 6β-hydroxytestosterone) in three different donors (lot #GKJ, lot #QBU, and lot #HJK), and each batch of rifampin or furmonertinib was compared with the vehicle; *P < 0.05, **P < 0.01, and ***P < 0.001. c Comparison of the magnitudes of changes in the protein expression and enzyme activity induced by 10 µM rifampin and 0.5 µM furmonertinib (e.g., lot #HJK); *P < 0.05 and **P < 0.01.

Full size image

Measurement of the enzyme activity of CYP3A4 in plated primary human hepatocytes

Three batches of PHHs were treated with the positive drug, different concentrations of furmonertinib, and the vehicle (0.1% DMSO) for 3 days. After incubation with the probe drug testosterone (200 µM), the upper supernatant was collected and tested for enzyme activity, and the plated cells were digested for absolute protein quantification. In general, the formation of 6β-hydroxytestosterone increased as the concentration of furmonertinib increased (Table 4). The rifampin- and 0.5 µM furmonertinib-treated groups showed similar magnitudes of the increased enzyme activity (Fig. 3b). However, several inconsistencies were observed between protein expression and enzyme activity. For example, in the cells from lot #HJK, 0.5 µM furmonertinib had an inducing capability (3.78-fold change in average) comparable with that of 10 µM rifampin (4.66-fold change in average) in terms of enzyme activity. However, the level of increase in the protein expression of the rifampin-treated group was much higher (25.37-fold change in average) than that of the 0.5 µM furmonertinib-treated group (5.79-fold change in average; Fig. 3c). The MRM chromatograms of the surrogate peptides of CYP3A4 (EVTNFLR and isotopic forms EVTNFLR [15N,13C]) are shown in Fig. 4.

Fig. 4: MRM chromatograms of the surrogate peptides of CYP3A4 (EVTNFLR and isotopic forms EVTNFLR [15N,13C]): blue and purple lines denote the transitions of the isotopic internal standard EVTNFLR [15N,13C]; red and green lines represent the unlabeled peptide EVTNFLR.
figure 4

Chromatograms of a standard peptide and isotopic internal standard, b peptides digested from cells induced by 10 µM rifampin, c peptides digested from cells induced by 0.5 µM furmonertinib, and d peptides digested from cells treated with 0.1% DMSO.

Full size image

Enzyme induction by furmonertinib and AST5902 in primary mouse/rat hepatocytes

The enzyme activity of putative CYP3A isoform induction in sandwich-cultured primary rat and mouse hepatocytes was evaluated by the measurement of testosterone metabolism and quantifying the formation of 6β-hydroxytestosterone. Dexamethasone was used as a positive drug in rat hepatocytes, but the rifampin group was set up, considering the conflicting views about whether rifampin is an inducer of CYP3A in rats [16,17,18,19]. As shown in Fig. 5, though furmonertinib increased the mRNA expression in primary mouse hepatocytes (Fig. 5a), the protein levels and enzyme activity had no obvious increase (Figs. 5b and 5c). Neither furmonertinib (Fig. 5d) nor AST5902 (data not shown) increased the enzyme activity in primary rat hepatocytes at any incubation concentrations. However, the level of testosterone metabolism increased by around two fold in primary rat and mouse hepatocytes treated with 10 µM rifampin and around five fold in rat hepatocytes treated with 100 µM dexamethasone (Fig. 5c, d).

Fig. 5: Induction effects of furmonertinib in rat and mouse.
figure 5

a Fold changes of furmonertinib-induced mRNA expression of CYP3A11 in mice; b furmonertinib-induced protein expression (relative bond density) of CYP3A11 in mice; c fold changes of furmonertinib-induced CYP3A11 activity in mice, with a representative gel band on the top; and d fold changes of furmonertinib-induced CYP3A1 activity in rats. ***P < 0.001.

Full size image

Protein quantification of CYP2C9 and CYP2C19 in plated primary human hepatocytes

The protein expression levels of CYP2C9 and CYP2C19 were quantified simultaneously. The calibration standards included the surrogate peptides of CYP3A4, CYP2C9, and CYP2C19 (Fig. 6a). However, for most samples, the concentrations of the surrogate peptide of CYP2C19 were below the limits of quantification by detecting the peptide GHFPLAER. For CYP2C9, the surrogate peptide is GIFPLAER [8]. Although no significant differences in the CYP2C9 protein expression were found between the groups treated with different furmonertinib concentrations, the protein expression increased in the cells treated with 10 µM rifampin (Fig. 6b). The average CYP2C9 protein concentrations in the cells treated with 10 µM rifampin and 0.1% DMSO were 2.96 ± 1.03 and 1.29 ± 0.61 pmol/mg total protein, respectively.

Fig. 6: Induction effects of furmonertinib on CYP2C family.
figure 6

a MRM chromatograms of the surrogate peptides of CYP2C19 (GHFPLAER), CYP3A4 (EVTNFLR), and CYP2C9 (GIFPLAER); b relative changes in the protein expression of CYP2C9 compared with that of the vehicle (0.1% DMSO) in two different donors (lot #GKJ and lot #HJK). *P < 0.05.

Full size image

Discussion

Our results provided robust evidence supporting that furmonertinib is a potent CYP3A4 inducer, which agrees with previously published mRNA data [2]. Furmonertinib has a prominent induction potential of CYP3A4 comparable with that of rifampin, an in vitro positive model inducer recommended in the US FDA guide. Thus far, the induction potential of furmonertinib for CYP3A4 has been evaluated systemically in three aspects: transcriptional level (mRNA expression), protein expression, and enzyme activity. We also verified that this compound could activate the PXR pathway, promoting the consequent transcription of CYP3A4 (Fig. 2). All these in vitro data confirm that furmonertinib is a strong CYP3A4 inducer, and it can be used as a positive control in sandwich-cultured PHHs. We attempted to determine whether an in vitro drug induction model might replace plated human hepatocytes to a certain extent, given that plateable human hepatocytes are expensive and scarce. Plated primary rat/mouse hepatocytes (Fig. 5) and HepG2 cell lines (data not shown) were tested. However, no significant induction of CYP3A enzyme activity was observed after the treatment with furmonertinib.

As an anti-cancer drug for patients with advanced tumor, furmonertinib is likely to be used in combination with other drugs. Given that this drug is metabolized mainly by CYP3A4 and can induce its enzyme activity, it has high drug–drug interaction risks in clinical practice. On the one hand, after multiple-dose administrations, furmonertinib can be cleared from the human body rapidly because of the increased metabolism rate through self-induction, which affects its own efficacy. On the other hand, if other drugs in a regimen are mainly metabolized by CYP3A4, which is highly likely to occur, their clearance can be accelerated. Although furmonertinib can raise safety concerns in vivo, it is still a potent in vitro CYP3A4 inducer that can be used as a positive control. We observed that furmonertinib promoted the mRNA transcription of CYP3A4, increased its protein expression, and consequently enhanced the enzyme activity.

Although protein expression can be measured with Western blot [20], this method needs a high amount of cells; furthermore, specific antibodies of CYP450 enzymes, especially CYP3A4 and CYP3A5, which share over 80% homology in their amino acid sequence, are hard to obtain [5]. Other attempts have also been conducted. Since 2000, LC-MS/MS-based proteomics has been developed from the evolution of MS instruments [21, 22]. In 2011, Williamson et al. [5] (from AB Sciex Ltd.) characterized several surrogate peptides that can specifically represent corresponding proteins and used this method in a drug induction assay; then, AB Sciex launched a human induction kit for CYP450 protein assay. Previous studies investigated different surrogate peptides of DMETs in two ways: top-down shot-gun proteomics analysis and in silico analysis based on available protein sequences [15, 23]. LC-MS/MS-based proteomics has been widely used in biological and pharmaceutical laboratories. In pharmacokinetics, laboratories in the University of Washington established a database of DMETs to quantify proteins in human tissues; this database is useful in studying physiologically based pharmacokinetic modeling and inter-individual differences in pharmacokinetics [24,25,26,27]. The peptides selected in our study, along with their stability and matrix effect, were verified by several laboratories [5, 8]. In summary, the peptides were stable in short-term periods (over 2 h at room temperature) and after three freeze–thaw cycles. In addition, when digested human serum albumin (1–2 mg/mL) was used, no distinct matrix effect was observed. However, the methods mentioned in all the reference articles above involved the isolation of microsomes from human tissues or cells in culture dishes.

In our model, specific protein expression could be measured in a single well of 48-well cell culture plates, and enzyme activity could be estimated simultaneously. In this manner, well-to-well or batch-to-batch deviation could be greatly reduced, and duplicate or triplicate samples could be prepared. Rifampin was used as the positive drug of CYP3A4 in the experiments, but several inconsistencies were observed in the magnitudes of rifampin-induced changes between CYP3A4 protein expression and its enzyme activity. Rifampicin caused a 25.37-fold change in the increased protein expression, and the enzyme activity showed a 5.79-fold change in induction compared with 0.1% DMSO in lot #HJK. Thus, using mRNA and protein changes induced by rifampin would result in the overestimation of drug–drug interaction risks and make it challenging for decision-makers in drug discovery and development to judge whether a drug candidate is a potential enzyme inducer which need more investigation. As such, potential drugs should be investigated thoroughly before they are commercially launched. As for furmonertinib, the CYP3A4 protein showed a similar magnitude of increased enzyme activity (3.78-fold change vs. 4.66-fold change in lot #HJK). Thus, with the consistency between protein changes and enzyme activity, drug–drug interaction risks could be easily evaluated. Savaryn et al. [11], who also observed similar results, proposed that such finding can be caused by the lack of necessary components (e.g., NADPH cytochrome P450 reductase and cytochrome b5) in enzymic catalytic reaction [11]. Another assumption was that apart from activating the PXR pathway and promoting DNA transcription, furmonertinib can stabilize CYP3A4 protein and enhance its activity, causing a comparable enzyme activity with notably less protein. However, additional studies should be conducted to elucidate the related induction mechanism. The measurement method for proteins in this study can yield important results. It is worth mentioning that furmonertinib may be used as a positive control only in in vitro studies. It also has some restrictions when applied in vivo. In clinical drug interaction studies, inducers should be applied to subjects as multiple doses to ensure a maximal induction. As an anti-cancer drug, furmonertinib is not recommended to be given to healthy subjects for a long term. Also, the dose of furmonertinib is only about 80 mg in clinical regimen, indicating that the plasma concentration could be too low to induce the enzyme activity. Moreover, the drug can covalently bind to plasma proteins.

The activation of PXR may lead to the induction of CYP2C enzymes. In vitro studies on CYP3A4 have revealed a positive induction result, and in such case, the US FDA recommends further induction studies on CYP2C8, CYP2C9, and CYP2C19. We attempted to quantify CYP2C9 and CYP2C19 via LC-MS/MS (Fig. 6), and the results showed that signals of CYP2C19 in some of the samples were below the quantification limits, whereas the fold change in CYP2C9 of the furmonertinib-treated groups was not significant. However, an induction trend was observed in the rifampin-treated group (Fig. 6). We are currently attempting to increase the signal-to-noise ratio and enhance the signals in MS to investigate the changes in CYP2C19.

Considering that plateable human primary hepatocytes are costly and scarce, we attempted to determine whether furmonertinib could induce CYP3A enzymes in other cells. Primary rat/mouse hepatocytes (Fig. 5) and HepG2 cell lines (data now shown) were tested. However, the results showed no signs of CYP3A induction (CYP3A4 in HepG2 cells, CYP3A1 in rat cells, and CYP3A11 in mouse cells). Choi et al. [28] investigated the inducibility of HepG2, Huh7, and NKNT-2 cell lines through enzyme assay, immunoblot analysis, and real-time PCR; they found that HepG2 cells are more responsive to various CYP inducers than other cell lines. However, HepG2 is a human liver cancer cell line. Furmonertinib, as an anti-cancer drug, is an EGFR tyrosine kinase inhibitor. Furmonertinib likely inhibits the normal growth and regular metabolism of HepG2 cell lines. In terms of enzyme activity, no obvious induction by furmonertinib was observed in primary mouse/rat hepatocytes. Although rifampin is considered to have no induction effect on rat CYP3A1 enzyme [18, 19], we observed that the rifampin group achieved a twofold increase in enzyme activity compared with that of the 0.1% DMSO group.

In conclusion, furmonertinib is a potent CYP3A4 inducer in PHHs in terms of three aspects: mRNA expression, protein expression, and enzyme activity. The induction potential of 0.5 µM furmonertinib is comparable with that of 10 µM rifampin in plated human hepatocytes but has no effect on other possible induction cell models (primary rat/mouse hepatocytes and HepG2 cells). The magnitudes of changes in protein expression and enzyme activity are notably more consistent in the 0.5 µM furmonertinib group than in the rifampin group. Therefore, the former is possibly a better positive model CYP3A4 inducer than the latter. However, the induction mechanism of this drug should be further explained.