Potent and Selective Inhibition of CYP1A2 Enzyme by Obtusifolin and Its Chemopreventive Effects

Obtusifolin, a major anthraquinone component present in the seeds of Cassia tora, exhibits several biological activities, including the amelioration of memory impairment, prevention of breast cancer metastasis, and reduction of cartilage damage in osteoarthritis. We aimed to evaluate the inhibitory effects of obtusifolin and its analogs on CYP1A enzymes, which are responsible for activating procarcinogens, and investigate its inhibitory mechanism and chemopreventive effects. P450-selective substrates were incubated with human liver microsomes (HLMs) or recombinant CYP1A1 and CYP1A2 in the presence of obtusifolin and its four analogs. After incubation, the samples were analyzed using liquid chromatography-tandem mass spectrometry. Molecular docking simulations were performed using the crystal structure of CYP1A2 to identify the critical interactions between anthraquinones and human CYP1A2. Obtusifolin potently and selectively inhibited CYP1A2-mediated phenacetin O-deethylation (POD) with a Ki value of 0.031 µM in a competitive inhibitory manner in HLMs, whereas it exhibited negligible inhibitory effect against other P450s (IC50 > 28.6 µM). Obtusifolin also inhibited CYP1A1- and CYP1A2-mediated POD and ethoxyresorufin O-deethylation with IC50 values of <0.57 µM when using recombinant enzymes. Our molecular docking models suggested that the high CYP1A2 inhibitory activity of obtusifolin may be attributed to the combination of hydrophobic interactions and hydrogen bonding. This is the first report of selective and potent inhibitory effects of obtusifolin against CYP1A, indicating their potential chemopreventive effects.

In humans, the CYP1A family is composed of two major isoforms, namely CYP1A1 and CYP1A2 [9]. CYP1A enzymes are preliminarily regulated by their aromatic hydrocarbon receptors and exhibit aryl hydrocarbon hydroxylation activity [10]. They are critical in activating procarcinogens such as aflatoxins and polycyclic aromatic hydrocarbons. Reactive carcinogenic intermediates (aflatoxin B1-exo-8,9-epoxide or benzo[a]pyrene-7,8epoxide) produced by the upon the action of CYP1A enzymes ultimately form adducts with DNA [11,12]. Previous reports suggest that CYP1A contributes to tumor formation. Therefore, selective and robust inhibitors of CYP1A enzymes may act as effective chemopreventive agents for cancer treatment. Notably, resveratrol and sulforaphane have been considered as potent chemopreventive agents because they directly inhibit CYP1A enzymes [13,14].
Therefore, we evaluated the CYP1A2 inhibitory potential of obtusifolin, chrysophanol, emodin, physcion, and rubrofusarin, which are the major components of Cassiae semen, in HLMs. The difference in CYP1A2 inhibitory ability of the five phytochemicals was analyzed based on the interaction between the active site structure of the CYP1A2 enzyme and phytochemicals. An inhibition mechanism was identified for obtusifolin, which exhibited the most substantial inhibition of CYP1A2 activity; the inhibitory potential of obtusifolin against other nine P450 isoforms was also investigated to elucidate its selectivity in terms of CYP1A2 inhibition. In conclusion, we have demonstrated that obtusifolin acts as a selective CYP1A1 and CYP1A2 inhibitor in HLMs and also inhibits the activity of recombinant CYP1A enzymes, and is a promising chemopreventive candidate.

Molecular Docking Simulation
The X-ray crystal structure of α-naphthoflavone bound to CYP1A2 was retrieved from the protein data bank (PDB) under accession number 2HI4 [19]. The protein structure was cleared of all water molecules except those at the active site. Building missing loops, energy minimization, and protonation in Discovery Studio Client (DSC) v19.1 produced the final protein structure with heme; this was used for further molecular docking analysis. The anthraquinone structures were drawn using ChemDraw 20.1.1. These structures were converted to a 3D format in DSC 19.1 by energy minimization after adding hydrogen atoms. The grid center was derived using α-naphthoflavone in the crystal structure, and 1000 docking runs were performed for obtusifolin with SMINA, a fork of AutoDock Vina using Gnina 1.0 [20]. The redundant final poses were filtered using root-mean-square deviation (rmsd) values. The other four active compounds of Cassiae semen (three anthraquinones and one naphthopyrone) were also docked by aligning to the obtusifolin-binding pose ("Align to substructure" option in DSC v19.1) to understand the structural and molecular interactions responsible for the variations in their inhibitory activity against CYP1A2. For the docked CYP1A2-ligand complexes, AutoDock VINA scores were calculated using the "score_only" option in Gnina 1.0. The molecular docking procedure was validated by re-docking α-naphthoflavone in the crystal structure; the rmsd value of the docked α-naphthoflavone and its crystal form was 0.3 Å , showing a high agreement ( Figure S2). The figures depicting the protein-ligand interactions were generated using DSC v19.1.

Time-Dependent Inhibition Assays
The IC50 shift approach was applied to assess the time-dependent inhibition of CYP1A2 by obtusifolin [18]. For 30 min, obtusifolin was pre-incubated with pHLMs in the presence of NGS. The reaction was started with 100 µ M phenacetin and then incubated for 15 min. Following reaction termination and centrifugation, aliquots of the supernatants were subjected to LC-MS/MS analysis.

Molecular Docking Simulation
The X-ray crystal structure of α-naphthoflavone bound to CYP1A2 was retrieved from the protein data bank (PDB) under accession number 2HI4 [19]. The protein structure was cleared of all water molecules except those at the active site. Building missing loops, energy minimization, and protonation in Discovery Studio Client (DSC) v19.1 produced the final protein structure with heme; this was used for further molecular docking analysis. The anthraquinone structures were drawn using ChemDraw 20.1.1. These structures were converted to a 3D format in DSC 19.1 by energy minimization after adding hydrogen atoms. The grid center was derived using α-naphthoflavone in the crystal structure, and 1000 docking runs were performed for obtusifolin with SMINA, a fork of AutoDock Vina using Gnina 1.0 [20]. The redundant final poses were filtered using root-mean-square deviation (rmsd) values. The other four active compounds of Cassiae semen (three anthraquinones and one naphthopyrone) were also docked by aligning to the obtusifolin-binding pose ("Align to substructure" option in DSC v19.1) to understand the structural and molecular interactions responsible for the variations in their inhibitory activity against CYP1A2. For the docked CYP1A2-ligand complexes, AutoDock VINA scores were calculated using the "score_only" option in Gnina 1.0. The molecular docking procedure was validated by re-docking α-naphthoflavone in the crystal structure; the rmsd value of the docked α-naphthoflavone and its crystal form was 0.3 Å, showing a high agreement ( Figure S2). The figures depicting the protein-ligand interactions were generated using DSC v19.1.

Time-Dependent Inhibition Assays
The IC 50 shift approach was applied to assess the time-dependent inhibition of CYP1A2 by obtusifolin [18]. For 30 min, obtusifolin was pre-incubated with pHLMs in the presence of NGS. The reaction was started with 100 µM phenacetin and then incubated for 15 min. Following reaction termination and centrifugation, aliquots of the supernatants were subjected to LC-MS/MS analysis.

Data Analysis
WinNonlin software (Pharsight, Mountain View, CA, USA) was used to calculate IC 50 values. Based on Lineweaver-Burk double reciprocal plots, secondary plots of Lineweaver-Burk plots versus obtusifolin concentrations, and visual inspection of Dixon plots, the type of inhibition and apparent kinetic parameters for inhibitory activity (K i ) were estimated.

Binding Modes of Obtusifolin Assessed by Molecular Docking Simulation
To explore the critical interactions between anthraquinones from Cassiae semen and human CYP1A2 from the perspective of protein-ligand interactions and to clarify the mechanism behind the enhanced CYP1A2 inhibitory activity of obtusifolin compared with other anthraquinones, molecular docking simulations were performed using a previously reported crystal structure of CYP1A2 (PDB ID: 2HI4) [19]. As shown in Figure 3, anthraquinones could be readily docked into the active site of the CYP1A2 enzyme, while the binding area in the active site was highly overlapped with that of phenacetin (PDB accession number: 3EBS), implying that anthraquinones could occupy the binding area of the CYP1A2 substrate and thereby serving as inhibitors of CYP1A2 enzyme.

Binding Modes of Obtusifolin Assessed by Molecular Docking Simulation
To explore the critical interactions between anthraquinones from Cassiae semen and human CYP1A2 from the perspective of protein-ligand interactions and to clarify the mechanism behind the enhanced CYP1A2 inhibitory activity of obtusifolin compared with other anthraquinones, molecular docking simulations were performed using a previously reported crystal structure of CYP1A2 (PDB ID: 2HI4) [19]. As shown in Figure 3, anthraquinones could be readily docked into the active site of the CYP1A2 enzyme, while the binding area in the active site was highly overlapped with that of phenacetin (PDB accession number: 3EBS), implying that anthraquinones could occupy the binding area of the CYP1A2 substrate and thereby serving as inhibitors of CYP1A2 enzyme.  The potential critical interactions between anthraquinones and CYP1A2 enzyme were comprehensively analyzed using the docking complex model. Obtusifolin binds to the CYP1A2 active site with the lowest docking score of −12.78 kcal/mol ( Table 2). As shown in Figure 3, obtusifolin hydrophobically interacts mainly with heme, Phe226, Phe256, Phe260, Gly316, Ala317, and Leu497 in the catalytic site of CYP1A2. It also forms a nonclassical hydrogen bond with Asp313 through its methoxy group. The complex structure of CYP1A2 and α-naphthoflavone exhibited similar hydrophobic contacts with Phe226, Gly316, Ala317, and Leu497; thus, these interactions may be conserved for CYP1A2 inhibition. Conversely, emodin was the second-best CYP1A2 inhibitor among the examined compounds. Its hydroxyl group interaction with heme, along with hydrophobic contacts, contributed to a high CYP1A2 inhibitory activity ( Figure S3). Chrysophanol harbors a methyl group in the same location as that in obtusifolin. However, it lacks a hydroxyl group that would facilitate its interaction with heme like emodin. ( Figure S4). Rubrofusarin, the least potent inhibitor, likely has steric hindrance with Leu382, as its methoxy group is within 1 Å of this amino acid ( Figure S5). Thus, the binding orientation of rubrofusarin might be altered owing to these differences from obtusifolin, thereby reducing its CYP1A2 inhibitory capacity. The bulky methoxy group in physcion, the second least effective inhibitor, might have lessened its inhibitory action on CYP1A2; however, the lack of a hydroxyl group, as in rubrofusarin, might have contributed to its superior CYP1A2 inhibitory function compared with that of rubrofusarin ( Figure S6).
As observed in rubrofusarin and physcion, the presence of bulky methoxy groups in disadvantageous positions in the molecules may result in steric hindrance with Leu382, which could result in inefficient CYP1A2 inhibition (Figures S5 and S6). However, as observed in emodin, small functional groups, such as the hydroxyl group, may result in moderate action against CYP1A2. The presence of a methyl group (equivalent to the 14th atomic position in obtusifolin) in the molecules may result in moderate inhibitory activity, as seen in the case of emodin and chrysophanol, if the methoxy group is absent in both the favorable and unfavorable locations of the molecules (Figures S3 and S4).
In summary, the docking models suggest that the high inhibitory activity of obtusifolin against CYP1A2 primarily results due to the combination of extensive hydrophobic interactions with three Phe residues (Phe226, Phe256, and Phe260) via its methyl group and the non-classical hydrogen bond with Asp313 via the methoxy group ( Figure 3). The methoxy group at this particular position in obtusifolin is not present in other molecules, suggesting its significant role in the effective inhibition of CYP1A2 ( Figure 3).

Characterizatin of Inhibition Kinetics of Obtusifolin against CYP1A2
Obtusifolin inhibited microsomal CYP1A2 activity with IC 50 values of 0.19 µM, therefore, we sought to clarify the underlying mechanism of this inhibition. We measured the inhibition constant of α-naphthoflavone, a well-known CYP1A2 selective inhibitor, to validate our experimental system. α-Naphthoflavone inhibited CYP1A2-mediated POD activity with a K i value of 0.0075 µM in HLMs. Our results with α-naphthoflavone are consistent with that of a previous study, which reported potent inhibition of CYP1A2 activity in HLMs by α-naphthoflavone with a K i value of 0.01 µM using phenacetin as the CYP1A2 probe substrate [28]. This highlighted the suitability of our experimental system for evaluating the inhibitory ability of obtusifolin against CYP1A2. According to the Lineweaver-Burk plot ( Figure 4A), obtusifolin presented a typical pattern of competitive inhibition for CYP1A2-mediated POD activity in HLMs, with a K i value of 0.11 µM ( Table 3). The secondary Lineweaver-Burk plot also demonstrated a linear correlation ( Figure 4B, R 2 = 0.997). The Dixon plot intersected above the X-axis, indicating that obtusifolin inhibited CYP1A2 in a competitive manner ( Figure 4C) [29]. Obtusifolin exhibited a stronger inhibitory potency than machilin A (K i = 0.71 µM) [30], isopimpinellin (K i = 1.2 µM) [31], and mollugin (K i = 3.74 µM) [27], but was less potent than α-naphthoflavone (K i = 0.01 µM), a well-known strong selective inhibitor of CYP1A2 [26]. Obtusifolin also inhibited CYP1A2 activity in the rCYP1A2 isoform with a K i value of 0.21 µM, which was similar to the value observed in HLMs (Table 3).
our experimental system for evaluating the inhibitory ability of obtusifolin against CYP1A2. According to the Lineweaver-Burk plot ( Figure 4A), obtusifolin presented a typical pattern of competitive inhibition for CYP1A2-mediated POD activity in HLMs, with a Ki value of 0.11 μM ( Table 3). The secondary Lineweaver-Burk plot also demonstrated a linear correlation ( Figure 4B, R 2 = 0.997). The Dixon plot intersected above the X-axis, indicating that obtusifolin inhibited CYP1A2 in a competitive manner ( Figure 4C) [29]. Obtusifolin exhibited a stronger inhibitory potency than machilin A (Ki = 0.71 μM) [30], isopimpinellin (Ki = 1.2 μM) [31], and mollugin (Ki = 3.74 μM) [27], but was less potent than α-naphthoflavone (Ki = 0.01 μM), a well-known strong selective inhibitor of CYP1A2 [26]. Obtusifolin also inhibited CYP1A2 activity in the rCYP1A2 isoform with a Ki value of 0.21 μM, which was similar to the value observed in HLMs (Table 3).  In addition, several CYP1A2 inhibitors, including antofloxacin [26], furafylline [32] and isopimpinellin [31], are time-dependent inhibitors of CYP1A2. We investigated the effect of incubation time on the IC50 values of obtusifolin against CYP1A2. We found that obtusifolin showed time-independent inhibition of CYP1A2-mediated POD activity with an IC50 shift value of 1.46 (IC50 values of 0.13 and 1.19 µ M, with and without NGS preincubation, respectively). A chemical with an IC50 fold-shift decrease of <1.50 is considered a time-independent inhibitor [33]. α-Naphthoflavone, a time-independent inhibitor [34], also exhibited an IC50 shift value of 0.83.  In addition, several CYP1A2 inhibitors, including antofloxacin [26], furafylline [32] and isopimpinellin [31], are time-dependent inhibitors of CYP1A2. We investigated the effect of incubation time on the IC 50 values of obtusifolin against CYP1A2. We found that obtusifolin showed time-independent inhibition of CYP1A2-mediated POD activity with an IC 50 shift value of 1.46 (IC 50 values of 0.13 and 1.19 µM, with and without NGS preincubation, respectively). A chemical with an IC 50 fold-shift decrease of <1.50 is considered a timeindependent inhibitor [33]. α-Naphthoflavone, a time-independent inhibitor [34], also exhibited an IC 50 shift value of 0.83.
In this study, we used in vitro experimental system such as human liver microsomes and recombinant P450 isoforms for enzyme inhibition studies. Since these systems have a limitation in that they can not reflect the exact intracellular concentration of the test compounds, it will be necessary to conduct enzyme inhibition studies using cultured cells, such as human hepatocytes.

Chemopreventive Effects of Obtusifolin
The CYP1A family plays a vital role in the metabolic activation of carcinogens, including mycotoxins, such as aflatoxin B1 [11], and polycyclic aromatic hydrocarbons, such as benzo[a]pyrene [12]. The most reactive metabolite, benzo[a]pyrene-7,8-diol-9,10-epoxide, which is responsible for tumor production in new born mice, is produced by CYP1A enzymes [36]. Therefore, the selective inhibition of CYP1A-mediated activation of procarcinogens is potentially a crucial chemopreventive strategy. Sulforaphane [14], resveratrol [13], and curcumin [37] are potent chemopreventive agents that directly inhibit CYP1A enzymes. The CYP1A family contains only two functional isoforms, namely CYP1A1 and CYP1A2. We evaluated it inhibitory potential against the CYP1A family to confirm the potential of obtusifolin as a chemopreventive compound. Obtusifolin was incubated with human rCYP1A1 and rCYP1A2 isoforms using 7-ethoxyresorufin as the substrate. Obtusifolin similarly inhibited the CYP1A1-and CYP1A2-mediated EROD activities with IC50 values of 0.39 and 0.57 µ M, respectively (Figure 7), indicating that obtusifolin can be considered a chemopreventive agent. The inhibitory potential of obtusifolin against CYP1A1 isoform was higher than those of resveratrol (IC50 = 23 µ M) [13] and curcumin (IC50 = 0.74 µ M) [38], which are well-known chemopreventive CYP1A inhibitors. However, it was less potent than ɑ-naphthoflavone (IC50 = 0.06 µ M), a well-known strong inhibitor of CYP1A1 [39]. In addition, in terms of CYP1A-mediated POD activity, obtusifolin exhibited over six-fold selectivity for rCYP1A1 (IC50 = 0.06 µ M) over rCYP1A2 (IC50 = 0.37 µ M) ( Table 5). Its selectivity for CYP1A1 inhibition over CYP1A2 is similar to that of 7-hydroxyflavone (six-fold) [40]; however, it was lower than resveratrol which demonstrated 51-fold selectivity [13]. Based on this obtusifolin-mediated selective and potent inhibition of CYP1A1 and CYP1A2 enzymes, the in vivo chemopreventive effects of obtusifolin should be evaluated in the future.

Chemopreventive Effects of Obtusifolin
The CYP1A family plays a vital role in the metabolic activation of carcinogens, including mycotoxins, such as aflatoxin B1 [11], and polycyclic aromatic hydrocarbons, such as benzo[a]pyrene [12]. The most reactive metabolite, benzo[a]pyrene-7,8-diol-9,10-epoxide, which is responsible for tumor production in new born mice, is produced by CYP1A enzymes [36]. Therefore, the selective inhibition of CYP1A-mediated activation of procarcinogens is potentially a crucial chemopreventive strategy. Sulforaphane [14], resveratrol [13], and curcumin [37] are potent chemopreventive agents that directly inhibit CYP1A enzymes. The CYP1A family contains only two functional isoforms, namely CYP1A1 and CYP1A2. We evaluated it inhibitory potential against the CYP1A family to confirm the potential of obtusifolin as a chemopreventive compound. Obtusifolin was incubated with human rCYP1A1 and rCYP1A2 isoforms using 7-ethoxyresorufin as the substrate. Obtusifolin similarly inhibited the CYP1A1-and CYP1A2-mediated EROD activities with IC 50 values of 0.39 and 0.57 µM, respectively (Figure 7), indicating that obtusifolin can be considered a chemopreventive agent. The inhibitory potential of obtusifolin against CYP1A1 isoform was higher than those of resveratrol (IC 50 = 23 µM) [13] and curcumin (IC 50 = 0.74 µM) [38], which are well-known chemopreventive CYP1A inhibitors. However, it was less potent than α-naphthoflavone (IC 50 = 0.06 µM), a well-known strong inhibitor of CYP1A1 [39]. In addition, in terms of CYP1A-mediated POD activity, obtusifolin exhibited over sixfold selectivity for rCYP1A1 (IC 50 = 0.06 µM) over rCYP1A2 (IC 50 = 0.37 µM) ( Table 5). Its selectivity for CYP1A1 inhibition over CYP1A2 is similar to that of 7-hydroxyflavone (six-fold) [40]; however, it was lower than resveratrol which demonstrated 51-fold selectivity [13]. Based on this obtusifolin-mediated selective and potent inhibition of CYP1A1 and CYP1A2 enzymes, the in vivo chemopreventive effects of obtusifolin should be evaluated in the future.

Evaluation of Drug Interaction Potential of Obtusifolin
It was estimated that an in vivo interaction potential via the inhibition of P450 would likely occur if the ratio of inhibitor Cmax/Ki exceeded 1.0, and would be possible if it was between 0.1 and 1.0 [18]. Based on obtusifolin's maximum concentrations (0.86 and 0.54 μM) in rat blood after a single oral administration of Semen Cassiae extracts (1.25 g/kg; contents: 5.01 mg/g obtusifolin) [41] and obtusifolin (1.3 mg/kg) [42], the respective values of Cmax/Ki were 7.82 and 4.91 from the data of pHLMs (Ki = 0.11 μM), suggesting that obtusifolin has possible drug interaction potential with CYP1A2 substrate drugs. Thus far, there have been no reports on the pharmacokinetics of obtusifolin in humans, therefore, it is difficult to estimate the drug interaction potential of obtusifolin for humans. However, rhein, which is one of the anthraquinone compounds like obtusifolin, has been reported to reach a maximum plasma concentration of 9.52 and 18.8 μM after oral dose of 50 and 100 mg, respectively, in humans [43]. Therefore, obtusifolin might have drug interactions with CYP1A2 substrate drugs, such as imipramine [44], olanzapine [45], and tizanidine [46]. Therefore, in vivo studies are necessary to determine whether drug interactions between obtusifolin and CYP1A2 substrates have clinical relevance.

Conclusions
In this study, we investigated the inhibitory potential of four anthraquinones and one naphthopyrone isolated from Cassiae semen on the activity of CYP1A2 isoform in

Evaluation of Drug Interaction Potential of Obtusifolin
It was estimated that an in vivo interaction potential via the inhibition of P450 would likely occur if the ratio of inhibitor C max /K i exceeded 1.0, and would be possible if it was between 0.1 and 1.0 [18]. Based on obtusifolin's maximum concentrations (0.86 and 0.54 µM) in rat blood after a single oral administration of Semen Cassiae extracts (1.25 g/kg; contents: 5.01 mg/g obtusifolin) [41] and obtusifolin (1.3 mg/kg) [42], the respective values of C max /K i were 7.82 and 4.91 from the data of pHLMs (K i = 0.11 µM), suggesting that obtusifolin has possible drug interaction potential with CYP1A2 substrate drugs. Thus far, there have been no reports on the pharmacokinetics of obtusifolin in humans, therefore, it is difficult to estimate the drug interaction potential of obtusifolin for humans. However, rhein, which is one of the anthraquinone compounds like obtusifolin, has been reported to reach a maximum plasma concentration of 9.52 and 18.8 µM after oral dose of 50 and 100 mg, respectively, in humans [43]. Therefore, obtusifolin might have drug interactions with CYP1A2 substrate drugs, such as imipramine [44], olanzapine [45], and tizanidine [46]. Therefore, in vivo studies are necessary to determine whether drug interactions between obtusifolin and CYP1A2 substrates have clinical relevance.

Conclusions
In this study, we investigated the inhibitory potential of four anthraquinones and one naphthopyrone isolated from Cassiae semen on the activity of CYP1A2 isoform in HLMs. Among five compounds tested, obtusifolin potently inhibited CYP1A2-mediated POD activity, with K i values lower than 0.5 µM in HLMs and rCYP1A2. Furthermore, obtusifolin selectively inhibited CYP1A1 and CYP1A2 enzymes; however, it had negligible inhibitory effects on other P450 isoforms, highlighting its potential chemopreventive effects. In conclusion, we confirmed the selective and potent inhibitory effects of obtusifolin against CYP1A enzymes in HLMs and recombinant CYP1A enzymes.