Propafenone analogue with additional H‐bond acceptor group shows increased inhibitory activity on P‐glycoprotein

P‐glycoprotein (P‐gp) is an ATP‐dependent efflux pump that has a marked impact on the absorption, distribution, and excretion of therapeutic drugs. As P‐gp inhibition can result in drug–drug interactions and altered drug bioavailability, identifying molecular properties that are linked to inhibition is of great interest in drug development. In this study, we combined chemical synthesis, in vitro testing, quantitative structure–activity relationship analysis, and docking studies to investigate the role of hydrogen bond (H‐bond) donor/acceptor properties in transporter–ligand interaction. In a previous work, it has been shown that propafenone analogs with a 4‐hydroxy‐4‐piperidine moiety exhibit a generally 10‐fold higher P‐gp inhibitory activity than expected based on their lipophilicity. Here, we specifically expanded the data set by introducing substituents at position 4 of the 4‐phenylpiperidine moiety to assess the importance of H‐bond donor/acceptor features in this region. The results suggest that indeed an H‐bond acceptor, such as hydroxy and methoxy, increases the affinity by forming a H‐bond with Tyr310.


| INTRODUCTION
P-glycoprotein (P-gp) is an extensively studied efflux pump belonging to the ABC (ATP binding cassette) transporter superfamily. [1] This transporter protein is integrated into the plasma membrane and uses the energy of adenosine triphosphate (ATP) hydrolysis to extrude a wide variety of substances from the intracellular to the extracellular compartment. [2,3] In the human body, P-gp is physiologically expressed in tissues with barrier and/or excretory functions, including the liver, kidney, gastrointestinal tract and blood-brain barrier, where it mediates the uptake and elimination of endogenous compounds and xenobiotics. [4] In addition, overexpression of P-gp in cancerous cells has been recognized as a major cellular mechanism of multidrug resistance. [5] On the basis of its expression profile and broad substrate specificity, P-gp is involved in the absorption, distribution, and excretion of therapeutic drugs. [6] Co-administration of a transported drug and a P-gp inhibitor can lead to elevated drug levels in the plasma and in the organs defended by blood-tissue barriers, which poses the risk of organ toxicity. [7,8]  capacity to inhibit P-gp and the analogous breast cancer-resistant protein BCRP. [9][10][11] Over the past four decades, a considerable effort has been made to characterize the molecular basis of the interaction between P-gp and its inhibitors. [12,13] Although within analogous sets of compounds (e. g. propafenone, tetrahydroisoquinoline, tariquidar and chalcone derivatives, alkaloids, and flavonoids), a clear structure-activity relationship (SAR) pattern was observed, a general and conclusive model is still missing. [14][15][16][17][18][19] Numerous SAR studies suggested that there is a strong correlation between lipophilicity and P-gp inhibitory potency and that active compounds commonly possess at least one aromatic ring, a basic nitrogen atom, and several hydrophobic regions. [12] Preference of P-gp toward lipophilic compounds can be explained by the widely accepted model of substrate transport, which proposes that P-gp extracts its ligands directly from the inner leaflet of the plasma membrane. [20,21] In addition, ligand-and structurebased approaches suggest that hydrogen-bond (H-bond) acceptor properties are important for P-gp inhibitors. [21,22] Propafenone analogs are potent inhibitors of P-gp mediated drug efflux, exhibit a well-defined SAR pattern, and therefore, represent an excellent tool for investigating the molecular features triggering P-gp inhibition. [23] Chiba et al. [24] performed quantitative structure-activity relationship (QSAR) analyses on a series of highly related propafenone analogs and found a significant correlation between lipophilicity and inhibitory activity (log[1/EC 50 ]). They also revealed that piperazine and piperidine analogous propafenone derivatives, which bear a 4-hydroxy-4-phenylpiperidine moiety, are generally 10-fold more active than equilipophilic compounds without a hydroxy group in position 4 of the piperidine ring. On the basis of the docking study of a small set of propafenone analogs into P-gp homology models, Klepsch et al. [23] proposed the formation of an H-bond between the 4-hydroxy group of the 4-hydroxy-4-phenylpiperidine compound GPV062 (4g) and the Y310 moiety of P-gp. This additional H-bond could account for a stronger interaction with P-gp, and thus for the high relative activity of 4g when compared to its calculated logP value (EC 50 = 0.07 µM and logP = 3.98 [24] ). [25] In the present study, we further investigate the structural requirements for P-gp inhibition, focusing on the role of potential H-bond acceptors at the 4-hydroxy-4-phenylpiperidine moiety. To this end, we synthesized six novel propafenone derivatives sharing the same scaffold with 4g, but bearing varying functional groups in position 4 of the 4-phenylpiperidine substituent and tested their inhibitory activity on P-gp in vitro.
Following Barker et al., [26] the commercially acquired amine 6 was protected by Boc 2 O to yield carbamate 7. The free hydroxy group of 7 was activated using sodium hydride and methylated by methyl iodide to S C H E M E 1 Synthetic route to compounds 4a-g give 8. For deprotection of carbamates, the method of DeGoey et al. [27] proved to give the best results. Thus, 8 was deprotected by HCl in dioxane to yield 4-methoxy-4-phenylpiperidine (2a).

| In vitro studies
To evaluate the P-gp inhibitory activity of 4a-g, the intracellular accumulation of daunorubicin in the P-gp overexpressing CCRF VCR1000 cells was measured by flow cytometry. As shown in Table 1, all compounds proved to be strong P-gp inhibitors with IC 50 values in the low nanomolar range (see Supporting Information for Figure S1 showing the dose-response curves values reported earlier for this compound (0.07, [24] 0.06, [29] and 0.06 µM [30] ), which demonstrates a high comparability with earlier experiments on series of propafenone analogs and thus allows to pool all compounds into one data set for subsequent QSAR analysis.

| QSAR analysis
As previously reported, [24] a hydroxy group located at the 4-phenylpiperidine position of propafenone derivatives increases P-gp inhibitory activity, compared to other analogs with the same lipophilicity. Docking studies of analog 4g into a protein homology model of P-gp indicated that this affinity increase is possibly due to an H-bond formed between the hydroxy group and Y310. [23] However, as all four compounds synthesized by Chiba et al. [24] bear To evaluate the SAR of our newly synthesized compound set (4a-f) and 4g in a broader context, we involved the propafenone analogs 5a-p synthesized and characterized by Chiba et al. [24] in the analysis (Table 2 and Scheme 4). A previous work showed that the activities (IC 50 values) of the sodium channel blocker propafenone and its analogs exhibit an excellent correlation with their octanol/water partition coefficient (logP value). [31] As it is

S C H E M E 4
Chemical structures of compounds 5a-p [24] F I G U R E 1 Correlation of P-glycoprotein inhibitory activity (expressed as pIC 50 values) and calculated logP values. Propafenone analogs (•) bearing and (▪) lacking a hydrogen-bond donor or acceptor group

| Molecular docking analysis
Due to the unavailability of the crystal structure of human P-gp, we used our recently published homology model [32] for docking experiments. As can be seen from Figure 2, both the methoxy analog 4a and the acetyl derivative 4e are positioned in the same way as reported previously for the hydroxy compound 4g. [23] However, though the methoxy-oxygen is perfectly located for an H-bond with

| Cell culture
The human T-lymphoblast cell line CCRF VCR1000, overexpressing Pgp, was generated by stepwise selection of CCRF-CEM cells in the vincristine-containing medium [34] and was kindly provided by V. Gekeler (Altana-Pharma AG, formerly Byk-Gulden, Konstanz, Germany). CCRF VCR1000 cells were cultured in RPMI-1640 medium containing 20% FBS and were treated regularly with vincristine (1 µg/ml) for 72 hr, followed by centrifugation (300 g, 5 min, RT) and resuspension in a normal cell culture medium. Cells were maintained at 37°C in an atmosphere containing 5% CO 2 with 95% relative humidity.

| Steady-state daunorubicin accumulation experiments
For P-gp inhibition studies, the steady-state accumulation of daunorubicin (3 µM) was performed as previously described, [14] where X is the log of compound concentration, Y is the response in fluorescent intensity units, "bottom" and "top" are the lower and the higher plateaus of the nonlinear fit curve, respectively, and HillSlope is a factor that describes the steepness of the curve.
To correct for the dependency of IC 50 apparent values on the expression level of P-gp and the pump-leak kinetics as reviewed by Stein, [35] the final IC 50 values were calculated using the following where "bottom" and "top" are the lower and the higher plateaus of the nonlinear fit curve, respectively, and therefore they refer to the fluorescence intensity at zero and infinite inhibitor concentration, respectively. [36] The mean IC 50 values ± standard deviations given were calculated from three independent experiments for each compound.

| Calculation of lipophilicity and QSAR studies
The logP values were calculated using the "consensus" method of the MarvinSketch software (ChemAxon). [37] The performance of the "consensus" method was verified on a set of 19 propafenone derivatives whose distribution coefficient was experimentally determined by Chiba et al., [31] using high-performance liquid chromatography (HPLC). The experimentally obtained and the calculated values were in good agreement (r = 0.98). QSAR studies carried out comprised multiple linear regression analyses performed in MS Excel.

| Molecular docking analysis
To further understand the association between the activity of the compounds (4a and 4e) and the molecular structure, we performed molecular docking studies. The LigPrep module of Schrödinger Suite [38] was used to generate the correct protonation states for the ligands, which were then used for the docking studies. The OPLS_2005 force field was applied for the minimization of the structures and different ionization states were generated by adding or removing protons from the ligand at a target pH of 7.0 ± 2.0 using Epik, version 3.1. [39,40] Tautomers were also generated for each ligand. To generate stereoisomers, the information on chirality from the input file for each ligand was retained as is for the entire calculation. This resulted in a data set of 12 ligands. Due to unavailability of the crystal structure of P-gp, we used the homology model published by Jain et al. [32] The protein was prepared using the Protein Preparation Wizard of the Schrödinger Suite (2015). [41,42] Hydrogen atoms were added, and optimal protonation states and ASN/GLN/HIS flips were determined. The binding site was defined as the complete transmembrane region. Docking was performed using the genetic algorithm-based docking program GOLD. [43,44] All sidechains were kept rigid and the ligand was treated flexible by performing 100 genetic algorithm runs per molecule. The implemented Gold scoring function, GoldScore, was used for the evaluation of the complexes. Top scored poses were then inspected for the presence of protein-ligand interactions as reported by Klepsch et al. [23] and successively a final pose for each ligand (4a and 4e) was selected.