Evaluation of Larger Side-Group Functionalities and the Side/End-Group Interplay in Ritonavir-like Inhibitors of CYP3A4

Abstract

A new series of 13 ritonavir-like inhibitors of human drug-metabolizing CYP3A4 was rationally designed to study the R₂ side-group and R₃ end-group interplay when the R₁ side-group is represented by phenyl. Spectral, functional and structural characterization showed no improvement in the binding affinity and inhibitory potency of R₁/R₂-phenyl inhibitors upon elongation and/or fluorination of R₃-Boc (tert-butyloxycarbonyl) or its replacement with benzenesulfonyl. When R₃ is pyridine, the impact of R₂-phenyl-to-indole/naphthalene substitution was multidirectional and highly dependent on side-group stereo configuration. Overall, the R₂-naphthalene/R₃-pyridine containing 2f (R/S) was the series lead compound and one of the strongest binders/inhibitors designed thus far (K_s = 0.009 μM; IC₅₀ = 0.10 μM). Introduction of a larger biphenyl or fluorene as R₂ did not lead to any improvements. Contrarily, fluorene-containing 13 was the series weakest binder and inhibitor (K_s = 0.734 μM; IC₅₀ = 1.32 μM), implying that the fluorene moiety is too large to allow unrestricted access to the active site. The R₂-biphenyl, however, can switch positions with R₃-Boc to enable heme ligation. Thus, for small and chemically simple end-groups such as Boc and pyridine, the R₂/R₃ interplay could lead to conformational rearrangement that would be difficult to foresee without structural information.

Graphical Abstract

Structure-activity relationship of 13 rationally designed CYP3A4 inhibitors differing in side- and end-groups was investigated using spectral, functional and structural approaches. The impact of the side/end-group interplay on the ligand binding mode, K_s and IC₅₀ was found to be multidirectional and largely dependent on the size/stereochemistry of the side group.

INTRODUCTION

Cytochrome P450 (CYP) monooxygenases are found in all kingdoms of life and metabolize a wide variety of endogenous and exogenous molecules. In humans, CYP3A4 is the major hepatic and intestinal isoform that clears over half of administered pharmaceuticals through their oxidation (Guengerich & Shimada, 1991; Manikandan & Nagini, 2018). CYP3A4 has a large and flexible active site capable of binding structurally diverse compounds, some of which could stimulate or inhibit its activity (Zhou, 2008). Generally, CYP3A4 inhibition is undesired because it could interfere with drug metabolism and lead to drug-drug interactions, toxicity and therapeutic failures. However, partial inhibition of CYP3A4 activity could prolong the half-life of quickly metabolized pharmaceuticals, thereby increasing their bioavailability and therapeutic efficiency. This pharmacoenhancing (boosting) effect is currently used in anti-retroviral and other therapies where the potent CYP3A4 inhibitors, such as ritonavir and cobicistat, are co-administered with drugs predominantly metabolized by CYP3A4 (Brayer & Reddy, 2015; Chen et al., 2020; Eisenmann et al., 2021; Gaur et al., 2022; Kao et al., 2018). Ritonavir was originally designed as an HIV-1 protease inhibitor (Figure 1), and its ability to effectively suppress the CYP3A4 activity was purely coincidental (Kempf et al., 1997). Cobicistat, on the other hand, is a derivative of ritonavir developed based on chemical structure-activity relationship (SAR) analysis (Mathias et al., 2010).

Figure 1.

Chemical structures of ritonavir and series VII analogues. The head-group, R₁/R₂ side-groups in different stereo configuration, and terminal R₃-moiety are indicated.

Knowledge of the CYP3A4 inhibitory mechanism is highly important because it could assist in designing safer and more effective drugs and pharmacoenhancers. The main obstacle is that many structurally diverse compounds can inhibit CYP3A4 via different mechanisms (competitive, noncompetitive, mechanism-based and mixed) (Zhou, 2008). To date, the interaction of CYP3A4 with several classes of inhibitors has been investigated, including quiniline (Pearson et al., 2011), biguanidine (Guo et al., 2017), pyrimidine (Mukherjee et al., 2015), imidazopyridine (Song et al., 2012), and steroid-based compounds (Paquin, Oufqir, Sevrioukova, Reyes-Moreno, & Berube, 2021). Although few strong CYP3A4 binders and inhibitors were identified, none were superior to ritonavir in terms of binding affinity and inhibitory potency. Moreover, ritonavir induces pronounced spectral changes upon binding and willingly co-crystallizes with CYP3A4. This largely facilitates determination of the dissociation constant and the ligand binding mode, which is required for structure-based design. For this reason, ritonavir analogues are the focus of our research.

Early studies on ritonavir analogues and cobicistat (Flentge et al., 2009; Kempf et al., 1997; Xu et al., 2010) showed that (i) removal of the backbone hydroxyl in ritonavir, strictly required for the inhibition of HIV-1 protease, does not affect the inhibitory activity against CYP3A; (ii) unhindered electron-rich nitrogen atom of the pyridyl or thiazolyl head-group directly interacts with the heme iron; (iii) hydrophobic side-groups are required for the high inhibitory activity toward CYP3A; and (iv) substitution of the valine side-chain with a bulky morpholine decreases cross-reactivity with other hepatic CYPs.

Our objective is to identify structural elements of ritonavir-like compounds that lead to potent inhibition of CYP3A4 and to test whether more effective inhibitors can be developed using structure-based design. Toward this goal, we investigated several series of analogues differing in the backbone and head/side/end-group functionalities to elucidate how biochemical properties and the binding manner relate to the inhibitory potency for CYP3A4. The initial SAR studies were conducted on desoxyritonavir analogues provided by Gilead Sciences and suggested that (i) coordination to the heme drives the inhibitor association, with the pyridine moiety being the strongest heme-ligating group; (ii) the desoxyritonavir backbone is more flexible and allows the inhibitors to adopt a more optimal conformation in the active site; (iii) the backbone amide nitrogen or carbonyl oxygen can stabilize the inhibitory complex by H-bonding to the active site Ser119; (iv) aromatic R₁ and R₂ side-groups are required for potent inhibition; and (v) a poly-functional end-group promotes the inhibitor binding but is not as critical as the side-groups (I. F. Sevrioukova & Poulos, 2010, 2012; I. F. Sevrioukova & Poulos, 2013). Based on these findings, we developed a pharmacophore model for potent CYP3A4-specific inhibitors (I. F. Sevrioukova & Poulos, 2014) and utilized a build-from-scratch approach for its evaluation. To date, we investigated the impact of the backbone length/composition, position of the heme-ligating N-heteroatom, the length of the pyridyl-R₁ and R₁-R₂ spacers, the size/hydrophobicity/stereochemistry of the R₁/R₂ side-groups, and the end-pyridine attachment (Kaur, Chamberlin, Poulos, & Sevrioukova, 2016; Samuels & Sevrioukova, 2019; E. R. Samuels & I. F. Sevrioukova, 2018; Samuels & Sevrioukova, 2020, 2021, 2022). No direct correlation was observed between the binding affinity and inhibitory potency for the investigated compounds, but both parameters were improved with an increase in side-group hydrophobicity (phenyl -> indole -> naphthalene) and the length of the pyridyl-R₁ spacer. Steric constraints imposed on the heme-ligating group were the key impact factor, which could be minimized with seven-atom separation (pyridyl-propyl linker) or by switching from meta- to para-N-pyridine for the shorter tether. The side-group stereochemistry was more impactful for the bulky R₁/R₂ substituents, with the R/S and S/S configuration being more and S/R least favorable.

The goal of this study was to further evaluate the impacts of the R₃ end-group substitution and increase in size of the R₂ side-group. The new set, series VII, contains 13 inhibitors, all of which except 2a have pyridyl-propyl linker (Figure 1). The first subset contains four analogues with R₁/R₂-phenyls and R₃ amide represented by either trifluoroacetyl glycyl (2a and 2b), trifluoroacetyl (2c), or benzenesulfonate (2d). The second subset includes five analogues with R₁-phenyl, R₃-pyridine and R₂ as indole (2e) or naphthalene (2f-i). Four R₂-naphtalene containing stereoisomers were produced to test the effect of side-group stereochemistry. The third subset has four compounds with R₁-phenyl and R₂ as biphenyl (5a-b and 6) or fluorene (13) and was designed to determine whether the bulkier R₂ could improve the binding and inhibitory strength. Interaction of the analogues with CYP3A4 was investigated using spectral and functional approaches, and crystal structures with eight inhibitors have been determined to assist interpretation of the experimental data. The properties of new analogues and their parent compounds (Figure 2) were compared to better decipher SARs. As we show here, elongation/fluorination of R₃-Boc or its replacement with R₃-benzylsulfonyl did not lead to notable improvements, whereas a decrease in the binding and inhibitory strength was observed upon introduction of the bulky R₂-biphenyl and, to a larger extent, R₂-fluorene. Nonetheless, even for small end-groups such as Boc and pyridine, the R₂/R₃ interplay was found to be an important factor that could significantly alter the ligand binding mode and modulate the inhibitory potency of ritonavir-like compounds.

Figure 2.

Previously characterized parent compounds from series III(Samuels & Sevrioukova, 2019), V(Samuels & Sevrioukova, 2021) and VI(Samuels & Sevrioukova, 2022) that were used for comparative analysis with series VII analogues. The K_s, IC₅₀ and ΔT_m values and percentage of heme destroyed by H₂O₂ in the inhibitor-bound CYP3A4 are indicated for comparison with the respective parameters for series VII inhibitors given in Table 1.

RESULTS AND DISCUSSION

Rationale for series VII analogues

Our previous studies (Kaur et al., 2016; Samuels & Sevrioukova, 2019; E. R. Samuels & I. F. Sevrioukova, 2018; Samuels & Sevrioukova, 2020, 2021, 2022) have shown that combining the flexible pyridyl-propyl spacer with R₁-phenyl and R₂-naphthalene in R/S or S/S configuration largely improves the binding affinity and inhibitory potency for CYP3A4. Thus far, the R₂ functionalities larger than naphthalene had not been investigated, and only alkyl pyridines were tested in place of the terminal Boc (used as the amine protective group during chemical synthesis). This study was set to determine the effects of R₃-Boc replacement and whether CYP3A4 can accommodate compounds with a bulkier R₂ group. Due to chemical inertness and higher hydrophobicity of fluorine, carbon-to-fluorine substitution is widely used in medicinal chemistry to improve metabolic stability, alter physicochemical properties or improve the binding affinity of compounds of interest (Richardson, 2016). Given that the isopropyl end-moiety of ritonavir is the primary site of CYP3A4-dependent metabolism (Gangl, Utkin, Gerber, & Vouros, 2002; Koudriakova et al., 1998), we synthesized three compounds with trifluoroacetyl groups to test the impact of this R₃ modification on the interaction with CYP3A4. The fluorinated compounds had R₁/R₂-phenyls and differed in the length of the head- and end-group linkers (2a-c; Figure 1). An analogue with R₃-benzenesulfonate (2d) was included in this subset to elucidate the effect of a polar, yet rigid end-group linkage. The R₃-pyridine containing 2e-i, in turn, were synthesized as counterparts of 3e-f and 10a (Figure 2), the tightest binders and most potent inhibitors from series V and VI (Samuels & Sevrioukova, 2021, 2022). The third subset was designed to determine whether the bulkier and more hydrophobic R₂ functionalities would be beneficial or not. Here, two substitutes were tested: elongated biphenyl, coupled with Boc- (5a-b) or pyridine end-groups (6), and voluminous fluorene linked through the central carbon atom (13). The R/S and S/S conformers with R₂-biphenyl were produced to assess the impact of stereochemistry. Interaction of CYP3A4 with new analogues was characterized using spectral, functional and structural approaches. The ligand binding kinetics were not measured because no correlation between the binding rate and inhibitory potency for CYP3A4 was observed for the previous series of compounds.

Interaction of CYP3A4 with the R₁/R₂-phenyl containing 2a-d

Spectral and inhibitory properties

As expected, 2a-d induced a red shift in the Soret band, indicative of the N-pyridine ligation to the heme iron (type II spectral change; Figure 3A-D). However, the shorter λ_max values were observed for both the ferric and ferrous forms of 2a-bound CYP3A4: 421/443 nm vs 422/444 nm for complexes with 2b-d (Table 1). This reflects differences in the electronic orbital overlap and confirms that the shorter pyridyl-ethyl linker does not allow the head-group to orient optimally for coordination to the heme. Further evidence for the weaker binding was a 10-fold difference in the spectral dissociation constant (K_s; measure of the binding affinity) for 2a and its counterparts with the pyridyl-propyl linker: 0.158 μM vs 0.011-0.018 μM, respectively (Table 1). Moreover, 2a had a 3-fold lower affinity for CYP3A4 than the nonfluorinated parent compound 4f from series III (K_s of 0.045 μM; Figure 2), meaning that elongation and fluorination of R₃ decreases rather than enhances the binding strength. In contrast, 2b-d bind as tightly as the R₃-Boc containing parent compound 3a from series V (Figure 2): K_s of 0.011-0.018 μM vs 0.015 μM, respectively (Table 1). Thus, depending on the head-group linker, extension/fluorination of R₃-Boc or its replacement with benzenesulfonate can either weaken or strengthen the binding to CYP3A4.

Figure 3.

Spectral properties and the binding modes of 2a-d. (A-D) Ligand-induced spectral changes in CYP3A4. Absorbance spectra of ferric ligand-free and inhibitor-bound CYP3A4 are in black and red, respectively. Spectra of ferrous ligand-bound CYP3A4 and its CO-adducts are in green and blue, respectively. Insets a show the difference spectra recorded during equilibrium titrations; insets b - titration plots with quadratic fittings. The derived spectral dissociation constants (K_s) are given in Table 1. (E-H) Crystallographic binding modes of 2a-d, respectively. Part of the adjacent I-helix and F304 in the inhibitory complexes and water-bound CYP3A4 (5VCC structure) are depicted in gray and black, respectively. Polder omit electron density maps contoured at 3σ level are shown as green mesh. Blue dotted lines are H-bonds between the backbone carbonyl oxygen and the hydroxyl group of the active site S119 residue. The end-group of 2a is disordered and displayed in arbitrary conformation. The 2a-, 2b-, 2c- and 2d-bound structures of CYP3A4 have PDB ID codes 9COR, 9COS, 9COT, and 9COU, respectively.

Table 1.

Properties of series VII inhibitors

Compound	λmax (nm) ferric/ferrous	Ks a (μM)	IC50 b (μM)	ΔTm c (°C)	Heme destroyed (%) d
pyridyl-ethyl head-group linker
R1/R2-phenyl
R3-trifluoroacetyl glycyl
2a (R, S)	421/443	0.158±0.023	0.24±0.02	3.1	50
pyridyl-propyl head-group linker
R1/R2-phenyl
R3-trifluoroacetyl glycyl
2b (R, S)	422/444	0.011±0.005	0.30±0.04	6.4	28
R3-trifluoroacetyl
2c (R, S)	422/444	0.012±0.003	0.31±0.02	5.8	29
R3-benzenesulfonyl
2d (R, S)	422/444	0.018±0.007	0.26±0.03	5.7	26
R1-phenyl/R2-indole/R3-pyridyl
2e (R, S)	422/444	0.032±0.010	0.20±0.03	6.3	29
R1-phenyl/R2-naphthalene/R3-pyridyl
2f (R, S)	422/444	0.009±0.001	0.10±0.02	7.7	21
2g (S, S)	421/444	0.011±0.004	0.18±0.01	5.8	26
2h (R, S)	421/444	0.037±0.003	0.32±0.04	4.3	29
2i (S, S)	421/443	0.015±0.003	0.22±0.03	4.6	34
R1-phenyl/R2-biphenyl/R3-Boc
5a (R, S)	422/444	0.030±0.001	0.42±0.03	3.6	34
5b (S, S)	421/444	0.016±0.002	0.38±0.04	3.5	27
R1-phenyl/R2-biphenyl/R3-pyridyl
6 (S, S)	421/444	0.053±0.003	0.62±0.05	3.0	43
R1-phenyl/R2-fluorene/R3-Boc
13 (R, rac)	421/443	0.734±0.052	1.32±0.11	4.5	35
Ritonavir e	422/443	0.017	0.22	5.4	31

^{a -}Spectral dissociation constants for the CYP3A4-inhibitor complexes determined from titration plots (insets a in Figures 3A-D, 5A-E and 7A-D).

^{b -}Inhibitory potency measured for the BFC debenzylase activity in a soluble reconstituted system.

^{c -}Ligand-dependent change in the melting temperature of CYP3A4.

^{d -}Percentage of heme destroyed by H₂O₂ within 2 h relative to that in ligand-free CYP3A4.

^{e –}Determined previously (I. F. Sevrioukova, 2024).

Inhibitory potency of new analogues was assessed by measuring the 7-benzyloxy-4-(trifluoromethyl)coumarin (BFC) O-debenzylase activity of recombinant CYP3A4. Since neither compound in this and other subsets has chemical groups that could undergo bioactivation and produce reactive metabolites, time-dependent inactivation of CYP3A4 was not investigated. The R₃ fluorination was found to increase the inhibitory potency of 2a by 3-fold compared to that of the parent compound 4f: IC₅₀ of 0.24 μM vs 0.68 μM, respectively. For 2b-d, there was a reverse trend and, relative to the parent 3a, the IC₅₀ increased by ~1.5-2-fold: from 0.16 μM to 0.26-0.31 μM (Table 1). There were also differences in thermostability of the inhibitor-bound CYP3A4, assessed by measuring changes in the melting temperature (T_m). Association of all compounds improved thermostability of CYP3A4. Compared to 2a, an increase in ΔT_m was 2-fold higher upon binding of 2b-d: 3.1 °C vs 5.7-6.4 °C, respectively (Table 1). However, all analogues in this subset were the weaker stabilizers than the parent compounds 4f and 3a, binding of which increased T_m by additional 1-2 °C (ΔT_m of 4.2 °C and 8.1 °C, respectively).

Ritonavir-like molecules inhibit CYP3A4 activity not only via direct heme ligation but also by clogging the active site and preventing substrates from accessing the catalytic center. As reported earlier (Samuels & Sevrioukova, 2021, 2022), the heme accessibility correlates with the inhibitory potency and can be probed using a small oxidizing agent, hydrogen peroxide. The H₂O₂-dependent heme bleaching assay showed that 2a had the smallest heme-protective effect, with half of the heme being destroyed within two-hour reaction vs 26-29% for 2b-d (Table 1). The respective values for the parent compounds 4f and 3a were 45% and 18%, respectively. Thus, the investigated modifications of R₃ did not improve the binding strength or heme protective ability and had a mixed effect on the inhibitory potency of R₁/R₂-phenyl containing compounds.

Crystal structures of inhibitory complexes

Crystal structures of CYP3A4 bound to 2a-d were solved to 2.55-2.80 Å resolution (Table S1). The ligand binding modes relative to the central I-helix are shown in Figure 3E-H. All compounds bind in a traditional orientation, with the R₁-phenyl inserted into a hydrophobic pocket adjacent to the I-helix (P1 site) and the R₂-phenyl placed near the heme and head-pyridine (P2 site). Polder omit electron density maps for all ligands were verified (Liebschner et al., 2017) (Table S3) and displayed as green mesh. Structural features of the inhibitory complexes are summarized in Table 2.

Table 2.

Structural features of CYP3A4-inhibitor complexes

Compound	Fe-N bond		pyridine ring rotation (°)b	I-helix shift (Å)c	H-bond with S119 (Å)d	S-π contact with F108 (Å)e	pyridine-R2 ring angle and overlap	F304-R1 ring angle (°) and overlap	End-group protein contacts
	distance (Å)	angle (°)a
2a (R, S)	2.13	10	25	0.98-1.18	2.96	3.9	48°; half	90°; edge	disordered
2b (R, S)	2.19	7	35	2.04-2.11	3.06	3.4	55°; half	43°; half	57, 106, 108, 374
2c (R, S)	2.19	5	40	1.98-1.96	2.83	3.7	55°; half	55°; partial	106, 108, 374
2d (R, S)	2.18	5	40	1.99-1.97	2.33	3.8	38°; half	45°; half	106, 108, 374
2e (R, S)	2.06	0	40	2.13-2.32	2.26	- f	55°; partial	54°; half	108
2f (R, S)	2.19	5	42	2.27-2.26	2.23	3.8	55°; half	45°; half	106-108, 221, 224
2g (S, S)	2.22	5	40	1.19-2.16	2.2.6	4.4	40°; half	42°; partial	304, 308, 482
5a (R, S)	2.10	0	43	0.70-1.88	2.51	-	-g	45°; half	105, 370, 372 g

^{a -}Deviation from perpendicularity.

^{b -}Angle between the planes passing through the pyridine ring and the NB-ND heme atoms

^{c -}Distance between the C_α-atoms of F304 and A305 in the inhibitor-bound and ligand-free CYP3A4 (5VCC structure).

^{d -}Distance between the inhibitor’s carbonyl oxygen and the hydroxyl group of S119.

^{e -}Distance from the backbone sulfur atom to the nearest atom of the F108 aromatic ring.

^{f -}No overlap.

^{g -}In 5a, the R₂ and R₃ groups switch places: R₃ is at P2 site near the heme, whereas R₂ faces the substrate access channel and is disordered.

To determine if/how the R₃ modification affects the ligand binding mode, new analogues were comparted with the parent compounds. Superposition of 2a and series III 4f is shown in Figure 4A. Due to structural disorder, the end-groups of these inhibitors are depicted in arbitrary conformations and cannot be compared. Even so, the distinct backbone curvatures are evident. The 2a core shifts aside, thereby reducing the depth of R₁ embedment into the hydrophobic P1 pocket and preventing R₂ from maximizing π-π and cation-π interactions with the heme-ligating pyridine and R105 guanidine, respectively, to the same extent as 4f does. This could explain the higher affinity of 4f for CYP3A4. That 2a has a higher inhibitory potency is likely due to its stronger ligation to the heme (Fe-N distance is 2.13 Å vs 2.29 Å for 4f), stronger H-bonding to S119 (2.96 Å vs 3.24 Å for 4f), and the ability to form an S-π interaction with F108, lacking for 4f.

Figure 4.

Structural comparison of 2a-d with the parent compounds 4f and 3a from series III and V, respectively, (Samuels & Sevrioukova, 2019), (Samuels & Sevrioukova, 2021) to demonstrate differences in the backbone, side- and end-group positioning. (A) Superposition of 2a (in lavender) and 4f (in black; 6DAC structure). Part of the I-helix is displayed to show the adaptive rotameric changes in F304. The disordered end-groups of 2a and 4f are shown in arbitrary conformations and cannot be compared. Even so, it is evident that 4f adopts a more compact conformation due to a different backbone folding, which triggers rotameric changes in F304. (B) Two views at superimposed 2b-d (in green, orange and magenta, respectively) and 3a (in black; 7KVH structure). In this subgroup, 3a adopts a more relaxed conformation, with the R₂-phenyl positioned slightly closer to the heme and with a larger overlap with the head-pyridine and R105 guanidine.

Structural superposition of 2b-d with the parent compound 3a shows a greater resemblance in the ligand fold (Figure 4B), which explains similarity in the binding affinity. One distinctive feature of 2d is the inward orientation of R₃-benzenesulfonate, where the benzene ring is directed toward the I-helix rather than the substrate channel, as observed for other end-groups. Also, as was noted for 4f, 3a has a distinct backbone curvature that allows to maximize interactions via the R₂-phenyl and the backbone sulfur atom, resulting in a larger overlap with the head-pyridine and the F108 ring, respectively. Moreover, 3a forms a shorter and near perpendicular Fe-N bond, with the length of 2.11 Å and 3° deviation from perpendicularity vs 2.18-2.19 Å and 5-10°, respectively, for 2b-d (Table 2). These features could explain the higher inhibitory potency of 3a.

Summing up, comparison of new R₁/R₂-phenyl containing analogues with the parent compounds shows that elongation, fluorination or replacement of R₃-Boc with benzenesulfonate alter the ligand binding mode without a uniform effect on K_s and IC₅₀. In 2a, due to a shorter head-group linker, the modified tail induces notable changes in the backbone curvature and the side-group placement, which decreases the binding affinity but improves the inhibitory potential. In contrast, having an elongated pyridyl-propyl linker, 2b-d are more flexible and seem to better adjust to structural changes imposed by the modified R₃ moiety. These adjustments decrease the IC₅₀ to some extent but do not affect K_s.

Interaction of CYP3A4 with the R₂-indole/naphthalene containing 2e-i

Spectral and inhibitory properties

Previously (Samuels & Sevrioukova, 2022), we introduced the meta-N-pyridine as R₃ in anticipation that this would lead to a productive heme coordination regardless of whether the analogue enters the active site cavity with the head or tail on. Indeed, two bi-pyridine compounds were found to ligate to CYP3A4 via the end-group to better optimize protein-ligand interactions. The di-pyridine containing 10a from series VI (Figure 2) was the lead compound, so the scaffold was used to synthesize 2e-i (Figure 1) for evaluation of larger R₂ functionalities with R₁ as phenyl.

As seen from Figure 5A-E, all compounds from this subgroup produced type II spectral shifts in CYP3A4. The R₂-indole containing 2e and the R₂-naphthalene containing R/R conformer 2h were the weakest binders, with the binding affinity 4-fold lower relative to that of the parent compound 10a (K_s of 0.032-0.037 μM vs 0.008 μM, respectively). For other stereoisomers, the K_s value was in the 0.009-0.015 μM range (Table 1). 2f (R, S) was not only the strongest binder but also the most potent inhibitor, while 2h (R, R) had the lowest inhibitory potency (IC₅₀ of 0.10 μM and 0.32 μM, respectively). Other analogues in the subgroup had intermediate IC₅₀ values: 0.18-0.22 μM vs 0.15 μM for the parent 10a. Thermal denaturation and heme bleaching assays showed that 2f was the most effective CYP3A4 protector as well: ΔT_m of 7.7 °C and 21% of heme decay. For other analogues, the T_m increase was 1.4-2.0 ° lower, while the heme loss was 5-13% higher (Table 1). Overall, 2f was the lead compound, yet inferior to 10a only in the ability to protect CYP3A4 from destruction by H₂O₂.

Figure 5.

Spectral properties and the binding modes of the R₂-indole/naphthalene containing analogues. (A-E) Spectral changes induced in CYP3A4 by 2e-i. Absorbance spectra of ferric ligand-free and inhibitor-bound CYP3A4 are in black and red, respectively. Spectra of ferrous ligand-bound CYP3A4 and its CO-adduct are in green and blue, respectively. Insets a show the difference spectra recorded during equilibrium titrations; insets b - titration plots with quadratic fittings. The derived K_s values are listed in Table 1. (F-H) Crystallographic binding modes of 2e-g. The central portion of the I-helix and F304 in the inhibitory complexes and water-bound CYP3A4 (5VCC structure) are depicted in gray and black, respectively. Polder omit electron density maps contoured at 3σ level are shown as green mesh. Blue dotted lines are H-bonds between the backbone carbonyl oxygen and the hydroxyl group of the active site S119. In 2g, an intramolecular H-bond is formed between the amide nitrogen of the head-group linker and the terminal carbonyl. (I) Superposition of 2e-g bound CYP3A4 showing distinct placement of the side- and end-groups. The 2e-, 2f-, and 2g-bound structures have PDB ID codes 9COV, 9COW, and 9COX, respectively.

Comparison of 2e-i with the respective Boc-containing 3e-i conformers from series V (Figure 2) was conducted to decipher the impact of R₃-Boc-to-pyridine substitution. For the R₂-indole containing R/S conformers 2e and 3e, introduction of the end-pyridine was not beneficial, as all parameters worsened by ~25-70%. For the R₂-naphthalene containing analogues, the impact of R₃ modification depended on the side-group stereochemistry. For the most favorable R/S configuration (2f), the binding affinity and inhibitory potency were improved by 60% relative to the parent compound 3f; ΔT_m did not change, whereas protection against H₂O₂ decreased by half. A similar trend was observed for the S/R counterpart, 2i, when compared to series V 3g. In contrast, all parameters were worsened to a different extent for 2g (S, S) and 2h (R, R) relative to the respective stereoisomers 3h and 3i (compare corresponding values in Table 1 and Figure 2). Thus, the impact of R₃-Boc-to-pyridine substitution is multidirectional and depends on the chemical nature and stereo configuration of the side-groups.

Crystal structures of CYP3A4 bound to 2e-g

To obtain further insights into the ligand binding modes and better understand the experimental results, we attempted to determine X-ray structures of the inhibitory complexes with all subset compounds but succeeded in co-crystallization of CYP3A4 only with 2e-g. The structures were solved to 2.30-2.72 Å resolution (Table S2), and the ligand orientations are shown and compared in Figure 5F-I.

All three bi-pyridine analogues ligate via the head-pyridine, have a traditional side-group orientation and establish strong H-bonds with S119 (Table 2). The R₂-indole containing 2e (R, S) coordinates most strongly to the heme, forming the shortest and perpendicular Fe-N bond. Further, 2e and the R₂-naphthalene containing 2f (R, S) cause the largest distortion in the I-helix (Table 2), but their R₁-phenyls insert into the P1 pocket without triggering rotameric changes in F304. The indole and naphthalene R₂-groups, in turn, orient perpendicular to each other (Figure 5I), possibly due to a long-range polar interaction formed between the 2e indole nitrogen and the carbonyl oxygen of R372 (3.7 Å away). Such indole orientation promotes cation-π interaction with the R105 guanidine, which is weakened for the 2f naphthalene because of its minimal overlap with the R105 side chain. However, due to a different backbone curvature, 2f but not 2e can form S-π interaction between the backbone sulfur atom and the F108 ring. Finally, in both analogues, the R₃-pyridine is directed toward the substrate channel and forms T-shaped π-π stacking interaction with F108.

The S/S conformer 2g orients distinctly from 2f (R, S). First, the 2g R₁-phenyl triggers rotameric changes in F304. As a result, the 2g tail bends toward the I-helix to fill the void, which promotes T-shaped π-π stacking interaction between the R₃-pyridine and F304 rings (Figure 5H, I). This conformation is stabilized by the intramolecular H-bond formed between the amide nitrogen of the head-group linker and the end-group carbonyl (2.7 Å distance). Second, the R₃ movement drags the R₂-naphthalene up and further from the heme and heme-ligating pyridine, weakening hydrophobic and aromatic interactions. The cation-π interaction with R105 is weakened as well due to unfavorable, near perpendicular orientation of the naphthalene ring. These distinctive features could explain why 2f is more potent than 2g and better protects CYP3A4 from thermal denaturation and hydrogen peroxide bleaching.

Pairwise comparison of 2e-g with stereoisomeric parent compounds provides further insights on the relative impact of the R₂ and R₃ functionalities on ligand binding. As seen from Figure 6A, folding of the R₂-indole containing R/S conformers 2e and series V 3e (Samuels & Sevrioukova, 2021) is very similar, meaning that, in this case, the R₃-Boc-to-pyridine substitution has no significant effect on the ligand orientation. For the R₂-naphthalene containing R/S pair, 2f and series V 3f (Figure 6B), the same modification triggers a 90° swing and relocation of R₃ from the side of the active site to the substrate channel. In the S/S conformer 2g, introduction of R₃-pyridine leads to more drastic conformational changes and alters the backbone curvature and the R₁, R₂ and R₃ placement.

Figure 6.

Superposition of 2e-g with stereoisomers 3e, 3f and 3h from series V (Samuels & Sevrioukova, 2021) and 10a from series VI ²⁰ to demonstrate how substitution of R₂ and R₃ functionalities changes the ligand binding mode. (A-C) Effects of R₃-Boc-to-pyridine substitution. For the R₂-indole containing 2e-3e pair (A), the R₃ replacement has no significant effect on the ligand binding mode. When R₂ is naphthalene (B, C), there is positional change in R₃-pyridine, most notable in the S/S conformer 2g, where the end-group swings in the opposite direction toward to the I-helix. Concurrently, both side-groups and the heme-ligating pyridine of 2g rotate by 90-130° and 30°, respectively. (D, E) Effects of R₂-phenyl-to-indole/naphthalene substitution when R₃ is pyridine. For the 2e-10a pair (D), positional shifts in R₂ and R₃ are evident and likely caused by the 2e R₂-indole movement toward R372 to establish a long-range polar interaction between the indole nitrogen and the main chain carbonyl. In contrast, as seen from the 2f and 10a overlay (E), replacement of R₂-phenyl with naphthalene has virtually no effect on the ligand conformation. Direction of conformational changes is shown by arrows. The previously reported 3e-, 3f-, 3h- and 10a-bound structures of CYP3A4 have PDB ID 7KVN, 7KVO, 7KVQ and 7UFE, respectively.

Relative to the R₃-Boc-containing stereoisomer 3h from series V (Figure 6C), the heme-ligating pyridine and both side-groups of 2g rotate by 30° and 90-130°, respectively, while the tail swings in the opposite direction toward the I-helix. When R₃ is pyridine, the ligand binding mode is also R₂-dependent. As seen from an overlay of 2e and series VI 10a, the R₂-phenyl-to-indole replacement alters positioning of R₃ and, to a lesser extent, R₂ groups (Figure 6D). In contrast, the R₂-phenyl-to-naphthalene switch (2f-10a pair) has virtually no effect on the ligand conformation (Figure 6E). This demonstrates the multifactorial control and interdependent abilities of the R₂ and R₃ functionalities to influence the ligand binding mode. Based on comparative analysis, it can be ultimately concluded that for small and chemically simple R₃ groups such as Boc and pyridine, the outcome is largely defined by the size and stereochemistry of the R₂ moiety.

Interaction of CYP3A4 with 5a-b, 6 and 13

Spectral and inhibitory properties

This subset was designed to test whether further increase in size/hydrophobicity of R₂ could improve the binding and inhibitory strength. One such functionality was elongated biphenyl in R/S and S/S configurations coupled with R₃-Boc (5a and 5b, respectively). The R₂-biphenyl/R₃-pyridine S/S stereoisomer 6 was also produced to obtain further insight into the R₂/R₃ interplay. The second tested substituent was fluorene, for added lateral bulk, coupled with R₃-Boc (compound 13 (R, rac)).

All analogues from this subset produced type II spectral changes in CYP3A4 (Figure 7). Based on spectral and functional data, 5b was the strongest binder/inhibitor in the subset; 13 was the weakest binder/inhibitor in the entire series (K_s and IC₅₀ of 0.734 μM and 1.32 μM, respectively), while 6 had the lowest stabilizing effect (ΔT_m of 3.0 °C) and the second lowest ability to protect the heme from destruction by H₂O₂ (Table 1). These findings suggest that the large fluorene moiety markedly decreases the ability of 13 to approach the heme, thereby lowering the binding affinity and, consequently, the inhibitory strength. The R₂-biphenyl, in turn, allows unrestricted access, but has a lower capacity to protect CYP3A4 from thermal denaturation and heme destruction as effectively as other R₂ substituents do, likely due to its distinct/sub-optimal orientation in the active site.

Figure 7.

Spectral changes induced in CYP3A4 by 5a-b, 6 and 13 (A-D, respectively). Absorbance spectra of ferric ligand-free and inhibitor-bound CYP3A4 are in black and red, respectively. Spectra of ferrous ligand-bound CYP3A4 and its CO-adduct are in green and blue, respectively. Insets a show the difference spectra recorded during equilibrium titrations; insets b - titration plots with quadratic fittings. The derived K_s values are listed in Table 1.

Crystal structure of 5a-bound CYP3A4

Determination of co-crystal structures of CYP3A4 with the biphenyl/fluorene containing compounds was highly desired as it could provide valuable insights into protein-ligand interactions. However, only the 2.65 Å structure of 5a-bound CYP3A4 was solved (Table S2), where R₂-biphenyl was not seen due to thermal disorder. Even so, it is evident that 5a adopts a nontraditional orientation, with R₃-Boc occupying the P2 site and R₂-biphenyl likely pointing toward the substrate access channel (Figure 8).

Figure 8.

The binding mode of 5a observed in the crystal structure (PDB ID 9COY). The R₂ and R₃ groups switch places: the Boc tail binds at the P2 site near the head pyridine, while R₂-biphenyl is disordered and likely points toward the substrate channel (shown in arbitrary orientation). The central part of the I-helix and F304 in the inhibitor- and water-bound CYP3A4 (5VCC structure) are depicted in gray and black, respectively. Polder omit electron density map contoured at 3σ level is shown as green mesh.

That R₃-Boc can occupy the P2 site was demonstrated earlier with analogues lacking R₂ or both side-groups (Kaur et al., 2016). Positional switch in R₂ and R₃ was also previously observed in the R₂-indole containing compounds, where the ligand conformation was stabilized by the newly formed H-bond between the indole nitrogen and the R372 carbonyl (E. R. Samuels & I. F. Sevrioukova, 2018). For 5a, the R₂/R₃ rearrangement is likely triggered by spatial limitations and the inability of biphenyl to fit into the P2 site without disrupting coordination to the heme. With the biphenyl pointing upward toward the substrate channel, 5a can ligate strongly to the heme, H-bond to S119, and establish aromatic interactions via the R₁-moiety similarly to other analogues (Table 2). Thus, the reverse R₂-biphenyl/R₃-Boc orientation is beneficial and allows 5a to maintain the relatively high binding affinity and inhibitory strength.

Comparison of 2f and ritonavir

2f has emerged as the series lead compound and one of the strongest binders/inhibitors designed thus far, with K_s and IC₅₀ of 0.009 μM and 0.10 μM, respectively. Considering the concentration of CYP3A4 in the functional assays (0.1 μM) and equimolar (1:1) stoichiometry of the inhibitory complex, the IC₅₀ would be limited to 0.05 μM under our experimental conditions. The dissociation constant, on the other hand, does not depend on protein concentration and could have any value. One problem is that the quadratic (Morrison) equation for tight ligand binding, used for spectral data analysis, can be reliably applied when K_s is ≥1/100th of enzyme concentration (Murphy, 2004). In our spectral assays, CYP3A4 concentration was 2 μM and, thus, the accuracy limit for K_s would be 0.020 μM, with the lower values being potentially underestimated.

One way to overcome this limitation is to lower protein concentration. However, this would increase the noise level and, hence, was not attempted. Another approach for evaluation and comparison of dissociation constants for tight binders is displacement titrations, where the protein is first complexed with a dissociable ligand and then titrated with a stronger binder. We utilized this approach for monitoring dissociation of various substrates from CYP3A4 upon addition of ritonavir (I. F. Sevrioukova, 2019). Mibefradil was most resistant to displacement by ritonavir and, therefore, this substrate was used for displacement titrations with 2f (Figure 9). It was anticipated that the dissociation constant of 2f for mibefradil-bound CYP3A4 (K_s^mib) would be substantially higher than for the ligand-free protein (K_s of 0.009 μM). However, the K_s^mib and K_s values were found to be equal, meaning that the binding affinity of 2f for ligand-free CYP3A4 is, indeed, underestimated. For comparison, for ritonavir, K_s^mib is >40-fold higher than K_s: 0.390 μM vs 0.019 μM. This implies that 2f binds to CYP3A4 much tighter and, as a result, displaces mibefradil more easily than ritonavir.

Figure 9.

Equilibrium titration of mibefradil-bound CYP3A4 with 2f and ritonavir. Mibefradil is a substrate that does not coordinate to the heme iron and can be displaced from the active site by stronger ligands. Direction of spectral changes is indicated by arrows. Left inset shows the difference spectra recorded during titration with 2f; analogous spectra for ritonavir were similar and not shown. Right inset shows titration plots with quadratic fittings. The derived K_s^mib values are indicated.

CONCLUSIONS

This work was a continuation of our studies on SARs of ritonavir-like compounds to identify structural attributes required for potent inhibition of human drug-metabolizing CYP3A4. Thirteen R₁-phenyl containing analogues differing in the side- and end-group functionalities and the length of the head-group linker were designed and spectrally and functionally characterized. Determination of crystal structures of the inhibitory complexes with eight compounds helped define the ligand binding modes and carry out structural comparisons with the parent compounds for deciphering the impact of R₂ and R₃ modifications.

As expected, the R₂-phenyl containing 2a was one of the weakest binders and stabilizers of CYP3A4 due to a shorter, suboptimal pyridyl-ethyl linker. Elongation and fluorination of R₃-Boc or introduction of benzenesulfonyl as the end-moiety altered the backbone curvature and the side-group placement but did not lead to notable improvements in the binding and inhibitory strength.

When R₃ was represented by pyridine, the phenyl-to-indole/naphthalene substitution in R₂ had a multidirectional impact, highly dependent on the side-group stereo configuration. In accord with the earlier findings, the R₂-naphtalene containing R/S and S/S conformers, 2f and 2g, had the lowest K_s and IC₅₀. Moreover, 2f (R, S) was identified as the series lead compound and one of the strongest inhibitors designed thus far, with a ≥2-fold lower IC₅₀ relative to that of ritonavir and other analogues.

Further increase in size/hydrophobicity of R₂ was achieved by introducing biphenyl and fluorene moieties but did not lead to any improvements. Conversely, the fluorene-containing 13 had the highest K_s and IC₅₀ in the series, meaning that the fluorene ring is too large to allow unrestricted access to the active site or has a suboptimal linkage site. The R₂-biphenyl/R₃-pyridine containing 6, in turn, had the lowest ability to protect CYP3A4 from thermal denaturation and peroxide bleaching. However, when coupled with R₃-Boc, R₂-biphenyl had a lower negative impact on all parameters tested. As structural data suggest, this could be due to reverse R₂/R₃ orientation, with the Boc end-group binding near the heme and the biphenyl side-group facing the substrate channel.

Collectively, the SAR analysis confirms the importance of an elongated head-group linker, needed for minimization of steric constrains on the heme-ligating group, identifies naphthalene as the most optimal R₂ functionality when coupled with R₁-phenyl in R/S configuration, and suggests that introduction of heterocyclic or other poly-functional substituents of R₃ attached through a flexible linker could further enhance the inhibitory potency for CYP3A4. Moreover, our findings clearly demonstrate the interdependent ability of the R₂ and R₃ functionalities to influence the ligand binding mode, with the side-group size and stereo configuration being the major defining factors. Once again, determination of crystal structures of the inhibitory complexes was critical for deciphering the R₂/R₃ interplay, which must be taken into consideration during analogue optimization.

METHODS

General chemistry methods

1H NMR spectra were recorded on Bruker DRX 400 MHz, Bruker DRX 500 MHz, or Bruker Avance 600 MHz spectrometer. Chemical shifts (δ) are reported in ppm for the solution of compound in CD₃OD or CDCl₃ (with internal reference TMS), and J values in hertz. NMR data were processed using TopSpin 3.5 software or ACD/Spectrus Processor 2018.2.5. LRMS and HRMS data were obtained via ESI LC-TOF on a Waters (Micromass) LCT Premier spectrometer (Waters), with PEG as the calibrant for HRMS. Optical rotation was recorded on a Rudolph Autopol III Automatic Polarimeter at room temperature in MeOH. Purity of final products was verified by NMR with TMS as a standard. TLC was performed using EMD Millipore silica gel 60 F₂₅₄ aluminum plates. Separation by column chromatography was conducted using Fisher silica gel 60 (230-400 mesh). All reactions were performed with commercially available reagents (Aldrich, Thermo-Fisher, Alfa Aesar, Acros, Oakwood, Millipore). Anhydrous solvents were acquired through a solvent purification system (Inert PureSolv and JC Meyer systems) or purified according to standard procedures.

Synthesis of Analogues

Compounds 1a-g were prepared as described previously (Series III, IV, V) (Samuels & Sevrioukova, 2019; Samuels & Sevrioukova, 2020, 2021), starting with commercially available or synthesized amino alcohol and D or L-α-thio-phenylalanine (E. Samuels & I. Sevrioukova, 2018) followed by coupling with an amino alkyl pyridine. Boc deprotection of 1a-g to form the intermediate material for 2a-i was also performed as described previously (Series VI) (Samuels & Sevrioukova, 2022).

General Procedure for Synthesis of Compounds 2a-b

The crude Boc deprotected, ethylpyridyl thioether (0.1g, 0.24 mmol) was dissolved in DMF (4 ml). To this solution, EDAC (0.07 g, 0.36 mmol, 1.5 eq) and HOBt (0.055 g, 0.36 mmol, 1.5 eq) were added, followed by the addition of N-(trifluoroacetyl)glycine (0.041 g, 0.24 mmol, 1 eq) - prepared from ethyl trifluoroacetate and glycine (Reay, Williams, & Fairlamb, 2015) - and DIPEA (0.093 g, 0.72 mmol, 3 eq). The reaction was stirred at room temperature overnight. Upon completion, the solvent was evaporated and the reaction mixture was diluted with ethyl acetate. The organic layer was then washed with saturated NaHCO3, water, and brine. The combined organic layers were dried over MgSO₄ and concentrated in vacuo to give the crude product, which was purified via column chromatography (95:5 EtOAc:MeOH). The pure product 2a was obtained as a light yellow fluffy solid (0.078g, 57%). TLC: EtOAc/MeOH 90:10 (Rf. 0.47). ¹H NMR (400 MHz, CDCl₃) δ 8.37 (d, J = 4.8 Hz, 1H), 8.28 (s, 1H), 8.07 (t, J = 6.2 Hz, 1H), 7.43 (d, J = 7.9 Hz, 1H), 7.31-7.14 (m, 7H), 7.10 (d, J = 7.1 Hz, 2H), 6.91 (d, J = 8.3 Hz, 1H), 6.48 (t, J = 5.7 Hz, 1H (NH)), 4.25 (q, J = 6.7 Hz, 1H), 3.94 (dd, J = 5.2, 16.7 Hz, 1H), 3.86 (dd, J = 4.8, 16.7 Hz, 1H), 3.50 (dd, J = 6.8, 13.6 Hz, 1H), 3.39 (m, 2H), 3.17 (dd, J = 7.2, 13.8 Hz, 1H), 2.88 (dd, J = 6.9, 14.1 Hz, 1H), 2.80 (d, J = 6.8 Hz, 2H), 2.71 (q, J = 6.3 Hz, 2H), 2.63 (dd, J = 5.0, 13.4 Hz, 1H), 2.50 (dd, J = 7.3, 13.2 Hz, 1H). HRMS m/z calculated for C₂₉H₃₂F₃N₄O₃S [M + H]⁺: 573.2147. Found: 573.2156. 1b afforded the pure product 2b as a yellow fluffy solid (0.087g, 64%). TLC: EtOAc/MeOH 90:10 (Rf. 0.48). 1H NMR (400 MHz, CDCl₃) δ 8.37 (d, J = 4.9 Hz, 1H), 8.33 (s, 1H), 8.15 (t, J = 5.8 Hz, 1H), 7.46 (d, J = 7.8 Hz, 1H), 7.29-7.16 (m, 7H), 7.11 (d, J = 7.0 Hz, 2H), 6.87 (d, J = 8.1 Hz, 1H), 6.55 (t, J = 5.8 Hz, 1H (NH)), 4.27 (q, J = 6.7 Hz, 1H), 3.93 (dd, J = 4.8, 16.6 Hz, 1H), 3.86 (dd, J = 4.6, 16.5 Hz, 1H), 3.49 (t, J = 7.7 Hz, 1H), 3.20 (m, 2H), 3.10 (dd, J = 6.4, 13.3 Hz, 1H), 2.92 (dd, J = 7.1, 13.7 Hz, 1H), 2.83 (t, J = 5.9 Hz, 2H), 2.73 (dd, J = 5.2, 13.7 Hz, 1H), 2.63 (dd, J = 6.6, 13.6 Hz, 1H), 2.48 (t, J = 7.5 Hz, 2H), 1.69 (quint, J = 7.3 Hz, 2H). HRMS m/z calculated for C₃₀H₃₄F₃N₄O₃S [M + H]⁺: 587.2303. Found: 587.2271.

Synthesis of Compound 2c

The crude Boc deprotected, propylpyridyl thioether (0.05 g, 0.12 mmol) was dissolved in MeOH (3 ml). To this solution, triethylamine (0.015 g, 0.14 mmol, 1.2 eq), followed by ethyl trifluoroacetate (0.02 g, 0.14 mmol, 1.2 eq) were slowly added at 0°C. The ice bath was removed and the reaction was left to stir at room temperature overnight. Upon completion, the solvent was evaporated and the reaction mixture was diluted with ethyl acetate. The organic layer was then washed with saturated NaHCO₃, water, and brine. The combined organic layers were dried over MgSO₄ and concentrated in vacuo to give the crude product, which was purified via column chromatography (95:5 EtOAc:MeOH). The pure product 2c was obtained as a light yellow solid (0.02g, 31%). TLC: EtOAc/MeOH 90:10 (Rf. 0.6). 1H NMR (400 MHz, CDCl₃) δ 8.42 (d, J = 4.7 Hz, 1H), 8.37 (s, 1H), 7.44 (d, J = 7.8 Hz, 1H), 7.37 (m, 1H), 7.31-7.13 (m, 10H), 6.07 (t, J = 6.0 Hz, 1H (NH)), 4.28 (sext, J = 6.7 Hz, 1H), 3.41 (t, J = 7.3 Hz, 1H), 3.23 (m, 2H), 3.14 (m, 1H), 2.93 (m, 3H), 2.80 (dd, J = 5.1, 13.8 Hz, 1H), 2.69 (dd, J = 7.2, 13.7 Hz, 1H), 2.49 (t, J = 7.6 Hz, 2H), 1.70 (q, J = 7.4 Hz, 2H). HRMS m/z calculated for C₂₈H₃₁F₃N₃O₂S [M + H]⁺: 530.2089. Found: 530.2090.

Synthesis of Compound 2d

The Boc deprotected, propylpyridyl thioether (0.05 g, 0.12 mmol) was dissolved in dry DCM (3 ml). To this solution, benzylsulfonyl chloride (0.032 g, 0.18 mmol, 1.5 eq) and triethylamine (0.036 g, 0.36 mmol, 3 eq) were slowly added at 0°C. The reaction was allowed to gradually come to room temperature overnight. Upon completion, the solvent was evaporated and the reaction mixture was diluted with ethyl acetate. The organic layer was then washed with saturated NaHCO₃, water, and brine. The combined organic layers were dried over MgSO₄ and concentrated in vacuo to give the crude product, which was purified via column chromatography (95:5 EtOAc:MeOH). The pure product 2d was obtained as a light orange solid (0.034g, 49%). TLC: EtOAc/MeOH 90:10 (Rf. 0.59). 1H NMR (400 MHz, CDCl₃) δ 8.43 (d, J = 4.8 Hz, 1H), 8.39 (s, 1H), 7.60 (d, J = 7.2 Hz, 1H), 7.52 (t, J = 7.5 Hz, 1H), 7.45 (d, J = 7.9 Hz, 1H), 7.38 (t, J = 7.8 Hz, 2H), 7.32 (m, 1H), 7.29-7.13 (m, 10H), 6.88 (m, 1H), 6.40 (t, J = 5.7 Hz, 1H (NH)), 5.04 (d, J = 7.1 Hz, 1H (NH)), 3.64 (t, J = 7.4 Hz, 1H), 3.57 (m, 1H), 3.23 (m, 3H), 2.93 (dd, J = 7.1, 13.7 Hz, 1H), 2.85 (dd, J = 5.0, 14.0 Hz, 1H), 2.66 (m, 3H), 2.49 (m, 2H), 1.71 (m, J = 7.4 Hz, 2H). HRMS m/z calculated for C₃₂H₃₆N₃O₃S₂ [M + H]⁺: 574.2198. Found: 574.2200.

General Procedure for Synthesis of Compounds 2e-i

The Boc deprotected, propylpyridyl thioether (0.025 g, 0.053 mmol) was dissolved in DMF (2 ml). To this solution, EDAC (0.015 g, 0.08 mmol, 1.5 eq) and HOBt (0.012 g, 0.08 mmol, 1.5 eq) were added, followed by the addition of nicotinic acid (0.0065 g, 0.053 mmol, 1 eq) and DIPEA (0.02 g, 0.16 mmol, 3 eq). The reaction was stirred at room temperature overnight. Upon completion, the solvent was evaporated and the reaction mixture was diluted with ethyl acetate. The organic layer was then washed with saturated NaHCO₃, water, and brine. The combined organic layers were dried over MgSO₄ and concentrated in vacuo to give the crude product, which was purified via column chromatography (90:10 EtOAc:MeOH). The pure product 2e was obtained as an off white solid (0.011g, 36%). TLC: EtOAc/MeOH 90:10 (Rf. 0.18). 1H NMR (400 MHz, CDCl₃) δ 8.86 (s, 1H), 8.71 (d, J = 4.8 Hz, 1H), 8.42 (d, J = 4.8 Hz, 1H), 8.34 (s, 1H), 7.98 (d, J = 7.8 Hz, 1H), 7.66 (d, J = 7.8 Hz, 1H), 7.42 (d, J = 7.6 Hz, 1H), 7.34 (m, 2H), 7.28-7.10 (m, 7H), 7.03 (s, 1H), 6.66 (d, J = 7.2 Hz, 1H), 6.30 (t, J = 5.9 Hz, 1H (NH)), 4.58 (quint, J = 6.5 Hz, 1H), 3.47 (m, 1H), 3.22 (dd, J = 7.7, 13.8 Hz, 2H), 3.15 (dd, J = 6.6, 10.5 Hz, 2H), 3.04 (quint, J = 6.6 Hz, 1H), 2.94 (dd, J = 6.9, 13.8 Hz, 1H), 2.86 (dd, J = 6.0, 13.7 Hz, 1H), 2.79 (dd, J = 6.3, 13.7 Hz, 1H), 2.44 (t, J = 7.7 Hz, 2H), 1.61 (quint, J = 7.6 Hz, 2H). HRMS m/z calculated for C₃₄H₃₆N₅O₂S [M + H]⁺: 578.2590. Found: 578.2594. 1d afforded the pure product 2f as a white fluffy solid (0.03g, 41%). TLC: EtOAc/MeOH 90:10 (Rf. 0.18). 1H NMR (400 MHz, CDCl₃) δ 8.88 (s, 1H), 8.70 (d, J = 4.8 Hz, 1H), 8.40 (d, J = 5.0 Hz, 1H), 8.30 (d, J = 8.6 Hz, 1=2H), 8.00 (d, J = 7.9 Hz, 1H), 7.84 (d, J = 8.2 Hz, 1H), 7.74 (d, J = 8.0 Hz, 1H), 7.58 (t, J = 7.6 Hz, 1H), 7.49 (t, J = 7.4 Hz, 1H), 7.35 (m, 4H), 7.24-7.12 (m, 5H), 6.99 (d, J = 7.2 Hz, 1H), 6.28 (t, J = 5.9 Hz, 1H (NH)), 4.63 (sext, J = 6.5 Hz, 1H), 3.55 (dd, J = 6.7, 13.9 Hz, 1H), 3.43 (t, J = 7.4 Hz, 1H), 3.35 (dd, J = 7.6, 13.9 Hz, 1H), 3.19 (m, 2H), 2.94 (m, 2H), 2.86 (m, 2H), 2.38 (t, J = 7.8 Hz, 2H), 1.56 (quint, J = 7.4 Hz, 2H). HRMS m/z calculated for C₃₆H₃₇N₄O₂S [M + H]⁺: 589.2637. Found: 589.2620. 1e afforded the pure product 2g as a white fluffy solid (0.032g, 44%). TLC: EtOAc/MeOH 90:10 (Rf. 0.17). 1H NMR (400 MHz, CDCl₃) δ 8.90 (s, 1H), 8.65 (d, J = 4.9 Hz, 1H), 8.38 (d, J = 5.0 Hz, 1H), 8.28 (s, 2H), 7.97 (d, J = 8.4 Hz, 1H), 7.88 (d, J = 8.0 Hz, 1H), 7.77 (d, J = 8.2 Hz, 1H), 7.58 (t, J = 7.5 Hz, 1H), 7.52 (t, J = 7.4 Hz, 1H), 7.42-7.24 (m, 5H), 7.24-7.13 (m, 5H), 6.70 (t, J = 5.6 Hz, 1H (NH)), 4.61 (sext, J = 6.5 Hz, 1H), 3.62 (t, J = 7.8 Hz, 1H), 3.46 (d, J = 6.9 Hz, 2H), 3.26 (dd, J = 8.7, 13.6 Hz, 1H), 3.15 (dd, J = 6.5, 13.5 Hz, 1H), 3.08 (dd, J = 6.6, 13.3 Hz, 1H), 2.93 (dt, J = 6.1, 12.6 Hz, 2H), 2.64 (dd, J = 6.2, 14.2 Hz, 1H), 2.38 (t, J = 7.8 Hz, 2H), 1.59 (m, J = 7.6 Hz, 2H). HRMS m/z calculated for C₃₆H₃₇N₄O₂S [M + H]⁺: 589.2637. Found: 589.2609. 1f afforded the pure product 2h as a white fluffy solid (0.023g, 56%). TLC: EtOAc/MeOH 90:10 (Rf. 0.16). 1H NMR (400 MHz, CDCl₃) δ 8.88 (s, 1H), 8.68 (d, J = 5.2 Hz, 1H), 8.39 (d, J = 4.4 Hz, 1H), 8.29 (d, J = 8.4 Hz, 2H), 7.97 (d, J = 8.0 Hz, 1H), 7.88 (d, J = 8.0 Hz, 1H), 7.78 (d, J = 8.3 Hz, 1H), 7.59 (t, J = 7.6 Hz, 1H), 7.53 (t, J = 7.3 Hz, 1H), 7.36 (m, 4H), 7.25-7.13 (m, 5H), 6.99 (d, J = 7.2 Hz, 1H), 6.59 (t, J = 5.8 Hz, 1H (NH)), 4.61 (sext, J = 6.6 Hz, 1H), 3.61 (t, J = 7.5 Hz, 1H), 3.47 (m, 2H), 3.27 (dd, J = 8.6, 13.8 Hz, 1H), 3.18 (sext, J = 6.7 Hz, 1H), 3.08 (sext, J = 6.6 Hz, 1H), 2.94 (dt, J = 6.2, 14.5 Hz, 2H), 2.65 (dd, J = 6.1, 14.0 Hz, 1H), 2.40 (t, J = 8.3 Hz, 2H), 1.61 (quint, J = 7.4 Hz, 2H). HRMS m/z calculated for C₃₆H₃₇N₄O₂S [M + H]⁺: 589.2637. Found: 589.2643. 1g afforded the pure product 2i as an off white fluffy solid (0.021g, 51%). TLC: EtOAc/MeOH 90:10 (Rf. 0.21). 1H NMR (400 MHz, CDCl₃) δ 8.88 (s, 1H), 8.70 (d, J = 4.8 Hz, 1H), 8.40 (d, J = 4.6 Hz, 1H), 8.30 (d, J = 8.9 Hz, 2H), 8.00 (d, J = 7.7 Hz, 1H), 7.84 (d, J = 7.8 Hz, 1H), 7.74 (d, J = 7.9 Hz, 1H), 7.58 (t, J = 7.6 Hz, 1H), 7.49 (t, J = 7.6 Hz, 1H), 7.35 (m, 4H), 7.24-7.12 (m, 5H), 6.93 (d, J = 6.9 Hz, 1H), 6.22 (t, J = 5.5 Hz, 1H (NH)), 4.63 (sext, J = 6.9 Hz, 1H), 3.56 (dd, J = 6.6, 13.9 Hz, 1H), 3.42 (t, J = 7.3 Hz, 1H), 3.35 (dd, J = 7.4, 14.2 Hz, 1H), 3.20 (m, 2H), 2.97 (dt, J = 6.8, 20.0 Hz, 2H), 2.87 (m, 2H), 2.39 (t, J = 8.3 Hz, 2H), 1.57 (quint, J = 7.4 Hz, 2H). HRMS m/z calculated for C₃₆H₃₇N₄O₂S [M + H]⁺: 589.2637. Found: 589.2653.

General Procedure for Synthesis of Compounds 5a-b

Compound 3 was prepared via reduction of L-biphenylalanine, followed by boc protection and tosylation using procedures described previously. ^{(Samuels & Sevrioukova, 2019; Samuels & Sevrioukova, 2020)} Compounds 4a-b were also prepared as described earlier, (Samuels & Sevrioukova, 2019; Samuels & Sevrioukova, 2020) with either D or L-α-thio-phenylalanine. (E. Samuels & I. Sevrioukova, 2018) Boc deprotection of 5b to form the intermediate material for 6 was performed as described elsewhere (Samuels & Sevrioukova, 2022).

Crude 4a (0.09 g, 0.18 mmol) was dissolved in DMF (3.5 ml). To this solution, EDAC (0.053 g, 0.27 mmol, 1.5 eq) and HOBt (0.041 g, 0.27 mmol, 1.5 eq) were added, followed by the addition of 3-(3-pyridyl)propylamine (0.029 g, 0.21 mmol, 1.2 eq) and DIPEA (0.07 g, 0.54 mmol, 3 eq). The reaction was stirred at room temperature overnight. Upon completion, the solvent was evaporated and the reaction mixture was diluted with ethyl acetate. The organic layer was then washed with saturated NaHCO3, water, and brine. The combined organic layers were dried over MgSO₄ and concentrated in vacuo to give the crude product, which was purified via column chromatography (95:5 EtOAc:MeOH). The pure product 5a was obtained as an light yellow solid (0.03g, 27%). TLC: EtOAc/MeOH 90:10 (Rf. 0.67). ¹H NMR (400 MHz, CDCl₃) δ 8.43 (d, J = 4.7 Hz, 1H), 8.36 (s, 1H), 7.57-7.49 (dd, J = 18.4, 8.3 Hz, 4H), 7.39-7.46 (q, J = 8.1 Hz, 3H), 7.37-7.16 (m, 9H), 6.44 (bs, 1H (NH)), 4.60 (m, 1H), 4.00 (m, 1H), 3.50 (t, J = 7.1 Hz, 1H), 3.32-3.10 (m, 3H), 2.98 (dd, J = 13.6, 7.0 Hz, 1H), 2.83 (d, J = 6.9 Hz, 2H), 2.70 (dd, J = 5.3, 13.7 Hz, 1H), 2.63 (dd, J = 5.8, 13.5 Hz, 1H), 2.46 (t, J = 7.9 Hz, 2H), 1.69 (quint (o), J = 7.4 Hz, 2H), 1.40 (s, 9H). HRMS m/z calculated for C₃₇H₄₄N₃O₃S [M + H]⁺: 610.3104. Found: 610.3105. 4b afforded the pure product 5b as a white solid (0.032g, 29%). TLC: EtOAc/MeOH 90:10 (Rf. 0.63). ¹H NMR (400 MHz, CDCl₃) δ 8.43 (d, J = 4.6 Hz, 1H), 8.36 (s, 1H), 7.59-7.49 (dd, J = 19.9, 8.2 Hz, 4H), 7.39-7.46 (q, J = 6.6 Hz, 3H), 7.36-7.13 (m, 9H), 6.88 (bs, 1H (NH)), 4.66 (d, J = 8.0 Hz, 1H), 4.00 (sext, J = 7.3 Hz, 1H), 3.54 (t, J = 7.0 Hz, 1H), 3.33-3.10 (m, 3H), 2.94 (dd, J = 13.8, 7.4 Hz, 1H), 2.73 (dd, J = 14.0, 7.6 Hz, 2H), 2.68 (d, J = 5.6 Hz, 1H), 2.49 (m (o), 3H), 1.69 (quint (o), J = 7.2 Hz, 2H), 1.37 (s, 9H). HRMS m/z calculated for C₃₇H₄₄N₃O₃S [M + H]⁺: 610.3104. Found: 610.3087.

Synthesis of Compound 6

The Boc deprotected, propylpyridyl thioether (0.03 g, 0.059 mmol) was dissolved in DMF (2 ml). To this solution, EDAC (0.017 g, 0.088 mmol, 1.5 eq) and HOBt (0.014 g, 0.088 mmol, 1.5 eq) were added, followed by the addition of nicotinic acid (0.007 g, 0.059 mmol, 1 eq) and DIPEA (0.023 g, 0.18 mmol, 3 eq). The reaction was stirred at room temperature overnight. Upon completion, the solvent was evaporated and the reaction mixture was diluted with ethyl acetate. The organic layer was then washed with saturated NaHCO₃, water, and brine. The combined organic layers were dried over MgSO₄ and concentrated in vacuo to give the crude product, which was purified via column chromatography (90:10 EtOAc:MeOH). The pure product 6 was obtained as an off white solid (0.014g, 39%). TLC: EtOAc/MeOH 90:10 (Rf. 0.17). ¹H NMR (400 MHz, CDCl₃) δ 8.94 (s, 1H), 8.70 (d, J = 4.7 Hz, 1H), 8.42 (d, J = 4.8 Hz, 1H), 8.35 (s, 1H), 8.03 (d, J = 7.8 Hz, 1H), 7.56 (t, J = 8.4 Hz, 4H), 7.44 (t, J = 7.5 Hz, 3H), 7.35 (m, 2H), 7.30-7.13 (m, 9H), 6.76 (d, J = 7.5 Hz, 1H), 6.65 (t, J = 5.7 Hz, 1H (NH)), 4.47 (sext, J = 7.0 Hz, 1H), 3.64 (t, J = 7.7 Hz, 1H), 3.32 (dd, J = 13.2, 7.5 Hz, 1H), 3.21 (q, J = 6.5 Hz, 2H), 3.06-2.92 (m, 3H), 2.85 (dd, J = 13.9, 6.4 Hz, 1H), 2.47 (t, J = 7.6 Hz, 2H), 1.68 (quint, J = 7.0 Hz, 2H). HRMS m/z calculated for C₃₈H₃₉N₄O₂S [M + H]⁺: 615.2794. Found: 615.2787.

Procedures for Synthesis of Compound 13

Synthesis of Compound 7

Compound 7 was synthesized as described by Goehring. (Goehring, 1994) In a three-neck flask, 60% NaH (2.2 g, 55.4 mmol, 2 eq) was washed 2x with hexanes (10 ml) to remove oil. Dry THF (20 ml) was added followed by slow addition of triethylphosphonoacetate (7.5 g, 33.2 mmol, 1.2 eq) in dry THF (10 ml). The solution was stirred for 1 hr at room temperature. After 1 hr, 9-fluorenone (5.0 g, 27.7 mmol), dissolved in THF (20 ml), was slowly added to the to the solution. The solution was stirred for 1 hr at room temperature, followed by reflux at 90C for 2 hrs. Upon reaction completion, the flask was cooled to room temperature and the solvent was evaporated in vacuo. H₂O (300 ml) was added and the crude product was extracted 3x with Et₂O (250 ml). The combined Et₂O layer was washed with brine, dried with MgSO₄, and concentrated in vacuo. The resulting oil was titrated with hexanes affording 7 as an orange solid (3.0g, 43%) which was used without further purification. LRMS m/z calculated for C₁₇H₁₅O₂ [M + H]⁺: 251.1. Found 251.3.

Synthesis of Compound 8

(1) Compound 7 (2.0 g, 8.0 mmol) was dissolved in dry ethyl acetate (25 ml). 10% w/w Pd/C (0.2 g) was added followed by H₂. The reaction was stirred overnight at room temperature. Upon completion, the solution was filtered and the solvent was evaporated in vacuo. The crude product was purified by column chromatography (9:1 Hexanes:EtOAc) affording a clear oil (0.9 g, 44%). LRMS m/z calculated for C₁₇H₁₆O₂Na [M + Na]⁺: 275.1. Found 275.2.

(2) In a three-neck flask, 1.0 g (4.0 mmol) of reduced 7 was dissolved in 20 ml dry toluene and allowed to cool to −78°C on a dry ice bath. 4.5 ml (4.5 mmol, 1.2 eq) 1M DIBAL-H was slowly added to the flask and the mixture was stirred for 2 hrs at −78°C. Upon completion, the reaction was quenched with 6 ml saturated NaHCO₃. The solution was then filtered and diluted with ethyl acetate, which was washed with saturated NaHCO₃, water, and brine. The combined organic layers were dried over MgSO₄ and concentrated in vacuo to give the crude product, which was purified via column chromatography (4:1 Hexanes:EtOAc) and preparative TLC. The pure product 8 was obtained as a clear oil (0.68g, 82%). LRMS m/z calculated for C₁₅H₁₂ONa [M + Na]⁺: 231.1, found 231.3.

Synthesis of Compound 9

Compound 9 was synthesized according to Steiger (Steiger, 1942).

Note: Synthesis of 9 from 8 was found to be more successful using Strecker amino acid synthesis rather than Bucherer–Bergs (KCN, ammonium carbonate) followed by hydrolysis of the hydantoin with NaOH.

Compound 8 (0.44 g, 2.1 mmol) was dissolved in MeOH (4 ml) and added all at once to a solution of sodium cyanide (0.11 g, 2.2 mmol, 1.05 eq) and ammonium chloride (0.12 g, 2.3 mmol, 1.1 eq) in H₂O (4 ml). The reaction was stirred at room temperature overnight. Upon completion, H₂O was added and extracted with ethyl acetate. The combined organic layers were then washed with saturated NaHCO₃, water, brine, dried over MgSO₄ and concentrated in vacuo to give the crude product as a light pink oil, which was used without further purification. LRMS m/z calculated for C₁₆H₁₄N₂Na [M + Na]⁺: 257.1, found 257.2.

The crude aminonitrile was added to dissolved in 6N HCl (15 ml). The mixture was refluxed at overnight. Upon completion the solvent was evaporated and the crude material was titrated with 1:1 hexanes:EtOAc. The precipitate was dried affording 9 as a green-white solid (0.13 g, 24%). LRMS m/z calculated for C₁₆H₁₆NO₂ [M + H]⁺: 254.1, found 254.3.

Synthesis of Compounds 10 and 11

Compounds 10 and 11 were prepared via reduction of the fluorene amino acid (DL-9-Fla-OH), 9 followed by boc protection and tosylation using previously described procedures (Samuels & Sevrioukova, 2019; Samuels & Sevrioukova, 2020). Compound 12 was also prepared as described earlier (Samuels & Sevrioukova, 2019; Samuels & Sevrioukova, 2020) with L-α-thio-phenylalanine (E. Samuels & I. Sevrioukova, 2018).

Synthesis of Compound 13

Crude 11 (0.07 g, 0.14 mmol) was dissolved in DMF (3.5 ml). To this solution, EDAC (0.04 g, 0.21 mmol, 1.5 eq) and HOBt (0.032 g, 0.21 mmol, 1.5 eq) were added, followed by the addition of 3-(3-pyridyl)propylamine (0.023 g, 0.17 mmol, 1.2 eq) and DIPEA (0.034 g, 0.42 mmol, 3 eq). The reaction was stirred at room temperature overnight. Upon completion, the solvent was evaporated and the reaction mixture was diluted with ethyl acetate. The organic layer was then washed with saturated NaHCO3, water, and brine. The combined organic layers were dried over MgSO₄ and concentrated in vacuo to give the crude product, which was purified via column chromatography (95:5 EtOAc:MeOH). The pure product 13 was obtained as a yellow oil (0.0031 g, 3.5%). Due to exceptionally low yield, the sample used for NMR analysis was further dried of residual solvent and used for biochemical analysis/crystallography. TLC: EtOAc/MeOH 90:10 (Rf. 0.63). ¹H NMR (400 MHz, CDCl₃) δ 8.45 (m, 1H), 8.38 (m, 1H), 8.09 (m, 1H), 7.81-7.64 (m, 2H), 7.60 (d, J = 8.7 Hz, 1H), 7.53-7.28 (m, 11H), 6.99, (bs, 1H (NH)), 4.67 (d, J = 8.0 Hz, 1H), 4.50 (t, J = 7.8 Hz, 1H), 4.28 (m, 1H), 4.02, (d, J = 6.3 Hz, 1H), 3.66, (dd, J = 3.7, 11.1 Hz, 1H), 3.55 (m, 1H), 3.29 (m, 2H), 2.98 (m, 1H), 2.79 (m, 2H), 2.42 (m, 2H), 2.35 (m, 1H), 1.87 (m, 2H), 1.70, (solvent overlapped, 9H). HRMS m/z calculated for C₃₈H₄₄N₃O₃S [M + H]⁺: 622.3104. Found: 622.3116.

Biochemical methods and crystal structure determination

Protein expression and purification

Codon-optimized full-length and Δ3-22 CYP3A4 were produced as reported previously (Paquin et al., 2024; I. F. Sevrioukova, 2017). The full-length protein was used for assays and the truncated form for crystallization.

Spectral binding titrations

Equilibrium ligand binding to CYP3A4 was monitored in a Cary 300 spectrophotometer at 23 C° in 0.1 M phosphate buffer, pH 7.4, supplemented with 20% glycerol and 1 mM dithiothreitol. Inhibitors were dissolved in dimethyl sulfoxide (DMSO) and added to a 2 μM protein solution in small aliquots, with the final solvent concentration <2%. Spectral dissociation constants (K_s) were determined from quadratic fits to the titration plots. For the displacement titrations, CYP3A4 was first preincubated for 10 min with 0.1 mM mibefradil and then titrated with the inhibitor. The dissociation constant for the mibefradil-bound form, K_s^mib, was determined from quadratic fits to the titration plots built based on the difference spectra.

Thermal denaturation

Melting curves were recorded in 0.1 M phosphate buffer, pH 7.4, in a Cary 300 spectrophotometer. CYP3A4 (1 μM) was mixed with the inhibitor (20 μM) and incubated for 15 min at room temperature. Thermal denaturation was monitored at 260 nm using a 0.2 °C measurement step, 0.9 °C/min ramp rate, and 50-75 °C temperature range. A denaturation midpoint (melting temperature; T_m) was determined from non-linear fittings to the melting curve as described earlier (Samuels & Sevrioukova, 2019).

Inhibitory assays

Inhibitory potency for the BFC O-debenzylase activity of CYP3A4 was evaluated fluorometrically in a soluble reconstituted system with P450 reductase and the IC₅₀ values were determined as described in detail elsewhere (Samuels & Sevrioukova, 2021).

H₂O₂-dependent heme bleaching

Heme bleaching in ligand-free and inhibitor-bound CYP3A4 (1.6 μM) was monitored at 23 C° in 0.1 M phosphate buffer, pH 7.4. After 10 min preincubation of CYP3A4 with 16 μM inhibitors, heme decay was initiated by adding 10 mM H₂O₂ (final concentration) and monitored at 420 nm for 2 h. The percentage of heme destroyed was calculated relative to that in ligand-free CYP3A4, considered as 100% decay.

Crystallization of the inhibitory complexes

Crystals of Δ3-22 CYP3A4 bound to inhibitors were grown using a sitting drop vapor diffusion method. Prior to crystal setup, CYP3A4 (55-72 mg/ml in 20-100 mM phosphate, pH 7.4) was incubated for 10 min with a two-to-five-fold excess of inhibitor and centrifuged to remove the precipitate. The supernatant containing the inhibitor-bound CYP3A4 (0.4-0.6 μl) was mixed with an equal volume of crystallization solution, containing 6-10% polyethylene glycol 3350 and either 3% taximate pH 6.0-7.0 (Hampton Research; for 2a-d), or 50 mM succinic acid pH 7.0 (for 2f/g). Complexes with 2e and 5a were crystallized against 70% Morpheus II - A7 solution (Molecular Dimensions), containing 0.21 M lithium, sodium and potassium sulphate (each), 0.07 M N-Bis(2-hydroxyethyl)-2-aminoethanesulfonate /triethanolamine buffer pH 7.5, 14% PEG 8000, and 28% 1,5-pentanediol (final concentration). Crystals grew at room temperature overnight and, upon harvesting, were cryoprotected with Paratone-N oil and frozen in liquid nitrogen.

Determination of crystal structures of inhibitory complexes

X-ray diffraction data were collected at the Stanford Synchrotron Radiation Lightsource beamlines 9-2 and 12-2, and the Advanced Light Source beamlines 8.2.1 and 8.2.2. Crystal structures were solved by molecular replacement with PHASER (McCoy et al., 2007) using the 5VCC structure as a search model. Ligands were built with eLBOW (Adams et al., 2010) and manually fit into the density with COOT (Emsley, Lohkamp, Scott, & Cowtan, 2010). The initial models were rebuilt and refined with COOT and PHENIX (Adams et al., 2010). Polder omit electron density maps and local correlation coefficients (Table S3) were calculated with PHENIX. Data collection and refinement statistics are summarized in Tables S1 and S2. The atomic coordinates and structure factors for the 2a-, 2b-, 2c-, 2d-, 2e-, 2f-, 2g- and 5a-bound CYP3A4 were deposited to the Protein Data Bank with the ID codes 9COR, 9COS, 9COT, 9COU, 9COV, 9COW, 9COX, and 9COY, respectively.

Scheme 1:

General Synthesis of 1a-g

Scheme 2:

Synthesis of 2a,b

Scheme 3:

Synthesis of 2c

Scheme 4:

Synthesis of 2d

Scheme 5:

Synthesis of 2e-i

Scheme 6:

Synthesis of 5a-b

Scheme 7:

Synthesis of 6

Scheme 8:

Synthesis of 13

Data Availability Statement

All experimental data generated during this study are included in this article and Supplementary Material. Coordinates and structure factors for the X-ray models of inhibitor-bound CYP3A4 are freely available at the Protein Data Bank (https://www.rcsb.org/).