Probing membrane enhanced protein–protein interactions in a minimal redox complex of cytochrome-P450 and P450-reductase

Abstract

Investigating the interplay in a minimal redox complex of cytochrome-P450 and its reductase is crucial for understanding cytochrome-P450’s enzymatic activity. Probing the hotspots of dynamic structural interactions using NMR revealed the engagement of loop residues from P450-reductase to be responsible for the enhanced affinity of CYP450 towards its obligate redox partner.

Cytochrome P450s (CYP450s) are monooxygenases from the heme protein family that play a vital role in the biosynthesis and metabolism of numerous endogenous compounds and xenobiotics.¹ Their unique ability for detoxification makes them an indispensable target in medicinal and pharmaceutical drug discovery. Although they differ in substrate specificity and enzymatic activity, all CYP450 isoforms share common features in their catalytic cycle. In the classical catalytic cycle of CYP450, the heterolytic cleavage of the dioxygen molecule is associated with simultaneous hydroxylation of substrates and formation of water. CYP450 reductase (CPR) provides the essential first electron whereas the second electron may come directly from CPR or via cytochrome-b₅ (cyt-b₅).^2,3 Despite the availability of high-resolution structures of CYP450 lacking the transmembrane domain,^4,5 the enzymatic activity of membrane-bound full-length (or wild-type) CYP450 and the roles of cell membrane are still unclear. Understanding the structural basis of the catalytic mechanism requires the exploration of dynamic protein–protein interactions in the redox complex machinery.

Mammalian microsomal CYP450s and their redox partners, CPR and cyt-b₅, are highly conserved membrane-bound proteins. CPR is a prototypic member of the diflavin oxidoreductase family containing the cofactors FMN (flavin mononucleotide) and FAD (flavin adenine dinucleotide) as electron carriers for its redox partners, CYP450, cyt-b₅ and heme oxygenase.⁶ Both charged pair-interactions and hydrophobic packing forces have been considered to be responsible for the protein–protein complex association in CYP450–CPR/cyt-b₅.^7–9 The electrostatically driven interaction¹⁰ of CYP450 17A1 with the FMN binding domain (FBD) of CPR supports the steroidogenic hydroxylation of progesterone and pregnenolone followed by the C_17,20-lyase reaction.^11–14 Although previous studies have only focused on the soluble domains of these proteins in the absence of a membrane, knowledge gained on the structure–function correlation of such metabolic enzymes has been useful to improve our understanding on the function of CYP450s.¹⁵ However, insights on the structure based functional mechanism of full-length membrane-bound CYP450s (fl-CYP450) is still limited due to the challenges posed by the lipid environment. Earlier investigations on CYP450 and co-partners have used detergent and phospholipid mixtures for structural and functional characterization.^16–18 On the other hand, it has now been demonstrated that the presence of detergent can inactivate the function of CYP450.¹⁹ To overcome this limitation, recently developed lipid nanodiscs – consisting of a lipid bilayer encased within a membrane scaffold protein (MSP)^20–22 or a peptide (4F or 22A)^23–25 – are well-suited membrane mimetics to study membrane proteins in a near-native environment. In addition to stabilizing the structural folding, the lipid membrane has been shown to facilitate the access of hydrophobic ligands to the active site of CYP450.²⁶ Our recent study on the membrane-anchored minimal redox P450–CPR complex (fl-CYP2B4–fl-FBD) demonstrated the importance of transmembrane domains and the lipid bilayer for functional–structural interactions and electron transfer kinetics.²⁷ Therefore, the membrane effect is indispensable and would be useful in extrapolating the in vitro data (binding, kinetics and structural information) to in vivo pharmacokinetics. As reported in earlier studies, we have used peptide-based nanodiscs as a model membrane mimetic due to their advantages over MSP nanodiscs in their ability to accommodate multiple proteins to study protein–protein interactions, which can occur between the transmembrane and/or globular domains.^27,28 Investigating the interacting hotspot regions of the functionally active membrane anchored full-length redox complex (fl-CYP450–fl-FBD) could aid in the development of novel drug molecules. Moreover, substrate dependent modulation of CYP450 conformation and its interaction with redox partners has been reported for differential activity.^14,29,30 Using NMR spectroscopy and other biophysical techniques, we have investigated the effect of various substrates on the protein–protein interacting interface of a minimal redox complex of CYP450–CPR reconstituted in nanodiscs.

Stepwise reconstitution of fl-CYP450 and fl-FBD in 4F peptide-based nanodiscs was ensured using size-exclusion chromatography (SEC) and dynamic light scattering (DLS) measurements (Fig. 1). The increase in the relative Stokes radius of empty nanodiscs (~4 nm) upon incubation with fl-CYP450 and fl-FBD (~7.0 nm) demonstrates the successful reconstitution of the redox complex in the lipid bilayer. In this study, the two di?erent isoforms of CYP450 (rabbit 2B4 and human 3A4) investigated were found to be functionally active in nanodiscs (as confirmed by a CO-bound reduced protein showing a maximum at 450 nm, Fig. S1, ESI^†) and stable at room temperature for several days.²⁷

Fig. 1.

Reconstitution of fl-CYP450–fl-FBD in peptide-based nanodiscs. Size exclusion chromatography (A) and dynamic light scattering (B) traces of fl-CYP2B4 (black) and fl-CYP2B4–fl-FBD (red) in 4F-DMPC nanodiscs.

Biotransformation processes catalyzed by CYP450 involve a change in the oxidation and spin states of the heme iron. After reconstitution in nanodiscs, we measured the changes in the spin states of the redox complex upon interaction with various substrates (Fig. 2). Based on the optical absorption spectra, substrates can be categorized into two types: type 1 (shift from 417 to 390 nm) and type 2 (shift from 417 to 420 nm). The characteristic Soret absorption band observed from CYP450s is typically a mixture of high-spin (HS) (absorption maximum at B390 nm) and low-spin (LS) states (absorption maximum at B417 nm) (Fig. 2);³¹ depending on the protein preparation and storage conditions, a CYP420 fraction derived from the irreversible loss of the thiolate bond was also present (Table S1, ESI^†). Changes in the UV-Vis absorption spectra of nanodisc-anchored fl-CYP450 (2B4 and 3A4) were monitored upon titration with type-1 substrates, benzphetamine (BZ), t-butylated hydroxytoluene (BHT) and testosterone (Fig. 2 and Fig. S2–S4, ESI^†). In all the titrations, a type I spin-shift was observed with an isosbestic point at 405 nm. All substrates were found to have the capability to shift the low spin state (observed at 417 nm) to a high spin state (observed at 391 nm) as reported previously.²⁸ The initial and final spin states are given in Table S1 (ESI^†). Furthermore, the effect of fl-FBD on the spin state shift behavior was monitored by adding a stoichiometric ratio of fl-FBD to the nanodisc reconstituted fl-CYP450 both in free and substrate bound forms (Fig. S2–S4, ESI^†). It should be noted that the addition of fl-FBD to substrate-free fl-CYP450 (reconstituted in nanodiscs) does not change the spin state equilibrium, unlike cyt-b₅,²⁸ however, a spin shift was observed in substrate-bound CYP450s (Fig. S2–S4, ESI^†). It is worthy to mention that the addition of the substrate (BZ) to the nanodisc reconstituted fl-CYP450 or the preformed fl-CYP450–fl-FBD complex did not alter the spin state of heme (Fig. S5, ESI^†).

Fig. 2.

Substrate induced changes in the spin-state of fl-CYP450. Differ-ential absorption spectra of the heme iron of CYP2B4 with type-1 substrates benzphetamine (A) and BHT (B). The UV-Vis spectra of CYP3A4 upon titration with testosterone are shown in the ESI^† (Fig. S4).

CYP450 has been reported to be ubiquitously present in all kingdoms of life.^32–34 Evolutionary conservational analysis of CYP450 (2B4) and FBD was conducted to probe the evolutionarily constrained residues.^35,36 Intriguingly, the structural representation of CYP450 and FBD (Fig. 3A) reveals that evolutionarily conserved residues (purple) are localized chiefly at the interaction interface of the redox complex, marked by a yellow circle in Fig. 3A. However, blue color residues are variable and localized distant from the functional site. Thus, extensive variability of residues on the distal site of CYP450 putatively reveals its capability to assimilate different types of substrates (Fig. 3A). Despite the very minimal amino acid sequence homology between the CYP450 isoforms (Fig. S6, ESI^†), the interacting interface between the redox partners was found to be highly similar, which confirms the highly conserved interacting interface among all CYP450s (Fig. 3A). Additionally, the FBD loop residues are evolutionarily more constrained (Fig. 3A and B) in comparison to other residues. Hence, we set out to structurally investigate the evolutionarily conserved loops on the fl-FBD site using ¹⁵ N-labeled fl-FBD reconstituted in nanodiscs via NMR spectroscopy. A 2D ¹H/¹⁵N TROSY-HSQC spectrum along with backbone chemical shift assignment of truncated FBD has been reported earlier.^27,37 The ¹H/¹⁵N TROSY-HSQC of fl-FBD in nanodiscs displays a well-dispersed spectrum for amide resonances (Fig. 3C), indicative of a well-folded protein.

Fig. 3.

Interacting interface between fl-CYP450 and fl-FBD revealed by NMR. (A) ConSurf analysis of CYP2B4 and FBD. Both proteins are colored based on the traditional ConSurf scale (Variable-Conserved: Blue-Purple). (B) Amino acid sequence of highly conserved loop residues of FBD with charge at pH = 7.4. The charge on the stretch of amino acids was predicted using http://protcalc.sourceforge.net/. (C) ¹H–¹⁵N TROSY-HSQC spectrum of ¹⁵N labeled fl-FBD anchored to peptide-based nanodiscs in the absence (red) and presence (black) of fl-CYP2B4 showing that the protein is well-folded. For simplification, only loop residues are labeled with assignment. (D) Top view of a cartoon representation of FBD showing the localization of FMN in loop residues.

The interacting interface of the protein–protein complex (CYP2B4–FBD and CYP3A4–FBD) is characterized by measuring the changes in the ¹H/¹⁵N TROSY-HSQC spectra of fl-FBD anchored to the nanodisc upon addition of 0.5 molar equivalents of unlabeled fl-CYP450 (Fig. 3C: CYP2B4, Fig. S7 and S8, ESI:^† CYP3A4). Addition of fl-CYP2B4 induced significant line broadening and negligible chemical shift perturbations (Fig. 3C). The signal intensity is slightly decreased for all FBD residues, whereas the loop residues at the interacting interface of the CYP450–FBD complex were extensively broadened (labeled in the spectrum, Fig. 3B and C), thereby suggesting their role in the interaction with CYP450 (Fig. 3C). This line broadening observed in the fl-FBD loop is mainly due to a change in the time scale of conformational dynamics from relatively fast to a slow exchange time regime upon interaction with fl-CYP450. Therefore, we focused on measuring the changes in the loop residues of fl-FBD, specifically present at the interacting interface of the CYP450–FBD complex (Fig. 3). The respective resonances of the residues in the four loops obtained from the TROSY-HSQC spectra of ¹⁵N-fl-FBD are plotted in Fig. 3 (fl-FBD in red and fl-CYP2B4–fl-FBD in black). Addition of unlabeled fl-CYP2B4 leads to differential line broadening of various loop residues. Residues in loops 1 and 2 are the most affected in comparison to those in loops 3 and 4 (Fig. 4). Intriguingly, loop 4 residues (D207, D208, G210 and N211) were observed to experience a slow exchange on an NMR time scale which resulted in peak splitting. This suggests that the membrane environment is driving the involvement of loop residues in the formation of a productive complex of fl-CYP2B4 and fl-FBD. A similar trend for the intermediate to slow chemical exchange was also observed in the membrane-bound fl-CYP3A4–¹⁵N-fl-FBD complex (Fig. S7 and S9, ESI^†).

Fig. 4.

(A) Representative excerpts from the overlapped ¹H–¹⁵N TROSY-HSQC spectra of loop residues of fl-FBD reconstituted in nanodiscs (red), fl-CYP2B4–fl-FBD with type-1 substrate benzphetamine (green). (B) Hotspot residues involved in CYP–FBD interactions obtained from NMR experiments.

Furthermore, it has been shown that the presence of substrates in the distal site of CYP450 can affect the conformation of the residues on its proximal site. Thus, we set out to probe the effect of substrate on the loop residues of ¹⁵N-labeled fl-FBD interacting with CYP450. The effect of substrates on the loop ¹⁵N-labeled residuesoffl-FBDin thefl-CYP450–¹⁵N-fl-FBD complex was observed and plotted in Fig. 4 and Fig. S9 and S10 (ESI^†). BZ, BHT and testosterone are examples of type-1 substrates that have been reported to modulate the conformation of CYP2B4 and CYP3A4.³⁸ To better understand the effects of these substrates, the interacting interfaces in fl-CYP2B4–15N-fl-FBD and fl-CYP3A4–¹⁵N-fl-FBD complexes were evaluated in the presence of each of the above-mentioned substrates. For the BZ loaded redox complex (BZ-fl-CYP2B4–¹⁵N-fl-FBD, green) (Fig. 4A), all the loop residues undergo extensive line broadening along with minute chemical shift perturbations. However, reappearance of the resonances (green) from residues of loops 1 and 2 demonstrates the change of chemical exchange from intermediate to slow time scales. Line broadening was observed along with peak splitting for loop 3 and 4 residues (G174, K176, T177, Y178, E179, D208, G210, N211 and L212), owing to the slow conformational exchange dynamics on the NMR time scale, indicative of their involvement in a tighter binding to CYP450 (Fig. 4A). Intriguingly, the line broadening is more pronounced in the case of BHT-saturated fl-CYP2B4 (BHT-fl-CYP2B4–¹⁵N-fl-FBD, green) (Fig. S10, ESI^†). Furthermore, the number of hotspot residues from ¹⁵N-fl-FBD in the minimal CYP450–CPR complex increases in the presence of both membrane and substrate (Fig. 4B) supporting the basis for a tighter interaction. A previous study on truncated forms of CYP17A1 and FBD revealed only loop 1 residues to be involved in the interaction.¹⁴ It is remarkable that all four loops of fl-FBD interact with fl-CYP450 when reconstituted in the membrane, which enhances the protein–protein interactions in the redox complex. The role of acidic residues in FBD loops has been reported to be critical for metabolism by CYP450.⁷ Furthermore, the effect of type-2 substrates (1-CPI and 4-CPI) was also probed on the loop residues of fl-FBD in the fl-CYP2B4–fl-15N-FBD complex (Fig. S11 and S12, ESI^†). The inhibitor (either 1-CPI or 4-CPI) loaded fl-CYP2B4–¹⁵N-fl-FBD complex causes significant line broadening and peak splitting for residues in loops 3 and 4 along with residues in loops 1 and 2 (Fig. S11 and S12, ESI^†). Intriguingly, peak attenuation was also observed for fl-CYP3A4–¹⁵N-fl-FBD when ¹⁵N-fl-FBD was titrated with testosterone loaded fl-CYP3A4 (Fig. S9, ESI^†). Earlier studies have also reported the ligand induced structural rearrangement for crystal structures of CYP2B4, and demonstrated the C-helix and C–D loop region on the proximal surface of CYP2B4 as one of the plastic regions.³⁹ Site directed mutagenesis studies have implied that residues on the C helix and the C–D loop are involved in direct interaction in the CYP2B4–CPR complex.⁹ The conformational rearrangement of CYP2B4 upon its interaction with BHT has been demonstrated.38 Both UV-Vis and NMR spectra illustrate that substrate-induced alteration in the heme spin state is associated with induction of conformational changes on the proximal surface of CYP450 towards improved binding affinity for fl-FBD of CPR.

In summary, we report the membrane induced interplay of loop residues of fl-FBD in the CYP450–FBD complex and its contribution towards higher affinity between the redox partners. NMR results show that the changes in the structural interaction between FBD and P450 facilitated the negatively charged residues to interact with CYP450. Intriguingly, such intermolecular interaction between the full-length CYP450–FBD complexes did not change the spin state of heme iron. However, introduction of substrates in functionally catalytic sites alters the spin state equilibrium for both CYP450 and CYP450–FBD. Thus, the substrate induced changes in the spin state and NMR spectra demonstrate that the conformational changes in the distal residues affect the residues on the proximal site of CYP450 for enhanced affinity towards its redox partner, CPR. Despite the low amino acid homology between 2B4 and 3A4, the interacting interfaces in the redox complexes are highly similar, which broadens the impact of our results. Taken together, our study shows that probing the residues in the hotspot region of the CYP450–FBD complex would contribute to better understanding of the mechanistic details of biotransformation by CYP450s and could be valuable in the development of novel therapeutics.