Exhaustive computational search of ionic-charge clusters that mediate interactions between mammalian cytochrome P450 (CYP) and P450-oxidoreductase (POR) proteins

Abstract

In this work, a model for the interaction between CYP2B4 and the FMN domain of rat P450-oxidoreductase is built using as template the structure of a bacterial redox complex. Amino acid residues identified in the literature as cytochrome P450 (CYP)–redox partner interfacial residues map to the interface in our model. Our model supports the view that the bacterial template represents a specific electron transfer complex and moreover provides a structural framework for explaining previous experimental data.

We have used our model in an exhaustive search for complementary pairs of mammalian CYP and P450-oxidoreductase (POR) charge clusters. We quantitatively show that among the previously defined basic clusters, the 433K–434R cluster is the most dominant (32.3% of interactions) and among the acidic clusters, the 207D–208D–209D cluster is the most dominant (29%). Our analysis also reveals the previously not described basic cluster 343R–345K (16.1% of interactions) and 373K (3.2%) and the acidic clusters 113D–115E–116E (25.8%), 92E–93E (12.9%), 101D (3.2%) and 179E (3.2%).

Cluster pairings among the previously defined charge clusters include the pairing of cluster 421K–422R to cluster 207D–208D–209D. Moreover, 433K–434R and 207D–208D–209D, respectively the dominant positively and negatively charged clusters, are uncorrelated. Instead our analysis suggests that the newly identified cluster 113D–115E–116E is the main partner of the 433K–434R cluster while the newly described cluster 343R–345K is correlated to the cluster 207D–208D–209D.

Introduction

The cytochromes P450 (CYP, EC 1.14.14.1) are a superfamily of heme-thiolate enzymes that catalyze the monooxygenation of hydrophobic endogenous and xenobiotic substrates (Bridges et al., 1998). CYP homologs have been sequenced from all lineages of life—including eukaryotes and bacteria (Sevrioukova et al., 1999b, Zawaira et al., 2008). The cytochrome P450 enzymatic cycle includes substrate binding, first electron transfer, oxygen binding, second electron transfer, substrate oxidation and finally, product dissociation. Hence the CYP reaction cycle involves two distinct electron transfer steps (Bridges et al., 1998, Kuznetsov et al., 2006). Biologically relevant CYP-catalyzed monooxygenation reactions include steroid metabolism and biogenesis, xenobiotic metabolism, fatty acid metabolism and antibiotic metabolism (Sevrioukova et al., 1999b). Bacterial CYPs are soluble proteins that have high substrate specificity while their mammalian cousins are endoplasmic reticulum-bound CYPs mostly involved in xenobiotic metabolism and have broad substrate specificity (Nelson and Strobel, 1987). Despite these differences, surveys of known CYP structures from across the different lineages of life show that the CYP structure/fold is highly conserved (Hasemann et al., 1995, Sansen et al., 2007, Schoch et al., 2004, Wester et al., 2004, Yano et al., 2004). A notable feature of this highly conserved CYP fold architecture is that a proximal surface, where the heme cofactor comes closest to the protein surface, can readily be discerned from the distal surface where the heme group is farthest from the protein surface (Hasemann et al., 1995, Poulos et al., 1987, Sansen et al., 2007, Schoch et al., 2004, Wester et al., 2004, Yano et al., 2004).

The cytochrome P450 enzyme sources electrons from redox partner systems. Bacterial and mitochondrial CYPs obtain catalytic cycle electrons from small soluble iron–sulfur electron transport proteins such as adrenodoxin (also known as ferredoxin-1, FDX1) and putidaredoxin (Sevrioukova et al., 1999b). On the other hand, mammalian microsomal CYPs source their electrons from NADPH dependent P450-oxidoreductase (POR). POR contains the flavin cofactors FAD and FMN (Sevrioukova et al., 1999b). POR is organised into four domains which are, from the protein N-terminus to the C-terminus, the FMN binding domain, the connecting domain, the FAD binding domain and the NADPH-binding domain (Wang et al., 1997). The FAD cofactor accepts electrons from NADPH and transfers them to the FMN cofactor, which in turn, transfers electrons to the heme iron of substrate-bound CYP (Sevrioukova et al., 1999b, Wei et al., 2007).

An important question in CYP research is the elucidation of the protein–protein interactions involved in the process of electron transfer from redox partners to substrate-bound CYP. Following the determination of the first structure of a CYP (CYP450Cam from Pseudomonas putida) in 1985 (Poulos et al., 1985), the most parsimonious answer to that question would have been: CYP is likely to interact with redox partners at the proximal side where the heme centre is most easily accessible. Early models for the CYP–POR complex supported the parsimonious view (Stayton et al., 1989).

The current mainstream views on CYP–POR interaction surfaces ultimately trace their origins to the studies on the reduction of cytochrome c by cytochrome b5 (Stonehuerner et al., 1979). Stonehuerner et al. (1979) showed that the interaction between cytochrome c and cytochrome b5 is mediated by complementary electrostatic interactions between specific lysine residues on cytochrome c and specific acidic residues (aspartate or glutamate) on cytochrome b5. Similar experiments have been performed demonstrating the role of electrostatic interactions in mammalian CYP–POR (Bernhardt et al., 1988, Davydov et al., 2000, Kelley et al., 2005).

Extensions from the cytochrome c–cytochrome b5 model system to CYP target systems by Poulos and co-workers (1989) and by Davydov et al. (1992) provided the first information about CYP–redox partner interfacial residues. Poulos and co-workers’ extensions to the bacterial putidaredoxin–CYP450Cam system suggested that the P450Cam residues Arg112, Lys344 and Arg364 formed salt bridges with putidaredoxin residues Glu48, Glu44 and Asp60 respectively (Stayton et al., 1989). The P450Cam residues Arg112, Lys344 and Arg364 are homologous to CYP2B4 residues Arg125, Lys421 and Arg443 (Bridges et al., 1998). Davydov et al. (1992) extensions to the CYP2B4–P450-oxidoreductase system suggested that CYP2B4 positions 121–145 are involved in POR recognition.

Several other workers have contributed to the identification of CYP–redox partner interfacial residues. These include Juvonen et al. (1992) who showed the involvement of Arg129 in the binding of CYP2A5 to cytochrome b5. Bridges et al. (1998) delimited the following CYP2B4 residues as CYP2B4–POR interfacial residues: R122, R126, R133, F135, K433, R422 and R443. Shen and Kasper (1995) searched for surface exposed acidic clusters within the FMN-binding domain of POR and identified two interfacial residue clusters comprising 207D, 208D and 209D and 213E, 214E and 215D. Zhao et al. (1999) identified the cluster comprising 142D, 144D and 147D.

Of the bacterial redox systems that have been used as model systems to address the problem of redox partner recognition in mammalian systems, the cytochrome P450-BM3 bears the closest resemblance to mammalian systems. Flavocytochrome P450-BM3 is a 119 kDa self-sufficient fatty-acid monooxygenase from Bacillus megaterium. The protein consists of a heme domain (BMP) and a FMN/FAD-containing P450-oxidoreductase domain linked together in a single polypeptide chain (Sevrioukova et al., 1999b). Sevrioukova et al., 1999a, Sevrioukova et al., 1999b have determined the structure of a P450-BM3 construct comprising the heme/FMN-containing domain and it has been debated whether this structure is a plausible prototypical representation of a specific electron transfer complex.

Here we investigate the relationship between CYP2B4–POR interfacial residues one can delimit using the P450-BM3 structure as template and those which have been delimited to date by the many various experimental approaches listed above. We have also used our model of the CYP2B4–POR complex to implement an exhaustive search for oppositely charged clusters mediating the ionic interactions between mammalian CYP and POR. Comparison of our pairings with Karyakin's experimentally delimited pairings (Karyakin et al., 2007) has been used to validate our assignments.