| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Pediatric Endocrinology, Diabetology & Metabolism, and Department of Clinical Research, University of Bern, CH-3010 Bern, Switzerland
Address all correspondence and requests for reprints to: Christa E. Flück, M.D., Pediatric Endocrinology and Diabetology, University of Bern, Freiburgstrasse 15, G3 812, CH-3010 Bern, Switzerland. E-mail: christa.flueck{at}insel.ch.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTS AND DISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTS AND DISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
-hydroxylase, 21-hydroxylase deficiency (OMIM 201750) and a broad spectrum of clinical characteristics ranging from severe neonatal skeletal malformations with genital ambiguity [known as Antley Bixler syndrome (ABS); OMIM 207410], to phenotypically minor polycystic ovary syndrome-like features (18, 19, 20, 21, 22, 23, 24). Subsequent to our initial report, Arlt et al. (25) identified another POR missense mutation, Y181D (reported as Y178D), and also reported three of the POR mutations (A287P, R457H, and C569Y) that we had originally described. In a larger follow-up study, we have examined the POR genes in 19 additional patients (19). Fifteen of 19 patients having abnormal genitalia and disordered steroidogenesis were homozygous or apparent compound heterozygous for POR mutations that destroyed or dramatically inhibited POR activity. Most missense mutations were found once, but A287P was found on 10 alleles, and R457H was found on seven alleles. Thus, these two missense mutations accounted for 17 of 34 (50%) of the identified POR missense mutations. The A287P mutation was found only in samples from subjects described as "white," "Caucasian," or "European." This same mutation was also found on two (of eight) alleles in our initial report (18) in individuals of European heritage. Thus, A287P is a frequent mutation causing POR deficiency in this group. The R457H mutation was found in four of eight alleles from Japanese patients. This same mutation was also found on one (of two) Japanese alleles in our initial report (18) and on two of four alleles (26) and 10 of 16 alleles (27) in two subsequent reports of Japanese patients. Thus, R457H is strongly associated with alleles of Japanese heritage (28). In the recent reports from other laboratories, several new POR variants have been described (28, 29, 30). A G5G variant of POR has recently been reported to be associated with increased breast cancer risk in African-American women (31).
POR-dependent, steroidogenic type II P450 enzymes include CYP17A1 (17-hydroxylase/17,20 lyase), CYP21B1 (21-hydroxylase), and CYP19A1 (aromatase). CYP17A1 is essential for the synthesis of glucocorticoids through its 17-hydroxylase activity and for the synthesis of all sex steroids through its 17-hydroxylase and 17,20 lyase activities (32). CYP21B1 catalyzes 21-hydroxylation of mineralocorticoids and glucocorticoids, and CYP19A1 converts androgens to estrogens, specifically androstenedione to estrone, testosterone to estradiol, and 16-hydroxytestosterone to estriol (33). CYP19A1 is encoded by CYP19 that belongs to the family 19 of the P450 proteins (14, 34). It is expressed in many different tissues including the ovaries, testes, placenta, adipose tissue, and bone osteoblasts (33). CYP19A1 is localized in the endoplasmatic reticulum of the cell and catalyzes a unique three-step reaction (Fig. 1
). The first two oxidative steps are cytochrome P450 hydroxylations, and the third step is a ferric peroxide removal of the aldehyde moiety, followed by aromatization of the steroid A-ring. Each step of the reaction requires a pair of electrons that are supplied from NADPH through POR.
|
-hydroxylase/17,20 lyase) activity assays revealed partially conflicting results indicating that different POR variants may impact supported enzymatic reactions to variable degrees (18, 19). There were three distinct features of the POR mutations that we have originally described in our first report and that we are now addressing in more detail. First, mutations in the FAD region that destroy the binding of FAD to POR resulted in total loss of activity in both CYP17A1 and cytochrome c-based assays and therefore, should be almost inactive with all substrates. Second, the mutation A287P, found mostly in European patients, is in a location in POR that is not directly involved in the function of the electron transport and has close to wild-type (WT) activity with cytochrome c assay but lower activity with CYP17A1 assays. So it is possible that A287 is involved in the interaction of POR to P450s and other electron acceptors, and depending on its role in interacting with a particular acceptor protein, it may have variable effects on the activity of particular proteins. Third, the mutations in the NADPH binding domain of POR showed a higher Michaelis constant (Km) for NADPH in cytochrome c-based assay. However, CYP17A1 was assayed with fixed NADPH concentration of 1 mM, which is more than 5000 times higher than the apparent Km of POR for NADPH in cytochrome c-based assays. Therefore it is conceivable that the variations in the NADPH concentration might play a role in activity of these mutants and that a P450 with a faster reaction or requiring higher amounts of NADPH may be affected to a greater extent by mutations in the NADPH binding domain of POR than most other POR-dependent proteins. To test these three hypotheses concerning different POR mutants, we coexpressed WT human CYP19A1, which requires three pairs of electrons for its activity, together with WT or mutant POR in yeast to obtain recombinant proteins for performing enzyme kinetic studies. We then compared kinetic data for POR-supported CYP19A1 activity with CYP17A1 activities. We used variations in assay conditions that limit the electron flux between POR and CYP19A1 to assess the impact on different mutations of POR. Finally, we built three-dimensional (3D) protein models for CYP19A1, CYP17A1, and POR and performed computational protein docking studies to understand better the structure-function relationship of various POR mutants and their impact on P450 activities.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTS AND DISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Kinetic studies of CYP19A1 reaction converting androstenedione to estrone revealed a Km of 126.3 nM and a maximum velocity (Vmax) of 1.70 pmol/min·mg protein (Table 1). Previously published enzyme kinetic studies found Km values for the conversion of androstenedione by human CYP19A1 ranging from 100 to 260 nM when assaying N-terminally modified recombinant human CYP19A1 expressed either in insect cells (39) or in E. coli (40). Thus our novel, yeast-based functional assay seemed a reasonable system for studying the influence of POR variants on CYP19A1 activity.
|
-hydroxylase and 17,20 lyase activities of CYP17A1 for the POR mutations R457H and V492E, a severe decrease of both activities for A287P, but only a mild inhibition for C569Y and V608F. Thus, our in vitro assay for CYP17A1:POR reflected the clinical findings in those patients. For POR-supported CYP19A1 activity, we found that compared with WT POR, A287P and C569Y inhibited CYP19A1 activity moderately and V608F significantly; however, when CYP19A1 activity was tested with POR R457H and V492E, enzymatic activity was completely lost (Fig. 2A). Calculations for enzyme kinetic constants revealed values that are summarized in Table 1
. The catalytic efficiency (Vmax/Km) of CYP19A1 supported by POR variant A287P was similar to WT POR, but catalytic efficiency of POR mutants C569Y and V608F was significantly lower and was almost unmeasurable for POR R457H and V492E (Table 1).
|
). For POR mutants R457H and V492E, which are located in a highly conserved FAD binding domain (Fig. 2, C and D), enzymatic activities of both the CYP17A1 and CYP19A1 were lost. In contrast, POR mutants A287P, C569Y, and V608F were found to affect the activities of CYP17A1 and CYP19A1 to different degrees (Fig. 2B). Mutation A287P exerted no inhibitory effect on CYP19A1 activity compared with WT POR but was found to decrease 17-hydroxylase to 20% and 17,20 lyase to 10% of WT activity (18). Conversely, we found a more prominent inhibitory effect of POR mutants C569Y and V608F, which are both located in the NADPH binding domain (Fig. 2C), on CYP19A1 activity than on CYP17A1 activities (25–50% vs. 50–80%; Fig. 2B and Ref. 18). Thus, different POR variants may affect diverse P450 activities to different degrees, explaining the broad spectrum of clinical phenotypes.
POR Mutants R457H and V492E Are Located in the FAD Domain and Disrupt Electron Transfer to Any Interacting P450
POR mutations R457H and V492E, which were found to cause a loss of both CYP19A1 and CYP17A1 activities, are located in the FAD domain of the POR protein (Fig. 2D). The FAD binding domain (residues 450–492) is highly conserved, and the sequence RYYSI (457–461) is invariant among human, rat, and yeast POR. The x-ray crystal structure of rat POR suggests that R457 forms a hydrogen bond with the pyrophosphate group of FAD, and Y459 contacts the FAD isoalloxazine ring and hydrogen bonds with the ribityl 3'-hydroxyl group (9, 42). Mutation R457H abolished all measurable activity of CYP19A1 similar to CYP17A1 activities (Fig. 2B). Similar results were obtained when these mutations were studied in rat POR (42, 43, 44). Residues 488–494, which lie in helix N, form hydrogen bonds with the FAD pyrophosphate. The mutation V492E, which is predicted to disrupt hydrogen bonds with FAD, has no activity. Thus, we conclude that mutations in the FAD domain of POR that directly interfere with the binding of FAD may affect all depending P450 enzyme activities severely because mutations in this location will hamper the ability of POR to transfer electrons to the FMN group of POR that ultimately provides electrons to P450s. Diminished FAD binding to R457H and V492E mutants of POR has recently been demonstrated (45).
3D Models of CYP19A1 and CYP17A1
Our CYP17A1 and CYP19A1 models were based on structures of mammalian P450 that have become available in the past couple of years. Both models showed that typical P450 fold around I helix and heme binding sites are in agreement with the structures of other mammalian P450 structures. CYP19A1 model had a fold typical of mammalian P450 structures with well-defined C, D, E, G, H, I, J, K, and L helices forming the structure. The helix F was in two distinct parts even after extensive MD refinement and could be a feature for providing access to different substrates of CYP19A1 because it was found to be quite flexible (Fig. 3). Models had energy values and structure parameters close to experimental structures (Table 2). Structural models of CYP19A1 based on bacterial P450 structures have been described previously (46, 47). In the later effort a consensus core structure based on structures of soluble bacterial P450s was employed to build the initial model, and less conserved residues and loops were added separately (47). Although the major features of the P450 structure like heme binding region and I-helix that constitutes the width of the molecule were found to be similar in the bacterial P450-based model, the substrate access channel and N-terminal helices were found to be slightly different from our model. A kink in the I-helix of CYP19A1 that is caused by sequence A-A-P was predicted in the earlier model and was also present in our model structure. More recently, structural models of CYP19A1 have been built based on mammalian P450 structures (48, 49). A model of CYP19A1 published recently that is based on the structure of rabbit P450 2C5 has similar features as our model (49). In the bacterial structure-based model, H105 was proposed to be pointing toward solvent and had similar position in our (CYP3A4-based) model and the CYP2C5-based model. However, K150 present on helix C that was proposed to interact with heme propionate was far away from heme in mammalian P450-based models, and this position is taken by R145 in our model and R130 in the CYP3A4 structure. One of the heme propionates in our CYP19A1 model is stabilized by K119 and R375, whereas R435 and R145 stabilize the other propionate; these positions are R105, R375, R130, and R440 in CYP3A4 structure. A cysteine at position 437 that provides the fifth ligand for heme iron in CYP19A1 is conserved across all P450s and was modeled similarly in the bacterial as well as mammalian P450-based models. A 3D model for CYP17A1 based on a bacterial P450 structure has been described previously (50). In our CYP17A1 model, R96, H373, R125, and R440 provide propionate ligands, whereas C442 is linked to heme iron. In the CYP17A1 model from Auchus and Miller (50), neither R96 nor R125 is pointing toward heme propionate, and only R441 was identified as the correct ligand for heme propionate. A mutation in H373 has been reported to have complete loss of CYP17A1 activities (51), but H373 does not bind heme propionate in the model of Auchus and Miller (50) but was a ligand of heme in our model. In the CYP21B1 model described by us previously, R91, R124, H365, and R426 were found to associate with heme propionates (52). These positively charged amino acids are conserved across all mammalian P450s (H369 and H368 in CYP2C5 and CYP2B4 structures), and use of a bacterial P450 template for model seems to have caused this discrepancy in the model of Auchus and Miller. Other major structural features like the presence of a T group at the catalytic center in I helix, an R group in K helix, and W in C helix were in the same positions in both our model and the model of Auchus and Miller. Cytochrome P450s have quite variable N-terminal regions as compared with C-terminal region containing the meander region/heme binding site and I, J, K, and L helices that are well conserved. Earlier modeling efforts that were based on bacterial P450 structures had to rely on extensive refinement of the variable region, which often had to be modeled separately and led to incorrect assignment for some critical residues. Another advantage of using P450 3A4 structure as template is that testosterone, a substrate of CYP19A1, is also metabolized by CYP3A4 and, therefore, may have the region around substrate access and binding that is likely to be closer to the CYP19A1 structure.
|
|
). But because A287 may play a role in the interaction with P450s, we analyzed the interaction of POR with CYP17A1 and CYP19A1 by molecular docking studies. Models of CYP19A1 and CYP17A1 were built based on 3D crystal structures of CYP3A4 and CYP2B4, respectively, and docking of POR with either CYP17A1 or CYP19A1 was performed. Model evaluation parameters are given in Table 2 and Fig. 3. Interaction of CYP17A1 with POR involves location of CYP17A1 behind the FMN and FAD domains of POR so that C-terminal arginines of CYP17A1 are in close proximity to negatively charged amino acids in the FMN domain of POR (Fig. 4A). Amino acid alanine 287 is located at the start of a loop in POR that is interacting with the C-terminus loop of CYP17A1, and possibly the interaction of these loops plays a role in correct positioning of CYP17A1 with POR (Fig. 4B). A change from alanine to proline will break the loop in POR at the beginning, and as a consequence the interacting loop in CYP17A1 will no longer be in close proximity with POR. In Fig. 4C, a close-up of the interface of the POR:CYP17A1 interaction is shown where the loop involving A287 and the C-terminal loop on CYP17A1 are visible. In contrast, interaction of POR with CYP19A1 does not seem to involve amino acid A287 of POR because CYP19A1 interacts with POR mostly from the front side of the FMN domain and the space between FMN and NADPH domain (Fig. 4D). This rules out any damaging effect of the POR A287P mutant on the activity of CYP19A1.
|
). In the 3D structure of rat POR as well as model of human POR, cysteine 569 interacts with NADPH with distances of 3.05 and 3.42 Å (for human POR). A change from cysteine to tyrosine will change the binding of NADPH and will result in reduction of efficiency for the mutated POR. Studies in which C569 of human POR was alkylated with iodoacetic acid eliminated activity, implicating a role for cysteine in binding of NADPH to human POR (3). But mutagenesis of the C569 to S in rat POR suggested that this residue was not essential, despite a 4.6-fold higher Km value for NADPH (55), and alkylation of pig POR did not change activity (56). The mutation C569Y in human POR, which we previously described in a patient with disordered steroidogenesis but without ABS (18) also had a high Km for NADPH in the cytochrome c assay, but retained 70% of 17,20 lyase activity and 80% of 17-hydroxylase activity. The C569Y mutation was found as a compound heterozygote with V608F (18) which retained 52% of 17,20 lyase activity and 53% of 17-hydroxylase activity. Valine 608 is next to tyrosine 607 and lysine 605 which have several contacts with NADPH, especially between OH group of tyrosine and first orthophosphate group of NADPH with a distance of 2.96 Å and second orthophosphate group of NADPH is at a distance of 2.49 Å from lysine 605. A change from valine to phenylalaline will disturb the NADPH binding site. However, an excess of NADPH may be able to overcome some of the binding defect because binding does not seem to be completely destroyed.
|
). In contrast, CYP17A1 activities for the conversion of both steps, first pregnenolone to 17-hydroxypregnenolone (by 17-hydroxylase), and second to dehydroepiandrosterone (by 17,20 lyase) require only one pair of electrons per reaction (15). Therefore, we thought that the differential inhibition of POR mutants C569Y and V608F on CYP17A1 and CYP19A1 might be due to a difference in cofactor requirement. So we assayed CYP19A1 activity supported by constant amounts of WT, A287P, C569Y, and V608F POR protein but different concentrations of NADPH (Fig. 5C). WT and A287P POR needed approximately 5 nM NADPH for half maximal CYP19A1 activity (calculated EC50). By contrast, C569Y and V608F required 76–108 nM NADPH. Consistent with that, mutations in the NADPH binding domain also show a higher Km for NADPH in cytochrome c assays (18, 19). Thus, POR mutations in the NADPH binding domain seem to impair CYP19A1 activity more when cofactor availability is limited. In line with these in vitro findings, the single reported patient harboring two POR mutations in the NADPH binding domain manifested with a very mild, polycystic ovary syndrome-like phenotype consistent with only CYP19A1 deficiency (18).
POR Mutants C569Y and V608F Have Diminished Capacity for NADH as Electron Donor Compared with WT POR
A change from cysteine to tyrosine seems to have higher impact on binding of NADH and results in reduced binding efficiency for the POR mutation C569Y. The mutation C569Y, which has a higher Km for NADPH in the cytochrome c assays, has almost no activity when NADH was substituted for NADPH as the electron donor (Fig. 5D). Valine 608 on the other hand is not directly involved in the binding of NADPH and has some activity with NADH as substrate, implying a less critical role for V608 than C569 in the cofactor binding. In general, both C569Y and V608F mutations that lie in the NADPH binding site of POR resulted in severely diminished activities when NADH is used as a substrate compared with NADPH. Normal and A287P mutant of POR also showed reduced activities with NADH as substrate, but the effects were much less compared with C569Y and V608F mutants of POR.
Effect of Changes in pH on Activity of CYP19A1 Supported by Normal and Mutant POR
We have used several different methods to vary the interaction between POR and CYP19A1 to assess the impact of variable electron flux on changes in the activity of CYP19A1 supported by different POR mutants. One such method is to change the pH of the reaction in the assay of CYP19A1 activity (57). The pH profile of CYP19A1 activity supported by normal POR follows a bell-shaped curve with optimum pH between 7.0 and 7.4, and mutant enzymes also show optimum pH around 7.4 (Fig. 6A). However, there are remarkable variations in the responses of different POR variants at pH values higher or lower than optimum pH. At pH 5.5, up to 35% of optimal CYP19A1 activity supported by normal POR (obtained at pH 7.4) is retained (Fig. 6B). However, CYP19A1 activity supported by C569Y mutant drops to less than 10% and mutant V608F has almost no activity at pH 5.5. At higher pH values, the effects are a little less remarkable but the same pattern is observed. At pH 8.5, normal POR retained more than 40% of its optimal activity, and activities of A287P, C569Y, and V608F mutants fell below 20% of their optimal values (Fig. 6B). Higher pH seems to destabilize A287P variant of POR to a greater extent than low pH conditions because at pH 5.5 and 6.0, the A287P mutant of POR retains greater than 50 and 80% activities, respectively.
|
). Effects on C569Y and V608F mutants were also in the same range. This suggests that changes in general conditions that limit the POR interaction with P450 do not differentiate between the variant POR studied by us. However, any mutations on the surface of POR that might be involved in interaction with P450 might show a different response to changes in membrane order that affect the POR-P450 interaction.
|
). However, C569Y and V608F variants showed much reduced activities around 200 mM KCl and lost more than 60% of their activities at 500 mM KCl (Fig. 6D). In comparison, WT POR retained more than 80% activity, and the A287P mutant retained more than 60% of its activity, suggesting a milder effect by higher KCl concentrations. Some destabilization in the A287P mutant may be responsible for greater reduction in activity compared with normal POR. Increased loss of activities by C569Y and V608F mutants suggests alterations in NADPH binding under these conditions that have greater impact on POR with mutations in the NADPH binding site. Reduction in activity of P450 at high KCl concentrations is attributed to disruption of charged pair electrostatic interactions between POR and P450, but the higher impact on mutations C569Y and V608F is probably due to diminished binding of NADPH or slower release of NADP in the cofactor binding pocket of POR. In summary, our detailed studies of POR-supported CYP17A1 and CYP19A1 activities suggest that different POR mutations can affect dependent P450s differently. Mutations in the FAD domain, which disrupt electron transfer, will affect all microsomal P450s. Mutations in the NADPH domain may affect P450s with high electron requirement more severely. Cofactor specificity of POR seems to be affected even more by mutations in the NADPH binding site. Normal POR has preference for NADPH over NADH by several orders of magnitude in the Km values, which are in millimolar range for NADH but lie in nanomolar to submicromolar range for NADPH. Similar results were observed in this study for CYP19A1 reaction. However, C569Y mutation seems to lose almost all capacity to bind and use NADH as an alternate electron donor. Limiting the cofactor availability by other methods of perturbation like inclusion of alcohol or salt enabled us to differentiate the effects in general POR-P450 interactions compared with specific changes in cofactor utilization. POR mutations C569Y and V608F can interact with P450s like normal POR as evidenced by relatively minor changes in apparent Km of P450 substrates for the P450s in both CYP17A1 and CYP19A1 reactions studied by us. Changing the POR-P450 interaction by membrane perturbation induced by octanol results in a general reduction of CYP19A1 activities. However, some changes by inclusion of octanol/KCl or change in pH could be due to altered protein conformation. Some indication of this is suggested from relatively higher loss of activity supported by A287P mutant of POR under low pH and high KCl concentrations. Further studies on folding and conformation of different POR mutants will shed more light on these differences.
Finally, some POR mutations may be located in the interaction face of some POR:P450 complexes, hampering the interaction between some partners but not affecting others. In reconstituted systems, P450 proteins form a complex with POR with an apparent Km of around 0.2 µM. The P450:POR interaction is influenced by many factors including type of P450, availability of substrates, and ionic strength of the system (58). The interaction of POR with cytochrome P450 and other electron acceptor proteins is based primarily on electrostatic charge pairing, although there is evidence for an additional hydrophobic component. The surface of the electron-donating (FMN) domain of POR is characterized by acidic residues (9, 59, 60), whereas the redox-partner binding site of microsomal P450 enzymes is typically characterized by basic residues (61, 62, 63). Chemical cross-linking and modification studies have shown that POR contains multiple carboxylate groups, presumably contributed by the acidic amino acids aspartate and glutamate (59, 64). These charge groups pair with basic amino acids (lysines, arginines) on the various electron acceptor proteins (65, 66, 67). Thus electrostatic interactions that pair charges are obviously important in governing the association of POR with a P450 enzyme. Chemical modification of acidic residues in the FMN site of POR reduces the activity of P450 (59, 64, 68). Perturbation in the membrane and electrostatic environment of POR-P450 reactions by changes in pH/KCl and inclusion of n-alcohols will be able to differentiate these mutations.
In addition, cytochrome P450 forms a dipole across the molecule, with the positive charge at the proximal face of the protein where the heme makes its closest approach to the surface. This is thought to be the surface most suitable for electron transfer from POR. Although electrostatic forces may serve to connect and orient the partners, hydrophobic forces contributed by nonpolar amino acids (leucine, tryptophan, valine, etc.) may be responsible for bringing the two proteins close enough together for electron transfer (69). Other electron acceptor proteins, such as cytochrome b5, heme oxygenase, and squalene monooxygenase, probably interact by the same mechanism. POR interacts with and supports the enzymatic activities of all 50 human microsomal P450 enzymes. The activities assessed in different P450 studies are not only a function of the specific P450 enzyme used as the read-out of the POR activity, but also a function of the form of POR used. Both the full-length POR used in our first report and the N-27 human POR used in our bacterial expression system remain associated with membranes and are able to support the reduction of a cytochrome P450 (41). By contrast, when rat or human POR is rendered wholly soluble by deleting 57 N-terminal residues, it can still reduce cytochrome c but cannot reduce a P450 (8, 11). Because the interactions of POR with different P450 enzymes will vary with the geometry of the redox-partner binding site of the P450, no assay based on a single P450 enzyme will reliably forecast all the consequences of a specific POR mutant. Only the residues that severely affect the FAD or FMN binding may lose all enzymatic activity and could be tested by employing the cytochrome c-based assays along with a single P450-based assay (41). By contrast, those POR mutations that are involved in POR:P450 interactions may have to be tested with many different P450 assays to establish the exact nature of their impact on individual P450 reactions. Because all 50 microsomal P450 enzymes depend on POR for electron supply, we are putting forward the hypothesis that defects in POR may cause disorders in hepatic drug and xenobiotic metabolism as well as affect the functions of all steroid metabolizing microsomal P450 enzymes.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTS AND DISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Expression of Human CYP19A1 and POR
Yeast strain W(B) lacking the endogenous yeast CPR1 gene (35, 71, 72) was propagated and transiently cotransfected with pYcDE2 expressing full-length WT or mutant POR cDNAs and pYeSF1-pgk-ARO expressing human WT CYP19A1. Yeast microsomes were prepared as described (37), and microsomal proteins were quantitated colorimetrically (Protein Assay Dye Reagent, Bio-Rad, Hercules, CA). To control for equal expression, Western blot analyses of microsomal proteins were performed using a rabbit polyclonal antibody against human placental CYP19A1 kindly provided by Dr. Nobuhiro Harada (73), and a rabbit polyclonal antibody against POR commercially available from Stressgen Bioreagents (LuBioScience GmbH, Lucerne, Switzerland).
CYP19A1 Activity Assay
The ability of POR to support the activity of CYP19A1 was assayed with yeast microsomal proteins containing human CYP19A1 and WT or mutant human POR as electron donor. CYP19A1 activity was measured by the release of tritiated water from 1ß-3H substrates during aromatization as described by Lephart and Simpson (36). Briefly, reactions were performed in 15-ml Falcon tubes in a final volume of 200 µl at 37 C. Yeast microsomes (40 µg microsomal protein/reaction) were incubated with 0.01–3 µM cold androstenedione, and a constant amount of [1ß-3H(N)]-androstene-3,17-dione (40,000 cpm/reaction) in 50 mM K-phosphate buffer (pH 7.4). Reactions were initiated by adding 1 mM NADPH (unless indicated differently). Reactions were stopped by adding 1 ml of chloroform. The mixture was vortexed for 30 sec, then 1 ml of water was added, and the mix was centrifuged at 1000 rpm for 5 min. Aliquots of the water phase (0.5 ml) were taken and mixed with an equal volume of 5% charcoal/0.5% dextran. After 40 sec of extraction, samples were centrifuged at 12,000 rpm for 15 min, and 0.5-ml aliquots of supernatants were collected for counting of 3H radioactivity. The measured cpm values were corrected for background radioactivity and multiplied by a factor of 4 because actual counted aliquots represented only one fourth of the total reaction volume. Once percentage conversion of androstenedione to estrone was assessed, further calculations were performed to describe the kinetic characteristics of WT CYP19A1 when supported by either WT or mutant POR. Nonlinear regression was used for this analysis. Curve fitting and calculation of Vmax and apparent Km values were performed using PRISM 3.02 (GraphPad Software Inc., San Diego, CA). Data represent the mean ± SD of two to four independent experiments, each performed in duplicate.
Buffers involving different pH values for measuring the effect of pH on CYP19A1 activities were prepared separately at twice the working concentration and substituted for normal buffer in the reactions involving measurements at alternate pH values. Experiments involving the presence of octanol were carried out by addition of octanol in the reaction mixture before addition of substrate and NADPH. Total alcohol volume did not exceed 5% of the reaction volume, and after an incubation of 5 min at room temperature the reaction was started by addition of substrate and NADPH. Similarly KCl was added from a stock solution of 5.0 M in the reactions involving measurement of CYP19A1 activity at different concentrations of KCl. In one set of experiments NADH was substituted for NADPH as the source of electrons.
3D Protein Models
3D structural models of CYP19A1, POR, and CYP17A1 were generated to study the potential impact of POR mutations on interaction with CYP19A1 and CYP17A1. A phi BLAST search of the protein structure database was performed against the CYP19A1 (NP_000094) and CYP17A1 (NP_000093) amino acid sequences, which revealed the recently solved structure of human P450 3A4 (74) (PDB no. 1tqn) (NP_059488) as the closest match for CYP19A1, and rabbit P450 2B4 structure (PDB no. 1suo) (75) was identified as the template for modeling human CYP17A1. We first performed a multiple sequence alignment of CYP19A1 with the human P450 sequences from PDB database including steroidogenic CYP17A1 and CYP21B1 by CLUSTALW (http://www.ebi.ac.uk/clustalw) (76) to analyze the structural features of the CYP19A1 protein sequence. We then performed a structural alignment of P4503A4 and CYP19A1 and used that for model building with the program MODELLER 8 (77). For the CYP17A1 model, we have used our previously described alignment for P450 2B4 structure and CYP21B1 and CYP17A1 sequences (52). Basic protocols for model building for CYP19A1 and CYP17A1 models were as described previously (52). The model was subjected to energy minimization and then checked by the programs WHATCHECK (http://swift.cmbi.ru.nl/gv/whatcheck) (78), Verify3D (79) (http://nihserver.mbi.ucla.edu/Verify_3D), and PROCHECK (80) (http://biotech.embl-ebi.ac.uk:8400/) and by Ramachandran plot analysis (81, 82) for abnormalities in the structure. Calculations for RMSD values were performed with DALI (83). Several rounds of model building and optimizations were performed to finally obtain a model that was energetically stable and passed the tests for the quality of the structure. CYP17A1 model was also generated in similar fashion based on 3D structure of rabbit P4502B4. For human POR, we used our previously described 3D protein model, which is based on the structure of rat POR (9) (PDB id 1AMO) (NP_113764) that shares 96% sequence similarity with human POR (NP_000932). The WT protein model was used to create mutants of POR with the program DEEPVIEW (84). Altered structures were subjected to energy minimization using molecular dynamics (MD) simulation and checked by WHATIF (http://swift.cmbi.kun.nl/whatif/) (85) and WHATCHECK (78) to remove bumps in the sidechains with respect to neighboring amino acids that were caused by in silico mutations. Coordinates of computer models are available as PDB files upon request. Models were depicted with Pymol (www.pymol.org) and rendered as ray-traced images with POVRAY (www.povray.org).
MD Simulation for Model Refinement
The MD simulations were performed using YASARA dynamics (86). A simulation cell (91 x 72 x 83 Å) was constructed around the CYP19A1 homology model with a 7.9 Å cutoff for Lennard-Jones forces and the direct space portion of the electrostatic forces, which were calculated using the Particle Mesh Ewald method. The pKa values of the ionizable groups in the model were predicted and assigned the protonation states based on pH 7.0. The cell was filled with water, and the AMBER99 (87) electrostatic potential was evaluated at all water molecules; the one with the lowest or highest potential was turned into a sodium or chloride counter ion until the cell was neutral. A short steepest descent minimization of all atoms was done to remove severe bumps in the protein and to achieve convergence until the maximum atom speed dropped below 2200 m/sec. Then a start-up simulation was run for 5 psec, using a multiple time step of 1 fsec for intramolecular and 2 fsec for intermolecular forces, with all heavy protein atoms fixed, so that the solvent molecules could smoothly cover the protein surface. Simulated annealing minimizations were started at 298 K, and velocities were scaled down every 10 steps for a total time of 5 psec in 500 steps. We then ran MD simulations with AMBER99 force field at 298 K and 0.9% NaCl in the simulation cell for 1000 psec to refine the models. All MD computations were run locally on a Dell computer server containing 4 intel CPUs running under SUSE Linux 10.1/Windows server 2003. A local installation of WHAT IF program was used to check the quality of models in combination with Verify3D analysis to check for abnormalities. An analysis of MD trajectories showed a significant improvement, and final model snapshot has a free energy of –19255 kJ/mol for CYP19A1 model. The C-RMSD dropped from an initial value of 2.7 Å, which is typical for a decent homology model to 1.4 Å, which corresponds to a medium-resolution x-ray structure. Using the MD simulation, we could refine the homology model close to an experimental-like structure. Modeling of P450 proteins has recently been quite successful, and some of the models of P450 show very high similarity when compared with the actual structures which have recently become available (88). An increase in the number of mammalian P450 structures that have become available recently combined with powerful computational hardware and software for modeling side-chains and loops in the variable regions of homology models could be attributed to this achievement.
Analysis of the Interaction of POR with CYP19A1 and CYP17A1 by 3D Docking
We used the programs Patchdock (89) and Hex 4.5 (90) for analysis of the interaction of WT and A287P mutant of POR with CYP19A1 and CYP17A1. We performed the 3D protein-protein interaction analysis of the interaction of POR with either CYP19A1 or CYP17A1 to understand the role of alanine 287 in P450:POR interactions and whether a change to proline would affect such interactions. Electrostatic interactions are known to be involved between positively charged groups on P450s and negatively charged acidic residues on POR and were taken into consideration while calculating the dockings. In the first step, a molecular shape representation was performed by computing the molecular surface of the molecule. Next, a segmentation algorithm was applied for detection of geometric patches, which were filtered, and then a geometric hashing and pose-clustering matching was performed to match the patches on two structures. Concave patches were matched with convex and flat patches with any type of patches. The complexes obtained were then examined, and all complexes with unacceptable penetrations of the atoms of the receptor to the atoms of the ligand were discarded. Finally, the remaining candidates were ranked according to a geometric shape complementarity score. In total, 50,000 calculations were performed per pair, and solutions were filtered for energetically stable conformations. The top 20 solutions from each round of calculations were manually checked for interaction sites and orientation of the molecules. Structures were edited with the program Pymol and depicted and ray-traced images were generated by POVRAY.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
This work was supported by Swiss National Science Foundation Grants 3232B0103178/179 (to C.E.F.) and 3100A0-113719 (to A.V.P.).
The nucleotide sequence of the human NADPH P450 oxidoreductase gene is available in the NCBI nucleotide database under NCBI accession no. NM_000941. The amino acid sequence of human POR can be accessed through NCBI Protein Database under NCBI accession no. NP_000932. The atomic coordinates for the crystal structure of rat POR are available in the Research Collaboratory for Structural Bioinformatics Protein Databank PDB no. 1AMO. The atomic coordinates for the crystal structure of FMN binding domain of human POR are available in the Research Collaboratory for Structural Bioinformatics Protein Databank PDB no. 1B1C. Online Mendelian Inheritance in Man, OMIM no. 124015, 201750. Enzyme collection number, EC 1.6.2.4.
Disclosure Statement: A.V.P., P.K., G.H., P.E.M., and C.E.F. have nothing to declare.
First Published Online June 26, 2007
1 A.V.P. and P.K. contributed equally to this work. 
Abbreviations: ABS, Antley Bixler syndrome; 3D, three-dimensional; Km, Michaelis constant; MD, molecular dynamics; NADPH, nicotinamide adenine dinucleotide phosphate; POR, NADPH P450 oxidoreductase; Vmax, maximum velocity; WT, wild type.
Received for publication May 10, 2007. Accepted for publication June 21, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTS AND DISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
-hydroxylation system of liver microsomes into three components. J Biol Chem 244:3714–3721-hydroxylase/17,20-lyase): insights into reaction mechanisms and effects of mutations. Mol Endocrinol 13:1169–1182-hydroxylase deficiency. J Biol Chem 268:25811–25817This article has been cited by other articles:
 |
|
 |
|
  M. Fukami, G. Nishimura, K. Homma, T. Nagai, K. Hanaki, A. Uematsu, T. Ishii, C. Numakura, H. Sawada, M. Nakacho, et al. Cytochrome P450 Oxidoreductase Deficiency: Identification and Characterization of Biallelic Mutations and Genotype-Phenotype Correlations in 35 Japanese Patients J. Clin. Endocrinol. Metab., May 1, 2009; 94(5): 1723 - 1731. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. G. Gomes, N. Huang, V. Agrawal, B. B. Mendonca, T. A. S. S. Bachega, and W. L. Miller Extraadrenal 21-Hydroxylation by CYP2C19 and CYP3A4: Effect on 21-Hydroxylase Deficiency J. Clin. Endocrinol. Metab., January 1, 2009; 94(1): 89 - 95. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. G. Gomes, N. Huang, V. Agrawal, B. B. Mendonca, T. A. S. S. Bachega, and W. L. Miller The Common P450 Oxidoreductase Variant A503V Is Not a Modifier Gene for 21-Hydroxylase Deficiency J. Clin. Endocrinol. Metab., July 1, 2008; 93(7): 2913 - 2916. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Huang, V. Agrawal, K. M. Giacomini, and W. L. Miller Genetics of P450 oxidoreductase: Sequence variation in 842 individuals of four ethnicities and activities of 15 missense mutations PNAS, February 5, 2008; 105(5): 1733 - 1738. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
, Red line, WT; 
, blue line, A287P; light green line, R457H; 
, yellow line, V492E; 
, green line, C569Y; 
, black line, V608F. B, Catalytic efficiency of CYP19A1 activity supported by WT or mutant POR compared with both 17
