PEDS Advance Access originally published online on August 30, 2006
Protein Engineering Design and Selection 2006 19(11):491-496; doi:10.1093/protein/gzl035
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Cytochrome P450 active site plasticity: attenuation of imidazole binding in cytochrome P450cam by an L244A mutation
Department of Pharmaceutical Chemistry, University of California 600 16th Street, San Francisco, CA 94158-2517, USA
1To whom correspondence should be addressed. E-mail: ortiz{at}cgl.ucsf.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResults and discussionConclusionsAcknowledgementsREFERENCES |
|---|
We have identified a P450cam mutation, L244A, that mitigates the affinity for imidazole and substituted imidazoles while maintaining a high affinity for the natural substrate camphor. The P450cam L244A crystal structure solved in the absence of any ligand reveals that the I-helix is displaced inwards by over 1 Å in response to the cavity created by the change from leucine to alanine. Furthermore, the crystal structures of imidazole-bound P450cam and the 1-methylimidazole-bound P450cam L244A mutant reveal that the ligands have distinct binding modes in the two proteins. Whereas in wild-type P450cam the imidazole coordinates to the iron in an orientation roughly perpendicular to the plane of the heme, in the L244A mutant the rearranged I helix, and specifically residue Val247, forces the imidazole into an orientation almost parallel to the heme that impairs its ability to coordinate to the heme iron. As a result, the imidazole is much more weakly bound to the mutant than it is to the wild-type enzyme. Despite the constriction of the active site by the mutation, previous work with the L244A mutant has shown that it oxidizes larger substrates than the wild-type enzyme. This paradoxical situation, in which a mutation that nominally increases the active site cavity appears to decrease it, suggests that the mutation actually increases the active site maleability, allowing it to better expand to oxidize larger substrates.
Keywords: active site conformation/azole drug resistance/cytochrome P450 inhibition/imidazole binding to P450/protein maleability
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionConclusionsAcknowledgementsREFERENCES |
|---|
Azole-based compounds such as fluconazole, ketoconazole and itraconazole are potent inhibitors of cytochrome P450 enzymes (Little and Ryan, 1982
; Loose et al., 1983; Murray and Ryan, 1983; Benedetti and Bani, 1999) and are employed in the treatment of topical and systemic fungal infections (Keating, 2005; Krcmery, 2005; Lacroix et al., 2005). The potential utility of azoles in other clinical contexts, for example in therapeutic approaches to tuberculosis (Guardiola-Diaz et al., 2001; McLean et al., 2002), has also received attention. However, azole-resistant fungal strains are now well known (Asai et al., 1999; Kelly et al., 1999; Cuenca-Estrella et al., 2005; Meneau and Sanglard, 2005). An understanding of the parameters that control the binding of azoles to hemoproteins is thus relevant to the development of azole agents that circumvent resistance, possess greater specificity for particular hemoproteins or address new targets. Because azole compounds are still the primary antifungal therapeutics in current use, an understanding of azole binding and resistance is particularly critical in this area.
Until recently, structural insights into the origins of mitigated azole affinity were derived primarily from homology models and experiments with model P450 enzymes, because no crystal structure existed of a therapeutically relevant target. The crystallization of CYP51 from Mycobacterium tuberculosis in 2001 provided the first structurally characterized potential drug target in the P450 family, as well as a crystallographic model for the CYP51 enzymes that are the actual targets in fungi (Podust et al., 2001). Azole-based compounds have been shown to bind and inhibit CYP51 (Podust et al., 2001; Matsuura et al., 2005). The insights derived from homology models of resistant strains of CYP51 revealed that mutations near the heme cofactor are likely to be responsible for the altered affinity of azole-based agents (Sheng et al., 2004).
In previous work, we constructed the L244A mutant of P450cam based on the hypothesis that the active site would be modestly expanded relative to that of the wild-type enzyme, because the alanine side chain is three carbons smaller than that of the normal leucine (De Voss et al., 1997). In accord with this hypothesis, and with the predictions of an associated in silico docking study, the L244A mutant was found to oxidize larger molecules than the wild-type enzyme. More recently, we reported a computational docking effort to predict the differential binding of substituted imidazoles to the active sites of cytochrome P450cam and its L244A mutant (Verras et al., 2004). The goal of the study was to determine whether computational approaches could identify ligands that selectively bound to the mutant but not to the wild-type even though the two proteins differed by only three carbons and six hydrogens owing to replacement of an isopropyl moiety by a hydrogen. As in the earlier studies, the model of the L244A mutant was constructed for this effort by simply deleting the three atoms corresponding to the mutation from the crystal structure of the wild-type enzyme. While P450cam is not itself a target for antifungal azole compounds, it is the most highly characterized P450 enzyme in both biochemical and structural terms (Poulos and Johnson, 2005). Crystal structures are available of P450cam complexed with phenylimidazole, metyrapone and an azole drug (Poulos and Howard, 1987; Raag et al., 1993), but no structures are available of this enzyme complexed with smaller imidazoles. The computational studies successfully provided qualitative predictions concerning the binding of mono- and di-substituted imidazoles to P450cam and its L244A mutant, but quantitative predictions were stymied, in part, by an unexplained 40-fold decrease in the affinity of the L244A mutant for imidazole itself, despite an essentially unchanged affinity for the natural substrate camphor.
We report here the structures of both the substrate-free and 1-methylimidazole-bound P450cam L244A mutant. We have also co-crystallized wild-type P450cam with imidazole in order to compare the structures and explore the basis for the decreased imidazole affinity of the L244A mutant. We have found that the single L244A mutation, despite inducing a very different crystal morphology and crystal space group, affords a structure highly similar to that of the wild-type protein. However, a shift in the distal helix caused by the L244A mutation contracts the active site and causes an imidazole to bind in a drastically different orientation relative to the heme iron atom compared with the orientation of imidazole in the wild-type enzyme. The difference readily rationalizes the difference in affinity and offers a model for a mechanism that could result in resistance to azole agents. In view of the earlier substrate oxidation studies, the results furthermore suggest that a primary effect of the mutation is to enhance the dynamic plasticity of the active site.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionConclusionsAcknowledgementsREFERENCES |
|---|
Protein expression
The L244A and C334A double mutant of P450cam was constructed as described elsewhere (Verras et al., 2004). The C334A mutation exists solely to prevent adventitious dimerization of the protein and thus to facilitate crystallization. It has no other discernible effect on the structure or activity of the protein and is referred to here as the wild-type protein (Nickerson and Wong, 1997; Nickerson et al., 1998).
Cytochrome P450cam and the P450cam L244A mutant were expressed in Escherichia coli as previously reported (Verras et al., 2004). The proteins were purified as before with the exception that camphor was not included in any wash, elution or loading buffer. No ammonium salt cut was performed in purifying the proteins, but an additional gel filtration column was run after the Q-Sepharose column. The proteins thus obtained were sufficiently pure for crystallization. The proteins were concentrated to 50 mg/ml in Amicon-2000 centrifugal concentrators prior to storage but, when possible, the proteins were used for crystallization without freezing.
Crystal screening, data collection and refinement
Crystal trays were set up at the temperatures indicated with mother liquor that consisted of 50 mM Kpi containing 250 mM KCl, 50 mM dithiothreitol and 36–52% ammonium sulfate (Poulos et al., 1982). The protein was combined in a ratio of 1 µl to 1 µl of mother liquor in the hanging drop. The cryoprotectant was 50 mM KCl, 25% ammonium sulfate and 30% glycerol. Imidazole and 1-methylimidazole were added to the mother liquor from a 1 M stock solution in water. The final concentration of the ligands in the mother liquor was 25 mM. At room temperature, crystals generally formed overnight and crystal growth was complete within 2 days.
The P450cam L244A crystals had significantly different morphologies from those of the wild-type enzyme. They formed overlapping plates and the edges of these plates were broken off and used in diffraction. The average size of the plates was 0.30 x 0.30 x 0.02 mm. The average size of the wild-type crystals was 0.3 x 0.3 x 0.2 mm. The temperature (16–25°C) of crystal tray incubations and the concentration (12–50 mg/ml) of the protein were varied for the L244A crystals, but overall the crystals still grew as overlapping plates.
Despite the presence of only a single point mutant at the center of the protein, the crystals of the wild-type and L244A mutant crystallized in different space groups. The wild-type refined as P212121 while the mutant refined as P21. The substrate-free L244A crystals were found to be twinned and were initially refined in the C212121 space group. The protein crystals were found to be 44% twinned by the Yeates test (Yeates, 1997) with a twinning operator of –h, –k, h + l. Calculation of the Matthews coefficient (Kantardjieff and Rupp, 2003) revealed two monomers in the asymmetric unit when solved in the P21 space group. When the protein was refined in the P21 space group using the inputs in CNS for treating partially hemihedrally twinned crystals and two monomers (related by a pure translation), the Rfree immediately dropped to 29%. Twinning fraction and operations were included in the refinement using CNS. Twinned data have been used for structure determination in the past, and, while it complicates the structure determination, it does not make it impossible (Redinbo and Yeates, 1993; Breyer et al., 1999; Contreras-Martel et al., 2001).
Diffraction data were collected at the Berkeley ALS (Advanced Light Source) on beamlines 8.3.1 and 8.3.2. The data reduction was done with DENZO (Otwinowski and Minor, 1997) and MOSFLM (Leslie, 1992). Protein unit cell rotational and translational searches were done with CCP4 [Collaborative Computational Project, Number 4 (1994)] and refinement was carried out with CNS (Brunger et al., 1998). Manual rebuilding of the proteins and visualization was accomplished with O (Jones et al., 1991). The figures were prepared with PYMOL (DeLano, 2002).
Protein Data Bank accession codes
The atomic coordinates for the new structures reported here have been deposited in the Protein Data Bank, Rutgers University, New Brunswick, NJ under accession numbers 2H7Q (wild-type + imidazole), 2H7R (L244A + 1-methylimidazole) and 2H7S (L244A apo).
|
Results and discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionConclusionsAcknowledgementsREFERENCES |
|---|
Binding of imidazole to P450cam and L244A P450cam
The binding affinities of P450cam and its L244A mutant for the natural substrate camphor are very similar, Ks = 1.3 ± 0.2 µM and Ks = 1.6 ± 0.3 µM, respectively. Furthermore, the spectroscopic shifts induced by titration of the protein with the substrate are identical for both proteins. Thus, beginning in both instances with a water coordinated heme, one observes a decrease at 416 nm and an increase in the high spin species at 390 nm.
The affinity of the protein for imidazole as judged by spectroscopic methods, however, is drastically different for the wild-type and L244A mutant, with Ks = 7.53 ± 1.4 µM and Ks = 300 ± 45 µM, respectively. Nevertheless, the spectroscopic shifts observed upon titration of the proteins with imidazole are indistinguishable. Beginning with the water-coordinated heme, a decrease is observed at 416 nm and an increase at 424 nm. The spectroscopic shifts indicate that in both cases the binding of the ligand involves coordination of the azole nitrogen to the heme iron, giving rise to a low spin species.
The spectroscopically determined constants for binding of 1-methylimidazole to the two enzymes diverge similarly, the affinities being Ks = 2.4 ± 0.4 µM and 55.9 ± 9.0 µM for the wild-type and L244A proteins, respectively. Again, the spectral shifts are the same for both proteins, resulting in a low spin 424 nm species. The L244A protein was crystallized with 1-methyl imidazole, which is larger and more asymmetric than imidazole, to facilitate the ligand placement within the density.
To understand how the mutation affects the binding affinity of large ligands, we measured the binding affinity of the wild-type and L244A proteins for ketoconazole. The affinities of the wild-type and L244A mutant for ketoconazole are Ks = 0.5 ± 0.1 µM and 0.6 ± 0.1 µM, respectively, which indicates that the large discrepancy in imidazole affinity is ablated in the presence of large iron-coordinating ligands. Prior work with 1,5-disubstituted imidazoles reinforces this finding (Verras et al., 2004).
Crystal structure of substrate-free L244A P450cam
The substrate-free L244A unit cell contained two molecules in the asymmetric unit. The density of monomer A was slightly clearer than that of monomer B, so when referring to structural characteristics, we refer to monomer A. The C
rmsd between the monomers was less than 1 Å.
While the crystal morphologies and unit cell dimensions are different between the wild-type (P212121) and the mutant protein (P21) (Table I), the proteins are highly similar to wild-type P450cam [PDB code 1PHC (Poulos et al., 1986, 1987)] with a global rmsd of 1.6 Å for all atoms. The primary differences between P450cam and the P450cam L244A mutant are within the active site. A difference map was constructed by using the structure factor file from the L244A and the wild-type (1PHC) as a model. Viewing the negative difference map (Fo–Fc) at –2.5 sigma revealed a change in cavity size smaller than expected for the deletion of an isopropyl moiety. When overlaying the structures of P450cam L244A, substrate-free wild-type P450cam and substrate-bound wild-type P450cam [PDB code 2CPP
[PDB]
(Poulos et al., 1987)] we see that the cause of the small negative density at position 244 is a shift of the I helix toward the active site by over 1 Å (Figure 1).
|
|
Crystal structure of imidazole-bound wild-type P450cam
To understand the structural differences between the L244A and wild-type proteins we co-crystallized the wild-type protein with imidazole. Imidazole is a small ligand, with few contacts in the active site. To date, only one crystal structure, that of CYP119 [PDB code 1F4U
[PDB]
(Yano et al., 2000)], has been resolved with an unsubstituted imidazole bound in the active site. The wild-type protein structure was determined to a resolution of 1.5 Å in the presence of imidazole (Table I). The overall structure is very similar to that of the substrate-free wild-type enzyme and, even in the active site, no backbone or side chain rearrangements are evident. The imidazole, as expected, is bound perpendicular to the heme (Figure 2a). The temperature factor (7.7) of the imidazole N3 is very low at occupancy of 1.0, whereas the other atom occupancies lie between 0.5 and 0.85 with temperature factors ranging between 18 and 22. This is in agreement with our knowledge of imidazole binding, suggesting that the position of the nitrogen in imidazole that coordinates to the heme is more tightly restricted than those of the other atoms of the ligand.
|
Crystal structure of the 1-methyl imidazole-bound L244A mutant
The L244A protein was crystallized in the presence of 1-methylimidazole, a larger and more asymmetric ligand than imidazole, to improve the fitting of the ligand to the active site density and the structure was determined to a resolution of 2.1 Å (Table I). While the overall structure is again very similar to that of wild-type P450cam (C rmsd of <1 Å), the active site density is significantly distinct between the substrate-free L244A and substrate-free wild-type protein. The density indicates that the ligand lies almost parallel to the plane of the heme (Figure 2b). As in the substrate-free L244A structure, the I helix is shifted toward the active site. Most notably Val247 shifts 1.8 Å toward the heme. The density of this residue is well defined. Because of the position of Val247, the 1-methylimidazole is unable to assume the orientation perpendicular to the heme that is observed in the wild-type protein. Because coordination of imidazole to the heme iron involves sigma bonding between the N3 imidazole nitrogen and heme iron, as well as 
backbonding from the electron deficient imidazole to the dxy and dxz iron orbitals, a perpendicular binding mode is necessary to optimize affinity. The steric restraints imposed by the shifted backbone, specifically Val247, prohibit optimization of the imidazole coordination and, thus, result in the observed mitigated affinity of 1-methylimidazole and imidazole for the L244A protein. We also crystallized and attempted to determine the imidazole-bound L244A structure (data not shown) and again found a large patch of ligand density almost parallel to the heme. However, the density was not sufficiently defined to permit conclusive placement of the ligand, so we report here the structure of the L244A protein crystallized with 1-methylimidazole.
In earlier work, we prepared the L244A mutant as part of a study of DOCKing methods for the prediction of substrate specificity in cytochrome P450 enzymes (De Voss et al., 1997). This mutant was prepared based on a computer graphics analysis of the crystallographic active site, which suggested that the mutation would delete three carbon atoms from the active site and, thus, expand the substrate-binding cavity by the volume of these three carbons and their associated hydrogen atoms. DOCKing was then carried out with the wild-type crystal structure and with the L244A model, without further refinement of the model, and substrates that docked within the mutant but not the wild-type active site were tested experimentally as ligands and substrates. In fact, the L244A mutant was found to bind and oxidize larger substrates than the wild-type enzyme, as would be expected for a larger active site cavity (De Voss et al., 1997). However, the present crystal structure of the L244A mutant indicates that the I-helix shifts inward, compensating for the volume vacated by the deleted atoms. The binding and oxidation of larger substrates by the L244A mutant, thus, implies that the mutant cavity is more maleable, and can expand to allow the binding of larger substrates, than that of the wild-type enzyme.
|
Conclusions |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionConclusionsAcknowledgementsREFERENCES |
|---|
We present here the first crystallographic analysis of the basis for mitigated imidazole affinity stemming from a point mutation in a P450 enzyme. P450cam is not an antifungal target, but because it is so well characterized it is an excellent model in which to study the phenomenon of drug resistance. As shown here, a single point mutant in the active site results in a 40-fold difference in binding affinity for imidazole and a 25-fold difference in affinity for 1-methyl imidazole.
The structural studies implicate a movement of the I helix as the source of the mitigated imidazole affinity. In the absence of a ligand the helix moves in to fill the cavity created by the leucine to alanine mutation, and in the presence of 1-methylimidazole the helix also shifts into the active site and prevents imidazole from assuming an orientation perpendicular to the heme. Our finding that the ketoconazole binding affinity is not decreased by the lessened affinity of the coordinating imidazole moiety is consistent with the inference that mutations have larger effects on small ligands that have few interactions with other parts of the protein. The observation that the helix shifts inward to diminish the active site cavity, yet the enzyme binds and oxidizes larger substrates than the wild-type enzyme, implies that the mutation increases the deformability of the active site, enhancing its ability to conformationally adjust to accommodate large substrates.
|
Footnotes |
|---|
Edited by Tom Alber
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionConclusionsAcknowledgementsREFERENCES |
|---|
This work was supported by grants GM25515 and GM56531 from the National Institutes of Health.
|
REFERENCES |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionConclusionsAcknowledgementsREFERENCES |
|---|
Asai K., Tsuchimori N., Okonogi K., Perfect J.R., Gotoh O., Yoshida Y. (1999) Antimicrob. Agents Chemother. 43:1163–1169.
Benedetti M.S. and Bani M. (1999) Drug Metab. Rev. 31:665–717.[CrossRef][Web of Science][Medline]
Breyer W.A., Kingston R.L., Anderson B.F., Baker E.N. (1999) Acta Crystallogr. D Biol. Crystallogr. 55:129–138.[CrossRef][Medline]
Brunger A.T., et al. (1998) Acta Crystallogr. D Biol. Crystallogr. 54:904–921.
Collaborative Computational Project, Number 4. (1994) Acta Crystallogr. D Biol. Crystallogr. 50:760–763.[CrossRef][Medline]
Contreras-Martel C., Martinez-Oyanedel J., Bunster M., Legrand P., Piras C., Vernede X., Fontecilla-Camps J.C. (2001) Acta Crystallogr. D Biol. Crystallogr. 57:52–60.[CrossRef][Medline]
Cuenca-Estrella M., Gomez-Lopez A., Garcia-Effron G., Alcazar-Fuoli L., Mellado E., Buitrago M.J., Rodriguez-Tudela J.L. (2005) Antimicrob. Agents Chemother. 49:1232–1235.
De Voss J.J., Sibbesen O., Zhang Z., Ortiz de Montellano P.R. (1997) J. Am. Chem. Soc. 119:5489–5498.[CrossRef]
DeLano W.L. (2002) PYMOL: The Molecular Graphics Systems(DeLano Scientific, South San Francisco).
Guardiola-Diaz H.M., Foster L., Mushrush D., Vaz A.D. (2001) Biochem. Pharmacol. 61:1463–1470.[CrossRef][Web of Science][Medline]
Jones T.A., Zou J.Y., Cowan S.W., Kjeldgaard M. (1991) Acta Crystallogr. A 47:110–119.
Kantardjieff K.A. and Rupp B. (2003) Protein Sci. 12:1865–1871.[CrossRef][Web of Science][Medline]
Keating G.M. (2005) Drugs 65:1553–1567.[CrossRef][Web of Science][Medline]
Kelly S.L., Lamb D.C., Kelly D.E. (1999) FEMS Microbiol. Lett. 180:171–175.[Web of Science][Medline]
Krcmery V.C. Jr. (2005) Med. Princ. Pract. 14:125–135.[CrossRef][Web of Science][Medline]
Lacroix C., de Kerviler E., Morel P., Derouin F., Feuilhade de Chavin M. (2005) Br. J. Dermatol. 152:1067–1068.[CrossRef][Web of Science][Medline]
Leslie A.G.W. (1992) Joint CCP4 and ESF-EAMCB Newsletter on Protein Crystallography 26:.
Little P.J. and Ryan A.J. (1982) J. Med. Chem. 25:622–626.[CrossRef][Web of Science][Medline]
Loose D.S., Kan P.B., Hirst M.A., Marcus R.A., Feldman D. (1983) J. Clin. Invest. 71:1495–1499.[Web of Science][Medline]
Matsuura K., Yoshioka S., Tosha T., Hori H., Ishimori K., Kitagawa T., Morishima I., Kagawa N., Waterman M.R. (2005) J. Biol. Chem. 280:9088–9096.
McLean K.J., Marshall K.R., Richmond A., Hunter I.S., Fowler K., Kieser T., Gurcha S.S., Bestra G.S., Munro A.W. (2002) Microbiology 148:2937–2949.
Meneau I. and Sanglard D. (2005) Med. Mycol. 43:(Suppl 1), S307–S311.
Murray M. and Ryan A.J. (1983) Chem. Biol. Interact. 43:341–351.[CrossRef][Web of Science][Medline]
Nickerson D., Wong L.L., Rao Z. (1998) Acta Crystallogr. D Biol. Crystallogr. 54:470–472.[CrossRef][Medline]
Nickerson D.P. and Wong L.L. (1997) Protein Eng. 10:1357–1361.
Otwinowski Z. and Minor W. (1997) Methods Enzymol. 276:307–326.[CrossRef][Web of Science]
Podust L.M., Poulos T.L., Waterman M.R. (2001) Proc. Natl Acad. Sci. USA 98:3068–3073.
Poulos T.L. and Howard A.J. (1987) Biochemistry 26:8165–8174.[CrossRef][Medline]
Poulos T.L. and Johnson E.F. (2005) In Ortiz de Montellano P.R. (Ed.). Cytochrome P450: Structure, Mechanism, and Biochemistry 3rd edn. , Plenum-, NY pp. 87–114.
Poulos T.L., Perez M., Wagner G.C. (1982) J. Biol. Chem. 257:10427–10429.
Poulos T.L., Finzel B.C., Howard A.J. (1986) Biochemistry 25:5314–5322.[CrossRef][Medline]
Poulos T.L., Finzel B.C., Howard A.J. (1987) J. Mol. Biol. 195:687–700.[CrossRef][Web of Science][Medline]
Raag R., Li H., Jones B.C., Poulos T.L. (1993) Biochemistry 32:4571–4578.[CrossRef][Medline]
Redinbo M.R. and Yeates T.O. (1993) Acta Crystallogr. D Biol. Crystallogr. 49:375–380.[CrossRef][Medline]
Sheng C., et al. (2004) J. Biomol. Struct. Dyn. 22:91–99.[Web of Science][Medline]
Verras A., Kuntz I.D., Ortiz de Montellano P.R. (2004) J. Med. Chem. 47:3572–3579.[CrossRef][Web of Science][Medline]
Yano J.K., Koo L.S., Schuller D.J., Li H., Ortiz de Montellano P.R., Poulos T.L. (2000) J. Biol. Chem. 275:31086–31092.
Yeates T.O. (1997) Methods Enzymol. 276:344–358.[Web of Science][Medline]
Received May 16, 2006; revised July 21, 2006; accepted August 1, 2006.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||