Insight Into Steroid Scaffold Formation from the Crystal Structure of Human Oxidosqualene

Insight into steroid scaffold formation from the crystal structure of human oxidosqualene cyclase

Ralf Thoma, Tanja Schulz-Gasch, Brigitte D’Arcy, Jörg Benz, Johannes Aebi, Henrietta Dehmlow, Michael Hennig, Martine Stihle, Armin Ruf

Address:

F. Hoffmann-La Roche AG, Pharma Research Discovery Chemistry, 4070 Basel, Switzerland

In higher organisms the formation of the steroid scaffold is catalyzed exclusively by the membrane bound oxidosqualene cyclase (OSC, lanosterol synthase). In a highly stereoselective cyclization reaction OSC forms lanosterol with seven chiral centers starting from the achiral linear substrate 2,3-oxidosqualene (Figure 1). Valuable data on the mechanism of the complex cyclization cascade has been collected during the last fifty years using suicide inhibitors, mutagenesis studies and homology modeling. Nevertheless it is still not fully understood how the enzyme catalyzes the reaction1,2. Due to the decisive role of OSC in sterol biosynthesis it represents a target for the discovery of novel drugs3 that could complement statins. Here we present two crystal structures of the human membrane protein OSC in complex with an inhibitor that showed cholesterol lowering in vivo and in complex with the reaction product lanosterol at 2.2Å and 2.1Å resolution respectively. The structures show the activation of the catalytic Asp455 by coordination with two cysteines and how the active site geometry enforces the boat conformation of the sterol B-ring. Product specificity is achieved by the confinement of polar groups to two limited areas in the predominantly hydrophobic active site.

The human OSC consists of two (/) barrel domains that are connected by loops and three smaller -structures. The large active site cavity is located in the center of the molecule between domains 1 and 2 (Figure 2). The N-terminal OSC region that is absent in other cyclase sequences fills the space between the two domains and could function in stabilizing their relative orientation (Figure 2). Five cyclase finger print QW-sequence motifs are located C-terminal of outer barrel helices. They assume the same conformation with stacked Gln and Trp side chains that was also observed in the bacterial squalene hopene cyclase (SHC) and is thought to contribute to cyclase fold stability during the highly exergonic cyclization reaction4,5. Human OSC is monomeric in the crystal which is consistent with analytical ultracentrifuge data6.

OSC is a monotopic membrane protein, i.e. it inserts into the membrane but does not span the bilayer7. The OSC membrane binding region in domain-2 can be detected by bound detergent molecules and by its hydrophobic surface (1.700Ų) which comprises 6% of the total enzyme surface (Figure 2). The membrane inserted surface consists of a plateau of 25Å diameter and a channel that leads to the active site cavity. This channel is supposed to allow the substrate oxidosqualene to enter the hydrophobic active site but a constriction site separates it from the active site cavity (Figure 2B)8,4. Passage of the substrate could be achieved either by a change in the side chain conformations of the residues Tyr237, Cys233 and Ile524 or by rearrangement of the strained loops 516-524 and 697-699 that contain the only two residues in the disallowed regions of the Ramachandran plot (Glu519 and Lys698). The electron density observed in the substrate entrance channel can be fitted with a tetradecyl carbon chain but not with the octyl chain of the detergent -OG and may belong to a partially ordered lipid (Figure 2B). In the crystals of other monotopic membrane enzymes detergents have been observed in the hydrophobic substrate entrance channels9,10. The approximate orientation of OSC in the membrane was inferred by aligning the glucose head groups of two bound -OG detergent molecules in the polar membrane layer and the octyl groups parallel to the fatty acid chains in the bilayer (Figure 2B). The two polar head groups are bound along one of the three helices that constitute the membrane binding region (Figure 2B). The non polar octyl tail of one -OG molecule interacts with the hydrophobic region of a neighboring OSC molecule in a crystal lattice contact. The defined binding and the involvement in a crystal lattice contact may explain why -OG was the only detergent that yielded OSC crystals6.

The catalytic mechanism for the polycyclization reaction involves several reaction steps2. First, (3S)-2,3-oxidosqualene adopts a pre-organized chair boat-chair conformation. Protonation of the epoxide ring then triggers a cascade of ring forming reactions to the protosterol cation. Skeletal rearrangement of this intermediate through a series of 1,2-hydride and -methyl group shifts and a final deprotonation step leads to the product lanosterol2. Product specificity and high stereoselectivity is believed to be achieved through several factors: (1) enforcing the substrate to occupy a prefolded conformation, (2) progression of the reaction via rigidly held, partially cyclized carbocationic intermediates and (3) stabilization of the intermediate carbocations by cation- interactions11-14, thus preventing early truncation of the cyclization cascade by deprotonation or nucleophilic addition of solvent molecules.

In order to gain more insight in the structural basis for this highly stereoselective cyclization, we cocrystallized OSC with its product lanosterol. The ligand was well defined in the difference map at 2.1Å resolution with clear electron density for the sterol scaffold and all lanosterol methyl substituents (Figure 3A). Lanosterol fits closely to the shape and physicochemical properties of the OSC active site cavity. One hydrogen bond is formed between the lanosterol 3-hydroxy group and the catalytic Asp455 that constitutes the polar cap of the mainly hydrophobic cavity. Based on this crystal structure the 2,3-oxidosqualene cyclization intermediates were modeled into the active site. This confirmed the hypothesis that initiation of the cyclization reaction is accomplished by Asp455 protonating the epoxide group of prefolded 2,3-oxidosqualene (Figure 1). 2,3-oxidosqualene is completely stable in glacial acetic acid for about one day and therefore Asp455 is expected to be further activated15. The X-ray structure assigns Cys456 and Cys533 to act as hydrogen-bonding partners with Asp455 (Figure 3) and thus contribute to the required increased acidity of Asp455 in OSC16. Reprotonation of Asp455 after completion of one catalytic cycle can be achieved either from the bulk solvent through a chain of water molecules and the carboxylate group of Glu459 (Figure 2) or by shifting the proton from the final deprotonation step back to Asp455.

The conserved side-chains of Phe444, Tyr503 and Trp581 are in an appropriate position and orientation to stabilize the intermediate tertiary cations at C6 and C10 after A-ring and B-ring formation via -cation-interactions (Figure 3A)1. Tyr98 is well positioned to enforce the energetically unfavorable boat conformation of 2,3-oxidosqualene required for lanosterol B-ring formation by pushing the methyl group at C10 below the molecular plain (Figures 1 and 3B). In the OSC sequence this is reflected by a one residue insertion above and a one residue deletion below the molecular plain compared to SHC that create the B-ring in chair conformation17. His232 and Phe696 side-chains are positioned to stabilize the anti-Markovnikov secondary cation created at C14 during C-ring formation (Figure 3) via -interactions. The OSC cyclization cascade stops with the tertiary protosterol cation at C20 after formation of the 5-membered D-ring (Figure 1), because OSC lacks an aromatic residue like Trp169 in SHC1 that stabilizes the secondary cation at C17 (protosterol cation numbering) required for hopene E-ring cyclization.

The protosterol C20 cation is converted to the lanosterol C8/C9 cation by skeletal rearrangement. The equilibrium between these cations is shifted, because the active site has a higher -electron density near C8/C9 with seven aromatic residues compared to three residues in 6 Å distance to C20. Truncation of the reaction by nucleophilic addition of water molecules is avoided by the predominantly hydrophobic nature of the active site cavity (Figure 3). In addition, the absence of basic residues in the proximity of the charge of the protosterol cation avoids premature truncation by deprotonation. As predicted in previous studies17,18 the highly conserved His232 is the only basic residue close enough to serve as a catalytic base in the specific deprotonation of the C8/C9 lanosterol cation that terminates catalysis (Figure 3C). However the hydroxygroup of Tyr503 that is hydrogen bonded to His232 would be in a proper position for the final deprotonation step.

Inhibitors of OSC as anticholesteremic drugs act downstream of farnesyl-pyrophosphate in the cholesterol pathway and do not interfere with the synthesis of isoprenoids and coenzyme Q. In hamsters pharmacologically active doses of an OSC inhibitor showed less adverse effects compared to a statin3. The structure of the bench mark inhibitor Ro 48-8071 (Figure 4) in complex with human OSC provides a structural base for the design of improved OSC inhibitors. A single molecule of the cocrystallized inhibitor Ro 48-8071 was well defined in the initial 2.2 Å (Fo-Fc)-electron density map. This is in agreement with competitive inhibitor binding and the 1:1 binding ratio derived from a fluorescence titration experiment6,19.

The basic nitrogen atom of the inhibitor Ro 48-8071 forms a charged hydrogen bond of 2.9Å distance with the Asp455 carboxylate that directs both oxygen atoms towards the active site cavity. The charged amino group is also stabilized through cation- interactions with several aromatic residues (Figure 4A). The fluoro phenylgroup of the inhibitor benzophenone moiety is stacked between the side chains of Phe696 (3.7Å) and His232 (3.1Å). The carbonyl group is hydrogen bonded to a water molecule which in turn interacts with the backbone amide nitrogen of Ile338 (Figure 4B). This water mediated interaction explains why the hydrogen bond donor or acceptor properties of groups substituting for benzophenone have no effect on IC5020. The terminal phenyl group interacts with Trp192 (3.6Å), Trp230 (3.7Å) and Phe521 (4.6Å) (Figure 4B). The π-electron rich pocket created by these residues is optimal for the stabilization of electron deficient aromatic systems21. The bromine atom of the inhibitor is interacting with the residues Ile524, Tyr237 and Cys233 close to the channel constriction site. The length of the linker between the amine nitrogen and the electron deficient aromatic group has a pronounced effect on inhibitor binding affinity22. The structure pinpoints that too long linkers lead to steric clashes or constrained inhibitor conformations and short inhibitors do not fully fill the binding site and loose binding affinity. Ro 48-8071 bound to OSC has a different conformation and different interactions than those observed in the complex with the bacterial SHC23 and illustrates why structure-based design based on SHC resulted in inaccurate predictions for OSC. The observation of the direct hydrogen bond between the catalytic Asp455 and the amino group of Ro 48-8071 prompted the design of novel structural classes of OSC inhibitors, such as amides that lack the charged amino group and therefore have improved physicochemical properties.

The high resolution crystal structures of human OSC provide for the first time a clear picture of the molecular mechanism of substrate entry and stereoselective cyclization reaction of sterol formation.The structure confirms the assignment of Asp455 as the catalytic acid, its further increase in acidity by the two cysteins Cys456 and Cys533 and His232 as the basic residue that terminates the cyclization reaction. Further biochemical investigation should be directed to confirm the proposed function of the hydrophobic substrate channel and aromatic residues that are crucial for the π-interaction with the reaction intermediates. Drug design will be facilitated by the elucidation of key interaction that can be used for the improvement of the inhibitory and physicochemical properties of drug molecules for lowering of cholesterol in men.

Methods:

Recombinant human OSC in a buffer containing 0.8% (w/v) -OG was used for crystallization as described6. For inhibitor cocrystals the protein was incubated with 5mM Ro 48-8071 over night. Crystals were grown at room temperature in hanging drops by mixing equal volumes of protein solution with the reservoir solution consisting of 25% (w/v) PEG 3350, 0.4M ammonium acetate, 0.1M Tris, pH8.5, 10% (v/v) ethylene glycol and the addition of seed crystals. Untwinned crystals were selected for data collection on the beam line PX06A (Swiss Light Source). Data were processed with DENZO and XDS24,25. The structure was solved by molecular replacement with the CCP426 program MOLREP27 using SHC10 (pdb-entry 2sqc ) as search model and was rebuilt by ARP/wARP28. After adding the inhibitor and detergents with MOLOC29 the model was refined with REFMAC30. The lanosterol structure was refined with BUSTER31. Difference electron density maps calculated with autoBUSTER31 from a model refined without a ligand clearly showed at the 3 sigma level the sterol scaffold featuring double substituents at C4 (Figure 3A). After refinement the electron density for the B- and C-rings of lanosterol remains weaker than for the rest of the molecule. Figures have been prepared with Ribbons32 and MOE 2003.02. Modeling of substrates and intermediates to the human OSC structure were done with MOE 2003.02 and MOLOC.

Acknowledgements We thank the staff at the beamline PX06A at the Swiss Light Source (SLS, Switzerland) for support and Clemens Vonrhein (Global Phasing Ltd.) for an early version of autoBUSTER. All colleagues at Roche Basel we thank for the supporting research atmosphere and especially O. Morand for stimulating discussions.

Correspondence and requests for materials should be addressed to A.R. (). The atomic coordinates have been deposited at the Protein Data Bank under the accession codes XX and XX.

References

1.Wendt, K.U., Schulz, G.E., Corey, E.J. & Liu, D.R. Enzyme Mechanisms for Polycyclic Triterpene Formation. Angew. Chem. Int. Ed. Engl. 39, 2812-2833. (2000).

2.Hoshino, T. & Sato, T. Squalene-hopene cyclase: catalytic mechanism and substrate recognition. Chem. Commun. (Camb) 4, 291-301. (2002).

3.Morand, O.H. et al. Ro 48-8071, a new 2,3-oxidosqualene:lanosterol cyclase inhibitor lowering plasma cholesterol in hamsters, squirrel monkeys, and minipigs: comparison to simvastatin. J. Lipid Res. 38, 373-390. (1997).

4.Wendt, K.U., Poralla, K. & Schulz, G.E. Structure and function of a squalene cyclase. Science 277, 1811-1815. (1997).

5.Poralla, K. et al. A specific amino acid repeat in squalene and oxidosqualene cyclases. Trends Biochem. Sci. 19, 157-158. (1994).

6.Ruf, A. et al. The monotopic membrane protein human oxidosqualene cyclase is active as monomer. Biochem. Bioph. Res. Co. 315 (2), 247-254 (2004).

7.Blobel, G. Intracellular protein topogenesis. Proc. Natl. Acad. Sci. USA 77, 1496-1500. (1980).

8.Milla, P. et al. Thiol-modifying inhibitors for understanding squalene cyclase function. Eur. J. Biochem. 269, 2108-2116. (2002).

9.Loll, P.J., Picot, D. & Garavito, R.M. The structural basis of aspirin activity inferred from the crystal structure of inactivated prostaglandin H2 synthase. Nat. Struct. Biol. 2, 637-643. (1995).

10.Wendt, K.U., Lenhart, A. & Schulz, G.E. The structure of the membrane protein squalene-hopene cyclase at 2.0 A resolution. J. Mol. Biol. 286, 175-187. (1999).

11.Corey, E.J., Matsuda, S.P.T., Baker, C.H., Ting, A.Y. & Cheng, H. Molecular cloning of a Schizosaccharomyces pombe cDNA Encoding Lanosterol Synthase and Investigation of Conserved Tryptophan Residues. Biochem. Bioph. Res. Co. 219, 327-331 (1996).

12.Zoltewicz, J.A., Maier, N.M. & Fabian, W.M.F. p-Cation and p-Dipole-Stabilizing Interactions in a Simple Model System with Cofacial Aromatic Rings. J. Org. Chem. 63, 4985-4990 (1998).

13.Miklis, P.C., Ditchfield, R. & Spencer, T.A. Carbocation-p Interaction: Computational Study of Complexation of Methyl Cation with Benzene and Comparisons with Related Systems. J. Am. Chem. Soc. 120, 10482-10489 (1998).

14.Gallivan, J.P. & Dougherty, D.A. Cation-p interactions in structural biology. Proc. Natl. Acad. Sci. USA 96, 9459-9464. (1999).

15.Corey, E.J. et al. Methodology for the Preparation of Pure Recombinant S. cerevisiae Lanosterol Synthase Using a Baculovirus Expression System. Evidence That Oxirane Cleavage and A-Ring Formation Are Concerted in the Biosynthesis of Lanosterol from 2,3-Oxidosqualene. J. Am. Chem. Soc 119, 1277-1288 (1997).

16.Gandour, R.D. On the importance of orientation in General Base catalysis by carboylate. Bioorganic. Chem. 10, 169-176 (1981).

17.Schulz-Gasch, T. & Stahl, M. Mechanistic insights into oxidosqualene cyclizations through homology modeling. J. Comput. Chem. 24, 741-753. (2003).

18.Corey, E.J. et al. Studies on the Substrate Binding Segments and Catalytic Action of Lanosterol Synthase. Affinity Labeling with Carbocations Derived from Mechanism-Based Analogs of 2,3-Oxidosqualene and Site-Directed Mutagenesis Probes. J. Am. Chem. Soc. 119, 1289-1290 (1997).

19.Abe, I., Zheng, Y.F. & Prestwich, G.D. Photoaffinity labeling of oxidosqualene cyclase and squalene cyclase by a benzophenone-containing inhibitor. Biochemistry 37, 5779-5784. (1998).

20.Brown, G.R. et al. Quinuclidine inhibitors of 2,3-oxidosqualene cyclase-lanosterol synthase: optimization from lipid profiles. J. Med. Chem. 42, 1306-1311. (1999).

21.Dehmlow, H. et al. Synthesis and structure-activity studies of novel orally active non-terpenoic 2,3-oxidosqualene cyclase inhibitors. J. Med. Chem. 46, 3354-3370. (2003).

22.Jolidon, S., Polak-Wyss, A., Hartmann, P.G. & Guerry, P. 2,3-Oxidosqualene-lanosterol cyclase: an attractive target for antifungal drug design. in Recent Advances in the Chemistry of Anti-infective Agents (ed. Bentley, P.H.) 223-233 (Royal Soc. Chemistry, Cambridge, U.K, 1993).

23.Lenhart, A., Weihofen, W.A., Pleschke, A.E. & Schulz, G.E. Crystal structure of a squalene cyclase in complex with the potential anticholesteremic drug Ro48-8071. Chem. Biol. 9, 639-645. (2002).

24.Otwinowski, Z. & Minor, W.M. Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307-326 (1997).

25.Kabsch, W. Automatic processing of rotation diffraction data from crystals of initially unknown symmetry and cell constants. J. Appl. Cryst. 26, 795-800 (1993).

26.CCP4 (Collaborative Computational Project, N. The CCP4 suite: programs for protein crystallography. Acta Crystallogr. D, 760-763 (1994).

27.Vagin, A. & Teplyakov, A. MOLREP: an automated program for molecular replacement. J. Appl. Cryst. 30, 1022-1025 (1997).

28.Morris, R.J., Perrakis, A. & Lamzin, V.S. ARP/wARP's model-building algorithms. I. The main chain. Acta Crystallogr. D 58, 968-975. (2002).

29.Gerber, P.R. Peptide mechanics: a force field for peptides and proteins working with entire residues as small unites. Biopolymers 32, 1003-1017 (1992).

30.Murshudov, G.N., Vagin, A.A., Lebedev, A., Wilson, K.S. & Dodson, E.J. Efficient anisotropic refinement of macromolecular structures using FFT. Acta Crystallogr. D 55, 247-255. (1999).

31.Roversi, P., Blanc, E., Vonrhein, C., Evans, G. & Bricogne, G. Modelling prior distributions of atoms for macromolecular refinement and completion. Acta Crystallogr. D 56, 1316-1323. (2000).

32.Carson, M. Ribbon. Methods Enzymol. 277, 493-505 (1997).

33.Laskowski, R.A., MacArthur, M.W., Moss, D.S. & Thornton, J.M. PROCHECK: a program to check the stereochemical quality of protein structure. J. Appl. Crystallogr. 26, 283-291 (1993).

Table 1: Data collection and refinement statistics2004-07-21809

Data Collection / Ro 48-8071 / Lanosterol
Wavelength (Å) / 0.92 / 0.98
Resolution1 (Å) / 20-2.20 (2.26-2.20) / 20-2.0 (2.09-2.0)
Unique reflections1 / 60002 (4884) / 79898 (9547)
Completeness (%)1 / 98.0 (97.5) / 99.2 (96.3)
Rmerge (%)1,2 / 9.9 (37.6) / 7.6 (48.9)
<I/1 / 7.7 (2.5) / 14.4 (3.1)
Unit Cell (Space group C2221) / 189.9 Å 202.4 Å 62.6 Å / 189.6 Å 201.5 Å 62.1 Å
Refinement / Refmac5 / autoBUSTER
Resolution (Å) / 20-2.2 (2.26-2.20) / 20-2.1 (2.155-2.10)
Rcryst1,3 / 19.2 (26.2) / 14.75.1 (18.8)
Rfree1,4 / 22.5 (30.2) / 18.89 (24.0)
Average B-factor (Ų) / 38.1 / 33.5
R.m.s. deviations from ideality Bond lengths (Å) / angles (°) / 0.01 / 1.1 / 0.01 / 1.6
Main chain dihedral angles (%) Most favored / allowed / disallowed 5 / 92.3 / 6.4 / 0.3 / 93.3 / 5.9 / 0.3

1 Values in parentheses refer to the highest resolution bins.