- Email: [email protected]
Antagonist binding in the rat muscarinic receptor
Computational Biology and Chemistry 28 (2004) 375–385
Antagonist binding in the rat muscarinic receptor A study by docking and X-ray crystallography Anna C. Tanczosa , Rex A. Palmerb , Brian S. Potterb , Jos´e W. Saldanhac , Brendan J. Howlina,∗ a
Department of Chemistry, School of Biomedical and Molecular Sciences, University of Surrey, Guildford, Surrey GU2 7XH, UK b School of Crystallography, Birkbeck College, University of London, Malet Street, London WC1E 7HX, UK c National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 1AA, UK
Received 9 September 2004; received in revised form 29 September 2004; accepted 29 September 2004
Abstract A series of agonists to the rat muscarinic receptor have been docked computationally to the active site of a homology model of rat M1 muscarinic receptor. The agonists were modelled on the X-ray crystal structure of atropine, which is reported here and the docking studies are shown to reproduce correctly the order of experimental binding affinities for the agonists as well as indicate where there appear to be inconsistencies in the experimental data. The crystal and molecular structure of atropine (tropine tropate; ␣-[hydroxymethyl]benzeneacetic acid 8-methyl[3.2.1]oct-3-yl ester C17 H23 NO3 ) has been determined by X-ray crystallography using an automated Patterson search method, and refined by full-matrix least-squares to a final R of 0.0452 for 2701 independent observed reflections and 192 parameters using Mo K␣ ˚ at 150 K. The compound crystallises in space group Fdd2 with Z = 16 molecules per unit cell. radiation, λ = 0.71073 A © 2004 Published by Elsevier Ltd. Keywords: Crystal structure; Receptor binding; Docking of atropine; Docking; Homology modelling
1. Introduction G-protein coupled receptors (GPCRs) are the largest group of cell surface receptors (Dong et al., 2001) and as such are targets for a number of therapy discovery programs. The signalling is a two-way process: (1) the binding of a ligand facilitates the binding of a G-protein on the inside of the cell and (2) it then elicits a second messenger response. Conversely, if a G-protein binds prior to ligand binding, the affinity for the ligand is altered. Many different ligands bind to GPCRs. These range from a single photon, in the case of the opsins, through to small molecules, which bind in a channel between the seven helices and progress in size to proteins, which bind to the extracellular loops and do not enter into the channel ∗
Corresponding author. Tel.: +44 1483 686834; fax: +44 1483 686851. E-mail address: [email protected] (B.J. Howlin).
1476-9271/$ – see front matter © 2004 Published by Elsevier Ltd. doi:10.1016/j.compbiolchem.2004.09.009
between the seven helices. There are thought to be at least two different conformations of GPCRs: an inactive (R) and active (R*) conformation which exist in equilibrium. It is possible that the various ligands either cause a difference in the relative amount of time spent in or the accessibility of these two states. In general, only four of the helices are involved directly with binding the ligand, namely 3, 5, 6 and 7 (Flower, 1999) although there are often other interactions. As a result of the difficulty of crystallising membrane proteins, there is only one set of atomic coordinates for this family of receptors, that of bovine rhodopsin (Palczewski et al., 2000). This has resulted in a rapidly increasing number of models being produced based on this one structure and refined using the vast databases of biological data available, such as mutagenesis data (Shacham et al., 2001). A recent paper has discussed homology models for four of the classes of GPCRs and concluded that they can be used for drug design studies
376
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385
(Vaidehi et al., 2002), although Archer et al. (2003) have concluded that care is needed in the construction of homology models and experimental data is useful to correctly guide these studies. 1.1. Agonists, antagonists and inverse agonists During its random sampling of conformational space, a receptor will at times adopt the active conformation without the presence of an agonist. The existence of the receptor in this state–activated without an agonist–normally occurs so infrequently that it is considered to be zero for most purposes. However, there are certain receptors, such as rat acetylcholine muscarinic M1 receptor (m1AChR) (Spalding and Burnstein, 2001; Jakubik et al., 1995), for which constitutive activity is observed when the receptor is present in larger concentrations than normal in a transfected system. That is to say that they attain the active state a significant number of times without the presence of an agonist (Parnot et al., 2002; Chen et al., 1999). The response to an agonist remains increased activation and a neutral antagonist elicits no response. However, some antagonists inhibit the constitutive activation; these are called inverse agonists. Although this phenomenon was initially evident only in systems with a high concentration of receptors it has been shown to exist in natural systems (Bakker et al., 2000). It has also been shown that many human diseases are a consequence of constitutive activity of receptors. This arises as a consequence of mutations in the receptor or increased expression. These diseases include hyperthyroidism (TSH receptor) (Parma et al., 1993), retinal degeneration (rhodopsin) (Robinson et al., 1992) and hyper-pigmentation (LH receptor) (Weinstein and Shenker, 1993). Inverse agonists as opposed to neutral antagonists are required to prevent the activation of these mutant receptors. Muscarinic receptors are G-protein coupled receptors found in the parasympathetic nervous system and extensively in the brain, particularly in the cerebral cortex and the hippocampus. Acetylcholine is the endogenous agonist for this receptor. One of the features of the binding of many of these ligands to muscarinic receptors is the quaternary ammonium moiety. The presence of the N-methyl group changes the activity preventing these ligands from crossing the blood–brain barrier and improves the binding affinity of a given ligand. A proposed mechanism for activation of the receptor (Lu et al., 2001) begins with the enclosure of this group within an aromatic cage that triggers an alteration in the hydrogen bonding network leading to helical movement and thus to the active state of the receptor. Modelling studies are important for examination of the mechanistic context of the data that shed light on these interactions. Here we look at the initial interaction of m1AChR with various atropine analogues and examine the role of these compounds, some of which are inverse agonists. The X-ray analysis of the inverse agonist, atropine (Fig. 1), has been carried out in order to provide an accurate model of the ligand for use in the receptor modelling and binding studies.
Fig. 1. Atropine chemical formula: atom numbering used in the X-ray analysis; ring labels are also shown. Drawn with Chemwindow (1999–2003).
2. Experimental 2.1. Model of rat muscarinic receptor The model used has been described previously (Lu et al., 2001). 2.2. X-ray structural analysis of atropine The discussion of the X-ray determination is collected in Appendix A. The X-ray data for the atropine crystal are shown in Table 1. The rest of the X-ray data appear in the tables in Appendix A. Interestingly, the ring conformations of the tropine ring are very similar to published work on Table 1 Crystal data and structure refinement for atropine Identification code
Atropine
Empirical formula Formula weight Temperature (K) ˚ Wavelength (A) Crystal system Space group Unit cell dimensions
C17 H23 NO3 289.36 100(2) 0.71073 Orthorhombic Fdd2 ˚ b = 39.538(8) A, ˚ a = 24.291(5) A, ˚ α = 90◦ , β = 90◦ , c = 6.4727(13) A, γ = 90◦ 6217(2) 16 1.237 0.084 2496 0.20 × 0.15 × 0.35 3.30 to 25.02 −28 ⇐ h ⇐ 28, −46 ⇐ k ⇐ 46, −7 ⇐ l ⇐ 7 14,353 2701 [R(int) = 0.0820] 99.7% Full-matrix least-squares on F2 2701/1/192 0.831 R1 = 0.0452, wR2 = 0.1206 R1 = 0.0558, wR2 = 0.1294 −0.1(16) 0.00108(14) ˚ −3 0.262 and −0.225 e A
˚ 3) Volume (A Z Density (calculated) ( Mg/m3 ) Absorption coefficient ( mm−1 ) F(0 0 0) Crystal size (mm3 ) Theta range for data collection (◦ ) Index ranges Reflections collected Independent reflections Completeness to theta = 25.02◦ Refinement method Data/restraints/parameters Goodness-of-fit on F2 Final R indices [I > 2sigma(I)] R indices (all data) Absolute structure parameter Extinction coefficient Largest diff. peak and hole
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385
the X-ray structures of two atropine-type alkaloid biosynthetic precursors, published by Yamashita, Endo, Higashi, Nakatsu, Yamada, Oda and Kato in the structure of tropinone reductase-II complexed with NADP+ and pseudotropine at ˚ resolution (Yamashita et al., 2003). The B1 B2 ring 1.9 A conformation (Fig. 1) is a distorted boat and the B1 B3 rings ˚ are in a sofa conformation. The resolution of our data is 0.7 A, so individual bonds can be resolved whereas the studies mentioned above are less than atomic resolution, so it is particularly gratifying that they are in agreement. 2.3. Docking of ligands to the receptor model Docking was carried out with the program AUTODOCK (Morris et al., 1998). Molecular models of the atropine analogues (Table 2) were modelled in Viewer Pro 4.2 (Accelerys, Inc.) using the atropine X-ray structure as a template, with replacement and addition of groups performed manually in the program. All structures were subjected to the CLEAN routine in ViewerPro 4.2 to remove unlikely geometrical features. These ligand models were transferred to AUTODOCK TOOLS (http://www.scripps.edu/∼sanner/python/adt) to prepare them for docking. Energy minimisation of the X-ray structure of atropine in Viewer Pro 4.2 showed no appreciable difference in conformation, indicating that the atropine conformation had not been modified by crystal packing efTable 2 Structures of ligands used in this study
377
fects and the structure could be used as solved. Therefore, all ligand structures were only subjected to the CLEAN routine in ViewerPro 4.2, so that there would be little deviation from the starting atropine model. AUTODOCK TOOLS was used to identify the torsion angles in the ligands, add the solvent model and assign partial atomic charges. The ligand models were docked to the model of rat muscarinic receptor using the default settings of AUTODOCK for the Lamarkian genetic algorithm. Hence, a flexible ligand was docked to a rigid protein. The fitness function that was optimised by the genetic algorithm was a function of the conformation of the ligand (defined by its flexibility) and its interaction with the protein (by non-bonded and charge interactions). The results of the docking studies were the 10 best docked ligands quantified in terms of the calculated free energy of binding. This process is essentially a flexible ligand docking to a rigid protein and ideally the docking should be carried out with a flexible ligand and a flexible protein, so we have approximated this by also carrying out simulated annealing and molecular dynamics of the receptor model. 2.4. Mutation studies Experimental data are also available on two mutants of m1AchR (Ward et al., 1999). These are Tyr381Phe and Tyr381Ala. We have simulated these experiments by making the relevant point mutations in DEEPVIEW (Peitsch et al., 1997), energy minimising the resulting mutant using the AMBER suite of programs (Weiner and Kollman, 1981) (running on Columbus, the Computational Chemistry Facility at the Rutherford Appleton Laboratory), and running the docking simulations for both cases. The calculated free energies of binding have then been correlated with the experimental values of pIC50 (Ward et al., 1999). The pIC50 value is the negative logarithm of the concentration of ligand that inhibits 50% of the activity. The experimental pIC50 values are reported to two significant figures; therefore for comparison purposes we have quoted our calculated binding energies to the same number of significant figures. It must be borne in mind, however, that this does not reflect on the reproducibility or accuracy of the calculated results. Repeating a docking run with the same models invariably gives the same result, so no estimate of error is possible with the present calculations. 2.5. Simulated annealing and molecular dynamics of the receptor Simulated annealing and molecular dynamics of the receptor were performed using the AMBER suite of programs. Simulated annealing was carried out for 5000 fs at a simulated temperature of 1000 ◦ C. After equilibration at 1000 ◦ C for 1000 fs, the temperature remained constant for a further 1500 fs and was then cooled in two stages, fast for 1000 fs and slow for the remaining 1500 fs with a distance-dependent
378
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385
Table 3 Ether linkage parameters used in AMBER Bond parameters Bond
˚ R0 (A)
˚ 2) K (kcal/mol/A
OS C OS CT
1.31 1.41
350 320
Angle
θ 0 (◦ )
K (kcal/mol/◦2 )
C OS CT OS CT CT OS C CT OS C O OS CT HC N CT HC C CT CA
109.5 109.5 120 123 109.5 109.5 109.5
60 70 123 145 57 49 49
Angle parameters
Fig. 2. ORTEP/POV-Ray view of the stereochemistry of the atropine molecule in the solid state.
Torsion parameters Torsion angle
k (kcal/mol/◦2 )
Periodicity, n
θ (◦ )
CT OS C CT O C OS CT
4.5 4.5
2 2
180 180
Atoms types given are AMBER atom types.
dielectric. No explicit solvent atoms or lipids were included in the calculation. The alpha carbon atoms of the protein were constrained with a force constant of 1 during the simulated annealing runs to prevent unfolding of the protein; essentially we are simulating different side chain conformations of the same active state by constraining the helices. This process was repeated 10 times, each time using the output of the previous run, both with and without the ligand. The docking studies were repeated with these structures. In order to examine unconstrained movement, molecular dynamics was performed, both with and without atropine present in the active site of the receptor. AMBER was parameterised for atropine by comparisons with the CVFF (BIOSYM Technologies Inc., 1990) force field for the missing parameters concerning the ether linkage. A correlation table was drawn up with known AMBER parameters versus known CVFF parameters. A scaling factor was applied to convert the known CVFF parameters to the known AMBER parameters and this factor was used to convert the known CVFF parameters for the ether linkage to the respective AMBER parameters. This process must be considered an approximate method but care was taken to ensure that the new parameters would not unbalance the force field. Parameters for the ether linkage used in AMBER are given in Table 3. RESP (Bayly et al., 1993) was used to fit the charges from an electrostatic potential grid produced from a single point calculation in GAMESS (Schmidt et al., 1993) using the 631G* basis set. A more complete study would involve the derivation of the respective parameters by molecular orbital calculations on suitable small molecules containing the respective linkage. However, our aim in this work was to use the atropine as a ‘bung’ to keep the shape of the active site of the receptor.
3. Results and discussion The refined atomic co-ordinates and isotropic parameters for non-hydrogen atoms of the atropine are given in Table A.1. Bond distances and bond angles are in Table A.2. An ORTEP/POV-Ray view of the molecular conformation of atropine is shown in Fig. 2. 3.1. Docking studies (native protein) The docked conformations of atropine are located in a site between transmembrane helices TM3, TM5, TM6 and TM7 (Fig. 3). The figure shows the shape of the binding site in the receptor. This is composed of a van der Waal’s surface (coloured dark blue) and showing the volume occupied by the amino acids in the helices. The backbone of TM3 is shown as a helical ribbon coloured yellow and the amino acid side chains of this helix which are shown in wire frame format are also coloured yellow. TM5 is coloured red, TM6 is green and TM7 is pink. An electrostatic potential is superimposed on the left-hand part of the van der Waal’s surface showing the
Fig. 3. Atropine docked into the binding site of the rat muscarinic receptor from the homology model.
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385
distribution of charges in the binding site. The red patches indicate areas of negative charge, which serve to attract the positively charged end of the ligand, which is shown in CPK format in the middle of the binding site. CPK format shows carbon atoms as grey spheres, nitrogen atoms as blue spheres and oxygen as red spheres. The ‘walls’ of this binding pocket are made up of the amino acid residues TM3: Asp105, Tyr106, Ser109 and Asn110; TM5: Thr192, Ala196, Phe197 and Tyr198; TM6: Phe374, Trp378, Tyr381, Asn382 and Leu386; TM7: Tyr404, Cys407 and Tyr408. These are all residues that have been implicated in the binding of inverse agonists (Lu et al., 2001; Ward et al., 1999). Our studies have indicated that there is a possibility of a ring-stacking interaction between the aromatic ring of the antagonists and Tyr106 and Phe197, but unfortunately such interactions are not modelled explicitly in the docking software. Molecular orbital calculations need to be used to study ring-stacking interactions and this will be an area of future study. All of the remaining atropine analogues are found to bind to this site. Acetylcholine, the natural agonist, displays a slightly different binding behaviour, as it is located deeper within the negatively charged area of the binding pocket than the atropine. The smaller size of acetylcholine, when compared to those of the inverse agonists, without the bulky aromatic tail offers a reason for this behaviour. The above argument suggests a rationale for the observation that both the Gbinding and pIC50 are lower for acetylcholine than for atropine and its analogues, it does not fit as snugly into the site as the larger antagonists and is thus not as energetically stabilised. This may be what distinguishes the agonists from the inverse agonists, the agonists allowing the receptor freedom of movement and thus a conformational change. Both agonist and inverse agonists are found to bind with the ammonium head group buried inside the receptor in the vicinity of Asp105, which contributes a negative charge to this area of the binding pocket. A plot of the pIC50 versus minimum free energy of binding is given as Fig. 4. A table of the values is given as Table 4. It is important to consider the correct concentrations of the ligands as this will affect the pIC50 values used. Some of the ligands were used as racemic mixtures of both isomers, so only half of the concen-
379
Table 4 Table of molecule docked with their calculated average and minimium G binding energies in kJ/mol and pIC50 values in mol Molecule
Ave. DG binding
Min. DG binding
pIC50
Benzilyltropine Benztropine N-Methylatropine Atropine N-Methylhomatropine dl-Homatropine N-Methylacetyltropine Acetylcholine Diphenylacetyltropine Phenylacetyltropine
−58.93 −55.64 −49.62 −50.27 −46.71 −46.92 −33.99 −29.30 −58.39 −48.14
−59.79 −56.27 −50.84 −51.67 −47.82 −47.15 −34.31 −29.66 −59.16 −48.28
10.05 9.75 9.56 9.19 7.61 7.37 5.74 4.78 8.42 6.94
pIC50 values are taken from reference 20 and those from racemic mixtures have been corrected by the method outlined.
tration is active. There are two ways to correct for this, either by decreasing the concentration of the non-racemic ligands or increasing the concentrations of the racemic ligands. We have tried both methods and found no substantial difference in the results. Therefore, for the purposes of this paper, those ligands with no chiral centres have had their pIC50 values corrected by only considering half of the concentration of ligand in order to normalise the data with those studied as a 50:50 racemic mix. These plots show the excellent rank order of the free energy of binding with the pIC50 values. The correlation between average free energy of binding and pIC50 is over 90% and that of minimum free energy of binding is 90%. This good correlation with experimental data shows that the approach used in this work is valid. It is therefore possible to produce good correlations with experimental data using a model of the receptor and models of the ligands. The use of X-ray crystallography here is important as it gives confidence in the three-dimensional structure of the atropine produced by the molecular modelling. It is also interesting to note that the last two ligands in Table 4 do not fit with the correlation produced by the others. On further investigation, it was found that these two compounds had been obtained from a different source than the rest of the ligands studied and were used in the experimental studies without further purification. For this reason, we have excluded these compounds from the study. 3.2. Docking studies (point mutations to the protein)
Fig. 4. Plot of pIC50 vs. minimum free energy of binding.
Two mutants of the protein have been studied experimentally: Tyr381Ala and Tyr381Phe; so it was of interest to see if we could also reproduce the docking to these mutants (Ward et al., 1999). The point mutations were made in DEEPVIEW using the mutate facility and the resulting structures energy minimised in AMBER. Docking of the same ligands was carried out to both mutants and the results again correlated with the experimental pIC50 data. The average of the best docking results for the Ala mutant had an r2 of 93% (Table 5) and that of the Phe mutant was slightly better at 95%. In the first mutant, both an aromatic ring and a potential hydrogen-
380
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385
Table 5 Table of molecule docked to the point mutations with their calculated average and minimium G binding energies in kJ/mol and pIC50 values in mol Ave. DG (kJ/mol)
Min. DG (kJ/mol)
Ave. best (kJ/mol)
pIC50
Tyr381Ala Benzilyltropine Benztropine N-Methylatropine Atropine N-Methylhomatropine dl-Homatropine N-Methylacetyltropine acetylcholine
−46.18 −44.37 −36.96 −37.36 −36.02 −36.15 −24.79
−47.40 −47.82 −38.20 −39.58 −38.66 −37.66 −25.90
−46.39 −44.56 −36.69 −37.15 −35.73 −36.32 −24.85
9.77 8.92 7.17 7.12 6.26 6.38 4.79
Tyr381Phe Benzilyltropine Benztropine N-Methylatropine Atropine N-Methylhomatropine dl-Homatropine N-Methylacetyltropine acetylcholine
−50.78 −47.25 −40.46 −41.21 −39.91 −38.38 −27.38
−51.51 −47.36 −41.88 −42.47 −41.97 −40.29 −27.87
−50.78 −47.25 −40.17 −41.13 −39.91 −38.33 −27.30
10.54 9.62 8.74 8.49 7.07 7.06 4.77
pIC50 values are taken from reference 20. Again the pIC50 values have been corrected for racemic mixtures.
bonding site are removed, thereby making the pocket larger and possibly less specific. This is reflected in the substantial drop in the observed pIC50 values and our calculated binding energies. The second mutation keeps the aromatic ring but loses the hydrogen-bonding group, so the drop in pIC50 values is much less and this is again reflected in the calculated binding energies. These results suggest that the mutations are only affecting the binding pocket and not the overall fold of the protein as it is unlikely that such a good correlation with experimental data would be achieved if the fold of the protein had been disrupted.
plication of this constraint whilst compensating for the lack of a lipid membrane or explicit solvent would also have the effect of preventing any major conformational changes which might be necessary for the production of alternative receptor conformations or states. Hence we can observe the movement of amino acid side chains to block the binding site in the simulations where the ligand is not present but to infer that this represents the inactive protein conformation would not be inferring too much. Further studies on the unconstrained model in a membrane with explicit solvent would be needed to explore this aspect further.
3.3. Simulated annealing 4. Conclusion The simulated annealing simulations of the protein–ligand complex show that the binding site moves little compared with the rest of the protein. Ring flips are observed on the exterior of the protein, but the conformations of the key residues within the binding area are maintained. The root mean square deviation (rmsd) of all the atoms in the lowest ˚ from the energy structure after simulated annealing is 0.84 A original model, whereas the key residues shaping the bind˚ The ing pocket, mentioned above, have an rmsd of 0.47 A. atropine molecule remained in a similar orientation and location. Simulated annealing of the protein alone results in a binding site altered in such a way that atropine cannot enter the charged area. This is mainly due to residues Phe197 and Trp378 moving into the space that the ligand would otherwise occupy. The latter is particularly frequently observed with the largest motion into this area, thus facilitating the movement of the helices probably by the disruption of a hydrogen-bonding network directly below the binding site. The protein retains its overall fold in the simulated annealing studies. This is not surprising as the structure was constrained by a force constant to prevent unfolding. The ap-
A good correlation was obtained for the docking studies of a number of related inverse agonists using models based on the crystal structure of atropine and a model of m1AchR produced previously. The experimental results could be reproduced for both the native protein and two point mutations that had been studied experimentally for a range of inverse agonists with different sizes and donor/acceptor groups. This provides confidence in this approach to computational drug design and provides a basis for the prediction of the inhibition constants of previously untested agonists for this system.
5. Supplementary material Crystallographic data (excluding structure factors) for the structure reported in this paper have been deposited with the Cambridge Crystallographic Data Centre as supplementary publication No. CCDC-252464. Copies of available material can be obtained, free of charge, on application to the Director, CCDC, 12 Union Road, Cambridge CB2 1EZ, UK (fax: +44 1223 336033 or e-mail: [email protected]).
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385
Acknowledgments
Table A.2 ˚ and angles (◦ ) for atropine Bond lengths (A)
The authors would like to thank the EPSRC for a grant of computer time on the computational chemistry facility at RAL, for collection of the single-crystal X-ray data at Southampton University and for a studentship for ACT. The authors would particularly like to thank Ed Hulme for his valued comments on a draft of this paper.
Appendix A The compound was supplied by Sigma Chemicals Ltd. as a racemic mixture. Space group Fdd2 has inversion elements (the d-glides) and thus contains both enantiomorphs. The crystal structure contains both configurations of the molecule although only one is input to the X-ray calculations. The known active enantiomorph was selected for the docking studies described here and its co-ordinates are given in Tables A.1 and A.4. Crystallisation was achieved by controlled slow evaporation from methanol at −4 ◦ C. A suitable crystal of dimension 0.20 mm × mm 0.15 × 0.35 mm was selected for X-ray analysis. Intensity data were collected on an Enraf-Nonius CCD diffractometer controlled with the COLLECT (Hooft, 1998) software, using monochromated Mo K␣ ˚ The diffractometer was equipped radiation, λ = 0.71073 A. with an Oxford Cryosystems “Cryostreams” cooler (Cosier and Glazer, 1986), enabling the data to be collected at 150 K. A total exposure time of 119.5 min recorded 474 images at 1◦ intervals. A total of 17,984 integrated reflections was collected of which 2377 were unique. The resolution range was ˚ The crystal showed no significant variation in 20.00–0.70 A. Table A.1 Atomic coordinates (×104 ) and equivalent isotropic displacement parame˚ 2 ) for atropine ters (103 × A C(1) C(2) C(3) C(4) C(5) C(6) C(7) C(8) C(9) C(10) C(11) C(12) C(13) C(14) C(15) C(16) C(17) O(3) O(1) O(2) N(1)
381
x
y
z
U(eq)
3241(1) 2878(1) 2791(1) 3063(1) 3420(1) 3510(1) 3897(1) 4467(1) 3652(1) 3098(1) 3459(1) 3365(1) 3455(1) 2879(1) 2519(1) 2528(1) 2663(1) 3356(1) 3699(1) 4832(1) 2774(1)
45(1) −228(1) −449(1) −406(1) −135(1) 93(1) 393(1) 283(1) 646(1) 1145(1) 1460(1) 1674(1) 1475(1) 1327(1) 1445(1) 1204(1) 2048(1) 876(1) 634(1) 556(1) 1766(1)
11581(3) 11832(3) 10227(4) 8390(4) 8128(3) 9734(3) 9458(3) 8719(3) 7952(3) 7774(3) 7857(3) 9785(3) 11818(3) 12313(3) 10510(3) 8655(3) 11252(3) 8984(2) 6112(2) 8568(2) 9805(3)
36(1) 45(1) 47(1) 41(1) 34(1) 28(1) 27(1) 30(1) 27(1) 28(1) 28(1) 30(1) 33(1) 33(1) 30(1) 29(1) 44(1) 32(1) 41(1) 37(1) 29(1)
U(eq) is defined as one-third of the trace of the orthogonalised Uij tensor.
˚ Bond lengths (A) C(1) C(6) C(1) C(2) C(1) H(1A) C(2) C(3) C(2) H(2A) C(3) C(4) C(3) H(3) C(4) C(5) C(4) H(4A) C(5) C(6) C(5) H(5A) C(6) C(7) C(7) C(9) C(7) C(8) C(7) H(7) C(8) O(2) C(8) H(8A) C(8) H(8B) C(9) O(1) C(9) O(3) C(10) O(3) C(10) C(16) C(10) C(11) C(10) H(10) C(11) C(12) C(11) H(11A) C(11) H(11B) C(12) N(1) C(12) C(13) C(12) H(12) C(13) C(14) C(13) H(13A) C(13) H(13B) C(14) C(15) C(14) H(14A) C(14) H(14B) C(15) N(1) C(15) C(16) C(15) H(15) C(16) H(16A) C(16) H(16B) C(17) N(1) C(17) H(17A) C(17) H(17B) C(17) H(17C) O(2) H(2) Bond angles (◦ ) C(6) C(1) C(2) C(6) C(1) H(1A) C(2) C(1) H(1A) C(3) C(2) C(1) C(3) C(2) H(2A) C(1) C(2) H(2A) C(4) C(3) C(2) C(4) C(3) H(3) C(2) C(3) H(3) C(3) C(4) C(5) C(3) C(4) H(4A) C(5) C(4) H(4A) C(4) C(5) C(6)
1.375(3) 1.401(3) 0.9500 1.375(3) 0.9500 1.371(3) 0.9500 1.390(3) 0.9500 1.392(3) 0.9500 1.525(2) 1.518(2) 1.527(2) 1.0000 1.399(2) 0.9900 0.9900 1.197(2) 1.339(2) 1.461(2) 1.515(3) 1.526(2) 1.0000 1.524(3) 0.9900 0.9900 1.480(2) 1.549(3) 1.0000 1.549(3) 0.9900 0.9900 1.532(3) 0.9900 0.9900 1.482(2) 1.534(3) 1.0000 0.9900 0.9900 1.482(3) 0.9800 0.9800 0.9800 0.8400 120.39(19) 119.8 119.8 119.88(19) 120.1 120.1 120.15(19) 119.9 119.9 120.2(2) 119.9 119.9 120.38(19)
382
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385 Table 2 (Continued )
Table 2 (Continued ) C(4) C(5) H(5A) C(6) C(5) H(5A) C(1) C(6) C(5) C(1) C(6) C(7) C(5) C(6) C(7) C(9) C(7) C(6) C(9) C(7) C(8) C(6) C(7) C(8) C(9) C(7) H(7) C(6) C(7) H(7) C(8) C(7) H(7) O(2) C(8) C(7) O(2) C(8) H(8A) C(7) C(8) H(8A) O(2) C(8) H(8B) C(7) C(8) H(8B) H(8A) C(8) H(8B) O(1) C(9) O(3) O(1) C(9) C(7) O(3) C(9) C(7) O(3) C(10) C(16) O(3) C(10) C(11) C(16) C(10) C(11) O(3) C(10) H(10) C(16) C(10) H(10) C(11) C(10) H(10) C(12) C(11) C(10) C(12) C(11) H(11A) C(10) C(11) H(11A) C(12) C(11) H(11B) C(10) C(11) H(11B) H(11A) C(11) H(11B) N(1) C(12) C(11) N(1) C(12) C(13) C(11) C(12) C(13) N(1) C(12) H(12) C(11) C(12) H(12) C(13) C(12) H(12) C(14) C(13) C(12) C(14) C(13) H(13A) C(12) C(13) H(13A) C(14) C(13) H(13B) C(12) C(13) H(13B) H(13A) C(13) H(13B) C(15) C(14) C(13) C(15) C(14) H(14A) C(13) C(14) H(14A) C(15) C(14) H(14B) C(13) C(14) H(14B) H(14A) C(14) H(14B) N(1) C(15) C(14) N(1) C(15) C(16) C(14) C(15) C(16) N(1) C(15) H(15) C(14) C(15) H(15) C(16) C(15) H(15) C(10) C(16) C(15) C(10) C(16) H(16A) C(15) C(16) H(16A) C(10) C(16) H(16B) C(15) C(16) H(16B)
119.8 119.8 119.03(16) 120.19(17) 120.78(17) 110.17(14) 109.99(15) 112.03(13) 108.2 108.2 108.2 112.15(14) 109.2 109.2 109.2 109.2 107.9 125.09(17) 125.11(17) 109.75(15) 107.68(14) 109.15(14) 112.66(14) 109.1 109.1 109.1 113.29(15) 108.9 108.9 108.9 108.9 107.7 106.75(14) 104.69(15) 113.14(14) 110.7 110.7 110.7 103.88(14) 111.0 111.0 111.0 111.0 109.0 104.08(15) 110.9 110.9 110.9 110.9 109.0 104.86(14) 106.50(14) 113.47(15) 110.6 110.6 110.6 113.84(15) 108.8 108.8 108.8 108.8
H(16A) C(16) H(16B) N(1) C(17) H(17A) N(1) C(17) H(17B) H(17A) C(17) H(17B) N(1) C(17) H(17C) H(17A) C(17) H(17C) H(17B) C(17) H(17C) C(9) O(3) C(10) C(8) O(2) H(2) C(12) N(1) C(17) C(12) N(1) C(15) C(17) N(1) C(15)
107.7 109.5 109.5 109.5 109.5 109.5 109.5 117.23(14) 109.5 111.58(15) 101.51(13) 111.81(15)
intensity during the course of data collection. Data were processed using DENZO (Otwinowski and Minor, 1997), correcting for Lorentz and polarisation effects, and absorption effects were applied using the program SORTAV (Blessing, 1997). The structure was solved by Patterson Search (Beurger, 1959) using the program PATTSEE (Egert and Sheldrick, 1985). The search model (Fig. A.1) comprised two rigid fragments: ring A (atoms 1–6) and atoms C(7), C(9), C(8), O(1) and O(3). The program determines and refines one linkage torsion angle, in this case about bond C(6) C(7). This model was searched for in the Patterson function, and possible solutions were refined to give a starting model for the structure, which was then expanded by Fourier methods and refined by full-matrix least-squares. Co-ordinates for the starting model fragments were built in the program Chem-X (Oxford Molecular, 1999). Location of the search model takes place in two steps: (1) rotation, where the best solution for the orienTable A.3 ˚ 2 ) for atropine Anisotropic displacement parameters (103 × A C(1) C(2) C(3) C(4) C(5) C(6) C(7) C(8) C(9) C(10) C(11) C(12) C(13) C(14) C(15) C(16) C(17) O(3) O(1) O(2) N(1)
U11
U22
U33
U23
U13
U12
29(1) 31(1) 30(1) 37(1) 30(1) 23(1) 31(1) 29(1) 30(1) 41(1) 31(1) 35(1) 35(1) 42(1) 28(1) 31(1) 66(1) 44(1) 55(1) 40(1) 37(1)
34(1) 47(1) 35(1) 30(1) 30(1) 23(1) 21(1) 24(1) 23(1) 19(1) 26(1) 25(1) 36(1) 30(1) 27(1) 24(1) 29(1) 22(1) 37(1) 38(1) 20(1)
45(1) 56(1) 76(2) 57(1) 42(1) 39(1) 28(1) 38(1) 27(1) 25(1) 27(1) 31(1) 29(1) 27(1) 36(1) 34(1) 36(1) 29(1) 32(1) 34(1) 30(1)
8(1) 21(1) 21(1) 3(1) 6(1) 5(1) −2(1) 2(1) 0(1) 3(1) 3(1) 2(1) −1(1) 4(1) 5(1) 4(1) −6(1) 2(1) −3(1) 2(1) −1(1)
9(1) 15(1) −2(1) −10(1) −2(1) −4(1) −2(1) 1(1) −1(1) −8(1) 2(1) −5(1) −7(1) 2(1) 7(1) −6(1) 6(1) 0(1) 0(1) −3(1) 1(1)
8(1) 9(1) −5(1) −4(1) 1(1) 7(1) 3(1) −3(1) −4(1) 4(1) 1(1) −7(1) −1(1) 1(1) 1(1) −3(1) 5(1) 10(1) 16(1) −15(1) 1(1)
The anisotropic displacement factor −2π2 [h2 a∗2 U 11 + · · · + 2hka∗ b∗ U 12 ].
exponent
takes
the
form:
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385 Table A.4 Hydrogen coordinates (×104 ) and isotropic displacement parameters ˚ 2 ) for atropine (103 × A H(1A) H(2A) H(3) H(4A) H(5A) H(7) H(8A) H(8B) H(10) H(11A) H(11B) H(12) H(13A) H(13B) H(14A) H(14B) H(15) H(16A) H(16B) H(17A) H(17B) H(17C) H(2)
x
y
z
U(eq)
3302 2691 2543 3007 3603 3940 4618 4432 3063 3850 3385 3600 3729 3582 2892 2739 2133 2370 2290 2269 2876 2769 4938
197 −259 −633 −562 −104 507 114 174 1069 1391 1600 1881 1292 1626 1077 1416 1485 984 1298 2102 2247 1980 611
12694 13108 10392 7293 6846 10826 9694 7350 6306 7807 6621 9755 11625 12938 12381 13643 10983 9082 7555 11224 10835 12654 9755
43 54 56 50 41 32 36 36 34 33 33 36 40 40 39 39 36 35 35 66 66 66 56
tation of the search model is output, and (2) translation of the rotated model to determine its position in the unit cell. The expanded structure containing all non-H atoms was refined using SHELX-97 (Sheldrick, 1997) implemented in the WinGX system of programs (Farrugia, 1999). Non-hydrogen atoms were refined anisotropically by full-matrix least-square methods. Hydrogens were added geometrically and refined in riding mode with isotropic temperature factors. Geometrical calculations were made with the programs PARST and PLATON (Spek, 1990) as implemented in WinGX. WinGX was also used to prepare publication material, including figures (ORTEP, Barnes, 1997 and POV-Ray, 2004) and tables. All diagrams of the molecules given in the paper are of the biologically active enantiomer (this is the enantiomer referred
383
Table A.5 Torsion angles (◦ ) for atropine C(6) C(1) C(2) C(3) C(1) C(2) C(3) C(4) C(2) C(3) C(4) C(5) C(3) C(4) C(5) C(6) C(2) C(1) C(6) C(5) C(2) C(1) C(6) C(7) C(4) C(5) C(6) C(1) C(4) C(5) C(6) C(7) C(1) C(6) C(7) C(9) C(5) C(6) C(7) C(9) C(1) C(6) C(7) C(8) C(5) C(6) C(7) C(8) C(9) C(7) C(8) O(2) C(6) C(7) C(8) O(2) C(6) C(7) C(9) O(1) C(8) C(7) C(9) O(1) C(6) C(7) C(9) O(3) C(8) C(7) C(9) O(3) O(3) C(10) C(11) C(12) C(16) C(10) C(11) C(12) C(10) C(11) C(12) N(1) C(10) C(11) C(12) C(13) N(1) C(12) C(13) C(14) C(11) C(12) C(13) C(14) C(12) C(13) C(14) C(15) C(13) C(14) C(15) N(1) C(13) C(14) C(15) C(16) O(3) C(10) C(16) C(15) C(11) C(10) C(16) C(15) N(1) C(15) C(16) C(10) C(14) C(15) C(16) C(10) O(1) C(9) O(3) C(10) C(7) C(9) O(3) C(10) C(16) C(10) O(3) C(9) C(11) C(10) O(3) C(9) C(11) C(12) N(1) C(17) C(13) C(12) N(1) C(17) C(11) C(12) N(1) C(15) C(13) C(12) N(1) C(15) C(14) C(15) N(1) C(12) C(16) C(15) N(1) C(12) C(14) C(15) N(1) C(17) C(16) C(15) N(1) C(17)
0.4(3) 0.5(3) −1.1(3) 0.8(3) −0.7(3) 178.97(16) 0.1(3) −179.57(16) −110.94(18) 68.7(2) 126.28(18) −54.0(2) 60.3(2) −176.83(15) −86.0(2) 38.0(2) 91.53(17) −144.50(15) −83.27(17) 36.3(2) −57.62(18) 57.00(19) 25.77(18) −90.08(17) 1.44(18) −28.21(18) 87.61(18) 84.44(17) −36.0(2) 56.77(18) −58.1(2) −3.7(3) 178.78(13) 139.48(15) −97.92(17) −164.07(15) 75.73(17) 76.70(16) −43.50(17) 44.70(17) −75.85(17) −74.36(18) 165.09(16)
Symmetry transformations were used to generate equivalent atoms.
to as (−)atropine). Anisotropic displacement parameters are in Table A.3. A.1. Molecular geometry
Fig. A.1. Search model used in the X-ray analysis comprising fragments 1 and 2 and rotation about bond C(6) C(7).
Bond lengths in the structure are determined approxi˚ and bond angles to ±0.1◦ . Both bond lengths mately ±0.02 A and bond angles, Table A.2, in the structure compare well with those found in organic compounds (Ladd and Palmer, 2003) and exhibit no unusual values. The linkage chain comprising atoms C(7), C(9), O(3) and C(10) is fully extended with a torsion angle of 178.8(1)◦ . The methyl-azabicyclooctyl ester head of the molecule contains three saturated rings: a six-membered piperidino ring B1 B2, a seven membered saturated carbon ring B1 B3 and a 5-membered pyrrolinyl
384
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385
not observed in the hydrobromide structure or in the present structure of the native molecule.
References
Fig. A.2. Ring conformations.
ring B2 B3. The conformations of these three rings are illustrated in Fig. A.2. Torsion angles are in Table A.5. Ring B1 B2 (Fig. A.2(a)) has a sofa conformation with a pseudo m-plane along N(1) · · · C(10). The asymmetry parameter (Duax and Norton, 1995) CsN(1)···C(10) = 0.7(2)◦ indicating a high degree of mirror symmetry. The ring is steeper at the N(1) end with torsion angles of around 90◦ . Ring B1 B3 (Fig. A.2(b)) has a boat-like conformation with a pseudo m-plane along C(10) to the bisector of bond C(13) C(14). The asymmetry parameter CsC(10) = 1.6(3)◦ indicating also a high degree of mirror symmetry. The sides of boat are steeper at the C(13) and C(14) end again with torsion angles of around 90◦ . Ring B2 B3 (Fig. A.2(c)) has a sofa conformation with a pseudo m-plane along N(1) to the bisector of bond C(13) C(14). The asymmetry parameter CsN(1) = 1.9(4)◦ again indicates a high degree of mirror symmetry. Atoms C(12), C(13), C(14) and C(15) are approximately coplanar with N(1) out of the plane. The three pseudo m-planes in the three rings are approximately coincident and the composite saturated ring system thus has a pseudo m-plane running from C(10) to N(1) and the bisector of bond C(13) C(14). The methyl group C(17)H3 is approximately equatorial to this ring system. A.2. Crystal packing and hydrogen bonding The crystal structure is held together by strong hydrogen bonds formed through O(2) H(2) · · · N(1)* linkages where * indicates the symmetry operation [x + 1/4, −y + 1/4, z + 1/4]. These H-bonds have the geometry O(2) · · · ˚ and O(2) H(2) · · · N(1) = 172.5(3)◦ . BeN(1)* = 2.740(6) A cause of the high symmetry in space group Fdd2 there are many equivalent H-bonds throughout the structure. A previous study of the X-ray structure of atropine has been published by Pauling and Petcher in 1970 but no details of the space group or co-ordinates were given. The crystal structure of atropine hydrobromide was reported by Pauling and Petcher (1970) although no corresponding data have been deposited. In this paper, the molecule is shown in the correct absolute configuration corresponding to the more potent enantiomer and to that derived and used here. These authors propose that the active form of the molecule involves an intermolecular H-bond O(2)H · · · O(1) although this is
Archer, E., Maigret, B., Escrieut, C., Pradayrol, L., Fourmy, D., 2003. Rhodopsin crystal: new template yielding realistic models of Gprotein-coupled receptors? Trends Pharmacol. Sci. 24, 36–40. Bakker, R.A., Wieland, K., Timmerman, H., Leurs, R., 2000. Constitutive activity of the histamine H-1 receptor reveals inverse agonism of histamine H-1 receptor antagonists. Eur. J. Pharmacol. 387, R5–R7. Bayly, C.I., Cieplak, P., Cornell, W.D., Kollman, P.A.A., 1993. A wellbehaved electrostatic potential based method using charge restraints for deriving atomic charges—the RESP model. J. Phys. Chem. 97, 10269–10280. Chemwindow, 1999–2003. Bio-Rad Laboratories, Inc., Informatics Division. Chen, G., Jayawickreme, C., Way, J., Armour, S., Queen, K., Watson, C., Ignar, D., Chen, W., Kenakin, T., 1999. Constitutive receptor systems for drug discovery. J. Pharmacol. Toxicol. 42, 199–206. CVFF, 1990. CVFF forcefield file shipped with Discover 2.6.0/InsightII 1.1.0/Insight 2.6, September 1990. Biosym Technologies, Inc. Dong, X., Han, S.K., Zylka, M.J., Simon, M.I., Anderson, D.J., 2001. A diverse family of GPCRs expressed in specific subsets of nociceptive sensory neurons. Cell 106, 619–632. Flower, D.R., 1999. Modelling G-protein-coupled receptors for drug design. Biochim. Biophys. Acta 1422, 207–234. Jakubik, J., Bacakova, L., El-Fakahany, E.E., Tucek, S., 1995. Constitutive activity of the M(1)M(4) subtypes of muscarinic receptors in transfected CHO cells and of muscarinic receptors in the heart cells revealed by negative antagonists. FEBS Lett. 377, 275–279. Lu, Z.-L., Saldanha, J.W., Hulme, E.C., 2001. Transmembrane domains 4 and 7 of the M-1 muscarinic acetylcholine receptor are critical for ligand binding and the receptor activation switch. J. Biol. Chem. 276, 34098–34104. Morris, G.M., Goodsell, D.S., Halliday, R.S., Huey, R., Hart, W.E., Belew, R.K., Olson, A.J., 1998. Automated docking using a Lamarckian genetic algorithm and an empirical binding free energy function. J. Comp. Chem. 19, 1639–1662. Palczewski, K., Kumasaka, T., Hori, T., Behnke, C.A., Motoshima, H., Fox, B.A., Le Trong, I., Teller, D.C., Okada, T., Stenkamp, R.E., Yamamoto, M., Miyano, M., 2000. Crystal structure of rhodopsin: a G-protein-coupled receptor. Science 289, 739–745. Parma, J., Duprez, L., Vansande, J., Cochaux, P., Gervy, C., Mockel, J., Dumont, J., Vassart, G., 1993. Somatic mutations in the thyrotropin receptor gene cause hyperfunctioning thyroid adenomas. Nature 365, 649–651. Parnot, C., Miserey-Lenkei, S., Bardin, S., Corvol, P., Clauser, E., 2002. Lessons from constitutively active mutants of G-protein-coupled receptors. Trends Endocrinol. Metab. 13, 336–343. Peitsch, M.C., Wilkins, M.R., Tonella, L., Sanchez, J.C., Appel, R.D., Hochstrasser, D.F., 1997. Large-scale protein modelling and integration with the SWISS-PROT and SWISS-2DPAGE databases: the example of Escherichia coli. Electrophoresis 18, 498–501. Robinson, P.R., Cohen, G.B., Zukovsky, E.A., Oprian, D.D., 1992. Constitutively active mutants of rhodopsin. Neuron 9, 719–725. Schmidt, M.W., Baldridge, K.K., Boatz, J.A., Elbert, S.T., Gordon, M.S., Jensen, J.J., Koseki, S., Matsunaga, N., Nguyen, K.A., Su, S., Windus, T.L., Dupuis, M., Montgomery, J.A., 1993. General atomic and molecular electronic-structure system. J. Comput. Chem. 14, 1347–1363. Shacham, S., Topf, M., Avisar, N., Glaser, F., Marantz, Y., Bar-Haim, S., Noiman, S., Naor, Z., Becker, O.M., 2001. Modeling the 3D structure of GPCRs from sequence. Med. Res. Rev. 21, 472–483. Spalding, T.A., Burnstein, E.S., 2001. Constitutively active muscarinic receptors. Life Sci. 68, 2511–2516.
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385 Vaidehi, N., Floriano, W.B., Trabino, R., Hall, S.E., Freddolino, P., Choi, E.J., Zamanakos, G., Goddard III, W.A., 2002. Prediction of structure and function of G-protein-coupled receptors. Proc. Natl. Acad. Sci. 99, 12622–12627. Ward, S.D.C., Curtis, C.A.M., Hulme, E.C., 1999. Scanning mutagenesis identifies amino acid side chains in transmembrane domain 5 of the M-1 muscarinic receptor that participate in binding the acetyl methyl group of acetylcholine. Mol. Pharm. 56, 1031– 1041. Weiner, P.K., Kollman, P.A., 1981. AMBER-assisted model-building with energy refinement—a general program for modeling molecules and their interactions. J. Comput. Chem. 2, 287–303. Weinstein, L.S., Shenker, A., 1993. G-protein mutations in human-disease. Clin. Biochem. 26, 333–338. Yamashita, A., Endo, M., Higashi, T., Nakatsu, N., Yamada, Y., Oda, J., Kato, H., 2003. Capturing enzyme structure prior to reaction initiation: tropinone reductase-II-substrate. Biochemistry 42, 5566– 5573.
Further reading
Barnes, C.L., 1997. ORTEP-3 for Windows – a version of ORTEP-III with a Graphical User Interface (GUI). J Appl. Cryst. 30, 568–568 [Based on ORTEP-III (v 1.0.3) by C.K. Johnson and M.N. Burnett].
385
Beurger, M.J., 1959. Vector Space. Wiley, New York. Blessing, R.H., 1997. Outlier treatment in data merging. J. Appl. Cryst. 30, 421–426. Cosier, J., Glazer, A.M., 1986. A nitrogen-gas-stream cryostat for general X-ray diffraction studies. J. Appl. Cryst. 19, 105–107. Duax, W.L., Norton, D.A., 1995. Atlas of Steroid Structure. Plenum Press, New York. Egert, E., Sheldrick, G.M., 1985. Search for a fragment of known geometry by integrated Patterson and direct methods. Acta Crystallogr. A 41, 262–268. Farrugia, L.J., 1999. WinGX. J. Appl. Cryst. 32, 837–838. Hooft, R., 1998. Nonius, B.V., COLLECT: X-ray data collection and processing software user interface. Ladd, M.F.C., Palmer, R.A., 2003. Structure Determination by X-ray Crystallography, 4th ed. Plenum Press, New York, Tables 7.24, 7.25. Otwinowski, Z., Minor, W., 1997. Methods in enzymology Part A. In: Carter Jr., C.W., Sweet, R.M. (Eds.), Macromolecular Crystallography, vol. 276. Academic Press, New York, pp. 307–326. Oxford Molecular (1999). http://www.oxmol.com:Chem-X. Pauling, P.J., Petcher, T.J., 1970. Nature 228, 673–674. POV-Ray, 2004. POV-Ray rendering engine for Windows. Version 3.1 g.watcom.win32 [Pentium II Optimized]. Sheldrick, G.M., 1997. SHELX-97: Program for Refinement of Crystal Structures. University of Gottingen. Spek, A.L., 1990. BYPASS—an effective method for the refinement of crystal-structures containing disordered solvent regions. Acta Crystallogr. A 46, 194–201.
Antagonist binding in the rat muscarinic receptor A study by docking and X-ray crystallography Anna C. Tanczosa , Rex A. Palmerb , Brian S. Potterb , Jos´e W. Saldanhac , Brendan J. Howlina,∗ a
Department of Chemistry, School of Biomedical and Molecular Sciences, University of Surrey, Guildford, Surrey GU2 7XH, UK b School of Crystallography, Birkbeck College, University of London, Malet Street, London WC1E 7HX, UK c National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 1AA, UK
Received 9 September 2004; received in revised form 29 September 2004; accepted 29 September 2004
Abstract A series of agonists to the rat muscarinic receptor have been docked computationally to the active site of a homology model of rat M1 muscarinic receptor. The agonists were modelled on the X-ray crystal structure of atropine, which is reported here and the docking studies are shown to reproduce correctly the order of experimental binding affinities for the agonists as well as indicate where there appear to be inconsistencies in the experimental data. The crystal and molecular structure of atropine (tropine tropate; ␣-[hydroxymethyl]benzeneacetic acid 8-methyl[3.2.1]oct-3-yl ester C17 H23 NO3 ) has been determined by X-ray crystallography using an automated Patterson search method, and refined by full-matrix least-squares to a final R of 0.0452 for 2701 independent observed reflections and 192 parameters using Mo K␣ ˚ at 150 K. The compound crystallises in space group Fdd2 with Z = 16 molecules per unit cell. radiation, λ = 0.71073 A © 2004 Published by Elsevier Ltd. Keywords: Crystal structure; Receptor binding; Docking of atropine; Docking; Homology modelling
1. Introduction G-protein coupled receptors (GPCRs) are the largest group of cell surface receptors (Dong et al., 2001) and as such are targets for a number of therapy discovery programs. The signalling is a two-way process: (1) the binding of a ligand facilitates the binding of a G-protein on the inside of the cell and (2) it then elicits a second messenger response. Conversely, if a G-protein binds prior to ligand binding, the affinity for the ligand is altered. Many different ligands bind to GPCRs. These range from a single photon, in the case of the opsins, through to small molecules, which bind in a channel between the seven helices and progress in size to proteins, which bind to the extracellular loops and do not enter into the channel ∗
Corresponding author. Tel.: +44 1483 686834; fax: +44 1483 686851. E-mail address: [email protected] (B.J. Howlin).
1476-9271/$ – see front matter © 2004 Published by Elsevier Ltd. doi:10.1016/j.compbiolchem.2004.09.009
between the seven helices. There are thought to be at least two different conformations of GPCRs: an inactive (R) and active (R*) conformation which exist in equilibrium. It is possible that the various ligands either cause a difference in the relative amount of time spent in or the accessibility of these two states. In general, only four of the helices are involved directly with binding the ligand, namely 3, 5, 6 and 7 (Flower, 1999) although there are often other interactions. As a result of the difficulty of crystallising membrane proteins, there is only one set of atomic coordinates for this family of receptors, that of bovine rhodopsin (Palczewski et al., 2000). This has resulted in a rapidly increasing number of models being produced based on this one structure and refined using the vast databases of biological data available, such as mutagenesis data (Shacham et al., 2001). A recent paper has discussed homology models for four of the classes of GPCRs and concluded that they can be used for drug design studies
376
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385
(Vaidehi et al., 2002), although Archer et al. (2003) have concluded that care is needed in the construction of homology models and experimental data is useful to correctly guide these studies. 1.1. Agonists, antagonists and inverse agonists During its random sampling of conformational space, a receptor will at times adopt the active conformation without the presence of an agonist. The existence of the receptor in this state–activated without an agonist–normally occurs so infrequently that it is considered to be zero for most purposes. However, there are certain receptors, such as rat acetylcholine muscarinic M1 receptor (m1AChR) (Spalding and Burnstein, 2001; Jakubik et al., 1995), for which constitutive activity is observed when the receptor is present in larger concentrations than normal in a transfected system. That is to say that they attain the active state a significant number of times without the presence of an agonist (Parnot et al., 2002; Chen et al., 1999). The response to an agonist remains increased activation and a neutral antagonist elicits no response. However, some antagonists inhibit the constitutive activation; these are called inverse agonists. Although this phenomenon was initially evident only in systems with a high concentration of receptors it has been shown to exist in natural systems (Bakker et al., 2000). It has also been shown that many human diseases are a consequence of constitutive activity of receptors. This arises as a consequence of mutations in the receptor or increased expression. These diseases include hyperthyroidism (TSH receptor) (Parma et al., 1993), retinal degeneration (rhodopsin) (Robinson et al., 1992) and hyper-pigmentation (LH receptor) (Weinstein and Shenker, 1993). Inverse agonists as opposed to neutral antagonists are required to prevent the activation of these mutant receptors. Muscarinic receptors are G-protein coupled receptors found in the parasympathetic nervous system and extensively in the brain, particularly in the cerebral cortex and the hippocampus. Acetylcholine is the endogenous agonist for this receptor. One of the features of the binding of many of these ligands to muscarinic receptors is the quaternary ammonium moiety. The presence of the N-methyl group changes the activity preventing these ligands from crossing the blood–brain barrier and improves the binding affinity of a given ligand. A proposed mechanism for activation of the receptor (Lu et al., 2001) begins with the enclosure of this group within an aromatic cage that triggers an alteration in the hydrogen bonding network leading to helical movement and thus to the active state of the receptor. Modelling studies are important for examination of the mechanistic context of the data that shed light on these interactions. Here we look at the initial interaction of m1AChR with various atropine analogues and examine the role of these compounds, some of which are inverse agonists. The X-ray analysis of the inverse agonist, atropine (Fig. 1), has been carried out in order to provide an accurate model of the ligand for use in the receptor modelling and binding studies.
Fig. 1. Atropine chemical formula: atom numbering used in the X-ray analysis; ring labels are also shown. Drawn with Chemwindow (1999–2003).
2. Experimental 2.1. Model of rat muscarinic receptor The model used has been described previously (Lu et al., 2001). 2.2. X-ray structural analysis of atropine The discussion of the X-ray determination is collected in Appendix A. The X-ray data for the atropine crystal are shown in Table 1. The rest of the X-ray data appear in the tables in Appendix A. Interestingly, the ring conformations of the tropine ring are very similar to published work on Table 1 Crystal data and structure refinement for atropine Identification code
Atropine
Empirical formula Formula weight Temperature (K) ˚ Wavelength (A) Crystal system Space group Unit cell dimensions
C17 H23 NO3 289.36 100(2) 0.71073 Orthorhombic Fdd2 ˚ b = 39.538(8) A, ˚ a = 24.291(5) A, ˚ α = 90◦ , β = 90◦ , c = 6.4727(13) A, γ = 90◦ 6217(2) 16 1.237 0.084 2496 0.20 × 0.15 × 0.35 3.30 to 25.02 −28 ⇐ h ⇐ 28, −46 ⇐ k ⇐ 46, −7 ⇐ l ⇐ 7 14,353 2701 [R(int) = 0.0820] 99.7% Full-matrix least-squares on F2 2701/1/192 0.831 R1 = 0.0452, wR2 = 0.1206 R1 = 0.0558, wR2 = 0.1294 −0.1(16) 0.00108(14) ˚ −3 0.262 and −0.225 e A
˚ 3) Volume (A Z Density (calculated) ( Mg/m3 ) Absorption coefficient ( mm−1 ) F(0 0 0) Crystal size (mm3 ) Theta range for data collection (◦ ) Index ranges Reflections collected Independent reflections Completeness to theta = 25.02◦ Refinement method Data/restraints/parameters Goodness-of-fit on F2 Final R indices [I > 2sigma(I)] R indices (all data) Absolute structure parameter Extinction coefficient Largest diff. peak and hole
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385
the X-ray structures of two atropine-type alkaloid biosynthetic precursors, published by Yamashita, Endo, Higashi, Nakatsu, Yamada, Oda and Kato in the structure of tropinone reductase-II complexed with NADP+ and pseudotropine at ˚ resolution (Yamashita et al., 2003). The B1 B2 ring 1.9 A conformation (Fig. 1) is a distorted boat and the B1 B3 rings ˚ are in a sofa conformation. The resolution of our data is 0.7 A, so individual bonds can be resolved whereas the studies mentioned above are less than atomic resolution, so it is particularly gratifying that they are in agreement. 2.3. Docking of ligands to the receptor model Docking was carried out with the program AUTODOCK (Morris et al., 1998). Molecular models of the atropine analogues (Table 2) were modelled in Viewer Pro 4.2 (Accelerys, Inc.) using the atropine X-ray structure as a template, with replacement and addition of groups performed manually in the program. All structures were subjected to the CLEAN routine in ViewerPro 4.2 to remove unlikely geometrical features. These ligand models were transferred to AUTODOCK TOOLS (http://www.scripps.edu/∼sanner/python/adt) to prepare them for docking. Energy minimisation of the X-ray structure of atropine in Viewer Pro 4.2 showed no appreciable difference in conformation, indicating that the atropine conformation had not been modified by crystal packing efTable 2 Structures of ligands used in this study
377
fects and the structure could be used as solved. Therefore, all ligand structures were only subjected to the CLEAN routine in ViewerPro 4.2, so that there would be little deviation from the starting atropine model. AUTODOCK TOOLS was used to identify the torsion angles in the ligands, add the solvent model and assign partial atomic charges. The ligand models were docked to the model of rat muscarinic receptor using the default settings of AUTODOCK for the Lamarkian genetic algorithm. Hence, a flexible ligand was docked to a rigid protein. The fitness function that was optimised by the genetic algorithm was a function of the conformation of the ligand (defined by its flexibility) and its interaction with the protein (by non-bonded and charge interactions). The results of the docking studies were the 10 best docked ligands quantified in terms of the calculated free energy of binding. This process is essentially a flexible ligand docking to a rigid protein and ideally the docking should be carried out with a flexible ligand and a flexible protein, so we have approximated this by also carrying out simulated annealing and molecular dynamics of the receptor model. 2.4. Mutation studies Experimental data are also available on two mutants of m1AchR (Ward et al., 1999). These are Tyr381Phe and Tyr381Ala. We have simulated these experiments by making the relevant point mutations in DEEPVIEW (Peitsch et al., 1997), energy minimising the resulting mutant using the AMBER suite of programs (Weiner and Kollman, 1981) (running on Columbus, the Computational Chemistry Facility at the Rutherford Appleton Laboratory), and running the docking simulations for both cases. The calculated free energies of binding have then been correlated with the experimental values of pIC50 (Ward et al., 1999). The pIC50 value is the negative logarithm of the concentration of ligand that inhibits 50% of the activity. The experimental pIC50 values are reported to two significant figures; therefore for comparison purposes we have quoted our calculated binding energies to the same number of significant figures. It must be borne in mind, however, that this does not reflect on the reproducibility or accuracy of the calculated results. Repeating a docking run with the same models invariably gives the same result, so no estimate of error is possible with the present calculations. 2.5. Simulated annealing and molecular dynamics of the receptor Simulated annealing and molecular dynamics of the receptor were performed using the AMBER suite of programs. Simulated annealing was carried out for 5000 fs at a simulated temperature of 1000 ◦ C. After equilibration at 1000 ◦ C for 1000 fs, the temperature remained constant for a further 1500 fs and was then cooled in two stages, fast for 1000 fs and slow for the remaining 1500 fs with a distance-dependent
378
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385
Table 3 Ether linkage parameters used in AMBER Bond parameters Bond
˚ R0 (A)
˚ 2) K (kcal/mol/A
OS C OS CT
1.31 1.41
350 320
Angle
θ 0 (◦ )
K (kcal/mol/◦2 )
C OS CT OS CT CT OS C CT OS C O OS CT HC N CT HC C CT CA
109.5 109.5 120 123 109.5 109.5 109.5
60 70 123 145 57 49 49
Angle parameters
Fig. 2. ORTEP/POV-Ray view of the stereochemistry of the atropine molecule in the solid state.
Torsion parameters Torsion angle
k (kcal/mol/◦2 )
Periodicity, n
θ (◦ )
CT OS C CT O C OS CT
4.5 4.5
2 2
180 180
Atoms types given are AMBER atom types.
dielectric. No explicit solvent atoms or lipids were included in the calculation. The alpha carbon atoms of the protein were constrained with a force constant of 1 during the simulated annealing runs to prevent unfolding of the protein; essentially we are simulating different side chain conformations of the same active state by constraining the helices. This process was repeated 10 times, each time using the output of the previous run, both with and without the ligand. The docking studies were repeated with these structures. In order to examine unconstrained movement, molecular dynamics was performed, both with and without atropine present in the active site of the receptor. AMBER was parameterised for atropine by comparisons with the CVFF (BIOSYM Technologies Inc., 1990) force field for the missing parameters concerning the ether linkage. A correlation table was drawn up with known AMBER parameters versus known CVFF parameters. A scaling factor was applied to convert the known CVFF parameters to the known AMBER parameters and this factor was used to convert the known CVFF parameters for the ether linkage to the respective AMBER parameters. This process must be considered an approximate method but care was taken to ensure that the new parameters would not unbalance the force field. Parameters for the ether linkage used in AMBER are given in Table 3. RESP (Bayly et al., 1993) was used to fit the charges from an electrostatic potential grid produced from a single point calculation in GAMESS (Schmidt et al., 1993) using the 631G* basis set. A more complete study would involve the derivation of the respective parameters by molecular orbital calculations on suitable small molecules containing the respective linkage. However, our aim in this work was to use the atropine as a ‘bung’ to keep the shape of the active site of the receptor.
3. Results and discussion The refined atomic co-ordinates and isotropic parameters for non-hydrogen atoms of the atropine are given in Table A.1. Bond distances and bond angles are in Table A.2. An ORTEP/POV-Ray view of the molecular conformation of atropine is shown in Fig. 2. 3.1. Docking studies (native protein) The docked conformations of atropine are located in a site between transmembrane helices TM3, TM5, TM6 and TM7 (Fig. 3). The figure shows the shape of the binding site in the receptor. This is composed of a van der Waal’s surface (coloured dark blue) and showing the volume occupied by the amino acids in the helices. The backbone of TM3 is shown as a helical ribbon coloured yellow and the amino acid side chains of this helix which are shown in wire frame format are also coloured yellow. TM5 is coloured red, TM6 is green and TM7 is pink. An electrostatic potential is superimposed on the left-hand part of the van der Waal’s surface showing the
Fig. 3. Atropine docked into the binding site of the rat muscarinic receptor from the homology model.
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385
distribution of charges in the binding site. The red patches indicate areas of negative charge, which serve to attract the positively charged end of the ligand, which is shown in CPK format in the middle of the binding site. CPK format shows carbon atoms as grey spheres, nitrogen atoms as blue spheres and oxygen as red spheres. The ‘walls’ of this binding pocket are made up of the amino acid residues TM3: Asp105, Tyr106, Ser109 and Asn110; TM5: Thr192, Ala196, Phe197 and Tyr198; TM6: Phe374, Trp378, Tyr381, Asn382 and Leu386; TM7: Tyr404, Cys407 and Tyr408. These are all residues that have been implicated in the binding of inverse agonists (Lu et al., 2001; Ward et al., 1999). Our studies have indicated that there is a possibility of a ring-stacking interaction between the aromatic ring of the antagonists and Tyr106 and Phe197, but unfortunately such interactions are not modelled explicitly in the docking software. Molecular orbital calculations need to be used to study ring-stacking interactions and this will be an area of future study. All of the remaining atropine analogues are found to bind to this site. Acetylcholine, the natural agonist, displays a slightly different binding behaviour, as it is located deeper within the negatively charged area of the binding pocket than the atropine. The smaller size of acetylcholine, when compared to those of the inverse agonists, without the bulky aromatic tail offers a reason for this behaviour. The above argument suggests a rationale for the observation that both the Gbinding and pIC50 are lower for acetylcholine than for atropine and its analogues, it does not fit as snugly into the site as the larger antagonists and is thus not as energetically stabilised. This may be what distinguishes the agonists from the inverse agonists, the agonists allowing the receptor freedom of movement and thus a conformational change. Both agonist and inverse agonists are found to bind with the ammonium head group buried inside the receptor in the vicinity of Asp105, which contributes a negative charge to this area of the binding pocket. A plot of the pIC50 versus minimum free energy of binding is given as Fig. 4. A table of the values is given as Table 4. It is important to consider the correct concentrations of the ligands as this will affect the pIC50 values used. Some of the ligands were used as racemic mixtures of both isomers, so only half of the concen-
379
Table 4 Table of molecule docked with their calculated average and minimium G binding energies in kJ/mol and pIC50 values in mol Molecule
Ave. DG binding
Min. DG binding
pIC50
Benzilyltropine Benztropine N-Methylatropine Atropine N-Methylhomatropine dl-Homatropine N-Methylacetyltropine Acetylcholine Diphenylacetyltropine Phenylacetyltropine
−58.93 −55.64 −49.62 −50.27 −46.71 −46.92 −33.99 −29.30 −58.39 −48.14
−59.79 −56.27 −50.84 −51.67 −47.82 −47.15 −34.31 −29.66 −59.16 −48.28
10.05 9.75 9.56 9.19 7.61 7.37 5.74 4.78 8.42 6.94
pIC50 values are taken from reference 20 and those from racemic mixtures have been corrected by the method outlined.
tration is active. There are two ways to correct for this, either by decreasing the concentration of the non-racemic ligands or increasing the concentrations of the racemic ligands. We have tried both methods and found no substantial difference in the results. Therefore, for the purposes of this paper, those ligands with no chiral centres have had their pIC50 values corrected by only considering half of the concentration of ligand in order to normalise the data with those studied as a 50:50 racemic mix. These plots show the excellent rank order of the free energy of binding with the pIC50 values. The correlation between average free energy of binding and pIC50 is over 90% and that of minimum free energy of binding is 90%. This good correlation with experimental data shows that the approach used in this work is valid. It is therefore possible to produce good correlations with experimental data using a model of the receptor and models of the ligands. The use of X-ray crystallography here is important as it gives confidence in the three-dimensional structure of the atropine produced by the molecular modelling. It is also interesting to note that the last two ligands in Table 4 do not fit with the correlation produced by the others. On further investigation, it was found that these two compounds had been obtained from a different source than the rest of the ligands studied and were used in the experimental studies without further purification. For this reason, we have excluded these compounds from the study. 3.2. Docking studies (point mutations to the protein)
Fig. 4. Plot of pIC50 vs. minimum free energy of binding.
Two mutants of the protein have been studied experimentally: Tyr381Ala and Tyr381Phe; so it was of interest to see if we could also reproduce the docking to these mutants (Ward et al., 1999). The point mutations were made in DEEPVIEW using the mutate facility and the resulting structures energy minimised in AMBER. Docking of the same ligands was carried out to both mutants and the results again correlated with the experimental pIC50 data. The average of the best docking results for the Ala mutant had an r2 of 93% (Table 5) and that of the Phe mutant was slightly better at 95%. In the first mutant, both an aromatic ring and a potential hydrogen-
380
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385
Table 5 Table of molecule docked to the point mutations with their calculated average and minimium G binding energies in kJ/mol and pIC50 values in mol Ave. DG (kJ/mol)
Min. DG (kJ/mol)
Ave. best (kJ/mol)
pIC50
Tyr381Ala Benzilyltropine Benztropine N-Methylatropine Atropine N-Methylhomatropine dl-Homatropine N-Methylacetyltropine acetylcholine
−46.18 −44.37 −36.96 −37.36 −36.02 −36.15 −24.79
−47.40 −47.82 −38.20 −39.58 −38.66 −37.66 −25.90
−46.39 −44.56 −36.69 −37.15 −35.73 −36.32 −24.85
9.77 8.92 7.17 7.12 6.26 6.38 4.79
Tyr381Phe Benzilyltropine Benztropine N-Methylatropine Atropine N-Methylhomatropine dl-Homatropine N-Methylacetyltropine acetylcholine
−50.78 −47.25 −40.46 −41.21 −39.91 −38.38 −27.38
−51.51 −47.36 −41.88 −42.47 −41.97 −40.29 −27.87
−50.78 −47.25 −40.17 −41.13 −39.91 −38.33 −27.30
10.54 9.62 8.74 8.49 7.07 7.06 4.77
pIC50 values are taken from reference 20. Again the pIC50 values have been corrected for racemic mixtures.
bonding site are removed, thereby making the pocket larger and possibly less specific. This is reflected in the substantial drop in the observed pIC50 values and our calculated binding energies. The second mutation keeps the aromatic ring but loses the hydrogen-bonding group, so the drop in pIC50 values is much less and this is again reflected in the calculated binding energies. These results suggest that the mutations are only affecting the binding pocket and not the overall fold of the protein as it is unlikely that such a good correlation with experimental data would be achieved if the fold of the protein had been disrupted.
plication of this constraint whilst compensating for the lack of a lipid membrane or explicit solvent would also have the effect of preventing any major conformational changes which might be necessary for the production of alternative receptor conformations or states. Hence we can observe the movement of amino acid side chains to block the binding site in the simulations where the ligand is not present but to infer that this represents the inactive protein conformation would not be inferring too much. Further studies on the unconstrained model in a membrane with explicit solvent would be needed to explore this aspect further.
3.3. Simulated annealing 4. Conclusion The simulated annealing simulations of the protein–ligand complex show that the binding site moves little compared with the rest of the protein. Ring flips are observed on the exterior of the protein, but the conformations of the key residues within the binding area are maintained. The root mean square deviation (rmsd) of all the atoms in the lowest ˚ from the energy structure after simulated annealing is 0.84 A original model, whereas the key residues shaping the bind˚ The ing pocket, mentioned above, have an rmsd of 0.47 A. atropine molecule remained in a similar orientation and location. Simulated annealing of the protein alone results in a binding site altered in such a way that atropine cannot enter the charged area. This is mainly due to residues Phe197 and Trp378 moving into the space that the ligand would otherwise occupy. The latter is particularly frequently observed with the largest motion into this area, thus facilitating the movement of the helices probably by the disruption of a hydrogen-bonding network directly below the binding site. The protein retains its overall fold in the simulated annealing studies. This is not surprising as the structure was constrained by a force constant to prevent unfolding. The ap-
A good correlation was obtained for the docking studies of a number of related inverse agonists using models based on the crystal structure of atropine and a model of m1AchR produced previously. The experimental results could be reproduced for both the native protein and two point mutations that had been studied experimentally for a range of inverse agonists with different sizes and donor/acceptor groups. This provides confidence in this approach to computational drug design and provides a basis for the prediction of the inhibition constants of previously untested agonists for this system.
5. Supplementary material Crystallographic data (excluding structure factors) for the structure reported in this paper have been deposited with the Cambridge Crystallographic Data Centre as supplementary publication No. CCDC-252464. Copies of available material can be obtained, free of charge, on application to the Director, CCDC, 12 Union Road, Cambridge CB2 1EZ, UK (fax: +44 1223 336033 or e-mail: [email protected]).
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385
Acknowledgments
Table A.2 ˚ and angles (◦ ) for atropine Bond lengths (A)
The authors would like to thank the EPSRC for a grant of computer time on the computational chemistry facility at RAL, for collection of the single-crystal X-ray data at Southampton University and for a studentship for ACT. The authors would particularly like to thank Ed Hulme for his valued comments on a draft of this paper.
Appendix A The compound was supplied by Sigma Chemicals Ltd. as a racemic mixture. Space group Fdd2 has inversion elements (the d-glides) and thus contains both enantiomorphs. The crystal structure contains both configurations of the molecule although only one is input to the X-ray calculations. The known active enantiomorph was selected for the docking studies described here and its co-ordinates are given in Tables A.1 and A.4. Crystallisation was achieved by controlled slow evaporation from methanol at −4 ◦ C. A suitable crystal of dimension 0.20 mm × mm 0.15 × 0.35 mm was selected for X-ray analysis. Intensity data were collected on an Enraf-Nonius CCD diffractometer controlled with the COLLECT (Hooft, 1998) software, using monochromated Mo K␣ ˚ The diffractometer was equipped radiation, λ = 0.71073 A. with an Oxford Cryosystems “Cryostreams” cooler (Cosier and Glazer, 1986), enabling the data to be collected at 150 K. A total exposure time of 119.5 min recorded 474 images at 1◦ intervals. A total of 17,984 integrated reflections was collected of which 2377 were unique. The resolution range was ˚ The crystal showed no significant variation in 20.00–0.70 A. Table A.1 Atomic coordinates (×104 ) and equivalent isotropic displacement parame˚ 2 ) for atropine ters (103 × A C(1) C(2) C(3) C(4) C(5) C(6) C(7) C(8) C(9) C(10) C(11) C(12) C(13) C(14) C(15) C(16) C(17) O(3) O(1) O(2) N(1)
381
x
y
z
U(eq)
3241(1) 2878(1) 2791(1) 3063(1) 3420(1) 3510(1) 3897(1) 4467(1) 3652(1) 3098(1) 3459(1) 3365(1) 3455(1) 2879(1) 2519(1) 2528(1) 2663(1) 3356(1) 3699(1) 4832(1) 2774(1)
45(1) −228(1) −449(1) −406(1) −135(1) 93(1) 393(1) 283(1) 646(1) 1145(1) 1460(1) 1674(1) 1475(1) 1327(1) 1445(1) 1204(1) 2048(1) 876(1) 634(1) 556(1) 1766(1)
11581(3) 11832(3) 10227(4) 8390(4) 8128(3) 9734(3) 9458(3) 8719(3) 7952(3) 7774(3) 7857(3) 9785(3) 11818(3) 12313(3) 10510(3) 8655(3) 11252(3) 8984(2) 6112(2) 8568(2) 9805(3)
36(1) 45(1) 47(1) 41(1) 34(1) 28(1) 27(1) 30(1) 27(1) 28(1) 28(1) 30(1) 33(1) 33(1) 30(1) 29(1) 44(1) 32(1) 41(1) 37(1) 29(1)
U(eq) is defined as one-third of the trace of the orthogonalised Uij tensor.
˚ Bond lengths (A) C(1) C(6) C(1) C(2) C(1) H(1A) C(2) C(3) C(2) H(2A) C(3) C(4) C(3) H(3) C(4) C(5) C(4) H(4A) C(5) C(6) C(5) H(5A) C(6) C(7) C(7) C(9) C(7) C(8) C(7) H(7) C(8) O(2) C(8) H(8A) C(8) H(8B) C(9) O(1) C(9) O(3) C(10) O(3) C(10) C(16) C(10) C(11) C(10) H(10) C(11) C(12) C(11) H(11A) C(11) H(11B) C(12) N(1) C(12) C(13) C(12) H(12) C(13) C(14) C(13) H(13A) C(13) H(13B) C(14) C(15) C(14) H(14A) C(14) H(14B) C(15) N(1) C(15) C(16) C(15) H(15) C(16) H(16A) C(16) H(16B) C(17) N(1) C(17) H(17A) C(17) H(17B) C(17) H(17C) O(2) H(2) Bond angles (◦ ) C(6) C(1) C(2) C(6) C(1) H(1A) C(2) C(1) H(1A) C(3) C(2) C(1) C(3) C(2) H(2A) C(1) C(2) H(2A) C(4) C(3) C(2) C(4) C(3) H(3) C(2) C(3) H(3) C(3) C(4) C(5) C(3) C(4) H(4A) C(5) C(4) H(4A) C(4) C(5) C(6)
1.375(3) 1.401(3) 0.9500 1.375(3) 0.9500 1.371(3) 0.9500 1.390(3) 0.9500 1.392(3) 0.9500 1.525(2) 1.518(2) 1.527(2) 1.0000 1.399(2) 0.9900 0.9900 1.197(2) 1.339(2) 1.461(2) 1.515(3) 1.526(2) 1.0000 1.524(3) 0.9900 0.9900 1.480(2) 1.549(3) 1.0000 1.549(3) 0.9900 0.9900 1.532(3) 0.9900 0.9900 1.482(2) 1.534(3) 1.0000 0.9900 0.9900 1.482(3) 0.9800 0.9800 0.9800 0.8400 120.39(19) 119.8 119.8 119.88(19) 120.1 120.1 120.15(19) 119.9 119.9 120.2(2) 119.9 119.9 120.38(19)
382
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385 Table 2 (Continued )
Table 2 (Continued ) C(4) C(5) H(5A) C(6) C(5) H(5A) C(1) C(6) C(5) C(1) C(6) C(7) C(5) C(6) C(7) C(9) C(7) C(6) C(9) C(7) C(8) C(6) C(7) C(8) C(9) C(7) H(7) C(6) C(7) H(7) C(8) C(7) H(7) O(2) C(8) C(7) O(2) C(8) H(8A) C(7) C(8) H(8A) O(2) C(8) H(8B) C(7) C(8) H(8B) H(8A) C(8) H(8B) O(1) C(9) O(3) O(1) C(9) C(7) O(3) C(9) C(7) O(3) C(10) C(16) O(3) C(10) C(11) C(16) C(10) C(11) O(3) C(10) H(10) C(16) C(10) H(10) C(11) C(10) H(10) C(12) C(11) C(10) C(12) C(11) H(11A) C(10) C(11) H(11A) C(12) C(11) H(11B) C(10) C(11) H(11B) H(11A) C(11) H(11B) N(1) C(12) C(11) N(1) C(12) C(13) C(11) C(12) C(13) N(1) C(12) H(12) C(11) C(12) H(12) C(13) C(12) H(12) C(14) C(13) C(12) C(14) C(13) H(13A) C(12) C(13) H(13A) C(14) C(13) H(13B) C(12) C(13) H(13B) H(13A) C(13) H(13B) C(15) C(14) C(13) C(15) C(14) H(14A) C(13) C(14) H(14A) C(15) C(14) H(14B) C(13) C(14) H(14B) H(14A) C(14) H(14B) N(1) C(15) C(14) N(1) C(15) C(16) C(14) C(15) C(16) N(1) C(15) H(15) C(14) C(15) H(15) C(16) C(15) H(15) C(10) C(16) C(15) C(10) C(16) H(16A) C(15) C(16) H(16A) C(10) C(16) H(16B) C(15) C(16) H(16B)
119.8 119.8 119.03(16) 120.19(17) 120.78(17) 110.17(14) 109.99(15) 112.03(13) 108.2 108.2 108.2 112.15(14) 109.2 109.2 109.2 109.2 107.9 125.09(17) 125.11(17) 109.75(15) 107.68(14) 109.15(14) 112.66(14) 109.1 109.1 109.1 113.29(15) 108.9 108.9 108.9 108.9 107.7 106.75(14) 104.69(15) 113.14(14) 110.7 110.7 110.7 103.88(14) 111.0 111.0 111.0 111.0 109.0 104.08(15) 110.9 110.9 110.9 110.9 109.0 104.86(14) 106.50(14) 113.47(15) 110.6 110.6 110.6 113.84(15) 108.8 108.8 108.8 108.8
H(16A) C(16) H(16B) N(1) C(17) H(17A) N(1) C(17) H(17B) H(17A) C(17) H(17B) N(1) C(17) H(17C) H(17A) C(17) H(17C) H(17B) C(17) H(17C) C(9) O(3) C(10) C(8) O(2) H(2) C(12) N(1) C(17) C(12) N(1) C(15) C(17) N(1) C(15)
107.7 109.5 109.5 109.5 109.5 109.5 109.5 117.23(14) 109.5 111.58(15) 101.51(13) 111.81(15)
intensity during the course of data collection. Data were processed using DENZO (Otwinowski and Minor, 1997), correcting for Lorentz and polarisation effects, and absorption effects were applied using the program SORTAV (Blessing, 1997). The structure was solved by Patterson Search (Beurger, 1959) using the program PATTSEE (Egert and Sheldrick, 1985). The search model (Fig. A.1) comprised two rigid fragments: ring A (atoms 1–6) and atoms C(7), C(9), C(8), O(1) and O(3). The program determines and refines one linkage torsion angle, in this case about bond C(6) C(7). This model was searched for in the Patterson function, and possible solutions were refined to give a starting model for the structure, which was then expanded by Fourier methods and refined by full-matrix least-squares. Co-ordinates for the starting model fragments were built in the program Chem-X (Oxford Molecular, 1999). Location of the search model takes place in two steps: (1) rotation, where the best solution for the orienTable A.3 ˚ 2 ) for atropine Anisotropic displacement parameters (103 × A C(1) C(2) C(3) C(4) C(5) C(6) C(7) C(8) C(9) C(10) C(11) C(12) C(13) C(14) C(15) C(16) C(17) O(3) O(1) O(2) N(1)
U11
U22
U33
U23
U13
U12
29(1) 31(1) 30(1) 37(1) 30(1) 23(1) 31(1) 29(1) 30(1) 41(1) 31(1) 35(1) 35(1) 42(1) 28(1) 31(1) 66(1) 44(1) 55(1) 40(1) 37(1)
34(1) 47(1) 35(1) 30(1) 30(1) 23(1) 21(1) 24(1) 23(1) 19(1) 26(1) 25(1) 36(1) 30(1) 27(1) 24(1) 29(1) 22(1) 37(1) 38(1) 20(1)
45(1) 56(1) 76(2) 57(1) 42(1) 39(1) 28(1) 38(1) 27(1) 25(1) 27(1) 31(1) 29(1) 27(1) 36(1) 34(1) 36(1) 29(1) 32(1) 34(1) 30(1)
8(1) 21(1) 21(1) 3(1) 6(1) 5(1) −2(1) 2(1) 0(1) 3(1) 3(1) 2(1) −1(1) 4(1) 5(1) 4(1) −6(1) 2(1) −3(1) 2(1) −1(1)
9(1) 15(1) −2(1) −10(1) −2(1) −4(1) −2(1) 1(1) −1(1) −8(1) 2(1) −5(1) −7(1) 2(1) 7(1) −6(1) 6(1) 0(1) 0(1) −3(1) 1(1)
8(1) 9(1) −5(1) −4(1) 1(1) 7(1) 3(1) −3(1) −4(1) 4(1) 1(1) −7(1) −1(1) 1(1) 1(1) −3(1) 5(1) 10(1) 16(1) −15(1) 1(1)
The anisotropic displacement factor −2π2 [h2 a∗2 U 11 + · · · + 2hka∗ b∗ U 12 ].
exponent
takes
the
form:
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385 Table A.4 Hydrogen coordinates (×104 ) and isotropic displacement parameters ˚ 2 ) for atropine (103 × A H(1A) H(2A) H(3) H(4A) H(5A) H(7) H(8A) H(8B) H(10) H(11A) H(11B) H(12) H(13A) H(13B) H(14A) H(14B) H(15) H(16A) H(16B) H(17A) H(17B) H(17C) H(2)
x
y
z
U(eq)
3302 2691 2543 3007 3603 3940 4618 4432 3063 3850 3385 3600 3729 3582 2892 2739 2133 2370 2290 2269 2876 2769 4938
197 −259 −633 −562 −104 507 114 174 1069 1391 1600 1881 1292 1626 1077 1416 1485 984 1298 2102 2247 1980 611
12694 13108 10392 7293 6846 10826 9694 7350 6306 7807 6621 9755 11625 12938 12381 13643 10983 9082 7555 11224 10835 12654 9755
43 54 56 50 41 32 36 36 34 33 33 36 40 40 39 39 36 35 35 66 66 66 56
tation of the search model is output, and (2) translation of the rotated model to determine its position in the unit cell. The expanded structure containing all non-H atoms was refined using SHELX-97 (Sheldrick, 1997) implemented in the WinGX system of programs (Farrugia, 1999). Non-hydrogen atoms were refined anisotropically by full-matrix least-square methods. Hydrogens were added geometrically and refined in riding mode with isotropic temperature factors. Geometrical calculations were made with the programs PARST and PLATON (Spek, 1990) as implemented in WinGX. WinGX was also used to prepare publication material, including figures (ORTEP, Barnes, 1997 and POV-Ray, 2004) and tables. All diagrams of the molecules given in the paper are of the biologically active enantiomer (this is the enantiomer referred
383
Table A.5 Torsion angles (◦ ) for atropine C(6) C(1) C(2) C(3) C(1) C(2) C(3) C(4) C(2) C(3) C(4) C(5) C(3) C(4) C(5) C(6) C(2) C(1) C(6) C(5) C(2) C(1) C(6) C(7) C(4) C(5) C(6) C(1) C(4) C(5) C(6) C(7) C(1) C(6) C(7) C(9) C(5) C(6) C(7) C(9) C(1) C(6) C(7) C(8) C(5) C(6) C(7) C(8) C(9) C(7) C(8) O(2) C(6) C(7) C(8) O(2) C(6) C(7) C(9) O(1) C(8) C(7) C(9) O(1) C(6) C(7) C(9) O(3) C(8) C(7) C(9) O(3) O(3) C(10) C(11) C(12) C(16) C(10) C(11) C(12) C(10) C(11) C(12) N(1) C(10) C(11) C(12) C(13) N(1) C(12) C(13) C(14) C(11) C(12) C(13) C(14) C(12) C(13) C(14) C(15) C(13) C(14) C(15) N(1) C(13) C(14) C(15) C(16) O(3) C(10) C(16) C(15) C(11) C(10) C(16) C(15) N(1) C(15) C(16) C(10) C(14) C(15) C(16) C(10) O(1) C(9) O(3) C(10) C(7) C(9) O(3) C(10) C(16) C(10) O(3) C(9) C(11) C(10) O(3) C(9) C(11) C(12) N(1) C(17) C(13) C(12) N(1) C(17) C(11) C(12) N(1) C(15) C(13) C(12) N(1) C(15) C(14) C(15) N(1) C(12) C(16) C(15) N(1) C(12) C(14) C(15) N(1) C(17) C(16) C(15) N(1) C(17)
0.4(3) 0.5(3) −1.1(3) 0.8(3) −0.7(3) 178.97(16) 0.1(3) −179.57(16) −110.94(18) 68.7(2) 126.28(18) −54.0(2) 60.3(2) −176.83(15) −86.0(2) 38.0(2) 91.53(17) −144.50(15) −83.27(17) 36.3(2) −57.62(18) 57.00(19) 25.77(18) −90.08(17) 1.44(18) −28.21(18) 87.61(18) 84.44(17) −36.0(2) 56.77(18) −58.1(2) −3.7(3) 178.78(13) 139.48(15) −97.92(17) −164.07(15) 75.73(17) 76.70(16) −43.50(17) 44.70(17) −75.85(17) −74.36(18) 165.09(16)
Symmetry transformations were used to generate equivalent atoms.
to as (−)atropine). Anisotropic displacement parameters are in Table A.3. A.1. Molecular geometry
Fig. A.1. Search model used in the X-ray analysis comprising fragments 1 and 2 and rotation about bond C(6) C(7).
Bond lengths in the structure are determined approxi˚ and bond angles to ±0.1◦ . Both bond lengths mately ±0.02 A and bond angles, Table A.2, in the structure compare well with those found in organic compounds (Ladd and Palmer, 2003) and exhibit no unusual values. The linkage chain comprising atoms C(7), C(9), O(3) and C(10) is fully extended with a torsion angle of 178.8(1)◦ . The methyl-azabicyclooctyl ester head of the molecule contains three saturated rings: a six-membered piperidino ring B1 B2, a seven membered saturated carbon ring B1 B3 and a 5-membered pyrrolinyl
384
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385
not observed in the hydrobromide structure or in the present structure of the native molecule.
References
Fig. A.2. Ring conformations.
ring B2 B3. The conformations of these three rings are illustrated in Fig. A.2. Torsion angles are in Table A.5. Ring B1 B2 (Fig. A.2(a)) has a sofa conformation with a pseudo m-plane along N(1) · · · C(10). The asymmetry parameter (Duax and Norton, 1995) CsN(1)···C(10) = 0.7(2)◦ indicating a high degree of mirror symmetry. The ring is steeper at the N(1) end with torsion angles of around 90◦ . Ring B1 B3 (Fig. A.2(b)) has a boat-like conformation with a pseudo m-plane along C(10) to the bisector of bond C(13) C(14). The asymmetry parameter CsC(10) = 1.6(3)◦ indicating also a high degree of mirror symmetry. The sides of boat are steeper at the C(13) and C(14) end again with torsion angles of around 90◦ . Ring B2 B3 (Fig. A.2(c)) has a sofa conformation with a pseudo m-plane along N(1) to the bisector of bond C(13) C(14). The asymmetry parameter CsN(1) = 1.9(4)◦ again indicates a high degree of mirror symmetry. Atoms C(12), C(13), C(14) and C(15) are approximately coplanar with N(1) out of the plane. The three pseudo m-planes in the three rings are approximately coincident and the composite saturated ring system thus has a pseudo m-plane running from C(10) to N(1) and the bisector of bond C(13) C(14). The methyl group C(17)H3 is approximately equatorial to this ring system. A.2. Crystal packing and hydrogen bonding The crystal structure is held together by strong hydrogen bonds formed through O(2) H(2) · · · N(1)* linkages where * indicates the symmetry operation [x + 1/4, −y + 1/4, z + 1/4]. These H-bonds have the geometry O(2) · · · ˚ and O(2) H(2) · · · N(1) = 172.5(3)◦ . BeN(1)* = 2.740(6) A cause of the high symmetry in space group Fdd2 there are many equivalent H-bonds throughout the structure. A previous study of the X-ray structure of atropine has been published by Pauling and Petcher in 1970 but no details of the space group or co-ordinates were given. The crystal structure of atropine hydrobromide was reported by Pauling and Petcher (1970) although no corresponding data have been deposited. In this paper, the molecule is shown in the correct absolute configuration corresponding to the more potent enantiomer and to that derived and used here. These authors propose that the active form of the molecule involves an intermolecular H-bond O(2)H · · · O(1) although this is
Archer, E., Maigret, B., Escrieut, C., Pradayrol, L., Fourmy, D., 2003. Rhodopsin crystal: new template yielding realistic models of Gprotein-coupled receptors? Trends Pharmacol. Sci. 24, 36–40. Bakker, R.A., Wieland, K., Timmerman, H., Leurs, R., 2000. Constitutive activity of the histamine H-1 receptor reveals inverse agonism of histamine H-1 receptor antagonists. Eur. J. Pharmacol. 387, R5–R7. Bayly, C.I., Cieplak, P., Cornell, W.D., Kollman, P.A.A., 1993. A wellbehaved electrostatic potential based method using charge restraints for deriving atomic charges—the RESP model. J. Phys. Chem. 97, 10269–10280. Chemwindow, 1999–2003. Bio-Rad Laboratories, Inc., Informatics Division. Chen, G., Jayawickreme, C., Way, J., Armour, S., Queen, K., Watson, C., Ignar, D., Chen, W., Kenakin, T., 1999. Constitutive receptor systems for drug discovery. J. Pharmacol. Toxicol. 42, 199–206. CVFF, 1990. CVFF forcefield file shipped with Discover 2.6.0/InsightII 1.1.0/Insight 2.6, September 1990. Biosym Technologies, Inc. Dong, X., Han, S.K., Zylka, M.J., Simon, M.I., Anderson, D.J., 2001. A diverse family of GPCRs expressed in specific subsets of nociceptive sensory neurons. Cell 106, 619–632. Flower, D.R., 1999. Modelling G-protein-coupled receptors for drug design. Biochim. Biophys. Acta 1422, 207–234. Jakubik, J., Bacakova, L., El-Fakahany, E.E., Tucek, S., 1995. Constitutive activity of the M(1)M(4) subtypes of muscarinic receptors in transfected CHO cells and of muscarinic receptors in the heart cells revealed by negative antagonists. FEBS Lett. 377, 275–279. Lu, Z.-L., Saldanha, J.W., Hulme, E.C., 2001. Transmembrane domains 4 and 7 of the M-1 muscarinic acetylcholine receptor are critical for ligand binding and the receptor activation switch. J. Biol. Chem. 276, 34098–34104. Morris, G.M., Goodsell, D.S., Halliday, R.S., Huey, R., Hart, W.E., Belew, R.K., Olson, A.J., 1998. Automated docking using a Lamarckian genetic algorithm and an empirical binding free energy function. J. Comp. Chem. 19, 1639–1662. Palczewski, K., Kumasaka, T., Hori, T., Behnke, C.A., Motoshima, H., Fox, B.A., Le Trong, I., Teller, D.C., Okada, T., Stenkamp, R.E., Yamamoto, M., Miyano, M., 2000. Crystal structure of rhodopsin: a G-protein-coupled receptor. Science 289, 739–745. Parma, J., Duprez, L., Vansande, J., Cochaux, P., Gervy, C., Mockel, J., Dumont, J., Vassart, G., 1993. Somatic mutations in the thyrotropin receptor gene cause hyperfunctioning thyroid adenomas. Nature 365, 649–651. Parnot, C., Miserey-Lenkei, S., Bardin, S., Corvol, P., Clauser, E., 2002. Lessons from constitutively active mutants of G-protein-coupled receptors. Trends Endocrinol. Metab. 13, 336–343. Peitsch, M.C., Wilkins, M.R., Tonella, L., Sanchez, J.C., Appel, R.D., Hochstrasser, D.F., 1997. Large-scale protein modelling and integration with the SWISS-PROT and SWISS-2DPAGE databases: the example of Escherichia coli. Electrophoresis 18, 498–501. Robinson, P.R., Cohen, G.B., Zukovsky, E.A., Oprian, D.D., 1992. Constitutively active mutants of rhodopsin. Neuron 9, 719–725. Schmidt, M.W., Baldridge, K.K., Boatz, J.A., Elbert, S.T., Gordon, M.S., Jensen, J.J., Koseki, S., Matsunaga, N., Nguyen, K.A., Su, S., Windus, T.L., Dupuis, M., Montgomery, J.A., 1993. General atomic and molecular electronic-structure system. J. Comput. Chem. 14, 1347–1363. Shacham, S., Topf, M., Avisar, N., Glaser, F., Marantz, Y., Bar-Haim, S., Noiman, S., Naor, Z., Becker, O.M., 2001. Modeling the 3D structure of GPCRs from sequence. Med. Res. Rev. 21, 472–483. Spalding, T.A., Burnstein, E.S., 2001. Constitutively active muscarinic receptors. Life Sci. 68, 2511–2516.
A.C. Tanczos et al. / Computational Biology and Chemistry 28 (2004) 375–385 Vaidehi, N., Floriano, W.B., Trabino, R., Hall, S.E., Freddolino, P., Choi, E.J., Zamanakos, G., Goddard III, W.A., 2002. Prediction of structure and function of G-protein-coupled receptors. Proc. Natl. Acad. Sci. 99, 12622–12627. Ward, S.D.C., Curtis, C.A.M., Hulme, E.C., 1999. Scanning mutagenesis identifies amino acid side chains in transmembrane domain 5 of the M-1 muscarinic receptor that participate in binding the acetyl methyl group of acetylcholine. Mol. Pharm. 56, 1031– 1041. Weiner, P.K., Kollman, P.A., 1981. AMBER-assisted model-building with energy refinement—a general program for modeling molecules and their interactions. J. Comput. Chem. 2, 287–303. Weinstein, L.S., Shenker, A., 1993. G-protein mutations in human-disease. Clin. Biochem. 26, 333–338. Yamashita, A., Endo, M., Higashi, T., Nakatsu, N., Yamada, Y., Oda, J., Kato, H., 2003. Capturing enzyme structure prior to reaction initiation: tropinone reductase-II-substrate. Biochemistry 42, 5566– 5573.
Further reading
Barnes, C.L., 1997. ORTEP-3 for Windows – a version of ORTEP-III with a Graphical User Interface (GUI). J Appl. Cryst. 30, 568–568 [Based on ORTEP-III (v 1.0.3) by C.K. Johnson and M.N. Burnett].
385
Beurger, M.J., 1959. Vector Space. Wiley, New York. Blessing, R.H., 1997. Outlier treatment in data merging. J. Appl. Cryst. 30, 421–426. Cosier, J., Glazer, A.M., 1986. A nitrogen-gas-stream cryostat for general X-ray diffraction studies. J. Appl. Cryst. 19, 105–107. Duax, W.L., Norton, D.A., 1995. Atlas of Steroid Structure. Plenum Press, New York. Egert, E., Sheldrick, G.M., 1985. Search for a fragment of known geometry by integrated Patterson and direct methods. Acta Crystallogr. A 41, 262–268. Farrugia, L.J., 1999. WinGX. J. Appl. Cryst. 32, 837–838. Hooft, R., 1998. Nonius, B.V., COLLECT: X-ray data collection and processing software user interface. Ladd, M.F.C., Palmer, R.A., 2003. Structure Determination by X-ray Crystallography, 4th ed. Plenum Press, New York, Tables 7.24, 7.25. Otwinowski, Z., Minor, W., 1997. Methods in enzymology Part A. In: Carter Jr., C.W., Sweet, R.M. (Eds.), Macromolecular Crystallography, vol. 276. Academic Press, New York, pp. 307–326. Oxford Molecular (1999). http://www.oxmol.com:Chem-X. Pauling, P.J., Petcher, T.J., 1970. Nature 228, 673–674. POV-Ray, 2004. POV-Ray rendering engine for Windows. Version 3.1 g.watcom.win32 [Pentium II Optimized]. Sheldrick, G.M., 1997. SHELX-97: Program for Refinement of Crystal Structures. University of Gottingen. Spek, A.L., 1990. BYPASS—an effective method for the refinement of crystal-structures containing disordered solvent regions. Acta Crystallogr. A 46, 194–201.