letters to nature 15. Kuhn, R., Schwenk, F., Aguet, M. & Rajewsky, K. Inducible gene targeting in mice. Science 269, 1427–1429 (1995). 16. Socolovsky, M. et al. Ineffective erythropoiesis in Stat5a 2/25b 2/2 mice due to decreased survival of early erythroblasts. Blood 98, 3261–3273 (2001). 17. Wadman, I. A. et al. The LIM-only protein Lmo2 is a bridging molecule assembling an erythroid, DNA-binding complex which includes the TAL1, E47, GATA-1 and Ldb1/NLI proteins. EMBO J. 16, 3145–3157 (1997). 18. Yamada, Y. et al. The T cell leukemia LIM protein Lmo2 is necessary for adult mouse hematopoiesis. Proc. Natl Acad. Sci. USA 95, 3890–3895 (1998). 19. Choi, K., Kennedy, M., Kazarov, A., Papadimitriou, J. C. & Keller, G. A common precursor for hematopoietic and endothelial cells. Development 125, 725–732 (1998). 20. Robertson, S. M., Kennedy, M., Shannon, J. M. & Keller, G. A transitional stage in the commitment of mesoderm to hematopoiesis requiring the transcription factor SCL/tal-1. Development 127, 2447–2459 (2000). 21. Faloon, P. et al. Basic fibroblast growth factor positively regulates hematopoietic development. Development 127, 1931–1941 (2000). 22. Terada, N. et al. Bone marrow cells adopt the phenotype of other cells by spontaneous cell fusion. Nature 416, 542–545 (2002). 23. Ying, Q. L., Nichols, J., Evans, E. P. & Smith, A. G. Changing potency by spontaneous fusion. Nature 416, 545–548 (2002). 24. Cantor, A. B., Katz, S. G. & Orkin, S. H. Distinct domains of the GATA-1 cofactor FOG-1 differentially influence erythroid versus megakaryocytic maturation. Mol. Cell. Biol. 22, 4268–4279 (2002).
Acknowledgements We thank D. Traver for discussions; J. Dailey and S. Lazo-Kallanian for cell sorting; K. Rajewsky for MxCre mice; and A. Williams and S. Galusha for assistance in generating the conditional SCL/tal-1 strain. H.K.A.M. received support from the Finnish Cultural Foundation and the Academy of Finland. This work was supported in part by a grant from the NIH to S.H.O., who is an Investigator of the Howard Hughes Medical Institute. Competing interests statement The authors declare that they have no competing financial interests. Correspondence and requests for materials should be addressed to S.H.O. (e-mail:
[email protected]).
..............................................................
Crystal structure of the human angiotensin-converting enzyme–lisinopril complex Ramanathan Natesh*, Sylva L. U. Schwager†, Edward D. Sturrock† & K. Ravi Acharya* * Department of Biology and Biochemistry, University of Bath, Claverton Down, Bath BA2 7AY, UK † Division of Medical Biochemistry and MRC/UCT Liver Research Centre, University of Cape Town Medical School, Observatory 7925, South Africa .............................................................................................................................................................................
Angiotensin-converting enzyme (ACE) has a critical role in cardiovascular function by cleaving the carboxy terminal His-Leu dipeptide from angiotensin I to produce a potent vasopressor octapeptide, angiotensin II. Inhibitors of ACE are a first line of therapy for hypertension, heart failure, myocardial infarction and diabetic nephropathy. Notably, these inhibitors were developed without knowledge of the structure of human ACE, but were instead designed on the basis of an assumed mechanistic homology with carboxypeptidase A1. Here we present the X-ray structure of human testicular ACE and its complex with one of the most widely used inhibitors, lisinopril (N 2-[(S)-1-carboxy-3-phenylpropyl]-L-lysyl-L-proline; also known as Prinivil or Zestril), at 2.0 A˚ resolution. Analysis of the three-dimensional structure of ACE shows that it bears little similarity to that of carboxypeptidase A, but instead resembles neurolysin2 and Pyrococcus furiosus carboxypeptidase3—zinc metallopeptidases with no detectable sequence similarity to ACE. The structure provides an opportunity to design domainselective ACE inhibitors that may exhibit new pharmacological profiles. NATURE | VOL 421 | 30 JANUARY 2003 | www.nature.com/nature
Angiotensin-converting enzyme (also known as peptidyl dipeptidase A, EC 3.4.15.1 or ACE) is a type-I membrane-anchored dipeptidyl carboxypeptidase that is essential for blood pressure regulation and electrolyte homeostasis through the renin–angiotensin–aldosterone system. There are two isoforms of ACE that are transcribed from the same gene in a tissue-specific manner. In somatic tissues it exists as a glycoprotein composed of a single, large polypeptide chain of 1,277 amino acids, whereas in sperm cells it is a lower-molecular-mass glycoform of 701 amino acids. The somatic form consists of two homologous domains (N and C domain), each of which contains an active site with a conserved HEXXH zincbinding motif 4, where the two histidines are zinc ligands, with a glutamate 24 residues downstream forming the third ligand5. The two domains differ in their substrate specificities, inhibitor and chloride activation profiles, and physiological functions6. There are two N-domain-specific substrates: the peptide N-acetyl-serylaspartyl-lysyl-proline, which regulates haematopoietic stem cell differentiation and proliferation; and the bradykinin-potentiating peptide angiotensin-(1-7)7. On the other hand, the active sites of both domains catalyse the hydrolysis of angiotensin I and the vasodilator bradykinin with similar efficiency. However, inhibition of the N domain with a phosphinic peptide RXP407 has no effect on blood pressure regulation7, and expression in transgenic mice of the N domain alone produces a phenotype similar to that seen in complete ACE knockout mice8. Thus, the C domain seems to be necessary and sufficient for controlling blood pressure and cardiovascular function, suggesting that the C domain is the dominant angiotensin-converting site. Testis ACE is identical to the C-terminal half of somatic ACE, except for a unique 36-residue sequence constituting its amino terminus9. We used a truncated version, tACED36NJ (a mutant with full enzymatic activity that is truncated at Ser 625 and that lacks the O-glycan-rich, N-terminal 36 residues and hydrophobic transmembrane domain), for structure determination. The carbohydrates were modified to minimize oligosaccharide-based heterogeneity10. We refer to this as tACE. The structure of tACE (residues 37–625) adopts an overall ellipsoid shape (dimensions approximately 72 £ 57 £ 48 A˚) with a central groove that extends for about 30 A˚ into the molecule and divides the protein into two ‘subdomains’ (labelled I and II in Fig. 1a). The boundaries of the groove are provided by helices a13, a14, a15, a17 and strand b4. The structure of the tACE–lisinopril complex was used to locate the active site of the molecule. On top of the molecule there is an N-terminal ‘lid’ formed by helices a1, a2 and a3 (all three containing several charged residues), which seems to restrict the access of large polypeptides to the active-site cleft and thereby accounts for the enzyme’s inability to hydrolyse large, folded substrates (Fig. 1b). The structure of tACE is predominantly helical with 27 helices (almost equally distributed in both subdomains), 20 a-helices and seven 310-helices (Fig. 1a, c). The only b-structure, accounting for 4% of all residues, occurs as six relatively short strands, two of which are located near the active site (Fig. 1a). The long central helix a15 packs diagonally across the two kinked helices a6 and a8 (both of them contain a proline residue). Residues Asp 40 (a1) and Gly 615 (five residues downstream of a20) define the N and C termini of the ectodomain, respectively, and this is in agreement with previous tACE mutagenesis and cleavage-secretion studies11. Residues Ser 435 to Gly 438 are disordered in the structure. A total of 504 water molecules were identified in the native structure. The active-site pocket is occupied by several ordered water molecules. All six glycosylation sites (g1–6, Fig. 1c) are located on the surface of the tACE molecule. Weak electron density was observed for N-linked carbohydrates in all of the glycosylation sites and modelled with an N-acetylglucosamine moiety. Zinc is an important catalytic component of ACE6,12. One highly ordered zinc ion (B-factor ¼ 16.3 A˚2 in tACE-native) is bound at
© 2003 Nature Publishing Group
551
letters to nature the active site. Helix a13 contains the HEXXH zinc-binding motif, with its two zinc-coordinating histidines (His 383 and His 387). Additional coordination is provided by Glu 411 on helix a14 and an acetate ion (from the crystallization medium). The role of a zinc ion in ACE catalysis was thought to be analogous to that in thermolysin4, and our structural data confirm that the zinc-binding sites in both proteins are indeed very similar (root mean square (r.m.s.) deviation of 0.52 A˚ (all atoms) for residues that are zinc ligands in both proteins), except for the acetate ion that is replaced by a water molecule in the coordination sphere in thermolysin. Substrate hydrolysis by ACE is activated by chloride ion in a substrate-dependent manner13,14. The C-domain active site, but not the N domain, in ACE is strongly activated by chloride ion15. The structure of tACE revealed the location of two buried chloride ions separated by 20.3 A˚ (Fig. 1a). The first (Cl1, 20.7 A˚ away from the zinc ion) is bound to four ligands—Arg 489 (NH1), Arg 186 (NE), Trp 485 (NE1) and water—and is surrounded by a hydrophobic shell of four tryptophans. The second (Cl2, 10.4 A˚ away from the zinc ion) is bound to Arg 522 (NE, 3.1 A˚), in agreement with a previous report indicating that Arg 1098 (the analogous Arg residue in the C domain of somatic ACE) is critical for the chloride dependence of ACE activity16. Tyr 224 and a water molecule are the other two Cl2 ligands. Thus, binding of each chloride in tACE involves ligands from both subdomains (Fig. 1c). It has been proposed that the chloride ion might interact with the substrate as well as the
enzyme16. However, the location of both buried chloride ions outside the active site makes such an interaction unlikely. The molecular mechanism of chloride activation is not readily apparent from the structure; however, the primary ligand for Cl2, Arg 522, lies on the same helix (a17) as two residues (Tyr 520 and Tyr 523) that interact with lisinopril and, presumably, substrate as well (Figs 1c and 2a). ACE belongs to the M2 family of zinc-binding metallopeptidases, within the MA clan. Apart from the HEXXH metal-coordinating motif, there is little sequence homology between ACE and other members of the MA clan. Structural comparison of tACE with other protein structures using the DALI server17 revealed significant homology with neurolysin2, a protein involved in neurotensin metabolism (Fig. 3a, b), and a carboxypeptidase from the hyperthermophilic archaeon P. furiosus3. Neurolysin is a member of the M3 family of oligopeptidases (a member of the MA clan), and P. furiosus carboxypeptidase is a member of the M32 family of carboxypeptidases. Similar to ACE, both belong to the family of metallopeptidases bearing the HEXXH active-site motif18 and consist of an abundance of a-helices with very little b-structure. The two proteins exhibit little amino-acid sequence similarity with tACE, yet when the two structures are optimally superimposed, there is a noticeable match with an r.m.s. deviation of 3.4 A˚ for 361 Ca atoms and 3.11 A˚ for 399 Ca atoms against neurolysin and P. furiosus carboxypeptidase, respectively. The core structures for
Figure 1 Overview of tACE structure. a, Stereo view of the ribbon representation of the molecule looking down on the active site. The molecule can be divided into two portions, as subdomains I and II (cyan and pink, respectively). The active-site zinc ion and the lisinopril molecule are shown in green and yellow, respectively. The two chloride ions are shown as red spheres. b, Molecular surface representation (negative and positive potentials in red and blue, respectively) showing the active-site groove. The view is at 908 (towards the observer) to a. For clarity, the molecular surface has been sliced. The buried
lisinopril molecule is shown in yellow. Helices a1, a2 and a3 forming the lid are shown. c, The structure–sequence relationship in tACE10. The secondary structure elements (subdomain I in cyan; subdomain II in pink) follow the same colour code as in a. a, a-helices; b, b-strands; H, 310 helices. The important residues that are involved in binding are indicated as follows: zinc ligands, green boxes; chloride-binding residues, orange (Cl1) and red (Cl2) boxes; lisinopril-binding residues, yellow boxes; and glycosylation sites, black boxes.
552
© 2003 Nature Publishing Group
NATURE | VOL 421 | 30 JANUARY 2003 | www.nature.com/nature
letters to nature
Figure 2 Details of the active site. a, Binding of lisinopril to tACE (stereo representation). Selected residues are shown in a ‘ball-and-stick’ representation with zinc atoms in green, chloride ions in red, water molecules in purple, and lisinopril (inhibitor) in yellow. Important secondary structure elements are marked. The lisinopril, Cl2, water hydrogen
bonds and zinc coordination are shown in cyan, red, purple and green dotted lines, respectively. b, Schematic view of lisinopril binding with distances marked in A˚. The different binding subsites are labelled.
these three proteins are similar, with significant differences in loops on the outer surface. The marked similarity also extends to the active-site region in neurolysin and P. furiosus carboxypeptidase, which consists of a deep narrow channel (about 20% larger accessible surface area than in tACE) that divides the molecule into two halves, and which only allows access to small peptide substrates. Notably, ACE also acts on neurotensin among its wide range of substrates19. Structural comparison of tACE with thermolysin (M4 family) revealed that there was significant, but far less marked, structural similarity between them (r.m.s. deviation of 3.2 A˚ for 170 Ca atoms). The structure of tACE bound to the potent inhibitor lisinopril6,20,21 ðK i ¼ 2:7 £ 10210 MÞ shows that the inhibitor binds in a highly ordered (overall B-factor of 15.26 A˚2), extended conformation, with the phenyl group stretching towards the lid (Fig. 1a, b) and the lysine side chain parallel to the a13 helix containing the HEXXH motif (Fig. 2a). The lisinopril molecule is buried about 10 A˚ inside the groove (as viewed in Fig. 1a), in agreement with the previous studies on biotinylated derivatives of lisinopril with ACE22. No significant rearrangement of active-site residues was observed on complex formation and the chloride ions are bound as in the native structure. The carboxyalkyl carboxylate of lisinopril is well positioned to bind to the active-site zinc atom20,21, and provides one coordinating ligand (provided by an acetate ion in the native structure). The other three ligands (two histidines and a glutamic acid) are the same in both structures. The second oxygen atom of this carboxylate is 2.6 A˚ away from the zinc atom and makes an
H-bond interaction with Glu 384 (OE2 atom, which appears to be protonated, as the complex was crystallized at about pH 4.7) of the HEXXH motif (Fig. 2a, b). The S1 phenylpropyl group makes van der Waals interactions with Val 518, and the lysyl amine forms a weak H-bond with Glu 162 (OE2 atom, 3.4 A˚) at the S1 0 subsite of tACE. Surprisingly, the C-terminal carboxylate, which was thought to interact with a positively charged arginine residue (based on chemical modification experiments23), instead binds to a lysine (Lys 511) as well as to Tyr 520. We have presented here a three-dimensional structure of tACE and the tACE–lisinopril complex. Our study provides a detailed picture of the active site, which could be used for the development of new, highly selective ACE inhibitors targeted to the C domain by structure-based, rational drug design. This approach should produce a new generation of ACE inhibitors with the potential for greater efficacy, fewer side effects and new treatment indications (for example, polycythaemia). Furthermore, the similarity of tACE with neurolysin and P. furiosus carboxypeptidase has shown a surprising structural conservation between these metallopeptidases. A
Methods Purification, crystallization and data collection Details of the production of recombinant tACE (tACED36NJ) and growth of crystals (by vapour diffusion) will be reported elsewhere (our own unpublished results). Briefly, 2 ml of the protein solution was mixed in 10 mM HEPES and 0.1% phenylmethylsulphonyl fluoride, with an equal volume of a reservoir solution containing 15% PEG 4000, 50 mM sodium acetate trihydrate, pH 4.7, and 10 mM ZnSO4·7H2O. tACE–lisinopril complex crystals were grown at 1:1.2 (protein/inhibitor) ratio of the inhibitor. Both native and complex crystals belonged to the P212121 space group (grown at 16 8C, cell dimensions: a ¼ 56.47, b ¼ 84.90, c ¼ 133.99 A˚) with one molecule in the asymmetric unit. All X-ray data collection was performed at 100K either at the European Synchrotron Radiation Facility (ESRF) or at the Synchrotron Radiation Source (SRS). Both native and lisinopril complex data sets were collected to 2.0 A˚ (R merge (native), 8.1% (last shell, 43.2%) with overall completeness 98.8% (2.07–2.0 A˚, 95.7%); R merge (tACE–lisinopril) 5.8% (last shell, 23.8%) with overall completeness 95.9% (2.08–2.01 A˚, 72.1%)). For structure solution, multiple anomalous dispersion (MAD) data were collected at the peak of the zinc-K edge (1.2825 A˚), inflection (1.2832 A˚) and a remote wavelength (0.9537 A˚), and at 1.7712 A˚, the closest possible wavelength at the sulphur K-edge. Out of 32 soaks with different heavy atoms, only 3 were found to be useful. These derivatives were prepared by soaking the crystals (approximately 10–60 min, 1–5 mM concentration) in the presence of K2PtCl4, K2PdCl4 and OsCl3, respectively. All data were processed using the program HKL2000 (ref. 24).
Structure determination and refinement
Figure 3 Comparison of tACE (a) and neurolysin (b) folds. The same view as in Fig. 1a is retained. NATURE | VOL 421 | 30 JANUARY 2003 | www.nature.com/nature
The crystal structure of the ACE–lisinopril complex was solved using MAD and Multiple Isomorphous Replacement with Anomalous Scattering (MIRAS) procedures. The position of the zinc atom was identified using anomalous difference Patterson maps calculated using diffraction data at peak wavelength. The MAD phases that we obtained were not very strong, and therefore additional phase information was obtained using MIRAS procedures with platinum (K2PtCl4), palladium (K2PdCl4) and osmium (OsCl3)
© 2003 Nature Publishing Group
553
letters to nature heavy-atom derivatives. The zinc site was used to obtain the starting phases for each derivative. Double difference Fourier maps calculated using the CCP4 program25 gave the first major heavy-atom site, and the phases from the combined zinc and heavy-atom sites were used to locate additional major/minor sites for each derivative. All heavy-atom sites and the zinc site were refined to 2.8 A˚ resolution using both MLPHARE25 and SHARP26. The overall figure of merit from SHARP was 0.38 (at 2.8 A˚) and was improved to 0.89 (at 2.0 A˚) by iterative solvent flattening, phase combination, and phase extension with the program SOLOMON27. We carried out model building and refinement using the programs O and CNS28, respectively. During the final stages of refinement, water molecules, zinc ion, acetate ion and the inhibitor molecule were inserted in the respective structures. Both the native and inhibitor complex structures were refined at 2.0 A˚ (see Supplementary Information) with 94% of the residues in the maximum allowed region and none in the disallowed region of the Ramachandran map. The final structures of ACE and the ACE–lisinopril complex had an R free value of 22.08% and 21.88%, and a final R cryst value of 18.29% and 18.14%, respectively. Both models have r.m.s. deviations in bond length of 0.005 A˚ and in bond angles of 1.28. Diagrams were computed using MOLSCRIPT25, RASTER3D29 (Figs 2a and 3a, b), POVRAY (Fig. 1a; see http:// www.povray.org) and GRASP30 (Fig. 1b).
16.
17. 18. 19. 20. 21. 22.
23.
Received 21 October; accepted 17 December 2002; doi:10.1038/nature01370. Published online 19 January 2003.
24.
1. Cushman, D. W., Cheung, H. S., Sabo, E. F. & Ondetti, M. A. Design of potent competitive inhibitors of angiotensin-converting enzyme. Carboxyalkanoyl and mercaptoalkanoyl amino acids. Biochemistry 16, 5484–5491 (1977). 2. Brown, C. K. et al. Structure of neurolysin reveals a deep channel that limits substrate access. Proc. Natl Acad. Sci. USA 98, 3127–3132 (2001). 3. Arndt, J. W. et al. Crystal structure of a novel carboxypeptidase from the hyperthermophilic archaeon Pyrococcus furiosus. Structure 10, 215–224 (2002). 4. Soubrier, F. et al. Two putative active centers in human angiotensin I-converting enzyme revealed by molecular cloning. Proc. Natl Acad. Sci. USA 85, 9386–9390 (1988). 5. Williams, T. A., Corvol, P. & Soubrier, F. Identification of two active site residues in human angiotensin I-converting enzyme. J. Biol. Chem. 269, 29430–29434 (1994). 6. Wei, L., Clauser, E., Alhenc-Gelas, F. & Corvol, P. The two homologous domains of human angiotensin I-converting enzyme interact differently with competitive inhibitors. J. Biol. Chem. 267, 13398–13405 (1992). 7. Junot, C. et al. RXP 407, a selective inhibitor of the N-domain of angiotensin I-converting enzyme, blocks in vivo the degradation of hemoregulatory peptide acetyl-Ser-Asp-Lys-Pro with no effect on angiotensin I hydrolysis. J. Pharmacol. Exp. Ther. 297, 606–611 (2001). 8. Esther, C. R. et al. The critical role of tissue angiotensin-converting enzyme as revealed by gene targeting in mice. J. Clin. Invest. 99, 2375–2385 (1997). 9. Ehlers, M. R. W., Fox, E. A., Strydom, D. J. & Riordan, J. F. Molecular cloning of human testicular angiotensin-converting enzyme: the testis isozyme is identical to the C-terminal half of endothelial angiotensin-converting enzyme. Proc. Natl Acad. Sci. USA 86, 7741–7745 (1989). 10. Yu, X. C. et al. Identification of N-linked glycosylation sites in human testis angiotensin-1 converting enzyme and expression of an active deglycosylated form. J. Biol. Chem. 272, 3511–3519 (1997). 11. Chubb, A. J., Schwager, S. L. U., Woodman, Z. L., Ehlers, M. R. W., Sturrock, E. D.. Biochem. Biophys. Res. Commun. 297, 1225–1230 (2002). 12. Ehlers, M. R. W. & Riordan, J. F. Angiotensin-converting enzyme: zinc- and inhibitor-binding stoichiometries of the somatic and testis isozymes. Biochemistry 30, 7118–7126 (1991). 13. Bu¨nning, P. & Riordan, J. F. Activation of angiotensin converting enzyme by monovalent anions. Biochemistry 22, 110–116 (1983). 14. Shapiro, R., Holmquist, B. & Riordan, J. F. Anion activation of angiotensin converting enzyme: dependence on nature of substrate. Biochemistry 22, 3850–3857 (1983). 15. Jaspard, E., Wei, L. & Alhenc-Gelas, F. Differences in the properties and enzymatic specificities of the
554
25. 26. 27. 28. 29. 30.
two active sites of angiotensin I-converting enzyme (Kininase II). J. Biol. Chem. 268, 9496–9503 (1993). Liu, X., Fernandez, M., Wouters, M. A., Heyberger, S. & Husain, A. Arg-1098 is critical for the chloride dependence of human angiotensin-1 converting enzyme C-domain catalytic activity. J. Biol. Chem. 276, 33518–33525 (2001). Holm, L. & Sander, C. Protein folds and families: sequence and structure alignments. Nucleic Acids Res. 27, 244–247 (1999). Corvol, P. & Williams, T. A. in Handbook of Proteolytic Enzymes (eds Barrett, A. J., Rawlings, N. D. & Woessner, J. F.) 1066–1076 (Academic, London, 1998). Skidgel, R. A., Engelbrecht, S., Johnson, A. R. & Erdos, E. G. Hydrolysis of substance P and neurotensin by converting enzyme and neutral endopeptidase. Peptides 5, 769–776 (1984). Patchett, A. A. et al. A new class of angiotensin-converting enzyme inhibitors. Nature 288, 280–283 (1980). Patchett, A. A. & Cordes, E. H. The design and properties of N-carboxyalkyldipeptide inhibitors of angiotensin converting enzyme. Adv. Enzymol. 57, 1–84 (1985). Bernstein, K. E., Welsh, S. L. & Inman, J. K. A deeply recessed active site in angiotensin converting enzyme is indicated from the binding characteristic of biotin-spacer-inhibitor reagents. Biochem. Biophys. Res. Commun. 167, 310–317 (1990). Bunning, P., Holmquist, B. & Riordan, J. F. Functional residues at the active site of angiotensin converting enzyme. Biochem. Biophys. Res. Commun. 83, 1442–1449 (1978). Otwinowski, M. & Minor, W. Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307–326 (1997). Collaborative computational project Number 4 The CCP4 suite: programs for protein crystallography. Acta Crystallogr. D 50, 760–763 (1994). De La Fortelle, E. & Bricogne, G. Maximum-likelihood heavy-atom parameters refinement in the MIR and MAD methods. Methods Enzymol. 276, 472–494 (1997). Abrahams, J. P. & Leslie, A. G. W. Methods used in structure determination of bovine mitochondrial F1 ATPase. Acta Crystallogr. D 52, 110–119 (1996). Bru¨nger, A. T. et al. Crystallography & N. M. R. system: a new software suite for macromolecular structure determination. Acta Crystallogr. D 54, 905–921 (1998). Merritt, E. A. & Bacon, D. J. Raster 3D: photorealistic molecular graphics. Methods Enzymol. 277, 505–524 (1997). Nicholls, A., Sharp, K. A. & Honig, B. Protein folding and association: insights from the interfacial and thermodynamic properties of hydrogen atoms. Proteins Struct. Funct. Genet. 11, 281–296 (1991).
Supplementary Information accompanies the paper on Nature’s website (ç http://www.nature.com/nature). Acknowledgements This work was supported by a Wellcome Trust grant to K.R.A., a Collaborative Research Initiative Grant to E.D.S. and K.R.A., and a National Research Foundation Grant to E.D.S. We thank M. Walsh, H. Belrahli and A. Thompson at ESRF, and L. Duke, J. Nicholson, M. McDonald, P. Rizkallah and M. Papiz at SRS for their help during X-ray data collection. We also thank R. Shapiro, J. Riordan and M. Ehlers for constructive criticism of the manuscript. Competing interests statement The authors declare that they have no competing financial interests. Correspondence and requests for materials should be addressed to K.R.A. (e-mail:
[email protected]) or E.D.S. (e-mail:
[email protected]). The X-ray coordinates of tACE and the tACE–lisinopril complex have been deposited in the Protein Data Bank under entry codes 1O8A and 1O86, respectively.
© 2003 Nature Publishing Group
NATURE | VOL 421 | 30 JANUARY 2003 | www.nature.com/nature
