Structure of -lactam synthetase reveals how to synthesize antibiotics instead of asparagine

The enzyme -lactam synthetase (-LS) catalyzes the formation of the -lactam ring in clavulanic acid, a clinically important -lactamase inhibitor. Whereas the penicillin -lactam ring is generated by isopenicillin N synthase (IPNS) in the presence of ferrous ion and dioxygen, -LS uses ATP and Mg²⁺ as cofactors. According to sequence alignments, -LS is homologous to class B asparagine synthetases (AS-Bs), ATP/Mg²⁺-dependent enzymes that convert aspartic acid to asparagine. Here we report the first crystal structure of a -LS. The 1.95 Å resolution structure of Streptomyces clavuligerus -LS provides a fully resolved view of the active site in which substrate, closely related ATP analog ,-methyleneadenosine 5'-triphosphate (AMP-CPP) and a single Mg²⁺ ion are present. A high degree of substrate preorganization is observed. Comparison to Escherichia coli AS-B reveals the evolutionary changes that have taken place in -LS that impede interdomain reaction, which is essential in AS-B, and that accommodate -lactam formation. The structural data provide the opportunity to alter the synthetic potential of -LS, perhaps leading to the creation of new -lactamase inhibitors and -lactam antibiotics.

Antibiotic resistance is a serious and growing problem in human health^{1, 2}. A principal bacterial resistance strategy is inactivation of -lactam antibiotics by -lactamases. These enzymes hydrolyze front line antibiotics, such as the penicillins and cephalosporins, destroying their ability to inhibit cell wall biosynthesis^{1, 3}. Clavulanic acid, a weak antibiotic isolated from Streptomyces clavuligerus, is a potent inhibitor of -lactamases and is used clinically in combination with amoxycillin and other penicillins^{4, 5}. The key -lactam ring of clavulanic acid is generated by a recently characterized enzyme, -lactam synthetase (-LS)^{6, 7, 8}, via a mechanism completely distinct from formation of the penicillin -lactam ring (Fig. 1). For penicillin, the enzyme isopenicillin N synthase (IPNS) catalyzes cyclization of a tripeptide to the bicyclic penicillin nucleus in the presence of ferrous ion and dioxygen^{9, 10}. In contrast, -LS converts N²-(carboxyethyl)-l-arginine (CEA) to deoxyguanidinoproclavaminic acid (DGPC) using ATP and Mg²⁺ (refs 6,7). A homologous enzyme, CarA, is proposed to perform similar chemistry in the biosynthesis of carbapenems^{11, 12}. These enzymes could be altered by genetic methods to improve production of clinically useful -lactamase inhibitors or to biosynthesize new -lactam antibiotics with enhanced resistance characteristics.

Figure 1. Reactions catalyzed by -LS, IPNS and AS-B. In clavulanic acid biosynthesis, N2-(carboxyethyl)-l-arginine (CEA) is cyclized to deoxyguanidinoproclavaminic acid (DGPC) by -LS. Subsequent chemical transformations by other enzymes result in the formation of clavulanic acid. In penicillin biosynthesis, IPNS catalyzes the cyclization of -(l-aminoadipoyl)-l-cysteinyl-d-valine (ACV) to isopenicillin N. In asparagine biosynthesis, glutamine and aspartic acid are converted to glutamic acid and asparagine. Full Figure and legend (15K)

Although -LS is an amide-synthesizing enzyme, its amino acid sequence does not resemble tRNA synthetases or nonribosomal peptide synthetases but is similar to class B asparagine synthetases (AS-Bs)^{13, 14}. These ATP/Mg²⁺-dependent enzymes, which belong to the broader family of Ntn glutamine amidotransferases¹⁵, catalyze transfer of the amide nitrogen of glutamine to aspartic acid, producing asparagine (Fig. 1). AS-B from Escherchia coli comprises two domains, each with an active site. Its N-terminal glutaminase domain hydrolyzes glutamine to glutamic acid and ammonia, and the C-terminal synthetase domain catalyzes the adenylation of aspartic acid and its subsequent reaction with ammonia. An extended tunnel provides a conduit for ammonia between the two sites¹⁶. -LS has homology to both domains but lacks the critical N-terminal Cys residue conserved in all AS-Bs¹⁷. Both -LS and AS-B mediate adenylation of an amino acid, but, in contrast to AS-B, -LS catalyzes an intramolecular rather than intermolecular amide bond formation. The basis for the apparent close evolutionary relationship is not clear. Detailed structural characterization of -LS is essential to exploit its potential in engineered antibiotic biosynthesis and to understand the connection to AS-B. Here we report the 1.95 Å resolution structure of Streptomyces clavuligerus -LS in the presence of substrate, CEA, and cofactors, close ATP analog , -methyleneadenosine 5'-triphosphate (AMP-CPP) and a single Mg²⁺ ion. In the crystal structure, these substrates are preorganized to favor the adenylation and -lactamization reactions catalyzed by -LS. In addition, comparison to E. coli AS-B reveals the evolutionary changes that have occurred in -LS to accommodate -lactam formation and provides new insight into the active site of AS-B.

-LS comprises two distinct domains, an N-terminal domain including residues 1−210 and a C-terminal domain composed of residues 211−513 (Fig. 2a). The N-terminal domain consists of two antiparallel -sheets, one five-stranded and the other six-stranded. Three short -helices pack against the six-stranded sheet. The outer surface of the five-stranded sheet is covered by two loop regions, the longer of which encompasses 28 residues and is preceded by a short -helix. Several residues in these loop regions are disordered and could not be modeled (see Methods). The linker to the C-terminal domain extends from the last strand in the six-stranded sheet. The C-terminal domain, which houses the active site, is composed of 11 -helices surrounding a five-stranded parallel -sheet. The active site is located in a cleft formed by four -strands and five -helices. In both molecules in the asymmetric unit, part of a loop covering the active site is disordered. The interface between the two domains is quite extensive, with 1,700 Ų buried surface area per domain. Interactions at this interface include several hydrophobic patches and two salt bridges linking Glu 78 with Arg 343 and Lys 159 with Glu 317.

Figure 2. Comparison of -LS and AS-B. a, Stereo view of -LS. Residues 21−24, 165, 166 and 445−453 have not been modeled. The bound CEA and AMP-CPP in the C-terminal domain are shown as stick representations, and the Mg2+ ion is depicted as a yellow sphere. Regions that differ significantly from the corresponding regions in AS-B are colored magenta and labeled by amino acid residue number. b, Stereo view of AS-B in the same orientation as -LS in (a). Regions that differ significantly between the two enzymes are colored magenta. Glutamine present in the N-terminal domain and AMP present in the C-terminal domain are shown as stick representations. c, Surface representation of the -LS active site color coded according to electrostatic potential: red corresponds to −80 kT; white, 0 kT; and blue, +80 kT. CEA and AMP-CPP are shown as stick representations, and the Mg2+ ion is obscured from view. The negatively charged region surrounding the CEA Arg side chain corresponds to Glu 382 and Asp 373. d, Surface representation of the AS-B active site color coded according to electrostatic potential: red corresponds to -80 kT; white, 0 kT; and blue, +80 kT. AMP is shown as a stick representation, and a bound uranium ion is shown as a yellow sphere. Negatively charged residues in the active site include Asp 351, Glu 352 and Asp 384. Full Figure and legend (129K)

The two -LS molecules in the asymmetric unit form a dimer with a buried surface area of 1,350 Ų per molecule. Two pairs of hydrogen bonds separated by 30 Å constitute the main intermolecular interactions. In each dimer, Tyr 109 and His 112 from one molecule interact with Glu 412 and Thr 368 from the second molecule. The interface size is well within the range observed for stable homodimers¹⁸, consistent with gel filtration and native gel analyses indicating that -LS is a dimer in solution at high concentrations (M.T.M. and A.C.R., unpublished data). The related enzyme AS-B, which functions as a dimer, dimerizes using a completely different interface¹⁶. The two molecules in the -LS asymmetric unit are very similar and can be superimposed with a root mean square (r.m.s.) deviation of 0.7 Å for C atom coordinates. The most prominent differences between the two -LS monomers in the asymmetric unit occur in residues 66−70 and residues 454−468. These deviations are due to crystal packing interactions. Residues 66−70 from both -LS monomers are involved in crystal lattice contacts, but the contacts are completely different for the two molecules. Residues 454−468 from only one -LS molecule participate in lattice interactions, accounting for the observed differences in this region.

As anticipated by sequence alignments⁶, -LS is structurally similar to E. coli AS-B. The two enzymes share many secondary and tertiary structural features, including distinct N- and C-terminal domains and a common overall fold (Figs 2, 3). However, a number of changes consistent with the different functions of the two enzymes are apparent. The N-terminal domain of -LS is substantially less organized than the corresponding glutaminase domain of AS-B (Fig. 2a,b). In AS-B, the critical N-terminal Cys residue, which was replaced with an Ala residue in the crystal structure, is buried at the beginning of the first -sheet and positioned precisely to attack the -amide of the glutamine substrate¹⁶. In -LS, this residue is replaced with Phe and preceded by nine additional residues. These extra residues fill the space occupied by the glutamine substrate in AS-B and are accommodated by a large outward shift in the loop spanning residues 52−61. In AS-B, the glutamine is stabilized by hydrogen bonding interactions with Arg 50, Asn 75, Glu 77 and Asp 99. Only two of these residues are conserved in -LS. The -LS equivalent to AS-B Glu 77, Glu 78, is hydrogen bonded to Arg 343, and the -LS equivalent to AS-B Asp 99, Asp 98, interacts with the amide nitrogen of Ile 79. These changes render the N-terminal domain of -LS chemically inactive, consistent with kinetic experiments showing that catalysis is unaffected by the addition of glutamine⁶. Other major differences between the N-terminal domains of -LS and AS-B are observed in regions distant from the glutamine binding site. The -LS -strands spanning residues 145−154 and 195−202 are shorter and longer, respectively, than their AS-B counterparts and are tipped inward. In addition, a large loop involving residues 163−183 is not conserved in AS-B, and the somewhat disordered loop encompassing -LS residues 20−29 is a helix in AS-B.

Figure 3. Structure-based sequence alignment of -LS and AS-B. The positions of the -LS secondary structure elements are superimposed on the -LS sequence, and the positions of the AS-B secondary structure elements are superimposed on the AS-B sequence. Helices are shown as green cylinders, and -strands are shown as blue arrows. Residues that are not observed in the crystal structures are highlighted in red. Full Figure and legend (69K)

The core regions of the C-terminal synthetase domains of -LS and AS-B are very similar, but a number of key differences are observed. In both enzymes, the active site region is formed by four -strands and five -helices (Fig. 2a,b). However, the -LS substrate binding cleft is relatively elongated (Fig. 2c,d). This difference in size is not surprising because the -LS binding site must accommodate CEA, which is substantially longer (12 Å) than aspartic acid (5 Å). The expansion of the active site is primarily due to the presence of a loop at residues 377−390 rather than the helix at the equivalent region in AS-B (Fig. 2a,b). In particular, residues 377−384 from -LS are shifted away from the cleft, and most of their bulky side chains, including Phe 377, Asp 378 and Asn 381, point away from the active site. The space vacated by these residues in -LS is filled in AS-B by the side chains of Leu 380, Tyr 383, Asp 384 and Arg 387. In AS-B, the visible part of the C-terminus is located at the domain interface adjacent to the helix containing these residues. The C-terminal region of -LS (residues 454−508), which ends in a surface helix, is completely different and does not restrict the position of residues 377−384 in a similar way. The two active sites in AS-B are connected by a molecular tunnel¹⁶. Cavity analysis of the -LS structure with the program VOIDOO¹⁹ indicates that a similar tunnel is not present, consistent with the absence of an N-terminal active site. Furthermore, a number of highly conserved residues that line the AS-B tunnel are not conserved in -LS.

The -LS active site

The -LS active site contains a full set of ligands, including the substrate CEA, the nonreactive ATP analog, AMP-CPP, and a Mg²⁺ ion (Fig. 4a−c). The guanidino group of CEA is situated between two negatively charged residues, Glu 382 and Asp 373. One CEA -nitrogen is hydrogen bonded to the side chain of Glu 382, and both -nitrogens are linked to the carbonyl oxygen of Asp 373 via a water molecule. The CEA Arg side chain is also held in position by the side chain of Tyr 326. The -carboxylate of CEA is stabilized by interactions with the amide nitrogen of Gly 349 and with a water molecule that is coordinated to the Mg²⁺ ion. This water molecule is also hydrogen bonded to the side chain of Asp 253, the carbonyl oxygen of Gly 347 and one of the -phosphate oxygens. The carboxyethyl group is not as well ordered as the rest of the CEA molecule, with B-values of 60 Ų as compared to 50 Ų for the -carboxylate and guanidino groups and 40 Ų for the AMP-CPP. Based on the visible electron density, the carboxyethyl group in monomer A was modeled in the eclipsed conformation, with the -carboxylate and -amino groups approximately lined up with one another, consistent with the proposed chemical mechanism of intramolecular -lactam formation⁷. The density in monomer B is less well defined, and several conformations of the carboxyethyl group are possible. The AMP-CPP−Mg²⁺ is very well defined in the electron density map. It is bound in close proximity to the CEA, with the -phosphate presenting itself for attack by the -carboxylate of CEA. The Mg²⁺ ion is coordinated by one side oxygen each from Asp 253 and Asp 351; by the -, - and -phosphate oxygens; and by a water molecule in a well-organized octahedral geometry. The phosphate oxygens also interact with Lys 423, Lys 443 and the side chain hydroxyl groups of Ser 249 and Ser 254. The adenosine hydroxyl groups interact with the amide nitrogens of Gly 347 and Tyr 348 and the carbonyl oxygen of Val 247. The adenine nitrogens are hydrogen bonded to the carbonyl oxygen and amide nitrogen of Met 273.

Figure 4. The active site. a, Stereo view of an Fo − Fc simulated annealing omit map contoured at 2 with the CEA, AMP-CPP and Mg2+ omitted. b, Stereo view of the -LS active site. Key residues that interact with the CEA, AMP-CPP and Mg2+ are shown. c, Schematic diagram of the -LS active site showing hydrogen bonding interactions. d, Schematic diagram of Asp, AMP-CPP and Mg2+ modeled into the AS-B active site. Predicted hydrogen bonding interactions and Mg2+ coordination environment are shown. e, Cyclization step in proposed chemical mechanism for -LS. The position of the CEA -amino group in the crystal structure favors this -lactamization reaction. Full Figure and legend (95K)

The active site details revealed by the -LS structure also provide new insight into the active sites of Ntn glutamine amidotransferases. In particular, the -LS active site can be used to model AMP-CPP−Mg²⁺ and Asp into the AS-B structure. One AMP molecule and three uranyl ions were found in the AS-B active site¹⁶. Notably, one of the uranyl ions occupies the same position as the Mg²⁺ ion in -LS and is also coordinated by two Asp residues, Asp 238 and Asp 351. By placing Mg²⁺ in the position occupied by this uranyl ion and using the existing AMP as a guide, Asp and AMP-CPP−Mg²⁺ were modeled into the E. coli AS-B active site (Fig. 4d). In this model, the Asp is situated similarly to the CEA in -LS, with its -carboxylate adjacent to the -phosphate of AMP-CPP. The amino nitrogen is hydrogen bonded to Asp 384, and the -carboxylate oxygen interacts with Arg 387. These two residues protrude into the AS-B active site cleft, filling part of the region occupied by CEA in -LS. A BLAST²⁰ analysis of the AS-B sequence reveals that both Asp 384 and Arg 387 are highly conserved in all AS-Bs. The aspartic acid can be positioned so that it interacts with Lys 376 instead of Arg 387, but this Lys residue is not conserved, suggesting that it is less likely to play an important role in substrate positioning. The model also predicts that Ser 234, Ser 239, Lys 429 and Lys 449 interact with the phosphate oxygens of the AMP-CPP. These findings may be useful in designing new inhibitors of AS-B, which is a therapeutic target for acute lymphoblastic leukemia (ALL)²¹.

The unreacted substrates in the -LS active site are bound in a geometry that favors both the adenylation and -lactamization reactions catalyzed by this enzyme. The proposed chemical mechanism involves adenylation of the CEA -carboxylate followed by cyclization via an oxoanion intermediate⁷ (Fig. 4e). The -carboxylate and -amino group of CEA are positioned relative to their respective electrophiles, the -phosphate of AMP-CPP and the hypothetical adenylate in a plausible near-attack conformation²². Notably, the -amine is oriented at a favorable angle with respect to the sp² -carboxylate, approaching the 110° Bürgi-Dunitz trajectory²³ for nucleophilic addition to a carbonyl group. As the reaction progresses, the nascent adenylate carbonyl can be bound in an oxoanion stabilizing Mg²⁺/H₂O environment, orienting the adenylate and favoring formation of the tetrahedral intermediate required for intramolecular closure of the -lactam bond (Fig. 4e). The striking preorganization of substrate molecules in the active site maximizes the energetic benefit from the hydrolysis of ATP in the formation of the strained four-membered ring. The structural features of -LS that impede attack by external nucleophiles coupled with the enlarged binding pocket provide an excellent starting point for protein engineering experiments to construct modified -lactam antibiotics.

Purification of -LS for crystallization.

-LS for initial crystallization trials and selenomethionine-substituted -LS were purified as described⁷. To generate the selenomethionine protein, the expression vector pBOB1 was transformed into E. coli B834(DE3), and transformed bacteria were grown in LeMaster media²⁴ supplemented with 50 mg l^-1 selenomethionine. The gene for S. clavuligerus -LS was also cloned into a pET24a(+) vector, which was used to transform E. coli BL21(DE3) cells. These transformed cells were grown in a 12 l fermenter, and protein expression was induced with 0.5 mM IPTG at an optical density of 0.6−0.8. Cells were harvested after 3 h, lysed with hen egg lysozyme and treated with DNase. After centrifugation for 30 min at 12,000 g, the supernatant was diluted 10-fold in 50 mM HEPES, pH 8.5, and loaded onto a Source 30Q column. A 0−330 mM KCl gradient was applied, and -LS eluted at 25 mM KCl. Fractions containing -LS were diluted 10-fold in 50 mM HEPES, pH 7.5, loaded onto an arginine Sepharose column and eluted with a step gradient of 5 mM KCl followed by 200 mM KCl. Fractions from the arginine Sepharose column were then applied to a Blue Sepharose column. The protein, which did not bind to the column, was concentrated and further purified on Superdex 200 equilibrated in 50 mM HEPES, pH 7.5, containing 200 mM KCl. The final purified material was concentrated to 20 mg ml^-1 in 50 mM Tris, pH 7.5, and used to grow crystals for substrate soaking experiments.

All -LS crystals were obtained by the hanging drop vapor diffusion method at room temperature using a precipitant solution containing 7% (w/v) PEG 4000, 200 mM MgCl₂, 7% (v/v) glycerol and 80 mM Tris, pH 8.0. Crystals with average dimensions of 0.01 0.05 0.20 mm³ appeared within 16 h and were transferred to a cryosolution composed of 24% (w/v) PEG 4000, 260 mM MgCl₂, 20% (v/v) glycerol and 80 mM Tris, pH 8.0. After 5 min, the crystals were frozen directly in liquid nitrogen. The substrate CEA was synthesized by acid hydrolysis of deoxyguanidinoproclavaminic acid (DGPC) as described²⁵. Soaking experiments were performed in the cryosolution supplemented with 5 mM each of CEA and the ATP analog ,-methyleneadenosine 5'-triphosphate (AMP-CPP). After 1 h, the crystals were frozen in liquid nitrogen. The -LS crystals belong to space group P2₁ with unit cell dimensions a = 61.0 Å, b = 98.2 Å, c = 81.3 Å and = 91.27°. All data were collected using synchrotron radiation and processed with DENZO and SCALEPACK²⁶ (Table 1).

Table 1. Crystallographic statistics Full Table

Data for MAD phasing were collected at three wavelengths using the reverse beam technique (Table 1). Selenium positions were determined with CNS²⁷, which located 11 of 14 possible sites in the asymmetric unit. MAD phasing and density modification with CNS²⁷ yielded an interpretable electron density map. The asymmetric unit contains two -LS molecules (labeled A and B) related by a noncrystallographic two-fold axis. The program XtalView²⁸ was used for model building, and the structure was refined with CNS²⁷ by iterative cycles of simulated annealing and individual B-value refinement. Noncrystallographic symmetry restraints were imposed until the final cycles of refinement to 2.3 Å resolution. This model was then used as a starting model for refinement of the substrate structure (Table 1). The final model for the substrate structure contains residues A4−A20, A25−A164, A167−A444, A454−A508, B2−B21, B24−B444 and B450−B508; 622 water molecules; five glycerol molecules; two AMP-CPP molecules; two CEA molecules; and two Mg²⁺ ions. No electron density was present for residues A21−A24, A165−A166, A445−A453, B22−B24 and B445−B449. According to Ramachandran plots generated with PROCHECK²⁹, the model exhibits good geometry with 92.7% of the residues in the most favored regions and all other residues in the additionally allowed regions. Figures were generated with MOLSCRIPT³⁰, Raster3D³¹, GRASP³² and BOBSCRIPT³³. Accessible surface area calculations were performed with the CCP4 program AREAIMOL³⁴.

The coordinates of -LS have been deposited in the Protein Data Bank (accession code 1JGT).

	Top

1. Neu, H.C. Science 257, 1064−1073 (1992). | PubMed | ISI | ChemPort |
2. Levy, S.B. N. Engl. J. Med. 338, 1376−1378 (1998). | Article | PubMed | ISI | ChemPort |
3. Walsh, C. Nature 406, 775−781 (2000). | Article | PubMed | ISI | ChemPort |
4. Baggaley, K.H., Brown, A.G. & Schofield, C.J. Nat. Prod. Rep. 14, 303−333 (1997).
5. Jensen, S.E. & Paradkar, A.S. Antonie van Leeuwenhoek 75, 125−133 (1999). | Article | PubMed | ISI | ChemPort |
6. Bachmann, B.O., Li, R. & Townsend, C.A. Proc. Natl. Acad. Sci. USA 95, 9082−9086 (1998). | Article | PubMed | ChemPort |
7. Bachmann, B.O. & Townsend, C.A. Biochemistry 39, 11187−11193 (2000). | Article | PubMed | ISI | ChemPort |
8. McNaughton, H.J. et al. Chem. Commun. 21, 2325−2326 (1998). | Article |
9. Roach, P.L. et al. Nature 387, 827−830 (1997). | Article | PubMed | ISI | ChemPort |
10. Burzlaff, N.I. et al. Nature 401, 721−724 (1999). | Article | PubMed | ISI | ChemPort |
11. Li, R., Stapon, A., Blanchfield, J.T. & Townsend, C.A. J. Am. Chem. Soc. 122, 9296−9297 (2000). | Article | ISI | ChemPort |
12. McGowan, S.J., Bycroft, B.W. & Salmond, G.P.C. Trends Microbiol. 6, 203−208 (1998). | Article | PubMed | ISI | ChemPort |
13. Scofield, M.A., Lewis, W.S. & Schuster, S.S. J. Biol. Chem. 265, 12895−12902 (1990). | PubMed | ISI | ChemPort |
14. Richards, N.G.J. & Schuster, S.M. Adv. Enzymol. Relat. Areas Mol. Biol. 72, 145−198 (1998). | PubMed | ISI | ChemPort |
15. Zalkin, H. & Smith, J.L. Adv. Enzymol. Relat. Areas Mol. Biol. 72, 87−143 (1998). | PubMed | ISI | ChemPort |
16. Larsen, T.M. et al. Biochem. 38, 16146−16157 (1999). | Article | ISI | ChemPort |
17. Boehlein, S.K., Richards, N.G.J. & Schuster, S.M. J. Biol. Chem. 269, 7450−7457 (1994). | PubMed | ISI | ChemPort |
18. Jones, S. & Thornton, J.M. Proc. Natl. Acad. Sci. USA 93, 13−20 (1996). | Article | PubMed | ChemPort |
19. Kleywegt, G.J. & Jones, T.A. Acta Crystallogr. D 50, 178−185 (1994). | Article | PubMed | ISI | ChemPort |
20. Altschul, S.F. et al. Nucleic Acids Res. 25, 3389−3402 (1997). | Article | PubMed | ISI | ChemPort |
21. Chakrabarti, R. & Schuster, S.M. Int. J. Pediatr. Hematol. Oncol. 4, 597−611 (1997). | ISI |
22. Bruice, T.C. & Benkovic, S.J. Biochemistry 39, 6267−6274 (2000). | Article | PubMed | ISI | ChemPort |
23. Bürgi, H.B. & Dunitz, J.D. Acc. Chem. Res. 16, 153−161 (1983). | ISI |
24. Hendrickson, W.A., Horton, J.R. & LeMaster, D.M. EMBO J. 9, 1665−1672 (1990). | PubMed | ISI | ChemPort |
25. Elson, S.W. et al. J. Chem. Soc. Chem. Commun. 15, 1212−1214 (1993). | Article |
26. Otwinowski, Z. & Minor, W. Methods Enzymol. 276, 307−326 (1997). | Article | PubMed | ISI | ChemPort |
27. Brünger, A.T. et al. Acta Crystallogr. D 54, 905−921 (1998). | Article | PubMed | ISI |
28. McRee, D.E. J. Struct. Biol. 125, 156−165 (1999). | Article | PubMed | ISI | ChemPort |
29. Laskowski, R.A. J. Appl. Crystallogr. 26, 283−291 (1993). | Article | ISI | ChemPort |
30. Kraulis, P.J. J. Appl. Crystallogr. 24, 946−950 (1991). | Article | ISI |
31. Merritt, E.A. & Bacon, D.J. Methods Enzymol. 277, 505−524 (1997). | Article | ISI | ChemPort |
32. Nicholls, A., Sharp, K.A. & Honig, B. Proteins 11, 281−296 (1991). | PubMed | ISI | ChemPort |
33. Esnouf, R.M. J. Mol. Graph. Model. 15, 132−134 (1997). | Article | PubMed | ISI | ChemPort |
34. Collaborative Computational Project, Number 4. Acta Crystallogr. D 50, 760−763 (1994). | Article | ISI |

	Top