Ushkaryov/Wilson (letter) - Nature

© 2000 Nature America Inc. • http://structbio.nature.com

letters


1. Gorio, A., Rubin, L.L. & Mauro, A. J. Neurocytol. 7, 193–202 (1978). 2. Lang, J., Ushkaryov, Y., Grasso, A. & Wollheim, C.B. EMBO J. 17, 648–657 (1998). 3. Liu, J. & Misler, S. J. Neurosci. 18, 6113–6125 (1998). 4. Rosenthal, L. & Meldolesi, J. Pharmacol. Ther. 42, 115–134 (1989). 5. Grasso, A., Alema, S., Rufini, S. & Senni, M.I. Nature 283, 774–776 (1980). 6. Deri, Z., Bors, P. & Adam-Vizi, V. J. Neurochem. 60, 1065–1072 (1993). 7. Davletov, B.A. et al. EMBO J. 17, 3909–3920 (1998). 8. Finkelstein, A., Rubin, L.L. & Tzeng, M.C. Science 193, 1009–1011 (1976). 9. Mironov, S.L., Sokolov, Y.V., Chanturiya, A.N. & Lishko, V.K. Biochim. Biophys. Acta 862, 185–198 (1986). 10. Dulubova, I.E. et al. J. Biol. Chem. 271, 7535–7543 (1996). 11. Frontali, N. et al. J. Cell Biol. 68, 462–479 (1976). 12. Kiyatkin, N.I., Dulubova, I.E., Chekhovskaya, I.A. & Grishin, E.V. FEBS Lett. 270, 127–131 (1990). 13. Ushkarev, Iu.A. & Grishin, E.V. Bioorg. Khim. 12, 71–80 (1986). 14. Gouaux, E. Curr. Opin. Struct. Biol. 7, 566–573 (1997). 15. van Heel, M. Ultramicroscopy 21, 111–124 (1987). 16. Dubochet, J. et al. Quat. Rev. Biophys. 21, 129–228 (1988). 17. Meldolesi, J., Madeddu, L., Torda, M., Gatti, G. & Niutta, E. Neuroscience 10, 997–1009 (1983).

The structural basis for molecular recognition by the vitamin B12 RNA aptamer Django Sussman, Jay C. Nix and Charles Wilson Department of Biology and Center for the Molecular Biology of RNA, Sinsheimer Laboratories, University of California, Santa Cruz, California 95064, USA.

Previous solution structures of ligand-binding RNA aptamers have shown that molecular recognition is achieved by the folding of an initially unstructured RNA around its cognate ligand, coupling the processes of RNA folding and binding. The 3 Å crystal structure of the cyanocobalamin (vitamin B12) aptamer reported here suggests a different approach to molecular recog-

18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34.

Misler, S. & Hurlbut, W.P. Proc. Natl. Acad. Sci. U. S. A. 76, 991–995 (1979). Scheer, H. et al. J. Physiol. (Paris) 79, 216–221 (1984). Dube, P., Tavares, P., Lurz, R. & van Heel, M. EMBO J. 12, 1303–1309 (1993). Volynski, K.E., Nosyreva, E.D., Ushkaryov, Y.A. & Grishin, E.V. FEBS Lett. 442, 25–28 (1999). Batchelor, A.H., Piper, D.E., de la Brousse, F.C., McKnight, S.L. & Wolberger, C. Science 279, 1037–1041 (1998). Jacobs, M.D. & Harrison, S.C. Cell 95, 749–758 (1998). Bork, P. Proteins 17, 363–374 (1993). Madeddu, L. et al. J. Neurochem. 45, 1708–1718 (1985). Gilbert, R.J. et al. Cell 97, 647–655 (1999). Ichtchenko, K. et al. EMBO J. 17, 6188–6199 (1998). van Heel, M., Harauz, G. & Orlova, E.V. J. Struct. Biol. 116, 17–24 (1996). van Heel, M. Optik 82, 114–126 (1989). Schatz, M., Orlova, E.V., Dube, P., Jager, J. & van Heel, M. J. Struct. Biol. 114, 28–40 (1995). Harauz, G. & van Heel, M. Optik 73, 146–156 (1986). Radermacher, M. J. Elect. Microsc. Tech. 9, 359–394 (1988). van Heel, M. & Harauz, G. Optik 73, 119–122 (1986). Jones, T.A., Zou, J.Y., Cowan, S.W. & Kjeldgaard, M. Acta Crystallogr. A. 47, 110–119 (1991).

nition in which elements of RNA secondary structure combine to create a solvent-accessible docking surface for a large, complex ligand. Central to this structure is a locally folding RNA triplex, stabilized by a novel three-stranded zipper. Perpendicular stacking of a duplex on this triplex creates a cleft that functions as the vitamin B12 binding site. Complementary packing of hydrophobic surfaces, direct hydrogen bonding and dipolar interactions between the ligand and the RNA appear to contribute to binding. The nature of the interactions that stabilize complex formation and the possible uncoupling of folding and binding for this RNA suggest a strong mechanistic similarity to typical protein–ligand complexes. Structures of antibody–antigen, receptor–hormone and enzyme–substrate complexes have revealed many of the mechanisms by which proteins recognize small ligands. Almost invariably, secondary structure elements create a stable framework upon which the binding site is built. In contrast, a range of RNA–ligand complexes studied by NMR have shown that recognition of peptides, amino acids, antibiotics, cofactors and nucleotides can be accomplished by binding sites formed exclusively with local inter-

Fig. 1 Overview of the cyanocobalamin aptamer structure. Strands forming the triplex are colored dark blue (strand 1; nucleotides 7–11), cyan (strand 2; nucleotides 15–22) and yellow (strand 3; nucleotides 22–30). Strands forming the duplex are colored green (nucleotides 12–14) and magenta (nucleotides 31–33). Overhanging bases on the 5' and 3' ends are colored gray (the conformation of all of these nucleotides is defined by intermolecular contacts in the crystal lattice). Inset shows a stereo view of the tertiary structure using the same color scheme and with vitamin B12 colored red. Figure prepared using Ribbons29.

nature structural biology • volume 7 number 1 • january 2000

53


letters a


b

Fig. 2 Structural motifs that stabilize the aptamer include a three-stranded zipper and a pair of unusual three-nucleotide loops. a, As shown in this stereo view, nucleotides in the first strand of the triplex (U9, G10, C11) interdigitate with base pairs formed by the other strands to form a three-stranded zipper. Experimental electron density is shown for these nucleotides contoured at 1σ. U9 and A20 form a canonical Watson–Crick base pair. The backbone linkage between U9 and G10 is extended, allowing G10 to skip over the C19:G26 base pair to form a base triple with C18:G28. The backbone then stretches past the U15:A13:C29 triple, permitting C11 to form a parallel base pair with G30. Figure generated using Raster3D30. b, The loop between the first two strands in the triplex (G12, C13, A14) also serves as the 5' half of the duplex. The constrained geometry at this site brings three backbone phosphates (shown in red) into close contact with each other (G12, C13 and U15) in a conformation likely stabilized by the high concentration of lithium. A14 is held in position by dyad-symmetric hydrogen bonding with A31 and by perpendicular stacking on C15. The A14 backbone phosphate lies outside the cluster formed by the other loop nucleotides and is positioned to specifically hydrogen bond with a propionamide side group on the ligand (Fig. 4a). Figure generated using Raster3D30. c, Loop 2. Nucleotides U23 and A25 stack upon another, forcing the intervening nucleotide C24 to be flipped out into solution. The conformation of C24 is fixed in the crystal by its formation of a base triple with one of the stacked base pairs in the middle of the triplex. Figure generated using Ribbons29.

c

actions1–5. In these examples, RNA helices clamp together bulges or loops that are responsible for binding but themselves do not interact to create a stable tertiary structure. As a result, the RNAs in these complexes generally display ‘adaptive binding,’ whereby the ligand nucleates folding of a previously unstructured molecule6. It is unclear, however, if this behavior represents a truly general mechanism for molecular recognition by RNA. For example, ample low-resolution evidence from large biological ligand-binding RNAs (for example, the ribosome, the group I intron, the spliceosome7–9) implies the existence of elaborate tertiary structures in the absence of ligand or substrates. Analysis of the crystal structure of a vitamin B12–RNA aptamer complex reported here shows how a ligand-binding site may be built from a stable framework of RNA secondary structure elements. This structure suggests that, rather than folding to surround its ligand, this aptamer creates a large surface binding site that docks with one face of its ligand. As such, this aptamer–ligand complex more closely resembles typical protein–ligand complexes. Overview The B12 binder was evolved in an effort to demonstrate that RNA can bind a large, complex ligand10. With a dissociation constant of 90 nM, it is one of the highest affinity aptamer–small molecule complexes characterized to date. High concentrations of lithium chloride were introduced during the in vitro selection experiment in an effort 54

to minimize nonspecific ionic interactions, and lithium has subsequently been shown to be required for binding. Diffraction from previously described crystals11 of the affinity-purified vitamin B12–aptamer complex was phased by a three-wavelength multiwavelength anomalous dispersion (MAD) experiment. The structure has been refined to 3 Å resolution with an R-factor of 20.8%.

Fig. 3 Base triples in detail. Bases in the triplex and on the same plane are boxed in the secondary structure. The hydrogen bonding between the bases is shown as dashed lines in the inserted detail. Major groove triples are proximal to the binding site and are linked to the distal minor-groove triples by two base pairs (boxed in gray).



letters a


b

Although technically a two-helix pseudoknot, the first helix is actually nested within a much larger, seven-tiered triplex (Fig. 1) that accounts for most of the conserved nucleotides within the aptamer. Helix 1 was previously predicted on the basis of an in vitro phylogenetic experiment in which partially randomized aptamers were reselected for function10. Nucleotides that covary in the in vitro selection experiment are observed to base pair in the crystal structure. Three-stranded zipper stabilizes the triplex The end of the triplex that contacts the ligand is stabilized by a novel three-stranded zipper (Fig. 2a). Three consecutive bases in the first strand (U9, G10 and C11) adopt an extended conformation that allows interdigitation by nucleotides from the other two triplex strands (U15 and G26). Intercalation locks these nucleotides in place, positioning them to hydrogen bond with other triplex nucleotides. Coaxial packing of bases from three different strands into a single stack effectively zippers the strands in place. Insertion of a base between adjacent nucleotides has been observed previously in both the transfer RNA and the theophylline aptamer structures12–14. In both of these cases, intercalating bases are left unpaired and the interaction is limited to two strands. The vitamin B 12 aptamer structure shows how this simple stabilizing interaction can be expanded to assemble multiple elements within an RNA. Formation of the zipper has a pronounced effect on the structure. The packing of U15 under C11 locks the intervening nucleotides into a tight loop, positioning them to form helix 2 and defining this helix’s orientation relative to the triplex (Fig. 2b). Constraints on the folding force the bases of nucleotides A14 and C11 to pack against each other at right angles. Although this arrangement has not been observed previously


Fig. 4 Chemical modification data and details of the binding site a, Stereo view of chemical modification data in context of the structure. The RNA (gray) is shown in our standard orientation (binding site in the top left corner). Colored spheres correspond to atoms that are protected from (blue) or sensitive to (red) Dimethyl sulfate (DMS) modification under binding conditions but in the absence of ligand. Atoms that become protected from modification reagents after ligand binding are shown as yellow spheres. Nucleotides 1–4 were not in the original selected aptamer and have not been shown. b, Cofactor (gray) rests in a cleft generated by the perpendicular stacking of the helix (blue) against the triplex (magenta). Hydrogen bonds between the RNA and the cyanocobalamin amide side groups are indicated in yellow. Figure generated by MIDAS26,27. c, van der Waals surfaces (colored by atom type) for the RNA aptamer and the vitamin B12 cofactor indicate extensive complementarity at their interface. Generated by Raster3D30.

c

in RNA structures, it resembles the herringbone packing found in the crystals of benzene and in certain nucleotide analogs. Similar aromatic side chain packing also stabilizes several proteins15–17. The opposite end of the triplex contains an unusual turn that allows the third strand to pack against the first and second while leaving no intervening unpaired nucleotides. C22 and U23, defining the ends of strands 2 and 3 respectively, lie on the same plane, forming a platform stabilized largely by base stacking with the next tier in the triplex rather than by direct contacts between the two nucleotides. U23 is stacked on the following A25, forcing the intervening C24 to adopt a highly exposed, looped-out conformation (Fig. 2c)(unsurprisingly, C24 is one of the few nucleotides in the aptamer core that is not absolutely conserved). At the opposite end of strand 3, G30 forms an unusual parallel Watson–Crick base pair with C11 to cap the triplex. Base triples Most of the absolutely conserved nucleotides within the aptamer are involved in the formation of base triples that define the triplex. This domain can be broken down into a distal region made from minor-groove triples, a central region of stacked Watson–Crick base pairs, and a proximal region containing major groove triples. As illustrated in Fig. 3, triples are generally stabilized by interbase hydrogen bonding and, in some cases, ribose–ribose and ribose–base hydrogen bonds. Hydrogen bonding between bases is generally limited within each triplex tier, and instead, base stacking between tiers appears to be the key driving force for folding. The first tier of the triplex contains U23 and the G7:C22 base pair arranged to form a minor-groove base triple stabilized by a single weak hydrogen bond. Stacked directly above this tier is a

55


letters Table 1 Diffraction data and refinement statistics Data set Anomalous peak Inflection point High-energy remote Figure of merit to 3.0 Å SHARP DM R-factor Rcryst3 Rfree4 R.m.s.d. from ideality Bond lengths Bond angles

I (Å)

dmin (Å)

1.6047 1.6057 1.5842

3.0 3.0 3.0

Observed reflections 28,819 28,808 28,921

Unique reflections 7,480 8,039 7,216

Completeness (%) 90.2 90.2 91.0

/

Rsym (%)1

< PP > 2

13.7 12.8 13.5

6.0 6.2 6.5

2.07 1.60 1.91

0.52 0.73 20.7 24.8 0.021 Å 1.31°

Rsym= ΣhΣi|Ii(h) - |/Σh, where Ii(h) is the ith measurement and is the weighted mean of all measurements of I(h). Phasing power is calculated for acentric reflections as the mean value of the heavy atom structure factor amplitude divided by the residual lack of closure error. 3R cryst = 100 × Σ|Fo - Fc|/Σ|Fo|, where Fo and Fc are the observed and calculated structure factors, respectively. 4R free is the R-factor obtained for 10% of the reflections not included in the refinement. 1


2

second minor-groove triple formed by G8:C21 .A25. The 2'hydroxyls from G8 and A25 each make base-specific hydrogen bonds to A25 and C21, respectively, to hold this base in position. An identical hydrogen-bonding pattern is observed in a minorgroove triple in the beet western yellow virus ribosomal frameshifting pseudoknot18, suggesting that this arrangment may represent a common motif for docking adenosines into the minor groove of an RNA helix. While the portion of the triplex farthest from the ligand is defined by strand 3 pairing with the minor groove of strands 1 and 2, its opposite end is defined by pairing of strand 1 with the major groove of strands 2 and 3. Two base pairs separate the halves of the triplex, and mediate this switch. The U9:A20 pair, linking strands 1 and 2, is followed up the triplex axis by C19:G26 to join strands 2 and 3. Stacking between these base pairs is blocked in part by docking of the looped-out base of C24 from another molecule to form a minor-groove base triple with C19:G26. The major groove portion of the triplex consists of three tiers: G10.C18:G28, A17.C29.U15 and A16.C11:G30. Only the first of these appears substantially stabilized by hydrogen bonding between nucleotides. The Watson–Crick face of G10 interacts with the major-groove face of the C18:G28 base pair to form three hydrogen bonds. A related base triple in tRNAPhe (G46.G22:C13) has been previously observed although the unpaired guanosine in this case is shifted toward one side of the base pair and a different hydrogen bonding pattern is observed. Stacked above this stable triple, the A17.C29.U15 tier is remarkable in the sense that it lacks conventional base pairing. The O2 and N4 atoms of C29 hydrogen bond to A17 and U15, respectively, to form the only direct interactions between these nucleotides. Separating U15 and A17 is a channel that continues up the triplex to the bound cofactor. While this channel appears wide enough to accommodate water molecules, no specifically bound solvent is visible in the current 3 Å electron density maps. The separation between the U15 and A17 effectively increases the width of the triplex to more effectively match that of the vitamin B12 bound at its end. In the final tier, the bases of C11, G30 and A16 all lie in a common plane below the vitamin B12. Insertion of the cyanide ligand to the central cobalt of the cofac56

tor effectively blocks direct interaction of A16 with the C11:G30 parallel base pair. High concentrations of lithium are required for binding. The low electron density of lithium (2 e- / ion) makes it essentially invisible at 3 Å resolution and thus impossible to directly observe in the crystal structure. The small radius of the lithium cation also gives it the highest positive charge density among all monovalent ions. It forms especially stable coordination complexes with enolates19 and thus can effectively neutralize electrostatic repulsion between RNA nucleotides. Two sites within the aptamer, corresponding to the junctions between the triplex and the helix domains (Fig. 2b), bring nonbridging phosphate oxygens into contact with each other and seem likely to require stabilization by associated counterions. The geometry of the phosphate oxygens in each of these two junctions matches that expected for hydration of a lithium ion20. Similar packing of negatively charged phosphates has been observed in both the 5S ribosomal RNA and group I intron P4–P6 fragment 21,22, but in both instances, tightly bound divalent metals are instead responsible for their stabilization. Subtle steric differences between these examples of close phosphate packing and those in the vitamin B12 aptamer may account for the differences in ionic requirements for folding. Binding site at the triplex/helix junction Perpendicular packing of the helix and the triplex exposes the five nucleotides that cap these elements. Their arrangement creates a large hydrophobic patch, devoid of backbone phosphates and ringed by hydrogen bond donors and acceptors, that serves as the vitamin B12 binding site. Chemical modification data suggest that this site may be assembled before ligand binding. Primer extension of dimethyl sulfate (DMS) and diethyl pyrocarbonate (DEPC) treated RNA shows that the majority of nucleotides in the aptamer core are protected in the absence of vitamin B12 (ref. 10), suggesting that the unbound RNA exists as a folded structure. Virtually all of these protected sites (17 of 19), spanning both the triplex and the flanking helix, also appear buried in the crystal structure of the bound state (Fig 4a). Conversely, highly modified sites in the unbound RNA are generally solvent accessible in the crystal structure with the excepnature structural biology • volume 7 number 1 • january 2000



letters tion of nucleotides surrounding the vitamin B12 binding site. Ligand-induced protections thus can be explained by direct cofactor occlusion rather than by ligand-induced conformational changes. In the absence of an experimentally determined structure for the unbound RNA, we can speculate on its conformation using the chemical modification data as a guide. The possibility remains that two significantly different structures exist in the bound and unbound states with the same global pattern of chemical reactivity across most of the nucleotides in the aptamer core. However, the simplest interpretation of the data is that the unbound and bound states share a common overall fold. Hydrophobic packing, direct RNA–ligand hydrogen bonding and electrostatic interactions appear to contribute to binding. Methyl, acetamide, and propionamide side groups protruding from both faces of vitamin B12 prevent contact of its core pyrroline rings by the aptamer. Instead, the aliphatic portions of seven of these side groups make direct contacts with the aromatic faces and ribose rings of binding site nucleotides. Axially and equatorially oriented methyls lining one edge of the vitamin B12 alternately pack against either side of the cleft between the triplex and duplex (Fig. 4c). While hydrophobic packing contributes to binding in several aptamer–ligand complexes (for example, ATP, FMN, arginine), previously observed interactions have consisted largely of direct stacking between planar portions of the ligands and nucleotide bases2–4,23. In contrast, bases and riboses in the vitamin B12 aptamer assemble to create a featured hydrophobic surface with many bumps and pockets that complement the ligand. A wide range of conjugated ringcontaining molecules (such as hematoporphyrin, chlorophyllin and pthalocyanine), each of which lack the side groups found on the vitamin B12 corrin ring, fail to competitively elute the RNA from a vitamin B12 agarose column, suggesting that steric complementarity is essential for binding. The amide side chains of the cofactor also participate in binding by directly interacting with the aptamer (Fig 4b). RNA functional groups involved in hydrogen bonding encompass a wide assortment of donors and acceptors, including phosphate oxygens, purine imidazole nitrogens, a ribose 2'-hydroxyl and a ribose ring oxygen. The cofactor’s acetamide and propionamide side chains are chemically identical to those of asparagine and glutamine, amino acids that often direct protein binding to nucleic acids. As shown in Fig 4b, bifurcation of the nitrogen and oxygen at the end of the A-ring acetamide side chain allows it to form a pair of hydrogen bonds. This forked arrangement involving simultaneous interaction with the RNA backbone and a nucleotide base resembles interactions observed in the Rev response element (RRE)–Rev peptide complex24,25 and several protein–DNA complexes. The axial cyanide coordinated to vitamin B12 lies at the end of the triplex in a pocket created by C11, A16 and G30. The sphybridized cyanide nitrogen lone pair creates a strong dipole moment aligned with the triplex axis. An electrostatic potential generated in part by unsatisfied hydrogen bond donors on the aptamer complements this dipole, potentially providing an additional driving force for binding 26,27. The vitamin B12 aptamer–ligand crystal structure shows how constrained folding of an RNA juxtaposes structural elements to create a complementary surface for molecular recognition. It is striking in some ways that high-affinity binding can be


achieved while only recognizing only one face of the ligand. The number of conserved nucleotides in aptamer core matches the effective complexity of the pool from which it was derived 10, raising the possibility that more elaborate approaches to ligand recognition involving the entire cofactor could be accessed using more complex pools of random sequence RNAs. Methods Data collection. Affinity-purified RNA was crystallized as described11. The structure was determined using a single C2221 crystal with the vitamin B12 cobalt ion providing the anomalous scattering used for crystallographic phasing. All data were collected at 100 K on beamline 5.0.2 at the Advanced Light Source, Lawrence Berkeley National Laboratory. Phase determination and structure refinement. A threewavelength MAD data set was collected at the cobalt K edge, and automated analysis of the resulting anomalous Patterson map identified a solution with four scatterers in the asymmetric unit28. Four-fold noncrystallographic symmetry averaging, solvent flattening and histogram matching improved the initial phase estimates and allowed all 35 nucleotides and the B12 ligand to be identified and traced in the resulting map. An atomic model containing four independent copies of the aptamer–vitamin B12 complex without bound water molecules or ions refines with a free R-factor of 24.8% (R-factor = 20.7%). The root mean square deviation (r.m.s.d.) between noncrystallographic symmetry mates ranges from 0.5 to 0.8 Å for all atoms (excluding residues 1–6, which lie outside the aptamer core and form crystal contacts). Coordinates. The structural coordinates have been deposited in the Protein Data Bank (accession code 1DDY).

Correspondence should be addressed to D.S. email: [email protected] or C.W email: [email protected]. Received 13 August, 1999; accepted 16 November, 1999. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30.

Fourmy, D., Recht, M.I. & Puglisi, J.D. J. Mol. Biol. 277, 347–362 (1998). Dieckmann, T., Suzuki, E., Nakamura, G.K. & Feigon, J. RNA 2, 628–640 (1996). Fan, P., Suri, A.K., Fiala, R., Live, D. & Patel, D.J. J. Mol. Biol. 258, 480–500 (1996). Jiang, L. & Patel, D.J. Nature Struct. Biol. 5, 769–774 (1998). Puglisi, J.D., Chen, L., Blanchard, S. & Frankel, A.D. Science 270, 1200–1203 (1995). Patel, D. J. et al. J. Mol. Biol. 272, 645–664 (1997). Golden, B.L., Gooding, A.R., Podell, E.R. & Cech, T.R. Science 282, 259–264 (1998). Mueller, F. & Brimacombe, R. J. Mol. Biol. 271, 545–565 (1997). Furman, E. & Glitz, D.G. J. Biol. Chem. 270, 15515–15522 (1995). Lorsch, J.R. & Szostak, J.W. Biochemistry 33, 973–982 (1994). Sussman, D., Greensides, D., Reilly, K. & Wilson, C. Acta Crystallogr. D 55, 326–328 (199). Zimmermann, G.R., Jenison, R.D., Wick, C.L., Simorre, J.P. & Pardi, A. Nature Struct. Biol. 4, 644–649 (1997). Suddath, F.L. et al. Nature 248, 20–24 (1974). Robertus, J.D. et al. Nature 250, 546–551 (1974). Burley, S.K. & Petsko, G.A. Science 229, 23–28 (1985). Cox, E.G., Cruickshank, D.W.J. & Smith, J.A.S. Proc. Soc. London Ser. A 247 (1958). Kim, S.H. & Rich, A. J. Mol. Biol. 42, 87–95 (1969). Su, L., Chen, L., Egli, M., Berger, J.M. & Rich, A. Nature Struct. Biol. 6, 285–292 (1999). Streitwieser, A.J. & Heathcock, C.H. Introduction to organic chemistry (Macmillan, New York, New York; 1985). Maetzke, T. & Seebach, D. Organometallics 9, 3032–3037 (1990). Cate, J.H., Hanna, R.L. & Doudna, J.A. Nature Struct. Biol. 4, 553–558 (1997). Correll, C.C., Freeborn, B., Moore, P.B. & Steitz, T.A. Cell 91, 705–712 (1997). Jiang, F., Kumar, R.A., Jones, R.A. & Patel, D.J. Nature 382, 183–186 (1996). Ye, X., Gorin, A., Ellington, A.D. & Patel, D.J. Nature Struct. Biol. 3, 1026–1033 (1996). Battiste, J.L. et al. Science 273, 1547–1551 (1996). Ferrin, T.E., Huang, C.C., Jarvis, L.E. & Langridge, R. J. Mol. Graphics 6, 13–27 (1988). Bash, P.A., Pattabiraman, N., Huang, C.C., Ferrin, T.E. & Langridge, R. Science 222, 1325–1327 (1983). Terwilliger, T.C. & Berendzen, J. Acta Crystallogr. D 53, 571–579 (1997). Carson, M. & Bugg, C.E. J. Mol. Graphics 4, 121–122 (1986). Merritt, E.A. & Murphy, M.P.E. Acta Crystallogr. D 50, 869–873 (1994).

57