Inhibition of proteinase K by methoxysuccinyl-Ala-Ala-Pro-Ala ...

2 downloads 0 Views 7MB Size Report
(0 1991 by The American Society for Biochemistry and Molecular Biology, Inc. ..... “oxyanion hole” in blue, inhibitor in red and substrate recognition site in pink. ... circles. a-Helices are drawn as cylinders, &pleated sheet strands as ribbons, S-S.
Vol. 266, No. 26, Issue of September 15, pp. 17695-17699.1991 Printed in U. S.A.

THEJOURNAL OF BIOLOGICAL CHEMISTRY (0 1991 by The American Society for Biochemistry and Molecular Biology, Inc.

Inhibition of Proteinase K by Methoxysuccinyl-Ala-Ala-Pro-Alachloromethyl Ketone AN X-RAY STUDYAT

2.2-A RESOLUTION* (Received for publication, February 27, 1991)

Wojciech M. Wolfs, Jurgen Bajorath, Alexander Muller, Srinivasan Raghunathan, Tej P. Singh, Winfried Hinrichs, and Wolfram Saengers From the Institutf u r Kristallographie, Freie Universitat Berlin, Takustrasse6, 0-1000 Berlin 33, Germany

The crystal structure of the transition state analog oxy-Ala-Ala-chloromethyl ketone (4)and its Ala-Phe derivacomplex formed covalently between proteinase K and tive ( 5 ) were investigated by crystallographicmethodsto methoxysuccinyl-Ala-Ala-Pro-Ala-chloromethyl locate the active site of the enzyme and to characterize the ketone was determinFd by x-ray diffraction methods substrate binding site. at a resolution of 2.2 A and refined by constrained least In the present study, we elongated the inhibitor to map and squares to anR factor of 19.8%for the 11864 structure to saturate the substrate binding site of proteinase K. The amplitudes greater than la^. The chloromethyl ketone search for an optimum inhibitor was guided by manual model group is covalently linked with the active site func- building using FRODO (6) and an Evans and Sutherland tional groups Hises(N.)and Ser224(0,).The former has graphics unit and subsequent energy minimizationstudies substituted for chlorine and the latter has attacked the ( 7 ) .These studies indiusing the BIOSYM program package carbon of the ketone group, thereby forming the tetrahedral carbon atomof the transition state analog. The cate thata chloromethyl ketone inhibitor, bound to the active peptide part of the inhibitor is in an extended confor- site residues Hisfiga n d S e P 4in a way similar to the already mation and fills subsites S1 to S5 of the substrate known shorter inhibitors (4,5 ) , fits completely into the subrecognition site. Its backbone hydrogens bond with strate binding site even if the amino acid in the subsite S2 strands 100-104 and 132-136 of the substrate recog- (according to the Schechter andBerger nomenclature (8))is nition site as the central strand of a three-stranded a proline, followed by two to four amino acidswith small side antiparallel &pleated sheet. This sheet formation is chains. To test thissuggestion obtained from model building associated with a movement by -1 A of strand 100- studies,theinhibitory power againstproteinase K of the 104 which is probably associated with the insertion of commercially available N-terminal protected methoxysuccithe bulky proline side chain. The methoxysuccinyl nyl-Ala-Ala-Pro-Ala-chloromethyl ketone was investigated by group is stacked on the phenolic side chain of Tyrlo4 kinetic assays. Since it strongly inhibits proteinase K, the two that is a part of the bottom of the recognition site. molecules were cocrystallized and the structure of the complex Biochemical studies show that shorter inhibitors of this determined by x-ray diffraction methods. type are less effective than the longer one, because there are fewerhydrogen bonding and van der Waals/ MATERIALS ANDMETHODS stacking interactions.

Proteinase K was obtained from SERVA, Heidelberg, and purified by Sephadex-G50 gel filtration in 50 mM Tris-HCI, pH7.5, containing 1 mM CaC12 to remove autolytic fragments, as previously described (9), and lyophilized. The inhibition of proteases by specifically designed moleThe inhibitors Ala-Ala-COCHyC1 (I), Ala-Ala-Phe-COCH2C1 (II), cules is of particular interest in view of the involvement of Phe-Pro-Arg-COCH,Cl (111), methoxysuccinyl-Ala-Ala-Pro-Alathese enzymes in biological processes which frequently are of COCH,Cl (IV) were purchased from SERVA, Heidelberg, and from medical importance. One classof inhibitors, the peptide chlo- Calbiochem, Bubendorf (Switzerland), andused without further puis in water, and the others were romethyl ketones, has been more intensively studied for two rification. The latter inhibitor soluble first dissolved in a minimum volume of methanol and then added to reasons: (i) they bind covalently to the catalytically active aqueous, buffered solutions of proteinase K. For inhibition studies, histidine and serine residues in the ubiquitous serine proteases inhibitors 1-111 were dissolved 0.5 pg/p1 in 50 mM Tris-HC1, 5 mM and are good transition state analogs, and (ii) their peptide CaC12,pH 8.0, containing 60% methanol. Stock solutionsof proteinsequence can be syxithesized according to the substratespec- ase K (0.2 pg/pl) and of the substratesuccinyl-(Ala),i-CO-NH-(CH,)ificity of the target enzyme (1-3). For the subtilisin-related NO, (1 mM) were prepared in the same buffer without methanol; in proteinase K, the binding of two such inhibitors, carbobenz- all experiments the molar ratio of inhibitor to enzyme was 1OO:l. Assays contained 0.7 nmol of proteinase K, 70 nmol of inhibitor, 0.75 pmol of substrate, 0.04-4% methanol. Afterenzyme and inhibitor * This work was supported by the Deutsche Forschungsgemeinschaft (Sonderforschungsbereich 9, Teilprojekt A 7), and by Fonds were incubated for 15 min at 25 “C, the substrate succinyl-(Ala):TCO-NH-(C6H4)-NOawas added and reacted 1 h. The reaction was der Chemischen Industrie. The costs of publication of this article were defrayed in part by the payment of page charges. This article then stopped with glacial acetic acid and the released p-nitrophenymust therefore behereby marked “aduertisernent” in accordance with lolate monitored a t 410 nm with a Beckman DU6 spectrophotometer. Relative inhibition rates are given in Table I. 18 U.S.C. Section 1734 solely to indicate this fact. For crystallization of the complex between proteinase K and methThe atomic coordinates and structure factorshave been deposited in the Protein Data Bank, Brookhaven National Laboratory, Upton, oxysuccinyl-Ala-Ala-Pro-Ala-COCH2C1, the lyophilized enzyme was dissolved a t 5% (w/v) in 50 mM Tris-HC1, 1 mM CaCl,, pH 7.5, with NY. 2-fold molar excess of the inhibitor. 20-4 dropsof this solution were $ O n leavefrom the Institute of GeneralChemistry,Technical University of Lodz, Lodz, Poland. equilibrated in the sitting drop vapor diffusion method against 0.75 § To whom correspondence should he addressed. M NaNO:! in the same buffer. Single crystals of size 0.8 X 0.6 x 0.6

17695

Structure of Peptide-inhibited ProteinaseK

17696

TABLEI Effectivity of several inhibitors against proteinaseK For conditions, see "Materials and Methods." Inhibitor

Relative inhibition rate

I Ala-Ala-COCH2C1 I1 Ala-Ala-Phe-COCH'C1 111 Phe-Pro-Arg-COCH&l IV Methoxysuccinyl-Ala-Ala-Pro-Ala-COCH&l

10 28 10 88

4

% 161

TABLEI1 Data collection, data processing, and refinement statistics Crystal habit tetragonal bipyramids; space group P432,2; cell constants a and b = 68.3(2) c = 108.4(2), CuK,, radiation ( X = 1.5418 A) from Elliott GX20 rotating anode generator, 40 kV, 70 mA, 0.2- X 2-mm' focus; Cea-Reflex 25 (Ceaverkaen) films in 3-packs; 59-mm film-to-crystaldistance, 2.2 Aresolution; crystalmountedalong c; 1.5" oscillation per exposure of 50 min; 53620 reflections collected, merged to 11864 2 l u ~overall , merging R,,, factor" = 9.8%; mean B factor from Wilson plot = 14.6 A'.

6

A,

Refinement statistics; rms deviations from ideal values

(A)

s

gles (A)

Bond Bond angles (") Torsion (") Trigonal atoms planarity (A) (carboxylates, amides, peptides) Planar groups Bad Chirality restrained to L-configurationfor all amino acids Final R factorb for 11.864 reflections I F > o(F) I in resolution range 10-2.2 A, R = 0.198 Mean temperature factorB = 15.0

(A)

0.009 1.824 24.470 0.017 0.025 0.074

FIG. 1. Difference electron density I F.-F, I for the methoxysuccinyl-Ala-Ala-Pro-Ala inhibitor, with final model superimposed. The -CH,-CH,- portion of the methoxysuccinyl group is less well defined butis visible at lower contour levels. We associate the low level electrondensity of the-CH2-CH2- group with high thermal mobility (temperature factors around 39 A') and/or with disorder of the zigzag structure over a t least two alternative positions. The map was computed at the endof the refinement, with inhibitor atoms not included in phase calculations. atoms, 31 inhibitor atoms, 170 water oxygen atoms, and one Ca2+ atom is 0.198, based on the 11864 reflections with F > UF. The atomic coordinates of the complex will be deposited with the Brookhaven protein data file. RESULTS AND DISCUSSION

In the preyious crystallographic studies on native proteinase K at 1.5-A resolution (11)and on proteinaseK complexed with the inhibitor carbobenzoxy-Ala-Ala-COCH2Cl (4), the A* catalytic site wasidentified a s t h e t r i a d A ~ p ~ ' - H i s ~ ' - S e r ~ ~ ~ , RBYm = ZhZ, I < F h > -F; I /&&Fa,. with the free C ~ S ~ ~ ( near S H ) the imidazole ring of His6'. The R = z h I F o b s I - I F C a l C I /zI Fobs 1. substrate recognition site is formed by the two peptide chains, 99-104 and 132-134, respectively (Fig. 2, a and b ) . mm:' grew within 1 day at room temperature. These peptide chains are oriented approximately parallel The crystals of diameter -0.8 mm are isomorphous tq crystals of and directly connected onlyby a hydrogen bond formed native proteinase K ( 1 0 , l l ) . X-ray datacollection to 2.2-A resolution between the Tyrlo4(07) and Gly13'j(0)(Fig. 2a). Thepreviously was carried out by film methods using an Arndt-Wonacott rotation studied peptide inhibitors (4, 5) and the one presently invescamera installed on an Elliott GX20 rotating anode x-ray generator recognition site such that an (see Table 11). The x-ray intensities on the films were digitized with tigated are inserted into the a n Optronics PlOOO film scanner and processed with the MOSCO antiparallel three-stranded pleated sheet formed. is The sheet program system (12). has a commonly observed overall left-handed twist. The coordinates of all the protein and two Ca2+ atoms of the 1.5In the reaction between proteinase K and the inhibitor A resolution structure of proteinase K (11)were used as a starting methoxysuccinyl-Ala281-A1a282-Pro283-Ala2&1-CH2C1, the active model in the structure determination. For the preliminary stages of site residues His6'(NC2) and SerZz4(0,) react with the terminal restrained least-squares refinement (13), only the 7809 data > 3 a ~ and within the 5-2.2-A resolution shellwere used, with an initial R factor -COCH2C1 group, to form a covalently bonded complex with of 0.283. After 8 cycles of refinement, theR factor converged a t 0.224 proteinase K as shown in Fig. 2b. This complex mimics the and a 2F0-F, electron density map was calculated using the CCP4 transition state with tetrahedral CZR4 atom, except for the suite of proteincrystallographicprogramsandinterpretedwith CH2- group which, in the transition state,would be the -NH FRODO (6) by means of an Evans and Sutherland PS330 graphics of the leaving group (the product). The-CH2-His6'(N,) bond display. The map clearly revealed the Ala-Ala-Pro-Ala portionof the inhibitor and the covalent bonds of the -CO-CH,- end group with restricts the orientation of this group so that the CZ8'(O-) oxygen cannot fully move into the "oxyanion hole" which is H~s"(N.~) and SerZ2'(0,), but themethoxysuccinyl group could not Asn"l(N), and Asn"jl(Na). In the native be located. A further 5 cycles of refinement with restrained covalent formed by SerZz4(N), enzyme-inhibitor bonds led to a map in which electron density for proteinase K, this hole is filledby a water molecule (11). the protecting group could be interpreted (Fig. 1).The subsequent Instead, the Ala284(O-) atom forms hydrogen bonds only to refinement was carried out with the reciprocal space least-squares SerZz4(N), and because it cannot move further into the "oxyprogram T N T (14) because it allowsfor betterdefinition of the anion hole" to approach and to bind to Asn"l(N) and restraints within the inhibitor binding site. At this stage all 11864 the Asnl6'(N6), the resultinggap is filled by a well defined water x-ray data above the laF level were successively included and the molecule Wat'" ( B = 23.4 A') in hydrogen bonding distance, resolution range extended to 10 A. Inspection of 2F0-Fc maps not only allowed for improvement of the position of the inhibitor, but see Table 111. also to locate 170 water molecules and the terminal six atoms of the The covalent binding geometry parameters within theresArglG7 sidechain which were not defined in the structure of native idues His6', S e P 4 , a n dAlaz8' were refined with constraints. proteinase K. The strong Ca2+ binding siteC a l is fully occupied hut the weak binding site Ca2 appears to be filled by a water molecule, As expected, the bond angles around atom AlaZa4(C) are in the range 107-112", the angle at Ser224(Oy) is 107", and that as indicated by the occupation/thermal parametersfor this site. The finalR factor for the refined structure containing2022 protein at the -CH2- atom AlaZa4(C)is 116"; the CH2-N bond distance

Structure of Peptide-inhibited Proteinase K

17697

FIG.2. a, stereoviewshowingthe substrate binding site and the catalytic site (left) of proteinase K, and the covalently bound inhibitor methoxysuccinylAla-Ala-Pro-Ala-C(O-)-CHz-. Red, inhibitor atoms;orange, CyP; black, water molecules; green, hydrogen bonds, blue, proteinatoms as intheproteinase Kinhibitor complex;brown, as in free and fully active proteinase K. C,-atoms are labeled. b, schematic drawing and atom numbering of the complex as described in a. S ' S 6 define the subsites according to (8), the correspondingnumbering scheme on the inhibitor would be P1 for P4 for Alam, P2 for Prozm,P3 for Ala2@, AlaB1, and P5 for succinyl 280. Drawn with SCHAKAL (16).

s5

s4

s3

s2 SI

TABLEI11 Hydrogen bonds D-H- . - A between inhibitor, water molecules, and proteinase K H atom Dositions were calculated with MOLEDT(7). Atoms involved D...A H...A D-H...A

see caption Fig. 2b), Ala2@(N);it donates an unsymmetrical bifurcated three-center bondwhere the N-H opposestwo oxygen atoms (15). The major component isto Ser'32(0)and the minor (weaker) component to SeP4(0,). In subsite P2, the oxygen Pro283(0)accepts a hydrogen bond from water A Wat448 which simultaneouslyinteracts with AlaZs2(0),and the latter oxygen accepts a second hydrogen bond from G~Y'~~(N). MSUzso(Ol)-Tyr104(N)a 3.31 159 2.28 MSUm(03)-Wat453b 2.99 is not only heldin its positions by the hydrogen bonds, Gly13s(N)-Wat*3 3.10 but it is also fixed by van der Waals contacts with the pep2.26 2.72 Ala2R1(N)-Gly'02(0) 103 tide carbonyl groupof G1y1O0 (Pro'"(CJ. . .Gly'"(O), 2.98 A). Ala2R1(0)-Gly'm(N) 2.77 3.47 122 It even appears that this interaction pushes away one of the Ala"2(N)-Gly'34(0) 3.07 2.11 147 recognition strands G1yIoo-QPto releave steric clash, as Ala282(0)-G19u(N) 3.14 2.21 143 Ala*(O)-Wat& 3.17 shown in Fig. 2a. With other amino acids in place of Proza3, pro28"(o)-WatMs 3.24 this interaction is not possible because in nonproline amino ) Ala2RA(N)-Se324(0 2.86 2.57 94 acids the side chains can adopt other conformations to avoid Ala"8"(N)-Ser132(Oj 2.93 1.86 168 steric clash. In addition, the N-C, bond is replaced byN-H, Ala'm(0)-SeP4(N) 2.71 2.00 142 and another hydrogen bond can be formed withthe recogni2.84 Alam-WaP A s ~ ' ~ ~ ( N & W ~ ~ ' ~ ~ 3.24 tion site of the proteinase K.At positions P3 and S3, Alam2 and Gly134form a hydrogen bonding motif characteristic for Additional hydrogen bonds antiparallel @-pleatedsheets, and a similar interaction is also stabilizing the activesite observed forP4 and S4, formed by Ala2*land Gly"', Fig. 2, a 2.79 and b. The N-terminal methoxysuccinyl groupof the inhibitor 2.69 is bound to the recognition site of proteinase K by two 2.91 hydrogen bonds, one direct between T y P ( N ) and the ester 'MSU, methoxysuccinyl. -0-CHIoxygen, the other mediated by a water molecule, Wat, water. succinylpeptide (C=O). .Wat4s3- (HN)GlyI3'j.The succinyl -NH-CO-CH2-CH2- moiety is stacked on the T y r 1 0 4 phenyl is 1.4 A, the C-0 distances are 1.35-1.36 and the C-C group, wit! intermolecular C C distances in the range of distances fall in the range 1.50-1.55 The inhibitor adopts 3.22-3.43 A. an extended conformation with trans Pro283,and all @,@ To detect any conformational changesthat occurred upon angles correspond to the &pleated sheet region in a Rama- binding of the inhibitor to proteinase K, the main chain atoms chandran plot (Alaza1:(b -loo", \k 178"; Alam2:r#J -146", N, C,, C, 0 of the complex weresuperimposedby least squares 163";Prom: (b +OD, \k 151";Ala2%r#J -125"). In addition to fittiig on the corresponding atoms of the native proteinase K the covalent bonds with activesite.residues Ser224 and His6', structure using the OVERLAY procedure of the TNT prothe inhibitor is anchored in thesubstrate recognition site by gram (14). The root mean square (rms) deviations of the 1362 superimposed ?toms and all the 2017 common atoms were several hydrogen bonds,details are given in Table 111. These hydrogen bonds involve, in subsite P1 (for definition 0.21 and 0.41 A, respectively. Except for the binding region,

-

A.

-

A,

*

17698

Structure of Peptide-inhibited Proteinase K

FIG. 3. Stereo plot of the covalent complex formed between proteinase K and methoxysuccinyl-AlaAla-Pro-Ala-C&. The orientation is similar as in Fig. 2a. Active site residues A s p 3 9 , His6’, Selu, and AsnIB1 forming the “oxyanion hole” in blue, inhibitor in red and substrate recognition site in pink. Water molecules in recognition site are filled blue circles. a-Helices are drawn as cylinders, &pleated sheet strands as ribbons, S-S bond by circles, (-0-0-), and Ca2’ is indicated by 285 CAL. Drawn with program (17).

no systematic and significant conformational changes were inhibitor binds into the recognit+ site, strand 100-104 is found. Close to the active site and just Ubelow”the plane of consequently pushedaway by -1 A to open the binding cleft, functionally important Hise imidazole, there is the free SH probably caused by the tight interaction with the side chain 2.98 A. group of C ~ S (Fig. ’ ~ 2a), the role of which is not yet under- of the inhibitor proline, Proz83(Ca). (O)GlyLW, The inhibition of proteinase K by the four different peptide stood. Its sulfur atom is engaged in several close contacts with chloromethyl ketone inhibitors I to IV is described in Table active site residues (HisG9(0),Se#“(O), Ser13’(0,) which do not change significantly their positions after inhibitor binding I. The dataindicate that the inhibition (binding of the inhibexcept forthe distance to His6’(0), which increasesby 0.7 A. itors) is better with longer peptide chains of the inhibitors. This increase in distance is caused by a movement of HisG9 The sample of inhibitors given in Table I is not sufficient to which is necessary to permit formation of the covalent bond propose a correlation between inhibitor sequence and binding affinity, but it is clear from this x-ray study that proline between the H~S”(N,~) and the inhibitor -CH2-groupby nucleophilic displacement of the chlorine atom. The move- interferes with peptide binding to proteinase K. It suggests ment of the H i s G g imidazole is accompanied by a similar that more effective inhibitors would have the proline substimovement of the Asp3’ carboxylate so that the hydrogen tuted by another amino acid to avoid the steric interference bonds between these two side chains are not disrupted; the with strand 100-104 of the recognition site. movement of the imidazole is also transmitted to the C, atom CONCLUSIONS so that the positions of the His69 mainchain atoms and of the associated amino acidsAmG1and GlfS are also affected. Complementary to our earlier publications on the crystal In contrast, binding of the inhibitor does not perturb sig- structures of complexes betweenproteinase K and dipeptide nificantly the conformation around the active SeP4,although inhibitors carbobenzoxy-Ala-Ala-C(O)CH&land carbobenzthe Se#24(0,) is covalently bonded withthe inhibitor ketone oxy-Ala-Phe-C(O)CH&l, we have now investigated a tetragroup to form a structure reminiscent of the tetrahedral peptide which, in contrast to the earlier, shorter inhibitors intermediate. Since this reaction also occurs with a “real” saturates the whole substrate binding site. There is antiparsubstrate and is indispensable for catalysis, it obviously can allel @-pleated sheet formation withall theamino acids of the proceed smoothly, with no major conformational re- inhibitor engaged in hydrogen bonding except for If arrangement. this proline were replaced by another amino acid, one more When the inhibitor binds to proteinase K, there is insertion hydrogen bond ProZm(N) GlyloO(0) should be possible after into the recognition site and hydrogenbonds are formed. minor conformational changes around position 283of the There may be two reasons why the position of strand 132- inhibitor and/or of the protein segment GlylW-Tyrlo4. 136 of the substrate recognition site is not significantly afThe contact surface betweenthe inhibitor and thesubstrate fected. First, it is tightly anchored with the bulk of the binding site is extensive.Besides the hydrogenbonds, it proteinase K structure (Fig. 3); its N terminus is linked with involves van der Waals interactions between the side chains a 9-stranded @-sheetthrough strand PI14 (ll),its C terminus of the Ala28’, Prom, and the substrate recognition site, and with a-helix a4, and its segment 134-137 forms an antiparallel the stacking between the succinylmoiety and the T y r 1 0 4 p-sheet (@I11(11))with amino acids 168-170. Second, the phenolic group.As the “bottom”of the proteinase K substrate Proa3 side chain pushes segment100-104 while segment 132- recognition site is predominantly hydrophobic dueto theside 136 is in the correct position to smoothly hydrogen bond to chains ofLeu’33 and TyrlQ4,it is not the sequence of the the inhibitor. Segment 100-104, although connected at its N substrate that is of importance in the recognition but only terminus through @-turnt9 with strand @I13 of the pleated the actual size of the side chains. Even larger side chains can sheet and, a t its C terminus, with a-helix a 3 (ll),is not be tolerated due to the flexibility of the segment Gly100-Tyr104, engaged in the secondary structure hydrogen bonding but as shown with ProzE3of the peptide inhibitor used in this exposed to solvent, and can move accordingto sterical require- study. In this light, the observation that the affinity of proments. When the methoxysuccinyl-Ala-Ala-Pro-Ala-CH~-C1 teinase K correlates with the length of the inhibitor is readily

--

Structure of Peptide-inhibited Proteinase K explained as thebinding to the substraterecognition site will be the stronger the more peptide- . .peptide hydrogen bonds are formed between inhibitorand enzyme, andthe more hydrophobic contacts are made.Any sequence dependence should be of minor influence, as actually shown by the unspecific nature of proteinase K which has only some preference for aromatic,bulky amino acid side chains. Acknowledgments-W. M. W. and T. P. S. thank the Alexander von Humboldt Stiftung for a fellowship. The help of Dirk Kostrewa with initial stages of the refinement is gratefully acknowledged. REFERENCES 1. Robertus, J. D., Alden, R. A,, Birktoft, J. J., Krant, J., Powers,

J . C., and Wilcox, P. E. (1972) Biochemistry 11, 2439-2449 2. Poulos, T. L., Alden, R. A., Freer, S. T., Birktoft, J. J., and Kraut, J. (1976) J. Biol. Chem. 251, 1097-1103 3. Bromme, D., Peters, K., Fink, S., and Fittkau, S. (1986) Arch. Biochem. Biophys. 244,439-446 4. Betzel, C., Pal, G . P., Struck, M., Jany, K.-D., and Saenger, W. (1986) FEBS Lett. 197, 105-110 5. Betzel, C., Bellemann, M., Pal, G. P., Bajorath, J., Saenger, W., and Wilson, K. S. (1988) Proteins Struct. Funct. Genet.4, 157164

17699

6. Jones, A. T . (1983) in Computational Crystallography (Sayre, D., ed) pp. 303-317, Oxford University Press, Oxford 7. Dayringer, H. E., Tramontano, A., Sprang, S. R., and Fletterick, R. J. (1986) J . Mol. Grph. 4, 82-87 8. Schechter, I., and Berger, A. (1967) Biochem. Biophys. Res. Commun. 27,157-162 9. Bajorath, J., Hinrichs, W., and Saenger, W. (1988) Bur. J. Biochem. 176,441-447 10. Pahler, A., Banerjee, A., Dattagupta, J. K., Fujiwara, T., Lindner, K., Pal, G. P., Suck, D., Weber, G., and Saenger, W. (1984) EMBO J. 3,1311-1314 11. Betzel, C . , Pal, G. P., and Saenger, W. (1988) Eur. J. Biochem. 178,155-171 12. Machin, P. A., Wonacott, A., and Moss, D. (1983) Daresbury Laboratory News 10,3-9 13. Hendrickson, W. A., and Konnert, J. H. (1981) in International Symposium onBiomolecular Structure (Srinivasan, R., ed) pp. 43-57, Pergamon Press, Oxford 14. Tronrud, D. E., Ten Eyck, L. F., and Matthews, B. W. (1987) Acta Crystallogr. A43, 489-501 15. Ceccarelli. C.. Jeffrev. " . G. A.. and Tavlor. " , R. .(1981) . J. Mol. Struct. 70,255-271 16. Keller. E. (1988) SCHAKAL88. KristallomaDhisches Institut der Albert-Ludwigs-Universitat, Freiburg, fed'eral Republicof Germany 17. Lesk, A. M., and Hardmann, K. D. (1982) Science 216, 539-540