multiple solvent crystal structures: probing binding sites, plasticity and hydration
TRANSCRIPT
doi:10.1016/j.jmb.2006.01.039 J. Mol. Biol. (xxxx) xx, 1–12
DTD 5 ARTICLE IN PRESS
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
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
Multiple Solvent Crystal Structures: Probing BindingSites, Plasticity and Hydration
CarlaMattos1*, Cornelia R. Bellamacina2, Ezra Peisach2, Antonio Pereira2
Dennis Vitkup2, Gregory A. Petsko2 and Dagmar Ringe2*
78
79
80
81
82
83
84
85
86
87
88
89
90
1Department of Molecular andStructural Biochemistry, NorthCarolina State UniversityCampus Box 7622, 128 PolkHall, Raleigh, NC 27695, USA
2Rosenstiel Basic MedicalSciences Research CenterDepartment of Chemistry andBiochemistry, BrandeisUniversity, 415 South StreetWaltham, MA 02254-9110USA
U0022-2836/$ - see front matter q 2006 P
Abbreviations used: MSCS, multistructures.E-mail addresses of the correspon
[email protected]; ringe@bran
YJMBI 57996—28/1/2006—12:12—SATHYA—
91
92
93
94
95
96
97
98
99
100
D PROOFMultiple solvent crystal structures (MSCS) of porcine pancreatic elastasewere used to map the binding surface the enzyme. Crystal structures ofelastase in neat acetonitrile, 95% acetone, 55% dimethylformamide, 80%5-hexene-1,2-diol, 80% isopropanol, 80% ethanol and 40% trifluoroethanolshowed that the organic solvent molecules clustered in the active site, werefound mostly unclustered in crystal contacts and in general did not bindelsewhere on the surface of elastase. Mixtures of 40% benzene or 40%cyclohexane in 50% isopropanol and 10% water showed no bound benzeneor cyclohexane molecules, but did reveal bound isopropanol. The clustersof organic solvent probe molecules coincide with pockets occupied byknown inhibitors. MSCS also reveal the areas of plasticity within theelastase binding site and allow for the visualization of a nearly completefirst hydration shell. The pattern of organic solvent clusters determined byMSCS for elastase is consistent with patterns for hot spots in protein–ligandinteractions determined from database analysis in general. The MSCSmethod allows probing of hot spots, plasticity and hydration simul-taneously, providing a powerful complementary strategy to guidecomputational methods currently in development for binding sitedetermination, ligand docking and design.
q 2006 Published by Elsevier Ltd.
Keywords: elastase; solvent mapping; organic solvents; protein bindingsites; multiple solvent crystal structures
101
102
*Corresponding authors
TE 103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
NCORRECIntroductionOne of the current challenges in structuralbiology is to understand the general features thatguide protein–ligand interactions, allowing theprediction of these interactions given the structureof the unbound components.1 The binding processis thought to occur in a step that involvesrecognition, followed by rearrangements thatoptimize packing at the interface.2–4 Therefore, animportant aspect in understanding interfaces andthe formation of complexes is to determine whatdistinguishes ligand-binding sites from other areason protein surfaces, addressing the question poseda decade ago: what makes a binding site a bindingsite?5 Indeed, the last decade has seen intenseefforts to characterize the features that distinguish
ublished by Elsevier Ltd.
ple solvent crystal
ding authors:deis.edu
198707—XML – pp. 1–12/TL
121
122
123
124
125
126
binding sites form other areas on protein surfaces.Initial database analysis determined that bindingsites exhibit no general patterns of hydrophobicity,shape or charge,6 although for small ligands there isa correlation with shape. Furthermore, it quicklybecame clear that plasticity plays a major role inprotein–ligand interactions.7 The dynamic com-ponent of ligand binding as well as the thermo-dynamic contribution of solvation effects havecontributed to the difficulty of predicting a priorispecific residues or binding pockets that are hotspots for ligand binding affinity.8 In addition, therole of bound water molecules in mediatingprotein–ligand interaction cannot be ignored.9 Inspite of the difficulties, however, a combination ofdatabase analysis,4 experimental observations onindividual complexes,10 the development of com-putational methods for predicting the location ofbinding sites11 and docking of binding partners12
have contributed to significant advancements inthis area of research.The multiple solvent crystal structures (MSCS,
elastase models from cross-linked crystals in
T
2 MSCS of Elastase
DTD 5 ARTICLE IN PRESS
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
UNCORREC
various solvent conditions: XLINK, aqueous sol-ution; ACN, neat acetonitrile; DMF, 55% dimethyl-formamide; HEX, 80% 5-hexene-1,2-diol; TFE1, 40%2,2,2-trifluoroethanol (cross-linked crystals werenot transferred to distilled water); TFE2, 40%2,2,2-trifluoroethanol; ETH, 80% ethanol; ACE,95% acetone; ISO, 80% isopropanol; IBZ, 50%isopropanol, 40% benzene; ICY, 50% isopropanol;40% cyclohexane) approach provides a robustexperimental method to locate and characterizeligand binding sites on proteins using organicsolvents.13,14 The method is fully developed hereusing porcine pancreatic elastase as a modelenzyme. As different protein crystals have beensuccessfully transferred to organic solvents forX-ray structure determination,15–21 the MSCSmethod is gaining recognition as a method tostudy ligand binding sites when well diffractingcrystals are available, as the organic solventmolecules probe the surface in a way that is notpossible within the constraints of a larger moleculethat occupies the entire binding site. The crystalstructures of thermolysin in different concentrationsof isopropanol20 and in three other organicsolvents21 reveal that the organic solvents clusterin the known active site and appear in some areas ofcrystal contacts. Furthermore, the authors used theMSCS method to qualitatively rank the affinity ofdifferent thermolysin subsites for isopropanol.20
The work on thermolysin includes a comparison ofthe experimental organic solvent binding sites withthose obtained computationally using the multiplecopy simultaneous search (MCSS)22 and the GRID23
methods. Only poor agreement was obtainedbetween the experimental and computational results.Indeed, these calculations areunrealistic because theydo not include solvation effects and the plasticity ofthe protein structure.24 More recently, a compu-tational counterpart to the MSCS method has showngreat success in predicting the location of the primarybinding pockets in enzymes.25,26
One long-term use of MSCS might be in liganddesign, where the organic solvents provide exper-imental positions for functional groups that can beincorporated into larger ligands.27–29 A strategy forligand design that has been discussed for manyyears in molecular modeling approaches relies onthe idea that functional groups can be optimizedindependently for different regions of a proteinbinding site.22,23 These functional groups can thenbe linked to form a ligand with high affinity andspecificity to the target protein.27,28 The proteinbinding affinity of the resulting molecule will be, inprinciple, the product of the binding constants forthe individual fragments plus a term that accountsfor changes in binding affinity due to the linkerportion of the larger ligand.30 Though viable inprinciple, this strategy has met with only limitedsuccess, notably in cases when fragment positionswere obtained experimentally,30 because a majordifficulty with the computational linked-fragmentbased approach is the lack of reliability in predict-ing optimal bindingmodes for the fragments tested.
YJMBI 57996—28/1/2006—12:12—SATHYA—198707—XML – pp. 1–12/TL
The use of organic solvent binding as an exper-imental approach to identify fragments for liganddesign has been previously suggested as animmediate consequence of the MSCS method.13,17
The focus in the present study is to further exploreways in which MSCS can be used to more fullycharacterize the surfaces of proteins. In addition tobinding at specific sites, the organic solventsprovide changes in the protein environment,inducing structural adjustments in areas of plas-ticity and influencing the way in which surfacewater molecules interact with the protein.31 TheMSCS method provides a unique experimentalapproach to characterize active sites of enzymes,while probing protein plasticity and surfacehydration simultaneously throughout the entirestructure. This is a critical step in understandingthe complex protein template targeted for liganddesign.
ED PROOF
Results
Porcine pancreatic elastase (from here on referredto as elastase) is a serine protease of the trypsinfamily, with 240 amino acid residues. It is composedof two b-barrel domains, with the catalytic triad(Ser206, His60 and Asp108) residing in the cleftbetween the domains.13,17,32,33 The fact that elastasehas been well studied makes it a particularlyappropriate model to explore the potential of theMSCS method to characterize the surfaces ofenzymes in general.
Elastase crystals grown in aqueous solution werecross-linked with glutaraldehyde and transferred toa series of solutions containing high concentrationsof organic solvents in water as described inMaterials and Methods. The crystal structures ofelastase in each of ten conditions were solved,revealing the way in which the organic solventmolecules interact on its surface. No cross-linkswere observed in any of the electron density maps.Furthermore, the cross-linking does not affect theoverall structure of elastase. Each type of solventmolecule binds to the surface of the protein in aunique fashion, reflecting its shape and chemicalproperties.
For the present analysis nine crystal structures ofelastase were added to the one previously solved inneat acetonitrile (ACN).17 We present here thecrystal structures of elastase solved in the presenceof 95% acetone (ACE), 55% dimethylformamide(DMF), 80% 5-hexene-1,2-diol (HEX), 80% isopro-panol (IPR), 80% ethanol (ETH) and two structuressolved in 40% trifluoroethanol (TFE1 and TFE2). Toobtain the TFE1 model, the cross-linking phosphatebuffer solution containing the elastase crystal wasexchanged directly with a 40% trifluoroethanolsolution, while for TFE2 the crystal was firstexchanged into distilled and de-ionized water. Inaddition, a crystal structure was solved in a mixtureof 40% benzene, 50% isopropanol and 10% water(IBZ) and one in 40% cyclohexane, 50% isopropanol
T
MSCS of Elastase 3
DTD 5 ARTICLE IN PRESS
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
UNCORREC
and 10% water (ICY). Isopropanol was bound toelastase in these last two structures, but there wereno bound benzene or cyclohexane moleculesobserved in the electron density maps. For theMSCS analysis, each of the ten crystal structureswas superimposed onto the previouslypublished structure of cross-linked elastase inaqueous solution.17
Comparison between the elastase models
The elastase crystals in organic solvents areisomorphous with those in aqueous solution. Theyhave the symmetry of space group P212121 and unitcell dimensions of approximately aZ51 A, bZ58 Aand cZ75 A. The unit cell parameters, datacollection and refinement statistics are presentedin Table 1 for each of the nine previouslyunpublished structures. All of the elastase modelscontain a calcium ion and one sulfate ion in thepositions observed in previous structures of elas-tase solved in aqueous solution34 (Table 1).A second sulfate ion is found in the oxy-anionhole in the ACE, DMF andHEXmodels. In the otherstructures, either a water molecule or one of theorganic solvent molecules is found at this site. Thenumber of water molecules found in each model isgiven in Table 1.
The root mean square deviations for pairs ofMSCS structures range between 0.12 and 0.33 basedon the Ca positions of elastase residues. Thisconfirms the visual observation that overall thereis no significant difference between elastase modelsin the various solvent environments. When compar-ing the structures in detail, it becomes clear thatthey vary primarily in the conformations of side-chains that have high B-factors in aqueous solution.However, a couple of notable exceptions involvingmain-chain atoms do occur. Residues 24–27(RNSW) and 122–123 (SY), two regions that interactwithin the structure, adopt one of two very distinctbackbone and side-chain conformations. In thepresence of solvents of higher dielectric constants(XLINK, ACN, DMF, HEX, and TFE2/water)several of the charged or polar groups are exposed,while in environments of lower dielectric constant(TFE1/buffer, ETH, ACE, IPR, IBZ, and ICY) theresulting conformation maximizes H-bondingbetween the protein atoms, with polar groupsturned away from the protein surface.
The organic solvent binding sites
In general, only a few organic solvent moleculeswere found in each of the models (Table 1), which isa far cry from the protein surface being solvated bythe organic solvents. Figure 1 shows a ribbondiagram of elastase with 38 organic solventmolecules taken collectively from the ten super-imposed crystal structures of elastase solved underdifferent conditions (see Table 2 for a list of sitesoccupied in each model). There are 16 unique sitesfor the organic solvents. Molecules occupying
YJMBI 57996—28/1/2006—12:12—SATHYA—198707—XML – pp. 1–12/TL
ED PROOF
the same site have the same number in allcoordinate sets. The binding sites are numberedconsecutively from 1001 to 1016 and the organicsolvent molecules are numbered accordingly in themodels. There are seven sites occupied by morethan one type of organic molecule, numbered 1001to 1007 (four of these are in the active site). Theother nine sites, 1008 through 1016, bind a singleorganic solvent molecule (two in the active site).Organic solvent binding sites outside of the activesite are located mostly at or near crystal contacts.There are two crystal contact areas that areparticularly populated by the organic solventprobes. The first crystal contact group includessulfate ion 290, which is invariably found in elastasecrystal structures. The organic solvent binding sitesin this general area of crystal contacts are 1005, 1015and 1016 (see Table 2 for the identity of the solventsbound at these sites). The organic solvent sites in theother crystal contact area are 1007, 1012 and 1014.Solvent sites 1010, 1011 and 1013 are also in crystalcontacts, but site 1006 is not. At this site, Lys234interacts with either an acetone or an isopropanolmolecule through its N3 group. In addition toH-bonding with Lys234, the probes at site 1006 arein van der Waal’s contact with Leu227, Val231 andthe aliphatic portion of Arg233. This site is 8.5 Afrom the S4 subsite (1002).
The active site
The active site of elastase consists of pockets thatbind four or five amino acid side-chains before(S1–S4 or S5) and three after (S1
0–S30) the scissile
peptide bond.32,35 These subsites have been wellcharacterized through kinetic experiments andseveral crystal structures of elastase/inhibitorcomplexes have been published.32,36–39 Small ali-phatic amino acid residues were shown to befavored in the S1 and S4 subsites, with S2 preferringLys. The S3 subsite is highly exposed to solvent andhas a less distinct preference for any given aminoacid residue.The organic solvent molecular probes provide a
map of the protein surface that distinguishes theactive site from all other areas. Clustering of organicsolvent molecules occurs prominently in the activesite, where five of the known subsites on theprotease are clearly delineated. Organic solventmolecules are found in S1 (1001), S4 (1002), theoxyanion hole (1003) and two sites on the leavinggroup side of the catalytic triad likely to be the S1
0
(1008) and S30 (1004) binding pockets. Four of these
subsites are observed to bind at least three differenttypes of organic solvent molecules. In addition,there is a single acetonitrile molecule at the entranceto the S2 (1009) pocket17 making a total of sixbinding sites for organic solvents in the active site.While there is a common set of hydrophobic
interactions with organic solvents within each of theS subsites, there is significant variability in the polarinteractions, depending on the H-bonding potentialof the molecule involved and its orientation within
UNCORRECTED PROOF
Table 1. Data collection and refinement statistics
HEX ETH TFE1 TFE2 IPR IBZ ICY ACE DMF
Data collectionSpace group P212121 P212121 P212121 P212121 P212121 P212121 P212121 P212121 P212121Unit cell (A, deg.) aZ52.60, bZ
58.27, cZ75.84,aZbZgZ90
aZ52.45, bZ58.58, cZ75.32,aZbZgZ90
aZ52.60, bZ58.07, cZ75.46,aZbZgZ90
aZ51.88, bZ57.84, cZ75.24,aZbZgZ90
aZ52.49, bZ58.24, cZ75.39,aZbZgZ90
aZ52.13, bZ57.92, cZ74.91,aZbZgZ90
aZ52.02, bZ57.92, cZ74.84,aZbZgZ90
aZ52.20, bZ58.12, cZ75.05,aZbZgZ90
aZ52.54, bZ58.01, cZ75.66,aZbZgZ90
Resolution (A) N to 2.2 N to 2.0 N to 1.9 N to 1.8 N to 2.2 N to 1.9 N to 1.9 N to 2.0 N to 2.2Temperature (8C) 25 25 0 0 25 25 25 0 4Number of unique reflections(% complete)
19,334 (85) 14,637 (48) 14,070 (40) 25,281 (62) 20,624 (91) 21,613 (63) 22,509 (65) 12,617 (42) 11,178 (44)
68 95 75 90 91 91 90 87 78
RefinementR factor (%) 16.5 18.4 17.1 16.7 15.4 16.1 16.4 16.9 16.5R-free (%) 20.6 22.4 21.6 19.3 19.8 20.0 19.7 22.9 22.6Restraints (rms obsvd)Bond lengths (A) 0.005 0.005 0.005 0.004 0.005 0.004 0.004 0.005 0.005Bond angles (deg.) 1.16 1.23 1.23 1.22 1.16 1.22 1.24 1.20 1.18Dihedral angles (deg.) 27 26 25 25 26 25 26 26 26Improper angles (deg.) 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.7Total no. protein atoms 1822 1822 1822 1822 1822 1822 1822 1822 1822Number of water molecules 149 134 179 178 149 161 150 126 147Number of sulfate ions 2 1 1 1 1 1 1 2 2Number of calcium ions 1 1 1 1 1 1 1 1 1Number of solvent molecules 2 8 4 2 3 3 5 4 5
RZP
h jFoðhÞKFcðhÞj=P
h FoðhÞ.
YJM
BI57996—
28/1/
2006—12:12—
SATHYA—198707—
XML–pp.1–12/
TL
4MSCSofElastase
DTD
5ARTICLEIN
PRESS
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
UNCORRECT
Figure 1. Organic solvent binding sites. Ribbondiagram of elastase showing the binding sites for organicsolvent molecules in a common frame of reference. Eachsite is numbered as described in the text. The number ofthe sites occupied by organic solvent molecules in each ofthe models is given in Table 2. The catalytic triad is shownexplicitly in the cleft between the two b-barrel domains:Ser203, His60, Asp108 are shown in gray. The b-strandsare shown in purple and the two a-helices are shown ingreen. The organic solvent molecules are color-coded asfollows: HEX, salmon; ETH, hot pink; TFE1, cyan; TFE2,orange; IPR, light green; IBZ, green; ICY, dark green; ACE,red; DMF, blue; ACN, yellow. Figures 1–4 were madeusing the program MOLSCRIPT.52
Table 2. Data collection and processing conditions
Organic solvent % Volume EquipmentaTempera
(8C)
HEX 80 3 25ETH 80 3 25
TFE1 40 2 4
TFE2 40 1 4IPR 80 3 25IBZ 40/50/10 1 25ICY 40/50/10 1 25
ACE 95 1 0
DMF 55 2 4
ACN 100 1 4
a Three different setups were used for data collection, designated 1on a Rigaku RU200 rotating anode generator operating at 50 kV andGX-6 rotating anode operating at 30 kV and 30 mA. (3) Siemens multiRigaku RU200 generator operating at 50 kV and 100 mA.
YJMBI 57996—28/1/2006—12:13—SATHYA—198707—XML – pp. 1–12/TL
MSCS of Elastase 5
DTD 5 ARTICLE IN PRESS
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
D PROOF
the pocket. The hydrophobic interactions withsolvent molecules in S1 are provided by threeresidues that line the bottom of this pocket:Thr221 (Cg2), Val224 and Thr236 (Cg2). In generalthe organic solvent molecules are oriented in such away that their more hydrophobic areas reach deepinto the S1 pocket, with their polar atomsH-bonding to atoms closest to the surface. The S4subsite is lined by Phe223, Val103, Ala104, the Cg2
atom of Thr182 and the aliphatic portion of Arg226.Both the S1 and S4 subsites provide extensivehydrophobic contacts to the bound molecules,consistent with the preference for substrates withapolar residues in these pockets.34,35 Two Leuresidues provide the hydrophobic contacts withineach of the S1
0 and S30 pockets. Leu156 contributes
to both pockets, while Leu149 is found in S10 and
Leu77 is found in S30. In contrast, the interactions
within the oxyanion hole are primarily polar innature. The models HEX, ACE and DMF have asulfate ion bound in the oxyanion hole, as reportedfor the structure in neat acetonitrile.17 The sulfateions were not included in Figure 1 for clarity. Of theorganic solvents, only the alcohols ethanol (ETH),trifluoroethanol (TFE1) and isopropanol (ICY) bindin the oxyanion hole (Figure 1). In all three cases theOH group H-bonds with the backbone N atoms ofGly201 and Ser203 as well as with the Og of theactive site Ser203, mimicking the interactions of thewater molecule found at this site in the native andcross-linked structures solved in aqueous sol-ution.17 In the TFE1 model the hydrophobictrifluoromethyl group is in van der Waal’s contactwith the trifluoromethyl group of the moleculebound in the S1
0 site.The binding of organic solvents in the S subsites
coincides with the way these sites are occupied byelastase inhibitors. Figure 2 shows the inhibitortrifluoroacetyl-Lys-Pro-p-isopropylanilide32 super-
EtureData processingPhases (PD
code)Binding sites(see Figure 2)
Xengen 1ELA 1001, 1004Xengen 1ELA 1001, 1002, 1003,
1004, 1005, 1010,1011, 1012
XDS 1ELC 1001, 1002, 1003,1008
Denzo 1ELC 1001, 1002Xengen 1ELA 1001, 1002Process 1ELA 1001, 1002, 1006Process 1ELA 1001, 1002, 1003,
1005, 1006Process 3EST 1001, 1013, 1014,
1006XDS 3EST 1004, 1005, 1007,
1015,1016Process 3EST 1001, 1007, 1009
, 2 and 3 as follows: (1) R-Axis II phosphoimaging plate mounted100 mA. (2) Siemens X100-A area detector mounted on an Eliot-wire area detector with Argonne “Mad” Interface mounted on a
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
T
OFFigure 2. Superposition of organic solvents and
inhibitor in the active site of elastase. The organic solventbinding sites are numbered and some of the active siteresidues are labeled. The protein atoms are depicted withN in blue, O in red and C in gray. The trifluoroacetyl-Lys-Pro-p-isopropylanilide inhibitor is in pink and the organicsolvents are color-coded as in Figure 1.
Figure 3. Plasticity in the active site of elastase. The S1site is shown with the MSCSmodels superimposed. Someof the active site residues are labeled. The protein atomsare depicted with N in blue, O in red, S in yellow and C ingray. The organic solvent molecules in the S1 subsite(1001) are color-coded as in Figure 1.
6 MSCS of Elastase
DTD 5 ARTICLE IN PRESS
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
UNCORREC
imposed on the organic solvent molecules observedin that region. The inhibitor binds with thetrifluoroacetyl group in the S1 subsite, the Lysresidue in the S2, Pro in S3 and the anilide group inS4. The functional groups of the inhibitor interactwith the same protein atoms described above forthe solvents. The protein interactions that arecommon to all ligands provide a core of hydro-phobic residues whose positions are invariant fromone structure to another, whereas the interactionsthat vary between ligands form the surroundingportions of the pocket that show plasticity inaccommodating each one. Two such residues inthe S1 subsite are Gln200 and Ser203. Figure 3 showsthe MSCS models of elastase clustered by leastsquares superposition and focused on the primaryspecificity pocket S1. The strands containing Thr221,Val224 and Thr236 cluster so tightly that it is nearlyimpossible to distinguish the individual models inFigure 3. This is also the case for the backbone ofresidues 200 through 203. The side-chains of Gln200and Ser203, however, are in slightly differentpositions in each of the models and provide twokey areas of plasticity on either side of the ligandwithin the binding pocket. A similar situation isobserved in the S4 subsite, with residues Phe223,Val103, Ala104 and Trp179 well clustered, whileArg226 and Glu65 provide plasticity.
Water binding sites
Table 1 shows that while few organic solventmolecules were found in each of the models, the
YJMBI 57996—28/1/2006—12:13—SATHYA—198707—XML – pp. 1–12/TL
ED PRO
number of crystallographic water molecules perstructure was similar to that found for the structuresolved in aqueous solution, with a range of 126 to179 water molecules in the MSCS models. When all11 models are superimposed over 400 uniquewater-binding sites are observed. It is clear thatthe superimposed models provide an enhancedpicture of the first hydration shell surrounding theprotein compared to what is normally obtainedwith a single model, even at high resolution. Ananalysis of the hydration pattern revealed by MSCSdistinguishes four categories of water binding sitesaccording to the way water molecules interact withthe protein: buried, channel, surface and crystalcontact.33 The different types of water moleculescan be distinguished by their average atomicB-factors, which correlate with disorder and occu-pancy of the sites. The average B-factor for buriedwater molecules is 16 A2 (similar for atoms found inthe interior of the protein), for channel it is of 33 A2,and for surface it is 38 A2. It is this last group ofwater binding sites that most often changes locationfrom one structure to another, mediating localadaptation of the protein structure to the bulksolvent environment.31
Interestingly, all of the water molecules found inthe active site are of the surface type. Each of themodels individually has somewhere between fiveand 11 crystallographic water molecules in theactive site. However, when all 11 structures aretaken together they total 21 unique sites. Figure 4shows these water molecules superimposed on theinhibitor discussed above. It is striking how well
T
Figure 4.Crystallographic water molecules in the activesite. The same region of the active site is shown as inFigures 2 and 3, with the same color code for proteinatoms and organic solvent molecules. Water moleculesare superimposed on the trifluoroactyl-Lys-Pro-p-isopropylanilide (pink). The water molecules are color-coded according to the model from which they weretaken: XLINK, white; HEX, salmon; ETH, hot pink; TFE1,cyan; TFE2, orange; IPR, light green; IBZ, green; ICY, darkgreen; ACE, red; DMF, blue; ACN, yellow.
MSCS of Elastase 7
DTD 5 ARTICLE IN PRESS
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
Cthese crystallographically visible water moleculescollectively trace the location of the inhibitorbinding sites. The active site of elastase is wellhydrated, with crystallographically visible watermolecules sampling all of the hydrogen bondsavailable to ligands. Although the average positionof a water molecule varies with the solventconditions in which the protein is immersed, it isexactly this variation that increases the scope of theinformation available from the MSCS models. Thisallows a view of how closely the hydration net-works form an imprint of the inhibitors andpresumably of the substrate in the active site.
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
UNCORREDiscussion
The MSCS method provides an experimentalapproach for locating and characterizing bindingsites on protein surfaces. Its power lies in thecollective analysis of several (typically five to ten)superimposed crystal structures of the protein, eachsolved in the presence of high concentration of aparticular organic solvent, to compensate for thefact that the probe molecules bind with relativelylow affinity to most sites. Elastase is a goodrepresentative of the group of extracellular enzymesthat have typically been used in solvent mappingexperiments to date. These proteins have welldefined binding sites that do not undergo largeconformational changes upon substrate bindingand therefore can be thought to represent the type
YJMBI 57996—28/1/2006—12:13—SATHYA—198707—XML – pp. 1–12/TL
ED PROOF
of protein–ligand interactions embodied by the“lock and key” model. In addition to analyzingthe patterns of organic solvents clustered in theactive site, the MSCS analysis presented here forelastase also focuses on the patterns of plasticityand on the hydration properties of thesite, providing a more complete picture of thecomponents involved in the binding process.
Characterizing the active site
The results presented here for elastase fullyestablish the MSCS method as a powerful tool tolocate binding sites on enzymes. The distribution oforganic solvent molecules unambiguously dis-tinguishes the binding pockets for side-chains ofthe peptide substrate from all other areas on theprotein surface. The organic solvent moleculescluster in four of the seven elastase pocketsknown to be important for binding. The side-chainbinding pockets are lined by hydrophobic residuessurrounded by polar groups that interact in aspecific way with each type of molecule found inthe pockets. In this manner a variety of functionalgroups can be accommodated in the pockets, withapolar regions facing inward and polar functionalgroups taking advantage of a subset of the availableH-bonding interactions offered by protein atomscloser to the surface within the site. The rotationalfreedom available to a small organic moleculedistinguishes it from larger, more specific ligandsthat must simultaneously satisfy a greater numberof constraints within an extended binding site. Inthis sense, a diverse group of small molecules cancollectively serve as true probes of the possibleinteractions available within a binding site.Rotational freedom is reflected in computationalsolvent mapping in that each type of molecule isobserved to bind in multiple conformations withina site, typically having a common binding inter-action for the apolar portion of the molecule and avariety of orientations for the polar groups.26
The entire cluster, rather than individual dockingconformers, is used to obtain an average free energyof interaction, thus accounting at least partially forthe entropic effects of binding. In the electrondensity maps produced in the experimental solventmapping by MSCS only a single conformation isobserved for each solvent, although in some caseshigh B-factors for the bound organic solventmolecules can be indicative of local disorder in thestructure. The fact that a major peak in the electrondensity is observed for a given conformer indicatesthat it is likely present at least 40% to 50% of thetime, but does not rule out the presence of lesspopulated conformations. This is one of the ways inwhich the computational solvent mapping providescomplementary information to MSCS. Interestingly,while each distinct organic solvent molecule has itspreferential set of interactions observed experimen-tally with protein atoms, computational solventmapping results for lysozyme and thermolysinindicate that multiple sets of H-bonding
T
8 MSCS of Elastase
DTD 5 ARTICLE IN PRESS
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
UNCORREC
interactions often exist for a particular solvent.25
The fairly diverse options of polar interactionswithin a binding site make possible the binding ofsmall molecules of different shapes and functionalgroups. Conversely, it is possible to experimentallydetect the diversity of polar interactions in a site bythe number of distinct types of organic moleculesfound to cluster within it in the MSCS experiments.
The fact that clustering of different organicsolvent molecules is observed by MSCS primarilyat the known binding site of elastase is in strikingcontrast to the abundant water binding sitesthroughout the entire surface of the protein. Atfirst sight it is remarkable that the organic solventmolecules, present in most cases at higher %V thanwater, is unable to solvate the protein in any way. Infact, the organic solvents are only able to competewith water where the protein evolved to interactwith ligands or other proteins. From this perspec-tive, the organic solvents behave much more likeligands than like solvents. It is probably thisinability to compete with water outside ligandbinding pockets, combined with a diverse set ofavailable polar interactions at binding sites thatmakes binding of organic solvents so distinctlyprominent in the active sites of enzymes. It is theirbehavior as ligands, rather than solvents, that makesmall organic molecules powerful locators ofbinding sites on protein surfaces.
Protein plasticity
It is well known that there is a great degree ofplasticity in the binding sites of proteins.7,40
Plasticity does not necessarily involve largerearrangement of main-chain or side-chain confor-mation, but can also reflect small movements of oneor more residues in response to ligand binding or toa residue mutation that changes the local environ-ment at a given site. Plasticity in response to ligandbinding has been observed in many enzymes andwas described for elastase in particular as the“subtle induced fit of the active site as a result ofligand binding”.34 By superimposing the proteinmodels taken from MSCS as shown in Figure 3, onecan use experimental solvent mapping to delineateareas of plasticity in the active site. For the S1 subsitein elastase, used here as the illustrative example,these areas include the side-chains of residuesGln200 and Ser203, which undergo subtle changesin position in response to the various bound organicsolvent molecules. This kind of information can beextremely important in ligand docking and in thedevelopment of ligand design strategies. The lack ofthis information has been shown to be a limitingfactor in computational solvent mapping, which isoften more successful in predicting consensusbinding sites when the conformation found in thecomplex (rather than the apo form of the protein) isused as a template for the calculations.26 It is oftencostly to include protein flexibility in computationalstudies of ligand docking in binding sites,especially when a large number of compounds
YJMBI 57996—28/1/2006—12:13—SATHYA—198707—XML – pp. 1–12/TL
ED PROOF
need to be tested. This expense could be greatlyreduced if one could use the information fromMSCS to allow only selected parts of the structure tobe dynamic during docking or ligand design.
Plasticity can also be observed at the level of theentire protein structure. TheMSCS provide a view ofthe protein in very different environments. Throughobserving how the protein structure adapts to theglobal changes, it becomes clear which areas of thestructure are malleable and contribute to the changesin conformation that optimize structure given theproperties of the solvent and the constraints of thecrystal environment. The bulk effects of organicsolvents on protein structure have been previouslyreviewed14 and a study of apolar solvents on theconformation of the switch II region of Ras41 providesa fundamental framework for explaining the localconformational changes seen in the elastase residues24–27 and 122–123, where polar interactions betweenprotein atoms are optimized in solvents of lowerdielectric constants, while in the higher dielectricenvironment polar groups are often extended towardthe solvent. Remarkably, with the exception ofthe region involving residues 24–27/122–123,the backbone of elastase is largely unperturbed bythe changes in solvent environment. Rather, it is thefirst hydration shell that is the primary mediatorbetween the bulk solvent and the protein, changingpositions with changes in the solvent andprobably helping to maintain the integrity of thenative state.31 Thus, the MSCS method provides aview of plasticity in the water structure as well as inthe protein.
It has been observed through database analysisthat enzyme complexes tend to involve anchorresidues that interact with structurally constrainedportions of the binding partner.3 These constrainedor anchor areas are thought to be involved in the firststep of recognition, which is hypothesized to befollowed by an induced fit process involvingsurrounding residues to form the high affinitycomplex. The pattern of organic solvents in thebinding site of elastase can be viewed as depictingbinding sites for anchor residues at a few locations inthe more extended active site. The pockets mappedexperimentally have areas that do not change withthe organic solvent environment surrounded byplasticity regions as exemplified in Figure 3. Theprobe molecules cluster in the active site in well-defined hot spots. Rather than forming a continuoussurface, the hot spots are distributed throughout thebinding site in pockets lined by hydrophobicresidues. This is also the pattern observed indatabase analysis of protein complexes, althoughthe hot spot residues are not always hydrophobic innature.4 The important point here is that patternsessential to binding site obtained through databaseanalysis are reflected in the structures of elastase inthe presence of organic solvents. The MSCS methodis able to depict both hot spot residues and theplasticity areas that compose the active site ofelastase, providing a consistent picture of someessential binding site properties.
MSCS of Elastase 9
DTD 5 ARTICLE IN PRESS
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
Hydration
The 21 active site water molecules observedcollectively in the MSCS experiment formH-bonding interactions with virtually all availablepolar groups in the active site. This includes thewalls and openings to the pockets as well as theb-strand that forms a classic antiparallel interactionwith peptide substrates or a parallel interactionwith some inhibitors in the active site.32 The patternof crystallographic water molecules delineating thepolar groups in the active site mimics the shape ofthe bound inhibitors. This leads to the idea thatwhile the organic solvents may identify areas ofhydrophobic hot spots on the surfaces of enzymes,the water molecule pattern surrounding the hotspots may provide clues to the general shape of theactive site that can be occupied by a ligand. If so,these two components of MSCS are indeed highlycomplementary. All of the crystallographic watermolecules observed in the active site of elastasemust be released upon complex formation. Thiswould be expected to contribute an entropicallyfavorable component to binding.
T
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
ORREC
Conclusions
A complete MSCS analysis of a protein surfacetakes into account patterns of organic solventmolecules indicating the location of hot spots, theareas of plasticity observed when superimposingthe protein models and the distribution of crystal-lographically visible water molecules. The resultspresented here provide a detailed map of the activesite of elastase containing all of these componentsand consistent with binding site properties deci-phered form database analysis of hundreds ofproteins. Because the properties being probedhave been shown to be component features ofbinding sites, MSCS is poised as a powerful methodfor analyzing the active sites of enzymes in general.Hot spots, plasticity and certain patterns ofhydration converge in active sites. The MSCSmethod probes these features simultaneously, andtherefore can serve not only to locate the bindingsites, but also to offer a fairly complete character-ization of their essential properties. The diversity ofpolar and hydrophobic interactions within a proteinbinding site can be mapped in context with theplasticity of the active site and the malleability ofthe water structure. This information is critical infilling the missing links currently existing incomputational ligand design methods.
C 11261127
1128
1129
1130
1131
1132
1133
1134
UNMaterials and Methods
Crystal growth, cross-linking and solvent soaks
Porcine pancreatic elastase was purchased from eitherCalbiochem, Inc. or Worthington Biochemicals and usedwithout further purification. Crystals were grown and
YJMBI 57996—28/1/2006—12:13—SATHYA—198707—XML – pp. 1–12/TL
cross-linked with gluteraldehyde as described.17 Thecross-linking buffer (100 mM sodium sulfate, 100 mMsodium phosphate, pH 7.5) was slowly exchanged withdistilled water in a stepwise fashion. At the end of thisprocess the cross-linked elastase crystals were immersedin 200 ml of distilled water. Cross-linked crystals weretransferred to different organic solvents and the solventconcentration decreased 5% from neat by the addition ofwater, to a maximum concentration that was notdetrimental to the diffraction of X-rays. A new crystalwas used for each test. Once the maximum viableconcentration of a particular solvent was determined,the distilled water containing the cross-linked elastasecrystals was exchanged with the organic solvent solutionin a stepwise fashion by repeated removal of 20 mlaliquots followed by addition of the same amount of thefinal solution every 5 to 10 min. The final step in theprocess involved the transfer of the crystals to a freshsolution containing the desired concentration of theorganic solvent. The one exception to the general protocolabove was the TFE1 structure, for which the cross-linkingphosphate buffer solution containing the elastase crystalwas exchanged directly with a 40% trifluoroethanolsolution, rather than being first transferred to distilledand de-ionized water. The solvents for which data werecollected are shown in Table 1 and described in Results.
ED PROOFData collection
Crystals were mounted in quartz capillaries for X-raydiffraction data collection. Data were collected to amaximum resolution ranging from 2.2 A to 1.8 A on oneof three detector/generator sources: (1) R-Axis II phos-phoimaging plate mounted on a Rigaku RU200 rotatinganode generator operating at 50 kV and 100 mA; (2)Siemens X100-A area detector mounted on an Eliot GX-6rotating anode operating at 30 kVand 30 mA; (3) Siemensmulti-wire area detector with Argonne “Mad” Interfacemounted on a Rigaku RU200 generator operating at 50 kVand 100 mA. The data were reduced using either theprogram Process,42 XDS,43 Xengen44 or Denzo.45 The datacollection temperature and equipment, as well as theprogram used for data processing, are given in Table 2 foreach of the structures used in the present MSCS analysis.
Structure refinement
For all structures the program XPLOR46 was used forinitial phase calculation and least squares positionalrefinement. Initial phases were calculated from thecoordinates of elastase solved from crystals grown inaqueous solution, with PDB codes as specified in Table 2.All water molecules, counter ions and inhibitor atomswere removed from the initial models. The refinementstrategy was to first make adjustments to protein atoms,then to add water molecules that appeared simul-taneously in an electron density map with coefficients2FoKFc, contoured at the 1s level, and in an electrondensity map with coefficients FoKFc contoured at the 3slevel. A Silicon Graphics workstation running theprogram O47 was used to manually rebuild the protein.In general, water molecules were only added to the modelonce the R-factor dropped to below 22%. Calcium andsulfate ions, as well as the first round of water moleculeswere selected from the coordinates of native elastasesolved in aqueous solution (PDB code 1ELA). Additionalwater molecules were added manually by visual inspec-tion using the program O. Only then were the organic
10 MSCS of Elastase
DTD 5 ARTICLE IN PRESS
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
solvent molecules included in the model. The alcoholmoiety in ethanol and isopropanol and the N atom inacetonitrile were positioned based on the chemicalenvironment of the protein site, in order to optimizehydrogen-bonding interactions. With the exception ofDMF and TFE, the topology and parameter files used torefine the organic solvent positions were based on similarfunctional groups already present in the XPLOR proteintopology and parameter files.46 The parameters for DMFused with XPLORwere developed by Jorgensen48 and theparameters for TFE were taken from CHARMM22.49
As a final check, all organic solvent molecules wereremoved from each model and a slow-cool simulatedannealing protocol was performed using CNS50 with 10%of the data set aside for cross-validation.51 The startingtemperature for the simulated annealing calculations was2500 K and the final temperature was set to 300 K, with a25 K drop per annealing cycle. The resulting FoKFc (3s)and 2FoKFc (1s) electron density maps were used tocheck the protein structure, the water molecule positionsand to validate the location of the organic solventmolecules. Only the probe molecules for which therewas clear difference density in the omit maps were kept.One or two rounds of positional and B-factor refinementwith the organic solvent molecules included followed,with the entire model being checked and adjusted ifnecessary. The final R-factors and corresponding R-freevalues are shown in Table 1.
T
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
NCORRECCoordinate numbering scheme
For the purpose of analysis, the DMFmodel was chosenas a reference onto which all other models were super-imposed using the least squares superposition function inthe program O.47 Every organic solvent molecule boundat a particular site on the protein has the same number inall of the coordinate sets. The numbering system for theprobe molecules and how it relates to the pockets in theactive site were described in Results.Similarly, equivalent water molecules have the same
residue numbers in all structures.Watermolecules retainedfrom the native elastase structure solved in aqueoussolution (PDB code 3EST) have numbers 301–345. Thisincludes 21 of the 23 buried water molecules.33 Theremaining two buried sites are numbered 406 and 418,respectively. Ten channel water molecules (305, 306, 308,311, 316, 318, 319, 323, 326 and 332) and 14 surface typewater molecules (325, 327, 329, 333, 335, 336, 337, 339 and340–345) were also retained from the original coordinatesand therefore fall into this numbering group. The active sitewater molecules, all of them classified as surface, arenumbered 351–371. Channel water molecules that did notoriginate from the 3EST coordinate set are numbered400–421, with the exception of 406 and 418, which areburied, as mentioned above. Crystal contact water mol-ecules, those that are within 4 A of a symmetry-relatedprotein atom, are numbered 500–558. All of them have thecharacteristics of the surface water type with the exceptionof 500, 503, 549 and554,whichare of the buried type (buriedbetween two protein molecules in the crystal). Surfacewater molecules that did not originate from the 3ESTcoordinate set, are away from crystal contacts and do notbind in the active site are numbered 600–892.
1256
1257
1258
1259
1260
UProtein Data Bank accession codes
The coordinates and structure factors for the tenmodels of elastase soaked in organic solvents have been
YJMBI 57996—28/1/2006—12:13—SATHYA—198707—XML – pp. 1–12/TL
deposited in the RCSB Protein Data Bank. The accessioncodes are given using the abbreviation for each model asspecified in the text and Tables: HEX PDB code aaaa; ETHPDB code bbbb; TFE1 PDB code cccc; TFE2 PDB codedddd; IPR PDB code eeee; IBZ PDB code ffff; ICY PDBcode gggg; ACE PDB code hhhh; DMF PDB code iiii.
Acknowledgements
We thank Karen Allen for help with the originalsoaks. We are grateful to Diana Griffith, DmitriIvanov, Cheryl Kreinbring and Marty Stanton fortechnical help during various stages of this project.Thanks also to Paul Swartz for redoing the Figuresfor publication. This research was supported in partby the NSF PECASE award to C.M. at NC StateUniversity, MCB-0237297.
ED PROOF
References
1. Vajda, S. & Camacho, C. J. (2004). Protein–proteindocking: is the glass half-full or half-empty? TrendsBiotechnol. 22, 110–116.
2. Selzer, T., Albeck, S. & Schreiber, G. (2000). Rationaldesign of faster associating and tighter bindingprotein complexes. Nature Struct. Biol. 7, 537–541.
3. Rajamani, D., Thiel, S., Vajda, S. & Camacho, C. J.(2004). Anchor residues in protein–protein inter-actions. Proc. Natl Acad. Sci. USA, 101, 11287–11292.
4. Keskin, O., Ma, B. & Nussinov, R. (2005). Hot regionsin protein–protein interactions: the organization andcontribution of structurally conserved hot spotresidues. J. Mol. Biol. 345, 1281–1294.
5. Ringe, D. (1995). What makes a binding site a bindingsite? Curr. Opin. Struct. Biol. 5, 825–829.
6. Jones, S. & Thornton, J. M. (1996). Principles ofprotein–protein interactions. Proc. Natl Acad. Sci.USA, 93, 13–20.
7. Najmanovich, R., Kuttner, J., Sobolev, V. & Edelman,M.(2000). Side chain flexibility in proteins upon ligandbinding. Proteins: Struct. Funct. Genet. 39, 261–268.
8. Ma, B., Wolfson, H. J. & Nussinov, R. (2001). Proteinfunctional epitopes: hot spots, dynamics and combi-natorial libraries. Curr. Opin. Struct. Biol. 11, 364–369.
9. Papoian, G. A., Ulander, J. & Wolynes, P. G. (2003).Role of water mediated interactions in protein–protein recognition landscapes. J. Am. Chem. Soc.125, 9170–9178.
10. Reichmann, D., Rahat, O., Albeck, S., Meged, R., Dym,O. & Schreiber, G. (2005). The modular architecture ofprotein–protein binding interfaces. Proc. Natl Acad.Sci. USA, 102, 57–62.
11. Kortvelyesi, T., Dennis, S., Silberstein, M., Brown, L.,3rd & Vajda, S. (2003). Algorithms for computationalsolvent mapping of proteins. Proteins: Struct. Funct.Genet. 51, 340–351.
12. Cummings, M. D., DesJarlais, R. L., Gibbs, A. C.,Mohan, V. & Jaeger, E. P. (2005). Comparison ofautomated docking programs as virtual screeningtools. J. Med. Chem. 48, 962–976.
13. Mattos, C. & Ringe, D. (1996). Locating and char-acterizing binding sites on proteins. Nature Biotechnol.14, 595–599.
T
MSCS of Elastase 11
DTD 5 ARTICLE IN PRESS
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
UNCORREC
14. Mattos, C. & Ringe, D. (2001). Proteins in organicsolvents. Curr. Opin. Struct. Biol. 11, 761–764.
15. Fitzpatrick, P. A., Ringe, D. & Klibanov, A. M. (1994).X-ray crystal structure of cross-linked subtilisinCarlsberg in water vs. acetonitrile. Biochem. Biophys.Res. Commun. 198, 675–681.
16. Yennawar, N. H., Yennawar, H. P. & Farber, G. K.(1994). X-ray crystal structure of gamma-chymotryp-sin in hexane. Biochemistry, 33, 7326–7336.
17. Allen, K. N., Bellamacina, C. R., Ding, X., Jeffery, C. J.,Mattos, C., Petsko, G. A. & Ringe, D. (1996). Anexperimental approach to mapping the bindingsurfaces of crystalline proteins. J. Phys. Chem. 100,2605–2611.
18. Schmitke, J. L., Stern, L. J. & Klibanov, A. M. (1997).The crystal structure of subtilisin Carlsberg inanhydrous dioxane and its comparison with those inwater and acetonitrile. Proc. Natl Acad. Sci. USA, 94,4250–4255.
19. Wang, T. & Zhou, J. (1998). 3DFS: a new 3D flexiblesearching system for use in drug design. J. Chem. Inf.Comput. Sci. 38, 71–77.
20. English, A. C., Done, S. H., Caves, L. S., Groom, C. R.& Hubbard, R. E. (1999). Locating interaction sites onproteins: the crystal structure of thermolysin soakedin 2% to 100% isopropanol. Proteins: Struct. Funct.Genet. 37, 628–640.
21. English, A. C., Groom, C. R. & Hubbard, R. E. (2001).Experimental and computational mapping of thebinding surface of a crystalline protein. Protein Eng.14, 47–59.
22. Miranker, A. & Karplus, M. (1991). Functionalitymaps of binding sites: a multiple copy simultaneoussearch method (MCSS). Proteins: Struct. Funct. Genet.11, 29–34.
23. Goodford, P. J. (1985). A computational procedure fordetermining energetically favorable binding sites onbiologically important macromolecules. J. Med. Chem.28, 849–857.
24. Smith, G. R. & Sternberg, M. J. E. (2002). Prediction ofprotein–protein interactions by docking methods.Curr. Opin. Struct. Biol. 12, 28–35.
25. Dennis, S., Kortvelyesi, T. & Vajda, S. (2002).Computational mapping identifies the binding sitesof organic solvents on proteins. Proc. Natl Acad. Sci.USA, 99, 4290–4295.
26. Silberstein, M., Dennis, S., Brown, L., Kortvelyesi, T.,Clodfelter, K. & Vajda, S. (2003). Identification ofsubstrate binding sites in enzymes by computationalsolvent mapping. J. Mol. Biol. 332, 1095–1113.
27. Eisen, M. B., Wiley, D. C., Karplus, M. & Hubbard,R. E. (1994). HOOK: a program for finding novelmolecular architectures that satisfy the chemical andsteric requirements of a macromolecule binding site.Proteins: Struct. Funct. Genet. 19, 199–221.
28. Verlinde, C. L., Rudenko, G. & Hol, W. G. (1992). Insearch of new lead compounds for trypanosomiasisdrug design: a protein structure-based linked-frag-ment approach. J. Comput. Aided Mol. Des. 6, 131–147.
29. Huc, I. & Lehn, J. M. (1997). Virtual combinatoriallibraries: dynamic generation of molecular andsupramolecular diversity by self-assembly [publishederratum appears in Proc Natl Acad Sci U S A 1997 Jul22;94(15):8272]. Proc. Natl Acad. Sci. USA, 94,2106–2110.
30. Shuker, S. B., Hajduk, P. J., Meadows, R. P. &Fesik, S. W. (1996). Discovering high-affinity ligandsfor proteins: SAR by NMR. Science, 274, 1531–1534.
YJMBI 57996—28/1/2006—12:13—SATHYA—198707—XML – pp. 1–12/TL
ED PROOF
31. Mattos, C. (2002). Protein–water interactions in adynamic world. Trends Biochem. Sci. 27, 203208.
32. Mattos, C., Giammona, D. A., Petsko, G. A. & Ringe, D.(1995). Structural analysis of the active site of porcinepancreatic elastase based on theX-ray crystal structuresof complexes with trifluoroacetyl-dipeptide-anilideinhibitors. Biochemistry, 34, 3193–3203.
33. Mattos, C. & Ringe, D. (2001). Solvent structure. InInternational Tables for Crystallography (Rossman, M. G.& Arnold, E., eds), vol. F, pp. 623–640, KluwerAcademic Publishers, Dordrecht.
34. Meyer, E., Cole, G., Radhakrishnan, R. & Epp, O.(1988). Structure of native porcine pancreatic elastaseat 1.65 A resolutions. Acta Crystallog. sect. B, 44, 26–38.
35. Atlas, D. (1975). The active site of porcine elastase.J. Mol. Biol. 93, 39–53.
36. Shotton, D. M., White, N. J. & Watson, H. C. (1972).Conformational changes and inhibitor binding at theactive site of elastase. Cold Spring Harbor Symp. Quant.Biol. 36, 91–105.
37. Meyer, E. F., Jr, Radhakrishnan, R., Cole, G. M. &Presta, L. G. (1986). Structure of the product complexof acetyl-Ala-Pro-Ala with porcine pancreatic elastaseat 1.65 A resolution. J. Mol. Biol. 189, 533–539.
38. Takahashi, L. H., Radhakrishnan, R., Rosenfield, R. E.,Jr, Meyer, E. F., Jr, Trainor, D. A. & Stein, M. (1988). X-ray diffraction analysis of the inhibition of porcinepancreatic elastase by a peptidyl trifluoromethylk-etone. J. Mol. Biol. 201, 423–428.
39. Takahashi, L. H., Radhakrishnan, R., Rosenfield, R. E.,Jr & Meyer, E. F., Jr (1989). Crystallographic analysisof the inhibition of porcine pancreatic elastase by apeptidyl boronic acid: structure of a reaction inter-mediate. Biochemistry, 28, 7610–7617.
40. Atwell, S., Ultsch, M., De Vos, A. M. & Wells, J. A.(1997). Structural plasticity in a remodeled protein–protein interface. Science, 278, 1125–1128.
41. Buhrman, G., de Serrano, V. & Mattos, C. (2003).Organic solvents order the dynamic switch II in Rascrystals. Structure (Camb), 11, 747–751.
42. Brunger, A. T. (1988). Crystallographic refinement bysimulated annealing. Application to a 2.8 A resolutionstructure of aspartate aminotransferase. J. Mol. Biol.203, 803–816.
43. Kabsch, W. (1993). Automatic processing of rotationdiffraction data from crystals of initially unknownsymmetry and cell constants. J. Appl. Crystallog. 26,795–800.
44. Howard,A. J.,Gilliland,G.L., Finzel, B.C., Poulos, T. L.,Ohlendorf, D. H. & Selemme, F. R. (1987). Use of animaging proportional counter in macromolecularcrystallography. J. Appl. Crystallog. 20, 383–387.
45. Otwinowski, Z. & Minor, W. (1997). Processing ofX-ray diffraction data collected in oscillation mode. InMethods in Enzymology: Macromolecular Crystallo-graphy, Part A (Carter, C. W. & Sweet, R. M., eds),vol. 276, pp. 307–326, Academic Press, New York.
46. Brunger, A. T. (1992). X-PLOR: A System for X-rayCrystallography and NMR, Yale University Press, HewHaven, CT.
47. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M.(1991). Improved methods for building proteinmodels in electron density maps and the location oferrors in these models. Acta Crystallog. sect. A, 47,110–119.
48. Jorgensen, W. L. & Swenson, C. J. (1985). Optimizedintermolecular potential functions for amides andpeptides. Structure and properties of liquid amides.J. Am. Chem. Soc. 107, 569–578.
12 MSCS of Elastase
DTD 5 ARTICLE IN PRESS
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
49. MacKerell, A. D., Jr, Bashford, D., Bellott, M.,Dunbrack, R. L., Jr, Evanseck, J., Field, M. J. et al.(1998). All-atom empirical potential for molecularmodeling and dynamics studies of proteins. J. Phys.Chem. B, 102, 3586–3616.
50. Brunger, A. T., Adams, P. D., Clore, G. M., Gros, P.,Grosse-Kunstleve, R. W., Jiang, J.-S. et al. (1998).Crystallographic & NMR system (CNS): a new
UNCORRECT
YJMBI 57996—28/1/2006—12:13—SATHYA—198707—XML – pp. 1–12/TL
software system for macromolecular structure deter-mination. Acta Crystallog. sect. D, 54, 905–921.
51. Kleywegt, G. J. & Brunger, A. T. (1996). Checking yourimagination: applications of the free R value. Struc-ture, 4, 897–904.
52. Kraulis, P. G. (1991). MOLSCRIPT: a program toproduce both detailed and schematic plots of proteinstructures. J. Appl. Crystallog. 24, 946–950.
1457
1458
1459
Edited by R. Huber 14601461
1462
(Received 3 September 2005; received in revised form 21 December 2005; accepted 5 January 2006)ED PROOF
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512