multiple solvent crystal structures: probing binding sites, plasticity and hydration

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 PROOF
Multiple 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
NCORRECIntroduction
One 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


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


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


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.


MSCS of Elastase 5


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-
Eture
Data 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


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


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


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
C
these 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


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


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


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


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 1126
1127

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


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


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
NCORREC
Coordinate 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


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


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.


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


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


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 1460
1461

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

multiple solvent crystal structures: probing binding sites, plasticity and hydration

Documents