Probing the structure of human protein disulfide isomerase by chemical cross-linking combined with mass spectrometry.

JPROT-01815; No of Pages 16 JOURNAL OF P ROTEOM IC S XX ( 2014) X XX–X XX

Available online at www.sciencedirect.com

ScienceDirect www.elsevier.com/locate/jprot

4Q1

Li Penga , Morten Ib Rasmussena , Anna Chailyana , Gunnar Houenb , Peter Højrupa,⁎

O

5

a

6

b

R O

2

F

3

Probing the structure of human protein disulfide isomerase by chemical cross-linking combined with mass spectrometry

1

Department of Biochemistry and Molecular Biology, University of Southern Denmark, Odense, Denmark Department of Clinical Biochemistry, Immunology and Genetics, Statens Serum Institut, Copenhagen, Denmark

P

7

AR TIC LE I NFO

ABSTR ACT

11 15

Article history:

Protein disulfide-isomerase (PDI) is a four-domain flexible protein that catalyzes the formation of 15

12 16

Received 10 February 2014

13 17

Accepted 24 April 2014

disulfide bonds in the endoplasmic reticulum. Here we have analyzed native PDI purified from 16 human placenta by chemical cross-linking followed by mass spectrometry (CXMS). In addition to 17

E

D

10 9

PDI the sample contained soluble calnexin and ERp72. Extensive cross-linking was observed 18

14 18

within the PDI molecule, both intra- and inter-domain, as well as between the different compo- 19 Keywords:

nents in the mixture. The high sensitivity of the analysis in the current experiments, combined 20

38

Cross-linking

with a likely promiscuous interaction pattern of the involved proteins, revealed relatively 21

39

BS3

densely populated cross-link heat maps. The established X-ray structure of the monomeric PDI 22

40

Protein structure

could be confirmed; however, the dimer as presented in the existing models does not seem to be 23

41

Calnexin

prevalent in solution as modeling on the observed cross-links revealed new models of dimeric 24

42

ERp72

PDI. The observed inter-protein cross-links confirmed the existence of a peptide binding area on 25

43

Placenta

calnexin that binds strongly both PDI and ERp72. On the other hand, interaction sites on PDI and 26

R

E

C

T

19 37 20

44

ERp72 could not be uniquely identified, indicating a more non-specific interaction pattern.

R

28

27 29

Biological significance

N C O

The present work demonstrates the use of chemical cross-linking and mass spectrometry 30 with the chaperones ERp72 and calnexin. The data shows that the dimeric structure of PDI may 32 be more diverse than indicated by present models. We further observe that the temperature 33 influences the cross-linking pattern of PDI, but this does not influence the overall folding pattern 34

50 49

1. Introduction

51

Elucidating the structure of proteins is an essential part of molecular structure-function studies. However, despite the fact that the amino acid sequence is known for most proteins of many species, the structure of many proteins remains unknown or only

52 53 54

35

of the molecule.

U

36 47 46 45 48

(CXMS) for the determination of a solution structure of natural human PDI and its interaction 31

© 2014 Published by Elsevier B.V.

partly characterized. This is due mainly to limitations in the availability of the proteins and in the methods used for protein structure elucidation. X-ray crystallography has for many years been the major method for detailed structure elucidation of proteins [1–6]. This method requires relatively large amounts of pure protein

⁎ Corresponding author. Fax: +45 65 50 24 67. E-mail address: [email protected] (P. Højrup).

http://dx.doi.org/10.1016/j.jprot.2014.04.037 1874-3919/© 2014 Published by Elsevier B.V.

Please cite this article as: Peng L, et al, Probing the structure of human protein disulfide isomerase by chemical cross-linking combined with mass spectrometry, J Prot (2014), http://dx.doi.org/10.1016/j.jprot.2014.04.037

55 56 57 58 59 60

2

75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120

127

2.2. Purification of PDI

137

PDI was obtained during purification of human placenta chaperones as described elsewhere [45]. In short, a placenta was homogenized with Triton X-114, the supernatant was separated by temperature-dependent phase separation, and the supernatant from this was precipitated by ammonium sulfate. The new supernatant was ultradiafiltrated, and the retentate was applied to a Q Sepharose Fast Flow column equilibrated with 20 mM Tris, 1 mM CaCl2, and pH 7.5. Proteins were eluted with a gradient of NaCl (0–0.5 M). Eluted proteins were monitored by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and enzyme-linked immunosorbent assay (ELISA) using antibodies directed against calreticulin, calnexin, ERp57, ERp72 and PDI. The fraction used for the current analysis corresponds to fractions 132–136 in Fig. 3 of [37]. Two different preparations were used, one for the room temperature experiment, and one for the low/high temperature experiments.

138

2.3. Cross-linking

155

50 μg sample (fraction 132–136) was transferred to phosphate buffer (25 mM, pH 7.5) to a concentration of 2 μg/μL using 10 kDa spin filter. Before adding cross-linkers, samples were equilibrated for 30 min under different temperatures, 0 °C on ice water, room temperature and 37 °C separately in a thermostat. Following equilibration, an equal volume of phosphate buffer containing 2500 μg BS3 was added to the samples to a final ratio of 1:50 w/w protein: cross-linker (approximate molar ratio of 1:376), and the cross-linking reactions were carried out under the same temperatures as the equilibration step. After 1 h, the reaction was terminated by the addition of 1 M NH4HCO3 to a final concentration of 20 mM NH4HCO3. After 20 min, the samples were reduced, alkylated and digested by 2% trypsin w/w at 37 °C overnight. Two different batches containing the same proteins in different amounts were used, one for the room temperature assay and another for the 0 °C and 37 °C experiments.

156

F

NaH2PO4, Na2HPO4, NH4HCO3, CH3CN and TFA were from Sigma Aldrich (St. Louis, MO). Bis[sulfosuccinimidyl] suberate (BS3) was obtained from Thermo Scientific Pierce (Rockford, IL). Porcine trypsin was a gift from Novo Nordisk (Copenhagen, Denmark). Spin filters were from Sertorius Stedim Biotech (Bohemia, NY). Poros Oligo R2 and R3 50 were from Applied Biosystems (Foster City, CA). ReproSil Pur C18 AQ 3 μm was from Dr Maisch GMBH (Ammerbuch-Entringen, Germany). Pure water was obtained from an Elga PureLab Flex (Marlow, UK).

O

74

126

R O

73

2.1. Materials

P

72

125 124

D

71

2. Materials and methods

E

70

by the methods described above and when handling proteins 121 from natural sources, that have not been obtained in a very 122 high purity. 123

T

69

C

68

E

67

R

66

R

65

O

64

C

63

and is limited to proteins, which can be made to form suitable crystals. Large amounts of protein can usually be obtained by recombinant technology; however, this often does not lead to suitable crystals. Obstacles to this include inherent mobility or flexibility in multi-domain proteins, heterogeneity, or a lack of a well-defined structure. In such cases, structural information may be obtained for crystallizable domains or for defined parts of the protein. Even so, X- ray crystallography can only give information about solid state structures of proteins and for many proteins, particularly enzymes, molecular dynamics is vital to the function, e.g. for PDI, which must be able to break improper disulfide bonds and to join the proper ones. Protein disulfide isomerase (PDI), which is a 57 kDa protein composed of four thioredoxin domains, two oxido-reductase domains (a and a′) with CGHC active sites, two non-catalytic domains (b and b′), a linker region between domains b′ and a′, and a C-terminal acidic tail [7–9]. The main responsibility of disulfide bonds is to stabilize protein structures, while in some proteins the cysteines perform redox regulation of enzymatic activity. Approximately one third of human proteins contain disulfide bonds [10], and the formation of disulfide bond limits the rate of protein synthesis. Protein disulfide isomerase (PDI), was published in 1963 as the first folding catalyst [11]. It catalyzes disulfide bond formation, breakage and rearrangement, and thus assists proper protein folding [12,13]. PDI is mainly expressed in the endoplasmic reticulum (ER), but also on the cell surfaces of lymphocytes, hepatocytes, and platelets [14,15]. In the ER It also takes part in peptide loading onto major histocompatibility complex class I [16], as well as regulating NAD(P)H oxidase [17]. Although PDI was discovered more than 50 years ago, a complete X-ray structure was not solved until 7 years ago for yeast PDI (yPDI) [18] and has only recently been obtained for human (hPDI) PDI [19]. On the way to solve the full structure, X-ray crystallography models for individual domains were solved and yielded important information about the active sites and substrate binding sites [18,20–22]. Nuclear magnetic resonance (NMR) spectroscopy can yield information about structure and dynamics for proteins in solution and can be used for proteins up to 100 kDa but also requires large amounts of pure protein, which must not aggregate in solution [23–32]. For PDI, NMR spectroscopy has yielded important information about the structure and dynamics of several domains [33–38] but not yet the whole enzyme. Small angle X-ray scattering (SAXS) can yield low resolution structural information for proteins in solution, but has the same requirements for sample amount and purity as NMR spectroscopy [31,39–43]. For PDI, SAXS has shown an annular arrangement of the four domains [44]. Thus, the above methods each have their advantages and disadvantages, but all are challenged by the requirement for sample purity. Here, we show, using PDI and other endoplasmic reticulum (ER) proteins, that chemical cross-linking in combination with mass spectrometry (CXMS) can yield relevant information about the structure, internal dynamics and intermolecular interactions for low concentrations of mixtures of native proteins in solution. CXMS is particularly useful in combination with known structural models obtained

N

62

U

61

JOUR NAL OF P ROTEOM ICS XX ( 2014) X XX–XX X


128 129 130 131 132 133 134 135 136

139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154

157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173

3

JOURNAL OF P ROTEOM IC S XX ( 2014) X XX–X XX

216

The desalted peptides were loaded onto a 100 μm inner diameter and 18 cm RP capillary column (homepacked with ReproSil Pur C18 AQ 3 μm sorbent) in buffer A (0.1% formic acid, 5% ACN) and run on a Proxeon 1100 easy-nLC system (Thermo Fisher Scientific/Proxeon Biosystems, Odense, Denmark). The peptides were eluted using either a 50 min gradient (protein identification) or an 80 min one (cross-link identification) from 0 to 34% B-buffer (95% ACN, 0.1% formic acid) at 350 nL/min flow rate and via nanoelectrospray introduced into an LTQ-Orbitrap XL mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). A full MS scan in the mass area of 300–1800 m/z was performed in the Orbitrap with a resolution of 30,000 FWHM and a target value of 1 * 10(6) ions. For each full scan the top five ions were selected for either collision induced dissociation (CID) for protein identification or for high energy collision dissociation (HCD) for cross-link identification. CID settings were as follows, activation time, 30 ms; isolation width, 2.5; normalized collision energy, 35; repeat count, 1; repeat duration, 30; and exclusion duration, 45. HCD settings were as follows: threshold for ion selection, 15,000; target value of ions used for HCD 2 * 10(5); activation time, 10 ms; isolation width, 2.5; normalized collision energy, 42; repeat count, 1; repeat duration, 30; and exclusion duration, 45.

217

2.6. Data analysis

218

Raw data were converted to mgf format by Proteome Discover 1.4 (Thermo Fisher Scientific). The data were further filtered by MGF Filter V0.104 to retain only the top 125 most intense peaks per scan (in-house built software, now included in the MassAI search tool). For protein identification the data were searched by the MassAI search engine against the Homo sapiens part of the UNIProt database with the parameters of 10 ppm MS accuracy, 0.1 Da MS/MS accuracy, trypsin as enzyme, three allowed missed cleavages, carbamidomethylation of cysteine as fixed modification, and oxidation of methionine as variable modifications. Based on the proteins

185 Q3 186 187 188 189

193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215

219 220 221 222 223 224 225 226 227 228

C

184

E

183

R

182

R

181

N C O

180

U

179

F

192

178

O

2.5. Mass spectrometry

177

R O

191

176

P

190

The samples were reduced by adding dithiothreitol to a concentration of 5 mM, incubating at 56 °C for 30 min followed by alkylation by iodoacetamide (10 mM) at room temperature for 30 min. Digestion of the protein was performed with 2% trypsin (w/w) overnight at 37 °C, and the resulting peptides were separated by strong cation exchange (SCX) chromatography using a polySULFOETHYL A, 150 × 1 mm, 5 μm particle size, 300 Å pore size column (PolyLC Inc., Columbia, MD). The buffer gradient was: 0–0% B (10 min); 0–40% B (30 min); 40–90% B (5 min); and 100% B (5 min) at a flow rate of 50 μL/min (A buffer, 30% ACN, 0.1% TFA; B buffer, 30% ACN, 0.1% TFA, 500 mM NaCl). The separation was performed on an Agilent 1200 HPLC (Agilent Technologies) monitored at 214/254/280 nm. The entire run was manually collected and divided into 4 fractions. Each fraction was desalted using homemade Poros Oligo R2/R3 (R2:R3 ratio: 4:1) RP microcolumns [45].

229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248

2.7. Modeling

249

Two complete structures of human PDI are publicly available, both determined by X-ray crystallography, known as oxidized PDI (4EL1) [19] and reduced PDI (4EKZ) [19]. As our experimental conditions closely match that of the oxidized version, all our further structure analyses are based on the PDI conformation solved in the above-mentioned redox state. The crystal structure of hPDI, 4EL1, shows co-crystallization of two PDI molecules with the a domain of one molecule nested between the a′ and the a domain of the other. This is likely a result of crystallization, and the dimer of the hPDI molecule is still missing. Whereas for the yeast, Tian et al. have shown that the yPDI has a propensity to dimerize in solution and have deposited the structure in PDB with the code 3BOA.

250

1. Dimer structure When aligning yeast and hPDI the sequence identity is 31%. Due to the low sequence identity of yeast and human PDI (not all cross-linked Lys residues are conserved in yeast) it is rather complicated to use it in our analysis, thus a model using the yeast conformation as a template has been constructed. We keep in mind, that the structure of full-length yPDI cannot represent the structure of hPDI, because they are not functionally equivalent and are dissimilar in several aspects, but due to the absence of structural information on the hPDI dimer, we overcome the problem by modeling the hPDI dimer based on yPDI. The following steps have been performed to obtain the above-mentioned dimer structure, in the following called hYeastDimer: 1. As a first step a sequence similarity search on the nr database March 28, 2014 using two iterations of CS-Blast has been performed. 2. We selected the top 100 multiple sequence alignment having an e-value lower than 1e-103. 3. The sequences obtained in the previous step together with the 2 structures i.e. pdb code 4EL1 and 3BOA have been used to produce the structurally correct sequence

263

D

175

identified, a database including only these proteins was generated. Results were further verified by searching the same dataset by the Mascot search engine (Matrix Science, UK). Searches for crosslinked peptides were conducted against this database using the CrossWork algorithm [46]. The initial searches were carried out using the original CrossWork software (www.massai.dk) with the following parameters, cross-linker: BS3; protease: trypsin with allowance for 3 missed cleavages, fragmentation mode: CID/HCD; MS tolerance: 0.010 Da; MS/MS tolerance: 0.025 Da; cross-link search modes included mass matching, level 1 and level 2 search, as described [46]. During preparation of this paper, the CrossWork algorithm itself was incorporated into the MassAI analytical software package (in-house developed; www.massai.dk). The latter was then used for constructing the cross-link heat maps, using search parameters identical to the CrossWork searches. When constructing the heatmaps, low scoring scans (score < 10) were excluded. Representative scans for the cross-links discussed were annotated and exported (Supplementary Fig. 3).

E

2.4. Strong cation exchange chromatography

T

174


251 252 253 254 255 256 257 258 259 260 261 262

264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285

4

300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 319 318 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346

3.1. Cross-linking

351

In the purification of ER chaperones, PDI was greatly enriched in some fractions, but was not completely separated from a number of other constituents as displayed by SDS PAGE (Fig. 1). Peptide mapping of the entire fraction revealed the presence of ERp72 (PDI A4), PDI and calnexin as the major constituent proteins with sequence coverages of 93, 92 and 89%, respectively (Supplementary Fig. 1), along with trace amounts of other proteins. The finding of calnexin was unexpected, as it is synthesized as a transmembrane type I protein, and the purification and characterization of this soluble form of the protein have been described elsewhere [45]. As previous experiments on yPDI have shown great variation in the orientation of the domains [18], the cross-linking experiment was carried out at three different temperatures, 0 °C, room temperature and 37 °C. Each experiment was divided into four fractions and performed in triplicate, thus a total of 45 runs were performed and over 220.000 scans (ms/ms spectra) were recorded.

352

F

350 349

O

299

3. Results

R O

298

348

P

297

low- 347

D

296

and

E

295

probabilities

T

294

solvation, side-chain rotamer weighted electrostatics energy.

C

293

E

292

R

291

R

290

O

289

C

288

alignment of the two above-mentioned structures using PROMALS3D. 4. The sequence alignment has subsequently been used to build the three-dimensional model of hPDI using Modeller (Fig. 6). 2. Dimeric structure cross-link list In order to exclude the cross-links that were observed to fit the monomer model, all the intra-protein Cα cross-link distances have been calculated using the hPDB structure 4EL1 (as a first step all the coordinates of missing residues in the 4EL1 structure have been modeled using Modeller [47]) and a hYeastDimer model with 30 Å threshold (the sum of the length of a BS3 cross-linker and lysine side chains and 5 Å for flexibility). Out of 261 total cross-links identified at room temperature only 123 fitted the monomeric structure of 4EL1 and 117 the hYeastDimer one. The remaining 138 and 144 cross-links were in discordance thus suggesting the existence of alternative conformations. 3. Cross-link ambiguity The 138 cross-links that were in discordance with the 4EL1 structure have been sorted according to the copy number in order to identify the most reliable ones. The total number of cross-links considered for further data analysis was equal to 10 (the number of cross-links seen more than 5 times; this number being a sum of mean (2.6) + sd (2.2) of the data). But since the hPDI is a homodimer, the cross-link list determined experimentally between the residues X and Y and obtained in the previous step can be represented by 2 scenarios: 1. an inter-protein cross-link between the residue X of chain A and the residue Y of chain B 2. an inter-protein cross-link between the residue X of chain B and the residue Y of chain A. Therefore, the total number of the most reliable 10 crosslinks should be multiplied by two resulting in 20 cross-links, but since one of the cross-links is symmetrical (the residue 54 does a cross-link with itself, there is no need to consider 2 scenarios i.e. residue 54 chain A–residue 54 chain B and vice versa as in the remaining 9 cases). 19 cross-links have been selected for the further conformational searches to produce models of the corresponding dimer complexes. 4. Dimer complex search Dimer models of PDI were created based on the monomeric unit of 4EL1 using the RosettaDock modeling suite (http:// www.rosettacommons.org). RosettaDock is a well-established docking protocol that incorporates physical and empirical energy functions and fast conformational searching techniques. The algorithm searches the rigid-body and side-chain conformational space of two interacting proteins, making it suitable for modeling protein complexes starting from the free monomers. It is a multi-scale Monte Carlobased algorithm that combines a low-resolution, a centroidmode, a coarse-grain stage and a high-resolution all-atom refinement stage that optimizes both rigid-body orientation and side-chain conformation. Scoring in the low-resolution phase includes residue–residue contacts and bumps, knowledge-based terms for residue environment and residue–residue pair propensities. In the high-resolution phase, the energy is dominated by van der Waals energies, orientation-dependent hydrogen bonding, implicit Gaussian

N

287

U

286


80 70

d

60

c

50

b a 40 30 25 20

15

1

2

3

4

5

1

Fig. 1 – SDS PAGE of the proteins used for cross-linking. Lane 1: molecular marker; lane 2: non-cross-linked proteins; lane 3–5: proteins cross-linked by BS3 at room temperature (lane 3) at 0 °C (lane 4) and at 37 °C (lane 5). Proteins in lane 2 are a, b: calnexin; c: PDI; d: ERp72.


353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369

18 1

18

1

31

14

42

25

57

40

65

48

69

52

71

54

81

64

103

86

114

97

130 113 131 114 162 145 170 153 195 178 200 183 207 190 208 191 222 205 230 213 247 230 254 237 263 246 271 254 276 259 283 266 285 268 308 291 309 292 326 309 328 311 350 333 352 335 366 349 370 353 375 358 385 368 386 369 401 384 409 392 415 398 424 407 436 419 444 427 467 450 468 451

31 14

13

42 25

57 40

100 6

1

65 48

69 52

71 54

81 64

103 114 130 131 162 170 195 200 207 208 222 230 247 254 263 271 276 283 285 308 309 326 328 350 352 366 370 375 385 386 401 409 415 424 436 444 467 468 502 505 86

97

113

114

145

153

178

183

13

30

118

102

42

31

11

16

4

7

50

44

71

27

4

14

6

2

6

5

3

18

5

8

83

1

12

3

27

20

11

7

47

N

1

18

5

12

9

57

11

4

8

14

5 1

U

190

191

205

26

6

2

213

230

237

246

254

259

266

291

292

309

311

3

3

5

2

8

10

3

6

13

6

2

5

6

4

8

10

9

268

333

335

349

353

1

358

368

6

3

2

2

369

384

392

427

450

2

3

13

5

2

8

1

2

1

2

3

4

398

407

419

451

485

488

2

2

3 2

1

C

1

O

4 1

1

1

2

8

1 2

1 1

1 1

1

6

5

4

6

9

9

16

2

4

7

2

10

6

5

26

22

2

8

7

34

102

8

13

2

1

3

1

7

5

2

3

1

1

2

1

7

1

1

4

2

2

2

5

9

1

1

2

2

13 5

4

2

2

1

10

4

2

2 2

2

1 1

3

2

1

3

3

39

3 2

R

4 1

1 1

4

1

2

2

5

R 3

1

1

1

5

6

4

11 1

7

2

1 6

13

E

2

C

2

8

15

5

15

1

3

3

1

1

2

1

1

1

1

2

3

1

2

2

13

5

4

6

1

2

12

12

14

6

18

1

5

4

10

3

6

10

20

7

8

T 1

E

2

5

1

1 9

8

1

1

1

2

3

1

1

D

1

3

2

2

1

1 2 5

1

1

1 3

1

3

1

3

1

1 1

2

4

28

29

5

9

10

1

8 6

P

6

7

1

1 1

18

1

10

5

2

2

15

6

17

4

53

4

1

2

3

11

R O

8

1

13 2

3

1

1

1 1

2

1

1

1

1

4

1

1

1

1 4 1

1

1

1

1

O

1

1

2

1

2



A

5 2

F

1 11 2

502 485 505 488

5

Fig. 2 – Heat maps showing all observed cross-link scans (spectral observations) in PDI obtained by cross-linking with BS3 and observed with a CrossWork score of at least 10. Total cross-link observations for all three temperatures and repeat experiments combined. Coloring is relative to the total number of observed scans. Hot spots, where at least 10 scans were observed, are shown in dark color. The horizontal and vertical lines are drawn to separate the four domains. The outermost numbering is the lysine numbers according to the nucleotide sequence, the inner numbering is after removal of the signal peptide. Heat maps for individual temperatures and inter-protein interactions are presented in Supplementary Fig. 2.

6

b′

a′

1067

336 36

295 111 325

8

13

146 25 207 33 15

13

396

3.2. Observed cross-links

397

Looking at the heat map of all observed cross-links, it is clear that most of the intra domain cross-link scans are present in the a domain (Fig. 2, Table 1). This was not caused by a larger number of amino groups (lysines + N-terminus) present, as the number of amino groups is similar to b′ and less than a′ (Table 1). The most likely reason could be greater detectability of the peptides, as cross-linked peptides by their nature are larger. Looking at the number of scans per peptide and sorting by mass, a correlation between number of observed scans and mass can be seen (Fig. 3A). Here is a clear trend for large peptides (>2000 Da) to be less represented, with only three peptides (m/z 2010, 2093 and 2639) being highly represented. Both peptides 2093 and 2639 are arginine-terminated, which could play a role due to the higher

379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394

398 399 400 401 402 403 404 405 406 407 408 409

C

378

E

377

R

376

R

375

O

374

C

373

N

372

U

371

3.3. Intradomain and a–b inter-domain cross-links

451

Comparing the observed cross-links with the X-ray structure of intact hPDI (PDB 4EL1) [19] shows that all observed intra-domain hot spot cross-links fit within the allowed Cα–Cα distance of 25 Å for the BS3 linker (Fig. 4; representative assigned ms/ms scans are shown in Supplementary Fig. 3). This is expected, as each domain has a diameter of 25–30 Å, and most amine groups should thus be reachable by cross-links created by BS3. Comparing links between the a and b domains, cross-links between Lys200, Lys208 and Lys222 in the b domain and the N-terminus and lysine 71, 81, 103 and 114 are abundant, but

452

T

395

Running the native and cross-linked proteins on an SDS-PAGE gel revealed some dimerization of PDI and the disappearance of gel bands representing calnexin and ERp72, indicating that there was interaction between the proteins separated in the gel, either as homo- or as hetero-mers. After chemical cross-linking and strong cation exchange chromatography, followed by mass spectrometry, we identified cross-linked peptides originating from all three proteins. Cross-linking of ERp72 and calnexin individually has been performed previously, and the intra-protein cross-links have been described elsewhere [18,46]. Most of the cross-links in these experiments were observed for PDI. Combining the 15 runs from each temperature triplicate experiment and plotting the observed number of scans (spectral observations) for each cross-link into a grid, we generated a heat-map for all observed cross-links (Fig. 2), along with heat-maps for each individual experiment (Supplementary Fig. 2A–C). Data for the observed interprotein cross-links (PDI/CNX, PDI/ERp72 and ERp72/CNX) are presented in Supplementary Fig. 2D–F. The heat maps were created in MassAI with varying cutoffs in CrossWork score [46] and by checking the related spectra manually. Higher CrossWork scores required for inclusion in the heat map, obviously increased the confidence of the observation, but increasing to more than 10 favored peptides with a particularly good fragmentation rather than increasing the confidence of identification.

F

b

O

370

a b b′ a′ NH2

a

R O

t1:7 t1:8 t1:9 t1:10 t1:11

Domain

410

P

t1:6

pKa of arginine. However, when looking at the observed peptides containing a dead-end BS3 linker (Fig. 3B), arginine containing peptides are not preferentially observed, and the loss of high-mass peptides is only first observed above MH+ 2700. As tryptic cross-linked peptides are typically triply or quadruply charged, these cross-peptides should be well within the mass range of the instrument (m/z 1800), but are clearly disfavored. The three high-occurrence high-mass peptides thus make almost no cross-links with other high-mass peptides, only with small mass peptides and among themselves. Two low mass peptides are not observed, MH+303 and 674. The peptide MH+303 contains Lys131 and will be seen as a dipeptide (Lys-Arg), which is filtered out by the search program as a peptide with high probability of false identification. Peptide MH+ 674 contains Lys505 and is neither observed as cross-linked nor as a linear peptide (Supplementary Fig. 1). This can be due to the low mass of the peptide, but may also be caused by its absence. The last residue observed with high confidence is residue Lys468, however, a lower confidence peptide spans up to residue Lys502, and it is thus likely that the KDEL sequence or a large part of the acidic C-terminal part is missing. Calnexin purified in the same fraction is missing the C-terminal transmembrane region [45], and for calreticulin, obtained by the same purification scheme, it has been observed that a fraction was missing the C-terminal 6 residues [48]. It thus seems that a part of the ER proteins from the placenta is C-terminally truncated. The low number of observed cross-links relating to the b domain is most likely related to the fact that of the eight lysine residues present, only three peptides with one missed cleavage have a mass between 500 and 2000 Da (Supplementary Table 1). Omitting the ten cross-links least observed (i.e. two small peptides, 8 large peptides, thus going from 48 amine groups to 38) the number of total observed scans decreases from 2581 to 2571 while the number of different observed cross-links drops from 340 to 333. As the number of potential cross-links drops from 1176 to 741 this means that 45% of all remaining potential cross-links are observed at least once, indicating that a number of short-lived interactions are likely, and only multiply observed cross-links with manually validated scans are thus considered in the following. Cross-links which are observed in at least 10 scans are in the following labeled ‘hot-spots’.

D

t1:5

Table 1 – Total number of intra and inter cross-links observed per domain for PDI with a CrossWork score > = 10. Bottom line shows number of amino groups in each domain (N-terminus and lysine residues).

E

t1:1 t1:2 t1:3 t1:4


Fig. 3 – A. Total number of observed cross-linked peptide scans sorted after mass (M + H+) of the smallest tryptic peptide containing the linking residue (i.e. tryptic peptide with one missed cleavage) and having a CrossWork score of at least 10. Arginine-terminated peptides are indicated with a red ‘R’. The domain relation of each peptide is shown with colored letters, yellow (a), blue (b), orange (b′) and green (a′). List of peptides are shown in Supplementary Table 1. B. The same peptides observed with dead-end cross-linkers. Please cite this article as: Peng L, et al, Probing the structure of human protein disulfide isomerase by chemical cross-linking combined with mass spectrometry, J Prot (2014), http://dx.doi.org/10.1016/j.jprot.2014.04.037

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

453 454 455 456 457 458 459 460 461

7


A

600

F

400

O

300

R O

Number of observations

500

200

R

R

RR

R

R

R

R

R

P

100

R

R

3885.54

3544.7

3082.47

2865.36

2709.46

2093...

2540.29

2093.05

R

2313.14

2019.03

R

2019...

D

1965.97

1826.09

E

1729.91

1466.69

1661.89

1451.7

1342.7

1330.7

1189.66

1094.66

938.49

988.49

808.46

674.37

303.13

a a a'ba'a'b'a'b'a b'a a a a a'b'b b'b'a a a'a'a'a'b'b'b'bb'a'a a'a'b'a'b'b'aa'b a b b ba'a' 0

800

C

B

T

Peptide mass

R

E

700

R

N C O

500

400

300

U

Number of observations

600

200

R

R

RR

R

R

R

R

R

3885...

3544.7

3082...

2865...

2709...

2540...

2313...

1965...

1826...

1729...

1661...

1466...

1451.7

1342.7

1330.7

1189...

1094...

988.49

938.49

808.46

674.37

0

303.13

100

Peptide mass


8

N

18

31

42

57

65

69

71

81

103

114

130

131

162

170

195

200

207

208

222

230

247

263

271

276

283

285

308

309

326

328

350

352

366

370

375

401

409

415

424

-

19.3

15.6

26.2

19

17.4

12.7

16.9

22.8

25.9

22.7

25.9

39.1

30

48.1

45.9

44.8

47

33.1

39.9

40.8

63.1

64.6

57.3

50.6

49.7

64.6

65

56.9

62.3

64.9

63.4

84.1

77.4

70.3

78.8

77.8

71

81.1

84.5

71.7

58.4

66.6

78.1

77

-

18.7

22

24.8

26.8

24.8

11.5

15.7

19.9

27.5

29

45.2

33.7

51.9

49.5

42.5

44.6

31.7

34.3

31.8

55.5

56.7

49.9

42.4

39.6

53

54.5

43.3

49.6

51.5

49.1

67.8

60.7

54.8

60.8

59.8

55.4

64

67.1

55

41.6

48.8

59.9

58.5

-

27

20.5

18.1

14.9

68

-

13.3

19.2

22

12.5

17.7

22.6

24.5

-

6.1

10

13.7

17.8

21.2

15.3

-

5.4

16.7

19.2

21.5

12.1

-

16.4

19.5

21.6

12.9

-

9.9

15.3

19.2

-

5.7

15.2

15.4

25.8

25

-

13.9

12.9

33.2

24.4

35.3

36.3

24.2

25.5

21.7

19.7

15.2

37.7

39.9

33.6

-

3.8

28.6

23.2

31.1

34.1

26.3

27.3

25.3

27

25.6

43.6

46.6

40.5

-

27.5

22.5

28.6

32.2

22.7

23.6

23.9

24.2

22.6

39.9

42.9

36.9

-

12.2

14.4

8.4

23.7

26.6

18.1

25.3

36.3

48.1

45.5

37.7

-

22.9

16.7

23.3

26.8

8.5

19.3

29.9

47

44.7

36.1

-

12.4

16.9

18.2

25.1

25.3

33.8

38.3

36.4

-

21.6

24.6

19.1

23.8

36.1

44.8

40.7

-

3.8

20.2

12.6

17.6

24.6

23.4

-

23.9

15.7

-

12.3

31 42 57 65 69 71 81 103 114 130 131 162 170 195 200 207 208 222 230 247 263 271 276 283 285 308 309 326 328 350 352 366 370 375 385 386 401 409 415 424 436 444 467 468

16

13.6

13.4

13.3

C 14.7

17.4

49.4

41.8

51.1

50.1

73.1

66.9

58.2

62.2

72.3

74.9

43.6

54.5

56.9

44.9

45.3

43

42.2

34

53.3

58.3

53.8

50

48.4

51

52.1

48

54.1

60

56.2

68.5

61.9

58.5

67.2

65.1

51.3

62.7

68.4

55.7

45.3

51.8

61.4

60.5

36

46.3

48.9

40.6

41.3

37.9

39.6

34.9

53.9

58.3

53

49.2

48.9

55.3

55.5

52.3

57.6

63.2

60.5

76.7

70.5

65

75.7

73.9

61.1

73

78

64.5

53.3

61.3

72.2

71.6 75.7

43 37.3

O

24.5

26.8

48.7

35.3

35

51.4

51.5

43.6

48.4

68.9

61.5

48.3

57.7

70.6

468

52

38

18.8

467

27.6

35.6

31.9

444

42.9

16.7

29.5

436

26.2

21.2

33

386

18.4

15.7

34.8

385

70

32.1

42.1

44.6

38.3

39.3

35.1

38

35.3

54

57.8

52

48.2

48.4

56.8

56.7

53.5

58.4

63.8

61.6

79.7

73.6

67.1

78.5

76.9

65

76.9

81.4

67.7

56.2

64.8

76.3

29

41.5

42.5

38.5

40

32.5

37.1

36.1

55.9

58.9

52.4

47.7

47.8

58.9

58.9

54.2

59.1

63.6

61.8

81.4

75.2

68.1

79.1

77.7

67.5

78.9

83

69.4

57.2

65.9

77.7

77

33.2

47.9

48

39.1

40.5

32.6

34.1

29.5

51.8

54.8

48.7

43

41.3

50.8

51.7

44.3

50.4

54.5

51.7

68.9

62.2

56.5

65.2

63.6

54.6

65

69.2

56.2

43.7

51.5

62.7

61.7

33.6

32.2

42.1

37.7

46.7

59

58.5

28.5

27.7

38.8

39.2

32.6

37.7

42.4

40.3

60.8

54.8

46.7

58.7

57.4

49.4

60.1

63

49.3

36.9

46.5

59.4

59.2

37.1

38

48

47.5

44.6

48.7

54.1

52.5

72.8

67.2

59.1

72.2

70.8

60.5

72.1

75.7

61.7

50.1

59.7

72.3

72.2

34

35.1

44.8

44.2

41.8

45.6

51.2

49.7

70.3

64.8

56.4

70.1

68.7

58.5

70.1

73.5

59.5

48.1

57.9

70.7

70.7

33.8

37.1

57.2

55.8

51.2

52.6

54

55.8

83.1

78.5

66.7

80.8

80.5

77.3

87

87.6

74.5

62.6

73.2

87.1

87.4

29.6

31.3

53.2

52.6

44.8

47.6

48.5

49.6

76.8

71.5

60.4

72.7

72.4

70.1

79.4

80.2

67.1

54.3

64.6

78.4

78.2

30.6

30.7

35.4

50

47.8

47.6

47.7

51

52.7

78

74.3

62.6

78.9

78.4

73

83.1

83.9

70.8

60.8

71.4

85.1

86

33.1

29.8

33.8

54.6

53

48.5

48.9

49.7

52.2

80.2

76.1

63.6

78.1

78.1

76.4

85.4

85.3

72.6

61.4

72

86

86.5

16.8

16.9

21.1

34

32.3

31.2

31.8

36.6

37.1

61.6

57.6

46.2

62.7

62

56.4

66.3

67.2

53.9

44.1

54.7

68.3

69.3

27.9

40.4

41.1

29.8

R

31.1

R

E

19.7

-

17.9

42.3

21.7

44.9

18.5

31

62.4

60.4

56

37.1

37.3

46.9

46.6

37.2

39.9

40.2

41.4

69.2

63.9

52.5

64.6

64.4

63.7

72.3

72.6

59.8

46.9

57.2

71.1

71

34.8

34.4

26.1

28.3

30.3

31

58.2

53.4

41.6

55.6

55.2

53.6

62.2

62.4

49.5

37.7

48.3

62.3

62.8

-

23.9

25.4

20.1

16.5

24.8

25

19.1

23.2

29.5

27.3

48.9

43.7

34.1

48.9

47.6

41

51

52.8

-

10.5

16.1

24.7

27

14.7

11.3

23.8

21.1

30.9

29.4

43.9

42

32.6

51.7

50.3

41.6

50.8

51.7

39.8

36.4

44.8

56.4

58.5

-

9.1

18.8

21.8

19.6

17.2

22.3

17.9

24.6

25.3

45.1

43.5

31.8

50.3

49.7

46.6

53.9

53

41.9

37.5

46.2

58.4

60.5

-

10.2

14.3

23.8

22.2

20.7

18.2

22.8

24.2

48.9

46.1

50.9

50.5

48.9

56.5

36.5

46.2

59.4

60.9

-

5.5

28.8

28.3

19.5

19.5

20.7

22.5

50.8

46.8

34

49

48.9

49.9

57.1

56

44

34

44.2

58

58.9

-

28.2

28.5

16.1

17.4

17.5

18.8

47.7

43.3

30.6

44.4

44.4

46.8

53.4

52.2

40.3

29.6

39.7

53.5

54.3

-

3.8

16.4

15.2

26.3

22.2

31.1

41.9

-

18.6

16.6

27.7

24.3

33.1

31.2

23.4

42.2

40.4

30.8

39.6

40.7

29

27.8

34.9

45.6

48

-

7

13.9

9.2

32.2

27.6

15.7

31.7

30.9

31.4

37.7

36.8

24.5

16.1

25.6

39

40.4

-

11.2

8.3

39.6

37.8

26.7

21.4

29.6

42.2

44

-

6.3

35.2

32.2

18.7

31.1

31.9

41.5

44

40.1

31.4

25.2

32.3

44.5

45.8

-

30.4

26.7

13.2

27.2

27.4

35.2

38.2

35

25.4

19.2

26.4

38.9

40.3

-

7.4

17.3

19.5

17.1

15

10.2

13.3

26.3

21.5

22.2

25.8

-

13.7

16.2

13.1

16.8

12.6

9.8

6.3

19.4

14.2

17.6

20.9

-

19.1

18

24.6

25.7

22.5

13.6

15.4

18.1

28.3

30.6

-

3.8

30.1

22.9

15.8

18.4

22.9

16.8

19.1

20.3

-

26.4

19.3

12.8

15.1

20.9

13.8

15.9

17.5

-

13.6

21.2

12.7

20.8

17.4

20.2

22.5

-

9.7

13.2

25.8

17.4

11.4

-

14.1

26.7

18.3

13.4

17

-

14.2

9.9

16.9

19.6

-

10.6

24.6

25.3

-

14

14.9

-

3.8

31.6

R O

28.5

15.4

O

33

32.5

27.4

34.4

21.2

36.1

66.3

50.3

23.2

36.9

65.2

63.8

12.8

38.8

55

60.3

22.2

20.2

61.8

49.6

11.2

33.5

62.7

58.2

30

27.7

45.6

59.7

18.5

30

56.7

49

38.8

P

31.3

43.4

27.1

D

30

46

41.9

E

32

41.4

30.1

17.1

22.7

35.8

24.2

T

16.3

42.8

14.2

C

21.7

38.8

55.6

37.3

53

39.1

43.8

25.3

44

28

24.3

54.4

38.5

67.9

52.1

69

52.7

44.2

15

-

F

Fig. 4 – Cα intra-protein distance map for N-terminal and lysine residues for PDI structure 4EL1. Distances are color-coded as follows: green < 20 Å; yellow 20–30 Å; orange 30–40 Å; red > 40 Å. The distances for Lys254 are based on modeling of residues 250–254, as these residues are not resolved in structure 4EL1. Numbering is the same as for Fig. 2.



U

0 18

9


522

3.5. b′–a′ inter-domain cross-links

523

536

Although a number of cross-links can be found between the b′ and the a′ domains, most of these are of low copy number. However, Lys375, which is situated in the interface between the a′ and the b′ domains, links strongly to five lysine residues on the nearby surface of b′: 308, 309, 326, 328, and 352, all of which are within reach in the 4EL1 structure. Lys375 also links to Lys285, which is on the edge of the far side of the b′ domain, which may only be reached if the a′ domain rotates and/or bends around the X-linker region. Lys385 in the b′ domain also cross-links to a lesser degree with the same lysines in the a′–b′ interface, but as these Cα distances are > 40 Å, it hints that the a′ domain may be able to rotate and/or shift relative to the plane of the other three domains.

537

3.6. Longer reaching inter-domain cross-links

538

10 hot spot cross-links are observed between the a and b′ domains (Fig. 2). Aside from the underrepresented long peptide lysine residues 247, 254, 263, 285, and 350, the linkages are quite evenly distributed with few strong hot spots. Lys114 and Asp18 (N-terminus) both link to Lys308 and Lys328 (distance 38–62 Å), and Lys328 in addition links to Lys31 (43 Å). These linkages are all impossible in the monomeric structure, and even comparing to the potential dimeric structures (distance maps in Supplementary Fig. 5) only a single linkage can be supported. Cross-links between domains a and a′ domains are somewhat scattered and without clear hot spots, except for Lys467 that links to the N-terminus and to Lys103. In the monomer, these residues are at a distance of more than 50 Å. In the hYeastDimer model there are interacting areas between a and a′ domains (Supplementary Fig. 5), but the cross-links found still have distances larger than 30 Å.

516 517 518 519 520

524 525 526 527 528 529 530 531 532 533 534 535

539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554

T

515

C

514

E

513

R

512

R

511

N C O

510

U

509

3.8. Modeling of the dimeric structure

584

As described above, most cross-links with multiple observations can be assigned to intra-domain distances. However, of the 27 inter-domain hot spots observed, 17 cannot be explained by the monomeric structure. Although the crystal structure 4EL1 shows two PDI molecules, these most likely do not represent a solution dimer, and the distance map (Supplementary Fig. 5A) does not support this either. Based on these conclusions, we modeled the PDI dimer using monomeric crystallized structure available in PDB using RosettaDock. We generated 20,000 rigid body structure predictions in a blind run of RosettaDock. For each of the 20,000 predicted structures, we checked the distances between cross-linked residues not satisfied in the monomer (Table 2) to find the PDI homodimer model structure that simultaneously satisfies the largest number of cross-links. In this docking we used a cross-linking distance of 30 Å in order to allow slightly more flexibility in the interacting molecules. Following this strategy we expected to identify the best among the 20,000 models as a result of an extensive exploration of the conformational space. Fig. 5 shows a heat map of the co-occurrence of all the 19 cross-links (10 cross-links A-B, 10 cross-links B-A, 1 of which is symmetrical). As can be seen, there are 2 large mirror (A-B and B-A) groups of cross-links, that tend to be satisfied together and few times or never with the other cross-links. In this big group we observe 6 cross-links which are satisfied at the same time (the maximum observed number) and are represented by

585

F

521

The cross-links between the b and b′ domains are mainly centered on Lys200 in the b domain which cross-links to Lys247, Lys276, and Lys 308. The distances to Lys247 and Lys276 are within linking distance in the crystal model, while Lys308 is located on the opposite side of the b′ domain (along with the relatively high copy-linked residues Lys326 and Lys328) indicating inter-protein interaction as no rotation about linker regions can bring these residues into close contact. In the X-ray structure these residues are on the convex ‘outside’ of the dimer, and cannot be accommodated by our dimer models (Supplementary Fig. 5), but necessitate a different form of dimer interaction. Further three 7–9 scan observations with a distance of 30–32 Å indicate some flexibility in the b–b′ interface.

556

O

508

505

The cross-linking of PDI was performed at three different temperatures, 0 °C, room temperature and 37 °C. When making a cutoff minimum score of 10, 525 scans were obtained in the 0 °C experiment, 1243 scans in the room temperature experiment and 813 scans in the 37 °C experiment. The differences in number of scans obtained are likely explained by the fact that the 22 °C experiment had a different composition than the 0 °C and 37 °C experiments, and for the two temperature experiments, there is a difference in the reactivity of the cross-linking reaction. Comparing the extremes of temperatures, 0 °C and 37 °C, only a few differences can be seen (Supplementary Fig. 2A and 2C). Going from zero to 37 °C, most hot spots in the a domain change, but this is mainly the result of the number of observed cross-links in each position, as the pattern is the same. As these residues are well within interaction range in the PDI monomer, the increased number of cross-links at 37 °C is likely a result of increased chemical activity of BS3, but may also be due to an increased flexibility of the PDI molecule, as the distribution of hot spots has changed slightly. For intra-linkages in the other domains the number of linkages is quite small and the changes minimal. Two residues, Lys81 and Lys208, show a much increased reactivity at increased temperature, both of which are stabilized in the crystal structure by salt bridges to glutamic acid residues (Glu23 and Glu151 respectively). Raising the temperature thus clearly diminishes this interaction, making the lysine residues available for chemical reactions and thus cross-linking.

R O

3.4. b–b′ inter-domain cross-links

504

555

P

507

503

3.7. Influence of temperature on cross-linking

D

506

also links to Lys31 and 130 are observed. Except for Lys103– Lys222 all Cα distances are more than 30 Å and most exceed 40 Å (Fig. 4A), showing that the domains have to move relative to each other to accommodate the cross-links. As Lys71 and Lys81 are located on the distant face of the a domain, this may even have to rotate in order for the linkage to occur.

502

E

501 500 499 498 497 496 495 494 493 492 491 490 489 488 487 486 485 484


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

586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611

10

615

Cross-links were also observed between both PDI and calnexin, and between PDI and ERp72 as well as all the three major components in the mixture, PDI, ERp72 and calnexin. The observed cross-links are presented as heat-maps in Supplementary Fig. 2A to 2C and the data in Supplementary Tables 3, 4 and 5 respectively.

T C E

291_191

R

191_291 268_358

R

358_268 450_86 450_14 86_450 14_450 54_311 14_311 1_191 1_183 14_183 311_54 311_14 54_54 191_1 183_1

183_1

183_14

191_1

54_54

311_54

311_14

14_450

86_450

450_14

450_86

268_358

358_268

291_191

183_14 191_291

620

O

619

C

618

N

617

U

616

R O

3.9. Inter-protein cross-links

P

614

E

613

3 different solutions that are shown in Fig. 6. Table 3 describes each one of them with the list of the satisfied cross-links.

O

F

6 13 9 8 6 9 7 13 8 7

D

1–183 1–191 14–183 14–311 14–450 54–54 54–311 86–450 191–291 268–358

1_183

612

Cross-link copy number

14_183

t2:6 t2:7 t2:8 t2:9 t2:10 t2:11 t2:12 t2:13 t2:14 t2:15

Residue number

1_191

t2:5

When cross-linking to calnexin both PDI and ERp72 show a very uneven distribution of crosslinks, with a very high number of observed cross-links to certain lysine residues and none to others. Plotting these lysine residues onto the structure of dog calnexin, which is conserved relative to the human calnexin (e.g. all lysine residues are conserved), a pattern can be seen where all the high copy number linkage positions are located on the same surface close to the peptide binding area as proposed by Chouquet et al. [49] (Fig. 7). For PDI, the cross-linking to calnexin is clearly dominated by interactions in the a domain (supplementary Fig. 2A), and presenting the high-copy number linkage positions on the structure (Fig. 8) clearly shows that the main interactions are to the a domain and the linking regions of b–b′ and b′–a′ (the x-linker region). For the interaction of PDI and ERp72, the linkage pattern is much more widely distributed, but the highest copy-number linkage positions are either identical to PDI-calnexin or are neighboring residues. For the linking to ERp72, the two most linking residues (Lys573 and Lys571) are located in the a′ domain and are the same for interaction with both calnexin and PDI (although most pronounced for calnexin). This is the same for the third most reacting residue, Lys 401 in the b–b′ domain, while the next most linking residues are scattered through the a and a′ domains but with no overlap. The main interaction thus seems to be the a′ domain for the interaction with either protein but with scattered

14_311

t2:4

Table 2 – Cross-links used for modeling the PDI dimer. These are the cross-links that do not fit the monomer PDI model and are observed more than five times.

54_311

t2:1 t2:2 t2:3


Fig. 5 – Heat map of satisfied (Cα distance of 30 Å) cross-link co-occurrences of 20,000 models. Heat colors: small values (co-occurrence of 2 cross-links that are under the above-mentioned cutoff) are dark blue, progressing to higher values as oranges and yellows. Please cite this article as: Peng L, et al, Probing the structure of human protein disulfide isomerase by chemical cross-linking combined with mass spectrometry, J Prot (2014), http://dx.doi.org/10.1016/j.jprot.2014.04.037

621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646

11


a

A.1

B.1

a

a

a

O

F

B.2 a

a

A.2

a

R O

a

D.1

a

a

E.1

E

D

a

P

C.1

C

T

a a

D.2

a

a

E.2

647 648 649 650 651 652

a

Fig. 6 – Models of the PDI dimer created using RosettaDock and the constraint of cross-links defined in Table 2. A: the two PDI molecules observed in PDB structure 4EL1. B: the model of the humanized dimer of the yeast PDI (hYeastDimer) based on 3BOA [21]. C–E: the three modalities observed after modeling. The white PDI molecule is shown in approximately the same position in all panels. The bottom panel (labeled X.2) shows the model above (X.1) rotated 180° about the vertical axis. The a domain of each PDI molecule is indicated.

U

Q2

a

N C O

a

R

R

E

C.2

a

a

interactions throughout the protein. These interactions are difficult to map onto a structure, as only theoretical models of ERp72 exist [46,50]. Mapping onto the top model reported [46] showed no clear pattern, with several of the lysines being partially buried in the structure leaving the results inconclusive (results not shown).

654 653

4. Discussion

655

The combined use of high-resolution methods for protein structure elucidation (e.g. X-ray crystallography, NMR

656

spectroscopy), low-resolution methods (SAXS, CXMS, electron microscopy, and others) and molecular biological methods is needed to gain a detailed understanding of protein structure and function. Each method has its advantages and limitations and they complement each other in several ways. An excellent example of this is PDI. This enzyme plays an important role in the proper folding and quality control of ER proteins and collaborates with many other ER chaperones and folding catalysts. Due to the importance of PDI, it has been suggested to be involved in the development of several diseases like AIDS [51,52], ovarian cancer [53], thrombosis [54–58], and neurodegenerative diseases [59–62] mainly due to its important chaperone functions.


657 658 659 660 661 662 663 664 665 666 667 668

12

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

59

89

84

75

4EL1_08502

50

26

49

15

56

91

77

89

71

49

4EL1_09304

23

16

30

27

74

107

62

102

81

93

Model number

462_183

462_191

475_183

475_311

4EL1_01827

10

28

28

47

78

59

89

4EL1_08502

9

23

27

28

68

49

69

4EL1_09304

29

25

24

65

100

64

103

547_450

652_291 84

F

515_311

86_911

729_358 75

62

32

88

95

O

475_450

54_772

P

R O

cannot be observed for the peptides derivatized by a hydrolysed BS3 molecule (dead-end linkers). In the case of PDI, particularly the a and a′ domains have been shown to be mobile relative to the b–b′ domains [13,38,66]. Thus we can expect a large number of low frequency cross-links spanning a large number of lysines. In the present experiment we cross-linked with BS3 digested with trypsin and after LC–MS/MS analysis we searched the data with the MassAI program, which has incorporated an updated version of CrossWork [46] using a score cutoff of 10. This resulted in 340 different PDI cross-links in 2581 observed scans (ms/ms spectra). To establish an overview of these results, they are presented as heat maps, colored according to the number of observed scans (Fig. 2, Supplementary Fig. 2). A large part of these cross-links is of low copy number and most likely represent short-term spatial arrangements of PDI, PDI-PDI interactions or false positives. For this reason we have focused on ‘hot spots’, i.e. cross-links observed in more than 10 scans. The present results have been compared to the full-length structure of oxidized hPDI structure, PDB code 4EL1 [7,19]. This structure is based on recombinant PDI expressed without the signal peptide, but with an N-terminal purification tag. However, as the PDI used in the current analysis was purified from human placenta, there may be slight structural differences. We also chose to compare to the oxidized form, as this has been analyzed in a tris–HCl, pH 8.0 buffer, while the reduced PDI (PDB 4EKZ) has been analyzed in the presence of 50 mM DTT. When compared to the reduced form (distance map in Supplementary Fig. 4), only minor differences can be seen, and no additional cross-link mapping can be assigned. Comparing the combined cross-links (Fig. 2) to the lysine Cα distances in PDI (Fig. 4A) it can be observed that all observed intra-domain hot-spot linkages fit with the 4EL1 structure. However, when comparing interactions between domains, only 10 out of 27 cross-links are still supported by the monomer model. The non-explained cross-links are most likely explained by the occurrence of a dimer, which is supported by the SDS PAGE gel (Fig. 1). In the X-ray crystal of hPDI (4EL1), two PDI molecules co-crystallize in a concave fashion. A dimerized yPDI was also suggested by Tian and co-workers [21] which also showed a dimer with the a domain nestled in the concave cavity of the opposite PDI. We subsequently modeled the hPDI sequence into the yeast dimer and calculated lysine–lysine distances.

D

677

78

The combined use of all the methods mentioned above has shown PDI to contain a large hydrophobic pocket suggesting domain b′ as the major substrate binding site [38,63–65]. Tian G et al. [18], revealed that at 4 °C the four thioredoxin-like domains of yPDI are arranged in a twisted ‘U’ shape with domains a and a′ facing each other and the inside surface of the ‘U’ shape containing a high abundance of hydrophobic residues, thus providing binding sites for substrate proteins, while at 22 °C the domain a rotates 123° around the junction between domains a and b, and hence the four domains form a ‘boat’ shape. It was demonstrated that the junction between domains a and b is even more flexible than the one between domains b′ and a′, which was critical for the enzymatic activity both in vivo and in vitro. Wang C et al. [19] revealed that the four domains of hPDI were arranged as a horseshoe shape, with two CGHC active sites, located in domains a and a′ respectively, facing each other, but hPDI exists in different conformations under oxidized and reduced states. The oxidized human PDI exists in an open state with more exposed hydrophobic areas and a larger cleft with potential for substrate binding, in which the four domains stay in the same plane, and the two active sites are 40.3 Å apart, while the reduced hPDI exists in a closed state, in which domains a, b and b′ line up in the same plane, whereas domain a′ twists around 45° out, and the distance between the two active sites is 27.6 Å. The redox-regulated open/closed conformations facilitate the protein's binding capacities for versatile targets. These conformation changes are facilitated by the mobility around the linker region [66]. Here we have analyzed hPDI co-purified with calnexin and ERp72 by chemical cross-linking and mass spectrometry. We have used the lysine-reacting compound bis[sulfosuccinimidyl] suberate (BS3) which cross-links amino groups (i.e. N-terminus and lysine residues). A feature of cross-linking with BS3 is that the reaction is relatively slow with reaction times of more than 1/2 h (in this experiment BS3 was allowed to react for one hour). If the protein has a very stable structure, we will observe specific interactions, i.e. only relatively few cross-links. However, if the structure is flexible, intra-domain cross-links are expected to dominate, as the linking points have a relatively stable position relative to each other, while inter-domain cross-links are less frequent if the domains are flexible and have active movements relative to each other while the reaction takes place. Another bias that has to be considered, is that high-mass peptides are disfavored in cross-linked peptides (Fig. 3A), which however

14_911

E

676

268_819

47

T

675

191_752

28

C

674

14_772

28

E

673

14_644

9

R

672

1_644

R

671

1_652

46

O

670

54_515

4EL1_01827

C

669

Model number

N

t3:4

Table 3 – Cross-links observed in each of the 3 different solutions for dimeric PDI obtained by modeling in RosettaDock. The Cα distances between cross-linked residues more than 30 Å are shown in black, whereas the ones less than 30 Å are highlighted in red.

U

t3:1 t3:2 t3:3



714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758

13

T

E

D

P

R O

O

F


762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783

R

761

However, as neither model could satisfy our cross-linking results, it led us to model alternative PDI dimer models using Rosettadock. We identified 10 cross-links that could not be accommodated by the monomer structure and were determined to be reliable and stable (Table 2). We generated 20,000 models and sorted them according to their fit to the cross-links (Fig. 5). This resulted in three modalities (Fig. 6) showing that instead of the dimer models, where the a domain of one PDI molecule is nestled inside the U form of another PDI molecule (Fig. 6A), we have interaction between the outside of the a–b domains. Either domain a against b in an antiparallel fashion (Fig. 6B and D), or slightly cross-wise centered on the a–b linker region (Fig. 6C). None of the dimer models satisfied all of the ‘missing’ cross-links (Table 3), indicating that we have a number of different interactions. Furthermore, we are likely to have alternative conformations of PDI, as the present modeling was performed based on a rigid-body docking that did not take intra-chain flexibility into consideration. As a large number of the observed cross-links indicate that we have considerable flexibility, it is likely that this could satisfy the missing cross-links. As the crystal structure of the yPDI showed distinct differences in structure when crystallized at 0 °C and at room temperature, we performed the cross-linking experiment at 0 and 37 °C. Based on the heat maps (Supplementary Fig. 2A and

N C O

760

U

759

R

E

C

Fig. 7 – Inter-cross-linking sites on calnexin model 1JHN. The general structure is shown as backbone, lysine residues are displayed in space filling mode. Low scan number cross-linking lysines are shown in blue. High scan number lysine residues unique for cross-links to PDI are shown in green, unique for ERp72 in cyan, common for both PDI and ERp72 are shown in orange.

2C) it can be concluded, that there are no great differences in the solution structures at either temperature. The main difference observed is a markedly lower cross-linking ability of residues Lys81 and Lys208 at 0 °C, which is likely caused by a stronger stabilization by salt bridges to glutamic acid residues at low temperature. A large number of inter-protein cross-links were also observed between the major components of the fractions (PDI, calnexin and ERp72). For all interactions involving calnexin, the cross-linking was clear to the surface of the molecule containing the peptide and carbohydrate binding area [49,67] (Fig. 7), thus confirming this part as the main peptide interaction region. For PDI the interactions with ERp72 and calnexin were mostly similar, with the a domain being by far the strongest cross-linker (Fig. 8). For ERp72 the situation was different, as the main interaction to PDI shared only part of the interaction with calreticulin. Although most of the participating lysines were cross-linking for both proteins, the distribution of the intensities was quite different with the interaction to calreticulin being more localized, in agreement with calreticulin being a smaller molecule with a single globular domain. The binding residues were the same, particularly in the a domain, but the distribution of linked residues was slightly different as observed for calreticulin, probably reflecting the size of the binding proteins.


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

14

D

P

R O

O

F


815 816 817 818 819 820 821 822 823

E

R

814

R

813

O

812

C

811

Although there seems to be no reports on a biological interaction of PDI, ERp72 and calnexin, the fact that they co-purify through a lengthy purification process, indicates that the interaction is both specific and stable. In our purification of placental chaperones, we have observed other protein mixtures, which may represent similar complexes. Another question is whether it is possible to analyze the binding PDI to substrate proteins by CXMS. In this case the interaction is expected to be rather short-lived, and the use of a fast reacting cross-linker is thus necessary. The most likely choice would be photoreactive cross-linkers, however, they are quite unspecific, which will lead to a lower yield of the individual cross-link and a much larger search space for the search program. Combined with the large size of PDI, this seems to be out of reach by the current technology.

N

810

U

809

C

T

E

Fig. 8 – Inter cross-linking sites on monomeric PDI model 4EL1. The general structure is shown as backbone, lysine residues are displayed in space-filling mode. Low scan number cross-linking lysines are shown in blue. High-scan number lysine residues unique for cross-links to calnexin are shown in green, unique for ERp72 in cyan, common for both calnexin and ERp72 in orange. The a domain is shown at the bottom right.

825 824

5. Conclusion

826

The current results show that it is possible to generate a wealth of data using standard cross-linking techniques when combined with sensitive mass spectrometric analysis and detection software. For the analyzed hPDI protein an X-ray structure of the intact protein exists, and the current experiments are for a large part consistent with the model. PDI is a very flexible molecule as the a and a′ domains can move through large angles relative to the b and b′ domains. X-ray experiments have shown a large variation in the orientation of the a and a′ domains relative to the b and b′

827 828 829 830 831 832 833 834 835

domains, both regarding temperature (for yPDI) and oxidation state (for hPDI). Based on the current results, we hypothesize that the different conformations observed in X-ray structures may be ‘frozen’ conformations that may be entered under most/all conditions, and which happen to be stabilized under the specific crystallization conditions used. Furthermore, the structure of the yeast dimer is unlikely to be representative of the hPDI dimer. Although the substrate interaction generally has been located to the b′ domain, the present results indicate that the in-solution homo-dimer does not bind in this area, but in the domain a–b interface. Our data thus suggest a convex dimer interaction rather than a concave. Interaction was observed between all the major components in the mixture. While the interaction to calnexin was clearly localized, the interaction to PDI was more diffuse over the a′ and part of the b′ domain. Interaction to Erp72 was difficult to determine, as a complete structural model is not available yet. In conclusion, cross-linking in combination with mass spectrometry is a valuable supplement to other structural methods and can be used to obtain information about protein structure, flexibility and interaction, as exemplified here for PDI, calnexin and ERp72.

836

Transparency document

860 859

837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858

The Transparency document associated with this article can 861 be found, in the online version. 862


15


865 Q4

868

The Danish Agency for Science, Technology and Innovation is gratefully acknowledged for financial support (LP). Sanne G. Boelt is acknowledged for scientific discussions. Dorthe T. Olsen and Elisabeth V. Jakobsen are acknowledged for technical assistance.

870 869

Appendix A. Supplementary data

871 872

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.jprot.2014.04.037.

87 Q73

REFERENCES

874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923

[1] Trincao J, et al. Dynamic structural science: recent developments in time-resolved spectroscopy and X-ray crystallography. Biochem Soc Trans 2013;41(5):1260–4. [2] Allen JP, Seng C, Larson C. Structures of proteins and cofactors: X-ray crystallography. Photosynth Res 2009;102(2–3):231–40. [3] Chayen NE, Saridakis E. Protein crystallization: from purified protein to diffraction-quality crystal. Nat Methods 2008;5(2):147–53. [4] Ilari A, Savino C. Protein structure determination by X-ray crystallography. Methods Mol Biol 2008;452:63–87. [5] Rossmann MG, et al. Combining X-ray crystallography and electron microscopy. Structure 2005;13(3):355–62. [6] Pusey ML, et al. Life in the fast lane for protein crystallization and X-ray crystallography. Prog Biophys Mol Biol 2005;88(3):359–86. [7] Alanen HI, et al. Defining the domain boundaries of the human protein disulfide isomerases. Antioxid Redox Signal 2003;5(4):367–74. [8] Darby NJ, Creighton TE. Functional properties of the individual thioredoxin-like domains of protein disulfide isomerase. Biochemistry 1995;34(37):11725–35. [9] Freedman RB, et al. Experimental and theoretical analyses of the domain architecture of mammalian protein disulphide-isomerase. Biol Chem 1998;379(3):321–8. [10] Hatahet F, Ruddock LW. Protein disulfide isomerase: a critical evaluation of its function in disulfide bond formation. Antioxid Redox Signal 2009;11(11):2807–50. [11] Venetianer P, Straub FB. The enzymic reactivation of reduced ribonuclease. Biochim Biophys Acta 1963;67:166–8. [12] Frand AR, Cuozzo JW, Kaiser CA. Pathways for protein disulphide bond formation. Trends Cell Biol 2000;10(5):203–10. [13] Freedman RB. Protein disulfide isomerase: multiple roles in the modification of nascent secretory proteins. Cell 1989;57(7):1069–72. [14] Jordan PA, Gibbins JM. Extracellular disulfide exchange and the regulation of cellular function. Antioxid Redox Signal 2006;8(3–4):312–24. [15] Tager M, et al. Membrane-bound proteindisulfide isomerase (PDI) is involved in regulation of surface expression of thiols and drug sensitivity of B-CLL cells. Exp Hematol 1997;25(7):601–7. [16] Peaper DR, Cresswell P. Regulation of MHC class I assembly and peptide binding. Annu Rev Cell Dev Biol 2008;24:343–68. [17] Janiszewski M, et al. Regulation of NAD(P)H oxidase by associated protein disulfide isomerase in vascular smooth muscle cells. J Biol Chem 2005;280(49):40813–9.

R O

P

D

E

T

C

E

R

R

N C O

867

U

866

[18] Tian G, et al. The crystal structure of yeast protein disulfide isomerase suggests cooperativity between its active sites (vol 124, pg 61, 2006). Cell 2006;124(5):1085–8. [19] Wang C, et al. Structural insights into the redox-regulated dynamic conformations of human protein disulfide isomerase. Antioxid Redox Signal 2013;19(1):44–53. [20] Kozlov G, et al. Crystal structure of the bb′ domains of the protein disulfide isomerase ERp57. Structure 2006;14(8):1331–9. [21] Tian G, et al. The catalytic activity of protein-disulfide isomerase requires a conformationally flexible molecule. J Biol Chem 2008;283(48):33630–40. [22] Wang C, et al. Human protein-disulfide isomerase is a redox-regulated chaperone activated by oxidation of domain a′. J Biol Chem 2012;287(2):1139–49. [23] Billeter M, Wagner G, Wuthrich K. Solution NMR structure determination of proteins revisited. J Biomol NMR 2008;42(3):155–8. [24] Bonvin AMJJ, Boelens R, Kaptein R. NMR analysis of protein interactions. Curr Opin Chem Biol 2005;9(5):501–8. [25] Foster MP, McElroy CA, Amero CD. Solution NMR of large molecules and assemblies. Biochemistry 2007;46(2):331–40. [26] Kay LE. NMR studies of protein structure and dynamics. J Magn Reson 2005;173(2):193–207. [27] Neira JL. NMR as a tool to identify and characterize protein folding intermediates. Arch Biochem Biophys 2013;531(1–2):90–9. [28] Shapiro YE. NMR spectroscopy on domain dynamics in biomacromolecules. Prog Biophys Mol Biol 2013;112(3):58–117. [29] Shi Y, Wu J. Structural basis of protein–protein interaction studied by NMR. J Struct Funct Genomics 2007;8(2–3):67–72. [30] Shin J, Lee W, Lee W. Structural proteomics by NMR spectroscopy. Expert Rev Proteomics 2008;5(4):589–601. [31] Sibille N, Bernado P. Structural characterization of intrinsically disordered proteins by the combined use of NMR and SAXS. Biochem Soc Trans 2012;40:955–62. [32] Tzakos AG, et al. NMR techniques for very large proteins and RNAs in solution. Annu Rev Biophys Biomol Struct 2006;35:319–42. [33] Darby NJ, Penka E, Vincentelli R. The multi-domain structure of protein disulfide isomerase is essential for high catalytic efficiency. J Mol Biol 1998;276(1):239–47. [34] Dijkstra K, et al. Assignment of 1H, 13C and 15 N resonances of the a′ domain of protein disulfide isomerase. J Biomol NMR 1999;14(2):195–6. [35] Kemmink J, et al. Structure determination of the N-terminal thioredoxin-like domain of protein disulfide isomerase using multidimensional heteronuclear 13C/15 N NMR spectroscopy. Biochemistry 1996;35(24):7684–91. [36] Kemmink J, et al. Nuclear magnetic resonance characterization of the N-terminal thioredoxin-like domain of protein disulfide isomerase. Protein Sci 1995;4(12):2587–93. [37] Kemmink J, et al. The structure in solution of the b domain of protein disulfide isomerase. J Biomol NMR 1999;13(4):357–68. [38] Nguyen VD, et al. Alternative conformations of the x region of human protein disulphide-isomerase modulate exposure of the substrate binding b′ domain. J Mol Biol 2008;383(5):1144–55. [39] Bernado P, Svergun DI. Structural analysis of intrinsically disordered proteins by small-angle X-ray scattering. Mol Biosyst 2012;8(1):151–67. [40] Blanchet CE, Svergun DI. Small-angle X-ray scattering on biological macromolecules and nanocomposites in solution. Annu Rev Phys Chem 2013;64:37–54. [41] Kathuria SV, et al. Minireview: structural insights into early folding events using continuous-flow time-resolved small-angle X-ray scattering. Biopolymers 2011;95(8):550–8.

F

Acknowledgements

O

864 863


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

16

[59]

[60]

[61]

[62]

E

D

[63]

[64]

C

E

[65] [66]

[67]

U

N

C

O

R

1076

F

[58]

O

[57]

R O

[56]

L-selectin shedding from leukocytes. J Immunol 2000;164(8):4120–9. Cho J, et al. A critical role for extracellular protein disulfide isomerase during thrombus formation in mice. J Clin Invest 2008;118(3):1123–31. Jasuja R, Furie B, Furie BC. Endothelium-derived but not platelet-derived protein disulfide isomerase is required for thrombus formation in vivo. Blood 2010;116(22):4665–74. Jasuja R, et al. Protein disulfide isomerase inhibitors constitute a new class of antithrombotic agents. J Clin Invest 2012;122(6):2104–13. Reinhardt C, et al. Protein disulfide isomerase acts as an injury response signal that enhances fibrin generation via tissue factor activation. J Clin Invest 2008;118(3):1110–22. Hoffstrom BG, et al. Inhibitors of protein disulfide isomerase suppress apoptosis induced by misfolded proteins. Nat Chem Biol 2010;6(12):900–6. Uehara T, et al. S-nitrosylated protein-disulphide isomerase links protein misfolding to neurodegeneration. Nature 2006;441(7092):513–7. Walker AK, Atkin JD. Mechanisms of neuroprotection by protein disulphide isomerase in amyotrophic lateral sclerosis. Neurol Res Int 2011;2011:317340. Walker AK, et al. Protein disulphide isomerase protects against protein aggregation and is S-nitrosylated in amyotrophic lateral sclerosis. Brain 2010;133:105–16. Klappa P, et al. The b′ domain provides the principal peptide-binding site of protein disulfide isomerase but all domains contribute to binding of misfolded proteins. EMBO J 1998;17(4):927–35. Byrne LJ, et al. Mapping of the ligand-binding site on the b′ domain of human PDI: interaction with peptide ligands and the X-linker region. Biochem J 2009;423(2):209–17. Denisov AY, et al. Solution structure of the bb′ domains of human protein disulfide isomerase. FEBS J 2009;276(5):1440–9. Wang C, et al. Plasticity of human protein disulfide isomerase: evidence for mobility around the X-linker region and its functional significance. J Biol Chem 2010;285(35):26788–97. Pocanschi CL, et al. Structural and functional relationships between the lectin and arm domains of calreticulin. J Biol Chem 2011;286(31):27266–77.

P

[55]

T

[42] Mertens HDT, Svergun DI. Structural characterization of proteins and complexes using small-angle X-ray solution scattering. J Struct Biol 2010;172(1):128–41. [43] Petoukhov MV, Svergun DI. Applications of small-angle X-ray scattering to biomacromolecular solutions. Int J Biochem Cell Biol 2013;45(2):429–37. [44] Li SJ, et al. Annular arrangement and collaborative actions of four domains of protein-disulfide isomerase — A small angle X-ray scattering study in solution. J Biol Chem 2006;281(10):6581–8. [45] Olsen DT, et al. Purification and characterization of a soluble calnexin from human placenta. Protein Expr Purif 2013;92(1):105–11. [46] Rasmussen MI, et al. CrossWork: Software-assisted identification of cross-linked peptides. J Proteomics 2011;74(10):1871–83. [47] Sali A, Blundell TL. Comparative protein modelling by satisfaction of spatial restraints. J Mol Biol 1993;234(3):779–815. [48] Hojrup P, Roepstorff P, Houen G. Human placental calreticulin characterization of domain structure and post-translational modifications. Eur J Biochem 2001;268(9):2558–65. [49] Chouquet A, et al. X-ray structure of the human calreticulin globular domain reveals a peptide-binding area and suggests a multi-molecular mechanism. PLoS One 2011;6(3):e17886. [50] Kozlov G, et al. Structure of the catalytic a(0)a fragment of the protein disulfide isomerase ERp72. J Mol Biol 2010;401(4):618–25. [51] Gallina A, et al. Inhibitors of protein-disulfide isomerase prevent cleavage of disulfide bonds in receptor-bound glycoprotein 120 and prevent HIV-1 entry. J Biol Chem 2002;277(52):50579–88. [52] Ryser HJ, Fluckiger R. Progress in targeting HIV-1 entry. Drug Discov Today 2005;10(16):1085–94. [53] Lovat PE, et al. Increasing melanoma cell death using inhibitors of protein disulfide isomerases to abrogate survival responses to endoplasmic reticulum stress. Cancer Res 2008;68(13):5363–9. [54] Bennett TA, et al. Sulfhydryl regulation of L-selectin shedding: phenylarsine oxide promotes activation-independent

R

993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 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

Probing the protein interaction network of Pseudomonas aeruginosa cells by chemical cross-linking mass spectrometry.

Compact conformations of human protein disulfide isomerase.

Resolution of protein structure by mass spectrometry.

Structural insight into the dimerization of human protein disulfide isomerase.

Protein disulfide isomerase a multifunctional protein with multiple physiological roles.

Mitochondrial protein interactome elucidated by chemical cross-linking mass spectrometry.

Formation of enzyme-substrate disulfide linkage during catalysis by protein disulfide isomerase.

Beyond gene sequencing: analysis of protein structure with mass spectrometry.

In-Depth Characterization of Protein Disulfide Bonds by Online Liquid Chromatography-Electrochemistry-Mass Spectrometry.

The flexibility and dynamics of protein disulfide isomerase.

Oxidative protein labeling with analysis by mass spectrometry for the study of structure, folding, and dynamics.

Active hydrogen by chemical ionization mass spectrometry.

Application of electrospray mass spectrometry in probing protein-protein and protein-ligand noncovalent interactions.

Structure of the substrate-binding b' domain of the Protein Disulfide Isomerase-Like protein of the Testis.

Combined ligand-observe (19)F and protein-observe (15)N,(1)H-HSQC NMR suggests phenylalanine as the key Δ-somatostatin residue recognized by human protein disulfide isomerase.

Global Membrane Protein Interactome Analysis using In vivo Crosslinking and Mass Spectrometry-based Protein Correlation Profiling.

Protein disulfide isomerase appears necessary to maintain the catalytically active structure of the microsomal triglyceride transfer protein.

deuterium exchange mass spectrometry.

Advances in protein complex analysis by chemical cross-linking coupled with mass spectrometry (CXMS) and bioinformatics.

Identification of protein disulfide isomerase 1 as a key isomerase for disulfide bond formation in apolipoprotein B100.

Emerging roles of protein disulfide isomerase in cancer.

Protein disulfide isomerase A6 controls the decay of IRE1α signaling via disulfide-dependent association.

Protein surface topology-probing by selective chemical modification and mass spectrometric peptide mapping.

Monitoring of human chemical signatures using membrane inlet mass spectrometry.