Discovery and Qualification of Serum Protein Biomarker Candidates

Jul 16, 2019 - Cite This:J. Proteome Res.2019XXXXXXXXXX-XXX ... Label-free shotgun proteomics was performed on depleted serum samples from 30 ...
1 downloads 0 Views 2MB Size
Subscriber access provided by BUFFALO STATE

Article

Discovery and Qualification of Serum Protein Biomarker Candidates for Cholangiocarcinoma Diagnosis Kassaporn Duangkumpha, Thomas Stoll, Jutarop Phetcharaburanin, Puangrat Yongvanit, Raynoo Thanan, Anchalee Techasen, Nisana Namwat, Narong Khuntikeo, Nittaya Chamadol, Sittiruk Roytrakul, Jason Mulvenna, Ahmed Mohamed, Alok K. Shah, Michelle M. Hill, and Watcharin Loilome J. Proteome Res., Just Accepted Manuscript • DOI: 10.1021/acs.jproteome.9b00242 • Publication Date (Web): 16 Jul 2019 Downloaded from pubs.acs.org on July 17, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 37 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

Journal of Proteome Research

1

Discovery and Qualification of Serum Protein

2

Biomarker Candidates for Cholangiocarcinoma

3

Diagnosis

4 5

Kassaporn Duangkumpha, †,‡ Thomas Stoll, ¥ Jutarop Phetcharaburanin, †,‡Puangrat

6

Yongvanit, ‡ Raynoo Thanan, † Anchalee Techasen, ‡,ǁ Nisana Namwat, †,‡ Narong Khuntikeo,

7

‡,€

Nittaya Chamadol, ‡,Ꜫ Sittiruk Roytrakul, Ω Jason Mulvenna, ¥ Ahmed Mohamed, ¥ Alok K. Shah, ¥ Michelle M. Hill, ¥,* and Watcharin Loilome †,‡,*

8 9 †

10

Department of Biochemistry, Faculty of Medicine, Khon Kaen University, Khon Kaen,

11

Thailand ‡

12

Cholangiocarcinoma Research Institute, Khon Kaen University, Khon Kaen, Thailand; ¥

13 ǁ

14 15 16



QIMR Berghofer Medical Research Institute, Queensland, Australia

Faculty of Associated Medical Sciences, Khon Kaen University, Khon Kaen, Thailand

Department of Surgery, Faculty of Medicine, Khon Kaen University, Khon Kaen, Thailand Ꜫ

Department of Radiology, Faculty of Medicine, Khon Kaen University, Khon Kaen,

17 18

Thailand Ω

Proteomics Research Laboratory, Genome Institute, National Center for Genetic

19

Engineering and Biotechnology, National Science and Technology Development Agency,

20

Pathum Thani, Thailand

21 22

Corresponding authors:

23

Associate Professor Watcharin Loilome, Email: [email protected]

24

Associate Professor Michelle Hill, Email: [email protected] 1 ACS Paragon Plus Environment

Journal of Proteome Research 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

25

Page 2 of 37

ABSTRACT

26

Cholangiocarcinoma ( CCA) is a major health problem in northeastern Thailand. The

27

majority of CCA cases are clinically silent and difficult to detect at an early stage. Although

28

abdominal ultrasonography ( US) can detect pre- malignant periductal fibrosis ( PDF) , this

29

method is not suitable for screening populations in remote areas. With the goal of developing

30

a blood test for detecting CCA in the at-risk population, we carried out serum protein biomarker

31

discovery and qualification. Label-free shotgun proteomics was performed on depleted serum

32

samples from 30 participants (n=10 for US-normal, US-PDF and CCA groups). Of 40 protein

33

candidates selected using multiple reactions monitoring on 90 additional serum samples (n=30

34

per group) , 11 discriminatory proteins were obtained using supervised multivariate statistical

35

analysis. We further evaluated 3 candidates using ELISA and immunohistochemistry (IHC) .

36

S100A9, thioredoxin ( TRX) and cadherin- related family member 2 ( CDHR2) were

37

significantly different between CCA and normal, and CCA and PDF groups when measured in

38

an additional 247 serum samples (p0.9 Pearson’s correlation. Then, normalized peptide intensities were averaged to protein

162

intensity, and a log2 transformation was performed to obtain a near-normal distribution needed

163

for statistical tests using in the computing environment R (Table S6). For statistical analysis,

164

missing values were replaced by the minimum detected intensity for each peptide 27.

165 166

Bioinformatics and statistical analysis

167

Protein interaction network analysis was generated using STRING software based on

168

the STRING database and Gene Oncology ( GO) term. All of analyses were constructed and

169

visualized in SPSS 19. 0 ( IBM, USA) , GraphPad Prism 5 and R statistical software. The

170

ANOVA test was conducted with Shiny MixOmics online software ( http: / / mixomics-

7 ACS Paragon Plus Environment

Journal of Proteome Research 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

Page 8 of 37

171

projects. di. uq. edu. au/ Shiny) . Principal component analysis ( PCA) and orthogonal signal

172

correction projection to latent structures discriminant analysis (O-PLS-DA) was conducted in

173

SIMCA 15.0 (Umetrics, Sweden). Described and detailed of statistical analysis are available

174

in the Supplementary Methods.

175 176

Antibody-based methods

177

The following primary antibodies were used for indirect enzyme- linked

178

immunosorbent assay ( ELISA) and immunohistochemical staining ( IHC) : S100A9 ( Cat.

179

#ab24111) , TRX ( Cat. #ab185329) purchased from Abcam ( Cambridge, MA) and

180

CDHR2/PCLKC (Cat. #orb158119) purchased from Biorbyt (San Fransisco, CA). Detailed

181

methods for ELISA and IHC are available in the Supplementary Methods.

182 183

RESULTS

184

Overview of biomarker workflow and baseline characteristics of samples

185

We performed a multi- phased biomarker discovery and development study as

186

illustrated in Figure 1. Participant recruitment, ultrasound, blood sample collection and

187

biobanking were completed before biomarker discovery.

188

‘triangular’ biomarker study design for development of early cancer biomarkers

189

small cohort for discovery using shotgun proteomics to measure a broad range of proteins, then

190

increased the sample size while reducing the number of biomarker candidates. For the

191

discovery and qualification phases, we selected age-sex matched participants with ultrasound-

192

confirmed normal liver and PDF pathology. CCA cases were confirmed by pathology

193

diagnosis. In the qualification cohort, smoking status and alcohol consumption were

194

significantly different (Table 1 and Table S9). Orthogonal investigation using antibody-based

According to the proposed 28

, we used a

8 ACS Paragon Plus Environment

Page 9 of 37 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

Journal of Proteome Research

195

methods were conducted in independent samples using ELISA and IHC techniques on serum

196

samples and tumor micro tissue arrays, respectively.

197 198

Figure 1 Generalized workflow diagram for serum protein biomarker discovery. A; Serum

199

samples from respective patient groups were stored at - 80ºc until analysis. B; Discovery

200

samples (n=30) were depleted for the top 12 serum proteins and spiked with internal standard

201

protein. Tryptic peptides were analysed by label- free proteomics and MaxQuant software.

202

Biomarker candidates were selected after analysis with Shiny MixOmics. C; A custom multiple

203

reaction monitoring-mass spectrometry (MRM-MS) was developed for biomarker verification

204

in an independent cohort of 90 participants. Data processing and analysis used Skyline, R,

205

SIMCA and SPSS. D; Antibody-based assays were used to validate peptide level MS data for

206

selected candidates at the protein level in additional independent cohorts, n= 247 for ELISA,

207

n=208 for IHC.

208 209 210

9 ACS Paragon Plus Environment

Journal of Proteome Research 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

211

Page 10 of 37

Table 1 Baseline characteristics of serum samples in the discovery and qualification phases Discovery phase (N=30) Normal US

PDF US

CCA

Sample size

10

10

10

Gender Male/Female

5/5

5/5

5/5

Age in year (Median ± SD)

60 ± 11

60 ± 12

61 ± 11

Diagnosed with Diabetes Yes No Unknown*

1 (10%) 5 (50%) 4 (40%)

2 (20%) 8 (80%) -

Smoking status Yes No Unknown*

1 (10%) 4 (40%) 5 (50%)

Alcohol consumption Yes No Unknown*

2 (20%) 3 (30%) 5 (50%)

Qualification phase (N=90) p-value

Normal US

PDF US

CCA

30

30

30

1.000

15/15

15/15

15/15

1.000

1.000

63 ± 5

65 ± 6

64 ± 5

0.497

9 (90%) 1 (10%)

0.376 6 (20%) 4 (13%) 6 (20%) 24(80%) 26 (87%) 18 (60%) 6 (20%)

0.738

3 (30%) 7 (70%) -

4 (40%) 5 (50%) 1 (10%)

0.649 13 (43%) 6 (20%) 14 (47%) 13 (43%) 22 (73%) 10 (33%) 4 (14%) 2 (7%) 6 (20%)

0.042

4 (40%) 6 (60%) -

6 (60%) 3 (30%) 1 (10%)

0.449 18(60%) 8 (27%) 21 (70%) 8 (27%) 20 (67%) 3 (10%) 4 (13%) 2 (6%) 6 (20%)

P0. 9 ( Pearson’ s correlation test) were selected and 12 ACS Paragon Plus Environment

Page 13 of 37 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

Journal of Proteome Research

251

converted to protein intensities and log2 transformed. The qualification data set was then

252

subjected to multivariate statistical analysis (Figure 3). PCA with Pareto scaling showed a

253

clear separation of the CCA group from the normal and PDF groups (Figure 3A). The first two

254

principal components expressed 35.9% for PC1 and 21.9% for PC2 in the PCA model (Figure

255

3B). Orthogonal signal correction projection to latent structures discriminant analysis (O-PLS-

256

DA) was applied to conduct pair-wise comparisons of normal versus PDF, normal versus CCA,

257

and PDF versus CCA ( Figure S2) . Although normal and PDF groups were unable to be

258

separated, the CCA group showed clear separation from the normal as well as PDF groups. The

259

O- PLS- DA regression model confirmed the above visual observations, with significant CV-

260

ANOVA for the normal versus CCA comparison, PDF versus CCA comparison, but not for

261

the normal versus PDF comparison (Table 2).

262

The loadings of the pairwise O- PLS- DA model identified significantly higher

263

normalized protein intensities for 11 proteins: Haptoglobin ( HP) , Alpha- 1- antichymotrypsin

264

( A1AC / SERPIN3) , Complement component C9 ( C9) , Intercellular adhesion molecule 1

265

(ICAM1), Protein S100A9 (S100A9), Thioredoxin (TRX/TXN), Aminopeptidase N (ANPEP),

266

Fumarylacetoacetase (FAH), Lipopolysaccharide-binding protein (LBP), Inter-alpha-trypsin

267

inhibitor heavy chain H3 (ITIH3), Cadherin-related family member 2 (CHDR2/PCLKC) in the

268

CCA group when compared with the normal and PDF groups (Table 2). Data for all measured

269

proteins are provided in Table S7.

270

13 ACS Paragon Plus Environment

Journal of Proteome Research 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

Page 14 of 37

271 272 273

Figure 3 Biomarker qualification and multivariate analysis. A and B; PCA score plot and

274

loading plot of MRM results of candidates that shows sample differentiation. The three

275

significant candidate proteins are shown in red circles.

276 277 278 279

14 ACS Paragon Plus Environment

Page 15 of 37 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

280

Journal of Proteome Research

Table 2 Summary of O-PLS-DA qualified serum biomarker candidate proteins O-PLS-DA Model CCA (-) vs Normal (+) R2X = 56.7%; Q2Y = 0.645; p