Differential tissue expression of extracellular vesicle‐derived proteins in prostate cancer

Abstract Background Proteomic profiling of extracellular vesicles (EVs) from prostate cancer (PCa) and normal prostate cell lines, led to the identification of new candidate PCa markers. These proteins included the nuclear exportin proteins XPO1 (also known as CRM1), the EV‐associated PDCD6IP (also known as ALIX), and the previously published fatty acid synthase FASN. In this study, we investigated differences in expression of XPO1 and PDCD6IP on well‐characterized prostate cancer cohorts using mass spectrometry and tissue microarray (TMA) immunohistochemistry to determine their diagnostic and prognostic value. Methods Protein fractions from 67 tissue samples (n = 33 normal adjacent prostate [NAP] and n = 34 PCa) were analyzed by mass spectrometry (nano‐LC‐MS‐MS). Label‐free quantification of EVs was performed to identify differentially expressed proteins between PCa and NAP. Prognostic evaluation of the candidate markers was performed with a TMA, containing 481 radical prostatectomy samples. Samples were stained for the candidate markers and correlated with patient information and clinicopathological outcome. Results XPO1 was higher expressed in PCa compared to NAP in the MS data analysis (P > 0.0001). PDCD6IP was not significantly higher expressed (P = 0.0501). High cytoplasmic XPO1 staining in the TMA immunohistochemistry, correlated in a multivariable model with high Gleason scores (P = 0.002) and PCa‐related death (P = 0.009). Conclusion High expression of cytoplasmic XPO1 shows correlation with prostate cancer and has added clinical value in tissue samples. Furthermore, as an extracellular vesicles‐associated protein, it might be a novel relevant liquid biomarker.


| INTRODUCTION
Biomarker discovery via extracellular vesicles (EVs; often referred to as exosomes) released by (cancer) cells, has been the focus of many research groups in the last decade. 1,2 Based on their biogenesis and secretion pathway, they contain low-abundant, cancer-specific proteins, and RNAs that could be of interest in identifying novel biomarkers. 3 With respect to prostate cancer (PCa), several EV-derived candidate biomarkers have been revealed. [4][5][6][7][8][9][10] Although multiple markers have been proposed as candidates for several malignancies, the majority has been identified and validated in EVs derived from cell culture. Few of the candidate biomarkers have been validated on larger groups of patient samples. Because this validation step is rarely taken, it remains difficult to elucidate the full potential of EV markers, which limits its translation and clinical implementation. 11,12 Our own efforts, by using state-of-the-art mass spectrometry, has led to the discovery of some candidate markers of which XPO1 (also known as CRM1), FASN, and PDCD6IP (also known as ALIX) were found to have the highest potential. 7 The objective of this study was to investigate whether the PCa EV-associated expression could be reproduced in tissue analyses of larger cohorts of patients. Result for FASN has been published previously. 13 2 | MATERIALS AND METHODS

| Mass spectrometry
Protein fractions from tissue RNA isolations with RNA-Bee of 67 PCa tissue samples (33 NAP and 34 PCa) were selected and stored at −80°C as described in Rodriguez et al. 13 Samples were thawed and 50 µL precipitated with cold acetone and microcentrifugation. After A total of 5 µL was diluted 40 times and subsequently transferred to LC vials for LC-MS analysis. Upon analysis, 2 µL was injected to the nano-LC. After preconcentration and washing of the sample it was loaded on to a C18 column (PepMap C18, 75 mm ID × 500 mm, 2 μm particle, and 100 Å pore size; Thermo Fisher Scientific, Bremen, Germany) using a linear 90 minutes gradient (4%-25% acetonitrile/ H20; 0.1% formic acid) at a flow rate of 250 nL/minute. The separation of the peptides was monitored by a UV detector (absorption at 214 nm). The nano-LC was coupled to a nanospray source of a Q-Exactive plus mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). Full scan MS spectra (m/z 400-1600) in profile mode were acquired in the Orbitrap with a resolution of 70 000 after the accumulation of an AGC target of 1 × 10 6 . The top 12 peptide signals (charge-state 2 + and higher) were isolated (1.6 Da window) and fragmented by HCD (higher-energy collision, normalized collision energy 28.0) and measured in the Orbitrap with an AGC target of 50 000 and a resolution of 17 500. Maximum fill times were 100 ms for the full scans and 60 ms for the MS/MS scans. The dynamic exclusion was activated, after the first time a precursor was selected for fragmentation it was excluded for a period of 30 seconds using a relative mass window of 10 ppm. Lock mass correction was activated to improve mass accuracy of the survey scan.
Label-free quantitation was performed using Progenesis LC-MS Software (version 3.0; Nonlinear Dynamics Ltd., Newcastle-upon-Tyne, UK) following our previously reported methodology. 14,15 To get quantitative data, we selected only proteins identified by three or more peptides for statistical analysis of protein abundance between groups. Duplicates in identified sequences as a consequence of peak tailing were removed to avoid false positives. Technical replicates of each sample were randomly analyzed within the measurement period and no significant changes in the number of identified proteins were observed between replicates and quality control measurements.

| Tissue microarray
A tissue microarray (TMA) was constructed as published previously. 16 Briefly, 481 men were selected from the European Randomized Study of Screening for prostate cancer (ERSPC), who had undergone radical prostatectomy for PCa. 17 From each patient sample, three representative cores (diameter 0.6 mm) were taken and placed in nine paraffin blocks. Patient information and clinical follow data were recorded via the ERSPC protocol and stored in a central study database.

| 1033
Based on previous IHC staining results, for XPO1 a score was assigned to both nuclear staining and cytoplasmic staining. 7 For PDCD6IP only the cytoplasmic expression was scored. In cases of staining heterogeneity, the highest expression levels were used for statistical analysis. After scoring, the average intensity for the triplicate cores was calculated. When a core was missing or no cancer was observed, this respective case was excluded from the analysis. In a combined session consensus on expression value was reached in all cases.
Statistical association of staining intensities and clinic-pathological features (PSA at diagnosis, Gleason score (GS), pT-stage, surgical margins, biochemical recurrence, local recurrence, overall death, and PCa-related death) were performed with SPSS (version 17, SPSS Inc., Chicago, IL) by using Pearsonʼs χ 2 tests and Studentʼs t tests. A multivariable analysis was performed to determine the contribution of each individual variable. A P-< 0.05 was considered to be statistically significant.

| Protein expression by mass spectrometry
We previously published a list with proteins (n = 263) that were identified in EVs from normal prostate (PNT2C2 and RWPE-1) and PCa (VCaP and PC346c) cell lines by using mass spectrometry. 7 From this list, 10 proteins were identified as higher expressed in PCa-derived EVs of which expression of 3 proteins (XPO1, FASN, and PDCD6IP) were further analyzed for EV and tissue expression. For a first validation, we compared the 263 proteins to a shotgun mass spectrometry database recently published which 34 PCa (n = 22 GS 6, n = 12 GS≥ 7) and 33 NAP tissues were compared using label-free quantification. 19 In this database, a total of 2865 proteins were identified from which 798 proteins were statistically significant differentially expressed between normal prostate and PCa (FDR < 0.01). When compared to the list of EV-derived proteins, 42 of these proteins showed overlap ( Figure 1A and Table 1).
F I G U R E 1 A, Overlap of proteins between the discovery set of extracellular vesicle-associated proteins (n = 263) 7 and the proteins differentially expressed between prostate cancer (PCa) and normal adjacent prostate (NAP) tissue (n = 798). 13 B, Protein expression (LOG10 normalized) of XPO1 and PDCD6IP in NAP (n = 33) and PCa tissue (n = 34) and in C, Gleason score

| Protein expression by tissue microarray immunohistochemistry
For independent validation of XPO1 and PDCD6IP, immunohistochemistry (IHC) was performed on the TMA. Patient characteristics and clinicopathological parameters are shown in Table 2 and were previously published by Hoogland et al. 16  Subsequently, we evaluated the association between protein expression intensities of our three candidate biomarkers and PSA at diagnosis, GS, pT-stage, surgical margins, biochemical recurrence, local recurrence, overall death and PCa-related death (Tables 3,4; Table S1). Nuclear XPO1 expression and PDCD6IP did not correlate with any clinicopathological parameter. High cytoplasmic XPO1 expression correlated with GS ≥ 7 (P = 0.002) and PCa-specific death after multivariate analysis (P = 0.009) (

| DISCUSSION
In this study we investigated whether previously identified PCa EVderived proteins were differentially expressed in PCa tissue from patients using mass spectrometry and immunohistochemistry. We found that XPO1 was associated with PCa in mass spectrometry and with higher GS using IHC on our TMA. PDCD6IP was not associated with adverse clinicopathological characteristics.
XPO1 (also known as CRM1) mediates nuclear export of proteins and RNAs and its differential expression has been linked to multiple types of cancer. [20][21][22] These transported proteins play a role in tumor signaling pathways, including the AR-pathway. 23,24 XPO1 has already been identified as a marker for several malignancies. 21 We identified this protein to be higher expressed in EVs derived from the VCaP PCa T A B L E 3 Scored staining intensity and distribution of the cytoplasmic and nuclear XPO1 in our tissue microarray Intensity was scored as negative (0; no staining), weak (1; only visible at high magnification), moderate (2; visible at low magnification), or strong (3; striking at low magnification). Staining intensities were correlated with patient characteristics after radical prostatectomy Abbreviation: PSA, prostrate specific antigen.
T A B L E 4 Univariate and multivariate correlation of clinicopathological parameters and staining intensities of PDCD6IP, nuclear XPO1, and cytoplasmic XPO1. A P < 0.05 was considered as statistically significant When correlated to clinicopathological parameters, we observed a significant correlation between higher XPO1 cytoplasmic expression with higher GS (7 or higher) and disease-specific death.

PSA
Therefore, this finding implies that there seems to be a clinical role as a tissue marker regarding prognosis for PCa. The correlation with disease-specific death could only be addressed in 12 patients.

| CONCLUSIONS
In this study, we investigated previously-identified EV-derived markers on large cohorts of patient tissue samples for validation of diagnostic and prognostic differential expression. High expression of cytoplasmic XPO1 shows a strong correlation with PCa progression, while no differential tissue expression of PDCD6IP was observed.
The increase in cytoplasmic XPO1 during the progression of PCa can explain the higher abundance in secreted EVs.