Circulating proteomic signature for detection of biomarkers in bladder cancer patients

The identification of clinically-relevant early diagnostic and prognostic protein biomarkers is essential to maximize therapeutic efficacy and prevent cancer progression. The aim of the current study is to determine whether aberrant plasma protein profile can be applied as a surrogate tool for early diagnosis of bladder carcinoma. Plasma samples from patients with low grade non-muscle invasive bladder cancer and healthy controls were analyzed using combined 2D-DIGE and mass-spectrometry to identify differentially expressed proteins. Validation was performed using western blotting analysis in an independent cohort of cancer patients and controls. Fifteen differentially-expressed proteins were identified of which 12 were significantly up-regulated and three were significantly down-regulated in tumors compared to controls. The Ingenuity Pathways Analysis revealed functional connection between the differentially-expressed proteins and immunological disease, inflammatory disease and cancer mediated through chemokine and cytokine signaling pathway and NF-kB transcription factor. Among the three validated proteins, haptoglobin was able to distinguish between patients with low grade bladder cancer and the controls with high sensitivity and specificity (AUC > 0.87). In conclusion, several biomarker proteins were identified in bladder cancer. Haptoglobin is a potential candidate that merit further investigation to validate its usefulness and functional significance as potential biomarkers for early detection of bladder cancer.

It is well perceived that proteins are the key regulators that execute cellular functions. Therefore the proteomic signatures will most closely reflect the physiological and the pathological alterations that occur in the carcinogenic process 10,11 . Blood derivatives, either serum or plasma, have been considered as promising sources for developing proteomic protocols and the discovery of non-invasive biomarkers as blood samples are relatively easy to obtain and are rich with circulating aberrant proteins that can be used as tumor biomarkers 12 . Two dimensional gel-electrophoresis (2D-E) combined with mass spectrometry (MS) have evolved greatly and are capable of analyzing and identifying multiple differentially expressed proteins with high sensitivity and specificity 13,14 . Recent years have witnessed tremendous expansion in circulating biomarkers analysis by liquid chromatography/MS to develop bladder cancer biomarkers for diagnosis and/ or prognosis [15][16][17] . In this context several single biomarker molecules, with mechanistic relevance to bladder cancer, have been identified including complement C3b, CYFRA 21-1, ApoA4, ApoE and soluble E-cadherin [17][18][19] . Also, extensive research has been performed on multi-markers model which has led to the identification of panels of biomarkers that are differentially regulated in bladder carcinoma 20,21 . This approach may provide a comprehensive understanding of the complex mechanisms underlying bladder tumorogenesis. However there is still relative lack of comprehensive studies exploring the landscape of differentially-expressed proteins in bladder cancer. In this study, we used 2D-DIGE followed by MALDI-TOF mass-spectrometry to identify potential plasma biomarkers and explore proteome profile changes that differentiate between a cohort of patients with low grade non-muscle invasive bladder cancer and healthy controls which may help in the clinical setting and the management of BC.

Results
Samples. Plasma samples (discovery set and validation set) were grouped into two groups: the first one for patient with low grade non-muscle invasive tumors and the second group comprise healthy controls (Table 1). All plasma samples were depleted to reduce the complexity and enrich the proteins that might be used as biomarkers ( Supplementary Fig. 1). For such reason, plasma pre-fractionation strategy was undertaken using immuno-depletion kit that removes 20 of the most abundant proteins, including immunoglobulins and albumin, from the plasma and the final fractions were collected. This step removes more than 95% of the highabundance proteins in serum; hence increase the number of detectable spots. After depletion protein fractions from each of the tumor group and the healthy group were labeled with Cy3 and Cy5 Dyes respectively. In order to remove gel-to-gel variation in 2D-DIGE, control sample (internal standard) was labeled with Cy2 dye and run in parallel with the other gels.
Identification of candidate biomarkers. Comparisons of individual plasma samples (discovery set) from patients with low grade BC (n = 4 patients) and healthy controls (n = 4) were equally loaded, labeled and separated on 2D-DIGE. Differentially expressed protein spots were excised and identified by MALDI-TOF mass spectrometry. Analysis of the discovery cohort revealed a total of 980 spots consistently detected and mapped on all gels between the two groups by 2D-DIGE. Representative 2D-DIGE gels of each condition are shown in Fig. 2A-D. A total of 22 protein spots with a statistical significant change (ANOVA-test p < 0.05; foldchange ≥ 1.5) in abundance between the patients with low grade BC and control groups were identified. These spots were selected and further analyzed by MALDI TOF. 15 protein spots were successfully identified and matched using the MASCOT PMF to entries in the SWISS-PROT database with high confidence. Of the 15 spots identified, 12 spots were up and three down in patients with low grade bladder cancer in comparison to controls ( Table 2).
Validation of differentially expressed plasma proteins. Although several differentially-expressed proteins were identified only three proteins selected for further validation for many reasons. Most importantly, www.nature.com/scientificreports/ the selected proteins exhibited significant p-value of increase or decrease in cancer compared to healthy controls, because of the higher protein score and sequence coverage of ceruloplasmin, haptoglobin and C6 compared with the other proteins identified and also due to time restraint and shortage of technical staff. Furthermore, published data indicated that changes in serum levels of all three proteins are remarkably altered in solid tumours and are associated with poor prognosis. Western blotting analysis was carried out to validate the data using plasma samples from independent cohorts (validation set) of low grade bladder tumors (n = 18) and healthy controls (n = 10). We focused on three proteins: two down-regulated proteins (ceruloplasmin and C6) and one up-regulated protein (haptoglobin) because these proteins show significant p-value difference between cancer and healthy controls. The experiments were repeated three times for each target protein and the results gathered for all three proteins confirm the trend seen in 2D-DIGE analysis. Haptoglobin level was significantly increased (p = 0.0006) in low grade bladder cancer patients when compared to healthy control group (Fig. 3A). This is consistent with 2D-DIGE data ( Table 2). Western blots of plasma samples probed with the ceruloplasmin antibody exhibited the tendency observed in 2D-DIGE and showed reduced expression of the 140 kDa isoform in low grade bladder cancer patients when compared to healthy controls (Fig. 3B). Interestingly, extra fragment (80 kDa) was observed across all tumor patients with low 140 kDa ceruloplasmin isoform expression. The intensity of the band (80 kDa) was inversely proportional to the level of the140 kDa band. The 80 kDa isoform is weakly expressed in only two controls but completely absent in the rest of healthy controls ( Supplementary  Fig. 2). This suggests that the uncharacterized ceruloplasmin isoform (80 kDa) may arise from a pre-spliced transcript or through proteolytic cleavage of the original protein. Validation for complement C6 expression in an independent cohort (Fig. 3C) demonstrated significant reduction in C6 protein levels in low grade bladder cancer patients when compared to healthy controls (p = 0.046) which is consistent with the 2D-DIGE protein expression data. Equal protein loading was confirmed using Coomassie blue staining (Fig. 3D). Immunoblot band intensity for haptoglobin, ceruloplasmin and complement protein C6 was measured by densitometry using ImageJ software and presented as relative band intensity ( Supplementary Fig. 3).

Principal component analyses (PCA) analyses.
We then sought to investigate the total protein abundance variation between both cohorts. All gels images were grouped into two groups according to whether they have cancer or not and principal component analysis was performed using Progenesis SameSpots software. Statistically significant protein spots that were identified by MS were selected for such analysis. PCA plot of the two first principal components revealed that the two groups (cancer and controls) clustered distinctly from one another and the selected spots exhibit 52% abundance variability between the two groups ( Supplementary  Fig. 4A,B).  www.nature.com/scientificreports/ interaction and functional characteristics of key proteins. We then sought to use the Ingenuity Pathways Analysis (IPA), which accurately predicts for interacting partner proteins and also determines the signal transduction in various pathways, to identify the biological and molecular protein network related to our proteins of interest. The IPA revealed the interaction of the differentially expressed proteins in different networks and biological processes associated mainly with immunological disease, inflammatory disease and inflammatory response with highest score 17 (Fig. 4A). In this network, the transcription factor (NF-kB) as well as the cytokines interleukin-6 (IL-6) and interferon gamma (IFN-γ), which are known to play a pivotal regulatory role in carcinogenesis, are connected to three of the enriched proteins. Functional annotation was carried out to identify the molecular functions of the differentially expressed proteins. The result revealed that 30% of the proteins are involved in catalytic activities, 20% are involved in transporter activities, 20% are involved in enzyme regulator activities, 20% in receptor activities and 10% in binding activities (Fig. 4B). In addition, the majority of these proteins are components of the cellular compartment, the organelles or the extracellular matrix compartment (Fig. 4B).
Diagnostic performance of the plasma biomarkers. Among the three identified circulating serum protein biomarkers haptoglobin displayed the highest AUC > 0.87 (confidence interval 72-99%, p = 0.002) followed by C6 and ceruloplasmin with an AUC = 0.68 (confidence interval 46-89%, p = 0.39) and 0.68 (confidence interval 42-95%, p = 0.45) respectively ( Supplementary Fig. 5). The combination of the haptoglobin, ceruloplasmin and C6 did not yield a better diagnostic efficacy for low grade cancer patients compared to individual biomarker alone (data not shown).

Discussion
Bladder cancer (BC) is the second most common genitourinary neoplasm after prostate cancer, in men, with steady increase incidence rate 1 . This situation necessitates an urgent protocol to develop biomarkers for early detection and risk stratification. The recent histopathological classification of bladder cancer cases by the World Health Organization (2016) into low grade and high grade has major implications on the risk stratification and the management of bladder cancer patients 22,23 . Interestingly, progression of low-grade NMIBC to MIBC is frequent (up to 20% of cases) and is critically important in the management of BC patients 24 . The high recurrence rates together with the disease heterogeneity and the limited treatment options for bladder cancer patients advocate the development of alternative approaches for early cancer detection.

Healthy controls LG Tumours
A wide range of approaches have utilized for biomarker discovery including RNA-seq, micro-RNA (miR), circulating tumor DNA (ctDNA), proteomics, metabolomics, and intact circulating tumor cells [25][26][27][28][29] . Interesting new biomarkers of disease activity have been identified however biomarker validation is difficult and only few have made a significant clinical impact to date 30 .
The body biological fluids, such as serum and plasma, represent ideal and important tools for the development of novel biomarker molecules distinguishing between health and disease.
In the current study we combined both 2D-DIGE and mass-spectrometry to investigate the proteomic plasma signature from patients with low grade NMIBC and healthy controls. Several plasma biomarkers were identified some of which were significantly up-regulated (Hemopexin, Haptoglobin, Kelch domain-containing protein 8B, E3 ubiquitin-protein ligase HECTD, DNA (cytosine-5)-methyltransferase 3A, Complement C1r subcomponent, Immediate early response 3-interacting protein 1, Integrator complex subunit 10, Complement component C6) other were significantly down-regulated (Ceruloplasmin, Complement C3) in plasma of patients with low grade bladder cancer compared to healthy controls ( Table 2). Three of these proteins (Haptoglobin, Ceruloplasmin and Complement component C6) were selected for further validation in an independent cohort. Haptoglobin showed a significantly high expression in tumor samples compared to healthy controls (p = 0.0006) by western blotting. Three forms of haptoglobin have been identified which may indicate numerous potential splices variants and post-translational modification of haptoglobin. Our finding is consistent with previous ones reporting elevated haptoglobin levels in Turkish and Polish bladder cancer patients 17  www.nature.com/scientificreports/ Under normal physiological conditions Haptoglobin stabilize hemoglobin (Hb) through binding and forming a complex with it. Haptoglobin has also significant protective properties against oxidative damage by Hb 32 . Is has been demonstrated that plasma haptoglobin level increased in several solid tumors including breast cancer, colon cancer, pancreatic cancer, lung cancer and ovarian cancer where it is considered as an independent prognostic marker 33-38 . Our results indicated that the Complement component C6 protein expression was significantly down-regulated in patients with low grade tumors compared to control in the discovery set (p = 0.001) and the validation set (p = 0.046). To our knowledge this is the first study describing the association between elevated level of C6 and bladder cancer. Members of the complement system, consisting of over 50 plasma proteins, are potent effector of innate immunity involved in host defense 39 . Activation of Complement system leads to the generation of membrane attack complex. The latter requires the sequential assembly of complement proteins including C6, to form multi-protein pore involved in a direct lyses of a wide range of Gram-negative bacteria 40 . Recent experimental evidence indicated that activation of complement system plays an important role in carcinogenesis, however the exact mechanism is yet to be elucidated 41 .
As for ceruloplasmin, although the difference in protein expression level between tumors and controls did not reach significance, although there is a trend toward decrease protein expression in tumor compared to healthy controls (p = 0.29). Western blots of plasma samples probed with ceruloplasmin antibody showed expected 140 kDa isoform in healthy controls. Interestingly, cancer patients express another 80 kDa band that is differentially expressed across cancer patients. Patients expressing no or low levels of 80 kDa band show strong expression of the 140 kDa band whereas patients expressing high levels of 80 kDa band exhibit no expression of the 140 kDa band. The low molecular weight isoform is completely absent in healthy controls. This suggests that the 80 kDa isoform of the ceruloplasmin protein represent a posttranslational cleaved product or produced in a pre-spliced transcript. It may also arise from un-characterized ceruloplasmin isoforms. Ceruloplasmin is a secreted glycoprotein member of the multi-copper oxidase family involved in a wide variety of cellular processes mainly in copper transport and ion metabolism 42 . It also catalyzes the oxidation of polyamines, catecholamines and polyphenols. Ceruloplasmin loss-of-function is involved in neurodegenerative syndrome, whereas high ceruloplasmin level is associated with copper toxicity, Alzheimer's disease and cardiovascular diseases 43,44 . As ceruloplasmin levels were significantly high in cancer, ceruloplasmin has been suggested to be potential biomarkers in numerous solid tumors [45][46][47][48][49][50] . The relationship between ceruloplasmin level and bladder cancer has not been documented previously. In a meta-analysis by Mao and Huang, the authors reported a marked increase in serum copper level in patients with bladder cancer compared with controls which might be caused by excess of Cu-binding components, such as ceruloplasmin 51 . Taken together, our findings support the possible involvement of high plasma ceruloplasmin level in increasing copper levels in the blood in bladder cancer patients. The exact mechanism of ceruloplasmin action in bladder carcinogenesis is yet to be elucidated. Furthermore, network analysis using IPA identified cancer-associated network associated with the differentially expressed proteins. The data presented show that three of seven proteins (ceruloplasmin, Complement C3 and Complement C6) are functionally implicated in immune disease and inflammatory response, DNA replication and cellular assembly and organization. Deregulation of these proteins has also been implicated in cancer development and progression.

conclusion
We believe that differential plasma proteome analysis of patients with low grade BC provides new insights on disease pathophysiology. The current study sheds the light on new plasma marker proteins that could potentially help in early detection of bladder cancer. The level of haptoglobin was found to be highly elevated in low grade NMIBC patients, which point to the potential involvement of this protein in early stages of bladder cancer development. Its utility as a non-invasive biomarker tool in the context of bladder cancer warrant further in-depth investigation. There are several limitations to this study. Although the analysis was performed using an independent validation cohort with a freshly collected plasma samples, the results should be considered as preliminary given that the small sample size hence the results merit validation in a larger multicenter study. Also, the clinicopathological features such as smoking status, treatment and survival were not available at the time of the analysis which prevents us from conducting more relevant correlations and consistent comparison. The strips were then equilibrated and separated on 12.5% (SDS-PAGE) gels using an Ettan Dalt Six device. The gels were scanned with a Typhoon 9400 scanner using appropriate wavelengths and filters for Cy2, Cy3 and Cy5. Difference in-gel electrophoresis (DIGE) images were analyzed using the Progenesis SameSpots v.3.3 software (Nonlinear Dynamics Ltd., UK) as described previously 53 . All spots were pre-filtered and manually checked before applying the statistical criteria (ANOVA test, p ≤ 0.05 and fold ≥ 1.5). Normalized spot volumes, instead of spot intensities, were used in statistical processing. Only those spots that fulfilled the abovementioned statistical criteria were submitted for MS analysis.

Methods
Protein identification by MALDI-TOF mass spectrometry (MS). Coomassie stained gel spots were excised manually, washed and digested overnight, at 37 °C according to the previously described protocol 52 . The digests were transferred to a 0.5 mL tube and the peptides were extracted by the addition of 30% CH 3 CN containing 0.1% CF 3 COOH 54 . An aliquot of the digestion solution (1 μL) was spotted onto a MALDI target together with 0.5 μL of a matrix CHCA in 1 μL of 30% CH 3 CN and 0.1% aqueous CF 3 COOH and then left to dry before MS analysis. Protein identification was assigned by peptide mass fingerprinting (PMFs) using an ultrafleXtreme time-of-flight mass spectrometer and confirmed by MS/MS analysis of at least three peptides in each sample using LIFT spectra performed by MALDI-TOF-MS/MS. The MS data were interpreted by using BioTools v.3.2, together with the Mascot search algorithm. Identified proteins were accepted as correct if they showed a Mascot score greater than 56 and p < 0.05. pathway analysis. The quantitative data were then imported into IPA software (Ingenuity Systems, https ://www.ingen uity.com). This software aids in determining the functions and pathways most strongly associated with the protein list by overlaying experimental expression data on networks constructed from published interactions.
Western blot analysis. Protein samples were separated on 8% SDS-polyacrylamide gels. Proteins were transferred to nitrocellulose membranes and blocked with 5% non-fat milk then incubated with the primary antibodies overnight at 4 °C. Membranes were then incubated with HRP-labeled secondary antibodies and bands were visualized with enhanced chemiluminescence reagent. To ensure equal gel loading, equal amounts of plasma samples were separated using 8% PAGE gels and stained with Coomassie blue stain.
Statistical analysis. Statistical

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.